Department of Internal Medicine I, Center for Internal Medicine, University Medical Center Ulm, Ulm, Germany
Department of Internal Medicine III and IZKF, University Hospital Aachen, Aachen, Germany
Address reprint requests to: Pavel Strnad, University Medical Center Aachen, Department of Internal Medicine III and IZKF, University Hospital Aachen, Pauwelsstr. 30, D-52074, Aachen, Germany. E-mail: firstname.lastname@example.org; fax: + 49 241 8082455.
Potential conflict of interest: Nothing to report.
Supported by the German Research Foundation grants STR 1095/2-1, STR 1095/4-1, SFB TRR57 and the Interdisciplinary Center for Clinical Research (IZKF) in Aachen.
Mallory-Denk bodies (MDBs) are protein aggregates consisting of ubiquitinated keratins 8/18 (K8/K18). MDBs are characteristic of alcoholic and nonalcoholic steatohepatitis (NASH) and discriminate between the relatively benign simple steatosis and the more aggressive NASH. Given the emerging evidence for a genetic predisposition to MDB formation and NASH development in general, we studied whether high-fat (HF) diet triggers MDB formation and liver injury in susceptible animals. Mice were fed a high-fat (HF) or low-fat (LF) diet plus a cofactor for MDB development, 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC). Additionally, we fed nontransgenic and K8 overexpressing mice (K8tg) with the HF diet. The presence of MDB and extent of liver injury was evaluated using biochemical markers, histological staining, and immunofluorescence microscopy. In DDC-fed animals, an HF diet resulted in greater liver injury and up-regulation of inflammation-related genes. As a potential mechanism, K8/K18 accumulation and increased ecto-5′-nucleotidase (CD73) levels were noted. In the genetically susceptible K8tg mice, HF diet triggered hepatocellular injury, ballooning, apoptosis, inflammation, and MDB development by way of 1) decreased expression of the major stress-inducible chaperone Hsp72 with appearance of misfolded keratins; 2) elevated levels of the transglutaminase 2 (TG2); 3) increased K8 phosphorylation at S74 with subsequent TG2-mediated crosslinking of phosphorylated K8; and 4) higher production of the MDB-modifier gene CD73. Conclusion: Our data demonstrate that HF diet triggers aggregate formation and development of liver injury in susceptible individuals through misfolding and crosslinking of excess K8. (Hepatology 2014;60:169–178)
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.
luminescent conjugated oligothiophene having heptameric thiophene backbone
human K8 overexpressing mice
nonalcoholic fatty liver disease
NAFLD activity score
quantitative real time polymerase chain reaction
Due to the widespread epidemic of obesity, nonalcoholic fatty liver disease (NAFLD) represents the most common subtype of chronic liver disorders in developed countries. It ranges from a relatively benign steatosis to the more aggressive nonalcoholic steatohepatitis (NASH), which may lead to development of liver cirrhosis and hepatocellular carcinoma and is associated with increased liver-related mortality. Despite its great clinical relevance, the factors driving the progression from simple steatosis to NASH remain poorly understood. NASH can be easily distinguished from simple steatosis by the additional presence of inflammation, hepatocellular ballooning, and Mallory-Denk bodies (MDBs).[1, 3]
MDBs are large eosinophilic hepatocellular cytoplasmic protein aggregates which are characteristic hallmarks of alcoholic steatohepatitis and NASH, but can also be found in other human diseases such as Wilson's disease, chronic cholestasis, Indian childhood cirrhosis/copper toxicosis, or hepatocellular carcinoma.[3, 4] In patients with chronic hepatitis C, MDBs are associated with insulin resistance and steatosis and likely indicate a presence of a metabolic comorbidity. In NASH, MDBs together with ballooning serve as a marker of a more active disease. A close link between MDBs and steatohepatitic injury is also highlighted by the fact that they often associate with fat droplet accumulation.
