To identify miRNA in the Plaur Mus musculus gene, we employed the following bioinformatic tools. Promoter 2.0² and ElemeNT³ computational tools were used to detect core promoter elements upon screening RNA polymerase III (Pol III) promoter regions (Sloutskin et al., 2015). Putative miRNA precursors were identified by using the miRNA Fold web service⁴ (Supplementary Figure 1; Tav et al., 2016). We selected one from the predicted stem-loop structures based on the stem-loop length, the free energy of the stem-loop formation and consensus motifs enhancing Drosha processing: a basal UG motif, a flanking CNNC motif, a mismatched GHG motif and an apical UGU/GUG motif (Lee and Shin, 2018). Stem-loop structures were visualized by implementing Quickfold (Supplementary Figure 2).⁵ In the selected stem-loops, Drosha cleavage sites were predicted by using MatureBayes (Gkirtzou et al., 2010) and putative mature miRNAs were identified (Supplementary Figure 3).

To identify the putative targets and binding sites of predicted miRNA, we employed a web-based target prediction algorithm, namely MR-microT in DianaTools (Supplementary Figure 4A; Reczko et al., 2012; Paraskevopoulou et al., 2013).⁶ The sequences of predicted miRNAs were used as an input. The mouse genome (Mus musculus, Ensembl v84) was used as a database for target mRNA prediction (Supplementary Figure 4B). We selected several predicted target mRNAs (Supplementary Tables 1, 2) for further verification. To verify the identified targets, small RNA sequencing using the Plaur-miR1 primer was performed. The obtained sequences were mapped in the Mus musculus genome using the BLAST algorithm. Genes that appeared both in the BLAST search and MR-microT prediction were selected for quantitative real-time polymerase chain reaction (qPCR) verification. To assess the specificity of Plaur-miR1-5p and Plaur-mir1-3p to the target gene promoter regions and introns, we took NCBI gene sequences and aligned them against Plaur-miR1-5p and Plaur-mir1-3p via the M-Coffee sequence alignment tool on the T-COFFEE Multiple Sequence Alignment Server web service (Supplementary Figure 4C).

Mouse Neuro2a neuroblastoma cells (ATCC^® CCL-131™ University Boulevard Manassas, VA, United States) not exceeding 20 passages were cultured in complete medium—Dulbecco’s Modified Eagle’s Medium (DMEM) (#21969035), 10% fetal bovine serum (FBS, Gibco, #10270-106, United Kingdom), 1 × Minimum Essential Medium (MEM) Non-Essential Amino Acids Solution (#11140050) and 1 × antibiotic-antimycotic solution (#15240062; all from Gibco, Life Technologies, Bleiswijk, Netherlands)—at 37^°C in an atmosphere with 5% CO₂. Cells were plated at a concentration of 1 × 10⁵ cells/ml. Neuro2a cells with uPAR knockout were obtained by using the CRISPR/Cas9 (Neuro2a KO-Plaur cells) genome editing tool as described previously (Rysenkova et al., 2018).

We had previously shown that Plaur gene expression is induced in various brain structures (Shmakova et al., 2020) in a model of pentylenetetrazole (PTZ)-induced seizures in mice. Here we used miRNA and mRNA isolated from previously obtained brain samples. A detailed description of methodology, animal facility and enabling documentation has been previously published (Shmakova et al., 2020). To assess the endogenous level of Plaur-miR1 induction and expression of its target genes, we selected brain regions with the most significant Plaur mRNA induction 3 h after PTZ (Sigma-Aldrich, cat. # P6500, Saint Louis, MO, United States) administration. Thus, posterior cortex (Plaur induction was 8.7 times higher than control) was enrolled to assess the level of Plaur-mir1-5p and Plaur-mir1-3p and the striatum (Plaur induction was 16 times higher than control)—for the target gene analysis (for further information, see Figure 2 in Shmakova et al., 2020).

