Retinoic acid receptor α dominant negative form causes steatohepatitis and liver tumors in transgenic mice

Authors


Abstract

Although attention has focused on the chemopreventive action of retinoic acid (RA) in hepatocarcinogenesis, the functional role of RA in the liver has yet to be clarified. To explore the role of RA in the liver, we developed transgenic mice expressing RA receptor (RAR) α– dominant negative form in hepatocytes using albumin promoter and enhancer. At 4 months of age, the RAR α– dominant negative form transgenic mice developed microvesicular steatosis and spotty focal necrosis. Mitochondrial β-oxidation activity of fatty acids and expression of its related enzymes, including VLCAD, LCAD, and HCD, were down-regulated; on the other hand, peroxisomal β-oxidation and its related enzymes, including AOX and BFE, were up-regulated. Expression of cytochrome p4504a10, cytochrome p4504a12, and cytochrome p4504a14 was increased, suggesting that ω-oxidation of fatty acids in microsomes was accelerated. In addition, formation of H2O2 and 8-hydroxy-2′-deoxyguanosine was increased. After 12 months of age, these mice developed hepatocellular carcinoma and adenoma of the liver. The incidence of tumor formation increased with age. Expression of β-catenin and cyclin D1 was enhanced and the TCF-4/β-catenin complex was increased, whereas the RAR α/ β-catenin complex was decreased. Feeding on a high-RA diet reversed histological and biochemical abnormalities and inhibited the occurrence of liver tumors. These results suggest that hepatic loss of RA function leads to the development of steatohepatitis and liver tumors. In conclusion, RA plays an important role in preventing hepatocarcinogenesis in association with fatty acid metabolism and Wnt signaling. (HEPATOLOGY 2004;40:366–375.)

Retinoids, most notably retinoic acids (RA), exert a wide variety of profound effects on vertebrate development, differentiation, and homeostasis.1–4 In the liver, retinoids have been reported to be involved in many pathobiological conditions, including regeneration, fibrosis, and cancer.5–7 Recently, supplementation with a retinoid analog has been shown to suppress the second primary tumor in patients with hepatocellular carcinoma (HCC).8 This finding is very impressive, because the prognosis of HCC is still poor and depends on the rates of tumor recurrence and occurrence of second primary tumors.9 However, the mechanism through which retinoids exert a chemopreventive effect is not fully understood. Therefore, it is very important to clarify the functional roles of RA in the liver.

RA functions are mediated by members of the nuclear receptor superfamily that act as ligand-dependent transcription factors.1 Receptors for RA consist of heterodimers of retinoic acid receptors (RARs) and retinoid X receptors (RXRs). RARs and RXRs are encoded by at least three distinct genes: RAR-α, -β, and -γ and RXR-α, -β, and -γ, respectively. The RAR/RXR heterodimers bind to the polymorphic cis-acting response elements of RA target genes and exert a variety of effects. To explore the functions of RA in vivo, knockout mice were generated10–12 that exhibited growth deficiency, early lethality, skeletal malformation, and testis degeneration. Saitou et al.13 found that the dominant negative form of RARα (RARE) can suppress the activities of endogenous RAR/RXR heterodimers. As a result, investigation into the physiological function of RA has been enabled by circumventing problems with embryonic lethality and increasing the likelihood of observing the effects of inhibition of retinoid signaling pathways in specific tissues.14–15 To establish the functional role of RA in the liver, we examined pathological changes of liver tissues and underlying biochemical abnormalities of RARE transgenic mice.

Abbreviations

RA, retinoic acid; RAR, retinoic acid receptor; HCC, hepatocellular carcinoma; RXR, retinoid X receptor; RARE, retinoic acid receptor α; BrDU, bromodeoxyuridine; 8-OHdG, 8-hydroxy-2′-deoxyguanosine; CPT, carnitine palmitoyltransferase; VLCAD, very long acyl-CoA dehydrogenase; LCAD, long acyl-CoA dehydrogenase; HCD, 3-hydroxyacyl-CoA dehydrogenase; AOX, acyl-CoA oxidase; BFE, bifunctional enzyme; PPAR, peroxisome proliferator-activated receptor; HCV, hepatitis C virus.

Materials and Methods

Animals.

