SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Mallory-Denk bodies (MDBs) are hepatocyte inclusions found in several liver diseases and consist primarily of keratins 8 and 18 (K8/K18) and ubiquitin that are cross-linked by transglutaminase-2. We hypothesized that genetic variables contribute to the extent of MDB formation, because not all patients with an MDB-associated liver disease develop inclusions. We tested this hypothesis using five strains of mice (FVB/N, C3H/He, Balb/cAnN, C57BL/6, 129X1/Sv) fed for three months (eight mice per strain) the established MDB-inducing agent 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC). MDB formation was compared using hematoxylin-and-eosin staining, or immunofluorescence staining with antibodies to K8/K18/ubiquitin, or biochemically by blotting with antibodies to transglutaminase-2/p62 proteins and to K8/K18/ubiquitin to detect keratin cross-linking. DDC feeding induced MDBs in all mouse strains, but there were dramatic strain differences that quantitatively varied 2.5-fold (P < 0.05). MDB formation correlated with hepatocyte ballooning, and most ballooned hepatocytes had MDBs. Immunofluorescence assessment was far more sensitive than hematoxylin-and-eosin staining in detecting small MDBs, which out-numbered (by ∼30-fold to 90-fold) but did not parallel their large counterparts. MDB scores partially reflected the biochemical presence of cross-linked keratin-ubiquitin species but not the changes in liver size or injury in response to DDC. The extent of steatosis correlated with the total (large+small) number of MDBs, and there was a limited correlation between large MDBs and acidophil bodies. Conclusion: Mouse MDB formation has important genetic contributions that do not correlate with the extent of DDC-induced liver injury. If extrapolated to humans, the genetic contributions help explain why some patients develop MDBs whereas others are less likely to do so. Detection and classification of MDBs using MDB-marker-selective staining may offer unique links to specific histological features of DDC-induced liver injury. (HEPATOLOGY 2008.)

Mallory-Denk bodies (MDBs) (previously called Mallory bodies) are characteristic hepatocellular inclusions observed in multiple liver diseases including alcoholic and nonalcoholic steatohepatitis (ASH and NASH, respectively); chronic cholestatic diseases; copper storage disorders, such as Wilson disease and Indian childhood cirrhosis; and hepatocellular carcinoma.1, 2 MDBs are defined by their morphological appearance and molecular composition; and consist primarily of the intermediate filament proteins keratins 8 and 18 (K8/K18), together with ubiquitin (Ub), p62, and chaperones.1, 2 MDBs represent the prototype epithelial inclusion bodies (though others occur such as those associated with α1-anti-trypsin deficiency) whereas a variety of nonepithelial aggregates with molecular composition and morphological features similar to MDBs are seen particularly in several neurodegenerative or muscle diseases.1, 3, 4

MDBs can be experimentally induced in livers of mice chronically fed griseofulvin or 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC).1, 5 These models have been instrumental in elucidating several essential aspects relating to the pathogenesis of MDBs. For example, an imbalance in the typical 1:1 protein ratio of K8-to-K18 to generate a K8>K18 cellular state, with subsequent cross-linking of excess K8 by tissue transglutaminase-2 (TG2) lead to MDB formation.6–11 MDBs also contain misfolded proteins1 that likely accumulate as a consequence of insufficient/defective protein degradation via the proteasome or autophagy, or accumulate because of inadequate protein repair mechanisms.12–14

The data on MDB prevalence varies dramatically depending on the detection method.15 In clinical contexts, MDBs are routinely detected by hematoxylin and eosin (HE) staining, but other alternative histological methods such as trichrome staining have been described.1, 15 Alternatively, immunofluorescence (Fl) or immunohistochemical staining that targets MDB constituents such as K8/K18, Ub, or p62 may offer better sensitivity for detection.2 In addition, electron microscopy is another sensitive but rarely-used method for MDB detection15 because it is not readily accessible and is labor-intensive, and is not practical for generalized assessment of MDB presence and extent. Furthermore, biochemical markers such as keratin cross-linking11 may be useful for MDB detection, but additional studies are needed to test their potential broad applicability.2

MDBs are characteristic but not constant morphologic features of several liver diseases, because they are not observed in all patients with a given MDB-associated liver disorder such as ASH or NASH or chronic cholestatic diseases. The reasons for MDB formation and accumulation in some but not all patients with a given liver disease are unknown.16, 17 MDBs might be a response to a complex pathological milieu, which is present in a subset of patients with a given liver disease. Also, genetic differences may predispose to MDB formation. To explore the latter possibility, the effect of the genetic background on MDB development and liver injury was tested by feeding DDC to five distinct and widely used inbred strains of mice.18, 19 In addition, we analyzed MDBs in the five mouse strains using standard histochemical evaluation, Fl staining with MDB-selective markers, and biochemical detection of keratin cross-linking and the up-regulation of MDB protein constituents.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Antibodies.

