High content image analysis reveals function of miR-124 upstream of Vimentin in regulating motor neuron mitochondria

microRNAs (miRNAs) are critical for neuronal function and their dysregulation is repeatedly observed in neurodegenerative diseases. Here, we implemented high content image analysis for investigating the impact of several miRNAs in mouse primary motor neurons. This survey directed our attention to the neuron-specific miR-124, which controls axonal morphology. By performing next generation sequencing analysis and molecular studies, we characterized novel roles for miR-124 in control of mitochondria localization and function. We further demonstrated that the intermediate filament Vimentin is a key target of miR-124 in this system. Our data establishes a new pathway for control of mitochondria function in motor neurons, revealing the value of a neuron-specific miRNA gene as a mechanism for the re-shaping of otherwise ubiquitously-expressed intermediate filament network, upstream of mitochondria activity and cellular metabolism.

To uncover the underlying molecular mechanism responsible for miR-124 activity, we performed transcriptome profiling, using next generation sequencing (NGS). Total RNA was extracted from primary motor neurons and 3′ cDNA libraries were constructed and sequenced. Hierarchical clustering analysis of mRNA expression depicted a unique expression profile for neurons that overexpressed miR-124, which was distinguishable from cells transfected with scrambled control oligos (dendrogram of Pearson correlation coefficient, Fig. 2a).
Sylamer analysis 45 of 6500 expressed mRNAs, from neurons expressing scrambled control mimics or miR-124, uncovered two enriched motifs, which matched the miR-124 'seed' sequence. However, such enrichment was not evident for any other miRNA gene (Fig. 2b). We conclude that miR-124 overexpression had a widespread and specific impact on motor neuron mRNA expression profile.
Gene ontology (GO) analysis was carried out for approximately 1100 mRNAs that were significantly up-or downregulated, following miR-124 overexpression (corrected P-value < 0.05), using DAVID 46 . Intriguingly, many of the gene ontology terms highlighted the potential relevance of mitochondria-related structure or function in response to miR-124 (Table 1, Fig. 2c). We therefore further hypothesized that miR-124 plays a role in the regulation of mitochondrial functions.
We studied mitochondria function and intracellular position by employing Mitotracker Deep Red FM, a mitochondrion-selective dye that accumulates in active mitochondria. We also used in parallel Tetramethylrhodamine ethyl ester (TMRE), which is a fluorescent indicator of mitochondria membrane potential. Both TMRE and Mitotracker signals were dampened by the activity of Oligomycin A 47 , a benchmark inhibitor of mitochondrial ATP synthase. Interestingly, miR-124 overexpression inhibited TMRE and Mitotracker signals reminiscent of Oligomycin A, leading us to conclude that miR-124 regulates mitochondrial function (Fig. 3).
Confocal immunofluorescence of the mitochondrial marker, ATP5A, and an ultrastructural study with transmission electron microscopy (TEM), revealed depletion of mitochondria in primary motor neuron axons, which overexpressed miR-124, relative to controls. (Sup. Figures 3, 4). Therefore, miR-124 regulates mitochondria position and activity in motor neurons.
To characterize the targets and pathways that are regulated by miR-124, we intersected the list of mRNAs that were repressed more than twofold by miR-124 overexpression in the NGS data, with the list of predicted miR-124 targets (TargetScan 48 ). Three genes that harbor conserved miR-124 binding sites, were also downregulated more Enrichment landscape plot for all 876 7mer motifs complementary to canonical mouse miRNA seed regions, gained by Sylamer analysis 45 . Sorted 6500 mRNAs expressed in primary motor neurons, ranked from down-to up-regulated after overexpressing of miR-124, or control mimic. Assessment of over-and under-represented miRNA recognition sequences (seed-matches) for all known miRNAs, identified two enriched motifs, both matching the miR-124 'seed' sequence (blue, 7mer-2; light blue, 7mer-1A). (c) Top seven terms (by gene count) from gene ontology (GO) analysis on all mRNAs significantly up-or down-regulated following miR-124 overexpression (corrected P-value < 0.05), using database for annotation, visualization and integrated discovery (DAVID) software 46 . This analysis revealed an enrichment for mitochondrial-related genes and is further described in Table 1. than twofold by miR-124 overexpression, namely, Polypyrimidine Tract Binding Protein 1 (Ptbp1, MGI:97791), Midkine (Mdk, MGI:96949) and Vimentin (Vim, MGI:98932; Fig. 4a). qPCR study validated that miR-124 overexpression inhibited Vim to ~1/3 its levels compared with control cells that were treated with scrambled oligos. Ptbp1, an established miR-124 target 49 , and Mdk were downregulated to ~1/2 their expression level (Fig. 4b).
Interestingly, Vim is a known regulator of mitochondria localization and activity 31,32 , with miR-124 binding sites that are conserved across several vertebrate species (Fig. 4c). Furthermore, we identified molecular evidence for direct interactions of miR-124 with the target Vim, in Argonaute CLIP studies 50,51 and a Vim 3′UTR reporter was inhibited by miR-124 mimics in hepatocellular carcinoma cells 52 .
