Research ArticleAmyotrophic Lateral Sclerosis

Human genetics and neuropathology suggest a link between miR-218 and amyotrophic lateral sclerosis pathophysiology

See allHide authors and affiliations

Science Translational Medicine  18 Dec 2019:
Vol. 11, Issue 523, eaav5264
DOI: 10.1126/scitranslmed.aav5264

A micro(RNA) contribution to ALS

Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disorder characterized by selective motor neuron degeneration and consequent progressive paralysis. Recent data have shown that microRNA-218 (miR-218) is enriched in motor neurons. However, whether neuronal miR-218 is modulated in ALS and plays a role in the disease is unknown. Now, Reichenstein et al. show that miR-218 controls neuronal activity by modulating the potassium channel Kv10.1 and its expression is reduced in motor neuron from patients with ALS. Screening of ALS genomes, the authors also identified miR-218 variants associated with reduced function, suggesting that this microRNA might play a role in ALS pathophysiology.


Motor neuron–specific microRNA-218 (miR-218) has recently received attention because of its roles in mouse development. However, miR-218 relevance to human motor neuron disease was not yet explored. Here, we demonstrate by neuropathology that miR-218 is abundant in healthy human motor neurons. However, in amyotrophic lateral sclerosis (ALS) motor neurons, miR-218 is down-regulated and its mRNA targets are reciprocally up-regulated (derepressed). We further identify the potassium channel Kv10.1 as a new miR-218 direct target that controls neuronal activity. In addition, we screened thousands of ALS genomes and identified six rare variants in the human miR-218-2 sequence. miR-218 gene variants fail to regulate neuron activity, suggesting the importance of this small endogenous RNA for neuronal robustness. The underlying mechanisms involve inhibition of miR-218 biogenesis and reduced processing by DICER. Therefore, miR-218 activity in motor neurons may be susceptible to failure in human ALS, suggesting that miR-218 may be a potential therapeutic target in motor neuron disease.


MicroRNA-218 (miR-218) is an endogenous small RNA that is enriched in motor neurons. Its relevance to motor neuron diseases was recently suggested by showing that miR-218 is essential for perinatal neuromuscular survival (1, 2), miR-218 is decreased in human amyotrophic lateral sclerosis (ALS) postmortem spinal cord (3, 4), and cell-free miR-218 can serve as a marker for motor neuron loss in a rodent model of ALS (4) and as a neuron-to-astrocyte signal (5). However, miR-218 was not yet studied in human motor neurons, and relevance to human ALS is still missing.

ALS is a fatal disease of the human motor neuron system, characterized by the selective degeneration of cortical and ventral spinal motor neurons. More than two dozen different genes have been associated with ALS in families or via genome-wide association studies. Mutations in these genes explain only a small fraction of the cases (69). Thus, ALS genetic variants in SOD1, NEK1, TARDBP, or FUS a are observed in <1 to 3% of cases, and the fraction of disease explained by the hexanucleotide repeat at the first exon of C9orf72 is <10% (6, 10). ALS-associated genes are ubiquitously expressed and therefore provide limited insight into why ALS shows motor neuron–selective vulnerability (8, 11).

Differential susceptibilities could be explained by the dysregulated activity of cell type–specific genes, including microRNAs (miRNAs). We and others have shown that miRNA dysregulation is involved in ALS (3, 1218).

In this study, we demonstrate (i) that miR-218 is specifically enriched in human spinal motor neurons and is down-regulated in ALS; (ii) that miR-218 orchestrates neuronal activity in a new pathway upstream of Kv10.1 (Kcnh1) voltage-gated potassium channel; and (iii) that rare genetic miR-218 variants, identified in patients with ALS, are detrimental to its biogenesis and function, providing a connection from human genetics to motor neuron–specific functions.


miR-218 is highly and specifically expressed in mature human and murine motor neurons

We sought to evaluate the relevance of miR-218 to human motor neuron and its relevance to ALS. First, miRNA in situ hybridization in human tissues depicted motor neuron–specific expression pattern of miR-218 in ventral motor neurons throughout the human spinal cord (Fig. 1A). In parallel, we differentiated human induced pluripotent stem cells (iPSCs) into motor neurons, following a protocol developed by Kiskinis et al. (19). Accordingly, several mRNA markers of motor neuron differentiation were up-regulated, namely, ISL1, HB9, and CHAT. miR-218 expression was up-regulated >2000-fold from undifferentiated pluripotent state to human motor neurons (Fig. 1B). We then assessed miR-218 expression in laser capture microdissection–enriched motor neurons from lumbar spinal cords of samples where neurological disease was not reported, by revisiting data that were generated in our previous work (3). miR-218 is specifically enriched in control motor neurons, relative to surrounding non-motor neuron tissue at the ventral horn of the human lumbar spinal cord, or relative to proprioceptive neurons at Clarke’s column. Furthermore, assessing miR-218 expression in laser capture microdissection–enriched surviving lumbar motor neurons of patients with ALS, who suffered from bulbar onset disease, revealed about twofold repression relative to control lumbar motor neurons (Fig. 1C and data file S1). We further tested another independent set of postmortem tissues with an orthogonal nanoString nCounter miRNA profiler. This RNA study revealed that miR-218 was the most down-regulated miRNA in lumbar ventral horns of sporadic ALS (sALS) nervous systems relative to non-neurodegeneration controls (Fig. 1D and data file S2). Reduced miR-218 in ALS may be explained by loss of motor neurons and/or by molecular down-regulation in motor neurons that are still present in the ventral horn. Accordingly, we have performed miR-218 in situ hybridization that revealed reduced numbers of miR-218+ cells in ALS patient tissue, relative to non-neurodegeneration controls (Fig. 1E and data file S1) and a reduction in the densitometric miR-218 in situ hybridization signal in ALS motor neurons (Fig. 1F). Last, we demonstrated that there is a global up-regulation (derepression) of miR-218-5p targets in human ALS spinal motor neurons by comparing the expression of top 100 predicted miR-218-5p mRNA targets [TargetScan (20)] in laser capture microdissection–enriched surviving motor neurons from lumbar spinal cords of patients with sALS, relative to all expressed mRNAs and to the expression in non-neurodegeneration controls (Fig. 1G) (21). Together, our results show that miR-218 is a highly sensitive marker of human spinal motor neurons, whose expression rises high in developing human motor neuron and is maintained in the adult. miR-218 expression is reduced in motor neuron disease because of both molecular down-regulation and motor neuron loss, and the mRNA targets of miR-218 are reciprocally up-regulated. Therefore, miR-218 might serve as a marker of motor neuron mass in the human ventral horn in ALS.

