The ROS-sensitive microRNA-9/9* controls the expression of mitochondrial tRNA-modifying enzymes and is involved in the molecular mechanism of MELAS syndrome

Mitochondrial dysfunction activates mitochondria-to-nucleus signaling pathways whose components are mostly unknown. Identification of these components is important to understand the molecular mechanisms underlying mitochondrial diseases and to discover putative therapeutic targets. MELAS syndrome is a rare neurodegenerative disease caused by mutations in mitochondrial (mt) DNA affecting mt-tRNA^Leu(UUR). Patient and cybrid cells exhibit elevated oxidative stress. Moreover, mutant mt-tRNAs^Leu(UUR) lack the taurine-containing modification normally present at the wobble uridine (U34) of wild-type mt-tRNA^Leu(UUR), which is considered an etiology of MELAS. However, the molecular mechanism is still unclear. We found that MELAS cybrids exhibit a significant decrease in the steady-state levels of several mt-tRNA-modification enzymes, which is not due to transcriptional regulation. We demonstrated that oxidative stress mediates an NFkB-dependent induction of microRNA-9/9*, which acts as a post-transcriptional negative regulator of the mt-tRNA-modification enzymes GTPBP3, MTO1 and TRMU. Down-regulation of these enzymes by microRNA-9/9* affects the U34 modification status of non-mutant tRNAs and contributes to the MELAS phenotype. Anti-microRNA-9 treatments of MELAS cybrids reverse the phenotype, whereas miR-9 transfection of wild-type cells mimics the effects of siRNA-mediated down-regulation of GTPBP3, MTO1 and TRMU. Our data represent the first evidence that an mt-DNA disease can directly affect microRNA expression. Moreover, we demonstrate that the modification status of mt-tRNAs is dynamic and that cells respond to stress by modulating the expression of mt-tRNA-modifying enzymes. microRNA-9/9* is a crucial player in mitochondria-to-nucleus signaling as it regulates expression of nuclear genes in response to changes in the functional state of mitochondria.

INTRODUCTION

Mitochondria are pivotal organelles for the life and death of the cell. They play a crucial role in ATP production, their primary function via the oxidative phosphorylation (OXPHOS) system, and in many other metabolic, regulatory and developmental processes. The mitochondrial genome encodes 13 key OXPHOS proteins and the tRNAs and rRNAs used for intra-mitochondrial protein synthesis. However, the vast majority of the mitochondrial proteins are encoded by the cell nucleus. Integration of mitochondrial function within the cell depends on anterograde and retrograde signaling pathways. Anterograde regulation arranges signals from external and internal stimuli at the nucleus to activate a genetic program that adjust mitochondrial function accordingly. Mitochondrial retrograde signaling includes diverse communication pathways from mitochondria to the nucleus that influence cellular and organismal activities in response to changes in the functional state of mitochondria. The components of the human retrograde pathways are mostly unknown (1–3). Their identification is crucial to understand the pathophysiology of mitochondrial diseases and find putative therapeutic targets.

Mutations of mitochondrial DNA (mtDNA) are responsible for a variety of rare human diseases, caused by mitochondrial dysfunction (2). Mitochondrial encephalomyopathy with lactic acidosis and stroke-like episodes (MELAS; OMIM# 540000) and myoclonus epilepsy associated with ragged-red fibers (MERRF; OMIM# 545000), two of the major clinical subgroups of the mitochondrial encephalomyopathies, are mostly due to mutations in the mitochondrial- (mt-) tRNA^Leu(UUR) and mt-tRNA^Lys genes, respectively, with m.3243A>G, in mt-tRNA^Leu(UUR), and m.8344A>G, in mt-tRNA^Lys, being the most frequent. By constructing cybrid cell lines, in which the mutant mtDNAs derived from patients are transferred into human cells lacking mtDNA, it has been possible to demonstrate the direct involvement of both mutations in mitochondrial dysfunction, and recapitulate features found in cells of patients, such as mitochondrial translation defects (4,5), oxidative stress (6), and diminished respiratory enzyme activity and oxygen consumption (4,7).

Despite the numerous studies published, a full view of the pathogenic mechanism of MELAS and MERRF mutations is lacking. The m.3243A>G mutation at mt-tRNALeu^(UUR), causative of MELAS, has been associated with a repertory of defects, including reduced 3-end processing and CCA addition (8), decreased aminoacylation (9), impaired transcription termination (10), decreased half-life of tRNALeu^(UUR) molecules (11), abnormal tRNA folding (12), and impaired wobble uridine (U34) modification (11). All these defects may affect mitochondrial translation and, accordingly, they could justify the functional perturbations of the OXPHOS system in MELAS and MERRF cells.

A correlation between the wobble modification deficiency in mutant mt-tRNAs^Leu(UUR) and the clinical features of MELAS disease has been demonstrated. The mt-tRNAs^Leu(UUR) harboring mutations identified in patients of mitochondrial diseases with different symptoms to MELAS (m.3242G>A, m.3250T>C, m.3254C>T and m.3280A>G) have been reported to show the normal taurine-containing modification at U34, whereas this modification was lacking in mt-tRNAs^Leu(UUR) bearing mutations detected in patients with MELAS symptoms (m.3243A>G, m.3244G>A, m.3258T>C, m.3271T>C and m.3291T>C) (13).

Taurine-containing modifications at U34 have been described in a set of human mt-tRNAs (14,15). Particularly, 5-taurinomethyluridine (τm⁵) and 5-taurinomethyl-2-thiouridine (τm⁵ s²) have been identified in human mt-tRNA^Leu(UUR) and mt-tRNA^Lys, respectively. It is assumed that the 5-taurinomethyl (τm⁵) modification is incorporated into human mt-tRNAs by the nucleus-encoded enzymes GTPBP3 and MTO1 in a way similar to their bacterial homologs MnmE and MnmG, which form an enzymatic complex in which both proteins are interdependent (16). By its side, the 2-thio modification is attributed to the TRMU protein (also named MTU1 and MTO2), the human homolog of the bacterial MnmA (17). Modifications at both 5′ and 2′ positions of U34 contribute to optimize the translation process yet their decoding properties are still a matter of debate (16,18–22).

Since the mutant mt-tRNA^Leu(UUR) and mt-tRNA^Lys molecules from MELAS and MERRF cells, respectively, lack the 5-taurinomethyl (τm⁵) and 5-taurinomethyl-2-thiouridine (τm⁵ s²) modifications, mutations m.3243A>G and m.8344A>G, located far away from U34, have been proposed to exert their pathogenic effect by altering the tRNA identity determinants required by the modifying enzymes, thus preventing U34 modification and causing a decoding disorder in mitochondrial translation (11,23). Curiously, the lack of U34 modifications produced by mutations in MTO1 (OMIM*614667) and TRMU (OMIM*610230) genes, which are, respectively, associated with hypertrophic cardiomyopathy and acute infantile liver failure, does not appear to consistently alter mitochondrial protein synthesis in patient cells (24–26). Even though compensatory mechanisms could explain for efficient translation in a context of hypomodified tRNAs, the point is that patient cells still exhibit mitochondrial dysfunction. Therefore, it is reasonable to hypothesize that functional disturbance of mt-tRNA-modifying proteins could also be associated with OXPHOS dysfunction in a translation-independent manner (25).

