Diverse regulation of NF-κB and peroxisome proliferator-activated receptors in murine nonalcoholic fatty liver

Authors

  • Laszlo Romics Jr.,

    1. Liver Center, Division of Gastroenterology, Department of Medicine, University of Massachusetts Medical School, Worcester, MA
    Search for more papers by this author
  • Karen Kodys,

    1. Liver Center, Division of Gastroenterology, Department of Medicine, University of Massachusetts Medical School, Worcester, MA
    Search for more papers by this author
  • Angela Dolganiuc,

    1. Liver Center, Division of Gastroenterology, Department of Medicine, University of Massachusetts Medical School, Worcester, MA
    Search for more papers by this author
  • Lucia Graham,

    1. Liver Center, Division of Gastroenterology, Department of Medicine, University of Massachusetts Medical School, Worcester, MA
    Search for more papers by this author
  • Arumugam Velayudham,

    1. Liver Center, Division of Gastroenterology, Department of Medicine, University of Massachusetts Medical School, Worcester, MA
    Search for more papers by this author
  • Pranoti Mandrekar,

    1. Liver Center, Division of Gastroenterology, Department of Medicine, University of Massachusetts Medical School, Worcester, MA
    Search for more papers by this author
  • Gyongyi Szabo

    Corresponding author
    1. Liver Center, Division of Gastroenterology, Department of Medicine, University of Massachusetts Medical School, Worcester, MA
    • Department of Medicine, University of Massachusetts Medical School, 364 Plantation Street, LRB 215, Worcester, MA 01605-2324
    Search for more papers by this author
    • fax: (508) 856-4770


Abstract

Fatty liver is highly sensitive to inflammatory activation. Peroxisome proliferator-activated receptors (PPAR) have anti-inflammatory effects and regulate lipid metabolism in the fatty liver. We hypothesized that fatty liver leads to endotoxin sensitivity through an imbalance between pro- and anti-inflammatory signals. Leptin-deficient, ob/ob mice and their lean littermates were challenged with single or double insults and pro- and anti-inflammatory pathways were tested on cytokine production and activation of nuclear regulatory factors NF-κB and peroxisome proliferator receptor element (PPRE). Ob/ob mice produced significantly higher serum tumor necrosis factor α (TNF-α) and interleukin (IL) 6 and showed increased hepatic NF-κB activation compared to lean littermates after stimulation with a single dose of lipopolysaccharide (LPS) or alcohol. In ob/ob mice, double insults with alcohol and LPS augmented proinflammatory responses mediated by increased degradation of inhibitory κB (IκB)-α and IκB-β and preferential induction of the p65/p50 NF-κB heterodimer. In lean mice, in contrast, acute alcohol attenuated LPS-induced TNF-α, IL-6 production, and NF-κB activation through reduced IκB-α degradation and induction of p50/p50 homodimers. PPRE binding was increased in fatty but not in lean livers after alcohol or LPS stimulation. However, cotreatment with alcohol and LPS reduced both PPRE binding and nuclear levels of PPAR-α in fatty livers but increased those in lean livers. In conclusion, our results show opposite PPRE and NF-κB activation in fatty and lean livers. PPAR activation may represent an anti-inflammatory mechanism that fails in the fatty liver on increased proinflammatory pressure. Thus, an imbalance between PPAR-mediated anti-inflammatory and NF-κB-mediated proinflammatory signals may contribute to increased inflammation in the fatty liver. (HEPATOLOGY 2004;40:376–385.)

Steatosis, a common initial pathology observed in both nonalcoholic and alcoholic liver disease, has the potential to progress to steatohepatitis and cirrhosis.1, 2 However, the mechanisms of this progression are yet to be fully understood. It has been proposed that cumulative insults (at least 2, according to the “2-hit model”) are necessary to cause progression from normal liver to steatosis and then to steatohepatitis.1 Initially, an adaptive response to ongoing metabolic derangements results in fatty liver, which then leads to a reduced capacity to withstand additional insults.3, 4 This would suggest that anti-inflammatory mechanisms are still capable of counteracting proinflammatory activation in the fatty liver to maintain a frail homeostasis. A balance between pro- and anti-inflammatory activation is fundamental in nonalcoholic steatohepatitis, where tumor necrosis factor α (TNF-α) plays a central role.1, 5 According to this model, the increased exposure of hepatocytes to TNF-α and the loss of their ability to protect themselves against TNF-α–induced cell death will lead to liver disease. It is known that disruption of the anti-inflammatory interleukin (IL) 10 gene potentiates TNF-α–induced liver injury.6 Previous studies have demonstrated increased susceptibility of the fatty liver to lipopolysaccharide (LPS)-induced injury.1, 3 LPS triggers both pro- and anti-inflammatory pathways and, depending on their balance, results in a self-limiting, transient, or prolonged inflammatory activation.7

