MEG3 Promotes Differentiation of Porcine Satellite Cells by Sponging miR-423-5p to Relieve Inhibiting Effect on SRF

Although thousands of long noncoding RNAs (lncRNAs) have been identified in porcine growth and development, the regulation mechanisms of functional lncRNAs have not been well explored. In this study, using 5′- and 3′-rapid amplification of cDNA ends (RACE) assays, we obtained two different variants of lncRNA maternally expressed gene 3 (MEG3), namely, MEG3 v1 and MEG3 v2, that were both highly expressed in porcine skeletal muscle and in the early stage of the differentiation of porcine satellite cells. Moreover, we identified the core transcript MEG3 v2. Functional analyses showed that MEG3 overexpression could effectively arrest myoblasts in the G1 phase, inhibit DNA replication, and promote myoblast differentiation, whereas MEG3 knockdown resulted in the opposite effects. Interestingly, the expression of serum response factor (SRF), a crucial transcription factor for myogenesis process, remarkably increased and decreased in mRNA and protein levels with the respective overexpression and knockdown of MEG3. Dual luciferase reporter assay showed that MEG3 could attenuate the decrease of luciferase activity of SRF induced by miR-423-5p in a dose-dependent manner. MEG3 overexpression could relieve the inhibitory effect on SRF and myoblast differentiation induced by miR-423-5p. In addition, results of RNA immunoprecipitation analysis suggested that MEG3 could act as a ceRNA for miR-423-5p. Our findings initially established a novel connection among MEG3, miR-423-5p, and SRF in porcine satellite cell differentiation. This novel role of MEG3 may shed new light on understanding of molecular regulation of lncRNA in porcine myogenesis.


Introduction
In mammalian genomes, only 2% of transcripts are translated into proteins. The vast majority of transcripts are noncoding RNAs, including microRNAs (miRNAs), Piwi-interacting RNAs, circular RNAs, and long noncoding RNAs (lncRNAs) [1]. Recently, an increasing number of researchers have focused on lncRNAs, which are a type of RNA with lengths of more than 200 nt and lack protein-coding potential [2]. They are characterized by less abundant, less evolutionarily conserved, and spatio-temporal specific expression profiles [3][4][5]. Thousands of functional lncRNAs have been identified to be involved in multiple biological processes, such as X chromosome inactivation,

RNA Oligonucleotide and Plasmid Construction
The miR-423-5p mimic, mimic negative control (NC) and the antisense oligonucleotide (ASO) oligo against MEG3 and scrambled oligo were purchased from RiboBio (Guangzhou, China). Small interfering RNA (siRNA) of SRF and scrambled oligo were designed and synthesized from GenePharma (Shanghai, China). Oligonucleotide sequences in this study are shown in Table S1.
For the overexpression plasmids, the full lengths of two different transcripts of porcine MEG3 gene were synthesized from Tsingke (Beijing, China) and cloned into the pZW1-son plasmid. The coding DNA sequence (CDS) of SRF was amplified by PCR and cloned into the pcDNA3.1 plasmid. The major primers used in this study are listed in Table S2. For the dual-luciferase reporter vector plasmid construction, about 300 bp wild-type and mutant sequences of MEG3 and SRF, containing miR-423-5p seed sequence target sites, were inserted into pGL3-Basic vector. Mutant plasmid of MEG3 was generated by changing the binding site of miR-423-5p from CTGCCCCT to GACGATAG; that of SRF was changed from CTGCCCCTCA to GACGGAGTAT.
All the recombinant plasmids were confirmed by sequencing (Sangon Biotech, Shanghai, China).

Cell Transfection
All transient transfections in porcine satellite cells or PK15 cells were performed with Lipofectamine 2000 reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions.

