Targeting of miR34a–NOTCH1 Axis Reduced Breast Cancer Stemness and Chemoresistance

The cancer stem cell hypothesis offers an important avenue in the field of cancer therapy. Cancer stem cells are thought to trigger tumor and cancer metastasis, drug resistance, and cancer recurrence (1). A subpopulation of cells with stem cell properties was identified in breast cancer cells, and this population of cells was called breast cancer stem cells (BCSC; ref. 2). BCSCs have an enriched CD44⁺CD24⁻ population and can form a mammosphere under specific growth conditions (1, 2). In recent studies, it was reported that Notch and Wnt signaling is the main cell fate regulator in breast cancer (3, 4), making them potential therapeutic targets in BCSCs.

miRNAs are a conserved class of noncoding small RNAs that regulate the expression of certain key genes involved in self-renewal, survival, and tumor progression (5). In recent studies, some miRNAs, e.g., let-7, miR451, miR128, and miR34, reportedly regulated cancer stemness and drug resistance in breast cancer (1, 6–8). In particular, members of the miR34 family, which includes miR34a and miR34b/c, are known as tumor suppressors that are inactivated in various tumors (9, 10). This reduced expression of members of the miR34 family has been associated with the hypermethylation of its neighboring CpG island or mutation of p53 (11). The miR34c gene is repressed by DNA methylation in breast cancer cells (12), and methylation-associated silencing of miR-34b/c has been shown to occur in gastric cancer cell lines (100%) and primary gastric cancer tissues (70%; ref. 13). More importantly, miR34 family genes are direct targets of p53, and dysregulation of the p53/miR34 axis increases the risk of cancer progression and invasion by controlling the expression of target genes involved in the cell cycle or DNA damage response in various types of cancer (6, 14).

The resistance of tumor cells to chemotherapy continues to be a critical issue in the field of clinical therapy for breast cancer. Previously, it was reported that MCF7/ADR cells contained a larger CD44⁺/CD24⁻ population and were more invasive than MCF7 cells in vivo (15). The MCF7/ADR cell line is breast adenocarcinoma cells and is resistant to doxorubicin (Dox; Adriamycin). In a recent study, miRNA expression patterns in these two cell lines were examined using miRNA microarrays. It was found that the expression levels of miR127, let-7, and miR34a were decreased in MCF7/ADR cells (7). Furthermore, in the study of Ogretmen and Safa (16), codons 126 to 133 in exon 5 of p53 were shown to be deleted in MCF7/ADR cells and the resulting mutated p53 protein had a longer half-life in the nucleus than wild-type p53. Taken together, our study will provide evidence to indicate a possible link between miRNA dysregulation and cancer drug resistance by controlling cancer stemness.

In this study, we found that the expression levels of members of the miR34 family were decreased in MCF7/ADR cells compared with MCF7 cells. We showed that the reduced expression of miR34a resulted in the overexpression of NOTCH1, which is a target gene of miR34a. The ectopic expression of miR34a decreased the stem cell properties of MCF7/ADR cells and sensitized these cells to Dox. In addition, xenograft tumors were reduced in nude mice that overexpressed miR34a following lentivirus treatment. The findings of this study indicate a novel role of miR34a in cancer stem cells involved in tumor progression. Collectively, this study proposes miR34a–NOTCH1 axis as a potential target for the treatment of drug-resistant breast cancer.

The cell lines MCF7 and MCF7/ADR were provided by Dr. YM Park (Roswell Park Cancer Institute, Buffalo). MCF7 and MCF7/ADR cell lines were tested using short tandem repeat markers for DNA fingerprinting analysis (Korean Cell Line Bank). We confirmed that the cell lines are not cross-contaminated. MCF7 and MCF7/ADR cells were maintained in DMEM supplemented with 10% FBS (Gibco). For mammosphere culture, cells (1,000 cells/mL) were cultured in suspension in serum-free DMEM F12 (WELGENE) supplemented with 1% penicillin, B27 (1:50, Gibco), 20 ng/mL epidermal growth factor (Prospec), 5 mg/mL insulin (Sigma), and 0.4% BSA (Sigma). After 10 to 12 days of culture, the plates were analyzed for mammosphere formation, which was quantified using a microscope (Olympus).

