Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
Potential conflict of interest: Nothing to report.
This work was supported by the National Institutes of Health (grant nos.: R01DK087696, K24DK02957, and K24DK02957.S1; to K.V.K.).
The aim of this study was to examine the relationship between the presence of hepatic iron deposition, apoptosis, histologic features, and serum markers of oxidative stress (OS) and cell death in nonalcoholic fatty liver disease (NAFLD). Clinical, biochemical, metabolic, and independent histopathologic assessment was conducted in 83 unselected patients with biopsy-proven NAFLD from a single center. Apoptosis and necrosis in serum was quantified using serum cytokeratin 18 (CK18) M30 and M65 enzyme-linked immunosorbent assays and in liver by terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining in situ. Serum malondialdehyde (MDA) and thioredoxin-1 (Trx1) levels were measured to evaluate OS. Presence of reticuloendothelial system (RES) cell iron in the liver was associated with nonalcoholic steatohepatitis (P < 0.05) and increased hepatic TUNEL staining (P = 0.02), as well as increased serum levels of apoptosis-specific (M30; P = 0.013) and total (M65; P = 0.006) CK18 fragments, higher MDA (P = 0.002) and lower antioxidant Trx1 levels (P = 0.012), compared to patients without stainable hepatic iron. NAFLD patients with a hepatocellular (HC) iron staining pattern also had increased serum MDA (P = 0.006), but not M30 CK18 levels or TUNEL staining, compared to subjects without stainable hepatic iron. Patients with iron deposition limited to hepatocytes had a lower proportion of apoptosis-specific M30 fragments relative to total M65 CK18 levels (37% versus ≤25%; P < 0.05). Conclusions: Presence of iron in liver RES cells is associated with NASH, increased apoptosis, and increased OS. HC iron deposition in NAFLD is also associated with OS and may promote hepatocyte necrosis in this disease. (HEPATOLOGY 2013)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Nonalcoholic fatty liver disease (NAFLD) affects approximately 30% of adults in the United States, closely mirroring the obesity epidemic and prevalence of metabolic syndrome.1 Nonalcoholic steatohepatitis (NASH), the severe form of NAFLD, is a multifactorial disease whereby the initial development of steatosis in the setting of insulin resistance is complicated by additional insults, such as oxidative damage, mitochondrial dysfunction, and endoplasmic reticulum stress.1 A potential contributing factor in many of these “second hits” is iron deposition.2 A recent study by our group showed that 35% of subjects enrolled in the NASH Clinical Research Network (NASH) had stainable hepatic iron.3 We also observed a relationship between the pattern of hepatic iron staining and disease severity in these patients; reticuloendothelial system (RES) cell iron staining alone was associated with advanced histologic features and a diagnosis of NASH, whereas iron staining exclusively in hepatocytes or a mixed hepatocellular (HC)/RES pattern was associated with comparatively less severe disease.3
Iron is known to increase cellular oxidative stress (OS) through production of reactive oxygen species (ROS) by catalyzing Fenton's reaction. ROS damages cell and organelle membranes through lipid peroxidation (LPO), causing altered membrane integrity and function.4 ROS can also cause oxidative damage to nucleic acids (e.g., strand breaks, base adducts, and molecular cross-links) and proteins (e.g., sulfhydryl oxidation, modification of prosthetic groups, fragmentation, or structural changes), contributing to the cytotoxic effect of cellular iron accumulation.5, 6
At different thresholds of oxidative damage, the processes of reparative autophagy, apoptosis, or necrosis can be induced by the release of lysosomal enzymes.7 Apoptosis can be induced by either extrinsic, death-receptor–mediated pathways or intrinsic, intracellular pathways. Extrinsic pathways, such as FAS and tumor necrosis factor receptor (TNFR), are thought to be dominant in NASH, but both extrinsic and intrinsic pathways are actuated by the mitochondrial release of cytochrome-c and initiation of apoptosis machinery by caspase-3 and -7.8, 9 Moderate lysosomal rupture has been shown to induce apoptosis, either through mitochondrial swelling and cytochrome-c release or direct caspase-3 activation and caspase-2 and -8 activation by lysosomal cathepsins.10 In contrast, large-scale lysosomal breakage is associated with necrosis and limited caspase activation.10 ROS-induced mitochondrial dysfunction initiates the intrinsic apoptosis cascade as a result of increased membrane permeability, release of cytochrome-c, and caspase-9/3 inhibition.11
