miR-101a-3p Impairs Synaptic Plasticity and Contributes to Synucleinopathy

Human samples

RNA extracted from human cortex was provided by the Newcastle Brain Tissue Resource, Institute of Neuroscience, Newcastle University, Campus for Ageing and Vitality, Newcastle upon Tyne, UK. Detailed sample information is presented in Supplementary Table 1.

Ethical approval

Newcastle Brain Tissue Resource Ethics Committee granted ethical approval for the DLB and control cases.

Animals

Male homozygous [A30P] aSyn Tg mice [24] and littermate controls were used at 6 months of age. Animals were housed in standard cages with 12 h light/12 h dark cycle and free access to food and water in the animal facilities of the University Medical Center, Göttingen, Germany. All animal experiments were approved by the Lower Saxony State office for Consumer Protection and Food Safety (license number: 33.9.42502-04-19/3213). In addition, midbrain samples of WT mice housed in either a standard or enriched environment for 12 months were included. For details on enriched housing conditions, see Wassouf et al. [25]. All procedures strictly adhered to the international standards for the care and use of laboratory animals and were approved by the local Animal Welfare and Ethics committee of the Country Commission Tübingen, Germany (license number: TVA HG 4/11).

Primary cortical neuronal cultures and immortalized cell lines

Mouse cortical neuronal mass cultures were prepared as previously described [12] and grown on coated plates including glass coverslips for imaging or custom-made microfluidic devices (MFDs). Details for immunocytochemistry and the MFDs are provided in the respective sections. For assessing the miRNA effect, cells were infected on day in vitro (DIV) 1 and harvested on DIV 14. For recombinant wt aSyn (raSyn) experiments, cells were exposed to 100 nM monomeric or oligomeric raSyn, or pre-formed fibrils (PFFs) on DIV 5 and harvested on DIV 25. Recombinant wt aSyn species were produced as in Vasili et al. [26]. Human embryonic kidney 293 cells were grown at standard conditions and transfected with plasmid DNA (1μg/ml culture medium) and microRNA mimic (4 nM final concentration) in a 1:3 ratio with FuGENE^® HD Transfection Reagent. Following transfection, cells were incubated for 24 h.

Plasmids

For the luciferase assay, fragments of the GABRB2 and DLGAP3 3’-untrsanslated region (UTR) sequences (GABRBR2 3’UTR fragment: 5’ -ATTTCCTTCATATGACTTTTTTTTTTTTAAATTTGCCAGCCAACAGTCACCATTCTGAAATAAATAATACAGGGTACTGAACATCTCTCTCAGCAAATCAATGCCTACAGTTCCTAAGAACATTGCACAGTTTGACTCTACAATAGTACTGTACCTCAGAATATGAGATGTTAACTAGTGTCTGCATTGCTTGTCAAATATCCATTTGTTCTTTTAAAGGACACAGATAACACTATCCATTGGTGATTTAAAATTGAAAAAAAAATTAGGAAAATGGGGAAAAATCCTCTTGAGAGGAGG- 3’; DLGAP3 3’UTR fragment: 5’ -CCCGCCCGGGCCGCCCAGTCCGACCCGGGCCCGCGGTTTTCTACCCGTACTGTACACCCAGCGTCGAGGTCACTGTGAACGCGGGCAGCTCCGTGCGCCCGCCCTGCCGGCACCGCACGCCCCGGCTTCTGCCCGCCGCGCTTTCGTGGGTTTTTTACCTTCCTGATCCCACGCAAAGGCGCCCGGGCTAGGCTGGGGGTCGTGCCTCTCCGCCCTGCGCCCCTCACTTGGAACTCCCATCTTCCTGGTCCGACGCTTTGACCCCCTCACCTTTTTCCCCCCATGGGCACCATCTCTGCC- 3’) (custom made, BioCat, Germany) were subcloned in a mammalian expression vector that encodes Gaussia luciferase with a KDEL sequence under the cytomegalovirus promoter. The vector expressing luciferase alone was used as a control vector. For assessing the miRNA effects, miR-101a-3p (5’ -CAAAAATACGTACTGATAACGAACTGCAGTTCAGTTATCACAGTACTGTA- 3’) or a scrambled miRNA sequence (5’ – CAAAAAATAGTATGCAATTCGAAGATCCTGCAGGATCTTCGAATTGCATACTA- 3’) (custom made, Metabion, Germany) was introduced in the pLKO.3G vector (Plasmid #14748, Addgene, MA, USA). The vector expressing GFP alone was used as a control vector. Lentivirus production was performed according to Follenzi and Naldini [27].