To clone Plaur-pre-miR1 (precursor of mature Plaur-miR1), we used a vector for mouse Plaur-pre-miR1 expression originally based on the pBl-U6-CMV-RFP (pBlueScriptII vector from Agilent Santa Clara, CA, United States) plasmid and encoding red fluorescent protein (RFP) for detection (Supplementary Figure 5). Mouse Plaur-pre-miR1 was amplified from Neuro2a genomic DNA using primers listed in Supplementary Table 1. For amplification, we used Phusion High-Fidelity PCR Master Mix (#F531L, Thermo Fisher Scientific, Vilnius, Lithuania) according to the manufacturer’s protocol. The program for template denaturation, primer annealing and primer extension was 40 cycles of 94°C for 15 s, 70^°C for 15 s and 72^°C for 25 s, respectively. The product length was 280 base pairs (bp). The sequences were cloned into the pBl-U6-CMV-RFP plasmid via BbsI (#R0539, New England Biolabs, Ipswich, MA, United States) restriction sites. The pBl-U6-Plaur-pre-miR1 plasmid sequence was verified using the seq u6 primer 5′-CCTATTTCCCATGATTCCTTCATATTTGC-3′ (Supplementary Figure 6; sequencing was performed by Evrogen, Moscow, Russia).

The pBl-U6-Plaur-pre-miR1 vector was transfected in Neuro2a cells with Lipofectamine 2000 according to the manufacturer’s protocol. The Neuro2a cell transfection efficiency was evaluated basing on the RFP fluorescence analysis 24 h after transfection using Leica DMI 6000 B fluorescent microscope and LAS X software (Wetzlar, Germany). Over a period of 48 h, Neuro2a-Plaur-pre-miR1-transfected cells were lysed and a fraction of small RNAs (mirVana miRNA Isolation Kit AM1560, Ambicon, Carlsbad, CA, United States) and total RNA (Quick-RNA MicroPrep R1051, Invitrogen, Freiburg, Germany) was purified to assess the Plaur-miR1 and its target genes expression level.

Short RNAs (< 200 nt small RNAs including pri-miRNAs, pre-miRNAs and mature miRNAs) were isolated from wild type Neuro2a cells, Plaur-deficient cells (Neuro2a KO-Plaur) (Rysenkova et al., 2018), Neuro2a cells overexpressing Plaur-pre-miR1 (three replicates per cell group), and from posterior cortex samples (three animals per group) 0 and 3 h after PTZ treatment according to the manufacturer’s protocol (mirVana miRNA Isolation Kit AM1560, Ambicon, Carlsbad, CA, United States). To generate complementary DNA (cDNA), 500 ng of small RNAs fraction and miScript II RT kit (#218160, Qiagen, Hilden, Germany) was used. PCR was carried out by using qPCR mix-HS SYBR (Evrogen) on a CFX96 real-time PCR device (Bio-Rad, Hercules, CA, United States). qPCR was employed to detect Plaur-miR1-5p and Plaur-miR1-3p from Plaur-pre-miR1. Primers for Plaur-miR1 were designed with NCBI Primer-blast⁷ and the IDT Oligo Analyzer tool (eu.idtdna.com/pages/tools/oligoanalyzer). For reverse primers, we used the commercially available 10 × miScript universal primer from the miScript SYBR^® Green PCR Kit (#218073, Qiagen, Hilden, Germany). All primers are listed in Supplementary Table 1. The thermal cycling program for template denaturation, primer annealing and primer extension was 40 cycles of 94^°C for 15 s, 57^°C for 15 s and 72^°C for 20 s, respectively. The relative transcript level of Plaur-miR1-5p was calculated using the 2^–ΔΔCt method with Snord95 as a reference; normalization was carried out by taking the average level of each transcript in the control as a unit. The PCR products of Plaur-miR1-5p obtained from Neuro2a cells and posterior cortex were cloned into 40 TA plasmids and subjected to Sanger sequencing (performed by Evrogen).

