Potential conflict of interest: Nothing to report.
Fatty liver is vulnerable to conditions that challenge hepatocellular energy homeostasis. Lipid-laden hepatocytes highly express uncoupling protein-2 (UCP2), a mitochondrial carrier that competes with adenosine triphosphate (ATP) synthesis by mediating proton leak. However, evidence for a link between UCP2 expression and susceptibility of liver to acute injury is lacking. We asked whether absence of UCP2 protects ob/ob mice from Fas-mediated acute liver damage. UCP2-deficient ob/ob mice (ob/ob:ucp2−/−) and UCP2-competent littermates (ob/ob:ucp2+/+) received a single dose of agonistic anti-Fas antibody (Jo2). Low-dose Jo2 (0.15 mg/kg intraperitoneally) caused less serum alanine aminotransferase (ALT) elevation and lower apoptosis rates in ob/ob:ucp2−/− mice. High-dose Jo2 (0.40 mg/kg intraperitoneally) proved uniformly fatal; however, ob/ob:ucp2−/− mice survived longer with less depletion of liver ATP stores, indicating that fatty hepatocytes may benefit from lack of UCP2 during Jo2 challenge. Although UCP2 reportedly controls mitochondrial oxidant production, its absence had no apparent effect on fatty liver tissue malondialdehyde levels augmented by Jo2. This finding prompted us to determine UCP2 expression in Kupffer cells, a major source of intrahepatic oxidative stress. UCP2 expression was found diminished in Kupffer cells of untreated ob/ob:ucp2+/+ mice, conceivably contributing to increased oxidative stress in fatty liver and limiting the impact of UCP2 ablation. In conclusion, whereas UCP2 abundance in fatty hepatocytes exacerbates Fas-mediated injury by compromising ATP stores, downregulation of UCP2 in Kupffer cells may account for persistent oxidative stress in fatty liver. Our data support a cell-specific approach when considering the therapeutic effects of mitochondrial uncoupling in fatty liver disease. (HEPATOLOGY 2006;44:592–601.)
Obesity has become exceedingly prevalent in the United States, and nonalcoholic fatty liver disease (NAFLD) associated with this trend is now present in over 20% of the population.1 Although considered mostly benign, NAFLD may result in significant morbidity when featuring histological signs of inflammation and fibrosis described as nonalcoholic steatohepatitis with a potential to progress into cirrhosis and hepatocellular carcinoma.2, 3 Fatty liver is also more susceptible to acute injury such as ischemia–reperfusion and endotoxinemia.4, 5 In a pilot study on patients with biopsy-proven nonalcoholic steatohepatitis, recovery from hepatic adenosine triphosphate (ATP) depletion induced by intravenously given fructose was less efficient than in healthy controls, suggesting impaired hepatic energy homeostasis in fatty liver disease.6 Although a follow-up study on healthy volunteers could not confirm a similar connection, baseline hepatic ATP stores in these individuals inversely correlated with higher body mass index.7 These observations indicate that diminished ATP reserves in fatty liver may account for its increased vulnerability.
Uncoupling protein 2 (UCP2) is an inner mitochondrial membrane carrier with debated evolutionary role and biological function.8, 9 UCP2 has wide tissue expression, but its overall abundance is typically very low.10 UCP2 has been shown to mediate proton leak in various mammalian cells when activated by superoxide and lipid peroxidation end products.11–13 By tapping into the proton gradient, activated UCP2 may compete with ATP synthase for the electrochemical energy of mitochondria and result in altered cellular ATP levels.14, 15 UCP2 expression in healthy liver is primarily localized to Kupffer cells16; however, UCP2 becomes strikingly abundant in hepatocytes of fatty liver.17–19 Although accumulation and peroxidation of lipids in fatty hepatocytes may advance the effects of UCP2, the biological importance of these changes is not well understood.20 In addition, there is no evidence that UCP2 abundance would indeed account for the hepatocellular energy compromise observed in fatty liver.
Although activated UCP2 may interfere with mitochondrial ATP synthesis, reduction of the mitochondrial membrane potential also limits superoxide production,21, 22 concordant with recent observations that UCP2 controls oxidative stress in a variety of cells.23–25 Reports on cell protection by excess UCP2 have raised significant interest in mitochondrial uncoupling as a potential therapeutic tool. However, concurrent effects of UCP2 on synthesis of ATP and production of reactive oxygen species (ROS) leave us with ambiguity as to whether promotion of mitochondrial uncoupling in fatty liver disease would protect against oxidative liver injury or would further perturb a compromised hepatocellular energy balance during acute challenges.
