fax: (323) 442-5425
Liver Biology and Pathobiology---Pathobiology
Role of TNF-α in ethanol-induced hyperhomocysteinemia and murine alcoholic liver injury
Article first published online: 13 AUG 2004
Copyright © 2004 American Association for the Study of Liver Diseases
Volume 40, Issue 2, pages 442–451, August 2004
How to Cite
Ji, C., Deng, Q. and Kaplowitz, N. (2004), Role of TNF-α in ethanol-induced hyperhomocysteinemia and murine alcoholic liver injury. Hepatology, 40: 442–451. doi: 10.1002/hep.20309
- Issue published online: 13 AUG 2004
- Article first published online: 13 AUG 2004
- Manuscript Accepted: 28 APR 2004
- Manuscript Received: 5 FEB 2004
- United States National Institute of Alcohol Abuse and Alcoholism (NIAAA). Grant Numbers: P50AA11999, R01AA014428
- Robert E. and May R. Wright Foundation. Grant Number: 263
We previously reported a link between ethanol-induced elevation of homocysteine, endoplasmic reticulum (ER) stress, and alcoholic liver injury in the murine model of intragastric ethanol feeding. We studied the role of TNFα in this setting by using TNFR1 knockout mice (C57 BL/6). There was a 7.4-fold increase of homocysteine in wild-type and a 6-fold increase in TNFR1 knockout mice with intragastric alcohol exposure for 4 weeks. Plasma TNFα increased in the wild-type (18.4 ± 3.3 pg/mL vs. 8.4 ± 1.3 pg/mL (control)) and in the knockouts (12.9 ± 1.4 pg/mL vs. 7.2 ± 1.6 pg/mL (control)). Similar extent of fatty liver was observed in both types. Increased ALT was observed in both groups. Necroinflammatory foci were increased significantly in ethanol-fed knockouts but not to the same extent as in the ethanol-fed wild type. Increase of hepatic apoptosis and reduction of S-adenosyl-L-methionine was detected in both types of animals fed ethanol. ER stress demonstrated by RT-PCR of mRNA of selective ER stress markers GRP78, CHOP, and SREBP1 was increased equivalently in both types of mice. Betaine administration decreased ER stress in conjunction with attenuation of the elevated plasma homocysteine in both types of animals. Betaine increased hepatic S-adenosyl-L-methionine by 28 fold in the knockouts and by 24-fold in wild type. In conclusion, TNFα makes a moderate contribution to the ALT elevation, necroinflammation, apoptosis, a small contribution to the fatty liver and no contribution to hyperhomocysteinemia and ER stress in intragastric alcohol fed mice. (HEPATOLOGY 2004;40:442–451.)
The pathogenesis of alcoholic liver disease is complex and not completely understood. We recently described the association of ethanol-induced hyperhomocysteinemia and ER stress and proposed a role for these effects of ethanol in the pathogenesis of experimental alcoholic fatty liver, necroinflammation, and apoptosis.1 An important aspect of this work was the demonstration that feeding with betaine, a methyl donor, markedly attenuates changes in homocysteine level, expression of ER stress response, and liver injury.
A widely held view is that tumor necrosis factor α (TNF-α) plays a major and obligatory role in the early stages of experimental alcoholic liver injury2–7. It has been proposed that ethanol induces gut permeabilization and endotoxin accumulation, which induces already primed Kupffer cells to release increased amounts of TNF-α.5, 6 The most compelling evidence in support of the importance of this pathway is that the fatty liver and necroinflammation observed in response to intrgastric ethanol feeding in mice has been reported to be completely abrogated in tumor necrosis factor receptor 1 (TNFR-1) receptor knockout mice.5 Therefore, any new evidence, such as our homocysteine/ER stress data, must be interpreted with and conform to the obligatory roles of TNF-α in alcohol-induced injury. In our previous work, we observed that betaine feeding, which prevented hyperhomocysteinemia, fatty liver, and injury, did not affect the up-regulation of TNF-α and CD14 (a molecular marker of Kupffer cell activation).1 This leads us to conclude that the increases in homocysteine and ER stress were not upstream of TNF-α, but most likely downstream; that is, TNF-α either directly or indirectly leads to hyperhomocysteinemia and/or ER stress. Theoretically, however, parallel independent pathways involving TNF-α and homocysteine would explain the findings but would not fit well with the general view that TNF-α must have an obligatory role. Therefore, to confirm the predicted role of TNF-α, we assessed homocysteine, ER stress, and liver injury in TNFR-1 knockout mice versus control mice fed intragastric ethanol.