Total RNA was extracted using easyBLUE (Intron Technology) according to the manufacturer's instructions. Reverse transcription was performed using a TaqMan microRNA assay and TaqMan reverse transcription kit (PN: 4366596; Applied Biosystems). qRT-PCRs were performed with the TaqMan universal PCR master mix (PN: 4440040; Applied Biosystems) on an ABI Real-time PCR 7500 system using human RNU6B or RNU48 as endogenous control. Primers for mature miR34a, miR34b, miR34c, RNU6B, and RNU48 were obtained from Ambion. All reactions were performed in triplicate.

MCF7 and MCF7/ADR cells were transfected for 48 hours with miRNA mimics (15 and 30 nmol/L) using the siPORT NeoFX transfection agent (AM4511; Ambion). Pre-miRNA and negative control precursors were obtained from Ambion. After seeding, MCF7 and MCF7/ADR cells were transfected with NOTCH1 and control siRNA (30 nmol/L; Santa Cruz Biotechnology) using RNAiMax (Invitrogen) according to the manufacturer's protocol. After 48 hours, the cells were collected for Western blotting or resuspended in mammosphere medium. MCF7 cells were transfected p53 dominant-negative vector (631922; Clontect) for 48 hours using the FuGENE transfection agent (Promega).

Proteins were extracted from adherent or mammosphere cells using a NucleoSpin RNA/Protein extraction kit (Macherey-Nagel). The samples were subjected to SDS-PAGE (10%–12%), and the separated proteins were immunoblotted with primary antibodies against NOTCH1 (Cell Signaling Technology) and β-actin (Enzo). The PVDF filters were washed with PBS/0.1% Tween-20, and bound proteins were detected using an enhanced chemiluminescence system (Amersham Pharmacia Biotech).

We assessed the expression of CD44 and CD24 surface markers after transfection with miR34a mimics (PM11030; Ambion) or NOTCH1 siRNA (Santa Cruz Biotechnology). Briefly, cells were washed with PBS and stained with anti-CD44 (APC-conjugated; BD Biosciences) or anti-CD24 (phycoerythrin-conjugated; BD Biosciences) antibody in PBS containing 1% FBS and incubated on ice in the dark for 30 minutes. The cells were washed again with cold PBS, and >10,000 cells were analyzed by flow cytometry using a BD Canto II flow cytometer (BD Biosciences). The data were analyzed using FACSDiva software (BD Biosciences).

Reporter constructs containing the 3′-untranslated region (UTR) of NOTCH1 were cloned into the pGL3-control vector. The 3′-UTR of NOTCH1 was amplified from genomic DNA of HEK293T cells. The seed sequences of miR34 on NOTCH1 were mutated using a PCR-based approach. These “mutated” reporter constructs were verified by sequencing. HEK293T cells were transiently transfected with the 3′-UTR reporter constructs (1.5 μg/well in 6-well plates) and 15 nmol/L of the miR34 family mimics (Ambion) using Lipofectamine 2000 (Invitrogen). The activity of the 3′-UTR reporter constructs was normalized to the activity of the cotransfected vector pCMV-hRL (40 ng/well in 6-well plates; Promega). After incubation for 24 hours, the cells were lysed with 1× passive lysis buffer (Promega) and their activity was assessed using a dual-luciferase assay kit (Promega) according to the manufacturer's protocol.

Cell survival analysis following treatment of the MCF7 and MCF7/ADR cell lines with Dox (Sigma) was performed as described previously (7). Briefly, cells were seeded and transfected with scrambled miRNA precursor or pre-miR34a precursor (30 nmol/L; Ambion) using the siPORT NeoFX transfection agent (Ambion). After 48 hours, the cells were reseeded in 24-well plates at a density of 5.0 × 10⁴ cells/well and treated with Dox (2.5–100 μmol/L) for 72 hours. Cell survival was analyzed using the cell proliferation kit II (Roche) and absorbance was read at 470 nm on a microplate reader (Bioteck).