Luciferase and β-galactosidase activity assay

HEK cells expressing target genes 3’UTRs, Gaussia luciferase and miRNA mimic (miScript miRNA mimics, Qiagen, Germany) were harvested with Glycylglycin lysis buffer, centrifuged for 15 min at 15,000× g at 4°C and the supernatant was collected.

For the luciferase assay, cell lysate was added in a 96-well plate and luciferase activity was estimated by measuring luminescence on the plate reader Infinite M200 PRO after injecting the Luciferase Assay Reagent (Promega, WIS, USA) with the Te-Injecttrademark reagent injector.

For the β-galactosidase assay, cell lysate and β-galactosidase reagent (1 mg/ml o-nitrophenyl-β-D-galactopyranoside, 25 mM Sodium Phosphate pH 7.1, 2μM MgSO₄) was incubated at ambient temperature until color development was evident. Reaction was stopped with 0.5 M sodium carbonate, and absorbance at 420 nm was measured on the plate reader Infinite M200 PRO.

RNA extraction

RNA isolation was performed with TRIzol reagent (Invitrogen, CA, USA) according to the manufacturer’s instructions. For liquid samples equal volume of sample and TRIzol were mixed. For cell culture, media was aspirated and TRIzol was added directly to the culture plate for cell lysis. For tissue, TRIzol and ceramic beads (1.4 mm diameter ceramic beads, 91-PCS-CK14B, PEQLAB Biotechnologie, Germany) were added in vials with frozen tissue and homogenization was performed with the Precellys 24 tissue homogenizer (Bertin Instruments) at 65,000× g for 2×30-s runs with a 30-s break in between. Following RNA extraction and precipitation, RNA was dissolved in RNAse-free water and concentration and quality was estimated with NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific, MA, USA) by measuring absorption at 260 nm and the ratios 260/280 nm and 230/280 respectively.

Small RNA sequencing and data analysis

Small RNA sequencing was performed using RNA from whole midbrain of 8 Wt mice and 7 Tg mice on an Illumina HiSeq 2000 system. Small RNA libraries were prepared from 1μg total RNA using the Illumina TruSeq Small RNA Sample Preparation kit. Analysis of sequencing data was performed using custom pipelines. For quality check and demultiplexing, the CASAVA 1.8.2 software (Illumina) was used. For alignment the mapper.pl from miRDeep2 was used. Reads were aligned to MiRBase version 19 with a cut off score of p < 5×10-7 and a single mismatch allowed. In order to compare the miRNA expression levels between samples, a differential expression analysis was performed using R and the DESeq2 package and the following thresholds were used: log2FC < >0.25, basemean ≥5, padjusted ≤0.05. Heatmaps were created using the ggplot2 package. Computational miRNA target prediction analysis was performed using Target Scan Mouse 6.2. For functional annotation of predicted miRNA targets the Panther database (version 15.0) was used.

For the RNA-seq data integration we used in-house scripts to detect the miRNAs and target genes interaction using information from 2 different databases: miRTar [28] and miRbase [29]. We selected the miRNA-gene interactions that were true in both databases and then correlated those to the RNA seq data.

Reverse transcription and real-time qPCR

RNA sequencing results were validated using qPCR with primers specific for miR-101a-3p and target genes (Supplementary Table 2). QuantiTect RT Kit and miScript RT II Kit (Qiagen, Germany) were used for reverse transcription according to the kit manuals. Reactions were performed on Eppendorf MasterCycler.

Quantification of relative mRNA and miRNA levels was performed with QuantStudio3 Real-Time PCR system (Applied Biosystems, CA, USA). For the miRNA qPCR, miScript SYBR Green PCR Kit (Qiagen, Germany) was used with miScript primer assay (Supplementary Table 2) and the expression was normalized to endogenous control RNU6. For the mRNA qPCR, Mesa blue qPCR Mastermix plus for SYBR assay (Eurogentec, Belgium) was used with custom primers (Supplementary Table 2) and mRNA expression was normalized to ACTB. Relative expression levels (fold change, FC) were calculated using the Livak method [30].