Peroxisome proliferator-activated receptors (PPARs) are nuclear hormone receptors and transcription factors that play a role in the regulation of both inflammation and lipid metabolism.8, 9 Three different PPARs are known, each with a unique tissue distribution but overlapping functions. PPAR-α is predominantly expressed in liver, heart, and kidney, and its activation induces transcription of enzymes involved in lipid metabolism.10, 11 PPAR-γ is primarily expressed in adipocytes, hematopoietic cells, and, to a lesser extent, in liver, spleen, skeletal muscle, and intestinal cells.12 PPAR-γ is involved in adipocyte differentiation, glucose metabolism, and lipid storage.13 Increasing evidence suggests that PPARs also contribute to the anti-inflammatory response. In vivo responses to inflammatory mediators are prolonged in mice lacking PPAR-α gene compared to wild type mice.14 Other studies have demonstrated that PPAR-γ ligands can inhibit inflammatory responses by decreasing IL-6, TNF-α, IL-1β secretion, and inducible nitric-oxide synthetase (iNOS) production in macrophages and Kupffer cells.15, 16 Both PPAR-α and PPAR-γ inhibit inflammation by interfering with nuclear factor-κB (NF-κB) and activator protein 1 (AP-1) transactivation through a direct protein-protein interaction with p65 or c-Jun, respectively.17, 18 In addition, PPAR-α activators up-regulate inhibitory κB (IκB)-α in many cell types.19 Therefore, activation of PPAR-α and PPAR-γ represent important anti-inflammatory elements in the control of inflammation.

The aims of this study were (1) to test the hypothesis that nonalcoholic fatty liver disease leads to increased endotoxin sensitivity through an imbalance between pro- and anti-inflammatory pathways and (2) to investigate the mechanisms that lead to increased inflammation after double insults of LPS, a bacterial stimulation, and acute alcohol, a trigger of oxidative stress. We demonstrated that the fatty liver of leptin-deficient, ob/ob mice produced an exaggerated inflammatory response to double insults. This was associated with reduced induction of nuclear protein binding to peroxisome proliferator receptor element (PPRE) and sustained up-regulation of NF-κB activation. In normal liver of lean mice, in contrast, acute alcohol attenuated endotoxin-induced inflammatory responses through inhibition of NF-κB activation and up-regulation of the anti-inflammatory PPRE binding.

Abbreviations

TNF-α, tumor necrosis factor α; IL, interleukin; LPS, lipopolysaccharide; PPAR, peroxisome proliferator-activated receptor; iNOS, inducible nitric-oxide synthetase; NF-κB, nuclear factor-κB; AP-1, activator protein 1; IκB, inhibitory κB; IL, interleukin; PPRE, peroxisome proliferator response element; EMSA, electrophoretic gel mobility shift assay; RPA, RNase protection assay; mRNA, messenger RNA.

Materials and Methods

Animal Studies.

