Rat hepatoma FaO cells (European Collection of Cell Cultures, Sigma-Aldrich Corp.) (30) were grown in Coon’s modified Ham’s F12 with 10% fetal bovine serum. Cells were grown until 80% confluence, incubated in high-glucose medium with 0.25% bovine serum albumin (BSA) to increase stability and solubility of FFA (31). A condition mimicking human steatosis (SS) was induced by incubating FaO cells for 3 h with oleate/palmitate mixture (2:1 M ratio, final concentration 0.75 mM). A SH condition was mimicked by incubating SS cells for 24 h with 10 ng/mL TNFα (32). After replacing the medium, both SS and SH cells were treated for 24 h with 50 µM silybin (S) [Istituto Biochimico Italiano (IBI), Lorenzini, Italy]. Silybin stock (10 mM) was prepared in dimethyl sulfoxide.

For both resazurin and sulforhodamine B (SRB) assays 1.5 × 10⁴ cells/well were seeded in 96-well plates. For resazurin assay, after treatment, the medium was removed and cells were incubated for 30 min with medium supplemented with 10 µg/mL resazurin. The reduction of resazurin to resorufin, indicative of metabolic activity, was measured fluorimetrically (λ_exc 570 nm; λ_em 600 nm) in Biotek-Cytation 3 reader (Biotek Instruments, Winooski, VT, USA) (33). For SRB assay, cells were fixed with ice-cold methanol containing 1% acetic acid for 30 min and then incubated with 0.5% SRB in 1% acetic acid for 1 h at 37°C. The unbound dye was removed with 1% acetic acid solution. The dye bound to proteins was extracted with 10 mM Tris-base solution pH 10, and absorbance was measured at 540 nm (34).

For apoptosis assessment, cells were collected and the pellet resuspended in 20 mM HEPES/NaOH pH 7.5, 250 mM sucrose, 10 mM KCl, 2 mM MgCl₂, 1 mM EDTA, 2 mM dithiothreitol (DTT), and 100 µM phenylmethylsulfonyl fluoride (PMSF), a protease-inhibitor cocktail (1 µg/mL of leupeptin, antipain, chymostatin, and pepstatin A). Caspase 3-like activity was measured in cell extracts containing 25 µg protein determined by the bicinchoninic acid method using BSA as a standard (35). For calibration, known concentrations of p-nitroanilide (pNA) were measured. Cell extracts were incubated for 1 h at 37°C in 25 mM Hepes pH 7.5, 10% sucrose, 10 mM DTT, 0.1% CHAPS, and 100 µM caspase substrate Ac-DEVD-pNA; the pNA released after cleavage from the substrate was measured. The results are expressed as amount of nanomoles of pNA released per microgram of protein (36).

Cells were scraped and lysed. Lipids were extracted using the chloroform/methanol (2:1) method (37). TG content was quantified spectrophotometrically by using “Triglycerides liquid” kit (Sentinel, Milan, Italy) in a Varian Cary-50Bio spectrophotometer (Agilent, Milan, Italy). Values were normalized to protein content. Data are expressed as percent TG content relative to controls.

Cells grown on coverslips were rinsed with phosphate-buffered saline (PBS) pH 7.4 and fixed with 4% paraformaldehyde for 20 min at room temperature. Slides were incubated with 1 µg/mL BODIPY 493/503 (Molecular Probes, Life technologies, Monza, Italy) in PBS for 30 min (38). After washing and mounting with 4′,6-diamidino-2-phenylindole (DAPI) (ProLong Gold medium with DAPI; Invitrogen) slides were examined by Nikon Eclipse E80i light microscope (Nikon, Tokyo, Japan) equipped with the standard epifluorescence filter set up. Images were captured under oil with a 100× plan apochromat objective. Analyses were performed on two independent experiments measuring at least 40 cells for each treatment using the ImageJ software.¹

Cells grown on μ-slides eight-well ibiTreat (Ibidi, Germany) were incubated with 50 nM tetramethylrhodamine ethyl ester (TMRE) and 1 µg/µL Hoechst for 30 min at 37°C in the dark. A lower concentration of TMRM⁺ was maintained in the medium to avoid leakage from mitochondria. Cells were observed under a Nikon Eclipse Ti-S microscope. Images were obtained through LSM Image Browser (39).

