Background
microRNAs are small non-coding, single-stranded RNA molecules that regulate gene expression by binding to the 3′UTRs of their target mRNAs through base pairing of their 6–8-nucleotide long seed region. They either initiate mRNA degradation or inhibit protein translation [
1,
2]. Regulation of gene expression in the nervous system is highly complex and is fine-tuned by microRNAs [
3]. In neurons, mRNAs and microRNAs are compartmentalized to specific subcellular regions, like the synaptodendritic compartment, to regulate local protein synthesis in response to neuronal stimuli [
4].
microRNA-137 (miR-137) is enriched in the human and mouse brain [
5], especially in cortical regions and in the hippocampus [
6]. It plays a role in cell proliferation and differentiation and is present at the synapse [
6,
7]. Furthermore, it regulates neuronal maturation, dendritic morphogenesis, and spine development [
8]. Homozygous knockout of
Mir137 is embryonically lethal in mice, indicating that embryonic development is dependent on at least one functional allele [
9]. This emphasizes the importance of miR-137 in early developmental processes.
Genetic alterations of
MIR137 have been associated with neurodevelopmental disorders. The rare chromosomal microdeletion 1p21.3 encompassing
MIR137 and
DPYD was identified in individuals with intellectual disability (ID), comorbid with autism spectrum disorder (ASD), and obesity [
6,
10,
11]. In addition, 19/65 (29%) highest ranked ASD risk genes in the Simons Foundation Autism Research Initiative gene database [
12,
13] (URL:
https://sfari.org/resources/sfari-gene#refs, accessed May 2017) are predicted or confirmed miR-137 targets (Additional file
1: Table S1). miR-137 target genes are enriched among ASD risk genes (29%, CI 18.6–41.8%) compared with the random frequency (2.9%, predicted in the miRDB database [
14]; CI 2.7–3.2%, 525/17860) (
P ≤ 0.000001, Fisher’s exact test, two-sided). Furthermore, the
MIR137 gene has been reported as a schizophrenia (SCZ) susceptibility locus because the common SNP rs1625579, located in an intron of
MIR137, was associated with SCZ in several studies [
15‐
18]. Putative miR-137 targets are enriched in SCZ risk loci [
18], suggesting that convergent pathways connected by this microRNA contribute to SCZ etiology. A meta-analysis of SNP data across five different neuropsychiatric disorders identified a shared risk locus association with SCZ and ASD [
19]. Taken together, these findings indicate that deregulation of miR-137 is the key to various psychiatric and neurodevelopmental disorders.
SHANK proteins (SHANK1, SHANK2, and SHANK3) are postsynaptic scaffolding proteins with an important role in the formation, maintenance, and function of synapses in glutamatergic neurons of the brain [
20].
SHANK genes converge with miR-137 on several levels: (i) both are expressed at the synapse, (ii) both influence dendrite and spine formation in glutamatergic neurons [
8,
21], and (iii) both have a strong link to neurodevelopmental and neuropsychiatric disorders like ID, ASD, and SCZ [
18,
22,
23]. This suggests that
SHANK genes might be targets of miR-137.
Methods
Luciferase assays
To validate a predicted miR-137 binding site in
SHANK2, luciferase reporter assays were carried out in human SH-SY5Y cells and mouse primary hippocampal neurons. The wild type
SHANK2-3′UTR was cloned into the psiCHECK™-2 vector (Promega, Mannheim, Germany) after PCR amplification on human genomic DNA from a healthy individual. PCR primers introduced the restriction sites (XhoI) which were used to introduce the PCR product into the psiCHECK™-2 vector via Gibson® Assembly (New England BioLabs, Frankfurt, Germany). Mutagenesis of the miR-137 binding site was performed on the wild type construct by PCR with the original cloning primers and mutagenesis primers, assembled using the SLiCE cloning system [
24]. The sequence of wild type and mutated SHANK2-3′UTR was validated by Sanger sequencing. Primer sequences are given in Additional file
1: Table S2.