Previously we found that absence of UCP2 had no apparent positive or negative laboratory and histological effects on obesity-related fatty liver in mice,19 suggesting either the lack of obvious biological action(s) associated with increased UCP2 expression or the presence of chronic compensatory mechanisms. However, the potential impact of UCP2 deficiency during acute liver injury has not been investigated. To study this problem, we used agonistic anti-Fas antibody (Jo2) to challenge ob/ob mice made deficient for UCP2 by crossbreeding with Ucp2−/− mice (ob/ob:ucp2−/−). The widely studied ob/ob mice develop massive steatosis with insulin resistance, and although no fibrosis or spontaneous progression into steatohepatitis is seen, this animal model provides an excellent tool for studying the metabolic and bioenergetic aspects of hepatic fat accumulation.26, 27 Thus, most studies on the potential link between UCP2 and fatty liver have been conducted on ob/ob mice.17–19, 28 Jo2 treatment predictably induces death receptor–mediated hepatocellular destruction and fulminant liver failure in mice.29 Recent work found that mice with fatty liver are increasingly sensitive to Fas-mediated injury.30 Here we report that the impact of Fas stimulation is attenuated in ob/ob:ucp2−/− mice in association with relative preservation of hepatic ATP stores. Our findings indicate that this beneficial effect of UCP2 ablation in ob/ob mice primarily pertains to hepatocytes, because UCP2 expression of Kupffer cells is already diminished in fatty liver. Thus, UCP2 abundance in fatty hepatocytes is not necessarily inert and may account for increased bioenergetic compromise during acute liver injury. In contrast, the primary impact of UCP2 on persistently increased oxidative stress seen in fatty liver may relate to its low expression level in Kupffer cells. As far as data from this animal model can be extrapolated, caution and a cell-specific approach is therefore advised when mitochondrial uncoupling is considered in the treatment of fatty liver disease.
Heterozygous founders of ucp2−/− mice were obtained from Dr. Bradford Lowell (Boston, MA), crossbred with ob/ob mice (Jackson Laboratories, Bar Harbor, ME), and genotyped as previously described.15 Because the availability of double homozygous offspring was much limited, 8- to 18-week-old obese female littermates of heterozygous parents that had both alleles for the ucp2 gene (ob/ob:ucp2+/+) or were nullizygous for it (ob/ob:ucp2−/−) were chosen to receive a single intraperitoneal injection of Jo2 antibody (BD PharMingen, San Diego, CA) at low (0.15 mg/kg) or high (0.40 mg/kg) doses. Control animals were injected with saline. Serum alanine aminotransferase (ALT) levels were monitored up to 24 hours unless the mice died earlier. Blood was obtained through the tail vein or by cardiac puncture. Livers were excised and snap-frozen in liquid nitrogen or processed for histological studies. The Lifespan Animal Welfare Committee of Rhode Island Hospital, Providence, RI, approved all animal experiments.
Histological Studies and Image Analysis.
Liver tissue was fixed overnight in 4% paraformaldehyde dissolved in phosphate-buffered saline at 4°C, then dehydrated, embedded in paraffin, and cut into 4-μm thickness. Hepatocellular apoptosis and necrosis was assessed by hematoxylin-eosin staining and by use of the terminal deoxynucleotidyl transferase-mediated nick-end labeling (TUNEL) assay (ApopTag Plus Peroxidase, Serologicals Corp., Norcross, GA). Labeling indices were determined without knowledge of the genotypes by counting TUNEL-positive nuclei per 100 hepatocytes. At least 2,000 hepatocytes were counted for each animal. To assess the combined extent of apoptosis and necrosis in the liver, we prepared digitized images of TUNEL slides (MicroPublisher 3.3 RTV, Qimaging, Burnaby, British Columbia). We recorded the area and intensity of peroxidase staining above a constant optical density threshold using Image Pro Plus 5.1 (MediaCybernetics, Silver Springs, MD) to calculate and express the integrated optical density in arbitrary units. Constant optical conditions were maintained along the entire morphometric evaluation.