Materials and Methods
Male TNFR-1 knockout (C57BL/6-Tnfrsf1a) and male C57BL/6 mice were purchased from Harlan (Indianapolis, IN). The intragastric ethanol infusion model has been described previously.1 Briefly, at 6–8 weeks of age, mice were aseptically operated upon under general anesthesia with ketamine and xylazine for implantation of a long-term gastrostomy catheter. After 1 week of acclimatization with infusion of a control high-fat diet, ethanol infusion was initiated at a dose of 18 g/kg/d, which was increased by 1.5 g every 2 days until it reached 29 g/kg/d.
Thereafter, the dose was increased by 0.4 g every 2 weeks. At the initial ethanol dose, total caloric intake derived from a diet and ethanol was set at 533 Cal/kg. The caloric percentages of ethanol, dietary carbohydrate (dextrose), protein (lactalbumin hydrolysate), and fat (corn oil) were 24.3%, 15.7%, 25%, and 35%, respectively. The highest ethanol dose at the end of 4 weeks accounted for 34.4% of calories. Adequate vitamin and salt mix was included at the recommended amounts by the Committee on Animal Nutrition of the National Research Council (AIN-76A, 4.42 g/L and 15.4 g/L, respectively; Dyets, Inc., PA). In some experiments, both ethanol and control groups were simultaneously fed with 1.0% (w/v) betaine (anhydrous; Sigma, St. Louis, MO). The animals were treated in accordance with the National Institutes of Health publication Guide for Care and Use of Laboratory Animals.
Hematoxylin-Eosin Staining and Terminal Deoxynucleotidyl Transferase-Mediated dUTP Nick End Labeling.
Mice were usually sacrificed 1.5 hours after removal of the gastrostomy catheter. At the time of sacrificing, small pieces of liver tissue were harvested and fixed immediately in 10% buffered formalin phosphate and 3% paraformaldehyde (Sigma). After paraffin embedding, 5-μm transverse sections were prepared and stained with hematoxylin-eosin. Liver sections were coded and fat was graded on a scale of 0–4, with 0 being normal liver architecture.1, 8, 9 Apoptotic hepatocytes were detected using terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) procedures with a TACS TdT Kit (R&D Systems, Inc., Minneapolis, MN); numbers were counted in 5 high-power fields. Clusters of inflammatory cells replacing areas of parenchyma in hematoxylin-eosin–stained sections were counted as necroinflammatory foci and number of apoptotic hepatocytes was also assessed in 5 high-power fields. Images of liver sections were captured with a Nikon Eclipse E600 microscope equipped with a digital camera (SPOT, Diagnostic Instruments, Sterling Heights, MI).
Confocal Fluorescence Microscopy.
Liver tissue was fresh frozen to ensure optimal cutting temperature (O.C.T., Sakura, Tokyo, Japan), sectioned at 5 μm, and then mounted on glass slides. The specimens on the slides were fixed in cold acetone for 10 minutes and washed in phosphate-buffered saline (PBS) three times with 5 minutes of gentle shaking per wash. The fixed specimens were blocked with PBS containing 5% bovine serum albumin and 2% secondary antibody serum for 30 minutes and washed 3 times in PBS. They were then incubated with primary antibody-rat against 160 kd glycoprotein expressed in mouse macrophages (F4/80) (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) in PBS containing 1.5% bovine serum albumin for 60 minutes and washed three times with PBS. Rabbit antiactivated caspase 3p20 (CM1; IDUN Pharmaceuticals, La Jolla, CA) and DAPI (Santa Cruz Biotechnology) were also used for apoptotic cell staining. Antibody binding was detected with FITC-conjugated secondary antibody and the Texas Red-conjugated goat anti–rabbit IgG (Southern Biotechnology Associates, Birmingham, AL). Microfilaments along the cell periphery were stained with rhodamine-phalloidin (1:100 dilution, 30 minutes). The stained tissues were examined with a Nikon PCM2000 confocal laser-scanning microscope, equipped with both argon blue laser and HeNe green laser and corresponding filters to give green and red signals, respectively. Images were compiled using Simple PCI 3.6 software (C-Imaging Systems, Cranberry Township, PA). The Kupffer cells were counted using a ×60 objective of the microscope.