MDBs consist primarily of ubiquitinated keratins 8/18 (K8/K18) and sequestosome 1/p62. MDBs can be induced in mice by feeding hepatotoxic drugs such as griseofulvin or 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC) for 2-5 months. They also form spontaneously in aging animals lacking K18 or transgenic mice overexpressing K8, which meshes well with the fact that excess K8 is crucial for their pathogenesis.[3, 4] To further highlight this fact, K8/K18 dysregulation is found in DDC-fed animals and patients with MDB-forming diseases,[3, 4] (Guldiken et al., unpublished data). In mice and men, MDB formation is paralleled by a profound redistribution or even disappearance of the fibrillar keratin network with appearance of empty cells, i.e., hepatocytes largely devoid of keratin fluorescence.[3, 4]
Over the years, the DDC model became the gold standard in MDB research and greatly enhanced our understanding of MDB pathogenesis. It includes misfolding of excess K8 with formation of cross β-sheets and K8 phosphorylation at S74, which facilitates its crosslinking through transglutaminase 2 (TG2). As a potential mechanism driving the conformational change of K8, murine MDB formation is associated with impaired chaperone function and decreased levels of Hsp72, the prototypical stress-inducible heat shock protein.[8, 9] Studies in different mouse strains also manifested a significant genetic predisposition to MDB formation and several modifier genes have been identified.[3, 10, 11]
MDBs represent a characteristic feature distinguishing simple steatosis from NASH; however, this feature is present only in a subset of patients. Given the known genetic component in both NASH development and MDB formation,[3, 10-12] we tested the hypothesis that a diet rich in saturated fatty acid, i.e., an established murine model of diet-induced obesity and a major alimentary factor responsible for NASH development, triggers MDB appearance and liver injury in susceptible individuals. Indeed, a high-fat (HF) diet promoted/induced MDB formation and development of hepatic injury/inflammation in both the DDC-fed animals as well as the K8 overexpressing mice. As underlying mechanisms, we observed an induction of the MDB-modifier gene CD73, increased K8 misfolding, and TG2-mediated K8 crosslinking.
Materials and Methods
To study the impact of HF diet on MDB formation, 3-month-old male C57Bl/6 mice were pretreated with low-fat (LF; i.e., containing 10% kcal% fat; D12450B, Research Diets, New Brunswick, NJ) or HF chow (i.e., containing 60% kcal% fat, D12492, Research Diets) for 4 weeks. After that, mice were fed with the respective 0.1% DDC (#137030, Sigma-Aldrich Chemie, Munich, Germany)-supplemented diet for 10 weeks. Control animals were exposed to the corresponding DDC-free chow for 14 weeks (Supporting Fig. 1). Alternatively, 2-month-old human K8 overexpressing males on an FVB/N background (K8tg) and their nontransgenic littermates were fed the LF/HF diet for 10 months. The animals were kept under standardized conditions (12 hours day/night cycle, 21-24°C, humidity ∼50%) and had free access to water and food. Mice were sacrificed by CO2 inhalation and blood samples were obtained by cardiac puncture. Serum parameters were measured in the Clinical Chemistry Department of University Hospital Ulm. Livers were removed, weighed, cut into pieces, and 1) fixed in 10% formaldehyde for histological/immunohistochemical staining; 2) snap-frozen in liquid nitrogen or precooled methylbutane (Sigma-Aldrich Chemie) for biochemical analyses and immunofluorescence staining, respectively; or 3) submerged into RNAlater Stabilization Reagent (Ambion, Life Technologies, Darmstadt, Germany) for RNA analysis. All mice received humane care, and the experimental protocols were approved by the State of Baden-Wurttemberg (Germany) and the University of Ulm (Germany) Animal Care Committee.
Total RNA was isolated using the RNeasy mini kit (Qiagen, Valencia, CA) and converted into complementary DNA (cDNA) using Superscript II reverse transcriptase (Invitrogen, Carlsbad, CA) and the oligo-dT method. qPCR was performed with a 7500 fast Real Time PCR Sequence Detection System (Applied Biosystems, Foster City, CA). Samples were analyzed in duplicate using specific primers (Supporting Table 1). Ribosomal protein L7 (Rpl7) was employed as an internal control and cDNA levels were normalized so that its expression was approximately equal in all tested samples. The transcript levels relative to Rpl7 were determined and reported as means ± SD.
Table 1. Liver, Serum Parameters, and Body Weights in DDC-Induced MDB Model
C57Bl/6 (− DDC)
C57Bl/6 (+ DDC)
LF, low-fat diet; HF, high-fat diet; DDC, 3,5-diethoxycarbonyl-1,4-dihydrocollidine; Values are expressed as mean ± SEM. P values (Student t test)
Total liver lysates were prepared by homogenization in 3% sodium dodecyl sulfate (SDS)-containing sample buffer. Alternatively, taking advantage of their poor solubility in nonionic detergents, high salt extraction was performed to enrich for keratins. For details, see the Supporting Methods section.
For immunofluorescence staining and subsequent MDB quantification, both acetone- and formaldehyde-fixed sections were used. For details, see the Supporting Methods section.