Adult (7-to-10-week old) female (ob/ob) mice and their lean littermates were purchased from Jackson Laboratories (Bar Harbor, ME). Animals were maintained in a temperature- and light-controlled animal facility at the University of Massachusetts Medical School on standard laboratory chow diet and water ad libitum. All animals received care in compliance with institutional requirements, and this study underwent full Institutional Animal Care and Use Committee (IACUC) committee review at the University of Massachusetts Medical School according to National Institutes of Health guidelines. Obese mice and their lean littermates were intraperitoneally injected with LPS (E. coli 0111:B4, Sigma, St. Louis, MO) at a dose of 0.5 μg/g body weight in 200 μL saline or saline alone. Ethanol was intraperitoneally administered to animals either as a single challenge or 1 hour prior to LPS administration as a combined stimulation. Obese mice received a dose of 350 μL and lean mice received 200 μL of 20% (vol/vol) concentration of ethanol or saline. These doses of ethanol resulted in comparable maximum blood alcohol levels (150–200 mg/dL) at 2 hours in ob/ob and lean mice (data not shown; Sigma). In control experiments, the highest total amount of LPS required for ob/ob mice was administered to lean animals, and it resulted in serum TNF-α and NF-κB activation comparable to the weight-adjusted dose of LPS in the lean animals (data not shown). Mice (3-5 per group) were sacrificed at 1, 2, 4, and 6 hours after stimulation, and serum and livers were collected and stored at −80°C. RNA was prepared as previously described20 and quantified by measuring absorbance at 260 nm and stored at −80°C. Preparation of cytosolic and nuclear protein extracts were carried out as previously described,21 and protein concentration of the cytosolic and nuclear extracts was determined by the Bio-Rad Dye Reagent Assay (BioRad, Hercules, CA).

Measurement of TNF-α and IL-6 Serum Levels.

Serum TNF-α and IL-6 levels were determined by specific enzyme-linked immunosorbent assays (Endogen, Woburn, MA).

Electrophoretic Gel Mobility Shift Assay (EMSA).

EMSA was performed as we previously described21 using equal amounts (5 μg) of nuclear protein and a 32P-labeled NF-κB consensus (Promega; Madison, WI) or PPRE consensus oligonucleotide sequence (Santa Cruz Biotechnology, Santa Cruz, CA). Anti-mouse p50, p65, cRel, PPAR-α, and PPAR-γ antibodies for supershift were purchased from Santa Cruz Biotechnology. A 20-fold excess of specific unlabeled double-stranded probe was added for cold competition.

RNase Protection Assay (RPA).

RNase protection assays were carried out using RiboQuant multiprobe assay system (BD PharMingen, San Diego, CA). Briefly, 32P-labeled RNA probes were transcribed with T7 polymerase using the multiprobe template set mCK-2b. RNA (10 μg) was hybridized with 3.5 x 105 cpm/μL of probe overnight at 56°C. Samples were then digested with RNase followed by proteinase K treatment, phenol:chloroform extraction, and ethanol precipitation and resolved on 5% acrylamide-bisacrylamide (19:1) urea gels. Dried gels were visualized using the Fuji FLA-5000 PhosphorImager system and analysis software (Image Gauge, Elmsford, NY).

Western Blot.

For Western blot analysis, 10 to 30 μg of protein extracts were run on 10% to 15% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred onto nitrocellulose blotting membranes,20 and then exposed to the primary antibody against IκB-α (1:200), IκB-β (1:200), PPAR-α (1:100), and PPAR-γ (1:100 [Santa Cruz Biotechnology]). Secondary antibody binding was visualized using an enhanced chemiluminescence assay (Amersham, Piscataway, NJ). The optical density of the specific protein bands was quantified by using the UVP BioImaging Systems (UVP, Inc., Upland, CA).

Statistical Analysis.

Statistical values were determined using paired Student t test (2-tailed distribution) using the Microsoft Excel program. Each experiment was performed on 3 to 6 mice.

Results

Opposite Regulation of Serum TNF-α and IL-6 Levels in Mice With Fatty and Normal Livers.

To evaluate overall inflammatory responses, serum IL-6 and TNF-α induction were tested in leptin-deficient ob/ob mice and their lean littermates after in vivo stimulation with LPS, alcohol, or their combination. LPS has been implicated as an “endogenous” trigger of inflammation in nonalcoholic liver disease.1, 2 Acute alcohol has been previously shown to attenuate inflammatory activation and to induce oxidative stress both acutely and chronically22 Alcohol stimulation was also relevant because increased gastrointestinal ethanol production in ob/ob mice has been implicated in the pathogenesis of fatty liver disease.23 Stimulation with either alcohol or LPS alone induced significantly higher IL-6 and TNF-α levels in ob/ob mice compared to lean littermates (Figs. 1A and B). To test responses after double insults, we administered acute alcohol followed by an LPS challenge 1 hour later. Alcohol significantly attenuated LPS-induced serum IL-6 and TNF-α levels in lean mice (P < .05). In LPS-stimulated ob/ob mice, in contrast, alcohol augmented IL-6 levels (P < .05) and failed to inhibit TNF-α production (Fig. 1A). The inhibitory effects of acute alcohol in lean mice were similar to that previously described in alveolar macrophages and human monocytes.20, 21, 24 These results suggested that acute alcohol sensitizes ob/ob mice with fatty liver to LPS-induced injury, but it attenuates the LPS-induced proinflammatory activation in mice with normal liver.

