Dynamic changes in murine forebrain miR-211 expression associate with cholinergic imbalances and epileptiform activity

Significance

Acute traumatic stress increases the sensitivity to develop epileptic seizures in certain people. It is therefore important to discover physiological mechanisms that avoid epilepsy. To test if rapidly inducible microRNAs (miRs) could mediate such protection, we combined mouse engineering, RNA sequencing, electric recording of brain activity, and learning tests. We discovered that miR-211, originating from an epilepsy-related genomic locus, may be involved, and therefore engineered mice produce a drug-suppressible excess of brain miR-211. In these mice, suppressing miR-211 excess to the original expression levels in normal brains led to electrically recorded epilepsy and hypersensitivity to epilepsy-inducing compounds; it also modified acetylcholine receptor composition. The functional impact of miR-211 dynamics on seizure threshold may enable future development of miR-211–directed therapeutics.

Abstract

Epilepsy is a common neurological disease, manifested in unprovoked recurrent seizures. Epileptogenesis may develop due to genetic or pharmacological origins or following injury, but it remains unclear how the unaffected brain escapes this susceptibility to seizures. Here, we report that dynamic changes in forebrain microRNA (miR)-211 in the mouse brain shift the threshold for spontaneous and pharmacologically induced seizures alongside changes in the cholinergic pathway genes, implicating this miR in the avoidance of seizures. We identified miR-211 as a putative attenuator of cholinergic-mediated seizures by intersecting forebrain miR profiles that were Argonaute precipitated, synaptic vesicle target enriched, or differentially expressed under pilocarpine-induced seizures, and validated TGFBR2 and the nicotinic antiinflammatory acetylcholine receptor nAChRa7 as murine and human miR-211 targets, respectively. To explore the link between miR-211 and epilepsy, we engineered dTg-211 mice with doxycycline-suppressible forebrain overexpression of miR-211. These mice reacted to doxycycline exposure by spontaneous electrocorticography-documented nonconvulsive seizures, accompanied by forebrain accumulation of the convulsive seizures mediating miR-134. RNA sequencing demonstrated in doxycycline-treated dTg-211 cortices overrepresentation of synaptic activity, Ca²⁺ transmembrane transport, TGFBR2 signaling, and cholinergic synapse pathways. Additionally, a cholinergic dysregulated mouse model overexpressing a miR refractory acetylcholinesterase-R splice variant showed a parallel propensity for convulsions, miR-211 decreases, and miR-134 elevation. Our findings demonstrate that in mice, dynamic miR-211 decreases induce hypersynchronization and nonconvulsive and convulsive seizures, accompanied by expression changes in cholinergic and TGFBR2 pathways as well as in miR-134. Realizing the importance of miR-211 dynamics opens new venues for translational diagnosis of and interference with epilepsy.

Epileptic seizures are generated when excitation/inhibition imbalances or abnormal afferent signals to a brain region lead to excessive or parallel nonspecific neuronal activation (1). Epilepsy has a wide range of manifestations, from brief loss of consciousness to overt convulsions. All mammalian brains, including healthy human brains, are prone to the generation of seizures, and diverse brain insults (ischemic, traumatic, infectious) reduce the threshold to seizures (2, 3). However, why some brains undergo epileptogenesis and develop recurrent unprovoked seizures following insults while others escape this pathophysiology remains unknown (4). Specifically, genes whose modulation mitigates these processes, and the neurotransmission and/or growth factor pathways involved, remain largely unexplored.

Recent reports point at microRNAs (miRs) as functional regulators of epileptogenesis. MiRs are small noncoding RNA molecules that regulate the expression levels of most protein coding genes in mammals and may orchestrate whole transcriptional pathways (5). Recent studies on status epilepticus (SE) animal models of epilepsy identified seizure-characteristic miR changes (6). Also, different miRs, including miR-132 (7), miR-34a (8), and miR-146a (9) emerged as actively involved in the pathogenesis of epilepsy. Of growing interest is miR-134, whose silencing exerts protective effects both in reducing the in vivo severity of SE (10, 11) and in in vitro-cultured hippocampal neurons (12), accompanied by numerous links to human epilepsy (13).

Disrupted ACh signaling and malfunctioning of the TGF-β pathway may both lead to excessive neuronal activation and seizures, and underlie some cases of epilepsy (14, 15). However, the molecular mechanisms linking these two pathways with epileptogenesis are only partially known. Disrupted ACh signaling leads to seizures due to the modulatory function of ACh over mammalian brain neurotransmitters such as glutamate or GABA, and affects synaptic transmission in large cortical networks. Cholinergic receptors include muscarinic metabotropic receptors, which function via G proteins, and nicotinic cation-channel receptors, which are exclusively excitatory (17); modulating the action of both receptor types results in seizure activity (18). In autosomal dominant nocturnal frontal lobe epilepsy, missense mutations in either the nAChRα4 or β2 nicotinic receptor subunits interrupt cholinergic regulation, leading to seizures (19). Also, focal hypercholinergic activity suffices as an initial event in generating network hyperexcitability and seizures (20). Additionally, hippocampal induction of hyperexcitability by cholinesterase inhibitors induces epileptogenesis in rodent models of epilepsy, and injecting the cholinergic agonist pilocarpine initiates epileptic events, seizures, or full SE (21). Nevertheless, mechanisms underlying the control of such modulatory changes and avoidance of epileptogenesis in the healthy brain are still sought after.

Cholinergic-mediated control of excitability may also target nonneuronal components of the neurovascular unit. ACh modulates inflammatory signaling in both the periphery and the brain, which together with blood–brain barrier integrity may critically control network excitability and seizure threshold. For example, brain insults such as trauma may lead to increased permeability of the microvasculature to serum albumin, which then binds TGFβR2 (TGFBR2) receptors on astrocytes, inducing a chain of epileptogenic events. Such events include activation and phosphorylation of Smad2 and downstream transcriptional effects, entailing neuroinflammation, excitatory synaptogenesis, and reduced neuronal inhibitory ques, all leading to chronic epilepsy (22). Concordantly, local or systemic application of losartan, a potential TGF-β–signaling blocker, can reduce albumin-derived TGF-β–signaling and thus block epileptogenesis (23).

Numerous genetic epileptic syndromes are hallmarked by cholinergic dysfunctions that control neuronal excitability and seizure threshold (18, 20). Therefore, we predicted that miR modulators of cholinergic activities may avoid neuronal hypersynchronization by performing coordinated surveillance over signaling networks that keep synchronized firing in check, and sought miRs that may play functional role(s) in inducing and/or suppressing seizure-related cholinergic signaling by affecting synaptic targets. For this purpose, we generated and tested transgenic mouse models, used continuous electrocorticography recordings to detect changes in neocortical activity, and used RNA sequencing (RNA-seq), bioinformatics analyses, and corresponding validations, all of which identified miR-211 as an attenuator of coregulated cholinergic and TGFBR2-associated seizures.

Results

Neuronal miR-211 Controls Cholinergic Synapse Transcripts in the Epileptogenic Brain.

To perform a nonbiased search for neuronal miRs regulating synaptic processes and responding to cholinergic seizure-related cues, we intersected publicly available transcriptional profiles of miRs regulating synaptic vesicle transcripts (24), overrepresented in healthy forebrain immune precipitates of neuronal Argonaute 2 (25) and differentially expressed following pilocarpine injection, which induces cholinergic-mediated seizures (26). Of the three miRs that emerged in all three profiles (miR-211-5p, -218-5p, and -27a-3p; Fig. 1A), we experimentally observed hippocampal miR-211 to be down-regulated in acute pilocarpine-injected model mice, compared with saline-injected controls (48 h following injection; Fig. 1B).

Fig. 1.

