miRNA Transcriptome of Hypertrophic Skeletal Muscle with Overexpressed Myostatin Propeptide

MicroRNAs (miRNAs) play an imperative role in cell proliferation, differentiation, and cell metabolism through regulation of gene expression. Skeletal muscle hypertrophy that results from myostatin depression by its propeptide provides an interesting model to understand how miRNA transcriptome is involved in myostatin-based fiber hypertrophy. This study employed Solexa deep sequencing followed by Q-PCR methods to analyze miRNA transcriptome of skeletal muscle of myostatin propeptide transgenic mice in comparison with their littermate controls. A total of 461 mature known and 69 novel miRNAs were reported from this study. Fifty-seven miRNAs were expressed differentially between transgenic and littermate controls, of which most abundant miRNAs, miR-133a and 378a, were significantly differentially expressed. Expression profiling was validated on 8 known and 2 novel miRNAs. The miRNA targets prediction and pathway analysis showed that FST, SMAD3, TGFBR1, and AcvR1a genes play a vital role in skeletal muscle hypertrophy in the myostatin propeptide transgenic mice. It is predicted that miR-101 targeted to TGFBR1 and SMAD3, miR-425 to TGFBR2 and FST, and miR-199a to AcvR2a and TGF-β genes. In conclusion, the study offers initial miRNA profiling and methodology of miRNA targets prediction for myostatin-based hypertrophy. These differentially expressed miRNAs are proposed as candidate miRNAs for skeletal muscle hypertrophy.


Introduction
Skeletal muscle growth and maintenance are essential for animal and human health as 40-60% of the total body mass is composed of skeletal muscles, which provide structural support and enable the body to maintain posture, to control motor movements, and to store energy [1]. Skeletal muscle displays a remarkable plasticity with great capacity to alter its size, structure, and function in response to various stimuli. It therefore plays a vital role in the whole body metabolism. The muscle fibers are established early in life, and muscle growth and development are regulated by an interesting protein named myostatin, a cytokine synthesized in skeletal muscle cells. Myostatin (MSTN), also known as growth and differentiation factor-8 (GDF-8), is a member of the transforming growth factor-(TGF-) superfamily and was identified by McPherron et al. [2]. Myostatin is expressed in skeletal muscle tissue where it functions to suppress myoblast proliferation and myofiber hypertrophy [3]. Genetic disruption of myostatin causes a dramatic increase in muscle mass. MSTN binds to activin receptor type IIB (ActRIIB) to exert its biological activity [4][5][6]. The binding of MSTN to its receptor leads to phosphorylation of transcription factors Smad2 and Smad3 along with subsequently forming a complex with Smad4, resulting in nuclear translocation of the Smad complex and consequent regulation of transcription of downstream target genes [7][8][9].
Myostatin mRNA is conserved across many mammalian species and massive muscle growth is observed in several livestock animal species with mutations in MSTN gene [10]. Previously, studies reported that MSTN-knockout mice show muscle fiber hyperplasia and hypertrophy in comparison with wild type mice [2]. Like other TGF-family members, myostatin is synthesized as a precursor protein, which 2 BioMed Research International undergoes two posttranslational cleavage events, generating an N-terminal and a C-terminal peptides. The N-terminal peptide is referred to as myostatin propeptide while the Cterminal peptide is the actual mature form of myostatin. Transgenic overexpression of myostatin propeptide cDNA in skeletal muscle increases animal growth and muscle mass [11,12]. Enhanced muscle mass phenotype in the propeptide transgenic mice primarily results from myofiber hypertrophy rather than myofiber hyperplasia. The size of fast-twitch, glycolytic muscle fiber at 9 weeks of age was increased by 60% compared with wild-type littermates [11]. By systemic comparisons of global mRNA expression in hypertrophic muscle of the transgenic mice using microarray and qRT-PCR techniques, the distinct expression patterns are identified which are comprised of enhanced expressions of myogenic regulatory factors and extracellular matrix components and differentially downregulated expressions of genes related to protein degradation and mitochondrial ATP synthesis [13,14].
MSTN gene also can regulate certain miRNAs expression through feedback autoregulation mechanism [45]. In Piedmontese cattle with mutated myostatin gene, MSTN gene was found to be targeted by miR-27b and miR-27b expression was correlated with hypertrophic phenotype of Piedmontese cattle [46]. A recent study further confirmed that MSTN regulates miR-27a/b via SMAD3 in feedback autoregulation manner. Increased expression of miR-27a/b was correlated with decreased expression of MSTN and vice versa both in vitro and in vivo mouse model. Loss of SMAD3 was associated with an increased level of MSTN [47].
Although the functional significances of miRNAs in controlling myogenesis have been documented and the majority of miRNAs are abundantly expressed, identifications of novel miRNAs that are expressed at low levels during skeletal muscle development have not been well defined. With robust approaches such as high-throughput deep sequencing technology, we can produce the whole miRNA transcriptomes [48]. In this study, we have attempted to identify candidate miRNAs in hypertrophic skeletal muscle with transgenic overexpressed myostatin propeptide by using transcriptome deep sequencing. Sequencing results have shown that a group of highly abundant known miRNAs were expressed in skeletal muscle. Using computational target prediction software we have identified differentially expressed miRNAs associated with muscle hypertrophy. An interaction network between miRNA and identified targets was constructed.

