Presented in part at the 2003 annual meeting of the American Association for the Study of Liver Diseases and published in abstract form [HEPATOLOGY 2003;38 (Suppl. 1): 718 and 722].
Potential conflict of interest: Nothing to report.
Nonalcoholic steatohepatitis/nonalcoholic fatty liver disease is considered to be a hepatic manifestation of various metabolic disorders. However, its precise pathogenic mechanism is obscure. Oxidative stress and consequent lipid peroxidation seem to play a pivotal role in disease progression. In this study, we analyzed the localization of oxidized phosphatidylcholine (oxPC), a lipid peroxide that serves as a ligand for scavenger receptors, in livers of patients with this steatotic disorder. Specimens of non-alcoholic fatty liver disease (15 autopsy livers with simple steatosis and 32 biopsy livers with steatohepatitis) were examined via immunohistochemistry and immunoelectron microscopy using a specific antibody against oxPC. In addition, scavenger receptor expression, hepatocyte apoptosis, iron deposition, and inflammatory cell infiltration in the diseased livers were also assessed. Oxidized phosphatidylcholine was mainly localized to steatotic hepatocytes and some macrophages/Kupffer cells. A few degenerative or apoptotic hepatocytes were also positive for oxPC. Immunoelectron microscopy showed oxPC localized to cytoplasmic/intracytoplasmic membranes including lipid droplets. Steatotic livers showed enhanced expression of scavenger receptors. The number of oxPC cells was correlated with disease severity and the number of myeloperoxidase-positive neutrophils, but not with the degree of iron deposition. In conclusion, distinct localization of oxPC in liver tissues suggest that neutrophil myeloperoxidase-derived oxidative stress may be crucial in the formation of oxPC and the progression of steatotic liver disease. (HEPATOLOGY 2006;43:506–514.)
Diseases associated with abnormal glucose or lipid metabolism are increasingly becoming major public health issues in many countries. The most serious problem is death caused by cardiovascular diseases, where metabolic abnormalities are frequent complications as well as predisposing disorders. Recent clinical and epidemiological investigations have shown these metabolic disorders also affect the liver. Ludwig et al.1 first identified such a type of hepatic disorder as the clinical entity, non-alcoholic steatohepatitis (NASH). Histologically, the affected livers show hepatic steatosis with necroinflammatory features resembling those of alcoholic steatohepatitis, but the patients have no medical history of excess alcohol intake. NASH is now recognized to be a hepatic manifestation of metabolic disorders.2
At present, NASH and simple steatosis not related to alcoholism are defined as one combined entity, non-alcoholic fatty liver disease (NAFLD), considered as the major cause of cryptogenic cirrhosis.3 Hence, effective therapy to prevent disease progression from NASH/NAFLD to cirrhosis has been sought.4 To establish appropriate therapy, pathological mechanism should be completely elucidated. However, the pathogenesis and progression of NASH/NAFLD have not been fully clarified. Day and James5 have suggested a 2-hit-theory in which the development of steatohepatitis requires an additional etiological factor other than steatosis. In their theory, oxidative stress is considered the most likely factor as the “second hit,” a key event for disease progression.4–7
Accumulating evidence suggests oxidative stress plays an etiopathogenic role in diseases of various organs/tissues. Atherosclerosis, which frequently occurs in association with metabolic disorders such as hyperlipidemia or diabetes, is one example where oxidative stress contributes to its pathogenic mechanism.8 Oxidatively modified low-density lipoprotein (oxidized low-density lipoprotein; oxLDL) is thought to be an atherogenic factor due to its localization in macrophage-derived foam cells accumulated in human atherosclerotic lesions.9–11 OxLDL has cytotoxicity and macrophage chemoattractivity, which may be important in the genesis of the inflammatory process associated with atherosclerotic plaques.12–14 Cellular uptake of oxLDL is mediated by the binding of oxidized phosphatidylcholine (oxPC) molecules, the principal oxidized phospholipid present in oxLDL, to scavenger receptors.15, 16
Recently, a specific monoclonal antibody against oxPC was developed by one of us.17, 18 Using this antibody, we have investigated the influence of oxidative stress on various diseases.9–11, 19 The aim of this study is to clarify the degree of oxidative stress in NASH/NAFLD by means of immunohistochemical examination using this antibody. Furthermore, we investigated the significance of oxPC localization in livers of NASH/NAFLD; its potential sources and its impact on disease progression.
