Analysis of MicroRNA Transcriptomes Insights into the miR-148a-3p Inducer of Rumen Development in Goats

Background: Rumen is an important digestive organ of ruminant. From fetal to adult stage, the morphology, structure and function of rumen have changed signicantly. But the intrinsic genetic regulation is still limited. We previously reported a genome-wide expression prole of miRNAs in prenatal goat rumens. In the present study, we rejoined analyzed the transcriptomes of rumen miRNAs during prenatal (E60 and E135) and postnatal (D30 and D150) stages. Results: A total of 66 differentially expressed miRNAs (DEMs) were identied in the rumen tissues from D30 and D150 goats. Of these, 17 DEMs were consistently highly expressed in the rumens at the preweaning stages (E60, E135 and D30), while down-regulated at D150. Noteworthy, annotation analysis revealed that the target genes regulated by the DEMs were mainly enriched in MAPK signaling pathway, Jak-STAT signaling pathway and Ras signaling pathway. Interestingly, the expression of miR-148a-3p was signicantly high in the embryonic stage and down-regulated at D150. The potential binding sites between miR-148a-3p and QKI were predicted by the TargetScan and veried by the dual luciferase report assay. The co-localization of miR-148a-3p and QKI was observed not in intestinal tracts but in rumen tissues by in situ hybridization. Moreover, the expression of miR-148a-3p in the epithelium was signicantly higher than that in the other layers, suggesting that miR-148a-3p involve in the development of rumen epithelial cells by targeting QKI. Subsequently, miR-148a-3p inhibitor was found to induce the proliferation of GES-1 cells. Conclusions: Taken together, these results identied the DEMs involved in the development of rumen and provided an insight into the regulation mechanism of goat rumens during development. Inhibition of in potential target gene of miR-148a-3p. Furthermore, suppression of miR-148a-3p could induced the proliferation of GES-1 cells. Our results indicate that miR-148a-3p may be an inducer of rumen development in goats by targeting QKI.


Background
Rumen is the biggest compartment of stomach in adult ruminants, it's special structure plays an important role in the digestion and absorption of nutrients. The development of rumen in goats includes the development of tissue morphology and metabolic function. The rumen wall is thin and slightly transparent, divided into ve layers, epithelial layer, lamina propria, submucosa, muscular layer and serous layer [1]. In mature ruminants, the rumen epithelium is involved in many importantly physiological functions, including absorption, transportation, volatile fatty acid (VFA) metabolism, and protection [2]. In addition, a large number of microorganisms in rumen also play a signi cant role in the processes of digestion and absorption [3]. Up to now, the effects of dietary composition [4] and weaning age [5] on rumen development have been widely investigated, while the roles of microRNAs in rumen development of goats is rarely described.
MicroRNA (miRNA), originally found in C. elegans, is a class of non-coding RNA [6]. It mainly regulates the expression of target gene by binding to the sequence of 3'-untranslated region (3'-UTR) [7]. miRNAs can regulate gene expression and participate in the regulation of almost every cellular process [8], such as cell proliferation and differentiation [9], apoptosis and metabolism [10], embryonic development [11]. For instance, it has been reported that overexpression of miR-148a-3p contributed to the proliferation of glomerular cells by targeting PTEN in lupus nephritis [12]. Overexpression of miR-148a-3p in broblastic osteosarcoma cells increased the number of cells in S phase as well as G2-M phase [13]. In gastric cancer cells, miR-148a might promote gastric cell proliferation by depressing expression of CDNK1B [14]. In addition, the QKI gene is regulated by a variety of miRNAs, thus affecting many biological processes. miR-214 inhibited angiogenesis via direct targeting of QKI and reduced the release of pro-angiogenic growth factor [15]. QKI regulated glial cell function by regulating the expression of speci c miRNAs [16]. During embryonic development, QKI was vital to the myelination and embryogenesis. QKI has also been shown to be an anti-apoptotic protein in cardiomyocyte [17].
In our previous study, the expression pro le of rumen miRNAs was revealed in the two embryonic stages (E60 and E135) [18]. Here, we investigated the DEMs identi ed in the pre-weaning and post-weaning stages (D30 and D150) and rejoined analyzed the RNA sequencing data. Furthermore, we identi ed miR-148a-3p regulated the expression of QKI by directly targeting 3′-UTR of QKI. Inhibition of the expression of miR-148a-3p could signi cantly improve the proliferation capacity of GES-1 cells, suggesting that miR-148a-3p may be involved in the process of rumen development in goats.