Nuclear and Cytoplasmic RNA Fractionation
Cells were prepared at both proliferative and differentiated periods. The procedure for separating the nuclear and cytoplasmic RNA fractionation was performed in accordance with previous published reports [40]. RNAs of cytoplasmic and nuclear fractions were extracted with RNAiso reagent (TaKaRa, Otsu, Japan). The locations of MEG3 (detection with the overlap region primers of MEG3 two transcripts), GAPDH (cytoplasmic marker gene) and Neat1 (a nuclear expression lncRNA) were analyzed by quantitative polymerase chain reaction (qPCR). Primer sequences for qPCR are listed in Table S2.
2.6. RNA Extraction, cDNA Synthesis, and Quantitative Polymerase Chain Reaction(qPCR) Total RNA was extracted from cells using RNAiso reagent (TaKaRa, Otsu, Japan) according to the manufacturer's instructions. The concentration and quality were measured by a spectrophotometer Complementary DNA (cDNA) synthesis for messenger RNA (mRNA) was performed using the PrimeScript RT Reagent Kit with gDNA Eraser (Perfect Real Time) (TaKaRa, Otsu, Japan). For miRNA, stem loop miRNA qRT-PCR primers specific for miR-423-5p and U6 were designed by Vazyme (Nanjing, China) and cDNA was synthesized with miRNA 1st Strand cDNA Synthesis Kit (by stem-loop) (Vazyme, Nanjing, China).
Quantitative polymerase chain reaction (qPCR) for mRNA was carried out on a Bio-Rad CFX96 Real-Time Detection System using TB Green Premix Ex Taq II (Tli RNase H Plus) (TaKaRa, Otsu, Japan). For miRNA, miRNA Universal SYBR qPCR Master Mix (Vazyme, Nanjing, China) was used and analyzed with the 2 −∆∆CT method. All primers for qPCR in this study were designed with Primer 5, and primer sequences are listed in Table S2. The 18S ribosomal RNA (18S rRNA) and U6 were used as internal controls.

5 and 3 RACE and Full-Length LncRNA Cloning
To obtain the transcription information and full-length sequences of MEG3, SMARTer RACE cDNA Amplification Kit (Clontech, Osaka, Japan) was used for 5 and 3 RACE according to the manufacturer's instructions. The gene-specific primers (GSP) for RACE PCR were designed for 5 RACE (GSP1) and 3 RACE (GSP2) amplification. The PCR products were inserted into the pRACE vector and sequenced by Tsingke Biological Technology (Wuhan, China). The gene-specific primers are shown in Table S2.

CCK-8 Assay
Porcine satellite cells were seeded in 96-well plates and transfected with ASO oligo against MEG3 or MEG3 overexpression vectors when cell confluence was 40-50%. Then, cell proliferation was monitored with the CCK-8 Cell Counting Kit (Vazyme, Nanjing, China) following the manufacturer's protocol. And the absorbance at 450 nm was measured in the spectrophotometer after being transfected for 12, 24, 36, and 48 h.

5-Ethynyl-20-deoxyuridine (EdU) Assay
Porcine satellite cells were transferred to culture medium with 50 µM EdU (RiboBio, Guangzhou, China) for 2 h at 37 • C after 36 h transfection. Afterwards, cells were fixed in 4% paraformaldehyde for 15 min at room temperature (RT), and then permeabilized with 0.3% Triton X-100 for 10 min. To block unspecific binding, cells were incubated in the blocking buffer (PBS containing 3% bovine serum albumin, 0.3% Triton X-100) for 1 h at RT. Then cells were incubated with a solution containing 10 mM EdU in dark for 30 min. The nuclei were stained with 10 µg/mL 4, 6-diamidino-2-phenylindole (DAPI, Invitrogen, Carlsbad, CA, USA) solution in dark for 10 min. Leica SP8 confocal microscope was used to capture three randomly selected fields to visualize the number of EdU-stained cells.

Flow Cytometry Analysis
For flow cytometry analysis of cell cycle, the Cell Cycle Detection Kit (Keygen, Nanjing, China) were used in line with the manufacturer's instructions. Briefly, cells were harvested and fixed in 70% ethanol overnight at 4 • C after 36 h transfection. Then cells were rinsed with PBS and centrifuged at 2500 rpm for 5 min. Subsequently, cells were stained with prepared propidium iodide (PI) solution, containing RNase A and PI at a volume ratio of 1:9, and then incubated in dark for 30 min at RT. Flow cytometry analysis was performed on Beckman Coulter FC500 Cytometer (Beckman Coulter, Miami, FL, USA) and data were processed by ModFit software (Verity Software House, Topsham, ME, USA).