Figure 1.

Acute alcohol attenuates LPS-induced serum IL-6 and TNF-α levels in lean but not in ob/ob mice. Ob/ob mice and their lean littermates were treated with saline, acute alcohol, LPS, or their combination (alcohol given 1 hour prior to LPS). (A) Serum IL-6 and (B) TNF-α levels were evaluated by enzyme-linked immunosorbent assay. IL-6 data from lean mice at 4 hours substitutes for the 2 hours time point. *P < .05 versus saline control. **P < .03 versus lean littermates with same stimulation (n = 5). EtOH, ethanol.

Acute Alcohol Treatment Attenuates LPS-Induced Expression of Proinflammatory Cytokines in Lean and Not Fatty Livers.

Serum cytokine levels reflect overall changes in cytokine production by various cellular sources. In order to correlate changes in serum cytokine levels with inflammatory activation in the liver, we evaluated local cytokine production in the liver. Consistent with the inflammatory activation reflected in elevated serum TNF-α and IL-6 levels after LPS stimulation, we found significantly increased messenger RNA (mRNA) induction of proinflammatory cytokines including IL-12, IL-1β, and interleukin-1 receptor antagonist (IL-1Ra) in the fatty livers of ob/ob mice compared to lean littermates (Fig. 2A). In contrast, mRNA levels of the anti-inflammatory cytokine, IL-10, were significantly reduced in the fatty liver.6 In normal livers, double insults with alcohol and LPS attenuated LPS-induced proinflammatory cytokine mRNA induction of IL-12, IL-1β and IL-6 (Fig. 2B). Consistent with attenuation of proinflammatory cytokines, the level of the anti-inflammatory IL-10 mRNA was increased by alcohol in the LPS-stimulated lean livers (P < .02). In fatty livers, alcohol failed to attenuate LPS-induced mRNA expression of proinflammatory cytokines or to increase mRNA levels of the anti-inflammatory IL-10 (data not shown). These observations supported the hypothesis that the fatty liver responds to a single LPS challenge with increased inflammatory activation, which can be augmented by the additional insult of acute alcohol administration.

Figure 2.

Acute alcohol decreases LPS-induced proinflammatory cytokine RNA levels in normal but not fatty liver. (A) Ob/ob mice and their lean littermates were treated with saline or LPS for 4 hours, and total liver RNA levels were assessed by RPA. Three animals per stimulation are shown. Separate IL-12p40 and IL-6 bands shown were digitally enhanced from the same blot for better visualization. Bar graph shows average densitometic values after normalization of the individual bands to L32 gene expression in the same animal. (B) Lean mice were treated with saline, LPS, or a combination of acute alcohol plus LPS, and liver RNA levels were measured by RPA at 4 hours. One representative gel is shown. Enhanced images of the IL-12p40, IL-10 and IL-6 bands are shown on the side. Bar graph represents average ± SD of the densitometric values of respective cytokine bans after normalization to L32 gene expression from 5 animals. IL-1Ra, interleukin-1 receptor antagonist; EtOH, ethanol; MIF, macrophage migration inhibitory factor.

Morphological analysis of hematoxylin-eosin–stained liver samples revealed marked steatosis in livers of ob/ob mice but not in the lean littermates (Fig. 3). After LPS stimulation, recruitment of small foci of inflammatory cells was seen in livers of both ob/ob and lean mice. After double insult with alcohol plus LPS, we found minimal, if any, inflammatory cell accumulation in livers of lean animals. In contrast, there were prominent foci of inflammatory cells in the fatty livers of ob/ob mice after double insult (Fig. 3).

Figure 3.

