Introduction
Globally, breast cancer is a major cause of morbidity and mortality. Oestrogen, which plays a key role in the growth and differentiation of normal breast tissue, is also implicated in the pathogenesis and progression of breast cancer [
1], with oestrogen receptor alpha (ERα) being expressed in about 70% of breast cancers. Endocrine therapy [tamoxifen and/or aromatase inhibitors (AI)] has been proven very successful in the clinical setting. However, de novo or acquired resistance to endocrine therapy is a major clinical problem and limits its use [
2,
3]. Several studies have identified potential mechanisms of resistance to endocrine therapies such as enhanced growth factor signalling, changes in the expression and/or action of the ER, altered expression of ER co-regulators, micro-RNA or long non-coding RNA interference and DNA methylation [
3‐
7]. In addition, mutations of the
ESR1 gene, encoding the ERα, have been increasingly recognised as an important mechanism of endocrine therapy resistance, mainly to AIs, with a prevalence of around 25% [
8,
9]. Transcriptomic profiling of MCF7 tamoxifen-sensitive and its tamoxifen-resistant variant revealed differential expression of genes involved in cell cycle control, transcriptional/translational machinery,
ESR1 regulation, dysfunctional mitochondrial and oxidative phosphorylation and altered metabolism, whereas proteomic analysis of acquired tamoxifen resistance demonstrated downregulated ER-signalling, activation of alternative survival pathways and enhanced cell motility and migration through regulation of the actin cytoskeleton [
3,
4,
10‐
12]. The adaptation of the tumour cell from oestrogen to androgen dependent has been recently demonstrated as another mechanism of resistance to ER-targeted therapies [
13]. Understanding the underlying molecular mechanisms that mediate resistance is required to improve the management of endocrine-unresponsive breast cancer, and the development of novel therapeutic strategies in the management of breast cancer.
Activating transcription factor-2 (ATF2), a member of the leucine zipper family of DNA binding proteins, regulates the transcription of various genes, including those involved in apoptosis, cell growth, proliferation, inflammation and DNA damage response [
14,
15]. Dependent on its binding partner, ATF2 binds to CRE (cAMP response element) consensus sequences (5-TGACGTCA-3) or to AP-1 (activator protein 1) consensus sequences (5-TGACTCA-3) [
14]. ATF2 also possesses an intrinsic histone acetyltransferase activity triggering its own DNA binding effectiveness [
16]. However, the exact role of ATF2 in breast cancer is still unclear. ATF2 has been found to increase the transcription of matrix metalloproteinase 13 (MMP13), which may help facilitate breast cancer bone metastasis [
17,
18]. In addition, cJun-ATF2 dimers have been shown to lead to the transcription of cyclin A, which increases cell proliferation [
19], providing further evidence for a possible oncogenic role for ATF2. Co-culture with malignant epithelial cells in primary human adipose fibroblasts obtained from breast cancer patients increased the levels of phosphorylated ATF2 (pATF2) at the promoter of the aromatase gene responsible for oestrogen synthesis [
20]. Furthermore, pATF2 has been shown to facilitate the transcription of MMP2, which increases migration in H-Ras-transformed MCF10A human breast epithelial cells indicating that ATF2 may play a role in breast cancer metastasis [
21]. ATF2 also forms a complex with c-Jun and c-Fos that mediates HER2’s induction of cyclooxygenase-2 (COX2), involved in cancer development and metastasis [
22]. Several studies reported that v-src causes ATF2 and CREB to bind the CRE/ATF site of the cyclin D1 gene, leading to transcription of cyclin D1 in MCF7 human breast cancer cells [
23,
24]. Together, these findings are strongly supportive of a role for ATF2 as an oncogene in breast cancer.
