New insights into radioresistance in breast cancer identify a dual function of miR‐122 as a tumor suppressor and oncomiR

Radioresistance of tumor cells gives rise to local recurrence and disease progression in many patients. MicroRNAs (miRNAs) are master regulators of gene expression that control oncogenic pathways to modulate the radiotherapy response of cells. In the present study, differential expression profiling assays identified 16 deregulated miRNAs in acquired radioresistant breast cancer cells, of which miR‐122 was observed to be up‐regulated. Functional analysis revealed that miR‐122 has a role as a tumor suppressor in parental cells by decreasing survival and promoting radiosensitivity. However, in radioresistant cells, miR‐122 functions as an oncomiR by promoting survival. The transcriptomic landscape resulting from knockdown of miR‐122 in radioresistant cells showed modulation of the ZNF611, ZNF304, RIPK1, HRAS, DUSP8 and TNFRSF21 genes. Moreover, miR‐122 and the set of affected genes were prognostic factors in breast cancer patients treated with radiotherapy. Our data indicate that up‐regulation of miR‐122 promotes cell survival in acquired radioresistant breast cancer and also suggest that miR‐122 differentially controls the response to radiotherapy by a dual function as a tumor suppressor an and oncomiR dependent on cell phenotype.

Radioresistance of tumor cells gives rise to local recurrence and disease progression in many patients. MicroRNAs (miRNAs) are master regulators of gene expression that control oncogenic pathways to modulate the radiotherapy response of cells. In the present study, differential expression profiling assays identified 16 deregulated miRNAs in acquired radioresistant breast cancer cells, of which miR-122 was observed to be up-regulated. Functional analysis revealed that miR-122 has a role as a tumor suppressor in parental cells by decreasing survival and promoting radiosensitivity. However, in radioresistant cells, miR-122 functions as an oncomiR by promoting survival. The transcriptomic landscape resulting from knockdown of miR-122 in radioresistant cells showed modulation of the ZNF611, ZNF304, RIPK1, HRAS, DUSP8 and TNFRSF21 genes. Moreover, miR-122 and the set of affected genes were prognostic factors in breast cancer patients treated with radiotherapy. Our data indicate that up-regulation of miR-122 promotes cell survival in acquired radioresistant breast cancer and also suggest that miR-122 differentially controls the response to radiotherapy by a dual function as a tumor suppressor an and oncomiR dependent on cell phenotype.

Introduction
Radiotherapy, in addition to surgery and chemotherapy, remains the core of the current clinical management of breast cancer. Although radiotherapy is effective in most patients, some of them will develop recurrent disease because of radioresistant tumor cells (Jameel et al., 2004). Radiotherapy is an extrinsic factor that affects the behavior of the breast cancer cells themselves. When cells avoid the effect cytotoxic of radiation, cell growth is induced and spreads, resulting in a progression or recurrence of tumors in patients (Moran and Haffty, 2002;Torres-Roca et al., 2015). To overcome this problem, it is necessary to detemrine the mechanisms of resistance to radiotherapy. Several studies have demonstrated that tumor recurrence and progression as a consequence of radioresistance can be regulated by microRNAs (miRNAs) (Arechaga-Ocampo et al., 2017;Metheetrairut and Slack, 2013). miRNAs are master regulators of gene expression; moreover, they have a role in the regulation of carcinogenesis and the control of response to chemo-and radiotherapy in breast cancer (Zhang et al., 2014). miRNAs are short, 18-25 nucleotides in length, noncoding RNA molecules that regulate gene expression by suppressing mRNA translation and reducing mRNA stability, usually via imperfect complementary base pairing to the 3 0 -UTR (Bartel, 2004). miRNAs in cancer are classically categorized as either tumor suppressive or oncogenic. Generally, oncogenic miRNAs (oncomiRs) are overexpressed in tumors, whereas tumor-suppressive miRNAs are repressed. When these tumor-suppressor miRNAs or oncomiRs are stimulated or inhibited, respectively, cancer cell growth, proliferation, metastasis and survival may be significantly reduced via the control of pro-oncogenic factors (Svoronos et al., 2016). miR-122 is frequently down-regulated in breast cancer and has been related to tumor suppressor activity in breast cancer. Up-regulation of miR-122 suppressed cell growth and cell-cycle progression in breast cancer cell lines and suppressed tumorigenesis in vivo by targeting IGF1R and regulating the PI3K/Akt/mTOR/p70S6K pathway (Wang et al., 2012). However, the crucial roles and underlying the mechanisms of miR-122 with respect to radioresistance of breast cancer remain unclear. In the present study, we report the generation of an isogenic model of acquired radioresistant human breast cancer cells, as well as functional approaches aiming to identify the molecular changes in miRNAs that may explain this phenotype. We demonstrate that miR-122 has a dual function in breast cancer because it has tumor suppressor activity as a result of sensitizing parental cells to radiation, although it functions as an oncomiR in radioresistant breast cancer cells by promoting cell survival.

