Potential conflict of interest: Nothing to report.
Supported by NIH R01 DK52951 and the Department of Veterans Affairs (to M.B.O.); NIH Michigan Gastrointestinal Peptide Research Center P30 DK34933; and NIH F32 DK093202 (to A.S.).
Mallory-Denk bodies (MDBs) are hepatocyte inclusions commonly seen in steatohepatitis. They are induced in mice by feeding 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC) for 12 weeks, which also causes porphyrin accumulation. Erythropoietic protoporphyria (EPP) is caused by mutations in ferrochelatase (fch), and a fraction of EPP patients develop liver disease that is phenocopied in Fechm1Pas mutant (fch/fch) mice, which have an inactivating fch mutation. fch/fch mice develop spontaneous MDBs, but the molecular factors involved in their formation and whether they relate to DDC-induced MDBs are unknown. We tested the hypothesis that fch mutation creates a molecular milieu that mimics experimental drug-induced MDBs. In 13- and 20-week-old fch/fch mice, serum alkaline phosphatase, alanine aminotransferase, and bile acids were increased. The 13-week-old fch/fch mice did not develop histologically evident MDBs but manifested biochemical alterations required for MDB formation, including increased transglutaminase-2 and keratin overexpression, with a greater keratin 8 (K8)-to-keratin 18 (K18) ratio, which are critical for drug-induced MDB formation. In 20-week-old fch/fch mice, spontaneous MDBs were readily detected histologically and biochemically. Short-term (3-week) DDC feeding markedly induced MDB formation in 20-week-old fch/fch mice. Under basal conditions, old fch/fch mice had significant alterations in mitochondrial oxidative-stress markers, including increased protein oxidation, decreased proteasomal activity, reduced adenosine triphosphate content, and Nrf2 (redox sensitive transcription factor) up-regulation. Nrf2 knockdown in HepG2 cells down-regulated K8, but not K18. Conclusion: Fch/fch mice develop age-associated spontaneous MDBs, with a marked propensity for rapid MDB formation upon exposure to DDC, and therefore provide a genetic model for MDB formation. Inclusion formation in the fch/fch mice involves oxidative stress which, together with Nrf2-mediated increase in K8, promotes MDB formation. (Hepatology 2012;56:322–331)
Erythropoietic protoporphyria (EPP) is an inherited disorder caused by mutations in the ferrochelatase (Fch) gene.1, 2 More than 40 molecular defects have been described in Fch gene in EPP patients.3 Mitochondrial ferrochelatase catalyzes the insertion of ferrous iron into protoporphyrin IX (PP-IX), thereby regulating heme biosynthesis.1 Reduced ferrochelatase activity in EPP causes excessive accumulation of PP-IX in red blood cells (RBCs), skin, and liver.4 The disease is characterized by cutaneous photosensitivity as PP-IX becomes phototoxic upon light exposure.4 Approximately 20% of patients exhibit hepatic manifestations, and 5%-10% progress to endstage liver disease.4 Genetic background has been suggested as a key determinant in the variable clinical symptoms in EPP.3
The Fechm1Pas mutant Balb/c mice (fch/fch) were previously reported by others.5 These mice harbor a point mutation in the ferrochelatase gene (resulting in 95% enzymatic activity loss) and suffer from phototoxicity, hemolytic anemia, and severe hepatic dysfunction.5 They have elevated levels of serum transaminases, bilirubin, and hyperlipidemia.6 Fechm1Pas mice develop biliary and parenchymal hepatic injury as evidenced by the presence of hepatocyte ballooning, acidophil bodies, necrosis, and Mallory-Denk bodies (MDBs).7 MDBs are markers of hepatocellular injury and are seen after feeding mice for 12 or more weeks with the porphyrinogenic compounds griseofulvin or 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC).8 Interestingly, MDB formation has not been described in EPP patients and the characterization and mechanism of MDB formation in the fch/fch mice has not been addressed.