Identifying MiR-211 as a synaptic candidate associated with cholinergic signaling-induced seizures. (A) Three candidate miRs (miR-211, -218, and -27a) emerged by intersecting rodent miRs whose levels modify following exposure to the cholinergic facilitator pilocarpine (145 miRs) (6); interact with the RNA-induced silencing complex (RISC) protein Argonaute 2 (AGO2) in CamK2a-expressing cells (83 miRs) (25); and target synaptic vesicle transcripts (94 miRs) (75), predicting involvement in cholinergic-related epileptic seizures. (B) qRT-PCR measurements show mmu-miR-211 decline in hippocampal RNA 24 h following exposure to pilocarpine. (C) Human MiR-211, as well as its in silico target, nicotinic nAChRα7, and five other genes localize to a 15q13.3 chromosomal region where heterozygote deletions entail cognitive impairments with recurrent seizures. (D) The seed domain of hsa-miR-211-5p shows sequence complementarity with the inflammation-regulating nicotinic nAChRα7. (E) Luciferase assay validated direct targeting by miR-211 of nAChRα7 in human embryonic kidney cells. Results were considered significant at *P < 0.05, ***P < 0.001, after correction for multiple testing when applicable.

Notably, miR-211 is a conserved intragenic miR located within the TRPM1 calcium channel gene (5), itself within the 15q13.3 locus where heterozygote microdeletions [Online Mendelian Inheritance in Man (OMIM) no. 612001] associate with mental retardation and recurrent epileptic seizures (27, 28), and homozygous deletions associate with severe neurodevelopmental problems, including epileptic encephalopathy (29). In proximity to the TRPM1 gene and within the 15q13.3 locus is the nicotinic receptor nAChRα7 (30) (Fig. 1C), a gain-of-function mutation that results in nicotine-induced seizures (31). In silico miR–target interaction analysis (32, 33) predicted that Homo sapiens (hsa)-miR-211 targets nAChRα7 via a 7-mer miR response element (Fig. 1D). Validating this prediction, miR-211 expression in cultured HEK293 cells directly down-regulated a luciferase expression construct containing the nAChRα7 3′ UTR (Fig. 1E); taken together, this indicated that miR-211 is both affected by and may control cholinergic transmission in an epilepsy-relevant manner.

DTg-211 Mice Present Spontaneous Emergence of Nonconvulsive Seizures upon miR-211 Suppression.

In the retina, the half-life of miR-211 is regulated via active degradation by light and neuronal activity (34). Therefore, using a constitutive, single time point expression system would have missed the dynamic dimension of its in vivo demise. To explore the in vivo impact of miR-211 decline on cholinergic signaling and seizure susceptibility, we used a double-transgenic tetracycline-repressible (Tet-Off) system, where engineered mice exclusively express miR-211 from the TRE insertion in CamK2a–expressing cells (i.e., forebrain neurons) and only in the absence of doxycycline (Dox), allowing temporal follow-up of the effects of introducing and removing overexpression. To overcome the possibility of phenotypic consequences due to the expression of the Tet activator, rather than miR-211 in CamK2a-expressing cells, we used littermate CamK mice (which express the transgenic inducer in the same cells and contexts, but with no miR-211 overexpression) as controls throughout the study. Briefly, we cloned the murine premiR-211 sequence under control of the tetracycline regulatory region (using a pTRE_Tight plasmid system), and established TRE-miR-211 mouse lines. These were crossed with Tet-Off CaMK2-transactivator (tTA) mice (referred to as CamK mice; see Table S1 and Methods). Progeny double-transgenic mice (dTg-211), unlike littermate CamK controls, overexpressed miR-211 in forebrain tissues, but only in the absence of Dox (scheme in Fig. 2A). When administered with Dox in drinking water for 6–8 wk, dTg-211 mice exhibited normally low forebrain miR-211 expression levels, indistinguishable from those in control mice (Fig. 2B). In contrast, Dox withdrawal induced miR-211 accumulation in the frontal cortex, hippocampus, and striatum, but not in the cerebellum (Fig. 2B), essentially as reported for other CamK:Tet mouse lines (35). When readministered with Dox, adult (2-mo-old) dTg-211 mice showed miR-211 decline to basal levels in the frontal cortex within 4 d (Fig. 2C), supporting the applicability of these mice for evaluating temporal attributes of miR-211 decreases in the forebrain.

Table S1.

Primers for PCR-based genotyping

No.	Gene	Primer	Sequence
1	PTRE-miR-211	PTRE-miR-211-F	GTGTACGGTGGGAGGCCTAT
2	PTRE-miR-212	PTRE-miR-211-R	GGGAGGTGTGGGAGGTTTT
3	CamK:tTA	oIMR8746	CGCTGTGGGGCATTTTACTTTAG
4	CamK:tTA	oIMR8747	CATGTCCAGATCGAAATCGTC

Fig. 2.

dTg-211 mice develop spontaneous nonconvulsive seizures following Dox-induced reduction of forebrain miR-211 excess. (A) DTg-211 mice carry the CamK2a promoter, followed by a tTA coding sequence and a pTRE-transgene inducing Dox-suppressible expression of mmu-miR-211 in forebrain neurons. (B) MiR-211 overexpression in the mouse forebrain but not cerebellum is Dox suppressible. Expression normalized to CamK controls. (C) Dox-suppressed miR-211 levels decline to basal levels within days. (D) DTg-211 mice were administered Dox before and after birth, preventing transgene overexpression during development. ECoG recordings in dTg-211, but not control, mice showed synchronous neuronal cortical activity after Dox treatment, parallel to declined miR-211 levels. (E) DTg-211 mice presented ECoG-recorded seizures exclusively after Dox administration. (F) ECoG plots showing number of seizures per day in single dTg-211 mice and controls (red, blue). Dashed gray line marks initiation of Dox administration. (G) Representative ECoG recording plot shows a seizure of a Dox-exposed dTg-211 mouse; corresponding heat map shows representative higher-power seizure of low-frequency oscillations (∼5 Hz) at the same time window. (H) Magnification of a single event (marked by asterisk in G), presenting an enlarged section of the seizure activity, with spike and wave form. Results were considered significant at *P < 0.05, **P < 0.01, ***P < 0.001, after correction for multiple testing when applicable. ns, not significant.

To directly test if miR-211 decline affects cortical neuronal hyperexcitability, we performed electrocorticography (ECoG) measurements on the mice before and following Dox-mediated miR-211 reduction. In this paradigm, both CamK controls and dTg-211 mice received Dox from conception, to prevent developmental effects due to miR-211 overexpression; Dox was later withdrawn to allow miR-211 accumulation in the forebrain. At 12 wk of age, ECoG recordings were initiated and 4–6 d after, Dox was readministered, once again reducing the elevated miR-211 levels (Fig. 2 C and D). During the subsequent 6 d, ECoG recordings demonstrated spontaneous nonconvulsive seizures in six of eight dTg-211 mice but in none of nine control CamK mice receiving similar Dox treatment (Fig. 2E). Seizures mostly initiated by the third or fourth day after Dox administration (Fig. 2F), parallel to the decline in miR-211 overexpression (Fig. 2C). Identified seizures showed a pattern of low frequency (∼5 Hz) and sharp activity (Fig. 2 G and H), without observed motor convulsions. Notably, similarly slow hypersynchronous cortical activity is reminiscent of several human syndromes manifested in epilepsy (36), altogether raising the possibility that forebrain miR-211 elevation may be protective against spontaneous nonconvulsive seizures, whereas its reduction induces them.

Forebrain miR-211 Suppression Exacerbates Long-Lasting Pentylenetetrazole-Induced Seizures and TGF-β Signaling.

The spontaneous seizures induced following Dox-suppression in dTg-211 mice suggested that miR-211 decline may also entail a sustained susceptibility to convulsions. To test such an effect, we treated dTg-211 mice with Dox for 6 d, and challenged them and matched controls with the seizure-provoking pentylenetetrazole (PTZ) agent 5 d after Dox has been removed, when the levels of cortical miR-211 are again elevated (Fig. 3A). Tested dTg-211 mice showed increased susceptibility to convulsive stimuli (Fig. S1A), and ECoG recording of consequent epileptiform spikes in dTg-211 mice showed an increase in spike counts and total numbers (Fig. 3 B and C), as well as seizure numbers and latency to first spike and to first seizure compared with CamK control mice (Fig. 3 D–F), where miR-211 levels were conspicuously lower (Fig. 3G), all pointing to long-lasting epileptiform impact of Dox in dTg-211 mice.