In our previous study, we observed that pATF2 predicts improved disease-free and overall survival in ER-positive breast cancer patients treated with tamoxifen [
25]. We also showed that silencing of ATF2 led to a loss in the growth-inhibitory effects of tamoxifen in the ER-positive, tamoxifen-sensitive MCF7 cell line and that tamoxifen treatment caused a dose-dependent phosphorylation of ATF2 within its activation domain, enhancing its transcriptional activity. That work suggested a tumour-suppressive role of ATF2 in ER-positive breast cancer [
25]. The apparent dual function of ATF2 has been also observed in skin tumourigenesis, where ATF2 has both oncogenic and tumour-suppressive activities [
15,
26,
27]. In the current study, we sought to investigate the in vitro role of ATF2 in acquired resistance to endocrine therapy.
Materials and methods
Breast cancer cell lines
MCF7 cells obtained from the Cancer Research UK Cell Services (Clare Hall Laboratories, South Mimms, Herts, UK) were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% foetal bovine serum (FBS), 5 mM L-glutamine and 1% penicillin/streptomycin (P/S). TAMR cells (MCF7-derived tamoxifen-resistant cell line), a kind gift from Professor RI Nicholson [
28] were maintained in DMEM phenol red-free supplemented with 5% charcoal/dextran-stripped FBS (CSS), 1% P/S and 100 nM 4-hydroxytamoxifen (4-OH-Tam) (all the reagents and media were from Sigma-Aldrich Company Ltd., Gillingham, UK). LCC2 (E2-independent, tamoxifen-resistant and ICI 182,780-sensitive subline of the MCF7) and LCC9 (ICI 182,780-resistant MCF7 variant and cross-resistant to tamoxifen) were kind gifts from Professor R. Clarke [
29,
30]. LCC2 and LCC9 were cultured similarly to TAMR cells but without tamoxifen.
ATF2-siRNA transfections, qRT-PCR and immunoblotting
Two small interfering RNA (siRNA) duplexes for ATF2 mRNA depletion (S3492 and S3493) and the non-targeting siRNA (NT-siControl) (Silencer® Negative Control: AM4635) were used in our experiments (Thermofisher Scientific, Paisley, UK). Cells (2.5 × 10
5 cells/well in 6-well plate) were transfected with 100 nM of each of the ATF2-siRNA duplexes including the NT-siControl using RNAiMAX, according to the manufacturer’s instructions (Thermofisher Scientific, Paisley, UK) and incubated in DMEM phenol red-free medium with 5% CSS. Following overnight incubation, the transfected cells were changed to their standard growth medium for 48 h. RNA was prepared using the Qiagen RNeasy kit (Crawley, UK) and converted to cDNA with the High-Capacity cDNA reverse transcription kit, followed by quantitative real-time PCR (qRT-PCR) using the TaqMan Gene Expression Assays, listed in supplementary Materials and Methods, Table
1 (all from Thermofisher Scientific, Paisley, UK).
Protein extracts and immunoblotting was performed as previously described [
25]. Briefly, following ATF2 knockdown, cell lysates were prepared in RIPA buffer (Sigma-Aldrich Company Ltd., Gillingham, UK) supplemented with protease/phosphatase inhibitors (Roche Diagnostics, Burgess Hill, UK). Proteins were separated in 12% SDS-PAGE and probed with primary antibodies at 4 °C overnight. Following secondary horseradish peroxidase-conjugated antibody incubation (Dako UK Ltd., Cambridgeshire, UK), membranes were developed with Super-signal West Pico Chemiluminescent Substrate (VWR International Ltd., Lutterworth, UK). The antibodies were ATF2, p-ATF2, TFF1, ERα, GREB1, PGR, NCOA3, HER2, ERK1/2 and pERK1/2 (Cell signalling, Danvers, MA, USA) and β-actin (Insight Biotechnology, Middlesex, UK), and their details are listed in supplementary Material and Methods, Table
2.
