miR-181a/b downregulation exerts a protective action on mitochondrial disease models.

Mitochondrial diseases (MDs) are a heterogeneous group of devastating and often fatal disorders due to defective oxidative phosphorylation. Despite the recent advances in mitochondrial medicine, effective therapies are still not available for these conditions. Here, we demonstrate that the microRNAs miR-181a and miR-181b (miR-181a/b) regulate key genes involved in mitochondrial biogenesis and function and that downregulation of these miRNAs enhances mitochondrial turnover in the retina through the coordinated activation of mitochondrial biogenesis and mitophagy. We thus tested the effect of miR-181a/b inactivation in different animal models of MDs, such as microphthalmia with linear skin lesions and Leber's hereditary optic neuropathy. We found that miR-181a/b downregulation strongly protects retinal neurons from cell death and significantly ameliorates the disease phenotype in all tested models. Altogether, our results demonstrate that miR-181a/b regulate mitochondrial homeostasis and that these miRNAs may be effective gene-independent therapeutic targets for MDs characterized by neuronal degeneration.

Introduction

Mitochondrial diseases (MDs) represent a relevant group of inherited disorders with a cumulative prevalence of about 1:5,000 individuals (Gorman et al, 2015). They are caused by mutations in either nuclear or mitochondrial genes resulting in oxidative phosphorylation (OXPHOS) impairment, leading to huge variability of symptoms, organ involvement, and clinical course. The clinical manifestations range from dysfunction of single tissue/structures such as the optic nerve in Leber's hereditary optic neuropathy (LHON, MIM535000), to syndromic multi‐organ conditions with a prominent involvement of the central nervous system (CNS), such as microphthalmia with linear skin lesions (MLS, MIM309801, 300887, 300952) and Leigh syndrome (LS, MIM256000). Neurons are particularly sensitive to mitochondrial dysfunction due to their highest energy demands, and defects in mitochondrial metabolism may lead to severe energy deficiency, increased reactive oxygen species (ROS), and neuronal death.

MDs are biochemically and genetically heterogeneous, and their complexity has so far prevented the development of effective treatments (Sanchez et al, 2016). In the past few years, specific modulation of either mitochondrial biogenesis/dynamics or mitochondrial clearance/quality control has been tested as possible therapeutic strategies in different MD models (Viscomi et al, 2011; Johnson et al, 2013; Cerutti et al, 2014; Civiletto et al, 2015, 2018). In spite of initial promising data, the latter approaches failed to be effective across different MD models [reviewed in Lightowlers et al (2015); Viscomi et al (2015)]. We hypothesize that a synergic and fine modulation of mitochondrial biogenesis and clearance pathways is necessary to ensure a more efficient neuroprotective effect.

MicroRNAs (miRNAs) are fundamental fine regulators of gene expression and represent promising therapeutic tools due to their capability of simultaneously modulating multiple molecular pathways involved in disease pathogenesis and progression. Modulation of miRNAs has been applied, with therapeutic purposes, to different disorders and has reached preclinical and clinical stages in specific instances (Broderick & Zamore, 2011; Janssen et al, 2013; Ling et al, 2013; Christopher et al, 2016). miRNAs play a key role in neuron survival, and accumulating evidence indicates that alterations of miRNA‐regulated gene networks increase the risk of neurodegenerative disorders (Hebert & De Strooper, 2009). miR‐181a and miR‐181b (miR‐181a/b) belong to a family of miRNAs highly expressed in different regions of brain and retina (Boudreau et al, 2014; Karali et al, 2016) and were recently reported to target genes involved in mitochondrial‐dependent cell death (Ouyang et al, 2012; Hutchison et al, 2013; Rodriguez‐Ortiz et al, 2014) and autophagy (He et al, 2013; Tekirdag et al, 2013; Cheng et al, 2016).

Here, we show that miR‐181a/b are involved in global regulation of mitochondrial function by controlling a group of genes involved in mitochondrial biogenesis and function, and redox balance. We demonstrate that downregulation of these two miRNAs protects retinal neurons from mitochondrial dysfunction, and ameliorates the phenotype of three different MD animal models with ocular involvement, indicating that miR‐181a/b could be new therapeutic targets for MDs.

Results

miR‐181a/b control mitochondrial turnover

Bioinformatic search (Gennarino et al, 2012) allowed us to identify PPARGC1A and NRF1, master regulators of mitochondrial biogenesis (Wu et al, 1999; Finck & Kelly, 2006), COX11 and COQ10B, involved in mitochondrial respiratory chain (MRC) assembly (Carr et al, 2002; Desbats et al, 2015), and PRDX3, an important mitochondrial ROS scavenger (Wonsey et al, 2002), as putative miR‐181a/b target genes. Interestingly, quantitative real‐time PCR (qPCR) analysis indicated increased levels of all of the above‐mentioned transcripts in SH‐SY5Y human neuroblastoma cells following miR‐181a/b silencing (Fig 1A). To validate the newly predicted miR‐181a/b targets, 3′‐UTRs of each human gene (PPARGC1A, NRF1, COX11, COQ10B, and PRDX3), including the predicted miR‐181 target site, were cloned in the pGL3‐TK‐luciferase plasmid, downstream the coding region of the luciferase reporter gene. We then tested the ability of transfected mimic‐miR‐181 to affect luciferase activity. The presence of the 3′‐UTR sequence of the analyzed genes inhibited luciferase activity in response to mimic‐miR‐181 (Fig 1B). In addition, point mutations in the miR‐181a/b binding site in the 3′‐UTR of each gene abolished luciferase repression, demonstrating that these miRNAs directly and specifically target NRF1, COX11, COQ10B, and PRDX3 (Fig 1B). Direct targeting of PPARGC1A was not validated (Fig 1B), indicating that the upregulation observed by qPCR after miR‐181a/b silencing (Fig 1A) could be the result of an indirect effect.

Figure 1

miR‐181a/b silencing increases mitochondrial biogenesis and mitophagy and protects retinal neurons from FCCP‐induced cell death

Data information: P‐values were calculated by one‐tailed Student's t‐test in (A–C, E and F), by two‐tailed Student's t‐test in (D), and by two‐way ANOVA with post hoc Tukey's analysis in (G); error bars are SEM.Source data are available online for this figure.

qPCR reveals that miR‐181a/b silencing leads to upregulation of miR‐181a/b predicted targets in SH‐SY5Y cells. N = 3 independent experiments.

miR‐181‐mimic transfection specifically inhibits luciferase activity of constructs containing WT 3′‐UTR predicted target sequences. Point mutations (mut) in miR‐181a/b binding sites abolish luciferase repression in all cases apart from PPARGC1A. Data are normalized to negative mimic transfection (dashed line). N = 6 independent experiments.

qPCR reveals upregulation of miR‐181a/b targets in the eyes of miR‐181a/b‐1 −/− versus miR‐181a/b‐1 +/+ animals. n ≥ 5 animals/genotype.

miR‐181a/b‐1 −/− mice show increased mtDNA content versus miR‐181a/b‐1 +/+ mice as measured by qPCR. N = 4 animals/genotype.

WB analysis (left panel) reveals increased levels of mitochondrial proteins in the eyes of miR‐181a/b‐1 −/− (‐/‐) versus miR‐181a/b‐1 +/+ (+/+) mice (quantified in the right panel). Data are normalized to either p115 or Gapdh. N ≥ 3 animals/genotype. Please note that all compared bands from +/+ and −/− samples are from the same blots, which were cropped and shown split for the sake of data presentation clarity.

WB analysis on mitochondrial and cytosolic fractions (left panel) shows increased levels of p62 and Parkin in mitochondrial fraction from the eye of −/− versus +/+ mice (quantified in the right panel). Data are normalized to Ndufb8 or Gapdh for mitochondrial and cytosolic fractions, respectively. N = 3 animals/genotype.

Cell death analysis shows that miR‐181a/b silencing protects SH‐SY5Y cells from FCCP treatment. N ≥ 7 independent experiments.

