miR-322/-503 rescues myoblast defects in myotonic dystrophy type 1 cell model by targeting CUG repeats

Myotonic dystrophy type 1 (DM1) is the most common type of adult muscular dystrophy caused by the expanded triple-nucleotides (CUG) repeats. Myoblast in DM1 displayed many defects, including defective myoblast differentiation, ribonuclear foci, and aberrant alternative splicing. Despite many were revealed to function in DM1, microRNAs that regulated DM1 via directly targeting the expanded CUG repeats were rarely reported. Here we discovered that miR-322/-503 rescued myoblast defects in DM1 cell model by targeting the expanded CUG repeats. First, we studied the function of miR-322/-503 in normal C2C12 myoblast cells. Downregulation of miR-322/-503 significantly hindered the myoblast differentiation, while miR-322/-503 overexpression promoted the process. Next, we examined the role of miR-322/-503 in the DM1 C2C12 cell model. miR-322/-503 was downregulated in the differentiation of DM1 C2C12 cells. When we introduced ectopic miR-322/-503 expression into DM1 C2C12 cells, myoblast defects were almost fully rescued, marked by significant improvements of myoblast differentiation and repressions of ribonuclear foci formation and aberrant alternative splicing. Then we investigated the downstream mechanism of miR-322/-503 in DM1. Agreeing with our previous work, Celf1 was proven to be miR-322/-503′s target. Celf1 knockdown partially reproduced miR-322/-503′s function in rescuing DM1 C2C12 differentiation but was unable to repress ribonuclear foci, suggesting other targets of miR-322/-503 existed in the DM1 C2C12 cells. As the seed regions of miR-322 and miR-503 were complementary to the CUG repeats, we hypothesized that the CUG repeats were the target of miR-322/-503. Through expression tests, reporter assays, and colocalization staining, miR-322/-503 was proved to directly and specifically target the expanded CUG repeats in the DM1 cell model rather than the shorter ones in normal cells. Those results implied a potential therapeutic function of miR-322/-503 on DM1, which needed further investigations in the future.