Total RNA was extracted from wild type Neuro2a cells, Plaur-deficient cells (Neuro2a-KO uPAR) and Neuro2a cells overexpressing Plaur-pre-miR1 (three replicates per each group) (Supplementary Figure 7A) as well as from the striatum of control mice (treated with saline) and mice 0.5, 1, 3, 6, 24, and 72 h after PTZ treatment (4–5 animals per group). The Quick-RNA MicroPrep kit with TRIzol (#R1051, Invitrogen, Carlsbad, CA, United States) was used according to the manufacturer’s protocol. The isolated RNA was treated with RNase-free DNAase I (Fermentas, Rockford, IL, United States) and then run on an agarose gel for quality control (Supplementary Figure 7E). To generate cDNA, 1μg of total RNA and the MMLV RT kit (Evrogen) were used. PCR was carried out using qPCR mix-HS SYBR (Evrogen) and the CFX96 Touch Real-Time PCR Detection System (Bio-Rad). qPCR was used to detect the expression of Plaur-miR1-5p target genes. Primers were designed using NCBI Primer-blast (see text footnote 8) and the IDT Oligo Analyzer tool.⁸ All primers are listed in Supplementary Table 1; the Plaur mRNA primers are specific to exon 4. The thermal cycling program for template denaturation, primer annealing and primer extension was 40 cycles of 94°C for 10 s, 60°C for 30 s, and 72°C for 15 s, respectively. The relative transcript level of mRNA was calculated using the 2^–ΔΔCt method with Actb (encodes β-actin) as a reference. The reactions were performed in instrumental triplicates; the results represent the mean of biological triplicates (unless otherwise stated) ± standard error of the mean (SEM).

Brain tissue samples was homogenized and lysed in an ice-cold RIPA lysis buffer as previously described (Shmakova et al., 2020). Proteins (45 μg) were resolved in 10% SDS-PAGE gels and transferred to PVDF membrane (GE Healthcare) in the transfer buffer (25 mM Tris, 192 mM glycine, 0.1% SDS and 20% methanol). Non-specific binding was blocked by 5% non-fat dried milk in phosphate buffered saline (PBS, Sigma-Aldrich), containing 0.1% Tween-20 at + 4°C overnight. Proteins were probed with the following primary antibodies in 1:1,000 dilution: rabbit anti-SNRNP200 (Sigma, HPA029321), rabbit anti-MEF2D (Cell signaling, 25621), rabbit anti-EMX2 (Abcam, ab94713), rabbit anti-β-actin (Cell signaling, 4970S, control of protein load) for 2h at room temperature. Membranes were washed with PBS containing 0.1% Tween-20 and incubated with appropriate peroxidase-conjugated secondary antibodies in 1:10,000 dilution for 1.5h at room temperature, followed by washing in PBS containing 0.1% Tween-20. Proteins were visualized using SuperSignal West Dura Chemiluminescent Substrate (Thermo Fisher Scientific) and ChemiDoc™ XRS + System (Bio-Rad) for Western blotting imaging and analysis. Densitometric analysis of blots at non-saturating exposures was performed using ImageJ. Values of SNRNP200, MEF2D, and EMX2 protein expression were normalized to β-actin. Original uncropped western blot images are presented in Supplementary Figure 10.

DNA fragments encoding the predicted binding sites for Plaur-miR1-5p (3′UTR for Mef2d [positions 3376-3499 and 5140-5276 in NM_001310587.1], Emx2 [positions 2191-2281 in NM_010132.2] and CDS for Snrnp200 [positions 3037-3114 in NM_177214.5]) were amplified from murine cDNA (see Supplementary Table 1 for primers) using Phusion High-Fidelity PCR Master Mix (Thermo Fisher Scientific, # F531L) and inserted in pGL3-promoter vector (Promega, #U47298). HindIII and NcoI restriction sites were used for CDS sequences, while XbaI and RigI restriction sites were used for 3′-UTR. The sequence of the resulting vectors was confirmed using Sanger sequencing.

Neuro2a cells were seeded onto 96-well plate, cultured into a monolayer, and co-transfected with PGL3 plasmids encoding Mef2d 3′-UTR sequences, Emx2 3′-UTR sequence, Snrnp200 CDS sequence or empty pGL3 vector (control) with plasmid pBl-U6-Plaur-pre-miR1 or pBl-U6 vector. 48 h after transfection cells were analyzed using Luciferase Reporter Assay Kit (Promega, Fitchburg, WI, United States). Luminescence was evaluated using Victor™ X3 Multilabel Plate Reader (Perkin-Elmer Inc., United States), the luciferase signal was calculated in Relative luciferase activity units. Data were normalized by luciferase signal in Neuro2a cells co-transfected with empty pGL3 vector and pBl-U6-Plaur-pre-miR1 vector.