MDBs are cytoplasmic hyaline inclusions that associate with various liver disorders, including alcoholic and nonalcoholic steatohepatitis.8 In the context of hepatitis C virus infection, MDB presence is associated with poor prognosis.9 Biochemically, MDBs are composed primarily of the intermediate filament proteins keratins 8 and 18 (K8/K18), ubiquitin (Ub), and the ubiquitin-binding protein p62.8 The major molecular processes involved in MDB formation include induction of K8/K18 expression with an increase in K8-to-K18 ratio, transglutaminase-2 (TG2) induction, resulting in transamidation and keratin crosslinking, and increased p62 levels.8 TG2 induction concomitant with an increase in K8-to-K18 ratio leaves excess K8 that is unbound to its obligate binding partner K18, which is significant because K8 appears to be a preferential TG2 substrate (compared with K18) and TG2-null mice do not form DDC-induced MDBs.10
Oxidative stress strongly associates with DDC-induced MDB formation in FVB/N mice (male>female) due to differences in DDC metabolite formation.11 Similarly, C3H mice form far fewer MDBs, as compared with C57BL mice, due to their ability to maintain levels and subcellular distribution of the protective enzymes glyceraldehyde-3-phosphate-dehydrogenase and nucleoside-diphosphate-kinase under oxidative stress.12 Additionally, S-adenosyl methionine prevents the formation of MDBs in DDC-primed mice by, in part, normalizing an otherwise compromised proteasomal activity.13 Expression-profile analysis of fch/fch mouse livers showed major alterations in several oxidative stress-response genes including glutathione-S-transferase, superoxide dismutase, and nuclear factor E2 derived 2-like-2 (Nrf2).6 Further, exogenous treatment of HepG2 cells with protoporphyrin increases intracellular H2O2, thereby contributing to oxidative stress.14
We hypothesized that fch mutation creates a molecular milieu that mimics experimental drug-induced MDBs. Upon testing this hypothesis we delineated the genetic susceptibility of Fechm1Pas mice to spontaneous and drug-induced MDB formation using various biochemical and molecular approaches. Our results support the importance of oxidative stress and Nrf2 in both spontaneous and DDC-induced MDB formation in Fechm1Pas mice.
K8 (Troma-I) (Developmental Studies Hybridoma Bank); glutathione-S-transferase-M-1 (GSTM1), Ub, and Nrf2 (Santa Cruz Biotechnology); mouse/human K8 and K18 (antibody-8592); mouse/human K18 (antibody-4668)11; p62 (Sigma-Aldrich); TG2, β-tubulin, human K8 (TS1), and human K18 (DC10) (Thermo Scientific).
All animal studies were approved by the Animal Use and Care Committee (University of Michigan). Fechm1Pas mice (Balb/c background; Jackson Laboratories) were screened using a polymerase chain reaction (PCR)-based protocol. The 13- and 20-week-old mice were used to study spontaneous MDB formation. MDB induction in 20-week-old mice was done by feeding with LabDiet 5001 (PMI Nutrition International) containing 0.1% DDC (Sigma-Aldrich) (3 weeks). Age-matched controls were fed a standard mouse diet. FVB mice were also fed 0.1% DDC (2 to 10 days) to examine oxidative stress and proteasome inhibition. Mice were euthanized by CO2 inhalation and intracardiac blood was collected for measurement of alkaline phosphatase (ALP), alanine aminotransaminase (ALT), and bile acids (BA) (using the Vetscan-vs2 instrument; Abaxis). Livers were harvested and apportioned for: hematoxylin-eosin (H&E) staining, biochemical and molecular analysis, and immunofluorescence staining.
RNA Extraction and Quantitative Real-Time PCR.
Total RNA was extracted utilizing Qiagen RNeasy kits. TaqMan Reverse Transcriptase kit (Applied Biosystems) was used to translate RNA into complementary DNA (cDNA), which was subjected to real-time PCR (MyiQ real time PCR detection system, Biorad) and amplified by Brilliant SYBR green master mix utilizing gene specific primers (Supporting Table 1).
Total Lysates, High Salt Extracts (HSE) Preparation, and Immunoblotting.
For total lysates, livers were homogenized in sodium dodecyl sulfate (SDS)-containing sample buffer. HSE were prepared as described15 to obtain keratin-enriched fractions. Equal protein amounts were loaded (4%-20% gradient SDS-polyacrylamide gels) and stained by Coomassie blue or transferred to polyvinylidene-difluoride membranes and immunoblotted using relevant antibodies.