Establishment of radioresistant breast cancer cells
MCF-7RR and MDA-MB-231RR cell lines were established from their parental MCF-7 and MDA-MB-231 cells. 1 9 10 6 parental cells were irradiated with a linear accelerator (Clinac 600; Varian Inc., Palo Alto, CA, USA) available at the National Institute of Cancer in Mexico City. Cells received 15 sequential fractions of 2 GyÁweek À1 , allowing irradiated cell populations a period of recovery between exposures. Non-irradiated controls were handled identically to the irradiate cells without radiation exposure. All of the experiments were performed within 4-10 passages after the final irradiation.

Clonogenic survival assay
About 3 9 10 5 cells were irradiated and, after 24 h of radiation, 1000 cells per well were seeded in six-well tissue culture plates. The cells were cultured for 10-12 days. Colonies were fixed with 7 : 1 methanol/acetic acid, stained with 0.05% crystal violet and counted. The surviving fraction (SF) was calculated according to Franken et al. (2006). The SF of cells was plotted on a log scale.

miRNAs expression profile analysis
Expression of 667 miRNAs was analyzed by a quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR) using the Megaplex TaqMan Low-Density Arrays (TLDAs), version 2.0, system (Applied Biosystems, Foster City, CA, USA). Briefly, 100 ng of total RNA was retro-transcribed using stemloop primers and a pre-amplification step was added so that the minimum amounts of miRNAs were detected. qPCR assays were performed using Gen-eAmp System 9700 (Applied Biosystems).

qRT-PCR
The expression of individual miRNAs was evaluated via qRT-PCR using the Stem-loop RT miRNA assay (Applied Biosystems). About 100 ng of total RNA was retro-transcribed using the looped RT primer (Applied Biosystems) in accordance with the manufacturer's protocol. Detection for miR-122, miR-10a, miR-222, miR-222*, miR135b, miR-135b*, miR-196b and miR-934 was performed using TaqMan Universal PCR Master Mix (Applied Biosystems). qPCR was carried out in 7500 Real-Time PCR System (Applied Biosystems). The expression of miRNA was determined using the comparative C t (2 ÀDDC t ) method. RNU44 was used as a control for normalization of data.

Transfections
MCF-7 and MDA-MB-231 cells were transfected with mimic-miR122 (Ambion, Austin, TX, USA) 10 nM, whereas MCF-7RR and MDA-MB-231RR cells were transfected with antagomiR-122 (Ambion) 30 nM. Mimic-miR122 and antagomiR-122 were diluted in Opti-Mem (Invitrogen, Carlsbad, CA, USA), scramble sequence was used as a control and Lipofectamine 2000 (Invitrogen) was used as transfection agent. The expression of miR-122 was evaluated 48 h post-transfection by qRT-PCR. After transfection, cells were irradiated with 4 Gy of irradiation. The subsequent clonogenic assay was performed as described previously in the section 2.3.

Microarray processing and data analysis
Total RNA was obtained from MCF-7RR and MCF-7RR cells transfected with antagomiR-122. Equimolar concentrations of total RNA from three independent experiments were mixed and the transcriptional profiles were analyzed using the Affymetrix GeneChip Human Gene 1.0 ST array (Affmetrix, Santa Clara, CA, USA) in accordance with the manufacturer's instructions. Arrays were scanned using a Genechip Scanner 3000 7G (Affmetrix). The data were analyzed with the Robust Multichip Analysis algorithm using the default analysis settings (Affmetrix) and global scaling as the normalization method. To define the differential expression profile, TRANSCRIPTOME ANALYSIS CONSOLE software (Affmetrix) was used. Genes with fold change > 1.3 or < À1.3 and with P < 0.05 by ANOVA were considered significantly altered between the conditions (MCF-7RR and MCF-7RR cells transfected with antagomiR-122). Microarray raw data tables have been deposited at the National Center for Biotechnology Information Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc= GSE120171).