The following antibodies were used: rat anti-K8 monoclonal antibody (Ab) (Troma I) (Developmental Studies Hybridoma Bank); rabbit anti-mouse/human K8 and K18 Ab-8592, rabbit anti-mouse/human K18 Ab-4668;20 mouse anti-Ub and goat anti-p62 Abs (Santa Cruz Biotechnology); rabbit anti-TG2 Ab, and mouse anti-heat shock protein 60 (Hsp60) Ab (NeoMarkers).

Animal Experiments.

The five mouse strains FVB/N, C3H/He, and Balb/cAnN (Taconic Farms), and C57BL/6 and 129X1/Sv (Jackson Laboratory) were used. The strains will henceforth be referred to, respectively, as: FVB, C3H, BALB, C57BL and 129X. Their selection was based on their distinct genetic backgrounds (Fig. 1) and their inclusion in prior MDB and other genetic studies. All mice were males (3 months old at the beginning of DDC feeding). To induce MDBs, eight mice per strain were fed a powdered chow (Formulab Diet 5008, Dean's Animal Feeds) containing 0.1% DDC (Sigma-Aldrich) for 3 months. Age-matched males (n = 15, three mice per strain) were kept on a standard mouse diet (Teklad Global Diet 2019; Harlan Teklad) and used as controls. In another set of experiments, four mice per strain were also fed a DDC-containing diet to insure reproducibility of the first experiment. Mice were euthanized by CO2 inhalation, and blood was collected by intracardiac puncture for subsequent measurement of aspartate aminotransferase (AST), alanine aminotransferase (ALT), total bilirubin (TB), and alkaline phosphatase (ALP). Livers were removed, weighed, and divided into 1-2 mm slices and then apportioned for fixation in 10% formaldehyde then HE staining and histological analysis, snap-freezing in liquid nitrogen for biochemical analysis, or embedded in optimal-cutting-temperature medium then frozen for subsequent Fl staining. All mice received humane care and their use was approved by the Institutional Animal Care Committee at the Palo Alto Veterans Administration.

thumbnail image

Figure 1. The genealogy of inbred strains of mice. The schematic (adapted with modification from Beck et al.18 and Petkov et al.19) depicts the genetic relationship between the mouse strains used herein based on previously published genealogies of mouse inbred strains and the seven major groups they can be divided into.18, 19 Unlabeled branches represent other strain derivatives (not shown). The five strains we used are highlighted in rectangles (C3H, BALB, FVB, C57BL, and 129X).

Download figure to PowerPoint

Histological Analysis.

HE-stained liver sections were assessed by an experienced clinical hepatopathologist (E.M.B.) for the parameters shown in Table 2. Descriptive assessments were noted for each liver; in addition, lesions of interest were semiquantitatively scored as follows: MDBs and ductular reaction were both scored as 0–4 (0, none; 1, rare; 2, moderate; 3, frequent; 4, abundant). MDB scoring was done for zone 1 or zones 1+3 because MDBs are found preferentially in zone 1 in cholestatic and copper storage liver diseases. Hepatocyte ballooning was scored as 0–3 (0, none; 1, rare; 2, frequent; 3, abundant) and zonal location noted. Bile plugs and non-bile pigment deposition, and intraductal lithiasis were noted for location and semiquantified (0–3 and 0–4, respectively). Periductal fibrosis was scored as absent (0) or present (1). Steatosis was scored as <1% (1), 1%–5% (2), > 5% (3), and location noted; and the number of acidophil bodies per eight power fields (20×) was counted. Parenchymal nodularity was noted if easily detectable.

Table 2. Histological Analyses of the Livers from DDC-Fed Mice
 FVBC3HBALBC57BL129X
  • HE-stained livers were evaluated for several histological parameters associated with DDC ingestion (see Materials and Methods). All values are expressed as mean ± SD. Note that C3H mice exhibit the least amount of MDBs and hepatocyte ballooning, FVB mice have the most acidophil bodies and BALB mice display the most prominent ductular reaction (see Fig. 2 for representative examples of the histological findings). In addition, concretions, which are defined as intraductal stone formation, are most frequently observed in BALB mice. Liver steatosis is best appreciated in 129X mice. None of the histological findings were noted in the normal diet fed mice, except for an occasional ballooned hepatocyte or concretion (not shown).

