Transcriptional regulation of bidirectional gene pairs by 17 β-estradiol in MCF-7 breast cancer cells

Using cDNA microarray analysis, we previously identified a set of differentially expressed genes in primary breast tumors based on the status of estrogen and progesterone receptors. In the present study, we performed an integrated computer-assisted and manual search of potential estrogen response element (ERE) binding sites in the promoter region of these genes to characterize their potential to be regulated by estrogen receptors (ER). Publicly available databases were used to annotate the position of these genes in the genome and to extract a 5’flanking region 2 kb upstream to 2 kb downstream of the transcription start site for transcription binding site analysis. The search for EREs and other binding sites was performed using several publicly available programs. Overall, approximately 40% of the genes analyzed were potentially able to be regulated by estrogen via ER. In addition, 17% of these genes are located very close to other genes organized in a head-to-head orientation with less than 1.0 kb between their transcript units, sharing a bidirectional promoter, and could be classified as bidirectional gene pairs. Using quantitative real-time PCR, we further investigated the effects of 17β-estradiol and antiestrogens on the expression of the bidirectional gene pairs in MCF-7 breast cancer cells. Our results showed that some of these gene pairs, such as TXNDC9/ EIF5B, GALNS/TRAPPC2L, and SERINC1/PKIB, are modulated by 17β-estradiol via ER in MCF-7 breast cancer cells. Here, we also characterize the promoter region of potential ER-regulated genes and provide new information on the transcriptional regulation of bidirectional gene pairs.

Estrogens are crucial for the development and maintenance of reproductive organs and play important roles in several target tissues, including bone and the cardiovascular and nervous systems.In the mammary gland, the estrogens mediate key physiological processes that are essential for normal growth, differentiation, and survival.A large body of evidence shows that estrogens, especially 17β-estradiol (E 2 ), also play a pivotal role in promoting mammary carcinogenesis (1).Most of the complex effects of estrogens on cell proliferation, differentiation, and survival are mediated by the estrogen receptors ERα and ERβ via the transcriptional regulation of ER target genes.In the classical mechanism of action, the estrogen receptor complex binds with high affinity to estrogen response elements (EREs) in the promoter region of target genes, interacts with the basal transcription machinery, and regulates gene expression.The consensus ERE, which was first identified in the promoter region of the Xenopus vitellogenin gene, is a perfect palindromic DNA motif composed of two inverted sequences of 5 bp separated by three nucleotides: 5'-GGTCAnnnTGACC-3' (2).Computational and experimental studies have revealed that most promoter regions in estrogen target genes in the human genome do not contain the consensus ERE, but contain degenerate palindromic sequences showing one or more substitutions, ERE half-sites, direct repeats of halfpalindromes, and other transcription factor binding sites, such as AP1, SP1, NF-κB, C/EBP, FoxA1, and octamers (2)(3)(4)(5).ER also interacts with EREs containing variations in base substitution, although some nucleotide substitutions can be more harmful than others (6).Some symmetrical substitutions on each side of the palindrome may more dramatically affect the interaction with the ER than a unilateral www.bjournal.com.brBraz J Med Biol Res 44(2) 2011 replacement (6).ERE half-site sequences associated with SP-1 sites, direct repeats of ERE half-sites, and EREs with different spacing between the palindromic sequences also display estrogen responses (7)(8)(9).
Genome-wide analysis has indicated that 10% of human genes are organized in a head-to-head orientation and located on opposite DNA strands in close proximity to the transcription start sites (TSS), separated from them by less than 1 kb (10,11).These genes are classified as bidirectional gene pairs and share an intervening sequence containing regulatory elements called bidirectional promoters (12,13).Bidirectional gene pairs are evolutionarily conserved, suggesting that the close proximity might have been maintained for some functional reason (14).One possibility is that genes sharing promoters might be coordinately regulated in different tissues; however, the mechanisms underlying the transcriptional regulation of bidirectional promoters are still poorly understood (15,16).Much evidence indicates that bidirectional gene pairs might play a role in cancer; DNA repair genes that are involved in tumorigenesis are frequently found to be organized as bidirectional gene pairs (10,11).The expression profiles of ovarian and breast cancer show an enrichment of bidirectional gene pairs that include DNA repair genes, such as BRCA1, BRCA2, CKEK1, and FANC family members (17).Furthermore, the silencing of bidirectional gene pairs by DNA methylation has been reported in various human cancer cell lines and primary breast tumors (18)(19)(20).
In the present study, we performed an integrated computational and manual search to identify potential ER binding sites in the promoter region of 83 genes previously identified as being differentially expressed in primary breast tumors based on the presence of estrogen and progesterone receptors (21).We identified a subgroup of genes arranged in a head-to-head orientation and separated by a small intergenic region that could be classified as bidirectional promoters.Selected candidate ER target genes were further examined for their potential to be modulated by 17β-estradiol in MCF-7 breast cancer cells.