Introduction
Myotonic dystrophy type 1 (DM1) is a dominant autosomal inherited neuromuscular disease. DM1 is caused by the expanded triple-nucleotides (CTG) repeats in the 3′UTR of the DMPK gene. The CTG repeats are transcribed into the expanded CUG repeats, called the toxic RNA 1 . The toxic RNA forms a hairpin-like secondary structure, resulting in MBNL1 sequestration and Celf1 upregulation. MBNL1 sequestration leads to ribonuclear foci formation and lowered functional MBNL1 levels 2 . Unlike MBNL1, the Celf1 level was enhanced by the PKC mediated hyperphosphorylation 3 and the decrease of miR-23a/b 4 . The aberrant levels of MBNL1 and Celf1 in DM1 caused aberrant splicing patterns of many genes, such as CLCN1, IR, PKM, and TNNT2, resulting in disease phenotypes [5][6][7][8][9][10] . Loss of MBNL1 or gain of Celf1 in mouse partially recaptured aberrant alternative splicing, muscle wasting, and defective heart function in DM1 [11][12][13][14] . Celf1 was reported to promote myogenic factor (Mef2A and p21) expressions 15,16 but significantly hindered myogenesis process 17 .
Currently, many targeted therapy strategies against DM1 were proposed. Ectopic MBNL1 expression by AAV infection or repression of Celf1 hyperphosphorylation by PKC inhibition were promising ways 18,19 . Another more radical strategy was directly targeting the expanded CUG repeats, which included degrading the CUG repeats via RNA interference and small chemical molecules 20,21 , or blocking the binding of RNA binding factors to the CUG repeats using small molecules and peptides 22 . Moreover, removal of the CTG repeats in the genome by CRISPR/ Cas9 reverted ribonuclear foci formation and aberrant splicing pattern in DM1 23,24 . Despite those potential therapy strategies, microRNAs (miRNAs) involved therapy strategies were rarely reported 25-27 . miRNAs are a group of 18~22 nucleotides noncoding RNAs, which are incorporated with RNA induced silencing complex to downregulate target mRNAs. The recognition of miRNAs to their target mRNAs is mainly determined by the fully complementary binding of miR-NAs' seed regions to the targets. Previously many miR-NAs were discovered to participate in DM1 pathology and therapy, such as miR-1 28,29 , miR-206 27,30 , miR-148a 30 , and miR-15b/16 30 . Among those, miR-206 and miR-148a were reported to directly target non-CUG repeat region of DMPK 3′UTR, while miR-15b/16 were shown to target the CUG repeat region based on molecular and biochemistry data, leaving biological functions in DM1 undetermined 30 . Together, previously reported miRNAs regulated DM1 mainly through manipulating MBNL1 and Celf1. The miRNAs that rescue DM1 defects by directly targeting the expanded CUG repeats were rarely reported to our best knowledge. miR-322/-503, a miRNA cluster on X chromosomes of mouse and human, is consisted of miR-322 (miR-424 in human) and miR-503. The expressions of both miRNAs are under the control of the same cis-elements. miR-322/-503 was reported to function mainly in cancer [31][32][33] , angiogenesis 34,35 , cardiovascular diseases [36][37][38] , and development fields [39][40][41] . Our previous study demonstrated that miR-322/-503 was enriched in early cardiac progenitors and promoted cardiac differentiation by repressing Celf1 39 . As to myoblast differentiation, miR-322/-503's role was under debate: one reference claimed that miR-322/-503 promoted myoblast differentiation by inhibiting cell cycle via targeting Cdc25A 40 , while another one argued that miR-322 inhibited myoblast differentiation by targeting SETD3 41 . Moreover, although miR-322/-503 was proven to target Celf1 during cardiac differentiation, miR-322/-503's function in DM1 was still illusive.
In this study, we discovered that miR-322/-503 could rescue myoblast defects by directly targeting the expanded CUG repeats in the DM1 cell model. Through gain-and loss-of-function analysis, we found that miR-322/-503 was essential in normal myoblast differentiation. Ectopic miR-322/-503 expression rescued defective myoblast differentiation, ribonuclear foci formation, and aberrant alternative splicing by directly targeting not only Celf1 but the expanded CUG repeats in the DM1 C2C12 cell model.

Materials and methods
Cell culture and flow cytometry C2C12 and HEK293T cells were purchased from Stem Cell Bank, Chinese Academy of Sciences, which were authenticated by STR profiling and free of mycoplasma contamination.
When doing flow cytometry to detect GFP expression, cells were trypsin digested into single cells and examined by FACSCanto II (BD Biosciences, CA, USA). All data were analyzed with FlowJo v10 software.
Plasmid transfections were performed using lipofectamine 2000 (Invitrogen, CA, USA). Before plasmid transfections, both C2C12 and HKE293T cells were seeded one day ahead to make their confluence to be 80-90% for HKE293T cells and 40-50% for C2C12 cells at transfection. The ratio of plasmids to lipofectamine 2000 was 1:2 (μg:μL). The cell culture medium was changed 24 h after transfection. G418 or puromycin selections were performed 48 h after transfection if stable cell lines were needed.

