ER is down-regulated in Twist over-expressing breast cancer cell lines
To characterize the role of Twist in breast cancer biogenesis, we initially analyzed the MCF-7/Twist cell line (Mironchik et al 2005
) for differential gene expression using microarray analysis (Affymetrix, Santa Clara, CA), and identified the ER transcript which was down-regulated by 13 fold. To confirm this finding, breast cancer cell lines were evaluated for Twist and ER expression by immunoblotting and qRT-PCR. As shown in , there was an inverse correlation between Twist and ER protein and mRNA levels within the cell lines tested.
Twist regulation of ER by direct promoter binding
Twist represses ER promoter activity in breast cancer cells
To functionally confirm the regulatory role of Twist in ER down-regulation, we carried out promoter-reporter assays in breast cancer cell lines. The 4 kb ER promoter has 26 canonical E-box sequences (CANN
TG) (Murre et al 1989
) to which Twist can potentially bind (). Transient transfections with Twist plasmids were carried out for promoter reporter assays in MCF-7 () and MCF-7/Twist cells (data not shown). Twist repressed the full-length ER promoter by 2.5 fold, while the other deletion constructs were repressed from 2.5 to 3 fold.
In order to confirm the role of the bHLH regions of Twist in binding the ER promoter, we used the full-length ER promoter and Twist bHLH deletion mutants to assay for ER promoter repression. As seen in , none of the Twist mutants demonstrated repression comparable to wild-type Twist, except for the deletion mutant Q161X, which was downstream of the bHLH domain.
Twist binds directly to E-boxes within the ER promoter
To address if Twist binds directly to the ER promoter, we carried out chromatin immunoprecipitation (ChIP) assays using MCF-7/Twist and MDA-MB-231 cell lines (). MDA-MB-231 is an ER negative breast cancer cell line with high levels of endogenous Twist. Chromatin immunoprecipitation was carried out as per instructions (Cell Signaling, Danvers, MA) using antibodies against Twist. Normal rabbit IgG was used as a negative control, while Histone H3 was used as a positive control antibody. Five pairs of primers were designed to span most of the ER promoter and the 26 putative Twist binding E-boxes (). Maximum chromatin immunoprecipitation binding was seen in the areas of E-box 6–9 in MCF-7/Twist and MDA-MB-231 cells. Significant binding was also seen in the area of E-boxes 16–17 in MCF-7/Twist cells. These results indicate that Twist binds directly or as part of a complex to the endogenous ER promoter.
Twist facilitates estrogen independence in breast cells
As we observed increased Twist expression in ER negative cell lines, we sought to determine if Twist promoted hormone independence by repressing ER expression. To confirm this observation, MCF-7 (ER positive) and MCF-7/Twist (ER negative) cells were grown for three days in estrogen depleted media containing 5% charcoal stripped serum (CSS) and cell cycle distribution was analyzed by flow cytometry (). Proliferation of MCF-7 cells was significantly reduced in CSS (S=5.4%) compared to untreated cells (S=15.6%, P<0.05)() but not in MCF-7/Twist cells (S=22.0% vs. 19.5%, P>0.05) (). Moreover, the percentages of cells in all three phases of the cell cycle was significantly altered between MCF-7 and MCF-7/Twist cells treated with CSS - G1=85.2% vs. 63.2%, P<0.05; S=5.4% vs. 22.0%, P<0.005; and G2=6.3% vs. 12.1%, P<0.05 (). The difference was insignificant in untreated controls of both MCF-7 and MCF-7/Twist. These results support our earlier data indicating that the down-regulation of ER by Twist in MCF-7 cells leads to estrogen independent growth.
Cell cycle profiles of MCF-7 and MCF-7/Twist cells
Twist promotes hormone resistance in breast cancer cells
To investigate if the loss of ER brought about by Twist caused hormone resistance in breast cells, we treated MCF-7 and MCF-7/Twist cells with the selective estrogen receptor modulator (SERM) tamoxifen and the selective estrogen receptor down-regulator (SERD) fulvestrant. As seen in , MCF-7 cells were significantly arrested in presence of tamoxifen (S=2.3%) compared to untreated cells (S=15.6%, P<0.005). On the other hand, MCF-7/Twist cells were largely unaffected by tamoxifen treatment (S=16.6% vs. 19.5%, P>0.05) (). Also, G1 and S phases of the cell cycle were significantly altered in MCF-7 and MCF-7/Twist cells treated with tamoxifen (G1=86.6% vs. 66.5%, P<0.005; S=2.3% vs. 16.6%, P<0.005; G2=8.9% vs. 14.2%, P>0.05) ().