Based on the above results, we reasoned that the downregulation of miR‐181a/b could stimulate mitochondrial biogenesis and decided to test this hypothesis in vivo. In mammals, miR‐181a and miR‐181b are organized in two clusters, namely miR‐181a/b‐1 and miR‐181a/b‐2, which are localized to different genomic loci. The mature forms of miR‐181a‐1 and miR‐181a‐2, as well as those of miR‐181b‐1 and miR‐181b‐2, display identical sequences. Furthermore, both miR‐181a and miR‐181b contain the same “seed” sequence (Ji et al, 2009), i.e., the region that is believed to play the most important role in target recognition (Bartel, 2009). We chose to analyze a mouse model harboring a targeted deletion of the miR‐181a/b‐1 cluster (Henao‐Mejia et al, 2013). This cluster accounts for most of the expression of mature miR‐181a/b in the retina, as demonstrated by RNA in situ hybridization, TaqMan assays, and the increase in several previously validated miR‐181a/b targets, such as Bcl2, Mcl1, Atg5, Erk2, and Park2 (Ouyang et al, 2012; He et al, 2013; Hutchison et al, 2013; Tekirdag et al, 2013; Rodriguez‐Ortiz et al, 2014; Cheng et al, 2016) (Appendix Fig S1, Fig EV1A).

Figure EV1

miR‐181a/b‐1 deletion leads to increased autophagic flux

Data information: P‐values were calculated by one‐tailed Student's t‐test; error bars are SEM.Source data are available online for this figure.

qPCR analysis reveals increased levels of miR‐181a/b targets Bcl2, Mcl1, Atg5, Erk2, and Park2 in the eyes of miR‐181a/b‐1 −/− versus miR‐181a/b‐1 +/+ animals. N ≥ 5 animals/genotype.

WB analysis (left panel) of Lc3‐I/Lc3‐II and p62 on protein extracts from eyes of animal in fed and starved conditions reveals decreased levels (quantified in the right panel) of both proteins in miR‐181a/b‐1 −/− (−/−) versus miR‐181a/b‐1 +/+ (+/+) mice. N = 2 mice for each genotype and condition.

qPCR analysis reveals no changes in the Sqstm1 (p62) and Map1lc3b (LC3) transcript levels between miR‐181a/b‐1 −/− and miR‐181a/b‐1 +/+ mouse eyes in both fed and starved conditions. These data indicate that the decreased levels of the autophagy markers Lc3‐II and p62 observed by WB analysis are due to increased autophagic flux in miR‐181a/b‐1 −/− eyes. N = 3 for each genotype and condition.

Interestingly, by qPCR, we observed increased expression levels of Nrf1, Cox11, Coq10b Prdx3, and Ppargc1a in the eye of miR‐181a/b‐1 −/− mice (Fig 1C). Upregulation of Nrf1 and Ppargc1a indicates enhanced mitochondrial biogenesis (Wu et al, 1999; Finck & Kelly, 2006). In line with this observation, we detected an increase of mitochondrial DNA (mtDNA), as measured by qPCR (Fig 1D), and of the protein levels of MRC complex subunits (Ndufb11 and CoxIV), mitochondrial matrix (citrate synthase [Cs]), and mitochondrial membranes (Mfn2 and Tim23), as assessed by Western blot (WB) analysis in the eye of miR‐181a/b‐1 −/− mice (Fig 1E). Overall, these data demonstrate that miR‐181a/b inactivation stimulates mitochondrial biogenesis in the CNS.

It was previously reported that miR‐181a/b regulate in vitro the expression of Atg5 and Park2, which are key players in autophagy and mitophagy (Tekirdag et al, 2013; Cheng et al, 2016). Therefore, we investigated whether miR‐181a/b inactivation enhances mitophagy in the mouse eye. First, we observed, by qPCR assays, that the transcript levels of the Atg5 and Park2 genes were upregulated in miR‐181a/b‐1 −/− eyes (Fig EV1A). Moreover, we demonstrated, by WB, enhanced recruitment of Park2/Parkin and of the autophagic adaptor Sqstm1/p62 in mitochondrial fractions (Fig 1F), indicating an increase of mitophagy in miR‐181a/b‐1 −/− eyes. We also analyzed general autophagy by evaluating the levels of Sqstm1/p62 and of the autophagic marker Map1lc3b/Lc3‐II in total protein extracts. By WB, we observed decreased levels of these two proteins, which is more evident in starved conditions (Fig EV1B), without changes in the corresponding transcript levels (Fig EV1C), indicating an increase in the autophagic flux rate in miR‐181a/b‐1 −/− eyes.

Taken together, these results uncover an important role of miR‐181a/b in the regulation of mitochondrial turnover through the coordination of mitochondrial biogenesis and clearance in vivo.

miR‐181a/b inhibition protects neurons from cell death and ameliorates the phenotype of in vivo models of MLS syndrome

Increased mitochondrial biogenesis and clearance were previously shown to exert protective effects in mitochondrial dysfunction (Viscomi et al, 2011; Johnson et al, 2013; Cerutti et al, 2014; Civiletto et al, 2015, 2018). We therefore decided to verify whether miR‐181a/b inactivation could protect cells from mitochondrial damage. First, we verified whether miR‐181a/b downregulation protects SH‐SY5Y neuron‐like cells from FCCP, a potent OXPHOS uncoupler. In control cells, 6 hours (h) of FCCP treatment induced a significant increase in the extent of cell death. Interestingly, miR‐181a/b‐silenced cells did not display any differences in cell death after FCCP treatment (Fig 1G), indicating that miR‐181a/b silencing protects cells from mitochondrial damage.

Based on the above results, we decided to evaluate the neuroprotective effect of miR‐181a/b inactivation in in vivo models of MDs. Toward this goal, we examined the consequences of miR‐181a/b downregulation in two fish models for a rare inherited form of MD, the MLS syndrome. MLS is a neurodevelopmental disorder characterized by microphthalmia, brain abnormalities, and skin defects in heterozygous females and in utero lethality in hemizygous males (Indrieri & Franco, 2016). The disease is due to mutations in key players of the MRC, such as the holocytochrome c‐type synthase (HCCS), involved in complex III function (Bernard et al, 2003; Wimplinger et al, 2006; Indrieri et al, 2013), and COX7B, the 7B subunit of cytochrome c oxidase (MRC complex IV) (Indrieri et al, 2012). We previously generated two medakafish (Oryzias latipes) models of MLS by knocking down, using a Morpholino(MO)‐based approach, hccs or cox7B expression (Indrieri et al, 2012, 2013). Both models (hccs‐MO and cox7B‐MO) showed a severe microphthalmic and microcephalic phenotype due to increased cell death in the CNS (Indrieri et al, 2012, 2013, 2016). In medaka, the mature forms of miR‐181a and miR‐181b are perfectly conserved with respect to their mammalian counterparts, in terms of both sequence identity (100%) and pattern of expression in the retina and brain (Carrella et al, 2015). We found that MO‐mediated silencing of miR‐181a/b in medaka leads to increased levels of the majority of targets involved in mitochondrial biogenesis and function, and in autophagy (Fig EV2). Interestingly, downregulation of miR‐181a/b in either of the above‐mentioned MLS medaka models led to a notable reduction of cell death in the eye and brain, as shown by TUNEL and caspase activation assays (Fig 2A–C). Accordingly, miR‐181a/b downregulation resulted in full rescue of the disease phenotype in about 50% of both hccs (Fig 3A–C and M) and cox7B morphants (Fig 3G–I and N). Notably, MO‐mediated silencing of miR‐181a/b did not cause any obvious morphological alteration in the controls (Fig EV3A and B). These data show that the downregulation of miR‐181a/b ameliorates the phenotype in both MRC complex III and IV defective models, indicating that the protective effect of miR‐181a/b silencing is gene‐independent.

Figure EV2

miR‐181a/b knockdown in medaka leads to upregulation of genes involved in mitochondrial function and autophagy

qPCR carried out on total RNA extracted from whole miR‐181a/b‐MOs medaka embryos to analyze the transcript levels of the miR‐181a/b targets involved in mitochondrial‐dependent cell death (bcl2, mcl1), autophagy (atg5, erk2), and mitochondrial biogenesis and function (cox11, coq10b, prdx3) and of the indirect target ppargc1a. N = 3. P‐values were calculated by one‐tailed Student's t‐test; error bars are SEM.

Figure 2

miR‐181a/b inhibition counteracts cell death in MLS medakafish models

Data information: P‐values were calculated by analysis of deviance for negative binomial generalized linear model in (B) and by one‐way ANOVA with post hoc Tukey's analysis in (C); error bars are SEM.