Luciferase assays
Luciferase assay vectors to test if miR-322/-503 targeted Celf1's 3′UTR were constructed as previously reported 39 . Celf1-3′UTR was constructed by ligating the nucleotides 1661~2261 region of Celf1' 3′UTR, which contained the predicted miR-322/-503 mutual binding sites, into pmir-GLO vector (Promega, WI, USA) using In-Fusion HD Cloning kit (Clontech, CA, USA). Celf1-3′UTR-mut was constructed by replacing the predicted miR-322/-503 seed regions' binding sequence-"GACTGCT" with "CTGACGA". The luciferase assays were performed utilizing the Dual-Luciferase Reporter Assay System according to the manufacturer′s protocol. (Promega) Total RNA extraction and real-time quantitative PCR (RT-qPCR) Total RNA samples were extracted using Total RNA Isolation Reagent (Biosharp, Hefei, China). For proteincoding gene expression quantification, reverse transcriptions were performed using FastKing RT Kit (Tiangen, Beijing, China) and Quantitative PCRs were performed using Powerup SYBR Master Mix (Applied Biosystems, CA, USA). GAPDH served as a normalized control. As to microRNA, cDNA samples were produced sequentially by adding poly(A) tail with E.coli Poly(A) Polymerase (New England Biolabs, MA, USA) and doing reverse transcription with SuperScript™ III Reverse Transcriptase (Invitrogen) and a universal reverse transcription primer (5′-CAGGTCCAGTTTTTTTTTTTTTTTVN-3′; "V" stands for A, C, or G, and "N" stands for A, T, C, or G.) as described previously [43][44][45] . Quantitative PCRs for micro-RNA were performed utilizing Powerup SYBR Master Mix (Applied Biosystems). microRNAs′ expression was relative to U6. All quantitative PCR primer sequences are provided in the supplement (Table S1).

Western Blot
Cells were lysed in the Cell Lysis Buffer (Byotime) with protease inhibitors (Roche, Basel, Switzerland). Protein concentrations were measured using the BCA protein assay kit (Biosharp) and adjusted to the same in each experiment set. Proteins were subjected to the electrophoresis on SDS-PAGE gels, which were consisted of 5% stacking gel and 10% separation gel (15% separation gel for LC3B western blots). The proteins then were transferred onto PVDF membranes. The membranes were blocked and incubated with primary antibodies overnight. On the next day, the membranes were incubated with secondary antibodies and reacted with chemiluminescent substrates (Biosharp) to produce signals. The signals were captured by scanning with Tanon 5200 Imaging system (Tanon, Shanghai

Immunostaining
Cells on slides or tissue culture plates were fixed with 4% paraformaldehyde (PFA) for MF-20 immunostaining, whereas cells were fixed with chilled methanol: acetone (1:1) mixed solution in 4°when doing MBNL1 immunostaining to detect ribonuclear foci. Following fixation, cells were blocked with the blocking solution (10% normal goat serum, 0.1% Triton X-100 in phosphate-buffered saline (PBS)). The cells were then incubated overnight in primary antibodies. On the next day, the cells were incubated in the fluorescence conjugated-secondary antibodies for 90 min at room temperature and stained with DAPI for 5 min. MF-20 immunostaining images were taken with an Olympus fluorescence microscope. No. A32727). Immunostaining images were analyzed by ImageJ2X software. The fusion index was the ratio of nuclei number in the cells with at least two nuclei versus total nuclei number. Myotube area was calculated as the ratio of the MF20 fluorescence positive area versus the whole image area in the immunostaining images. Fluorescence intensity was calculated as the ratio of total fluorescence intensity within cells vs. the total cell area.
RNA fluorescence in situ hybridization (RNA FISH) RNA FISH was performed according to the protocol described previously 30 . Cells were fixed with 4% PFA at 4°f or 20 min. Following fixation, the cells were washed three times with PBS and permeabilized with 0.5% Triton X-100 in PBS supplemented with 2 mM ribonucleoside vanadyl complex (RVC) for 7 min. Next, the cells were incubated in 30% formamide and 2× SSC for 10 min. As to hybridization, the cells were incubated in the hybridization buffer (30% formamide, 2× SSC, 0.02% bovine serum albumin, 66 µg/ml yeast tRNA, 10% dextran sulfate, 2 mM RVC, and 2 ng/µl probes) for 24 h. Following hybridization, the cells were washed with 30% formamide and 2× SSC at 45°for 30 min and then 1× SSC at 37°for another 30 min. The cells were mounted in Antifade Mounting Medium with DAPI (Beyotime) and subjected to observation and image capture using a Zeiss ApoTome.2 fluorescence microscope. Probes used in the study were as following: CAG probe for the expanded CUG repeats detection, 5′-CAGCAGCAGCAGCAGCAGCAG-3′ with 5′-FAM label and 2′-O-methyl modification at the first two nucleotides; miR-322 probe for miR-322 detection, 5′-TCCAAAACATGAATTGCTGCT-3′ with 5′-Cy3 label; miR-503 probe for miR-503 detection, 5′-AGTACTGTTCCCGCTGCTA-3′ with 5′-Cy3 label. RNA FISH images were analyzed by the Colocalization plugin in ImageJ2X software. The colocalization ratio of miR-322/-503 and the expanded CUG repeats was calculated as the ratio of miRNAs and the CUG repeats dualpositive area versus the area that was at least one probe positively stained.