  • *

    P < 0.05 compared with all the other strains.

  • P < 0.05 compared with FVB and C57BL.

  • P < 0.05 compared with 129X.

  • §

    P < 0.05 compared with C3H.

  • P < 0.05 compared with BALB and 129X1.

  • P < 0.05 compared with FVB, C3H and C57BL.

MDBs (zone 1)1.1 ± 0.60.4 ± 0.2*1.0 ± 0.51.5 ± 1.00.9 ± 0.2
MDBs (zone 1 + 3)3.6 ± 1.02.1 ± 0.93.3 ± 0.84.3 ± 1.63.1 ± 1.1
Hepatocyte ballooning1.5 ± 0.70.3 ± 0.31.6 ± 1.22.1 ± 0.80.3 ± 0
Acidophil bodies16.3 ± 10.87.4 ± 6.07.0 ± 5.612.6 ± 12.03.6 ± 2.0
Periductal fibrosis0.9 ± 0.4§0.1 ± 0.40.8 ± 0.50.7 ± 0.50.3 ± 0.5
Steatosis1.6 ± 0.71.8 ± 0.92.4 ± 0.81.4 ± 0.72.9 ± 0.4
Pigment deposition1.9 ± 0.83.0 ± 0.51.6 ± 0.93.1 ± 0.61.7 ± 1.0
Ductular reaction3.3 ± 0.73.3 ± 0.95.1 ± 1.13.3 ± 1.04.0 ± 0.6
Concretions (# with/total # mice)4/81/86/84/82/7
thumbnail image

Figure 2. HE staining of livers from DDC-fed mice. HE-stained livers from mice fed DDC for 3 months demonstrate several histological features associated with DDC ingestion, the extent of which varies significantly with the strain (Table 2). (A,C) FVB mice. Arrows highlight (A) MDBs and (C) acidophil bodies. (B) In comparison, the C3H mice exhibited only a few MDBs (arrow). (D) BALB mice had the most prominent ductular reaction, the periportal proliferation of cholangiocytes and cholangiolar structures within a fibrous stromal matrix. Inset: intraductal deposits and ductal hyperplasia were common in this strain. (E) Steatosis of hepatocytes was best appreciated in livers of 129X mice; a few droplets are highlighted by the arrows. (F) Marked brown pigment deposition was present in the C57BL livers (arrows). Scale bars are included with each image.

Download figure to PowerPoint

Fl Staining.

Liver tissues were sectioned (6 μm) and fixed with acetone for 10 minutes (−20°C). Nonspecific binding was blocked with a buffer containing 5% bovine serum albumin, followed by incubation with the primary and secondary Abs as described.20 Stained sections were viewed using confocal microscopy (Zeiss 510-Meta or Bio-Rad MRC 1024ES). To quantify the extent of MDB and ballooned hepatocyte formation, a 20× lens was used and the number of cells with K8/K18 and Ub-positive aggregates were counted (20 fields were analyzed per liver specimen).

Protein Isolation and Analysis.

Total liver lysates were prepared using a homogenization buffer (0.187 M Tris-HCl [pH 6.8], 3% sodium dodecyl sulfate [SDS], and 5 mM ethylene diamine tetraacetic acid), and samples were subsequently diluted to a desired protein concentration with 4× reducing Laemmli sample buffer. Equal amounts of proteins were separated by SDS-polyacrylamide gel electrophoresis and transferred to polyvinylidene fluoride membranes. Immunoblotting was performed as described,20 and the resulting antigen-Ab complexes were detected by enhanced chemiluminescence (PerkinElmer Life Sciences).

Statistical Analysis.

The values were expressed as the mean ± standard deviation. Data were analyzed using the Kruskal Wallis H-test. The post hoc test was performed by the Mann Whitney U-test with Bonferroni correction. A P value < 0.05 was considered statistically significant.

Results

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

DDC-Induced Liver Hypertrophy and Hepatic Injury Vary Markedly in Different Mouse Strains.