ERE binding site search
The NCBI (http://www.ncbi.nlm.nih.gov),USCS Genome Bioinformatics (http://genome.uscs.edu)and ENSEMBL (http://www.ensembl.org)databases were used to annotate the position of 83 genes in the genome and to extract sequences 2 kb upstream to 2 kb downstream of the TSS for binding site analysis.The search for EREs and other binding sites was performed using several publicly available programs (Transfac 6.0, http://www.gene-regulation.com,and Dragon Genome Explorer, http://research.i2.a-star.edusg/promoter/).A list of ERE sequences was prepared from the reports extracted from PubMed and used for a manual search (6,(22)(23)(24).The present study was approved by the Institutional Ethics Committee.

Cell lines and culture conditions
Human breast cancer cells, MCF-7 and MDA-MB-231, were obtained from the American Type Culture Collection (ATCC).Cells were cultured at 37°C in phenol red-free RPMI supplemented with 10% FCS and 1% penicillin-streptomycin (Gibco, USA) in an atmosphere of 5% CO 2 /95% air.Before the treatments, cells were washed twice with PBS to remove residual serum and were grown in phenol red-free media containing 5% charcoal-stripped FCS for 48 h.After hormone deprivation, the cells were incubated with 10 nM 17β-estradiol for 2, 6, and 24 h or treated with 1 µM 4-hydroxytamoxifen or 1 µM ICI 182.780 (Faslodex) alone or in combination with 17β-estradiol for 24 h.Cells maintained in phenol red-free media and supplemented with 5% FBS were used as controls.

RNA extraction
After the treatments, the cells were washed twice with PBS and harvested.Total RNA was extracted from cultured cells using the guanidine isothyocyanate method (25).RNA samples were analyzed using 1% agarose gel electrophoresis and ethidium bromide staining.The quality of the RNA samples was estimated by the ratio of 28S:18S (2.0 and higher).All RNA samples were treated with DNaseI for 1 h at 37°C to eliminate genomic DNA contamination.
The relative gene expression was normalized using GAPDH expression as an internal control.MCF-7 cells maintained in phenol red-free medium and supplemented with 5% FCS served as calibrator samples for hormone-treated cells.The results are reported as n-fold differences in the expression of the target gene relative to the expression of GAPDH and a calibrator sample.The relative expression was calculated by 2 -∆∆CT , where CT is the fluorescence threshold value, ∆CT = CT of the target gene -CT of the reference gene, and ∆∆CT = ∆CT of the tumor sample -∆CT of the calibrator sample.

Statistical analysis
Statistical analyses were performed by one-way ANOVA using the SPSS 10.0 software (Statistical Package for the Social Sciences; SPSS Inc., USA), and P values were calculated using the Bonferroni test and considered to be significant at P < 0.05.