The SH-SY5Y cells were seeded in antibiotic-free medium and transfected after 24 h with 200 ng of empty psiCHECK™-2 vector and wild type or mutated SHANK2 3′UTR constructs with 12 nM miR-137 or control mimics (mirVana® miRNA hsa-miR-137 mimic MIMAT0000429 and Pre-miR™ miRNA Precursor Molecules Negative Control #2, AM17111, Thermo Fisher Scientific, Darmstadt, Germany) in technical triplicates using Lipofectamine®2000 (Thermo Fisher Scientific, Darmstadt, Germany). miRNA mimics are small, double-stranded RNA molecules which mimic endogenous, mature microRNA molecules when introduced into the cells. Luciferase activity was measured 48 h posttransfection using the dual luciferase reporter assay system (Promega) and the luminometer Centro LB960 (Berthold Technologies, Bad Wildbad, Germany). Renilla luciferase activity was normalized against firefly luciferase activity, and the background was subtracted by normalization to the empty psiCHECK™-2 vector activity.
Primary hippocampal neurons were transfected on day in vitro (DIV) 5 using the same experimental setup as in the SH-SY5Y cells for the mimic experiments with 25 nM mimics. The amount of miR-137 mimics had to be optimized for each cell type separately. For the inhibitor experiments, the primary mouse hippocampal neurons were treated with 25 nM miR-137 or power inhibitor control (miRCURY LNA™ Power microRNA hsa-miR-137 4101446-111 or negative control B 199007-111 inhibitor 5′-Fluorescein-labeled, Exiqon, Vedbaek, Denmark) added drop-wise to the cells on DIV5. The miR-137 power inhibitor sequesters both the human hsa-miR-137 and mouse mmu-miR-137, thereby preventing it from binding to its targets
. The control inhibitor does not influence miR-137 and was used to control for unspecific effects of the treatment. Experimental conditions were identical between both groups regarding cell number and cell types (Additional file
1: Figure S1). The cells were transfected with the luciferase reporter constructs with Lipofectamine®2000 on DIV6. Luciferase activity was measured 48 h posttransfection as described above.
Cell culture
The SH-SY5Y cells were obtained from the DSMZ (Leibniz Institute German collection of Microorganisms and Cell Cultures, Braunschweig, Germany, ACC209). The cells were kept in supplemented Dulbecco’s Modified Eagle’s Medium (DMEM) (15% fetal bovine serum (FBS), 1% penicillin/streptomycin (P/S), 1% non-essential amino acids). Hippocampal cultures were prepared from gestating CD-1® IGS mice (Charles River, Sulzfeld, Germany) at embryonic stage 15.5. The cells were seeded in plating media (DMEM + 10% FBS, 0.25% P/S, 2 mM L-glutamine) on poly-L-ornithine (Sigma Aldrich, Schnelldorf, Germany) coating. On DIV 1, a complete medium change to supplemented Neurobasal medium (Gibco® Neurobasal® media + 2% Gibco®B-27® supplement, 0.25% P/S, 0.5 mM L-glutamine) was performed. On DIV3, a half medium change was carried out with 5 μM cytosine β-D-arabinofuranoside in supplemented Neurobasal media, except for the cells that were harvested on DIV3. From DIV5 until the final culture day, a half medium change was performed with supplemented Neurobasal media every other day, except for the cells that were transfected or harvested.
Nucleofection of mouse hippocampal cells
Hippocampal cultures were transfected with the optimized hsa-miR-137 mimic concentration of 300 nM and negative miRNA control on DIV5 using the AD1 4D-Nucleofector™ Y Unit system (Lonza, Basel, Switzerland) according to manufacturer’s instructions. Cell types and cell numbers showed no difference after nucleofection between the two conditions (Additional file
1: Figure S1).
RT-qPCR
To determine the effect of miR-137 on endogenous
Shank2 expression real-time quantitative PCR (RT-qPCR) was performed. For the quantification of
Shank2 expression, RNA was isolated from mouse primary hippocampal neurons (DIV8) 72 h after transfection with using the Quick-RNA™ MicroPrep Kit (Zymo Research, Freiburg, Germany) and was transcribed with the SuperScript®
VILO™ cDNA Synthesis Kit (Thermo Fisher Scientific, Darmstadt, Germany). RT-qPCR was performed to quantify
Shank2 expression normalized against two reference genes (
Gapdh and
Hprt1). Primer sequences are provided in Additional file
1: Table S2.
The RNAqueous®-Micro Total RNA Isolation Kit (Thermo Fisher Scientific) was used for total RNA isolation. Ten nanograms of total RNA were transcribed using the miRCURY LNA™ microRNA PCR Universal cDNA synthesis Kit (Exiqon) with a spike-in loading control according to manufacturer’s instructions. RT-qPCR was performed with miRCURY LNA™ microRNA PCR ExiLENT SYBR® Green PCR sets for hsa-miR-137, U6, and spike-in (Exiqon), measuring the endogenous miR-137 expression profile in primary hippocampal neurons over a time period of 11 DIV. miR-137 expression levels were analyzed with the same method using 21 different human tissue RNA samples (specified in Additional file
1: Table S3).