Four lines of albumin-RARE transgenic mice were developed as previously described.16 We concentrated on male mice of the TG.943 line, which demonstrated the greatest abnormalities on initial screening for histological changes. Mice were fed a CE-2 diet (CLEA Japan, Tokyo, Japan), which contains 0.5 mg/kg (w/w) all-trans retinoic acid. The high-RA diet contains 50.5 mg/kg (w/w) all-trans retinoic acid (Sigma-Aldrich Japan, Tokyo, Japan) in a CE-2 diet. For the high-RA diet experiment, 20 transgenic mice began feeding on a high-RA diet at 3 weeks of age; 8 of the 20 mice were sacrificed at 4 months of age, and the other 12 were sacrificed at 12 months of age. The daily intake of high-RA diet per mouse was almost equal to that of the CE-2 diet. The mice were kept under pathogen-free conditions and were maintained in a temperature-controlled room with a 12-hour light/dark illumination cycle. All animals received humane care in accordance with study guidelines established by the Tottori University Subcommittee on Laboratory Animal Care.

Histological Analysis of Tissues, Tumors, and Hepatocyte Proliferation.

The fixed tissues were used in 4% paraformaldehyde, but oil red O staining was performed with frozen sections. The degree of hepatic steatosis was classified into 4 grades: 0 = no or few fat droplets in the lobule; 1 = fat droplets restricted to zone 1; 2 = fat droplets in zones 1 and 2; 3 = numerous fat droplets in zones 1 and 2. For immunohistochemical analysis, sections were incubated with mouse anti–proliferating cell nuclear antigen monoclonal antibody (Novacastra Laboratories, Newcastle, UK) and rabbit anticatalase polyclonal antibody, then stained with the avidin–biotin–peroxidase complex (Vector Laboratories, Burlingame, CA). HCC was diagnosed according to criteria described previously.17 For bromodeoxyuridine (BrDU) experiments, 50 mg/kg of BrDU (Sigma-Aldrich Japan) was used as previously described.18

Electron Microscopy.

After perfusion with 2.5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4), liver tissues were postfixed in 0.1% osmium tetroxide for 2 hours. They were dehydrated in ethanol of increasing concentrations and finally embedded in Epon 812 resin. Ultrathin sections made on an ultramicrotome (Reihert Ultracut UCT, Leica, Austria) were stained with lead citrate and uranyl acetate and were observed with a Hitachi-7100 transmission electron microscopy (Hitachi Ltd, Tokyo, Japan).

Real-Time Reverse-Transcriptase Polymerase Chain Reaction of Messenger RNA.

Total RNA was extracted from liver tissues using ISOGEN (Nippongene Co., Toyama, Japan) and was digested with DNase (Nippongene Co.). Complementary DNA was synthesized using an RNA polymerase chain reaction kit (TaKaRa Bio, Inc., Kyoto, Japan). Real-time reverse-transcriptase polymerase chain reaction was performed five times with a LightCycler using a SYBR Green I Kit (Roche Diagnostics, Mannheim, Germany) according to the manufacturer's instructions.19 The primers amplifying the genes coding the enzymes are described in Table 1. Reported copy numbers were the geometric average of five determinations derived from independent polymerase chain reactions performed from the same complementary DNA. The values were reported as the normalized quotient, which is derived by dividing the copy number of each gene by the GAPDH copy number.

Table 1. Primers of the Genes Coding for Enzymes Involved in Fatty Acid Metabolism
Genes5′-primer3′-primer
  1. Abbreviations: MCAD, middle chain acyl CoA dehydrogenase; ECH, enoyl-CoA hydratase; KCT, 3-ketoacyl-CoA thiolase; CYP4a10, cytochrome p4504a10; CYP4a12, cytochrome p4504a12; CYP14, cytochrome p4504a14; GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

CPT ITCCCCACTCAAGATGGCAGAGGCTCTTCCGTGTGGCTCAGGGGTTTAC
CPT IIACCTCAGTGCACAGAAGCACCAGGTAGGCTCCATAC
VLCADTGCAAGGGCTGTATGGCCGGCACATGACCTTGCCAG
LCADGTGGACAGCTGTCTGCAGCCCCCCCTTTTCCAGGTT
MCADCGATGCATCACCCTCGTGCTTCCTTGACAAGCCGCC
ECHGAGCAGTCGGCAAATCGCAGAAGACGGGGTGAGCCT
HCDTACACCCTTGGTGCCTTGTACACCCTTGGTGCCTTG
KCTTGAGTGGCATCCCAGAGACCTGTACCATCCCTCTCCAGACAC
AOXACTCGAAGCCAGCGTTACACTCGAAGCCAGCGTTAC
BFECGTACCTTGGCAATGGCTCGTACCTTGGCAATGGCT
CYP 4a10TCCTGCTGCTGGTGAAAGAGTGGAAGGCTGGGGTTA
CYP 4a12CTGGAAGCCCAGCAGAAGGTGTGAGTCCTATGAAAGAGTGCC
CYP 4a14CCACAGGGACATGCAGATTAGCACACAGAGCTCGGAAGACC
PPAR-αTGTGATCGGAGCTGCAAGCATCTTGGCCACAAGCGT
PPAR-βTTTTGGCAGCTGGGGCGTCCCTGGGTCTGCTTT
PPAR-γACGAGCTGGGGGATTCTGTTTGCTGGCATCCCTCGC
GAPDHACCCAGAAGACTGTGGATGGCACATTGGGGGTAGGAACAC

