Effect of low-dose BPA on the nuclear localization of ERα in MDECs
Environmental chemicals, such as BPA, are known to act as estrogenic ligands that activate or deactivate gene transcription in breast epithelial cells (Soto et al., 2006
; Dairkee et al., 2008
). To determine whether BPA causes an estrogen-like effect, we performed immunofluorescence analyses in MDECs (un-exposed) transiently treated with different doses (ranges: 1 – 1,000 nM) of BPA at 0, 5, 30, 60, and 120 min (). BPA, as a weak estrogenic ligand, caused maximized ERα internalization at 30 min in a higher dose (1,000 nM) of exposure.
Fig. 1 Preexposure of MDECs to bisphenol A (BPA) and immunofluorescence analysis of nuclear ERα. (A) Subcellular localization of ERα in MDECs on acute BPA treatment. MDECs were exposed to BPA (1000 nM) for the indicated time periods. The observed (more ...)
Because prolonged exposure of breast progenitor cells to xenoestrogens also causes ERα internalization in their differentiated progeny (Hsu et al., 2009
), we determined whether BPA has this effect. Progenitor-containing mammospheres from an individual (#124) were continuously exposed to 4 nM BPA for 3 weeks. After the exposure, BPA was removed, and progenitor cells underwent epithelial differentiation in the collagen-coated dishes for 2–3 weeks. Immunofluorescence analysis showed an increase of ERα-positive population during the mammosphere and MDEC stages (). After 7-d preexposure, the majority (>90%) of mammospheres were ERα-negative (yellow bar in ). However, cell lineages were greatly shifted from ERα-negative to ERα-positive (green-plus-red bar, > 80%) after 42-d incubation. Among ERα-positive cells at 42-d, nuclear expression of ERα (green bar/green-plus-red bar, 78%) in BPA-preexoposed MDECs was increased compared to that of control (DMSO) cells (11.6%), suggesting that BPA preexposure contributes to ERα internalization in MDECs (). As a control, we also observed similar effect in E2 (70 nM)-preexposed MDECs.
When the analysis was extended to different primary MDECs (n = 9), we noticed individual variations in response to this low-dose BPA preexposure. As shown in , five (#96, 124, 98, 99, and 117) of these MDEC sets exhibited greater effects (up to 80%) of ERα internalization compared to the other four sets (#111, 120, 113, and 119) showing lesser effects (18 – 40%). This initial result suggests that as a weak estrogenic ligand, high-dose BPA (at least 1000 nM) is needed to acutely activate ERα-mediated signaling while chronic exposure of a lower dose (4 nM) can similarly bring about this signal transduction in breast epithelial cells. Furthermore, our observations indicate that the genetic background of individuals may influence differential responses to the exposure of low-dose BPA.
It is known that other exogenous stimulants, such as growth factors, may act through mitogen-activated protein kinase (MAPK) or Akt pathways to promote the nuclear internalization of ERα for transcriptional regulation in proliferating cells (Lannigan, 2003
; Murphy et al., 2009
). In this regard, we observed an increased level of phospho-p42/44 MAPK, likely attributed to this internalization in BPA-preexposed MDECs without further ligand stimulation ().
Effect of low-dose BPA on differential gene expression in MDECs
To investigate whether this effect altered gene expression, we conducted microarray analysis in ten sets of preexposed (BPA, 4 nM) and control MDECs using the Affymetrix Human Genome U133 Plus 2.0 Array. One set of the MDECs (#119) was removed from gene expression analysis because of the low nuclear localization of ERα. Differential expression of genes at P < 0.05 within 9 set samples was scored, yielding a total of 2,234 candidate loci (1,162 down-regulated and 1,072 up-regulated) likely influenced by this BPA preexposure. Scatter plots and the number of differentially expressed genes for individual MDECs are presented in . Consistent with the observation of ERα internalization, we observed individual variations of gene expression in these primary MDECs preexposed to low-dose BPA. In this regard, greater numbers of differentially expressed genes were seen in #124, 99, 100, and 128 (i.e., the high-responder group) while the rest of six primary MDECs had fewer changes of expression (i.e., the low- responder group). We additionally compared these expression profiles with the status of ERα internalization available for seven MDEC sets (#124, 99, 117, 120, 111, 113 and 119). Though not statistically significant, we observed a general trend that greater degrees of ERα internalization seemed to be associated with increased numbers of differentially expressed genes in MDECs.