Human liver specimens obtained at autopsy from 15 patients with simple steatosis and at biopsy from 32 patients with NASH were studied. None of the patients had hepatitis virus infection, excess alcohol consumption (>40 g/day), iron overload, or autoimmune-related disorders. The other clinical characteristics of the patients are summarized in Table 1. Most patients had 1 or more cardiovascular risk factors, namely, diabetes mellitus, hyperlipidemia, or obesity. Histological assessment was performed according to the scoring system of Brunt et al.20 [necroinflammatory grade (0-3); fibrosis stage (0-4)]. Then we confirmed that (1) in all liver samples more than 5% hepatocytes were involved in steatosis; (2) the autopsy livers were simple steatosis (both necroinflammatory grade and fibrosis stage = 0); and (3) the biopsy livers fulfilled the diagnostic criteria of NASH (both necroinflammatory grade and fibrosis stage ≥ 1). In addition, 11 autopsy livers without any pathological changes other than mild congestion were used as normal controls. Informed consent was obtained in every case (from the patient in cases of biopsy, and from a family member in cases of autopsy), and the study was approved by the hospital ethical committee.
Table 1. Clinical Characteristics of Patients Studied
Abbreviations: AST, aspartate aminotransferase; ALT, alanine aminotransferase; NA, not available.
All autopsies were performed within 5 hours after death, and the liver specimens were sampled in a fresh state. Each autopsy liver specimen was divided into 2 pieces. One was snap-frozen to be cut into frozen sections for immunohistochemistry and real-time polymerase chain reaction (PCR) analysis. The other was cut into formalin-fixed paraffin sections for histological, histochemical, and immunohistochemical examinations. The biopsy specimens were fixed with formalin immediately after sampling and were cut into paraffin sections for histopathological examinations.
The degree of liver fibrosis was confirmed by performing Azan Mallory stain on the liver sections. Hepatic iron deposition was examined by Perl's Prussian blue stain and graded according to the scoring system of our previous report; degree of iron deposit was assessed separately in hepatocytic (0-3), sinusoidal (0-3), and portal iron deposit (0-3).21
Primary antibodies used in this study are listed in Table 2. For the identification of oxPC localization, a newly developed mouse monoclonal antibody, DLH3, was used.17, 18 The methods of antibody production and specificity testing have been reported previously.17, 18 4-Hydroxynonenal (4HNE), another representative lipid peroxide produced by oxidative injury,22 was also examined using a specific antibody. For the identification of 3 different scavenger receptors [scavenger receptor class A (SR-A), CD36, and scavenger receptor class B type-1 (SR-B1)], frozen sections were used, because monoclonal antibodies against these scavenger receptors work well on frozen sections only. As summarized in Table 2, immunohistochemical markers for lymphocytes, macrophages, neutrophils [neutrophil elastase and myeloperoxidase (MPO)], hepatocytes, and apoptotic cells [single-stranded DNA (ssDNA)] were identified with the specific antibodies.
For paraffin sections, immunohistochemical analyses were performed using a 3-step staining procedure consisting of sequential incubation with primary and secondary antibodies and streptavidin–biotin complex with horseradish peroxidase. Horseradish peroxidase activity was visualized with 3-amino-9-ethylcarbazole, and the sections were faintly counterstained with hematoxylin. The specificity and results obtained with the immunostaining were checked by omission of the primary antibodies and use of nonimmune mouse or rabbit serum (Dako Corp., Carpinteria, CA) as a negative control. For frozen sections, we adopted Envision+ system (Dako) as a second step. The number of oxPC-positive cells and of inflammatory cells (lymphocytes, macrophages, and neutrophils) were evaluated by computer-aided morphometery, as reported previously.23 Data were expressed as mean cell counts per 1 mm2 tissue area.