Transcriptome Sequencing and Differential Expression Analysis
To further study the involvement of MEG3 in porcine myogenesis, we thoroughly analyzed RNA-seq data from MEG3 knockdown and control groups in porcine satellite cells differentiated for 30 and 40 h. In total, 3 µg of RNA for each sample was used to construct sequencing libraries. The libraries were sequenced on the Illumina HiSeq X-ten platform and 150 bp paired-end reads were generated. Then, we used FastQC software (Nanjing Agricultural University, Nanjing, China) to evaluate the quality of obtained sequence data and used Trimmomatic tool (version 0.3.2, Nanjing Agricultural University, Nanjing, China) to trim. Next, HISAT2 (version 2.0.1, Iowa State University, Ames, IA, USA) was used to obtain the qualified and clean reads mapped to the pig reference genome (Sus scrofa 11.1) and StringTie (version 1.3.4, Johns Hopkins University, Baltimore, MD, USA) was used to assemble the mapped reads with default parameters. HTSeq-count (version 0.9.1, European Molecular Biology Laboratory, Heidelberg, BW, Germany) was used to count reads mapped to the genome and the annotation file. Subsequently, differentially expressed genes were identified utilizing the R packages DESeq2 (Tsinghua University, Beijing, China). A transcript will be considered as differentially expressed between two groups if the absolute value of log2 (fold-change) > 1, p-value < 0.05 and false discovery rate (FDR) < 0.05. In order to query each protein-coding gene and understand their functions, we performed gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis by running queries for each protein-coding gene against the DAVID database. GO terms or KEGG pathways with corrected p-value < 0.05 were considered to be enriched clusters. Because of the limitation of genes annotation in Sus scrofa, all genes were converted into human homologous genes using BIOMART from Ensembl.

RNA Immunoprecipitation Assay MEG3
To identify specific RNA molecules associated with Argonaute2 (Ago2), a key effector of small RNA mediated gene silencing [41], we performed RNA immunoprecipitation (RIP) assay with EZ-Magna RIP Kit (Millipore, Billerica, MA, USA) according to the manufacturer's protocol. Briefly, porcine satellite cells, differentiated for 48 h, were collected and lysed in RIP lysis buffer. Then, cell lysates were incubated with A/G magnetic beads conjugated with anti-Ago2 antibody (Boster, Wuhan, China). Then, the immunoprecipitated RNA was isolated and qPCR was performed to detect the abundance of MEG3, SRF, and miR-423-5p; 18S rRNA and U6 were used as internal controls. Primer sequences for qPCR are listed in Table S2.

Statistical Analysis
Generally, results are presented as the means ± standard error of the mean (SEM). Statistical comparison between two different groups were assessed by two-tailed Student's t-test. The p value < 0.05 was considered to be statistically significant.

Expression of MEG3 LncRNA
To investigate the regulatory mechanism in pigs, we first evaluated the full-length cDNA of MEG3 in porcine satellite cells using 5 and 3 RACE ( Figure 1A). We designed gene-specific primers (GSP) for RACE and identified two polyadenylated MEG3 transcripts: MEG3 variant 1 (MEG3 v1), 1430 nt in length, and MEG3 variant 2 (MEG3 v2), 1380 nt in length. Sequence analysis showed that these two different variants shared the same 508-nt front part ( Figure 1B). Interestingly, MEG3 showed remarkably higher mRNA expression levels in brain and muscle tissues, such as longissimus dorsi and gastrocnemius muscle, than other multiple tissues ( Figure 1C). Notably, the expression of MEG3 v2 showed a substantial dominance in porcine muscle tissues compared with MEG3 v1 ( Figure 1C). We collected porcine satellite cells at the proliferation stage every 6 h and at the differentiation stage every 12 h to distinguish the expression levels of these two MEG3 variants in different periods of myogenesis. Overall, two variants of MEG3 had similar expression trends. Both variants were upregulated in the proliferation and early stage of differentiation and reached their peaks in porcine satellite cells differentiated for 48 h, but they gradually decreased afterward ( Figure 1D), suggesting that MEG3 could be a promyogenic factor during the early differentiation in porcine satellite cells. Obviously, MEG3 v2 was the most abundant in porcine satellite cells with dominant expression level compared with that of MEG3 v1 ( Figure 1D). Therefore, MEG3 v2 was the core transcript in porcine satellite cells. Moreover, nuclear-cytoplasmic RNA fractionation experiments demonstrated that MEG3 was mainly located in the nuclear compartment of proliferating myoblasts (76.86%) and differentiated myotubes (59.99%). Interestingly, its proportion in the cytoplasm had increased in myotubes (in a range from 23.14% to 40.01%) ( Figure 1E). NEAT1 is a known nuclear lncRNA, and GAPDH is a cytoplasmic-enriched gene. The relative RNA levels were normalized to those of the control 18S rRNA. Error bars represent mean ± SEM of three biological replicates. Statistical significance of differences was assessed by Student's t-test. NC, negative control.

