Abstract

   Calpain-1 knock-out (KO) mice exhibit enhanced susceptibility to
   neurodegeneration due to the lack of the neuroprotective function of
   calpain-1. Dicer has been shown to play a fundamental role in the
   biogenesis of most miRNAs. Here, we identified 45 differentially
   expressed miRNAs (DE miRNAs) in the brain of calpain-1 KO mice, as
   compared to wild-type mice. In particular, among all the DE miRNAs, 7
   neurodegeneration-related miRNAs were found to be down-regulated in
   calpain-1 KO mice. We also found that Dicer is cleaved by calpain-1 in
   mouse brain, which generates an active fragment of Dicer with RNAse III
   activity and increases miRNA formation. Levels of active Dicer were
   reduced in brain homogenates from calpain-1 KO mice and incubation with
   calpain-1 and calcium restored Dicer activity and miRNA expression. Our
   results indicate that calpain-1 deletion results in decreased levels of
   active Dicer and changes in neurodegenerative-related miRNAs. These
   findings could account for some of the pathological changes found in
   brain of various mammals, including humans, with calpain-1 mutations or
   down-regulation.

Graphical abstract

   Schematic representation of the effects of calpain-1 down-regulation on
   miRNA generation. Left. The activation of Dicer by calpain-1-mediated
   cleavage contributes to miRNA synthesis. Levels of active Dicer and of
   several miRNAs are decreased in brain of calpain-1 KO mice. Incubation
   of brain homogenates from calpain-1 KO mice with calpain-1 restores
   active Dicer and enhances miRNAs synthesis. Red fonts represent miRNAs
   which were validated by RT-qPCR.

   [30]Image, graphical abstract
   [31]Open in a new tab

1. Introduction

   In recent years, a new class of non-coding small RNAs, microRNAs
   (miRNAs), has been shown to play critical roles in regulating gene
   expression in various tissues [32][1]. So far, more than 700 miRNAs
   have been identified in humans, with an estimated target of >30% of all
   human genes [[33]2,[34]3]. Alterations in miRNAs have been implicated
   in several neurodevelopmental disorders [35][4], and more recently in
   neurodegenerative disorders [36][5]. In particular, a subset of miRNAs
   appears to be commonly deregulated in the majority of neurodegenerative
   diseases, including Alzheimer's disease (AD) [37][6], Parkinson's
   disease (PD) [38][7], amyotrophic lateral sclerosis (ALS) [39][8] and
   Huntington's disease (HD) [40][9]. Because miRNAs control the
   expression of large networks of genes, understanding the pattern of
   miRNA deregulation in these diseases might provide new clues regarding
   their underlying complex mechanisms. Moreover, this understanding could
   lead to the identification of specific sets of miRNAs that could be
   used to develop new therapeutic strategies.

   The machinery involved in miRNA biogenesis is well conserved across
   species and consists in several stages: i) miRNAs are first transcribed
   from genes localized within introns of both protein coding or
   non-coding genes, generating pri-miRNAs [41][10]; ii) these pri-miRNAs
   are cleaved within the nucleus by an RNase called Drosha, to produce
   pre-miRNAs; iii) pre-miRNAs are exported from the nucleus and further
   processed into mature miRNAs by another RNase, Dicer [[42]10,[43]11].
   In neurons, this cleavage can take place in many subcellular
   compartments, including dendrites, axons or cell bodies. The
   double-stranded miRNAs then separate into single-stranded miRNAs, which
   are stabilized by the Argonaute-2 protein and incorporated into the
   RNA-induced silencing complex (RISC), which then inhibits the
   translation of the complementary gene. Alterations of miRNA levels
   could thus be due to a variety of different mechanisms related to the
   production or degradation of miRNAs [44][12].