MCF7 and MCF7/ADR cells were seeded at 5.0 × 10⁵ cells in 60-mm culture dishes. The cells were treated with miR34a (30 nmol/L) or scrambled miRNA using the siPORT NeoFX transfection agent; Dox (10 or 25 μmol/L; Sigma) was added to the cells after 24 hours. After incubation for 48 hours, the cells were lysed and the whole cell lysates were analyzed using a caspase-3/CPP32 colorimetric assay kit (BioVision). The whole cell lysates were then incubated with 25 μmol/L aspartylglutamylvalinylaspartyl-7-amino-4-trifluoromethylcoumarin (DEVD-AFC; a fluorogenic substrate) at 37°C in a humid chamber under 5% CO₂ atmosphere for 2 hours. Finally, the amount of proteolytically cleaved AFC was measured at 405 nm using a fluorescence microplate reader (BioTek Instruments, Inc.) to calculate relative caspase-3 activation.

Apoptosis of MCF7 and MCF7/ADR cells was assessed after transient transfection with scrambled miRNA or miR34a precursor using the siPORT NeoFX transfection agent. After 24 hours, the cells were treated with Dox (25 μmol/L) and then cultured for additional 48 hours. After a final 72-hour culture period, the cells were incubated with an antibody mixture [Annexin V–FITC, propidium iodide (PI), and DAPI were dissolved in 1× binding buffer] for 15 minutes at 37°C in a humid chamber under 5% CO₂ atmosphere. For cell fixation, the cells were incubated with a 2% paraformaldehyde solution for 30 minutes at room temperature and then mounted with fluorescence VECTASHIELD mounting medium (Dako). Images were captured using the MetaMorph software and statistically analyzed. Quantitative graphs were obtained by calculating the number of PI-stained cells.

All studies involving the use of nude mice were approved by the Animal Care and Use Committee of Yonsei University Medical School (2013-0339) and performed in specific pathogen-free facilities and in accordance with the Guidelines for the Care and Use of Laboratory Animals of YUMS. Primary cells from nude mice were infected with control or miR34a lentivirus (GenTarget Inc.; multiplicity of infection, 3–5) in the presence of 5 μg/mL polybrene. Mice were anesthetized with 150 μL of saline/Zoletil/Rompun (7:1:1) and inoculated subcutaneously with 1 × 10⁶ stable MCF7/ADR primary cells into each flank. Mice were randomized into groups (n = 5 per group) and killed at 33 days after tumor cell inoculation. From palpable tumor formation until termination, the tumor size was measured every 2 to 3 days using calipers, and the tumor volume was calculated with the following formula: length × width² × 0.5236. Mice were killed by asphyxiation in a CO₂ chamber, and tumors were harvested for immunohistochemical and other analyses.

The one-tailed t test or one-way ANOVA was performed using GraphPad Prism6 software (Graphpad). A P value of <0.05 was considered significant.

Previous reports showed that MCF7/ADR cells had a better ability to form mammospheres than MCF7 cells and had a larger CD44⁺CD24⁻ population and side-population cell fractions (15). To confirm these results, we performed a mammosphere formation assay using MCF7 and MCF7/ADR cells. We found that MCF7/ADR cells formed larger mammospheres (Fig. 1A) and contained more highly expressing cells with the CD44⁺CD24⁻ phenotype than MCF7 cells (data not shown; MCF7/ADR: 25.8%, MCF7: 4.4%). In addition, stemness-related genes such as OCT4 and SOX2 were highly expressed in MCF7/ADR mammospheres (Supplementary Fig. S1). From these data, we conclude that chemoresistant MCF7/ADR cells have a greater cancer stem cell population than MCF7 cells.

We verified that the expression of three members of the miR34 family (a, b, and c) was significantly reduced in MCF7/ADR cells compared with MCF cells using a TaqMan assay in adherent and mammosphere status (Fig. 1B and Supplementary Fig. S3A). Recently, the miR34 family was found to be directly regulated by the tumor suppressor p53 (6, 17). We hypothesized that the reduced expression of members of the miR34 family was due to the p53 mutation of MCF7/ADR cells because MCF7 cells express wild-type p53, whereas MCF7/ADR cells express mutant p53 protein that is more stable and expressed at a higher level (16). We also confirmed the p53 mutation in our cell line. Codons 126 to 133 in exon 5 were deleted in MCF7/ADR cells, but this region was not mutated in MCF7 cells (data not shown). To confirm our hypothesis, we determined the miR34a level after treatment of MCF7 cells with p53 siRNA or dominant-negative p53 vector. Notably, p53 knockdown or p53 mutation in MCF7 cells led to a decrease in primary or mature miR34a levels (Fig. 1C and D). Taken together, these results suggest that the reduced levels of miR34 are related to the dysregulation of p53 in chemoresistant MCF7/ADR cells.

