Introduction
Many breast cancers initially respond well to chemotherapy, however, in a significant proportion of cases, and following a cancer-free period, metastases at distant sites appear and these are accompanied by a poor response to therapy, normally with fatal consequences. It is thus imperative to improve our understanding of the mechanisms leading to cancer spread and therapy response with the aim of developing new markers and therapeutic strategies [
1].
Epithelial cells undergoing epithelial-to-mesenchymal transition (EMT, the first step in the metastatic cascade) lose E-cadherin expression and their characteristic epithelial morphology. These events are accompanied by an increase in motility and acquisition of the ability to degrade the extracellular matrix and invade other tissues. Loss of E-cadherin, a hallmark of the EMT programme, liberates β-catenin (normally associated with the C-terminus of E-cadherin), which migrates to the nucleus and induces the expression of genes orchestrating the EMT programme. Importantly, the EMT programme generates cancer cells with stem cell-like characteristics (cancer stem cells, CSCs). EMT and CSCs are often associated with the acquisition of drug resistance. Thus, these three processes are tightly interlinked [
2‐
4].
Downregulation of E-cadherin by transcriptional repressors (Snail and Slug, as well as ZEB1, ZEB2 and Twist) is well-studied. However, transcriptional activators of E-cadherin, which help maintain the cell in an epithelial state, such as EP300, FOXA1 and RUNX1, are less well characterized [
5]. EP300 was originally discovered as a transcriptional co-activator that plays pivotal roles in integrating and mediating multiple signal-dependant transcription events [
6]. The most studied function of EP300 is as a histone acetyltransferase, regulating transcription via chromatin remodelling [
7], and it has important roles in cell proliferation, transformation and differentiation [
8]. Loss of EP300 heterozygosity has been described in breast carcinomas [
9], somatic
EP300 mutations have been identified in several malignancies [
10] and EP300-deficient colon carcinoma cells show phenotypic changes characteristic of EMT [
11].
We have recently described a novel EMT/CSC/drug resistance regulatory axis controlled by the miR-106b~25 cluster [
12], which is associated with aggressive basal-like, oestrogen receptor-negative, grade 3 breast cancers [
13]. The three miRs in the cluster target
EP300 mRNA and downregulate its expression, leading to increased motility and invasive properties as well as the generation of doxorubicin and radiation-resistant derivatives [
12]. In bladder cancer cells, experimental downregulation of EP300 also leads to doxorubicin and cisplatin resistance [
14], [
15]. Using a minimally transformed mammary epithelial cell model [
16], we have also demonstrated that cells in which this pathway has been experimentally downregulated acquire a multidrug resistance phenotype with evasion from apoptosis [
17].
Here we show that experimental modulation of EP300 alters paclitaxel sensitivity and the generation of paclitaxel resistance. EP300 silencing is also associated with increased in vitro tumorigenicity and CSC-like markers, whilst its ectopic expression in basal-like breast cancer cells partly rescues the epithelial, differentiated and paclitaxel-sensitive phenotype. Gene expression profiling identifies down-stream EP300 targets associated with drug resistance, EMT and CSCs. Finally, immunohistochemical analysis reveals a strong downregulation of EP300 in metaplastic breast cancer, a rare, but aggressive form of invasive breast cancer with histological evidence of EMT, which has a poor clinical outcome.
Discussion
We have recently described a novel pathway controlling drug resistance in breast cancer cells. Upstream of the pathway, three miRs (miR-106b, miR-93 and miR-25) transcribed from the same cluster, target EP300, a transcriptional activator of E-cadherin. Upregulation of these miRs, found in aggressive basal-like, oestrogen receptor-negative, grade 3 breast cancers [
13] and drug-resistant cell models, leads to a downregulation of EP300 and E-cadherin with initiation of an EMT [
12]. However, miRs regulate the expression of many genes, and the contribution to EMT by miR targets other than EP300 cannot be ruled out. Here we describe that experimental modulation of EP300 in breast and colon cancer cell models alters paclitaxel sensitivity and the generation of resistant cells. Importantly, EP300 is also associated with stemness. Its downregulation is associated with increased in vitro tumorigenicity and cancer stem cell-like markers, whilst its ectopic expression in basal breast cancer cells partly rescues the epithelial, differentiated, paclitaxel-sensitive phenotype. Transcriptome analysis identifies key molecules associated with these phenotypes such as
CEACAM5 (adhesion),
CAPN9 (cytoskeletal remodelling),
ABCG2 (stemness),
BCL2 (apoptosis),
ITAG2 and
ITAG3 (migration) and
TGFB2 (metastasis). Lastly, we have unveiled that EP300 is downregulated in metaplastic breast cancer, a rare aggressive form of the disease with histological evidence of EMT and poor clinical outcome.