Histological and Immunofluorescence Studies.
H&E-stained liver sections were assessed for MDBs. For immunofluorescence studies, liver sections were fixed in acetone, air-dried, and stained as described.16 MDBs were counted using a 40× lens and seven fields were examined per liver specimen.
Mitochondria Isolation, Protein Oxidation Assay, and Adenosine Triphosphate (ATP) Content.
Liver mitochondria were isolated as described.17 Mitochondrial lysates were analyzed for oxidized proteins using an OxyBlot Protein Oxidation Detection Kit (Millipore). DNP signals were quantified using ImageJ software and expressed as relative intensity values (mean ± standard error of the mean [SEM]). Mitochondrial ATP content was measured using an ATP determination kit (Invitrogen).
Nuclear Extracts and 20S Proteasome Activity.
Nuclear extracts were prepared using nuclear and cytoplasmic extraction reagent kit (Pierce Biotechnology). Extracts were immunoblotted with anti-Nrf2 antibody. Liver 20S proteasomal activity was determined using the 20S proteasome assay kit (Cayman Chemical).
Cell Culture and Transfections.
HepG2 hepatoblastoma cells were transfected with control or Nrf2 small interfering RNA (siRNA) (Santa Cruz Biotechnology) using Lipofectamine RNAiMax (Invitrogen) (48 hours), then used to prepare RNA extracts and protein lysates.
One-way analysis of variance (ANOVA) with Dunnett's post test or unpaired t test was used for statistical analysis. Differences were considered significant at P < 0.05.
Fechm1Pas Mice Exhibit Spontaneous MDB Formation.
The presence of MDBs was previously reported in a mouse model of EPP,7 although the underlying mechanisms are unknown. We carried out detailed biochemical, serologic, and histological analyses of the livers of these animals. Initial assessment of 20-week-old fch/fch mice and their wildtype (wt/wt) and heterozygous (wt/fch) controls indicated no body weight differences, whereas the percent liver-to-body weight ratio was 1.7 to 1.8-fold higher in fch/fch mice relative to wt/wt and wt/fch mice (Table S2). Serological analysis of the 20-week-old fch/fch mice revealed increased ALP, ALT, and BA levels (Fig. 1A), in agreement with previously reported hepatobiliary injury in fch/fch mice.6, 18
Given that an essential component of drug-induced MDB formation is an up-regulation of keratins with a K8>K18 ratio and formation of high molecular weight K8-containing crosslinked species, we examined if the same features occur in the genetic fch model. As shown in Fig. 1B, K8 and K18 messenger RNA (mRNA) levels were increased in fch/fch mice with a K8-to-K18 ratio of 1.94. Biochemical analysis of fch/fch livers revealed increased K8/K18 expression (per mg of liver tissue) as shown by Coomassie staining of the HSE in Fig. 1C. Immunoblotting showed the presence of K8-ubiquitin high-molecular-weight complexes and increased expression of p62 and TG2 in fch/fch mice (Fig. 1C). These data provide biochemical support for spontaneous MDB formation in 20-week-old fch/fch mice. Densitometric analysis showed ≈2-fold increase in K8/K18 protein in fch/fch livers (Fig. 1D).
Susceptibility of younger Fechm1Pas mice (13-week-old) to MDB formation was also examined. Similar to the 20-week-old mice, 13-week-old fch/fch mice showed a 1.8-1.9 higher percent liver/body weight (Table S2), and had elevated levels of ALP, ALT, and BA (Supporting Fig. S1A). The 13-week-old mice showed K8 and K18 mRNA induction, with a K8:K18 ratio of 1.25 (Fig. S1B) that is less than the 1.94 ratio seen in the 20-week-old fch/fch mice (Fig. 1B). K8, K18, and TG2 protein levels were increased in 13-week-old fch/fch mice (Fig. S1C). These mice exhibited K8 dimer formation in the absence of K8 high molecular weight complexes.
MDB Formation in Fechm1Pas Mice Is Age-Dependent and Gender-Independent.