Protein extraction, electrophoresis, and immunoblotting

Proteins were extracted in ice cold RIPA buffer (50 mM Tris pH 8.0, 150 mM NaCl, 0.1% sodium-dodecyl-sulphate, 1% Nonidet NP40, 0.5% sodium-deoxycholate, and protease inhibitor (completetrademark Protease Inhibitor Cocktail, Roche, Switzerland)). Cell culture protein extracts were prepared by incubating on ice for 20 min to allow complete protein extraction from cell membranes following centrifugation at 13,000× g for 15 min at 4°C to pellet debris. For tissue protein extracts, 100μl RIPA buffer was used per 10 mg tissue and homogenization was in vials with ceramic beads on the Precellys 24 tissue homogenizer at 65,000× g for 2×30 s runs with a 30 s break in between. Following homogenization, the sample was allowed to rest for 20 min on ice to allow protein extraction followed by ultracentrifugation to remove debris at 100,000× g for 1 h at 4°C in a Sorvalltrademark MTX 150 Micro-Ultracentrifuge (Thermo Fisher Scientific, MA, USA). Protein concentration was estimated by the Bradford assay (Bradford 1976), a standard curve was prepared with standard concentrations of bovine serum albumin (BSA) and triplicates of all samples and standards were measured at 595 nm using the plate reader Infinite M200 PRO (Tecan Ltd., Maennedorf, Switzerland).

Protein extracts were mixed with Laemmli buffer (250 mM Tris pH 6.8 (Carl Roth, Germany), 10% SDS, 1.25% Bromophenol Blue (Sigma-Aldrich, MO, USA), 5% β-mercaptoethanol (Fluka Analytical, Germany), 50% Glycerol (Sigma-Aldrich, MO, USA)) and incubated at 95°C for 5 min. Electrophoresis was performed on 12 or 15% SDS-PAGE gels and proteins were transferred onto nitrocellulose membrane (Thermo Fisher Scientific, MA, USA) using the iBlot 2 Gel Transfer device (Thermo Fisher Scientific, MA, USA) by applying 20 V constantly for 10 min.

Immunoblotting was performed with of one of the primary antibodies: Syn1, mouse, BD; anti-β-actin, mouse, Sigma; anti-GABA Aβ2, rabbit, Abcam; anti-GAPDH, rabbit, Cell Signaling; anti-PSD-95, rabbit, Cell Signalling, anti-SAPAP3, rabbit, Abcam; anti-SNAP-25, rabbit, Synaptic System; anti-Synaptophysin, mouse, Sigma. Protein bands were visualized with a chemiluminescence imaging system (Fusion FX, Vilber Lourmat, Marne-la-Vallée, France). Semi-quantitative analysis of protein levels was performed with densitometric quantification using the Fiji software [31].

Golgi cox staining

Freshly dissected brain tissue was washed once with H₂O, cut in sagittal halves and each half was incubated in a glass vial with Golgi Cox solution (5% w/v K₂Cr₂O₇, Merck, Germany; 5% w/v HgCl₂, Carl Roth, Germany; 5% K₂CrO_4, Carl Roth, Germany) overnight at room temperature in the dark. The next day tissue was transferred in fresh Golgi Cox solution in a new vial and incubated for 10 days at room temperature in the dark. After that, tissue was transferred in a new vial and incubated with tissue protection solution (0.87 M sucrose; 1% w/v polyvinylpyrrolidone, Sigma-Aldrich, MO, USA; 30% v/v ethylene glycol, Sigma-Aldrich, MO, USA; 0.05 M phosphate buffer pH 7.2, Geyer, Germany) at 4°C overnight in the dark. The next day, tissue was transferred to fresh tissue protection solution in a new vial and incubated at 4°C for 5 days. Tissue sectioning was performed with a Leica VT1000 S Vibrating blade microtome vibratome, Leica, Germany. Tissue was embedded in 4% agarose (PEQLAB Biotechnologie, Germany) blocks and sectioned in sagittal, 100μm thick sections, on vibration frequency 60 Hz, at a speed of 15 mm/s and placed onto gelatin coated glass slides. Sections were left to dry overnight before developing the staining. Slides were first washed in distilled water twice for 5 min, then 50% EtOH for 5 min and developed by 8 min in 3:1 ammonia solution, washed with distilled water twice for 5 min and 5% sodium thiosulfate for 10 min in the dark following by a quick 1 min wash with distilled water. Slides were then dehydrated in sequential 6 min incubations with 70, 95, 100% EtOH and xylol. Lastly, when the slides dried, a few drops of quick-hardening mounting medium Eukitt (Fluka Analytical, Germany) were added to apply the cover glass. Slides were left for 48 h in the dark to dry and brightfield images were acquired with confocal microscope.

