Animal samples, used in this study, were collected using a protocol approved by the “Animal Care and Use Committee” of the College of Animal Science and Technology, Northwest A&F University (No. NWAFAC1117).

All the Qinchuan cattle used in this study had similar genetic background, and animals were housed under similar management, nutritional, and environmental conditions at the experiment farm of the National Beef Cattle Improvement Center (NBCIC, Yangling, China). At each developmental stage (0, 6, 12, 18, 24, and 60 months of age), three cattle were randomly selected to collect the dorsal subcutaneous adipose tissue (backfat), which is deposited between the 12th and 13th rib. The adipose tissue was collected after animals were slaughtered from the mechanized line at the Shaanxi Qinbao Animal Husbandry Development Co. Ltd. In addition, perirenal and groin fat tissue samples were collected from the newborn calves. All samples were immediately stocked in liquid nitrogen and subsequently stored at −80°C until RNA isolation.

HEK293A cell line was purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA). The bovine primary preadipocytes were isolated from the Qinchuan cattle and maintained at NBCIC (Wang Y. et al., 2018). HEK293A cells and bovine preadipocytes were cultured in DMEM-high glucose and DMEM-F12 (HyClone, Logan, UT, USA) medium, respectively, supplemented with 10% fetal bovine serum (FBS, Gibico, Carlsbad, CA, USA) and 1% penicillin-streptomycin (HyClone). The cells were incubated at 37°C under saturated humidity and 5% CO₂.

The bovine preadipocytes were transfected at 70–80% confluence with miR-424 mimics (50 nM) or miR-424 inhibitor (100 nM) or their respective negative control (NC) (RiboBio, Guangzhou, China). Co-transfection was performed with STK11 3' UTR related plasmid (100 ng/μl) using Lipofectamine 3000 (Invitrogen, Carlsbad, CA, USA) reagent according to the manufacturer’s instructions.

The short hairpin RNA (shRNA) targeting STK11 was designed and synthesized as follows: sense, 5'-GACGACGAGCTCTTCGACATTTCAAGAGAATGTCGAAGAGCTCGTCGTCTTTTTTAGATCTG-3' and antisense, 5'-GATCCAGATCTAAAAAAGACGACGAGCTCTTCGACATTCTCTTGAAATGTCGAAGAGCTCGTCGTC-3'. The recombinant adenovirus pAd-sh-STK11 was constructed using AdMax^TM recombinant adenovirus packaging system, as previously described (Zhang et al., 2019). The virus titer was detected by enhanced green fluorescent protein (EGFP) labeled method (Le et al., 2004). An empty adenovirus vector (pAd-EGFP) was used as the control. The recombinant adenoviruses were infected at optimal multiplicity of infection (MOI) value into bovine preadipocytes, and all samples were collected 48 h post-infection for quantitative RT-PCR (qRT-PCR) or western blot assay.

Bovine preadipocytes were differentiated as described previously (Wang et al., 2018). Following differentiation for 4 or 8 days, the cells were visually analyzed using BODIPY or Oil red O (ORO) staining of the lipid droplets (LDs), respectively. For the ORO staining, the cells were washed 3 times with PBS and fixed with 4% paraformaldehyde for 30 min. They were stained with ORO solution (0.3% ORO, 60% isopropanol, and 40% PBS) for 30 min in the dark. For the BODIPY staining, the cells were incubated for 30 min after fixation, in a 1:1,000 dilution of 1 mg/ml BODIPY 493/503 (Invitrogen) in PBS at room temperature. We used 4,6-diamidino-2-phenylindole (DAPI) to identify the nuclei. Images were visualized with a Live Cell Imaging System (Nikon Instruments, Europe BV, Kingston, Surrey, England). Lipid accumulation was quantified from the fluorescence intensity parameter of the BODIPY dye (green fluorescence) using ImageJ software.

Total RNA was isolated from tissues and cells using RNAiso regent (Takara, Dalian, China). The qRT-PCR experiments were performed with the ABI7500 Fast Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). Expression of miR-424 was quantified using the miRcute Plus miRNA First-Strand cDNA Kit and miRcute Plus miRNA qPCR Kit (SYBR Green) (Tiangen, Beijing, China). The mRNAs were quantified using PrimeScript^TM RT Reagent Kit with gDNA Eraser and SYBR Premix Ex Taq^TM II Kit (Takara); 2^−∆∆Ct method was used to determine the relative mRNA and miRNA expression levels as described earlier (Livak and Schmittgen, 2001). The genes GAPDH and ACTB (for mRNA) or RPS18 and U6 (for miRNA) were used as internal controls for qRT-PCR. The experiments were performed in three biological and technical replicates for each assay. The specific primer sets used in the qRT-PCR are listed in Table S1 .