The above data suggest that young mice begin to show some of the initiating events of MDB formation (increased keratins/TG2) but have not yet progressed to mature MDBs that can be visualized histologically, because H&E staining revealed the presence of MDBs in 20-week (Fig. 2A, panels c,d), but not in 13-week-old, fch/fch mice (Fig. S2, panels c,d). This was further confirmed biochemically by the limited formation of high-molecular-weight K8-containing complexes and lack of induction of p62, another key constituent of MDBs (Fig. 2B). Immunofluorescence staining revealed prominent keratin and ubiquitin colocalized deposits in 20-week-old (Fig. 3, panels c,d) but not in 13-week-old fch/fch livers (Fig. S3), supporting our histological and biochemical findings. Keratin and ubiquitin staining for 20-week-old livers is shown in Fig. S4. Taken together, these data indicate that MDB formation in Fechm1Pas mice occurs in an age-dependent manner.
Given that male FVB/N mice are more susceptible to drug-induced MDB formation,11 potential gender differences in spontaneous MDB formation were analyzed. Notably, Fechm1Pas mice (Balb/c background) did not show a gender-based (M>F) predilection to MDB formation which was confirmed using biochemical (Fig. 2C) and histological (not shown) means.
Fechm1Pas Mice Have Increased Susceptibility to Drug-Induced MDBs.
The possibility that the underlying genetic predisposition of spontaneous MDBs renders Fechm1Pas mice more susceptible to drug-induced MDBs was tested using a modified, 3-week DDC feeding protocol, which is significantly shorter than the standard 12-16 weeks.8 Short-term DDC feeding markedly induced K8/K18 mRNA in all three genotypes relative to their non-DDC-fed controls (Fig. 4A). Biochemically, wt/fch mice respond by a dramatic up-regulation of K8 and K18 (compare Fig. 4B, lanes 2+3 with 5+6), which parallels increased K8 crosslinking (Fig. 4B). There was a significant increase in MDB numbers in wt/fch mice (13.2 ± 0.4) after DDC feeding, whereas MDB numbers was 16.7 ± 1.9 in non-DDC-fed fch/fch mice and 19.2 ± 2.4 in DDC-fed fch/fch mice. There was also an increase in TG2 levels in DDC-fed mice (Fig. 4B). Histological analysis confirmed the increase in DDC-induced MDBs in wt/fch and fch/fch mice (Fig. 5). These data indicate that the Fechm1Pas mice have a markedly enhanced propensity to form drug-induced MDBs.
Mitochondrial Oxidative Stress Is Altered in Fechm1Pas Mice.
Prior expression-profile analysis of livers from Fechm1Pas mice revealed alterations in genes involving heme metabolism, inflammation, and oxidative stress.6 Because mitochondria play an important role in reactive oxygen species generation, we examined several mitochondrial oxidative stress markers. fch/fch livers had increased protein oxidized by-products relative to wt/wt and wt/fch livers (Fig. 6A). Additionally, glutathione-S-transferase-mu-1 (GSTM1, induced during oxidative stress19) was markedly enhanced under basal conditions in the fch/fch livers (Fig. 6B, compare lanes 9+10 with 1+2 or 5+6).
Oxidative damage to mitochondria can lead to injury through ATP depletion. Notably, there was a significant decrease in ATP content in fch/fch livers relative to wt/wt and wt/fch livers, and DDC feeding led to further reduction of ATP, particularly in wt/fch and fch/fch livers (Fig. 6C). Given that increased oxidative stress has been shown to decrease proteasomal activity, which is associated with MDB formation,13, 20 we tested and observed a 2-fold inhibition of the 20S proteasomal activity in the fch/fch mice and also in the DDC fed wt/wt, wt/fch, and fch/fch mice (Fig. 6D). These results suggest that oxidative stress in fch/fch mice decreases proteasomal activity.
Fechm1Pas Mice Show Increased Expression of Nrf2.