Synaptosome isolation

Freshly dissected tissue was homogenized in 3 ml homogenization solution (0.32 M sucrose; 1 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid, HEPES, Carl Roth, Germany, pH 7.4; 1 mM MgCl₂, Fluka Analytical, Germany; 0.5 mM CaCl₂, Sigma-Aldrich, MO, USA) in a 5 ml glass-Teflon homogenizer with 12 strokes at 900 rpm on a motorized stirrer (Heidolph Instruments, Germany). Homogenate was transferred to a falcon tube and centrifuged at 1.400× g for 10 min at 4°C. Supernatant was saved and pellet was resuspended with 3 ml homogenization solution and re-homogenized (5 strokes, 900 rpm) to ensure maximum synaptosome extraction. The new homogenate was centrifuged at 710× g for 10 min at 4°C, supernatants were pooled and centrifuged at 13,800× g for 10 min at 4°C. Produced pellet was collected, resuspended in 2 ml resuspension solution (0.32 M sucrose; 1 mM HEPES pH 7.4) with the use of the homogenizer (5 strokes, 900 rpm) and loaded on a sucrose gradient. Sucrose gradient was prepared in Ultra-Clear centrifuge tubes (Beckman Coulter, CA, USA) with 4 ml 1.2 M sucrose, 3 ml 1.0 M sucrose and 3 ml 0.85 M sucrose. Ultracentrifugation was performed at 82,000× g for 2 h at 4°C in a TH641 swinging bucket rotor in an ultracentrifuge (Sorvall Discovery 90SE Thermo Fisher Scientific, MA, USA) and the synaptosome fraction was collected from the 1.0/1.2 M sucrose interface. All solutions were supplemented with RNase inhibitor (ProtectRNAtrademark RNase Inhibitor 500×Concentrate, Thermo Fisher Scientific, MA, USA). Purified synaptosomes were immediately processed for RNA extraction.

Compartmentalized cultures of primary cortical neurons

For optimal separation of neuronal cell bodies and dendrites, cells were cultured in custom made microfluidic devices (MFDs) manufactured from polymethylsiloxane (PDMS) poured in molds designed with soft-lithography in a clean room at the IV Physics Institute, Georg-August-University Göttingen similar to Epple et al. [32]. MDFs consist of 2 chambers connected with microgrooves (500×7500μm). Following polymerization, PDMS design was cut out of the molds, washed with soap water overnight, rinsed excessively with water, washed with EtOH for 2 h and mounted on glass slides by plasma application with a tesla coil. After mounting MFDs were washed with water, EtOH, incubated under UV light to sterilize and coated with PLO. Finally, MFDs were washed and left with culture medium to equilibrate at 37°C at least for 2 h prior cell seeding. Neuronal cells were plated in the MFDs by pipetting 200,000 cells in 60μl culture medium to the top well on one side. After a 10 min incubation, cells have moved to the lower well through the channel and stabilized enough for addition of 300μl neuronal culture medium per well. Cells in MFDs were infected on DIV1 with equimolar amounts of control, scrambled or miR-101a-3p lentivirus by removing the media from the bottom well, incorporation of the viral particles and reapplication on the top well so that viral particles are enriched in the culture channel. MFDs were incubated in petri dishes with a smaller dish with sterile double distilled H₂O to prevent media evaporation from the device. In case evaporation is observed media is added from a sister mass neuronal culture in a standard plate treated in the same manner. Neurons grow in the MFDs for 10 days before further processing for imaging purposes. We did not observe cellular death upon infection.

Immunocytochemistry

Cells grown on coverslips were fixed with 2% PFA at ambient temperature for 20 min. Following incubation with blocking buffer (3% BSA; 0.1% Triton-X100, Sigma-Aldrich, MO, USA; PBS) at ambient temperature for 1 h, cells were incubated at 4°C overnight in anti-MAP2 antibody (rabbit, Proteintech) diluted 1:500 in blocking buffer. Cells were washed with PBS and incubated with Alexa Fluor 555 anti-rabbit IgG secondary antibody (goat, Invitrogen) diluted 1:1000 in blocking buffer at ambient temperature for 2 h or at 4°C overnight for the MFDs. For visualization of the nuclei cells were stained with Hoechst 33258 (2μg/mL, pentahydrate (bis-benzimide), Thermo Fisher Scientific, MA, USA); PBS). Finally, coverslips were mounted on glass slides (Knittel Glass, Germany) with Mowiol mounting medium (1.18 M 1,4 Diazabicyclo (2.2.2) octane, Sigma-Aldrich, MO, USA; 3.6 M glycerol; 0.13 M Tris pH 7.2) and MFDs were stored with PBS at 4°C until imaging.