MELAS and MERRF cells are characterized by increased levels of reactive oxygen species (ROS) (6,27–29). Mitochondrial ROS can originate from ineffective transfer of electrons through the OXPHOS system (2,30,31), and operate at a low to intermediate concentration as key messengers in mitochondria-nucleus crosstalk, leading to an adjustment of the gene expression according to the stress condition (2,32,33). However, excessive ROS accumulation due to overwhelmed cellular defense is able to influence genetic and epigenetic cascades, and thus, the gene expression pattern (2,33–35). In this respect, Park and collaborators have recently reported in MELAS cells that a retrograde signaling pathway involving ROS contributes to diminish mRNA abundances of OXPHOS proteins encoded by the nuclear genome, thereby aggravating the mitochondrial dysfunction (36). Interestingly, a reduction in the steady-state levels of the nuclear-encoded TRMU enzyme was reported to accompany the 2-thiolation defect at the wobble uridine of mutant tRNA^Lys in MERRF cells (25). In addition, low TRMU levels were observed in MELAS cells, even though the tRNA harboring the mutation in this case (tRNA^Leu(UUR)) is not a natural substrate of TRMU. These findings suggest that regulation of TRMU responds to the altered functional state of MERRF and MELAS mitochondria. The information on mechanisms regulating the expression of mt-tRNA-modifying enzymes is scarce. Therefore, we decided to explore whether the expression of the proteins involved in the U34 modification of mt-tRNAs (TRMU, GTPBP3 and MTO1) is subject to a ROS-directed down-regulation in MELAS cells and, if so, whether such a down-regulation is involved in the pathogenic mechanism.

microRNAs (miRNAs) are highly conserved small ∼22 nt noncoding RNAs that have emerged as important post-transcriptional regulators of gene expression. These tiny RNAs often act by directly binding to target mRNAs, usually in the 3′-untranslated region (3′UTR), with the consequent translational repression and/or target-mRNA degradation. The resulting effect depends on the strength of the interaction between the miRNA and its targets as well as the relative abundance of each (37). Since a given miRNA can regulate the expression of multiple targets at the same time, miRNAs can potentially provide rapid coordinated changes in gene expression in response to stimuli and stresses (38,39). In this respect, cell exposure to oxidative stress-inducing agents is able to change the expression of several miRNAs (40–42). Moreover, the increased production of ROS observed in several mtDNA disorders has been hypothesized to cause a miRNA-mediated response, although no evidence has been provided to date (34). These data prompted us to investigate whether enhanced ROS levels in MELAS cells induce a post-transcriptional miRNA-mediated response that is responsible for regulation of the mt-tRNA-modifying enzymes.

In this study, we demonstrate that miRNA-9/9* is induced through a ROS/NFkB signaling pathway in MELAS cybrid cells. This miRNA down-regulates the post-transcriptional expression of the GTPBP3, MTO1 and TRMU genes, leading to mt-tRNA hypomodification and contributing to mitochondrial dysfunction in MELAS cybrids.

RESULTS

GTPBP3, MTO1 and TRMU genes are down-regulated in MELAS cybrid cells

The mRNA and protein expression levels of the GTPBP3, MTO1 and TRMU genes in MELAS and wild-type (WT) cybrids were evaluated by qRT–PCR and western blot, respectively (Fig. 1). A pronounced reduction in the expression of the three genes was found in MELAS cybrids when compared with WT controls (Fig. 1A–D). This reduction was not promoted by a lower content of mtDNA copies as no differences were observed between both cell types after the relative quantification of the mitochondrial MT-CO2 gene to the nuclear SDHA gene (Fig. 1E).

Figure 1.

GTPBP3, MTO1 and TRMU genes are down-regulated in MELAS cybrids by a post-transcriptional mechanism. (A–C) Western blot analysis of (A) GTPBP3, (B) MTO1 and (C) TRMU protein expression in wild-type (WT) and MELAS cybrids. The blots were also probed for VDAC1 (porin) as a loading control. Marker protein sizes are shown in kDa. Quantifications of at least three independent biological replicates are included below a representative blot for each protein. (D) qRT–PCR analysis of GTPBP3, MTO1 and TRMU mRNA expression in WT and MELAS cybrids. (E) Relative quantification of the mitochondrial-encoded CO2 gene to the nuclear-encoded SDH gene in WT and MELAS cybrids by qPCR. (F) Transcriptional rates of GTPBP3, TRMU, MTO1 and SOD1 genes by RNA polymerase II-chromatin immunoprecipitation. Results are the Mean ± SD of at least three independent biological replicates and are expressed as fold change respect to the WT values. *P < 0.05, **P < 0.01, ***P < 0.001.

To assess whether the decrease in GTPBP3, MTO1 and TRMU mRNAs is due to transcriptional regulation, we evaluated the transcriptional rate of the three genes in both WT and MELAS cybrids by RNAPol-ChIP (Fig. 1F). The transcriptional rate of the superoxide dismutase 1 (SOD1) gene was included as a positive control based on its previously reported overexpression in MELAS cybrids (6). We found that GTPBP3, MTO1 and TRMU exhibited a similar transcriptional rate in MELAS and WT cybrids (Fig. 1F), which indicates that a post-transcriptional rather than a transcriptional mechanism is responsible for the down-expression of these genes in MELAS cells.

Increased levels of miR-9/9* in MELAS cells down-regulate the expression of GTPBP3, MTO1 and TRMU

miRNAs are well-known post-transcriptional regulators. The expression of some miRNA species has been reported to change in response to different oxidative scenarios (41–43). To study a possible role of ROS-sensitive-miRNAs in the down-regulation of the GTPBP3, MTO1 and TRMU genes, we first looked for miRNA-binding-sites in the 3′UTR of their mRNAs. A list of miRNAs predicted to target each gene was prepared in accordance with three databases, miRWalk, Targetscan and miRanda (44–46) (Supplementary Data). From these lists, we selected the ROS-related-miRNAs described so far (miR-200a/b/c, miR-141, miR-429, miR-210, miR-215, miR-27, miR-217, miR-34, miR-15, miR-424, miR-155, miR-183, miR-872, miR-199a, miR-30b, miR-30d, miR-16, miR-26b, miR-23a/b, miR-9, miR-125b, miR-146a, miR-128, miR-335, miR-338 and miR-21) (40–43,47–50). The three databases indicated a predicted miR-9 binding site in the 3′UTR region of GTPBP3 mRNA (Supplementary Data). A seed sequence for this miRNA was also found in 3′UTR of TRMU mRNA by the miRanda database (Supplementary Data). Interestingly, miR-9*, the complementary strand to miR-9, was predicted to target MTO1 mRNA by the miRWalk and miRanda databases (Supplementary Data). Although other ROS-related-miRNAs apart from miR-9/9* were identified among the list elaborated for each gene, neither themselves nor their complementary strand was found to be a common predicted regulator for the three genes. An alignment of the 3′UTR regions of GTPBP3, TRMU and MTO1 mRNAs revealed an evolutionarily conservation of miR-9/9* seed sequences in mammals (Fig. 2A) (51). Therefore, miR-9/9* was selected as a candidate for binding to the 3′UTRs of GTPBP3, MTO1 and TRMU mRNAs and to down-regulate their expression.

Figure 2.

The highly conserved miR-9/9* is over-expressed in MELAS cybrids. (A) Conservation of miR-9/9* target sequences (in blue) in the 3′UTR region of GTPBP3, MTO1 and TRMU mRNAs in mammals. Sequences were obtained from NCBI and miRNA/target duplex prediction was performed with RNAhybrid (51). (B) qRT–PCR analysis of miR-9 and miR-9* expression in MELAS and WT cybrids. Data show the Mean ± SD based on three independent biological replicates. ***P < 0.001.

