Male ZDF rats (Fa/Fa, n = 30) and control lean rats (+/Fa, n = 7) were obtained from the Laboratory Animal Center of Vital River (Beijing, China; license number, SYXK (Yue) 2011-0074). The average weight of ZDF and lean rats at 8 weeks of age was 286 and 233 g, respectively. Rats were housed in an identical room with free access to a high-fat diet and purified water throughout the experiment. A high-fat diet (Purina 5008, PA) was provided from 8 weeks of age to the end of the study. After four weeks of induction, the glucose levels of the rats ranged between 19.3.3 and 33.3 mmol/L (average 26.53 ± 0.69 mmol/L). At of 12 weeks age, the ZDF rats were randomly divided into four groups: 1) diabetic control group (PC, n = 7); 2) diabetic group treated with insulin glargine (INS, n = 7); 3) diabetic group treated with the saxagliptin inhibitor dipeptidyl peptidase 4 (DPP-4) (Saxag, n = 7); and 4) diabetic group treated with the GLP-1 analogue liraglutide (Lirag, n = 9). The normal control group (NC, n = 7) and PC group were treated with 0.9% normal saline. Insulin glargine was intravenously injected into the INS group once a day and the dosage was set according to the changes in glucose levels of the Lirag group throughout the experiment. The Saxag group received saxagliptin (AstraZeneca, London, England) by intragastric gavage at a dosage of 5 mg/kg/24 h. The Lirag group was administered liraglutide (Novo Nordisk, Copenhagen, Denmark) at a dose of 200 μg/kg/12 h for 8 weeks, as previously described (Hayes et al., 2011). Glucose levels, body weight and food intake were measured every week. At the age of 20 weeks, the rats were anesthetized with 2.5% pentobarbital sodium and the left ventricular was punctured for blood sampling. The livers of rats were fixed in 4% paraformaldehyde or frozen in liquid nitrogen. All animal experiments were approved by the Institutional Animal Care and Use Committee of the Southern Medical University (Guangzhou, China).

Normal hepatocytes LO2 and hepatoma HepG2 cells were obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). Exendin-4 was purchased from MedChemExpress (United States). Exendin Fragment 9-39 (Exendin-9) was obtained from Sigma-Aldrich (United States). All the cells were cultured in DMEM (Invitrogen; Invitrogen; Paisley, United Kingdom) with 10% fetal bovine serum (FBS; Gibco-BRL, Invitrogen; Paisley, United Kingdom) in a 5% CO2 incubator at 37°C. ShPPARα (targeting 5′-GAC​TCA​AGC​TGG​TGT​ATG​A-3′) was purchased from Sangon Biotech (Shanghai, China). Exponential growth phase cells were transfected with 2.5 μg shPPARα in reduced serum medium (OPTI-MEM-I) according to the manufacturer’s protocol at 50–70% confluence. To build the hepatocyte steatosis model, the cells were exposed to sodium palmitate (PA) at a final concentration of 0.3 μM. The cells were divided into the following groups: 1) normal control group (NC): the cells were cultured in DMEM for 24 h; 2) PA group: the cells were exposed to PA (0.3 μM) in DMEM for 24 h; 3) PA + Exendin-4 group (PA + Ex-4): the cells were co-treated with PA (0.3 μM) and Exendin-4 (100 nM) in DMEM for 24 h; 4) PA + Exendin-4+Exendin-9 group (PA + Ex-4+Ex-9): the cells were cultured in PA (0.3 μM) and Exendin-9 (100 nM) for 1 h and were then co-administered with Exendin-4 (100 nM) for 24 h. 5) Insulin group (Insulin): the cells were exposed to Insulin (100 nM) in DMEM for 24 h; 6) PA + Insulin group (PA + Insulin): the cells were co-treated with PA (0.3 μM) and Insulin (100 nM) in DMEM for 24 h; 7) PA + Exendin-4+Insulin group (PA + Ex-4+Insulin): the cells were co-treated with PA (0.3 μM), Exendin-4 (100 nM) and Insulin (100 nM) in DMEM for 24 h; 8) sh PPARα group: the cells transfected with shRNA duplexes targeting PPARα were cultured in DMEM for 48 h; 9) PA + Exendin-4+shPPARα group (PA + Ex-4+shPPARα): the cells transfected with shRNA duplexes targeting PPARα were cultured in DMEM for 48 h and were then co-administered with PA (0.3 μM) and Exendin-4 (100 nM) for 24 h.

After overnight fasting, blood samples were collected from left ventricular puncturing and subsequently centrifuged at 3,000 rpm for 10 min after standing at room temperature for 3 h. Total cholesterol (mM), triglycerides (mM), HDL cholesterol (mM), LDL cholesterol (mM), AST (IU/L) and ALT (IU/L) were determined by an automated biochemistry analyzer (Cobas Integra 400 Plus; Roche Diagnostics, Basel, Switzerland).

