Rosiglitazone (Rosi) was purchased from Cayman Chemical Company. miR-21-5p mimic, inhibitor and negative control (NC) were synthesized by Shanghai GenePharma Co., Ltd. Superoxide dismutase (SOD) assay kit, methane dicarboxylic aldehyde (MDA) assay kit, glutathione peroxidase (GSH) assay kit and oxidized glutathione (GSSG) assay kit were purchased from Beijing Solarbio Science & Technology Co., Ltd. miRNA Detection kit was purchased from GeneCopoeia, Inc. ChIP kit was from Merck KGaA. Dual-luciferase reporter assay system was from Promega Corporation. Human miR-21-5p promoter region or mutations in PPARγ binding sites were synthesized by Sangon Biotech Co., Ltd. The 3′-UTR of SFRP5 mRNA or mutant binding site was synthesized by Sangon Biotech Co., Ltd.

A total of 20 liver tissue samples from patients with NASH and 20 normal individuals were obtained from Banan People's Hospital(Chongqing, China) between April 2020 and November 2020. All procedures were conducted in accordance with the ethical rules stated in the Declaration of Helsinki and were approved by the Ethics Committee of Banan People's Hospital (approval number BNRMYY20200016). A written informed consent was obtained from each subject before their participation. The clinical characteristics of the patients are summarized in Table I.

Liver cancer cell line HepG2 was purchased from China Center for Type Culture Collection and the cell line was authenticated by STR profiling. Cells were cultured in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum (both Gibco; Thermo Fisher Scientific, Inc.), streptomycin (100 mg/ml) and penicillin (100 U/ml) at 37°C in a 5% CO₂ humid incubator. HepG2 cells were treated with 1 mm free fatty acid for 16 h to establish a NASH cell model and then the cells were treated with DMSO, 10 M Rosi or 10 µm Rosi in the presence of miR-21-5p mimic or the combination of miR-21-5p mimic and inhibitor for 24 h, the cells were harvested for further experiment.

Male C57BL/6J mice (six weeks old) were purchased from Beijing Huafukang Bioscience Co., Ltd. and housed in specific-pathogen-free (SPF) laboratory. To induce NASH (14,15), mice were fed with methionine and choline deficient (MCD) diet (Research Diets A02082002B). Normal mice were fed with a control diet (Research Diets A02082003B). After MCD treatment for 5 weeks, mice were randomly divided into three groups (n=5): i) NASH + DMSO: These mice were fed with MCD diet and a treated with vehicle saline containing 0.1% DMSO intraperitoneally daily for 3 weeks; ii) NASH + Rosi: These mice were fed with MCD diet and treated with 10 mg/kg Rosi intraperitoneally daily for 3 weeks; iii) NASH + Rosi + miR-21-5p mimic: These mice were fed with MCD diet and were received intraperitoneal injection of 10 mg/kg Rosi daily and were injected via the tail vein daily with 8 mg/kg miR-21-5p mimic for 3 weeks (16,17). The experiment lasted for 8 weeks in total and mice were euthanized by cervical dislocation following mild (4 ml) diethyl ether (Sigma Aldrich, USA) anesthesia. The liver tissues were removed for hematoxylin and eosin (H&E) staining and the blood was collected from abdominal aorta for biochemical detections. All animal experiments were performed according to Guide for the Care and Use of Laboratory Animals 8th edition (18) and approved by the Animal Care and Use Committee of Banan People's Hospital (approval no. BNRMYY20200029).

HepG2 were seeded into 24-well plates at a density of 1×10⁵ cells/well. The cells were grown to 70% confluence in 24-well plates before transfection at 37°C. Next, cells were transfected with the pmirGLO reporter plasmid (empty plasmid) (Promega Corporation), pmirGLO reporter plasmid containing 3′-UTR of SFRP5 mRNA, miR-21-5p mimic (50 nm; 5′-GUCUCAAUAUUAGGCUAAGCUAU-3′), miR-21-5p inhibitor (50 nm; 5′-AUAGCUUAGCCUAAUAUUGAGAC-3′), mimic negative control (NC; 5′-UUCUCCGAACGUGUCACGUUU-3′) or inhibitor NC (5′-CAGUACUUUUGUGUAGUACAA-3′) (Shanghai GenePharma Co., Ltd), using Lipofectamine^® 2000 (Invitrogen, Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions at room temperature. Subsequently, the transfected cells were cultured for 24 h at 37°C in a 5% CO₂ humid incubator. Then, the luciferase activity was detected according to the instructions of the Dual-Luciferase reporter gene assay kit (Thermo Fisher Scientific, Inc.). Renilla luciferase activity was normalized to firefly luciferase activity and the result was shown as relative luciferase activity. Transfection experiments were performed 3 times in triplicate. Data were represented as the ratios of luciferase activities/Renilla activities.