During the initiation of the metastatic cascade, cells undergo reprogramming to a less differentiated, therapy-resistant and morphologically changed phenotype [
32]. As EP300 functions as a histone acetyltransferase [
7], with the potential to modulate the expression of a plethora of genes, it is not surprising that its experimental downregulation in MCF-7 cells alters the expression of more than 4000 genes. Although the main function of EP300 as a transcriptional activator is well-established, such as in the case of
CDH1 (E-cadherin) transactivation [
5], it may also lead to gene repression [
33,
34]. Early work on MCF-7 cells expressing ribozymes specific for EP300 indicates that DNA damage-induced apoptosis is impaired in EP300-deficient cells [
35]. In bladder cancer cells, experimental downregulation of EP300 leads to doxorubicin and cisplatin resistance [
14,
15] and, using a minimally transformed model of mammary epithelial cells in which EP300 has been downregulated, we have recently found an impaired caspase-9 and caspase 3/7 activation following paclitaxel treatment [
17]. EP300-deficient cells show multidrug resistance to a variety of structurally and functionally different drugs and γ-radiation, a phenotype that is ABC transporter independent [
12,
17]. The EP300 signature presented here highlights the importance of
BCL-
2 in the apoptotic evasion reported in EP300-deficient cells. Although a definitive mechanistic insight in the regulation of apoptosis by EP300 has not yet been established, EP300 has been described to be associated with SATB1, leading to the repression of
CYBB, the key component of the phagocyte NADPH oxidase [
36]. SATB1 has also been described to bind the
BCL-
2 promoter where it has a negative transcriptional regulatory function [
37]. Other EP300-associated factor, PCAF, accelerates apoptosis by repressing a GLI/BCL2/BAX axis in hepatocellular carcinoma [
38]. It is thus tempting to speculate that these mechanisms may also play a role in breast cancer cells.
EP300-deficient cells increase their motility and invasive properties, both in colon and breast cancer cells [
11], [
12]. This phenotype involves major cellular reprogramming highlighted in the EP300 signature described here with downregulation of cell adhesion molecules, such as EPCAM5 and EPCAM6, and upregulation of collagens (COL12A1 and COL4A5) and mesenchymal OB-cadherin (CDH11), a characteristic EMT marker [
39]. Upregulation of fibulins (EFEMP1 and HMCN1), which bind EGFR and regulate cell adhesion and migration, correlates with tumour progression and poor prognosis in ovarian cancer [
40]. Upregulation of EPHA4, a receptor tyrosine kinase which binds ephrin family ligands, modulates cell morphology and promotes migration in glioma cells [
41]. Consistently, high levels of EPHA4 correlate with a reduced overall survival in breast cancer patients [
42]. Downregulation of ARHGAP20, a GTPase activator of Rho-type GTPases, favours Rho to be in the active GTP-bound state and promotes motility [
43]. Reorganization of the actin cytoskeleton is also important for cellular remodelling, with the downregulation of plastin (PLS3) and other actin-binding proteins (CNN2), as well as upregulation of WIPF1, which intervenes in the disassembly of stress fibres in favour of filopodia formation. Importantly, high expression of WIPF1 is associated with poor prognosis in breast cancer [
44]. Cytoskeletal remodelling is also regulated by CAPN9, a member of the calpain family. Low CAPN9 expression has been associated with poorer clinical outcome in breast cancer patients following endocrine therapy [
45].
TGFB2 is one of the three highly homologous isoforms of TGF-β found in humans. TGF-β is a potent inducer of EMT in mammary cells, the acquisition of CSC properties and chemotherapy resistance [
46,
47]. TGF-β also induces FGFR switching, causing murine mammary epithelial cells to become sensitive to FGF2 [
48]. A similar effect is observed in the EP300 signature, with upregulation of
TGFB2 and
FGFR2. Importantly, FGF2 induces mesenchymal OB-cadherin (
CDH11) expression [
49], also observed in EP300-deficient MCF-7 cells. Importantly, metaplastic breast cancer is enriched in the markers of CSCs and EMT [
50] and we show here low EP300 expression when compared to non-cancerous breast epithelium. Transcriptome profiling of metaplastic breast cancer indicates downregulation of epithelial phenotypes, remodelling of the extracellular matrix and EMT [
51,
52]. Of note that the small sample analysed by Lien et al. indicates a concordance with differentially regulated genes in the MCF-7 EP300 signature, downregulation of
CDH1, CEACAM6 and
ARHGAP8, another Rho GTPase-activating protein, as well as upregulation of
BCL2A1, EFNB1 (ephrin B1) and osteoprotegerin (
TNFRSF11B), a soluble TRAIL decoy receptor secreted by inflammatory and invasive breast cancer cells that induces aneuploidy, cell proliferation and angiogenesis [
53].
In conclusion, we demonstrate that tumour suppressor EP300 is poorly expressed in metaplastic breast cancer and is a master regulator of EMT, CSCs and drug resistance.