Bioinformatic analysis
Verified miRNAs targets were obtained by miR-TarBase (http://mirtarbase.mbc.nctu.edu.tw) and predicted miRNAs targets from the miRWalk database (http://zmf.umm.uni-heidelberg.de/apps/zmf/mirwalk). Only miRNA-target interactions identified by at least three algorithms were considered. DAVID, version 6.7 (https://david-d.ncifcrf.gov) and PANTHER PATHWAY (http://www.pantherdb.org/pathway) were used to identify components of signaling pathways and Gene Ontology (GO) for biological processes or molecular functions. The analysis of biological network enrichment of the modulated genes obtained by microarrays assays was performed by CYTOSCAPE (https://cytoscape. org) using the Key Pathway Miner App (Alcaraz et al., 2014). In this analysis, a value of K = 6 was used.

The Cancer Genome Atlas (TCGA) data analysis
The RNA sequencing data from samples of 491 breast cancer patients were downloaded from the TCGA database (https://portal.gdc.cancer.gov). First-line treatment and/or additional radiotherapy, tumor status and follow-up days were considered. Total population was stratified according to low or high expression of mir-122, ZNF611, ZNF304, RIPK1, TNFRSF21, DUSP8 and HRAS. Kaplan-Meier analysis was used for relapse-free survival (RFS) curves and log-rank tests were employed to analyze the differences between curves. The results were confirmed by Cox proportional-Hazard regression analyses.

Statistical analysis
All results were derived from three independent experiments, which were plotted as the mean AE SD. The comparison between the groups was performed using ANOVA for all analyzes. P ≤ 0.05 was considered statistically significant. All statistical analyses were performed using SPSS, version 17.0 (SPSS Inc., Chicago, IL, USA).

Establishment of radioresistant MCF-7 and MDA-MB-231 breast cancer cells
To determine a mean lethal dose of irradiation, MCF-7 and MDA-MB-231 cells were evaluated by MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide] and clonogenic assays following increased radiation dose (2-8 Gy). The results showed that cell proliferation was significantly reduced at different time points (0, 24, 48 and 72 h) and at increased radiation doses (0, 2, 4, 8 Gy) compared to control cells. Proliferation diminished to 53 AE 0.07% in MCF-7 (Fig. 1A) and 58 AE 0.05% in MDA-MB-231 ( Fig. 1B) cells at 4 Gy at 48 h. The results from clonogenic assays showed a significant reduction in survival ( Fig. 1C) of MCF-7 (0.48 AE 0.005) and MDA-MB-231 (0.44 AE 0.005) cells at 4 Gy compared to non-irradiated cells. These results indicated that 4 Gy of radiation represents a mean lethal dose for MCF-7 cells and MDA-MB-231 cells. Therefore, a radiation dose of 4 Gy was used for all subsequent experiments. To establish the isogenic model of radioresistance, MCF-7 and MDA-MB-231 cells were exposed to 2 Gy-fractionated irradiations to a cumulative dose of 30 Gy (Fig. 1D). At the end of this process, radioresistance of the resulting cell population, designated as MCF-7RR and MDA-MB-231RR, was confirmed by clonogenic assays after single doses of 4 Gy of radiation. Acquired radioresistance of MCF-7RR and MDA-MB-231RR cells was further indicated by an increase of survival after 4 Gy of radiation compared to parental cells. The SF of the MCF-7RR (Fig. 1E) and MDA-MB-231RR (Fig. 1F) cells was 0.70 and 0.75, respectively; whereas, for parental MCF-7 and MDA-MB-231 cells, the SF was 0.47 and 0.45, respectively. These data confirmed that the population resulting from the cells exposed to long-term therapeutic fractionated irradiation developed a radioresistant phenotype.

miR-122 increases radiosensitivity in parental breast cancer cells and in breast cancer patients treated with radiotherapy
miR-122 has been described as a tumor suppressor and its downregulation is a common event in breast cancer (Wang et al., 2012). Conversely, in the present study, overexpression of miR-122 was observed in both radioresistant breast cancer cells. To investigate whether miR-122 might increase the radioresistance of breast cancer cells, we performed assays of gain-offunction in parental MCF-7 and MDA-MB-231 cells. We were able to overexpress miR-122 in parental MCF-7 (Fig. 3A) and MDA-MB-231 cells (Fig. 3B) using a mimic-miR122. Next, we evaluated the survival potential of these cells in response to radiotherapy by clonogenic assays. The results showed that, in nonirradiated cells, the overexpression of miR-122 alone but not scrambled transfected or control untransfected significantly decreased the survival MCF-7 (SF = 0.63) (  102 patients who received radiotherapy as a first-line treatment obtained from the TCGA database were randomly categorized into two groups according to positive or null expression of miR-122. The results revealed that patients with a positive expression of miR-122 and who had received radiotherapy had a significantly better RFS than those with negative expression of miR-122 ( Fig. 3E), suggesting that patients with null expression of miR-122 were significantly associated with poor prognoses after radiotherapy. These results were in agreement with our findings obtained in vitro, in which miR-122 sensitized breast cancer cells to irradiation. Furthermore, the results obtained in vivo suggested that the expression of miR-122 might be predictor biomarker for RFS in breast cancer patients treated with radiotherapy.