Results
In a previous study using cDNA microarray, we identified differentially expressed transcripts between ER/ progesterone receptor (PR)-positive and ER/PR-negative primary breast tumors, providing a new set of potential target genes for the ER signaling pathway.In the present study, in order to identify DNA binding sites that mediate the estrogen response, we performed a computer-assisted and manual search for putative ERE binding sites in the promoter region extracted from 83 of these potential ER target genes.The manual search was performed using a list of putative ERE binding sites based on data available in the literature.A summary of the results obtained from this analysis is shown in Figure 1.Overall, 43% (36/83) of the candidate genes showed a good potential for being regulated by estrogen via ER.The promoter regions of these genes contain perfect or imperfect palindromic sequences, SP1, AP1, and/or FOXA1 binding sites.
Interestingly, during the annotation of the 83 potential ER target genes, we identified 17 genes located in close proximity to other genes and organized in a headto-head orientation, possibly sharing the same promoter region.Using criteria of an intergenic distance of <1000 bp spanning the TSS of the divergent gene pair to define bidirectional promoters (11), we identified 14 (16.8%)bidirectional gene pairs (Table 1).The size of the intergenic sequences ranged from 18 to 973 bp, and all of these bidirectional promoters contain CpG-rich sequences.Figure 2 shows a schematic representation of three of the bidirectional gene pairs identified.We also investigated whether the identified bidirectional gene pairs maintained their divergent head-to-head organization in other species.Although limited, our analysis showed that the majority of the bidirectional gene pairs found in this study is evolutionarily conserved, as shown in Figure 3 for the GALNS/TRAPPC2L divergent gene pair.
Our computer-assisted search for EREs revealed that the promoter regions of the majority of the bidirectional gene pairs showed good potential to be regulated by estrogen.As shown in Table 1, all 14 bidirectional promoters analyzed contain putative ERE binding sites (at least half EREs) and other binding sites for other transcription factors, such as SP1 and AP1 or FOXA1 binding sites, indicating that they have the potential to be regulated by ER.The putative EREs were also found to be conserved in different species (Figure 3).In addition, approximately 80% of these bidirectional promoters contain GABP binding sites (CCGGAA(GTG)).We further evaluated the expression of five of these bidirectional gene pairs in response to 17β-estradiol in MCF-7 breast cancer cells.The gene pairs GALNS/TRAPPC2L, TXNDC9/EIF5B, and SERINC1/PKIB were modulated by estradiol (Figure 4).However, the expression of the gene pairs GTPBP2/MAD2L1BP and DIDO1/C20orf11 was not affected in the MCF-7 cells exposed to 17β-estradiol or to the antiestrogens tamoxifen and ICI 182,780 (data not shown).The transcripts of GALNS and TRAPPC2L showed increased expression in MCF-7 cells maintained in stripped serum (deprived of steroids and growth nih.gov) and USCU Genome Bioinformatics (http://www.genome.uscs.edu)databases were used to annotate the position of the 83 differentially expressed genes in the genome and to extract sequences 2 kb upstream and 2 kb downstream of the start sites for binding site analyses.The search for EREs and other binding sites was performed using several publicly available programs (Transfac 6.0, http://www.gene-regulation.com,and Dragon Genome Explorer, http://research.i2.a-star.edusg/promoter/). 1 Palindromic ERE consensus derived from vitellogenin A2 (GGTCAnnnTGACC) separated by 1 to 7 bp. 2 Imperfect palindromic ERE carrying a single substitution. 3Imperfect palindromic ERE carrying two substitutions. 4Direct repeats separated by 3 bp.factors) compared to the cells grown in normal serum (control cells) and were negatively modulated in the presence of 10 nM 17β-estradiol for 24 h (Figure 4A).A similar reduction in these transcripts was observed in MCF-7 cells treated with 1 μM tamoxifen or 1 μM tamoxifen plus 100 nM 17β-estradiol compared to the expression in MCF-7 cells maintained in stripped serum.However, in MCF-7 cells treated with 1 μM ICI 182,780 alone the GALNS/ TRAPPC2L transcripts showed increased expression compared to control cells and remained elevated in cells treated with 1 μM ICI 182,780 plus 100 nM 17β-estradiol.
The transcripts of the bidirectional gene pair TXNDC9 and EIF5B were positively regulated by 10 nM 17β-estradiol after 6 h of treatment (Figure 4B).In contrast to the gene pair GALNS/TRAPPC2L, the TXNDC9 and EIF5B transcripts were not significantly altered in MCF-7 cells grown in stripped serum compared to control cells, but both transcripts were significantly induced (>2-fold) in MCF-7 cells exposed to 10 nM 17β-estradiol for 6 h.After 24 h of 17β-estradiol treatment, the expression of these transcripts was similar to that observed in control cells.However, the transcripts of the TXNDC9 and EIF5B gene pair were not affected by treatment with the antiestrogens tamoxifen and ICI 182,780 alone or in combination with 17β-estradiol for 24 h compared to cells growing in normal or stripped serum.Interestingly, the transcripts of the bidirectional gene pair SERINC1/PKIB were divergently modulated by 17β-estradiol (Figure 4C).Expression of the PKIB transcripts was reduced in MCF-7 cells maintained in stripped serum and was substantially stimulated by 10 nM 17β-estradiol after 6 h of treatment, whereas the SERINC1 transcripts increased in MCF-7 cells grown in stripped serum and expression was significantly inhibited in the presence of 17β-estradiol.In addition, cells treated with 1 μM tamoxifen showed a reduction of protein kinase inhibitor beta (PKIB) levels, which was reversed in the presence of 100 nM 17β-estradiol.Treatment with ICI 182,780 dramatically reduced the expression of PKIB but did not affect the expression of SERINC1 transcripts compared to cells grown in stripped serum.
To determine whether 17β-estradiol or www.bjournal.com.brBraz J Med Biol Res 44(2) 2011 negative breast cancer cells, MDA-MD-231 cells was also treated with 17β-estradiol and/or antiestrogens for 6 and 24 h.Estrogen or antiestrogens were unable to regulate the transcripts of these bidirectional gene pairs (data not shown) in MDA-MB-231 cells.