RNA Isolation Analysis and One-step Reverse-Transcriptase Polymerase Chain Reaction.
Total hepatic RNA was isolated from fresh liver tissues or cultured primary hepatocytes using a NucleoSpin RNA II Purification kit from ClonTech (Palo Alto, CA) following the manufacturer's instructions and with an addition of 500 units of an RNase inhibitor (RNAguard, Amersham Phamacia Biotech, Piscataway, NJ) to each starting material (usually 300 mg of liver tissue or 5 × 106 cultured cells). RNA was stored at −80°C until used. The One-step reverse-transcriptase polymerase chain reaction (RT-PCR) kit from ClonTech was applied to examine the level of messenger RNA (mRNA), which allowed complementary DNA synthesis and PCR to be performed in a single buffer. The primer sequences used were as follows:
Betaine-homocysteine methyltransferase (BHMT): 5′-GCGTGAGCCAGACGCCTTCATACCTTAG-3′, 5′-CCTTTCTGGGGCCAACTCCTCTGCAATC-3′
Cystationine β-synthase (CBS): 5′-TCGCCTTTGCCCGCATGCTCATCGCAC-3′, 5′-GATGTGAGAGAGTGTGCCCAGCGTGTC-3′
Methionine synthase (MS): 5′-CCGAGGGATGGAAGCCATTCGAGAAGCA-3′, 5′-GTGGCCAACAGCCTTCTTCATGACACGG-3′
Sterol regulatory element-binding protein (SREBP-1): 5′-AGCTCGACCCCACACCTATCCCTCGCTACC-3′, 5′-CTACTTTGACTTTTCCAAACTTTATTTTCA-3′
CD14: 5′-CTG CAG CAG TGG CTA AAG CCT GGA CTC A-3′, 5′-CCA CTT GGG GCA GCT CAT CTG GGC TAG-3′
TNFR-1 was checked according to the protocol provided by the Jackson Laboratory.10 The PCR primer pairs for 78-kd glucose-regulated protein (GRP-78) and C/EBP homologous protein (CHOP) and growth arrest and DNA damage-inducible protein 153 (GADD153) were purchased from ClonTech. The PCR optimal cycle number for each gene was determined empirically to obtain detectable but nonsaturating PCR product. To determine the fold-change in mRNA abundance, the intensity of the unknown sample was determined with the PhosphorImage (Molecular Dynamics, Sunnyvale, CA) following the manufacturer's instructions; the relative expression was compared and normalized to the expression of β-actin in that same sample. The mean fold increase and SEM were calculated from analyses of 4–6 mice per group. The absolute value for basal gene expression between genes was not compared.
Proteins were extracted according to the method reported previously.1 Briefly, 50 μg of protein/lane was run in 12% denatured polyacrylamide gel and electrophoresed on a nitrocellulose membrane for detection by primary and secondary antibodies. The goat polyclonal antibody (sc-1050) raised against GRP-78, mouse monoclonal antibody (sc-7351) against GADD153, rabbit polyclonal antibody (sc5627) against caspase 12, and rabbit polyclonal antibody (sc-8984) against SREBP-1 were purchased from Santa Cruz Biotechnology. After the incubation with primary antibody, the membrane was washed and incubated with corresponding horseradish peroxidase–labeled secondary antibody from Santa Cruz Biotech or alkaline phosphatase–labeled secondary antibody from Cell Signaling Technology (Beverly, MA) for 45 minutes. Proteins were visualized using LumiGLO Reagent (Cell Signaling Technology) on CL-Xposure films (PIERCE, Rockford, IL) at an optimized time point. Relative signal intensity was quantified with the PhosphorImage.
Plasma alanine aminotransferase (ALT) and homocysteine levels were analyzed using methods described previously.1 The levels of triglycerides and cholesterol in mouse plasmas and livers were determined using the Sigma Diagnostics Triglyceride and Infinity Cholesterol Reagent. Mouse plasma TNF-α levels were quantified using commercial enzyme-linked immune assay kits from BD Biosciences (San Diego, CA). Concentrations of ethanol in blood were determined using the Sigma Diagnostic Alcohol Reagent.
BHMT Enzyme Activity and Hepatic S-adenosyl-L-methionine, S-adenosylhomocysteine, and Glutathione.