Muscle Tissue Sample Collection.
Myostatin propeptide transgenic mice were generated by standard microinjection techniques using B6SJL F1 females (Taconic, Germantown, NY) as zygote donors using the transgene MLC-pro construct [11]. The MLC-pro construct consists of myosin light chain (MLC) promoter and enhancer, myostatin cDNA from the 50 untranslated regions to the nucleotides encoding the propeptide cleavage site (RSRR), and SV40 PolyA tail signal sequence. Transgenic mice and wild-type littermate mice were obtained from offspring of MLC-pro transgenic mice mating with B6SJL F1. All procedures and animal care were in accordance with the institution guidelines and approved by the Institutional Animal Care and Use Committee of the University of Hawaii, Hawaii, USA. Myostatin propeptidetransgenic mice were generated by standard microinjection techniques, which has been previously described [9]. Male mice (hemizygous genotype for the transgene) from the high-expressing line were mated with B6SJL F1 wild-type females to produce offspring mice, which were used in this study. Mice were housed in cages; room temperature was maintained at 22 ∘ C and 12 h light/dark cycle. Mice were weaned at 4 weeks of age and given free access to a chow diet (10% kcal fat, ME3.85 kcal/g). Male mice at 4 months age were sacrificed for muscle tissue dissections and sampling after 8 hr fasting. Gastrocnemius muscle samples The miRNA expression level of each library was normalized by the following formula: Normalized reads count = Reads count × 1, 000, 000 total clean reads count. ( The Student's -test was used to calculate the value and analyze the expression difference between different groups [59]. The value was calculated by fdrtool in package [60].

Novel miRNA Prediction.
The known and novel miRNAs were identified from deep sequencing data using miRDeep software v.2.1.2 package [61]. All sequences were mapped to the mouse genome using megaBLAST, and exactly matched sequences were processed for further analysis. Potential precursor miRNA sequences were extracted from the mouse genome and the secondary structures were predicted by RNAfold [62] (Figure 4).

Target Prediction.
To observe the potential function of miRNAs with significantly differential expression in two groups (transgenic versus wild-type control), we used miRDB (Version 6.2) to predict putative target genes of miRNAs with homologous human miRNAs and target prediction score ≥ 60 [63] (http://mirdb.org/miRDB/index.html). Validated targets module of miRWalk was also used to investigate hosts experimentally verified target genes of differentially expressing miRNAs in the study (http://www.umm.uni-heidelberg.de/apps/zmf/mirwalk/) [64].

Gene Ontology and Pathway
Analysis. Gene ontology analysis was performed by analyzing KEGG pathways using "database for annotation, visualization, and integrated discovery" (DAVID) website [65,66] with the following parameters: Count = 2 and EASE = 1.0. "Count" means the threshold of minimum gene counts belonging to an annotation term, and "EASE" is a modified Fisher Exact value.  (Figure 1), and these estimates were in the accordance with the known 21-23 nt range for miRNAs [67]. To assess the efficiency of deep sequencing for miRNA detections, all sequence reads were annotated and classified by analyzing the sequence tags in relation to the data from miRBase (miRBase V20). The sequence tag annotation demonstrated that known Mus muculus miRNAs accounted for 59% of all sequence reads in the MSTN and control libraries ( Figure 2). These results indicate that the deep sequencing data were highly enriched  for mature miRNA sequences, suggesting that the data are reliable for expression profiling of known miRNAs and deep mining for novel miRNAs. Comparative expression of miRNAs in transgenic and control mice skeletal muscles revealed that 461 miRNAs obtained in the sequencing data matched perfectly with 1908 known Mus musculus miRNAs in the miRBase.