MEG3 Inhibits Myoblast Proliferation
MEG3 expression levels were upregulated during the myoblast proliferation phase, indicating that MEG3 could be involved in the regulation of myoblast proliferation. In functional deficit and acquisition experiments, optimized phosphorothioate-modified antisense oligodeoxynucleotide (ASO) against the overlapping region sequence of the two MEG3 variants and pZW1-MEG3 v1 or pZW1-MEG3 v2 plasmid were respectively transfected into porcine satellite cells. In the 5-ethynyl-2′-deoxyuridine (EdU) staining assays, the interference of MEG3 with ASO showed higher mitotic activity with an increase in EdU incorporation ( Figure 2A). On the contrary, the overexpression of MEG3 v1 (p < 0.05) or MEG3 v2 (p < 0.01) showed lower mitotic activity with a decrease in EdU positivity ( Figure 2B). The CCK-8 assay showed that the MEG3 knockdown for 24, NEAT1 is a known nuclear lncRNA, and GAPDH is a cytoplasmic-enriched gene. The relative RNA levels were normalized to those of the control 18S rRNA. Error bars represent mean ± SEM of three biological replicates. Statistical significance of differences was assessed by Student's t-test. NC, negative control.

MEG3 Inhibits Myoblast Proliferation
MEG3 expression levels were upregulated during the myoblast proliferation phase, indicating that MEG3 could be involved in the regulation of myoblast proliferation. In functional deficit and acquisition experiments, optimized phosphorothioate-modified antisense oligodeoxynucleotide (ASO) against the overlapping region sequence of the two MEG3 variants and pZW1-MEG3 v1 or pZW1-MEG3 v2 plasmid were respectively transfected into porcine satellite cells. In the 5-ethynyl-2 -deoxyuridine (EdU) staining assays, the interference of MEG3 with ASO showed higher mitotic activity with an increase in EdU incorporation (Figure 2A). On the contrary, the overexpression of MEG3 v1 (p < 0.05) or MEG3 v2 (p < 0.01) showed lower mitotic activity with a decrease in EdU positivity ( Figure 2B). The CCK-8 assay showed that the MEG3 knockdown for 24, 36, or 48 h could dramatically accelerate cellular proliferation ( Figure 2C). Inversely, the overexpression of MEG3 v1 or MEG3 v2 substantially suppressed the proliferative ability of porcine satellite cells compared with the negative control ( Figure 2D). The propidium iodide flow cytometry assays indicated a considerable reduction of cell quantity in the G0/G1 phase and a remarkable increase of cell quantity in the S phase after MEG3 knockdown ( Figure 2E,F). Conversely, the overexpression of MEG3 v1 or MEG3 v2 showed an opposite effect ( Figure 2G,H). These findings validated that MEG3 could inhibit the proliferation of porcine satellite cells.
Cells 2020, 9, x 8 of 20 36, or 48 h could dramatically accelerate cellular proliferation ( Figure 2C). Inversely, the overexpression of MEG3 v1 or MEG3 v2 substantially suppressed the proliferative ability of porcine satellite cells compared with the negative control ( Figure 2D). The propidium iodide flow cytometry assays indicated a considerable reduction of cell quantity in the G0/G1 phase and a remarkable increase of cell quantity in the S phase after MEG3 knockdown ( Figure 2E,F). Conversely, the overexpression of MEG3 v1 or MEG3 v2 showed an opposite effect ( Figure 2G,H). These findings validated that MEG3 could inhibit the proliferation of porcine satellite cells.

MEG3 Promotes Myoblast Differentiation
The above results demonstrated that MEG3 is extremely important for myoblasts to be able to withdraw from the cell cycle, a crucial step in myoblast differentiation. In addition, the expression profile of MEG3 prompted its association with myoblast differentiation. We used qPCR, Western blot, and immunofluorescence staining to test the changes of three established myogenic marker genes (MyoD, MyoG, and myosin heavy chain (MyHC)). qPCR results showed that MEG3 was successfully knocked down in the ASO-MEG3 group myotube differentiated for 48 h ( Figure 3A). Meanwhile, the mRNA and protein levels of MyoD, MyoG, and MyHC were remarkably downregulated after MEG3 knockdown compared with the control group ( Figure 3A,B). Consistently, immunofluorescence staining of MyoG and MyHC showed that MEG3 knockdown notably reduced the proportion of MyoG + and MyHC + cells ( Figure 3C,D). To further confirm the above observation, we successfully overexpressed MEG3 v1 and MEG3 v2 in porcine satellite cells in the differentiation phase ( Figure 3E). As expected, the mRNA and protein levels of MyoD, MyoG, and MyHC substantially increased after MEG3 v1 and MEG3 v2 overexpression in myotube ( Figure 3F,G). Likewise, MEG3 overexpression showed that more cells proceeded with differentiation and myotube formation than the control group in the immunofluorescence staining of MyoG and MyHC ( Figure 3H,I). Taken together, our results confirmed that MEG3 is required for myoblast differentiation.