In order to address the potential effect of the genetic background on predisposition to MDB formation, we selected five strains of mice (Fig. 1) that are widely used and genetically distinct.18 The mice (eight per strain; Table 1) were fed a DDC-containing diet for 3 months in order to induce MDBs.1 The non–DDC fed animals of all strains (with identical age and sex) had livers of similar size (4.1%–5.2% range of liver-to-total body weight ratio) and no obvious abnormal liver histology was detected (not shown). Average AST, ALT, and ALP levels were in the normal range (<100 IU/L) and total bilirubin was <0.5 mg/dL under basal conditions for all mouse strains (not shown).

Table 1. Analysis of Liver Enzymes and Liver Weights in the DDC-Fed Mouse Strains
 FVBC3HBALBC57BL129X
  • Values are expressed as mean ± SD. Note that DDC-treated FVB mice exhibit the highest ALT, ALP, and AST levels and have the highest liver-to-total body weight ratio, whereas BALB mice had the highest bilirubin levels. In nontreated animals, average AST, ALT, and ALP levels were < 100 IU/L for all mouse strains and total bilirubin was < 0.5 mg/dL (not shown). TB, total bilirubin; ALP, alkaline phosphatase.

  • *

    One of the 129X mice died after 52 days of DDC treatment and was excluded from the analysis.

  • P < 0.05 compared with all the other strains. Similar findings (without any animal deaths) were obtained from an independent experiment using four mice per genotype (not shown).

Number of mice88887*
Liver/body weight ratio (%)17.3 ± 1.312.7 ± 1.813.1 ± 1.411.8 ± 1.213.9 ± 0.4
AST (IU/L)2730 ± 8961023 ± 284892 ± 216735 ± 260833 ± 141
ALT (IU/L)2943 ± 7541332 ± 3961967 ± 3121137 ± 3651589 ± 360
TB (mg/dL)1.0 ± 0.30.6 ± 0.32.5 ± 1.40.6 ± 0.21.1 ± 0.2
ALP (IU/L)1092 ± 255601 ± 81388 ± 90281 ± 135370 ± 207

DDC feeding was well tolerated in all strains; only a single 129X mouse died after 52 days of DDC exposure. As described previously,1 DDC induced hepatomegaly, pronounced liver injury, and a variety of characteristic histological changes (Fig. 2). The extent of liver injury and hepatomegaly varied markedly among the different strains. FVB mice exhibited the most pronounced hepatomegaly with an average liver-to-total body weight ratio > 17% and the highest serum AST, ALT, and ALP level compared with the other mouse strains (P < 0.05) (Table 1). In line with that, FVB mice also exhibited the most acidophil bodies (likely reflecting increased liver apoptosis21) (Table 2). On the other hand, BALB mice displayed the highest TB levels (P < 0.05 when compared with the other strains; Table 1) and the most pronounced intraductal stone (concretion) formation (Table 2, Fig. 2). Furthermore, BALB mice developed the most prominent ductular reaction, manifested by cholangiocyte proliferation and ductule formation within fibrous matrix and inflammation (Table 2, Fig. 2D). All DDC-fed strains showed marked non-bile pigment deposition, consistent with porphyrin, which was most obvious in C57BL mice. Steatosis was not prominent and was greatest in the 129X mice (Table 2, Fig. 2E).

Histological Evaluation Reveals Strain Differences in Predisposition to MDB Formation.

In order to characterize MDB formation, HE-stained livers were evaluated by zones (periportal and perivenular) for the presence and quantity of MDBs. DDC feeding induced abundant MDB formation in all mouse strains. Similar to some diseases in the human situation, MDBs were preferentially seen in the perivenular hepatocytes (zone 3) of the hepatic lobules in all animals (Table 2). However, the extent of MDB formation varied significantly among the mouse strains. C57BL mice had the highest and FVB mice the second highest MDB scores. On the other hand, C3H mice had the least MDBs both overall (P < 0.05 when compared with the FVB and C57BL strains) (Fig. 2) and in zone 1 of the hepatic lobule (P < 0.05 compared with all the other strains; Table 2). The extent of hepatocyte ballooning correlated reasonably well with MDB scores. For example, the C57BL strain had the highest MDB counts and most prominent hepatocyte ballooning, whereas C3H mice displayed the lowest MDB amounts together with minimal hepatocyte ballooning (Table 2).

Analysis of MDB Formation and Hepatocyte Ballooning Using Fl Staining.

