lncRNA identification in porcine LD muscle
To verify the effect of lncRNAs in LD muscle development during porcine embryonic period, we examined the DELs of LD muscles between HN and LW pigs at the three embryonic-development stages. In total, there were 1.94 billion clean reads in these 18 samples, and 79.63% mapped to the porcine reference genome (Sscrofa10.2), 12.87% were multiply mapped, 66.76% were uniquely mapped, 33.34% mapped to “+”, 33.47% mapped to “–”, 19.24% were splice reads, and 47.52% was nonsplice reads (Additional Table.1). Expression correlation between the three samples in the same treatment was from 0.953 to 0.969 (Additional Fig. 1), indicating that sample selection was reasonable, and experiment results were reliable.
lncRNA properties in porcine LD muscle
In total, 2131 novel lncRNAs and 2057 annotated lncRNAs were detected from 18 LD muscle samples (Fig. 1A), and only lincRNA(89.9%)and intronic lncRNA were found in the novel lncRNAs (Fig. 1B). The annotated lncRNAs’ average length was shorter than the novel lncRNAs’ average length, but there was no significant difference in open reading frame (ORF) length and exon number (Fig. 1C). For the novel lncRNAs, antisense lncRNAs’ average length was longer than lincRNAs’ average length (Fig. 1D). Information on the novel lncRNAs is shown in Additional Table 2.
Expression difference of lncRNAs between HN and LW pigs
The expressional level of lncRNAs were lower than that of mRNAs (Fig. 2A), and lncRNAs FPKM distribution of LW pigs at 78 dpc was higher than that of the others (Fig. 2B). Systematic cluster analysis was carried out to analyze the 18 LD muscle libraries’ relationship, and results indicated that three replicates of the same sample were very conservative (Fig. 2C). The lncRNAs expression differences between the different stages were larger than those of the two breeds. There were 1155 and 751 DELs during muscle development in HN and LW pigs, respectively. In HN pigs, 579 and 809 DELs were identified at 58 vs 38 dpc and 78 vs 58 dpc, and 213 DELs were shared in these two comparisons. In LW pigs, there were 579 and 809 DELs at 58 vs 38 dpc and 78 vs 58 dpc, respectively, and 167 DELs were shared. At 78 vs 58 dpc, there were more DELs in HN than in LW pigs. There were 291, 305, and 683 DELs between these two breeds at 38, 58, and 78 dpc, and 37 DELs were shared by these three stages (Fig. 2D and Additional Fig. 2 and Table 3).
During embryonic muscle development, the GO enrichment of DEL-co-expressed genes in HN showed that muscle tissue development and the response to oxygen-containing compound were upregulated in 58 dpc. In LW pigs, only the response to the oxygen-containing compound was upregulated at 58 dpc, and muscle cell differentiation, myofibril assembly were only upregulated at 78 dpc; other muscle development-related pathways were continuously upregulated in both breeds. At 58 vs 38 dpc, muscle cell differentiation and myofibril assembly were only upregulated in HN at 58 dpc. Skeletal muscle tissue development was only upregulated in LW at 78 dpc (Table 1 and Additional Table 3).
KEGG enrichment of DEL-co-expressed genes shown that the cyclic adenosine monophosphate (cAMP), glyoxylate and dicarboxylated metabolism, metabolic pathways were only upregulated in HN pigs at 58 dpc. Fatty acid degradation and fructose and mannose metabolism was only upregulated in LW pigs at 78 dpc. The adipocytokine pathway was upregulated in HN but downregulated in LW pigs at 58 dpc. The MAPK pathway was upregulated only at 58 dpc in both breeds, and fructose and mannose metabolism was upregulated only at 78 dpc in both breeds. DEL-co-expressed genes related to fatty acid degradation, cAMP, glyoxylate and dicarboxylated metabolism, fatty acid metabolism, metabolic pathways continuously changed only in HN or LW pigs. Genes associated with metabolism of alanine, aspartate and glutamate, and Rap1 with higher expressional level in HN pigs at 78 dpc, while genes in protein digestion and absorption with a higher level in LW pigs at 78 dpc (Table 2 and Additional Fig. 3).
The results of GO enrichment of DELs between HN and LW pigs at the same stage showed that no pathway was enriched between these two breeds at 38 dpc. At 58 dpc, genes in muscle cell differentiation and eight other pathways had a lower expression level in LW pigs. At 78 dpc, the genes in seven pathways were upregulated in LW pigs. Genes associated with muscle cell differentiation, skeletal muscle tissue/organ development, muscle organ development pathways showed, higher expressional level at 58 dpc, but a lower expression level in HN pigs than that of in LW pigs (Table 1 and Additional Table 3).
KEGG analysis indicated that DEL-co-expressed genes involved in insulin, calcium, and the cAMP signaling pathway were continuously highly expressed in HN pigs. Genes in steroid biosynthesis showed a higher level in HN pigs at 58 and 78 dpc. Compared with LW pigs, genes in hypoxia-inducible factor 1 (HIF-1) and adipocytokine signaling pathway were upregulated in LW pigs at 38 and 78 dpc, but downregulated at 58 dpc. Genes involved in insulin secretion and Rap1 were downregulated at 58 dpc and upregulated at 78 dpc in HN pigs. The expression trends of biosynthesis of amino acids, fructose and mannose metabolism, and fatty acid metabolism were completely opposite (Table 2 and Additional Fig. 3).