To evaluate whether miR-9/9* regulates the GTPBP3, MTO1 and TRMU steady-state mRNA levels, we first determined the relative mature miR-9/9* expression in MELAS when compared with WT cells by qRT–PCR. As shown in Figure 2B, miR-9/9* was significantly up-regulated (∼2.5-fold) in MELAS cells. A similar miR-9 fold change has been reported in different situations, such as serum deprivation in cancer cells (52) and retinoic acid (RA)-induced differentiation of human neuroblastoma SH-SY5Y cells (53).

Next we explored whether the transfection of MELAS cells with a miRNA-9 antagonist (anti-miR-9) reverses the effect of miR-9 in these cells by increasing the GTPBP3 and TRMU expression. A decrease of about 50% in the miRNA-9 endogenous levels was achieved after transfection of MELAS cells with the anti-miR-9 (Fig. 3A), which was accompanied by a noticeable increase in the mRNA and protein levels of GTPBP3 and TRMU (Figure 3B and C). These results indicate that the miR-9 overexpression in MELAS cells is responsible for the down-regulation of GTPBP3 and TRMU. Considering that the RNAPol-ChlP experiments (Fig. 1F) indicated that the down-regulation of these genes in MELAS cells does not occur at transcriptional level, the increase of TRMU and GTPBP3 mRNA levels in the anti-miR-9-transfected cells also suggests that miR-9 controls expression of the mt-tRNA-modifying proteins through mRNA destabilization.

Figure 3.

miR-9 is a direct post-transcriptional regulator of the GTPBP3, TRMU and MTO1 genes. (A–C) miR-9 levels (A), and GTPBP3 and TRMU mRNA (B) and protein (C) levels in MELAS cybrids transfected with either anti-miR-9 or the negative control NC-anti-miR. In (C), a representative western blot (left) of those used for quantification of protein levels (right) is shown. (D and E) mRNA (D) and protein (E) levels of GTPBP3 and TRMU in 143B (WT cybrids), HEK-293 and HeLa cells transfected with either pre-miR-9 or the negative control NC-pre-miR. Protein levels in (E) were analyzed by western blot. (F and G) Effects of miR-9 and miR-9* transfection on the activity of luciferase reporter constructs containing the GTPBP3-, TRMU- and MTO1–3′-UTRs in the direct (+) (F) or reverse (−) (G) direction. In (C) and (E), marker protein sizes are shown in kDa. Data show the Mean ± SD of at least three independent biological replicates. *P < 0.05, **P < 0.01, ***P < 0.001, n.s. (non-significant differences).

Then we investigated whether transfection of WT cells with a pre-miR-9 mimic reduces the expression of GTPBP3 and TRMU. To this end we transfected different cell lines with pre-miR-9 or negative control (NC-pre-miR) oligonucleotides. We observed a clear reduction in the mRNA and protein levels of GTPBP3 and TRMU in the pre-miR-9-transfected cells when compared with the negative control (Figure 3D and E). These experiments reveal the anti-correlated expression patterns of miR-9 and their targets, which is a typical feature of regulation by miRNAs (39).

Finally, we used the luciferase reporter gene assay to confirm that miR-9/9* directly targets GTPBP3, MTO1 and TRMU. To this end, we first cloned the 3′UTR of the respective mRNAs in a direct or reverse direction downstream of a luciferase reporter gene. Then we co-transfected these constructs in HeLa cells together with the miR-specific precursor or the negative control. As shown in Figure 3F, the co-transfection of the GTPBP3 and TRMU reporters together with the pre-miR-9 mimic resulted in ∼50 and 30% reduction in luciferase activity, respectively, when compared with the NC-pre-miR-transfected cells. On the other hand, the MTO1 reporter showed a 50% reduction in luciferase activity when it was co-transfected together with pre-miR-9*. These effects were lost when the reporters carried the 3′UTR cloned in the reverse direction (Fig. 3G).

Taken together, the present data indicate that TRMU, GTPBP3 and MTO1 are directly regulated by miR-9/9*, and that the induced levels of this miRNA in MELAS cells are responsible for the down-regulation of the mentioned mt-tRNA-modifying enzymes. It is noteworthy to point out that a 2-fold reduction in protein level as that found for the mt-tRNA-modifying protein in MELAS cells (Fig. 1A–C) is a standard effect of regulation by microRNAs, which often act as fine-tuners of gene expression (39).

ROS control the expression of miR-9, GTPBP3, MTO1 and TRMU in MELAS cells

Given that MELAS cells exhibit an increased amount of ROS (6) and that miR-9/9* has been previously reported to be induced by ROS-generating neurotoxic metal sulfates (42), we speculated whether ROS themselves are capable of controlling the miR-9 expression in MELAS cells and thus down-regulating the GTPBP3 and TRMU gene expression. To test this hypothesis, WT cells were treated for 3, 5 and 8 h with 5 mm hydrogen peroxide (H₂O₂) or 1 μm plumbagin, which are known to activate the endogenous production of ROS (54–56). As shown in Figure 4A, the miR-9 expression increased significantly after the 8-h H₂O₂ treatment, while the miR-9 expression was affected after 5 h with plumbagin. Consistently, we found reduced levels of GTPBP3 and TRMU mRNAs in WT cells treated with both oxidants for 8 h (Fig. 4B). Note that treatment with H₂O₂ for 8 h and plumbagin for 5 h led to GTPBP3 and TRMU mRNA levels similar to those observed in MELAS cells.

Figure 4.

ROS play a critical role in miR-9, GTPBP3 and TRMU expression. (A) qRT–PCR analysis of miR-9 expression during treatment of WT cybrids with 5 mm H₂O₂ and 1 μm Plumbagin for the times indicated. (B) GTPBP3 and TRMU mRNA expression levels in WT cybrids after treatment with H₂O₂ (5 mm) for 8 h and with Plumbagin (1 μm) for 5 and 8 h. Expression in MELAS cybrids is used as a control. (C) Determination by flow cytometry of superoxide anion (⁠${{\text{O}}_2}^{-}$⁠) in WT, MELAS and MELAS cybrids treated for 48 h with 1 mm NAC and 100 μm MnTBAP. (D) Determination by flow cytometry of hydrogen peroxide (H₂O₂) in WT, MELAS and MELAS cybrids treated for 48 h with 1 mm NAC and 100 μm MnTBAP. (E and F) qRT–PCR analysis of miR-9 (E), and GTPBP3 and TRMU mRNA (F) expression after 48 h treatment of MELAS cybrids with 1 mm NAC and 100 μm MnTBAP. (G) Western blot analysis of GTPBP3 and TRMU protein expression after 48 h treatment of MELAS cybrids with 1 mm NAC and 100 μm MnTBAP. Marker protein sizes are shown in kDa. Mean ± SD of at least three independent biological replicates were represented. *P < 0.05, **P < 0.01, ***P < 0.001, n.s. (non-significant differences). In (A), (B), (E) and (F), data are expressed as fold change respect to the control sample. a.u., arbitrary units.