Fatty Acid β-Oxidation Activity.

Fatty acid β-oxidation activity was measured as described previously.20 It was done by using three kinds of 105-cpm radioactive fatty acid substrates (55 mCi/mmol; American Radiolabeled Chemicals, St. Louis, MO): [1-14C] behinic acid (C22:0), [1-14C] palmitic acid (C16:0), and [1-14C] lauric acid (C12:0).

H202 Measurement.

Liver samples were homogenized (1%w/v) in lysis buffer (0.2 M Tris-HCl, pH 8.0, 0.1 M EDTA, 2% sodium dodecyl sulfate containing 20 mmol/L NaN3 to inhibit endogenous catalase), and H2O2 was measured using the phenol red method.21

Quantitation of 8-Hydroxy-2′-Deoxyguanosine in the Liver.

DNA was isolated from the liver using a QIAmp DNA Mini Kit (Qiagen, Valencia, CA). Quantitative analysis of 8-hydroxy-2′-deoxyguanosine (8-OHdG) was accomplished using LC/MS according to the methods described previously using an HP-1100 system (Hewlett Packard Company, Palo Alto, CA).22

Analysis of Liver β-Catenin, c-myc, and Cyclin D1 Contents.

Equal amounts of protein homogenates (100 μg) from RARE transgenic and wild-type mice were separated via sodium dodecyl sulfate polyacrylamide gel electrophoresis, and proteins were transferred to polyvinylidene difluoride membrane. This was followed by immunoblot analysis with anti–β-catenin (Cell Signaling Technology, Inc., Beverly, MA), anti–c-myc (Santa Cruz Biotechnologies, Santa Cruz, CA), anti–cyclin D1 (Santa Cruz Biotechnologies), and anti–β-actin (Santa Cruz Biotechnologies) antibodies. Membranes were incubated with horseradish peroxidase–conjugated secondary antibodies followed by detection with enhanced chemiluminescence (Amersham Bioscience Corporation, Piscataway, NJ).

Immunoprecipitation of TCF-4 and RAR-α and Western Blot Analyses.

Equal amounts of precleared lysate proteins (500 μg) were immunoprecipitaed overnight with anti–TCF-4 (Upstate Cell Signaling Solutions, Charlottesville, VA) and anti–RAR-α (Santa Cruz Biotechnologies) antibodies at 4°C, followed by 2 hours of incubation with 100 μL (50% vol/vol) of protein A Sepharose. The immunoprecipitates were analyzed using Western blot analysis with anti–β-catenin antibody (Cell Signaling Technology, Inc.).

Statistics.

Values were expressed as means ± SD. Means were compared using the Mann-Whitney U test. The incidence of tumor development was compared using the Fisher exact test. A P value of less than .05 was considered to be significant.

Results

Development of Steatohepatitis.

The growth of RARE transgenic mice was normal, and they did not show any retardation or abnormalities of organs other than the liver. The body weights of transgenic mice were similar to those of wild-type mice up to 15 months of age; however, body weights of transgenic mice were lighter than those of wild-type mice at 18 months of age. The liver/body weights in RARE transgenic mice were heavier than those in wild-type mice at all time points. At 4 months of age, RARE transgenic mice exhibited severe and diffuse microvesicular fatty changes (Fig. 1A). The appearance of cytoplasm of some liver cells filled with numerous fat droplets was hydropic. The typically vacuolated appearance of lipid-laden hepatocytes was confirmed by demonstration of increased intrahepatic lipid on oil red O staining of frozen sections (Fig. 1B). The lipid deposition stained in red was diffusely distributed in liver lobules of transgenic mice. Semiquantitative evaluation of fatty deposition revealed that, of 8 transgenic mice, 2 exhibited grade 2 hepatic steatosis and the other 6 exhibited grade 3 hepatic steatosis, whereas of 8 wild type mice, 1 exhibited grade 1 and the other 7 exhibited grade 0 (Table 2). Focal necrosis of hepatocytes was scattered in the lobule (Fig. 1C). The sizes of focal necrosis were diverse. The infiltrating cells consisted mostly of polymorphonuclear neutrophils (Fig. 1D). Serum alanine aminotransferase levels in RARE transgenic mice were approximately 11-fold higher than those observed in wild-type mice (Table 3).