Cytokeratin (or keratin) 18 (CK18) is a cytoskeletal intermediate filament protein found in simple epithelial cells, especially of the digestive tract, and, together with keratin 8, are the only keratin intermediate filaments in hepatocytes and variants of these genes are associated with susceptibility to a variety of liver diseases.12 Circulating CK18 fragments generated by apoptosis and necrosis, though detectable in serum of nonliver disease controls,13-17 are elevated in a variety of liver diseases,12-14 but have been most widely studied in NAFLD; numerous studies have validated CK18 as a serum biomarker of NAFLD and/or to distinguish NASH from simple steatosis.13-21 The commonly used M30 antibody identifies an apoptosis-specific neoepitope at CK18 aspartic acid residue 396, generated by cleavage of early caspase-9 and caspase-3 and -7 during the execution phase of apoptosis. The use of two monoclonal antibodies, M5 and M6 (called M65), allows for measurement of all CK18 fragments resulting from loss of cell membrane integrity from necrosis and/or apoptosis by caspase activity.22 Thus, concurrent measurement of M30 and M65 assays permits quantification of the relative contributions of apoptosis and necrosis to cell death in a particular etiology.22
The aim of the present study was to determine the relationship between serum markers of OS and apoptosis/necrosis (fragmented M30 and total M65 CK18 levels) and histologic measurement of NAFLD and apoptosis among NAFLD subjects with different hepatic iron pattern phenotypes. We aimed to test the hypothesis that increased NAFLD severity among patients with hepatic RES iron deposition is associated with increased apoptosis and systemic OS, possibly as a consequence of iron loading in macrophages and other RES cells.
The study population comprised patients with NAFLD enrolled in a NAFLD registry and repository between 2003 and 2011. Liver biopsies were performed as part of routine clinical evaluation to confirm and stage the diagnosis of NAFLD and exclude another cause for liver disease. All patients underwent standard clinical assessment with clinical, laboratory, and serologic testing to exclude other causes of liver disease. Those who had previously undergone liver transplantation, had a diagnosis of concomitant viral hepatitis, hemochromatosis, or secondary iron overload, or had any histopathologic diagnosis other than NAFLD were excluded from the study. Demographic data, standard laboratory tests, and serum were collected within 6 months of biopsy. Seventy-nine consecutive patients who had given written informed consent to participate in the NAFLD registry were selected for inclusion into the study based on the availability of complete clinical and laboratory data, liver biopsy specimens for independent pathologic review, and terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining and stored serum for measurement of apoptosis and OS markers. Four patients with duplicate biopsies (mean time elapsed: 5 years 29 days) and complete associated data and specimens were included in the study as independent data points. All biopsies were stained for iron using Perls' stain and were classified as no iron, HC iron, or RES iron (including patients with a mixed HC/RES pattern). The study was approved by the institutional review board of the Benaroya Research Institute (Seattle, WA).
Histologic assessment for features of NAFLD using the NASH CRN scoring system, including presence and grade of steatosis, lobular inflammation, fibrosis, and ballooning, was completed by a single hepatopathologist with expertise in NASH (M.M.Y.) who was blinded to all clinical and laboratory data.23 The NAFLD activity score (NAS), determined by the sum of the steatosis, lobular inflammation, and ballooning scores, was calculated. In addition, a diagnosis of definite NASH, or not NASH (including borderline or possible NASH) was rendered. Biopsies were scored for the presence and grade of HC and RES iron using Perls' staining, as previously described.3
Measurement of Cell Death and OS in Blood.