Fig. 1 miR-218 is expressed in the human spinal motor neurons and is down-regulated in human ALS.

(A to F) Three orthogonal miRNA quantification studies in human motor neurons from 20 ALS cases and 14 non-neurodegeneration controls. (A) miR-218 chromogenic in situ hybridization depicting broad expression along the cervical, thoracic, and lumbar regions of the adult human spinal cord. (B) qPCR analysis of miR-218, ISL1, HB9, and CHAT in human iPSCs and differentiated motor neurons. miR-218 normalized to U6 expression. mRNAs normalized to average of hypoxanthine-guanine phosphoribosyltransferase (HPRT) and β-actin expression, presented on a log scale; n = 3 independent wells per time point. (C) miR-218 expression in laser capture microdissection–enriched motor neurons from non-neurodegeneration controls (n = 7 human lumbar spinal cords) relative to surrounding non–motor neuron anterior horn tissue (n = 10), to Clarke’s column proprioceptive neurons (n = 4), or to ALS motor neurons (n = 9 sporadic and 2 familial nervous systems carrying the SOD1 A4V mutation).TaqMan qPCR analysis of miR-218 normalized to the average of RNU48/SNORD48, RNU44/SNORD44, and U6 in the same sample and to the average miR-218 expression in the anterior horn. One-way ANOVA, followed by Newman-Keuls multiple comparisons test, was performed on log-transformed data (means ± SD). (D) Volcano plot of relative miRNA expression in ALS lumbar ventral horns (n = 5) versus non-neurodegeneration controls (n = 2; x-axis log2 scale), screened by NanoString nCounter platform. Y axis depicts the differential expression P values (−log10 scale). Black dots indicate P < 0.05; light gray dots are nonsignificant. miR-218 is the most down-regulated miRNA in ALS nervous systems. Data normalized to the average of five control mRNAs (ACTB, B2M, GAPDH, RPL19, and RPLP0). (E) Reduced miR-218+ cell numbers in sALS patient anterior horns (n = 4) relative to non-neurodegeneration controls (n = 5) and representative miRNA in situ hybridization micrographs. Two-way ANOVA followed by Bonferroni’s multiple comparisons test (means ± SEM). (F) Chromogenic miR-218 in situ hybridization signal densitometry in motor neurons at different spinal cord levels (non-neurodegeneration control/ALS cases: cervical: n = 151 control cells, 85 ALS cells; thoracic: n = 54 control cells, 75 ALS cells; lumbar: n = 189 control cells, 92 ALS cells). One-tailed Mann-Whitney test (means ± SEM) was performed. OD, optical density. (G) Cumulative distribution function (CDF) plot of top 100 predicted miR-218-5p targets [TargetScan (20)], or all expressed mRNAs, in laser capture microdissection–enriched surviving motor neurons from lumbar spinal cords of patients with sALS with rostral onset and caudal progression (n = 13) relative to non-neurodegeneration controls [n = 6; (21)] and box plot (inset) depicting median, upper and lower quartiles, and extreme points. P value was calculated using Kolmogorov-Smirnov test comparing miR-218-5p target subset distribution to all genes. *P < 0.05; ***P < 0.001; ****P < 0.0001.

miR-218 regulates motor neuron network activity

To study miR-218 function, we moved to rodent models, whereby miR-218 is specifically expressed in mouse motor neurons, without any preference to motor neuron subtypes (fig. S1) (1, 2). We first performed ontology analysis (22) of predicted miR-218 targets (20). This study identified enrichment in biological processes related to potassium ion transmembrane transport (Fig. 2A). Therefore, we tested the hypothesis that miR-218 regulates primary motor neuron gene expression and activity. Dissociated embryonic mouse spinal cords were enriched for motor neurons via OptiPrep gradient sedimentation (23) and transduced with lentiviruses encoding miR-218 overexpression (OE) or miR-218 knockdown (KD). Next-generation sequencing (NGS) of RNA revealed that predicted miR-218 targets [TargetScan; (20)] were significantly down-regulated after OE of miR-218 (P < 0.0001; Fig. 2B). Accordingly, enrichment for two miR-218-5p seed matches was depicted among mRNAs that were down-/up-regulated after miR-218 OE/KD, respectively [Sylamer study (24); Fig. 2, C and D]. No signatures were identified for the target set of any other miRNA. Therefore, the vectors used were specifically affecting miR-218 expression or silencing functions. Expression data are available at Gene Expression Omnibus (GSE136409).

Fig. 2 miR-218 controls motor neuron network activity.