The HepG2 cells were seeded into 6-well plates, when the confluence of HepG2 cells was 70%, small interfering RNA (siRNA) against PPARγ mRNA (100 nm; cat. no. sc-29455) and NC siRNA (100 nm; scrambled siRNA, sc-37007; both Santa Cruz Biotechnology) were transfected into HepG2 cells using Lipofectamine^® 3000 reagent (Invitrogen; Thermo Fisher Scientific, Inc.) at 37°C according to the instructions of the manufacture. Subsequently, 24 h after cell transfection different treatments and examinations were performed.

Total RNAs were extracted from cells (1×10⁶ cells/well) or liver tissues (100 mg) with TRIzol reagent (Thermo Fisher Scientific, Inc.). Reverse transcription of total RNA was performed using Primescript RT reagent kit (Takara Bio, Inc.) according to the manufacturer's protocol. Subsequently, qPCR was performed with SYBR green qPCR Master Mix (Promega Corporation) according to the manufacturer's protocol using the ABI 7500 system (Thermo Fisher Scientific, Inc.), and the relative mRNA expression levels were normalized to β-actin. miRNAs were extracted with E.Z.N.A. microRNA kit (Omega Bio-Tek, Inc.) and then the expression of miR-21-5p was determined using All-in-One miRNA qRT-PCR Detection kit (GeneCopoeia, Inc.) according to the manufacturer's instructions and the relative expression level was normalized against U6 small nuclear RNA (U6 snRNA). The PCR reaction conditions were as follows: Initial denaturation at 95°C for 2 min, followed by 40 cycles of 95°C for 15 sec and 60°C for 30 sec. The experiments were performed three times. miRNA and mRNA expression levels were quantified using the 2^−ΔΔCq method (19). Primer sequences are listed in Table II.

Whole proteins from cell/tissue samples were lysed with RIPA lysis buffer (Beyotime Institute of Biotechnology). Protein concentrations were determined using the BCA assay(Beyotime Institute of Biotechnology). The proteins (50 µg) were separated by 12% SDS- polyacrylamide gel electrophoresis and transferred onto PVDF membranes (EMD Millipore). Subsequently, the membranes were blocked with 5% bull serum albumin (Beyotime Institute of Biotechnology) at room temperature and then incubated at 4°C overnight with primary antibody for SFRP5 (1:1,000; Abcam; cat. no. ab230425), PPARγ (1:500; Santa Cruz Biotechnology, Inc.; cat. no. sc-271392) or β-actin (1:2,000; Abcam; cat. no. ab8227). Following washing with TBST (10 M Tris, 150 mm NaCl, 0.05% Tween-20), the membranes were incubated with HRP-conjugated secondary antibody (goat anti Rabbit IgG; Abcam; cat. no. ab205718, 1:10,000) and incubated at 37°C for 1 h. Finally, the blots were visualized using the enhanced chemiluminescence detection reagents (Thermo Fisher Scientific, Inc.) The grey value of protein bands was analyzed using ImageJ software 1.46r (National Institutes of Health).

The putative PPARγ response element (PPARE) in miR-21-5p promoter region was predicted by NUBIScan (nubiscan.unibas.ch). The potential binding sites of miR-21-5p in 3′-UTR of human SFRP5 mRNA were predicted using the online databases miRBase (mirbase.org), TargetScan (targetscan.org) and miRanda (microrna.org).

ChIP assays were performed using EZ-ChIP kit (EMD Millipore) according to the manufacturer's protocol. Briefly, HepG2 cells (1×10⁷) were treated with 0.1% DMSO or 5 µm Rosi at 37°C for 24 h and then harvested after fixation with 1% formaldehyde for 10 min at room temperature to cross-link the nuclear proteins to DNA. Then, the cells were lysed in SDS lysis buffer at room temperature for 10 min and the chromatin was sonicated with Ultrasonic Sonicator at 30% of maximum power 10 times for 10-sec pulses on ice, with 2 min interval. to shear the DNA to an average length between 200 and 1,000 bp, followed by immunoprecipitation with the antibody directed against PPARγ (Santa Cruz Biotechnology, Inc.; cat. no. sc-271392×), taking IgG as a negative control. The precipitated DNAs were purified and subjected to PCR amplification that cover the PPARγ response element (PPARE) in miR-21-5p promoter region (−276 to −127), taking ‘input’ (the total DNA extract) as a positive control while ‘no DNA’ as a negative control (20).

