Dimethyl sulfoxide (DMSO, D2438), berberine (BBR, purity over 98%, B3251), oleic acid (OA, purity over 99%, O1383), bovine serum albumin (BSA, B2064), sodium L-lactate (L7022), and sodium pyruvate (P8574) were purchased from the Sigma-Aldrich Co. LLC (St. Louis, MO). Fetal bovine serum (FBS, #10100154), minimum essential medium (MEM, #11095080), protease inhibitors (#78442), and kits for cell protein extraction (#89900) and quantification (#23225) were purchased from Thermo Fisher Scientific (China) Inc (Shanghai, China). The rabbit or mouse monoclonal antibodies against AMPKα (#2793), phosphorylated-AMPKα (p-AMPKα) (Thr172) (#2535), AMPKα1 (#2795), p-AMPKα1 (Ser485) (#2537), AMPKα2 (#2757), p-ACC (#3661), P-Raptor (#2083), and β-actin (ACTB) (#4970) were purchased from Cell Signaling Technology, Inc (Danvers, MA). The antibody against CD36 (ab133625) was purchased from Abcam (Cambridge, MA). The AMPKα1 CRISPR/Cas9 KO Plasmid (sc-400104), AMPK 1 HDR (homology-directed repair) Plasmid (sc-400104-HDR), and puromycin dihydrochloride (sc-108071) were purchased from Santa Cruz Biotechnology, Inc (Dallas, TX). The Lipofectamine^™ 3000 Transfection Kit (L3000150) and the Amplex^™ Red Glucose/Glucose Oxidase Assay Kit (A22189) were purchased from Invitrogen (Grand Island, NY). The Cyclex^® AMPK Kinase Assay Kit (cy-1182) was purchased from Cyclex Co. Ltd. (Nagano, Japan). The commercially available kits for the assay of L-lactate (GX8127T) and glucose (GS121T) were purchased from Beijing Strong Biotechnologies, Inc (Beijing, China). The Steatosis Colorimetric Assay Kit (10012643) was purchased from Cayman Chemical (Ann Arbor, MI) and the Tissue/Cell Triglyceride (TG) Assay Kit (E1013) was from Applygen Technologies Inc. (Beijing, China). Reagents and kits for cellular RNA extraction (#9109), reverse transcription (RT) (RR037A), and quantitative real-time PCR (RR820A) were purchased from Takara Bio Inc (Shiga, Japan).

HepG2 cells were routinely cultured in MEM + 10% FBS at 37°C in a 5% CO₂ incubator. Before transfection, cells were trypsinized with 0.25% trypsin-ethylenediaminetetraacetic acid (EDTA) and seeded onto six-well plates with 1×10⁶/well. When the cells reached about 70–80% confluence, the transfection experiment was performed. Briefly, for one well of the six-well plate, 7.5 μl of the Lipofectamine^™ 3000 reagent was added to 125 μl of serum-free MEM and mixed well. Meanwhile, 1 μg of the AMPKα1 CRISPR/Cas9 KO Plasmid and 1 μg of the AMPKα1 HDR Plasmid were diluted with 125 μl of serum-free MEM, then, 4 μl of the P3000^™ reagent was added and mixed well. The diluted plasmids were then added to the diluted transfection reagent at 1:1 ratio and mixed well. After incubation for 5 min at room temperature, the master mix (250 μl) was added to cells. CRISPR/Cas9 plasmid induced double-strand breaks (DSB) of target gene (AMPKα1). Target-specific HDR Plasmid provided a DNA repair template for a DSB and enabled the insertion of a Red Fluorescent Protein (RFP) gene to visually confirm transfection and an antibiotic resistance gene (puromycin) for selection of cells containing a successful CRISPR/Cas9 double-strand break.

After 48 h of transfection, cells were trypsinized, diluted, and cultured in 96-well plate, and stable monoclonal transfectants were selected in medium containing 0.5 μg/ml of puromycin for about 1 month. DNA transfection was confirmed by visualization of red fluorescence under an Olympus IX71 inverted fluorescence microscope (Olympus Corporation, Tokyo, Japan). And the disruption of target gene was confirmed by DNA sequencing (Invitrogen) and western blot.

The HepG2 AMPKα1^-/- cells were maintained in MEM + 10% FBS containing 2 μg/ml puromycin. Together with wild type, the cells were seeded onto 6-well plates, 24-well plates (4×10⁵/well), or 96-well plates (5×10⁴/well) before experiments and starved in serum-free MEM for 24 h before treatment. BBR (dissolved as 10 mM in DMSO) was used to treat the cells at indicated concentrations for 24 h in serum-free MEM. In some cases, OA (0.6 mM) dissolved in phosphate buffered saline (PBS) plus 0.5% BSA was also used to treat the cells, either alone or in combination with BBR.

