Intracellular microRNA expression patterns influence cell death fates for both necrosis and apoptosis

We reveal that intracellular miRNA expression patterns are dramatically altered in necrosis and apoptosis using a floxuridine (FUdR)‐induced mouse mammary tumor cell death model. We found that knockdown of Dicer in FUdR‐treated necrosis‐fated cells caused a partial shift from necrosis to apoptosis. These findings suggest that the expression of miRNA(s) regulates cell death fate toward necrosis or apoptosis.

MicroRNAs (miRNAs) are small noncoding RNA molecules that interact with target mRNAs at specific sites to induce cleavage of the mRNA or inhibit translation. Such miRNAs play a vital role in gene expression and in several other biological processes, including cell death. We have studied the mechanisms regulating cell death (necrosis in original F28-7 cells and apoptosis in their variant F28-7-A cells) in the mouse mammary tumor cell line FM3A using the anticancer agent floxuridine (FUdR). We previously reported that inhibition of heat-shock protein 90 by the specific inhibitor geldanamycin (GA) in F28-7 cells causes a shift from necrosis to apoptosis. In this study, we investigated the intracellular miRNA expression profiles of FUdR-treated F28-7 cells (necrotic condition), GA plus FUdR-treated F28-7 cells (apoptotic condition), and FUdR-treated F28-7-A cells (apoptotic condition) through miRNA microarray analysis. In addition, we knocked down Dicer, a key molecule for the expression of mature miR-NAs, in F28-7 cells to examine whether it modulates FUdR-induced cell death. Our analysis revealed that the miRNA expression patterns differ significantly between these cell death conditions. Furthermore, we identified miRNA candidates that regulate cell death. Knockdown of Dicer in FUdR-treated necrosis-fated cells caused a partial shift from necrosis to apoptosis. These findings suggest that modulation of miRNA expression patterns influences the decision of cell death fate toward necrosis or apoptosis. Our findings may serve as a basis for further study of the functions of miRNAs in cell death mechanisms.
Two major categories of noncoding RNA (ncRNA), that is, miRNA and long noncoding RNA (lncRNA), play important roles in gene expression, cell death, and several physiological processes [17]. The miRNAs are endogenous small ncRNAs (length: 21-25 nucleotides) that function as gene silencers by binding to the specific sites of target mRNAs, inhibiting the initiation of protein synthesis and/or promoting mRNA cleavage [18,19]. Importantly, miRNAs regulate many biological processes, including cell development, differentiation, and cell death [18][19][20][21]. They are excised in a stepwise process from primary miRNA transcripts [18,19]. The primary miRNA is cleaved by nuclear RNase III Drosha to release hairpin-shaped precursor miRNAs (pre-miRNAs) [18,19]. These pre-miRNAs are subsequently exported to the cytoplasm and further processed by Dicer to mature miRNAs [18,19].
In this study, we showed that the intracellular miRNA expression patterns were dramatically altered in cell death (necrosis and apoptosis) using the cell death model. We found that knockdown of Dicer in FUdR-treated necrosis-fated cells caused a partial shift from necrosis to apoptosis. These findings suggest that the expression of miRNA(s) regulates cell death fate toward necrosis or apoptosis.

Reagents
Floxuridine and GA were obtained from Sigma (St. Louis, MO, USA). FUdR was stored as 2 mM stocks in ultrapure water at À20°C. GA was stored as 2 mM stock in dimethyl sulfoxide at À20°C with protection from light. 4',6-Diamidino-2-phenylindole dihydrochloride was obtained from Invitrogen (Carlsbad, CA, USA).

RNA extraction
RNA extraction was performed as previously described [10,11,14]. For microarray analysis, the total RNA fraction was isolated from the individual cell lines using QIAshredder spin columns and an RNeasy Mini Kit, according to the instructions provided by the manufacturer (QIAGEN, Hilden, Germany).

Microarray analysis of miRNA expression
Microarray analysis of miRNA expression was performed as previously described [14]. Biotin-labeled RNA was prepared using a FlashTag TM Biotin RNA Labeling Kit according to the instructions provided by the manufacturer (Genisphere, Hatfield, PA, USA). The labeled RNA product was mixed in hybridization cocktail. The hybridization cocktails were added to GeneChipÒ miRNA 2.0 Array (Affymetrix, Santa Clara, CA, USA) in the GeneChip Hybridization Oven 640 under constant rotation (60 r.p.m.) at 48°C for 16 h. After the hybridization, GeneChips were washed and stained with GeneChip Fluidics Station 450 (Affymetrix) and scanned with a GeneChip Scanner 3000 7G (Affymetrix). Scanned GeneChip images were analyzed using the Affymetrix GeneChip Command Console software and miRNA QCTool (Affymetrix). Microarray data were analyzed using the GENESPRING software (Agilent Technologies, Santa Clara, CA, USA).

Observation of cell morphology
The observation of cell morphology was performed as previously described [13,14] using an Olympus (Tokyo, Japan) BX61 fluorescence microscope.

Statistical analysis
Determining the statistical significance of differences among groups was conducted using the Student's t-test and oneway analysis of variance. P < 0.05 denoted statistically significant differences.