Cell growth assay
The SulfoRhodamine B (SRB) cell proliferation assay was used for cell growth determination (GeneCopoeia, MD, USA). For ATF2-siRNA and siControl transfections, 10 μl of siRNA-Lipofectamine complexes were added per 3 × 103 cells/well in a 96-well plate with 150 μl of DMEM phenol red-free medium and 5% CSS. Standard growth medium, with/without ligands, was added to the plates the following day. SRB was performed according to the manufacturer’s protocol and absorbance was measured using a Tecan Infinite M200 (TECAN UK Ltd., Reading, UK) plate reader at 492 nm. Results are expressed as the mean ± standard deviation (SD) of triplicate wells from three independent experiments.
Soft agar gel-I (1% SA-I) was made from 2% low melting temperature agarose (LMA) (Sigma-Aldrich Company Ltd., Gillingham, UK) mixed with an equal amount of standard medium for siRNA-transfected cells, while soft agar gel-II (1% SA-II) was made from 2% LMA mixed with equal amount of DMEM phenol-red free and 5% CSS with 100 nM tamoxifen for siRNA-transfected cells. SA-I and SA-II were plated at 2 ml/well in a 6-well plate and placed at 4 °C to solidify (basement gel layer). Cells were harvested separately, adjusted with SA-I or SA-II to 1 × 104 cells/ml, plated on top of the pre-set basement gel layer and placed at 4 °C for 10 min. Once set, the cells were incubated at 37 °C in a humidified atmosphere (5% CO2) for 28 days. After 28 days, cells were fixed and stained with 5% MTT (3-(4,5-Dimethylthiazol-2-yl)-2,5-Diphenyltetrazolium Bromide (Sigma-Aldrich Company Ltd., Gillingham, UK), for 4 h and counted using an Optronix Gel Count (Oxford Optronix, Oxford, UK). The number of colonies for siControl, ATF2-siRNA1 and ATF2-siRNA2 transfected cells was calculated relative to the un-transfected (vehicle) control. Results are expressed as the mean ± SD of triplicate wells from three independent experiments.
Invasion and scratch-wound assay
To test the invasive and migratory ability of ATF2-silenced cells, ATF2 knockdown was performed as previously described. The invasion assay was performed in Transwell® plates with polycarbonate membrane inserts pre-coated with Matrigel™ (BD Biosciences, Oxford, UK). The cells were seeded onto the upper compartment filled with 100 μl standard culture medium with 2% (v/v) FBS. The lower compartment was filled with 500 μl medium with 10% (v/v) FBS. The invasive cells that crossed the membrane in a 24-h period were fixed, stained with a 1% Crystal Violet Reagent (Sigma-Aldrich Company Ltd., Gillingham, UK). Cell counting was calculated by light microscope for siControl, ATF2-siRNA1 and ATF2-siRNA2 transfected cells and was calculated relative to the un-transfected (vehicle) control. Results are expressed as the mean ± SD of triplicate wells from three independent experiments.
ATF2-silenced cells were also seeded in 6-well plates at a density of 2.5 × 105 cells/well and an artificial gap was created with a yellow pipette tip. The cells were rinsed several times with the appropriate medium to remove dislodged cells and images of living cells were captured at the indicated time points of 0, 12 and 24 h at a magnification of × 4 using an inverted microscope (Nikon Eclipse TE 2000-U). Images were analysed by ImageJ 1.48v software (National Institute of Health, USA) averaging the position of the migrating cells at the wound edges. Results are expressed as the mean ± SD of triplicate wells.