Different liver histopathology in ob/ob and lean mice. Ob/ob mice and their lean littermates were treated with saline, acute alcohol, LPS, or their combination (alcohol given 1 hour prior to LPS), and liver samples were collected 24 hours after LPS stimulation for hematoxylin-eosin staining. Arrow heads indicate foci of inflammatory cell recruitment.

LPS-Induced NF-κB Activation Is Attenuated by Acute Alcohol in Normal but Not Fatty Livers.

NF-κB activation plays a central role both in inflammatory responses and in liver injury.1, 25 Because induction of TNF-α and IL-6 genes is partially regulated through the NF-κB sequence localized in their promoter region,26, 27 we investigated NF-κB activation in the fatty liver of ob/ob mice. Upon LPS stimulation, we found 1.7 ± 0.14 times higher NF-κB binding in fatty compared to normal livers (P < .02; Fig. 4A), and this obervation was consistent with previous reports.28, 29 Acute alcohol attenuated LPS-induced NF-κB activation in livers of lean (P < .008) but not ob/ob mice. We found 7.26 ± 1.45 times greater NF-κB binding (P < .009) upon alcohol plus LPS stimulation in fatty compared to lean livers (Fig. 4A). These changes in NF-κB were consistent with inhibition of LPS-induced inflammatory cytokine protein and mRNA induction by acute alcohol in the livers of lean and not ob/ob mice. These data supported the hypothesis that acute alcohol regulates NF-κB binding differently in fatty and normal livers and suggested that hyperelevated NF-κB activation may contribute to increased proinflammatory responses triggered by LPS and alcohol plus LPS in fatty livers.

Figure 4.

Acute alcohol prevents LPS-induced NF-κB binding by decreasing p65/p50 heterodimer activation in normal liver but not in fatty liver. (A) Ob/ob mice and their lean littermates were treated with saline, acute alcohol, LPS, or their combination for 1 hour, and NF-κB binding in liver nuclear extracts was evaluated by EMSA. One representative gel and average densitometry results from 5 animals are shown. Comp, competition. (B) NF-κB supershift analysis of liver nuclear proteins in ob/ob and lean mouse liver. (C) Percent ratio of p65/p50 and p50/p50 in the NF-κB binding complex in fatty and normal livers. Average of 5 animals is shown. *P < .007 versus lean in percent ratio of p65/50. P < .01 indicates differences in p50/p50 between LPS- and ethanol (EtOH) + LPS-stimulated lean livers; Ab, antibodies.

NF-κB DNA-binding complexes consist of different dimers of the NF-κB/Rel family.29–31 Analysis of the NF-κB/Rel complexes in supershift experiments revealed the presence of p65/p50 heterodimers and p50/p50 homodimers (Fig. 4B). Previous reports have indicated that unlike p65, p50 lacks the DNA transactivation domain and, thus, fails to induce cytokine transcription upon binding to the NF-κB DNA sequence.30, 31 Analysis of the composition of the NF-κB complexes showed that the relative proportion of the p65/p50 heterodimer within the NF-κB complex was higher in the fatty liver than in lean livers after LPS or alcohol plus LPS stimulation (P < .007; Fig. 4C). At the same time, the relative ratio of the p50/p50 homodimer was significantly increased (22.18 ± 3.93%; P < .01) compared to the p65/p50 heterodimer in lean and not fatty livers after alcohol plus LPS stimulation. Thus, preferential induction of p65/p50 heterodimers as opposed to p50/p50 heterodimers may contribute to increased NF-κB–mediated inflammatory activation in fatty livers.

NF-κB activation is regulated by means of degradation of the cytoplasmic IκB proteins, IκB-α and IκB-β.32, 33 IκB-α degradation results in rapid, transient changes in NF-κB induction, whereas IκB-β degradation is associated with prolonged NF-κB activation.32–34 We found that LPS stimulation induced a significant decrease in IκB-α levels in both ob/ob and lean livers (P < .01). Consistent with the observation that acute alcohol alone triggered inflammation in fatty and not lean livers, acute alcohol treatment reduced IκB-α levels in fatty (P < .01) but not normal livers. Combined treatment with alcohol plus LPS further increased IκB-α degradation in the fatty liver. In the normal liver, in contrast, alcohol partially prevented LPS-induced IκB-α degradation (P < .04; Fig. 5A). Furthermore, we report for the first time that cytoplasmic IκB-β levels were significantly reduced in fatty but not in lean livers after alcohol plus LPS stimulation (Fig. 5B; P < .01). Such increased degradation of IκB-β after double insults suggests a unique mechanism that may contribute to prolonged and increased NF-κB activation found in the fatty liver.

