All animal care, slaughter and sample collection procedures was carried out in strict accordance with the protocol approved by the Institutional Animal Care and Institutional Ethic Committee of Northwest A&F University (NWAFU-314020038).
RNA extraction and RNase R treatment
The longissimus dorsi and soleus muscle samples were taken from three 180-day-old healthy large white pigs of the same parity and body weight. Total RNA was extracted from muscle samples by the TRIzol reagent according to the manufacturer’s instructions (Takara, Kyoto, Japan). The quality and concentration of RNA were detected by a NanoDrop 2000 instrument (Thermo Scientific, Waltham, MA, USA). Subsequently, ribosomal RNA (rRNA) was removed by Epicenter Ribo-zero rRNA Removal Kit (Epicenter, USA). Then, the residue RNAs were treated with RNase R (Epicenter, USA) to degrade the linear RNAs and were purified using RNeasy MinElute Cleanup Kit (Qiagen, Germany).
Library construction and sequencing
Ribosome RNA and linear RNA were removed to retain circRNAs. We used fragmentation buffer (Ambion, Foster City, CA, USA) to fragment the enriched circRNA and reverse transcribed the circRNA into cDNA using random primers. The second strand of cDNA was synthesized by DNA polymerase I, RNase H, dNTPs and buffer. The double-stranded cDNA was purified with the QiaQuick PCR extraction kit (Qiagen, USA), end repaired, polyadenylated, and ligated to Illumina sequencing adapters. Then, we used Uracil-N-Glycosylase (UNG) to digest the second-strand cDNA. The fragments were purified by VAHTSTM DNA Clean Beads and enriched by PCR amplification. Finally, the library products were sequenced via Illumina HiSeqTM2500.
Quality control, sequence mapping and circRNA prediction
The original image file (BCL) obtained by sequencing was base-recognized and converted into raw data of FASTQ format. Then, clean data was obtained by filtering raw data, de-joining sequences, treating low-quality reads, quality assessing, and removing ribosome RNA. The clean data were compared with the reference genome twice. First, a normal alignment was performed by Tophat2 software, and the fusion gene was not included for comparison so that the reads that could not be directly compared to the genome were obtained. Then the Tophat-fusion module of Tophat2 software was used to perform fusion gene alignment on unmapped reads, to obtain potential circular reads. The Tophat-fusion results were filtered according to circRNA reads with reverse splicing points and sequences of splicing sites that are usually GT/AG, to identify circRNA more accurately [14, 15, 16].
Expression profiling and analysis of differentially expressed circRNAs
We used HTseq software to calculate the count value for each circRNA to calculate its reads per million mapped reads (RPM) value, and count the number of splicing site sequences. DESeq was used in subsequent analysis to calculate the p-value and q-value of the gene in the comparison group, and the differentially expressed circRNAs between the samples were selected by the difference multiplier (log2 (Fold change) > 1) and the significance level (q-value < 0.05).
Network construction and KEGG enrichment analysis
We used miRanda, RNAhybrid, and TargetScan software to predict miRNA binding to exonic circRNA, and then we used Cytoscape software to construct a circRNA-miRNA interaction network. Additionally, miRanda, pita, and RNAhybrid software was used to predict the target genes of these miRNAs, and then biological pathway enrichment analysis was conducted on the target gene sets based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) biological pathway database.
In vivo injection of circMYLK4-AAV
Construction of the circMYLK4 overexpression vector and virus packaging were completed by Guangzhou geneseed. The circMYLK4-AAV virus titer packaged by geneseed is 1×1013 GC/mL. We selected five 12-day-old piglets for gastrocnemius injection. Each piglet was injected with 50 μL of virus titer of 1×1013 GC/mL in the left leg (diluted 50 μL of virus into 2 mL for easy injection), and the same amount of control virus was injected in the right leg. The piglets were sacrificed four weeks after injection, and the gastrocnemius, semitendinosus, semimembranosus and soleus muscles were separately collected for subsequent experiments.
Real-time quantitative polymerase chain reaction (RT-qPCR) and Western blot
RT-qPCR was performed on an Applied Biosystems real-time quantitative PCR machine using SYBR green master mix (Vazyme, China) according to the manufacturer’s protocol. Protein extraction and Western blot analysis of tissues were performed according to our previous methods . Antibodies against MyHCⅠ were from Developmental Studies Hybridoma Bank (DSHB, USA). Antibodies against β-tubulin were from Boster.
The experimental data are expressed as the mean ± SEM. All experimental data were analyzed by ANOVA and significance test using GraphPad Prism 7. *P < 0.05, **P < 0.01.