Ethanol and fish oil induce NFκB transactivation of the collagen α2(I) promoter through lipid peroxidation–driven activation of the PKC-PI3K-Akt pathway

Authors

  • Natalia Nieto

    Corresponding author
    1. Department of Medicine, Division of Liver Diseases, Mount Sinai School of Medicine, New York, NY
    • Department of Medicine, Division of Liver Diseases, Mount Sinai School of Medicine, Box 1123, 1425 Madison Avenue, Room 11-76, New York, NY 10029
    Search for more papers by this author
    • fax: 212-849-2574


  • Potential conflict of interest: Nothing to report.

Abstract

To analyze whether fish oil, as a source of polyunsaturated fatty acids from the n-3 series, could synergize with ethanol to promote collagen I upregulation in vivo, collagen α2(I) promoter-βGal (COL1A2-βGal) transgenic mice were fed a diet enriched in fish oil in the presence of ethanol (ethanol group) or dextrose (control group). Ethanol-fed mice showed mild steatosis, increased alanine aminotransferase (ALT), aspartate aminotransferase (AST), nonsterified fatty acids, and plasma alcohol levels along with elevated cytochrome P450 2E1 activity, lipid peroxidation end products, and low glutathione (GSH) levels, which suggested enhanced oxidant stress and liver injury. Increased transactivation of the COL1A2 promoter assessed by βGal activity was shown in vivo and by transfection with deletion constructs for the collagen α1(I) promoter (COL1A1) and COL1A2 promoters in vitro. Transcriptional regulation of both COL1A1 and COL1A2 promoters was validated by nuclear in vitro transcription run-on, northern blot analysis, and quantitative polymerase chain reaction, which was followed by the subsequent upregulation of collagen I protein with no changes in matrix metalloproteinase 13 (MMP 13). To further analyze the potential mechanism for collagen I upregulation, an in vitro coculture model was designed with primary stellate cells seeded on the bottom plate of a Boyden chamber and the rest of the liver cells plated on a cell culture insert, and fish oil or fish oil plus ethanol were added. The combination of fish oil plus ethanol increased nuclear factor κB binding to the COL1A2 promoter both in vivo and in the cocultures and also resulted in increased phosphorylation of protein kinase C, activation of PI3 kinase, and phosphorylation of Akt. The in vitro addition of vitamin E prevented such activation and collagen I increase. Furthermore, inhibitors of all 3 kinases blocked the increase in collagen I and NFκB binding to the COL1A2 promoter; the latter was also prevented by vitamin E. Conclusion: These results suggest that fish oil (mainly n-3 polyunsaturated fatty acids [PUFAs]) can synergize with ethanol to induce collagen I, transactivating the COL1A2 promoter through a lipid peroxidation-PKC-PI3K-Akt-NFκB-driven mechanism in the absence of overt steatosis and inflammation. (HEPATOLOGY 2007;45:1433–1445.)

There is considerable evidence that the amounts and types of fats in the diet are key determinants of lesions in alcoholic liver disease.1, 2 Saturated fatty acids may be protective against alcoholic liver disease, whereas PUFAs promote alcoholic liver disease.3, 4 Long-chain PUFAs from the n-3 series, present in fish oil (FO), interact with nuclear receptor proteins, thereby influencing the transcription of regulatory genes. They are natural ligands of the nuclear peroxisome proliferator-activated receptor α (PPARα), which modulates lipid metabolism.5 Low levels of circulating n-3 PUFAs, with a consequent increase in the n-6/n-3 ratio, modify PPARα activity in the liver. Previous studies have shown that diets enriched with n-3 PUFAs increase insulin sensitivity in rats, reduce the level of intrahepatic triglycerides, and ameliorate steatohepatitis in both mice and rats.6

The oxidation of n-3 PUFAs in the endoplasmic reticulum involves the initial formation of ω-hydroxy or (ω-1)-hydroxy fatty acids catalyzed by microsomal cytochrome P450 (CYP4A1 and CYP2E1). Dicarboxylic fatty acids, derived from ω-hydroxy and (ω-1)-hydroxy fatty acids, are increased in states of impaired mitochondrial fatty acid β-oxidation. Microsomal ω-hydroxylation by means of CYP2E1 and CYP4A1 is increased in alcohol-fed rats.7, 8 Evidence in the literature suggests that fatty acids and their ω-oxidation products activate PPARs.9, 10 PPARs are actively involved in regulating genes involved in fatty acid metabolism, such as enzymes of the extramitochondrial fatty acid oxidation pathways (e.g., peroxysomal fatty acyl CoA oxidase, CYP4A1, and liver fatty acid binding protein).11–13 The induction of hepatic extramitochondrial pathways of fatty acid oxidation by means of PPARα serves to provide the liver cell with alternative means for the catabolism of fatty acids under conditions of markedly increased fatty acid flux and fatty acid overload.14 PPARα appears to act as a cellular transducer that senses the presence of fatty acid overload states and directs the appropriate adaptive hepatocellular gene response.14

A key feature of liver fibrosis is the increase in collagen I protein.15 Collagen is a heterotrimeric protein composed of 2 α1 chains and 1 α2 chain encoded by the COL1A1 and COL1A2 genes. Both COL1A1 and COL1A2 genes are highly sensitive to reactive oxygen species (ROS) and acetaldehyde, a product from alcohol metabolism.16–25 It has been reported that the COL1A2 promoter contains at least 2 putative NFκB binding sites.26 Oxidant stress is a major factor inducing the phosphorylation of IκB, which releases NFκB, translocating it then to the nucleus to activate the transcription of target genes.27 In view of the potential link between oxidative stress and the activation of the COL1A1 and COL1A2 genes in hepatic stellate cells (HSCs), it was speculated that feeding FO plus alcohol for a short time could lead to the activation of collagen I even in the absence of overt steatosis and inflammation.