TUNEL assays on eye and brain sections of stage (st)30 medakafish control and MO‐injected embryos reveal a decrease in the number of apoptotic cells in both hccs‐MO/miR‐181a/b‐MO‐ and cox7B‐MO/miR‐181a/b‐MO‐injected compared to hccs‐MO‐ and cox7B‐MO‐injected embryos. Scale bars are 20 μm.

TUNEL‐positive cells/eye. n ≥ 5 eyes for each model.

Caspase assays show restored levels of caspase‐3 and caspase‐9 activities in hccs‐MO/miR‐181a/b‐MO‐ and cox7B‐MO/miR‐181a/b‐MO‐injected embryos with respect to hccs‐MO‐ and cox7B‐MO‐injected embryos. n ≥ 5 embryos for each model.

Figure 3

miR‐181a/b downregulation ameliorates the phenotype of MLS medakafish models

Representative images of st30 medaka embryos injected with hccs‐MO (B) and cox7B‐MO (H) alone or co‐injected with miR‐181a/b‐MOs (C, I). Co‐injection of miR‐181a/b‐MOs rescues microphthalmia and microcephaly in both hccs‐MO and cox7B‐MO embryos. (D–F, J–L) hccs‐MO/miR‐181a/b‐MO‐ and cox7B‐MO/miR‐181a/b‐MO‐injected embryos were treated with Baf‐A1, PD98059, or HA14‐1. Baf‐A1 treatment counteracts the protective effect of miR‐181a/b downregulation in both hccs‐MO/miR‐181a/b‐MO and cox7B‐MO/miR‐181a/b‐MO embryos (D, J). PD98059 treatment does not interfere with the modulation of the MLS phenotype mediated by miR‐181a/b downregulation (E, K). HA14‐1 treatment counteracts the effect of miR‐181a/b downregulation in the hccs‐MO/miR‐181a/b‐MO model but not in the cox7B‐MO/miR‐181a/b‐MO embryos (F, L). Scale bars are 100 μm.

Percentage of embryos with or without MLS phenotype in the different conditions illustrated in (A–L) in hccs‐MO (M) and cox7B‐MO (N) embryos. N ≥ 300 embryos/conditions. P‐values were calculated by analysis of deviance for generalized linear model; error bars are SEM.

Figure EV3

Absence of abnormal phenotypes following MO‐mediated silencing of miR‐181a/b and drug administration

Embryos injected with miR‐181a/b‐MOs only (B) show no phenotypic differences compared to control embryos (A). Scale bars are 100 μm.

Bafilomycin A (Baf‐A1) (C), PD98059 (D), and HA14‐1 (E) do not induce any morphological alterations in control embryos at the concentrations used in this study. Scale bars are 100 μm.

The concentration of Bafilomycin A used for the study blocks autophagy as demonstrated by lc3‐I/II Wb analysis.

Source data are available online for this figure.

Current evidence supports the protective role of autophagy and mitophagy in MDs as well as in other neurodegenerative disorders associated with impairment of mitochondrial functions (Lightowlers et al, 2015; Viscomi et al, 2015; Civiletto et al, 2018). Since we showed that miR‐181a/b regulate mitophagy and autophagy in the eye, we tested whether the latter processes are implicated in the amelioration of the phenotype mediated by miR‐181a/b downregulation. For this purpose, we treated hccs‐MO/miR‐181a/b‐MO‐ and cox7B‐MO/miR‐181a/b‐MO‐injected embryos with Bafilomycin A1 (Baf‐A1), a general autophagy inhibitor, at 50 nM. This concentration is able to block autophagy without inducing any obvious morphological abnormality in control embryos (Fig EV3C and F). Interestingly, Baf‐A1 treatment abolished the protective effect of miR‐181a/b downregulation in a significant number of both hccs‐MO/miR‐181a/b‐MO‐ (Fig 3D and M) and cox7B‐MO/miR‐181a/b‐MO‐injected embryos (Fig 3J and N), indicating that increased autophagy/mitophagy contributes to the phenotype amelioration observed in MLS medaka models.

Since miR‐181a/b downregulation also increased the mRNA levels of erk2, a key component of the MAPK/ERK cascade, and of bcl2 and mcl1, members of the Bcl2 anti‐apoptotic family (Fig EV2), we also tested whether the latter two pathways were involved in the amelioration of the MLS phenotype. Therefore, we treated hccs‐MO/miR‐181a/b‐MO‐ and cox7B‐MO/miR‐181a/b‐MO‐injected embryos with PD98059, a selective inhibitor of the MAPK/ERK pathway (Alessi et al, 1995), or with HA14‐1, an inhibitor of Bcl‐2 proteins (Wang et al, 2000). PD98059 and HA14‐1 were used at concentrations that did not induce any morphological alterations in control embryos [(Carrella et al, 2015) and Fig EV3D and E]. PD98059 treatment did not have any effect on the extent of phenotypic rescue in neither MLS models (Fig 3E, K, M and N), suggesting that the MAPK pathway is not primarily involved in the protective effect exerted by miR‐181a/b downregulation in the analyzed models. On the other hand, HA14‐1 significantly reduced the extent of phenotypic rescue in hccs‐MO/miR‐181a/b‐MO‐injected embryos, but not in cox7B‐MO/miR‐181a/b‐MO‐injected embryos (Fig 3F and L–N).

Taken together, these data suggest that increased autophagy/mitophagy are involved in the amelioration of the MLS phenotype in both models, indicating a primary role for this pathway in miR‐181a/b downregulation‐mediated neuronal protection. Moreover, increased levels of bcl2 play a role in the amelioration of hccs‐defective embryos only, indicating that the contribution of the Bcl2‐mediated apoptotic pathway may vary depending on the specific mitochondrial defect.

miR‐181a/b inactivation ameliorates the phenotype of a drug‐induced mouse model of LHON

To validate the efficacy of miR‐181a/b inactivation in mammalian models of MDs as well, we first exploited a drug‐induced mouse model of LHON, a non‐syndromic form of mitochondrial optic neuropathy characterized by degeneration of retinal ganglion cells (RGCs) that leads to loss of central vision. LHON represents one of the most frequent forms of MDs with a prevalence of 1:30,000 individual with an onset occurring between the ages of 15 and 35 (Carelli et al, 2004). Similar to other MDs, LHON is characterized by high genetic heterogeneity being caused by mutations in multiple genes (Carelli et al, 2004; Meyerson et al, 2015). In about 95% of LHON patients, mutations are located in the mitochondrially encoded ND1, ND4, or ND6 genes, which encode complex I subunits (Carelli et al, 2004; Meyerson et al, 2015). Intravitreal injection of rotenone, an MRC complex I inhibitor, leads to the damage of RGCs, and injected mice are considered a reliable drug‐induced LHON model (Carelli et al, 2013). In order to assess whether the genetic inactivation of miR‐181a/b exerts a protective effect against rotenone‐induced LHON, miR‐181a/b‐1 +/+ and miR‐181a/b‐1 −/− mice were intravitreally injected with rotenone unilaterally. As internal controls, contralateral eyes were injected with DMSO. As previously reported (Heitz et al, 2012), miR‐181a/b‐1 +/+ control mice injected with rotenone displayed a notable reduction in the number of RGCs (Fig 4A and B). In contrast, no difference was observed in the number of RGCs between rotenone‐ and DMSO‐injected retina in miR‐181a/b‐1 −/− mice, both 1 and 2 weeks after injection (Fig 4A and B). Interestingly, rotenone‐injected miR‐181a/b‐1 −/− mice also showed preserved MRC complex I activity in RGCs, as demonstrated by the staining of NADH dehydrogenase activity, with respect to rotenone‐injected miR‐181a/b‐1 +/+ mice (Fig 4C). Accordingly, CoxIV immunofluorescence analysis showed that miR‐181a/b‐1 depletion protects mitochondrial network integrity from rotenone‐induced damage (Fig 4D). Finally, we also tested retinal function by optokinetic response (OKR). For this analysis, miR‐181a/b‐1 +/+ and miR‐181a/b‐1 −/− mice were bilaterally injected with rotenone or DMSO. Interestingly, miR‐181a/b‐1 +/+ mice showed a severe decrease in visual performance starting 1 week after rotenone administration, whereas miR‐181a/b‐1 −/− mice were significantly protected against rotenone‐induced toxicity (Fig 4E). Ablation of miR‐181a/b‐1 rescued visual dysfunction even 2 weeks after rotenone injection (Fig 4E). Importantly, miR‐181a/b‐1 ablation did not cause per se neither abnormalities in retinal morphology and RGC number [as shown by immunofluorescence analysis with different retinal cell markers (Fig EV4A–J and M)] nor alteration of retinal function [as assessed by electroretinogram (ERG) analysis (Fig EV4K and L)]. Altogether, these data indicate that miR‐181a/b silencing exerts a protective effect in a drug‐induced mammalian model of LHON and therefore represents a potential therapeutic strategy for this condition.

