Potential conflict of interest: Nothing to report.
Nonalcoholic steatohepatitis (NASH) may cause fibrosis, cirrhosis, and hepatocellular carcinoma (HCC); however, the exact mechanism of disease progression is not fully understood. Angiogenesis has been shown to play an important role in the progression of chronic liver disease. The aim of this study was to elucidate the role of angiogenesis in the development of liver fibrosis and hepatocarcinogenesis in NASH. Zucker rats, which naturally develop leptin receptor mutations, and their lean littermate rats were fed a choline-deficient, amino acid–defined diet. Both Zucker and littermate rats showed marked steatohepatitis and elevation of oxidative stress markers (e.g., thiobarbital acid reactive substances and 8-hydroxydeoxyguanosine). In sharp contrast, liver fibrosis, glutathione-S-transferase placental form (GST-P)-positive preneoplastic lesions, and HCC developed in littermate rats but not in Zucker rats. Hepatic neovascularization and the expression of vascular endothelial growth factor (VEGF), a potent angiogenic factor, only increased in littermate rats, almost in parallel with fibrogenesis and carcinogenesis. The CD31-immunopositive neovessels were mainly localized either along the fibrotic septa or in the GST-P–positive lesions. Our in vitro study revealed that leptin exerted a proangiogenic activity in the presence of VEGF. In conclusion, these results suggest that leptin-mediated neovascularization coordinated with VEGF plays an important role in the development of liver fibrosis and hepatocarcinogenesis in NASH. (HEPATOLOGY 2006;44:983–991.)
Nonalcoholic steatohepatitis (NASH), a progressive metabolic liver disease, is one of the major consequences of the current obesity epidemic.1 It lies on a spectrum of nonalcoholic fatty liver disease that ranges from simple steatosis to cirrhosis.2 Whereas simple steatosis seems to be a benign and nonprogressive condition, NASH is recognized as a potentially progressive disease that may cause cirrhosis, an end-stage liver disease, and hepatocellular carcinoma (HCC).3, 4 The pathogenesis of NASH is known to be a multifactorial process.3, 5 The current concept in the pathogenesis of NASH involves a “two-hit” theory in which an initial metabolic disturbance, such as insulin resistance, causes steatosis, and a second pathogenic stimulus causes oxidative stress and reactive oxygen species, leading to steatohepatitis. It has been reported that hepatic thiobarbital acid reactive substances (TBARSs) and 8-hydroxydeoxyguanosine (8-OHdG), as reliable markers of lipid peroxidation and oxidative DNA-damage, respectively, were significantly increased both in the animal models and human samples of NASH.6–8 However, it is not known whether reactive oxygen species can result in progression of liver fibrosis and hepatocarcinogenesis, beccause, like diabetes mellitus, NASH is almost certainly a polygenic disease affected by several factors, with disease pathogenesis related to multiple “hits.”3
Animal models of hepatic steatosis and steatohepatitis have improved the understanding of the pathogenesis of NASH. The different animal models of NASH have been extensively reviewed.9 Among them, we employed a choline-deficient, amino acid–defined (CDAA) diet model. This model induces histological processes similar to those of human NASH—namely, hepatic steatosis, steatohepatitis, liver fibrosis, cirrhosis, and HCC development associated with elevation of several markers of reactive oxygen species, such as TBARSs and 8-OHdG.10 A downside of the CDAA model is that it does not induce first-hit features (obesity, insulin resistance, and hyperglycemia).9 Zucker rats, which lack functional leptin receptor due to a miss-sense mutation, have these first-hit conditions.11 In the present study, a choline-deficient, amino acid–defined diet was given to Zucker rats to observe the pathological sequences of NASH under first-hit conditions. Several studies have shown impaired liver fibrogenesis in Zucker rats and ob/ob mice, which are leptin-deficient.12–14 However, the effect of leptin signal impairment on HCC based on liver cirrhosis, which is a further sequential progressive manifestation of NASH, has not yet been elucidated.