miR-122 knockdown overcomes acquired radioresistance in breast cancer cells
We have demonstrated that miR-122 is up-regulated in both radioresistant breast tumor cells. To obtain additional insight into the biological function of miR-122 in acquired radioresistance, we performed  (Fig. 4D). Hence, the knockdown of miR-122 in radiotherapy-induced resistant cells is able to revert the radioresistance of the cells via delayed cell survival. Furthermore, these results suggest that miR-122 promotes survival pathways with respect to maintaining a radioresistant phenotype in breast cancer cells. Accordingly, we evaluated the expression of miR-122 in parental cells treated with radiotherapy. The results showed that radiation promotes overexpression of miR-122 in parental breast cancer cells (Fig. 4E). We hypothesized that overexpression of miR-122 was maintained during the adaptive biological reprogramming in response to continuous application of radiation (i.e. during the transition from a cancer cell to a radioresistant cancer cell) (Fig. 4F). Moreover, it is likely that miR-122 could gain an oncogenic role in radioresistant cells,  therefore having a dual function in breast cancer cells, acting either as a tumor suppressor or an oncogene depending on the cellular context (Fig. 4F). The adaptive transcriptional reprogramming includes the possibility that target genes of the miRNAs could be also changed, producing oncogenic or tumor-suppressive effect. In this  way, miR-122 should regulate different target genes to act as a positive regulator of the survival pathways, thus favoring an oncogenic function in radioresistant cells. To explore this hypothesis, we evaluated the expression of IGF1R, which has previously been reported as a target gene of miR-122 in breast cancer (Wang et al., 2012).
Remarkably, the results obtained via qRT-PCR showed that IGF1R is down-regulated in parental MCF-7 cells with gain-of-function of miR-122 and up-regulated in MCF-7RR cells with loss-of-function of miR-122 (Fig. 4G), which is consistent with targeted activity of miR-122. These results suggest that radiosensitivity observed in MCF-7RR cells with loss-of-function of miR-122 could be independent of the IGF1R function. Taken together, these results indicate that miR-122 has an oncogenic role in the acquired radioresistance of breast cancer cells.

Genes associated with loss-of-function of miR-122 are involved in the outcome of breast cancer patients treated with radiotherapy
To determine the prognostic value of genes associated with loss-of-function of miR-122, we evaluated RFS in 491 breast cancer patients treated with radiotherapy, as obtained from the TCGA database. Patients were categorized according to the median expression of each gene; therefore, we obtained different groups of patients according to the low and high expression of an individual gene. Characteristic populations are shown in Tables S2-S7. In the analysis for RFS, differential expression of ZNF611, ZNF304, RIPK1, DUSP8, TNFRSF21 and HRAS genes was associated with outcome for breast cancer patients who received radiotherapy. Kaplan-Meier curves showed that increased expression of ZNF611 (P = 0.0269; Fig. 5E) and ZNF304 (P = 0.0081; Fig. 5F), as well as lower expression of RIPK1 (P = 0.0047; Fig. 5G), TNFRSF21 (P = 0.0058; Fig. 5H) and DUSP8 (P = 0.048; Fig. 5I) and HRAS (P = 0.0138; Fig. 5J), was associated with longer RFS. These results were in accordance with the experimental evidence obtained in MCF-7RR cells in which an increase of ZNF611 and ZNF304 expression, in addition to a decrease of RIPK1, TNFRSF21, DUSP8 and HRAS expression, induced by knockdown of miR-122 was correlated with radiosensitivity in vitro.  (Fig. 6A). For experimental validation of the transcriptome results, and also to test whether modulation of ZNF611, ZNF304, RIPK1, DUSP8, HRAS and TNFRS21 following knockdown of miR-122 might be a common event in radioresistant breast cancer cells, we performed western blot assays in transfected MCF-7RR and MDA-MB-231RR cells with antagomiR-122. As expected, we verified the results obtained by microarrays assays in MCF-7RR cells. Levels of proteins ZNF304 and ZNF611 were up-regulated, whereas RIPK1 and DUSP8 were repressed, when we inhibited miR-122 (Fig. 6B). Levels of HRAS and TNFRS21 levels were observed with non-significant changes. The results showed a direct correlation among the expression of miR-122 and ZNF304, ZNF611, RIPK1 and DUSP8 levels. As a control, we compared the abundance of these proteins in parental MCF-7 cells transfected with mimic-miR122 (Fig. 6C). The results showed that the expression of ZNF304, RIPK1, DUSP8 and TNFRS21 was not changed when we forced overexpression of miR-122, whereas the expression of HRAS was higher compared to non-transfected and scrambled transfected cells (Fig. 6C). Notably, ZNF611 was downregulated when miR-122 was overexpressed (Fig. 6C). These findings suggested that miR-122 could target ZNF611 but was unable to modulate ZNF304, RIPK1, DUSP8 and TNFRS21 protein levels in parental MCF-7 cells. Although not all of the western blot results were similar in MDA-MB-231RR compared to MCF-7RR, we observed that, in MDA-MB-231RR cells, knockdown of miR-122 was correlated with up-regulation of ZNF611, DUSP8 and HRAS (Fig. 6D). Moreover, in parental MDA-MB-231 cells, forced overexpression of miR-122 was also correlated with inhibition of ZNF611, DUSP8 and HRAS levels Table 3. Genes differentially regulated in knockdown miR-122-MCF-7RR cells with a 3 0 -UTR-canonical binding site to miR-122.