Figure 4

Inactivation of miR‐181a/b protects RGCs from cell death in a rotenone‐induced mouse model of LHON syndrome

Data information: P‐values were calculated by one‐way ANOVA with post hoc Tukey's analysis; error bars are SEM.

Immunofluorescence analysis with anti‐NeuN antibody in the retina of miR‐181a/b‐1 +/+ and miR‐181a/b‐1 −/− mice intravitreally injected with rotenone or DMSO. RGCs are preserved in miR‐181a/b‐1 −/− rotenone‐injected mice with respect to controls at both one and 2 weeks after injection. Scale bars are 50 μm.

Number of RGCs/mm (indicated as %) at 1 and 2 weeks post‐injection. N = 10.

NADH dehydrogenase histochemical reaction on retinal sections of miR‐181a/b‐1 +/+ and miR‐181a/b‐1 −/− mice intravitreally injected with rotenone or DMSO. At 1 week post‐injection, NADH dehydrogenase activity is lost in RGCs (GCL, areas within dashed lines) of miR‐181a/b‐1 +/+ rotenone‐injected eyes, while it is preserved in those of miR‐181a/b‐1 −/− rotenone‐injected eyes. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer. Scale bars are 50 μm.

Immunofluorescence analysis, with an anti‐CoxIV antibody, in the retina of miR‐181a/b‐1 +/+ and miR‐181a/b‐1 −/− mice, injected with either rotenone or DMSO, showed preserved mitochondria in miR‐181a/b‐1 −/− rotenone‐injected eyes at 1 week post‐injection. Dashed boxes indicate the area of magnifications shown in right panels. Scale bars are 10 μm.

Graphical representation of the results of the optokinetic tracking assays reported as fold change of cycles/degree. Visual acuity is preserved in miR‐181a/b‐1 −/− rotenone‐injected mice with respect to controls at both 1 and 2 weeks after injection. N = 10.

Figure EV4

miR‐181a/b‐1 deletion does not cause significant alterations of retina morphology and function in mouse

Data information: P‐value was calculated by two‐tailed Student's t‐test; error bars are SEM. ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer.

Immunofluorescence staining of different retinal markers [(A, B) Rhodopsin (Rhod) for rods; (C, D) Cone arrestin (C‐Arr) for cones; (E, F) Pax6 for RGCs and amacrine cells; (G, H) GS6 for Muller cells; (I, J) Syntaxin (Synt) for synaptic structures in the inner plexiform layer] reveals no abnormalities in cell layer organization in the retina of miR‐181a/b‐1 −/− with respect to control mice. Scale bars are 50 μm.

Electroretinographic analysis reveals no difference in a‐wave (K) and b‐wave (L) patterns between miR‐181a/b‐1 +/+ and miR‐181a/b‐1 −/− mice, indicating that miR‐181a/b ablation does not cause retinal dysfunction. N = 10 for each genotype.

Graphical representation of the number of RGCs/mm (indicated as %) shows no significant differences between miR‐181a/b‐1 +/+ and miR‐181a/b‐1 −/− mice. N = 10 for each genotype.

miR‐181a/b inactivation rescues the LHON phenotype in the Ndufs4 −/− mouse model

Rotenone‐injected mice are considered a reliable model of LHON (Carelli et al, 2013). However, the latter represents an acute model in which it is difficult to dissect the mechanism by which miR‐181a/b downregulation exerts its protective function. Therefore, we decided to exploit the Ndufs4 −/− mouse model (Kruse et al, 2008) to further characterize and validate the efficacy of miR‐181a/b inactivation in mammalian models of MDs. Homozygous mutations in NDUFS4 are responsible for one of the most severe forms of Leigh syndrome (LS), an inherited neurodegenerative condition, characterized by poor prognosis as patients typically die before 3 years of age. NDUFS4 is a nuclear DNA‐encoded MRC complex I subunit located in the NADH dehydrogenase domain of the complex, responsible for its catalytic activity (Breuer et al, 2013; Stroud et al, 2016; Zhu et al, 2016). Mice with homozygous Ndufs4 inactivation develop a severe encephalopathy resembling LS that results in death between 50 and 60 postnatal (P) days (Kruse et al, 2008; Breuer et al, 2013). In addition, these mice also show a LHON phenotype characterized by RGC dysfunction starting from P16 to P20 and leading to a detectable RGC death after P42 (Yu et al, 2015; Song et al, 2017), which makes the Ndufs4 knockout line a good model to study the effect of miR‐181a/b inactivation in LHON. By crossing miR‐181a/b‐1 −/− with Ndufs4 −/− mice, we obtained Ndufs4 −/−/miR‐181a/b‐1 −/− animals that we exploited to study the effect of miR‐181a/b inactivation on the retinal phenotype. As previously reported (Yu et al, 2015), Ndufs4 −/− mice showed a significant reduction in the number of RGCs at P55, as confirmed by NeuN‐positive cell count (Fig 5A, B and E). This reduction was rescued in Ndufs4 −/−/miR‐181a/b‐1 −/− retinas (Fig 5C and E), in which the number of NeuN‐positive cells was indistinguishable from both WT and miR‐181a/b‐1 −/− mice (Fig 5D and E). Moreover, the OKR test showed a significant amelioration of visual acuity in Ndufs4 −/−/miR‐181a/b‐1 −/− compared to Ndufs4 −/− mice (Fig 5F) at P25, indicating a remarkable effect of miR‐181a/b downregulation on both RGC dysfunction and death. Finally, we also analyzed the ERG response, previously reported to be altered in Ndufs4 −/− mice with the absence of b‐wave response (Kruse et al, 2008; Yu et al, 2015). We observed a significant amelioration of the b‐wave (both in scotopic and in photopic ranges) in Ndufs4 −/−/miR‐181a/b‐1 −/− with respect to Ndufs4 −/− mice (Fig 5G and H), indicating a significant improvement of retinal response to light stimuli at P30. Altogether, these results indicate that in the Ndufs4 −/− model miR‐181a/b downregulation preserves RGCs from death and ameliorates the visual function before the miRNAs exert their activity on RGC death.

Figure 5

miR‐181a/b depletion protects RGCs and ameliorates visual function in Ndufs4 −/− mice

Data information: P‐values were calculated by one‐way ANOVA with post hoc Tukey's analysis in (E and F) and by two‐way ANOVA repeated measures with post hoc analysis in (G). P‐values in (G) refer to 20 log cd.s/m2 and photopic points. P‐values for the other points are reported in Appendix Table S1; error bars are SEM.

Immunofluorescence analysis with anti‐NeuN antibody in the retina of WT, Ndufs4 −/−, Ndufs4 −/−/miR‐181a/b‐1 −/−, and miR‐181a/b‐1 −/− mice. RGCs are preserved in Ndufs4 −/−/miR‐181a/b‐1 −/− with respect to Ndufs4 −/− mice. Scale bars are 50μm.

Number of RGCs/mm (%). N = 5.

Graphical representation of the results of the optokinetic tracking assays reported as fold change of cycles/degree. Visual acuity is preserved in Ndufs4 −/−/miR‐181a/b‐1 −/− with respect to Ndufs4 −/− mice. N ≥ 8.

Electroretinographic analysis reveals amelioration of b‐wave patterns in Ndufs4 −/−/miR‐181a/b‐1 −/− with respect to Ndufs4 −/−mice. N ≥ 5.