It is now known that angiogenesis plays an important role in many physiological and pathological events.15, 16 We have previously reported that angiogenesis plays an important role in the development of liver fibrosis and hepatocarcinogenesis.17, 18 Angiogenesis is regulated by the net balance of proangiogenic factors and angiogenic inhibitors. To date, many positive and negative angiogenic-modulating factors have been identified. Among these, vascular endothelial growth factor (VEGF), also known as vascular permeability factor, is the most potent factor in angiogenesis.19, 20 Emerging evidences have shown that VEGF plays a pivotal role in many steps of physiological and pathological angiogenesis. VEGF is not only an angiogenic factor—it is also known as a survival factor for endothelial cells (ECs). VEGF expression increases stepwise during liver fibrosis development and hepatocarcinogenesis, and suppression of the VEGF signaling cascade attenuates these pathological sequences.17, 18 Although the original findings of leptin indicated that it plays a crucial role in the regulation of appetite and the amount of body fat mass, some recent studies have revealed that leptin elicits multiplicity of biological effects.21 Recently, it has been shown that leptin exerts a potent proangiogenic activity both in vitro and in vivo.22, 23 It has also been shown that functional leptin receptor is expressed on ECs and that leptin induces proliferation of ECs in a dose-dependent manner.23 Furthermore, leptin has induced neovascularization in the corneas of normal rats but not in Zucker rats.23 Synergistic stimulation of angiogenesis by leptin with VEGF, and impairment of VEGF production in leptin receptor–deficient (db/db) mice have also been reported.24, 25 However, the role of leptin-mediated angiogenesis in fibrogenesis and carcinogenesis of the liver including NASH has not yet been elucidated.
We examined the role of leptin-mediated neovascularization in the development of liver fibrosis and HCC in a rat NASH model, especially in conjunction with oxidative stress. We also performed a set of in vitro experiments to elucidate the possible mechanisms involved in these processes.
Male Zucker (fa/fa: Z) rats and their littermates (+/?: L), aged 6 weeks, were purchased from Japan SLC Inc. (Hamamatsu, Shizuoka, Japan). They were housed in stainless steel mesh cages under controlled conditions of temperature (23 ± 3°C) and relative humidity (50 ± 20%), with 10 to 15 air changes per hour and light illumination for 12 hours a day. The animals were allowed access to tap water ad libitum throughout the experimental periods. A CDAA and a choline-sufficient amino acid–defined (CSAA) diet were purchased from CLEA Japan Inc. (Tokyo, Japan). The details of both diets are described elsewhere.26 Recombinant human leptin and VEGF, and VEGF-neutralizing monoclonal antibody were purchased from R&D systems (Minneapolis, MN).
Short- and long-term experiments were performed. The experimental period of the short-term experiment was 16 weeks. The Zucker rats that fed on CSAA and CDAA diets were designated as group 1 and group 2, respectively. Group 3 and group 4 were counterpart groups (i.e., littermate rats that fed on the CSAA and CDAA diet, respectively). Each group consisted of 15 rats. In the long-term experiment, the Zucker rats and littermate rats, which fed on the CDAA diet for 80 weeks, were designated as group 5 and group 6, respectively. Each group consisted of 10 rats. All rats were anesthetized, the thoracic cavity was opened, and blood samples were withdrawn via cardiac puncture. Alanine aminotransferase and glucose levels were determined via routine laboratory methods. All animal procedures were performed according to standard protocols and recommendations for the proper care and use of laboratory animals.
Histological and Immunohistochemical Examinations.
In all experimental groups, the first section was routinely stained with hematoxylin-eosin for histological examination. Another section was stained with Sirius red for detecting fibrosis development. Immunohistochemical staining of enzyme-altered preneoplastic lesions—namely, the placental form of glutathione-S-transferase (GST-P) (MBL Co. Ltd., Nagoya, Japan), and 8-OHdG (NOF Corp., Tokyo, Japan) were performed as described previously.27, 28 For determination of neovascularization, we employed immunohistochemical detection of CD31 (BD Biosciences, CA), which is widely used as a marker of neovascularization, using frozen sections as previously described.28 Immunopositive quantitation of GST-P, 8-OHdG, and CD31 was evaluated with Adobe Photoshop software and National Institutes of Health image software as previously described.29
Hepatic Hydroxyproline, Lipid Peroxidation, and VEGF Expression.
Hepatic hydroxyproline content was determined as previously described in 200 mg of frozen samples.17 After equalization of the protein content, TBARSs and VEGF in the liver were determined using the OXI-TEK TBARS Assay Kit (ZEPTOMETRIX Corp., Buffalo, NY) and an ELISA kit (R&D systems) according to the manufacturers' instructions.
In Vitro Angiogenesis Assay.
The in vitro angiogenesis was assessed as formation of capillary-like structures of human umbilical vein ECs cocultured with human diploid fibroblasts as previously described.30 The experimental procedure was performed according to the instructions provided with the angiogenesis kit (Kurabo, Tokyo, Japan). Briefly, the cells were treated with human leptin (10 ng/mL), VEGF (10 ng/mL), and/or VEGF-neutralizing monoclonal antibody (500 ng/mL) on day 1, and the medium was replaced on days 4, 7, and 9. On day 11, the cells were fixed, and human umbilical vein ECs were stained using an antihuman CD31 antibody (Kurabo) according to the manufacturer's protocol. Computer-assisted quantitation of tubule formation was performed in the same way as for the in vivo assay.