We analyzed qPCR data with GraphPad Prism 8.01 (GraphPad Software Inc., San Diego, CA). For cells we analyzed qPCR data using an unpaired t-test, treating the wild type and experimental cells as two independent groups. For comparisons with more than two groups, we used analysis of variance (ANOVA) followed by Dunnett’s multiple comparisons test. Relative luciferase activity data were compared using two-way ANOVA followed by Šídák’s multiple comparisons test. The data are presented as the mean ± SEM. The level of significance was set at P < 0.05.

We first hypothesized that small RNAs are encoded within Plaur as individual genes under specific Pol III promoters. Bioinformatic screening of Plaur employing ElemeNT (Sloutskin et al., 2015) and Promoter 2.0 failed to reveal any specific binding sites for Pol III. Hence, no predicted transcription start sites (TSS) for miRNA genes within Plaur that could give rise to miRNAs via the classical biogenesis pathway were identified. Since there is considerable evidence indicating that miRNAs can be located in the introns of protein-coding genes—called mirtrons (Dokanehiifard et al., 2015, 2017)—we next tested the hypothesis that Plaur can contain such structures. The sequences of predicted miRNAs are summarized in Table 1.

We employed the miRNA Fold⁹ web-based prediction service to reveal miRNA precursors in Plaur (GRCm39, Gene ID: 18793). miRNA Fold predicted 256 stem-loop structures, which are potential miRNA precursors, located within Plaur. We filtered the most stable predicted stem-loops with free energy of formation ≤ −15 kJ/mol (Xue et al., 2005). Such structures occur infrequently, although they are frequently non-random because evolution should theoretically reject them. Among that putative pre-miRNA, three ones had a high probability to exist in vivo based on motifs enhancing Drosha processing: Plaur-pre-miR1 (located in intron 3), Plaur-pre-miR2 (located in intron 3) and Plaur-pre-miR3 (located in intron 6) (Figure 1 and Table 1).

We focussed on Plaur-pre-miR1 since it contains three motifs that enhance Drosha processing (Lee and Shin, 2018), namely the 5′ GUG motif on the apical loop, an unpaired GHG in the downstream part of the stem and a UG motif in the base of the stem-loop structure. It is located upstream of exon 4 (Figure 2A), with free energy of formation −47.6 kJ/mol, indicating its high stability. Due to the presence of these consensus sequences critical for Drosha processing activity, we anticipated a high probability of Plaur-pre-miR1 being a substrate for Drosha processing (Figure 2B).

Hence, we conducted in silico Drosha cleavage of Plaur-pre-miR1 to locate the miRNA precursor within its sequence. We predicted Drosha cleavage sites as well as mature miRNA sequences using MatureBayes. According to the processing rules of the Drosha enzyme, the mature miRNA sequences located at the hairpin 5′ end (5p miRNAs) and predicted in silico corresponded to those predicted by the MatureBayes program (Supplementary Figure 3). However, the sequences located at the 3′ end (3p miRNAs) were mis-predicted by the MatureBayes program since they were located at the apical loop region therefore contradicting the Drosha-mediated processing mechanisms. Hence, we predicted 3p miRNA sequences following the Drosha processing rules: to protrude two nucleotides at the 3′ end of each mature miRNA (Figure 2C).

To verify Plaur-miR1-5p and Plaur-miR1-3p sequences predicted by MatureBayes and to confirm our predictions based on the Drosha processing pattern, we performed multiple sequence alignment of the Plaur region corresponding to Plaur-pre-miR1 in different vertebrate species: Bos taurus, Mus musculus, Canis lupus familiaris, Homo sapiens, Pogona vitticeps, and Poecilia latipinna (Figure 2D). Similar alignment results for Plaur-miR2 and Plaur-miR3 are presented in Supplementary Figure 8. The region corresponding to Plaur-miR1-5p showed maximum occupancy in possible miRNA seed sequence regions; together with a high consensus percentage (50–90%) the Plaur-miR1-5p sequence appeared to be conserved among all analyzed vertebrate species. There was also a TG sequence complementary to UG in RNA among all species in a highly homologous region with respect to the hairpin base. This is consistent with the literature on consensus sequences contributing to the Drosha/DGCR8 microprocessor complex function, located approximately 13 nt downstream the Drosha 5p cleavage site (Auyeung et al., 2013). In contrast, low occupancy and consensus ≤ 50% indicate that the Plaur-miR1-3p sequence was not conserved. Thereby, we assumed that Plaur-miR1-3p is a passenger strand and Plaur-miR1-5p is a novel guide strand. Although Plaur-miR1-5p showed similarity to mmu-miR-7672-3p in miRDataBase encoded within the gene PDE12 (Chr14:26390702-26390763 bp, GRCm39,—strand, according to Blast alignment), an identical miRNA for Plaur-miR1 has not yet been reported in miRbase. Overall, the high Plaur-miR1-5p sequence homology among the analyzed organisms suggests that it has been a subject to natural selection due a potentially important biological function. Assumingly, the 5p mature form performs a guiding function, while the 3p form degrades. In this regard, we have focused primarily on Plaur-miR1-5p and its targets.