Gene Expression Profile of MEG3 Knockdown in Porcine Satellite Cells
To further study the involvement of MEG3 in skeletal muscle development, we thoroughly analyzed RNA-seq data from MEG3 knockdown and control groups in porcine satellite cells differentiated for 30 and 40 h and identified differentially expressed genes between the two groups. In total, we obtained over 40 million raw reads from each library. Then we removed low-quality sequences, and the clean reads mapped more than 95% of the raw data. Next, we aligned all clean reads to the porcine Sscrofa11.1 reference genome and found that more than 70% clean reads could be uniquely mapped to the genome (Table S3). The principal component analysis (PCA) score plots showed that datasets from four groups (30 hNC, 30 hASO and 40 hNC, 40 hASO groups for MEG3 knockdown) were clustered separately ( Figure 4A). Hierarchical clustering was performed for differential expressions of protein-coding genes and obtained a global overview of gene expression profile among 40 hNC group and 40 hNC ASO group for MEG3 knockdown ( Figure 4B). Finally, the differentially expressed genes were used to perform gene ontology (GO) enrichment analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis. In detail, we identified 273 and 207 differentially expressed protein-coding genes from MEG3 knockdown for the 30 and 40 h groups, respectively, compared with the corresponding negative control groups. At 30 h comparison groups, GO enrichment analysis of these differentially expressed genes indicated that they were mainly associated with muscle-related processes, including muscle myosin complex, skeletal muscle contraction, and positive regulation of myoblast differentiation, skeletal muscle cell differentiation, and structural constituent of muscle ( Figure 4C). KEGG pathway enrichment analysis was used to explore the biological pathways for differentially expressed protein-coding genes. The results showed that these genes participated in the PI3K-Akt signaling pathway, oxytocin signaling pathway, focal adhesion, and adrenergic signaling in cardiomyocytes ( Figure 4D). The GO enrichment analysis at 40 h indicated that the differentially expressed genes were also related to muscle-related processes, such as Z disc, actin cytoskeleton, myofibril, muscle contraction, skeletal muscle cell differentiation, and structural constituent of muscle ( Figure 4E). KEGG pathway enrichment analysis showed that these differentially expressed genes were mainly involved in the PI3K-Akt signaling pathway, regulation of actin cytoskeleton, focal adhesion, and MAPK signaling pathway ( Figure 4F). Among these differentially expressed genes, CCND1 [42], PLCB1 [43], MEF2C [44], and FOXO3 [45,46] have a known function in regulating myogenesis. As a consequence, MEG3 could specifically have important effects on skeletal muscle development. To validate the reliability of our sequencing results, we randomly selected several differentially expressed genes in 30 and 40 h groups for qPCR to detect their expression levels. The expression levels of CAMK1, ITGA3, PLK2, CCND1, EIF4E, ITGA7, ITPR3, and PLCB1 were upregulated, whereas those of ACACB, FOXO3, MEF2C, MYL4, MYLK4, and TPM2 were downregulated ( Figure 4G,H).
Consistently, immunofluorescence staining of MyoG and MyHC showed that MEG3 knockdown notably reduced the proportion of MyoG + and MyHC + cells ( Figure 3C,D). To further confirm the above observation, we successfully overexpressed MEG3 v1 and MEG3 v2 in porcine satellite cells in the differentiation phase ( Figure 3E). As expected, the mRNA and protein levels of MyoD, MyoG, and MyHC substantially increased after MEG3 v1 and MEG3 v2 overexpression in myotube ( Figure  3F,G). Likewise, MEG3 overexpression showed that more cells proceeded with differentiation and myotube formation than the control group in the immunofluorescence staining of MyoG and MyHC ( Figure 3H,I). Taken together, our results confirmed that MEG3 is required for myoblast differentiation. 100 µm. Porcine satellite cells were harvested after transfection with MEG3 ASO or overexpression plasmid and differentiation for 48 h. The relative mRNA levels were normalized to those of the control 18S rRNA. The relative protein levels were normalized to those of the control β-tubulin. Error bars represent mean ± SEM of three biological replicates. Statistical significance of differences was assessed by Student's t-test. * p < 0.05, ** p < 0.01. N.S. means that there was no significant difference. NC, negative control. qPCR, quantitative polymerase chain reaction.