After treatment, cells cultured in six-well plates were rinsed with PBS and total proteins were extracted as described before (Zhang et al., 2018). For each sample, about 15 μg of proteins were subjected to 10% SDS-PAGE, and protein bands were transferred onto PVDF membranes (Millipore, Billerica, MA) through a Wet Transfer Cell (Bio-Rad, Hercules, CA). For the detection of target protein expression levels, membranes were probed with specific rabbit or mouse monoclonal antibodies with ACTB as an internal control. After washing and incubation with appropriate secondary antibodies, the bands were visualized with a Chemi-Luminescent Horseradish Peroxidase (HRP) Kit (Millipore). The signal intensities were quantified with Gel-Pro Analyzer 4.0 software (Media Cybernetics, Inc, Rockville, MD).

After treatment, cells were lysed and AMPK activities were measured according to the protocol of the kit. Briefly, samples were added to a plate which was coated with an AMPK-substrate, and the reactions were started by adding of Mg²⁺ and ATP. After proper incubation and washing, a monoclonal antibody specific for the phosphorylated form of the substrate was added. After another round of incubation and washing, a HRP-conjugated secondary antibody was added to the well. The color was developed by a chromogenic substrate and the signals were measured densitometrically at 450 nm. The AMPK inhibitor compound C was used as an “inhibitor control.” After subtracting OD450 of compound C-treated parallel samples, the OD450 values of the samples were used as the “relative AMPK activity” and were presented as percentage of control cells.

The cells were seeded onto 96-well plates and there were five to six replicate wells for each treatment. After treatment, culture mediums were collected and centrifuged at 1,000 rpm for 5 min. Glucose levels were assayed by a kit which was based on the glucose oxidase method, and glucose consumption was calculated as described in our previous report (Shan et al., 2013). Meanwhile, the concentration of L-lactate in the supernatant was also determined by a commercial kit according to the protocol.

The cells were seeded onto 24-well plates and there were four replicate wells for each treatment. After treatment, cells were washed with PBS and glucose production medium which contained 20 mM of sodium L-lactate and 2 mM of sodium pyruvate was added (100 μl/well). After 3.5 h of incubation, glucose levels were determined by a kit, normalized to protein concentrations, and presented as fold of vehicle-treated wild type cells, which was designated as 1 (Zhang et al., 2012).

The cells were seeded onto 24-well plates with three replicate wells for each treatment; after treatment, cells were subjected to ORO staining as described before (Wang et al., 2016). After photograph, the wells were let dry, ORO was extracted by the Lipid Droplets Assay Dye Extraction Solution and the absorbance values at 490 nm (OD490) were read by a plate reader (PerkinElmer, Inc, Waltham, MA). In parallel experiments, intracellular TG levels were determined by a kit and normalized to protein concentrations.

The cells were seeded onto six-well plates, after treatment, total cellular RNAs were isolated and 1 μg was used as a template to reverse transcribed into cDNAs in a 20 μl of reaction system as described previously (Shan et al., 2013). For quantitative PCR, a 50 μl of reaction system was set up which contained 25 μl of SYBR Premix Ex Taq II (2×), 2 μl of forward primer (Table 1), 2 μl of reverse primer (Table 1), 1 μl of ROX Reference Dye (50×), 4 μl of cDNA, and 16 μl of water. The reactions were performed on an Applied Biosystems 7500 Fast Real-Time PCR System (Foster City, CA) and the conditions were as following: 95°C, 30 sec, followed by 40 cycles of 95°C, 5 sec + 60°C, 30 sec. Relative quantification of the levels of target genes (Table 1) was conducted by using ACTB as an internal control.

After validation of the test for homogeneity of variance (Bartlett’s test), results were examined by one-way ANOVA followed by the Newman-Keuls test for multiple comparisons. P < 0.05 was considered to be statistically significant.

In order to clarify whether or not AMPK is indispensable for BBR to stimulate glucose metabolism, AMPKα1^-/- HepG2 cells were generated by disruption of the AMPKα1 gene and stable monoclonal transfectants were selected. As shown in Figure 1A, while wild type HepG2 cells had no fluorescence, the AMPKα1^-/- HepG2 cells presented a strong red fluorescence due to co-transfection of an HDR plasmid, which verified the success of DNA transfection.

The disruption of AMPKα1 gene was verified by DNA sequencing (data not shown) and western blot analysis of target proteins. As shown in Figure 1B, the protein of AMPKα1 completely disappeared in the AMPKα1^-/- HepG2 cells. In wild type HepG2 cells, BBR had no influence on the level of AMPKα1, but stimulated its phosphorylation at Ser485 significantly (p < 0.05 vs. DMSO). In AMPKα1^-/- HepG2 cells, the phosphorylation of AMPKα1 protein was not observed and the activity of BBR was completely abolished (p < 0.001 vs. wild type).

When total AMPKα was detected, the result showed that its protein level declined to about 45–50% of wild type (p < 0.01) in AMPKα1^-/- HepG2 cells (Figure 1B). BBR had no influence on the protein level of total AMPKα in wild type or AMPKα1^-/- HepG2 cells, but increased its phosphorylation at Thr172 significantly in wild type cells (p < 0.05 or p < 0.01 vs. DMSO). In AMPKα1^-/- HepG2 cells, the phosphorylation of total AMPKα declined to less than 10% of wild type cells (p < 0.001) and the stimulating activity of BBR on AMPKα phosphorylation was also abolished (p < 0.001 vs. wild type) (Figure 1B). Consistent with this, BBR increased the expression of P-ACC and P-Raptor compared to the vehicle group in wild-type but not AMPKα1^-/- HepG2 cells (Figure 1B).