   We recently reported that 354 genes were differentially expressed in
   brain of calpain-1 knock-out (KO) mice [45][13]. Most of them were
   down-regulated and could be classified into 10 KEGG pathways, including
   the Alzheimer's disease pathway [46][13]. Expression of several
   transcription factors was also significantly altered in brain of
   calpain-1 KO mice. Interestingly, Lugli et al. [47][14] reported that
   calpain could cleave Dicer to produce a fragment with RNAse activity,
   allowing for the rapid production of miRNAs from circulating
   pre-miRNAs. Since the nature of the calpain isoform was not identified
   in their study, we were interested in evaluating the possibility that
   at least some of the changes in mRNAs we observed in calpain-1 KO mice
   could be related to changes in the levels of active Dicer and
   consequently in the expression of miRNAs. We were particularly
   interested in miRNAs related to neurodegenerative disorders, since
   calpain-1 is neuroprotective, as down-regulation of calpain-1 is
   associated with enhanced susceptibility to neurodegeneration in mice,
   dogs and humans [48][15], [49][16], [50][17], [51][18], [52][19]. In
   particular, calpain-1 KO mice exhibit enhanced apoptosis throughout the
   brain during the postnatal period [53][20], and exhibit more
   neurodegeneration in a model of acute glaucoma [54][17], in a model of
   traumatic brain injury [55][21] and following Kainic acid-induced
   status epilepticus [56][22]. Our results indicate that Dicer is cleaved
   by calpain-1 and that active Dicer levels are decreased in brain of
   calpain-1 KO mice, resulting in decreased levels of miRNAs, with
   several of them previously found deregulated in neurodegenerative
   disorders.

2. Material and methods

2.1. Mice

   Animal experiments were performed in accordance with NIH guidelines.
   All protocols were approved by the Institutional Animal Care and Use
   committee, Western University of Health Sciences. Calpain-1 KO mice on
   C57BL/6 background were provided by Dr. Chishti (Tufts University).
   C57BL/6 mice were purchased from Jackson Labs and used as wildtype (WT)
   controls. Two to four month-old male mice were used in all experiments.

2.2. RNA isolation and miRNA sequencing

   Total RNA including miRNA was extracted from whole brain of WT and
   calpain-1 KO mice using miRNeasy mini kit (QIAGEN) according to the
   manufacturer's instruction. Quality and quantity of RNAs were assessed
   by NanoDrop 2000 spectrophotometer and agarose gel. miRNA library
   preparation was performed from 300 ng of total RNA following the
   Illumina protocol with minor modifications (City of Hope, California).
   Briefly, pooled total RNA was ligated to the sRNA 3′ adaptor with T4
   RNA Ligase 2, truncated (New England BioLabs) for 1 h at 22°C, and
   subsequently ligated to a 5′ adaptor with T4 RNA ligase1 (New England
   BioLabs) for 1 h at 20°C. The constructed library was first
   reverse-transcribed using GX1 (5′- GGAGTTCCTTGGCACCCGAGA) as the RT
   primer, then subjected to PCR amplification for 12 cycles, followed by
   6% TBE PAGE gel purification with size selection. The purified library
   was quantified using qPCR and sequencing was done using the HiSeq 2500
   platform (Illumina, USA).

2.3. Processing of raw data and identification of differentially expressed
miRNAs

   The first 3 nucleotides and adapters were removed by Cutadapt [57][23],
   a software, which can find and remove primers, adapter sequences and
   other unwanted sequences from high-throughput sequencing reads. The
   clean small RNA reads were aligned to the mouse mm10 genome by using
   Bowtie [58][24]. The read abundance of miRNAs was normalized to CPM
   (Count-Per-Million). Genes with CPM values greater than 5 in 50% of the
   samples were considered for statistical analysis, and Log[2](CPM+1)
   expression values were used for visualization. Principal component
   analysis (PCA) of the dataset was done using R software version 3.3.
   The differential expression was performed with limma-voom [59][25] and
   miRNAs with p-value less than 0.05 between WT and calpain-1 KO mice
   were defined as differentially expressed miRNAs.

2.4. Analysis of miRNA target prediction

   To predict possible targets for each miRNA and to reduce the number of
   false positives, miRNA target analysis was performed using three
   different target prediction online tools, consisting of TargetScan
   ([60]http://www.targetscan.org/mmu_72/), miRWalk [61][26] and miRmap
   [62][27]. The results obtained with the three tools were compared, and
   only the targets predicted by all three tools were considered.

   To understand miRNAs regulatory networks in mouse brain, we identified
   the target transcription factors (TF) interactions for each DE miRNAs
   from the TransmiR v2.0 database [63][28]. The network was visualized
   using Cytoscape software version 3.7.2 ([64]https://cytoscape.org/).
   KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway enrichment
   analysis for target genes was performed by KOBAS 3.0
   ([65]http://kobas.cbi.pku.edu.cn/kobas3/?t=1).