Fig. 3.

Dox-treated dTg-211 mice show sustained susceptibility to PTZ-induced convulsions alongside TGFBR-associated gene changes in RNA sequence. (A) Scheme of PTZ injection 4 d after 5-d Dox administration, aimed to examine long-term susceptibility to this convulsant. See Fig. S3 for increased manual convulsions index scores in dTg-211 mice. (B) ECoG recording shows larger spikes per minute counts, reflecting seizure susceptibility in PTZ-exposed dTg-211 mice compared with CamK controls. (C) Number of seizures. (D) Latency to first spike. (E) Number of seizure events by neuronal networks analysis. (F) Latency to first seizure. (G) DTg-211 mice regained miR-211 overexpression after Dox removal, at the time of PTZ test. (H) Luciferase validation tests of miR-211 targeting of the murine TGFBR2 3′-UTR but not a control sequence. (I) Reduced TGFBR2 protein concentration (twofold) in dTg-211 frontal cortex (ELISA, n = 7 + 7, P < 0.001). (J) Increased TGFBR2 mRNA levels following Dox administration. (K) Fold-change volcano plot differences for dTg-211 with/without Dox (Right) compared with dTg-211/CamK brains (Left). Dots represent genes, with positive or negative twofold change (orange), passing cutoff threshold for significance (red), both (green), or unmodified (black). (L) ECDF plots show differential expression (P values) following Dox of reduced (orange) but not elevated genes (green) in dTg-211 cortices or in all genes (gray). (M) Cortical genes up-regulated in dTg-211 are reduced (red) following Dox compared with (N) all genes. (O) Per-gene fold changes following Dox for TGF-β–signaling genes modified 12 h following status epilepticus (38). Results were considered significant at *P < 0.05, **P < 0.01, ***P < 0.001, after correction for multiple testing when applicable. ns, not significant.

Fig. S1.

dTg-211 mice with reduced forebrain miR-211 levels show higher susceptibility to PTZ-mediated seizures. (A) dTg-211 mice showed higher convulsive scores following PTZ injection, after a Dox administration paradigm. (B) Sequence complementarity of the murine TGFBR2 3′-UTR and the seed sequence of mmu-miR-211-5p, in a conserved 8-mer binding site (TargetScan, P_CT = 0.46). (C) Frontal cortex samples of non–Dox-treated dTg-211 showed reduction of TGFBR2 transcript levels by RT-qPCR. ***P < 0.001.

Given the reported role of TGFβR signaling in epileptogenesis, we next examined if the TGFBR pathway genes that change following status epilepticus were modified in the epilepsy-susceptible dTg-211 mice following Dox. The 3′-UTR of the murine TGFBR2 gene shows seed sequence complementarity with the mature Mus musuculus (mmu)-miR-211 (Fig. S1B), and an in vitro psiCHECK assay showed direct down-regulation of the murine TGFBR2 3′-UTR reporter by mmu-miR-211, validating this miR–target link (Fig. 3H). Correspondingly, non–Dox-treated frontal dTg-211 cortices showed twofold lower TGFBR2 transcript levels compared with controls (Fig. S2B), alongside ∼40% reduced TGFBR2 protein (Fig. 3I; ELISA). Reciprocally, administration of Dox induced a step-wise fourfold increase in TGFBR2 mRNA within 4 d (Fig. 3J). To explore if TGFBR2 pathway genes are globally changed, we turned to unbiased RNA-seq of dTg-211 cortical tissue RNAs (without and with Dox suppression of miR-211 overexpression) compared with matched control tissues. The cDNA libraries showed overall similar sequencing reads distribution across expression level and interrelated tag-wise normalized variance and expression levels (Fig. S2 A and B). Nevertheless, comparing dTg-211 brains before and after 5 d of Dox administration showed substantially higher numbers of differentially expressed genes than comparing naïve dTg-211 brains to CamK controls (Fig. 3K), suggesting that Dox-induced suppression of miR-211 overexpression may entail an extensive physiological change. Notably, grouped miR-211 target transcripts, as predicted in silico by the TargetScan algorithm (37), showed significant, albeit mild, increases and decreases following Dox administration [empirical cumulative distribution function (ECDF) plots] (Fig. S3). We conclude that the bulk of transcriptome changes induced by miR-211 perturbations occurred in secondarily affected transcripts and that the transcriptional footprint of miR-211 reduction was greater than that of its sustained overexpression.

Fig. S2.

RNA sequencing quality checks. (A) Illumina-compatible cDNA libraries show overall similar sequencing depth and distribution across expression level. (B) Tag-wise normalized variance predictably correlated to expression levels.

Fig. S3.

Forebrain miR-211 targets are preferentially modulated under miR-211 decline. Target transcripts of miR-211, as predicted in silico by the TargetScan algorithm (37) and grouped by MRE site type, showed mild albeit significant modification under miR-211 decline, with increased probability depending on site length following Dox administration; ECDF plot.

Predictably, genes that were down-regulated in dTg-211 forebrains compared with controls were more commonly differentially expressed following Dox administration, in contrast to all genes or all up-regulated genes (Fig. 3L; ECDF, Kolmogorov–Smirnov test, P < 0.001). Also, those genes that were down-regulated in dTg-211 brains showed preferred post-Dox up-regulation (Fig. 3 M and N). Numerous TGF-β pathway genes that were modified 12 h following SE (38) were also changed following Dox, either by up- or down-regulation (Fig. 3O), including the Myc protooncogene protein (MyC), the Chordin (chrd), inhibitor of DNA binding 2, HLH protein (id2), and SMAD family members 1 and 9 (smad1, smad9). Together, these findings suggested that the Dox-induced release of TGFBR2 from miR-211–mediated suppression impacted forebrain TGF-β signaling. Murine miR-211 thus emerged as a modulator of epilepsy, whose forebrain suppression induces long-lasting hyperexcitability and epileptiform activity alongside TGF-β pathway alterations.

MiR-211 Decline Induces Transcriptome Changes of Endothelial, Synaptic, and Cholinergic Functions.

To explore the global transcriptional changes following miR-211 suppression, and test if they relate to specific brain cell types, we compared dTg-211 brains’ transcripts to controls. Specifically, we tested highly expressed genes characteristic of neurons, astrocytes, oligodendrocytes and their progenitor cells (OPCs), microglia, and endothelial cells (39) (Fig. 4 A and B). None of the cell type marker groups showed changes in dTg-211 brains compared with controls. However, comparing dTg-211 transcripts in brains with and without Dox (Fig. 4C) demonstrated that 19 of 21 endothelial cell markers, but none of the other cell type markers showed an increase following Dox administration (P < 0.05, perturbation analysis; Methods). Thus, miR-211 reaction to Dox appeared to potentiate endothelial gene expression, predicting functional relevance for neurovascular unit activities.

Fig. 4.

MiR suppression in dTg-211 mice alters cell type marker genes and cholinergic receptors, and a cholinergic mouse model shows concordant miR changes alongside increased seizure susceptibility. (A) Experimental setup: CamK:TtA mice bred with Tg-pTRE-211 mice generated dTg-211 mice and littermate CamK controls. Illumina-compatible libraries from frontal cortex RNA of mice before or under doxycycline (color-coded squares) were sequenced. (B) Sustained cell-type marker (39) in dTg brains. (C) Elevated endothelial marker genes following Dox. (D) Modified muscarinic and nicotinic cholinergic receptors in Dox-treated dTg-211 brains. (E) Scheme of cholinergic receptors and regulators (shown in D) in brain cholinergic synapses. Note Dox-induced down-regulation of cholinergic receptors suppressing synaptic transmission: CHRM2 and CHRM4 (m2 and m4); and up-regulation of facilitators CHRM5 (m5), CHRNA5, and CHRNA7 (α5 and α7). (F) MiR-134 up-regulation in dTg-211 mice following Dox administration parallels the time frame of seizure induction in this model. (G) Scheme of the synaptic and nonsynaptic AChE transcript variants and corresponding protein forms. (H) Mice overexpressing the nonsynaptic cholinergic enzyme AChE-R (TgR) (76) show higher propensity with (I) shorter latency for status epilepticus event following pilocarpine injection, alongside (J) miR-211 reduction and (K) miR-134 elevation in prefrontal cortex (PFC) and hippocampus (Hipp) of TgR mice. Results were considered significant at *P < 0.05, **P < 0.01, ***P < 0.001, after correction for multiple testing when applicable.