Figure 1.

Histopathological findings of the liver from RARE transgenic mice. (A) Severe and diffuse microvesicular steatosis at 4 months of age (original magnification ×100). (B) Increased intrahepatic lipid demonstrated by oil red O staining of frozen sections (original magnification: ×400). (C) Focal necrosis of hepatocytes scattered in the liver lobule (original magnification: ×100). (D) Inflammatory response consisting mostly of polymorphonuclear neutrophils (original magnification: ×200).

Table 2. Grades of Hepatic Steatosis in Three Groups
GradeTg (n = 8)Wt (n = 8)Tg + RA (n = 8)
  1. Abbreviations: Tg, retinoic acid receptor α dominant negative form (RARE) transgenic mice; Wt, wild-type mice; Tg + RA, RARE transgenic mice fed the high-RA diet (3 months of age).

0076
1012
2200
3600
Table 3. Serum Alanine Aminotransferase, Fatty Acid β-Oxidation, H2O2, and 8-OHdG Contents
 TgWtTg + RA
  • NOTE: Data are expressed as mean ± SD of 6 mice at 3 months of age.

  • *

    P < .01 compared with wild-type mice.

  • P < .05 compared with wild-type mice.

  • P < .01 compared with RARE transgenic mice.

  • §

    P < .05 compared with RARE transgenic mice.

  • Abbreviations: Tg, retinoic acid receptor α dominant negative form (RARE) transgenic mice; Wt, wild-type mice; Tg + RA, RARE transgenic mice fed the high-RA diet (3 months of age).

Serum alanine aminotransferase (IU/L)353 ± 120*32 ± 641 ± 24
Fatty acid β-oxidation (nmol/min/liver)   
 C12:021.4 ± 3.136.2 ± 5.829.5 ± 6.5
 C16:09.1 ± 1.518.4 ± 3.510.9 ± 1.1
 C22:07.7 ± 1.36.1 ± 0.26.2 ± 0.3
H2O2 content (nmol/mg protein)20.2 ± 18.14.4 ± 2.85.9 ± 2.1§
8-OHdG content (per 105 dG)1.10 ± 0.360.78 ± 0.120.78 ± 0.14§

The proliferating activity of hepatocytes was assessed via proliferating cell nuclear antigen expression and BrDU incorporation. The hepatocytes positive for proliferating cell nuclear antigen increased in RARE transgenic mice approximately 20-fold compared with wild-type mice (2.0 ± 0.1% vs. 0.1 ± 0.05%, respectively; P < .01) (Fig. 2A). The labeling index of BrDU in transgenic mice was much higher than that in wild-type mice (1.1 ± 0.2% vs. 0.04 ± 0.01%, respectively; P < .01) (Fig. 2B). Expression of catalase, the enzyme specifically localized to peroxisome, was greatly enhanced in hepatocytes from RARE transgenic mice (Fig. 2C), although it was scarcely detectable in livers from wild-type mice, suggesting that peroxisomes proliferated in RARE transgenic livers. Besides steatosis and focal necrosis, liver cell dysplasias such as nuclear anisocytosis, hyperchromasia, pleomorphism, and increased nuclear cytoplasmic ratio were frequently observed in transgenic mice at 4 months of age (Fig. 2D). These histological abnormalities tended to be enhanced with age. Electron microscopic analysis revealed that transgenic hepatocytes displayed a substantial increase in lipid deposition within the hepatocyte cytoplasm (Fig. 3A). Interestingly, many fat droplets existed in contact with mitochondria. On the other hand, hepatocytes from wild-type mice showed no lipid deposition (Fig. 3B). The transgenic hepatocytes had almost normal mitochondria ultrastructure, because they showed neither mitochondrial swelling nor membrane damage.

Figure 2.