The paraffin-embedded samples were cut into 3-μm thick sections and stained with hematoxylin and eosin (H&E) for histological examination. Images were acquired on a Leica light microscope (Leica, Solms, Germany) equipped with a digital camera and Leica Application Suite software V4.1 (Leica Microsystems, Heerbrugg, Switzerland). Stained sections were assessed independently by two experienced hepatopathologists (H.D. and J.H.) for hepatocyte ballooning, steatosis, MDB formation, and inflammation using a semiquantitative scoring system (0, none; 1, rare; 2, frequent; and 3, abundant). The presence of steatosis, lobular inflammation, and ballooning was analyzed using an established scoring system and the values for all parameters were added up to obtain the NAFLD activity score (NAS).
Protoporphyrin IX (PPIX) Assay
PPIX assay was performed as described. Briefly, liver tissue homogenates were extracted with 20 volumes of 0.9N perchloric acid / ethanol (1:1 mixture), PPIX levels were measured based on their autofluorescence (400 nm excitation and 605 nm emission), and normalized to liver tissue weight. PPIX standards were used to generate a calibration curve.
Data Analysis and Statistical Methods
Morphometric quantification of immunofluorescence stained sections was carried out with the ImageJ program (National Institutes of Health, Bethesda, MD). To quantify the signal intensities, immunoblots were scanned and pixel intensities were quantified with Adobe Photoshop CS5 v. 12.0 (Adobe System). Based on normality test, data were analyzed with an unpaired two-tailed Student t test and Mann-Whitney test, respectively. All reported P values are two-tailed and values less than 0.05 were considered statistically significant.
Combination of DDC and HF Diet Promotes MDB Formation and Development of Cholestatic Liver Injury
In the first series of experiments, C57Bl/6 mice were exposed to LF/HF diet alone or in combination with DDC. The treatment was well tolerated, and no lethality was observed. HF diet alone did not lead to an obvious liver phenotype, while DDC-feeding alone resulted in marked liver enlargement with elevated alanine aminotransferase (ALT) and alkaline phosphatase (AP) levels but no bilirubin increase (Table 1). When compared to DDC-fed animals on LF diet, mice on HF+DDC chow displayed significantly increased bilirubin and AP levels, while hepatomegaly was slightly, but significantly, attenuated (Table 1; Supporting Fig. 2A). As shown previously, chronic DDC supplementation caused a massive ductular reaction, oval cell expansion and hepatic progenitor cell-dependent liver regeneration, which were more prominent in the HF+DDC group as evidenced by K19 immunofluorescence staining, higher K19 levels, as well as elevated mRNA of the oval/progenitor cell marker Ly6a (Fig. 1).
Since DDC is a porphyrinogenic drug, we evaluated whether the difference in cholestatic injury might be due to alterations in porphyrin metabolism. As expected, DDC-treated mice displayed marked PPIX accumulation. However, its magnitude did not differ between the LF+DDC and HF+DDC subgroups (Supporting Fig. 3). On the other hand, and in agreement with the stronger cholestatic injury, DDC-fed animals on HF diet displayed a prominent induction of various chemokines, chemokine receptors, and other inflammation-related genes (Supporting Fig. 4).
Histological evaluation demonstrated a higher extent of liver injury in the DDC-fed mice on HF versus LF diet. In particular, these animals exhibited more prominent steatosis, more MDBs, and higher NAS scores (Table 1, Fig. 2A). No MDBs/ballooning and only minimal steatosis was observed in the DDC-free treatment regimens (Table 1, Fig. 2A). The histological data were confirmed by immunofluorescence staining, which revealed more large MDBs in the HF+DDC animals, while LF+DDC mice had more empty cells (Fig. 2B-D; Supporting Fig. 2B). As a potential explanation for the more pronounced MDB formation, CD73 mRNA and K8/K18 levels were significantly increased in HF+DDC versus LF+DDC mice (Fig. 3) and the latter observation meshes well with the lower amount of empty cells in the HF mice.
Administration of HF Diet Promotes Liver Injury and MDB Formation in K8tg Mice
To test whether HF diet is sufficient to trigger MDB formation and development of liver injury in a genetically susceptible setting, we used transgenic animals overexpressing K8 (K8tg). Of note, 1-year-old K8tg mice on LF diet did not show an apparent liver phenotype and displayed normal liver enzyme levels (Table 2, Fig. 4A). However, HF feeding was not well tolerated in K8tg animals. They experienced significant weight loss, liver enlargement, and elevated ALT and AP levels, whereas only moderately enlarged livers and slightly elevated ALT were seen in nontransgenic mice on HF diet (Table 2).