Results and Discussion
The miRNA expression profiles of necrosis in F28-7 cells and apoptosis in F28-7-A cells after treatment with FUdR F28-7 and F28-7-A cells were treated with 1 lM FUdR for 0 h (no drug, no incubation) and 8 h to profile changes in the pattern of miRNA expression in FUdR-induced necrosis and apoptosis. The treatment induced necrosis in F28-7 cells and apoptosis in F28-7-A cells (Fig. 1) [9,11]. We investigated the miRNA expression profiles in necrotic and apoptotic dying cells through miRNA microarray analysis. For this purpose, three independent analyses were performed with the Affymetrix GeneChip miRNA 2.0 Array, which contains a 722-probe set for mouse mature miRNAs and 690-probe set for mouse pre-miRNA, allowing analyses of the expression levels of mouse mature miRNAs. Previously, we reported the miRNA expression profiles in these sister cells at the untreated stage (no drug) by microarray analyses [14]. We analyzed the differentially expressed miRNAs in the dying stage vs the untreated stage in both F28-7 and F28-7-A cells. Using a 1.5-fold cutoff (P < 0.05), the analysis identified 14 differentially expressed miRNAs. In necrosis (FUdRtreated F28-7 cells), 3 and 11 mature miRNAs were expressed at higher and lower levels, respectively, than in the untreated stage (F28-7 cells). Table 1 lists the names and sequence information of these 14 mature miRNAs. The miRNA expression analysis identified two differentially expressed miRNAs. In apoptosis (FUdR-treated F28-7-A cells), two mature miRNAs (i.e., miR-1199-5p and miR-691) were expressed at higher and lower levels, respectively, than in the untreated stage (F28-7-A cells). Table 2 lists the names and sequence information of these two mature miRNAs.

The miRNA expression profiles of apoptosis in F28-7 cells after cotreatment with GA and FUdR
Previously, we also demonstrated that inhibition of Hsp90 by the Hsp90 inhibitor GA in F28-7 cells causes a shift from original necrosis to apoptosis (Fig. 1) [11]. We investigated the miRNA expression profiles of original necrosis in F28-7 cells treated with FUdR alone and apoptosis in F28-7 cells treated with the combination of GA and FUdR by miRNA microarray analysis. Using a 1.5-fold cutoff, the analysis identified 25 differentially expressed miRNAs. In apoptosis (GA-and FUdR-treated F28-7 cells), 2 and 23 mature miRNAs were expressed at higher and lower levels, respectively, than in the untreated stage (F28-7 cells). Table 3 lists the names and sequence information of these 25 mature miRNAs. As shown in Table 4, the analysis identified 24 differentially expressed miRNAs in GA-treated F28-7 cells compared with untreated F28-7 cells.
Association with miRNA expression and two types of cancer cell death modes Two major members of ncRNA, namely miRNA and lncRNA, play important regulatory roles in gene expression and numerous important physiological processes, which include cell death [17]. Several reports indicated that ncRNA, including miRNA and lncRNA, function as prodeath or antideath signals in various cell death modes, that is, apoptosis and programmed necrosis as necroptosis [17]. We identified necrosis-related or apoptosis-related miRNAs by Venn diagram of the microarray data from FUdR-treated F28-7-A cells (apoptotic dying cell condition), FUdRtreated F28-7 cells (necrotic dying cell condition), GA plus FUdR-treated F28-7 cells (apoptotic dying cell condition), and GA-treated F28-7 cells (Fig. 2). Analysis showed the presence of necrosis-related miRNA candidates (pro-necrotic: miR-93-5p, miR-743a-3p, and miR-182-5p and anti-necrotic: miR-1195 and miR-181b-5p) and apoptosis-related miRNAs candidates (pro-apoptotic: miR-1199-5p and anti-apoptotic; miR-691). We will further examine the roles of these cell death-related miRNAs in the cell death mechanisms of necrosis or apoptosis using synthetic miRNA mimics and miRNA inhibitors. Importantly, apoptosis-related miRNAs were not shared by two types of apoptotic dying condition under the FUdR-treated F28-7-A and GA plus FUdR-treated F28-7 cells (Fig. 2). This finding suggests that the cell death fate of necrosis and apoptosis is determined by a miRNA expression patterns of the normal cell condition (untreated condition) or the early stages (within 8h after FUdR treatment) on cell death mode.
We previously reported that two unique miRNAs (i.e., miR-351-5p and miR-743a-3p) may regulate the processes of cell death, necrosis, and apoptosis [14]. These miRNAs exhibited higher expression in apoptotic cells (F28-7-A) than in necrotic cells (F28-7). Transfection of a miR-351-5p or miR-743a mimic in necrosis-fated F28-7 cells resulted in a shift of the cell death mode from necrosis to apoptosis [14]. In addition, we found that miR-351-5p regulated the expression of nuclear scaffold lamin B1, its previously identified regulator of cell death [10,12], and mediated FUdR-induced apoptosis [14]. Recently, we showed that miR-351-5p directly interacts with a lamin B1  mRNA using the cell-free in vitro miRNA and mRNA-binding evaluation system [15]. We considered that both miR-351-5p and miR-743a-3p function as pro-apoptotic miRNAs in the cell death mechanisms. Interestingly, miR-351-5p was downregulated by 0.5fold after cotreatment with GA and FUdR vs the untreated condition (Table 3). Also, miR-743a-3p was upregulated by 1.5-fold after treatment with FUdR vs the untreated condition (Table 1). These data suggest that the expression of cell death regulators miR-351-5p and miR-743a-3p plays an important role in the decision of cell death fate under normal condition. Furthermore, the expression of miR-351-5p and miR-743a-3p may be important very early in the cell death fate-determining process. Notably, Dicer is the predicted target of miR-351-5p in miRDB (http://mirdb. org/index.html). Through this miRNA microarray analysis, we revealed that the intracellular miRNA expression patterns differ between FUdR-treated F28-7 cells (necrotic dying cell condition), GA plus FUdR-treated F28-7 cells (apoptotic dying cell condition), and FUdR-treated F28-7-A cells (apoptotic dying cell condition). Interestingly, we revealed that miR-1199-5p (a proapoptotic miRNA candidate) is specifically  upregulated by 1.5-fold after treatment with FUdR in F28-7-A cells (i.e., under the apoptotic dying cell condition) vs the untreated condition (Table 2). Of note, the roles of miR-1199-5p in cell death mechanisms are poorly understood. Diepenbruck et al. reported that miR-1199-5p acts as a repressor of epithelial-mesenchymal transition, tumor migration, invasion, and lung metastasis [22]. In addition, mechanistically, miR-1199-5p functions in a reciprocal double-negative feedback loop with the epithelial-mesenchymal transition transcription factor zinc finger E-box-binding homeobox 1 (Zeb1) [22]. We will further investigate the functions of miR-1199-5p in the regulatory mechanisms of necrosis-apoptosis switching using synthetic miRNA mimics and miRNA inhibitors. Of note, Calabrese et al. reported that the expression of miR-1199-5p is affected by knockdown of Dicer (a miRNA processing-related RNase III-like enzyme) in mouse embryonic stem (ES) [23]. In addition, knockdown of Dicer induces alterations in the expression pattern of miRNAs in mouse ES cells [23].