To identify target genes of miR34a, we searched for predicted target genes using TargetScan and miRBase. As a result, several predicted target genes of miR34a, e.g., Prealdolase A, Protein kinase D1, and NOTCH1, were ranked high in the obtained list. “Stem cell genes,” e.g., Notch, Hedgehog, Bmi-1, and Wnt/B-catenin, are important in cancer progression because they are involved in the regulation of self-renewal, differentiation, and survival of cancer stem cells (18). NOTCH1 is a stem cell gene that is related to mammary stem cells and mammary carcinogenesis (19). MiR34a was most downregulated in MCF7/ADR cells among various breast cancer cell lines (data not shown). As expected, the expression level of NOTCH1 was dramatically increased in MCF7/ADR cells and in MCF7/ADR mammospheres compared with those of MCF7 cells (Fig. 2A and B). Furthermore, all members of the NOTCH family were highly expressed in MCF7/ADR mammospheres (Supplementary Fig. S2).

To further examine whether the miR34 family affects NOTCH1 expression, we transfected HEK293T cells with miR34a, -b, and -c mimics and scrambled miRNA (negative control). The targeting function of miR34a to the 3′-UTR of NOTCH1 mRNA was examined using luciferase constructs that were cloned into the pGL3-control vector. In case of the wild-type NOTCH1 3′-UTR, the luciferase activity was decreased following ectopic miR34a expression; however, that was not observed in the mutant constructs (Fig. 2C and Supplementary Fig. S3B). In addition, the mRNA and protein levels of NOTCH1 were reduced following transfection with miR34a mimics compared with miR34b or -34c mimics (Fig. 2D and Supplementary Fig. S3C). Therefore, we focused on the functions of miR34a in subsequent experiments. Collectively, these results suggest that NOTCH1was negatively regulated by miR34a in chemoresistant MCF7/ADR cells.

Cancer stem cells enhance the tumor-initiating capacity and seem to be more drug resistant than other cancer cells. In many previous studies, cancer stem cells have been shown to regulate the expression of some miRNAs. In particular, members of the miR34 family were reportedly downregulated in prostate cancer stem cells, pancreatic cancer stem cells, and BCSCs (10, 12, 20). To confirm the role of miR34a in cancer stemness, we validated the effect of miR34a restoration on the self-renewal capacity of BCSCs using a mammosphere formation assay. As expected, cells expressing ectopic miR34a formed fewer mammospheres than cells transfected with negative control mimics (Fig. 3A and B). The control mammospheres were larger and grew more rapidly than those generated following miR34a restoration. Furthermore, silencing NOTCH1 using siRNA also reduced mammosphere formation (Fig. 3C and D). Next, we examined the CD44⁺/CD24⁻ population (a marker of BCSCs) following transfection with miR34a mimics or NOTCH1 siRNA. FACS analysis indicated that the overexpression of miR34a or downregulation of NOTCH1 reduced the CD44⁺/CD24⁻ population (Fig. 3E). These results provide evidence that miR34a can suppress the self-renewal capacity of BCSCs by targeting NOTCH1.