Serum levels of malondialdehyde (MDA), a by-product of LPO, were determined using the Cayman TBARS Assay Kit (Cayman Chemical Company, Ann Arbor, MI), following the manufacturer's instructions. Thioredoxin-1 (Trx1), a small protein with antioxidant and anti-apoptotic properties, was also measured in serum by enzyme-linked immunosorbent assay (ELISA), following the manufacturer's instructions (Thioredoxin-1 kit; Northwest Life Science Specialties, Vancouver, WA). Serum levels of fragmented CK18 from apoptosis (M30) and total CK18 levels from the combination of apoptosis and necrosis (M65) were determined by ELISA, following the manufacturer's instructions (Peviva, Stockholm, Sweden). Cell death resulting from apoptosis was calculated by the following equation: % Apoptosis = [(M30 CK18/M65 CK18)*100]; likewise, by definition, the remaining amount of cell death is considered to be the result of necrosis and was calculated according to the formula, 100 − [(M30 CK18/M65 CK18)*100].
TUNEL was performed using the ApopTag Plus Peroxidase In Situ Apoptosis Detection Kit (Chemicon, Billerica, MA), following the manufacturer's instructions; diaminobenzidine (DAB; Invitrogen, Carlsbad, CA), prepared just before use, and 5 μL of NiCl2 (8% in distilled water) per milliliter of DAB was substituted for the peroxidase substrate to enhance visualization of dye. Specimens were counterstained using 2% methyl green for 4 minutes at room temperature, followed by three washes each of distilled water, absolute ethanol, and xylene. Quantification of apoptosis was determined by counting apoptotic cells in five random fields per patient at 400× magnification. Because the number of cells in each field varies widely between patients with ranging levels of steatosis, it was important to obtain a measure of cellularity for each patient. To estimate the total number of cells per field, two representative fields per patient were selected and overlaid with a 10 × 10 grid in ImageJ (National Institutes of Health, Bethesda, MD). Ten squares within the grid were analyzed for the number of cells and this number was multiplied by 10 to approximate the total number of cells per field. Percent apoptosis was expressed as the average number of apoptotic cells/average total number of cells for each group. Scoring was performed in a blinded fashion through two independent assessments, and the mean of these two scores was used for final analysis.
Demographic, clinical, and laboratory characteristics were recorded as number and percentage for categorical data and means and standard error or median and interquartile range for continuous data. Categorical data were analyzed using Fisher's exact test. Continuous variables, including laboratory measures, were not normally distributed and were analyzed using nonparametric statistics, including Mann-Whitney's rank-sum test and Spearman's rank-correlation test or linear regression modeling adjusting for sex as a potential confounding variable. Differences in mean histological scores, such as HC and RES iron grade, steatosis grade, fibrosis stage, lobular inflammation grade, and NAS between groups, were analyzed using ordinal logistic regression. Multivariate stepwise linear regression (cutoff: P < 0.20) analysis was used to investigate the independent association of HC iron and percent CK18 levels resulting from necrosis after adjustment for sex, age, body mass index (BMI), alanine aminotransferase (ALT), and presence of diabetes. All statistical analysis was performed using STATA version 12.1 (StataCorp LP, College Station, TX).
Clinical and laboratory data of the study cohort are shown in Table 1. Forty-seven patients (57%) were negative for hepatic iron staining and were categorized as “no iron”; 27 patients each stained positive for HC iron or RES iron, including 18 positive for both HC and RES iron staining (i.e., a mixed HC/RES phenotype). Patients with either HC or RES iron deposits were more likely to be male and had a lower BMI, compared to patients without iron. No significant differences were observed between either the iron group and the iron-negative group in age, proportion of Caucasians, or presence of diabetes mellitus (DM) or obesity. Initially there were significant differences in several lab measurements (e.g., ALT and aspartate aminotransferase [AST]), but most of these differences failed to remain significant after adjusting for sex in the analysis. Interestingly, even after adjusting for sex, serum ferritin values were still significantly higher in both iron groups, compared to the no-iron group (see Table 1). Subjects with either HC or RES iron had higher overall mean NAFLD histologic scores, compared to those without iron; the NAS index and ballooning scores were significantly higher among patients with RES iron, compared to those with no iron (Table 2). Subjects with RES iron staining were also significantly more likely to have a definitive diagnosis of NASH, compared to subjects without iron (76% versus 34%; P < 0.05).
