miR-128 is one of the most abundant and highest enriched miRNA in the adult mouse and human brain ((1, 2) (Fig. S1A). The expression of miR-128 in the mouse brain increases gradually during postnatal development and peaks in adulthood ((3, 4) (Fig. S1B)). miR-128’s expression in diverse brain regions (Fig. S1D) suggests an important role for this miRNA in processes that are common to many neuronal cell-types.

The indication of a potent regulatory role for miR-128 in brain function came from our observation of early-onset fatal epilepsy in mice deficient in miR-128 (Fig. 1A). miR-128 is encoded by two separate genes, miR-128-1 and miR-128-2, on mouse chromosomes 1 and 9 (Fig. S2A, B) or human chromosomes 2 and 3, respectively. In mice, germline miR-128-2 deficiency results in an 80% reduction of miR-128 expression in the forebrain, whereas ablation of the miR-128-1 gene eliminates only 20% of miR-128 (Fig. S2A, B). The profound decline in miR-128 expression levels in miR-128-2^−/− but not miR-128-1^−/− mice is associated with the development of hyperactivity and increased exploration at 4 weeks of age (Fig. 1A, Fig. S2C, D). The juvenile hyperactivity in miR-128-2^−/− mice progresses quickly to severe seizures and death at 2–3 months of age (Fig. 1A, B, movie S1). The lethal impact of miR-128 deficiency in mice can be prevented by treatment with the anticonvulsant drug valproic acid (Fig. 1C), thus demonstrating the causal role of seizures in the animals’ death.

The hyperactivity and fatal epilepsy in miR-128-2 deficient mice reflects the ability of miR-128 to control the excitability of postnatal neurons. Selective inactivation of the miR-128-2 gene in forebrain neurons (Camk2a-cre; miR-128-2^fl/fl) leads to a reduction of miR-128 expression, followed by early onset hyperactivity, seizures, and death, as observed in miR-128-2^−/− mice (Fig. 1B, D, Fig. S3A). Moreover, correction of miR-128 deficiency by ectopic miR-128-2 expression in neurons normalizes motor activity and prevents the seizure-induced death (Fig 1E, Fig. S4A, C).

To gain an understanding of the mechanism that mediates miR-128-dependent control of motor activity, and to avoid interference between phenotypes caused by the loss of miR-128 in diverse neuronal cell-types, we restricted the miR-128-2 deficiency to dopamine responsive neurons that regulate motor behavior in mice and humans. There are two major dopamine responsive Camk2a-expressing neuron types in the mouse forebrain, which have distinct contributions to motor activity (5). While activation of the dopamine 1 receptor expressing neurons (D1-neurons) increases locomotion, activation of dopamine 2 receptor expressing neurons (D2-neurons) reduces locomotion in mice (6). We found that miR-128 deficiency in D1-neurons (Drd1a-cre;miR-128-2^fl/fl), but not in D2-neurons (A2a-cre;miR-128-2^fl/fl), leads to juvenile hyperactivity followed by lethal seizures at around 5 months of age (Fig. 1F, Fig. S3B, C).

To identify miR-128 targets that are responsible for the abnormal motor activity, we analyzed mRNAs associated with the RNA induced silencing complex (RISC) in adult neurons in vivo. The RISC-bound mRNAs represent the pool of cellular mRNAs that become a subject of miRNA-mediated suppression (7). We used mice that express the epitope-tagged RISC component Argonaute 2 (Ago2) (8) in Camk2a-neurons (Fig. S5A). Immunoprecipitation of Ago2 (HITS-CLIP, (9)) from the forebrain of these mice yielded the neuron-specific RISC-associated mRNAs (Fig. S5A, B). The perfect base pairing of at least six nucleotides between the miRNA seed sequence and the 3′untranslated region (3′UTR) of the RISC-associated mRNAs (10) was considered to be the minimal requirement for any potential miRNA-mediated mRNA suppression (Fig. S5B, C). Using these criteria, we found that the miR-128 seed target sequence (ACUGUG) is the most represented hexamer among all RISC-associated mRNAs (Fig. S5C, Table S1)), and identified a total of 1061 potential miR-128 target mRNAs in adult neurons (Table S2).

We investigated these miR-128 target genes by analyzing their expression in neurons deficient for miR-128. We expected that mRNA transcripts that are targeted directly by miR-128 in neurons would show an increased in mRNA expression and subsequent ribosome association in the miR-128 deficient cells. We reasoned that the relative homogeneity of the D1-neuron population might provide the most accurate assessment of miR-128-dependent target genes that are responsible for controlling motor activity. The impact of miR-128 deficiency on mRNA expression was evaluated by D1 cell-type specific Translating Ribosome Affinity Purification (TRAP) in mice (11). The TRAP approach allows a direct comparison between ribosome-associated mRNAs from wild-type and miR-128-deficient D1-neurons in vivo (Fig. S6A). Using Sylamer analysis (12), we confirmed the expected enrichment of potential miR-128 binding sites among the most upregulated genes in miR-128 deficient D1-neurons (Fig. S6B). We found that the deficiency of miR-128 in D1-neurons results in a significant up-regulation of 154 of the predicted RISC-associated miR-128 target genes (Fig. 2A, Table S3). The fact that only ~15% of the potential RISC-associated miR-128 targets display increased expression is likely to reflect the known redundancy among miRNAs. Many mRNAs are regulated by more than one miRNA (13, 14) thus limiting the actual impact of individual miRNA deficiency on the expression of miRNA targets in vivo.