Cells grown on glass chamber slides (Lab-Tek, Nunc, 177380) were washed in 0.1 M cacodylate buffer and fixed in 0.1 M cacodylate buffer containing 2.5% glutaraldehyde for 1 h at room temperature. Cells were post-fixed in osmium tetroxide for 2 h and 1% uranyl acetate for 1 h. Samples were dehydrated through a graded ethanol series and flat embedded in resin (Poly-Bed; Polysciences, Inc., Warrington, PA, USA) for 24 h at 60°C. Ultrathin sections (50 nm) were cut parallel to the substrate until reaching the apical surface, stained with 5% uranyl acetate in 50% ethanol, and observed with CM10 electron microscope (Philips, Eindhoven, the Netherlands), images were taken with a Megaview 3 camera. Mitochondrial number and size were assessed in 12 randomly taken images at 25k magnification for each treatment. The mitochondrial length (major axis) and width (minor axis) were measured with iTEM software package (Olympus-SYS). Statistical analysis for mitochondrial diameter was performed using the Mann–Whitney test. The total number of mitochondria and mitochondrial cristae was manually scored. Statistical analysis was performed with Student’s t-test.

Catalase activity was evaluated spectrophotometrically in both 12,000 × g of supernatant and pellet of hepatocyte lysates (40). Catalase specific activity was expressed as micromoles of decomposed H₂O₂ per minute per milligram of sample protein. Data are expressed as percent relative to controls.

Lipid peroxidation was determined through the thiobarbituric acid reactive substances assay based on the reaction of malondialdehyde (MDA; 1,1,3,3-tetramethoxypropane) with thiobarbituric acid (TBA) (41). Briefly, 1 vol of cell suspension was incubated for 45 min at 95°C with 2 vol of TBA solution (0.375% TBA, 15% trichloroacetic acid, 0.25 N HCl) and, then, 1 vol of N-butanol was added and the absorbance of the organic phase was measured spectrophotometrically. Results were expressed as picomoles MDA per milliliter/milligram protein. Data are expressed as percent relative to controls.

The 8-OHdG excreted into the medium after DNA repair/turnover was quantified using DNA/RNA Oxidative Damage ELISA kit (Cayman Chemical Company, MI, USA). Samples were analyzed in duplicate. Standard 8-OHdG was assayed over a concentration range of 10.3–3,000 pg/mL (42). Results were expressed as picogram 8-OHdG per milliliter.

RNA was isolated using Trizol reagent, cDNA was synthesized, and real-time qPCR performed in quadruplicate using 1× IQ™SybrGreen SuperMix and Chromo4™ System apparatus (Bio-Rad, Milan, Italy) (43). The relative quantity of target mRNA was calculated by the comparative Cq method using glyceraldehyde 3-phosphate dehydrogenase (GAPDH) as housekeeping gene and expressed as fold induction with respect to controls (44). Primer pairs designed ad hoc starting from the coding sequences of Rattus norvegicus² and synthesized by TibMolBiol (Genova, Italy) are listed in Table 1.

DNA extracted using Genomic-tip 20/G kit (Qiagen, Valencia, CA, USA) was quantified using the PicoGreen dsDNA reagent (Invitrogen, Milan, Italy). Relative mtDNA copy numbers were measured by qPCR using iQ5 System Apparatus (Bio-Rad) and corrected by simultaneous measurement of nuclear DNA. Amplification of mitochondrial cytochrome c oxidase subunit II (COII, mitochondrial-encoded gene) and β-actin (nuclear-encoded gene) was performed using the primers listed in Table 1. The mtDNA content was calculated using ΔCt = average Ctnuclear DNA − average CtmtDNA and, then, was obtained using the formula mtDNA content = 2(2ΔCt).

In nuclear homogenates, NF-kB/p65 was detected by Western blot. Cellular pellet was suspended in 400 µL ice-cold Buffer A (20 mM Tris-HCl pH 7.8, 50 mM KCl, 10 µg/mL Leupeptin, 0.1 mM DTT, 1 mM PMSF) and 400 µL Buffer B (Buffer A plus 1.2% Non-indet P40). After vortex and centrifugation (14,000 × g for 30 s, 4°C) the nuclear pellet was washed with 400 µL Buffer A, resuspended in 100 µL Buffer B, mixed in ice for 15 min, and centrifuged (14,000 × g for 20 min, 4°C). The supernatant containing nuclei was collected, and 30–50 µg proteins were electrophoresed on SDS polyacrylamide gel electrophoresis (45). Membrane was blocked for 1 h in 5% fat-free milk/PBS (pH 7.4), probed overnight using rabbit anti-human NF-kB p65 (SC-109; Santa Cruz Biotechnology, DBA, Milan, Italy) in PBST buffer (PBS 0.1% Tween 20) at 4°C (46), and then washed and incubated with horseradish peroxidase-conjugated rabbit anti-mouse IgG (Sigma-Aldrich) in PBST for 1 h at room temperature. Immune complexes were visualized using chemiluminescence western blotting analysis system (Bio-Rad ChemiDoc XRS System). Films were digitized and band optical densities were quantified against the actin band and expressed as Relative Optical Density (ROD, arbitrary units). ROD of each band was expressed as percentage respect to control.