Abbreviations

ALT, alanine aminotransferase; AST, aspartate aminotransferase; COL1A1, collagen α1(I) promoter; COL1A2, collagen α2(I) promoter; EtOH, ethanol; FO, fish oil; GSH, glutathione; 4-HHE, 4-hydroxyhexenal; 4-HNE, 4-hydroxynonenal; HSC, hepatic stellate cell; LPO, lipid peroxidation; MMP, matrix metalloproteinase; PPARα, peroxisome proliferator-activated receptor α; PUFA, polyunsaturated fatty acid; ROS, reactive oxygen species; TIMP, tissue inhibitor of metalloproteinases; TNFα, tumor necrosis factor α; VLDL, very low density lipoproteins.

Material and Methods

Transgenic Mice.

Transgenic mice harboring the −17 kb to +54 bp of the proximal promoter of the mouse COL1A2 gene cloned upstream of the E. coli β-gal reporter gene (LacZ) were used. These transgenic mice were obtained from Dr. Benoit de Crombrugghe (Department of Molecular Genetics, University of Texas, M.D. Anderson Cancer Center, Houston, TX).28–33 Mice were bred in our institution and received humane care in compliance with the guidelines of the National Institutes of Health and the Animal Care Committee of Mount Sinai School of Medicine.

Diets and Experimental Design.

The composition of the diets is described in Table 1. The FO (Sigma F8020) used for preparing the diets is derived from menhaden fish and contains the following fatty acids: 14:0 myristic acid, 6%-9%; 16:0 palmitic acid, 15%-20%; 16:1 palmitoleic acid, 9%-14%; 18:0 stearic acid, 3%-4%; 18:1 oleic acid, 5%-12%; 18:2 linoleic acid, <3%; 18:3 linolenic acid, <3%; 18:4 octadecatetraenoic acid, 2%-4%; 20:4 arachidonic acid, <3%; 20:5 eicosapentaenoic acid, 10%-15%; and 22:6 docosahexaenoic acid, 8%-15%. This sum is approximately 80% of the fatty acids, whereas the remaining 20% represents unidentified fatty acids. Diets were made fresh daily. The FO was stored in air-tight containers, filled with nitrogen, in a cold room at 4°C. Lipid peroxidation (LPO) was measured to exclude the possibility of autoxidation. To adapt the mice, the FO diet alone was administered for 3 days to all mice because switching from chow diet pellets to any liquid diet requires ∼2 days of adjustment. Ethanol was progressively incorporated into the diet so that mice did not repel the diet because of an aversion to the smell of alcohol. Ethanol was incorporated into the FO diet as follows: 4 days with 10% of calories given as ethanol, 7 days with 20% of calories given as ethanol, and 14 days with 35% of calories given as ethanol. A control group drinking only ethanol was not possible because mice would not drink pure alcohol, even if diluted in water. The control group for the experiment was the FO group.

Table 1. Composition of the Diets
Ingredients (per 100ml of diet)FOFO+EtOH
AIN-93G (g) (salt mix, Bio-Serv F8538)0.930.93
AIN-93VX (g) (vitamins, Bio-Serv F8001)0.260.26
Choline bitatrate (g) (Sigma C1629)0.050.05
D,L-Methionine (g) (Sigma M9500)0.110.11
Lactoalbumin (g) (Bio-Serv 1275)5.755.75
Dextrose (g) (Sigma D9434)11.751.75
Fish Oil (Sigma F8020)3.663.66
Susp. agent K (g) (Bio-Serv 7945)0.450.45
Ethanol (ml)0.007.24

The animals were sacrificed, and both the liver and body weight values were recorded. Blood was collected from the abdominal aorta and centrifuged at 3000 rpm for 3 min, and the serum was assayed for the activity of ALT and AST with kits from Thermo Electron Corp. (Waltham, MA) and for the ethanol concentration with an alcohol dehydrogenase kit from Sigma (St. Louis, MO). Each liver was excised into different fragments for biochemical assays, paraffin embedding and staining, and frozen sections.

Histochemistry.

Liver samples were fixed overnight in 10% buffered formalin and embedded in paraffin. Five-micrometer sections were stained with hematoxylin and eosin (H&E) to be evaluated by a liver pathologist, who was blinded from the experimental information. The Sirius red/Fast green staining, used to assess collagenous proteins, was carried out as described.34

Hepatic Antioxidant Defense.

Liver homogenates were prepared in 10 volumes of ice-cold pH 7.4, 50 mM/l Tris-HCl/sodium phosphate buffer and 1.15% KCl with 1 mM/l ethylenediaminetetraacetic acid. For total glutathione, 50 μg of liver were homogenized in 5% trichloroacetic acid at a ratio of 1:10 (wt/vol) and centrifuged for 5 minutes at 8000 rpm and 4°C. GSH levels were determined in the protein free extract by the recycling method of Tietze.35 The total glutathione peroxidase, catalase, and total superoxide dismutase activities were measured according to the methods of Flohé and Günzler,36 Claiborne,37 and Paoletti and Mocali,38 respectively. Protein concentrations were determined with the method of Lowry et al.,39 and the enzyme activities are expressed as units per milligram of protein.

General Procedures.

Serum tumor necrosis factor α (TNFα) was determined by an enzyme-linked immunosorbent assay (Biosource, Camarillo, CA). Nonesterified fatty acids were analyzed with a kit from Wako Chemicals (Richmond, VA). Liver triglycerides were measured with a kit from Roche. Rat liver microsomes were prepared by differential centrifugation, and the catalytic activity of CYP2E1 as determined in earlier publications.20 Nuclear in vitro run-on transcription assays were carried out with nuclei isolated from total livers as described.20, 21 A Northern blot was carried out as described.20, 21 Liver RNA was extracted with the RNeasy mini kit (Qiagen, Chatsworth, CA) and treated with DNase. RNA (1 μg) was reverse-transcribed with first-strand cDNA synthesis with random primers (Promega, Madison, WI). Quantitative real-time PCR was performed in a Roche Lightcycler 480 using the following PCR primers: COL1A1 forward 5′-CCT CAA GGT TTC CAA GGA CC-3′ and COL1A1 reverse 5′-CAA TCC ATC CAG ACC GTT GTG-3′ and COL1A2 forward 5′-CCT CAA GGT TTC CAA GGA CC-3′ and COL1A2 reverse 5′-CAA TCC ATC CAG ACC GTT GTG-3′. All values were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH).