To assess the statistical significance of the intergroup differences in the quantitative data, Bonferroni's multiple comparison test was used after one-way ANOVA. This was followed by Barlett's test to determine the homology of variance.
The data concerning the effective numbers of rats and final body weights in all experimental groups of the short-term experiment are shown in Table 1. There were no significant differences in final body weight between groups 1 and 2. On the contrary, and in accordance with previous reports, the final body weight of the CDAA-treated littermate rats (group 4) was lighter than that of the CSAA-treated rats (group 3).26 Obesity was significantly more remarkable in Zucker rats than in littermate rats regardless of diet (i.e., CDAA or CSAA). Zucker rats also showed significant hyperglycemia. Elevation of serum alanine aminotransferase levels was observed in CDAA-treated Zucker and littermate rats at a magnitude similar to that seen in CSAA-treated rats. Neither ascites nor other organ abnormalities were observed at the end of the experiment in all groups (data not shown).
Table 1. Experimental Group Details Given CDAA and CSAA Diet
Abbreviations: CDAA, Choline-deficient, amino acid-defined diet; CSAA, Choline-sufficient, amino acid-defined diet; Z and L, Zucker and littermate rats, respectively.
Steatohepatitis and Oxidative Stress.
The histological findings of the liver in the CDAA-treated rats of the short-term experiment are shown in Fig. 1. Both Zucker rats (A; group 2) and littermate rats (B; group 4) developed marked steatohepatitis of almost similar magnitude, which reflected serum alanine aminotransferase elevations. We next examined 8-OHdG–immunopositive cells in all short-term experimental groups. As shown in Fig. 2A, the number of 8-OHdG–immunopositive cells was significantly more increased in the CDAA-treated groups (groups 2 and 4) than in the CSAA-treated groups (groups 1 and 3), respectively, although there was a statistically significant difference between groups 2 and 4. Similarly, the hepatic lipid peroxidation was significantly more augmented in the CDAA-treated rats than in the CSAA-rats in both Zucker rats and littermate rats (Fig. 2B).
Liver Fibrosis and HCC.
We next examined the development of fibrosis and preneoplastic lesions in all short-term experiment groups. As shown in Fig. 3B, liver fibrosis development with fatty accumulation was marked in the CDAA-treated littermate rats (group 4) as previously reported.10 Contrary to the oxidative stress increment, no fibrosis development could be observed in the CDAA-treated Zucker rats (group 2) (Fig. 3A). CSAA treatment induced neither liver fibrosis nor pre-neoplastic lesions in Zucker rats (group 1) or littermate rats (group 3) (data not shown). The results of computer-assisted semiquantitative analysis of the fibrotic area almost corresponded to the histological findings (Fig. 3C). Similar to the fibrotic area, the hepatic hydroxyproline content significantly increased only in group 4 compared with the other groups (Supplementary Fig. 1). The preneoplastic lesion development matched the liver fibrosis development. As shown in Fig. 4, marked GST-P–positive preneoplastic lesions developed in the CDAA-treated littermate rats (group 4) (Fig. 4B), whereas very few positive lesions were observed in the Zucker rats (group 2) (Fig. 4A). Semiquantitative analysis of the GST-P–positive lesions confirmed the immunohistochemical findings, and no preneoplastic lesion development was found in the CSAA-treated Zucker or littermate rats (Fig. 4C). The ameliorative effects of liver fibrosis development and GST-P–positive preneoplastic lesions were almost matching.
We next examined whether CDAA treatment for Zucker rats ameliorated HCC development rather than the development of preneoplastic lesions. Figure 5 shows the typical features of the liver in the CDAA-treated Zucker and littermate rats in the long-term experiment. After CDAA treatment for 80 weeks, the macroscopic examination revealed that the liver in littermate rats had a marked development of hepatic nodules with large, reddish tumors (Fig. 5D). Histological examination revealed that all CDAA-treated littermate rats (group 6) developed adenoma (10/10), and half of the rats developed HCC (5/10) (Fig. 5E–F) on top of liver cirrhosis with fatty change (Fig. 5C). In sharp contrast, there were no macroscopic nodular changes in the liver of Zucker rats (group 5) (Fig. 5A), and only steatohepatitis could be observed via histological examination (Fig. 5B). No macroscopic changes were observed in the CSAA-treated Zucker or littermate rats (data not shown).