BHMT enzyme activity was determined according to the method described by Finkelstein and Mudd.11 One unit of BHMT enzyme activity was defined as the conversion of one nmol of betaine-14CH3 by one mg of BHMT enzyme extract in one minute. Hepatic S-adenosyl-L-methionine (SAMe) and S-adenosylhomocysteine (SAH) were determined with high performance liquid chromatography (HPLC).12 Briefly, liver tissues (300 mg) were homogenized in 500 μL of 0.5 M HClO4 (perchloric acid [PCA]). The homogenates were centrifuged at 14,000 rpm for 5 minutes. To 200 μL of the supernatant 10 μL of the 10N KOH was added, mixed, and recentrifuged. One hundred μL of the resultant supernatant was injected into the HPLC. One hundred μM of 50 μM SAMe or SAH standards were injected into the HPLC for reference. Hepatic glutathione (GSH) was measured using the DTNB-glutathione reductase recycling assay.13
Assessment of Cell Viability and Cell Death by Cytochemical Staining of Apoptotic Cells.
Isolation of hepatocytes was performed as described previously.1 Hepatocytes on the culture dish were doubly stained with Sytox green (1 μmol/L; Molecular Probes, Eugene, OR) and Hoechst 33258 dye (8 μg/mL; Sigma) for 30 minutes at 37°C. Quantitation of total and apoptotic cells was performed according to a previously described method.14, 15
Experiments were performed routinely with 4–6 mice per group with values presented as mean ± SD. All the studies were replicated with representative data shown. Statistical analysis was performed using Student's t test for unpaired data or ANOVA and the Tukey-Kramer Multiple Comparisons Test post hoc. A P value < .05 was considered significant.
Characterization of Liver Injury in Alcoholic Mice: Wild Type Versus TNFR-1 Knockout.
To investigate the importance of TNF-α in experimental alcoholic liver disease, we used TNFR-1 knockout animals in our intragastric ethanol infusion model. The absence of TNFR-1 in knockout animals was confirmed by RT-PCR using specific primers for TNFR-1 (Fig. 1a). After 4 weeks of intragastric feeding of a high-fat diet, both the wild-type and knockout animals that received ethanol exhibited significant increased numbers of hepatic macrophages (Fig. 1b) and increased CD-14 expression (see Fig. 1a). However, the increase in the number of cells was less in knockout mice compared with wild-type mice (Fig. 1c). Plasma TNF-α levels increased significantly in ethanol-fed mice (2.2-fold in wild-type mice and 1.8-fold in TNFR-1 knockout mice), and again the increase in TNF was modestly decreased in knockout mice versus wild-type mice fed ethanol (Fig. 1d).
Average weight gains of control-fed wild-type and knockout mice were 5.6 g and 5.8 g, respectively; those of alcohol-fed wild-type and knockout mice were 6.3 g and 6.0 g, respectively. No significant difference of weight gain was found between wild-type and knockout mice and between alcohol-fed and alcohol/betaine-fed animals (Table 1). In alcohol-fed wild-type animals, a significant increase was detected in liver weight relative to body weight, plasma ethanol concentrations, serum ALT, and histological features of fatty liver with scattered necrosis and apoptosis (Fig. 2; see Table 1), which were similar to the observations we reported previously.1 Increases of liver weight to body weight, plasma ethanol levels, ALT, fatty liver, and apoptotic (TUNEL-positive) cells were also observed in alcoholic TNFR-1 knockout animals. Apoptosis was confirmed by active caspase 3 staining, which was absent in pair-fed controls and present in an equivalent amount in wild-type and knockout mice fed ethanol. There was no significant difference between wild-type and knockout animals in the plasma ethanol levels, and betaine treatment did not have a significant effect on blood ethanol levels. However, the data in Table 1 demonstrate that ALT in the alcoholic knockout mice was approximately half of that in the wild-type mice; the amount of liver triglycerides was reduced by 25% in TNFR-1 knockouts mice compared with wild-type mice; and the number of inflammatory foci and apoptotic cells in alcohol-fed wild-type animals was reduced by 35% and 50%, respectively, compared with that in TNFR-1 knockouts. These data suggest that TNF-α contributed modestly to fat accumulation and hepatic inflammation and apoptosis in alcoholic mouse liver.