The AMPK activity assay was used to verify the effects of BBR and results were in agreement with those of western blot. As shown in Figure 2, BBR greatly induced AMPK activity in wild type HepG2 cells (p < 0.05 or p < 0.01 vs. DMSO), and the activity was completely blocked by AMPKα1 gene KO (p < 0.001 vs. wild type). Taken together, the above results proved that AMPKα1 gene was successfully knocked-out in HepG2 cells and the stimulating activity of BBR was blocked.

Figure 3A showed that BBR greatly enhanced glucose consumption in wild type HepG2 cells after 24 h of treatment (p < 0.05 or p < 0.01 vs. DMSO). However, in AMPKα1^-/- HepG2 cells, the stimulating effect of BBR on cellular glucose consumption was completely abolished (p < 0.05 or p < 0.01 vs. wild type). As BBR was reported to stimulate glucose consumption through induction of glycolysis (Yin et al., 2008), we determined the concentrations of L-lactate in the culture supernatant after treatment. As shown in Figure 3B, the stimulating effect of BBR on lactate release in wild type HepG2 cells was completely blocked when AMPKα1 gene was knocked-out (p < 0.01 vs. wild type).

The activation of AMPK was able to suppress gluconeogenesis in hepatocytes (Yadav et al., 2017), therefore, we determined glucose production after BBR treatment. As shown in Figure 4, in wild type HepG2 cells, BBR reduced glucose production significantly after 24 h of treatment (p < 0.05 or p < 0.01 vs. DMSO). The baseline glucose production of the AMPKα1^-/- HepG2 cells was greatly elevated as compared to wild type cells (p < 0.05). Furthermore, the suppressing effect of BBR on gluconeogenesis was completely abolished after disruption of the AMPKα1 gene (p < 0.01 vs. wild type).

In summary, the above results prove that in HepG2 cells, the activation of AMPKα1 is essential for BBR to stimulate glucose consumption and inhibit gluconeogenesis.

In order to investigate the effects of BBR on AMPK in the presence of OA, wild type and AMPKα1^-/- HepG2 cells were co-administered with OA and BBR for 24 h, and cell proteins were extracted for western blot analysis. As shown in Figure 5, OA or BBR treatment did not influence the protein level of AMPKα1 or total AMPKα in wild type or AMPKα1^-/- HepG2 cells. In wild type cells, BBR greatly stimulated the phosphorylation of AMPKα1 and total AMPKα (Figure 5) and enhanced AMPK activity (Figure 6) in the presence of OA (p < 0.01 vs. DMSO). In AMPKα1^-/- HepG2 cells, the AMPKα1 protein and its phosphorylation disappeared (p < 0.001 vs. wild type) and the protein level of total AMPKα declined to about 50% of wild type cells (p < 0.01) (Figure 5). The phosphorylation of total AMPKα (Figures 1B, 5) and AMPK activity (Figures 2, 6) of AMPKα1^-/- HepG2 cells were greatly reduced to less than 10% of wild type cells (p < 0.001) with or without OA. In AMPKα1^-/- HepG2 cells treated with OA, the stimulating effects of BBR on AMPKα1 phosphorylation (Figure 5), total AMPKα phosphorylation (Figure 5), and AMPK activity (Figure 6) were totally abolished (p < 0.001 vs. wild type). Consistent with this, BBR increased the expression of P-ACC and P-Raptor compared to the OA-treated group in wild-type but not AMPKα1^-/- HepG2 cells (Figure 5).

OA treatment induced lipid accumulation in wild type and AMPKα1^-/- HepG2 cells, as determined by ORO staining (Figure 7A) and intracellular TG assay (Figure 7B) (p < 0.001 vs. DMSO). In wild type HepG2 cells, BBR administration greatly reduced lipid accumulation, as indicated by an improvement of ORO staining and a reduction of intracellular TG (p < 0.05 or p < 0.01 vs. OA alone). The anti-lipid effect of BBR was totally blocked in AMPKα1^-/- HepG2 cells (p < 0.05 or p < 0.01 vs. wild type). Accompanied with lipid accumulation, cellular mRNA expression levels of lipogenic genes like carbohydrate responsive element-binding protein (ChREBP) and fatty acid synthase (FAS) increased greatly (p < 0.01 vs. DMSO) (Figure 8). Co-administration of BBR prevented the up-regulation of ChREBP and FAS induced by OA (p < 0.05 or p < 0.01 vs. OA alone), and the inhibitory effects were abolished upon AMPKα1 knockout (p < 0.05 or p < 0.01 vs. wild type). Taken together, the above results prove that in HepG2 cells, AMPKα1 is also essential for BBR to improve lipid metabolism.