Oxygen consumption was measured at 37°C using Seahorse XFe96 Extracellular Flux Analyzer (Seahorse Bioscience, Germany). About 2 × 10⁴ cells/well were seeded in 96-well plates. In parallel, an XFe96 sensor cartridge for each cell plate was placed in a 96-well calibration plate containing 200 µL/well calibration buffer and left to hydrate overnight at 37°C. Medium was replaced with 200 µL/well of low-buffered serum-free medium pH 7.4 left at 37°C for 1 h without CO₂ to allow de-gassing the plate. A final concentration of 3 µM oligomycin, 1 µM carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone (FCCP), and a mixture of 1 µM rotenone and 1 µM antimycin were added sequentially to cells from prot A, B, and C, respectively. A total of 25 µM of each compound was pre-loaded into the ports of each well in the XFe96 sensor cartridge. The sensor cartridge and the calibration plate were loaded into the XFe96 Extracellular Flux Analyzer for calibration, and then, the calibration plate was replaced with the study plate. Three baseline rate measurements of oxygen consumption rate (OCR) were made using a 3 min mix and 3 min measure cycle. The compounds were injected pneumatically by the XFe96-Analyzer into each well, mixed, OCR measurements made using the 3 min mix and 3 min measure cycle (47).

Data are expressed as mean ± SD of at least three independent experiments in triplicate. Statistical analysis was performed using ANOVA with Tukey’s post-test (GraphPad Software, Inc., San Diego, CA, USA).

Intracellular TG content increased significantly and to a similar extent in steatotic (SS) (+109%; p ≤ 0.001) and in SH (+90%; p ≤ 0.001) cells compared to control (Figure 1A). Exposure to silybin 50 µM did not affect the TG content of control cells (data not shown) but significantly decreased to a similar extent the TG content in both SS and SH cells (−38 and −30%, respectively; p ≤ 0.001 vs their counterpart).

Lipid accumulation was associated to increased number and size of LDs, as visualized by BODIPY staining. At fluorescence microscopy, control cells showed few (about 72 LDs/cell) and small (0.9 µm average diameter) LDs, which were dispersed in the cytosol (Figures 1B,C). In SS, we observed an increase in both number (294 LDs/cell) and size (1.5 µm) of LDs, with average diameter increasing of about +67% (p ≤ 0.001) compared to controls, and sylibin reduced the LD diameter to a value similar to controls (1.1 µm; p ≤ 0.001) (Figures 1B,C). In contrast, SH cells displayed a LD number similar to that of SS cells but a LD size slightly lower (1.3 µm) (Figures 1B,C); silybin did not affect LD size.

The mRNA levels of ADRP and IkBip, two markers of LDs and NAFLD progression, respectively, were analyzed by qPCR (Figure 1D). Excess lipid accumulation in SS cells resulted in up-regulation of ADRP expression (2.4-fold induction vs control; p ≤ 0.05) that was further stimulated by silybin (4.2-fold induction compared to control; +77% vs SS; p ≤ 0.01). In SH cells, ADRP expression was at a level similar to control and it was unaffected by silybin. In regard to IkBip expression, we observed only a small, non-significant up-regulation in SS cells, while it was dramatically up-regulate in SH cells (4.6-fold induction compared to control; p ≤ 0.001). Exposure to silybin greatly and comparably reduced IkBip expression in both SS (−59% with respect to SS; p ≤ 0.05) and SH (−57% with respect to SH; p ≤ 0.001) cells. Sylibin did not affect ADRP and IkBip expression in control cells (data not shown).