Fig. 2 Gene expression profiling of ten sets of the MDECs. A total of genes in control (DMSO) and BPA-preexposed cells are shown in the scatter plot. The number of significant down-regulated (↓, green dots) and up-regulated (↑, red dots) genes (more ...)
Effect of BPA-influenced gene signatures in ERα-positive breast cancer
analysis was conducted to determine whether specific expression profiles of BPA-influenced genes are associated with the development of breast cancer. When the 2,234 candidate loci were compared with those of two microarray datasets, BCC48 (Neve et al., 2006
) and GSE2109 (International Genomics Consortium, http://www.intgen.org/expo
), we found a total of 170 BPA-influenced genes (57 up-regulated and 113 down-regulated), the aberrant expression of which may contribute to breast tumorigenesis (; see also Supplemental Table S2
). Hierarchical clustering of 48 breast cancer cell lines (i.e., BCC48) and 244 breast tumors (i.e., GSE2109) revealed that specific up- and down-regulated patterns of these 170 genes are distinctly related to ERα-positive cell lines (). This observation further indicates that 1) BPA may aberrantly regulate gene expression through an ERα-dependent pathway and 2) this regulatory mechanism may be epigenetically imprinted in ERα-positive breast cancer.
Fig. 3 Comparison of gene expression profiles in BPA-preexposed MDECs, 48 breast cancer cell lines (Neve et al. 2006) and breast tumor samples (GSE2109). (A) Venn diagram showing 170 common genes identified in separate analyses of the three different data sets, (more ...)
To validate this potential imprinting effect, we choose 15 down-regulated genes for expression analysis (). The reason to focus on these loci was that the BPA-influenced repression might be associated with hypermethylation of their CpG islands, which are located in the transcription start sites of these selected genes. First, RT-qPCR was used to confirm the expression status of these loci in the aforementioned six MDECs preexposed to BPA (4 nM). The expression of these loci was consistently down-regulated in three high-responders, #124, 99, and 128 (P < 0.05). Down-regulation of these loci, however, could not be confirmed in one high-responder (#100), likely attributed to a small sampling of down-regulated loci. Though this down-regulation was also seen in the low-responder group by the sensitive RT-qPCR assay, the repressive effect was usually less apparent (e.g., #120 and 113). In the rest of low-responders (#111, 129, 117, and 119), significant changes of expression between pre-exposed and control MDECs were not noted.
Fig. 4 Validation of differentially expressed loci by RT-qPCR. Gene-specific RT-qPCR on 10 independent sets of DMSO and BPA-preexposed MDECs was conducted to validate 15 down-regulated loci. Data were analyzed by ΔΔCt method using 36B4 as the (more ...)
Epigenetic repression of a BPA-influenced locus, LAMP3, in ERα-positive breast cancer cells
To further investigate a potential role of epigenetic repression, we focused the expression analysis on a candidate gene, lysosomal-associated membrane protein 3
), in the well-characterized ERα-positive MCF-7 cell line. (see functional analysis of this gene in MCF-7 cells in supplemental Figure S1
). A low level of LAMP3
expression was detected in MCF-7 cells. To determine whether this reduced expression is mediated by epigenetic mechanisms, we treated these cells with the demethylating agent DAC (1 μM) and/or the histone deacetylase inhibitor TSA (0.5 μM), known to reactivate epigenetically repressed genes (Dworkin et al., 2009
; Huang et al., 2009
). As shown in (lanes 1–4), the expression of LAMP3
was significantly reactivated by single treatments (i.e., DAC or TSA, P
< 0.01). Furthermore, synergistic re-expression of this gene was observed in cells with the combined treatment (DAC plus TSA, P <
0.001). Additional results of 7 other repressed genes are presented in supplemental Figure S2
Fig. 5 Epigenetic reactivation of ERα-mediated LAMP3 repression in MCF-7. (A) MCF-7 were treated with DAC (1 μM), TSA (1 μM) and/or ERα antagonist, ICI182780 (ICI, 1 μM) 6 hr before E2 stimulation. Total RNA was subjected (more ...)