MEG3 Acts As a ceRNA for miR-423-5p
Given that lncRNAs can regulate multiple biological functions by sponging regulatory miRNAs, the seed regions of miR-423-5p were predicted to be complementary with MEG3 core transcript MEG3 v2 (abbreviated as MEG3), and another potential target gene (i.e., SRF) of miR-423-5p was obtained using TargetScan (http://www.targetscan.org/vert_71/) and RNAhybrid software (https://bibiserv. cebitec.uni-bielefeld.de/rnahybrid/) ( Figure 5A). SRF is a crucial transcription factor that regulates muscle proliferation and differentiation [47,48], and miR-423-5p is also involved in skeletal muscle development and regeneration [49]. Therefore, we hypothesized that MEG3 and SRF are functional targets of miR-423-5p, that is, MEG3 modulates SRF by competing for miR-423-5p. To determine the binding sites between miR-423-5p and its target gene MEG3 or SRF, we constructed wild-type and mutant versions of MEG3 and SRF-3 UTR using pGL3-Basic vectors. As shown in Figure 5B, the wild-type and mutant reporter vectors contained miR-423-5p binding sites and mutated recognition sequences, respectively. Different vectors were co-transfected into PK15 cells with miR-423-5p mimic. In a dual luciferase reporter assay, we found that the luciferase activity of wild-type MEG3 and SRF remarkably decreased with the overexpression of miR-423-5p compared with the negative control groups (p < 0.01). However, no significant change was observed in the luciferase activity of the mutant group lacking miR-423-5p binding site ( Figure 5C,D). Interestingly, MEG3 could attenuate the decrease of luciferase activity induced by miR-423-5p in a dose-dependent manner ( Figure 5E). Consistently, RNA immunoprecipitation (RIP) assay using an antibody against Ago2 precipitated the Ago2 protein from our cellular extract of porcine satellite cells differentiated for 48 h. The RIP-Western blot result suggested that MEG3 had the potential to combine with miRNAs ( Figure 5F). The following qPCR results showed that MEG3 ( Figure 5G), SRF ( Figure 5H), and miR-423-5p ( Figure 5I) were significantly enriched in Ago2 pellet, confirming the interaction among MEG3, miR-423-5p and SRF. Western blot results indicated that MEG3 overexpression markedly increased SRF expression, while miR-423-5p showed opposite effect on SRF. It is worth noting that the inhibitory effect of miR-423-5p on SRF could be relieved by co-transfection with MEG3 overexpression plasmid ( Figure 5J). Consistent with the Western blot results in Figure 3B,G, the knockdown or overexpression of MEG3 could respectively downregulate or upregulate the protein of SRF (p < 0.01). These findings confirmed the potential of MEG3 to combine with miRNAs, revealed the connection among MEG3, SRF, and miR-423-5p, and verified the sponge role of MEG3 for miR-423-5p.
To further confirm the function of MEG3 as a ceRNA for miR-423-5p regulating myoblast differentiation, we performed qPCR and Western blot experiments. The results demonstrated that miR-423-5p overexpression significantly inhibited MEG3, SRF, MyoD, MyoG, and MyHC expressions at the mRNA and protein levels ( Figure 5K,L). Meanwhile, the immunofluorescence staining assay of MyoG and MyHC demonstrated that the number of MyoG + and MyHC + cells was dramatically reduced after miR-423-5p overexpression (p < 0.01); however, transfection with MEG3 overexpression plasmid could substantially relieve the inhibitory effect on myoblast differentiation ( Figure 5M,N). These findings verified that MEG3 regulates myoblast differentiation via abrogating the role of miR-423-5p.
To verify the function of SRF, the target gene for miR-423-5p, we transfected the small interfering RNA (siRNA) of SRF or SRF overexpression plasmid into porcine satellite cells and induced differentiation for 48 h. qPCR results demonstrated that SRF knockdown remarkably downregulated MEG3 mRNA level ( Figure 6A). Meanwhile, mRNA and protein expression levels of MyoD, MyoG, and MyHC were considerably decreased, suggesting that si-SRF could inhibit the differentiation of porcine satellite cells ( Figure 6A,B). As expected, SRF overexpression considerably upregulated the expression of MEG3 and promoted the differentiation of porcine satellite cells ( Figure 6C,D). Collectively, MEG3 acts as a ceRNA for miR-423-5p to attenuate the inhibitory effect on SRF, thereby promoting the differentiation of porcine satellite cells (Figure 7).  The numbers below the Western blots (J) mean the fold change of SRF protein quantities related to the mimic NC group. The relative mRNA levels were normalized to those of the control 18S rRNA. The relative protein levels were normalized to those of the control β-tubulin. Error bars represent mean ± SEM of three biological replicates. Statistical significance of differences was assessed by Student's t-test. * p < 0.05, ** p < 0.01. NC, negative control. UTR, untranslated regions. RIP, RNA immunoprecipitation. qPCR, quantitative polymerase chain reaction. downregulated MEG3 mRNA level ( Figure 6A). Meanwhile, mRNA and protein expression levels of MyoD, MyoG, and MyHC were considerably decreased, suggesting that si-SRF could inhibit the differentiation of porcine satellite cells ( Figure 6A,B). As expected, SRF overexpression considerably upregulated the expression of MEG3 and promoted the differentiation of porcine satellite cells ( Figure 6C,D). Collectively, MEG3 acts as a ceRNA for miR-423-5p to attenuate the inhibitory effect on SRF, thereby promoting the differentiation of porcine satellite cells (Figure 7).