Figure 5.

Acute alcohol attenuates LPS-induced IκB-α degradation in normal liver and increases IκB-β degradation in fatty liver. Ob/ob mice and their lean littermates were treated with saline, acute alcohol, LPS, or their combination. (A) IκB-α and (B) IκB-β protein levels were determined by Western blot in liver cytoplasmic fractions 1 and 6 hours after stimulations, respectively. One representative immunoblot and average densitometry results from 3 animals are shown. *P < .05 versus saline control. EtOH, ethanol.

Different PPRE Activation in Fatty and Normal Livers After Single or Double Insults.

Increasing evidence suggests that PPAR activation mediates anti-inflammatory effects that can counteract pathways induced by NF-κB activation.14, 17 Analysis of PPRE binding in the nuclear fractions of liver extracts revealed lower baseline PPRE binding (16 ± 0.39% decrease, statistically not significant) in the fatty compared to normal livers (data not shown). After single stimulation with alcohol or LPS, we found opposite trends in PPRE binding in fatty and normal livers (Fig. 6A). In normal livers, alcohol or LPS stimulation decreased PPRE binding (2 hours, P < .05; 4 hours, P < .01). In fatty liver, in contrast, a single alcohol or LPS administration increased PPRE binding (P < .05). Opposite regulation of PPRE binding in fatty and lean livers was also evident after double insults. The combination of alcohol and LPS resulted in persistent reduction in PPRE binding compared to LPS stimulation alone in fatty livers (P < .03; Fig. 6B), whereas double stimulation resulted in increased PPRE binding in normal livers (2 hours, P < .05; Fig. 6B). These results suggested dysregulation of PPRE activation in fatty livers where PPRE binding was induced by a single insult but multiple stimuli failed to further increase nuclear protein binding to PPRE.

Figure 6.

Different regulation of PPRE binding in fatty and normal livers. Ob/ob and lean mice were treated with (A) saline or acute alcohol and LPS or (B) LPS and a combination of LPS and alcohol. PPRE binding was evaluated by EMSA in liver nuclear fractions 2 and 4 hours after stimulation. One representative EMSA and average percent changes in PPRE binding (densitometry analysis) are shown from 6 experiments. *P < .05. **P < .03. Comp, competition.

PPAR-α and PPAR-γ Nuclear Protein Levels Correlate With the Diverse PPRE Binding in Fatty and Normal Livers.

The participation of PPAR-α or PPAR-γ in the protein complexes bound to the PPRE element was evaluated in supershift experiments. Data in Fig. 7A demonstrate that both PPAR-α and PPAR-γ were involved in the PPRE binding complexes. Interestingly, PPAR-α appeared to be the predominant component in the PPRE binding complexes in the fatty livers because it was present in both the lower and higher molecular-weight complexes.

Figure 7.

PPAR-α and PPAR-γ protein levels correlate with different PPRE binding in fatty and normal livers. (A) PPRE binding complexes were assessed by supershift analysis for PPAR-α and PPAR-γ antibody given (1) 30 minutes prior to oligo, (2) at the same time, or (3) 30 minutes after the oligo. (B) PPAR-α and (C) PPAR-γ protein levels in liver nuclear extracts at 4 and 2 hours, respectively, after stimulation determined by Western blot analysis of 20 μg nuclear protein per stimulation group. Average densitometry of 3 experiments is shown. EtOH, ethanol; comp, competition.