Microscopy

Imaging of MFDs was performed with a Leica DMI6000 B microscope equipped with a Leica DFC320 color camera (Leica, Germany). Imaging of primary cell culture and brain sections was performed with a TCS SP5 confocal microscope (Leica, Germany). Images were analyzed with Fiji software.

Statistical analysis

Data analysis was performed using the Graphpad software version 5.03 for Windows (GraphPad Software, La Jolla California USA, https://www.graphpad.com). For group comparisons, one-way ANOVA with Dunnet’s post-hoc test was used, while comparisons of two groups of means were done with unpaired t test. All data are expressed as mean±SEM. Differences are considered significant with p < 0.05 (^*p < 0.05, ^**<0.01, ^***<0.001).

Altered midbrain miRNAome in A30P aSyn mice

We used an established Tg mouse model of synucleinopathy expressing familial PD-associated mutant A30P aSyn in order to identify RNA-mediated synaptic alterations. This Tg model displays motor and cognitive impairments at 12 months of age, while accumulation of insoluble aSyn forms is already evident at 3–4 months [24, 34]. Transcriptional analysis of the mouse midbrain at 6 months shows downregulation of biological pathways related to neuronal development and synaptic signaling [35]. Thus, we assessed the miRNAome of the mouse midbrain at 6 months using small RNA-sequencing. Small RNA mapped reads in all samples consisted of 90.27% miRNA and principal component analysis (PCA) indicated a clear separation group. Although two samples in the Wt group are further from the rest, they are clearly separated from the Tg group and were not identified as significant outliers (Supplementary Figure 1a, b). 124 miRNAs were differentially expressed, of which 99 were downregulated and 25 upregulated (Fig. 1a). Gene set enrichment analysis (GSEA) was performed for the predicted miRNA target genes to determine affected processes. Development and morphogenesis were identified as the major processes affected by upregulated miRNAs, while downregulated miRNAs target genes were identified to mostly regulate secretion and, in particular, protein exocytosis (Fig. 1b).

OPEN IN VIEWER

Fig. 1

A30P αsyn expression alters midbrain miRNA expression. a) Heatmap illustrating differentially expressed miRNAs between Wt and Tg mice. The color key indicates expression levels ranging from lower (blue) to higher (red) expression. b) Enrichment ratio of overrepresented Gene Ontology terms of upregulated and downregulated miRNA targets. c) Heatmap illustrating negative interactions among deregulated miRNAs and mRNAs. The color key indicates expression levels ranging from lower (blue) to higher (red). d) Histogram plotting top interacting miRNAs according to number of target genes.

For further analysis we identified the predicted target mRNAs of the deregulated miRNA and correlated them with the deregulated mRNAs from the previously published data set [35]. Then, we selected the negative miRNA-mRNA interactions. This means, only downregulated mRNAs were appointed to upregulated miRNAs and contrariwise. Only those valid miRNA-mRNA interacting pairs were considered for downstream analysis. Finally, in order to select the most relevant pairs, the miRNAs were filtered according to number of target genes, and we selected the top 10 interacting miRNAs (Fig. 1d). This approach obviously restricted the candidate miRNAs but significantly increased the confidence of validity, reducing the possibility of identifying mRNAs targeted via non-canonical interactions or low proficiency matches with low biological impact [36, 37]. We identified the down regulated let-7b-3p (0.19±0.1 FC, p = 0.0003) and miR-486a-3p (0.20±0.08 FC, p = 0.0002) as the top interacting miRNAs with 138 and 104 upregulated target mRNAs respectively (Fig. 1d). Those miRNAs showed a low number of reads indicative of lower expression in the mouse midbrain reducing the chances of producing a strong phenotype alone (Supplementary Figure 1c). Considering the large number of validated targets, we cannot exclude the possibility of synergistic effects from other miRNAs targeting the same genes. Following the top 2 interacting miRNAs, a group of 4 miRNAs showed above 50 target genes. The upregulated miR-101a-3p (2.71±0.44 FC, p = 0.0009) with 60 target genes and the downregulated let-7c-1-3p (0.17±0.08 FC, p = 0.00007), miR-344d-3p (0.27±0.1 FC, p = 0.0004), and miR-543-3p (0.26±0.09 FC, p = 0.00005) with 57, 55, and 53 target genes respectively. Of those, miR-543-3p and miR-101a-3p are the miRNAs with the highest expression levels providing confidence of producing a strong phenotype (Supplementary Figure 1c).