|Parameter||Wild Type||TNFR-1 Knockout|
|Control||EtOH||Control||EtOH||EtOH + Betaine|
|Initial liver weight (g)||23.7 ± 1.3||24.2 ± 2.2||23.3 ± 1.7||23.2 ± 1.4||23.6 ± 0.6|
|Final liver weight (g)||29.3 ± 5.3||30.5 ± 2.3||29.1 ± 1.6||29.2 ± 2.0||29.8 ± 1.0|
|Liver/Body (%)||0.05 ± 0.01||0.08 ± 0.01*||0.05 ± 0.01||0.08 ± 0.01*||0.06 ± 0.01†|
|Steatosis score||0.23 ± 0.44||3.10 ± 0.54*||0.17 ± 0.36||2.63 ± 0.54*‡||0.83 ± 0.68†|
|Blood ethanol levels (mg/dL)||14 ± 8.6||309 ± 37*||6.3 ± 2.1||298 ± 31*||292 ± 25*|
|Triglycerides (mg/mg protein)||0.08 ± 0.021||0.42 ± 0.071*||0.07 ± 0.02||0.34 ± 0.07*‡||0.17 ± 0.07†|
|ALT (U/L)||12.43 ± 3.35||127.34 ± 65.55*||15.4 ± 5.5||59.57 ± 30.19*§||44.3 ± 33.75†|
|Necroinflammatory foci∥||0.15 ± 0.38||1.2 ± 0.68*||0.2 ± 0.41||0.67 ± 0.50‡||0.17 ± 0.50|
|Apoptosis¶||0.62 ± 0.65||6.8 ± 4.0*||0.27 ± 0.46||3.27 ± 2.4*‡||1.17 ± 1.9|
|Homocysteine (μM)||3.02 ± 0.86||22.3 ± 2.84*||3.17 ± 0.67||19.46 ± 4.05*||3.17 ± 0.98|
Effects of TNF-α on Homocysteine Metabolism in Ethanol-Fed Mice.
Hyperhomocysteinemia was observed in both alcohol-fed wild-type and TNFR-1 knockout animals. Plasma homocysteine increased 7.4-fold in alcoholic wild-type mice and 6.1-fold in alcoholic TNRF-1 knockout mice (see Table 1). The difference of plasma homocysteine between TNFR-1 knockout and wild-type animals was not significant, indicating a minimal contribution of TNF-α to hyperhomocysteinemia induced by alcohol. Betaine, a methyl donor used in the methylation of homocysteine to methionine, significantly reduced liver/body ratio, fatty liver, ALT, and cell death and reduced more than 85% of the elevated plasma homocysteine in TNFR-1 knockout mice and 75% in wild-type ethanol-fed mice (Table 2).
|Parameters||Wild Type||TNFR-1 Knockout|
|Control||EtOH||EtOH + Betaine||Control||EtOH||EtOH + Betaine|
|SAMe (μmol/gfw)||0.11 ± 0.02||0.07 ± 0.03||2.69 ± 1.39*||0.12 ± 0.05||0.11 ± 0.03||3.43 ± 0.46*|
|SAH (μmol/gfw)||0.05 ± 0.01||0.04 ± 0.01||0.14 ± 0.04*||0.08 ± 0.01||0.06 ± 0.02||0.10 ± 0.02|
|GSH (μmol/gfw)||5.54 ± 0.48||4.90 ± 0.53||4.95 ± 0.59||5.75 ± 0.50||5.24 ± 0.22||5.20 ± 1.05|
|Homocysteine (mM)||3.02 ± 1.13||21.3 ± 3.14*||5.33 ± 1.41†||3.25 ± 0.7||21.7 ± 4.88*||3.27 ± 0.98‡|
|BHMT (U)||8.16 ± 1.94||7.12 ± 1.85||9.33 ± 2.40||8.31 ± 1.84||7.77 ± 1.39||8.75 ± 2.68|
To know whether the protective role by betaine is due to decreased homocysteine or to increased SAMe and whether or not SAMe has a role in modulating TNF-α biosynthesis, liver SAMe, SAH, and GSH were examined. Table 2 shows that a significant reduction of the liver SAMe level in wild-type mice and a slight reduction of SAMe level in TNFR-1 knockout animals were observed in response to ethanol exposure. Simultaneous administration of betaine in the ethanol-fed animals increased liver SAMe levels by more than 24-fold in wild-type mice and by more than 28.6-fold in TNFR-1 knockout mice. Betaine also increased SAH by 1.8-fold in wild-type animals. SAH levels tended to be lower in response to ethanol exposure, but no significant reduction was detected in either wild-type or TNFR-1 knockout animals. Interestingly, significant changes of liver GSH levels were not detected in the betaine-treated animals, suggesting that betaine and SAMe do not exert a protective role by increasing the antioxidant GSH in this experimental model.