(A) Seven most enriched gene ontology terms (22) of predicted miR-218 targets (20). P value of term enrichment (−log10; dashed orange line indicates P = 0.05). RISC, RNA-induced silencing complex. (B) CDF plot of miR-218 predicted targets, relative to all expressed mRNAs, after OE of miR-218 and box plot (insets), depicting median, upper and lower quartiles, and extreme points. P value was calculated using Kolmogorov-Smirnov test comparing miR-218-5p subset distribution to all genes. ****P < 0.0001. (C and D) Binding site enrichment of all known miRNAs, in ~10,000 expressed mRNAs, was tested after (C) miR-218 OE or (D) miR-218 KD relative to control virus. Significant enrichment for two miR-218-5p seed matches (blue and red) and lack of enrichment for any other miRNA (gray) via a Sylamer study (24). (E) Diagram of calcium transient imaging in embryonic rat spinal motor neurons, transduced with lentiviruses encoding control vector, miR-218 OE, or a miR-218 KD. E14, embryonic day 14. (F to H) Neuron time lapse micrographs (F), representative traces (G), and quantification (H) of spontaneous calcium spike frequencies (∆F/F > 0.5) from Fluo2 HighAff AM study after 12 days in vitro. Recorded from 58/76/41 control/OE/KD cells, respectively. Box plot depicting median, upper and lower quartiles, and extreme points. ***P < 0.001, Kruskal-Wallis test followed by Dunn’s multiple comparisons test. This experiment was repeated three independent times with similar results.

We then monitored intracellular calcium transients in primary rat motor neurons that overexpressed (about eightfold) or knocked down (~50%) miR-218. Calcium dynamics were monitored on days 12 and 13 in vitro using the Ca2+-sensitive dye Fluo2 HighAff AM, setting the spike threshold for activity as ΔF/F > 0.5 over baseline (Fig. 2, E and F). miR-218 OE increased the frequency of spontaneous calcium bursts by ~70% compared to cells that were transduced with control viruses, whereas miR-218 KD attenuated neuronal Ca2+ transient by ~80% relative to control (Fig. 2, G and H). Changes in miR-218 expression did not alter motor neuron viability or morphology (fig. S2). Therefore, miR-218 regulates neuronal activity.

miR-218 regulates neuronal intrinsic excitability

To test whether miR-218 is involved in the regulation of active or passive conductance in neurons, we further used patch clamp. However, because primary motor neurons displayed an elevated resting membrane potential of >−50 mV in our hands, consistent with a previous study (25), we were forced to use primary rat hippocampal neurons as alternative, a well-established cell type for patch-clamp studies, which expresses miR-218, though less than spinal motor neurons (2628). Current-clamp electrophysiological experiments were performed with CNQX (6-cyano-7-nitroquinoxaline-2,3-dione; AMPA/kainate blocker) and APV (d,l-2-amino-5-phosphonovaleric acid; N-methyl-d-aspartate receptor blocker) on culture days 15 to 21. In response to current injection (300 pA, 500 ms), neuronal firing frequency was about twofold higher with miR-218 OE relative to miR-218 KD (17.9 ± 1.3 Hz versus 9.4 ± 1.9 Hz, P < 0.01; fig. S3, A and B), and rheobase, the current input required to generate an action potential (500 ms −100 to +500 pA steps in 20 pA increments), was ~35% lower in miR-218 OE relative to miR-218 KD (157 ± 12 pA versus 242 ± 19 pA, P < 0.001; fig. S3, C and D). Mean voltage threshold for triggering the first spike was unchanged between the different conditions (fig. S3E), and resting membrane potential correlated in a bidirectional manner with miR-218 expression (miR-218 OE: −58.8 ± 0.7 mV, n = 65; control: −60.9 ± 0.7 mV, n = 44; miR-218 KD: −63.7 ± 0.7 mV, n = 30; P < 0.001; fig. S3F). Together, network and intrinsic activity studies support the hypothesis that miR-218 regulates neuronal excitability, at least in rat hippocampal neurons.

Kv10.1 functions downstream of miR-218 in motor neurons

To gain molecular insight into the mechanisms by which miR-218 regulates network activity, we next focused on a selected set of relevant targets in the context of neuronal activity. This group includes the potassium channels Kv4.2 (Kcnd2) and Kv10.1 (Kcnh1), γ-aminobutyric acid (GABA) receptor subunits Gabrb2 and Gabrg1, GABA transporter GAT1 (Slc6a1), and the calcium channel β subunit Cacnb4. The changes in the expression of the above six targets, in response to miR-218 OE, were validated in an independent set of experiments using quantitative polymerase chain reaction (qPCR) on RNA extracted from rat primary motor neurons (Fig. 3A).

Fig. 3 The potassium channel Kv10.1 acts downstream of miR-218.

(A) qPCR measuring the expression of mRNA targets after miR-218 OE (n = 15). Data normalized to control virus (n = 12) and to average expression of HPRT and β-actin; two technical duplicates, two-sided Student’s t test (means ± SEM). (B) Representative traces of individual motor neurons and (C to H) quantification of spontaneous calcium spike frequencies (∆F/F > 0.5) of embryonic rat spinal motor neurons, transduced with lentiviruses encoding a control vector, or miR-218 KD and further transfected with siRNA for specific target KD, or a nontargeting siRNA control (minus sign). Cells (≥55) were recorded per experimental condition; n ≥ 2 independent experimental repeats with similar results, Kruskal-Wallis test followed by Dunn’s multiple comparisons test. ns, nonsignificant. (I) Relative Renilla luminescence upstream of a WT Kv10.1 3′UTR or a mutated 3′UTR that is insensitive to miR-218, normalized to coexpressed firefly luciferase and to a negative control miRNA vector. n = 3 independent wells per experimental condition, one-way ANOVA followed by Bonferroni’s multiple comparisons test (means ± SEM). (J) miR-218:Kv10.1 3′UTR chimera from an AGO2 CLEAR-CLIP experiment in mouse cortex (29). (K) miR-218 expression (qPCR: n = 3, normalized to U6) and (L) Kv10.1 protein expression (Western blot: n = 5), upon miR-218 lentiviral KD or OE, in primary rat motor neurons and a representative blot detected with anti-Kv10.1 and anti–tubulin β-III (TUBB3) antibodies. Box plots depict median, upper and lower quartiles, and extreme points; one-way ANOVA followed by Newman-Keuls multiple comparisons test. a.u., arbitrary units. (M) Kv10.1 mRNA expression, as log2-normalized counts, from NGS study of induced ALS motor neurons (n = 4 different donors in duplicates) or non-neurodegeneration controls [n = 3 different donors in duplicates; (30)]. Box plots depict median, upper and lower quartiles, and extreme points (DESeq analysis). (N) Kv10.1 mRNA expression, as reads per kilobase million (RPKM) from NGS study of laser capture microdissection–enriched surviving motor neurons from lumbar spinal cords of patients with sALS with rostral onset and caudal progression (n = 12) and non-neurodegeneration controls [n = 8; (21); GSE76220]. Box plots depict median, upper and lower quartiles, and extreme points. *P < 0.05; **P < 0.01; ***P < 0.001, two-sided Student’s t test.