Synaptic miR-101a-3p is deregulated in brain of PD mouse model and DLB patients

In order to validate our findings from bioinformatic analysis we performed qPCR. The relative levels of miR-543-3p were decreased in Tg mice, in agreement with the small RNA-seq data (0.26±0.09 FC, p = 0.00005), but the decrease did not reach significance (0.68±0.13 FC, p = 0.2518) due to variability in control animals (Supplementary Figure 1e). The levels of miR-101a-3p showed significant increase in Tg mice (2.71±0.44 FC, p = 0.0009) (Fig. 2a), in agreement with the small RNA-seq data (7.78±1.71 FC, p = 0.0006), indicating miR-101a-3p for downstream analysis based on our selection approach.

OPEN IN VIEWER

Fig. 2

miR-101a-3p is a synaptic miRNA dysregulated in PD mouse model and DLB patients. a) Real time qPCR validation of miR-101a-3p levels in Wt (n = 8) and Tg (n = 7) mouse midbrain. b) Fold enrichment of overrepresented Gene Ontology terms among miR-101a-3p target genes. GAPDH is used for normalization. c) Bar graph of miR-101a-3p enrichment ratio in cytosol and synaptosomes compared to total homogenate. d) Representative immunoblot of synaptic markers in the crude midbrain homogenate (hom), cytosolic fraction (cyt) and purified synaptosomes (syp) used for miRNA assessment e. miR-101a-3p levels quantified by real-time qPCR in cortex of 8 DLB patients and 14 control samples of healthy individuals.; All data are expressed as mean±SEM; Student’s t-test; ^*p < 0.05, ^**p < 0.01, ^***p < 0.001).

Next, GSEA was performed for the confirmed miR-101a-3p target genes based on the RNA-seq correlation. Analysis of the cellular component unveiled dendritic spines and post synapses as the affected compartments (Fig. 2b), in agreement with the cellular component from the bulk data set (Supplementary Figure 1d).

We then asked if miR-101a-3p acted locally, and specifically at the synapse. To address this, we assessed the levels of miR-101a-3p by qPCR in midbrain cytosolic and synaptosomal fractions of 6-month-old Wt mice, and calculated the miRNA enrichment ratio [18]. Cytosolic miR-101a-3p levels were similar to total homogenate while miR-101a-3p levels in synaptosomes were enriched by 2-fold (2.42±0.26) (Fig. 2c, d). This highlights that miR-101a-3p plays a regulatory role near synapses even in physiological conditions.

The Tg mouse we used displays cortical pathology and cognitive impairments that resemble those present in DLB [24, 33]. For this, we assessed by qPCR the relative expression levels of miR-101a-3p in the cortex of DLB patients obtained from the Newcastle Brain Tissue Resource (Supplementary Table 1) and found a significant increase in DLB samples (4.994±0.93 FC, p = 0.0027) compared to healthy individuals (Fig. 2e) indicating a potentially common epigenetic mechanism with the mouse model driven by miR-101a-3p.

miR-101a-3p effectively targets Dlgap3 and Gabrb2 and down regulates translation in vivo and in vitro

To validate the miRNA-driven synaptic deficits in the Tg mice, we focused on the miR-101a-3p target mRNAs and the post-synapse, as the data analysis indicated. Four downregulated target mRNAs associated with the post-synapse were identified: Dag1 (0.70±0.06 FC, p = 0.00017), Dlgap3 (0.66±0.07 FC, p = 0.0012), Shisa6 (0.53±0.05 FC, p = 0.000001), and Gabrb2 (0.48±0.07 FC, p = 0.000001) (Supplementary Figure 1f). The levels of the mRNAs were assessed in mouse midbrain by qPCR. The levels of Dlgap3 (0.60±0.06 FC, p < 0.0001) and Gabrb2 (0.64±0.14 FC, p = 0.01) were found significantly downregulated (Fig. 3a), while the levels of Dag1 (1.19±0.06 FC, p = 0.28) and Shisa6 (1.28±0.11 FC, p = 0.045) were not altered (Supplementary Figure 1g).