To investigate possible changes in gene expression contributing to the elevated plasma homocysteine levels in the alcoholic mice, we examined the mRNA expression of three major genes that might affect homocysteine metabolism: MS, BHMT, and CBS. Figure 3a shows RT-PCR results of samples from wild-type mice. mRNA abundance for BHMT as well as MS was reduced at 2 and 4 weeks in ethanol-fed mouse liver compared with that in control-fed mouse liver. No reduction was detected for CBS mRNA at 4 weeks after ethanol feeding. These results indicate that ethanol affects gene expression of the homocysteine metabolizing genes, and a significant but slight reduction in BHMT mRNA was observed after 4 weeks of ethanol feeding. Similar results were obtained with the TNFR-1 knockout mice at 4 weeks after ethanol feeding (Fig. 3b). Because changes in mRNA levels may not correspond to enzyme activities, and because others have already reported that enzyme activity of MS was reduced,16, 17 we measured the BHMT activity and found no significant difference between pair-fed control and ethanol-fed wild-type or TNFR-1 knockout animals (see Table 2). These results indicate that the regulation of BHMT by ethanol most likely occurs at a transcriptional level and that TNF-α has a minimal role in the in vivo regulation of homocysteine metabolizing genes. Overall, this data along with published work16, 17 suggests that down-regulation of MS is the major mechanism for increased homocysteine.
Effects of TNF-α on ER Stress Response in Ethanol-Fed Mice.
To study the role of TNF-α in the ER stress response, we examined the effects of alcohol feeding ± betaine on the expression of selective markers of ER stress in TNFR-1 knockout mice as compared to wild-type mice (Fig. 4). In both wild-type and TNFR-1 knockout animals, increased mRNA of GRP78 and GADD153 was detected in response to ethanol exposure for 4 weeks; mRNA of SREBP-1 was increased significantly in ethanol-fed wild-type mice and showed a similar trend in knockout mice that did not reach significance; no consistent increase or decrease of SREBP-2 mRNA was obtained in response to ethanol (data not shown). In betaine-treated mice, the levels of mRNA of all the ER stress markers remained normal (see Fig. 4, lane 6). Immunoblot analysis confirmed that CHOP/GADD153, a marker of ER stress, was increased in both wild-type and TNFR-1 null mice in response to ethanol feeding and was reversed by betaine feeding in both types of animals (Fig. 5).
To further assess the effect of homocysteine, ER stress agents, and TNF-α on induction of ER stress and proapoptotic responses, we treated primary cultured mouse hepatocytes for 36 hours and examined ER stress marker CHOP and caspase 3 (Fig. 6). Tunicamycin plus brefeldin increased CHOP and active caspase 3 (Fig. 6a) and induced approximately 19% apoptosis (Fig. 6b). Homocysteine was less potent but did increase CHOP and caspase 3 and caused approximately 5.5% apoptosis (fourfold above control). The positive control and homocysteine also both exhibited decreased procaspase 12, which is indicative of activation cleavage. TNF-α alone did not affect any of these parameters. Interestingly, ER stress in response to TNF-α plus homocysteine was not different from homocysteine alone with respect to increased CHOP and caspase 3 active cleavage product, decreased procaspase 12, or apoptosis, suggesting that TNF-α did not alter the ER stress response to homocysteine.7
TNF-α has been implicated in the pathogenesis of alcoholic liver disease.2–5 We reported previously that ER stress may contribute to fatty liver and injury in the murine intragastric ethanol feeding model and suggested that increased homocysteine may be a key factor.1 When homocysteine was lowered by feeding with betaine, the animals were protected against fatty liver and injury from ethanol. However, betaine feeding did not prevent up-regulation of TNF-α or CD14, suggesting that its effects were downstream of Kupffer cells and TNF-α. Because of the profound protection against ethanol in TNFR-1 knockout mice, we tested the hypothesis that TNF-α is involved in the causation of the hyperhomocysteinemia, ER stress, and injury in ethanol fed mice.