Table 1. Clinical and Serologic Differences of Patients According to Hepatic Iron Stain Status
Values are presented as numbers and percentages or medians and interquartile ranges, except for BMI and age (means and standard deviations).
Table 2. Histological Differences Between Patients According to Hepatic Iron Stain Status
Values are presented as mean ± standard error of the mean, except for presence of NASH diagnosis (numbers and percentages). All P values were determined by ordinal regression, except the presence of NASH diagnosis (Fisher's exact test).
NAFLD Patients With RES Iron Have Increased TUNEL Staining and OS.
TUNEL staining was performed to investigate the relationship between hepatic iron and apoptosis (Fig. 1). Patients with RES iron showed an increased mean percentage of TUNEL-positive cells, compared to no-iron patients (6.9 versus 4.8; P = 0.02). However, no significant differences were observed between patients with HC iron, compared to iron-negative patients. There was a trend toward a positive association between percentage of TUNEL-stained cells and grade of RES iron (r = 0.33; P = 0.01), but not HC iron. There was also a trend toward a positive association between percentage of TUNEL-stained cells and total M65 CK18 levels (r = 0.36; P = 0.004), but not M30 CK18 levels.
To investigate the relationship between the presence and pattern of hepatic iron stores and OS in vivo, levels of the LPO product, MDA, and the antioxidant/antiapoptotic protein, Trx1, were assayed in serum (Fig. 2). MDA levels were increased in subjects with either HC (P = 0.006) or RES iron deposits (P = 0.002), compared to iron-negative subjects. Both HC and RES iron-positive patients also demonstrated lower Trx1 levels, and the differences were statistically significant in patients with RES iron (P = 0.012). Overall, in both HC and RES iron-positive subjects, there was an inverse relationship between MDA and Trx1 (r = −0.50; P < 0.0001), suggesting opposing effects on these markers by hepatic iron loading; iron-positive subjects showed relatively increased MDA and decreased Trx1, whereas the opposite was true in patients without hepatic iron loading (Fig. 2). There was also a trend for a positive association between MDA levels and both percentage of TUNEL staining (r = 0.30; P = 0.02) and RES iron grade (r = 0.32; P = 0.01).
CK18 Levels Suggest More Cell Death in Patients With RES Iron, but a Higher Proportion Resulting From Necrosis in Patients With HC Iron.
A comparison of fragmented CK18 levels from apoptosis alone (M30) and total CK18 levels from the combination of apoptosis and necrosis (M65) for each group, relative to the total M65 level in subjects without iron staining, is shown in Fig. 3. Patients with RES iron had the highest M30 and M65 levels, which were more than twice the levels in patients without stainable iron (P = 0.013 and 0.006, respectively). The level of M65 (P = 0.043), but not M30, was significantly different between patients with HC iron and no iron.
There was a significant difference in the calculated percentage of CK18 levels resulting from apoptosis or necrosis (% apoptosis = [(M30 CK18/M65 CK18)*100]) between the iron phenotype groups (P = 0.047; Fig. 4). Cell death estimated from CK18 M30/M65 levels in patients without hepatic iron was almost exclusively the result of apoptosis (97%). In contrast, the proportion of CK18 resulting from apoptosis in patients with hepatic iron deposition ranged from 63% to 80%. Patients with HC-only iron had the highest proportion of CK18 attributable to necrosis (37%). To determine whether the presence of HC iron was independently associated with increased cell death resulting from necrosis, we performed stepwise multivariate linear regression analysis and adjusted for the following potential confounding variables known to be associated with NASH severity: age, sex, BMI, ALT, and presence of diabetes. In this statistical model, HC-only iron was the only variable independently associated with necrosis-associated CK18 (P = 0.011).