Histochemical study of the liver from RARE transgenic mice. (A) Hepatocytes positive for proliferating cell nuclear antigen clustered in the liver lobule (original magnification: ×200). (B) BrDU incorporated into some transgenic hepatocytes (arrow). (C) Granular staining of catalase protein present within the cytoplasm of the hepatocytes (original magnification: ×400). (D) Liver cell dysplasia (nuclear anisocytosis, hyperchromasia, and pleomorphism) at 4 months of age (original magnification: ×100).

Figure 3.

Electron microscopic analysis. (A) Lipid droplets present in the cytoplasm of a transgenic hepatocyte (original magnification: ×4,000). Lipid droplets existed in contact with mitochondria; mitochondrial membranes seemed to be intact. (B) No lipid deposition was observed in the cytoplasm of wild-type mouse hepatocytes (magnification: ×4,000).

Fatty Acid β-Oxidation.

Transgenic mice exhibited severe fatty deposition; the causes of steatohepatitis were examined. Because transgenic mice did not show obesity, we focused on fatty acid β-oxidation, which is the primary system of fatty acid oxidation. Compared with wild-type mice, the basal levels of fatty acid β-oxidation in RARE transgenic mice were depressed using palmitic acid (C16:0) and lauric acid (C12:0) as the substrates (see Table 3). Thus the β-oxidation of the long-chain and middle-chain fatty acids, which are usually oxidized in mitochondria,23, 24 was impaired, leading to the formation of steatosis. When behinic acid, one of the very long-chain fatty acids, was used as the substrate, the fatty acid β-oxidation activity in RARE transgenic mice was significantly increased, suggesting that peroxisomal β-oxidation is up-regulated instead in RARE transgenic mice. The increased β-oxidation of fatty acids in peroxisomes coincided with peroxisomal proliferation, as was observed in immunohistochemistry via anticatalase antibody (see Fig. 2C).

Messenger RNA Expression of Fatty Acid Oxidation–Related Enzymes.

For mitochondrial enzymes involved in fatty acid β-oxidation, the expression levels of carnitine palmitoyltransferase (CPT) II, very long acyl-CoA dehydrogenase (VLCAD), long acyl-CoA dehydrogenase (LCAD), and 3-hydroxyacyl-CoA dehydrogenase (HCD) were greatly decreased in RARE transgenic mice compared with those in wild-type mice (P < .01 each, Fig. 4). The levels of CPT I and medium acyl-CoA dehydrogenase were higher in RARE transgenic mice (P < .05 and P < .01, respectively), however, the differences were small. For peroxisomal enzymes, expression levels of acyl-CoA oxidase (AOX) and bifunctional enzyme (BFE) were much greater in RARE transgenic mice than those in wild-type mice (P < .01 each), supporting the data that peroxisomal β-oxidation activity was up-regulated in transgenic mice. For microsomal enzymes involved in fatty acid ω-oxidation, levels of cytochrome p4504a10, cytochrome p4504a12, and cytochrome p4504a14 were much higher in RARE transgenic mice than those in wild-type mice (P < .01 each), suggesting that microsomal ω-oxidation of fatty acids was greatly increased. These data suggest that mitochondrial β-oxidation of fatty acids was depressed; however, peroxisomal β-oxidation and microsomal ω-oxidation were increased instead in RARE transgenic livers. Expression of peroxisome proliferator-activator receptor (PPAR) α and β was enhanced by the dominant negative effect of RARE; PPAR-α and PPAR-β increased twofold and 19-fold, respectively (see Fig. 4). PPAR-γ expression in transgenic mice was similar to that in wild-type mice.

Figure 4.

Real-time reverse-transcriptase polymerase chain reaction analysis of the enzymes associated with β-oxidation and ω-oxidation of fatty acids. In each enzyme, by assuming that the lowest expression value is 100 among wild-type mice, transgenic mice, and transgenic mice fed a high-RA diet, the other two values were expressed as the ratios to the lowest value. The open columns, closed columns, and cross-hatch columns represent wild-type mice, transgenic mice, and transgenic mice fed a high-RA diet for 3 months, respectively. CPT, carnitine palmitoyltransferase; VLCAD, very long acyl-CoA dehydrogenase; LCAD, long acyl-CoA dehydrogenase; MCAD, medium acyl-CoA dehydrogenase; ECH, enoyl-CoA hydratase; HCD, 3-hydroxyacyl-CoA dehydrogenase; KCT, 3-ketoacyl-CoA thiolase; AOX, acyl-CoA oxidase; BFE, bifunctional enzyme; CYP4a10, cytochrome p4504a10; CYP4a12, cytochrome p4504a12; CYP4a14,cytochrome p4504a14; PPAR, peroxisome proliferator-activated receptor.