Western Blot Analysis.

A Western blot analysis was carried out with liver homogenates, cell lysates, and a cell culture medium. Antibodies to 4-hydroxyhexenal (4-HHE), malon dialdehyde, and 4-hydroxynonenal (4-HNE) were provided by Dr. Uchida (Nagoya University, Japan). Antibodies raised against CYP2E1, GCLC (glutamate cysteine ligase catalytic subunit) and GGLM (glutamate cysteine ligase modulatory subunit), and collagen I were gifts from Dr. Lasker (Hackensack Biomedical Research Institute, Hackensack, NJ), Dr. Kavanaugh (University of Washington, Seattle), and Dr. Schuppan (Harvard Medical School, Boston, MA), respectively. Anti–α-Sma and anti–β-tubulin were from Sigma. Antibodies for p50 and p65 and for kinases (p38, extracellular signal-regulated kinase ½ (ERK½), PKC, and Akt 1/2/3) and their phosphorylated forms were from Santa Cruz Biotechnologies (Santa Cruz, CA). The PKC antibody recognizes all PKC isoforms. Anti-Timp1 and anti-MMP13 were from Chemicon International (Temecula, CA). All primary antibodies were used at a dilution of 1/2000 to 1/5000. Goat anti–rabbit IgG and goat anti–mouse IgG (Chemicon; both at 1/5000) were used as secondary antibodies.

β-Gal Staining and Quantification.

β-Gal staining was performed with frozen sections from livers collected in optimal cutting temperature compound and immediately frozen with 2-methylbutane on dry ice. Proteins were also extracted from individual livers, and the β-gal activity was measured with a chemiluminescent reporter assay (Galactolight Plus, Tropix).

NFκB Binding Activity.

Electrophoretic mobility shift assays to determine the binding activity of NFκB were performed as described in previous studies.40 Equal amounts of nuclear proteins were incubated with a 5′32 P-labeled oligonucleotide containing the NFκB consensus site. The incubation mixtures were separated in a 7% nondenaturing polyacrylamide gel, and bands were detected by autoradiography. The specificity of binding was determined by the prior addition of a 100-fold excess of an unlabeled competitor consensus oligonucleotide and supershift analysis (not shown).

Coculture Model.

The coculture model, shown in Fig. 6A, includes a Boyden chamber and a cell culture insert with a membrane of 8 μm of pore size that allows for the diffusion of soluble mediators from the upper compartment to the lower compartment. Details of the coculture setting and cell isolation protocols are described in earlier publications for hepatocytes and HSCs18, 19, 40, 41 and for Kupffer cells and HSCs.41 The only difference from these previous studies was the incorporation of sinusoidal endothelial cells into the insert. They were isolated from the nonparenchymal cell fraction by elutriation at 18 mL/minute. The cell ratios were similar to those found in the liver. After 1 day of incubation, the medium was removed, and the cell-culture inserts containing the hepatocytes, sinusoidal endothelial cells, and Kupffer cells were transferred onto the HSCs. Fresh Dulbecco's modified Eagle's medium (DMEM)-F12 with 50μM/well of FO or FO plus 50 mM ethanol was added. In Fig. 6D, each fatty acid was added at a concentration of 30 μM for 24 hours.

Figure 6.

(A) Scheme of the coculture model: after liver perfusion and elutriation, HSCs are seeded onto polystyrene plates at a density of 150,000 cells, and the remainder of the liver cell populations (Kupffer cells, endothelial cells, and hepatocytes, the latter isolated from separate mice) at a cell-to-cell ratio proportional to that found in the liver are plated onto the cell culture inserts. Dulbecco's modified Eagle's medium (DMEM)-F12 medium (4 mL) is then added to each system. One day later, the medium is discarded, and the inserts containing the other cell populations or empty inserts (to be considered the coculture controls) are transferred onto the HSCs. Fresh medium (4 mL) is added with FO or FO plus EtOH, and samples of HSC lysates are collected at selected time points according to the experiment. (B) To validate the coculture model as a tool for understanding the mechanism of COL1A2 induction, the activity of β-Gal was quantified in the cocultures with cells isolated from the COL1A2-β-Gal mice. The results are expressed as arbitrary units of chemiluminescence/milligram of protein. ***P < 0.001 for FO plus EtOH versus FO. (C) LPO end products in the coculture were assessed by the use of an anti–4-HHE antibody. The results are expressed as arbitrary units of densitometry and are means ± SEM (n = 6). ***P < 0.001 for FO plus EtOH versus FO. (D) The cocultures showed increased collagen I protein expression in HSCs mainly when incubated in the presence of FO plus EtOH. The cocultures were incubated with selected fatty acids from each series (18:1n-9, 20:4n-6, 20:5n-3, and 22:5n-3) or with a saturated fatty acid (16:0), all of which are present in FO. The results are expressed as arbitrary units of densitometry under the blots. (F) HSCs were transfected with a series of deletion constructs of the COL1A2 promoter linked to the Luc reporter gene, as shown on the left side of each panel. Twenty four hours later, the cell culture inserts were transferred, and media containing FO or FO plus EtOH were added. Firefly Luc activity for the COL1A2 promoter transgenes in HSCs alone or in the coculture is shown after 24 hours. The results are corrected by transfection efficiency with the Luc activity from a cotransfected Renilla Luc reporter vector (pRL-null) and by the protein content and are normalized to the corrected firefly Luc activity detected in cells transfected with the Luc reporter vectors pGL3-Luc or PXP1-basic, in which the recombinant expression vectors were created. The results are expressed as average values of n = 3 determinations ± SEM. *P < 0.01 and ***P < 0.001 for the coculture versus HSCs and • • • P < 0.001 for FO plus EtOH versus FO or versus untreated (control non FO-treated cells).