The treated cells were fixed with 4% paraformaldehyde for 30 min at room temperature and washed three times with PBS. Next, cells were soaked in 60% Oil Red O stock solution (0.25 g Oil Red O/100 ml isopropanol) diluted by distilled water for 30 min at 25°C. The stained cells were washed using PBS until the background became clear (21). Finally, images were captured under a light microscope (Olympus Corporation) 10 randomly selected fields of view were evaluated at ×200 magnification.

The mouse liver tissues were harvested and gently cut into 0.3×0.5×0.5 cm cubes and fixed with 4% polyoxymethylene at 4°C for 24 h. The fixed liver tissues were dehydrated with gradient ethanol solutions at 25°C. After washing with xylene two times at 25°C, 1.5 h each, the samples were embedded in paraffin. The paraffin blocks were sectioned at 5 µm. Sections from each paraffin block were stained with hematoxylin for 3 min and counterstained with 1% Eosin Y for 10 min at 25°C to examine the pathologic structures of the tissues. Histological steatosis, inflammation and fibrosis were graded according to the NASH activity score (NAS) (22,23) by a certified pathologist. Images were captured by light microscopy (200× magnification; Olympus Corporation); ≥10 randomly selected fields of view were assessed.

Statistical analyses and graphical representations were performed using GraphPad Prism 5.0 software (GraphPad Software Inc). Data are presented as the mean ± standard deviation. Comparisons among multiple groups were analyzed using one-way ANOVA followed by Tukey's post hoc test. Comparisons between two groups were analyzed using a paired t-test. The correlation analysis was performed using Pearson's correlation coefficient. P<0.05 was considered to indicate a statistically significant difference.

It has been reported that PPARγ and miR-21-5p are associated with progression of NASH and have been proposed as potential targets for NASH treatment. Hence, the relationship between PPARγ and miR-21-5p was explored in NASH in mice and humans. As shown Fig. 1A and B, PPARγ was lowly expressed while miR-21-5p was highly expressed in the liver tissue of mouse model of NASH and the human tissue sample from patients with NASH (Fig. 1D and E). Furthermore, there was a significantly inverse correlation between the level of PPARγ and miR-21-5p in NASH in both mice and humans (Fig. 1C and F).

To investigate whether PPARγ is involved in the regulation of miR-21-5p, HepG2 cells were treated with the PPARγ agonist Rosi and the expression of miR-21-5p was detected. As shown in Fig. 2A, activation of PPARγ with Rosi decreased miR-21-5p expression in a dose-dependent manner. Knockdown of PPARγ by siRNA (cat. no. sc-29455; Santa Cruz Biotechnology, Inc.), in which the silencing effect was high (Fig. 2B), significantly attenuated the Rosi-mediated downregulation of miR-21-5p (Fig. 2C). It is well established that lipid droplet accumulation, hepatic inflammation and oxidative stress serve important roles in the progression of NASH (24–26). Therefore, the levels of lipogenesis, inflammatory cytokines and oxidative stress were evaluated in an in vitro model of NASH, in which the cells were treated with PPARγ agonist Rosi and miR-21-5p mimic or miR-21-5p inhibitor. As shown in Fig. 2D, the number of lipid particles was markedly decreased in cells treated with Rosi, in which PPARγ was activated, relative to the cells treated with DMSO. However, the number of lipid particles was increased in the cells treated with miR-21-5p mimic, which the miR-21-5p was overexpressed after transfected with miR-21-5p mimic (Fig. 2E). In addition, the expression levels of inflammatory markers TNF-α, IL-6 and monocyte chemotactic protein (MCP)-1 (Fig. 2F) and oxidative stress (expression levels of oxidative stress markers including MDA, SOD and GSH/GSSG ratio; Fig. 2G-I) in NASH cells was inhibited by activation of PPARγ, but the inhibitory effect was attenuated by miR-21-5p mimic (Fig. 2F-I). Inhibition of miR-21-5p by miR-21-5p inhibitor significantly enhanced the PPARγ-mediated inhibition of lipogenesis (Fig. 2D), inflammation (Fig. 2F) and oxidative stress in NASH cells (Fig. 2G-I). Taken together, these results indicated that PPARγ repressed progression of NASH via downregulation of miR-21-5p.