Discussion
In this study, we demonstrated that MEG3 acted as a key regulator in myogenesis and revealed a novel molecular mechanism by which MEG3 regulated the miR-423-5p-SRF axis. Our results suggested that MEG3, as a ceRNA, promotes the differentiation of porcine satellite cells by sponging miR-423-5p to relieve the inhibiting effect on SRF (Figure 7).
Proliferation and differentiation of myoblasts are crucial processes for skeletal muscle development, which determine the quality and quantity of agricultural animal meat production. Elucidating the regulatory mechanisms of myogenesis helps to find therapeutic targets for muscle disease and improve meat traits in animal production. Therefore, understanding the underlying mechanisms of myogenesis is particularly important. Although thousands of lncRNAs have been identified [38,50], a minority of functional lncRNAs, such as TncRNA [51], lncMD [52], H19 [52-56],

Discussion
In this study, we demonstrated that MEG3 acted as a key regulator in myogenesis and revealed a novel molecular mechanism by which MEG3 regulated the miR-423-5p-SRF axis. Our results suggested that MEG3, as a ceRNA, promotes the differentiation of porcine satellite cells by sponging miR-423-5p to relieve the inhibiting effect on SRF (Figure 7).
Proliferation and differentiation of myoblasts are crucial processes for skeletal muscle development, which determine the quality and quantity of agricultural animal meat production. Elucidating the regulatory mechanisms of myogenesis helps to find therapeutic targets for muscle disease and improve meat traits in animal production. Therefore, understanding the underlying mechanisms of myogenesis is particularly important. Although thousands of lncRNAs have been identified [38,50], a minority of functional lncRNAs, such as TncRNA [51], lncMD [52], H19 [52][53][54][55][56], and lncIRS1 [57], are involved in mammal myogenesis. MEG3, a differentially expressed lncRNA during postnatal skeletal muscle development in pigs, has four crucial polymorphism sites associated with back fat thickness [39]. Li et al. first confirmed the two overlapping fragment isoforms of MEG3 with respective lengths of 1160 and 1219 bp in Yorkshire and Korean native pigs [27]. Our RACE results confirmed the two variants in pigs and showed that they were 1430 and 1380 bp in full length. Subsequently, the tissue expression profile showed the advantage of their expression and the high abundance of MEG3 v2 in skeletal muscle tissues, which was consistent with the results of a previous study in various tissues and six developmental stages of longissimus dorsi muscle during porcine postnatal development [58]. In addition, the high expression level of MEG3 during the early stage of myogenesis in pigs and cattle collectively indicated its important role in myogenesis and muscle development [35]. Our research results demonstrated that MEG3 knockdown remarkably decreased the expression of myogenic marker genes in mRNA and protein levels; however, the overexpression results of the two transcripts of MEG3 were opposite. The findings conceivably indicated that MEG3 could act as an accelerator in porcine myogenic differentiation, which is also consistent with a previous study in cattle [35]. Furthermore, we further assessed differential gene expression after MEG3 knockdown, with differentiation for 30 and 40 h, and provided global insights into gene functions during myogenesis using an RNA-seq approach. KEGG pathway and GO term enrichment analysis found the involvement in muscle-related processes, such as Z disc, actin cytoskeleton, myofibril, muscle contraction, skeletal muscle cell differentiation, and structural constituent of muscle. As a consequence, it is reasonable to infer that MEG3 plays an important role in regulating porcine skeletal muscle development.
To reveal the underlying molecular mechanism of MEG3, we explored a large number of existing studies. We found that several functional lncRNAs have been characterized, and their function can be affected by multiple mechanisms. Partial lncRNAs, such as the new lncRNA SYISL [25], lncRNA Mata1 [59], and Linc-YY1 [24], regulate myoblast differentiation and skeletal muscle regeneration by recruiting chromosome modification complexes to the promoters of target genes. Other lncRNAs, such as Linc-RAM [40] and Myoparr, can recruit transcription factor MyoD and RNA-binding protein complex Ddx17/PCAF, respectively, to myogenic marker gene promoters to further promote myogenic differentiation and regeneration [60]. Also, LncMyoD