Because neovascularization has been shown to play an important role in the liver fibrogenesis and hepatocarcinogenesis, we elucidated the possible involvement of neovascularization in the current study. First, we performed immunohistochemical analysis of CD31 in the rats of the short-term experiment. Marked immunopositive vessels could be observed in the CDAA-treated littermate rats (group 4) (Fig. 6A), whereas no augmentation was found in the CDAA-treated Zucker rats (group 2) (Fig. 6B). Semiquantitative analysis confirmed the histological findings, and no augmentation of CD31 expression was observed in the CSAA-treated Zucker or littermate rats (Fig. 6C). Hepatic VEGF expression was also significantly higher in group 4 than in group 2. The expression level of VEGF in group 2 was of similar level to those in the CSAA-treated Zucker and littermate rats (group 1 and group 3, respectively) (Supplementary Fig. 2).
To identify the localization of CD31-immunopositive neovascularization, we next compared Sirius red staining with CD31 immunostaining in serial sections. As shown in Fig. 7, neovascularization was mainly observed along the fibrotic septa, and it was also disseminated in the hepatic lobules. We also compared GST-P and CD31 immunostaining in serial sections. The CD31-immnopositive neovascularization in the hepatic lobules was mainly observed in the GST-P–positive preneoplastic lesions compared with the nonneoplastic adjacent lesions (Supplementary Fig. 3).
To elucidate the direct interaction between leptin and VEGF, we performed a set of in vitro experiments. Similar to previous reports,30 VEGF treatment increased EC tubular formation. Leptin treatment also stimulated neovascularization, although it was not of a similar magnitude to that of VEGF. The combination treatment with VEGF and leptin exerted a more potent proangiogenic activity compared with single treatment by VEGF or leptin (Fig. 8). Semiquantitative analysis of the microvessel length confirmed these results, and the combination effect of leptin and VEGF was almost abolished by treatment with VEGF-neutralizing monoclonal antibody (Supplementary Fig. 4). These results indicated that leptin exerted a proangiogenic activity in the presence of VEGF.
In this study, we showed that, even in the absence of the leptin-signaling cascade, steatohepatitis and augmentation of oxidative stress could be observed in the rat NASH experimental model. However, without leptin signaling, neither fibrosis nor HCC developed, suggesting that leptin plays a pivotal role in the progression of fibrogenesis and carcinogenesis in NASH. Leptin-mediated neovascularization, which coordinated with VEGF, increased significantly almost in parallel with the progression of liver fibrogenesis and carcinogenesis.
To date, no animal model can completely reproduce the pathological features of human NASH. It is well established that the administration of CDAA diet to rats results in development of similar pathological sequences of NASH, although it does not induce first-hit features (obesity, glucose intolerance, and insulin resistance). Because Zucker rats exert these first-hit features, the pathological sequences of NASH under the first-hit conditions are supposed to be observed in the liver of CDAA-treated Zucker rats. However, without leptin signaling, only steatohepatitis develops. The “two-hit” theory is very widely recognized as an explanation of the pathogenesis of NASH. The second hit, which occurs after the first hit based on several metabolic disorders such as insulin resistance, leads to the progression of simple steatosis to steatohepatitis. Several studies have suggested that oxidative stress is the leading culprit of the second hit. There is strong evidence suggesting that oxidative stress is involved in the progression from steatosis to steatohepatitis.5, 6 As such, we found that both steatohepatitis and augmentation of oxidative stress occurs regardless of the existence of the leptin signaling pathway in the rat NASH model. However, without leptin signaling, further progression (i.e., fibrosis and HCC development) could not be observed. It has been reported that increased production of oxidative stress occurred very early in the histological spectrum of nonalcoholic fatty liver disease.7 Another report has shown that, even at the earliest stage of nonalcoholic fatty liver disease, there is already considerable oxidative stress production.31 Taken together, it is feasible that fibrogenesis and carcinogenesis of NASH from steatohepatitis can be initiated by introducing a leptin-mediated additional factor based on augmentation of the oxidative stress.
Leptin is now recognized as a multifunctional protein.21 Several studies have shown that leptin exerts a profibrogenic activity that acts directly on hepatic stellate cells.14, 32 Liver fibrosis has been significantly attenuated in ob/ob mice, animals with inborn leptin deficiency, and Zucker rats in several types of experimental models.12–14 It has been reported that leptin expression in the local hepatic environment during liver fibrogenesis increases significantly, and that it also augments platelet-derived growth factor–dependent proliferation of hepatic stellate cells.33 In the present study, we found that leptin signal deficiency led to poor response in hepatocarcinogenesis as well as in fibrogenesis.