Then we evaluated the effect of reducing ROS levels in MELAS cells by antioxidant treatment with N-acetyl-cysteine (NAC) and Mn(III)tetrakis (4-benzoic acid) porphyrin (MnTBAP). NAC exerts its antioxidant activity by inducing superoxide dismutase activity, which is responsible for converting superoxide (⁠${{\text{O}}_2}^{-}$⁠) into oxygen (O₂) and hydrogen peroxide (H₂O₂) (57,58), whereas MnTBAP is a superoxide dismutase mimetic (59). After treating MELAS cells with both antioxidant reagents for 48 h, the superoxide anion (⁠${{\text{O}}_2}^{-}$⁠) and H₂O₂ levels were evaluated by flow cytometry using hydroethidine and dihydrorhodamine 123, respectively. The ${{\text{O}}_2}^{-}$ levels were reduced in the antioxidant-treated MELAS cells when compared with the untreated MELAS cells (Fig. 4C), whereas the H₂O₂ levels increased in treated cells (Fig. 4D) as a result of ${{\text{O}}_2}^{-}$ dismutation. Under these conditions, the miR-9 expression was reduced in the treated MELAS cells (Fig. 4E), which was accompanied by a recovery of GTPBP3 and TRMU mRNA and protein expression (Fig. 4F and G, respectively). These results indicate that enhanced superoxide production in MELAS cells contributes to increasing the miR-9 levels and to consequently deregulating the expression of GTPBP3 and TRMU. In contrast, these data suggest that H₂O₂ is not directly responsible for miR-9 induction after H₂O₂ or plumbagin treatment (Fig. 4A) given that the increase of H₂O₂ by the antioxidant treatments of MELAS cells, detected with dihydrorhodamine (Fig. 4D), was parallel to a decrease of the miR-9 levels (Fig. 4E). Therefore, superoxide and other ROS, but not H₂O₂, appear to be responsible for the miR-9 induction.

Induction of miR-9 occurs via an NFkB signaling pathway

In humans, miR-9 is transcribed from three independent miR-9 gene loci, designated as pri-miR-9-1, pri-miR-9-2 and pri-miR-9-3, which encode three different hairpin precursors with the same mature miR-9 sequences (60). The control of the transcriptional activity of the miR-9 gene loci has been assigned to different transcription factors like REST, MYC, CREB and NFkB (52,61,62). In particular, NFkB has been reported to be recruited to pri-miR-9-1 promoter 24 h after serum deprivation in human hepatoma cells (52), while usage of NFkB inhibitors has demonstrated the NFkB dependence of miR-9 induction in human monocytes and neutrophils after exposure to bacterial lipopolysaccharides (63).

Since enhanced ROS levels are an inherent condition to MELAS cells and as the ROS generated by mitochondria have been demonstrated to activate NFkB (64), we wondered whether increased NFkB activity is responsible for the induced levels of miR-9/9* in MELAS cells, and consequently for the down-regulation of the GTPBP3 and TRMU genes. To address this point, we first investigated the activation state of NFkB in WT and MELAS cybrids by analyzing the levels of its inhibitor IκBα. NFkB activation is initiated by the signal-induced degradation of IκB repressor proteins, which is stimulated by cytokines such as TNFα (65). Degradation of the repressor triggers the translocation of NFkB from the cytoplasm to the nucleus and activates the transcription of the target genes. We found that the IkBα expression is compromised in MELAS cells (Fig. 5A), treated or not with TNFα, which suggests enhanced NFkB activity in these cells.

Figure 5.

Treatment of MELAS cybrids with NFkB inhibitors reduces miR-9 levels and recovers TRMU and GTPBP3 mRNA expression. (A) Western blot analysis of IkBα protein expression in WT and MELAS cybrids treated or not with TNFα (10 ng/ml) for 30 min. The filter was also probed with β-actin as a loading control. Marker protein sizes are shown in kDa. (B and C) MELAS cybrids were treated for 8 h with MG-132 (10 μm) and PDTC (300 μm). The miR-9 (B), and GTPBP3 and TRMU mRNA (C) levels were determined by qRT–PCR analysis. Mean ± SD of three independent biological replicates were plotted. *P < 0.05, **P < 0.01, ***P < 0.001.

Then we studied the NFkB dependence of miR-9 induction and the GTPBP3- and TRMU-expression using a potent NFkB inhibitor, pyrrolidinedithiocarbamic acid (PDTC) (66), and a general inhibitor of proteases, carbobenzoxy-Leu-Leu-leucinal (MG-132) (67), which ablates IκBα degradation and NFkB activation. After treatment with each agent, the expression of miR-9, GTPBP3 and TRMU mRNAs was evaluated by qRT–PCR (Fig. 5B and C). We found that the miR-9 levels were severely reduced in the treated MELAS cells when compared with untreated cells (Fig. 5B), whereas the GTPBP3 and TRMU mRNA expression was almost recovered upon treatment (Fig. 5C). These results suggest that the NFkB pathway participates in the induction of miR-9 by ROS in MELAS cells, and therefore indirectly controls the expression of GTPBP3 and TRMU.

miR-9 overexpression decreases the 2-thiouridylation of mt-tRNA^Glu and mt-tRNA^Lys in MELAS cells

Previous reports have demonstrated that the pathogenic m.3243A>G point mutation in tRNA^Leu(UUR) causes a τm⁵U-modification deficiency at U34 (13). As we found a lower expression for GTPBP3, MTO1 and TRMU enzymes in MELAS cells, we wondered whether this situation leads to wobble hypomodification of mt-tRNAs other than the mutated tRNA^Leu(UUR). The evaluation of the 2-thiolation at the wobble uridine of the mt-tRNA^Glu and the mt-tRNA^Lys promoted by the TRMU protein can be performed by a modified Northern Blotting version (APM-Northern blotting) (68). The p-(N-acrylamino)-phenyl]mercuric chloride (APM) included in the gel binds to the thiol group and determines a band shift. As shown in Figure 6A and B, mt-tRNA^Glu and mt-tRNA^Lys exhibited an increased non-thiolated mt-tRNA fraction in MELAS cybrids when compared with WT cells. Quantification of the band intensities corresponding to both mt-tRNA fractions (thiolated and non-thiolated mt-tRNAs) from three independent experiments revealed a significant 25–30% reduction in the 2-thiolated mt-tRNAs^Glu and mt-tRNAs^Lys levels in MELAS cells when compared with WT cells. Our results demonstrate for the first time that mt-tRNAs other than the mutant mt-tRNA^Leu(UUR) exhibit hypomodification levels in MELAS cells.

Figure 6.

miR-9 regulates the 2-thiolation status of mt-tRNAs Glu and Lys. (A and B) APM-Northern analysis of the 2-thiolation status of mt-tRNA^Glu (A) and mt-tRNA^Lys (B) molecules from WT and MELAS cybrids. The same amount of total RNA (7.5 μg) was run in a denaturing polyacrylamide-urea gel in the presence (+) or absence (−) of APM. After transfer to a Nylon membrane, hybridation was performed with probes for mt-tRNA^Glu and mt-tRNA^Lys. The thiolated tRNAs were detected as retarded bands in the presence of APM. Representative northern blots are shown. Quantification of the relative abundance of thiolated and non-thiolated tRNAs represents the Mean ± SD of at least three independent biological replicates. (C and D) APM-Northern analysis of the 2-thiolation status of mt-tRNA^Glu (C) and mt-tRNA^Lys (D) molecules from WT cybrids transfected with NC-Pre-miR, Pre-miR-9 and Pre-miR-9*. Total RNA (7.5 μg) from transfected samples were processed as described in (A). Retarded bands (2-thiolated tRNAs) and non-retarded bands (non-thiolated tRNAs) are indicated in the APM panels. **P < 0.01.