2.5. Preparation of whole brain homogenates and Western blot analysis

   Whole brains from WT and calpain-1 KO mice were homogenized in a lysis
   buffer (10 mM Tris, pH 8, 1 mM EDTA, 140 mM NaCl, 0.5 mM EGTA, 1%
   NP-40, 0.5% sodium deoxycholate, and 0.1% SDS) containing a cocktail of
   protease inhibitors. Homogenates were centrifuged at 13,000 g for 10
   min at 4°C and the supernatant was transferred to a new tube for
   further analysis. Protein concentration was determined using the BCA
   protein assay (Pierce Biotechnology). Aliquots of proteins were
   separated on 8% SDS-PAGE and Dicer protein was detected with an
   anti-Dicer antibody (sc-136979, Santa Cruz Biotechnology) used at 1:500
   dilution followed by secondary antibody goat anti-mouse IgG IRDye 800CW
   (1:10,000; LI-COR Biosciences). β-actin was visualized using the
   anti-β-actin antibody (A5441; 1:10,000; Sigma-Aldrich) and was used as
   a loading control. Western blots were analyzed with the LI-COR Odyssey
   system (LI-COR Biosciences). The quantification of the relative levels
   of protein expression was performed using ImageStudio software.

2.6. Treatment of whole brain homogenates with purified calpain-1

   Mouse brains were homogenized in a lysis buffer (ThermoFisher
   Scientific). The homogenates were centrifuged at 13,000 g for 10 min at
   4°C. Equal concentrations of proteins were incubated with purified
   human calpain-1 (Calbiochem) in the presence or absence of 5 mM calcium
   for 15 min at 37°C. Calpain-1 reaction was stopped by adding loading
   buffer as previously described [66][29], followed by boiling the
   samples at 95°C for 10 min and the samples were then subjected to
   Western blot. For miRNA expression experimental validation, calpain
   activity was stopped by adding 5 mM EDTA. Reactions continued for
   another h at 37°C before RNA extraction.

2.7. miRNA expression validation by RT-qPCR

   Two miRNAs, miR-146a-5p (Assay ID: 000468) and miR-29a-3p (Assay ID:
   002112), were selected for RT-qPCR validation; both have been reported
   as the most often dysregulated miRNAs in neurodegenerative diseases
   [67][30]. To confirm their expression, 10 ng total RNA samples were
   reverse-transcribed using MultiScribe reverse transcriptase and
   stem-loop primers from TaqMan microRNA reverse transcription kit and
   TaqMan microRNA assay (ThermoFisher Scientific), respectively. The
   snoRNA202 (Assay ID: 001232) was used as the endogenous control for
   mature miRNAs. The 2^−∆∆Ct method was adopted for quantification.
   RT-qPCR was performed with 4-6 biological replicates.

2.8. Statistical analysis

   Statistical and correlational analyses were carried out by using
   GraphPad Prism 7.0 software. The results of independent biological
   experiments are expressed as means ± SD. Statistical evaluation was
   performed using unpaired Student's t-test and a p < 0.05 was considered
   significant, *p < 0.05, **p < 0.01.

3. Results

3.1. Overall miRNA expression profile in brains of WT and calpain-1 KO mice

   The average sequence reads from miRNA-seq for WT and calpain-1 KO
   samples were 20,554,629 and 17,855,127, respectively. After quality
   control and normalization, we found 394 miRNAs expressed in brains from
   both WT and calpain-1 KO mice (Table S1). The most abundant miRNAs
   found from mouse brains were miR-26a-5p, miR-124-3p, miR-9-5p,
   let-7g-5p and miR-99a-5p ([68]Fig. 1A). The expression of the top 10
   most abundant miRNAs was relatively similar between WT and calpain-1
   KO. To screen target genes of these top 10 most abundant miRNAs, we
   performed pathway enrichment analysis. Interestingly, the mTOR and Rap1
   signaling pathways were the top 2 significantly enriched pathways in
   the target genes ([69]Fig. 1B). A clear separation between calpain-1 KO
   and WT miRNA expression was identified by displaying the relationships
   among miRNA expression patterns using PCA (Fig. S1). miRNA expression
   profiles were grouped by PC1 (principal component 1), which accounted
   for 99.81% of the overall miRNA expression difference, whereas PC2
   (principal component 2) did not distinguish between the samples. By
   using the separation with PC1 and applying a criterion of p < 0.05 for
   the difference in expression, we identified 45 differentially expressed
   miRNAs, including 11 up-regulated and 34 down-regulated DE miRNAs in
   brains of calpain-1 KO, as compared to WT mice ([70]Table 1). The
   relative expression of the DE miRNAs between WT and calpain-1 KO mice
   is displayed as a heat map in [71]Fig. 2A.