Gene expression microarray analysis
Microarray analysis was performed to identify genes that were differentially regulated by ATF2 in TAMR cells when compared to the tamoxifen-sensitive MCF7. ATF2 knockdown and RNA extraction were performed as described above. Four biological replicates were prepared for each experimental condition. Gene expression analysis was carried out on Illumina Human HT12 version 4 arrays and deposited to
Gene Expression Omnibus (GEO) repository (GSE145548). All data analyses were performed on R using Bioconductor packages [
31]. Raw intensity data from the array scanner was processed using the BASH and HULK algorithms as implemented in the bead array package [
31]. Log2 transformation and quantile normalisation of the data was performed across all sample groups. Differential expression analysis was performed using the limma package [
32]. Differentially expressed genes were selected using a
p value cut-off of < 0.05 after application of false discovery rate (FDR) correction for multiple testing applied globally to correct for multiple contrasts. Differentially expressed genes were analysed by the ‘transcription factor network’ prediction tool EnrichR [
33] to identify top driver transcription factors and pathway analysis (KEGG and REACTOME) was performed using the publically available STRING (Functional protein association networks) database v11 (
string-db.org).
DNA methylation
ATF2 knockdown was performed as described and DNA/RNA was extracted using the DNeasy/RNeasy kits (Qiagen, Crawley, UK), respectively. Total RNA was reverse transcribed using the Quantitect RT kit (Qiagen, Crawley, UK). Pyrosequencing primers were designed using Pyromark Assay Design 2.0 software (Qiagen, Crawley, UK) to measure the DNA methylation levels of
ESR1 and
PGR after ATF2 knockdown and synthesised by Eurofins MWG Operon (Ebersberg, Germany). The primer sequences are listed in Supplementary Material and Methods, Table
3. Genomic DNA was treated with sodium bisulphite using the EZ DNA methylation Kit (Zymo Research, CA, USA). PCR amplifications were performed in a final volume of 25 μl using HotStarTaq Master Mix (Qiagen, Crawley, UK), 200 nM biotinylated primer, 400 nM non-biotinylated primer and 60 ng of bisulfite-treated genomic DNA. The thermal profile was 95 °C for 5 min followed by 40 cycles of denaturation at 94 °C for 30 s, annealing at 51–56 °C for 30 s and extension at 72 °C for 30 s. The PyroMark Gold Q96 SQA Reagents and the PyroMark Q96 ID instrument (Qiagen, Crawley, UK) were used for pyrosequencing analysis following the supplier’s protocol. The methylation index for each promoter was calculated as the mean value of mC/(mC + C), where C is unmethylated cytosine and mC is 50′ methyl-cytosine, for all examined CpGs in the target sequence.
Statistical methods
Student’s t test (two-sided) was used to compare any differences observed between each experimental group and the control group. All the t tests and the 95% CI (confidence interval) were calculated by GraphPad Prism 5. A p value less than 0.05 was considered significant.
Discussion
The development of endocrine therapy resistance appears to involve multiple divergent mechanisms including epigenetic changes affecting the expression of ER and its target genes or deregulated ER-mediated gene transcription [
3‐
10]. Therefore, a better understanding of the complexity of resistance will help us to identify novel targets able to revert or abolish resistance to endocrine therapy. Studies utilising cell lines and patient-derived xenografts models of acquired resistance have indicated impaired ER-mediated transcriptional activity, enrichment for genes involved in cell growth, cell survival and apoptosis and, more recently, activation of oestrogen non-genomic signalling, involving mainly the ERa/Src/PI3K complex [
38‐
41]. Blocking the formation of the ERa/Src/PI3K complex by a competitive peptide partially restored tamoxifen sensitivity in the resistant cells [
41]. Moreover, blocking the AP1 transcription complex (Jun, Fos, Maf and ATF2 protein subfamilies) leads to increased tumour sensitivity to endocrine therapy and delayed onset of resistance through the inhibition of both proliferative and survival signals [
40]. A recent study highlighted that ATF2 promotes the development and progression of urothelial cancer via cooperation with androgen receptor and EGFR/ERBB2/ERK pathway signalling and suggested that ATF2 inhibition in association with AR inactivation may be a potential effective therapeutic approach for urothelial cancer [
42]. Although there is conflicting data regarding the prognostic role of AR expression in endocrine treatment response, an in vitro effect of AR expression on tamoxifen resistance has been observed, potentially mediated through activation of EGFR signalling pathways and the adaptation of the tumour cells from oestrogen to androgen dependent has been demonstrated as another mechanism of resistance to ER targeted therapies [
13,
43,
44]. Therefore, silencing ATF2 may be beneficial to endocrine therapy resistance.