Discussion
In the present study, we employed an integrated computer-assisted and manual search to identify EREs in the promoter region of a set of candidate ER target genes previously identified as being differentially expressed in primary breast tumors based on their estrogen and progesterone status (21).Using this approach, we characterized the promoter region of 83 genes for their potential to be regulated by ER.Although only a small percentage of the candidate ER target genes showed the presence of an ERE consensus sequence in their promoter region, more than 40% of these genes were found to have putative ERE, SP1, AP1, and/or FOXA1 binding sites in their promoter regions, indicating the potential to be regulated by estrogen.The maximum ER transcriptional transactivation is thought to be displayed by the interaction of a homo-or heterodimer of ER with the ERE consensus sequence (CAGGTCAnnnTGACCTG).However, the ERE consensus sequence is rarely found in the promoter regions of human genes (2).The promoter regions of estrogen-responsive genes usually contain more than one imperfect palindromic sequence separated by one, two, three, or more interstitial bases, direct repeats of ERE half-sites, and widespread ERE half-sites (2).Half-EREs have been shown to act synergistically, but the degree of synergy between EREs is modulated by the physical distance between EREs and by the distance between the ERE and the TATA-box (26).In addition, perfect or imperfect ERE half-sites adjacent to AP1 or SP1 binding sites in the promoter region of known estrogen-regulated genes, such as HSP27 and PR, were also found to be effective in conferring estrogen responsiveness (27,28).
During the process of gene annotation, we identified a subgroup of divergent gene pairs organized in a head-tohead orientation that could be classified as bidirectional gene pairs.Surprisingly, the prevalence of bidirectional gene pairs found in the set of candidate ER target genes analyzed was higher than the predicted frequency of 9.4-11% found in the human genome by computational genome-wide studies (11,29).An incremental prevalence of gene pairs organized in a head-to-head orientation has also been observed among the genes that encode proteins that function in DNA repair (10,17).
Bidirectional gene pairs show evolutionary conservation, which suggests a selective pressure to facilitate the expression of functionally related genes or those that need to be co-regulated by the same stimulus or in a cell-specific manner (10,11,29).In agreement with the literature, the majority of the bidirectional gene pairs identified in the present study were shown to be conserved during evolution, keeping the same organization in vertebrates and indicating that these genes organized in a head-to-head orientation play important roles (29,30).Our data also showed that some of the putative ERE binding sites were conserved in the bidirectional gene pairs in other species.Different degrees of conservation of ERE or ERE-SP1 sites have been reported in the literature (4,9,31).In addition, the majority of the bidirectional promoters analyzed also contain GABP binding sites, which were previously shown to be over-represented in bidirectional promoters (32,33).However, whether or not they are all likely to be functionally conserved is still being questioned.
The bidirectional head-to-head orientation suggests that the potential promoter regions of these gene pairs share cis-elements, such as ERE and SP1 sites, that could be transcriptionally regulated by estrogen.To validate our results, we evaluated the effects of estrogen and antiestrogens on the transcriptional regulation of five bidirectional gene pairs in the breast cancer cell line MCF-7.To our surprise, the expression of two of the bidirectional gene pairs, GTPBP2/ MAD2L1BP and DIDO1/C20orf11, was not modulated in MCF-7 cells exposed to estrogen and antiestrogens.Since the activity of transcription factors in transcriptional transactivation depends on the state of the chromatin (34,35), we can speculate that, in MCF-7 cells, the E 2 -ER complex has no accessibility to the binding sites in the promoter regions of these genes, possibly due to the state of chromatin organization.Further studies using different cell lines and agents that affect chromatin organization are required.On the other hand, distinct patterns of gene regulation by estrogen and antiestrogens were observed for the bidirectional gene pairs GALNS/TRAPPC2L, TXNDC9/EIF5B, and SERINC1/PKIB in MCF-7 cells.In contrast, in ER-negative breast cancer cells, MDA-MB-231, 17β-estradiol or antiestrogens (tamoxifen or ICI) were unable to modulate the mRNA expression of these bidirectional gene pairs (data not shown), indicating that the presence of ER is essential for the transcriptional transactivation of these bidirectional promoters by estrogen.In MCF-7 cells, two of the gene pairs analyzed displayed a positive correlation, with both gene transcripts being up-regulated (TXNDC9/EIFB5) or down-regulated (GALNS/TRAPPC2L) by estrogen.The differences observed in the effects of the antiestrogens tamoxifen and ICI 182,780 on the transcriptional regulation of these bidirectional gene pairs can be attributed to their mechanisms of action.Tamoxifen is a selective estrogen modulator that acts by inhibiting estrogen binding to its main target estrogen receptor; however, it also acts as an estrogen agonist in certain tissues or in specific regulation of ER-responsive genes (36).Compared to tamoxifen, ICI 182,780 is considered to be a pure antagonist that blocks ER-mediated transactivation down-regulating ER protein levels.Additionally, the gene pair SERINC1/PKIB showed a negative correlation with the up-regulation of PKIB tranwww.bjournal.com.brBraz J Med Biol Res 44(2) 2011 scripts and down-regulation of SERINC1 transcripts when MCF-7 cells were exposed to estrogen.The SERINC1/PKIB gene pairs may have distinct elements in the bidirectional promoter and further studies using reporter assays are needed to better define the critical sequences involved in this divergent regulation.On the other hand, in this study we evaluated the presence of EREs in the proximal promoter region only; however, the presence of EREs located in distal regions or of other cis-regulatory elements regulated by other factors may be necessary to modulate the divergent expression of this bidirectional gene pair.
Although genome-wide analysis showed that 45-68% of the gene pairs organized in a head-to-head orientation are co-expressed (16,29), the mechanisms involved in the transcriptional regulation of these bidirectional gene pairs appear to be complex and still poorly understood (13,29,37,38).Diverse patterns of gene regulation have been reported for different bidirectional gene pairs, includ-ing co-regulation, and divergent or asymmetric regulation (20,30,39,40).
In the present study, we characterized the promoter region of potential ER-regulated genes and provided new information on the transcriptional regulation of gene pairs organized in a head-to-head orientation.Further clinical and experimental studies are required to determine whether these genes play a role in breast cancer.The identification of ER target genes may improve our understanding of the role played by estrogens in breast cancer, permitting a better tailored hormone dependence and improved patient response to hormonal therapy.