Previous studies have shown an association between the presence and pattern of hepatic iron deposition and disease severity in NAFLD.2, 3 However, the mechanisms behind this association remain to be fully elucidated. The aim of the current study was to investigate the hypothesis that hepatic iron deposition in RES cells was associated with increased apoptosis and serum markers of oxidative stress in NAFLD. As in our previous multicenter study of 849 NAFLD patients (NASH CRN), we found that RES iron was associated with NASH and more-severe histologic features, whereas patients with HC iron deposition more frequently had milder pathology.3 We found that the presence of both HC and RES hepatic iron in NAFLD patients was associated with increased OS (i.e., greater LPO in the presence of decreased antioxidant capacity), as suggested by higher MDA levels and decreased Trx1 levels, compared to NAFLD patients without iron staining. However, only RES iron was associated with significantly increased apoptosis, as shown by greater numbers of TUNEL-positive cells in the liver and higher serum levels of total and apoptosis-specific CK18 fragments. We also found a positive association between RES iron grade and both MDA levels and amount of TUNEL-positive cells, providing indirect evidence that RES iron-induced OS may mediate apoptosis. Subjects with HC iron had lower CK18 levels, compared to subjects with RES iron, but with a higher proportion attributable to necrosis. This observation supports our previous hypothesis that HC iron contributes to NAFLD through a direct cytotoxic effect.3
Several in vitro and in vivo studies provide evidence that exogenous iron induces apoptosis by increasing OS.24-27 Moreover, addition of the iron chelators, baicalin,24 quercertin,24 deferasirox,25 or deferoxamine,24, 25 all were shown to reduce various markers of OS and prevent apoptosis. Lin et al. have recently provided further mechanistic insight into iron-mediated cell death; adiponectin (APN) gene therapy in C57BL/6J mice before intraperitoneal injection of iron dextran, limited ROS production and protected animals from apoptosis.26 APN treatment resulted in phosphorylation of adenosine-monophosphate–activated protein kinase, peroxisome proliferator-activated receptor alpha (PPAR-α) activation, and, finally, induction of heme oxygenase-1 (HO-1) through binding of PPAR-α to a PPAR-α-responsive element in the promoter of HO-1.26 Interestingly, APN treatment also resulted in decrease hepatic iron staining, compared to iron challenge alone, which was reversed by inhibition of HO-1 with tin protoporphyrin, demonstrating the role of HO-1 in iron recycling.26 Iron exposure of isolated murine hepatocytes also led to nuclear factor kappa B (NF-κB) activation and phosphorylation of p38 and extracellular signal-related kinase mitogen-activated protein kinases concomitant with ROS generation, reduced mitochondrial membrane potential, cytochrome c release, activation of the caspase cascade, and cleavage of poly (ADP-ribose) polymerase protein.27
A novel finding of this study was the greater degree of apoptosis with RES iron deposition, despite similar levels of serum indicators of OS in both HC and RES iron subjects. Apoptosis is usually observed in NASH patients concomitant with onset of inflammation and is also thought to develop by several signaling mechanisms, including OS, as well as increased proinflammatory cytokine production.8, 9, 28 Kupffer cells (KCs), the main constituent of RES, are key effectors of the phenotype of surrounding cells, including other KCs, hepatocytes, stellate cells, and neutrophils, by the production of ROS, cytokines, chemokines, nitric oxide, and prostaglandins.29-31 For example, KCs are a major source of tumor necrosis factor alpha (TNF-α) in the liver, which promotes recruitment of neutrophils and other inflammatory cells, induces OS, cytotoxicity, and ultimately apoptosis in hepatocytes and other cells.29, 30 Ribeiro et al. have shown, in human NASH, that TNF-α-mediated induction of the caspase cascade by the death receptor, TNF-R1, is a major apoptotic pathway and is correlated with NF-κB expression as well as disease severity.28 In addition, Peng et al. have