Breeding and identification of miR-22 knockout mice:
Mice carrying a knockout (null) mutation of mir-22 were purchased from the Jackson laboratory (Mir22tm1.1Arod, Stock No: 018155) . Mice were mated in trios (one male and two females). For breeding, either a single heterozygous male was mated with two homozygous mutant female mice or a single homozygous male mated with two heterozygous females. Wild type mice were also mated in trios. Mice were housed in a climate-controlled biomedical facility on a 12-hour light/dark cycle with food and water provided ad libitum. Primers for genotyping were (Forward (common), 5’ TGG GAC TTG GGT TCT ACA CC 3’; Reverse (wildtype), 5’ TCC TAA AAG GAA GGG GAG GA 3’; Reverse (miR-22 mutant), 5’ TGC TTT AGG TGG AGG GAA AG 3’.
Status epilepticus and epilepsy monitoring:
Status epilepticus was induced by unilateral intraamygdala microinjection of kainic acid, as described [16, 24]. Briefly, animals were anesthetized with isoflurane, placed in a stereotaxic frame and maintained normothermic by means of a feedback-controlled heat blanket. A craniotomy was performed and a guide cannula positioned over the surface of the dura. For acute EEG recordings, mice were equipped with three skull-mounted screw electrodes (Bilaney Consultants, U.K.) and connected to a Grass 40 channel lab-based digital EEG. For long-term epilepsy monitoring, mice were equipped with an EEG telemetry device (Model: F20-EET, Data Systems International) inserted subcutaneously and connected to skull screws. After baseline EEG recordings, a cannula was inserted to inject kainic acid (0.3 µg/0.2 μl in phosphate-buffered saline; PBS) or PBS into the basolateral amygdala nucleus. After 40 minutes, all mice received an intraperitoneal injection of lorazepam (8 mg/kg) to curtail seizures and reduce morbidity and mortality. EEG recordings during acute status epilepticus were analysed as before , including total power, amplitude and number of electrographic seizures. For long-term EEG monitoring, the number and duration of spontaneous seizures were manually scored, with epileptic seizures being defined as high-amplitude (>2X baseline) polyspike discharges of ≥10 seconds duration . Mice were killed by pentobarbital overdose and perfused with PBS. Brains were then either microdissected and frozen or retained whole for sectioning.
Rotarod and elevated plus maze
The rotarod test was used to assess motor coordination and balance. After a habituation trial day, mice were placed on a horizontally oriented, rotating suspended cylinder and the length of time that mice stayed on the rotating rod was measured during a gradual acceleration from 0 to 40 rpm over 120 seconds. The average time each mouse spent on the accelerating rotating rod was calculated. The elevated plus maze apparatus comprised two open arms (25 cm length x 5 cm width) and two closed arms (25 cm length x 5 cm width x 16 cm height walls to enclose the arms) placed in the centre of the testing room, 50 cm above the floor. The test consisted of a single trial lasting five minutes. Mice were individually placed in the centre facing one of the open arms. The number of entries into the open and closed arms and the time spent was recorded. These parameters were used to calculate the index of anxiety-like behaviour.
Immunohistochemistry was performed as described , using post-fixed fresh frozen or paraformaldehyde-perfused sections. Sections were blocked in goat serum followed by incubation with primary antibodies against NeuN, glial fibrillary acidic protein (GFAP), ionized calcium binding adaptor molecule 1 (Iba1) or parvalbumin (all from Cell Signaling Technology). Negative controls were included that omitted the primary antibody. Slides were washed then incubated with secondary antibodies, rinsed again and mounted. Positive cells were quantified and were the average of two adjacent sections. Neuronal damage was assessed based on a graded five-point system under 40X magnification: level 0, no damage; level 1, 0–25%; level 2, 25–50%; level 3, 50–75%; level 5, 75% cell loss.
Analysis of miRNA expression
Expression of individual miRNAs was performed using Taqman individual miRNA assays as described . Levels of miR-22 and a selection of miRNAs enriched in neurons (miR-124a, miR-134), microglia (miR-342, miR-150) and astrocytes (miR-146a, miR-29a) was based on in situ hybrisation and in vitro cell culture studies [21-23, 38]. Five microliters of RNA (500 ng/µl) was mixed with RT buffer, RNAase inhibitor, multiscribe reverse transcriptase enzyme, dNTPs and the microRNA specific RT primer (Applied Biosystems) and placed in a thermocycler. Tested miRNAs were: miR-124a, miR-134, miR-150, miR-29a, miR-146a, miR-342. The generated cDNA was diluted, mixed with TaqMan Fast Universal PCR Master Mix and miRNA-specific PCR primers and ran in triplicate and analysed using the QuantStudio™ 12K Flex PCR system as described .
RNA sequencing preparation and analysis
Wildtype and miR-22-/- mice (n = 4 each) were subjected to status epilepticus as described above. Twenty-four hours later, mice were euthanised and the ipsilateral hippocampus was obtained. Total RNA was extracted and library preparation was performed using an Illumina TruSeq Stranded mRNA Sample Prep Kit (poly-A enrichment). The generated RNA-Seq data were deposited in the Sequence Read Archive under GEO accession number GSE147466. Follow-up analysis including QC, alignment (reference genome GRCm38, Annotation_Ensembl70), mapping and raw analysis was performed by a service provider (Exiqon A/S, Vedbaek, Denmark). On average 41.3 million reads were obtained for each sample and the average genome mapping rate was 94.4%. Numbers of identified genes ranged from 21,260 to 22,315.
To reflect the mostly subtle effects of miRNAs on gene expression, only genes with small fold changes (less than two times, p < 0.01) were included in the pathway enrichment analysis. A total of 67 genes were searched against Reactome  and KEGG databases  using ReactomePA and EnrichR. Genes involved in immune-associated pathways of the top five most significantly (p < 0.05) enriched pathways from each analysis were selected and combined to analyse further. Potential transcription factors for these genes were retrieved from Tf2DNA database .
To identify targets of mmu-miR-22, experimentally validated targets were retrieved from miRTarBase Release 7.0, TarBase v.8 and miRecords while predicted targets were retrieved from TargetScan Release 7.2  and miRDB Version 6.0  and processed as described previously , with some modifications. Briefly, prediction scores of TargetScan targets were rescaled between 0 and 1 while those of miRDB targets were rescaled between 0.5 and 1 (since original miRDB database excluded all targets with scores < 50). Then, any targets with rescaled prediction scores < 0.5 were removed from further analysis. All data processing, analyses and visualisation were performed using R and bash scripts in Unix working environment.
Data analysis: Data are presented as mean ± SEM. Comparison of data was performed using ANOVA with appropriate post hoc test. EEG variables (e.g., total power, frequency) were analysed using the parametric Student's t-test, considering P < 0.05 as significant. For the long-term experiments, Zero-inflated Poisson regression was used to compare daily seizure counts between treatment groups. Robust variance estimation was used to correct for the effects of measurements within animals. Significance was accepted when p < 0.05.