miRNA Transcriptome Analysis Demonstrated the Presences of Several Highly Abundant miRNAs in Mouse Skeletal Muscle.
Almost all known muscle-specific miRNAs (myomiRs) of the more abundant miRNAs were identified in the analyzed muscle samples. The most abundant expressed miRNA was mmu-miR-22a which was represented by approximately 1637847 and 924589 reads in the small RNA libraries of the wild-type mice (CN148 and CN150) and 755381, 1123582, and 3994285 reads in the small RNA libraries of transgenic mice (TN126, TN135, and TN 329), respectively ( Table 2, see Additional File 1 in the Supplementary Materials available online at http://dx.doi.org/10.1155/2014/328935). The predominance of miR-133a was consistent with its well established function during skeletal muscle development; miR-378a is also reported to play important role during Mus musculus myogenesis. Two other myomiRs, miR-26a, and miR-27b were high-count sequences in both libraries of transgenic and wild-type mice. MyomiR 208 and 499 were expressed at a very low level in current study.

Identification of miRNA Isoforms.
The miRNAs also represent heterogeneity at 5 and/3 end ( Figure 3) and these variations from their miRBase reference sequences are referred to as isomiRs [68,69]. Examples of isomiRs are presented in Figure 3. In one case, the majority isoform, mmu-miR-16a-2, the most abundant isoform is identical to the reference in miRBase ( Figure 3, Additional File 2). In some other cases, such as mmu-miR-181b-2, mmu-miR-199a-2, and mmu-miR-16a-2, more than one highly abundant isoform was present (Figure 3), indicating that some miRNAs have more than one isoformin specific tissues.

Novel miRNAs Were Less Abundant and Less Conserved.
In addition to profiling known miRNAs, deep sequencing is a powerful strategy for discovering novel miRNAs that may not have been detected by traditionalmethods from sequencing cDNA libraries. We identified a total of 69 putative novel Mus musculus miRNAs from wild-type controls and transgenic mouse sequence tags (Additional File 3). The putative novel miRNAs were less abundant than known miRNAs (Additional File 3). Two novel miRNAs NMmu-14 and NMmu-36 having more than 100 read counts were validated and NMmu14 was found to be relatively abundant in comparison with miR NMmu36 (Figure 4) but both of these had read counts greater than 100 in the small RNA library from TN 329 library (Table 3). The novel miRNAs were less evolutionarily conserved (Additional File 3).

Identifications of Differentially Expressed miRNAs in
Hypertrophic Muscle. To identify miRNAs associated with skeletal muscle hypertrophy, we compared the miRNA transcriptoms of skeletal muscle between MSTN propeptide transgenic mice and their littermate control mice. Of the 461 known miRNAs, 57 differentially expressed miRNAs were identified (Additional File 4). The miRNAs, which satisfied the criteria log2FoldChange ≥ 1 or ≤ −1 and ≤ 0.05, value 0.1, were denoted as differentially expressed miRNAs. Cluster  analysis was generated to check the expression of miRNAs in different samples ( Figure 5).

BioMed Research International
NMmu-14 Figure 4: Predicted secondary structure of novel miRNAs: novel miRNAs secondary structure was predicted using miRDeep Software, where red color indicates the mature sequence, yellow color indicates loop sequence, blue color indicates the predicted star sequence, and purple color indicates miRNA star sequence.

Target Prediction for Differentially Expressed miRNAs.
To ascertain the query how miRNAs function in concert with their target genes in muscle tissue, we identified targets of the differentially expressed miRNAs. Of the 57 differentially expressed miRNAs, 20 abundant miRNAs were selected to predict the targets and 4,413 annotated mRNA transcripts were identified (Additional File 5).

Gene Ontology and Pathway
Analysis. Predicted 4,413 targets were analyzed for pathway analysis and the results showed that approximately 110 pathways are involved in corresponding miRNAs. Out of 110 pathways, 16 pathways are found to be associated with the most number of targeted genes (Figure 7), including mTOR signaling, insulinlike growth factor, MAPK signaling, and TGF-signaling. We also found that 20 differentially expressed miRNAs target many important genes such as FST, SMAD3, SMAD4, TGFBR1, ACRVR1a, ACVR1C, ACVR2a, ACVR, AKT3, and MEF2C (Table 4), which have been known to play a vital role in MSTN signalling pathway.