DNA Fragmentation Assay.
DNA fragmentation was assessed by the accelerated apoptotic DNA laddering protocol,31 using slight modification. Briefly, liver tissue was homogenized in lysis buffer (50 mmol/L Tris-HCl, pH 7.5, 1% NP-40, 20 mmol/L EDTA). The lysate was pelleted at 16,000g (5 minutes, 4°C), and the supernatant was subject to one round of phenol:chloroform:isoamyl alcohol (25:24:1; pH 7.4; 0.5 mL) extraction. Apoptotic DNA fragments were precipitated from the liquid phase by adding 50 μL of 3 mol/L sodium acetate (pH 5.2), 1 μL nuclease-free glycogen (Roche, Mannheim, Germany) and 0.6 mL isopropanol. After incubation on ice for 5 minutes, the precipitated nucleic acids were pelleted by centrifugation at 12,900g (10 minutes, 4°C). After washing with 70% ethanol, the pellet was reconstituted in TE buffer, and the DNA concentration was measured by spectrophotometry. Equal amounts of DNA were digested with RnaseOne (Promega, Madison, WI). After digestion, 5× Orange G dye was added to each sample, and the apoptotic DNA fragments were resolved by 1.8% TAE agarose gel electrophoresis.
We used a commercially available kit to measure serum ALT levels (Infinity ALT Reagent, Sigma, St. Louis, MO). To assess generation of ROS in the liver, we measured the tissue content of malondialdehyde (MDA), a terminal product of lipid peroxidation.32 Briefly, snap-frozen liver tissue was homogenized in 0.9% (w/v) saline containing 1% (v/v) butylated hydroxytoluene to determine MDA levels by a commercially available kit (Bioxytech MDA-586, OxisResearch, Portland, OR). To assess liver tissue ATP content, snap-frozen liver tissue was homogenized in a sample buffer (20 mmol/L glycine, 50 mmol/L MgSO4, 4 mmol/L EDTA, pH 7.75) and a luciferin-luciferase assay was performed (ATPLite Luminescence ATP Detection System, Perkin-Elmer Life Sciences, Boston, MA). Unless otherwise specified, all chemicals were obtained from Sigma.
Western Blot Analysis.
Liver tissue lysates were prepared in a mixture of 30 mmol/L Tris (pH 7.5), 2 mmol/L EDTA, 150 mmol/L NaCl, 0.1% SDS, 1% NP-40, 10% glycerol, 0.5% Na-deoxycholate containing protease inhibitors (Complete Mini EDTA-free Protease Inhibitor Cocktail, Roche, Indianapolis, IN). Protein concentrations of the lysates were determined using the BCA Protein Assay Reagent Kit (Pierce Biotechnology, Rockford, IL). Protein extracts were fractionated by 10% SDS-PAGE and transferred to a nitrocellulose membrane (Perkin-Elmer). Immunoblots were performed using polyclonal rabbit antibodies (1:1,000) against the murine Fas receptor (CD95) (Abcam, Cambridge, MA). Secondary antibodies were conjugated with horseradish peroxidase and immunoblots detected by ECL (Perkin-Elmer). Equal loading was confirmed using beta-actin antibody.
Liver Perfusion and Cell Separation.
Parenchymal cells (PC) and nonparenchymal cells (NPC) were isolated from the livers of 8- to 12-week-old UCP2-competent ob/ob mice and their lean (wt/wt) littermates, by a modified two-step in situ collagenase perfusion method.33 Briefly, the animals were anesthetized, laparotomy was performed, the portal vein was cannulated and perfused for 10 minutes with Krebs-Ringer buffer (KRB; containing 20 mmol/L glucose and 0.1 mmol/L EGTA), followed by perfusion for an additional 10 minutes with KRB (supplemented with 20 mmol/L glucose, 1.37 mmol/L CaCl2, and 150 U/mL type I collagenase, Worthington, Lakewood, NJ) at a flow rate of 2 to 3 mL/min. To obtain PC fraction, the livers were removed from the carcass, minced, filtered through a 70-μm cell strainer (BD Falcon, San Jose, CA), and centrifuged at 50g for 5 minutes to pellet the hepatocytes. To obtain NPC fraction, the filtered liver tissue digest was subjected to postperfusion digestion in KRB containing collagenase (100 U/mL), pronase (0.02%), and DNAse (0.005%) for an additional 10 minutes. After filtering and pelleting the PC fraction, the supernatant containing the crude NPC fraction was centrifuged at 1,000g for 10 minutes and the pellet resuspended in 20% iodixanol (Opti-Prep, Axis-Shield, Norton, MA), with a 17% iodixanol and Gray's balanced salt solution (GBSS) layered on top of each other according to the manufacturer's instructions. After centrifugation at 500g for 15 minutes, cells floated into the interface between 17% iodixanol, and GBSS were harvested and resuspended in GBSS. This NPC suspension was washed twice, checked for viability using trypan blue, flash-frozen in liquid nitrogen, and kept at −80°C until use. For subsequent real time PCR, we extracted total RNA from the PC and NPC fractions using the PARIS Kit (Ambion, Austin, TX) and performed reverse transcription as described.34
Laser Capture Microdissection.