To gain insight into the molecular mechanisms underlying the amelioration of the miR‐181a/b inhibition‐induced Ndufs4 −/− retinal phenotype, we analyzed the expression levels of miR‐181a/b targets in Ndufs4 −/−/miR‐181a/b‐1 −/− mice and we observed a significant increase in the Nrf1 and Park2 transcripts as compared to Ndufs4 −/− (Fig EV5A). As previously mentioned, Park2 and Nrf1 are important players in the control of, respectively, mitophagy (Narendra et al, 2008) and mitochondrial biogenesis (Wu et al, 1999; Finck & Kelly, 2006), i.e., pathways that we showed to be upregulated in miR‐181a/b‐1 −/− versus WT mice (Fig 1). We therefore hypothesized that the activation of these pathways significantly contributes to the phenotypic improvement observed in Ndufs4 −/−/miR‐181a/b‐1 −/− mice. Notably, the levels of Parkin and p62 were increased in mitochondrial fractions from Ndufs4 −/−/miR‐181a/b‐1 −/− with respect to Ndufs4 −/− mice (Fig EV5C and D) indicating enhanced mitophagy activation in response to miR‐181a/b depletion. Electron microscopy (EM) studies showed lack of accumulation of autophagosomes/autolysosomes containing undigested material excluding a possible mitophagy block at the level of autophagosome recruitment or autophagosome/lysosome fusion (Fig 6B). Moreover, consistent with Nrf1 upregulation, we showed by EM increased number of mitochondria at both P30 and P55 in RGCs of Ndufs4 −/−/miR‐181a/b‐1 −/− with respect to Ndufs4 −/− mice (Fig 6A–C). The augmented mitochondrial biogenesis was also confirmed by increase of mtDNA in Ndufs4 −/−/miR‐181a/b‐1 −/− versus Ndufs4 −/− mice, as measured by qPCR (Fig 6D). Interestingly, the increase in number of mitochondria in RGCs of Ndufs4 −/−/miR‐181a/b‐1 −/− mice was accompanied by a notable amelioration of their morphological structure, such as increased number of cristae and more electron‐dense matrix, with respect to Ndufs4 −/− mice (Fig 6A and B). Notably, NRF1 downregulation significantly abolished the miR‐181a/b‐mediated rescue in SH‐SY5Y cells treated with FCCP, indicating that indeed NRF1 is implicated in the neuroprotective effect exerted by miR‐181a/b downregulation (Fig EV5B). Finally, we also analyzed the biochemical activity of MRC complexes I, II, and IV. The latter analyses showed an increase in complex I and II activities, and a similar trend for complex IV activity, normalized to the activity of the Krebs cycle enzyme citrate synthase in the Ndufs4 −/−/miR‐181a/b‐1 −/− mice compared with the Ndufs4 −/− mice at P30 (Fig 6E), indicating that miR‐181a/b downregulation is partially able to counteract the complex I biochemical defect of the MRC in the Ndufs4 −/− mouse.

Figure EV5

miR‐181a/b depletion increases Nrf1 and Park2 transcript levels and enhances mitophagy in Ndufs4 −/− retina

Data information: P‐values were calculated by one‐tailed Student's t‐test in (A), one‐way ANOVA with post hoc analysis in (B) and one‐tailed Student's t‐test in (D). Error bars are SEM.Source data are available online for this figure.

qPCR reveals upregulation of the miR‐181a/b targets Nrf1 and Park2 in the eyes of Ndufs4 −/−/miR‐181a/b‐1 −/− versus Ndufs4 −/− animals. N = 4 animals/genotype.

Cell death analysis shows that NRF1 downregulation abolishes the miR‐181a/b‐mediated protection in SH‐SY5Y cells treated with FCCP. N = 4.

WB analysis on mitochondrial fractions shows increased levels of p62 and Parkin (quantified in D) in mitochondrial fraction from the eye of Ndufs4 −/−/miR‐181a/b‐1 −/− versus Ndufs4 −/− mice. Data are normalized to citrate synthase (Cs). N = 3 animals/genotype.

Figure 6

miR‐181a/b depletion enhances mitochondrial biogenesis Ndufs4 −/− mice

Data information: P‐values were calculated by two‐tailed Student's t‐test; error bars are SEM.

Electron microscopy analysis shows amelioration of mitochondrial morphology and increase of mitochondrial number in Ndufs4 −/−/miR‐181a/b‐1 −/− versus Ndufs4 −/− mice. Scale bars are 1 μm. The quantitative increase in mitochondria is reported in (C) as number of mitochondria/RGC, at p30 and p55. N ≥ 2 animals/genotype.

Ndufs4 −/−/miR‐181a/b‐1 −/− mice show increased mtDNA content versus Ndufs4 −/− mice as measured by qPCR. N ≥ 4 animals/genotype.

Biochemical activity of MRC complexes I, II, and IV normalized by the percentage of citrate synthase (CS) activity in Ndufs4 −/−/miR‐181a/b‐1 −/− versus Ndufs4 −/− mice. N ≥ 3 animals/genotype.

Altogether, these data indicate that miR‐181a/b downregulation exerts its protective effect mainly through the simultaneous stimulation of mitophagy and mitochondrial biogenesis, which leads to enhancement of mitochondrial turnover in the retina that ultimately results in amelioration of the mitochondrial morphological and biochemical phenotypes (Fig 7).

Figure 7

Proposed model for the mechanism of action of miR‐181a/b inactivation in MDs

In MDs, a portion of unhealthy mitochondria is degraded by mitophagy resulting in a decreased number of mitochondria, which is not efficiently balanced by mitochondrial biogenesis. Inactivation of miR‐181a/b leads to the simultaneous increase in mitophagy and mitochondrial biogenesis (red arrows), inducing a significant enhancement of mitochondrial turnover and activity and ensuring a more efficient protection from cell death.

Discussion

In this report, we provide a proof of principle for the genetic inactivation or downregulation of miR‐181a/b exerting a protective effect from mitochondrial‐mediated neurodegeneration in both in vitro and in vivo MD models, regardless of the underlying etiopathogenetic events thus highlighting these two miRNAs as potential, gene‐independent, therapeutic targets for MDs characterized by neuronal degeneration.

We found that miR‐181a/b control genes implicated in mitochondrial biogenesis, functionality, and antioxidant response. In particular, our data demonstrate that these miRNAs control mitochondrial biogenesis in the retina by direct targeting of Nrf1 and, together with the already known effect on general autophagy and mitophagy (Ouyang et al, 2012; Tekirdag et al, 2013; Cheng et al, 2016), uncover a fundamental role of miR‐181a/b in the control of mitochondrial turnover in the CNS. Overall, these data point to miR‐181a/b as possible hubs in the gene pathways underlying mitochondrial homeostasis in both physiological and pathological conditions. We propose that the broad and beneficial impact of miR‐181a/b downregulation on MD models relies on the ability of these two miRNAs to simultaneously and finely modulate the above‐mentioned mitochondrial‐related pathways, as opposed to more drastic and individual modulations of each of them.

Increased autophagy and mitophagy have been shown to exert protective effects in MDs and in other neurodegenerative diseases (Decressac et al, 2013; Martinez‐Vicente, 2017). Of note, Parkin is one of the main players in the clearance of damaged mitochondria via the autophagy pathway (Geisler et al, 2010; Vives‐Bauza et al, 2010). Moreover, treatment with the mTOR inhibitor rapamycin, which activates autophagy, significantly delayed both neurodegeneration progression and the fatal outcome of the Ndufs4 −/− mouse (Johnson et al, 2013) and of a mouse model of complex IV deficiency (Civiletto et al, 2018), although a general consensus on the mechanisms underlying the effects of rapamycin on primary mitochondrial dysfunction is still lacking. Here, we show that downregulation of miR‐181a/b leads to increased autophagy and mitophagy in the retina. In line with these data, inhibition of general autophagy in both hccs/miR‐181a/b‐MOs‐ and cox7B/miR‐181a/b‐MOs‐injected embryos reduces the neuroprotective effect of miR‐181a/b downregulation on the MLS phenotype, indicating that increased autophagy/mitophagy significantly contributes to the phenotypic rescue in both models. Moreover, analysis of Ndufs4 −/−/miR‐181a/b‐1 −/− mice shows increased mitophagy in the retina, confirming the importance of the activation of this process in the amelioration of MD neurodegenerative phenotypes in response to miR‐181a/b downregulation. In addition, Parkin overexpression may be beneficial for neuronal cell survival by acting on other pathways such as mitochondrial dynamics, through enhancement of proteasomal activity (Khandelwal & Moussa, 2010; Rana et al, 2013) and transcriptional control of key life/death genes such as p53 (Alves da Costa et al, 2018).