In a previous study from our group, we showed that ATF2 silencing leads to a loss in the growth-inhibitory effects of tamoxifen in the ER-positive, tamoxifen-sensitive MCF7 cell line and that tamoxifen treatment caused a dose-dependent phosphorylation of ATF2 within its activation domain, enhancing its transcriptional activity, suggesting a tumour-suppressive role of ATF2 in ER-positive breast cancer [
25]. In the current study, we demonstrated in vitro that the transcription factor ATF2, part of the AP1 complex, represents a key factor in acquired endocrine resistance by facilitating a shift towards ER-independent transcription and silencing of ATF2 leads to the reversal of this resistant phenotype.
Firstly, we demonstrated that knockdown of ATF2 had a differential effect on the growth and proliferation of MCF7 and the endocrine-resistant cell lines TAMR, LCC2 and LCC9. Interestingly, transient knockdown of ATF2 significantly inhibited the growth of TAMR, LCC2 and LCC9 but had little effect on the growth of MCF7 cells. In addition, ATF2 silencing was also associated with a significant reduction in colony formation in the endocrine-resistant cell lines in comparison to the MCF7-sensitive cells. These results indicate that ATF2 plays an important role in regulating growth of tamoxifen-resistant breast cancer cell lines, without affecting the growth of the tamoxifen-sensitive cells, by altering signalling pathways that lead to reduced cell proliferation and/or increased apoptosis and by affecting their anchorage-independent growth and tumorigenic potential. The effect of siATF2 in the induction of apoptosis in MCF7 and TAMR cells is presented in supplementary Fig.
5 whereas the lack of difference in the phosphorylation status of ATF2 in vitro between MCF7 and TAMR cells before or after siATF2 (supplementary Fig.
1E) highlights further the effect of ATF2 as a transcription factor and not due to its activity. We previously showed that high expression of p-ATF2 (69/71) was associated with longer disease-free (DSS) and breast cancer-specific survival (BCSS) in ER-positive high-risk patients exposed to tamoxifen [
25]. However, these patients do not represent a tamoxifen-resistant model to compare it to the current in vitro work. Our data is similar to a number of studies mainly in melanoma, pancreatic and oesophageal cancer where inhibition or silencing of ATF2 leads to an induction of apoptosis and inhibition of tumour growth and metastasis [
45‐
48].
We further showed that silencing ATF2 alters the chemotactic and migratory properties of the resistant cell lines and reverts the migratory capacity of the invasive TAMR, LCC2 and LCC9 cells with no effect in the less invasive MCF7 cells. This highlights further the distinct transcriptional activities between the endocrine resistant and sensitive cells and the importance of ATF2 in reverting the malignant phenotype and restoring sensitivity to endocrine therapy.
