The views expressed in this article do not necessarily represent those of the U.S. Food and Drug Administration.
Cancer cells that develop resistance to chemotherapeutic agents are a major clinical obstacle in the successful treatment of breast cancer. Acquired cancer chemoresistance is a multifactorial phenomenon, involving various mechanisms and processes. Recent studies suggest that chemoresistance may be linked to drug-induced dysregulation of microRNA function. Furthermore, mounting evidence indicates the existence of similarities between drug-resistant and metastatic cancer cells in terms of resistance to apoptosis and enhanced invasiveness. We studied the role of miRNA alterations in the acquisition of cisplatin-resistant phenotype in MCF-7 human breast adenocarcinoma cells. We identified a total of 103 miRNAs that were overexpressed or underexpressed (46 upregulated and 57 downregulated) in MCF-7 cells resistant to cisplatin. These differentially expressed miRNAs are involved in the control of cell signaling, cell survival, DNA methylation and invasiveness. The most significantly dysregulated miRNAs were miR-146a, miR-10a, miR-221/222, miR-345, miR-200b and miR-200c. Furthermore, we demonstrated that miR-345 and miR-7 target the human multidrug resistance-associated protein 1. These results suggest that dysregulated miRNA expression may underlie the abnormal functioning of critical cellular processes associated with the cisplatin-resistant phenotype.
Breast cancer is the most common malignancy in women. In the United States, the incidence of invasive breast cancer, the most serious form of breast cancer, was estimated as 182,460 new cases and 40,480 deaths in 2008.1 Despite advances in understanding the molecular mechanisms of breast cancer biology, as well as advances in early detection and treatment, in 50% of cases cancer cells will either rapidly acquire resistance against numerous cytotoxic drugs or are intrinsically resistant.2 Furthermore, ∼30% of all patients with early-stage breast cancer will have recurrent disease, which becomes predominantly metastatic and resistant to treatment.3, 4 It is believed that resistance to chemotherapy or administration of ineffective chemotherapeutic agents causes treatment failure in 90% of patients with metastatic cancer.3 Currently, acquired drug resistance and metastasis are major obstacles in the successful treatment of breast cancer.5, 6 Thus, increasing tumor cell sensitivity to chemotherapeutic agents and predicting chemotherapeutic agent effectiveness without developing drug resistance in individual patients are attractive goals for improving the clinical management of cancer.
Recent evidence suggests that drug-induced dysregulation of microRNA (miRNA) function may modulate the sensitivity of cancer cells to chemotherapeutic agents and may be involved in the acquisition of cancer cell resistance to chemotherapy,7, 8 including breast cancer drug resistance,9–11 and breast cancer metastasis.12, 13 In addition, there is increasing evidence that there are great similarities between drug-resistant and metastatic cancer cells; particularly, in terms of profound resistance to apoptosis and enhanced invasiveness.14 With this in mind, the aims of this study were as follows: (i) to determine whether or not the acquired drug-resistant phenotype of breast cancer cells to cisplatin (cis-dichlorodiammine platinum (II) (CDDP)), a major chemotherapeutic agent used to treat a range of human malignancies,15 is associated with secondary miRNA alterations and (ii) to determine whether or not these miRNA abnormalities are associated with specific known mechanisms of cisplatin-based resistance.
ATCC: American Type Culture Collection; CDDP: cisplatin (cis-dichlorodiammine platinum (II); DAPI: 4′,6-diamidino-2-phenylindole; DNMTs: DNA methyltransferases; DNMT3A: DNA methyltransferase 3A; EMT: epithelial to mesenchymal transition; IC50: inhibitory concentration to produce 50% cell death; LOWESS: locally weighted regression; MBD: methyl-CpG-binding domain proteins; MeCP2: methyl-CpG-binding protein 2; miRNA: microRNA; MRP1: multidrug resistance-associated protein 1; qRT-PCR: quantitative real-time PCR
Material and Methods
Cell lines and cell culture
The MCF-7 and MDA-MB-231 human breast cancer lines were obtained from the American Type Culture Collection (ATCC, Manassas, VA) and were maintained according to ATCC's recommendations. Cisplatin (CDDP) is a chemotherapeutic agent used to treat a range of cancers including ovarian and breast cancer.16, 17 The MCF-7 drug-resistant variant to CDDP (MCF-7/CDDP) cell line was established by stepwise selection after prolonged (>6 months) treatment of MCF-7 cells to increasing concentrations of CDDP (Sigma) at a range of 0.5–25 μM in the medium.18 After 6 months of culturing in the presence of CDDP, the IC50 (inhibitory concentration to produce 50% cell death) values were 94 and 12 μM of CDDP for the MCF-7/CDDP and parental MCF-7 cells, respectively. Cells were seeded at a density of 0.5 × 106 viable cells per 100 mm plate, and the medium was changed every other day for 6 days. Trypsinized cells were washed in phosphate-buffered saline and immediately frozen at −80°C for subsequent analyses. The experiments were independently reproduced twice, and each cell line was tested in triplicate.