Nuclear factor E2 derived 2-like-2 (Nrf2) is a redox sensitive transcription factor expressed in the liver.21 Nrf2 regulates multiple antioxidant pathways21 and its mRNA is up-regulated in Fechm1Pas mice based on unvalidated microarray studies.6 Therefore, mRNA levels and nuclear protein expression of Nrf2 in control and DDC-fed wt/wt, wt/fch, and fch/fch livers were analyzed. There was increased Nrf2 liver mRNA expression (≈3 fold) in fch/fch relative to wt/wt mice (Fig. 7A). Short-term DDC feeding induced liver Nrf2 mRNA levels in all mice. Consistent with the mRNA findings, control and DDC-fed fch/fch mice had significant induction in nuclear Nrf2 protein (Fig. 7B). These results indicate Nrf2 up-regulation in control and DDC-fed fch/fch mice, likely in response to oxidative stress. To further examine the sequence of events, FVB mice were fed DDC for 2-10 days. Analysis of the DDC-fed livers showed an increase in Nrf2 mRNA and protein expression after 5 days of DDC feeding (Fig. S5A,B). However, there was no inhibition in proteasome activity in 2-day and 5-day DDC-fed livers (Fig. S5C), suggesting that there is another mechanism involved in increasing Nrf2 expression. Further, we examined the mitochondrial oxidized proteins, and Oxyblot analysis (Fig. S5D) showed an increase in oxidized protein by-products in 5-day DDC-fed mice, suggesting that oxidative stress is upstream of proteasome inhibition. Nrf2 protein levels were greatly stabilized in M132-treated HepG2 cells, suggesting that proteasome inhibition also help stabilize Nrf2 protein (Fig. S5E).
Nrf2 Regulates K8 but Not K18 Expression.
Nrf2 was shown previously to regulate keratin-16 (K16) gene expression in human keratinocytes.22 Therefore, we examined the effect of Nrf2 on K8 and K18 expression in HepG2 cells using siRNA knockdown. Nrf2 siRNA but not control siRNA significantly inhibited K8 but not K18 mRNA and protein expression (Fig. 7C,D). These findings suggest that the K8 gene is a target for Nrf2.
Relevance of MDBs to EPP in Humans and Mouse Models of EPP.
Porphyrias are inherited metabolic disorders that occur due to defects in the heme biosynthetic pathway, and are divided into acute and nonacute porphyrias.23 EPP is a nonacute porphyria caused by mutation in the Fch gene. Two mouse models of EPP have been described to date. The first model was generated by an Fch exon deletion which led to embryo-lethality of the homozygous mice, whereas heterozygous mice developed mild protoporphyria without liver disease.24 The second model (Fechm1Pas) was generated by inducing a point mutation in the Fch gene,5 which has been a valuable animal model for EPP because it mimics the most severe hepatic forms of the human disease.
The presence of MDBs has been described in the Fechm1Pas mouse strain and also in the griseofulvin-induced mouse model of EPP.7, 25 However, MDBs have not been reported in EPP patients. It should be noted that MDBs are visualized in clinical contexts using histological approaches (e.g., H&E staining), which lack sensitivity for detecting small and early MDBs.16 It is possible that some patients with EPP may form MDBs if sensitive methods are employed for their detection, and if indeed found then the genetic background of an individual may be a contributing factor. For example, Hispanics appear to be more predisposed to MDB formation,9, 26 and clear differences in MDB susceptibility are also noted in different mouse strains.12, 16 Other potential factors that may contribute to the apparent absence of MDBs in EPP patients (versus the animal models) may be related to differences in porphyrin metabolites across species. In this context, the contribution of porphyria per se to MDB formation is unclear. Certainly, porphyrin accumulation leads to oxidative stress,14 although other factors are likely important. The range of accumulation of protoporphyrin-IX (PP-IX) in DDC-fed male (≈550 nmol/g protein) and female (≈1,900 nmol/g protein) mice11 is similar to the porphyria in the fch/fch mice (1,000-5,000 nmol/g protein6, 27). By comparison, patients with EPP have PP-IX levels in the range of 3,000-8,000 nmol/g tissue,1 although the genetic modifiers that result in severe human liver disease in the 5%-10% of affected EPP patients are not known.
fch/fch Mice Manifest Spontaneous MDB Accumulation and Enhanced Susceptibility to Drug-Induced MDB Formation.
Our findings herein characterize the formation of spontaneous and DDC-induced MDBs in fch/fch mice and showed that MDBs formation in these mice is age-dependent but gender-independent. The lack of gender-dependency in this model, in contrast to the FVB/N drug-induced model, may be due to genetic mouse strain differences (Balb/c versus FVB/N) or may be related to the increased porphyria that might overcome genetic gender-related influences.