To investigate whether this epigenetic repression could be attributed to an estrogen-mediated pathway, MCF-7 cells were additionally treated with E2 and/or an ERα antagonist, ICI182780. The subsequent E2 treatment led to re-silencing of LAMP3, suggesting a role of estrogen signaling in mediating this epigenetic repression (, lanes 5–8). Treatment of ICI182780 abolished the down-regulation, additionally indicating that this regulation is partly mediated through an ERα-dependent pathway (, lane 9). The repression was partially attenuated in the presence of additional epigenetic treatments (i.e., DAC and TSA, lanes 10–12).
Based on the results of these pharmacological experiments, our observations suggest that 1) estrogen signaling initiates the repression of the BPA-influenced loci in breast epithelial cells; 2) this repression is partly mediated trough an ERα-dependent pathway; and 3) persistent repression of the BPA-influenced loci in cancer cells may be further maintained by DNA methylation and histone modifications.
LAMP3 repression in MDECs preexposed to other estrogen-like chemicals
To determine whether long-term exposure of other estrogen-like chemicals can additionally initiate this epigenetic repression, mammospheres were exposed to diethylstilbestrol (DES, 70 nM), daidzein (10 μM), 1,3,5-tris(4-hydroxyphenyl)-4-propyl-1H-pyrazole (PPT, 0.1 nM), 4-nonylphenol (NP, 1 μM), N-butyl-benzyl phthalate (BBP, 10 μM), di(2-ethylhexyl)-phthalate (DEHP, 10 μM), and 4,4′-dichloro-biphnyl (PCB, 0.1 nM) for 3 weeks. After the exposure, MDECs were subjected to RT-qPCR analysis for LAMP3 expression. As shown in , downregulation of LAMP3 was confirmed in MDECs preexposed to these estrogen-like chemicals (2.3 to 12.5-fold decrease). Suppressive effects varied for the different environmental exposures, indicating differential sensitivity of progenitors to these chemicals.
Promoter hypermethylation of LAMP3 in ERα-positive breast cancer
To confirm the in vitro
epigenetic findings, we conducted DNA methylation analysis in the promoter CpG island regions of LAMP3
loci, in 48 breast cancer cell lines, 484 primary breast tumors (Taiwan cohort, n
= 336; US cohort, n
= 148), and 10 noncancerous breast tissues as normal controls. Pyrosequencing analysis of LAMP3
(9 CpG sites) revealed that DNA methylation levels were significantly increased in breast cancer cell lines relative to those of normal controls (Supplemental Figure S3B
). Moreover, promoter hypermethylation of LAMP3
= 0.008) was significantly associated with the ERα-positive status. In close agreement with these results, hypermethylation of LAMP3
was observed in ERα-positive tumors in the US cohort (P
< 0.0001) (, and Supplemental Figure S3A
) and in the Taiwan cohort (, and Supplemental S3A
). The cut-off points of age groups used in the further analysis were based on menopausal status - premenopausal (age <50 years) and postmenopausal (age >50 years) groups. The young age group defined by age <35 years appears to have distinct biological characteristics and display poor prognosis compared to those ≥35 years. Interestingly, while the hypermethylation event occurred in both age groups (35–50 and >50 years old) in the US cohort, this trend was only seen in the old age group (>50 years old) of the Taiwan cohort. Association of this hypermethylation with other clinicopathological features of patients was not apparent.
Fig. 6 DNA methylation analysis of LAMP3. Quantitative methylation profiles of tumor samples from the US and Taiwan cohorts are shown in supplemental figure S3A. (A) Box plots indicate that the level of LAMP3 promoter methylation is positively correlated with (more ...)