miRNA microarray expression analysis
Total RNA was extracted from MCF-7 and MCF-7/CDDP cells using TRIzol reagent (Invitrogen, Burlington, ON) according to the manufacturer's instructions. The miRNA microarray analysis was performed by LC Sciences (Houston, TX). Ten micrograms of total RNA was size-fractionated (<200 nucleotides) by using a mirVana kit (Ambion, Austin, TX) and labeled with Cy3 and Cy5 fluorescent dyes. Dye switching was performed to eliminate dye bias. Pairs of labeled samples were hybridized to dual-channel microarrays. Microarray assays were performed on a μParaFlo microfluidics chip with each of the detection probes containing a nucleotide sequence of coding segment complementary to a specific miRNA sequence and a long nonnucleotide molecule spacer that extended the detection probe away from the substrate. A miRNA detection signal threshold was defined as twice the maximum background signal. The maximum signal level of background probes was 180. Normalization was performed using a cyclic LOWESS (locally weighted regression) method to remove system-related variations.10, 19 Data adjustments included data filtering, log2 transformation, gene centering and normalization. T-test analysis was conducted between MCF-7 and MCF-7/CDDP samples, and miRNAs with p-values < 0.05 were selected for cluster analysis. The clustering analysis was performed using a hierarchical method and average linkage and Euclidean distance metrics.10, 20
Quantitative real-time PCR analysis for miRNA expression
The qRT-PCRs were performed by using SuperTaq Polymerase (Ambion, Austin, TX) and a mirVana qRT-PCR miRNA Detection Kit (Ambion) following the manufacturer's instructions. Reactions contained mirVana qRT-PCR primer sets (Ambion) specific for human miR-127, miR-126, miR-200c, miR-29a, miR-29b, miR-206 and miR-345. Human 5S rRNA served as an internal control. QRT-PCR was performed on a SmartCycler (Cepheid, Sunnyvale, CA), and each cell line was run in triplicate. The level of each miRNA expression was measured using the 2−ΔΔCt method.10, 21 The results are presented as fold change of each miRNA in the MCF-7/CDDP cells relative to the parental MCF-7 cells. Indicated changes are significant at 95% confidence level (p < 0.05).
Western blot analysis of protein expression
The protein levels of DNA methyltransferase 3A (DNMT3A), methyl-CpG-binding protein 2 (MeCP2), ZEB1, E-cadherin, MRP1 and β-actin in the MCF-7, MCF-7/CDDP cells were determined by Western blotting using protocols described previously.10, 18, 22
Expression of ZEB1 and E-cadherin in the MCF-7 and MCF-7/CDDP cells was detected by immunofluorescence. Cells were cultured on glass coverslips for 24 hr and fixed in PBS containing 0.4% paraformaldehyde. Fixed and permeabilized cells were incubated with the primary ZEB1 (1:50, Santa Cruz Biotechnology) and E-cadherin (1:50, Cell Signaling). After washing, the cells were incubated with Alexa Fluor secondary antibodies and counterstained with 4′,6-diamidino-2-phenylindole (DAPI).
Cytosine DNA methylation analysis
The extent of global DNA methylation was evaluated with a cytosine extension assay as described previously.18, 23
Analysis of the invasiveness of cisplatin-resistant MCF-7 cells
MCF-7 and MCF-7/CDDP cells were plated in the upper chamber of BD BioCoat Matrigel Invasion Chambers (BD Biosciences, Bedford, MA) and grown as recommended by the manufacturer. After 48 hr, cells on the bottom side of membrane were stained and mounted onto glass slides, and the mean number of cells in 4 replicates was determined. Representative images (magnification 125×) of stained membranes are shown.