Gene Function
Up-regulated IGLON5 Neuronal cell-adhesion protein NUP62CL Nucleocytoplasmic transporter activity ACAA1 Acetyl-CoA C-acyltransferase activity KLHL5 Ubiquitin-protein transferase activity FBXO48 Decreased interleukin-8 secretion ZNF304 Activation of KRAS and silencing of several tumor suppressor genes VIPR2 Pituitary adenylate cyclase activation CCDC127 Cell surface transport ZNF611 Transcriptional regulation Down-regulated LCE2C Precursors of the cornified envelope of the stratum PRH2 Protective and reparative environment for dental enamel RAB40AL Mediates the ubiquitination and subsequent proteasomal degradation of target proteins MICAL2 Actin binding and flavin adenine dinucleotide binding SEMA5A Semaphorin gene family that encodes membrane proteins containing a semaphorin domain and several thrombospondin type-1 repeats. May promote angiogenesis by increasing endothelial cell proliferation and migration and inhibiting apoptosis SLC25A18 Involved in the transport of glutamate across the inner mitochondrial membrane PODNL1 Small leucine-rich repeat protein family DUSP8 Phosphatase activity with synthetic phosphatase substrates and negatively regulates mitogenactivated protein kinase activity DDR2 Tyrosine kinase that functions as cell surface receptor for fibrillar collagen and regulates cell differentiation, remodeling of the extracellular matrix, cell migration and cell proliferation C2CD4A Involved in inflammatory process. May regulate cell architecture and adhesion KIAA0825 Cell surface transport ZNF793 Transcriptional regulation DLX4 Play a role in determining the production of hemoglobin S. May act as a repressor LIN28A Protein that acts as a posttranscriptional regulator of genes involved in developmental timing and self-renewal in embryonic stem cells. Disrupting the maturation of certain miRNAs SP7 Transcriptional activator essential for osteoblast differentiation TFCP2L1 Transcriptional suppressor, cellular self-renewal CITED4 Acts as transcriptional coactivator for TFAP2/AP-2. Enhances estrogen-dependent transactivation mediated by estrogen receptors TRPM1 Cation channel essential for the depolarizing photoresponse of retinal ON bipolar cells. It is Cytokine receptor that belongs to the interleukin-1 receptor family VEPH1 Downregulation of Wnt pathway after Wnt3A stimulation ARHGAP23 GTPase activator for the Rho-type GTPases by converting them to an inactive GDP-bound state C4B Complement component 4B, including the ZB transcript in the same orientation, complexing with C2 to form the C3/C5 convertase, classical pathway CNIH3 Regulates the trafficking and gating properties of AMPA-selective glutamate receptors DEAF1 Cell proliferation, arresting cells in the G0 or G1 phase. Required for neural tube closure and skeletal patterning. Regulates epithelial cell proliferation and side-branching in the mammary gland EPPIN Serine protease inhibitor that plays an essential role in male reproduction and fertility HEATR3 Role in ribosomal protein transport and in the assembly of the 5S ribonucleoprotein particle (5S RNP). The encoded protein also may be involved in NOD2-mediated NF-kappa B signaling NUB1 Protein that functions as a negative regulator of NEDD8, a ubiquitin-like protein that conjugates with cullin family members in order to regulate vital biological events RIPK1 Serine-threonine kinase, which transduces inflammatory and cell-death signals (programmed necrosis) following death receptors ligation, activation of pathogen recognition receptors and DNA damage. Activates the MAP3K5-JNK apoptotic cascade KRT77 Responsible for the structural integrity of epithelial cells (Fig. 6E). These results suggested that ZNF611 and DUSP8 might be targeted by miR-122 in the MDA-MB-231RR model. It should be noted that miR-122 could modulate dissimilar pathways in radioresistant breast cancer cells compared to parental breast cancer cells, which might partially explain the dual function as a tumor suppressor or oncomiR depending on the cellular context. On the other hand, the variability of the results among the radioresistant cells could be a result of a difference in cellular context with respect to the origin of the cell lines because MDA-MB-231 is a model of the triple-negative breast cancer (TNBC) subtype, whereas MCF-7 is a model of a luminal tumor.
In this sense, we evaluated the prognostic value of the ZNF611, ZNF304, RIPK1, TNFRSF21, DUSP8 and HRAS genes according to tumor subtypes. The RFS of patients with luminal breast cancer and TNBC treated with radiotherapy was evaluated by Kaplan-Meier analysis. The results showed that higher levels of ZNF611 (P = 0.0338), as well as lower levels of RIPK1 (P = 0.0024) and DUSP8 (P = 0.0165), were associated with longer RFS in luminal subtypes (Fig. 6F). However, although these genes were not significantly associated with RFS in TNBC subtypes, higher levels of ZNF611 and ZNF304 and lower levels of HRAS showed a trend for an association with a longer RFS (Fig. 6G).