H2O2 and 8-OHdG Contents.

Because up-regulation of peroxisomal β-oxidation and microsomal ω-oxidation of fatty acids leads to overproduction of H2O2 and 8-OHdG,23, 24 we measured the contents of H2O2 and 8-OHdG. Hepatic H2O2 content increased approximately fivefold in RARE transgenic mice compared with wild-type mice (P < .05; see Table 3). Hepatic content of 8-OHdG was also higher in RARE transgenic mice than in wild-type mice (P < .05).

Liver Tumor Development.

To examine liver tumor development, RARE transgenic mice and wild-type mice were sacrificed at 4, 8, 12, 15, and 18 monhs of age (Fig. 5). No tumors were observed at 4 and 8 months. At 12 months of age, 50% of RARE transgenic mice developed liver tumors. The incidence of tumor development was gradually increased with time; 40% of RARE transgenic mice developed liver tumors at 15 months of age and 80% did so at 18 months. Liver tumors were composed of HCC and adenoma. The percentages of HCC at 12, 15, and 18 months of age were 13%, 20%, and 27%, respectively. The HCC at 12 months of age was clearly discriminated from the surrounding tissues in one mouse, but no capsule formation was observed (Fig. 6A). Microscopic analysis of a tumor in this mouse revealed that it the tumor sarcomatous HCC (Fig. 6B). At 18 months of age, a 15-mm-long tumor with a smooth surface and whitish color developed (Fig. 6C). The acinar formation of neoplastic hepatocytes were evident, round to oval nuclei had prominent nucleoli, and eosinophilic cytoplasm was abundant (Fig. 6D). Tumor size and multiplicity tended to be increased with age.

Figure 5.

Liver tumor incidence in RARE transgenic mice (•) and wild-type mice (□). Tumor incidence of the mice at the ages of 4, 8, 12, 15, and 18 months is shown. Open circles (○) represent RA transgenic mice fed a high-RA diet. The numbers of the mice that developed tumors and total mouse numbers are shown at the bottom. *P < .05 between transgenic mice and transgenic mice fed a high-RA diet. Tg, transgenic; Wt, wild-type; RA, retinoic acid.

Figure 6.

Histopathological findings of a RARE transgenic mouse. (A) The tumor, whose edge is clearly discriminated from the surrounding tissue, developed at 12 months of age (original magnification: ×100). (B) Histological examination showed that the tumor in panel A was sarcomatous HCC (original magnification: ×200). (C) A RARE transgenic mouse developed a large tumor whose diameter was 1.5 cm at 18 months of age. (D) Microscopic appearance of panel C. The neoplastic hepatocytes formed an acinar formation (original magnification: ×400). T, tumor; NT, nontumor.

β-Catenin Expression.

Because PPAR-β is a target molecule of TCF/β-catenin,25 and transgenic mice expressing active β-catenin in the skin induced the formation of hair follicle tumors,26 expression of β-catenin and its target genes for TCF/β-catenin (such as c-myc and cyclin D1) was examined (Fig. 7A and 7B). Expression of β-catenin and cyclin D1 was enhanced in transgenic livers. Expression of c-myc was not altered. The TCF-4 complex with β-catenin increased; however, that of RAR-α with β-catenin decreased in transgenic livers (Fig. 7C). These data suggest that a Wnt signal was activated as a result of a loss of RA signal.

Figure 7.

Expression of β-catenin and related molecules. (A) β-catenin expression in transgenic and wild-type mice. Expression of β-catenin was enhanced in transgenic mice. β-actin was used as an internal control. (B) Cyclin D1 and c-myc expression in transgenic and wild-type mice. Expression was examined in liver tissues from two mice from both groups. Expression of cyclin D1 in transgenic mice was greater than that in wild-type mice; however, that of c-myc in transgenic mice was similar to that in wild-type mice. (C) TCF-4/β-catenin complex and RAR-α/β-catenin complex were examined. The amount of TCF-4/β-catenin complex in transgenic mice was greater than that in wild-type mice; however, that of RAR-β/β-catenin in transgenic mice was smaller than that in wild-type mice. Tg, transgenic; Wt, wild-type; RAR-α, retinoic acid receptor α.

Effects of High-RA Feeding on Transgenic Livers.