Because miR-218 enhances neuronal activity, we hypothesized that relevant mRNA targets potentially encode for proteins acting downstream of miR-218 in inhibiting neuronal activity. Thus, their KD should increase bursting, reminiscent of miR-218 OE, and concomitant KD of both miR-218 and its target may rescue neuronal activity.

We therefore analyzed the frequency of spontaneous calcium transients in primary motor neurons after candidate target KD, with small interfering RNA (siRNA) nanoparticles that exhibited 20 to 80% target mRNA KD (fig. S4). Nontargeting siRNAs were used as control. KD of either Kv10.1 (Kcnh1) or Kv4.2 (Kcnd2) enhanced the frequency of spontaneous calcium transients and was sufficient to rescue neuronal excitation upon miR-218 inhibition (Fig. 3, B to D). In addition, we tested the calcium channel Cacnb4 and GABA pathway components Gabrb2, Gabrg1, and GAT1, which did not obey the requirements to be considered as epistatic downstream effectors of miR-218 in the motor neuron system, under our experimental conditions (Fig. 3, E to H).

To substantiate the evidence for the relevance of voltage-gated potassium channels, we performed a series of additional studies that collectively increased our confidence in the relevance of Kv10.1 and were not sufficiently supportive of Kv4.2 in this context. We demonstrated that both Kv10.1 (Kcnh1) and Kv4.2 (Kcnd2) mRNAs can be directly targeted by miR-218 by measuring the luminescence of a Renilla reporter, harboring the 3′ untranslated region (3′UTR) of either Kv10.1 (Kcnh1) or Kv4.2 (Kcnd2). miR-218 silencing was abrogated by mutated miRNA recognition sequences (Fig. 3I and fig. S5A). We also mined miRNA-mRNA chimera data from AGO2 cross-linking and immunoprecipitation study in the mouse cortex (29). This study revealed miR-218 binding to the 3′UTR of Kv10.1 in vivo in the unmanipulated cortex (Fig. 3J). We next transduced primary rat motor neurons with viral vectors that either overexpress or knockdown miR-218 (Fig. 3K). miR-218 expression reciprocally correlated with Kv10.1 protein under miR-218 KD (Fig. 3L, fig. S6, and data file S3), as could be expected from a genuine target. miR-218 OE did not affect Kv10.1 expression, which might be due to the high basal miR-218 expression in motor neurons. Last, to test whether Kv10.1 is up-regulated in ALS, along with miR-218 down-regulation, we mined human NGS data, which revealed higher Kv10.1 mRNA expression in ALS, both in induced human motor neurons of patients with ALS (Fig. 3M) (30) and in laser capture microdissection–enriched surviving motor neurons from lumbar spinal cords of patients with sALS with rostral onset and caudal progression (Fig. 3N) (21). A parallel analysis of Kv4.2 expression and regulation was not equally supportive, and it is therefore less likely to be regulated by miR-218 in motor neurons and ALS (fig. S5, B to F, and data file S4). Therefore, Kv10.1 appears as a relevant miR-218 target in vitro and in vivo and might be relevant also in human ALS.

Rare miR-218 genetic variants are detected in human patients with ALS

To examine the relevance of miR-218 to human disease, we screened for rare genetic variations (minor allele frequency < 0.01) in the human genes, hsa-mir-218-1 (Chr. 4) and hsa-mir-218-2 (Chr. 5), in ALS and control cohorts of Project MinE ALS Sequencing Consortium data (31). We observed six unique rare variants in the precursor miR-218-2 (pre-miR-218-2) gene and a single variant (rs371622197) in pre-miR-218-1 (Fig. 4A and table S1) in multinational cohorts, which were matched geographically and for ancestry (see Materials and Methods). None of these variants were harbored within the ~22 nucleotides of mature miR-218-5p [miRBase v20 (32)]. Region-based rare variant association testing by the Optimized Sequence Kernel Association Test (SKAT-O) (33) was nonsignificant (adjusted P > 0.05). However, odds ratio (OR) was 1.93 with 95% confidence interval (CI) of 0.42 to 8.96 (Fig. 4B). We then performed an independent replication study on additional cohorts of Genomic Translation for ALS Care (GTAC), the ALS Sequencing Consortium and the New York Genome Center (NYGC) ALS Consortium for rare miR-218-2 variant association.

Fig. 4 Rare genetic miR-218 variants disrupt its ability to regulate neuronal activity.

(A) Diagrams of hsa-miR-218-2 pri-miRNA (top) and the pre-miRNA hairpin (bottom), with demarcation of DROSHA, DGCR8, and DICER binding and arrows, revealing variant nucleotides (V1 to V6). Guide RNA is shown in red. (B) Table and forest plot depicting odds ratio (OR) estimates with 95% confidence intervals (CIs), across study cohorts and P values, calculated with SKAT-O or χ2 test with Yates’ correction. Vertical dotted line denotes OR = 3. (C) Representative motor neuron traces and (D) quantification of spontaneous calcium spike frequencies (∆F/F > 0.5) in embryonic rat spinal motor neurons, transduced with lentiviruses encoding WT or mutated human miR-218-2. Number of cells recorded in a single experiment: control: n = 131; WT miR-218-2: n = 114; single variant V2: n = 137; single variant V5: n = 119; multiple variant Vall: n = 118; unprocessable miR-218-2 Vdead: n = 111. n = 4 independent times with similar results. ***P < 0.001, Kruskal-Wallis test followed by Dunn’s multiple comparisons test.