Oral glucose tolerance test was conducted on rats at the age of 20 weeks after an overnight fasting. After testing fasting blood glucose (FBG), rats were intragastrically given glucose solution at 2 g/kg body weight. The blood samples were collected from the rats’ tail vein and the plasma glucose was measured by a glucose meter 0, 30, 60, and 120 min after glucose administration. The area under the curve of glucose was calculated using the trapezoidal method (AUC = 1/4 fasting glucose +1/2 30 min glucose +3/4 60 min glucose +1/2 120 min glucose).

Liver tissues were fixed in 4% paraformaldehyde for 24 h and maintained in 70% alcohol for subsequent paraffin embedding. Paraffin sections were cut at a thickness of 3 μm and hematoxylin-eosin (HE) staining was performed after removal of paraffin. The hepatic steatosis of each section was observed by optical microscope using a model BX53 instrument (Olympus, Tokyo, Japan). IHC was performed to investigate the expression of proteins in human hepatic tissues. The sections were incubated overnight with 1:100 dilutions of primary antibodies against PPARα (bs-3614R; Bioss Antibodies, Boston, MA, United States) and GLP-1R (bs-1559R; Bioss) overnight at 4°C. This was followed by incubation with goat anti-rabbit secondary antibody conjugated with 3,3′-diaminobenzidine (DAB) chromogen. Samples were examined by optical microscopy. The slides were reviewed concerning low or high expression in a blinded manner by two pathologists.

LO2 cells were stained with Oil red O (ORO, Solarbio, Beijing, China) solution to visualize the accumulation of intracellular lipids. In brief, LO2 cells were washed with PBS for three times before fixing with 10% formalin for 20 min. After washing with 60% isopropanol for 5 min, the cells were stained with ORO solution for 20 min at room temperature. Then, the cells were washed by distilled water to remove the excess dyestuff and counterstained with hematoxylin. The red-stained lipid droplets were observed and photographed using a light microscope.

Total RNA was extracted using Trizol reagent (Invitrogen Life Technologies, Carlsbad, CA, United States) according to the manufacturer’s instructions. The total RNA was transcribed into cDNA using PrimeScript RT reagent Kit with gDNA Eraser (DRR047A, TaKaRa). Real-time quantitative PCR (qPCR) was carried out using SYBR Green PCR master mix (Applied Biosystems; Foster City, CA) on ABI 7500HT system. GAPDH was chosen as endogenous control. All the primers shown in Table 1 were synthesized from Sangon Biotech (Shanghai, China). The expression level of each targeted gene was normalized as fold change compared to control or reference group. Fold changes were calculated through relative quantification (2^−ΔΔCT).

Western blot analysis was conducted to investigate the expression of proteins and their phosphorylation level in hepatic tissues and cells. Hepatic tissues were ground into powders with liquid nitrogen and hepatic cells were homogenized in RIPA lysis buffer (1% leagene PMSF, 1% protease and Phosphatase Inhibitor Cocktail). Proteins extracted from each group were separated by SDS-PAGE and transferred to PVDF membranes. Nonspecific binding was blocked using a blocking reagent (0.1% Tween 20 and 5% Bovine Serum Albumin in TBS) and the membranes were then incubated with specific primary antibodies overnight at 4°C. Primary antibodies (1:1000) used in this manuscript are: anti-AMPK alpha 1 and anti-p-AMPK alpha 2 (S345) obtained from Biohua (Hangzhou, China); anti-GAPDH purchased from Santa Cruz (Biotechnology, Santa Cruz, CA, United States); anti-PPARα obtained from Proteintech (Chicago, CA, United States), anti-ACC was obtained from Abcam (Cambridge, England); anti-GLP-1R (Beijing, China, Bioss), and anti-p-ACC purchased from Affinity Biosciences (OH, United States). After incubating the membrane with the appropriate secondary antibodies for 1 h at room temperature, Perkinelmer western lightning Plus-ECL was used to detect the specific bands. The immunoblots were quantified by densitometric analysis using Quantity One software (Bio-Rad, Hercules, United States).

Intracellular accumulation of ROS was measured using a Reactive Oxygen Species assay kit (Beyotime Institute of Biotechnology) containing fluorescent probe, 2′,7′-dichloro-dihydro-fluorescein diacetate (DCFH-DA). In brief, cells were seeded in serum-free medium containing 10 μM DCFH-DA (v/v, 1:1,000) and incubated at 37°C for 30 min. The residual DCFH-DA solution was removed and cells were washed with serum-free medium three times for 5 min, and then washing with PBS. Cells were observed under a fluorescence microscope (Olympus Corporation). Moreover, intracellular fluorescence intensity were analyzed by flow cytometry (FACSCalibur, Becton–Dickinson, United States) in PBS dissolved 10,000 cells of each sample. The total intracellular ROS level is calculated as the percentage of control cells and is directly proportional to the fluorescence intensity.