Western blot analysis
To determine the effect of miR-137 on Shank2 protein expression, cellular protein was isolated using RIPA buffer from cell cultures on DIV10, and the lysates were run on Novex™WedgeWell™4–12% Tris-Glycine Gels (Thermo Fisher Scientific) and then blotted onto a PVDF membrane (Immobilon-FL, Millipore, Billerica, Massachusetts, USA) as recommended by the manufacturer. Primary antibodies anti-beta-III-tubulin (mouse, G7121 Promega) and anti-Shank2 (rabbit, ABIN656710 antibodies-online) were used followed by the secondary IRDye 680 donkey anti-mouse and IRDye 800CW donkey anti-rabbit antibodies (1706515 and 1706516, respectively, Li-COR). The membrane was scanned with the Odyssey® Infrared Imaging System (Li-COR) and quantified using ImageJ software39. Shank2 was normalized to beta-III-tubulin. All experiments were performed sequentially over several weeks. Hippocampal neurons were isolated and differentiated at different time points.
CommmonMind data
The tissue used for RNA sequencing was obtained from the following institutes: Mount Sinai School of Medicine, University of Pennsylvania, University of Pittsburgh, National Institute of Mental Health, F. Hoffman-La Roche Ltd., Takeda Pharmaceuticals Company Limited, Sage Bionetworks, Duke University, and University of North Carolina. Generation and analysis of the RNA sequencing data has been previously described in detail by the CommonMind Consortium [
25]. To analyze the expression levels of the validated miR-137 target genes and miR-137 precursor, we selected the average expression levels, the log
2-fold change and the
P values for the respective genes published on
https://synapse.org in the “Results Explorer” on the “CommonMind Consortium Knowledge Portal” (data accessed June 2017). A weighted linear regression analysis had been performed for each gene considering covariates that influence expression levels. For each gene, the SCZ disease status coefficient had been statistically tested for being nonzero, implying an estimated effect for SCZ, above and beyond any other effect from covariates. This test produced a
t statistic and a corresponding
P value. Details about the statistical analysis are provided in the CommonMind Consortium paper [
25].
Selection of experimentally validated miR-137 targets and network analysis
Validated targets of miR-137 were collected from miRTarBase (
http://mirtarbase.mbc.nctu.edu.tw/, accessed January 2017) and PubMed (
https://www.ncbi.nlm.nih.gov/pubmed/, accessed January 2017) database searches. miRTarBase targets were included if they were categorized as “strong evidence,” i.e., confirmed by experimental methods of validation including luciferase reporter assay, RT-qPCR, and/or Western blot analysis. Targets from the PubMed database search were considered as validated if a direct link was established by luciferase reporter assay, RT-qPCR, and/or Western blot analysis. Duplicates between databases were removed. A total of 71 validated targets were obtained; most of them have been investigated in cancer cells, whereas only 12 targets have been validated in neurons (details are shown in Additional file
1: Table S4). RNA sequencing data for the miR-137 target genes
HCRT,
SLC6A3, and
SNAI1 was not available in the CommonMind dataset. The target gene network was analyzed through the use of QIAGEN’s Ingenuity®Pathway Analysis (IPA®, QIAGEN Redwood City,
www.qiagen.com/ingenuity).
Statistical analysis
Statistical analyses were performed using the SPSS software (IBM Corp.; IBM SPSS Statistics for Windows, Version 22.0.; Armonk, NY, USA). Two-way ANOVAs were performed considering the influence of the experiment and the respective condition. The data is shown as mean ± SEM. The validated miR-137 target genes (including
SHANK2) were ranked according to their point-wise
P values from the CommonMind analysis, and the Benjamini-Hochberg method was used to correct for multiple testing with a false discovery rate of 10% (Additional file
1: Table S5). A two-sided
Χ2 test with Yates correction was used to compare the expression of validated target genes of miR-137 and five control microRNAs. The differentially expressed targets of the five control microRNAs were pooled together. Genes that were targets of more than one control microRNA were only counted once. Eight target genes that were not differentially expressed in SCZ and control individuals overlapped between miR137 and the five controls and were excluded from analysis.