Fig. 1.

   [72]Fig 1
   [73]Open in a new tab

   Overview of miRNA-seq data. (A) Distribution of the top 10 most
   abundant miRNAs in mouse brain. miRNA expression was similar in WT and
   calpain-1 KO mice for the top 10 most abundant miRNAs. The y-axis
   represents the mean normalized number of reads, with the error bars
   representing SD. (B) Heatmap of the KEGG pathway enrichment analysis of
   the target genes (528) of the top 10 most abundant miRNAs. The top 10
   pathways are shown. The color indicates the enrichment score for each
   KEGG pathway.

Table 1.

   Differentially expressed miRNAs in brains of Calpain-1 KO vs. WT mice.
   Listed are the miRNA significantly different *p < 0.05) between C1KO
   and WT mice.
        miRNA      C1KO % Change      miRNA      C1KO % Change
   mmu-miR-350-3p  down -13.6    mmu-miR-582-5p  down -16.9
   mmu-miR-29c-3p  down -16.2    mmu-miR-19a-3p  down -28.1
   mmu-miR-1251-5p down -24.1    mmu-miR-30b-3p  up   +17
   mmu-miR-146a-5p down -16      mmu-miR-33-5p   down -16
   mmu-miR-132-5p  down -11      mmu-miR-193b-3p up   +26.3
   mmu-miR-144-3p  down -23.7    mmu-miR-130b-5p down -21.1
   mmu-miR-451a    down -16.6    mmu-miR-99b-5p  up   +8
   mmu-miR-301a-3p down -21.6    mmu-miR-10b-5p  down -43.5
   mmu-miR-142a-5p down -21      mmu-miR-32-5p   down -11.7
   mmu-miR-142a-3p down -15.4    mmu-miR-101a-3p down -19.3
   mmu-miR-10a-5p  down -26.9    mmu-miR-30e-5p  down -11.7
   mmu-miR-3068-5p down -12      mmu-miR-429-3p  down -47.5
   mmu-miR-770-3p  up   +13      mmu-miR-200a-3p down -56.6
   mmu-miR-136-5p  down -9       mmu-miR-200b-3p down -47.9
   mmu-miR-136-3p  down -12.9    mmu-miR-106b-5p down -11.8
   mmu-miR-370-3p  up   +7.3     mmu-miR-29a-3p  down -11.4
   mmu-miR-379-5p  up   +10.2    mmu-miR-708-3p  down -11
   mmu-miR-323-5p  up   +15.2    mmu-miR-34b-5p  down -15.2
   mmu-miR-323-3p  up   +19.6    mmu-miR-184-3p  down -17.6
   mmu-miR-1193-3p up   +20.8    mmu-miR-362-3p  down -16.3
   mmu-miR-667-3p  up   +15.4    mmu-miR-384-3p  down -11.3
   mmu-miR-377-5p  up   +14.9    mmu-miR-325-3p  down -10.6
   mmu-miR-153-3p  down -15.6
   [74]Open in a new tab

Fig. 2.

   [75]Fig 2
   [76]Open in a new tab

   Integrative analysis of DE miRNAs and DEGs expression profiles in mouse
   brain. (A) Heatmap displaying the relative expression values of
   differentially expressed miRNAs between WT and calpain-1 KO. Red
   indicates up-regulation and blue indicates down-regulation. (B) Venn
   diagram of the mRNA potentially targeted by the up-regulated miRNAs in
   calpain-1 KO compared to the RNA-seq DEGs. (C) Venn diagram of the mRNA
   potentially targeted by the down-regulated miRNAs in calpain-1 KO
   compared to the RNA-seq DEGs.