The mechanisms by which ATF2 regulates the growth and migratory ability of TAMR cells in comparison with MCF7 was evaluated by microarray gene expression analysis of both cell lines before and after ATF2 silencing. Following ATF2 silencing, analysis was carried out to identify genes that were differentially expressed in TAMR cells without changing in MCF7. This revealed a number of ER-responsive genes and pathways regulated by ATF2 in tamoxifen resistance and highlighted a shift from the ER-enriched towards an ATF2-enriched transcription factor network. Although the endogenous levels of ATF2 were not found to be significantly different between the cell lines according to our microarray data (reconfirmed by qRT-PCR and western blot), enhanced expression of
ATF3 and
ATF4 (members of the ATF family) was observed in TAMRs, suggesting enhanced expression and activity of ATF transcription factors in tamoxifen resistance. Knockdown of ATF2 in TAMR cells differentially inhibited the expression of genes that are well documented in the literature to play key roles in mediating tamoxifen resistance such as
VEGFA,
ABL1,
FOXM1 and
DNMT1 [
3,
7,
11,
12,
49,
50]. Moreover, ATF2 knockdown also inhibited the expression of
CDK4 and
ATF3 which were found to be overexpressed in TAMR cells relative to MCF7. Many of the genes differentially expressed in TAMRs cells after ATF2 silencing (Supplementary file
3) are drug targets and could be potentially used to overcome the endocrine-resistant phenotype. For instance, inhibition of ABL in ER-positive breast cancer resulted in sensitization to anti-oestrogen therapies [
49,
51,
52] whereas CDK4/6 inhibitors have demonstrated clear clinical efficacy in the treatment of ER-positive, HER2-negative breast cancer when combined with endocrine therapy [
53,
54]. DNMT1 overexpression in TAMR cells leads to aberrant methylation of the PTEN promoter resulting into loss of PTEN expression and activation of the PI3K/AKT pathway. Therefore, drugs targeting DNMT1 (azacitidine, decitabine) may be of potential clinical use [
55,
56]. CTGF overexpression has been correlated to decreased survival and endocrine resistance in ER-positive breast cancer patients and identified as a potential therapeutic target to overcome resistance [
57] whereas ruxolitinib blocked the EMT process and VEGF production through the JAK/STAT3 pathway, consequently suppressing tamoxifen-resistant cell migration and angiogenesis [
58]. We also demonstrated IGF1R expression was upregulated after silencing ATF2. Of note, a randomised study targeting IGF1R with ganitumab in combination with endocrine therapy for hormone-receptor positive locally advanced or metastatic breast cancer demonstrated a significantly worse overall survival in those women receiving ganitumab as compared to placebo [
59]. Previously, it had been reported that IGF1R is reduced in tumour biopsy at the time of recurrence or resistance to tamoxifen and time to progression was significantly increased for IGF1R rich patients [
60]. These observations alongside ours would support a role for agonising IGF1R in ER-positive breast cancer.
Knockdown of ATF2 also resulted in increased expression of
ESR1 and ER-target genes, including
TFF1,
GREB1,
NCOA3 and
PGR in TAMR cells but not in MCF7 cells. This was also reconfirmed in the protein levels indicating that knockdown of ATF2 in TAMR cells enhances ER-regulated gene expression and activity. Using the Kaplan-Meier Plotter (
https://kmplot.com/analysis/) [
61] for breast cancer we observed that mRNA expression of NCOA3, GREB1 but not ATF2 and TFF1 correlated with progression-free survival (PFS) whereas NCOA3, GREB1 and TFF1 correlated with overall survival (OS) in ER-positive breast cancer patients systemically-treated with tamoxifen only (Supplementary Fig.
6). The KM plotter data for ATF2 is in agreement with our previous work utilising clinical samples [
25]. However, it cannot distinguish between tamoxifen-sensitive and tamoxifen-resistant patients. In the current study, we did not examine any clinical samples as out work was purely in vitro models of endocrine resistance.
Pathway analysis of genes upregulated in TAMRs after ATF2 silencing showed very strong enrichment for adherens junction, oestrogen and nuclear receptor signalling and drug metabolism whereas similar analysis of genes downregulated in TAMRs after ATF2 silencing showed enrichment for the ribosome (Supplementary files
1 and
2). The pathway analysis confirms our functional studies of anchorage-independent cell growth and migration and is in agreement with a previous study that observed an increased activity of mitochondrial biogenesis and expression of ribosomal genes in the anchorage-independent cell growth gene signature [
62].
Finally, DNA methylation has been implicated in the development of tamoxifen resistance [
6,
7]. However, in our study, although tamoxifen-resistant cells showed slightly increased global methylation compared to parental MCF7 at basal level, ATF2 knockdown did not trigger any significant changes indicating that ATF2 does not affect the methylation profile of these cell lines.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.