Although the removal of dysfunctional mitochondria would limit the risk of apoptosis, an uncontrolled activation of mitophagy without an appropriate compensatory mitochondrial biogenesis may contribute to further mitochondrial dysfunction resulting in loss of mitochondrial mass, energetic collapse, and cell death (Zhu et al, 2013; Villanueva Paz et al, 2016). Furthermore, enhanced mitochondrial biogenesis could be per se protective in MDs. Variability in mitochondrial DNA content and mitochondrial biogenesis is indeed associated with incomplete penetrance in non‐manifesting carriers of LHON (Giordano et al, 2014), and increased biogenesis was found to improve the phenotype of different in vitro and in vivo MD models (Komen & Thorburn, 2014). However, some discrepant and contradictory results obtained using compounds that trigger mitochondrial biogenesis (including bezafibrate, resveratrol, and AICAR) (Komen & Thorburn, 2014) clearly indicate that additional work is needed to effectively exploit modulation of mitochondrial biogenesis to increase mitochondrial functionality in patients. The tight balance between mitophagy and mitochondrial biogenesis is emerging as a critical aspect in the determination of cell viability in diseases characterized by mitochondrial dysfunction (Zhu et al, 2013). Here, we show that miR‐181a/b can simultaneously control both processes. Concomitantly to the increased mitophagy observed following miR‐181a/b inactivation, we detected increased levels of several mitochondrial proteins and of mtDNA, indicating enhanced mitochondrial biogenesis. Accordingly, miR‐181a/b‐depleted animals showed enhanced MRC complex I activity and preserved mitochondrial integrity in the rotenone‐induced LHON model. Interestingly, consistent with the higher levels of Nrf1, the Ndufs4 −/−/miR‐181a/b‐1 −/− model showed a significantly increased number of mitochondria in the retina compared to Ndufs4 −/− mice indicating enhanced mitochondrial biogenesis. Moreover, Ndufs4 −/−/miR‐181a/b‐1 −/− mice show amelioration of mitochondrial morphological structure with respect to Ndufs4 −/− mice. In addition, analysis of the biochemical activity of MRC complexes showed an increase in the activity of complex I in Ndufs4 −/−/miR‐181a/b‐1 −/− compared with Ndufs4 −/− mice. This increase, although significant, is still very far from compensating the profound complex I deficiency shown by the Ndufs4 −/− mice. However, in Ndufs4 −/−/miR‐181a/b‐1 −/− animals, we have observed a concomitant increase in the activity of complex II, which contributes to the amelioration of the phenotype by increasing the electron flux through complexes III and IV. We hypothesize that this enhancement of complex activities is mediated by miR‐181a/b action on degradation of the most dysfunctional mitochondria and enhanced mitochondrial biogenesis (proposed in Fig 7), which translates in higher MRC activities.

We cannot exclude that other mitochondrial‐related gene pathways may contribute to the miR‐181a/b downregulation effect. For example, another important pathway modulated by miR‐181a/b is the mitochondrial‐dependent cell death in which the Bcl‐2 protein family has a key regulatory role (Rasola & Bernardi, 2007; Tait & Green, 2010). Although the contribution of apoptosis to the pathogenesis of primary MDs is not univocally established, increasing the levels of anti‐apoptotic Bcl‐2 proteins may represent an effective strategy to prolong cell survival and slow down disease progression. Notably, overexpression of the Bcl‐2 family protein Bcl‐xL counteracted increased cell death and ameliorated the phenotype of the hccs‐deficient MLS medaka model (Indrieri et al, 2013). Here, we show that miR‐181a/b inhibition leads to increased levels of Bcl‐2 family members, such as Bcl‐2 and Mcl‐1, and rescues the phenotype of both MLS models tested. Moreover, HA14‐1, a potent inhibitor of Bcl‐2 proteins (Wang et al, 2000), significantly abolishes miR‐181a/b‐mediated amelioration of the MLS phenotype in the hccs‐defective, but not in the cox7B‐defective model. These data indicate that a single pathway, e.g., mitochondrial‐dependent cell death, and its modulation, may have different effects on highly related forms of MDs, depending on the underlying genetic defect. In addition, our data indicate that the amelioration of the phenotype observed in Ndufs4 −/− mice upon miR‐181a/b downregulation preceded the effect on RGC death which is detectable only after P42. We showed that miR‐181a/b depletion improves mitochondrial number, morphology, and MRC complex activities and rescues the visual function before P30, thus indicating that the beneficial effects are mainly due to mitochondrial turnover enhancement rather than activation of the Bcl2‐mediated cell death pathway. This observation strengthens our hypothesis that the simultaneous modulation of more than one pathway represents a more effective strategy to treat MDs.

Nevertheless, additional efforts are necessary for the exhaustive dissection of the molecular mechanisms underlying the remarkable efficacy of miR‐181a/b downregulation in MDs, and to evaluate its safety. It is to be noted that miR‐181a/b‐1 −/− mice show normal life‐span (Henao‐Mejia et al, 2013) and no obvious abnormalities in the eyes (Fig EV4), which provides initial support to the safety of the downregulation of these two miRNAs.

MDs are genetically heterogeneous disorders. Although each individual disorder is rare, collectively the total prevalence of adult MDs is estimated to be of approximately 1 in 4,300 representing one of the most prevalent groups of inherited neurological diseases (Gorman et al, 2015). In this study, we demonstrate that downregulation of miR‐181a/b exerts a neuroprotective effect in multiple models of MDs. Our results are of particular interest for genetically heterogeneous conditions, such as LHON and LS for which a gene‐independent therapeutic strategy is highly desirable. In summary, we believe that the modulation of miR‐181a/b may represent effective gene‐independent therapeutic targets for MDs characterized by neuronal degeneration.

Materials and Methods

Cell lines

HeLa and SH‐SY5Y cells were obtained from ATCC and maintained in culture with Dulbecco's modified Eagle's medium (Gibco) or a 1:1 mixture of Dulbecco's modified Eagle's medium and Nutrient Mixture F‐12, supplemented with 10% FBS (EuroClone) and 1% penicillin/streptomycin, respectively, as suggested by the vendor.

Animal studies

Ethical statement: All studies on fish and mice were conducted in accordance with ARRIVE guidelines for animal research and approved by the Italian Ministry of Health, Department of Public Health, Animal Health, Nutrition and Food Safety, in accordance with the law on animal experimentation (article 7; D.L. 116/92; protocol number: 389/2015‐PR and 575/2017‐PR). All animal treatments were reviewed and approved in advance by the Institutional Ethics Committee at the Telethon Institute of Genetics and Medicine (Pozzuoli, Italy).

Medakafish (O. latipes) of the cab strain were maintained in the in‐house TIGEM facility (28°C on a 14/10 h light/dark cycle). The C57BL/6‐miR‐181a/b−/− mice were provided by Prof. Richard A. Flavell (Department of Immunobiology, Yale University School of Medicine, New Haven, Connecticut, USA). The B6.129S4‐Ndufs4 +/− strain described in Kruse et al (2008) was obtained from The Jackson Laboratory and backcrossed with WT C57BL/6 mice. Mice (male and females, 2–8 weeks old) were housed at the TIGEM animal facility (22°C on a 12/12‐h light/dark cycle, humidity‐controlled) and kept on an ad libitum normal diet (VRF1, Special Diet Services) and free access to tap water.

Ndufs4 +/− mice were crossed with miR‐181a/b‐1 −/−  animals to generate Ndufs4 +/−/miR‐181a/b‐1 +/− mice. To obtain background homogeneity, Ndufs4 +/−/miR‐181a/b‐1 +/− mice were then backcrossed with miR‐181a/b‐1 +/+ or miR‐181a/b‐1 −/− derived from the same littermates. Ndufs4 +/−/miR‐181a/b‐1 +/+ were crossed with Ndufs4 +/−/miR‐181a/b‐1 +/+ mice to obtain Ndufs4 −/−/miR‐181a/b‐1 +/+, and Ndufs4 +/−/miR‐181a/b‐1 −/− were crossed with Ndufs4 +/−/miR‐181a/b‐1 −/− to obtain Ndufs4 −/−/miR‐181a/b‐1 −/−. WT (Ndufs4 +/+/miR‐181a/b‐1 +/+) controls were age‐matched and obtained both from littermates and separate litters that were co‐housed.