Among the possible biological functions of leptin, we focused on neovascularization. Recently, leptin has been found to stimulate angiogenesis both in vitro and in vivo.22, 23, 25 We have reported previously that angiogenesis increases significantly during the development of liver fibrosis and hepatocarcinogenesis.17, 18 It has been shown that “capillarization” and phenotypic changes of the hepatic sinusoidal ECs occur during the development of liver fibrosis.34, 35 This capillarization of sinusoidal ECs results either from development of neovessels or from alteration of pre-existing sinusoids, both of which are possibly induced in response to angiogenic stimulation.36 In the current rat NASH model, we found that the angiogenesis was significantly augmented during the development of liver fibrosis only in the existence of the leptin signaling cascade, and neovascularization was mainly along the fibrotic septa. The increment of neovascularization was also almost in parallel with the development of GST-P–positive preneoplastic lesions, and the CD31-positive vessels were observed in the preneoplastic lesions rather than in the adjacent nonneoplastic areas. These results are consistent with our previous finding that angiogenesis increased stepwise during HCC development.18 It has been reported that impairment of neovascularization in Zucker rats was restored by treatment with leptin and functional leptin receptor gene transfer, and that leptin treatment recovered the fibrogenic activity of the ob/ob mice.13, 22 Further studies are required to examine whether re-establishment of leptin signaling salvages hepatocarcinogenesis associated with the development of neovascularization.
It is now recognized that in vivo angiogenesis status is determined by a balance of several regulators rather than a single regulator. Leptin has been shown to coexpress with other angiogenic factors such as VEGF and basic fibroblast growth factor in several tissues, indicating that leptin exerts a proangiogenic activity with these factors.25, 37 As such, it has been reported that leptin synergistically stimulates angiogenesis with VEGF and basic fibroblast growth factor.25 We have shown that VEGF expression increases significantly during liver fibrogenesis and carcinogenesis almost in parallel with the augmentation of neovascularization.17, 18 In the current rat NASH model, VEGF expression was markedly upregulated along with the development of liver fibrosis and preneoplastic lesions.
We also observed potent augmentation of angiogenesis with combination of leptin and VEGF in the EC tube formation assay. This effect was almost abolished by treatment with VEGF-neutralizing monoclonal antibody. The decreased angiogenesis in Zucker rats was accompanied by an altered expression of VEGF signaling pathway and EC dysfunction.22, 38 It has been reported that the angiogenic activities of several factors such as hepatocyte growth factor and basic fibroblast growth factor took place only in a VEGF-dependent manner in vitro.30 In the liver, we have noticed that the proangiogenic activity of angiopoietin-2 was exerted only in the presence of VEGF in HCC.39 Similar VEGF dependence of leptin may occur in the liver in the rat NASH model.
To date, the effect of leptin on tumor growth is still controversial. It has been shown that leptin causes growth proliferation in breast and prostate cancer cells, whereas it inhibits proliferation of pancreatic cancer cells.37, 40 Contrary to our data, several reports have suggested a negative regulatory effect of leptin on HCC.41–43 In one report, leptin exerted an antineoplastic effect on Hep3B HCC cells in athymic nude mice,41 whereas another report showed that leptin had no effect on the proliferation of SMMC-7221 HCC cells.42 Regarding hepatocarcinogenesis, a negative regulatory effect of leptin has been suggested in ob/ob mice.43 However, no fibrotic stimulation was employed in this study. It is well known that most of the human cases of HCC that develop from NASH are accompanied by a marked development of liver fibrosis.3, 4 The existence of liver fibrosis has been shown to accelerate rat experimental hepatocarcinogenesis.44 Moreover, the risk of HCC increases along with the progression of hepatic fibrosis in humans.45 The different effect of leptin on hepatocarcinogenesis between our study and the previous report may be due in part to the existence of fibrosis in the base liver. Experiments to elucidate the interaction between leptin and angiogenesis during rat hepatocarcinogenesis under the condition of first-hit status and liver cirrhosis are currently being conducted in our laboratory.
In conclusion, the present study showed that the deficiency of the leptin-signaling cascade ameliorated not only liver fibrogenesis but also hepatocarcinogenesis in a rat NASH model. Furthermore, leptin-mediated neovascularization in the liver, which took place only in the presence of VEGF, significantly increased in parallel with the progression of NASH. These results suggest that leptin-mediated neovascularization, which coordinates with VEGF, plays a pivotal role in the progression of liver fibrosis and HCC in NASH.