Experimental Animals and rumen tissue collection
The goats used in this experiment were provided by the Jianzhou Da'er Goat Breeding Center (Sichuan, China). Rumen tissues were collected from goats at the four different periods (E60, E135, D30 and D150) to perform the high-throughput sequencing. The rumen tissues were frozen in liquid nitrogen immediately after collection, and then stored at -80 °C. All experiments involving animals was conducted according to the Regulations for the Administration of Affairs Concerning Experimental Animals (Ministry of Science and Technology, China, revised in June 2004).

RNA Isolation, Library Construction, and Sequencing
Total RNA was extracted from rumen tissues using TRIzol, according to the manufacturer's instructions. The purity, concentration, and quality of RNA were determined by Nanodrop 2000 spectrophotometer, Qubit 3.0 Fluorometer, and Agilent 2100 Bioanalyzer, respectively. Libraries were constructed with RNA using the TruSeq small RNA Sample Pre Kit (Illumina, San Diego, USA). Finally, the libraries were sequenced on Illumina Hiseq 2500 platform.
miRNAs Data Analyses and qPCR validation Two data sets, the new obtained raw data (D30 and D150) and our previous data (E60 and E135) were jointly analyzed in the present study. The analyses procedures of miRNA identi cation, expression, function and targets prediction were performed according to the methods [19][20][21][22][23]. DEMs were identi ed using the criteria of fold change ≥ 1 and P-value ≤ 0.01. To verify the high-throughput sequencing data, we randomly selected 8 miRNAs from the identi ed DEMs by qRT-PCR. Total RNA (5 μg) was reverse transcribed into cDNA using the Mir-XTM miRNA First-Strand Synthesis Kit (TaKaRa, Dalian, China). The primer sequences for qRT-PCR are listed in Additional le 1: Table S1. qRT-PCR was performed using the SYBR PrimeScript miRNA RT-PCR Kit (Takara) on a CFX Connect Real-Time PCR Detection System (Bio-Rad Laboratories, Singapore). Brie y, a volume of 10 µL including 5 µL of SYBR Green Real Time PCR Master Mix (TaKaRa), 0.2 µL each of forward and reverse primer, 3.8 µL of RNase-Free ddH 2 O, and 0.8 µL of cDNA. qPCR reactions were performed using the following condition: 95.0 °C for 30 sec, and 40 cycles of 95.0 °C for 5 sec, Tm for 30 sec. U6 was used as reference gene. The expression level of miRNA was calculated with the 2 −ΔΔCT method [24]. All samples were run in triplicate.

Expression vector construction
The putative binding sites of miR-148a-3p and QKI 3'-UTR were predicted the TargetScan website (www.targetscan.org). The wild-type (wt) 3'-UTR of QKI was ampli ed by PCR. Sequences of the primers used in vector construction assays are listed in Supplemental Table 1. PCR ampli cation system contained a 30 µL volume including 2× Master Mix Taq, forward and reverse primers 2 µL, cDNA 1 µL and ddH 2 O 12 µL. The reaction cycler conditions: 95.0 °C for 5 min; 95.0°C for 30 sec, Tm for 30 sec, 72 °C for 30 sec, 35 cycles; 72 °C 5 min. The ampli ed product and the psiCHECK-2 vector (Promega, USA) were digested. The digestion system was placed at 37 °C for 2 h. The digestion products were recovered and incubated at 16 °C overnight using T4 ligase. The DH5α (TIANGEN, Beijing, China) competent cells (50 μL) were added to the ligation, culled from a single colony, identi ed and veri ed by sequencing. The psi-CHECK2 plasmid containing the mutant type (mut) 3′-UTR of QKI was obtained from GeneCreate Biological Engineering (Wuhan, China).
Cell culture, transfection and expression detection HEK293T cells were used to detect luciferase activity, and GES-1 (BNCC, Beijing, China) cells were used to detect the effect of miR-148a-3p on cell proliferation. Cells were maintained in DMEM supplied with 10% FBS at a 37 °C incubator containing 5% CO 2 . Lipofectamine TM 2000 (Invitrogen, Carlsbad, CA, USA) was used for cell transfection following the manufacturer's procedure. GES-1 cells were transfected with miR-148a-3p mimics, miR-148a-3p inhibitor, miR-148a-3p-NC, QKI siRNA or QKI-NC at the concentration of 50 nM, respectively. miR-148a-3p mimics and mimic control were chemically synthesized by Ribo Bio Co. Ltd (Guangzhou, China). The expression levels of miR-148a-3p and QKI isoforms in each group were detected by qPCR at 48 h post-transfection. QKI5 QKI6 and QKI7 primers are designed in their corresponding speci c sequence positions (Additional le 2: Figure S1).