Rare miR-218-2 variants were enriched in cases (P = 0.048 by SKAT-O; OR, 3.06; 95% CI, 0.86 to 10.84). Meta-analysis of both discovery and replication cohorts P value was 0.067 by SKAT-O (34), and a joint analysis P value was 0.0195 (χ2 with Yates’ correction; OR, 2.87; 95% CI, 1.11 to 7.40; Fig. 4B). Therefore, the burden of variants showed nominal association to the trait (P < 0.05), although it did not reach genome-wide significance (P = 5.0 × 10−8) with ALS in our study. Last, we assessed an independent large cohort of 62,784 non-ALS genomes from National Heart, Lung, and Blood Institute’s (NHLBI’s) Trans-Omics for Precision Medicine (TOPMed). This validation effort yielded a joint P value of 0.0002 by χ2 test with Yates’ correction with OR of 3.02 (95% CI, 1.65 to 5.52), which confirmed the robustness of the findings (Fig. 4B). This modest excess of rare pre-miR-218-2 variants in ALS did not survive genome-wide statistical correction. Together, individuals harboring miR-218-2 sequence variants have a risk that is almost three times as high to suffer from ALS relative to the general population.

Rare miR-218 genetic variants disrupt its ability to regulate neuronal excitability

miRNA genes exhibit high evolutionary conservation, and sequence mutations may be detrimental to their function. We sought to test the impact of mutated miR-218 on neuronal activity by intracellular calcium transient recording. The variants were aggregated in two main domains, namely, in the loop region, which is supposed to bind DGCR8 (35), and in the miRNA 3′ terminal, which is cleaved by DROSHA (35, 36) and then becomes an important element of recognition by DICER (37, 38). To test these variants functionally, we created vectors that represent loop and 3′ terminal variants. Then, we transduced primary rat motor neurons with the following miR-218-2 vectors: (i) control; (ii) wild-type (WT) human miR-218-2; (iii) the predominant pre-miRNA loop variant (Chr5:168,195,207, V2); (iv) the most abundant patient variant at the miRNA 3′ terminal (Chr5:168,195,174, V5); (v) a miR-218-2 version, harboring a collection of variants, superimposed from cases (Vall); or (vi) a miR-218-2 sequence that we designed to be resistant to DROSHA activity, which yields no mature miR-218 (Vdead). WT miR-218 increased spontaneous calcium burst frequency as expected, whereas miR-218 with variant sequences failed to up-regulate neuronal Ca2+ transient frequency (Fig. 4, C and D).

Rare miR-218 genetic variants inhibit its biogenesis

We tested the hypothesis that miR-218 variants impair neuronal bursting through inhibition of biogenesis or creation of abnormal forms of the mature miRNA. We used HEPG2 cells, which do not express the endogenous miR-218 gene, to overexpress WT or mutated forms of primary miR-218 (pri-miR-218). In addition, we cotransfected miR-214-3p mimics, which served as spike-in control for downstream normalization.

We performed small RNA NGS on RNA extracted from transfected HEPG2 cells (Fig. 5A). miRNAs were the dominant RNAs in the libraries (56%; Fig. 5B) at about a million miRNA reads/library and complexity of ~160 different miRNA species (Fig. 5C). The expression of mature miR-218 after transfection was comparable with the most abundant endogenous miRNAs in HEPG2 cells (Fig. 5D). miR-218-5p dominated the expression profile, whereas sequences aligned to the loop or miR-218-3p were less prevalent, as expected (Fig. 5E). Furthermore, the isomiR-218 profile was comparable across different variants (Fig. 5F). The expression of mature miR-218, derived from mutated forms of pri-miR-218, was lower compared to the WT form (Fig. 5G). We validated the drop in mature miR-218 expression, when harboring variants, with quantitative real-time PCR (Fig. 5H). We also detected the accumulation of pre-miR-218 forms after transfection with a vector harboring the most abundant variant (V5; Fig. 5I), a hallmark of failed biogenesis. The inhibition score (3), describing the ratio of DICER substrate (pre-miR-218) to product (mature miR-218), was increased by 3.4-fold for the predominant pre-miRNA loop variant (V2) and 3.1-fold for the most abundant variant (V5), relative to WT miR-218, demonstrating inhibition of miR-218 biogenesis (Fig. 5J). Together, mutated miR-218 exhibits impaired biogenesis, providing a conceivable mechanism for insufficient regulation of neuronal activity.

Fig. 5 Rare genetic variants in miR-218 inhibit biogenesis.

(A) Diagram of experimental design. HEPG2 cells transfected with WT miR-218-2 or miR-218-2 genetic variants and RNA processing for NGS and qPCR studies. (B) Pie chart of relative representation of different RNA families in NGS data [percentage of reads aligned to miRNA: 56%; transfer RNA (tRNA): 20%; ribosomal RNA (rRNA): 13%; other RNA types: 11%]. (C) The number of expressed miRNAs was comparable across samples (means ± SEM). (D) MA plot of miRNA expression in HEPG2 cells transfected with WT miR-218-2 relative to control vector. Abundance (x axis; presented on a log scale) against ratio of miRNA in cells overexpressing WT miR-218 versus a control vector (log2 fold change). (E) Histogram of number of reads per base for WT miR-218-2 sequences, aligned over the genomic sequence. (F) Bar graph of miR-218-5p isotypes (isomiR-218-5p, sequence denoted) in HEPG2 transfected with WT miR-218-2 or V2/V5 variants. (G and H) Relative expression of mature miR-218-5p from NGS (G) or TaqMan qPCR studies (H), normalized to miR-214-3p spike-in mimics. (I) Pre-miR-218-2 expression from NGS. (J) Ratio of pre-miR-218-2 (substrate) to mature miR-218-5p (product), defined as “inhibition score.” Inhibition score approximates a value of 1 in the WT condition, whereas a value of >1 reflects reduced DICER activity. Control: n = 3; WT miR-218-2: n = 5; single variant V2: n = 4; single variant V5: n = 4; multiple variant Vall: n = 5; unprocessable miR-218-2; Vdead: n = 3. Box plots depict median, upper and lower quartiles, and extreme points. *P < 0.05; **P < 0.01; ***P < 0.001, one-way ANOVA followed by Bonferroni’s multiple comparisons test performed on data (I) or log-transformed data (G, H, and J).