Luciferase reporter assay for targeting MRP1-3′-UTR
Cloning of the UTR was based on transcript NM_004996 for the MRP1(Abcc1) gene. UTR was defined as the sequence between 4772 and 6564 bps. For the luciferase reporter experiments, a 3′-UTR segment of the MRP1 gene (nucleotides 4742–6564) that contains putative binding regions for hsa-miR-345 (nucleotides 4819–4838, NM_004996) and hsa-miR-7 (nucleotides 5346–5368) was amplified by PCR from human genomic DNA using primers that included an XbaI and EcoRI tails on the 5′ and 3′ strand, respectively. PCR products were restricted with both XbaI and EcoRI restriction endonucleases and then gel purified. The amplified 3′-UTR of MRP1 contains an XbaI restriction site; therefore, MRP1-3′-UTR was ligated into the pGL3-control vectors (Promega, Madison, WI) by using the XbaI site located immediately downstream of the luciferase stop codon. In parallel, we mutated the miR-7 (nucleotides 5361–5368) and miR-345 (4830–4837) seed sequence-binding regions in the 3′-UTR segment of the MRP1 gene to 5′-AGAGGAAG-3′ and 5′-AAGAGAGA-3′, respectively.
HEK293 cells were transfected with firefly luciferase MRP1-3′-UTR construct or mutated MRP1-3′-UTR construct, a control Renilla luciferase pRL-TK vector (Promega) and synthetic precursors of miR-345 and miR-7 (Ambion) using lipofectamine 2000 reagent according to the manufacturer's protocol (Invitrogen, Carlsbad, CA). Scrambled oligonucleotides or unrelated miRNA that is not predicted to target MRP1 (miR-127) (Ambion) served as controls.
Twenty-four hours after transfection, cells were lysed with a 1× passive lysis buffer, and the activity of both Renilla and firefly luciferases was assayed using the dual-luciferase reporter assay system (Promega) according to the manufacturer's instructions.10
Analysis of the effect of miR-7 and miR-345 on the cellular MRP1 levels
MCF-7/CDDP cells were transfected with miR-7, miR-345 or scrambled RNA oligonucleotides (100 nM). Twenty-four hours after transfection, the cellular levels of MRP1 were detected by Western immunoblotting using anti-MRP1 antibodies (Abcam) according to the manufacturer's instructions.
Statistical analysis was performed using MS Excel 2007 and JMP5 software packages.
Results and Discussion
Microarrays were used to analyze the expression of miRNAs in the MCF-7 human breast adenocarcinoma cell line and its cisplatin-resistant variant MCF-7/CDDP. Cluster analysis revealed that the MCF-7/CDDP cells were characterized by significant changes in miRNA expression. We identified 103 miRNA genes (46 upregulated and 57 downregulated) that were differentially expressed (p < 0.01) in the MCF-7/CDDP cells compared to the parental MCF-7 cells (Supporting Information Fig. 1). Microarray data were confirmed by qRT-PCR. Table 1 lists a number of miRNAs that exhibited pronounced changes in expression in the MCF-7/CDDP cells when compared with the parental MCF-7 breast cancer cells. Specifically, the MCF-7/CDDP cells were characterized by the greatest alterations in miRNAs involved in the control of several indispensable cellular processes and pathways, including cell signaling, cell survival and apoptosis, invasiveness and DNA methylation (Table 1). The most upregulated miRNAs in the MCF-7/CCDP cells were miR-146a, miR-10a and miR-221/222 (Table 1). These miRNAs regulate the cellular levels of breast cancer-associated BCRA1 protein, homeobox family HOXD10, tumor suppressor p27 and estrogen receptor α.11, 12, 26–29 The role of these proteins in the pathogenesis of breast cancer, cancer drug resistance and metastasis is well established.11, 12, 26–29
Table 1. miRNA expression profile in MCF-7 and MCF/CDDP breast cancer cells
Additionally, several miRNAs (miR-29a, miR-29b, miR-132 and miR-194), upregulated in MCF-7/CDDP cells, are known to target DNA methyltransferases (DNMTs) and MeCP2 and therefore affect DNA methylation patterns.30–32 Specifically, we found a significant upregulation of miR-29a and miR-29b (14.6 and 52.4 times, respectively), which target de novo DNMT3A and -3B.30 DNMT3A and -3B are de novo methyltransferases responsible for setting up DNA methylation patterns in the cells. Recently, Fabbri et al. have shown that the cellular levels of miR-29a and miR-29b are inversely correlated to DNMT3A and -3B in cancer tissues, and that miR-29a and miR-29b directly target both DNMT3A and -3B. Overexpression of these miRNAs led to significant decreases in the DNMT3A and -3B levels.30
In addition, MCF-7/CDDP cells were characterized by a significant (56.9 times) increase in the levels of miR-132 that targets MeCP2.31 MeCP2 is a transcriptional repressor that belongs to a family of methyl-CpG-binding domain proteins (MBD). MeCP2 selectively recognizes methylated DNA and plays a central role in chromatin remodeling. Recently, Klein et al. reported that MeCP2 translation is directly regulated by miR-132. Furthermore, blocking miR132-mediated repression effectively increased cellular MeCP2 levels.31
In light of this, we analyzed whether or not the upregulation of miR-29a, miR-29b and miR-132 in the drug-resistant MCF-7/CDDP cells resulted in altered cellular levels of DNMT3A and MeCP2. As predicted, we found a significant decrease in the cellular levels of DNMT3A and MeCP2 in the MCF-7/CDDP drug-resistant cells (Fig. 1a).