Dual-color Fluorescence in situ Hybridization
Brie y, tissue sections were depara nized, dehydrated, and incubated in room temperature for 15 min, digested in pepsin solution for 3-30 min at 37°C. Denaturation, hybridization, and post-hybridization washings were carried out by following the manufacturer's instructions. After DAPI staining for 10 min, wash with 0.5 × SSC for 15 min; wash with 0.2 × SSC for 15 min at 37 °C. miR-148a-3p was labeled with red, while QKI was labeled with green. Images were captured by an Olympus BX51T uorescence microscope (Olympus, Japan) and analyzed by the Image-Pro Plus software (Media Cybernetics, USA).

Statistical analyses
Statistical analyses were conducted using SPSS 20.0 software. Data were analyzed by one-way analysis of variance (ANOVA). The differences between the groups were analyzed using Student's t-test and multiple comparison tests. The differences between the means were considered statistically signi cant when P < 0.05.

Results miRNAs expression pro le and DEMs in goat rumens
Twelve rumen samples representing four developmental stages (E60, E135, D30 and D150) were performed RNA-sequencing and jointly analyzed to pro le the miRNAs expression pattern of rumen tissues in this study. An average of 18.59 million single-end clean reads was generated, and approximately 50.6% mapped to the goat reference genome ARS1 (Additional le 3: Table S2). A total of 1,488 miRNAs were identi ed in the rumen samples, including 431 known miRNAs and 1,057 novel miRNAs (Additional le 4: Table S3). The top-10 miRNAs are listed in Table 1. Seven miRNAs (miR-143-3p, let-7i-5p, miR-148a-3p, let-7f-5p, miR-21-5p, miR-26a-5p and let-7g-5p) were consistently highly expressed in the rumen tissues at the four stages.
In postnatal period, 35 up-regulated DEMs and 31 down-regulated DEMs were observed between the rumens at D30 and D150 (Fig. 1a, Additional le 5: Table   S4). The hierarchical cluster of these 66 DEMs was shown in Fig. 1b. Interestingly, there were comparable DEMs in the pairwise E60 vs. D30 (27 up-regulated and 12 down-regulated) and E135 vs. D30 (19 up-regulated and 25 down-regulated) comparisons, while in the pairwise E60 vs. D150 (53 up-regulated and 29 down-regulated) and E135 vs. D150 (67 up-regulated and 44 down-regulated) the number of DEMs was markedly increased (Additional le 6: Figure S2). Venn diagrams (Fig. 1c) showed the DEMs in each comparison of all the stages. In general, the number of DEMs was increased followed by the development of goat rumen, especially for the functional transition of the rumen epidermal layer from maternal glucose dependence in the fetal period to utilization of short chain fatty acids (SCFA) in feed after weaning. The expression patterns of the selected DEMs were veri ed by qPCR and consistent with the tendency of the small RNA-sequencing data (Fig. 2).