miR-218 was recently put in the spotlight for its roles in motor neuron development (1, 2). The link between perinatal death of mice deficient of miR-218 and a potential deleterious effect in adult humans motivated our investigations. In the current work, we demonstrated miR-218 relevance to human motor neurons in a systematic effort that explains how miR-218 contributes to a previously unappreciated facet of motor neuron specificity and disease susceptibility. ALS neuropathology establishes miR-218 as a marker of human motor neuron mass and well-being that is down-regulated in ALS. Accordingly, mRNA targets of miR-218 are up-regulated/derepressed.

We identified rare sequence variants in the miR-218-2 gene that impair miR-218 biogenesis and its ability to regulate motor neuron activity. These sequence variants are relevant for the understanding of motor neuron health and disease. We suggest that miR-218-2 variants are suboptimal for a DICER-dependent step of biogenesis, thus reducing mature miR-218 expression and contributing to selective motor neuron vulnerability. Subtle miR-218 down-regulation in humans plausibly contributes to failed homeostasis in adults, potentially due to broad up-regulation (derepression) of dozens of miR-218 targets in human motor neurons. Furthermore, because miR-218 expression is down-regulated in motor neurons of sporadic and familial patients with ALS, individuals harboring miR-218 variants suffer two sequential hits to miR-218 expression and function. Therefore, miR-218 is a relevant candidate for genetic screening in additional ALS genetics cohort.

A previously unrecognized pathway downstream of miR-218 controls neuronal activity by regulating the voltage-gated potassium channel Kv10.1. Altered motor neuron excitability and ion channel dysfunction have been reported in patients, rodent, and ALS iPSC models (3951) and drugs such as ezogabine (retigabine) (52) or riluzole, which control potassium and sodium channels, respectively, elute to the relevance of therapeutically altering neuronal activity in ALS.

In addition, increased expression of voltage-gated potassium channel subtypes has been reported in iPSC-derived ALS motor neurons with FUS and SOD1 mutations, and targeting potassium currents with 4-aminopyridine, a potassium channel blocker, recovered neuronal activity patterns in culture (53). These observations resonate with miR-218 activity upstream of voltage-gated potassium channel and suggest that aberrant neuronal activity is an important contributing factor at the ALS milieu.

Our study does not rule out that additional targets may play parallel roles in controlling neuronal activity downstream of miR-218. miR-218 is a member of an expanding class of miRNAs that regulate neuronal activity in flies (54, 55) and mammals (5659), including miR-128 (57), miR-101 (28), and miR-324-5p (60). The emerging regulation of neuronal activity by miRNAs depends on their capacity to fine-tune the expression of dosage-sensitive proteins locally at dendrites, axons, and synapses.

miR-218 regulates a myriad of targets designated Target218 (1). Our work, along with reported specific targets in astrocytes and neuronal progenitors (2, 5), contributes to deconvoluting the Target218 network. Amin et al. (1) recently showed by a patch-clamp study in lumbar spinal slices that miR-218 contributes to inhibiting neuronal activity. Reconciling this observation with ours requires new conditional miR-218 alleles that will allow uncoupling miR-218 roles in interneuron differentiation (2) and plausibly in establishing interneuron–motor neuron circuitry from miR-218 roles in adult motor neurons. Furthermore, developmental loss of miR-218 causes motor neuron death, further complicating the comparison to the moderate KD in postmitotic motor neurons.

In summary, motor neuron–enriched miR-218 might serve as a marker of motor neuron mass in the human ventral horn in ALS, and miR-218 functions uncover previously unappreciated facets of motor neuron specificity that may be particularly susceptible to failure in human patients with ALS. Currently, it is not clear whether the global miR-218 down-regulation in human neuropathology is a consequence of DICER inhibition (3) and how such a down-regulation might affect non–cell-autonomous effects of miR-218 (5). Mouse modeling can be beneficial for exploring miR-218 allele genetic interactions with other ALS-associated mutations and the functional implications of the discovered variants in the miR-218-2 gene sequence. Therefore, the study contributes to an emerging view of ALS as a disease with a prominent RNA component and suggests that miR-218 is a potential therapeutic target for motor neuron disease (graphically summarized in fig. S7).


Study design

The overall objective of our study was to investigate the relevance of motor neuron–specific miR-218 to human motor neuron specificity and disease (summarized in fig. S7) by using molecular, neurogenetic, and neuropathology approaches. First, we performed four orthogonal miRNA quantification studies in human motor neurons: (i) chromogenic miR-218 in situ hybridization in human spinal cord, (ii) NanoString nCounter, (iii) miR-218 qPCR, and (iv) analysis of mRNA expression of miR-218 targets from laser capture microdissection–enriched surviving motor neurons from lumbar spinal cords of patients with sALS. These experiments established miR-218 as a marker of human motor neuron mass and well-being. To test whether miR-218 regulates motor neuron activity, we transduced primary motor neuron with lentiviruses encoding miR-218 OE or KD and monitored intracellular calcium transients and intrinsic activity by patch-clamp electrophysiological experiments. A series of bioinformatics and experimental steps collectively directed us to conclude that Kv10.1 is a direct target of miR-218 in this system. Using statistical genetics and burden studies of rare variants, we identified miR-218 genetic variants in large ALS cohorts. The variants were shown to inhibit biogenesis and impair miR-218 function. Experimentalists were blinded while analyzing data. Outliers were excluded if deviated ±2 SDs away from the mean. Twenty ALS cases and 14 non-neurodegeneration control human motor neuron systems were taken for neuropathology. For neurogenetics, we studied 7738 ALS and 71,656 control genomes. These numbers reflect the maximal availability at the time of the study.