The enhanced susceptibility to drug-induced MDB formation, even in the heterozygous wt/fch mice, renders these mice very useful for studying MDBs given that MDBs can be induced in these mice within 3 weeks as compared with the standard 12 to 16-week DDC regimen. Hence, these mice are even more susceptible to MDB formation when compared with another genetic model, which is mice that overexpress K8 (which develop MDBs within 6 weeks of DDC exposure).28 The spontaneous formation of MDBs in fch/fch mice is also reminiscent of their formation in the K8-overexpressor mice and in the K18-null mice upon aging (both have a K8>K18 ratio).28, 29
Oxidative Stress, Proteasome Function, and Nrf2 in MDB Formation in fch/fch Mice.
Oxidative stress plays a major role in several protein aggregation disorders.30, 31 It is also involved in MDB formation in FVB/N and K8-overexpressing mice in response to DDC feeding.11 Our findings demonstrate a role for oxidative stress in MDB formation in fch/fch mice, and support previously reported alterations in oxidative stress genes in fch/fch mice and in EPP patients utilizing microarray analysis.3, 6 In addition, griseofulvin-induced EPP in mice results in ultrastructural changes in liver mitochondria, accumulation of large amounts of protoporphyrin, and low-energy coupling in mitochondria, which suggest blunting of mitochondrial function.32 Moreover, treatment of HepG2 cells with protoporphyrin in the absence of light increases intracellular H2O2 levels, which supports the role of protoporphyrin in causing oxidative stress.14 Our data are consistent with the presence of mitochondrial oxidative stress in the fch/fch livers, as demonstrated by decreased mitochondrial ATP and increased liver oxidized proteins (Fig. 6; Fig. S5), possibly due to accumulation of protoporphyrin deposits.
The time course of DDC feeding to FVB animals demonstrated that oxidative stress is upstream of proteasome inhibition and correlates with the increase Nrf2 expression, which becomes further magnified upon proteasome inhibition. Therefore, the observed 50% decrease in proteasome function in fch/fch mice under basal conditions are likely intimately linked to the increased oxidative milieu in the livers of these animals. Reduction in mitochondrial ATP content impairs proteasomal activity, which in turn leads to mitochondrial dysfunction in various disease states.20 Furthermore, proteasome inhibition in vivo accelerates MDB formation from 6 weeks to 2 weeks in K8-overexpressing mice,33 and a decrease in proteasome activity became normalized upon S-adenosylmethionine administration to DDC-fed mice in a short-term rechallenge model.13
Nrf2 is a redox-sensitive transcription factor that protects cells from oxidative stress.21 It is sequestered in the cytoplasm by Keap-1 (Kelch-like ECH associating protein 1) under nonstress conditions and is targeted for ubiquitination and proteasomal degradation.21 During stress, Nrf2 escapes Keap-1-mediated repression and translocates to the nucleus to up-regulate various genes such as glutathione-S-transferases and heme oxygenase-1.21 Notably, the phytochemical sulforaphane, which activates Nrf2, up-regulates epidermal keratins K16 and K1734 and Nrf2 up-regulates K16 in human keratinocytes.22 Furthermore, autophagy deficiency and up-regulation of p62 are known to stabilize Nrf2, resulting in activation of Nrf2 target genes.35 Our findings showed a decrease of K8, but not K18, mRNA and protein upon Nrf2 knockdown in HepG2 cells, and also showed Nrf2 nuclear accumulation in fch/fch mice under basal conditions and in fch/fch, wt/fch, and wt/wt livers after exposure to DDC (Fig. 7). Furthermore, sequence analysis of the 5′-proximal promoter region of K8 and K18 revealed the presence of Nrf2 consensus binding sequence (not shown). This was consistent with the study showing the Nrf2 binding sites on K16 promoter region.22 Based on the findings herein, and the established importance of K8 in MDB formation8 and of PPIX in oxidative stress,14 we propose a model (Fig. 8) whereby accumulation of PPIX leads to increased mitochondrial oxidative stress that inhibits the Keap-1-mediated Nrf2 degradation, thereby resulting in translocation of Nrf2 to nucleus and up-regulation of K8 and other target genes.