Luciferase assays

Human 3′‐UTR sequences containing miR‐181a/b predicted binding sites were amplified by PCR and inserted in the pGL3‐tk‐luciferase vector. Constructs containing mutagenized miR‐181a/b binding sites were obtained using the QuikChange Site‐Directed Mutagenesis Kit (Stratagene).

Primer sequences are reported in Appendix Table S2. Plasmids were transfected in HeLa cells (PolyFect reagent, Qiagen). After 7 h, cells were transfected with 100 nM of miRIDIAN negative mimic or mimic‐miR‐181 (Dharmacon) using Interferin (Polyplus). After 24 h, luciferase activities were quantified using Dual‐luciferase Reporter Assay (Promega).

RNA extractions

Tissue samples were processed in QIAzol lysis reagent (Qiagen). Total RNA from tissue and cell samples was extracted using the RNeasy Extraction Kit (Qiagen), according to manufacturer's instructions.

Quantitative real‐time PCR

For qPCR experiments, cDNAs were generated using QuantiTect Reverse Transcription Kit (Qiagen), according to manufacturer's instructions.

Primers for qPCRs were designed to span two different exons to avoid genomic DNA amplification using in silico tools (www.basic.northwestern.edu/biotools/oligocalc.html) to predict their melting temperature (T m) and to avoid the possibility of self‐annealing or primer dimerization. The specificity of the designed primers was tested in silico using the BLAT or BLAST tool in Genome Browser (https://genome.ucsc.edu/) or Ensembl (http://www.ensembl.org/index.html). Primers were tested as described (Bustin et al, 2009). Primer sequences are reported in Appendix Table S3. Quantification data, obtained in qPCRs on cDNAs obtained from different treatments, are expressed in terms of cycle thresholds (Ct). The HPRT and GAPDH genes were used as endogenous reference controls for experiments. The Ct values were averaged for each in‐plate technical triplicate. The averaged Ct was normalized as difference in Ct values (ΔCt) between the analyzed mRNAs and each reference gene in each sample analyzed. The ΔCt values of each sample were then normalized with respect to the ΔCt values of the control (ΔΔCt). The variation was reported as fold change (2−ΔΔ). Each plate was performed in duplicate, and all the results are shown as means ± SEM of at least three independent biological assays.

SH‐SY5Y transfections and FCCP treatment

SH‐SY5Y cells were transfected with 100 nM of negative control miRNA inhibitor and miR‐181a and miR‐181b inhibitors (Dharmacon) using a lipid‐based transfection reagent (DharmaFECT, Dharmacon). Upon 72 h of transfection, SH‐SY5Y cells were treated with 25 μM FCCP. After 6 h of FCCP treatment, dead cells were stained with trypan blue and counted with the automated cell counter (LUNA, Logos Biosystems). For double transfection, SH‐SY5Y cells were transfected with 100 nM of negative control miRNA inhibitor or miR‐181a and miR‐181b inhibitors, and with 100 nM control siRNA or siNRF1 (Dharmacon) using Lipofectamine® RNAiMAX transfection reagent (Invitrogen). Upon 72 h of transfection, SH‐SY5Y cells were treated with 25 μM FCCP. After 6 h of FCCP treatment, dead cells were stained with BOBO™‐3 Iodide (Thermo Fisher Scientific) and counted with Operetta microplate imaging reader, the Acapella and Columbus image data processing system, and software (PerkinElmer).

Mature miRNA quantitative assay

For mature quantitative detection of miR‐181a and miR‐181b, we used hydrolysis probes (TaqMan, Applied Biosystem). The cDNAs for mature miRNA analysis were generated using TaqMan MicroRNA Reverse Transcription Kit with miRNA‐specific primers, according to manufacturer's instructions. The quantification data, obtained in TaqMan‐PCRs on TaqMan‐cDNAs from different treatments, are expressed in terms of cycle thresholds (Ct). A TaqMan probe for the RNA sno234 was used as endogenous control for the experiments. The Ct values were analyzed as described in the “Quantitative Real‐Time PCR” section. Each plate was performed in duplicate, and all results are shown as means ± SEM of three independent biological replicates.

RNA in situ hybridization (ISH)

Mouse eyes were fixed overnight in 4% PFA, cryoprotected with 30% sucrose, and embedded in OCT. Twenty‐micrometer cryosections were treated with 5 μg/ml proteinase K for 15 min. After washes with 2 mg/ml glycine and post‐fixation with 4% PFA/0.2% glutaraldehyde, sections were prehybridized with 50% formamide, 5× sodium saline citrate buffer (SSC) and citric acid to pH 6.1% sodium dodecyl sulfate (SDS), 500 μg/ml yeast tRNA, and 50 μg/ml heparin. The miRCURY detection miR‐181a and miR‐181b locked nucleic acid probes (Exiqon) were used at a final concentration of 30 nM. Probe hybridization was performed overnight at 42°C. Hybridized sections were washed with 50% formamide, 2 × SSC at the hybridization temperature. Sections were blocked for 1 h in MABT (100 mM maleic acid, 150 mM NaCl, and 0.1% Tween‐20) containing 1% blocking reagent (Roche) and 10% sheep serum and incubated with alkaline phosphatase (AP)‐labeled anti‐digoxigenin antibody (1:2,000; Roche) in MABT/1% blocking reagent overnight at 4°C. Sections were stained with NBT‐BCIP (Roche) and photographed under a Leica DM5000 microscope.

Quantitative mitochondrial DNA content analysis

SYBR Green qPCR was performed using primers for the MT‐CO1 gene (mtDNA) and the RNaseP (nuclear gene reference), as described (Viscomi et al, 2009). mtDNA relative copy number was calculated from threshold cycle value (ΔCt), and mtDNA copy number/cell was calculated as 2 × 2−ΔCt to account for the two copies of RNaseP in each nucleus.

Protein isolation and Western blotting (WB)

Tissue and cells samples for total protein extraction were homogenized in RIPA buffer with Protease Inhibitor Cocktail Tablet (Roche). Protein extract concentrations were determined using the Bio‐Rad protein assay (Bio‐Rad). A total of 30 μg protein from each sample was loaded in SDS–PAGE. Mitochondrial and cytosolic extracts were obtained from whole eyes after lens removal, as described (Comitato et al, 2014). Thirty micrograms of proteins from mitochondrial and cytosolic fractions from each sample was loaded in SDS–PAGE. For WB, gels were electroblotted onto PVDF filters (Millipore) and sequentially immunostained with the following primary antibodies: anti‐p62 (Abnova, clone 2C11, 1:1,000); anti‐LC3B (Novus, 1:500); anti‐Mfn2 (Abcam, ab56889, 1:1,000); anti‐citrate synthase (Abcam, ab9660, 1:1,000); anti‐p115 (Santa Cruz, sc‐48363, 1:3,000); anti‐Tim23 (Santa Cruz, sc‐13298, 1:250); anti‐Ndufb11 (Proteintech, 1:1,000); anti‐Ndufb8 (Abcam, Total OXPHOS Rodent WB Antibody Cocktail ab110413, 1:250); anti‐CoxIV (Cell Signaling 4844, 1:1,000); anti‐Parkin (Cell Signaling 4211, 1:500); anti‐Gapdh (Santa Cruz sc‐32233, 1:3,000).

Proteins of interest were detected with horseradish peroxidase‐conjugated goat anti‐mouse or anti‐rabbit IgG antibody (1:3,000, GE Healthcare) visualized with the Luminata Crescendo substrate (Millipore) or the Super Signal West Femto substrate (Thermo Scientific), according to manufacturer's protocol. WB images were acquired using the ChemiDoc‐It imaging system (UVP), and band intensity was calculated using the ImageJ software. The signals for each protein staining were quantified and then normalized for GAPDH, p115 or Ndufb8, CS (mitochondrial fractions) in the same sample (internal normalization). These normalized values were then compared to the values in the control sample. Only bands on the same blot were compared. The average of the normalized values from three different biological replicates is reported as the relative fold change.

Medakafish and morpholino

Medakafish were staged as described (Iwamatsu, 2004). Design, specificity, and inhibitory efficiency of MOs were previously described (Indrieri et al, 2012, 2013; Carrella et al, 2015). hccs‐MO, cox7B‐MO, miR‐181a‐MO, and miR‐181b‐MOs were injected into fertilized one‐cell medaka embryos at a concentration of 0.3, 0.1, 0.075, and 0.075 mM, respectively.