In the in vitro study, higher doses of RA over 100-fold greater than the physiological concentration have been reported to overcome the dominant negative effect of RARE.13 We examined effects of high-RA feeding on phenotypic changes of transgenic livers. Dietary supplementation with high-RA diets (approximately 101-fold increase of normal all-trans retinoic acid content) improved steatosis of transgenic livers; out of 8 mice, 6 exhibited grade 0 and 2 exhibited grade 2 (see Table 2). Feeding transgenic mice on a high-RA diet for 3 months decreased serum alanine aminotransferase levels to those of wild-type mice (see Table 3). Histological disorders, including massive steatosis, focal necrosis, and liver cell dysplasia, were also improved. A high-RA diet reversed fatty acid β-oxidation activity when palmitic acid (C16:0) and lauric acid (C12:0) were used as the substrates (see Table 3). Thus, the β-oxidation activity of the long-chain and middle-chain fatty acids, which are usually oxidized in mitochondria, was retrieved. When behinic acid (C22:0), one of the very long-chain fatty acids, was used as a substrate, increased β-oxidation activity was normalized. Expression of depressed mitochondrial fatty acid β-oxidizing enzymes including CPT II, VLCAD, LCAD, and HCD was also normalized (P < .01 each; see Fig. 4). In addition, increased expression of peroxisomal enzymes such as AOX and BFE (P < .01 each) and microsomal ω-oxidation enzymes such as cytochrome p4504a10, cytochrome p4504a12, and cytochrome p4504a14 was normalized (P < .01, P < .01, and P < .05, respectively). Expression of PPAR-α and PPAR-β was also normalized by high-RA feeding (P < .01 each). Importantly, high-RA supplementation decreased H2O2 and 8-OHdG levels within the normal limits. Thus, normalization of liver histology by high doses of RA seems be due to normalization of impaired fatty acid metabolism and decreased production of H2O2 and 8-OHdG. Long-term feeding of transgenic mice on a high-RA diet for 12 months apparently decreased tumor incidence (6/12 vs. 0/12, P < .05; see Fig. 5), indicating that steatohepatitis and tumor formation were actually due to loss of RA signaling.

Discussion

In the present study, the transgenic mice expressing the RARE specific to hepatocytes developed massive hepatic steatohepatitis and liver tumor. This is the demonstration that the RA signal is indispensable for normal development of liver architecture and that functional loss of RA leads to occurrence of severe steatosis, cellular dysplasia, and cancer. The previous clinical trial suggests the important roles of RA in chemopreventive action on the post-resection recurrence of HCC.8 Therefore, RA is a promising candidate as a chemopreventive agent for HCC.

Biochemical analysis of RARE transgenic mice revealed that fatty acid β-oxidation using lauric acid (C12:0) and palmitic acid (C16:0) as substrates was significantly depressed compared with that in wild-type mice (P < .05 each). Because it is known that mitochondria catalyze the β-oxidation of the bulk of short-, medium-, and long-chain fatty acids,23 these data suggest that β-oxidation of fatty acids in mitochondria was depressed. Expression of mitochondrial enzymes involved in fatty acid β-oxidation such as VLCAD, LCAD, and HCD was suppressed in RARE transgenic mice. Because fatty acid oxidation in mitochondria depends on expression levels of β-oxidation–related enzymes and amount of their substrates,23, 24 decreased β-oxidation in mitochondria seems to be due to down-regulation of its related enzymes. Thus, hepatic steatosis was induced by functional loss of RA as a result of impaired β-oxidation of fatty acids in mitochondria.