To analyze oxPC localization in macrophages/Kupffer cells, we performed immunodouble staining for oxPC/macrophage (CD68) on the liver sections. Briefly, DLH3 [mouse immunoglobulin M (IgM) antibody] and PG-M1 (mouse IgG antibody) were applied as primary antibodies, followed by incubation with biotinylated anti-mouse IgM and alkaline phosphatase-conjugated anti-mouse IgG (both from Dako). The biotinylated antibody was visualized with horseradish peroxidase and 3-amino-9-ethylcarbazole (red: oxPC) and alkaline phosphatase with fast blue BB (blue: macrophages). A double-positive result yielded purple color. To analyze oxPC localization in apoptotic cells, double immunostaining for oxPC/ssDNA was also performed using fluorescent-labeled (green: for oxPC) and rhodamine-labeled (red: for ssDNA) second antibodies. To confirm the reported expressions of CD36 on activated stellate cells and SR-B1 on hepatocytes,24, 25 double immunostainings for CD36 (red)/smooth muscle actin (green) and SR-B1 (red)/hepatocytes (green) were performed using the same technique. A double-positive result was seen as yellow fluorescence.
The intracellular site of oxPC localization was examined by immunoelectron microscopy. Pre-embedding immunostaining technique was adopted. Briefly, 10-μm-thick sections cut from NASH/NAFLD liver tissue were placed on slides and stained with immunoperoxidase for oxPC. Immunoreaction was visualized with 3,3′-diaminobenzidine. The stained sections were post-fixed in osmium tetraoxide and embedded in epoxy resin. The slides were flamed briefly and the resin blocks were removed from the slides. Ultrathin sections were cut from the resin blocks, faintly stained with uranyl acetate and lead citrate, and observed under an electron microscope (JEM-1200 EX; JEOL, Tokyo, Japan).
In addition to the immunohistochemical examination, the expression of SR-B1 was investigated at the transcripts level. SR-B1 mRNA in the autopsy livers was quantified using a PCR-based technique. DNA-free RNA was extracted from homogenates of each liver sample by a modified acid guanidium/phenol/chloroform method with deoxyribonuclease (DNase I; Roche Diagnostics, Mannheim, Germany) digestion. The extracted RNA was converted into cDNA by reverse transcription (TaqMan reverse transcription reagents, Applied Biosystems, Foster City, CA). The cDNA solutions were analyzed by TaqMan technology, a real-time PCR assay (TaqMan PCR reagent kit, Applied Biosystems). Oligonucleotide primers and fluorescent dye–labeled probes used in this PCR assay are sense primer, 5′-GCG GCG GTG ATG ATG GA; antisense primer, 5′-GTG AAT GCC AAG GTC ATG ATG AG; and dye-labeled probe, 5′-CAG GGT CAT GGG CTT AT. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNA was also quantified using a specific primer-probe set (Applied Biosystems) to standardize the sample RNA amount. The PCR amplification was performed according to the manufacturer's recommendation. The PCR assay was performed in triplicate, and no template controls were prepared for each assay to check that cross-contamination did not occur.
The quantitative value was finally expressed as the thermo-cycling number at which the amount of PCR products reached a threshold level (Ct-value). As described in our previous report,26 we took the ΔCt value calculated by Ct(GAPDH) − Ct(SR-B1) as the relative amount of SR-B1 mRNA.
Statistical analyses were performed using nonparametric tests (StatView ver. 4.0, Abacus Concepts, Berkeley, CA). Differences between 2 groups were tested by Mann-Whitney U-test. Differences among more than 3 groups were tested by Kruskal-Wallis test. Associations between 2 parameters were evaluated using Spearman rank correlation coefficient (Rs value). P values of < .05 were considered statistically significant.
In normal livers, immunoreactivity for oxPC was observed in only a few macrophages/Kupffer cells (Fig. 1A). In livers of NAFLD, both simple steatosis and NASH, hepatocytes with lipid droplets were frequently stained for oxPC (Fig. 1C and Fig. 2A). Occasional swollen or ballooned hepatocytes also showed oxPC immunoreactivity in NAFLD (Fig. 2F). Similarly, 4HNE-positivity occurred predominantly in steatotic lesions and partially overlapped with oxPC-positive cells/areas but was more widely distributed than oxPC-positivity (Fig. 1B, D). Hence, oxPC-positivity seemed to be more specific to hepatocyte degeneration (steatosis and ballooning) than 4HNE-positivity.