In order to further explore whether hypothiolation in MELAS cybrids is the direct consequence of the TRMU down-regulation elicited by the increased miR-9 levels, we transfected WT cells with pre-miR-9 and evaluated the effect on 2-thiolation (Fig. 6C and D). An increase in the non-thiolated tRNA^Glu and tRNA^Lys fractions was observed in pre-miR-9 transfected cells in comparison with WT cells transfected with the control (NC-pre-mir-9). No effect was observed for the transfection with pre-miR-9*, thus supporting the notion that TRMU is a target for miR-9, but not for miR-9*. Accordingly, we conclude that the miR-9-mediated down-regulation of the TRMU expression observed in MELAS cells has direct consequences for the mt-tRNAs modified/unmodified ratio.

miR-9 induction contributes to mitochondrial dysfunction in MELAS cells

In a previous work, we demonstrated that silencing the GTPBP3 gene in HEK-293 cells by transitory transfection with siRNAs results in mitochondrial dysfunction, which was characterized by reduced oxygen consumption rate, cellular ATP content, mitochondrial membrane potential and mitochondrial protein synthesis (69). Other authors have reported a decrease of oxygen consumption rate and mitochondrial membrane potential in HeLa cells transfected with TRMU siRNAs (17). In the present work, we found that the knocking-down of TRMU by transient transfection of 143B cells with siRNAs led to a decrease of both ATP levels (∼50%) and thiolated fractions of tRNA^Glu and tRNA^Lys (Fig. 7). In addition, we found that siRNA-dependent MTO1 knocking-down caused a significant decrease of ATP levels (Supplementary Data). Given that miR-9/9* regulates the expression of TRMU, GTPBP3 and MTO1, we decided to investigate whether transfection of WT cells with pre-miR-9 has similar effects on mitochondria function to those produced by GTPBP3-, TRMU- and MTO1-knockingdown, and whether transfection with anti-miR-9 can reverse the mitochondrial dysfunction of MELAS cybrids. A decrease of about 50% in cellular ATP levels was obtained in pre-miR-9-transfected WT cybrids (Fig. 8A), and also in HEK-293 cells (Supplementary Data), whereas transfection of MELAS cybrids with anti-miR-9 significantly increased the cellular ATP levels when compared with the control (Fig. 8B). Moreover, when MELAS cells were treated with antioxidants, which reduce the ROS level and, consequently, the miR-9 expression (Fig. 4E), a slight but significant increase in ATP content was observed (Supplementary Data). Altogether these results indicate that the overexpression of miR-9 contributes to the reduced ATP content exhibited by MELAS cells.

Figure 7.

Knockdown of TRMU by siRNAs produces mitochondrial dysfunction in 143B cells. (A) qRT–PCR analysis of TRMU mRNA expression in 143B cells (WT cybrids) transfected during 72 h with TRMU-specific siRNAs and the negative control NC-siRNA. (B) Western blot analysis of TRMU protein expression in 143B cells (WT cybrids) transfected during 72 h with one of the TRMU-specific siRNAs (siRNA_2) and the negative control NC-siRNA. Marker protein sizes are shown in kDa. (C) Cellular ATP determination in TRMU-specific siRNAs transfected 143B cells. (D and E) APM-Northern analysis of the 2-thiolation status of mt-tRNA^Glu (D) and mt-tRNA^Lys (E) molecules from 143B cells (WT cybrids) transfected during 72 h with either a TRMU-specific siRNA or the negative control NC-siRNA. Representative northern blots are shown. Quantification of the relative abundance of thiolated and non-thiolated tRNAs represents the Mean ± SD of at least three independent biological replicates. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 8.

Enhanced miR-9 levels contribute to mitochondrial dysfunction in MELAS cybrids. (A and B) Cellular ATP determination in NC-Pre-miR- and Pre-miR-9-transfected WT cybrids (A), and NC-anti-miR- and anti-miR-9-transfected MELAS cybrids (B). (C) Determination of mitochondrial membrane potential by flow cytometry in NC-Pre-miR- and Pre-miR-9-transfected WT cybrids, and in NC-anti-miR- and anti-miR-9-transfected MELAS cybrids. WT cybrids treated for 30 min with sodium azide at 25 mm were included in the analysis as a positive control. Mean ± SD of three independent biological replicates were plotted. *P < 0.05, **P < 0.01, ***P < 0.001, n.s. (non-significant differences).

Then we analyzed the effect of miR-9 deregulation on the mitochondrial membrane potential by flow cytometry using MitoTracker Red. As shown in Figure 8C, transfection of WT cells with pre-miR-9 significantly diminished the membrane potential, while the transfection of MELAS cells with anti-miR-9 increased this potential at the WT levels.

Altogether these results indicate that the up-regulation of miR-9 in MELAS cells contributes to mitochondrial dysfunction, which likely occurs via down-regulation of the mt-tRNA-modifying proteins.

Considering that oxidative stress is also a feature of MERRF cells (70), we explored the levels of miR-9/9*, TRMU and GTPBP3 in MERRF cybrids. We found that miR-9/9* was up-regulated, whereas TRMU and GTPBP3 were down-regulated (Supplementary Data). This anti-correlative expression pattern of miR-9 and their targets in MERRF cybrids is reminiscent of that found in MELAS cybrids, and suggests that miR-9/9* also contributes to the MERFF phenotype by fine-tuning the expression of TRMU, GTPBP3 and MTO1.

DISCUSSION

In this study, we demonstrate that miRNA-9/9* is induced in MELAS cybrids through a ROS- and NFkB-dependent pathway, and down-regulates the expression of mt-tRNA-modifying enzymes TRMU, GTPBP3 and MTO1. Transfection of MELAS cells with an antagonist of miR-9 (anti-miR-9) partially reversed the phenotype, whereas transfection of WT cells with a pre-miR-9 mimic caused mitochondrial dysfunction as the ATP content and membrane potential decreased. Notably, these features were similar to those resulting from separately knocking-down TRMU, GTPBP3 and MTO1 with siRNAs (17,25,69, this work). Therefore, we conclude that the induced levels of miR-9/9* in MELAS cybrids contribute to the phenotype of these cells via the down-regulation of TRMU, GTPBP3 and MTO1.

We have found that miR-9/9* expression also increased in MERRF cybrids, whereas the levels of proteins GTPBP3 and TRMU lowered when compared with WT cells. Thus, miR-9/9* also appears to be an active player in the molecular mechanism underlying MERRF disease. Our data show for the first time that two mtDNA disorders directly affect miRNA expression (34).

miRNAs confer robustness to biological process; i.e. they form part of the cell's strategies to maintain its functions in spite of internal or external perturbations (39,71). However as the cell cannot cope with a particular stress, a persistent dysregulation of miRNA expression can contribute to stress-related chronic diseases (71). This appears to be the case in MELAS and MERRF cells. Oxidative stress, due to impaired OXPHOS system functioning as likely a result of the decoding defect, induces miR-9/9*, leading to an aggravation of the mitochondrial dysfunction by reducing the amounts of mt-tRNA-modifying enzymes (Fig. 9).

Figure 9.