3.2. Correlation between miRNA and mRNA expression

   As miRNAs are post-transcriptional regulators, we further compared the
   changes in miRNA expression with the changes in mRNA expression between
   WT and calpain-1 KO mice we recently reported [77][13], in order to
   determine whether changes in mRNAs in calpain-1 KO mice could be
   related to the changes in miRNAs. We utilized the predictive algorithms
   TargetScan, miRWalk and miRmap to identify putative mRNA targets of the
   DE miRNAs. The 11 miRNAs exhibiting increased expression in calpain-1
   KO brains target 47 out of the 270 genes, which were down-regulated and
   5 out of the 84 genes which were up-regulated in calpain-1 KO
   ([78]Fig. 2B). Likewise, the 34 miRNAs exhibiting decreased expression
   in calpain-1 KO samples target 8 out of the 84 genes, which were
   up-regulated and 43 out of 270 genes which were down regulated in
   calpain-1 KO ([79]Fig. 2C). Thus, the 45 DE miRNAs target 103 out of
   the 354 (38%) DE mRNAs in brains of calpain-1 KO mice. On the other
   hand, the predicted target genes of ~50% DE miRNAs could not be related
   to any differentially expressed genes (DEGs) in brains of calpain-1 KO
   mice.

3.3. Relationship between DE miRNAs and human neurodegenerative diseases

   Previous reports have suggested that dysregulation of miRNA expression
   is associated with various human diseases. As we have demonstrated that
   calpain-1 activation is linked to neuroprotection, we further
   investigated whether the DE miRNAs in brains of calpain-1 KO mice could
   be related to those miRNAs implicated in neurodegenerative diseases. We
   identified 7 miRNAs, which are down regulated in calpain-1 KO brains,
   and also dysregulated in AD, PD, ALS, HD and prion disease (Fig. S2A).
   Thus, miR-29a-3p, miR-10b-5p, miR-144-3p, miR-146a-5p, miR-451a,
   miR-29c-3p and miR-132-5p are dysregulated in these diseases and
   down-regulated in calpain-1 KO, as shown in a Venn diagram (Fig. S2A).
   Many of these miRNAs appear to be regulating transcription factors
   (TFs), as predicted by TransmiR v2.0 database, and these interactions
   are summarized in the gene network depicted in Fig. S2B, which
   highlights several TFs, including Egr1, Nr4a1, Fosl2, Xbp1, Atf4,
   Bhlhe40 and Fos. By comparing the expression pattern of those miRNAs
   with their interacting TFs, we found that both miRNAs and their
   interacting TFs were down-regulated in brains of calpain-1 KO mice. The
   algorithms we used are only predicting the potential targets of the
   miRNAs, and most TF-miRNA interactions were classified into level 1
   (predicted, but with low confidence). TFs and miRNAs can form feed-back
   or feed-forward loops, making the analysis of their relationships quite
   complicated. Because of the limitation of computational predictions,
   further validation is needed to better understand the role of calpain-1
   in changes in gene expression in various neurodegenerative conditions.

3.4. Decreased levels of Dicer and Dicer-associated canonical miRNAs in
calpain-1 KO brain

   We next sought to determine potential mechanisms by which
   calpain-1-related signaling could contribute to changes in miRNA and in
   gene expression. As mentioned above, it has been shown that calpain, by
   cleaving Dicer, generates a breakdown product with RNAse III activity,
   resulting in the processing of pre-miRNAs into mature miRNAs [80][14].
   To determine whether calpain-1 or calpain-2 is responsible for Dicer
   cleavage, we assessed Dicer levels in brains of WT and calpain-1 KO
   mice by Western blots ([81]Fig. 3A). In brain homogenates from WT mice,
   two major bands were labeled at 250 kDa and 150 kDa. Interestingly, in
   calpain-1 KO homogenates, there was a decrease in the levels of the 250
   kDa band and an almost complete disappearance of the 150 kDa band
   ([82]Fig. 3A). The ratio of the 150 kDa over the 250 kDa band was
   significantly decreased in calpain-1 KO mice, suggesting that levels of
   active Dicer were decreased in calpain-1 KO mice. These results could
   therefore account for the RT-qPCR data demonstrating a significant
   decrease in various miRNA expression levels (miR-146a-5p and
   miR-29a-3p) in calpain-1 KO mouse brain, as compared with WT. These
   results are also consistent with the initial miRNA sequencing analysis
   ([83]Fig. 3B, [84]Table 1).

Fig. 3.