Next, we searched for Dox-induced changes in the expression of cholinergic receptor genes. The excitatory muscarinic ACh receptor-5 (30), a positive effector of cholinergic synaptic transmission (31), was elevated by fourfold (Fig. 4D). Likewise, the excitatory nAChRα-1 neuronal nicotinic receptor and α-5 nicotinic receptor (40, 41), and the ionotropic α-7 nicotinic receptor, responsible for post- and presynaptic excitation (17) and blocker of inflammation (42), were all elevated. In contrast, the metabotropic muscarinic ACh receptors 4 and 2 (mAChR4, mAChR2) were both twofold reduced following Dox administration (Fig. 4D and scheme in Fig. 4E). MAChR4 is located on both pre- and postsynaptic sites in brain cholinergic synapses, and exerts inhibitory effects on synaptic firing (43) with a role in locomotion. Additionally, we noted increases in butyrylcholinesterase (BChE; Fig. S4B), which hydrolyzes ACh in the brain alongside AChE and is elevated in AD brains (44). In contrast, we noted decrease of ATCAY/BNIP-H (Fig. S4 C and D), an ataxia-related brain-specific scaffold protein, which was recently found to recruit choline acetyltransferase (ChAT) to neurite terminals, and promote cholinergic signaling (45, 46). Furthermore, within 4 d following Dox administration, dTg-211 mice presented fourfold increases in the forebrain levels of miR-134 (Fig. 4F), known to be causally involved with the induction of convulsive seizures (7, 10, 11, 47).

Fig. S4.

Bidirectional cholinergic gene changes under forebrain miR-211 decline. (A) Increased alpha-1 neuronal nicotinic receptor (Chrna1) and (B) BChE following Dox administration in the forebrain of dTg-211 mice. (C and D) Decrease of ATCAY/BNIP-H expression with Dox administration (RNA-seq and qRT-PCR, respectively). (E) Muscarinic acetylcholine receptor M1 (mAChR1) is elevated in dTg-211 mice 4 d following Dox administration. Results were considered significant at *P < 0.05, **P < 0.01, ***P < 0.001, after correction for multiple testing when applicable. ns, not significant.

To examine how extrasynaptic cholinergic imbalance would affect miR-211 expression and the risk of epilepsy, we used transgenic AChE-R (TgR) mice overexpressing the soluble, nonsynaptic stress-induced splice variant of AChE from which the 3′-untranslated region (3′-UTR), which contains the miR regulatory element, had been deleted (Fig. 4G, scheme). TgR mice, which constitutively overexpress AChE-R that catalyzes ACh breakdown in extrasynaptic sites and show chronic stress behaviors, are hypersensitized to nicotine administration (48). Intriguingly, these mice also experienced higher susceptibility to seizures, manifested as larger fraction of mice presenting full SE after pilocarpine injection (Fig. 4H); this was accompanied by shorter latency until SE was observed (Fig. 4I; P < 0.05), reduced miR-211 expression in the hippocampus and frontal cortex compared with controls (Fig. 4J), and overexpression of miR-134 in the prefrontal cortex and hippocampus (Fig. 4K), possibly in relation to their hypersynchronous state. Thus, modified cholinergic regulation in TgR mice reciprocally modified forebrain miR-211 and miR-134 levels in a similar fashion as well as exacerbated susceptibility to epileptic seizures.

Dox-Induced Decline of Forebrain miR-211 Impairs Numerous Signaling Pathways.

To test for phenotypic outputs that may contribute to the synergic effects of transcript changes in functionally related proteins, we established protein–protein interaction (PPI) networks (49). We used genes that were both highly (top 3,000 genes) and differentially expressed [false discovery rate (FDR) < 0.01] following Dox to generate a network of PPI-annotated partners defined by 61 core proteins, and that included first neighbor-interacting proteins of 515 edges and 434 nodes (Fig. 5A). Enriched up-regulated and down-regulated genes and pathways within this network included numerous synaptic vesicle cycle genes (e.g., reduced Vamp2, Snap25; Fig. 5B). Intriguingly, 19 of the 61 core proteins within this network related to the cholinergic synapse KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway 04725 (P < 10^-8; Fig. 5C). Also, numerous neuronal-related pathways were either up- or down-regulated (Fig. 5 D and E; see Table S2 for full list of pathways), including calcium-signaling (KEGG 04020) and synaptic vesicle cycle (KEGG 04721) pathways.

Fig. 5.

Protein–protein interaction network of Dox-induced differentially expressed synaptic vesicle and cholinergic genes. (A) Protein–protein interaction-based interconnected network of stringently defined 134 differentially expressed node genes and overall 427 genes. (B) Fold changes ± SEM of the synaptic vesicle cycle pathway genes within the network. (C) Fold changes ± SEM of the cholinergic synapse genes within the network. (D and E) Enriched biological process GO terms for PPI networks for genes differentially expressed following Dox, either down or up. (Fold enrichment, asterisks denote significance P value based on permutation analysis. Results were considered significant at *P < 0.05, **P < 0.01, ***P < 0.001, after correction for multiple testing when applicable.)

Table S2.

Full list of up- or down-regulated KEGG pathways in the PPI network of differentially expressed genes following Dox administration

Pathway	Total	Expected	Hits	Ratio enrichment	FDR	P value
Cholinergic synapse	97	3.01	8	2.7	0.045	0.010
Oocyte meiosis	109	3.38	9	2.7	0.031	0.006
Fc gamma R-mediated phagocytosis	92	2.85	8	2.8	0.035	0.007
Epstein–Barr virus infection	114	3.54	10	2.8	0.015	0.003
Natural killer cell-mediated cytotoxicity	123	3.82	11	2.9	0.009	0.001
Focal adhesion	199	6.17	18	2.9	0.000	0.000
Measles	99	3.07	9	2.9	0.019	0.003
Cell cycle	127	3.94	12	3.0	0.004	0.001
TGF-β–signaling pathway	84	2.61	8	3.1	0.022	0.004
Renal cell carcinoma	61	1.89	6	3.2	0.048	0.011
Leukocyte transendothelial migration	111	3.44	11	3.2	0.004	0.001
Osteoclast differentiation	110	3.41	11	3.2	0.004	0.001
MAPK-signaling pathway	265	8.22	27	3.3	0.000	0.000
Salmonella infection	68	2.11	7	3.3	0.024	0.005
HTLV-I infection	223	6.92	23	3.3	0.000	0.000
Prostate cancer	87	2.70	9	3.3	0.009	0.001
Herpes simplex infection	105	3.26	11	3.4	0.003	0.000
Acute myeloid leukemia	57	1.77	6	3.4	0.037	0.008
Dopaminergic synapse	129	4.00	14	3.5	0.000	0.000
GnRH-signaling pathway	92	2.85	10	3.5	0.004	0.001
Insulin-signaling pathway	137	4.25	15	3.5	0.000	0.000
Pathways in cancer	306	9.49	34	3.6	0.000	0.000
Gap junction	88	2.73	10	3.7	0.003	0.000
Cocaine addiction	43	1.33	5	3.8	0.045	0.010
Notch-signaling pathway	49	1.52	6	3.9	0.020	0.004
Salivary secretion	49	1.52	6	3.9	0.020	0.004
Bacterial invasion of epithelial cells	57	1.77	7	4.0	0.010	0.002
Gastric acid secretion	55	1.71	7	4.1	0.009	0.001
Pancreatic cancer	70	2.17	9	4.1	0.003	0.000
Long-term potentiation	69	2.14	9	4.2	0.003	0.000
Apoptosis	82	2.54	11	4.3	0.000	0.000
Wnt-signaling pathway	147	4.56	20	4.4	0.000	0.000
Endocrine and other factor-regulated calcium reabsorption	44	1.36	6	4.4	0.013	0.002
ErbB-signaling pathway	87	2.70	12	4.4	0.000	0.000
Melanogenesis	100	3.10	14	4.5	0.000	0.000
Neurotrophin-signaling pathway	126	3.91	18	4.6	0.000	0.000
Basal cell carcinoma	47	1.46	7	4.8	0.004	0.001
Endometrial cancer	44	1.36	7	5.1	0.003	0.000
Glioma	66	2.05	11	5.4	0.000	0.000
Viral myocarditis	29	0.90	5	5.6	0.010	0.002
Thyroid cancer	28	0.87	5	5.8	0.010	0.001
Chronic myeloid leukemia	74	2.30	15	6.5	0.000	0.000
Adherens junction	71	2.20	15	6.8	0.000	0.000
Colorectal cancer	49	1.52	11	7.2	0.000	0.000
Amphetamine addiction	64	1.99	15	7.5	0.000	0.000
Synaptic vesicle cycle	17	0.53	4	7.6	0.010	0.002
African trypanosomiasis	20	0.62	5	8.1	0.003	0.000
Arrhythmogenic right ventricular cardiomyopathy (ARVC)	12	0.37	5	13.4	0.000	0.000
Prion diseases	20	0.62	9	14.5	0.000	0.000