Diverse degrees of inflammatory cell infiltration were seen in livers of NASH/NAFLD. Immunohistochemically, the inflammatory cell infiltration was composed mainly of lymphocytes (Fig. 2B) and macrophages (Fig. 2C). Neutrophils, which were positive for both neutrophil elastase and MPO, were also involved in inflammatory infiltration (Fig. 2D-E). These MPO-positive neutrophils were aggregated nearby the oxPC-positive cells/areas (Fig. 2A, D-F), and were frequently accompanied by necroinflammatory damage of hepatocytes. In contrast, hepatic iron deposition was frequently seen in NASH/NAFLD but was not identical with oxPC-positive areas (not shown).
Morphometric analysis clarified that oxPC-positive cells were significantly (P < .05) increased in NAFLD livers (0.8-111.0/mm2; median, 5.5/mm2) compared with normal livers (0-8.8/mm2; median, 3.2/mm2). Moreover, the positive cell number was significantly (P < .05) greater in NASH livers (1.6-111.0/mm2; median, 7.7/mm2) than in simple steatosis (0.8-15.8/mm2; median, 3.7/mm2). The NAFLD/NASH livers showed various disease severities, which were related to the number of oxPC-positive cells (Fig. 3A-B). In addition, the oxPC-positive cell number showed positive correlation with the neutrophil number (Fig. 4), but not with the lymphocyte or macrophage/Kupffer cell number. The neutrophil number was also correlated with both necroinflammatory grade (Rs = 0.73, P < .0001) and fibrosis stage (Rs = 0.77, P < .0001). Concerning hepatic iron deposition, hepatocytic iron score was significantly (P < .05) greater in NAFLD livers (simple steatosis, 0.8 ± 1.2; NASH, 1.3 ± 1.2) than in normal livers (0.2 ± 0.6); however, no significant correlations were seen between the iron scores of any sites (hepatocytic, sinusoidal, and portal) and the oxPC-positive cell number.
Immunoelectron microscopy showed detail of hepatocytic oxPC localization. In hepatocytes with lipid droplets, oxPC immunoreactivity was concentrated at the rim of the lipid droplets (Fig. 5A-B). Mitochondria of the hepatocytes with oxPC-positive lipid droplets tended to vary in size. Enlarged mitochondria, more than 2 μm in diameter, were occasionally seen adjacent to the oxPC-positive lipid droplets (Fig. 5C). OxPC immunoreactivity in swollen/ballooned hepatocytes (Fig. 5D) was observed on cytoplasmic and intracytoplasmic membranes (Fig. 5E-G). Ultrastructures of both types of oxPC-positive hepatocytes showed degenerative natures, compared to oxPC-negative intact hepatocytes (Fig. 5H). These oxPC-positive hepatocytes, especially ballooned hepatocytes, often had shrunken nuclei and markedly dilated endoplasmic reticulum (Fig. 5B, E). Furthermore, immunodouble staining demonstrated that a small number of oxPC-positive hepatocytes in NASH/NAFLD had ssDNA-positive nuclei, a hallmark of apoptosis (Fig. 6A). Such oxPC/ssDNA double-positive cells were not seen in normal livers.
OxPC localization in macrophages/Kupffer cells (Fig. 1A, C) was confirmed by immunodouble staining for oxPC/CD68 (Fig. 6B). This was prominent in NASH/NAFLD livers, and was linked to enhanced expression of scavenger receptors in macrophages/Kupffer cells in NASH/NAFLD livers (Fig. 7) compared with normal livers. The main expression sites of the scavenger receptors were macrophages/Kupffer cells and sinusoidal endothelial cells. CD36 expression also was seen in occasional activated stellate cells (Fig. 7B, inset), and SR-B1 expression was in some hepatocytes (Fig. 7C, inset). A real-time PCR assay revealed that the expression level of SR-B1 mRNA was significantly (P < .05) higher in NAFLD livers (1.48 ± 2.18) than in normal livers (0.04 ± 0.97), but was not correlated with the oxPC-positive cell number.