Statistical analysis

Statistics were performed with Prism Origin (GraphPad Software Inc.). Shapiro-Wilk test was used to assess normality of the data. Pairwise comparisons passing normality test were analyzed with Student’s t test, whereas the Mann-Whitney test was used for pairwise comparison of nonparametric data. Multiple-group comparisons passing normality test were analyzed using analysis of variance (ANOVA) with post hoc tests, whereas nonparametric multiple-group comparisons were analyzed using Kruskal-Wallis test with Dunn’s post hoc testing, when ANOVA assumptions were not met. Statistical P values <0.05 were considered significant. Data are presented as specified in the figure legends. Data are shown as means ± SEM or SD or graphed using box plots, as noted in the text. Individual subject-level data are reported in data file S1.


Materials and Methods

Fig. S1. miR-218 is highly and specifically expressed in human and murine spinal motor neurons.

Fig. S2. High content analysis of neuronal morphology after miR-218 perturbation.

Fig. S3. miR-218 regulates intrinsic excitability.

Fig. S4. qPCR validation of miR-218 target KD.

Fig. S5. Evaluation of miR-218 upstream of the mRNA encoding for the potassium channel Kv4.2 (Kcnd2).

Fig. S6. Kv10.1 (KCNH1) protein quantification by Western blot after miR-218 KD.

Fig. S7. A summary diagram of key observations.

Table S1. Identified hsa-miR-218-2 variants.

Table S2. DsiRNA sequences used in the study.

Table S3. Synthetic miR-218 sequences used for cloning into pMA-T vectors.

Table S4. Primers used for quantitative real-time PCR.

Data file S1. Individual-level data for miR-218 expression.

Data file S2. NanoString nCounter data for miRNAs measured in lumbar ventral horns.

Data file S3. Source data for Kv10.1 (KCNH1) Western blot studies.

Data file S4. Source data for Kv4.2 (KCND2) Western blot studies.

References (6172)