Transfection Experiments and Reporter Assays.

Reporter DNA constructs containing upstream sequences of the human COL1A2 promoter linked to the Luc gene were provided by Dr. Francesco Ramirez (Hospital for Special Surgery, New York, NY).42 The human COL1A2 promoter sequences span from −3500 to +58 bp with 5′ endpoints of −3500, −772, −378, −183, and −108 bp.42 Details on the transfection protocol can be found in earlier publications.41 HSCs were incubated in the presence of the transfection mix for 48 hours, after which the medium was replaced and the inserts containing the other cells or empty inserts were transferred onto the HSC plates. The cocultures were treated with FO or FO plus ethanol. Samples for the Luc activity were collected, and the reaction was run using a kit from Promega as described.43

Statistical Analysis.

Data were analyzed in Figs. 1–5 by a Student t test and in the rest of the figures by a 2-factor analysis of variance, as suggested by the Department of Statistics at Mount Sinai School of Medicine. The values are expressed as means plus or minus the SEM (n = 3-5).

Figure 1.

(A) Time-dependent evolution of the body weight in COL1A2-β-Gal mice fed the FO and the FO plus EtOH diets. All mice were fed the FO diet for 3 days, and then EtOH was incorporated into the FO plus EtOH group progressively from 10%-35% of the total calories. The body weight was monitored weekly. The results are expressed as average values. ***P < 0.001 for FO plus EtOH versus FO. The average liver weight and the liver-to-body weight ratio are shown on the right. (B) H&E staining. Mice fed the FO plus ethanol diet showed centrilobular steatosis (arrows; original magnification = 200×). (C) The activity of ALT and AST, NEFA, and plasma ethanol levels are presented as units per liter, milliequivalents per milligram of protein, and millimoles per liter, respectively, and are the means plus or minus the SEM (n = 6). ***P < 0.001 for FO plus EtOH versus FO.

Figure 2.

(A) The catalytic activity of CYP2E1 was assessed measuring the rate of oxidation of p-nitrophenol to p-nitrocatechol. The results are expressed as picomoles per minute per milligram of microsomal protein and are average values ± SEM (n = 6). A Western blot analysis for CYP2E1 is shown below the graph. (B) Western blot analysis showing the expression of GCLC and GCLM in total liver of COL1A2-β-Gal mice. For the Western blot analysis in parts A and B, the results are corrected by the β-tubulin signal and are expressed as arbitrary units of densitometry under the blots, being average values of n = 3 per group. The expression of each protein in mice fed the FO diet was assigned a value of 1. LPO byproducts in livers were estimated with blotting for (C) 4-HHE, (D) MDA, and (E) 4-HNE residues by use of specific antibodies and by quantification by scanning. The results are expressed as arbitrary units of densitometry and are the means plus or minus the SEM (n = 6). In all panels, ***P < 0.001 for FO plus EtOH versus FO.

Figure 3.

(A) Mouse COL1A2 promoter scheme showing the enhancer region and the proximal promoter in which the transforming growth factor β (TGFβ)-, TNFα-, and ROS-responsive elements are located. (B) The β-Gal staining shows activation of the mouse COL1A2 promoter in mice fed the FO plus EtOH diet (arrows), whereas no activation is found in mice fed the FO diet (magnification = 200×). (C) The activity of β-Gal, an indicator of induction of the mouse COL1A2 promoter, was quantified by chemiluminescence, and the results are expressed as arbitrary units of chemiluminescence per milligram of protein. ***P < 0.001 for FO plus EtOH versus FO. (D) Transcription of COL1A1 and COL1A2 in HSCs in a coculture: nuclear in vitro transcription run-on assays were performed with nuclei isolated from total liver of COL1A2-β-Gal mice. A representative blot (n = 2 experiments in duplicate each) of labeled nuclear RNAs hybridized to COL1A1, COL1A2, and GAPDH cDNA is shown. ***P < 0.001 for FO plus EtOH versus FO.

Figure 4.

(A) COL1A1 and COL1A2 mRNA were analyzed by Northern blot using total RNA isolated from livers of COL1A2-β-Gal mice. The quantification of the intensity of the signal shown under the blots was made with ImageQuant software. The results were corrected by the intensity of the GAPDH signal and are expressed as average values, with the signal for the FO mice assigned a value of 1 for each mRNA. (B) Quantitative real-time PCR for both COL1A1 and COL1A2 mRNA. The results, given as average values, are corrected for the intensity of the GAPDH signal and are expressed as a fold increase over the signal for the FO group, which was assigned a value of 1. (C) The total collagen content was visualized by Sirius red/Fast green staining. More collagen deposition (red color, arrows) was observed in mice fed the FO plus EtOH diet than in mice fed the FO diet and was distributed in the periportal and pericellular areas (magnification = 200×). (D) The total collagenous proteins were eluted from the slides and quantified by spectrophotometry. The results are given as average values and are expressed as micrograms of collagenous proteins per milligram of total protein. (E) Western blot analysis for intracellular collagen I, α-Sma, and MMP13 expression. The numbers under the blot indicate arbitrary units of densitometry and are corrected by β-tubulin expression. The expression of each protein in mice fed the FO diet was assigned a value of 1. In all panels, ***P < 0.001 for FO plus EtOH versus FO (n = 3-6).

Figure 5.