A miR-9/9*-based new model for the pathogenic mechanism of MELAS disease. The circled numbers on Figure 9 try to facilitate the comprehension of the events, although it may be ample opportunity for crosstalk between the different steps. Initially, the hypomodification of the mt-tRNAs^Leu(UUR) bearing mutation m.3243A>G (steps 1 and 2) might not only affect the synthesis of specific proteins encoded by the mitochondrial genome (step 3), but may also trigger a retrograde signaling pathway (step 4), which alters the synthesis of nuclear-encoded mitochondrial proteins (step 5). In any case, the outcome would be the functional disturbing of the OXPHOS system (step 6) and increased ROS production (step 7) which, in turn, induces miR-9/9* (steps 8, 9 and 10). Induction of this miRNA leads to the down-regulation of mt-tRNA-modifying proteins (step 11), which extends hypomodification to other mt-tRNAs (step 12) and further activates the retrograde pathway (steps 4 and 5). This feedback mechanism would amplify the original effect of the MELAS mutation. Our data evidence a new mitochondrial retrograde signaling pathway in which miR-9/9* is a crucial player.

miR-9 has been found to be expressed at high level in brain and, to a lower extent, in insulin-secreting cell lines and in pancreatic islets (72,73). Moreover, it has been involved in pancreas, and therefore β-cell, development (74). Diabetes mellitus is a metabolic disorder that manifests when insulin production by the pancreas is insufficient or when the body cannot effectively use the secreted insulin. Interestingly, miR-9 overexpression in insulin-secreting cell lines causes drastic impairment of glucose-stimulated insulin release (72,73). Thus miR-9 is relevant to diabetes, a condition that is often associated with MELAS mutation m.3243A>G (75), and less frequently with MERRF mutation m.8344A>G (76). Our finding that miR-9/9* is induced in MELAS and MERRF cells may provide a molecular basis for understanding the manifestation of diabetes in a group of patients carrying the aforementioned mutations.

As stated above, miR-9/9 is highly expressed in the central nervous system of vertebrates (60,77). How this feature may affect the expression of TRMU, MTO1 and GTPBP3 as well as the modification status of mt-tRNAs in brain is currently unknown. In any case, it should be kept in mind that the effect of a specific miRNA in a cell type depends on many factors, including the relative expression levels of their mRNA targets and competition by other mRNA targets for the miRNA; i.e. the set of genes that a miRNA regulates in a cell depends on the transcriptional program of that particular cell, which, however, may change under pathological conditions (37,39,71,78). Considering that MELAS and MERRF are neurodegenerative diseases (79), future work should explore the contribution of miR-9/9* to neurological symptoms.

Several data suggest that miR-9 may also be a crucial regulator of a gene expression program that modulates the cardiac hypertrophy response (80). Notably, MTO1 mutations are associated with hypertrophic cardiomyopathy (24,81), and some cases of MELAS and MERRF as well. Therefore, the regulatory role of miR-9/9* (and other miRNAs) in the mitochondria-nucleus crosstalk appears as an exciting research topic to a better understanding of mitochondrial diseases.

The effect of a given miRNA on the level of a target's protein is usually mild (<2-fold) (82,83), as indeed observed in the fine-tuning of TRMU, GTPBP3 and MTO1 by miR-9/9*. Nevertheless, there are several mechanisms by which the impact of an miRNA can increase, including the targeting of genes involved in the same pathway or protein complex (39). Based on the data obtained with their Escherichia coli homologs, it is assumed that human proteins GTPBP3 and MTO1 form an enzymatic complex that is responsible for the modification of an mt-tRNA set (16,84). Our finding that miR-9 controls GTPBP3, whereas miR-9* regulates the MTO1 expression, could well be the second example of how the bifunctional miR-9/9* acts on two members of the same complex (85). Interestingly, miR-9 also targets MTHFD1L [methylentetrahydrofolate dehydrogenase (NADP⁺-dependent) 1–like] and MTHFD2 [methylentetrahydrofolate dehydrogenase (NADP⁺-dependent) 2] (86), which belong to folate metabolism and participate in the synthesis of methylene-tetrahydrofolate. This one-carbon form of tetrahydrofolate has been proposed to be the source of the methylene moiety directly attached to the C5 atom of U34 by the bacterial homologs of GTPBP3 and MTO1 (87). If this were so, miR-9/9* would control not only the expression of GTPBP3 and MTO1 in humans, but also the synthesis of one of the substrates of the modification reaction.

The miR-9/9*-dependent decrease (∼50%) of TRMU, GTPBP3 and MTO1 observed in MELAS cells may have a biological effect because a ∼50% decrease in the amount of each protein, after knocking down the respective gene with siRNAs, results in mitochondrial dysfunction. This phenotype may be due to the limiting amounts of proteins to properly carry out their functional roles. But what exactly are these roles and their consequences? We have observed an increase in non-thiolated tRNA^Glu and tRNA^Lys fractions in both MELAS cybrids and pre-miR-9 transfected cells. These results reveal that the modification status of mt-tRNAs is dynamic and that it responds to miR-9/9*, which is regulated by ROS. Recent works have shown that alterations in the extent of RNA modifications play critical roles in gene expression, often via mRNA decoding, although the mechanisms regulating the ratio of modified versus non modified RNA molecules are unclear (88–94). Our work discloses for the first time a stress-response mechanism that post-transcriptionally regulates the abundance of tRNA-modifying enzymes and may consequently affect the tRNA modification status, as seen for the TRMU-mediated thiolation.

The impact of mt-tRNA hypomodification on mitochondrial translation is unclear. Some TRMU mutations drastically affecting the tRNA modification activity were found to be associated with impaired translation but others were not, and translation appeared to be normal in TRMU-knocked-down cells (25,26). Moreover, MTO1 mutations have been reported to not affect mitochondrial protein synthesis in patient cells exhibiting altered biochemical activity of several respiratory complexes (24). Finally, we have previously found that the transient knocking-down of GTPBP3 in HEK-293 with siRNAs diminished the incorporation of [³H]leucine into mtDNA-encoded proteins by about 20% and reduce the cell ATP content (69). Modifications of U34 like those depending on GTPBP3/MTO1 and TRMU are important for modulating the translation rate of cognate codons and preventing ribosomal frameshifting (16,95,96). Therefore, it is reasonable to propose that inactivation of the tRNA-modifying function of these enzymes affects mitochondrial translation to some extent, but that compensatory mechanism(s) modulated by the genetic background and cell type may lower the level of affectation, making it difficult to detect a significant change in mitochondrial translation. Compensatory mechanisms of translational defects can be of different types, like those including the presence of mutant tRNAs (97,98) or the overexpression of either specific tRNAs (95,99,100), or tRNA synthetases (101–103). Even though compensatory mechanisms may improve the efficiency of the mitochondrial translation, the function of the OXPHOS system can still remain impaired, as seen in patient cells carrying MTO1 and TRMU mutations (24,25,69,81,104). This suggests that other mechanisms, in addition to the altered translation, might contribute to mitochondrial dysfunction associated with mt-tRNA modification defects.

The possibility that human proteins TRMU, GTPBP3 and MTO1 have additional roles to mt-tRNA modification cannot be ruled out. Accumulating data indicate that other mt-tRNA-modifying proteins are involved in different cellular functions including nuclear transcription regulation, tumor suppression, mt-tRNA biogenesis and cell signaling (14,105–108). Interestingly, in yeast the effects of MTO1 inactivation on mitochondrial-mRNA maturation and translation have been reported to be more severe than those produced by the inactivation of the GTPBP3 homolog (109–111), which suggests that these proteins, apart from their shared role in mt-tRNA modification, may play independent roles in other mitochondrial processes.