   [85]Fig 3
   [86]Open in a new tab

   Calpain-1 dependent cleavage of Dicer and levels of two miRNAs in brain
   of WT and calpain-1 KO mice. (A) Western blot of Dicer in brain
   homogenates from WT and calpain-1 KO mice. Levels of full length Dicer
   and the band around 150 kDa were decreased in calpain-1 KO mic. The
   ratio of 150 kDa/250 kDa was significantly decreased in calpain-1 KO
   mice. ** p < 0.01; Student's t test; n = 3. (B) RT-qPCR analysis of
   miR-29a-3p and miR-146a-5p in brain from WT and calpain-1 KO. Data were
   normalized to snoRNA202. The y-axis represents the relative miRNA
   expression level. * p < 0.05; n = 6. (C) Venn diagram of down-regulated
   miRNAs in brain of calpain-1 KO mice and in Dicer knockout human cells.

   To further confirm the role of Dicer activity in the regulation of
   miRNA expression, we compared the expression profile of the DE miRNAs
   in calpain-1 KO mice with that reported in Dicer knockout models.
   Previous studies found that canonical miRNA expression was markedly
   reduced in Dicer-deleted human cells, suggesting an essential
   contribution of Dicer in the canonical miRNA pathway. Our analysis
   identified 13 miRNAs (Average CPM > 10) that were down-regulated in
   Dicer-deleted cells and present in the DE miRNAs found in brains of
   calpain-1 KO mice ([87]Fig. 3C, Table S2). Among them, 12 out of these
   13 miRNAs were down-regulated in Dicer-deleted human HCT116 cells
   [88][31] and brains of calpain-1 KO mice.

   These 12 down-regulated miRNAs are predicted to target 18 DEGs in
   calpain-1 KO brains. Based on analysis of pathway enrichment of these
   18 target mRNAs, these genes are related to dopaminergic synapses,
   Alzheimer's disease and endocytosis pathways (Table S3). We also
   compared the miRNA expression profile in brains of calpain-1 KO mice
   with those reported in mouse embryonic stem (ES) cells with Dicer
   deletion and skeletal muscles from tamoxifen-inducible Dicer1 knockout
   mice. However, there were no similarities between these 3 different
   profiles.

3.5. Restoration of active Dicer and miRNA expression by calpain-1 treatment
of brain homogenates from calpain-1 KO mice

   To further investigate whether the 150 kDa cleavage product of Dicer is
   generated by calpain-1, we incubated brain homogenates from calpain-1
   KO mice with calpain-1 and calcium (since calpain-1 requires calcium
   for activation) at 37°C for 15 min and determined changes in Dicer by
   Western blot. This treatment resulted in a significant increase in the
   levels of the active 150 kDa Dicer band ([89]Fig. 4A).

Fig. 4.

   [90]Fig 4
   [91]Open in a new tab

   Restoration of active Dicer and miRNA levels by calpain-1 treatment of
   brain homogenates from calpain-1 KO mice. (A) Treatment of homogenates
   from calpain-1 KO mouse brain with purified calpain-1 in the presence
   of 5 mM Ca^2+ for 15 min increased the level of active Dicer. Left:
   Western blot; right: quantification of the ratio of 150 kDa/250 kDa.
   Results are means ± SD of 5 experiments. * p < 0.05, Student's t test.
   (B) Levels of miR-29a-3p and miR-146a-5p in brain homogenates following
   treatment with calpain-1. Calpain-1 KO mouse brain homogenates were
   treated with calpain-1 in the absence or presence of 5 mM Ca^2+ for 15
   min at 37°C, then 5 mM EDTA was added to stop calpain activity and the
   incubation was continued for 1 h at 37°C. Reactions were stopped by
   extracting RNA with QIAzol Lysis Reagent. miRNA expression was
   determined by RT-qPCR and the ratio of miRNA levels between
   calpain-treated samples and control was calculated. Relative miRNA
   expression levels were normalized to snoRNA202. ** p < 0.01; n = 4. (C)
   Correlation between miRNA levels for miRNA-29a-3p and miR-146a-5p and
   the ratio of 150 kDa/250 kDa in the various samples. Significant
   correlation was observed for both miRNA (r^2 = 0.66, p < 0.01 for
   miRNA-29a-3p and r^2 = 0.74, p < 0.01 for miR-146a-5p; n = 8).