Mice were inoculated through the tail vein with 200 μL sterilized India ink diluted 1:100 in normal saline 24 hours before removal of the liver. Laser capture was performed using an Arcturus Autopix automated LCM system of the Molecular Pathology Core Laboratory at Rhode Island Hospital. Kupffer cells were identified by their inert carbon content and captured within 1 hour of cutting the sections. Approximately 500 Kupffer cells were captured from each sample on CapSure macro LCM caps. RNA was extracted from the captured cells using a PicoPure RNA isolation kit (Arcturus, Mountain View, CA). Spin columns were eluted in 10 to 15 μL nuclease-free water or elution buffer from the kit, and RNA was protected with RNAse free DNAse 1 (Qiagen, Valencia, CA). We assessed RNA quality by a 6,000 Total RNA Picochip in an Agilent Bioanalyser, which provides qualitative evaluation down to 200 pg/μL. For subsequent real-time PCR, 50% to 100% of the entire extract was reverse transcribed with a Sensiscript RT kit (Qiagen), intended for use with less than 50 ng RNA per reaction.
Real-time quantitative PCR was performed using an iCycler iQ Multi-Color Real Time PCR Detection System (Bio-Rad, Hercules, CA). To detect UCP2, we created a reference plasmid by amplifying a 929-bp-long fragment of the mouse UCP2 gene (primers: forward, 5′-GATCCATATGGTTGGTTTCAAGGCCAC-3′; reverse, 5′-ATGAAGCTTTCAGAAAGGTGCCTCCC-3′) and subsequently inserting it into the pCR2.1 vector by TA cloning. Successful cloning was confirmed by sequence analysis. We used the TATA box-binding protein (TBP) as reference gene.35 The full-length mouse TBP gene (957 bp) was amplified (primers: forward, 5′-ATGGACCAGAACAACAGCCTTC–3′; reverse: 5′-CTATGTGGTCTTCCTGAATCCCTTT-3′) and cloned into pCR2.1. Serial dilutions of the UCP2 and TBP plasmids were used to create standard curves. Thermal cycling conditions to amplify UCP2 from samples involved 45 cycles, with denaturation at 95°C for 15 seconds, annealing at 60°C for 30 seconds and extension at 72°C for 30 seconds using 0.4 μmol/L (final concentration) each of the intron-spanning primers (forward, 5′-AGCCCTTGACTCTCCCCTTG-3′; reverse, 5′-GCATTGCAGATCTCATCACTTTCC-3′; amplicon length, 51 bp). For TBP, the PCR reaction mix contained 0.6 μmol/L each of the intron-spanning primers (forward, 5′-ACTTCGTGCAAGAAATGCTGAA-3′; reverse, 5′-TGTCCGTGGCTCTCTTATTCTCA-3′; amplicon length, 75 bp). Each sample was run in duplicates and normalized using its TBP content as endogenous reference gene, and data were expressed in arbitrary units as relative abundance of UCP2 mRNA over TBP.
We expressed data as mean ± SEM and performed statistical analysis with unpaired Student t test or ANOVA when multiple comparisons were made. Association between categorical groups was evaluated by the Fisher's exact probability, Mann-Whitney U test, and binomial exact calculations. Differences with calculated P values of less than .05 were regarded as significant.