(A) Western blot analysis of phosphorylated PKC, phosphorylated ERK1/2, PI3K, phosphorylated Akt1/2/3, and PPARα. The numbers under the blots indicate arbitrary units of densitometry and are corrected by the corresponding total protein or by β-tubulin expression. The expression of each protein in mice fed the FO diet was assigned a value of 1. In all panels, ***P < 0.001 for FO plus EtOH versus FO (n = 4). (B) Degradation of IκBα and (C) translocation of p50 and p65 from the cytosol into the nucleus. (D) Electrophoretic mobility shift assay with nuclear extracts from total liver of COL1A2-β-Gal mice fed either the FO diet or the FO plus EtOH diet using a radiolabeled oligonucleotide containing the NFκB binding site in the COL1A2 promoter.

Results

Liver Histopathology.

To determine the potential activation of the COL1A2 promoter in vivo, COL1A2-βGal transgenic mice were fed a diet enriched in FO with either dextrose (control group) or ethanol (ethanol group). At the endpoint of the experiment, there was a 30% increase in the body weight and a 20% increase in the liver weight in the ethanol group in comparison with the control group (Fig. 1A). None of the animals in the control group showed major evidence of pathological damage (Fig. 1B, left panel). Mice fed the FO diet plus ethanol showed some degree of liver injury as assessed by H&E staining (mild steatosis with minimal inflammatory infiltrates; Fig. 1B, right panel), ALT (135 ± 12.9 versus 55 ± 8.3 U/L, P < 0.001), and AST (180 ± 9.1 versus 97.8 ± 6.6 U/L, P < 0.001). Plasma alcohol levels in the ethanol-fed group were ∼26.1 ±0.6 mmol/L (Fig. 1C). The levels of plasma nonesterified fatty acids were 8.5 ± 0.2 versus 3.5 ± 0.7 mEq/mg protein (P < 0.001; Fig. 1C), and the levels of hepatic triglycerides were 11.2 ± 0.1 versus 14.4 ± 0.1 (P < 0.01; not shown). It is possible that the NEFA (non-esterified fatty acids) could be derived from adipose triglyceride stores with increased mobilization in the mice fed ethanol plus FO. It is also likely that under these conditions there is not much increase in triglyceride synthesis in the liver and that ethanol can decrease β-oxidation of fatty acids, so there may be an accumulation of NEFA, which, if not stored as triglycerides, may spill over to the blood.

Antioxidant/Prooxidant Status.

To assess the antioxidant defense of the liver, the activity of total superoxide dismutase, catalase, and total glutathione peroxidase was analyzed. All 3 enzymes remained similar in both groups (Table 2). The GSH values in mice fed the FO plus ethanol diet were approximately 40% lower than those in mice fed the FO diet (P < 0.01, Table 2). De novo GSH synthesis is mediated by glutamate-cysteine ligase, which has 2 subunits, glutamate cysteine ligase catalytic subunit (GCLM) and glutamate cysteine ligase modulatory subunit (GCLC).43 A western blot analysis showed a decrease in GCLM and GCLC proteins in mice fed the FO plus ethanol diet in comparison with mice fed the FO diet (Fig. 2B). The data suggest that the FO plus ethanol diet may lower the GSH content and likely its synthesis in the liver.

Table 2. Antioxidant Defense in Liver
AntioxidantFOFO+EtOH
  • NOTE. Mice were fed either a diet enriched in fish oil (FO) or diet enriched in FO plus ethanol (FO+EtOH). Levels of GSH were measured and expressed as pmol/mg of protein. The activities of superoxide dismutase (SOD), catalase (CAT), and total glutathione peroxidase (GPX) were measured and expressed as U/mg of protein. Results are means ± S.E.M (n=6).

  • **

    P<0.01 for FO+EtOH versus FO.

GSH3.2 ± 0.41.9 ± 0.1**
SOD22.1 ± 2.423.6 ± 1.9
CAT4.3 ± 0.14.2 ± 0.2
GPX30.1 ± 1.928.8 ± 1.1

The CYP2E1 activity, determined as the rate of oxidation of p-nitrophenol to p-nitrocatechol, was higher in mice fed the FO plus ethanol diet than in mice fed the FO diet (1930 ± 45 versus 380 ± 49 pmol/minute/mg of protein, P < 0.001; Fig. 2A). Likewise, a Western blot analysis showed higher CYP2E1 expression in mice fed the FO plus ethanol diet than in mice fed the FO diet (P < 0.001; Fig. 2A, bottom).

FO is high in n-3 PUFAs and low in n-6 PUFAs. n-3 PUFAs produce large amounts of 4-HHE during peroxidation,44 and n-6 PUFAs mostly generate MDA and 4-HNE.45 There were 8-fold, 3-fold, and 3.5-fold increases in LPO end products (4-HHE, MDA, and 4-HNE, respectively) in mice fed the FO plus ethanol diet in comparison with mice fed the FO diet (P < 0.001; Fig. 2C–E).

FO Plus Ethanol Activate the COL1A2 Promoter In Vivo.

Transgenic mice harboring the −17 kb to +54 bp of the proximal promoter of the mouse COL1A2 gene cloned upstream of the E. coli β-gal LacZ reporter gene have been shown to be valuable in studies of the activation of the COL1A2 gene in vivo.28–33 The administration of FO plus ethanol in the diet activated the COL1A2 promoter, as shown by the β-gal positive staining (Fig. 3B, right panel). Quantification of the β-gal activity by chemiluminescence showed an approximately 37-fold increase in the β-gal activity by the FO plus ethanol diet (P < 0.001; Fig. 3C). The enhanced COL1A1 and COL1A2 expression induced by FO plus ethanol occurred most likely through a transcriptional mechanism (Fig. 3D). Newly transcribed COL1A1 and COL1A2 mRNA increased 3.8-fold and 2.5-fold, respectively, in mice fed the FO plus ethanol diet in comparison with mice fed the FO diet (P < 0.001; Fig. 3D).

Collagen I Expression.