Acknowledgments: We gratefully acknowledge the contributions of all participants and the investigators who provided biological samples and data for Project Mine ALS Sequencing Consortium, the GTAC, the ALS Sequencing Consortium, the NYGC ALS Consortium, and the TOPMed (Trans-Omics for Precision Medicine) of the NHLBI ( Samples used in this research were, in part, obtained from the UK National DNA Bank for MND Research, funded by the MND Association and the Wellcome Trust. We would like to thank people with MND and their families for participation in this project. We acknowledge sample management undertaken by Biobanking Solutions funded by the Medical Research Council at the Centre for Integrated Genomic Medical Research, University of Manchester. We would like to thank T. Hajbi and O. Amram for veterinary services and husbandry. We thank A. Mauricio, J. Flautero, I. Ulitsky, M. Segal, B. Atalli, A. Yaron, U. Asheri, R. Tal, and G. Lippi for insightful comments on the manuscript. We are grateful to N. Sadeh and M. Segal for training in calcium imaging; H. Barr, N. Kozer, and A. Plotnikov for assistance with automated imaging; T. Shalit and Z. Melamed for assistance with bioinformatics; K. Cohen-Kashi Malina for assistance with patch clamp; and M. J. Rodriguez for histology. The Hornstein laboratory was supported by friends of S. Brenner. E.H. is the head of Nella and Leon Benoziyo Center for Neurological Diseases and incumbent of Ira and Gail Mondry Professorial Chair. Project MinE ALS Sequencing Consortium collaborators: C. E. Shaw (Maurice Wohl Clinical Neuroscience Institute and UK Dementia Research Institute, Department of Basic and Clinical Neuroscience, Department of Neurology, King’s College London, London, United Kingdom), P. van Damme and W. Robberecht [Laboratory of Neurobiology, VIB, Vesalius Research Center and Department of Neurosciences, Experimental Neurology and Department of Neurology, University Hospitals Leuven and Research Institute for Neuroscience and Disease (LIND), KU Leuven–University of Leuven, Leuven, Belgium], J. D. Glass (Department of Neurology and Emory ALS Center, Emory University School of Medicine, Atlanta, GA, USA), R. L. McLaughlin (Population Genetics Laboratory, Smurfit Institute of Genetics, Trinity College Dublin, Dublin, Republic of Ireland), J. S. Mora Pardina (ALS Unit, Hospital San Rafael, Madrid, Spain), and M. Povedano Panadés [Biomedical Network Research Center on Neurodegenerative Diseases (CIBERNED), Institute Carlos III and Functional Unit of Amyotrophic Lateral Sclerosis (UFELA), Service of Neurology, Bellvitge, University Hospitalet de Llobregat, Spain]. GTAC Consortium collaborator: D. Goldstein (Institute for Genomic Medicine, Columbia University Irving Medical Center, New York, NY, USA; supported by a grant from Biogen). Funding: The work was funded by Target ALS (118945 to E.H., J.M.R., and S.L.P.), Legacy Heritage Fund, Bruno and Ilse Frick Foundation for Research on ALS, Teva Pharmaceutical Industries Ltd. as part of the Israeli National Network of Excellence in Neuroscience (NNE), and Minna-James-Heineman Stiftung through Minerva. The research leading to these results has received funding to E.H. from the European Research Council under the European Union’s Seventh Framework Programme (FP7/2007–2013)/ERC grant agreement number 617351; the Israel Science Foundation; the ALS-Therapy Alliance; the AFM Telethon (20576 to E.H.); the Motor Neuron Disease Association (United Kingdom); the Thierry Latran Foundation for ALS research; the ERA-Net for Research Programmes on Rare Diseases (FP7); A. Alfred Taubman through IsrALS, Yeda-Sela, Yeda-CEO, Israel Ministry of Trade and Industry; Y. Leon Benoziyo Institute for Molecular Medicine, Kekst Family Institute for Medical Genetics; David and Fela Shapell Family Center for Genetic Disorders Research; Crown Human Genome Center; Nathan, Shirley, Philip and Charlene Vener New Scientist Fund; Julius and Ray Charlestein Foundation; Fraida Foundation; Wolfson Family Charitable Trust; Adelis Foundation; Merck (United Kingdom); M. Halphen; and Estates of F. Sherr, L. Asseof, L. Fulop, E. and J. Moravitz. S.M.K.F. was supported by the ALS Canada Tim E. Noël Postdoctoral Fellowship. C.E. and I.M. were supported by scholarship from Teva Pharmaceutical Industries Ltd. as part of the Israeli National Network of Excellence in Neuroscience (NNE). Work at the C. Gross laboratory was supported by NIH grant R01NS092705 (to C.G.). Work at the T. M. Miller laboratory was supported by grants from Project5 for ALS, Target ALS, the National Institute of Neurological Disorders and Stroke (R01NS078398 to T.M.M. and F31NS092340 to M.L.H. and T.M.M.), the Robert Packard Center for ALS Research, the University of Missouri Spinal Cord Injury/Disease Research Program, and the Hope Center for Neurological Disorders. Work at the M. B. Harms laboratory was funded by grants from ALS Association and Biogen. R.H.B. was funded by ALS Association, ALS Finding a Cure, Angel Fund, ALS-One, Cellucci Fund, and NIH grants (R01 NS104022, R01 NS073873, and NS111990-01 to R.H.B.). A.A.-C. was supported through the following funding organizations under the aegis of JPND [; United Kingdom, Medical Research Council (MR/L501529/1; MR/R024804/1)] and through the Motor Neurone Disease Association. This study represents independent research part funded by the National Institute for Health Research (NIHR) Biomedical Research Centre at the South London and Maudsley NHS Foundation Trust and King’s College London. A.N.B. was supported by the Suna and Inan Kirac Foundation. J.E.L. was supported by the NIH/NINDS (R01 NS073873). H.P. was supported by a grant from the ALS Association. Author contributions: I.R. and C.E. led the project. I.R. and C.E. contributed to research conception, design, and interpretations and wrote the manuscript with E.H. C.E., T.O., M.L.H., T.M.M., C.A.M.M., M.B.H., R.H.B., S.M.K.F., and K.R.v.E. contributed to human genetics analysis of rare miR-218 variants in human patients with ALS. C.E., G.H., I.M., and T.O. performed NGS studies. N.R. performed reporter assays. S.D.-G. performed human histology studies. J.K.R. and G.C. performed protein quantification by Western blots. J.L., R.L., A.S., R.R., B.T., A.D.M., and E.Y. helped in performing the research. G.B. and E.A. performed hiPSC experiments. J.M.R. and T.M. provided human autopsy material. M.A.B., N.M.B., and K.A.L. developed siRNA oligos. S.S., O.Y., C.G., and S.L.P. provided reagents and expertise. E.H. conceived and supervised the study and wrote the manuscript. All coauthors provided approval of the manuscript. Competing interests: M.A.B., N.M.B., and K.A.L. are employed by Integrated DNA Technologies Inc. (IDT), which manufactures reagents similar to some described in the manuscript. M.A.B. owns equity in DHR, the parent company of IDT. T.M.M. holds licensing agreement with Ionis Pharmaceuticals and with C2N, and he is on the Advisory Board of Ionis Pharmaceuticals and Biogen and is consulting to Cytokinetics. T.M.M. and M.L.H. are inventors on patent/patent application (PCT/US2016/019602, now U.S. patent application number 15/553,922) submitted by Washington University that covers methods to detect motor neuron disease using miRNAs and to target motor neuron disease miRNAs. T.M.M. is an inventor on patent/patent application (PCT/US2015/053283, now U.S. patent application number 15/515,909) submitted by Washington University that covers tau kinetic measurements. T.M.M. is an inventor on patent/patent application (PCT/US2013/031500, now U.S. patent application number 16/298,607, with corresponding national stage applications or issued patents in Australia, Canada, Europe, and Japan) that is jointly owned with Ionis Pharmaceuticals that covers methods for modulating tau expression for reducing neurodegenerative syndromes. T.M.M. is an inventor on patent/patent application (issue number 10,273,474) that is jointly owned with Ionis Pharmaceuticals that covers methods for modulating tau expression for reducing seizure and modifying a neurodegenerative syndrome. T.M.M. is an inventor on patent/patent application (61/547,890) submitted by Washington University that covers metabolism of SOD1 in the CSF. I.R. and E.H. are inventors on pending patent family PCT/IL2016/050328 entitled “Methods of treating motor neuron diseases.” The other authors declare that they have no competing interests. Data and materials availability: Gene Expression Omnibus accession number: GSE136409. miR-218-2-5p isotype counting code: Human miR-218 precursors [miRBase v20; (32)] are at Chr5:168195173-168195236 and Chr4:20529922-20529986 of human genome build 19 (hg19). Human genetics data are publicly available from the sequencing consortia: Project MinE, GTAC, ALS Sequencing Consortium, NYGC, and NHLBI’s TOPMed. All other data used for this manuscript are available in the manuscript.

Stay Connected to Science Translational Medicine

Navigate This Article