GO and KEGG pathway analyses of putative target genes
To elucidate the biological roles on development of rumens, the target genes of the DEMs were subjected to Gene Ontology (GO) analysis. Fifty-two GO terms were identi ed, including 24 biological processes, 17 cellular components, and 11 molecular functions (Fig. 3a). The analysis of biological process showed that the most genes were involved in cellular process and biological regulation. On the molecular function level, GO terms were mainly involved in binding and catalytic activity. On the cellular component level genes were mainly related to organelle and cell part. Meanwhile, the biological pathways involved in rumen development were revealed by the Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis based on all the DEMs. In total, 188 signaling pathways were considered as signi cantly enriched (Additional le 7: Table S5), and involved in MAPK signaling pathway, Rap1 signaling pathway, Ras signaling pathway and Jak-STAT signaling pathway, which were linked to cell proliferation (Fig. 3b). In addition, six speci c GO terms and thirty-ve KEGG categories were revealed between embryonic and post-weaning stages (Additional le 8: Table S6). Furthermore, eight KEGG pathways were identi ed between pre-weaning and post-weaning periods, which enriched in tryptophan metabolism and glycosaminoglycan biosynthesis (Additional le 8: Table S6).
Candidate miR-148a-3p expression and target gene analysis The expression pattern of miR-148a-3p in the rumens was shown in Table 1, representing a signi cant high expression level in pre-weaning stages (E60, E135 and D30). A signi cant low expression was observed in the rumen at post-weaning period (D150), indicating miR-148a-3p acts an especially important role in the development of rumen epithelium cells.
The potential target genes (Additional le 9: Table S7) of miR-148a-3p were estimated by the Targetscan analysis. As for correlation of expression of miR-148a-3p and QKI was signi cantly highest, QKI was selected as the target gene to be further investigated. The binding sites of QKI 3'-UTR are highly conserved in various species, including goat, sheep and cow. The predicted minimum free energy of miR-148a-3p and QKI mRNA was found to be -22.9 kcal/mol (Fig.  4a). The recombinant plasmid was constructed by the length of 257 bp from the 3'-UTR of QKI and an approximate 6.2 KB fragment of psiCHECK-2 vector (Fig. 4b). Furthermore, we employed a luciferase reporter assay to identify the binding of miR-148a-3p and 3'-UTR of QKI. We constructed luciferase reporter plasmids containing wild-type and mutant 3′-UTR of QKI (Fig. 4c), and then co-transfected with miR-148a-3p mimics or negative control in HEK293T cells. The luciferase reporter activity of co-transfected miR-148a-3p mimics and wild type was signi cantly lower than that of co-transfected miR-148a-3p negative control and wild type (Fig. 4d, P < 0.01). The luciferase reporter activity of co-transfected miR-148a-3p mimics and mutant was not signi cantly different from that of co-transfected miR-148a-3p negative control and mutant. In addition, the luciferase reporter activity of co-transfected miR-148a-3p mimics and wildtype mimics was signi cantly lower than that of co-transfected miR-148a-3p mimics and mutant mimics (P < 0.01). These results indicated that miR-148a-3p could bind to the 3'-UTR of QKI, thereby inhibiting the expression of QKI.
Co-localization of miR-148a-3p and QKI in goat digestive tissues The co-localization of miR-148a-3p and QKI in rumen, reticulum, omasum and abomasum was performed by double-color uorescence in situ hybridization (Fig. 5). The signal locations of miR-148a-3p and QKI were consistent, and miR-148a-3p and QKI were expressed in all layers of the gastric wall, but signi cantly higher in the epithelial layer than in other layers. Therefore, we speculated that miR-148a-3p play an important role in the development of rumen epithelial cells by targeting QKI. In order to further explore the potential functions of miR-148a-3p and QKI in intestinal tract, we also investigated their colocalization in duodenum, jejunum, ileum and colon (Additional le 10: Figure S3). The results showed that miR-148a-3p and QKI were located inconsistently in many structures, and the expression levels of miR-148a-3p and QKI in various types of intestinal tract were relatively low. In duodenal sections, miR-148a-3p was higher in the muscularis than that in the villus, while QKI was highly expressed in the villus than that in the muscularis. The signal intensity of miR-148a-3p and QKI on the vascular wall of colon was signi cantly higher than that of other structures. Thus, miR-148a-3p did not bind QKI in intestinal tract, while colocalization on the vascular wall suggested that miR-148a-3p may be involved in some of certain functions.