Treatment with fulvestrant exhibited comparable results to those of tamoxifen. MCF-7 cell growth was significantly affected by treatment (S=3.9%) compared to untreated controls (S=15.6%, P<0.005) () while MCF-7/Twist cells were unaffected by the treatment (S=16.6% vs. 19.5%, P>0.05) (). Similarly, MCF-7 cells were significantly affected by the fulvestrant treatment compared to MCF-7/Twist cells (G1=86% vs. 67.7%, P<0.005; S=3.9% vs. 16.6%, P<0.0005; G2=5.8% vs. 13.3%, P<0.05) (). There were no significant differences in cell cycle phases of untreated MCF-7 and MCF-7/Twist (data not shown).
Hormone resistance is reversed by down-regulation of Twist in breast cancer cells
In order to study if the loss of Twist caused a reversion to the estrogen dependent hormone sensitive phenotype, we down-regulated Twist expression in MCF-7/Twist and MDA-MB-231 cells using a lentiviral delivered shRNA based approach. As seen in , MCF-7/Twist shTwist cells showed significantly higher cell death following growth in CSS or on exposure to tamoxifen compared to parental MCF-7/Twist cells, which remain unaffected. Similar effects were also seen in MDA-MD-231 shTwist cells (). This would indicate that ER independence and hormone resistance exhibited by MCF-7/Twist and MDA-MB-231 cells is at least partially due to Twist expression and can be reversed by down-regulation of Twist.
Effect of Twist down-regulation on cell cycle
Twist promotes growth of breast tumors in the absence of estrogen
In order to strengthen our finding that over-expression of Twist induced estrogen independence in vivo, we orthotopically injected MCF-7/Twist cells into the mammary fat pad of SCID mice, which were not supplemented with estrogen. As seen in , MCF-7/Twist xenografts produced large tumors (greater >250 mm3) within four to five weeks of incubation. These results confirmed that MCF-7/Twist cells are estrogen independent in vivo. In order to confirm that the expression of Twist and ER in tumors was similar to that of MCF-7/Twist cells, we isolated RNA from four tumors and performed qRT-PCR using Twist and ER primers. As seen in , expression of Twist was inversely correlated with levels of ER transcripts.
In vivo growth characteristics of MCF-7 and MCF-7/Twist cells
Next, we injected mice (n=10) with MCF-7/Twist and MCF-7 cells in the presence of estrogen (17β-estradiol pellet implanted in the back). After 3–4 weeks of growth, all mice were implanted with a tamoxifen pellet. As seen in , MCF-7 tumors regressed to pre-treatment levels, while MCF-7/Twist tumors were unaffected by tamoxifen.
Twist increases vascular volume and vascular permeability of breast tumors in mice
Functional magnetic resonance imaging (fMRI) was used to non-invasively analyze the vascular volume (VV) and permeability-surface (PS) area product values in vivo. display representative false color-coded MRI generated 3-D transverse slices of xenograft tumors using MCF-7 and MCF-7/Twist cells in mice. The average tumor VV in MCF-7 (+estrogen)(E2) and MCF-7/Twist (−E2) xenografts was 6.2 and 14.9 μl/g respectively (). The average tumor PS in MCF-7 (+E2) and MCF-7/Twist (−E2) xenografts was 0.66 and 1.60 μl/g·min respectively (). Both results were significant according to the Scheffe test (F=15.9 and 7.04 respectively). VV and PS values in MCF-7 vector control xenografts were comparable to those in MCF-7 xenografts (data not shown), and were consistent with the previous report (36). VV and PS in MCF-7/Twist (+E2) xenografts were 21.1 ul/g and 1.66 ul/g.min respectively. These values were significantly higher than those in MCF-7 (+E2) controls (F=5.48 and 6.23 respectively). There was no significant difference between estrogen supplemented and non-supplemented MCF-7/Twist xenografts for VV and PS (F=3.00 and 0.03 respectively).
Twist down-regulation causes re-expression of ER
To investigate the mechanism of ER regulation by Twist, we transiently down-regulated Twist in MCF-7/Twist cells and up-regulated Twist in MCF-7 and T-47D. As seen in , down-regulation of Twist in MCF-7/Twist cells caused a significant drop in mRNA levels of Twist accompanied by an increase in ER expression. Transient expression of Twist, on the other hand, caused a significant drop in ER protein in MCF-7 and T-47D cells (). We also demonstrated that Her-2/neu
protein levels were low in MCF-7/Twist cells, which indicates that the effect of Twist on ER is not mediated by Her-2/neu
(Supplementary Figure 1
). Furthermore, we determined that the reactivation of ER in the Twist down-regulated clones was functionally active. For this purpose, the ERE-luc construct was used as a functional reporter system (kind gift of Nancy Davidson) for the in vitro
studies. As seen in , MCF-7/Twist cells show a significant drop in the activation of the reporter indicating the lack of ER functionality in these cells. Importantly, the re-expression of ER by down-regulating Twist in MCF-7/Twist cells increased reporter activity, an indication of functional ER proteins.