We sought to inhibit Vim independently of miR-124, and test its effect on cell morphology and mitochondria activity. Lentiviral transduction of primary motor neurons was very efficient (Sup. Figure 5) and allowed us to effectively knockdown Vim by shRNA 53 . Vim shRNA reduced Vim mRNA and protein levels (Fig. 4d,e, Sup. Figure 6a). Accordingly, high content image analysis of Vim knockdown depicted reduction in neurite outgrowth and branching (Fig. 4f), recapitulating miR-124 activity. In addition, Vim knockdown resulted in inhibition of mitochondria activity in axons, but not in the soma (Fig. 4g-i). Differences in the effect of miR-124 on soma may be the result of transfection efficiency. Intriguingly, miR-124 is primarily peri-nuclear 54 . Therefore, the soma compartment may be less amendable to manipulation, whereas, in the axon, where the miRNA is expressed at lower levels, the effects of miR-124 overexpression were consistent across all observations. In conclusion, Vim pheno-copies miR-124 functions, further suggesting that both genes are engaged in the same pathway.
Based on the above observations, we hypothesized that Vim is a novel effector of miR-124 in a pathway that regulates mitochondria function. Therefore, we tested whether upregulating Vim levels is sufficient for recovering mitochondrial activity. Doxycycline-induced expression of Vim, from a lentiviral vector that does not harbor the 3′UTR and hence is not inhibited by miR-124, upregulated Vim mRNA and protein levels ( Fig. 5a,b, Sup. Figure 6b). Exogenous Vim was also sufficient to alleviate miR-124-dependent inhibition of mitochondrial function, relative to miR-124 alone (Fig. 5c-e). Furthermore, when counting the numbers of mitochondria in axons, we observed rescue of axonal mitochondria occupancy by exogenous Vim (Fig. 5f-h; We also noted some effect without Dox that is probably due to leaky Vim expression). Finally, we used live imaging microscopy to test if the new miR-124-Vim axis regulates mitochondria motility in axons. We identified two subtypes of motile mitochondria: mitochondria that were continuously running in an uninterrupted manner along the axon over a distance of >10 µm at average speed >0.2 µm/sec., or mitochondria that displayed intermittent pauses in the same location for ≥3 frames in succession. Intriguingly, miR-124 caused mitochondria to pause more frequently on anterograde route, than in control samples, but did not affect retrograde transport. This may explain the re-distribution of mitochondria after overexpression of miR-124 and relative axonal depletion. Exogenous Vim expression rescued this phenotype and normalized mitochondria running/pausing dynamics (Fig. 6a-c). Furthermore, measurements of pause duration depicted asymmetry in anterograde vs. retrograde (Fig. 6d,e). We noted some effect without Dox that is probably due to leaky Vim expression. Parameters of run length and mean speed, were not changed by miR-124 and Vim (Sup. Figure 7). Therefore, anterograde mitochondria transport is influenced by the number of mitochondria pausing and the typical time interval spent resting during their run. We conclude that the mechanism for control of mitochondria running/pausing propensities in motor neurons involves Vim and miR-124 in a fashion affecting anterograde but not retrograde transport.
Taken together, we characterized a novel pathway in motor neurons downstream of miR-124 in regulation of mitochondria dynamics, distribution and activity. In this pathway, Vim, functions as an important effector of miR-124, revealing a surprising mechanism for controlling energy metabolism in motor neurons by neuronal miRNAs and intermediate filaments.

Discussion
In the current study, we employed high content image analysis, next generation sequencing and molecular approaches for discovery of a new pathway that is affecting motor neuron axon morphology and mitochondria homeostasis. An initial screen led us to focus on miR-124, which was the only miRNA that exhibited abnormal axonal morphology, out of nine miRNAs that were tested. miR-124 is one of the most abundant miRNA in many neuronal subtypes and is conserved from insects to mammals. miR-124 drives neuronal differentiation, promoting neuroblasts cell-cycle exit 41,55 , neuron-specific alternative splicing and chromatin remodeling via silencing of Ptbp11 49 and Actin like 6 A/BAF53a 56 , respectively. Furthermore, miR-124 levels remain high in postmitotic neurons, suggesting that it plays a role also in maintenance of the differentiated state of neurons. However, miR-124 roles in motor neuron were not thoroughly investigated. We demonstrated that miR-124 regulates mitochondrial activity and localization. The cluster of mitochondrial genes that responded to miR-124 overexpression was not particularly enriched in direct targets of miR-124, suggesting indirect regulation. The intermediate filament Vim is a key effector of miR-124 upstream of mitochondria function and localization. Vim knockdown by other means, pheno-copied miR-124 overexpression and an exogenous Vim that does not harbor miR-124 binding sites, rescued the mitochondrial phenotype.