Differentially Expressed miRNA Responsible for Skeletal
Muscle Hypertrophy. A total of 57 differentially expressed miRNAs were identified; out of these miR-133a, miR-378a, and miR-26 were highly abundant. MiR-133a [23,38], miR-378a [50], miR-26a [51], miR-27b [52,53], miR-21a [56], miR-29a [44], miR-148 [58], and miR-103 are skeletal muscle specific miRNAs and play a vital role in muscle differentiation and proliferation as reported in previous studies. miR-181 is upregulated during myocyte differentiation and represses homeobox protein Hox-A11, a repressor of muscle-cell differentiation, thereby allowing new muscle growth [81]. The expression of miR-133 (miR-133a, miR-133b), miR-1, and miR-181 (miR-181a, miR-181b, and miR-181c) was profiled in muscle from patients affected by myotonic dystrophy type1 and it was observed that they were specifically induced during myogenesis [82]. miR-148a promotes myogenic differentiation and downregulates ROCK1 gene at the translational level and plays a positive role in skeletal muscle development through the RhoA/ROCK pathway [58]. MiR-1 and miR-133a were proposed to contribute in muscle hypertrophy by the removal of their transcriptional inhibitory effect on growth factors such as IGF-1. Likewise, a regulatory feedback loop was demonstrated in vitro where IGF-1 downregulated miR-1 via the Akt/FoxO3a pathway [55]. It was shown that FoxO3a increased the levels of miR-1 resulting in reduced IGF-1 protein levels. The results from this study suggest that a number of differentially expressed miRNAs that could exert novel functions in skeletal muscle development. We identified targets for 20 highly abundant   Figure 8: (a) Interaction network of differentially expressed miRNAs and their targets: network construction can be divided into two components: miRNA-target and target-target interaction, and candidate miRNA targets were predicted by mRDB. Target-target pairs interaction were searched from STRING database. In the network yellow nodes represented miRNAs and pink nodes represented targets, red lines denote miRNA-target interaction, and blue line denotes target-target interaction. These differentially expressed miRNAs predicted to be involved in TGFbeta signaling pathway. (b) MSTN signaling in skeletal muscle and miRNAs involved in the pathway: TGFb legend binds to the activin receptor two beta which interact with the TGFBR1 and they form a receptor complex, and mature myostatin inhibited by follistatin binds to the receptor complex which in turn phosphorylates regulatory SMADs, such as SMAD2 or 3. Phosphorylated regulatory SMADs are recognized by and associated with mediator co-SMADs, such as SMAD4, and translocated into the nucleus to interact with target gene promoter cis-elements and to regulate myostatin-specific gene expression. miRNAs were predicted to regulate the genes in TGFb signaling. differentially expressed miRNAs and predicted genes were analyzed to search the pathways. Sixteen pathways were selected including mTOR signaling, insulin-like growth factor, MAPK signaling and TGF-signaling and other pathways such as ubiquitin mediated proteolysis, focal adhesion, Wnt signaling pathway, regulation of actin cytoskeleton, cytokine-cytokine receptor interaction, tight junction, and cell cycle which are known to be involved in cell and tissue structure. Of these, we particularly focused on TGFsignaling, as many studies reported that TGF-family plays important roles in regulating muscle growth and development [83,84]. Pathways analysis predicted that miR-101a is targeting TGFBR1 and SMAD3, miR-582 is targeting TGFB2, SMAD1, miR-425 is targeting TGFBR2 and FST, miR-199a is targeting ACVR2a, miR-148 is targeting ACVR1, and miR-103 is targeting BMP2 in TGF-signaling pathway. A network was constructed to show the miRNA-target and target-target interaction in TGF-signaling (Figure 8(a)). Myostatin (MSTN) is predominantly expressed in skeletal muscle and plays a crucial role in muscle development and also signals through the TGF-branch [85] (Figures 8(a) and 8(b)). Myostatin binds to the TGFBR1 and makes a receptor complex for binding with ActRIIB, and activation of this receptor complex leads to the phosphorylation of SMAD2 and SMAD3.

Conclusion
This study was carried out to identify differentially expressed miRNA transcriptome in myostatin propeptide transgenic mice and to evaluate the miRNAs associated with hypertrophic skeletal muscle. Predicted target and pathway analysis concluded that miR-101a is targeting TGFBR1 and SMAD3, miR-425 is targeting TGFBR2 and FST, miR-582 is targeting SMAD1 and TGFB2, miR-148 is targeting ACVR1, and miR-199a is targeting AcvR1a gene in the TGF-signaling pathway. As far as we have known, these miRNAs and their gene targets, which are crucial to myostatin signaling pathway, have not been reported earlier. Annotations of these major miRNAs to the myostatin signal pathway in relation to skeletal muscle hypertrophy can provide important guidance for the application of miRNA approaches to enhancing skeletal muscle growth and development.