Cells were lysed in radioimmunoprecipitation assay (RIPA) lysis buffer supplemented with proteinase inhibitors. An equal amount of total protein for each sample was separated on a 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) following conventional western blotting procedures as previously described (Bass et al., 2017). Primary antibodies against STK11 (1:1,000 (LS−C388827, Lifespan Biosciences, Seattle, USA)), β-actin (1:5,000) (NB100-56874, Novus, Littleton, USA), and horseradish peroxidase (HRP)-conjugated secondary antibody (1:5,000) (D110056, Sangon, Shanghai, China) were used. The protein bands were visualized using Gel DocTM XR+ Gel Documentation System (Bio-Rad, Hercules, CA). Band intensity was quantified using the ImageJ software.

Total RNA was isolated from the bovine perirenal and groin fat tissues and then reverse transcribed using the "hybrid" primer (QT). The first round of amplification was performed using STK11 gene-specific outer primer (GSP1) and a unique oligonucleotide sequence (Qo), and the second round of amplification cycle was carried out using a gene-specific inner primer (GSP2) and "nested" primers (QI) to quench the amplification of nonspecific products. Finally, all the PCR products were cloned into a pMD19-T vector (Takara) and sequenced. The 3' UTR sequences are listed in Figure S1 . All the primers used in this method are listed in Table S1 .

A fragment of 3' UTR from cattle STK11 transcript, containing the predicted miR-424 recognition element (MRE) or the potential RBP binding sites (AREs and GREs), were cloned into a psi-CHECK2 vector. Further, the miR-424 predicted target site (UGCUGCU) was mutated to CAUCAUG. All plasmids were verified by sequencing. Sequence information for the primers is listed in Table S1 .

For the luciferase reporter assay, cells at 60% confluence in 24-well plates were transfected using Lipofectamine 3000 reagent. Co-transfections were performed with 100 ng of empty, or wild type or mutant STK11 3' UTR luciferase plasmids, and 50 nM miR-424 mimics or NC mimics per well (in triplicates). Empty vectors were used as negative controls. After 36 h, the cells were collected, lysed, and assayed for luciferase activity using the Dual-Luciferase Reporter Assay System (Promega, Madison, WI, USA) according to the manufacturer’s instructions. Firefly luciferase activity was used as an internal reference standard.

All data are presented as the mean ± SEM and analyzed with GraphPad Prism 7 software. Two groups were compared using the unpaired two-tailed Student’s t-test. A value of p < 0.05 was considered to be statistically significant.

To screen for novel miRNAs regulating bovine adipogenesis, we re-analyzed the dataset GSE48569 from the GEO database (Sun et al., 2014). The differentially expressed miRNAs (DEMs) in backfat tissues between fetal (GSM1181371) and adult (GSM1181372) stages of Qinchuan cattle were obtained based on the value of transcripts per million (TPM). Compared to the fetal, the adult had 27 upregulated miRNAs and 28 downregulated miRNAs. We focused on miR-424 (unless otherwise stated, miR-424 stands for bta-miR-424-5p in this manuscript) because of its higher differential expression ( Figure 1A ). The results of qRT-PCR assay confirmed that the expression of miR-424 was noticeably increased between 12 and 60 months of age in the backfat tissues of Qinchuan cattle ( Figure 1B ). In addition, the bovine preadipocytes were differentiated into lipid-filled adipocytes in vitro. The qRT-PCR result showed that the expression of miR-424 increased gradually during the differentiation process ( Figure 1C ). Collectively, these results indicate that miR-424 is up-regulated during bovine adipogenesis in vivo and in vitro. Thus, miR-424 appears to be involved in the regulation of bovine adipogenesis and was selected as the candidate miRNA for subsequent experiments.

To understand the regulation and functional activity of miR-424 during adipocyte differentiation, bovine preadipocytes were transfected with either miR-424 mimics, inhibitor, or their respective negative controls. The transfection efficiency was monitored using Cy3-labeled NC mimics and further confirmed by qRT-PCR assay ( Figure 2A ). Next, the cells were induced to differentiate in vitro. On day 3, the mRNA expression of known early markers of differentiation such as CEBP/α, CEBP/β, PPARγ, and FABP4, were analyzed ( Figure 2B ). We found that overexpression of miR-424 markedly increased the expression of CEBP/α, PPARγ, and FABP4, whereas miR-424 inhibitor significantly reduced the expression of CEBP/α and PPARγ. Interestingly, miR-424 failed to influence the expression of CEBP/β. Furthermore, we assessed the cellular accumulation of LDs on day 4 of differentiation by staining with BODIPY 493/503 and found that the fluorescence signal of BODIPY was notably increased in cells treated with miR-424 mimics when compared with NC mimics. Conversely, the accumulation of LDs in cells transfected with miR-424 inhibitor showed the opposite trend ( Figure 2C ). These results demonstrate that miR-424 promotes the differentiation of bovine preadipocytes.