Lipid metabolism is regulated by PPARs, the main transcription factors for lipolytic and lipogenic genes (48). The mRNA level of PPARs was assessed by qPCR (Figure 2A). As compared to control cells, SS cells showed a significant increase in expression of PPARα and PPARγ (1.29-fold induction, p ≤ 0.05; 2.8-fold induction, p ≤ 0.001, respectively), and a down-regulation of PPARδ (about 0.55-fold induction; p ≤ 0.05). The lipid-lowering action of silybin was sustained by increased expression of PPARα (2.6-fold induction compared to controls; +99% compared to SS, p ≤ 0.01), decreased expression of PPARγ (1.6-fold induction compared to controls, −42% compared to SS; p ≤ 0.05), and increased expression of PPARδ (1.6-fold induction compared to controls; +193% compared to SS, p ≤ 0.001). In SH hepatocytes, the profile of PPARα and PPARγ expression was similar to SS, whereas PPARδ expression showed values similar to controls. Exposure of SH cells to silybin reduced transcription of both PPARα (1.1-fold induction compared to controls; −39% compared to SH, p ≤ 0.05) and PPARγ (1.5-fold induction compared to controls; −48% compared to SH, p ≤ 0.01), without affecting PPARδ expression.

The lipid-lowering action of silybin could be sustained by increased TG secretion, since SS cells released more TGs into culture medium than controls (+157%; p ≤ 0.001). The addition of silybin 50 µM did not modify this picture. In SH cells, a significant reduction in TG secretion was observed compared to SS (−28%; p ≤ 0.001). The addition of sylibin counteracted this condition as TG secretion was stimulated (+103% with respect to SH; p ≤ 0.001) (Figure 2B).

In contrast, TG accumulation might be also reduced by stimulation of oxidative pathways taking place in mitochondria, peroxisomes, and ER (43). Thus, mitochondrial CPT-1 and peroxisomal AOX for β-oxidation and microsomal cytochrome P450 (CYP) 2E1 for ω-oxidation were assessed. The mRNA levels of CPT-1 were significantly up-regulated in both SS and SH cells (2.44- and 2.91-fold inductions, respectively; p ≤ 0.05) and further increased upon incubation with silybin (+199% and +164 with respect to SS and SH cells, respectively; p ≤ 0.001) (Figure 2C). CYP2E1 expression was up-regulated in SS cells (2.20-fold induction; p ≤ 0.001) and was further increased upon incubation with silybin (+63% with respect to SS; p ≤ 0.001), whereas in SH hepatocytes, CYP2E1 expression was at basal level and it was markedly increased by silybin (+518 with respect to SH cells; p ≤ 0.01) (Figure 2C). In peroxisomes, AOX activity was stimulated in SS cells (+53%, p < 0.001 compared to controls), while silybin did not change it. We observed a decrease in AOX activity in SH cells compared to SS (−56% with respect to SS; p < 0.001) and an increase upon silybin treatment (+90% with respect to SH cells; p < 0.001) (Figure 2D).

Up-stream of FA oxidation there is ATGL, a lipase controlling lipid mobilization from LDs. A significant increase in ATGL transcripts was observed in SS cells (2.29-fold induction with respect to control; p ≤ 0.05) and a larger increase in SH cells (11.7-fold induction with respect to controls; p ≤ 0.001) (Figure 2E). ATGL mRNA expression was reduced after incubation of SH cells with silybin (−85% compared to SH; p ≤ 0.001).

We observed no significant changes in cell viability in SS cells, whereas a reduced viability was observed in SH cells (−24% compared to control; p ≤ 0.001) and silybin did not counteract this effect (Figure 3A). Similar results were supplied by cell density analysis: no significant decrease in cell density occurred in SS cells, while cell density was reduced in SH cells (−26% compared to control; p ≤ 0.001) and silybin did not counteract this effect (Figure 3A). In regard to apoptosis, we observed an increase in the caspase 3-like activity in SS cells (+126% compared to control; p ≤ 0.001) and a larger increase in SH cells (+219% compared to control; p ≤ 0.001). Of note, incubation of both SS and SH cells with silybin significantly reduced caspase 3-like activity (−68 and 158%, respectively, in SS + sil and SH + sil; p ≤ 0.01 and p ≤ 0.001) (Figure 3B).