Figure 1 .
Figure 1.Summary of the results of the estrogen response element (ERE) search for 83 potential estrogen receptor target genes.The NCBI (http://www.ncbi.nlm.nih.gov) and USCU Genome Bioinformatics (http://www.genome.uscs.edu)databases were used to annotate the position of the 83 differentially expressed genes in the genome and to extract sequences 2 kb upstream and 2 kb downstream of the start sites for binding site analyses.The search for EREs and other binding sites was performed using several publicly available programs (Transfac 6.0, http://www.gene-regulation.com,and Dragon Genome Explorer, http://research.i2.a-star.edusg/promoter/).1 Palindromic ERE consensus derived from vitellogenin A2 (GGTCAnnnTGACC) separated by 1 to 7 bp.2Imperfect palindromic ERE carrying a single substitution.3Imperfect palindromic ERE carrying two substitutions.4Direct repeats separated by 3 bp.

Figure 3 .
Figure 3. Evolutionarily conserved estrogen receptor binding elements (EREs) in the potential promoter region of the bidirectional gene pair GALNS/TRAPPC2L across vertebrates.A, Schematic presentation of the bidirectional gene pair GALNS/ TRAPPC2L in humans, chimpanzees, dogs, rats, and chickens.The black arrows and numbers 1 to 5 indicate the location of conserved EREs relative to the position of the transcript start site in the TRAPPC2L gene.B, 1 to 5 ERE sequences; EREs 2 (C) and 5 (D) and the surrounding sequences are shown in the alignment.The half-EREs are in bold.Asterisks denote highly conserved bases.

Table 1 .
List of the bidirectional gene pairs identified.