Functional effects of Twist on ER expression, promoter methylation, and histone deacetylation
To further characterize the functionality of Twist-mediated ER loss in breast cancer cells, we performed quantitative real-time PCR and immunoblotting for downstream targets of ER such as Cyclin D1, p21, p27 (Foster and Wimalasena 1996
), p14 (Cho et al 2006
), and Cathepsin D (Morisset et al 1986
) in MCF-7, MCF-7/Twist, and MCF-7/Twist shTwist cells. As seen in , p21 and Cathepsin D were up-regulated when Twist was down-regulated in MCF-7/Twist cells and were down-regulated when Twist was up-regulated in MCF-7 cells. This would indicate that regulation of ER by Twist is functionally relevant. Furthermore, we also probed for intrinsic levels of ER target genes in MCF-7 and MCF-7/Twist cells. As seen in , all the ER targets genes were down-regulated in MCF-7/Twist cells. This suggests that over-expression of Twist, in this system, can repress ER levels, which in turns affects the expression of its downstream target genes.
Twist induces hyper-methylation of the ER promoter
A common mechanism of ER gene silencing is that of promoter hypermethylation and occurs in 5–49% of patient samples (Ottaviano et al 1994
, Yan et al 2001
). Indeed, we did find a significant increase in ER promoter methylation in MCF-7/Twist cells as observed by MS-qPCR analysis (). Subsequently, we used cell lines with transiently up- and down-regulated Twist to validate our earlier observations. As seen in , Twist over-expression in T-47D cells caused an increase in ER promoter methylation. On the other hand, Twist down-regulation in MCF-7/Twist and MDA-MB-231 caused a significant decrease in ER promoter methylation. In order to reverse the Twist induced methylation of the ER promoter, we treated MCF-7/Twist cells with the demethylating agent 5-azacytidine. We found a significant increase in ER transcript and protein levels as seen by qRT-PCR and immunoblotting (). To decipher the mechanistic cause of the increased methylation of ER brought about by Twist, we analyzed for recruitment of DNA methyltransferases. The de novo methyltransferase DNMT3B was co-immunoprecipitated by Twist from MCF-7/Twist lysates (). Other methyltransferases such as DNMT1 and DNMT3A were not co-immunoprecipitated by Twist (data not shown).
Twist causes histone deacetylation of the ER promoter
Regulation of genes via methylation is accompanied, in some cases, by an increase in histone deacetylase (HDAC) activity (Cameron et al 1999
). Mechanistically, we determined that Twist recruited the histone deacetylase HDAC1 () to the ER promoter leading to deacetylation, which resulted in lowered expression of ER. In order to functionally study the role of HDACs in the regulation of ER, we treated MCF-7/Twist cells with the HDAC inhibitor valproic acid (VPA). As seen in , we found a significant increase in ER expression in these cells when treated with the inhibitor. Use of AZA and VPA in combination was able to rescue ER to a higher degree as compared to either of them alone.
Twist recruits HDAC1 and DNMT3B to the ER promoter
In order to confirm our co-immunoprecipitation results, we carried out ChIP experiments on MDA-MB-231 cells using HDAC1 and DNMT3B antibodies. As seen in , DNMT3B was recruited to all the E-box sites of the ER promoter, while HDAC1 was significantly recruited to E-box 6–9. Taken together, the largest enrichment of DNMT3B and HDAC1 binding was in the E-box 6–9 regions. This would point to an important role of E-box 6–9 region in the regulation of ER by Twist via the interaction with HDAC1 and DNMT3B.
Twist and ER are inversely correlated in breast cancer patients
To confirm the inverse correlation between Twist and ER expression seen in breast cancer cell lines, Twist and ER mRNA levels in human breast tumors were quantified by qRT-PCR. A total of 73 primary breast cancers (grade 1, n=16; grade 2, n=22; grade 3, n=35) and four normal breast samples were analyzed (). We examined correlation in normalized (to normal breast tissue) expression between TWIST and ER genes, overall, and within each grade, using the nonparametric, Spearman rank test. Twist expression levels were significantly different from ER levels (). We also observed an increase in Twist expression levels with increasing tumor grade while ER expression levels showed a decrease with increasing tumor grade for grades 1 and 2. An inverse relation was observed with increased Twist expression associated with decreased ER expression, which was statistically significant in grades 1 and 2 () but not in grade 3 (). Overall, these results suggest that Twist expression inversely correlates with ER expression in human breast cancers of grades 1 and 2.
Correlation between Twist and ER mRNA levels in human breast tumors