OPEN IN VIEWER

Fig. 3

miR-101a-3p regulates postsynaptic proteins GABA(A)b2 and SAPAP3 in vivo and in vitro. a) Real time qPCR assessment of miR-101a-3p target genes Gabrb2 and Dlgap3 levels in Wt (n = 8) and Tg (n = 7) mouse midbrain. b) miR-101a-3p targeting 3’UTRs of Gabrb2 and Dlgap3 validation by Gaussia luciferase assay (n = 6). Gaussia luciferase activity is normalized to β-galactosidase activity. c) Representative immunoblots and quantification of GABA A b2 and SAPAP3 levels in Wt (n = 4) and Tg (n = 4) mouse midbrain. β-actin is used for normalization. d) Real time qPCR assessment of miR-101a-3p levels and e) target genes Gabrb2 and Dlgap3 in neurons infected with control vector expressing only GFP (control) (n = 4), vector expressing GFP and scrambled miRNA sequence (scrambled) (n = 4) and vector expressing GFP and miR-101a-3p (miR-101a-3p) (n = 4). f) Representative immunoblots and quantification of GABA A b2 and SAPAP3 levels in infected neurons (n = 4). β-actin is used for normalization. All data are expressed as mean±SEM; Student’s t-test; ^*p < 0.05, ^**p < 0.01, ^***p < 0.001).

The interaction of miR-101a-3p and its targeting sequence in the miRNA 3^′-UTR was assessed by the Gaussia luciferase reporter assay. Luciferase activity was significantly inhibited by miR-101a-3p for the constructs containing the Dlgap3 and Gabrb2 3’UTR but not the naïve construct (Fig. 3b), confirming the post-transcriptional regulation of the target genes via the miR-101a-3p binding. Importantly, translation of Dlgap3 and Gabrb2 was significantly reduced as shown by the immunoblotting analysis of the produced proteins SAPAP3 and GABA(A)b2 that showed a significant decrease of 53% and 64% respectively in the mouse midbrain.

To confirm our observations were due to miR-101a-3p increase, we infected neurons with a lentiviral construct expressing miR-101a-3p (Fig. 3d). Infection efficiency and expression levels were equal among conditions (Supplementary Figure 2a, b). Importantly, post-transcriptional downregulation of mRNAs Dlgap3 (0.61±0.20 FC, p = 0.049) and Gabrb2 (0.51±0.24 FC, p = 0.02) was verified (Fig. 3e) and translation of the mRNAs was inhibited resulting in a significant decrease of 54% and 30% for GABA(A)b2 and SAPAP3 respectively (Fig. 3f), suggesting miR-101a-3p specifically targets Gabrb2 and Dlgap3 and successfully inhibits translation.

miR-101a-3p alters dendritic morphology in vivo and in vitro

Bioinformatic analysis identified dendritic spines and post synapses as the compartments affected by miR-101a-3p-mediated processes thus we sought to visualize the mouse midbrain dendritic spines with the Golgi-Cox staining (Fig. 4a, Supplementary Figure 3a). Although the absolute number of dendritic spines was not altered among Wt (6.471±0.1583 spines / 10μm) and Tg animals (6.485±0.3969 spines / 10μm) (Fig. 4b), classification of dendritic spines showed that in Tg mice the mushroom and thin spines, which are considered the more mature and plastic structures, were reduced by 24% (p = 0.0429) and 35% (p = 0.0007) respectively, while the more immature stubby spines, were increased by 69% (p = 0.0002) (Fig. 4c). Representative classification image is presented in Supplementary Figure 3b.

OPEN IN VIEWER

Fig. 4

miR-101a-3p reduces dendritic length and alters dendritic spine morphology in vivo and in vitro. a) Representative Golgi-Cox staining images of the analyzed brain region in Wt mouse; brightfield; scale bar = 200μm; and dendritic spine segments of Wt and Tg animals; brightfield; scale bar = 10μm. b) Quantification of total dendritic spines per 10μm dendrite and c. classified as mushroom, thin and stubby in Wt (n = 3) and Tg (n = 3) mice. (n≥45×10μm segments per condition manually counted with Fiji software). d) Photo of the custom made MFD; upper wells and chamber are filled with red dye and the bottom with blue dye. Fluorescent image of neurons cultured in the bottom chamber and the dendrites growing to the upper chamber through the microgrooves; MAP2 - red; scale bar = 200μm. e) Bar graph of average length (μm) of distal dendrites growing through the microgrooves of MFDs of neurons infected with control vector expressing only GFP (control), vector expressing GFP and scrambled miRNA sequence (scrambled) and vector expressing GFP and miR-101a-3p (miR-101a-3p) (n = 3 individual experiments×3 MFDs per condition). f) Bar graph of average length (μm) of apical dendrites in sparsely cultured infected neurons (n≥25 cells from 4 individual experiments per condition). g) Representative images of dendrites at the exit point to the upper chamber of the MFDs; MAP2 - red; scale bar = 50μm, single infected neurons; GFP – green; MAP2 - red; Hoechst – blue; scale bar = 10μm and dendritic spine segments; MAP2 – red; scale bar = 50μm. h) Quantification of total dendritic spine number and i) classification of mushroom, thin and stubby dendritic spines per 10μm dendrite of infected neurons (n≥32×10μm segments per condition manually counted with Fiji software). All data are expressed as mean±SEM; Student’s t-test; ^*p < 0.05, ^**p < 0.01, ^***p < 0.001).