A possible drawback for examining transcriptional effects using Tet-Off mice is that Dox administration may itself impact the transcriptional profile. To test for this possibility, we examined if cholinergic synapse and TGF-β pathway genes change in publically available transcriptional data of other Dox administration models (Fig. S5). The outcome profile was unique and nonoverlapping, indicating that miR-211 decline underlines diverse, rapid changes in forebrain cholinergic synapses of dTg-211 mice that are unrelated to the effect of Dox itself. To conclude, miR-211 suppression led to selective transcriptional changes in specific functional groups that control cholinergic synaptic transmission and/or affecting hyperexcitability.

Fig. S5.

MiR-211 decline exerts distinct RNA-seq changes from those of Dox administration. (A) In cultured cells (84) TGF-β pathway genes (red), but not cholinergic synapse genes (green), show down-regulation following Dox administration. (B) In the frontal cortex of dTg-211 mice following Dox administration, genes most differentially expressed in the Du et al. (84) dataset [first (red) and five upper (green) percentiles low P value in ref. 84] show similar distribution of up- and down-regulation as all genes. (C) Genes modified in cultured cells following Dox are unchanged in the forebrain of dTg-211 mice administered with Dox. Shown is a scatterplot of per-gene fold change following Dox in both experiments. Dots represent genes: gray, unmodified; green, modified in the mouse forebrain; blue, modified in cells; red, modified in both. Note very few genes along the diagonal, and few genes to be modified in both experiments.

MiR-211 Excess Associates with Impaired Learning Strategy in dTg-211 Mice.

Cholinergic transmission in the mammalian brain has important roles in learning and memory through numerous circuits feeding into the cortex (17). Concordantly, enrichment analysis using the RNA sequencing data identified functional Dox-induced outputs in the dTg-211 cortical datasets. Gene Ontology (GO) (24) showed 166 genes with log fold change less than 0.5 or greater than 2, FDR-corrected P value < 0.01, and absolute expression levels greater than log cpm of 6.2 (Dataset S1). GO annotation enrichment analysis using PANTHER (49, 50) highlighted mostly neural-related pathways (Fig. 5E; see Table S3 for full list), with top GO annotations consisting of processes responsible for synaptic events (four terms); neurotransmitter functioning (six); higher brain functions (two); transporter activity (three), and learning/memory (Fig. 6A), predicting cognitive miR-211–mediated transcript differences.

Table S3.

List of GO annotation enrichment analysis using PANTHER

GO biological process complete	mmu	No. of genes	Expected no. of genes	Overrepresented	Fold enrichment	Enrichment P value
Regulation of synaptic vesicle priming (GO:0010807)	9	4	0.06	+	>5	0.004
Negative regulation of microtubule polymerization or depolymerization (GO:0031111)	36	5	0.25	+	>5	0.0421
Regulation of microtubule polymerization or depolymerization (GO:0031110)	59	6	0.4	+	>5	0.0282
Regulation of calcium ion transmembrane transport (GO:1903169)	61	6	0.42	+	> 5	0.0341
Regulation of neurotransmitter transport (GO:0051588)	65	6	0.44	+	>5	0.0488
Neurotransmitter secretion (GO:0007269)	88	7	0.6	+	>5	0.0216
Regulation of synaptic plasticity (GO:0048167)	136	10	0.93	+	>5	0.00032
Regulation of exocytosis (GO:0017157)	165	10	1.12	+	>5	0.00187
Regulation of ion transmembrane transporter activity (GO:0032412)	167	10	1.14	+	>5	0.00208
Regulation of transmembrane transporter activity (GO:0022898)	172	10	1.17	+	>5	0.00272
Synaptic transmission (GO:0007268)	320	18	2.18	+	>5	8.03E-08
Regulation of transporter activity (GO:0032409)	182	10	1.24	+	>5	0.00452
Regulation of synapse structure or activity (GO:0050803)	237	13	1.61	+	>5	8.98E-05
Exocytosis (GO:0006887)		11	1.44	+	>5	0.00204
Modulation of synaptic transmission (GO:0050804)	289	14	1.97	+	>5	0.000113
Regulation of protein complex assembly (GO:0043254)	301	14	2.05	+	>5	0.000186
Learning or memory (GO:0007611)	223	10	1.52	+	>5	0.0272
Cognition (GO:0050890)	250	11	1.7	+	>5	0.0105
Regulation of vesicle-mediated transport (GO:0060627)	396	16	2.7	+	>5	0.000125
Single-organism behavior (GO:0044708)	418	16	2.85	+	>5	0.000262
Cell-cell signaling (GO:0007267)	537	20	3.66	+	>5	7.17E-06
Secretion by cell (GO:0032940)	359	13	2.44	+	>5	0.00979
Secretion (GO:0046903)	451	16	3.07	+	>5	0.00073
Behavior (GO:0007610)	564	20	3.84	+	>5	1.64E-05
Positive regulation of secretion (GO:0051047)	376	13	2.56	+	>5	0.0162
Regulation of ion transmembrane transport (GO:0034765)	380	13	2.59	+	>5	0.0182
Regulation of transmembrane transport (GO:0034762)	397	13	2.7	+	4.81	0.0291
Regulation of neuron projection development (GO:0010975)	400	13	2.72	+	4.77	0.0315
Regulation of cell projection organization (GO:0031344)	508	16	3.46	+	4.62	0.00352
Regulation of cellular component biogenesis (GO:0044087)	637	20	4.34	+	4.61	0.000123
Regulation of ion transport (GO:0043269)	612	19	4.17	+	4.56	0.000344
Regulation of secretion by cell (GO:1903530)	629	17	4.28	+	3.97	0.0122
Regulation of secretion (GO:0051046)	680	18	4.63	+	3.89	0.00789
Neuron development (GO:0048666)	654	17	4.45	+	3.82	0.0204
Cell migration (GO:0016477)	658	17	4.48	+	3.79	0.0221
Cytoskeleton organization (GO:0007010)	779	20	5.31	+	3.77	0.00305
Generation of neurons (GO:0048699)	1,291	33	8.79	+	3.75	3.03E-07
Single-organism cellular localization (GO:1902580)	669	17	4.56	+	3.73	0.0275
Regulation of neurogenesis (GO:0050767)	683	17	4.65	+	3.65	0.0361
Positive regulation of transport (GO:0051050)	889	22	6.05	+	3.63	0.00141
Localization of cell (GO:0051674)	731	18	4.98	+	3.62	0.0216
Cell motility (GO:0048870)	731	18	4.98	+	3.62	0.0216
Locomotion (GO:0040011)	940	23	6.4	+	3.59	0.000882
Neuron differentiation (GO:0030182)	821	20	5.59	+	3.58	0.00684
Regulation of nervous system development (GO:0051960)	781	19	5.32	+	3.57	0.0134
Neurogenesis (GO:0022008)	1,372	33	9.34	+	3.53	1.45E-06
Regulation of cell development (GO:0060284)	851	20	5.8	+	3.45	0.0118
Regulation of transport (GO:0051049)	1,676	39	11.41	+	3.42	5.74E-08
Nervous system development (GO:0007399)	1,800	41	12.26	+	3.34	2.8E-08
Negative regulation of multicellular organismal process (GO:0051241)	977	21	6.65	+	3.16	0.0255
Movement of cell or subcellular component (GO:0006928)	1,029	22	7.01	+	3.14	0.0158
Regulation of cellular localization (GO:0060341)	1,153	24	7.85	+	3.06	0.00823
Regulation of cellular component organization (GO:0051128)	2,095	41	14.27	+	2.87	2.99E-06
Regulation of localization (GO:0032879)	2,246	43	15.3	+	2.81	1.92E-06
Positive regulation of signaling (GO:0023056)	1,326	25	9.03	+	2.77	0.0277
Positive regulation of cell communication (GO:0010647)	1,446	27	9.85	+	2.74	0.0128
Regulation of biological quality (GO:0065008)	2,643	47	18	+	2.61	2.47E-06
Regulation of signaling (GO:0023051)	2,486	42	16.93	+	2.48	0.000135
Regulation of cell communication (GO:0010646)	2,620	44	17.84	+	2.47	6.48E-05
Single-organism localization (GO:1902578)	2,664	44	18.14	+	2.43	0.000107
Regulation of molecular function (GO:0065009)	2,260	35	15.39	+	2.27	0.0215
Single-organism transport (GO:0044765)	2,467	38	16.8	+	2.26	0.00811
Localization (GO:0051179)	4,100	59	27.92	+	2.11	2.24E-05
System development (GO:0048731)	3,499	49	23.83	+	2.06	0.00226
Establishment of localization (GO:0051234)	3,295	45	22.44	+	2.01	0.017
Positive regulation of biological process (GO:0048518)	4,910	67	33.44	+	2	8.53E-06
Transport (GO:0006810)	3,170	43	21.59	+	1.99	0.0372
Negative regulation of cellular process (GO:0048523)	3,811	49	25.95	+	1.89	0.0283
Positive regulation of cellular process (GO:0048522)	4,363	56	29.71	+	1.88	0.00394
Cellular component organization or biogenesis (GO:0071840)	4,260	54	29.01	+	1.86	0.0106
Anatomical structure development (GO:0048856)	4,183	53	28.49	+	1.86	0.0143
Multicellular organismal development (GO:0007275)	4,036	51	27.49	+	1.86	0.027
Cellular component organization (GO:0016043)	4,121	51	28.06	+	1.82	0.0497
Regulation of cellular process (GO:0050794)	9,809	102	66.8	+	1.53	5.42E-05
Biological regulation (GO:0065007)	10,699	109	72.86	+	1.5	1.74E-05
Regulation of biological process (GO:0050789)	10,256	103	69.84	+	1.47	0.000353
Cellular process (GO:0009987)	13,249	122	90.23	+	1.35	0.000246
Single-organism process (GO:0044699)	12,188	111	83	+	1.34	0.0171
Unclassified (UNCLASSIFIED)	1,652	8	11.25	−	0.71	0