As dysregulation of DNA methyltransferases and methyl-binding proteins is often associated with altered levels of DNA methylation, we next studied the status of global DNA methylation in parental MCF-7 cells and its MCF-7/CDDP drug-resistant variant. A well-established cytosine extension assay confirmed that decreased cellular levels of DNMT3A and MeCP2 were accompanied by a significant decrease in global DNA methylation in the MCF-7/CDDP cells (Fig. 1b).
It has been suggested that one of the features of cancer drug-resistant cells is enhanced invasiveness.14 Indeed, we detected altered expressions of several miRNAs, including miR-10a, miR-10b, miR126, members of miR-200 family (miR-200b, miR-200c, miR-141 and miR-429) and miR-205, whose aberrant expression has been linked to increased metastatic properties of breast cancer cells.12, 13, 33 Amongst these miRNAs, the most pronounced changes were specific for miR-200c and miR-200b, whose levels in MCF-7/CDDP cells were 497 and 1,000 times lower, respectively, when compared with parental MCF-7 cells (Table 1). These miRNAs play a major role in defining cellular epithelial phenotype by suppressing the expression of ZEB1/deltaEF1 and SIP1/ZEB2.33 Both ZEB1 and SIP1 are transcriptional repressors of E-cadherin and have been implicated in the epithelial to mesenchymal transition (EMT). Manipulation of the miR-200 family can induce EMT.34 EMT has been implicated in tumor progression and metastasis, and a major step in this process is the downregulation of E-cadherin.35, 36 A recent study by Gregory et al. provided evidence that ZEB1 and SIP1 expression is controlled by the miR-200 family and that downregulation of miR-200 is an essential early step in tumor metastasis.33
In the MCF-7/CDDP cells, the decreased levels of the miR-200 family (Table 1) were associated with increased levels of ZEB1 protein and consequently with the decreased expression of E-cadherin (Fig. 2a). Although ZEB1 is a nuclear protein, we noticed that immunofluorescence for ZEB1 still indicated some weak cytoplasmic staining for MCF-7 cells. In contrast, MCF-7/CDDP cells exhibited strong nuclear staining and some relatively strong cytoplasmic staining. Some weak cytoplasmic staining is sometimes observed in certain tumor cells.37–39 This may be indicative of some basal levels of translation. Indeed, some basal low levels of ZEB1 are also seen on the ZEB1 Western blot (Fig. 2b). Furthermore, the MCF-7/CDDP cells exhibited increased invasiveness when compared with the MCF-7 cells (Fig. 2b). Interestingly, the extent of the invasiveness of the MCF-7/CDDP cells was similar to the highly invasive MDA-MB-231 breast cancer cells (data not shown). More importantly, the IC50 (inhibitory concentration to produce 50% cell death) values for the MDA-MB-231 cells to CDDP (100 μM) were similar to that observed in the MCF-7/CDDP cells.
Regulation of drug efflux is yet another key mechanism involved in drug resistance.40 Specifically, one of the mechanisms of resistance to platinum-based chemotherapeutic drugs in cancer cells is the formation of platinum and glutathione intracellular complexes24, 25, 41; however, these complexes are themselves toxic. It has been suggested that MRP1 and MRP2 may mediate the resistance of cancer cells to cisplatin via enhancing the transport of cisplatin-glutathione-S-conjugate out of cancer cells.24, 25 Indeed, the results of Negoro et al. recent study demonstrated the important role of MRP1 and MRP2 upregulation in the establishment of a cisplatin-resistant cell line.42
Using Western immunoblotting, we determined that cisplatin-resistant MCF-7/CDDP cells were characterized by a significant upregulation of MRP1 when compared with the sensitive MCF-7 cells (Fig. 3).