Next, using the DianaTools (v84) web server, we performed computational prediction of Plaur-miR1-5p and Plaur-miR1-3p target genes. The following mature miRNA sequences were entered into the search bar: Plaur-miR1-5p 5′-UGGUGAUUGGCUGCCAGGUUC-3′ and Plaur-miR1-3p 5′-AGAACCUGGCCGCCAACA-3′. The search results are presented in Supplementary Table 2 for Plaur-miR1-5p and Supplementary Table 3 for Plaur-miR1-3p, as well as in primary screenshots of the web portal (Supplementary Figure 4B). Since miRNAs engage with various targets, 800 Plaur-miR1 targets were obtained, among which 50 targets had a binding score > 0.9 (set as the screening threshold).

We have previously shown that uPAR plays a role in neuronal cell differentiation and survival (Rysenkova et al., 2020). Moreover, we have found a correlation between high expression of uPAR and induction of neuronal migration to the outer layers of cerebral cortex, as well as Plaur function as an early response gene in the brain, a characteristic that possibly determines uPAR as a morphogen (Shmakova et al., 2021). Hence, we analyzed the possible targets obtained by DianaTools to study in further detail the role of Plaur-miR1-5p and Plaur-miR1-3p with a special focus on targets potentially involved in neuronal differentiation or surveillance, as well as the maturation of brain structures. Emx2 (empty helix homeobox 2) was the target for Plaur-miR1-5p with the highest score (0.998). Emx2 is a transcription factor that plays a pivotal role in the developing brain, determining cell fate in the embryonic central nervous system (Gulisano et al., 1996). For Plaur-miR1-3p, Mef2d (myocyte enhancement factor 2D) was a target with a highest score (0.989). Mef2d is a transcription activator that plays a key role in the regulation of neuronal apoptosis (Wang et al., 2009; Assali et al., 2019). Moreover, Mef2d also was a target for Plaur-miR1-5p with a score (0.485) as detected by DianaTools (Supplementary Figure 4C). Therefore, we analyzed Emx2 and Mef2d mRNA expression in Neuro2a control cells, Neuro2a KO-Plaur and Neuro2a-Plaur-miR1 cells (ectopic Plaur-miR1 expression) to assess the impact of Plaur-miR1-5p and Plaur-miR1-3p on their targets. Moreover, Plaur-miR1-5p and Plaur-miR1-3p have a recognition site in the Emx2 and Mef2d gene promoters (Figure 3A) and introns (Figure 3B and Supplementary Figure 4C), suggesting that Plaur-miR1 regulates the expression of these target genes at the nuclear level, including their expression induction. To examine the Plaur-miR1-5p specificity to the promoter region [600 nt upstream of the TSS, according to Ensembl genome browser]¹⁰ and target gene introns, we used intronic sequences from NCBI and aligned them with Plaur-miR1-5p and Plaur-miR1-3p via the M-Coffee tool of the T-COFFEE Multiple Sequence Alignment Server web service.

The obtained bioinformatic data on the existence of murine Plaur-miR1-5p and Plaur-miR1-3p was verified in Neuro2a cells and posterior cortex of C57BL/6J mice. Since the studied miRNAs are located in Plaur introns we used previously obtained Neuro2a cells with CRISPR/Cas9n-mediated Plaur knockout (Neuro2a KO-Plaur) (Rysenkova et al., 2018; Semina et al., 2020) as a negative control and wild type Neuro2a cells (Neuro2a WT) endogenously expressing Plaur for measuring the miRNA levels in Neuro2a cells.