acts as a competitive binding regulator to attenuate the binding ability of IMP2 for its target genes and inhibit myoblast differentiation [22]. Notably, subcellular localization determines the regulatory mechanism. Many cytoplasm-located lncRNAs can act as a ceRNA to sponge miRNA and relieve the inhibitory effect on target genes. Muscle-specific lncRNA, Linc-MD1, serves as a molecular sponge of miR-133 and miR-135 to relieve the repression of MAML1 and MEF2C and induce skeletal muscle differentiation [19]. Overexpression of lncRNA MAR1 can promote myogenic differentiation by effectively weakening the inhibitory effects of miR-487b on Wnt5a [61]. MEG3 is mainly found in the nucleus of porcine satellite cells, and very rarely is located in the cytoplasm [33,62]. However, the proportion of MEG3 in the cytoplasm was remarkably increased in the myotube in the present study. This result led us to hypothesize that MEG3 may act as a ceRNA to regulate the differentiation of porcine satellite cells similar to cytoplasm-located lncRNAs. Similarly, miR-9 can regulate the expression of nuclear lncRNA MALAT1 by directly binding with miRNA recognition elements and in an Argonaute-2-dependent manner in human L428 and U87MG cells [63]. In addition, MALAT1 is translocated from the nucleus into the cytoplasm during the G2/M cell cycle phase by interacting with heterogenous nuclear RNP C in the cytoplasm [64], where MALAT1 acts as a ceRNA for miR-133 and modulates SRF to promote the differentiation of C2C12 cell line [59]. This may be a reasonable explanation for the increased proportion of MEG3 in the cytoplasm from primary porcine myoblast to myotube in the present study. Therefore, we further affirmed the molecular sponge role of MEG3.
Software prediction analysis revealed that MEG3 certainly shared the same miRNA recognition sites for miR-423-5p with SRF, a crucial transcription factor for myogenesis process. Previous research showed functional SRF is required for the differentiation of C2C12 cells and the regulation of MyoD expression [47]. Deletion of SRF severely suppresses the muscle formation of muscle progenitors in mammalian embryonic development process [48], blocks cell fusion, and inhibits the synthesis of MyoD, MyoG, and MyHC, exerting severe muscle atrophy [47,65]. Similarly, SRF mutant mice died from severe skeletal muscle myopathy characterized by a deficiency in muscle growth during the perinatal period by inhibiting the recruitment of myocardin-related transcription factors [48,66,67]. Our observations in SRF siRNA and overexpression group provided evidence on the role of SRF in myogenesis. In addition, we found that SRF had conservative complementary sites with miR-423-5p, which is a potential regulator of myogenesis and plays a negative role during myoblast proliferation and differentiation by targeting the suppressor of fused homolog [49]. Because the Ago-RIP method described in Werfel et al. had been performed to seek ceRNA for a specific miRNA [68], we carried out an RNA immunoprecipitation assay using antibody against Ago2. The RIP-Western blot result showed the potential of MEG3 to combine with miRNAs. The enrichment of MEG3, SRF and miR-423-5p in Ago2 pellet revealed the connection among them. Subsequently, luciferase activity assays verified that MEG3 acted as a molecular sponge to adsorb miR-423-5p. MEG3 overexpression could effectively recover the reduced luciferase activity of wild-type MEG3 and SRF induced by miR-423-5p. Consistently, the inhibitory effect of miR-423-5p on SRF and myoblast differentiation could be abolished by MEG3 overexpression. These findings verified the novel functional mechanism that MEG3 acts as a ceRNA to sponge miR-423-5p, which weakens the suppression on SRF, thereby promoting the differentiation of porcine satellite cells. In addition, the fact that MEG3 was detected mainly in the nucleus implies that it can regulate myogenesis though other mechanisms, which remain to be further explored.
In conclusion, MEG3 is a vital regulator that inhibits myoblast proliferation and promotes myoblast differentiation in porcine satellite cells. Our findings suggested the novel functional mechanism of MEG3, which acts as a molecular sponge of miR-423-5p to upregulate the target gene SRF expression level during the differentiation of porcine satellite cells. Our research provides new insights into the molecular mechanisms of MEG3 in porcine myogenesis and contributes to a better understanding of molecular regulation of lncRNA in multiple pathways.