Discussion
In this study, we identified
SHANK2 as a novel direct target of miR-137. We discovered that physiological levels of miR-137 regulate
SHANK2 expression, most likely by repressing translation of SHANK2 protein rather than inducing mRNA degradation. Inhibiting effective protein translation is a known mechanism for local fine tuning of gene expression at postsynaptic sites and is in line with previously reported direct miR-137 targets, including Ephrin B2 (
EFNB2) and the AMPA receptor subunit GluA1 (
GRIA1) [
35,
36]. miR-137 regulates the expression of several proteins that function at glutamatergic synapses, e.g., EFNB2, GluA1, and Mib1 and thereby influences neuronal maturation and signal transduction [
8,
35,
36]. AMPA receptors are anchored to the postsynaptic Shank scaffold via PSD-95/Stargazin proteins [
37]. SHANK2 and AMPA receptors are both regulated by miR-137 and are important for synaptic maturation and plasticity; therefore, this control by miR-137 may have synergistic effects on synaptic regulation. Beside the regulation of postsynaptic genes, miR-137 has also been shown to regulate presynaptic genes and presynaptic neurotransmitter release [
38]. This is supported by the finding that altered miR-137 levels impact synaptic function and neuronal network formation in the mouse hippocampus [
36].
Morphological effects caused by altered Shank2 levels in hippocampal neurons do not correlate to the morphological changes found when miR-137 levels are altered. Overexpression of SHANK2 increased spine volume while its downregulation led to reduced spine volume and increased dendritic arborization [
39]. In contrast, miR-137 overexpression or inhibition did not affect dendritic spine morphology, whereas miR-137 inhibition only led to a reduced spine density [
36]. This difference might be due to a more subtle regulatory effect of miR-137 on Shank2 protein expression compared to an RNAi-mediated knockdown as microRNAs are “fine-tuners” of protein expression. miR-137 regulates the expression of multiple glutamatergic synapse proteins; therefore, subtle regulation of
SHANK2 expression by miR-137 is likely to be physiologically relevant. miR-137 also has subtle effects on the protein expression of its other targets [
36]. Different postsynaptic proteins are probably regulated by miR-137 at the same time, which may increase the intensity of the neuronal response.
We showed a regulatory influence of miR-137 on SHANK2 expression in mouse hippocampal neurons, whereas the relevance in other brain regions and in human neurons warrants future investigation.
miR-137 has been linked to various disorders including ID, ASD, and SCZ [
6,
10,
11,
18]. Impaired synaptic plasticity and glutamatergic neurotransmission have been postulated as underlying pathological mechanisms [
40]. A heterozygous microdeletion on 1q21.3 encompassing the genes
MIR137 and
DPYD was previously described in ID and ASD patients [
6,
10,
11]. Reduced levels of precursor and mature miR-137 concurrently with significantly increased levels of downstream target genes (
MITF,
EZH2, and
KLF4) were determined in lymphoblastoid cells isolated from two patients with this 1q21.3 microdeletion [
6]. Based on our experimental results, we speculate that reduced miR-137 expression may also increase
SHANK2 levels in these patients, which may contribute to the ID and ASD phenotype seen in these patients.
To further investigate the link between miR-137 and SCZ, we analyzed the expression of miR-137 precursor and known miR-137 target genes in postmortem DLPFC samples of SCZ individuals using the CommonMind gene expression data resource [
25]. No difference in miR-137 precursor expression was found between SCZ and control individuals. Our study was limited to miR-137 precursor expression analysis as no data of mature miR-137 expression was available. Previous studies have shown no difference in the levels of precursor and mature miR-137 between SCZ and control individuals [
31,
41,
42]. miR-137 expression has been investigated in fibroblasts and fibroblast-derived neurons isolated from individuals homozygous for four schizophrenia-associated SNPs at the MIR137 locus. Increased endogenous miR-137 levels were identified in the minor compared to the major allele SNP group in the neurons, but not in fibroblasts [
38]. This indicates that regulation of miR-137 expression varies in different cell types. Analyzing miR-137 expression in specific cortical layers, cell types, or even cell compartments might reveal distinct local alterations in the DLPFC of SCZ patients.