Therefore, as a next step, we analyzed the profile of miRNA expression in MCF-7 and MCF-7/CDDP lines to identify any novel miRNAs that may potentially target efflux pumps. The results of miRNA microarray and qRT-PCR analyses demonstrate that one of the miRNAs with the greatest difference in MCF-7/CDDP expression was miR-345, whose expression was 17.0 times lower when compared with parental MCF-7 cells. While examining potential targets of miR-345 using several in silico methods for target gene predictions, such as Sanger miRNA database43 and miRGen miRNA database,44 we noted that this miRNA may target the 3′-UTR of the human MRP1 (Abcc1) gene. Interestingly, the MRP1 gene was also predicted to be targeted by miR-7. MiR-7 levels are 2.3 times downregulated in MCF-7/CDDP cells when compared with MCF-7 cells. Detailed in silico analysis revealed that the 3′-UTR of MRP1 contains putative regions (nucleotides 4819–4838, NM_004996 and nucleotides 5346–5368, NM_004996) that match the sequences of hsa-miR-345 and hsa-miR-7, respectively (Fig. 4a).
To validate these predicted miRNA-target interactions, the segment of MRP1-3′-UTR (nucleotides 4742–6564) containing the miR-345 and miR-7 complementary sites was cloned into the 3′-UTR of a luciferase reporter system. Additionally, we have mutated the miR-7 and miR-345 seed-interacting sequences in the MRP1-3′-UTR. The resulting reporter vectors were transfected into HEK293 cells together with miRNA that do not have binding sites within the 3′-UTR of the MRP1-luciferase UTR-report vector and either with miR-345 or miR-7 alone or with both miR-345 and miR-7. No change in the luciferase reporter activity was observed in HEK293 cells that were cotransfected with negative control (scrambled oligonucleotides) or unrelated miRNAs such as miR-127. In contrast, transfection of the HEK293 cells with either miR-345 or miR-7, or both miRNAs together, significantly inhibited luciferase activity from the construct with the MRP1-3′-UTR segment. No luciferase inhibition was noted in the constructs containing mutated MRP1-3′-UTR segments (Fig. 4b).
Most importantly, transfection of the MCF-7/CDDP cells with miR-345 and miR-7 resulted in a significant decrease in the cellular levels of MRP1 24 hr after transfection (Fig. 4c) and an increased sensitivity of MCF-7/CDDP to CDDP, as evidenced by a decrease in IC50 values from 94 μM in MCF-7/CDDP cells to 43 μM in MCF-7/CDDP cells transfected with miR-345 and miR-7 (Fig. 4d).
Furthermore, one of the predicted targets of miR-489, downregulated in MCF-7/CDDP cells, is MRP2 (Table 1). Thus, we concluded that miRNAs may be important regulators of the cellular levels of efflux pump proteins such as MRP1 and MRP2.
Furthermore, this data further substantiate the previously reported role of miR-451 in the regulation of another efflux pump—MDR1, in another cancer drug-resistant cell line model—breast adenocarcinoma cells resistant to doxorubicin.10
Overall, acquired drug resistance is a multifactorial phenomenon, involving multiple mechanisms and processes that include decreased uptake of drugs, alterations in cell cycle and signal transduction pathways, increased repair of DNA damage, reduced apoptosis, increased efflux of hydrophobic drugs, DNA damage tolerance and altered DNA methylation and chromatin structure.
The results of our study demonstrate that dysregulation of miRNA expression is associated with abnormal functioning of some of the critical cellular processes associated with the drug-resistant phenotype in MCF-7/CDDP cells. Specifically, we identified miRNA changes associated with increased efflux of drugs, changed DNA methylation and altered DNA repair. We also found miRNAs that may contribute to increased EMT and the invasiveness of CDDP cells. However, these miRNA alterations are not necessarily indicative of the causative role of miRNA dysregulation in cancer drug resistance development. The ultimate goal of future studies is to address and identify these roles during the stepwise acquisition of molecular changes during development of drug resistance, among other questions.
Our study thus serves as a roadmap for the future analysis of the roles of miRNAs in drug resistance.
This research was supported by Alberta Health Services/Alberta Cancer Board grant to O.K. J.F. was a recipient of NSERC and AHFMR Graduate Scholarships. The authors are grateful to Dr. Valentina Titova and Ms. Diane Harms for careful proofreading of this manuscript.