miR-148a-3p affects cell proliferation by targeting QKI
To investigate the effect of miR-148a-3p on cell proliferation, we co-transfected miR-148a-3p mimics, inhibitor or negative control, and QKI siRNA or negative control within GES-1 cells, respectively. As shown in Fig. 6, expression of miR-148a-3p was determined by qRT-PCR after transfection, miR-148a-3p expression was signi cantly higher than that of the negative control (Fig. 6a). proliferation of GES-1 cells was then assessed at 4 h, 24 h (D1), 48 h (D2) and 72 h (D3) by the CCK8 kit, there was no signi cant difference in cell proliferation between the GES-1 cells overexpressing miR-148a-3p and the negative control (Fig. 6b). As shown in Fig. 5c, the expression levels of QKI5, QKI6 and QKI7 in the GES-1 cells overexpressing miR-148a-3p were all down-regulated, and the expression of QKI5 and QKI6 was signi cant down-regulation. Conversely, expression of QKI5, QKI6 and QKI7 was signi cantly down-regulated in GES-1 cells transfected with QKI siRNA (Fig. 6e). However, there was no signi cant effect on the proliferation of GES-1 cells (Fig. 6d). Taken together, these data suggest that the endogenous high expression of miR-148a-3p in GES-1 cells may be the reason why overexpression of miR-148a-3p has no signi cant effect on cell proliferation, while QKI, as the target gene of miR-148a-3p, is inhibited in cells with overexpression of miR-148a-3p, so silencing of QKI also has no signi cant effect on cell proliferation.
Intriguingly, the expression of miR-148a-3p was signi cantly deceased after transfection with miR-148a-3p inhibitor (Fig. 6f). As shown in Fig. 6g, the proliferation of GES-1 cells was signi cantly increased after 48 h transfection. Compared to negative control, expression level of QKI5 were slightly up-regulated, while the expressions of QKI6 were signi cantly down-regulated (Fig. 6h). These data suggest that miR-148a-3p could inhibit the expression of QKI5 by binding to its 3'-UTR. Furthermore, inhibition of miR-148a-3p expression could promote the proliferation ability of GES-1 cells.

Discussion
Expression pro les of goat rumens during pre-and post-weaning periods In the present study, we mainly focused on the expression characteristics of genome-wide microRNAs involved in rumen development and their potentially affects especially on the functional transition of the rumen epidermal layer in goats. Seven highly expressed miRNAs mainly involving in cell proliferation, growth and apoptosis were identi ed at the four tested periods. Among them, miR-143-3p has been veri ed to be related with the regeneration of skeletal muscle cells by targeting IGFBP5 [25], and it could target the CTGF and inhibit the Akt/mTOR signaling pathway to regulate the proliferation and apoptosis of broblasts [26]. miR-21-5p and miR-26a-5p have also been shown to be associated with the proliferation of melanoma and hepatocellular carcinoma cells, respectively [27,28]. As for the members of let-7, expressions of let-7i-5p, let-7f-5p and let-7g-5p have been revealed to regulate the regeneration rate of liver in mice [29].
Compared with the rumens from post-weaning period (D150), a similarity number of DEMs (82 in E60 and 111 in E135) were detected the rumens at prenatal stages (Additional le 6: Figure S2). It was notable that the rumen function was absorption maternal glucose in embryonic stage, while solely obtain energy from SCFA in feed after weaning. From fetal to birth and adult stage, the morphology, structure and function of rumen have signi cantly changed, especially the surface area of the rumen papillae of goats under supplementary feeding signi cantly increased during the pre-and post-weaning periods [30]. 66 DEMs were observed between the rumens at D30 and D150 (Fig. 1a). We previously reported that miRNAs expression in embryonic rumens and less number of DEMs (22 up-regulated and 20 down-regulated) was identi ed between E60 and E135 [18].
Histomorphological analyses have found that rumen development in cattle [31] and sheep [32] were slightly faster than that in goats during embryonic period. During weaning, the length and surface area of the rumen papillae of goats and calves increased, and rumen development signi cantly changed [33,34].
Accompanied by the development of rumen size, rumen papillae and muscular tissues, gene expression were also remarkably altered though transcriptional and post-transcriptional regulation. Plenty of differential expressed genes (DEGs) have been identi ed in cattle between pre-and post-weaning [2,35,36].
During weaning, the DEGs in calf rumen epithelium were mainly involved in the processes of lipid metabolism, cell morphology, growth and proliferation, molecular transport [2]. Later, more DEGs (4,104) were revealed between the pre-and the post-weaning periods in bovine rumen and several highly DEGs with immune functions (LY6D, MUC1, EMB and EYA2) were pointed out. Similarity, immunity and lipid metabolism were also signi cantly enriched for DE genes in the rumen [35]. Subsequently, 122 DEMs were identi ed from 260 known and 35 novel miRNAs in the same rumen tissues. Among them, six miRNAs (miR-143, miR-29b, miR-145, miR-493, miR-26a and miR-199) were identi ed as the key regulators of rumen development in cattle [36]. However, the studies about miRNAs in goat rumens still limited.