Another way in which cellular functions can be affected by the down-regulation of TRMU, GTPBP3 and MTO1 is if hypomodified tRNAs perform a signaling function. Loss of conserved wobble uridine modification in yeast cytosolic tRNAs has been found to exert an effect on gene expression due to perturbation of cell signaling, which appears to be independent of any codon-specific translation effects (94). These results suggest an unconventional role for tRNA modifications in regulating gene expression. Accordingly, we speculate that hypomodified mt-tRNAs in MELAS cells may generate signals that alter the expression of nuclear-encoded mitochondrial proteins, leading to an aggravated mitochondrial dysfunction (Fig. 9).

There is recent evidence that MELAS mutation m.3243A>G induces a retrograde signaling pathway involving ROS, kinase JNK, retinoid X receptor α and transcriptional coactivator PGC1α (36). This pathway contributes to diminish the mRNA abundances of nuclear-encoded OXPHOS enzymes via transcriptional regulation. Our results evidence a new retrograde pathway based on the ROS-mediated induction of miR-9/9* that is able to post-transcriptionally down-regulate the expression of mt-tRNA-modifying enzymes, and likely other nuclear-encoded mitochondrial proteins, thus aggravating the effects of the MELAS mutation.

The significance of the miR-9/9-dependent retrograde pathway in the context of physiologic homeostasis is currently undetermined. In light of findings indicating the important role of exported mitochondrial peptides in immunity and other functions (112), the possibility that perturbations of translation caused by mt-tRNA hypomodification may generate peptides that can be exported from the organelle as signaling factors cannot be ruled out. On the other hand, the putative roles of hypomodified mt-tRNAs and mt-tRNA-modifying proteins other than those directly related to translation remain to be explored. Induction of miR-9/9* by ROS, via NFkB, and the subsequent down-regulation of the mt-tRNA modification enzymes could trigger different regulatory signals, enabling the cell to adjust and respond to the stress situation.

MATERIALS AND METHODS

Materials

Hydrogen peroxide solution 30% (w/w) in H₂O, N-acetylcysteine, MG-132, pyrrolidine dithiocarbamate (PDTC) were purchased from Sigma-Aldrich. MnTBAP was purchased from Merck. The APM ([p-(N-acrylamino)-phenyl]mercuric chloride) was synthesized and kindly provided by Prof. Stephane Vincent (68). Oligonucleotides (Supplementary Data) were purchased from Sigma-Aldrich and Qiagen.

Cell culture

Cybrids were generated from platelets derived from two patients carrying m.3243A>G and m.8344A>G mutation, respectively. Platelet cells were isolated from a blood sample and fused to a large excess of mtDNA-less human osteosarcoma 143B (TK−) cells, as described previously (113). Different cybrid clones were obtained by culturing the fusion mixture in selective Dulbecco's modified Eagle's medium (DMEM) (Biological Industries, Kibbutz Beit Haemek, Israel) containing glucose (4.5 g/l), pyruvate (0.11 g/l), 10% dialyzed fetal bovine serum (FBS) (Invitrogen, Carlsbad, CA, USA) and 100 μg/ml BrdU (5-bromo-2-deoxyuridine; without uridine) to prevent growth of TK+ donor cells. Twenty different clones were selected and analyzed by polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP) to obtain cells containing 100% mtDNA with m.3243A>G mutation in the tRNA^Leu(UUR) gene (MELAS cybrids) or m.8344A>G mutation in tRNA^Lys (MERRF cybrids), or the wild-type counterpart (WT cybrids). Cells were tested periodically for the presence or absence of mutation using PCR-RFLP assay (114). All cybrids were cultured in high glucose Dulbecco's modified Eagle medium (Gibco) containing 10% FBS, 1 mm sodium pyruvate, 100 U/ml penicillin, 100 μg/ml streptomycin, 2 mm glutamine and 1 mm non-essential amino acids. Human HEK-293 and HeLa cells were grown in minimum essential medium (Sigma) supplemented with 10% heat-inactivated FBS, 100 U/ml penicillin and 100 μg/ml streptomycin. Cell cultures were kept at 37°C in a humidified atmosphere with 5% CO₂.

Anti-miR and pre-miR transfection

Cells were seeded at 250 000 cells/well in 6-well plate. After 24 h, cells were transfected with one of the RNA oligonucleotides (Anti-miR-9, Negative Control (NC)-anti-miR, pre-miR-9, pre-miR-9* or Negative Control (NC)-pre-miR; Ambion-Applied Biosystems) at the 50 nm final concentration, using 5 μl of the Lipofectamine 2000 reagent (Invitrogen) and Opti-MEM medium. After 6 h of transfection, the medium was changed to fresh growth medium. The optimal concentration of Pre-miR-9 used for transfections was determined empirically following the expression of an mRNA target (TRMU) after transfection. We found that transfection with pre-miR-9 at 50 nm produced similar TRMU mRNA levels to those observed in MELAS cybrids.

APM-northern blotting analysis

The procedure was performed essentially as described (25). Briefly, 7.5 μg of total RNA were run on a 15% polyacrylamide gel containing 7 m urea and 10 μg/ml APM followed by transfer to positively charged Nylon membrane (Manheim Boehringer). Pre-hybridization and hybridization were performed with Dig Easy Hyb (Roche) according to the manufacturer's instructions. mt-tRNA^Glu and mt-tRNA^Lys were detected with specific DIG-labeled synthetic oligodeoxynucleotides (5′-GACTACAACCACGACCAATGATATGAAA-AAC-3′ for mt-tRNA-Glu and 5′-TGGTCACTGTAAAGAGGTGTTGGT-3′ for mt-tRNA-Lys). Quantification of the non-radioactive signals was performed with ImageQuant TL v8.1 (GE Healthcare Life Sciences). The fraction of thiolated and non-thiolated tRNAs was calculated as described (26). Briefly, the quantification of each fraction was expressed as a percentage of the thiolated or the non-thiolated signal from the thiolated + non-thiolated signals (as detected on the (−) APM gel).

Luciferase reporter assays

The 3′UTRs of MTO1 and TRMU genes were amplified by PCR and cloned downstream of the luciferase reporter gene into the XbaI site of the pGL3 Luciferase Reporter Vector (Promega), whereas the 3′UTR of GTPBP3 was cloned downstream of the luciferase reporter gene into the HindIII site of the pMIR-REPORT Luciferase Vector (Applied Biosystems). All constructs were verified by DNA sequencing. HeLa cells were seeded at 50 000 cells/well, and the following day, 500 ng of Luciferase reporter construct and, as an internal control, 25 ng of Renilla Luciferase control vector (Promega) were co-transfected together with one of the RNA oligonucleotides at the 50 nm final concentration, using 1 μl of the Lipofectamine 2000 reagent (Invitrogen) and Opti-MEM medium as described above. After 48 h, cells were lysated and Firely and Renilla luciferase activities from the cell extracts were measured with the Dual-luciferase Reporter Assay System (Promega) according to the manufacturer's instructions.