β-galactosidase staining
Female heterozygous miR-322/-503 LacZ knock-in mice were generated by initial mating male Mirc24 tm1Mtm /Mmjax (MMRRC Stock No: 36306-JAX) mice with female Tg(Sox2-Cre) mice and further mating with other wild-type mice. E10.5 embryos, which were LacZ genotyping positive, were collected and fixed for 30 min in 4% PFA. The embryos were then washed three times with wash buffer, which was PBS supplemented with 0.02% NP-40 and 0.01% sodium deoxycholate. Afterwards, the embryos were stained with staining solution, which was PBS supplemented with 5 mM K 3 Fe (CN) 6 , 5 mM K 4 Fe(CN) 6 , 0.02% NP-40, 0.01% sodium deoxycholate, 2 mM MgCl 2 , 5 mM EGTA, and 1 mg/ml X-gal. The embryos were transferred to wash buffer after specific staining appeared and ready for image capture. All mice related experimental procedures were approved by the Institutional Animal Care and Use Committee (IACUC).

RT-PCR for alternative splicing test
Total RNA samples were extracted using Total RNA Isolation Reagent (Biosharp). The RNA samples were reverse transcribed by RevertAid Reverse Transcriptase (Thermo Fisher, MA, USA) to produce cDNA. cDNA samples were then amplified by Phusion Hot Start II High-Fidelity PCR Master Mix (Thermo Fisher) with designed primers, the sequences of which were provided in the supplement (Table S2). The PCR products were subjected to the electrophoresis on 3% agarose gel and scanned with the Tanon gel image system 1600. The densities of bands on gel images were measured by Tanon gel image system integrated software.

Statistical analysis
Three biological replicates and three technical replicates were performed for all assays except where otherwise stated. Significance was determined by the two-sided t test and p < 0.05 was considered to be statistically significant. All data were presented as mean ± SD.

miR-322/-503 was required in myoblast differentiation
miR-322/-503 was revealed to promote cardiac differentiation by targeting Celf1 in early cardiac progenitor cells 39 , but its function in myoblast differentiation was still under debate 40,41 . We performed β-galactosidase staining on miR-322/-503 lacZ knock-in mouse E10.5 embryos. miR-322/-503 was specifically expressed in the heart region and somites, suggesting that miR-322/-503 might be involved in not only the heart but the skeletal muscle developments (Fig. 1A). Then we performed myoblast differentiation of C2C12 cells and measured the expression patterns of miR-322 and miR-503. Both miR-322 and miR-503 displayed expression peaks at day 2, suggesting their elevation at day 2 might be needed to ensure the normal differentiation process (Fig. 1B).
In conclusion, we here discovered that miR-322/-503 could rescue myoblast defects in DM1 by directly targeting the expanded CUG repeats. By gain-and loss-offunction analysis, we found that miR-322/-503 was required by normal myoblast differentiation. In the DM1 cell model, ectopic miR-322/-503 expression rescued myoblast differentiation defects, ribonuclear foci, and aberrant alternative splicing. Celf1 and the expanded CUG repeats were proven to be miR-322/-503's targets. Moreover, miR-322/-503 specifically targeted the expanded CUG repeats in DM1 rather than in normal cells, resulting in ribonuclear foci dissolution. The specificity implied a potential therapeutic function of miR-322/-503 against DM1, which needed further examination (Fig. 8).