miRNAs related with the pathways involved in rumen development
With regard to biological functions of the target genes of DEMs, four relevant pathways were enriched signi cantly in goat rumens, including the MAPK signaling pathway, Jak-STAT signaling pathway, Rap1 signaling pathway, Ras signaling pathway and PI3K-Akt signaling pathway (Fig. 2b). Several factors could effect cell proliferation and differentiation mediated by the epidermal growth factor receptor (EGFR) with respond to EGF and transmembrane transforming growth factor (TGF) through the MAPK signaling pathway [37], it promoted the growth of gastric epithelium during development and the renewal of the gastric tissue [38]. It has been shown that inhibition of Jak-STAT3 signaling pathway could inhibit cell proliferation in gastric cancer cells [39]. Ras signaling was a downstream pathway of EGFR and played an important role in the cell fate and proliferation of intestinal epithelial cells [40]. In addition, phosphorylation of PI3K-Akt signaling pathway can promote ghrelin-mediated intestinal cell proliferation [41]. These ndings indicated that epithelial cell proliferation during rumen development is closely related to MAPK signaling pathway, Ras signaling pathway and PI3K-Akt signaling pathway.
Compared with prenatal rumens (E60 and E135), the differently enriched pathways at the post-weaning stage (D150) were mainly clustered into carbohydrate digestion and absorption, vitamin digestion and absorption, adherens junction, gap junction, amino acid metabolism (Additional le 8: Table S6).
Subsequently, the speci c pathways including tryptophan and metabolism, ether lipid metabolism and glycosaminoglycan biosynthesis were revealed in comparison of pre-and post-weaning rumens (D30 and D150). During the processes of rumen development, the histomorphology changes and functional transitions of rumen should be followed by gene regulation [42]. These results indicated that some DEMs may be involved in epithelial cell junctions, digestion and metabolism of nutrients.
To date, many investigations have focused on the molecular mechanism of rumen epithelial cell proliferation and related transporter regulatory pathways, such as insulin-like growth factor (IGF) and epidermal growth factor (EGF) involved in the regulation of glucose transport, NHE, MCTs and GPR involved in the transport of SCFA in rumen epithelial cells [43,44]. With respect to other external effects, diet composition and nutrient level can affect rumen development, and the expression of epithelial cell genes can also affect gastrointestinal development. Malmuthuge and colleagues demonstrated that the VFAs produced by microbiota could stimulate processes of rumen tissue metabolism and epithelium development via interacting with the host transcriptome and microRNAome. Approximately half of miRNAs and over 25% of mRNAs were correlated with the diversities of VFAs in neonatal calves [45]. In addition, ratio of non-brous carbohydrate/neutral detergent ber (NFC/NDF) in diet could affect the the length and width of rumen papillae. One differently expressed bta-miR-128 induced by NFC/NDF was found to regulate the rumen development via regulating PPARG and SLC16A1 genes [46]. Moreover, in grass-and grain-fed cattle could also resulted in alternations of miRNA expression potentially in uencing rumen function, as well as mRNA expression and DNA methylation [47]. miR-148a-3p functions in proliferation, differentiation and apoptosis In the current study, we have found that miRNA-148a-3p was highly expressed in rumen tissues, especially in D30 (2-fold), while its underlying regulatory mechanism in rumen development remains unclear. FISH results (Fig. 6) demonstrated that miR-148a-3p and QKI were co-expressed in rumen, reticulum, omasum and bomasum, compared with in other gastrointestinal tract samples, miR-148a-3p and QKI are expressed in different locations (Additional le 10: Figure S3). Furthermore, a high correlation of miR-148a-3p and QKI expression was found at all the four periods (Additional le 9: Table S7). As revealed in GES-1 cells, miR-148a-3p suppressed the cell proliferation by regulating QKI, suggesting that a similarity effects on the rumen epithelial cells.
A previous study has revealed that miR-148a-3p was signi cantly up-regulated during fetal liver development, and its target gene also participated in the differentiation process of liver cells [48]. miR-148a-3p could induce hepatocytic differentiation by inhibiting the IKKα/NUMB/NOTCH signaling pathway [49]. In domestic animals, miR-148a-3p was highly expressed in fetal bovine skeletal muscle, but decreased in the growing myocytes. It was found that the overexpression of miR-148a-3p could inhibit the proliferation of myocytes and promote the proliferation of myocytes after interference [50]. In rabbits, miR-148a-3p is not only highly expressed in white adipose tissue at early growth stage, but also gradually increased in the differentiation process of preadipocytes cultured in vitro [51]. In pigs, miR-148a-3p was always highly expressed during the development of skeletal muscle, which was speculated to be related to the proliferation, generation and apoptosis of muscle cells [52].
Here, QKI was con rmed to be a target gene of miR-148a-3p in goat rumens. QKI is a kind of RNA binding protein, which is a member of RNA signal transduction and activation family and plays an important role in mammalian embryo development. QKI protein has a maxi-KH domain, which is transcriptionally controlled by QKI-RNA recognition element CUAAC sequence, which is involved in many RNA metabolic processes, and has a target regulatory role in the localization, translation, transport and stability of mRNA [53]. The main subtypes of QKI are QKI5, QKI6 and QKI7, in which QKI5 mainly exists in the nucleus, QKI6 is distributed in all parts of the cell, and QKI7 mainly exists in the cytoplasm. The expression of QKI was regulated by various miRNAs, which affected its biological functions. For example, miR-214 can target and regulate the expression of QKI and participate in the differentiation of vascular smooth muscle cells [54]. miR-148a-3p could affect the proliferation of cancer cells by targeting QKI5 [55]. In addition, QKI can also act on miRNAs precursor and feedback regulate miRNAs expression [16].
Previous studies reported that the expression of miR-148a-3p was affected by methylation. The expression of QKI was also affected by the methylation level of its promoter region, which contained CpG islands in the upstream region of pre-miR-148a-3p [56]. Meanwhile, miR-148a-3p also regulated the expression of DNMT1 and affected the physiological processes of different cells [57][58][59]. The binding site of miR-148a-3p was also con rmed to be existance in the 3′-UTR of DNMT1. Thus, It is speculated that the expression of miR-148a-3p should be regulated by epigenetics [60]. In this study, we found that the expression of miR-148a-3p was signi cantly different in the rumen tissues between embryonic stage and after weaning. miR-148a-3p may affect cell proliferation by binding to QKI5 3'-UTR and inhibiting its expression. Therefore, during rumen development, DNA methylation whether involve in and how to regulate the expression of miR-148a-3p and QKI remains to be further studied.