In examining the pathological features of alcohol on the liver, we observed some protection in TNFR-1 knockout mice. However, these effects were modest and partial. There was a small decrease in the steatosis and triglyceride accumulation and an approximately 50% decrease in ALT, necroinflammatory foci, and apoptotic cells. There was also a modest decrease in Kupffer cell macrophage infiltration. These data are much less impressive than previous reports that suggested nearly complete protection.5 The supplier and strain of mice, feeding protocol, and diet composition are virtually identical in our studies compared with prevous work in TNFR-1 null mice fed ethanol. Nevertheless, we did see a minimal decrease in fatty liver and a modest decrease in injury, suggesting a contribution of TNF-α to injury in this model.
The contribution of TNF-α to homocysteine and ER stress was negligible. No difference in homocysteine levels and expression of ER stress genes was observed in ethanol-fed wild-type mice versus TNFR-1 knockout mice. Thus, we conclude that the cause of these changes in response to ethanol is not related to the effects of TNFR-1. Furthermore, betaine was as effective in reversing these changes in TNFR-1 knockout mice as in wild-type mice.
We also observed no differences in changes in homocysteine metabolizing gene expressions in knockout mice. Our results indicate that ethanol feeding, independent of TNF-α, decreases mRNA of MS and BHMT—the latter only slightly, however. The inhibitory effects of ethanol on MS gene expression and activity have been published previously.17–22 Our results, in which BHMT showed no significant change in enzyme activity, are different from data from rats fed a Lieber-DiCarli diet, which lead to a compensatory increase in BHMT activity.18 Although our findings suggest that the predominant cause for hyperhomocysteinemia is down-regulation of MS, the mechanism has yet to be identified. Furthermore, future studies will need to examine the availability of cofactors and substrates for these enzymes and metabolic flux to fully understand the molecular pathogenesis of hyperhomocysteinemia.23–29
To further assess the possible role of TNF-α in ER stress, we examined ER stress in hepatocyte culture. TNF-α alone did not induce ER stress, whereas homocysteine did. Furthermore, TNF-α did not potentiate ER stress or apoptosis induced by homocysteine. The additive effects of homocysteine and TNF-α on hepatocyte apoptosis in vitro suggest that cytotoxicity of homocysteine is independent of TNF-α signaling. Thus, alcohol-induced increased production of TNF-α and elevation of homocysteine may both contribute to the development of alcoholic liver damage.
An important caveat of the present work is the mechanism of protection by betaine. Betaine appears to have a dual effect of lowering homocysteine and raising SAMe. Thus, at present we cannot be certain as to whether one or both of these effects are important. A modest decrease in SAMe was observed in ethanol-fed wild-type mice but not in knockout mice. This suggests that decreased SAMe is not a factor in hyperhomocysteinemia or ER stress, because these were not altered in knockout mice. Moreover, because the SAMe/SAH ratio was not altered by ethanol feeding, the importance of the small change in SAMe is unclear. Nevertheless, the marked increase in SAMe in the liver of betaine-treated mice could contribute to lowering homocysteine (by activating CBS) or exerting other unknown protective effects. Against the former (i.e., lowering homocysteine) is the surprising finding that GSH levels were not increased. We would anticipate that SAMe induced flux through the transsulfuration pathway would increase GSH. Ultimately, more work is needed to prove the mechanism of protection by betaine, to link homocysteine to ER stress, and to link ER stress to liver injury. We cannot exclude that betaine and/or increased SAMe leads to protection by a mechanism independent of homocysteine or ER. Nevertheless, the data at present are consistent with the hypothesis that lowering homocysteine prevents ER stress and alcoholic liver disease.
In conclusion, we have observed modest protection against alcoholic necroinflammatory and apoptotic liver injury in TNFR-1 knockout mice, suggesting an important contribution of other pathways. TNF-α is not involved in the hyperhomocysteinemia and ER stress in alcohol-fed mice and makes a small contribution to fatty liver. Thus our data indicate that the pathogenesis of alcoholic liver disease is complex and its initiation cannot be explained simply by a single proximate pathway involving TNF-α. It appears that increased homocysteine and ER stress represent a parallel pathway independent of TNF-α.
The authors would like to thank The Animal Core of the USC/UCLA Research Center for Alcoholic Liver and Pancreatic Diseases for providing mice of intragastric infusion model; the Morphology Core for TUNEL staining; and the Cell Biology Core for confocal image processing, isolation of mouse hepatocytes, and HPLC analysis.