Results and Discussion
UCP2 Deficiency Affects Survival, Serum ALT Levels, and Liver Injury in Jo2-Treated ob/ob Mice.
Administration of low-dose Jo2 antibody caused no lethality among ob/ob:ucp2−/− mice and all but one ob/ob:ucp2+/+ mouse survived by 24 hours (Fig. 1A, left panel). By contrast, all mice in both groups died within 24 hours when exposed to high-dose anti-Fas antibody (Fig. 1A, right panel). However, even within the short course of fulminant liver failure, ob/ob:ucp2−/− mice survived longer (571 ± 81 minutes vs. 390 ± 50 minutes, P < .05), indicating that UCP2 deficiency may confer partial protection against the consequences of Fas stimulation.
To assess Jo2-induced acute liver injury, we monitored serum ALT levels up to 24 hours or until the animals became moribund, whichever occurred earlier. After the administration of low-dose Jo2, ALT levels in ob/ob:ucp2+/+ mice were higher at each examined time point when compared with ob/ob:ucp2−/− mice (Fig. 1B, left panel). As expected, treatment with high-dose Jo2 caused more rapid and severe ALT elevation, but to a lesser extent in ob/ob:ucp2−/− mice (Fig. 1B, right panel). Histological evaluation of acute liver damage indicated limited changes after the treatment of ob/ob:ucp2+/+ and ob/ob:ucp2−/− mice with low-dose Jo2 (Fig. 1C). Hematoxylin-eosin staining identified many apoptotic hepatocytes, primarily located in zone 2, more so in ob/ob:ucp2+/+ (Fig. 1C, panel 1) than in ob/ob:ucp2−/− mice (Fig. 1C, panel 2). In contrast, high-dose Jo2 treatment caused massive hemorrhagic necrosis in the livers of ob/ob:ucp2+/+ mice when assessed at the time of death (Fig. 1C, panel 3). Necrosis appeared less extensive, and sparing of the periportal areas was more apparent in ob/ob:ucp2−/− mice (Fig. 1C, panel 4). These results correlate with the survival data and indicate that UCP2 deficiency attenuates Fas-mediated acute liver injury in ob/ob mice.
UCP2 Deficiency Reduces Fas-Mediated Liver Cell Death in ob/ob Mice.
To further assess the impact of UCP2 deficiency on Fas-mediated liver injury, we determined and compared the extent of hepatocellular apoptosis in different treatment groups (Fig. 2A). Importantly, we identified a small number of apoptotic hepatocytes by TUNEL assay in the livers of saline-treated animals (Fig. 2A, panel 1) and found that baseline apoptosis rates were significantly lower in livers of untreated ob/ob:ucp2−/− mice (Fig. 2A, panel 2). Low-dose Jo2 treatment markedly increased the number of apoptotic hepatocytes, but to a lesser degree in ob/ob:ucp2−/− mice (Fig. 2A, panel 4) than in ob/ob:ucp2+/+ mice (Fig. 2A, panel 3). This finding is concordant with lower serum ALT levels seen in ob/ob:ucp2−/− mice in response to low-dose Jo2. High-dose Jo2 treatment of ob/ob:ucp2+/+ mice induced dramatic changes in the appearance of liver with numerous apoptotic cells engulfed by confluent areas of necrotic hepatocytes that also stained positive by the TUNEL assay (Fig. 2A, panel 5). Many liver cells within these areas displayed a diffuse staining pattern consistent with the necrotic process as opposed to the dark-staining and shrunken nuclei that are typical of apoptosis. Numerical evaluation (Fig. 2B) confirmed that hepatocellular apoptosis rates in ob/ob:ucp2−/− mice are significantly lower at baseline and after low-dose Jo2 treatment. Although the number of apoptotic cells in the livers of ob/ob:ucp2−/− mice treated with high-dose Jo2 was lower (Fig. 2A, panel 6), this difference was not significant (P = .086).