Plaur-miR1-5p and Plaur-miR1-3p were detected in Neuro2a WT cells but not in Neuro2a KO-Plaur cells (Figure 4A). These findings suggest that these miRNAs are expressed in Neuro2a cells and their expression level is Plaur dependent. Moreover, transfection of Neuro2a cells with the pB1-miR plasmid to overexpress Plaur-miR1-5p and Plaur-miR1-3p increased the content of these miRNAs in Neuro2a cells (Neuro2a WT Plaur-miR1 cell in Figure 4A) and restored their expression in Neuro2a KO-Plaur cells (Neuro2a KO WT Plaur-miR1 cell in Figure 4A). Plaur mRNA expression was verified by qPCR (Figure 4B). The specificity of the observed effects (Figure 4A) was further confirmed by the lack of change in uPAR expression level in Neuro2a cells transfected with the pB1-miR plasmid for Plaur-miR1-5p and Plaur-miR1-3p overexpression (Neuro2a WT Plaur-miR1 and Neuro2a KO uPAR Plaur-miR1 cells in Figure 4B).

Subsequently, we analyzed the Plaur-miR1-5p and Plaur-miR1-3p expression in the mouse brain (posterior cortex) in control conditions and 3 h after PTZ-induced Plaur expression (Shmakova et al., 2020). Surprisingly, we revealed not only Plaur-miR1-5p and Plaur-miR1-3p expression in the posterior cortex, but their expression was increased by 1.9 and 2.6 times, respectively, after PTZ treatment (Figure 4C).

qPCR using a small RNA matrix (< 200 nt) with primers for Plaur-miR1-5p yielded a single product with an approximate size of 50 nt (Supplementary Figure 9). To confirm the specificity of qPCR performed with Plaur-miR1-5p primers and to establish the nucleotide sequence of all PCR products in this reaction, we sequenced the qPCR product with the Plaur-miR1-5p primers using miRNA samples from posterior cerebral cortex. Sequencing of PCR products and their cloning into TA vector was carried out by Evrogen. The results of 19 vector clones of TA containing Plaur-miR1-5p sequences are shown in Figure 4D. Clones 1, 9, 11, and 16 demonstrate the sequence similarity in 15 out of 22 nucleotides to Plaur-miR1 (Figure 4E) and are 22–24 nt in length. All other sequences range in size from 31 to 134 nt suggesting that they may be related to other RNA fragments in small RNA fraction (< 200 nt).

The sequencing results revealed that only 20% (4 out of 19 clones) were the target products of Plaur-miR1-5p. Actually, an accurate detection of the relative Plaur-miR1-5p expression in posterior cortex may not be feasible at this stage and may be masked due to the presence of by-products. One of the possible reasons would be different levels of the Plaur and Plaur-miR1-5p induction (8.8-folds for Plaur, refer to Figure 2 in Shmakova et al. (2020); 2.9-folds for Plaur-miR1-5p (Figure 4C) in posterior cortex. Beyond that, the induction difference could stem from rapid degradation of Plaur-miR1-5p in the cytoplasm due to interaction with its targets (Agrawal et al., 2003).

Nevertheless, the qPCR product sequencing results allowed us to determine the possible targets for Plaur-miR1-5p using experimental approach. For that, we aligned the obtained sequencing products > 31 nt in length against Mus musculus genome GRCm39 using the BLAST algorithm. Two genes appeared in the BLAST search: Snrnp200 (U5 small nuclear ribonucleoprotein) and Nrip3 (nuclear receptor interacting protein 3). We then analyzed the mRNA expression levels of these genes by qPCR. Of note, both Snrnp200 and Nrip3 had rather low score values according to the prediction of targets in DianaTools: 0.48 and 0.38, respectively.

The established Plaur-dependent expression of Plaur-miR1-3p and Plaur-miR1-5p in the mouse brain prompted us to evaluate the expression of Plaur-miR1-3p and Plaur-miR1-5p target genes—Emx2 and Mef2d identified via DianaTools—as well as Snrnp200 and Nrip3, identified in qPCR products from posterior cortex samples. The target gene expression was assessed in the striatum, which, as we have previously reported, exhibited the highest increase in Plaur expression after PTZ treatment: 16 times compared with the control group, as presented in Figure 3 in the paper by Shmakova et al. (2020). Mouse striatum samples were enrolled for analysis at the baseline and 0.5, 1, 3, 6, 24, and 72 h after PTZ administration. We found a significantly increased expression of Mef2d, Emx2, and Snrnp200 mRNA (Figures 5A–C). Moreover, the induction dynamics was consistent with the Plaur expression dynamics after PTZ treatment, as shown in Figure 2 by Shmakova and co-authors (Shmakova et al., 2020). Meanwhile, the Mef2d and Snrnp200 expression remained elevated by more than 2-folds 72 h after PTZ treatment. Emx2 expression was elevated up to for 6 h after PTZ treatment, with a maximum increase of 2.97-folds after 3 h as compared with endogenous Plaur expression. Nrip3 expression remained unchanged at all the tested time points (Figure 5D).