It is also known that β-oxidation of very long-chain fatty acids occurs in peroxisomes.24, 27 In the present study, the activity of β-oxidation using behinic acid (C22:0) as a substrate increased significantly, suggesting that peroxisomal β-oxidation increased as well. Indeed, expression of AOX, the rate-limiting enzyme of peroxisomal fatty acid β-oxidation, and BFE was enhanced. Overexpression of catalase also suggests that peroxisomes proliferated in transgenic mice. This finding was supported by the report that expression of AOX was induced in association with peroxisomal proliferation.28 Because the chain-shortened fatty acyl-CoA by β-oxidation in peroxisomes (mostly short- and medium-chain fatty acids) then enter the mitochondrial β-oxidation system for completion,23, 24 increased peroxisomal β-oxidation and decreased mitochondrial β-oxidation may accelerate lipid deposition in liver tissue. In the process of peroxisomal β-oxidation and microsomal β-oxidation of fatty acids, a large amount of H2O2 is produced.23 H2O2 is converted to hydroxyl radicals in the presence of Fe2+,29 and 8-OHdG, a DNA adduct by reactive oxygen species, is formed mainly by hydroxyl radicals.30 In the present study, formation of H2O2 and 8-OHdG was significantly increased (P < .01 and P < .05, respectively), indicating that an increase in peroxisomal β-oxidation and microsomal β-oxidation caused formation of H2O2 and 8-OHdG. Because dicarboxyl acids, the products of β-oxidation, exert suppressive effects on β-oxidation in mitochondria23 and can be used as the substrates of β-oxidized in peroxisomes,24 they eventually accelerate peroxisomal β-oxidation and suppress mitochondrial β-oxidation. Taken together, as our hypothesis is shown in Fig. 8, reactive oxygen species may play an important role in progression of hepatic lesions in this mouse model. In addition, because the pathological and biochemical abnormalities observed in these mice have many similarities with those of nonalcoholic steatohepatitis,31 the RARE transgenic mice may be used as a model animal of nonalcoholic steatohepatitis.

Figure 8.

Proposed mechanism of RA function in this mouse model. The mice expressing RARE cannot bind to RA at its physiological concentration, and no signals from RA were transduced. Functional loss of RA causes many abnormalities in mitochondria, peroxisomes, microsomes, and nuclei. The most prominent effects seem to be due to overproduction of reactive oxygen species in peroxisomes and microsomes. In addition, Wnt signal activation may promote hepatocarcinogenesis. These biochemical and molecular abnormalities may explain phenotypic changes observed in the livers of RARE transgenic mice. RXRs, retinoid X receptors; RARE, retinoic acid receptor α; RA, retinoic acid; PPAR, peroxisome proliferator-activated receptor; FA, fatty acid; 8-OHdG, 8-hydroxy-2′-deoxyguanosine.

Because peroxisomal β-oxidation is regulated by acyl-CoA, PPAR-α and dicarboxylic acids,23, 24 PPAR-α overexpression in the present study seemed to be one cause of peroxisomal β-oxidation. How was PPAR-α up-regulated in these mice? Increased production of dicarboxylic acids may cause PPAR-α overexpression, because the dicarboxylic acids which were generated by cytochrome p4504a enzyme-mediated ω-oxidation act as PPAR-α ligands.32 Because cytochrome p4504a genes are also transcriptionally regulated by PPAR-α,33 PPAR-α–stimulated peroxisomal β-oxidation and microsomal ω-oxidation of fatty acids. How, then, was PPAR-β regulated? Because PPAR-β is a target molecule of TCF/β-catenin, enhanced expression of PPAR-β may be due to activation of the Wnt signal. In the present study, higher expression of β-catenin and cyclin D1 was observed. In addition, increased formation of TCF-4/ β-catenin complex and decreased formation of RAR-α/ β-catenin complex strengthen the idea that the Wnt signal was activated by loss of RA signal.25 These were also supported by the previous report that RA exerts its biological activity partly through RAR-α's competion with TCF for β-catenin binding.25, 34, 35

In conclusion, the pathological and biochemical changes of our mice were very similar to the mice lacking AOX.36 The AOX−/− mice exhibited severe fatty metamorphosis of the liver, spontaneous hepatic peroxisomal proliferation, increased peroxisomal fatty acid β-oxidation, and liver tumor. The AOX−/− mice produced tremendous amounts of H2O2, suggesting the roles of H2O2 in inflammation, cell proliferation, and carcinogenesis.37 PPAR-α was overexpressed in livers of the AOX−/− mice, although expression of PPAR-β was not mentioned in the report. The RARE transgenic mice were also similar to hepatitis C virus (HCV) core transgenic mice in that the HCV core protein transgenic mice developed hepatic steatosis early in life, and after 16 months of age, the mice developed liver tumors.38 The mechanisms responsible for hepatic steatosis and tumor formation in HCV core transgenic mice are supposed to be associated with decreased mitochondrial fatty acid β-oxidation, suppressed very low density lipoprotein secretion and production of reactive oxygen species.39, 40 However, there was no inflammation in HCV core transgenic mice. Recently, it has been reported that transgenic mice expressing the full-length of HCV open reading frame displayed hepatic steatosis and liver tumors.41 However, these mice did not show any inflammations, either. The similarities between RARE and HCV transgenic mice suggests the importance of RA in HCV-related liver diseases.

Ancillary