Potential function of DELs
The expressional level of 37 shared DELs between these two breeds at three stages are shown in Fig. 3A, B. The ceRNA regulatory network of these shared DELs was constructed and visualized using Cytoscape software, including 30 lncRNAs, 27 miRNAs, 27 mRNAs, and 24 pathways (Fig. 3C). LncRNAs had up to 7 interacting miRNAs, such as ALDBSSCT0000006192, and miR-199a-5p had the most target lncRNAs, seventeen target lncRNAs for each miRNA. Considering the abundance and transcript length, IMFlnc1 was selected for subsequent verification among miR-199a-5p target lncRNAs. IMFlnc1 is located on porcine chromosome 7, and includes two exons. The IMFlnc1 expressional level in intramuscular adipose tissue was higher than that in LD muscle tissue (Fig. 4A), so it was speculated that IMFlnc1 might play an important role in intramuscular adipose tissue, so we firstly verified its regulation role in adipogenesis of porcine intramuscular adipocytes.
IMFlnc1 promotes adipogenesis of intramuscular adipocytes
RT-qPCR showed that IMFlnc1 showed the highest expressional level in the gut and lungs, followed by in intermuscular fat. Its expression level in subcutaneous fat was lower than it was in intermuscular fat, but higher than it was in LD muscle (Fig. 4A). Moreover, time-course analysis showed that IMFlnc1, CAV-1 and PPARgama were upregulated during the differentiation of porcine intramuscular preadipocyte, which was similar with PPARgama expressional trends (Fig. 4B).
Analyzed by CPC software, IMFlnc1 showed very low coding potential, similar to a well-known lncRNA—ADNCR [22], (Fig. 4C). To explore the function of IMFlnc1 in adipogenesis, we performed knockdown IMFlnc1 in intramuscular adipocytes by lncRNA smart silencer, its RNA level was significantly reduced, and CAV-1 and PPARgama (adipogenic markers) significantly downregulated (Fig. 4D). The downregulation of CAV-1 by IMFlnc1 siRNA was confirmed by Western blot (Fig. 4E). Oil Red O staining indicated that adipogenesis was inhibited by knockdown of IMFlnc1 (Fig. 4F).
IMFlnc1 and CAV-1 were miR-199a-5p’s target genes
IMFlnc1 is mainly localized in preadipocyte cytoplasm (Fig. 5A), so it might participate in the regulation of adipogenesis through ceRNA mechanism. The IMFlnc1-miR-199a-5p-CAV-1 pathway was selected from the ceRNA network to verify its function in adipogenesis. Bioinformatics analysis of the RNAhybrid showed that there exists a binding site of miR-199a-5p in IMFlnc1 and CAV-1 (Fig. 5B), and the binding site in CAV-1 is conservative in different animals (Fig. 5C). The results of RT-qPCR showed that the expressional trends of IMFlnc1 and CAV-1 has a positive correlation in porcine LD muscle (Fig. 5D, R2 = 0.590).
Luciferase assay indicated that miR-199a-5p significantly reduced the fluorescence activity of the sensor and psiCHECK2-IMFlnc1 (p < 0.01, Fig. 5E). However, miR-199a-5p couldn’t reduce the fluorescence activity of psiCHECK2-IMFlnc1-Mut, which indicated that miR-199a-5p target IMFlnc1. Similarly, miR-199a-5p also target CAV-1 (Fig. 5F).
IMFlnc1 participates in adipogenesis by increasing CAV-1
To verify whether IMFlnc1 might sponge miR-199a-5p, the miR-199a-5p sensor was transfected with pcDNA3.1+, pcDNA-miR-199a-5p or pcDNA-IMFlnc1. The pcDNA-miR-199a-5p significantly reduced the fluorescence activity of the miR-199a-5p sensor (p<0.01), but the fluorescence activity could be recovered by IMFlnc1 in a dose-dependent manner (p<0.01, Fig. 5G). However, pcDNA-IMFlnc1-Mut (the binding site of miR-199a-5p was mutated) couldn’t recover the fluorescence activity. It was indicated that IMFlnc1 could sponge miR-199a-5p.
From Fig. 4D and 4E, mRNA and protein of CAV-1 was downregulated during the reduction of IMFlnc1. To further determine whether IMFlnc1 regulated CAV-1 through miR-199a-5p, psiCH2-CAV-1 was co-transfected with pcDNA3.1, pcDNA-IMFlnc1 or pcDNA-IMFlnc1Mut, respectively. The Rluc activity of CAV-1 was improved by the overexpression of IMFlnc1, but the overexpression of IMFlnc1 with the mutated miR-199a-5p binding sites no longer elicited similar effect (Fig. 5H). And Oil Red O staining confirming that, compared with inhibition of IMFlnc1, the number of mature adipocyte was increased after inhibition of miR-199a-5p and IMFlnc1. (Fig. 5I). In summary, these results indicated that IMFlnc1 might promote adipogenesis by sponging miR-199a-5p.