Fig. 6.

Mmu-miR-211–expressing mice show reduced memory abilities in the Morris water maze, and hsa-miR-211 is overexpressed in Alzheimer’s disease patient brains. (A) PANTHER classification of GO shows Dox-induced enrichment of differentially expressed gene groups, mainly regulation (Reg.) of neuron-related pathways in the dTg-211 brain transcripts (Methods). (B) Time to reach platform in the Morris water maze shows reduced learning ability in the first and second training days for dTg-211 mice. (C) Search strategy scores divided by trials and days for individual dTg-211 and CamK mice. Fewer trials of dTg-211 mice in the last days showed focal or directed strategy. (D) Loss of preference of the platform quadrant, reflecting impaired reference memory for dTg-211 compared with CamK mice in probe trials. (E) Higher miR-211 levels (∼twofold) in postmortem Alzheimer’s entorhinal cortices (73) compared with nondemented controls, n = 7 each, P < 0.05, Student’s t test. Results were considered significant at *P < 0.05, **P < 0.01, ***P < 0.001, after correction for multiple testing when applicable. ns, not significant.

To examine whether miR-211 excess perturbs learning functions, we used a recent Morris water maze protocol. In the first and second day of trials, dTg-211 mice expressing miR-211 took longer than littermate CamK controls to reach a submerged platform, a mild underperformance that was mitigated in the latter 2 d of trials (Fig. 6B). The search strategy of dTg mice in these trials was also affected, with slower strategy progression toward directed platform search over days (Fig. 6C). In a probe trial examining reference memory with the platform removed, CamK controls showed clear quadrant preference but dTg-211 mice lost the capacity to locate the platform quadrant (Fig. 6D), possibly indicating short memory duration even once the task had been learned. In contrast, dTg-211 mice performed similarly to CamK controls in locomotion-based tests for anxiety and exploratory behavior, such as the elevated plus maze (Fig. S6 A and B) and the open field test (Fig. S6 C and D). Thus, dTg-211 mice presented a specific functional phenotype of impaired allocentric navigation, a disability attributed to functional impairments in the hippocampus, entorhinal cortex, and some surrounding structures, the functioning of which is strongly modulated by the cholinergic system (51, 52).

Fig. S6.

DTg-211 mice do not show elevated anxiety. (A) No difference in time spent in open and closed arms of the elevated plus maze for dTg-211 compared with controls. (B) No difference in the fraction of time spent moving. (C) In the open field test, dTg mice showed no difference in time spent in border of the maze or (D) overall distance moved. (E) Temporal lobe samples from Alzheimer’s disease (AD) patients show hsa-miR-211 levels comparable to those of individuals without AD symptoms (CNT) or nondemented individuals with increased postmortem-observed BRAAK indexes (PATH nonD).

Notably, both memory loss and perturbed cholinergic function are reminiscent of Alzheimer’s disease (53). Also, miR-211 levels are elevated in epileptic human brain tissues (47), and epileptic events are arguably common in the AD brain (54). Therefore, the possibility of a miR-211 involvement in this condition should be explored. Indeed, postmortem entorhinal cortices from AD patients showed twofold increases in miR-211 levels compared with matched controls (Fig. 6E; P < 0.05). Of note, in temporal lobe samples of a discrete cohort of AD patients, no significant change in miR-211 levels was noted (Fig. S6B). Together, these findings may indicate multileveled roles of miR-211 dynamics in cholinergic-mediated neuronal functions.

Discussion

Our study implicated miR-211 as functionally involved in cholinergic surveillance that may limit the risk of neuronal hypersynchronization in the mammalian brain. We combined forebrain RNA-seq analyses with electrocorticography tests of engineered mice presenting conditional miR-211 excess, or overexpressing a miR refractory AChE-R. This exploration linked forebrain miR-211 decreases with spontaneous ECoG-detected nonconvulsive seizures, exacerbated susceptibility to PTZ-induced seizures and increased miR-134 levels, accompanied by comodified synaptic, cholinergic, and TGFBR2 pathways, with a selective impact on endothelial brain cells, indicating involvement of the neurovascular unit (55). The changes associated with miR-211 decline may all add to the risks of convulsive seizures in a long-lasting manner, together attributing several hallmarks of the dynamics and efficacy of cholinergic-mediated susceptibility for seizures in the adult mammalian brain to miR-211 suppression.