Discussion
To gain insight into the molecular adaptations underpinning the radioresistance of breast cancer cells, we report the development of isogenic radioresistant breast cancer cell lines, miRNome landscape analysis and functional analysis of miR-122. An isogenic model of radioresistance was developed by cumulative exposure of MCF-7 and MDA-MB-231 cells to 30 Gy-fractionated radiation, resulting in the generation of a subline with significantly increased survival potential compared to sham control cells, as reported previously in a model of radioresistant lung cancer cells (Arechaga-Ocampo et al., 2017). The development of isogenic chemo-and radioresistant cell lines has been used successfully to investigate the molecular changes associated with acquired resistance to therapy and tumor aggressiveness in cancer (McDermott et al., 2014). Exposure of tumors to fractionated radiation schedules can select a cancer cell subpopulation with an increased capacity to overcome the anti-proliferative effects of radiotherapy (Zaider and Hanin, 2011) by modulating the abundance and functions of molecules, including miRNAs (Arechaga-Ocampo et al., 2017;Metheetrairut and Slack, 2013;Zhang et al., 2014). In the present study, we identified a set of miR-NAs related to acquired radioresistance. Wang et al. (2014b) have shown that the expression of a group of miRNAs establishes a useful molecular signature with respect to differentiating radioresistant from nonradioresistant tumors, even though they originate from the same histological type. In this sense, we identified the miRNome of the isogenic MCF-7RR cell line compared to the parental MCF-7 cell line and found 16 miRNAs to be differentially expressed (13 up-and three down-regulated). Among them, miR-184 , miR-424* (Zhang et al., 2017b), miR-218 , miR-222 (Wei et al., 2017) and miR-10a (Rong et al., 2016) have been reported in relation to chemotherapy resistance in several types of tumors. Others, such as as miR-135b* (Wang et al., 2014a,b), miR-223* (de Melo Maia et al., 2016, miR-135b (Han et al., 2017) andmiR-196b (Ren et al., 2017), have been reported as oncomiRs. The aberrant expression of miRNAs in the radioresistance of human tumors is a result of their function as negative regulators of gene expression. miRNAs can control the expression of genes that are components of cell survival pathways, apoptosis, the immune response and cell differentiation, amongst others (Bartel, 2004). The results of our bioinformatic analysis of the GO and biological pathways revealed that the set of miRNAs is implicated in proliferation and survival pathways, the immune response and transcriptional control, suggesting that signaling pathways involved in acquired radioresistance can be directed by the coordinated action of these molecules. It is remarkable that deregulation of the miRNAs identified in MCF-7RR cells is conserved in MDA-MB-231RR cells. Specially, miR-196b, miR-222 and miR-122 show similar expression in both radioresistant cell lines, which suggests that they might have a significant role in the phenotypic evolution of cancer cells to acquired radioresistance. Notably, miR-122 is known to act as a tumor suppressor in breast cancer by targeting IGF1R (Wang et al., 2012) and ADAM10 genes (Erg€ un et al., 2015). In addition, miR-122 has the same role in liver and glioma tumors, in which it suppresses mechanisms to promote tumor progression and survival (Wang et al., 2014a,b). In the present study, we demonstrate that miR-122 significantly reduced the survival of the parental cells, although this effect was enhanced when the cells were irradiated. This result is consistent with previous reports of the tumor suppressor function of miR-122 in breast cancer (Erg€ un et al., 2015;Wang et al., 2012) and it also highlights its potential as a radiosensitizer. Similarly, a previous study showed that miR-122 induces radiosensitivity in lung cancer cells exposed to different doses of radiation. The ectopic overexpression of miR-122 in lung cancer cells decreased anchorage-dependent invasion and inhibited cell growth by knockdown of target genes related to tumor survival and the cellular stress response . The role of miR-122 as a radiosensitizer in lung cancer is consistent with our findings in breast cancer. Recently Zhang et al. (2017a,b) reported the function of miR-122-3p in response to radiation in TNBC cells. It was demonstrated that miR-122-3p promotes sensitivity to radiation by modifying cellular apoptosis, migration and invasion as a result of modulation of the PTEN/PI3K/AKT pathway. It should be noted that miR-122-3p (miR-122*) is the passenger strand of leading strand miR-122-5p (miR-122); therefore, miR-122-3p could target different genes, giving it an independent function compared to miR-122-5p.