TUNEL assay

Whole‐mount TUNEL assay was performed on stage(st)30 medakafish embryos using In Situ Cell Death Detection Kit (Roche) as described (Indrieri et al, 2013). After staining, embryos were embedded in a mix of BSA/Gelatine and vibratome‐sectioned. Sections were analyzed with a Leica DM‐6000 microscope, and TUNEL‐positive cells were manually counted in the entire retina.

Caspase assays

Twenty embryos for sample were dechorionated at st32 and frozen in liquid nitrogen for at least 24 h. Caspase assays were performed on protein lysates using caspase‐3/7‐ and caspase‐9‐GLO luciferase reagent (Promega) as previously reported (Indrieri et al, 2013). The emitted luminescence signals were normalized for the protein concentration of each sample.

Drug treatments in medakafish

st24 embryos were dechorionated and incubated for 24 h in 50 nM Bafilomycin A (Sigma), 25 μM PD98059 (Cell Signaling), or 6 μM HA14‐1 (Sigma). Drugs were diluted in 3% DMSO/embryo medium. Medium was refreshed twice a day. Control embryos were grown in 3% DMSO/embryo medium.

Rotenone injections

Mice were injected intravitreally with 2 μl of rotenone (Sigma) 5 mM, as reported (Heitz et al, 2012). For histological analysis of RGCs, one murine eye was injected with rotenone and the contralateral eye with DMSO (internal control). For optokinetic tests, mice were bilaterally injected with rotenone 5 mM or DMSO.

Optokinetic tracking

Visual acuity of rotenone‐injected mice was assessed by using the optomotor system (OptoMotry; Cerebral Mechanics) as described (Chadderton et al, 2013).

Immunofluorescence and immunohistochemistry analysis

For immunofluorescence analysis, mouse eyes were fixed in 4% PFA, cryoprotected with 30% sucrose, embedded in OCT, and cryosectioned. Sections were permeabilized by boiling in sodium citrate buffer or by incubation with 1% NP‐40 for 15 min for cone‐arrestin immunostaining. The following primary antibodies were incubated overnight: anti‐Rhodopsin (Abcam, ab3267, 1:5,000); anti‐cone arrestin (Millipore, AB15282, 1:1,000); anti‐Pax6 (Covance, 42‐6600, 1:250); anti‐GS6 (Millipore, MAB 302, 1:100); anti‐Syntaxin (Sigma, S1172, 1:100); anti‐NeuN (Millipore, clone A60, 1:400); and anti‐CoxIV (Cell Signaling 4844, 1:400). Sections were then incubated with the Alexa Fluor secondary antibodies (1:1,000; Invitrogen). Sections were counterstained with DAPI (Vector Laboratories). RGCs stained using anti‐NeuN antibody were counted on eight different slides for each eye in the areas surrounding the optic nerve. Sections were photographed under a LSM710 Zeiss confocal microscopy.

Electroretinogram (ERG)

Mice were dark‐reared for 3 h and anesthetized. Flash electroretinograms (ERGs) were evoked by 10‐ms light flashes generated through a Ganzfeld stimulator (CSO, Costruzione Strumenti Oftalmici, Florence, Italy) and registered as previously described (Surace et al, 2005; Botta et al, 2016). ERGs and b‐wave thresholds were assessed using the following protocol. Eyes were stimulated with light flashes increasing from −5.2 to + 1.3 log cd*s/m2 (which correspond to 1 × 10−5.2 to 20.0 cd*s/m2) in scotopic conditions. For ERG analysis in scotopic conditions, the responses evoked by 11 stimuli (from −5.2 to + 1.3 log cd*s/m2) with an interval of 0.6 log unit were considered. To minimize the noise, three ERG responses were averaged at each 0.6 log unit stimulus from −5.2 to 0.0 log cd*s/m2, while one ERG response was considered for higher stimuli (from 0.0 to +1.3 log cd*s/m2). a‐ and b‐wave amplitudes recorded in scotopic conditions were plotted as a function of increasing light intensity (from −4 to + 1.3 log cd*s/m2). The photopic ERG was recorded after the scotopic session by stimulating the eye with ten 10‐ms flashes of 20.0 cd*s/m2 over a constant background illumination of 50 cd/m2.

Electron microscopy and mitochondrial count

Mice were deeply anesthetized and perfused with 1% glutaraldehyde and 2% PFA in 200 mM HEPES buffer, pH 7.3, through the heart. Eyes were removed and left for 1 h in the same fixative solution. Specimens of retina were post‐fixed in 1% osmium tetroxide, dehydrated, and embedded in epoxy resin. Retina samples were cut on ultra‐microtome Leica EM UC7 and collected on the single‐slot oval grids and analyzed with FEI electron microscope. Mitochondrial number/cell was determined using FEI software, by counting the number of mitochondria in RGCs (N ≥ 10 cells for each animal).

Mitochondrial enzymatic activity measurements

Mitochondria were isolated from the two eyes of one mouse as described (Comitato et al, 2014). The mitochondrial pellets were kept frozen at −80°C until the measurements were performed. Each pellet was resuspended in washing buffer (250 mM sucrose, 20 mM HEPES‐KOH, pH 7.4, 1 mM EDTA), and one cycle of freeze (in liquid N2) and thawing (at 37°C) was performed. The activities of MRC complex I (NADH‐CoQ1 oxidoreductase), complex II (succinate‐CoQ1‐DCIP oxidoreductase), and complex IV (cytochrome c oxidase) and of citrate synthase were measured spectrophotometrically in a 96‐well plate reader, as described (Tiranti et al, 1995; Kirby et al, 2007).

Problem

Mitochondrial dysfunction underlies the pathogenesis of a variety of human neurodegenerative disorders, including mitochondrial diseases (MDs), a heterogeneous group of devastating and often fatal disorders due to defective oxidative phosphorylation. The complexity and genetic heterogeneity of these disorders have so far prevented the development of effective therapeutic strategies. Therefore, the establishment of gene‐independent therapeutic strategies for MDs characterized by neuronal degeneration represents a priority.

Here, we identify two microRNAs, miR‐181a and miR‐181b, as new global regulators of mitochondrial turnover in the central nervous system. We demonstrate that their modulation ameliorates the disease phenotype in in vivo models of primary mitochondrial diseases, namely microphthalmia with linear skin lesions and Leber's hereditary optic neuropathy.

Impact

Altogether, our findings indicate that these microRNAs may represent novel gene‐independent therapeutic targets for mitochondrial diseases and other neurodegenerative disorder associated with mitochondrial dysfunction for which an effective therapy is not available to date.

Statistical analysis

Sample sizes were estimated on the basis of power analysis assuming significance when P < 0.05 and power ≥ 0.8, and on the basis of our previous experience. In general, for each experiment we used ≥ 3 animals per genotype, in order to obtain statistically suitable values. Animals/samples were randomly allocated in the different groups based on the appropriate genotype/conditions/treatments. Data were excluded from the analysis if an animal died during experiments. Because of the obvious clinical and morphological phenotypes of most of the animals, the study was not based on blinding of the investigators. The number of experimental replicates is indicated in each figure legend. In all experiments, significance of differences between groups was evaluated by one‐way or two‐way ANOVA with post hoc Tukey's analysis, one‐tailed or two‐tailed Student's t‐test, and analysis of deviance for generalized linear model or analysis of deviance for negative binomial generalized linear model as reported in figure legends. P < 0.05 was considered significant. Quantitative data are presented as the mean ± SEM (standard error of the mean) of at least three independent experiments.

Author contributions

AI and SC conceived the study, designed the experiments, analyzed the data, and wrote the manuscript. AI, SC, AR, AS, SBar, MP, YE, FMG, LC, and RT carried out experimental work. EM and EMS performed intravitreal injection and visual tests in mice. EF‐V and MZ contributed to the characterization of Ndufs4 mouse model. EDL and MZ provided expertise and feedback. JH‐M, AW, and RAF provided critical reagents for studies in the mouse. SBan and BF conceived the study, wrote the manuscript, and provided funding. All authors discussed the results and had the opportunity to comment on the manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.

Appendix

Click here for additional data file.

Expanded View Figures PDF

Click here for additional data file.

Source Data for Expanded View

Click here for additional data file.

Review Process File

Click here for additional data file.

Source Data for Figure 1

Click here for additional data file.