Table 2. Liver, Serum Parameters, and Body Weights in K8 Overexpression-Induced MDB Model
K8tg (FVB/N Background)
LF, low-fat diet; HF, high-fat diet; TG, triglyceride. Values are expressed as mean ± SEM. P values (Student t test)
Histological evaluation of HF-treated subgroups demonstrated an increase in hepatocellular apoptosis and more inflammation but less steatosis in K8tg animals (Table 2, Fig. 4A). No obvious histological alterations were seen in K8tg animals on LF diet, while elevated collagen mRNA was found in HF-fed K8tg mice, although no or only minimal liver fibrosis was observed (Table 2 and not shown). In agreement with the higher extent of liver injury, K8tg mice subjected to HF diet demonstrated high levels of various chemokines, chemokine receptors, and other inflammation-related genes (Supporting Fig. 5). No differences between nontransgenic mice and K8tg animals on LF diet were noted (Supporting Fig. 5).
Unlike animals overexpressing mouse K8, no MDBs and no hepatocyte ballooning was observed in aging K8tg mice kept on LF diet (Table 2, Fig. 4), which is likely due to the lower overall K8 expression levels found in this line (not shown). HF diet was not sufficient to induce MDB formation/hepatocellular ballooning in nontransgenic animals but resulted in the development of macrovesicular steatosis and mild inflammation (Table 2, Fig. 4). In contrast, the administration of HF diet led to a prominent ballooning and MDB formation in K8tg mice, which was confirmed by morphometrical analysis of histological sections as well as by immunofluorescence staining (Fig. 4, Table 2). MDB formation in K8tg mice subjected to HF diet was associated with accumulation of p62, crosslinked and ubiquitinated proteins (particularly in the insoluble fraction which is enriched in MDBs), as well as with elevated K8/K18 levels (Fig. 5). Of note, HF feeding up-regulated K8/K18 levels only in K8tg mice but not in nontransgenic animals. Neither increased K8 crosslinking nor accumulation of ubiquitinated proteins was seen in the remaining experimental groups, whereas a minor increase in insoluble p62 was detected in nontransgenic mice fed with HF diet (Fig. 5). Finally, K8tg mice on HF but not on LF chow displayed prominent labeling with the β-sheet marker h-HTAA, revealing the presence of misfolded proteins (Supporting Fig. 6). Moreover, h-HTAA signal largely overlapped with the other MDB markers.
To determine the mechanisms responsible for increased MDB formation seen in K8tg mice exposed to HF diet, we analyzed several additional mechanisms implicated in this process. HF-fed K8tg mice displayed lower Hsp72 levels, which likely render them more susceptible to β-sheet formation (Fig. 6A). CD73 mRNA, which was previously shown to contribute to MDB development, was significantly increased in both HF-fed subgroups but was the highest in K8tg animals (Fig. 6B). Finally, TG2 was exclusively elevated in K8tgs fed the HF diet, and K8 phosphorylation at S74, which is known to facilitate TG2-mediated K8 crosslinking, was also increased preferentially in this subgroup. As a potential responsible enzyme for this observation, p38 kinase, which phosphorylates K8 on this residue, was also activated.
In summary, using two independent mouse models, our study demonstrates that an HF diet induces/promotes MDB formation in susceptible individuals but is not sufficient to precipitate aggregate formation alone. The factors involved in this process involve increased levels of the MDB-modifier gene CD73, keratin misfolding, and TG2-mediated K8 crosslinking. For the latter, HF diet-triggered, p38-mediated K8 phosphorylation at S74 is of importance, while lower Hsp72 levels probably contribute to the higher extent of misfolding with β-sheet formation.
Using two established mouse models, our study shows that a diet rich in saturated fatty acids promotes/induces MDB formation in susceptible animals (Fig. 6D). These findings are in agreement with the fact that MDBs are established NASH markers and that saturated fatty acids promote NASH development.[1, 14] However, HF diet alone is not sufficient for MDB formation and needs the presence of additional hits (Fig. 6D). This is not surprising, since previous studies have demonstrated that MDB formation represents a complex process which requires multiple factors for its occurrence. While MDB formation was associated with more severe injury and increased expression of inflammation-associated genes in both models, we observed no correlation with the extent of steatosis, suggesting that steatosis and MDB formation are independent. This is supported by human data that detected no correlation between the steatosis grade and MDB accumulation in NAFLD subjects.
