Identification of microRNA-21 target genes associated with hair follicle development in sheep

Target prediction

Target prediction of microRNA-21 was performed using the algorithms TargetScan (http://www.targetscan.org/), Pita (http://genie.weizmann.ac.il/pubs/mir07/mir07_dyn_data.html) and miRanda (http://www.microrna.org/microrna/home.do). These algorithms are available from a previous report (Sun et al., 2014). In brief, TargetScan predictions are based on the matching of sites between the seed regions of microRNAs and mRNAs and miRanda predictions are based on the double-stranded free energy and species conservation. Pita predictions are based on target-site accessibility. All the prediction processes were conducted using custom-written executable files that computed the parameters between microRNAs and mRNAs based on the inherent algorithms and the set thresholds. The thresholds for the algorithms were as follows: the scores ≥70 for TargetScan; Δ ΔG ≤-5 for Pita; and seed ≥7 for miRanda. The intersections of the output results of the three algorithms were used as prediction results for the differentially expressed microRNAs (Sun et al., 2016).

Dual-luciferase reporter vector construction and luciferase reporter assays

The reaction conditions were as follows (landing PCR): 98 °C for 2 min of denaturation; cycling with denaturation at 98 °C for 10 s, annealing every loop drop 1 °C. From 65 °C, extension at 72 °C for 30 s; followed by 10 cycles starting with denaturation again at 98 °C for 10 s, annealing at 60 °C and then extension at 72 °C for 30 s. Finally, 15 cycles were performed, the extension was continued at 72 °C for 3 min after PCR cycles and the reactions were then stored at 4 °C.

PCR primers were designed based on the seed region sequences of candidate target genes (Ovis aries) with 3′ UTRs provided by the National Center for Biotechnology Information (NCBI). Primers targeting the gene segments containing the seed sequences were designed with Xho I and Not I restriction enzyme sites incorporated. In addition, mutagenic primers were designed to specifically mutate 7 bases in the seed region of the candidate target genes. The primer sequences are listed in Table 1. Then, candidate target sequences containing wild-type and mutant seed regions were synthesized. The wild-type and mutant sequences were cloned into pmiR-RB-REPORT™ to construct wild-type and mutant reporter plasmids, respectively. The recombinant vector was sent to Guangzhou Ruibo Biotechnology Co., Ltd. for sequencing.

MiR-21 mimics (5 pmol) or negative control (NC 5 pmol) miR (Shanghai GenePharma Co., Ltd., China) of sheep were cotransfected with 2 ng of the candidate target gene-containing wild type, mutant or empty pmiR-RB-REPORT™ vector (negative control) into the 293T human renal epithelial cell line. The sequences are listed in Table 2. After culture for 48 h, Renilla luciferase and firefly luciferase activities were measured using a chemiluminescence reader.

RNA extraction and quantitative real-time reverse transcription polymerase chain reaction (RT-qPCR)

In October, January and February, to capture the different stages in hair follicle growth, four sheep were randomly selected. Total RNA from skin collected from the sides of the sheep was extracted using Trizol reagent (Invitrogen, USA) in accordance with the instruction manual provided by the manufacturer. Reverse transcription of miR-21 was conducted using the PrimeScript™ RT reagent Kit (TaKaRa, Japan). Reverse transcription reactions were prepared as follows: 2 µl of 5x PrimeScript Buffer, 0.5 µl of PrimeScript RT Enzyme Mix I, miR-21 RT primer (10 µM), 0.5 µl of U6 RT mix, and RNase-free ddH2O up to 10.0 µl. Reactions were performed with the following conditions: 37 °C, 15 min; 42 °C, 2 min; and 85 °C, 5 s. The reaction products were stored in a −80 °C freezer.

RT-qPCR was performed using SYBR Premier Dimer EraserTM (TaKaRa, Japan) and cDNA as the template according to the manufacturer’s instructions. The reaction mixes were prepared as follows: 0.5 µl of upstream primer (10 pmol/L), 0.5 µl of downstream primer (10 pmol/L), 8.0 µl of ddH2O, 1 µl of template and 10 µl of SYBR Premix Ex Taq II (2x). Reactions were performed with the following conditions: 95 °C for 30 s; and 40 cycles of 95 °C for 5 s followed by 60 °C for 30 s. The U6 small nuclear RNA (snRNA) was used as an internal control for miR-21, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an internal control for other genes. Experimental data were analysed using a relative quantification method (2^−△△CT). All reactions were performed in triplicate. The RT-qPCR primers are described in detail in Table 3.

Statistical analysis

All quantitative data and luciferase reporter data are presented as the means ± SDs, and differences were assessed using one-way ANOVA followed by the Bonferroni multiple-comparison test. Values of p < 0.01 were considered statistically significant, and all analyses were performed using SPSS 20.0.

Prediction of miR-21 target genes

According to the rules of bioinformatics for predicting target genes, miR-21 starts from the second base pair, and the three genes at the 3′ UTR sequence are matched with each other. The number of paired bases is more than seven, as shown in Fig. 1. The combined free energy is −16.7, −10.0 and −14.5 kcal/mol, respectively. Seed sequences are strictly conserved in species such as humans, cattle, sheep, and rats. Therefore, these three genes can be used as candidate target genes for later identification and verification.