ACh-related mechanisms are widely involved in, and sufficient for, generating epileptic seizures (56), with cholinergic imbalances hallmarking several epileptic syndromes (57) and cholinergic agonists inducing limbic seizures (58). Recent independent reports implicated miRs in both initiating and changing the response to seizure-inducing agents (59), but did not explore the putative regulatory role of miRs controlling neuronal cholinergic transmission (57) in avoiding such seizures and affecting the susceptibility to epilepsy. To address these questions, we intersected datasets for miRs targeting synaptic functions (24), experimentally found in neurons (25) and modified following pilocarpine injection (26). This search identified miRs that are directly or consequentially affected by epileptic perturbations, and that potentially function in the synapse. Of these miRs, we focused on miR-211, which was significantly reduced in our hands, either directly or consequentially by pilocarpine injection, and is one of those few genes that are lost in the 15q13.3 deletion syndrome (27, 60) (OMIM no. 612001), which is commonly manifested in epilepsy (27, 28). We also demonstrated miR-211 regulation over the human nAChRα7 nicotinic receptor, mapped within this genomic locus (61), and whose gain-of-function mutation associates with seizure induction in patients (29). Additionally, miR-211 targets the angiopoietin-1 (ANGPT1) gene, where a single nucleotide variation disrupting its binding site associates with elevated risk of ischemic stroke (62), a condition with diverse links to epileptic seizures (4). The miR-211 candidate thus emerged as putatively linked to epilepsy in several ways.

Notably, our work differs from many other miR-focused reports, which depict functional regulation by miRs by assessing their steady-state levels. Rather, we observed limited transcriptional changes in dTg-211 forebrains overexpressing the miR compared with controls, yet found dramatic transcriptional and electrophysiological changes in the brain following the dynamic reduction of miR-211 to its basal levels. That the decline of miR-211 has a more extensive effect than its steady-state level suggests that the function of this miR can be only partially assessed by single time-point measurements. Thus, Dox-induced suppression of the miR-211 transgene in dTg-211 mice induced both nonconvulsive seizures and entailed increased susceptibility to convulsive seizures while elevating the TGFBR2 transcript and protein levels. Because these ECoG-documented changes were absent in control mice, they unlikely reflect the mere impact of Dox, but rather point to a sustained epileptiform effect of miR-211 down-regulation, with possible relevance to human epileptogenic conditions. Additionally, RNA-seq showed a selective elevation in 19 of 21 epithelial marker genes, indicating cellular composition changes of the microvasculature (63), reported to hallmark and contribute to epileptogenesis, with possible gross changes in the epithelial transcriptional landscape.

Intriguingly, the observed seizures were characterized by sharp 5-Hz activity oscillations, with similarity to spike and wave seizures reminiscent of those of absence epilepsy patients. These human syndromes and their corresponding mouse models are manifested in epilepsy, which affects GABAergic modulatory neurotransmission but lacks observed motor convulsions (36, 64). Also, we noted expression changes concordant with systematic facilitation of cholinergic synaptic activation following Dox; specifically, this involved simultaneous elevation in the excitatory nicotinic receptors 5 and 7 and muscarinic receptor 5, and down-regulation of the inhibitory muscarinic receptors 2 and 4. We hypothesized that the muscarinic receptor 1 (mAChR1) will be concordantly modified; supporting this notion, qRT-PCR assays showed significant mAChR1 reduction in dTg-211 mice, and elevation following Dox suppression (Fig. S4E). That we did not observe a parallel change in the sequencing experiment may reflect mRNA editing of this gene’s transcripts (15), and for which alignment of sequencing reads is sensitive. We also observed elevation of the ataxia-related scaffold protein ATCAY/BNIP-H gene, which affects shuttling of the ACh synthesizing enzyme ChAT to neurite terminals (45, 46), and mutations in which associate with variable psychomotor retardation phenotypes and cerebellar dysfunction, together indicating multileveled exacerbation of cholinergic signaling.

After Dox has been removed and forebrain miR-211 levels were reelevated, we observed increased susceptibility of dTg-211 mice to PTZ-induced convulsions, suggesting an apparent link between miR-211 expression dynamics to convulsions. Alternative interpretations may, however, exist. For example, abrupt miR-211 reduction may function as a priming event for broad neurological changes, including susceptibility for convulsions, such as those implied by the selective elevation of epithelial marker genes. Alternatively, or in addition, sustained miR-211 elevation may have a supportive or even a merely ancillary role in the complex genetic makeup protecting the neuronal network from hyperexcitability, whereas its perturbation is functionally potent enough to disrupt that balance. Further work would be required to explore these possibilities.

At the cellular level, miR-211 reacts to endoplasmic reticulum stress by targeting the proximal promoter of the proapoptotic transcription factor chop/gadd153, which attenuates apoptotic processes, allowing the cell to reestablish homeostasis and avoid cell death (65); this may indicate attenuation by miR-211 of stressful cellular signals, inversely proposing that its down-regulation may exacerbate such stress-induced reactions. If indeed the observed miR-211 elevation in the Alzheimer’s entorhinal cortex relates to the disease pathophysiology, it may suggest an explanation to the greater risk for unprovoked seizures in early onset patients and those with more severe disease (66), where miR-211 increases would prove insufficient in avoiding seizures. This expression change, not seen in temporal lobe samples, may relate to the prominent cholinergic innervation of the entorhinal cortex by the medial septum.

Neuronal networks operate over a wide range of activity levels, and their imbalance underlies many neurological diseases, including epilepsy. Our study indicates that miR-211–induced transcription changes may be involved. Concordantly, PPI network analysis showed modified cholinergic synapse pathway genes (KEGG 04725) under miR-211 decline. Additionally, this decline affected protein kinase C gamma and inositol 1,4,5-trisphosphate receptor type 1, which modulate ion metabolism at the cholinergic synapse, and are linked to ataxia (67, 68), indicating globally modified synaptic composition and function under reduced miR-211 expression. Supporting this notion, a Caenorhabditis elegans gain-of-function mutation in a neuronal ACh receptor, acr-2(gf), causes an epileptic-like convulsion behavior, and the behavioral and physiological effects of acr-2(gf) require the activity of the TRPM1-similar channel GTL-2 (69). Thus, the functional link among ACh, the miR-211 host gene TRPM1, and epilepsy spans many species.

TgR mice overexpressing the nonsynaptic AChE-R splice variant show both enhanced seizure susceptibility and miR-211 reduction, which agrees with the role of miR-211 as both a regulator of and affected by cholinergic signaling-mediated seizures. Compatible with this notion, we identified a time-dependent induction of miR-134 in both TgR mice and within the 4 d following Dox administration to dTg-211 mice, alongside the susceptibility to seizures. Notably, miR-134 levels increase in temporal lobe samples from temporal lobe epilepsy patients and from mice 24 h after induction of status epilepticus (10). Moreover, antagomir silencing of miR-134 reduces recurrent seizures and development of epileptogenesis, and miR-134 affects long-term depression via the Pumilio-2/Polo-like kinase 2 (Pum2/Plk2) pathway, leading to inhibition due to altered subunit composition of plasma membrane-localized AMPA receptors (70), perhaps indicating reversibility of miR-211 decreases.

Pathway analysis performed on all genes that were differentially expressed in dTg-211 cortices following Dox indicated modified cognition, learning, and memory-related GO terms, suggesting that mice expressing miR-211 may have altered memory abilities, possibly due to the role of synaptic cholinergic signaling in memory (71). In the Morris water maze, dTg-211 mice presented delayed progression in search strategy and reduced reference memory abilities in probe trials. The impaired allocentric navigation phenotype may be attributed to impaired hippocampal, entorhinal cortex and some surrounding structures, the functioning of which is strongly modulated by the cholinergic system (51, 52, 72).

In conclusion, we show that dynamic reduction of miR-211 overexpression in forebrain neurons leads to major transcriptional change, including elevated levels of endothelial marker genes and enriched GO pathways relating to synapse genes, neurotransmitter functioning, higher brain functions, and transporter activity. MiR-211 decreases further induce overall potentiation of cholinergic synapses, and spontaneous ECoG documented nonconvulsive seizures in the brains of tested mice. Our findings present the dynamic changes in miR-211 and miRs at large as possible mediators and attenuators of seizures induced by cholinergic transmission imbalances. The long-lasting functional effects of miR-211 on the process of epileptogenesis and its implications for diagnostic and therapeutic avenues thus require further research.