In the DDC model, MDB formation was associated with greater cholestatic injury which is in agreement with the fact that MDBs are seen in patients with chronic cholestatic disorders such as primary biliary cirrhosis,[3, 7] but in contrast with previous studies demonstrating less MDBs but stronger cholestatic damage in mice lacking TG2. These data indicate that cholestatic injury is not as intimately linked to MDB formation as previously expected.
The factors responsible for increased MDB formation seen in the HF diet-fed animals include increased CD73 levels, accumulation of K8/K18 and p62, K8 crosslinking through TG2, as well as decreased Hsp72 levels which contribute to the observed protein misfolding. HF feeding alone did not alter K8/K18 levels but induced K8/K18 accumulation both in K8tg and DDC-fed animals. This is not surprising since keratins are established stress-inducible genes elevated in multiple liver disorders. While Nrf2 was previously implicated in K8/K18 induction, no Nrf2 activation was seen in our animals (data not shown). On the other hand, given the increased inflammation seen in our mice on HF diet, interleukin (IL)-6 signaling, which up-regulates K8/K18 in hepatic and intestinal cells (Guldiken N, unpublished data) might be of importance. In addition, posttranscriptional factors are likely to be involved. In this respect, posttranslational mechanisms are responsible for K8/K18 accumulation and associate with MDB occurrence in patients with primary biliary cirrhosis. Furthermore, autophagy contributes to MDB and K8/K18 degradation and it has been shown to be inhibited by prolonged exposure to HF diet.[29, 30]
CD73 was significantly up-regulated in both animal models, which is not surprising given that CD73 was associated with development of alcohol/HF-diet associated steatosis. It is also in agreement with the findings of a recent study which demonstrated that CD73 promotes MDB development. However, further studies are needed to clarify the precise role of CD73 in HF-driven MDB production. Crosslinking by way of TG2 is another essential event in MDB formation. To that end, TG2 accumulates in NASH subjects and DDC-fed animals and promotes MDB formation through K8 transamidation.[7, 32] Therefore, elevated TG2 levels found in K8tg mice on HF diet are likely responsible for the observed increase in K8 crosslinking and MDB formation seen in these animals. The detected elevated K8 phosphorylation at S74 further facilitates this process since it regulates K8 crosslinking at glutamine 62, the major K8 transamidation site. Protein kinase p38 mediates K8 S74 phosphorylation and it is induced not only in K8tg mice on HF diet (this study) but also in NASH as a downstream target of leptin. Of note, p38 was previously reported to colocalize with MDBs and to be up-regulated after DDC feeding. In agreement, a p38 inhibitor was shown to attenuate MDB formation in vitro and this finding meshes well with the fact that p38 is able to induce a profound redistribution of the K8/K18 network.
Decreased Hsp72 levels observed in HF diet-fed K8tg animals likely contribute to MDB appearance, given that Hsp72 is an established K8/K18-associated protein and an MDB constituent.[4, 36] Hsp72 is the major stress-inducible chaperone and thereby likely prevents the protein misfolding seen within MDBs.[8, 22] Of note, lower Hsp72 expression was detected in DDC-fed mice while Hsp72 dysfunction was noted in animals administered alcohol and HF diet.[9, 37] While Hsp72 overexpression protected against polyglutamine and α-synuclein aggregation,[38, 39] further studies are needed to precisely delineate the role of Hsp72 in MDB formation.
Additionally, p62, which accumulates in K8tg mice on HF diet, is a ubiquitin-binding protein and a major constituent of multiple human cytoplasmic aggregates. Elevated p62 levels are seen in DDC-administered mice, and p62 overexpression accelerated MDB formation in primary hepatocytes. As a potential underlying mechanism, p62 is known to regulate the compaction of its cargo and may promote aggregate formation by way of self-polymerization.[3, 41]
In summary, our study implicates HF diet as a crucial factor contributing to MDB formation in predisposed individuals. The underlying mechanisms include K8/K18 accumulation and K8 hyperphosphorylation with subsequent TG2-mediated K8 crosslinking. Decreased chaperone levels together with elevated p62 and CD73 are other relevant factors involved in this process.
We thank Kristina Diepold, Elke Preiβ, and Claudia Laengle for their expert technical assistance and Peggy Schwarz and Mariia Lunova for their overall experimental assistance.
Study concept and design: OK, PS; Acquisition of data: OK, NG, YC, VU, AEH; Analysis and interpretation of data: OK, NG, YC, HD, JH, PS; Drafting of the article OK, CT, PS; Critical revision of the article for important intellectual content: all authors; Statistical analysis: OK; Obtained funding and study supervision: PS; Technical or material support: JH.