Although the precise pathogenic mechanism of NASH/NAFLD is still obscure, oxidative stress has been considered to play a crucial role in disease progression.4–7 A sophisticated 2-hit theory of the pathogenesis of NASH/NAFLD emphasizes oxidative stress as the most important factor that causes a “second hit.”4–7 As a consequence of oxidative tissue injury, many kinds of oxidation-related products, oxidatively modified lipids, proteins, and nucleotides, are generated.27 Thus, these products can be used as markers of oxidative stress.28, 29 It is known that lipid peroxides are not only markers of oxidative stress but also toxic substances themselves.30 Hence, lipid peroxides have been considered a potential cause that accelerates liver injury in steatosis. In the current study, we found that 4HNE, a representative lipid peroxide produced by oxidative injury,22 was more extensive in steatotic livers than in control livers, suggesting a potential contribution of lipid peroxides to the pathogenesis of steatotic liver disorders.
Oxidative stress and lipid peroxides also have been considered harmful and pathogenic in organs/tissues other than the liver. Especially in cardiovascular pathology, oxLDL has been recognized to play a pivotal role in the development and progression of atherosclerosis.8–10 OxLDL is phagocytosed by macrophages via binding of oxPC molecules present in oxLDL to macrophage scavenger receptors and changes the macrophages into foam cells.8–10 We recently developed a novel specific antibody against oxPC,17, 18 and using this antibody, we demonstrated the localization of oxLDL in macrophages accumulated in human atherosclerotic lesions.9–11 In the current study, we investigated the localization of oxPC in normal and NASH/NAFLD liver tissues using the same antibody. As a result, we found oxPC predominantly in steatotic hepatocytes and macrophages/Kupffer cells, and was more abundant in NASH/NAFLD livers than in normal control livers. These findings suggest a pathological relevance for oxPC localization to NASH/NAFLD.
One question arising from our immunohistochemical findings is the source of oxPC in the diseased livers. A plausible explanation is that fat stored in lipid droplets of hepatocytes may be oxidized by oxidation catalysts or reactive oxygen species.6, 31 Our immunoelectron microscopic observation, where oxPC immunoreactivity was seen at the rim of lipid droplets, is in accordance with this explanation. Another possibility is that oxPC localization in steatotic livers is the result of cellular uptake of oxLDL into macrophages/Kupffer cells and hepatocytes.25, 32 Our immunohistochemical findings on scavenger receptors, which serve for oxLDL uptake,25 expressed not only on macrophages/Kupffer cells but also on hepatocytes in NASH/NAFLD livers support the speculation. The expression level of SR-B1 mRNA was higher in NASH/NAFLD livers than in normal livers, indicating the strengthened potency of oxPC uptake of NASH/NAFLD livers. However, because SR-B1 mRNA expression was not correlated with the amount of oxPC immunoreactivity, the source of oxPC in NASH/NAFLD livers cannot fully be explained only by the uptake of the oxidized lipid.
oxPC can be generated not only from oxidative modification of LDL but also from oxidation of phosphatidylcholine in cytoplasmic and intracytoplasmic membranes by oxidative injury.16, 19, 33 We recently demonstrated oxPC localization in apoptotic pneumocytes in interstitial pneumonia.19 Similarly, in the current study, we found oxPC localization in apoptotic hepatocytes in steatotic livers, indicating that oxPC was actually formed in oxidatively damaged hepatocytes. Our immunoelectron microscopic observation further revealed the localization of oxPC on cytoplasmic and intracytoplasmic membranes of swollen or ballooned hepatocytes. These findings suggest that oxPC localized in steatotic livers is, at least in part, generated from membrane phospholipids of oxidatively damaged hepatocytes. The recent advance in lipid research suggests the existence of limiting membranes around intracellular lipid droplets.34 Thus, oxPC immunoreactivity at the rim of lipid droplets also may reflect the oxidation of membrane phospholipids.
According to a concept of Hazen and Chisolm16 concerning the cellular uptake of oxidized lipid products, oxidatively damaged and apoptotic hepatocytes as well as oxLDL are thought to be phagocytosed by macrophages/Kupffer cells via scavenger receptor–oxPC binding. Phagocytosis itself can intensify macrophage bio-activity.35, 36 These mechanisms may contribute to the initiation and promotion of inflammatory processes in NASH/NAFLD. We noted a significant correlation between the amount of oxPC immunoreactivity and the severity of liver damage, suggesting that oxPC may play an etiopathogenic role in steatosis-related hepatic inflammation.