Because most liver tissue slides after high-dose Jo2 treatment displayed mixed features of apoptosis and necrosis, we assessed the extent of hepatocellular death by image analysis in which both apoptotic and necrotic cells were included, and we found that high-dose Jo2 treatment caused significantly less damage in the liver of ob/ob:ucp2−/− mice when evaluated by this method (Fig. 2C). Because apoptosis is energy-dependent and necrosis may prevail as a form of cell death in energetically compromised hepatocytes,36 these findings suggest that hepatocytes of ob/ob:ucp2−/− mice may retain more ATP, accounting for relatively higher fraction of cells dying through apoptosis. Nonetheless, assessment of DNA fragmentation, a hallmark of the apoptotic process, clearly showed that DNA laddering is diminished in the liver tissue of ob/ob:ucp2−/− mice when compared with ob/ob:ucp2+/+ mice after treatment with both low- and high-dose Jo2 (Fig. 2D). Because upregulation of Fas contributes to increased Fas-mediated liver injury in mice with diet-induced obesity,30 we assessed Fas expression by immunoblot analysis in livers of ob/ob:ucp2+/+ and ob/ob:ucp2−/− mice and found no difference (Fig. 2E), indicating that altered Fas expression could not explain the effects of UCP2 deficiency.
UCP2 Deficiency Results in Limited Loss of Hepatic ATP Content in Jo2-Treated ob/ob Mice.
To assess the impact of UCP2 on liver tissue energy homeostasis, we measured hepatic ATP content of ob/ob mice exposed to Jo2. Because UCP2 in the liver of ob/ob mice is expressed several folds higher than in lean littermates,17, 19 hepatocellular ATP synthesis may become impaired in ob/ob:ucp2+/+ mice, leading to increased vulnerability during acute challenges. We found that hepatic ATP content in saline-treated ob/ob:ucp2−/− mice and ob/ob:ucp2+/+ controls was similar, indicating that UCP2 abundance in fatty liver does not compromise ATP synthesis under resting conditions (Fig. 3A). This is not surprising in light of the increasing evidence that ambient UCP2 has no perceptible biological effect unless activated.9 Interestingly, hepatic ATP content remained essentially unaltered after treatment with low-dose Jo2, suggesting correction by hepatocellular homeostasis and in line with the unanimous survival of these animals. By contrast, hepatic ATP content decreased dramatically after treatment with high-dose Jo2, and UCP2 deficiency had a significant impact on this process. Thus, residual hepatic ATP content remained approximately threefold higher in the livers of moribund ob/ob:ucp2−/− mice. These results indicate that liver ATP stores are better preserved in the absence of UCP2 and support the notion that increased UCP2 expression contributes to increased vulnerability of fatty liver under certain conditions.
UCP2 Deficiency and Oxidative Stress in Fas-Mediated Liver Injury of ob/ob Mice.
Because regulation of mitochondrial ROS production as a major function of UCP2 is increasingly accepted,37, 38 we analyzed the effect of UCP2 deficiency on oxidative stress in livers of Jo2-treated ob/ob mice. As predicted, increasing doses of Jo2 induced significant elevation in the degree of lipid peroxidation assessed by accumulation of MDA in the liver of ob/ob:ucp2+/+ and ob/ob:ucp2−/− mice (Fig. 3B). Surprisingly, however, UCP2 deficiency had no discernible impact on liver tissue MDA levels, suggesting that altered oxidative stress is not a major modulator of Fas-mediated liver injury in ob/ob:ucp2−/− mice. This finding is at variance with reports in which reduced presence or absence of UCP2 led to increased oxidative stress in a variety of cells under various conditions.25, 39, 40 Therefore, we addressed this problem in subsequent experiments.
Differential Changes in UCP2 Expression of Hepatocytes and Kupffer Cells in ob/ob Mice.
Liver tissue expresses the rather ubiquitous UCP2 mainly in Kupffer cells and has little or no UCP2 present in hepatocytes under healthy conditions.16 This pattern of cell-specific UCP2 expression appears to change in fatty liver associated with diet-induced or genetically determined obesity in which hepatocytes display markedly increased amounts of UCP2 mRNA and protein.17, 28 UCP2 expression is decreased in peritoneal macrophages of obese rodents,28 although whether a similar reduction of UCP2 occurs in Kupffer cells associated with obesity and fatty liver remains unknown. We speculated that downregulation of UCP2 expression in Kupffer cells of ob/ob mice could provide an explanation for the lacking effect of UCP2 deficiency on liver tissue oxidative stress induced by Jo2. We reasoned that UCP2 ablation might have little further impact on the function of Kupffer cells, a major source of ROS production in the liver, if these cells have an already diminished presence of UCP2 in ob/ob mice.