To understand the molecular mechanism through which miR-424 promotes bovine adipogenesis, we used three softwares (TargetScan/Miranda/Mireap) to identify its putative targets that are involved in the regulation of adipocytes differentiation. STK11, a master upstream kinase in the AMPK cascade, was consistently screened out as a potential target of miR-424. Next, we constructed the wild-type and mutant luciferase reporter vectors for STK11-miR-424-MRE and performed dual-luciferase reporter assay. The results showed that miR-424 significantly suppressed the activity of wild-type reporter, while it did not affect that of the mutant reporter ( Figure 3A ). This assay confirmed that miR-424 binds to the 3' UTR of bovine STK11.

To gain insights into the direct effect of miR-424 on STK11 gene expression, we examined the levels of STK11 mRNA and endogenous protein levels 48 h post-transfection. Unexpectedly, as shown in Figure 3B , the STK11 expression level increased following the treatment with miR-424 mimics, while it decreased following the treatment with miR-424 inhibitor. These results suggest that the positive regulatory effect of miR-424 on STK11 is not through the conventional process of promoting mRNA degradation or blocking mRNA translation.

To directly validate the role of STK11 in the pro-adipogenic effect of miR-424 in bovine preadipocytes differentiation, the cells were infected with adenovirus pAd-sh-STK11 and pAd-EGFP, and the green fluorescent protein was analyzed 48 h post-infection to assess the infection efficiency ( Figure 4A ). As expected, specifically knockdown endogenous STK11, using pAd-sh-STK11, significantly decreased the levels of STK11 mRNA and protein expression as compared to the pAd-EGFP ( Figure 4B ). The mRNA levels of the adipogenic markers were measured by qRT-PCR on day 3 of differentiation. We found that these genes (CEBP/α, PPARγ, and FABP4) were downregulated in response to miR-424 overexpression ( Figure 4C ), suggesting that the knockdown of STK11 blocked bovine preadipocytes differentiation. This argument was further supported by the fact that knockdown of STK11 expression led to an impaired accumulation of LDs in the mature adipocytes ( Figure 4D ).

Next, we investigated whether suppressing the expression of STK11 could weaken the effects of miR-424 in preadipocytes differentiation. To address this, we performed a “rescue” assay. As illustrated by the ORO staining results ( Figure 4D ), there was a discernible difference in the accumulation of LDs in the mature adipocytes following the enforced expression of miR-424, while miR-424 could not stimulate adipogenesis in the STK11 knockdown cells. Therefore, we concluded that miR-424 specifically promotes STK11 expression, which in turn contributes (at least in part) to the regulatory role of miR-424 in adipocyte differentiation.

Given the positive effects of miR-424 on STK11 mRNA and protein expression as described above, we reasoned that miR-424 might unconventionally regulate the expression of STK11. Several studies have revealed that the post-transcriptional upregulation by miRNAs is selective, based on the target mRNA specific sequences, and associated with RBPs (Vasudevan et al., 2007). It is not clear whether the observed role of miR-424 in positively mediating STK11 expression is due to direct interaction with specific STK11 mRNA sequence, or indirect interaction through unknown RBPs, or both.

To identify the mechanism involved in the miRNA-mediated gene upregulation, we performed 3′ rapid amplification of complementary DNA (cDNA) ends (3' RACE) assay, both in the bovine perirenal and groin fat tissues, and successfully isolated and sequenced a band of 699 bp which aligned downstream of STK11 coding regions ( Figure 5A ). Interestingly, as indicated in Figure 5B , putative RBP binding sequences (GREs and AREs) were found by scanning the entire length of 3' UTR of STK11. We postulated that the putative RBP binding sequences located in 3' UTR of STK11 might be involved in the post-transcriptional regulation of STK11.

To evaluate this hypothesis, we constructed dual-luciferase reporter plasmids with or without the putative RBP binding sequences (F-ΔMRE or F-MRE) (illustrated in Figure 5C ) followed by co-transfection into bovine preadipocytes with miR-424 inhibitor or NC inhibitor. Reporter assays showed that the luciferase activity of the “F-FL” and “F-ΔMRE” reporters was significantly suppressed when miR-424 was absorbed by a specific inhibitor, but resulted in a modest yet significant increase in luciferase activity of the “F-MRE” ( Figure 5D ). These results indicate that the presence of AU/GU rich sequences in the 3' UTR of STK11 might disrupt the miR-424/STK11 incorporation into the RNA-induced silencing complexes, suggesting that post-transcriptional regulation of STK11 by miR-424 occurs in a potential RBP binding site-dependent manner.