   We then examined whether calpain-1-induced Dicer cleavage could also
   positively affect miRNA biosynthesis in brain homogenates.
   Interestingly, treatment of brain homogenates from calpain-1 KO mice
   with calpain-1 and calcium resulted in increased levels of miR-146a-5p
   and miR-29a-3p to levels similar to those found in WT mice
   ([92]Fig. 4B). To better understand the relationship between Dicer
   activity and miRNA expression, levels of miRNAs in homogenates of WT,
   calpain-1 KO and calpain-1 KO mice treated with calpain-1 and calcium
   were plotted against the ratio of the 150 kDa over the 250 kDa bands in
   the same animals ([93]Fig. 4C). Relative levels of miR-146a-5p and
   miR-29a-3p were positively correlated with relative levels of active
   Dicer (r^2 = 0.74, p < 0.01 and r^2 = 0.66, p < 0.01, respectively).
   These results are consistent with a model in which calpain-1 activation
   is required for Dicer activation and the formation of mature miRNAs,
   thus accounting for the decrease in several miRNAs in brains of
   calpain-1 KO mice.

4. Discussion

   Our results indicate that calpain-1 deletion leads to miRNA
   dysregulation in mouse brain. miRNAs are abundant in the central
   nervous system where they play important roles in the regulation of
   gene expression and brain function [94][32]. In addition, miRNA
   dysregulation has been implicated in many diseases, and in particular,
   in many neurological diseases [[95]33,[96]34]. Our results indicate
   that the target genes of the 10 most abundant miRNAs in mouse brain are
   enriched in the mTOR and Rap1 signaling pathways, which are involved in
   many brain functions, such as learning and memory [97][35] and in
   various neurological disorders [98][36]. By comparing miRNAs in brains
   of WT and calpain-1 KO mice, we identified 45 differentially expressed
   miRNAs. Of these, 7 have been previously linked with various
   neurogenerative diseases. Decreased levels for 2 of these 7 DE miRNAs,
   miR-146a-5p and miR-29a-3p, were further validated with RT-qPCR, and
   the results were in good agreement with the results of miRNA-seq.
   Recently, impaired expression of miR-146a in adult mouse brain has been
   shown to be associated with learning and memory deficits [99][37], and
   similar learning and memory impairment have been observed in calpain-1
   KO mice [100][38]. Thus, our findings indicate that calpain-1 is
   intimately linked with miRNA regulation and that calpain-1
   down-regulation is associated with miRNA dysregulation and related
   neurological diseases.

   Our previous study has shown that calpain-1 deletion results in changes
   in the expression of 354 genes in the brain [101][13]. By analyzing the
   target genes of the DE miRNAs in calpain-1 KO mice, we noticed that
   single DE miRNAs could regulate multiple DEGs and that certain DEGs
   were regulated by more than one miRNA. The DE miRNAs identified in this
   study were predicted to regulate the expression of 38% of the DE mRNAs
   previously identified [102][13]. This complex regulation of gene
   expression suggests that miRNAs are responsible for fine-tuning the
   expression of a variety of genes, and that calpain-1 plays a critical
   role in this regulation. In addition, some of these DE miRNAs-DEGs
   interactions exhibit opposite expression profiles. For instance, Nr4a1
   and Arc, are downregulated in brains of calpain-1 KO mice and could be
   targeted by miR-323-5p and miR-377-5p, respectively, which were
   up-regulated in calpain-1 KO mice. These two up-regulated miRNAs could
   therefore be involved in learning impairment and in AD via the
   downregulation of Nr4a1 and Arc. On the other hand, we also found some
   DE miRNAs with positive regulation of gene expression, as several
   miRNAs, whose expression was decreased in calpain-1 KO mice were
   predicted to target genes that are down-regulated in calpain-1 KO mice.
   Although miRNAs are generally assumed to inhibit gene expression,
   several mechanisms could explain why the miRNA/mRNA interactions could
   be positive as well as negative. First, the interactions between miRNAs
   and their target genes depend on many factors. In most cases, miRNAs
   target the 3’ untranslated region (3’ UTR) of mRNAs to stimulate their
   degradation and translational repression [103][39]. Increasing evidence
   indicates that miRNAs can also interact with the 5’ UTR, enhancer and
   promoter regions, resulting in gene or translation activation
   [104][40], [105][41], [106][42], [107][43], [108][44]. Second, the
   miRNA targeting mainly relies on the seed region base pairing, with a
   perfect complementarity of miRNA seed sequence to the 3’ UTR of mRNA
   leading to transcript degradation [109][45]; however, a single base
   mismatch with the target could change the interaction, resulting in
   protection of the mRNA from degradation [[110]46,[111]47]. Based on the
   GWAS catalog, tens of single-nucleotide polymorphisms (SNPs) are
   associated with AD, PD and ALS [112][48] and the potential impact of
   poly-miRTSs (polymorphisms/mutations in miRNA target sites) in human
   diseases has been discussed [[113]45,[114]49]. If SNPs occur in the
   miRNA/mRNA binding site in calpain-1 KO mice, the introduction of SNPs
   into the miRNA seed region may change the binding efficiency, resulting
   in target loss or gain, therefore affecting miRNA/mRNA interactions and
   miRNA function [115][47].