To determine whether changing UCP2 expression follows a cell-specific pattern in fatty liver, we obtained fractions of parenchymal cells (PC) and non-parenchymal cells (NPC) by in situ liver perfusion of UCP2-competent ob/ob mice and their wild-type littermates (wt/wt) and compared UCP2 expression in these cellular fractions. Analysis of UCP2 expression in the PC fraction showed several-fold higher UCP2 mRNA (Fig. 4A), in line with previous observations in which exposure of primary hepatocytes to large amounts of fatty acids caused UCP2 upregulation.41 By contrast, UCP2 mRNA levels were significantly lower in the NPC fraction obtained from ob/ob livers when compared with lean controls (Fig. 4B). Because the NPC fraction also contains non-Kupffer cells, such as sinusoid endothelial cells and stellate cells in which the presence or absence of UCP2 has not been reported, we also analyzed UCP2 expression in Kupffer cells identified by India ink phagocytosis and collected by laser capture microdissection. To avoid the effect of potential contamination from non-Kupffer cells, we corrected UCP2 mRNA to the transcript levels of Kupffer cell–specific CD68. Similar to the crude NPC fraction, UCP2 expression in Kupffer cells was significantly reduced in ob/ob mice when compared with lean control (Fig. 4C).
Toward a Model of Cell-Specific Modulation of UCP2 in Fatty Liver.
Our results indicate that downregulation of UCP2 previously demonstrated in peritoneal macrophages of ob/ob mice28 also occurs in Kupffer cells of these animals. Although the sample size and low UCP2 protein abundance did not allow us to conduct Western blot analysis, the changes seen in Kupffer cell UCP2 mRNA levels associated with fatty liver are potentially very important and reported for the first time to our knowledge. In vitro inhibition of UCP2 in rat liver NPC fractions results in increased ROS production,40 a finding consistent with Kupffer cell activation. Downregulation of UCP2 in Kupffer cells, rather than its upregulation in hepatocytes, may be a key contributor to increased oxidative stress in fatty liver of ob/ob mice. This process, however, may not escalate in ob/ob:ucp2−/− mice if the impact of UCP2 ablation is negligible on Kupffer cells with already diminished UCP2 expression. Accordingly, UCP2 deficiency has no significant effect on liver tissue MDA content in Jo2-treated ob/ob mice. By contrast, the effect of UCP2 ablation is considerable in fatty hepatocytes, where UCP2 deficiency appears to spare ATP stores and attenuate sensitivity of ob/ob:ucp2−/− mice to Fas-mediated injury.
Altogether, our findings obtained in ob/ob mice are best explained with a model in which cell-specific regulatory changes of UCP2 expression have their separate contribution to the pathogenesis of fatty liver (Fig. 5). Thus, diminished UCP2 in Kupffer cells would primarily increase ROS output and promote oxidative stress, whereas excess UCP2 in hepatocytes would primarily impair ATP production and hepatocellular energy balance. Because UCP2 requires specific activators,13 its amplified presence in fatty hepatocytes may only manifest on acute challenges. When this occurs, however, fatty hepatocytes may benefit from the lack of UCP2 as seen in Jo2-treated ob/ob:ucp2−/− mice. Whether other animal models of fatty liver exhibit the paradigm of cell-specific transcriptional regulation of UCP2 as described in ob/ob mice remains unclear. Little is known about hepatic UCP2 expression in human NAFLD and in these experimental conditions. Based on our findings, however, the therapeutic value of stimulating UCP2 expression in fatty liver remains questionable unless it is specifically directed to prevent decreased UCP2 function in Kupffer cells. Although the latter seems a desirable goal, the negative transcriptional regulation of UCP2 in macrophages, including Kupffer cells, awaits elucidation.
Note added in proof.
Since the submission of this manuscript, genipin, the active component of a Chinese/Japanese herbal medicine, has been identified as a specific inhibitor of UCP2 (Zhang et al., Cell Metabolism 2006; 3:417-427), while earlier work showed that Fas-mediated apoptosis in mouse liver is suppressed by genipin (Yamamoto et al., Gastroenterology 2000; 118:380-389). Thus, our work provides a link between these findings and indicates that genipin and its derivatives may have a therapeutic use in fatty liver.