To study the direct effects of miR-101a-3p on neuronal morphology, neurons were cultured in the custom-made MFDs to allow the separation and visualization of neuronal dendrites which growing through the microgrooves away of the cell bodies (Fig. 4d). miR-101a-3p induction resulted in significantly shorter dendrites (537.2±17.4μm) compared to the scrambled miRNA (683.8±22.11μm) (Fig. 4e). As these numbers account for the distal dendrites, we cultured neurons sparsely in order to assess the apical dendrites which were significantly shorter upon miR-101a-3p expression (54.6±2.8μm) compared to scrambled miRNA (136.6±6.7μm) (Fig. 4g, f).

Dendritic spine assessment in vitro indicated absolute spine number was significantly decreased upon miR-101a-3p expression (5.8±0.4 spines / 10μm) compared to scrambled miRNA (7.3±0.3 spines / 10μm) (Fig. 4g, h). Classification of dendritic spines showed 30% reduction of mushroom spines which accounted for the total spine number reduction as the thin and stubby spines remained unchanged (Fig. 4i). Representative classification image is presented in Supplementary Figure 3c. Collectively, these data confirm miR-101a-3p increase drives the observed post-transcriptional deregulation and the synaptic phenotype linked with early synaptic damage or synaptic plasticity impairments. A direct link between miRNA deregulation and aSyn pathology remains to be examined.

miR-101a-3p expression is triggered by aSyn species and miR-101a-3p shows a dynamic role in neuronal plasticity

To confirm the mechanistic role of miR-101a-3p on synaptic impairments involved in synucleinopathy, we assessed a direct correlation between aSyn and miR-101a-3p. For this, primary neurons were exposed to forms of wt raSyn thought to accumulate during the aggregation process (monomeric, oligomeric, and PFFs) as in Mahul-Mellier et al. [12]. Following incubation, a significant miR-101a-3p increase was identified by qPCR in neuronal cells exposed to monomeric and oligomeric aSyn species (1.75±0.38 FC and 1.75±0.09 FC respectively) but not to PFFs (1.02±0.06 FC) (Fig. 5a). This coincided with synaptic loss in the PFF exposed neurons, estimated by synaptic markers PSD95 and synaptophysin (Fig. 5b), suggesting miR-101a-3p that is enriched in synapses is lost parallel to synaptic loss. With this, we confirm a direct effect of aSyn levels/pathology on the miRNA levels while the opposite interaction was not confirmed as miR-101a-3p induction failed to induce any change in aSyn levels (Supplementary Figure 2c).

OPEN IN VIEWER

Fig. 5

miR-101a-3p is reduced in a model of synaptic plasticity and induced by aSyn. a) Real-time qPCR assessment of miR-101a-3p levels in neurons exposed to monomeric or oligomeric aSyn species and PFFs. b) Representative Immunoblot of synaptic markers in neurons exposed to the different aSyn species. β-actin is used for normalization. c) Real-time qPCR assessment of miR-101a-3p levels in midbrain of mice grown in standard environment (n = 4) or enriched environment (n = 4). All data are expressed as mean±SEM; Student’s t-test; ^*p < 0.05, ^**p < 0.01, ^***p < 0.001).

Finally, to confirm the role of miR-101a-3p in synaptic physiology, we performed a proof-of-concept experiment by assessing miR-101a-3p levels in a model of enhanced synaptic plasticity. We used midbrain from 12-month-old Wt animals housed in an enriched environment (EE) or in a standard environment (SE), and found a significant reduction in the EE group (0.6235±0.02611 FC) (Fig. 5c) confirming the role of this miRNA in synapse physiology and its potential to be modulated by environmental factors and lifestyle choices.