Drug resistance is an important issue for patients with cancer receiving cancer chemotherapy. Cancer stem cells are naturally resistant to chemotherapy, which enables them to survive treatment with anticancer drugs and supports the re-growth of other cancer cell populations (18, 21). Thus, we examined whether the downregulation of the miR34a expression contributed to Dox resistance in MCF7/ADR cells using an XTT assay. As expected, MCF7/ADR cells were more resistant than MCF7 cells to Dox treatment (2.5–100 μmol/L; Fig. 4A). The ectopic expression of miR34a in MCF7/ADR cells resulted in a significant reduction of the percentage of viable cells following treatment with Dox (Fig. 4A). Moreover, we measured caspase-3 activity, which is directly associated with increased apoptosis, using a colorimetric approach. We found that miR34a expression increased caspase-3 activity in MCF7/ADR cells following treatment with Dox (Fig. 4B). Next, to determine the percentage of Annexin-FITC/PI double-positive cells, we performed immunocytochemistry/fluorescence analysis (Fig. 4C and D). The percentage of double-positive apoptotic cells was decreased following transfection with miR34a mimics, indicating that miR34a mimics resensitized MCF7/ADR cells to Dox. Collectively, miR34a increases the sensitivity of chemoresistant MCF7/ADR cells to Dox, leading to apoptosis.

Next, we assessed the effect of restored miR34a expression on tumor formation in an in vivo model. Human cancer cells contain cancer stem cells and noncancer stem cells (22). Results of secondary xenograft assays showed that cancer stem cell populations can survive after dissociation from a primary tumor (22). As shown in Supplementary Fig. S4A and S4B, miR34a expression was increased; however, NOTCH1 expression was decreased in miR34a lentivirus-treated cells compared with control lentivirus–treated cells. Next, to establish secondary xenografts, stable cells (1 × 10⁶ cells) were injected into nude mice. As shown in Fig. 5A and B, at 33 days after injection, miR34a-expressing tumors were smaller and grew more slowly than control tumors. We determined the miR34a and NOTCH1 levels in MCF7/ADR xenograft tumors. As shown in Fig. 5C, the miR34a level was higher in miR34a lentivirus-treated tumors mass than in control tumors. Immunohistochemistry staining revealed that the expression of NOTCH1 was markedly decreased in miR34a lentivirus-treated xenografts compared with vehicle control–treated xenografts (Fig. 5D). Moreover, the percentage of Ki-67–positive cells was slightly reduced in miR34a-overexpressing tumor tissues. Our findings suggest that the reduced expression of miR34a contributes to breast cancer stemness and chemoresistance by regulating the expression of NOTCH1 in drug-resistant MCF7/ADR cells (Fig. 5E).

Cancer stem cells may generate other cancer stem cells and populations of cells forming the bulk of the tumor (2). In many patients with cancer, hidden cancer stem cells contribute to treatment resistance and relapse. Targeting cancer stem cells has some advantages by eliminating the root cause of tumors and minimizing side effects. Notch and Wnt pathways have been reported as main cell fate regulators in BCSCs and may be therapeutic targets (4). Mammary tumor-initiating cells directly regulate tumorigenesis through Notch1 signaling in vivo (23). In colorectal cancer stem cells, miR34a sets the “sweet spot” with Notch1 and this spot regulates the cell fate choice between stem cell maintenance and differentiation (24). As shown in Fig. 3, restoration of miR34a or inhibition of NOTCH1 reduced cancer stem cell properties in mammosphere cells. The miR34a–NOTCH1 axis may also play a crucial role in the asymmetric fate choice in breast cancers. Thus, anticancer stem cell therapies, including targeting miR34a–NOTCH1 signaling, will reduce the tumor recurrence rate and improve chances of long-term survival of many patients with cancer.

The miR34 family targets several genes such as CDK4/6, cyclin E2, MET, and Bcl-2 and is related to cell-cycle arrest, senescence, and apoptosis in many cancer types (11, 14, 25). In breast cancer, miR34c is related to self-renewal and epithelial–mesenchymal transition, and miR34a/c regulates breast cancer migration and invasion (12, 26). Interestingly, miR34a directly targets CD44 and suppresses tumor development and metastasis in prostate cancer stem cells (20). BCSCs also express the cell surface markers ESA and CD44 (27). We thought that miR34a might target crucial oncogenes and play a role as tumor suppressor miRNA in breast cancer.