Our network analysis revealed that many validated miR-137 target genes are linked to developmental, neurological, and psychiatric disorders, particularly SCZ. Therefore, we analyzed the expression of these genes in the DLPFC, a brain region that is affected in SCZ individuals. Expression of 23% (16/69) of the analyzed target genes was altered in the DLPFC of SCZ compared with control individuals. This change in expression was above the level of target enrichment in a pooled sample of five control microRNAs. This indicated that the miR-137 signaling network might be altered in the DLPFC of SCZ individuals. Five of these differentially expressed target genes have been associated with SCZ
(TCF4,
SIRT1,
XIAP,
GRIA1,
ZNF804A) and two with ASD (
RORA,
KDM5B), suggesting some overlap between these two disorders. Four miR-137 target genes (
RORA,
CPLX1,
TCF4,
SIRT1) even show significant differences on the whole-transcriptome level, according to data provided by the CommonMind Consortium [
25]. Furthermore, analysis of the CommonMind RNA sequencing data confirmed elevated
GRIA1 mRNA levels in the DLPFC of SCZ individuals [
32], indicating impaired synaptic signaling in the DLPFC. The majority of differentially expressed miR-137 target genes (12/16) were only modestly elevated in SCZ individuals, suggesting that small effects of single genes accumulate in the DLPFC, presumably leading to a general impairment of miR-137 signaling. It is important to note that expression of miR-137 target genes is not regulated by miR-137 alone, but by multiple factors, e.g., other microRNAs, epigenetic mechanisms, subcellular localization, synaptic activity, medication, or other environmental factors. We analyzed miR-124 and miR-128, which act cooperatively with miR-137, and obtained no evidence that these two microRNAs influence the differential expression of miR-137 targets. miR-124 precursor expression was not different between SCZ and control individuals and the differentially expressed miR-137 target genes may only slightly be co-regulated by miR-124 and miR-128. Only 23% of the analyzed miR-137 target genes were differentially expressed on the mRNA level between SCZ and control individuals. Most miR-137 targets have been investigated in cancer cell lines and only 13 miR-137 target genes (including
SHANK2) have been confirmed in neuronal cells; therefore, some of the confirmed targets may not be miR-137 targets in the brain. Expression of
SHANK2 mRNA was not different in the DLPFC of SCZ and control individuals, which may mean that
SHANK2 is not a relevant miR-137 target in the context of SCZ. However, we and others have identified neuronal miR-137 target genes which were regulated on protein [
35,
36] and not on mRNA level. Therefore, we conclude that changes in mRNA expression do not reveal the full regulatory potential of miR-137. In this regard, proteomic data from postmortem human brains will be of high value for future SCZ studies.
Acknowledgements
A.dS.C., F.-B.C., and S.B. are members of the Hartmut Hoffmann-Berling International Graduate School of Molecular and Cellular Biology (HBIGS). S.B. is indebted to the Baden-Württemberg Stiftung for the financial support of this research project by the Eliteprogram for Postdocs. G.A.R. is a member of CellNetworks Cluster of Excellence of the University of Heidelberg, Germany. We thank Birgit Weiss and Diana Porras Gonzalez for technical assistance. The study was supported by the EcTop3 Project of EXC81 CellNetworks (BMBF).
CommonMind Consortium––The data were generated as part of the CommonMind Consortium supported by funding from Takeda Pharmaceuticals Company Limited, F. Hoffman-La Roche Ltd., and NIH grants R01MH085542, R01MH093725, P50MH080405, R01MH097276, R01MH075916, P50MH096891, P50MH084053S1, R37MH057881 and R37MH057881S1, HHSN271201300031C, AG02219, AG05138, and MH06692. The brain tissue for the study was obtained from the following brain bank collections: the Mount Sinai NIH Brain and Tissue Repository, the University of Pennsylvania Alzheimer’s Disease Core Center, the University of Pittsburgh NeuroBioBank and Brain and Tissue Repositories, and the NIMH Human Brain Collection Core. CMC Leadership: Pamela Sklar, Joseph Buxbaum (Icahn School of Medicine at Mount Sinai), Bernie Devlin, David Lewis (University of Pittsburgh), Raquel Gur, Chang-Gyu Hahn (University of Pennsylvania), Keisuke Hirai, Hiroyoshi Toyoshiba (Takeda Pharmaceuticals Company Limited), Enrico Domenici, Laurent Essioux (F. Hoffman-La Roche Ltd.), Lara Mangravite, Mette Peters (Sage Bionetworks), Thomas Lehner, Barbara Lipska (NIMH).