shown that Fas ligand (FasL) gene expression is mediated by NF-κB and inhibition of NF-κB-attenuated apoptosis, but not TNF-α expression.32 In addition to expression of TNF-α and activation of the NF-κB pathway, increased ROS and OS also promote apoptosis by activation of the Jun N-terminal kinase (JNK)/activation protein 1 serine kinase-signaling cascade.33, 34 Tsukamoto and others have shown that addition of iron activates KCs both in vitro and by erythrophagocytosis, inducing LPO, NF-κB activation, and NF-κB-mediated TNF-α expression and release, which was abrogated by iron chelation treatment.31, 35-39 Last, phagocytosis by KCs results in expression of TNF-α and death receptors FasL and TNF-related apoptosis-inducing ligand, suggesting a feed-forward amplification of apoptosis.40 Taken together, these studies suggest a role of KC iron in apoptosis by the FAS, JNK, and TNFR pathways through production of ROS, cytokines, NF-κB and TNF-α, which could then be amplified through phagocytosis of erythrocytes and iron-containing apoptotic hepatocytes.
There are a number of cellular conditions that are thought to favor either necrosis or apoptosis, which potentially could explain our observations that HC iron may promote greater necrosis, compared to the other iron phenotypes. Apoptosis is a deliberate, adenosine triphosphate (ATP)-dependent process that usually occurs gradually, whereas necrosis is a rapid event involving plasma membrane rupture subsequent to ATP depletion; thus, availability of ATP is recognized as a key determinant for which mode of cell death predominates.10 Iron-mediated mitochondrial LPO contributes to pore formation in mitochondrial membranes or mitochondria permeability transition (MPT) and subsequent release of mitochondrial ROS.41, 42 Both necrosis through ATP depletion or caspase-dependent apoptosis induced by cytochrome c release are consequences of MPT, but the degree of mitochondria involvement may determine the extent of ATP depletion and hence the development of necrosis or apoptosis.29 Depletion of the antioxidant, glutathione (GSH), in both the mitochondrial and cytosolic compartments has been shown to promote OS-induced necrosis, whereas selective cytosolic GSH depletion sensitizes hepatocytes to TNF-α-induced apoptosis independent of OS.43, 44 Several studies have investigated the origins of cell death by necrosis or apoptosis in cultured hepatocytes or using in vivo animal models subsequent to chemically induced superoxide formation using menadione or diquat.33, 45 Evidence from these studies suggests that when extensive oxidant damage overwhelms the cellular antioxidant capacity, necrosis may result, whereas with moderate OS, apoptotic pathways predominate.
There are some limitations of our study worth noting, such as the possible effect of elevated MDA levels after prolonged serum storage,46 potentially explaining higher levels in subjects with hepatic iron. However, using linear regression modeling, including MDA level, duration of serum storage, and presence of iron, only the presence of iron (beta = 0.45; P < 0.001) and not serum storage duration (beta = 0.03; P = 0.786) was significantly associated with MDA levels. We also recognize that this cross-sectional study cannot imply causation, and that additional mechanistic studies will be required to definitively establish the role of RES cell iron in the induction of various apoptotic pathways.
In conclusion, we found that both HC and RES iron deposition are associated with increased OS, compared to patients without stainable iron. However, RES, but not HC iron deposition, in this disease is characterized by increased apoptosis in the liver, as shown by both in situ TUNEL staining in the liver and circulating CK18 levels and suggests a mechanism for increased disease severity. By contrast, our data suggest that selective HC iron deposition in NAFLD may preferentially promote cell necrosis versus apoptosis. Longitudinal studies will also help delineate the dynamic relationship of iron and OS and the role of these in NAFLD progression.
The authors thank Andrew Rhieu and Jilene Chua for their work in TUNEL assay development and imaging and Priya Handa for her helpful discussions.