Next, to assess whether changes in PPRE binding represented nuclear translocation of PPAR-α and/or PPAR-γ, we assessed nuclear levels of PPAR-α and PPAR-γ by Western blot analysis. Consistent with the decrease in PPRE binding in fatty livers, nuclear PPAR-α protein levels were reduced after double insult with alcohol and LPS compared to LPS stimulation alone (P < .05; Fig. 7B). Nuclear PPAR-γ levels did not decrease significantly in the same fatty livers (Fig. 7C). Consistent with the reduced PPRE binding after LPS stimulation alone in lean livers, both PPAR-α and PPAR-γ nuclear protein levels were decreased compared to saline controls (Figs. 7B and C). In lean livers, PPAR-α nuclear protein levels were up-regulated after double insult with alcohol and LPS compared to LPS stimulation alone (P < .04), and the increase in PPAR-γ was not statistically significant (Figs. 7B and C). These results suggested that changes in nuclear translocation of PPAR-α and PPAR-γ are likely to contribute to alcohol- and LPS-induced changes in PPRE binding.

Discussion

Although various mechanisms have been proposed to contribute to the increased susceptibility of the fatty liver to hepatotoxic insults, the complexity and interaction of these pathways remain to be fully understood. Here we report that ob/ob mice with fatty liver respond to a single insult of LPS or alcohol with increased proinflammatory cytokine production and NF-κB activation in the liver. Double insults with acute alcohol and LPS triggered hyperelevated proinflammatory cytokine and NF-κB activation in the fatty liver as opposed to attenuation of these markers of inflammatory activation in lean livers. Mechanisms of NF-κB activation involved diverse activation of IκB-α and IκB-β in fatty and lean livers. We also found that DNA binding of PPARs, the nuclear hormone receptors that modulate the intensity, duration, and consequences of inflammatory events,8, 9, 35 was different in fatty and lean livers. Our results suggested that PPAR activation may represent an anti-inflammatory mechanism that fails in the fatty liver under conditions of increased proinflammatory activation.

NF-κB plays a central role in the proinflammatory activation.1, 2 In the fatty liver, up-regulation of NF-κB was associated with increased susceptibility to LPS-induced stimulation.29 Our experiments revealed that an additional insult with acute alcohol further increased NF-κB and inflammatory cytokine induction in the LPS-challenged fatty but not lean livers. While chronic alcohol is known to up-regulate NF-κB binding in the liver, the effects of acute alcohol stimulation are less well characterized.36, 37 The present study demonstrated that, in the lean liver, acute alcohol inhibited LPS-induced proinflammatory responses mirrored by decreased NF-κB activation, reduced IκB-α degradation, decreased liver proinflammatory cytokine mRNA (IL-12p40, IL-6, IL-1β) induction, and attenuation of serum TNF-α and IL-6 levels. This inhibitory effect of acute alcohol on LPS-induced proinflammatory pathways in lean livers was similar to that described both in vivo and in vitro in alveolar macrophages, Kupffer cells, and human monocytes.20, 21, 24, 38 In the fatty liver, in contrast, acute alcohol triggered and augmented all of these proinflammatory responses. These significant differences in NF-κB activation between fatty and normal livers are reflective of the altered homeostasis of the fatty liver.

The present study revealed 2 novel mechanisms that directly contribute to increased NF-κB activation in the fatty and not the lean liver. First, we found that acute alcohol significantly increased the ratio of the p50/p50 homodimer at the expense of the p65/p50 heterodimer in normal but not fatty liver. It has been shown that while both p65 and p50 have nuclear localization and DNA binding sites, only p65—not p50—contains a DNA transactivation domain.30, 31 Consequently, DNA binding of the p50/p50 homodimer fails to induce gene activation.30, 31 Thus, preferential induction of the p50/p50 homodimer after acute alcohol treatment in normal liver may act as a functional inhibitor of NF-κB–driven gene activation. In contrast, preferential p65/p50 activation as found in the fatty liver would support proinflammatory activation. Second, our data indicate a unique involvement of IκB-β degradation in the fatty liver. Previous studies have demonstrated that degradation of IκB-β is associated with prolonged NF-κB activation.34 Thus, isolated increase in IκB-β degradation upon LPS stimulation in fatty but not lean livers suggest that IκB-β degradation may contribute to increased NF-κB activation. Differences in IκB-α degradation were also reflective of the increased NF-κB binding in fatty livers and the decreased NF-κB activation in lean livers after alcohol plus LPS stimulation. IκB-α expression is regulated by NF-κB and, according to recent reports, it is also induced by PPAR-α.19 Thus, in the fatty liver, reduced PPAR-α nuclear levels after multiple stimuli may contribute both to reduced IκB-α expression and increased NF-κB activation.