Conclusion
In summary, genome-wide expression pro le of miRNAs and speci c pathways involved in the development of rumen were identi ed in goats. FISH and luciferase experiments veri ed that QKI is a potential target gene of miR-148a-3p. Furthermore, suppression of miR-148a-3p could induced the proliferation of GES-1 cells. Our results indicate that miR-148a-3p may be an inducer of rumen development in goats by targeting QKI.

Declarations
Ethics approval and consent to participate The protocol for collecting ruminal biopsies was approved by the Institutional Animal Care and Use Committee of the College of Animal Science and Technology of Sichuan Agricultural University, Sichuan, China, under the permit of No. DKY-B20131403.

Consent for publication
Not applicable.

Availability of data and material
Goat rumen RNA-Seq data are accessible at SRA database (PRJNA596079).

Competing interests
The authors declare no con icts of interests.    a Alignment of binding sites among different species and the predicted secondary structure of miR-148a-3p and targeting QKI. b Electrophoresis of the recombinant plasmid. c Base complementarity between miR-148a-30 and the 3'-UTR of QKI. d Luciferase reporter assay of HEK293T cells co-transfected with QKI wild-type or QKI mutant-type and miR-148a-3p mimics Co-location of miR-148a-3p and QKI in different stomach chambers. Bar. 500 μm. Arrows indicates the epithelial layer.