Methods

Transgenic Mice Generation and Experiments.

We generated pTRE-miR-211 mice by cloning the premmu-miR-211 sequence into a pTRE- tight vector, followed by pronuclear injection at the Weizmann Institute of Science Animals Facility. Mice were held in specific pathogen-free conditions at the Hebrew University, an Association for the Assessment and Accreditation of Laboratory Animal Care (AAALAC) International accredited institute. All procedures, including animal tests, were approved by the Institutional Ethics Committee of The Hebrew University of Jerusalem (ethics no. NS-16-14729-4), with concordance to AAALAC International guidelines. tTA-CamKa strain was acquired, backcrossed, and crossed and housed as in SI Methods. dTg mice and CamK littermates were housed together and administered Dox in parallel. The cholinergic muscarinic agonist pilocarpine (Sigma; 290–340 mg/kg) was i.p. injected to mice as in ref. 73. Behavioral tests, including Morris water maze, elevated plus maze, and open field tests were performed as described in SI Methods.

Human-Derived Samples.

Postmortem samples of entorhinal cortex from AD and age-matched controls were obtained from the Netherlands Brain Bank (NBB) at the Netherlands Institute for Neuroscience (73). Samples were collected following written informed consent for a brain autopsy by the NBB.

Statistical Analyses.

Differential expression in sequencing experiments was derived from adjusted P values in DESeq (74) on R platform, after FDR correction. For specific genes presented, fold change and ratio values are shown as mean ± SEM. For cholinergic receptors, t test was used on normalized count data. For enrichment of cell type marker genes, P value was defined based on permutation analysis. Box and whisker plots show second and third quintiles for box, and 1.5 quantile distances from median for whiskers, as by convention. Results were considered significant if *P < 0.05, **P < 0.01, ***P < 0.001, after correction for multiple testing when applicable. For ECoG seizure and spike observations, a Mann–Whitney test was used.

Additional methods are described in SI Methods.

SI Methods

Transgenic Mice.

For genotyping, tail-tip PCR was used (sequences given in Table S1). B6;CBA-Tg(Camk2a-tTA)1Mmay/J (CamK mice) were purchased from the Jackson Laboratory. Mice were backcrossed to FVB/N background for three generations and further crossed to generate double-transgenic mice and littermate controls.

Luciferase Assay.

MiR-211-5p direct targeting was predicted in silico and experimentally assessed using in vitro luciferase measurement with the Dual-Luciferase Kit (Promega) 48 h after transfection. Specifically, a 3′-UTR fragment of human nAChRα7 OR murine TGFBR2 transcripts (both similar in mouse and human), was cloned into a luciferase reporter vector. Reporter and miR-211 expression vector (Genecopia; MmiR3291-MR04) were transfected into HEK293-T cells (ATCC) using polyethylenimine, and luciferase activity measured and normalized according to manufacturer’s instructions.

ECoG Recordings and Analysis.

ECoG recordings were performed as previously described (77). Briefly, mice were placed in a stereotaxic frame under deep isoflurane anesthesia (1–3%), after shaving and disinfecting the dorsal aspect of the head; the skull was exposed by a longitudinal incision. Holes were drilled in coordinates 3 mm caudal and 2 mm lateral relative to bregma. Stainless steel screws were fixed to the holes. After placing a wireless transmitter (Data Science International) within a pocket formed s.c. in the dorsal aspect of the body, the electrodes were connected to the screws and isolated with bone cement. Before termination of anesthesia, buprenorphine was administered (i.p., 0.05 mg/kg). Following recovery, animals were moved to a behavior room with 12-h light/dark cycle and had access to food and water ad libitum. After 4 d of habituation we began continuous ECoG recording using a homemade Matlab-based program that allows reliable unbiased detection of seizures (23). The results were revised manually and blindly. ECoG spikes were implemented via the wavelet transform algorithm (78) following a bandpass filtration of the ECoG signal between 1 and 45 Hz. After automated detection and clustering using the Matlab program, blind human revision was performed to reassure the results.

PTZ-Induced Seizures.

Animals were placed individually in Plexiglas boxes, and seizure behavior was observed for 30 min following PTZ injection (50 mg/kg, s.c.). Seizure intensity was evaluated (as in ref. 79). Parallel ECoG recording was analyzed as mentioned above.

RNA Samples Collection for RT-qPCR and Sequencing-Compatible Libraries.

Mice were anesthetized with isoflurane before cervical dislocation, and their brain regions, including frontal cortices, were dissected and collected in liquid nitrogen. For RNA sequencing, RNA was extracted by miRNeasy kit (QIAGEN); RNA quality determined with RNA 6000 Nano Bioanalyzer (Agilent), and samples with RIN values ranged between 8 and 8.9 were used. RNA was extracted for qRT-PCR validation analysis using TRIzol Reagent (Invitrogen), as reported (80). cDNA and PCR were as described herein.

High-Throughput RNA Sequencing.

Sequencing-compatible poly(A)-terminated single-end libraries were generated using an RNA Library prep kit (NEBNext Multiplex, E7330S; New England BioLabs) following manufacturer’s instructions, with 12 amplification rounds. Libraries were barcoded and sequenced on a NextSeq Series Sequencing System (HUJ Center for Genomic Technologies) using two Illumina flow cells (Illumina 500 NextSeq High Output v2 Kit, FC-404-2005; Illumina). Raw cluster densities for samples ranged between 170 and 189 K/mm². Reads were aligned (90% mapping) to the mouse transcriptome (TopHat2) (81), and expression analysis was performed using the DESeq (74) software via R platform (82). Libraries from all tissues were overall similar in depth, with a similar distribution of transcript numbers per expression level and tag-wise normalized variance predictably correlated to expression levels (Fig. S2). A 6.2 log cpm value for PPI thresholding represents the upper 10 percentiles of 18,570 genes in datasets.

Bioinformatics, Pathway Analysis, and Luciferase Assays.

Cell type marker genes (39) were selected as mouse orthologs name-wise and their levels plotted as fold change; P value presented after Bonferroni correction and permutation analysis were used to determine significance per marker. ECDF plots were generated in R via the stats package (82) after thresholding genes by counts [mean log(cpm) > 1] and subsetting by significance of differential expression. PANTHER (Protein ANalysis THrough Evolutionary Relationships) (50) version 10, GO version 1.2, annotated June 22, 2016, was used for GO of differentially expressed genes for the KEGG database.

ELISA.

Mice frontal cortex samples were used for ELISA using an EIAab kit (for murine TGFBR2, catalog no. E9935m). Cortices were homogenized in radioimmunoprecipitation assay buffer containing protease inhibitor (1:200) and centrifuged. Sample protein content was measured with the Lowry assay (Thermo Scientific). Samples were further diluted 1:3 in PBS and used for quantifying TGFBR2 by EIAab kit (catalog no. E9935m) as per manufacturer’s instructions.

Morris Water Maze.

Mice were subjected to a Morris water maze to assess learning and memory. Briefly, mice were released to search for a submerged platform in a fiberglass water tank, 1.2 m in diameter. Water temperature was fixed, and colored white for opaqueness. Lighting remained fixed; platform directions were queued with visible marks to allow for allocentric navigation, and systematic release directions. Each mouse was tested in four trials a day for 4 d, followed by a single day of probe trial, in which mice search for the platform that had been removed. Experiments were performed blind, using alias numbers per mouse, recorded from a monochromatic ceiling-fixed video camera, and automatic movement tracing and quantitative data analysis was performed using EthoVision (83) XT (Noldus version 8; Information Technology).

cDNA and qRT-PCR.

A qScript Kit (Quanta) was used for reverse transcription. Real-time PCR performed on a CFX-96 machine (Bio-Rad), and quantification performed using the ΔΔCt method, with snoRD47 as a loading control for miRs and β-actin for long transcripts.