Precise mitochondria localization is critical for maintaining energy and calcium homeostasis in neurons. Appropriate mitochondria localization is essential for neurite outgrowth [57][58][59] .Minin et al., have shown that Vim regulates mitochondria activity and motility 31,32,60 in other cell types, which is consistent with the discovery of a new miR-124 -Vim axis for unidirectional control of mitochondria transport in axons. miR-124 and Vim  asymmetric action is further evocative of kinesins and dynein motor proteins that reciprocally serve anterograde and retrograde transport. However, a direct molecular link to the classic kinesin/dynein system is still missing. miRNA dysregulation 7,61-63 and mitochondrial impairments [63][64][65] , are repeatedly observed in ALS. Our analysis reveals that miR-124 overexpression may be disadvantageous to primary motor neurons, in accordance with reported miR-124 upregulation in late ALS stages in mouse brains 61 and with injurious miR-124 overexpression in adult hippocampus/prefrontal cortex 66 . In summary, we propose that miR-124 expression levels should be tightly kept within defined margins and that a novel miR-124 -Vim pathway reveals a mechanism, by which miR-NAs regulate of axonal mitochondria transport.

Materials and Methods
Primary motor neuron culture. All experiments were performed in accordance with relevant guidelines and were approved by the Institutional Animal Care and Use Committee (IACUC) at Weizmann Institute of Science. Primary motor neurons were isolated and cultured as described 34 with the following modification: spinal cords were dissected from ICR mouse embryos at embryonic day 13.5 (E13.5). Motor neurons were dissociated with papain (2 mg/ml, Sigma), separated through Optiprep gradient (Sigma) and seeded either on 13 mm coverslips (200,000 cells/coverslip, Thermo scientific) or on 384 multiwell plates (7500 cells/well, Griener bio-one, cat# 781091) using a liquid handling device (GNF Systems), pre-coated with poly-ornithine (3 µg/ml) or poly L Lysine (Sigma P4707, 0.002% in Borate buffer 0.1 M ph 8.5, Sigma) and then Laminin (3 µg/ml, Gibco | Thermo Fisher Scientific). Motor neurons were cultured with Neurobasal (Gibco | Thermo Fisher Scientific)/B27 (Gibco | Thermo Fisher Scientific) medium supplemented with 2% horse serum (Sigma), and 1 ng/ml CNTF, 1 ng/ml GDNF (Peprotech) at 37 °C. For live imaging motor neurons were isolated as in 67,68 . miRNA mimics transfection. miRNA mimics, were dsRNA oligonucleotides (Integrated DNA Technologies, Inc.), as described in Table 2. dsRNA encapsulated in Neuro9 ™ nanoparticles (Precision NanoSystems, Inc.) 33 . Mimics (0.5 ng/µl) were transfected 24 hrs after seeding of primary motor neurons manualy or by using Bravo automated liquid handling robot.
Lentiviruses. Vimentin shRNA lentiviruses were described in 53  Imaging. For high content image analysis, eight/two micrographs taken in 24 /384 well plate setups, respectively, using automated fluorescence microscope (ImageXpress Micro and MetaXpress software, Molecular Devices). Motor neuron were defined by positive to Tuj1 staining (Alexa488, FITC channel) and nuclear stain (DAPI). Phenotypic parameters were quantified with relevant MetaXpress High-Content Image Acquisition modules (Neurite Outgrowth, MWCS). Confocal microscopy performed on Carl-Zeiss 710. Mitochondria Live imaging was performed on Nikon Eclipse Ti Spinning disc confocal with Yokogawa CSU X-1, 60X oil-immersion lens, Andor iXON3 EMCCD camera under controlled environment (37 °C, 5% CO2). Axonal transport analysis was carried by analysis of time-lapse images using imageJ or MATLAB, as in 67,68 . For transmission electron microscopy, motor neuron were prepared, following 69 and electron micrographs were captured with a FEI Tecnai SPIRIT transmission electron microscope (FEI, Eidhoven, Netherlands), operated at 120 kV and equipped with an EAGLE CCD Camera. RNA analysis and Next Generation Sequencing. Total RNA was isolated with miRNeasy micro kit (Qiagen), assessed with Nano Drop ND-1000 Spectrophotometer (Peqlab) and reverse transcribed to cDNA. qPCR, performed with SYBR Green (Thermo Fisher Scientific or Qiagen). miRNA/mRNA levels were normalized to U6/hypoxanthine phosphoribosyltransferase 1 (Hprt), respectively. Primer sequences are described in Table 2.
Statistical analysis. Analysis was performed manually or with GraphPad Prism 6 for Student's t-test, test of proportion or one way ANOVA with post-hoc Newman-Keuls or one way ANOVA with Duncan's new multiple range test (MRT), as indicated. Results are given as mean ± standard error of the mean (s.e.m). The null hypothesis was rejected at the 0.05 level ( * ), 0.01 ( ** ) or 0.001 ( *** ). Non-significant values on statistical test are not mentioned in the figures. Gene Ontology analysis was performed using DAVID 46 . miRNA mimics, designed as dsRNA oligonucleotides r-RNA bases; r_ * -Phosphorothioated RNA base; m-2' O-methyl RNA base