Immunofluorescence staining was performed according to the standard protocol described previously (Liao et al., 2017). Polyclonal rabbit primary antibody to GLP-1R (Bioss, bs-1559R; 1:100), HRP-conjugated secondary antibody and DAB staining kit (CWBIO, Beijing, China) were used in the experiment. Images were captured with a laser confocal microscope (FV10-ASW, Olympus, Tokyo, Japan).

Propidium iodide (PI) and Annexin V-FITC-flow cytometry assay (BD Pharmingen, CA) was used to detect the apoptosis in the cells. Cells were stained with FITC-conjugated annexin V and PI (KeyGEN BioTECH, China, KGA107) according to the manufacturer’s instructions. After staining, the cells were analyzed by flow cytometry (FACS Calibar; Becton-Dickinson) using Cell Quest software.

TUNEL assay was processed using the in situ cell death detection kit (KeyGEN BioTECH, China, KGA7071). Hepatic tissues were stained according to the manufacturer’s instructions. The apoptosis index was captured using a fluorescence microscope (Olympus IX73, Japan).

Data were analyzed using SPSS version19.0 (SPSS, Chicago, IL, United States). The Student’s t-test and one-way ANOVA test were carried out for RT-PCR. Statistical significance was set at p < 0.05.

To confirm the effects of GLP-1 on glycaemic control, both liraglutide and saxagliptin (dipeptidyl peptidase 4 inhibitor, which can increase the secretion of endogenous GLP-1), were used to treat ZDF rats. The food intake of both compounds resulted in decreased food intake at most time points, especially 4 weeks after antidiabetic drug treatment (Figure 1A, first panel). At the end of the treatment, the cumulative food intake of ZDF rats treated with liraglutide and saxagliptin had increased to a level similar to that of the positive control group. However, the food intake of the liraglutide group remained lower than that of the other groups.

The body weights of ZDF rats were higher than those of lean rats at all time points. Insulin significantly and continuously produced an increased weight gain of the ZDF rats. Notably, liraglutide decreased the weight of ZDF rats, especially after 9 weeks of treatment (Figure 1A, second panel). However, a similar decrease in body weight was not detected in ZDF rats treated with saxagliptin.

As expected, blood glucose levels increased significantly in ZDF rats compared with lean rats. Insulin, liraglutide and saxagliptin dramatically alleviated the blood glucose level in ZDF rats (Figure 1A, third panel). The oral glucose tolerance test analysis showed that lean rats were tolerant to glucose, whereas ZDF rats had an impaired tolerance to glucose (Figure 1D). Although the fasting glucose levels in all ZDF rats subjected to different treatments were the same, liraglutide-treated rats showed a significant decrease the glucose level at each time point after oral glucose intake (Figure 1A, fourth panel). Saxagliptin also improved glucose tolerance in ZDF rats.

We also examined serum lipid profiles that included total cholesterol (TC), TG, LDL-cholesterol (LDL-C), and HDL-C in ZDF and lean rats, including. Administration of liraglutide and saxagliptin decreased the levels of TC, TG, and LDL-C in ZDF rats. The effects of liraglutide were more dramatic than those of saxagliptin (Figure 1B). Both liraglutide and saxagliptin slightly decreased HDL-C levels (Figure 1B, third panel). Conversely, insulin dramatically increased the levels of LDL-C and HDL-C in ZDF rats. The livers of ZDF and lean rats were dissected and observed by optical microscopy after HE staining. Compared with lean rats, ZDF rats exhibited severe hepatic steatosis. To our surprise, big and confluent regions of lipid were detected in the liver of insulin treated ZDF mice. In these mice, hepatic steatosis was more severe compared to control ZDF rats. Administration of liraglutide and saxagliptin completely prevented the establishment of hepatic steatosis (Figure 1C). The GLP-agonist Ex-4 significantly alleviated PA induced lipid accumulation in LO2 cells (Figure 1D). Similar to in vivo experiments, insulin aggravated PA induced lipid accumulation, while Ex-4 dramatically relieved the insulin aggravated lipid accumulation.

The levels of aspartate aminotransferase (AST) and alanine aminotransferase (ALT) levels were detected in the blood serum of ZDF and lean rats to study liver injury. Liraglutide treatment produced decreased levels of AST and ALT levels in ZDF rats, while only AST levels were significantly downregulated in the insulin treated group (Figure 2A). The terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay was also performed on hepatic tissues of ZDF and lean rats. Compared with lean rats, the apoptosis of liver cells was dramatically increased in ZDF rats. Administration of liraglutide completely abolished the apoptosis of liver cells in ZDF rats (Figure 2B). Flow cytometry assays indicated that incubation in the presence of PA dramatically increased the apoptosis of LO2 cells, while Ex-4 significantly alleviated PA induced apoptosis (Figure 2C). Although insulin did not influence the apoptosis of liver cells, it aggravated PA induced hepatic cell apoptosis.