Our results indicate that PPARs represent a novel component of the malfunctioning anti-inflammatory mechanism of the fatty liver. Both PPAR-α and PPAR-γ exert profound anti-inflammatory activity.8, 9, 35 Mice lacking PPAR-α have prolonged inflammatory response and ear swelling with leukotriene B4 (LTB4)-induced insults.14 Total absence of the PPAR-α gene is lethal.39 Recent reports indicate increased susceptibility of PPAR-γ+/− animals to experimentally induced arthritis and inflammatory bowel disease.40 A single stimulation with acute alcohol or LPS increased PPRE binding in fatty livers, suggesting activation of this compensatory anti-inflammatory pathway. However, we found decreased PPRE binding after combined alcohol plus LPS stimulation in fatty livers, suggesting either exhaustion or inhibition of PPRE binding. Our results showing reduced PPAR-α and PPAR-γ protein levels in the nuclear extracts of fatty livers after double insults suggest faulty nuclear trafficking as a possible mechanism for reduced PPRE binding. A recent report demonstrated that anti-inflammatory mechanisms mediated by PPAR-γ involve nuclear export of RelA (p65), thereby preventing NF-κB–induced proinflammatory gene activation.41 An explanation could be differential phosphorylation of PPAR-α or PPAR-γ between fatty and lean livers that would affect their PPRE binding and transactivation capacity.42, 43 It has been shown that the ability of PPARs to regulate inflammatory responses is a result of their transactivation and transrepression capacities.35. Most of the anti-inflammatory properties of PPARs arise through their ability to antagonize NF-κB and AP-1 signaling pathways to repress expression of genes involved in inflammatory responses. These include cytokines, cell adhesion molecules, and other proinflammatory signaling molecules such as iNOS.15, 17, 44 Further, different regulation of PPRE binding in fatty and lean livers may result in different recruitment of coactivators and dissociation of corepressors upon combined stimulation with alcohol and LPS. PPARs were shown in in vitro studies to inhibit activation of various proinflammatory cascades and nuclear regulatory factors (mitogen-activated protein kinase, NF-κB, AP-1, and nuclear factor of activated T cells).14, 40 Based on the previously identified direct interactions between p65 and PPAR-α or PPAR-γ, it is conceivable that decreased PPAR nuclear protein levels may fail to counteract p65-mediated proinflammatory gene activation in the fatty liver.

It remains to be determined whether it was alcohol itself or its metabolites that mediated the observed changes in PPRE binding. However, it has been suggested that PPRE binding is inhibited by acetaldehyde.45 Both alcohol and acetaldehyde were shown to have regulatory effects on NF-κB.21, 46 It has also been shown that PPAR-α induction increases uncoupling protein 2 expression in hepatocytes, inhibiting apoptosis and increasing hepatocyte survival.47 Up-regulation of the uncoupling protein 2 enzyme in the fatty liver results in partial depolarization of the mitochondrial membranes and, upon exposure to secondary stimuli (e.g., LPS or TNF-α), leads to depletion of adenosine triphosphate and hepatocyte necrosis.3, 4 In mice with dietary steatohepatitis, PPAR-α–dependent hepatic lipid turnover plays an important role and correlates with the presence of hepatic steatosis and necroinflammation.48

The cellular sources of the observed changes in the fatty livers remain to be determined. Based on previous studies in normal livers, PPAR-α levels most likely represent changes in hepatocytes, and PPAR-γ is localized to Kupffer cells and endothelial cells.49 A recent report raises the possibility that the composition of PPAR isoforms may change in the fatty liver because analysis of mRNA expression suggested a significant increase, particularly in PPAR-γ gene expression in fatty livers of ob/ob mice.50 Nevertheless, results from our experiments failed to reveal significant differences in baseline protein expression of PPAR-α or PPAR-γ in the liver nuclear extracts.

In summary, our study suggests that upon exposure to the double insults of alcohol and LPS, the fatty liver is unable to balance proinflammatory activation due to significantly decreased anti-inflammatory PPRE activation and sustained NF-κB activation. Thus, the imbalance between pro- and anti-inflammatory pathways may contribute to the pathophysiology of fatty liver disease.

Ancillary