Northern blot analysis and real-time PCR for COL1A1 and COL1A2 mRNA showed about 4-fold and 3-fold increases, respectively, in mice fed the FO plus ethanol diet in comparison with mice fed the FO diet (P < 0.001; Fig. 4A, B). Liver sections from COL1A2-βGal transgenic mice fed the FO plus ethanol diet showed moderate pericentral and pericellular fibrosis (Fig. 4C, right panel), which was later quantified by spectrophotometry after color elution (P < 0.001; Fig. 4D). A Western blot analysis revealed a 3.6-fold increase in collagen type I protein with no changes in MMP13, the metalloproteinase that specifically degrades collagen I, in mice fed the FO plus ethanol diet in comparison with mice fed the FO diet. HSC activation was apparent from the increase in α-Sma in mice fed FO plus ethanol (P < 0.001; Fig. 4E).

The PKC-PI3K-Akt Pathway Is Activated by FO plus Ethanol.

Because most kinases are sensitive to ROS46–48 and collagen type I expression can be modulated by the activation of stress kinases,28, 49, 50 we analyzed the expression of some key kinases in the livers of mice fed the FO diet or the FO plus ethanol diet. There was at least a 2-fold up-regulation of phosphorylated PKC, PI3K, and its downstream effector phosphorylated Akt1/2/3, whereas phosphorylated ERK1/2 was lower in mice fed the FO plus ethanol diet than in mice fed the FO diet (P < 0.001; Fig. 5A). PPARα expression remained similar in both groups.

Activation of NF-κB and Degradation of IκBα.

The COL1A2 promoter contains 2 putative NFκB binding sites.26 To evaluate whether FO plus ethanol could increase collagen levels by transactivating the COL1A2 promoter, we analyzed the NFκB binding activity in electrophoretic mobility shift assays using nuclear extracts from whole livers. To determine whether the activation of NFκB might be the result of degradation of IκBα, the protein expression of IκBα in whole liver homogenates was measured by a western blot analysis (Fig. 5B). In livers from the mice fed FO plus ethanol, a marked decrease in the IκBa protein was observed (Fig. 5B). Results similar to those seen with IκBa were also seen with p50 and p65 (Fig. 5B). Nuclear localization of NFκB was increased in the FO plus ethanol group in comparison with the FO group (Fig. 5C). NFκB binding to the COL1A2 promoter was significantly higher in mice fed FO plus EtOH in comparison with mice fed the control diet (Fig. 5D). Thus, the loss of the IκBα protein coincided with NFκB activation in the mice fed FO plus EtOH. The protein/DNA complex was further characterized with competition and supershift assays. A 100-fold excess of nonradioactive NFκB, AP1, or Sp1 oligonucleotides was added to the binding reaction containing the nuclear extracts from the mice fed FO plus ethanol. The addition of the NFκB oligonucleotide completely abrogated complex formation, whereas the addition of an AP1 or Sp1 oligonucleotide had no effect (not shown). Antibodies against p50 and p65 were used for supershift assays to show the specificity of the NFκB complex (not shown).

A Coculture Model Replicates the In Vivo Activation of Collagen.

To gain further insight into the mechanism by which FO plus ethanol activate collagen I, a coculture model was designed in which primary mouse HSCs were cocultured in the presence of a pool of primary Kupffer cells, endothelial cells, and hepatocytes isolated from either COL1A2-βGal transgenic mice or wild-type mice, depending on the experiment (Fig. 6A). Primary HSCs isolated from the COL1A2-β-Gal mice were cocultured with pooled liver cells and showed a 24-fold increase in the β-Gal activity when incubated in the presence of FO plus ethanol versus FO alone (P < 0.001; Fig. 6B). Similarly, the coculture setting showed an increase in LPO (4-HHE) in the presence of FO plus ethanol (P < 0.001; Fig. 6C), suggesting a likely LPO-mediated mechanism for collagen I induction, as occurred in vivo. To validate the synergistic role of ethanol on the effects mediated by the FO diet, HSCs were incubated alone or in a coculture in the presence of FO, ethanol, or a combination of both. A Western blot analysis suggested a synergistic effect (Fig. 6D, lane 8 versus lanes 6 and 7). Furthermore, to determine the specific contribution of each series of fatty acids to the effects observed, the cocultures were incubated with palmitic acid (16:0), oleic acid (18:1n-9), arachidonic acid (20:4n-6), eicosapentaenoic acid (20:5n-3), and docosahexaenoic acid (22:6n-3) with 0-50 mM ethanol. The induction of collagen I protein expression was greater when n-3 series PUFAs were added to the cocultures and was even greater when ethanol was present.

To analyze whether the increase in COL1A2 transactivation observed in vivo could be due to the elevated binding activity of any particular transactivator (i.e., NFκB, AP1, and Sp1), a series of COL1A2-Luc deletion constructs were transfected onto HSCs cultured alone or with pooled liver cells and treated with either FO alone or FO in combination with ethanol. The luciferase activity indicated increased transactivation of the −378 and −772COL1A2-Luc constructs in the presence of FO plus ethanol in comparison with the FO-treated cells (P < 0.001; Fig. 6F).

FO plus Ethanol Increase Collagen I in Stellate Cells in a Coculture by a LPO-PKC-PI3K-Akt Mechanism.