The expression of members of the miR34 family is decreased in many cancers because of inactivating mutations of p53 or epigenetic inactivation (11). MiR34a is more correlated with transcriptional activation by p53 than miR34b/c because the miR34a encoding site exists near the p53 binding site (11, 17). The miR34a level is decreased in cell lines derived from triple-negative (ER⁻/PR⁻/HER2⁻) breast tumors (28). This could be explained by common mutations in p53 in triple-negative breast cancer. MiR34a is a direct transcriptional target of p53 that can be related to p53-mediated apoptosis (17). We confirmed the p53 mutation status in our cell lines; as a result, by sequencing of full-length p53, we found that codons 126 to 133 in exon 5 within the conformational domain were deleted in MCF7/ADR cells (data not shown). We hypothesized that the decreased level of miR34a was due to the mutation of p53. Furthermore, because the half-life and nuclear localization of mutated p53 were prolonged, it might be involved in drug resistance (16). As shown in Fig. 1C and D, the level of primary or mature miR34a was decreased in MCF7 cell treated with p53 siRNA or p53 mutant clone. We suggest that restoration of miR34a may be useful to prevent tumor progression, especially in patients with breast cancer with p53 mutation.

Li and colleagues (29) reported that NOTCH1 was a target of miR34a and miR34a modulated chemosensitivity in MCF7/ADR cells. In this study, we demonstrated that the miR34a–NOTCH1 axis mainly regulates the formation of breast cancer stem cells. As shown in the mammosphere assay and in vivo xenograft assay (Figs. 3 and 5), cancer stem cell properties were reduced by re-expression of miR34a or inhibition of NOTCH1. Reduction of the BCSC population through targeting the miR34a–NOTCH1 axis did not only modulate chemoresistance but also controlled tumor progression (Fig. 5E). We suggest that targeting miR34a is useful to prevent chemoresistance, including resistance to Dox.

Chemotherapy is the prevailing form of therapy for patients with cancer; however, there are still having many problems due to side effects and chemoresistance (30). So far, drug-resistant cell lines such as lung-A549/taxol and MCF7/ADR cell are more often used in vitro test than in vivo experiments. The present study revealed that the MCF7/ADR cell line was suitable to investigate chemoresistance in vitro or in vivo because MCF7/ADR cells had cancer stem cell characteristics and formed larger tumors than MCF7 cells (15). Interestingly, human cancer stem cell–containing cell populations survived in the secondary xenograft assay in immunodeficient mice and subsequently extended to solid tumors because of their self-replicating ability (22). In this study, we tested the functions of miR34a by establishing secondary xenografts after dissociation of cancer stem cells from a primary tumor mass in nude mice. The data showed that tumor volume was decreased by about 30% to 40% in mice treated with miR34a-expressing lentivirus (Fig. 5). In addition, we found that the MCF7/ADR cell line is suitable to test both cancer stemness and chemoresistance in vivo.

In conclusion, approaches to cancer therapy are shifting from conventional therapies to targeted therapies, including siRNA and miRNA. We have shown that the miR34a expression is reduced in tumorspheres of MCF7/ADR cells. NOTCH1 is a target of miR34a, which is involved in BCSC self-renewal. Moreover, restoration of miR34a and inhibition of NOTCH1 reduced cancer stem cell properties and chemoresistance. In addition, re-expression of miR34a delayed tumor progression in xenograft tumors. We propose that miR34a is one of the master tumor suppressor miRNAs and the p53–miR34a–NOTCH1 axis plays a crucial role in cancer stem cell survival. Thus, modulating miR34a may serve as a novel therapeutic strategy for drug-resistant breast cancers by targeting BCSCs.

No potential conflicts of interest were disclosed.

Conception and design: E.Y. Park, E. Song, J.H. Park

Development of methodology: Y.m. Woo, J.H. Park

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H.-W. Lee, H.-G. Kang, K.-H. Chun, H. Suzuki, J.H. Park

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E.Y. Park, E.S. Chang, K.-H. Chun, E. Song, J.H. Park

Writing, review, and/or revision of the manuscript: E.Y. Park, E.S. Chang, Y.m. Woo, E. Song, J.H. Park

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E.S. Chang, E.J. Lee, H.-W. Lee, H.K. Kong, J.Y. Ko, E. Song, J.H. Park

Study supervision: E. Song, J.H. Park

Other (performed mouse experiment): H.-G. Kang

This study was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP; 2013R1A2A1A01011908).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.