We also confirmed an increased expression of MEF2D protein in the mouse brain: MEF2D expression peaked 6h after PTZ administration (Figure 6A). The level of SNRNP200 and EMX2 proteins remained unchanged 3 and 6 h after PTZ administration (Figures 6B,C) suggesting that these genes may be subject to a different transcriptional and translational regulation, which requires further investigation.

To further expolore the relationship between the expression of Plaur-miR1 and its target genes, we analyzed Nrip3, Snrnp200, Emx2, and Mef2d mRNA expression in relation to Plaur and Plaur-pre-miR1 expression in Neuro2a cells using qPCR. Plaur knockout markedly reduced Mef2d and Snrnp200 expression compared with Neuro2a WT cells: 50-folds for Mef2d and 16.4-folds for Snrnp200 (Neuro2a KO cells in Figure 7). Emx2 and Nrip3 expression in the Neuro2a cell was below the detection level. Neuro2a control cells transfected with the pBl-U6-Plaur-pre-miR1 plasmid to overexpress Plaur-miR1-3p and Plaur-miR1-5p showed significantly increased Mef2d expression but not Snrnp200 (Neuro2a WT Plaur-miR1 cells in Figure 7). Moreover, the Plaur-miR1-3p and Plaur-miR1-5p expression restored Mef2d and Snrnp200 expression (in Neuro2a KO uPAR Plaur-miR1 cells in Figure 7), indicating that the expression of the Plaur-miR1-3p and Plaur-miR1-5p targets in Neuro2a cells is strongly dependent on the expression of Plaur and Plaur-miR1-3p and Plaur-miR1-5p transcribed from the Plaur gene. Thus, we have identified two small RNAs, Plaur-miR1-3p and Plaur-miR1-5p, encoded in intron 3 of the Plaur gene, as well as the Plaur-miR1-3p and Plaur-miR1-5p target genes—Emx2, Mef2d, and Snrnp200.

Since the sequencing of PCR products in cerebral cortex of murine brain (posterior cortex) with primers for Plaur-miR1-5p revealed Plaur-miR1-5p expression, we next addressed if there is a direct interaction between Plaur-miR1-5p and its target genes Mef2d, Snrnp200 and Emx2 . Using DianaTools (v84) web server we predicted microRNA response elements (MRE) for the Mef2d and Emx2 genes in their 3′-UTRs and for the Snrnp200 gene in its CDS sequence (Supplementary Figure 11). We cloned 3′-UTR for Emx2, Mef2d and CDS for Snrnp200 into a pGL3-reporter vector downstream of Renilla luciferase ORF and co-transfected Neuro2A cells with these constructs and pBL-Plaur-pre-miR1 plasmid; pBl-control vector was used as a control (Figure 8). We found that co-transfection of PGL3 vector containing Mef2d 3′-UTR sequence (positions 3376-3499 in NM_001310587.1) with pBL-Plaur-pre-miR1 resulted in a significant ∼1.4-fold increase in luciferase activity (Figure 8, p = 0.0178) compared to co-transfection with pBl-control vector. Co-expression of pBL-Plaur-pre-miR1 plasmid with pGL3-reporter vectors containing 3′-UTR sequence of Mef2d outside the mentioned region (positions 5140-5276 in NM_001310587.1), 3′-UTR sequence of Emx2 (positions 2191-2281 in NM_010132.2) or CDS sequence of Snrnp200 (positions 3037-3114 in NM_177214.5) revealed no significant change in luciferase activity compared with pBl-control vector (Figure 8). These results confirmed that Plaur-miR1-5p can specifically increase the Mef2d expression through its 3′-UTR site.