The results of the studies by Zhang et al. (2017a,b), together with our results, suggest that miR-122* and miR-122 might function cooperatively to promote the sensitivity of breast cancer cells exposed to radiation. However, in our isogenic model of acquired radioresistant breast cancer cells, we did not observed the aberrant expression of miR-122-3p and so the role of miR-122-3p in acquired radioresistance in breast cancer still remains unknown. Besides its function in vitro, we show that the expression of miR-122 in breast tumors was significantly associated with a favorable response to radiotherapy compared to those patients who did not express miR-122. In such patients, miR-122 has a tumor-suppressive role for treatment-na€ ıve tumors (i.e. in patients who have not yet received radiation therapy and therefore radioresistance has not yet developed). With these results, we demonstrate that miR-122 is related to the sensitivity to radiotherapy in parental breast cancer cells both in vitro and in vivo in a set of breast tumors, which, to our knowledge, has not been reported previously. Unexpectedly, in MCF-7RR and MDA-MB-231RR cells, we observed overexpression of miR-122. Its downregulation by antagomiR-122 was able to revert the resistance to radiotherapy by counteracting cell survival. By contrast to functioning as a tumor suppressor, miR-122 appears to have an oncogenic role in breast cancer cells that have acquired radioresistance. It is known that miRNAs can act as a tumor suppressor or as an oncogene depending on the scenario. Svoronos et al. (2016) provide an excellent review in which they discuss the dual role of miRNAs in cancer cells themselves and as extrinsic factors. They also report that dual function could be dependent on different cell phenotypes, tumor microenvironment, immune evasion and by selective pressure induced by therapy treatments, including radiotherapy. We showed that miR-122 is up-regulated in response to radiation treatment, suggesting that continuous fractionated irradiation maintains the overexpression of miR-122 until a radioresistant phenotype is attained. Based on our results, we propose that miR-122 could acquire an oncogenic function under the pressure exerted by radiation in breast cancer, which allows evolution of the cell and adaption to radiotherapy. To explore the genes and biological pathways modulated by miR-122 in radioresistant cells, we analyzed the transcriptome of MCF-7RR cells with loss-of-function of miR-122. The results showed 158 genes to be Fig. 7. Schematic representation of miR-122-mediated radioresistance in breast cancer cells. miR-122 has an oncogenic role in the acquired radioresistance of luminal and TNBC cells by differentially controlling the expression of ZNF611 and ZNF304 transcription factors, as well as modulating the expression of genes involved in RAS-MAPK and TNF pathways to promote survival. differentially modulated, of which 27 were increased and 131 were decreased. Analyses of the biological network by CYTOSCAPE revealed protein outliers that have been reported in radiotherapy resistance in cancer and survival, as well as proliferation and epigenetic regulation, such as EGFR (Lee et al., 2011), MOV10 (El Messaoudi-Aubert et al., 2010, ELAVL1 (Mehta et al., 2016), SIRT7 Tang et al., 2017), APP (Lim et al., 2014) and EWSR1 (Suzuki et al., 2012). Many studies have reported that radiotherapy can induce transcriptional reprogramming to enable acquired resistance and to avoid the toxicity triggered by radiotherapy (Doan et al., 2018;Ma et al., 2013). We determined that miR-122 modulates genes related to molecular processes such as transcriptional regulation and the signaling of receptors coupled to G proteins, as well as the MAPK and TNF pathways. This diversity of cellular processes is consistent with numerous regulatory mechanisms associated with the response to radiation and radioresistance in breast cancer (Kaidar-Person et al., 2013). In the transcriptomic landscape induced by knockdown miR-122, we identified genes involved in these cellular processes such as ZNF611, ZNF304, RIPK1, DUSP8, HRAS and TNFRSF21. Moreover, these genes were prognostic factors in breast cancer patients treated with radiotherapy. The clinical findings were in accordance with the results of survival assays in vitro. Among the set of up-regulated genes, we found that ZNF611 and ZNF304 contain canonical miR-122-binding sites in their 3 0 -UTR region; in addition, we verified that the protein levels of these transcription factors also increase when miR-122 is inhibited in MCF-7RR cells. ZNF304 and ZNF611 are transcription factors that belong to the C2H2 zinc finger family possessing Kr€ uppel associated box and are related to transcriptional silencing by the recruitment of epigenetic complexes (Aslan et al., 2015;Pengue and Lania, 1996). ZNF304 has also been reported as a regulator of the RAS pathway by recruitment of an epigenetic silencing complex in tumor suppressor genes in colorectal cancer (Serra et al., 2014) and as a promoter of cell survival in ovarian cancer (Aslan et al., 2015). ZNF611 has not yet been reported in cancer. On the other hand, RIPK1, DUSP8, HRAS and TNFRS21 were downregulated when we knocked-down miR-122. RIPK1 (Miyamoto, 2011) and TNFRS21 (Benschop et al., 2009) are components of the TNF pathway, whereas DUSP8 (Keyse, 2008) and HRAS (Knobbe et al., 2004) act on the RAS-MAPK pathway. Alteration in the activity of transcription factors could be critical for the development and maintenance of radioresistance in breast cancer by control genes of the survival pathways, such as the RAS-MAPK and TNF pathways (Ishihara et al., 2015;Zhao et al., 2018). Many of the genes that we observed to be down-regulated in the transcriptome analysis might be part of an epigenetic network of transcriptional silencing driven by axis miR-122-ZNF611 or miR-122-ZNF304. This is a research topic that is currently under investigation; however, we have obtained preliminary data strongly suggesting that RIPK1 and DUSP8 possess elements of response to ZNF611 and ZNF304 in their promoter regions (data not shown). Therefore, we propose that the dual function of miR-122 in the isogenic model of radioresistant breast cancer cells could be result of transcriptional reprogramming controlled by the modulation of ZNF611 and ZNF304 by miR-122.