As a transcription factor, PPARγ usually regulates gene transcription by binding to the PPARE in the promoter region of target gene (27). The present study found a putative PPARE ER4 (TGTCCTAATAAGGACT, −188 to −173) in miR-21-5p promoter region by online bioinformatics analysis (NUBIScan; Fig. 3A). To investigate the mechanism by which PPARγ downregulates miR-21-5p, the luciferase reporter plasmids containing wild-type (pGL3-PPARE-WT) or mutant PPARγ (pGL3-PPARE-Mut) binding sites were constructed (Fig. 3A) and the luciferase reporter assays were performed. As shown in Fig. 3B, activation of PPARγ significantly decreased the luciferase activity of construct pGL3-PPARE-WT, whereas the luciferase activity of construct pGL3-PPARE-Mut was not affected, indicating that PPARγ suppressed the transcriptional activity of miR-21-5p promoter. In addition, the ChIP assay demonstrated that PPARγ could directly bind to PPARE ER4 of the miR-21-5p promoter region (Fig. 3C). These results reveal that PPARγ downregulated miR-21-5p by binding to its promoter region.

miRNAs regulate gene expression by pairing interactions with mRNAs, which can lead to degradation or translation repression of target mRNAs (28). The online databases (TargetScan, miRBase, miRanda) predicted SFRP5, an anti-inflammatory adipokine that regulates progression of NASH, as a potential target gene of miR-21-5p (Fig. 4A). To investigate whether miR-21-5p is involved in the regulation of SFRP5, HepG2 cells were treated with miR-21-5p mimic or inhibitor. Treatment with miR-21-5p mimic significantly reduced the mRNA and protein levels of SFRP5, which was clearly attenuated by miR-21-5p inhibitor (Fig. 4B and C). In addition, SFRP5 was lowly expressed while miR-21-5p highly expressed in human liver tissue samples from NASH patients and there was a significant inverse correlation between SFRP5 and miR-21-5p levels (Fig 4D). miR-21-5p mimic notably decreased the luciferase activity of pmir-SFRP5-UTR-WT (the luciferase reporter plasmid containing the predicted miR-21-5p wild-type binding sites but not the binding site mutant 3′-UTR), which was alleviated by miR-21-5p inhibitor (Fig. 4E). These results suggested that SFRP5 is one of the target genes of miR-21-5p. It was found that activation of PPARγ enhanced the expression of SFRP5 and overexpression of miR-21-5p markedly inhibited the PPARγ-induced SFRP5 expression (Fig. 4F and G). These results indicated that activation of PPARγ suppresses expression of miR-21-5p and upregulates its target gene SFRP5.

NASH was induced in C57Bl/6 mice by the MCD diet and this NASH model was used for evaluating the effects of the PPARγ/miR-21-5p/SFRP5 pathway on NASH progression. Hepatic histopathology analysis with H&E staining demonstrated that various degrees of liver steatosis and numerous lipid droplets of different sizes were observed in MCD-fed mice. However, these pathological conditions were significantly alleviated by Rosi treatment. In addition, treatment of miR-21-5p mimic, which the miR-21-5p was overexpressed in the liver by caudal vein injection (Fig. 5A), clearly reduced the PPARγ-mediated inhibition of liver steatosis and lipid accumulation in NASH model (Fig. 5B). In addition, the levels of inflammatory cytokines and oxidative related genes were tested in the liver tissues of mice with MCD-induced NASH. As shown in Fig. 5C, activation of PPARγ dramatically suppressed the expression of proinflammatory cytokines such as TNF-α, IL-6 and MCP-1. Similar trends were observed in the levels of oxidative stress-related genes including MDA, SOD and GSH/GSSG ratio in the liver tissue of mice with MCD-induced NASH (Fig. 5D-F). However, overexpression of miR-21-5p markedly suppressed PPARγ-mediated inhibition of hepatic inflammation and oxidative stress (Fig. 5D-F). In addition, activation of PPARγ markedly decreased miR-21-5p while increased SFRP5 in MCD-fed mice (Fig. 5G and H). These results imply that activation of PPARγ inhibited the progression of NASH by manipulating the miR-21-5p/SFRP5 axis in vivo.