Modulation of FUdR-induced necrosis by silencing of Dicer expression
Previous studies have shown that some RNase III-like enzymes, such as Drosha and Dicer, are involved in miRNA biogenesis [18,19]. Especially, pre-miRNAs are cleaved by the RNase III-like enzyme Dicer to generate mature miRNAs [18,19]. We investigated the association with the expression of miRNAs and FUdR-induced cell death modes. We performed transfection of the Dicertargeted specific siRNA mixture to examine whether gene silencing of Dicer (a key molecule for the expression of mature miRNAs) in F28-7 cells modulates FUdR-induced cell death. Quantitative real-time PCR analysis was performed for the RNA fraction in F28-7 cells at 48 h after the transfection of electroporation buffer alone, siNS, or Dicer-targeted siRNAs (Fig. 3A). The mRNA expression levels of Dicer in F28-7 cells were reduced < 50 % in the Dicer-targeted siRNAtransfected cells compared with those observed in the siNS -transfected cells. As shown in Fig. 3B, transfection of either siNS or Dicer-targeted siRNAs in F28-7 cells did not change the cell viability after treatment with or without FUdR. We examined the cell morphology of F28-7 cells with low expression of Dicer after treatment with FUdR. Transfection alone of the siNS or the Dicer-targeted RNAs did not have an impact on the cell morphology (Fig. 4A, upper diagram). The necrotic cell morphology in F28-7 cells and apoptotic cell morphology in F28-7-A cells were characteristically observed 21 h after treatment with FUdR [9]. At that time point, the transfection of siNS in F28-7 cells showed a typical necrotic cell feature, namely cell swelling. In contrast, transfection of Dicer-targeted siRNAs in F28-7 cells partially showed typical apoptotic cell features, namely cell blebbing and chromatin condensation (Fig. 4A, bottom). The microscopic observation showed that the distribution of cell morphologies was 10% normal, 60% necrosis, and 30% apoptosis in the Dicer-siRNA-transfected F28-7 cells after treatment with FUdR (Fig. 4B). In the control experiment, the transfection of siNS in F28-7 cells showed that the fraction of cell morphologies was 30% normal and 70% necrosis after treatment with FUdR (Fig. 4B). We further elucidate the regulatory mechanisms of the cell death switch from necrosis to apoptosis and the altered miRNA expression by the Dicer knockdown. Our findings suggest that gene silencing of Dicer partially shifts FUdR-induced necrosis to apoptosis, indicating that the modulation of miRNA expression patterns influences the determination of cell death fate toward necrosis or apoptosis.

Conclusions
Using the FUdR-induced cell death model, we revealed that the expression patterns of miRNAs differ between necrosis and apoptosis. The alterations in the miRNA expression pattern by knockdown of Dicer in necrosis-fated cells changed the cell death mode from necrosis to apoptosis. The present findings may be important in studying the functions of ncRNAs in cell death mechanisms.