Increased oxidative stress has been identified as a major cause of liver injury (Lirussi et al., 2007). A reactive oxygen species (ROS) assay kit was used to detect the ROS level in LO2 cells with different treatments. PA treated hepatic cells displayed more intense DCFH-DA signals compared with the control group when examined by fluorescence microscopy. Liraglutide dramatically reduced the DCFH-DA signals in the PA treated hepatic cells (Figure 2D). Flow cytometry indicated that Ex-4 remarkably reduced the number of DCFH-DA-positive cells in PA treated hepatic cells. Insulin dramatically increased oxidative stress (Figure 2E).

PPARα is a nuclear transcription factor that can activate the expression of enzymes related to fat metabolism (Pawlak et al., 2015). To confirm the involvement of PPARα in GLP-1 induced metabolism, the expression of PPARα and its target genes were detected in ZDF and lean rats. The RNA expression of PPARα was dramatically decreased in the liver of ZDF rats compared with that in lean rats. Although the expression of PPARα was increased in the liver of insulin- and liraglutide-treated ZDF rats at the transcriptional level, PPARα protein was only overexpressed in the liraglutide group (Figure 3A). This was confirmed by IHC data. Compared with lean rats, stronger PPARα staining was detected in liver sections of liraglutide-treated ZDF rats (Figure 3B). The expression of acetyl CoA carboxylase (ACC) and carnitine palmitoyl transferase la (CPT1a), two PPARα targeting genes related to fat metabolism, were studied in both hepatic tissues and cell lines. The liraglutide group showed decreased hepatic expression of ACC and p-ACC, but increased expression of CPT1a. In contrast, insulin stimulated the hepatic expression of ACC and p-ACC, but inhibited that of CPT1a (Figure 3C,D). In PA treated LO2 and HepG2 cells, administration of Ex-4 dramatically stimulated the expression of PPARα and its target gene CPT1a, and inhibited the expression of ACC (Figure 3E,F). The GLP-1 antagonist Ex-9 reversed the effects of Ex-4 on the expression of ACC and CPT1a.

To further confirm the mediating role of PPARα in GLP-1 stimulated hepatic lipid metabolism, we silenced PPARα using short hairpin (sh)PPARα in LO2 and HepG2 cells. Administration of shPPARα dramatically inhibited the expression of PPARα protein in both cell lines (Figure 4A). Silencing of PPARα removed the inhibitory effect of GLP-1 on lipid accumulation in hepatic cells (Figure 4B, upper panel). Silencing of PPARα also aggravated insulin stimulated lipid accumulation in hepatic cells (Figure 4B, lower panel). Moreover, the basal expression of ACC was increased and CPT1a was decreased in the two hepatic cell lines when PPARα was silenced. Knockdown of PPARα suppressed the Ex-4 stimulated expression of CPT1a, while stimulation of Ex-4 inhibited the expression of ACC in hepatic cells (Figure 4C).

As detected by the ROS assay, silencing of PPARα reversed the Ex-4 suppressed oxidative stress, and the number of DCFH-DA-positive cells in the shPPARα group was even larger than that in the PA group (Figure 4D). Flow cytometry revealed that silencing of PPARα reversed the Ex-4 rescued apoptosis of hepatic cells (Figure 4E).

GLP-1 receptor expression was confirmed in both hepatic tissues and cells. Among the ZDF rats, the liraglutide group showed the highest expression of GLP-1R (Figure 5A). IHC staining of GLP-1R showed the same trends of expression in lean rats and ZDF rats with different treatments (Figure 5B). High GLP-1R expression was also detected in LO2 and HepG2 cells detected by western blot and immunofluorescence staining (Figure 5C). Ex-4 dramatically stimulated the expression of GLP-1R in hepatic cells, while Ex-9 totally abolished the Ex-4 induced GLP-1R increase (Figure 5D). The PA diet suppressed the expression of GLP-1R. Interestingly, insulin significantly inhibited the expression of GLP-1R in a dose-dependent manner (Figure 5E). H89, an inhibitor of protein kinase A, blocked the basal and Ex-4 stimulated expression of PPARα (Figure 5F). The AMPK pathway is involved in GLP-1R mediated downstream signaling both in vivo and in vitro. AMPK signaling was active in the hepatic tissues of liraglutide-treated ZDF rats compared with lean rats and other antidiabetic drug treated ZDF rats (Figure 5G, upper panel). Consistent with this, Ex-9 completely blocked the Ex-4 activated AMPK signaling (Figure 5G, lower panel).