Discussion
DM1 is one of the most prevalent adult inherited diseases. The fundamental cause of DM1 is the expanded CUG repeats, which results in spliceopathy and muscular dystrophy. Many miRNAs have been reported to function in DM1. In this study, we established the required role of miR-322/-503 in normal myoblast differentiation and its therapeutic potential of rescuing muscular defects in DM1 through directly targeting the expanded CUG repeats.
Firstly, we discovered that miR-322/-503 is required by the normal myoblast differentiation process. In the lossof-function test, although the losses of miR-322 and miR-503 both repressed myoblast differentiation, the loss of miR-503 caused more severe differentiation repression, suggesting miR-503 might be relatively more important in regulating myoblast differentiation. In the gain-offunction test, we identified a significant improvement of myoblast differentiation with miR-322/-503 overexpression. In previous studies, miR-322/-503′s role in myoblast differentiation was in arguing. One study stated that miR-322/-503 promoted myoblast differentiation by targeting Cdc25A 40 , which agreed with our findings. However, the other recent report argued that miR-322 repressed myoblast differentiation by targeting SETD3 41 . In this paper, the authors used a myoblast DM without insulin supplement, differing from our culture conditions. According to references, miR-322/-503 was reported to interact with insulin involved pathways 38,51 , and directly regulate insulin resistance 52 . Therefore, the difference in DM might bias the role of miR-322/-503 in myoblast differentiation.
To study myoblast differentiation in DM1, we adopted C2C12-CUG5 cells as the normal model and C2C12-CUG200 cells as the DM1 model. C2C12-CUG200 cells displayed defective myoblast differentiation, ribonuclear foci, and aberrant alternative splicing, which reproduced the pathologic phenotypes of DM1 skeletal muscle. In C2C12-CUG200 cell differentiation, we noticed a dramatic decrease of miR-322/-503. Similarly, we found that the levels of miR-424 (miR-322 orthologue) and miR-503 level also tended to be lower (although not significantly) in DM1 patient serums through analyzing a publicly available RT-qPCR data of healthy individuals (n = 26) and DM1 patients (n = 24) from a published reference (data not shown) 53 , suggesting that miR-424/-503 might be downregulated also in DM1 myoblast differentiation in humans. The mechanism of how miR-322/-503 was downregulated in DM1 was still unknown. The possible reason was the disturbed miRNA biogenesis in DM1. It was reported that miR-1 was downregulated in the heart (see figure on previous page) Fig. 5 miR-322/-503 specifically antagonized the expanded CUG repeats. GFP-CUG5 and GFP-CUG200 were used as reporters. A Schematic diagram showing the predicted bindings of the seed regions of miR-322 and miR-503 to CUG repeats. B GFP mRNA level was significantly repressed by miR-322/-503 in the GFP-CUG200 group in HEK293T cells. mRNA level was quantified by RT-qPCR. All expression levels were normalized to the Blank vector/GFP-CUG5 group. C, D GFP levels were significantly repressed by miR-322/-503 in the GFP-CUG200 group in HKE293T cells. The GFP levels were measured by flow cytometry. Mean GFP fluorescence intensities of four were quantified by the FlowJo v10 software. E, F GFP protein level was significantly repressed by miR-322/-503 in the GFP-CUG200 group in HKE293T cells. However, the GFP protein level was not affected by miR-322/-503 in the GFP-CUG5 group in HKE293T cells. Protein levels were measured by western blots. G GFP mRNA level was significantly repressed by miR-322/-503 in the GFP-CUG200 group in C2C12 cells. However, the GFP mRNA level was not affected by miR-322/-503 in the GFP-CUG5 group in C2C12 cells. mRNA level was quantified by RT-qPCR. All expression levels were normalized to the Blank vector/C2C12-CUG5 group. H miR-322/-503 induction by Dox treatment was verified by RT-qPCR in C2C12-CUG200/pCW57-miR-322/-503 cells. All expression levels were normalized to no Dox group. I, J Autophagy, marked by LC3B-II/LC3B-I