A Western blot analysis of HSCs cultured alone or with pooled liver cells showed an increase in the expression of phosphorylated PKC (3.8 ± 0.2 versus 1 ± 0.1, P < 0.001), PI3K (2.1 ± 0.2 versus 1 ± 0, P < 0.01), phosphorylated Akt (2.3 ± 0.5 versus 1 ± 0.1, P < 0.001), and intracellular collagen I (3.1 ± 0.4 versus 1 ± 0.3, P < 0.001) in the cocultures treated with FO plus ethanol versus FO alone (Fig. 7A). These effects were almost completely prevented by the addition of vitamin E, an antioxidant that prevents LPO-derived reactions, validating the role of LPO (Fig. 7B,C). The addition of inhibitors of PKC phosphorylation (Ro 31-8425), PI3K (LY294002), and Akt phosphorylation (triciribine) prevented intracellular and extracellular collagen I up-regulation in the cocultures incubated with FO plus ethanol, lowering collagen I to levels found in the cocultures incubated with FO (Fig. 7D). Finally, to define whether the aforementioned kinases could have an effect on collagen I expression modulating NFκB binding to its consensus site on the COL1A2 promoter, electrophoretic mobility shift assays were carried out with an oligonucleotide containing the binding site for NFκB in the COL1A2 promoter in the presence or absence of vitamin E and the specific inhibitors of each of the kinases. The binding of NFκB to its consensus sequence decreased in the presence of all inhibitors as well as vitamin E (Fig. 7E). In summary, a potential link was established between LPO-derived reactions and the phosphorylation of PKC, PI3K activation, phosphorylation of Akt, increased NFκB binding, and collagen I expression in the presence of FO plus ethanol.

Figure 7.

(A) Western blot analysis of phosphorylated PKC, PI3K, phosphorylated Akt1/2/3, and intracellular collagen I in cell lysates from HSCs in a coculture incubated with a medium containing FO or FO plus EtOH. The numbers under the blots indicate arbitrary units of densitometry and are corrected by the corresponding total protein or by β-tubulin expression. The expression of each protein in the coculture with FO was assigned a value of 1. In all panels, **P < 0.01 and ***P < 0.001 for FO plus EtOH versus FO (n = 4). (B) Western blot analysis of phosphorylated PKC, PI3K, phosphorylated Akt1/2/3, and intracellular collagen I in cell lysates from HSCs in a coculture incubated with a medium containing FO or FO plus EtOH in the presence or absence of 50 μM vitamin E. The numbers under the blot indicate arbitrary units of densitometry and are corrected by the corresponding total protein or by β-tubulin expression. The expression of each protein in the coculture with FO was assigned a value of 1. (C) LPO was assessed as blotting for 4-HNE residues by the use of specific antibodies and quantified by scanning. The results are expressed as arbitrary units of densitometry and are means plus or minus the SEM (n = 6). ***P < 0.001 for FO plus EtOH versus FO and • • • P < 0.01 and • • • P < 0.001 for vitamin E–treated versus non-vitamin E–treated. (D) Western blot analysis of intracellular and extracellular collagen I expression in the presence of inhibitors of the phosphorylation of PKC, PI3K, and Akt from HSCs cocultured in the presence of FO plus EtOH. The numbers under the blot indicate arbitrary units of densitometry and are corrected by the corresponding total protein or by β-tubulin expression. The expression of each protein in the coculture with FO was assigned a value of 1. (E) Gel mobility shift assay with nuclear extracts from cocultured HSCs incubated with FO or FO plus EtOH in the absence or presence of inhibitors of phosphorylation of PKC, PI3K, and Akt as well as vitamin E. A radiolabeled probe was made with the region of the COL1A2 promoter containing both NFκB binding sites.

Discussion

Dietary fat affects the severity of liver disease. Previous studies have shown that diets containing saturated fat protect against alcohol-induced liver injury, whereas n-6 PUFAs enhance the toxic potential of alcohol.3, 4 Alcohol is one of many known causes of impaired mitochondrial long-chain fatty acid oxidation,51 resulting in fatty acid overload, by which fatty acids that accumulate in the cell are diverted into esterification (triglyceride synthesis) and extramitochondrial fatty acid oxidation.52–54 The latter involves ω-oxidation and ω-1-oxidation by CYP2E1 and CYP3A4 in the endoplasmic reticulum and β-oxidation in the peroxisomes.52, 54, 55 The fatty acid overload hypothesis indicates a role for PUFAs in the hepatotoxic sequelae associated with alcohol, that is, impaired fatty oxidation.

Whereas many studies have been carried out with n-6 series PUFAs, less is known about the role of n-3 series PUFAs in the development of alcoholic liver disease. Previous work by others suggests the following:

  • 1n-3 PUFAs elicit hypotriglyceridemic effects by coordinately suppressing hepatic lipogenesis, reducing levels of SREBP-1c, upregulating fatty oxidation in the liver and skeletal muscle through PPARα activation, and enhancing the flux of glucose to glycogen through down-regulation of HNF-4α56; the net result is the repartitioning of metabolic fuel from triglyceride storage toward oxidation, thereby reducing the substrate available for very low density lipoprotein (VLDL) synthesis.
  • 2By downregulating genes encoding proteins that stimulate lipid synthesis and upregulating genes encoding proteins that stimulate fatty acid oxidation, n-3 PUFAs are more potent hypotriglyceridemic agents than n-6 PUFAs on a carbon-for-carbon basis.57
  • 3The peroxidation of n-3 PUFAs reduces VLDL secretion by stimulating apolipoprotein B degradation.58
  • 4n-3 PUFAs may act by enhancing postprandial chylomicron clearance through reduced VLDL secretion and by directly stimulating lipoprotein lipase activity.57, 58
  • 5n-3 PUFAs compete with n-6 PUFAs, lowering the induction of proinflammatory eicosanoids such as leukotriene B4, favoring leukotriene B5 synthesis, and lowering inflammation.59, 60

These effects support the use of n-3 PUFAs as a valuable tool for the treatment of liver disease. Indeed, the supplementation of n-3 PUFAs to high-fat-diet–treated rats restored hepatic adiponectin and PPARα expression, reduced hepatic TNFα levels, and ameliorated the fatty liver and the degree of liver injury.61 However, the administration of n-3 PUFAs may have the potential downside of inducing LPO reactions and the subsequent collagen I upregulation. Multiple studies have suggested an important role for LPO in the pathogenesis of alcoholic liver disease.62–65 4-HHE, MDA, and 4-HNE are the major aldehydes generated by microsomal peroxidation of n-3 and n-6 PUFAs, respectively, both present in FO.66 They are highly toxic and have been shown by in vivo and in vitro experiments to inhibit biological functions of rat liver microsomes and mitochondria and to alter rat liver membrane structure.67, 68

Excessive collagen I accumulation is the histopathological hallmark of liver fibrosis. Central to the development and progression of fibrosis are LPO end products derived from PUFAs.21, 69 The current study focused on analyzing potential mechanisms by which the coadministration of a diet enriched in FO (mostly n-3 series PUFAs) plus ethanol, which generated abundant LPO products (i.e., 4-HHE, MDA, and 4-HNE), could increase collagen I deposition. This model may be useful in studying the occurrence of fibrogenesis in the absence of overt steatosis and inflammation.