qRT–PCR

Total RNA was extracted using Trizol reagent (Invitrogen). For miRNA quantification, 10 ng of total RNA were reverse transcribed in 15 μl total reaction using the MultiScribe reverse transcriptase and specific stem-loop RT primers (Applied Biosystems). Then 1.33 μl of cDNA was subjected to a TaqMan miRNA assay (Applied Biosystems) in 20 μl total reaction using specific primers and probes for human miR-9, miR-9* and U6 snRNA according to the manufacturer's protocol. Expression values were calculated using the comparative CT method. U6 snRNA was used as an endogenous control. To obtain an estimate of the copy numbers of miR-9 in MELAS cybrids, we calculated a relative expression value based on the deltaCt-values between miRNA-9 and U6 snRNA according to the formula: estimated relative expression = 2^{(Ct U6 snRNA−Ct miR-9)}. The relative expression of miR-9 and miR-9* was ∼0.009- and 0.001-fold less abundant than U6 snRNA, respectively. Given that there are on average 250 000 copies of U6 snRNA per cell (115,116), we estimate the absolute abundance of miR-9 and miR-9* to be at ∼2250 and 250 copies/cell, respectively, an amount which exceeds the necessary threshold level of miRNA expression (100 copies/cell) proposed for significant target suppression (117). To measure mRNA levels, one-step qRT–PCRs were performed. 150 ng of total RNA were reverse transcribed in 20 μl total reaction and amplified by qPCR using specific primers, SYBR Green Master Mix and MultiScribe reverse transcriptase (Applied Biosystems) according to the manufacturer's instructions. The efficiency values obtained for qPCR amplifications were very near to 2. The expression values were also calculated using the comparative Ct method. ACTB gene was used as an endogenous control. Considering that there are on average 2500 copies of ACTB mRNA per cell (118), the relative abundance of TRMU, GTPBP3 and MTO1 mRNAs was estimated to be at ∼10 copies/cell. PCR reactions were run in an Applied Biosystems StepOne Real-Time PCR System.

mtDNA copy number quantification

mtDNA quantification was performed by qPCR as previously described (119).

Flow cytometry studies

For ROS analysis, cells were seeded at 100 000 cells/well in 12-well plate. On the following day, the medium was changed to fresh growth medium containing or not the corresponding antioxidant. After 48 h of treatment, cells were incubated for 1 h at 37°C with 2.5 μm hydroethidine or 2.5 μm dihydrorhodamine 123, washed twice with PBS, detached with trypsin-EDTA and resuspended in fresh growth medium. Red (620 ± 20 nm band pass filter) or green (525 ± 20 nm band pass filter) fluorescence corresponding to superoxide anion (⁠${{\text{O}}_2}^{-}$⁠) or hydrogen peroxide (H₂O₂) detection was measured in 10 000 cells using a Cytomics FC 500 flow cytometer (Beckman Coulter). Mitochondrial membrane potential was determined in cells transfected or not with the RNA oligonucleotide [Anti-miR-9, Negative Control (NC)-anti-miR, pre-miR-9, pre-miR-9* or Negative Control (NC)-pre-miR; Ambion-Applied Biosystems] at 50 nm final concentration. After 48 h from transfection, cells were washed with PBS, detached with trypsin-EDTA and resuspended in fresh growth medium. Cells in suspension (10⁶ cells/ml) were incubated with 100 nm Mitotracker Red CMXRos for 30 min at 37°C and the emitted red fluorescence (620 ± 20 nm band pass filter) was measured in 10 000 cells. Data were collected and analyzed using a Cytomics FC 500 flow cytometer (Beckman Coulter).

Measurement of intracellular ATP

Cellular ATP content was measured using an ATP bioluminescence assay kit (HSII; Roche), according to manufacturer's protocol. Measurements were performed 48 h after antioxidant treatment or 72 h after transfection. Cells were detached with trypsin-EDTA, resuspended with fresh medium, centrifuged at 100g and resuspended again with kit dilution buffer at 10⁶ cells/ml. After adding the kit reagent, luminescence was determined using a Wallac Victor 2 1420 Multilabel HTS Counter (PerkinElmer, Life Sciences).

RNA polymerase II-chromatin immunoprecipitation (RNA pol II-ChIP)

Chromatin immunoprecipitation (IP) was performed basically as described previously (120). Briefly, cultured cells were fixed in 1% formaldehyde for 10 min in the same growth media and unreacted formaldehyde was quenched with 0.141 m Glycine for 5 min. Cells were washed with PBS containing protease inhibitors, treated with SDS lysis buffer and then lysates were sonicated to prepare chromatin fragments (average DNA size of 500–1000 bp DNA as assessed by agarose gel electrophoresis) using Diagenode's Bioruptor. Pre-cleared samples were then incubated overnight with RNA-pol II antibody (sc-899, Santa Cruz Biotechnology) directed against a peptide mapping at the N-terminus of RNA-pol II. An aliquot of the cross-linked chromatin was processed in the same fashion, but in the absence of the antibody (NoAb fraction), and the first supernatant after the preclearing was saved as input fraction. Then protein A dynabeads (Invitrogen) were used to isolate the immune complexes, followed by DNA purification by phenol/chloroform extraction. Chromatin IP specificity was routinely checked by determining the presence of RNA polymerase II in the IP fraction by western blot analysis. The DNA isolated from IP, NoAb and input was subjected to qPCR using the primers designed to amplify fragments of the coding region of the studied gene and the Syber Green PCR Master mix (Applied Biosystems). Reactions were run in triplicate in an Applied Biosystems StepOne Real-Time PCR System. The coding region of the ACTB gene was also analyzed as a control. For each gene, the results of the NoAb samples were subtracted from those of the IP samples, and they were divided by their corresponding input. The data of the GTPBP3, MTO1, TRMU and SOD1 genes were divided by those of the control ACTB gene and relative values to wild type were represented.

Western blot

Cells were collected and lysated in RIPA buffer (50 mm Tris pH 8, 150 mm NaCl, 2 mm EDTA, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 0.1 mm leupeptin and 1 mm phenylmethylsulfonyl fluoride) at 4°C. Cellular suspensions were frozen-thawed at least eight times and centrifuged at 10 000g for 10 min at 4°C. Supernatants were collected and protein concentration was determined with Bio-Rad protein assay. Western blot was performed as described (121). The primary antibody against MTO1 was purchased from Protein Tech, while those against VDAC1, TRMU, ACTB and IKBα were purchased from Santa Cruz Biotechnology. Of note, western blot analysis with anti-IKBα antibody (sc-847, which recognizes the full-length protein) detected two bands in 143B (TK⁻) cells as occurs in other cell lines (122,123), suggesting that IKBα may be partially phosphorylated in these cells. Anti-GTPBP3 was purified from GTPBP3-His-inoculated rabbit serum (69). The anti-rabbit and anti-mouse IgG-horseradish peroxidase-conjugated secondary antibodies were obtained from Sigma.

Statistical analysis

Pair-wise comparisons among groups were analyzed by the one-tailed Student's t-test. The statistically significant differences between means were indicated by asterisks [P < 0.05 (*), P < 0.01 (**) or P < 0.001 (***)], and non-significant differences by n.s.

AUTHORS' CONTRIBUTIONS

S.M. and M.-E.A. conceived the project; S.M. performed most of the experiments; A.M.-Z. helped with experiments; E.G.-A. and A.L.A. generated cybrid cells and helped with the interpretation of data; S.M. and M.-E.A. analyzed the data and wrote the manuscript.

SUPPLEMENTARY MATERIAL

Supplementary Data.

FUNDING

This work was supported by grants from the Spanish Ministry of Economy and Competitiveness (grant number BFU2010-19737), and the Generalitat Valenciana (grant numbers ACOMP/2012/065; PROMETEO/2012/061) to M.-E.A., and a PhD fellowship from the Instituto de Salud Carlos III to A.M.-Z.

ACKNOWLEDGEMENTS

We thank Dr S. Vincent for his generous gift of APM.

Conflict of Interest statement. None declared.