   Computational algorithms predict up to thousands of target genes for
   individual miRNAs. In our study, we used three available prediction
   algorithms to predict miRNA/mRNA target interactions, and it is
   possible that we identified false positive (or false negative)
   interactions. Thus, the predicted miRNA target genes we identified need
   to be validated by further experiments.

   In our search for DE miRNAs in brain of calpain-1 KO mice, we found
   that a subset of these DE miRNAs are potentially regulated by Dicer,
   since their levels are significantly decreased in Dicer-deficient human
   cells [116][31]. Dicer is an essential protein required for miRNAs
   biogenesis, and it has previously been reported that Dicer can be
   cleaved by calpain, resulting in enhanced RNAse III activity of Dicer
   [117][14]. Our results clearly show that calpain-1 KO mice exhibit
   decreased levels of Dicer breakdown product, and decreased RNAse III
   activity, which could inhibit the formation of miRNAs and lead to
   reduced miRNA levels in calpain-1 KO mice. Intriguingly, this is
   different for what has been reported in diabetic platelets. Elgheznawy
   et al. reported that Dicer is predominantly a calpain-2 substrate and
   that a calpain inhibitor prevented loss of Dicer as well as the
   decrease in miRNA levels observed in platelets from subjects with
   diabetes mellitus [118][50]. Thus, these results suggest that the
   regulation of Dicer levels and activity in different tissues could be
   mediated by different calpain isoforms.

   Our results further indicate that brain Dicer is continuously cleaved
   by calpain-1 under normal conditions, and that miRNA formation is
   therefore under calpain-1 regulation. Down-regulation or mutations in
   calpain-1 results in decreased Dicer function and in significant
   alterations in the levels of many miRNAs and consequently, as we
   previously reported, in the expression of many genes [119][13]. Our
   results further expand our previous findings and indicate that
   down-regulation of calpain-1 could be involved in a number of
   neurodegenerative diseases, including AD, PD, and ALS. In this optic,
   it would be interesting to reevaluate the various human families that
   have been identified with calpain-1 mutations [[120]20,[121]51] and the
   potential links of these mutations to these neurodegenerative diseases.

5. Conclusions

   Our results clearly indicate that calpain-1 chronically regulates Dicer
   RNAse activity through partial truncation in mouse brain. Thus,
   calpain-1 contributes to the regulation of numerous miRNAs and the
   expression of a large number of genes in the brain. Deletion of
   calpain-1 in the brain results in changes in a number of miRNAs that
   have been shown to be involved in neurodegenerative diseases. These
   findings further demonstrate the role of calpain-1 in neurodegeneration
   and point to new research directions to explore in human patients with
   calpain-1 null mutations.

Funding

   This work was supported by the National Institutes of Health Grant
   R01NS104078 to MB. XB is supported in part by funds from the Daljit and
   Elaine Sarkaria Chair.

Data availability statement

   The raw datasets are publicly available on the NCBI database under
   accession number PRJNA670454.

CRediT authorship contribution statement

   Wenyue Su: Investigation, Data curation, Writing – original draft,
   Writing – review & editing. Xiaoning Bi: Conceptualization, Funding
   acquisition, Writing – review & editing. Yubin Wang: Investigation,
   Data curation. Michel Baudry: Conceptualization, Funding acquisition,
   Project administration, Writing – review & editing.

Declaration of Competing Interest

   The authors declare that they do not have any conflict of interest.

Acknowledgment