Mitochondrial respiration was analyzed by using the Seahorse Extracellular Flux Analyzer (Figure 4). Basal respiration remained unchanged across control, SS, and SH cells; silybin did not affect basal respiration of SS cells, but significantly reduced that of SH cells (−47% with respect to control; p ≤ 0.001) (Figure 4A). The proton leak was significantly reduced in both SS and SH cells with respect to controls (−42% and −31%, respectively; p ≤ 0.001 and p ≤ 0.05), and silybin did not influence this picture (Figure 4B). Maximal respiration remained unchanged between control, SS cells, and SS cells exposed to sylibin, but it was significantly increased in SH cells (+12% respect to control, p ≤ 0.05) and silybin dramatically reduced it (−70%, compared to SH; p ≤ 0.001) (Figure 4C). Also, the oxygen consumption associated with ATP production did not change in SS cells, while it increased in SH cells (+20% respect to control; p ≤ 0.05). Silybin exerted an opposite effect by increasing ATP production in SS cells (+20% respect to SS; p ≤ 0.05) and decreasing it in SH cells (−56% respect to SH cells; p ≤ 0.001) (Figure 4D).

To assess possible mitochondria damage, we evaluated the mtDNA copy number and expression of the two main genes involved in mtDNA repair by qPCR (Figure 5). The mtDNA copy number did not change in SS cells, but it was reduced in SH cells (−43% respect to control; p ≤ 0.05), and silybin was not able to rescue this damage (Figure 5A). With regard to the enzymes for DNA repair, SH cells exhibited an increase in mRNA expression of both APE1 (1.50-fold induction compared to control; +68% with respect to SS, p ≤ 0.001) and POLG (1.45-fold induction compared to control; +75% with respect to control, p ≤ 0.001), and silybin counteracted this effect (−30% for APE1 and −48% for POLG respect to SH; p ≤ 0.001) (Figure 5B).

In contrast, fluorescence microscopy of TMRE-stained mitochondria was performed to assess the potential of mitochondrial inner membrane (Figure 5C). We observed a higher fluorescent signal in SS cells than in control cells, and fluorescence was reduced and less organized in SH cells where it appeared rather diffuse and irregular. Treatment of both SS and SH cells with silybin increased TMRE signal with cells appearing more brilliant compared to untreated counterparts.

Ultrastructural morphology of mitochondria was assessed by EM analysis to measure mitochondrial number, major and minor axe diameters, number, and organization of cristae (Figures 6A–C). Our results showed that both mitochondrial morphometry and cristae number did not change in SS cells. Upon treatment of SS cells with silybin mitochondria showed a significant increase in size (both major and minor axes, p ≤ 0.001) and the mitochondrial cristae increased in number and were better organized and parallel to each other’s (Figures 6A–C). In SH cells, mitochondrial size did not change (Figures 6A,B) and the mitochondrial cristae were largely disorganized and significantly reduced in number compared to SS cells (p ≤ 0.05) (Figure 6C). Interestingly, in SH cells, treatment with silybin rescued both the number (p ≤ 0.001) and the parallel organization of mitochondrial cristae compared to SS cells (Figure 6C).

Increased fat oxidation produces excess of ROS, which react with cellular structures leading to lipid peroxidation and DNA oxidative damage, which are classical markers for cellular oxidative stress (18). By measuring lipid peroxidation, we found increased MDA levels in both SS and SH cells with respect to controls (+195 and +255%, respectively; p ≤ 0.001), which decreased upon incubation with silybin (−49 and −52% with respect to SS and SH cells, respectively; p ≤ 0.001) (Figure 7A). When extracellular levels of 8-OHdG were measured as a marker for oxidative DNA damage, a slight but no significant increase was observed in both SH and SS cells, and silybin reduced 8-OHdG levels in both conditions (−49% in SS and −50% in SH cells with respect to their counterparts; p ≤ 0.05) (Figure 7B).

The ability of silybin in protecting cells from fat-induced oxidative stress was confirmed by measuring the activity of the antioxidant enzyme catalase and expression of the transcription factor NF-κB, which is activated by oxidative stress. Catalase activity increased in both SS and SH cells with respect to control (+145% and +140%, respectively; p ≤ 0.01) and decreased upon incubation with silybin (−38% and −65% with respect to SS and SH cells, respectively; p ≤ 0.01 and p ≤ 0.001) (Figure 7C). NF-κB activation did not change significantly in SS and SH cells with respect to controls, but a significant reduction was observed upon exposure to silybin (−29% and −31% with respect to SS and SH cells, respectively; p ≤ 0.05) (Figure 7D).