Conclusions
In conclusion, the data obtained in the present study contribute to our understanding of the mechanisms of molecular adaptation to radiotherapy in breast cancer cells. The evidence indicates the aberrant expression of a set of miRNAs linked to carcinogenesis and the molecular control of pathways related to the response to therapy in cancer. Particularly, the overexpression of miR-122 maintains the radioresistant phenotype in breast cancer cells by promoting cell survival, from the regulation of several genes to the downstream effects of these genes, which confers it with an oncogenic function (Fig. 7). through targeting YAP1 in non-small cell lung cancer. Mol Carcinog 56, 821-832. Zhao W, Sun M, Li S, Chen Z and Geng D (2018) Transcription factor ATF3 mediates the radioresistance of breast cancer. J Cell Mol Med 22, 4664-4675.

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Fig. S1. Validation of the expression of a set of miR-NAs deregulated in MCF-7RR cells. Expression of a set of miRNAs was evaluated by qRT-PCR in MCF-7RR cells. The expression data were normalized using the parental MCF-7 cells. All values were normalized using RNU44 as an internal control. Data for the qRT-PCR assays were compared with the results from global expression profiles by the TLDAs system. Data are presented as the mean AE SD of three independent experiments. *P < 0.01 by Student's t-test. Table S1. Differentially expressed genes in the MCF-7RR cells transfected with antagomiR-122. Table S2. Characteristics of the population of breast cancer patients based on expression of the ZNF611 gene. Table S3. Characteristics of the population of breast cancer patients based on expression of the ZNF304 gene. Table S4. Characteristics of the population of breast cancer patients based on expression of the RIPK1 gene. Table S5. Characteristics of the population of breast cancer patients based on expression of the TNFRSF21 gene. Table S6. Characteristics of the population of breast cancer patients based on expression of the DUSP8 gene. Table S7. Characteristics of the population of breast cancer patients based on expression of the HRAS gene.