As a first approach, transgenic mice harboring the mouse COL1A2 promoter linked to the β-gal reporter gene were fed a diet containing FO with or without ethanol, and COL1A2 promoter activation and collagen I deposition were detected. Moderate injury was observed by increased levels of transaminases, H&E staining, NEFA, triglycerides, elevated LPO and CYP2E1 levels along with depleted GSH, the activation of HSCs, increased endogenous total collagen deposition, and the induction of COL1A1 and COL1A2 transcription. Increased CYP2E1 activity and low GSH levels are factors that significantly exacerbate the cascade of LPO reactions.70 Others have shown that the increase in CYP2E1 activity secondary to ethanol administration is dependent on the type of dietary fat, with the highest levels occurring in rats fed FO and ethanol.71, 72 Our group has shown that hepatocytes overexpressing CYP2E1 increased collagen I expression in HSCs.18, 19 Therefore, alcohol-induced perturbations in the cytochrome P450-dependent metabolism of fatty acids are of considerable interest because metabolites of fatty acids generated through P450-dependent oxidative pathways, such as dicarboxylic acids and other long-chain fatty acid metabolites, are potential regulators of gene expression.73 One possible reason for the lower GSH levels found could involve effects of de novo GSH synthesis as both GCLC and GCLM were down-regulated by FO plus ethanol.

Among the potential mechanisms likely to contribute to the effects on COL1A2 transactivation and collagen I protein deposition in mice fed FO plus ethanol are the involvement of LPO-derived products in the activation of protein kinases and NFκB binding, both of which are sensitive to ROS. An analysis of stress-activated kinases and NFκB binding in total mouse liver from mice fed FO plus ethanol revealed a clear induction of phosphorylation of PKC, PI3K, and Akt1/2/3 as well as the activation of NFκB binding to its consensus sequence in the COL1A2 promoter in comparison with mice fed FO only.

To further analyze the cascade of events leading to the upregulation of collagen I in mice fed FO plus ethanol, a second approach was taken. HSCs were cocultured in the presence of the rest of the liver cells, and FO or FO plus ethanol were added to the culture medium. This model allowed the validation of the synergistic role of ethanol with FO and the identification of the n-3 series of PUFAs (20:5n-3 and 22:6n-3) as the major contributors to the collagen I up-regulation, likely mediated by LPO. It is important to note that the protective effect of 18:1n-9 in olive oil, as consumed in the typical Mediterranean diet, may have antifibrogenic properties, especially in the presence of ethanol, because of its antioxidant properties. The coculture setting replicated results of the activation of the COL1A2 promoter as assessed by β-gal activity, validating the model as a tool for carrying out mechanistic studies. HSCs were transfected with a series of deletion constructs for the COL1A2 promoter, which contains 2 NFκB binding sites located downstream from the −378 and −772 bp from the transcription start site. The −378 region of the COL1A2 promoter has been previously described by others to be sensitive to acetaldehyde and ROS,21, 22 which play a pivotal role in the development of alcoholic liver disease.16, 17, 74 There was transactivation of the COL1A2 promoter under both culture conditions with 2.5-fold and 4-fold inductions for the combined treatment over the FO treatment when the −378 and −772 COL1A2 deletion constructs were used, respectively. This result suggests a potential role for NFκB in the transactivation of the COL1A2 promoter under FO plus ethanol.

To assess whether the up-regulation of collagen I by FO plus ethanol could be elicited, at least in part, by LPO-mediated activation of kinases, we first analyzed the expression of the activated kinases in vivo using primary HSCs in a coculture and found a pattern of expression similar to that found in vivo, which indicated that all 3 kinases could play a role in the mechanism by which FO plus ethanol activate the COL1A2 promoter and increase collagen I protein. Because LPO could prompt kinase activation, the cocultures were pretreated with vitamin E. The administration of vitamin E protected them from LPO-derived reactions and prevented the increased phosphorylation of PKC, PI3K, and Akt, lowering collagen I and establishing a link between LPO and the stress-activated kinases (PKC, PI3K, and Akt) in modulating collagen I protein under the FO plus ethanol treatment. Similarly, the addition of specific inhibitors of all 3 kinases reduced collagen I expression in HSCs in a coculture treated with FO plus ethanol, validating the role for the PKC-PI3K-Akt cascade in modulating this effect. Moreover, the addition of vitamin E and of the kinases inhibitors prevented NFκB binding to the COL1A2 promoter, suggesting an essential role for NFκB binding to the COL1A2 promoter in the fibrogenic response mediated by FO plus ethanol. In summary, these results suggest that an important mechanism by which n-3 PUFAs may synergize with ethanol to increase the profibrogenic response may be mediated by elevated LPO, the activation of the PKC-PI3K-Akt pathway, and increased binding of NFκB to the COL1A2 promoter, resulting in matrix deposition. In consideration of the link between lipid peroxidation and the activation of collagen gene expression in hepatic stellate cells, it is tempting to speculate that a longer period of feeding fish oil and ethanol could lead to fibrogenesis, despite the lack of overt steatosis and inflammation.

Ancillary