Bioinformatic network and pathway analyses of the miR-128 target genes indicates the ability of miR-128 to affect molecular processes that are intrinsically linked to the regulation of neuronal excitability and motor behavior in mice and humans (Fig. 2B). In particular, miR-128 regulates the expression of numerous ion channels and transporters, as well as genes that contribute to neurotransmitter-driven neuronal excitability and motor activity (Fig. 2B, Table S3, S4). Several of these genes are linked to epilepsy in humans, some of which, including the neurotransmitter GABA transporter Slc6a1, the high affinity glutamate receptor Slc1a1, the voltage gated sodium channels Scn2b and Scn4b, the voltage-dependent calcium channels Cacna2d3 and Cagn2, as well as the carbonic anhydrase Car7, are potential targets of clinically approved anti-seizure drugs (Table S3, S4) (15). The high abundance of extracellular signal regulated kinase (ERK1/2) signaling network components among the miR-128 targets underscores the potential of this miRNA to control signaling processes associated with neuronal excitability (Fig. 2B). Moreover, many of the neuronal signaling proteins and channels that we identified as direct miR-128 target genes are involved in the regulation of upstream signaling events, which can affect ERK activity (Tables S3, S4). While ERK1 and ERK2 are not directly targeted by miR-128, the ERK network appears to be at the center of the miR-128-controlled signaling circuit in neurons. The protein expression levels of potent ERK network regulators, which are directly targeted by miR-128, such as Pea15a (16), D4Ertd22e/Szrd1 (17), and the TARPP protein that is encoded by the long splice variant of the miR-128-2 host gene Arpp21 (18, 19), are increased in the striatum of mice with a D1-neuron specific deficiency in miR-128 (Fig. 2C, S7). Furthermore, mice with a D1-neuron specific deficiency of miR-128-2 display an increase in ERK2 activation as compared to their littermate controls (Fig. 2D). Notably, only ERK2, but not ERK1, displays increased phosphorylation (Fig. 2D). Deficiency of miR-128 in D1-neurons appears to specifically activate ERK2 phosphorylation, without affecting the activation of other MAP kinase pathways components such as the stress-activated protein kinase/Jun-amino-terminal kinase (SAPK/JNK) or protein kinase B (AKT) (Fig. S8). Electrophysiological studies in striatal slices from Drd1a-cre; miR-128-2^fl/fl mice revealed an increase in D1-neuron excitability. The miR-128 deficient D1-neurons show normal membrane excitability at the soma (Fig. S9A), but display enhanced dendritic excitability (Fig. 3A) as well as a ~20% increase of functional dendritic spines (Fig. 3B, S9B). These findings are consistent with a critical role of the ERK2 network in neuronal excitability and synaptic plasticity (20, 21).

Enhanced ERK2 activation is linked to increased motor activity and seizures in mice (22–24). The hyper-activation of ERK2 and concomitant increase in D1-neuron sensitivity to dopamine occurs also during Parkinson-like disease in mice caused by chemically induced depletion of dopamine in the mouse striatum (25–27). The reduced levels of dopamine and concurrent increase of D1-neuron sensitivity result in hyper-responsiveness to the motor activity-inducing effects of dopamine (26–28). In humans, the D1-neuron hyper-responsiveness is one of the major causes of dyskinesia, a side effect of L-Dopa treatment in Parkinson’s disease (25–27).

We found that miR-128 deficiency in striatal D1-neurons mimics the hypersensitivity of D1-neurons in mice suffering from Parkinson’s-like syndrome. The deficiency of miR-128 in D1-neurons enhances motor activity in response to Drd1-specific agonist treatment in mice (Fig. 3C). The D1-neuron hyper-responsiveness to the Drd1-agonist is also associated with an increase in ERK2 phosphorylation in the striatum of Drd1a-cre; miR-128-2^fl/fl mice (Fig. 3D). The increase in dopamine sensitivity and enhanced ERK2 activation in mice with Parkinson’s-like disease are accompanied by increased expression of dopamine-induced immediate early genes (IEG) in D1-neurons (25–27). Similarly, Drd1-agonist treatment enhances IEG expression in miR-128 deficient D1-neurons as compared to the D1-neurons of control mice (Fig. 3E). The increased locomotor activity characteristic of Drd1a-cre; miR128-2^fl/fl mice was normalized by pharmacological inhibition of the mitogen-activated protein kinase kinase MEK1, a major activator of ERK2 in neurons. In vivo administered MEK1-specific inhibitor SL327 does not affect motor activity in wild type mice (22) but does normalize ERK2 phosphorylation and motor activity in the mutant mice (Fig. 4A). In turn, overexpression of miR-128 in Camk2a-neurons is associated with reduced ERK2 activation (Fig. S10A) and decreased motor activity (Fig. S4B) in mice. The effect of increased miR-128 expression in adult neurons protects mice against abnormal motor activities associated with chemically-induced Parkinson’s disease (Fig. 4B, S10B) and seizures (Fig. 4C).

In summary, we have identified miR-128 as a modulator of signaling pathways that control neuronal excitability and motor activity in mice. The human miR-128-2 gene on chromosome 3p lies within a region that has been linked to idiopathic generalized epilepsy (29, 30). It is tempting to speculate that changes in miR-128 or miR-128 target gene expression could be a potential cause of increased neuronal excitability and epilepsy in humans. Our understanding of miR-128’s role in neuronal signaling could prove advantageous in the design of novel therapeutics for epilepsy and motor disorders.