ratio, was not affected with Dox-induced miR-322/-503 overexpression in C2C12-CUG200/pCW57-miR-322/-503 cells. LC3B was determined by western blots. K, L miR-322/-503 regulated GFP levels independent of autophagy in C2C12-CUG200/ pCW57-miR-322/-503 cells. All expression levels were normalized to the no treatment group. Control, pLL4.0 vector stable transfection; miR-322/-503, pLL4.0-miR-322/-503 stable transfection; pLL4.0, pLL4.0 vector transient transfection; pLL4.0-miR-322/-503, pLL4.0-miR-322/-503 transient transfection; Rapa rapamycin, Dox Doxycycline. * statistically significant (p < 0.05); ns not statistically significant.
because of the impaired Dicer activity resulting from MBNL1 sequestration in DM1, suggesting that the Dicer involved miRNA processing might be dysregulated in DM1 29 . That might be the cause of miR-322/-503 dysregulation in DM1 as well. This hypothesis needed experiments to verify, which was beyond the scope of this study.
Next, we found that ectopic miR-322/-503 expression successfully rescued the myoblast differentiation defects, ribonuclear foci formation, and aberrant alternative splicing in the DM1 cell model through targeting the expanded CUG repeats and Celf1. Celf1 is an important pathogenic factor in DM1. Transgenic Celf1 overexpression in mice heart and skeletal muscle was able to display similar DM1 phenotypes 13,14 . Since our previous study suggested that miR-322/-503 targeted Celf1 in regulating cardiac differentiation 39 , we asked if miR-322/-503 rescued muscular defects through targeting Celf1. Celf1 was proven to be miR-322/-503's target and its knockdown improved myoblast differentiation defects in DM1 to some extent, agreeing with the previous reference 17,39 . However, the Celf1 knockdown was unable to repress the formation of ribonuclear foci, differing from the outcomes of miR-322/-503 overexpression in DM1, suggesting that other miR-322/-503 involved regulatory mechanisms existed in DM1. Then we revealed that miR-322/-503 directly targeted the expanded CUG repeats. Through expression, reporter, and colocalization assays, we found that miR-322/-503′s targeting the CUG repeats only restricted to the expanded ones in DM1 rather than in normal cells.
Aberrant alternative splicing was caused by the expanded CUG repeats in DM1, which directly mediated the disease phenotypes of DM1. As we proved that miR-322/-503 targeted the expanded CUG repeats, we wondered if the miR-322/-503 corrected aberrant alternative splicing in DM1. Through RT-PCR, we found the aberrant splicing patterns of Anxa7, Atp2a1, Insr, MBNL1, Ldb3, CAPZB, FXR1, and MFN2 in DM1 were rescued with the ectopic miR-322/-503 expression. Among those genes, the alternative splicing of CAPZB, FXR1, and MFN2 were specifically regulated by Celf1. These The exons that caused band size variations in each gel image were specified. GAPDH served as an internal control. B The percentage of exon inclusions of Anxa7, Atp2a1, Insr, MBNL1, Ldb3, CAPZB, FXR1, and MFN2 were plotted according to band densities in (A). The optical densities of agarose gel bands were quantified using ImageJ2X software. The optical densities of both exon inclusion and exclusion bands were normalized to corresponding GAPDH bands. The exon inclusion percentage was calculated as follows: exon inclusion% = normalized optical density of exon inclusion/ (normalized optical density of exon inclusion + normalized optical density of exon exclusion). Control, C2C12-CUG200/control cells; miR-322/-503, C2C12-CUG200/miR-322/-503 cells; * statistically significant (p < 0.05).
suggested that miR-322/-503 rescued aberrant alternative splicing in DM1 via targeting both the expanded CUG repeats and Celf1.
In summary, we here discovered that miR-322/-503, which is required by normal myoblast differentiation, could rescue muscular defects in DM1 mainly by targeting the expanded CUG repeats. miR-322/-503's targeting toward CUG repeats displayed high specificity to the length in DM1 rather than normal cases. With those results, future studies on whether miR-322/-503 could be applied to DM1 therapy would be attractive.