Identification and expression of CNKSR2 target genes

To obtain the gene sequences, landing PCR was used to amplify CNKSR2 (Fig. 2A) and sequences with a length of 759 bp were identified by agarose gel electrophoresis; this length is consistent with the length provided in the nucleic acid database. The size of the 3′ UTR fragment obtained after digestion was confirmed to be the 1,000 bp expected size (Fig. 2B). The forward and reverse sequencing maps showed that the 3′ UTR sequence in the recombinant vector was exactly the same as the length of the PCR product (Fig. 2C) and the length provided in the GenBank database. The CDS sequence of the provided gene was the same, so the reporter recombinant vector was named CNKSR2-WT. After a similar procedure, the target sequence ATAAGCT was mutated to TATTCGA (Fig. 2D), and the remaining sequences were not changed. The reporter recombinant vector was named CNKSR2-Mut.

Under normal circumstances, the reporter fluorescence of miRNA mimics a 30% downregulation or more compared with that of the NC controls; after the corresponding binding site is mutated, the inhibition degree of the miRNA mimic reporter fluorescence is significantly reduced or eliminated compared with that of the mutant vector. We can infer that miRNA mimics can regulate reporter gene expression through this site. The dual-luciferase reporter gene expression analysis showed that the oar-miR-21 mimic significantly downregulated the reporter fluorescence of the three wild-type vectors. The reduction of the fluorescence rate of CNKSR2 was 36%. After the corresponding binding site was mutated, the reporter fluorescence in the mutant vector was significantly restored. Under this experimental model, oar-miR-21 may regulate the expression of the reporter gene through the corresponding 3′ UTR site on CNKSR2 (Fig. 2E). When the mRNA expression levels of miR-21 and CNKSR2 in the two extreme phenotypes were negatively correlated, highly significant differences were observed (Fig. 2F).

Identification and expression of KLF3 target genes

To obtain the gene sequences, landing PCR was used to amplify KLF3 (Fig. 3A). Fragments with a length of 607 bp were identified by agarose gel electrophoresis; this length is consistent with the length provided in the nucleic acid database. The expected size of the 3′ UTR fragment of the 850 bp obtained after digestion was confirmed (Fig. 3B). The forward and reverse sequencing maps showed that the 3′ UTR sequence in the recombinant vector was exactly the same as the length of the PCR product (Fig. 3C) and the length provided in the GenBank database. The CDS sequence of the provided gene was the same, so the reporter recombinant vector was named KLF3-WT.

After a similar procedure, the target sequence ATAAGCT was mutated to TATTCGA (Fig. 3D), and the remaining sequences were not changed. The reporter recombinant vector was named KLF3-Mut. The reduction of the fluorescence rate of KLF3 was 26%. After the corresponding binding site was mutated, the reporter fluorescence in the mutant vector was significantly restored. Under this experimental model, oar-miR-21 may regulate the expression of the reporter gene through the corresponding 3′ UTR site on KLF3. When the mRNA expression levels of miR-21 and KLF3 in the two extreme phenotypes were negatively correlated, highly significant differences were observed (Fig. 3F).

Identification and expression of TNPO1 target genes

To obtain the gene sequences, landing PCR was used to amplify TNPO1 (Fig. 4A). Fragments with a length of 680 bp were identified by agarose gel electrophoresis; this length is consistent with the length provided in the nucleic acid database. The expected size of the 3′ UTR fragment of the 950 bp obtained after digestion was confirmed (Fig. 4B). The forward and reverse sequencing maps showed that the 3′ UTR sequence in the recombinant vector was exactly the same as the length of the PCR product (Fig. 4C) and the length provided in the GenBank database. The CDS sequence of the provided gene was the same, so the reporter recombinant vector was named TNPO1-WT.

After a similar procedure, the target sequence ATAAGCT was mutated to TATTCGA (Fig. 4D), and the remaining sequences were not changed. The reporter recombinant vector was named TNPO1-Mut. The reduction of the fluorescence rate of TNPO1 was 48%. After the corresponding binding site was mutated, the reporter fluorescence in the mutant vector was significantly restored. Under this experimental model, oar-miR-21 may regulate the expression of the reporter gene through the corresponding 3′ UTR site on TNPO1. When the mRNA expression levels of miR-21 and TNPO1 in the two extreme phenotypes were negatively correlated, highly significant differences were observed (Fig. 4F).

The expression pattern of miR-21 and three target genes in hair follicle development

RT-qPCR was used to obtain the expression pattern, 2^−ΔΔCt method was used to quantify diferential expression level. The expression level of CNKSR2/KLF3/TNPO1 in October (HF flourishing period) are significantly higher than other months (p < 0.01) (Fig. 5A).The expression level of miR-21 in October are significantly decreases compared to other months (p < 0.01) (Fig. 5B). The expression level in January (HF degenerative period) and February (HF resting period) have no significant difference (p < 0.05).