Oxidative stress, which forms oxPC in damaged cells, is a process caused by reactive oxygen species, inflammation, and oxidation catalysts such as heavy metals and peroxidase.37, 38 Impaired mitochondrial oxidation has been considered to provide intracellular sources of oxidative stress in NASH/NAFLD.4–7, 31, 39 We found enlarged mitochondria adjacent to oxPC-positive lipid droplets. Although these mitochondria could not always be recognized as megamitochondria under light microscopy,40, 41 the abnormal morphology seemed to reflect impaired mitochondrial function, which may contribute to form oxPC-positive lipid droplets. Even in early stage of the disease without inflammatory infiltration, oxidative stress is believed to arise from the intracellular sources.
Excess hepatic iron can act as a pathogenic factor in several liver diseases, presumably through catalyzing hydroxyl radical production.21, 42, 43 NASH/NAFLD has been thought to be one of the diseases that are aggravated by excess hepatic iron.4, 7, 31 Oxidative tissue injury caused by excess iron may induce and augment lipid peroxidation and consequent liver fibrosis. In this study, we found increased hepatocytic iron deposits in NASH/NAFLD livers compared with normal livers, but there was no correlation between the degree of iron deposits and the oxPC-positive cell number. As noted in a previous report,44 hepatic iron overload was closely associated with insulin resistance and NASH/NAFLD. However, such excess hepatic iron may not be a source of oxidation for oxPC formation.
In contrast, MPO-positive neutrophil infiltration clearly showed both positional and quantitative associations with oxPC localization, and related to the severity of the liver disease. Neutrophil MPO is a potent oxidation enzyme and is considered to be the most powerful extracellular source of oxidative stress.45 Potentially, macrophages/Kupffer cells may also be involved in the MPO-positive inflammatory cells, as reported previously.46 In any case, increased MPO activity in steatotic livers may intensify oxdative stress and promote oxPC formation. Because oxPC can contribute to neutrophil recruitment,47 neutrophil accumulation around oxPC-positive hepatocytes may be intensified by positive feedback and may consequently exacerbate oxidative liver injury. Collectively, these data suggest that neutrophil MPO seems to play a pivotal role in oxidative liver injury in NASH/NAFLD.
The current study, as a whole, demonstrated that marked oxPC localization in NASH/NAFLD livers was accompanied by apparent steatosis and dense MPO-positive neutrophil infiltration, and was closely related to the severity of liver fibrosis. These findings suggest that predisposing steatosis and subsequent MPO-positive neutrophil infiltration are substantial factors in the oxPC formation and the fibrogenic process of NASH/NAFLD. Indeed, steatosis and neutrophil infiltration are known to be morphological characteristics of NASH/NAFLD.1, 4, 20, 41 Neutrophils have been reported to induce lipid peroxidation and stimulate collagen synthesis in human stellate cells.48 Alernatively, oxLDL and necrotic/apoptotic hepatocytes may exhibit a fibrogenicity triggered by stimulation of scavenger receptors expressed on stellate cells.24, 49 This was supported, at least in part, by our immunohistochemical result that showed CD36 expression on activated stellate cells in NASH/NAFLD livers. Taking these evidences into consideration, oxPC may be a key mediator connecting hepatic steatosis and neutrophil MPO-derived oxidative stress to hepatic fibrogenesis in NASH/NAFLD. To prove the hypothesis, however, further studies are required.
In conclusion, the current study, using immunohistochemical single and double staining methods and immunoelectron microscopic examination, revealed distinct oxPC localization in liver tissues of NASH/NAFLD. Neutrophil MPO-derived oxdative stress seems to be crucial in oxPC formation, and the formed oxPC potentially contributes to the progression of the inflammatory and fibrogenic processes via binding to scavenger receptors in NASH/NAFLD.
The authors thank Dr. K. Nakatani, Department of Anatomy, Osaka City University Graduate School of Medicine, for appropriate advice, and H. Nakagawa for technical assistance with immunoelectron microscopy.