In order to analyze MDB formation in a more quantitative fashion, Fl staining for K8/K18, Ub, and p62 (the major MDB constituents) was performed (Fig. 3). Total MDB counts (defined as any inclusion, be it small or large) were estimated by counting cells with all dots/aggregates that double-stained with Ub and K8/K18. Only a negligible number of dots did not double-stain with both keratins and Ub (not shown). Using this method of MDB quantification, 129X had the highest, C3H had the second highest, and C57BL had the lowest number of cells with MDBs (Table 3).

thumbnail image

Figure 3. Assessment of MDB formation using immunofluorescence staining. Mice of the indicated strains were fed DDC for 3 months followed by removal of their livers and double-labeling using antibodies to K8/K18 (red) and ubiquitin (green). MDBs are seen as yellow dots due to colocalization of the two epitopes. The (B) C57BL and (E) BALB livers had the lowest level of total MDBs when compared to the remaining strains (Table 3), but as noted in (E) (and Table 3) BALB livers had many prominent large MDBs. Arrow in (F) points to a large MDB seen in a ballooned hepatocyte. Scale bars: (A-E), 200 μm; (F), 20 μm.

Download figure to PowerPoint

Table 3. Quantification of MDBs and Hepatocyte Ballooning After Immunofluorescence Staining of Livers from DDC-Fed Mice
 FVBC3HBALBC57BL129X
  • Liver sections (from the same mice used in Table 1) were double-labeled with antibodies to K8/K18 and ubiquitin, then ballooned cells or cells with aggregates positive for both antibodies were counted. Small MDBs were defined as those with a dot-like appearance in the cytoplasm. MDBs were assigned as large when they were clearly located in cytoplasm and their size exceeded at least half the size of an adjacent well-defined nucleus. The ratio of small-to-large MDB-ratio was calculated as: (total − large = small)/(large) for each liver. Hepatocytes were considered as ballooned when their diameter exceeded 1.5 times the diameter of neighboring cells.

  • *

    P < 0.05 compared with C3H and 129X.

  • P < 0.05 compared with FVB.

  • P < 0.05 compared with FVB, BALB and C57BL.

  • §

    P < 0.05 compared with FVB and C57BL. All values are expressed as mean ± SD. Similar findings (without any animal deaths) were obtained from an independent experiment using the same five genotypes (four mice per genotype; not shown).

Total # cells with MDBs104 ± 32117 ± 2685 ± 3479 ± 18*148 ± 36
Cells with large MDBs3.7 ± 0.91.5 ± 0.72.8 ± 1.52.4 ± 0.92.6 ± 0.8
Small/large MDBs27 ± 6.190.6 ± 41.933.9 ± 14.436.8 ± 17.459.6 ± 20.2
Hepatocyte ballooning2.4 ± 0.90.9 ± 0.4§1.8 ± 1.22.8 ± 1.00.9 ± 0.2§

In addition, we carried out a careful subanalysis by counting the number of large MDBs which were defined as cytoplasmic aggregates being at least one-half the size of the nucleus. The FVB strain had the highest number of large MDBs whereas C3H had the lowest number (Table 3). A comparison of the small to large MDBs showed that C3H livers had the highest ratio whereas FVB livers had the lowest ratio (Table 3). The extent of MDBs, quantified by Fl or HE staining, did not correlate with mouse strain liver-to-total body weight ratios after DDC feeding (Tables 1–3).

We also analyzed the extent of hepatocyte ballooning using Fl staining. Ballooned cells were defined as those having at least 1.5 times the diameter of neighboring cells. Using such criteria, 70%-91% of ballooned hepatocytes had MDBs (example shown in Fig. 3F) (91% [BALB], 89% [C3H], 86% [129X], 79% [FVB], and 70% [C57BL]). There was no correlation between the presence of large MDBs and hepatocyte ballooning in the different strains. For example, C57BL livers had a prominent number of ballooned hepatocytes despite having only average counts of cells with large MDBs (Table 3).

Biochemical Markers of MDB Formation Partially Correlate with Histological and Fl-Based MDB Scores.

The important parameters and biochemical predictors of MDB accumulation include the formation of K8-containing and Ub-containing high molecular weight species and the up-regulation of keratin, TG2, and p62 expression.1, 2, 11 We evaluated these biochemical markers in the livers of the five mouse strains. In all cases, DDC feeding resulted in the induction of K8/K18, formation of K8-containing and Ub-containing high molecular weight cross-linked species, and induction of p62 (to variable levels) (Fig. 4).

thumbnail image

Figure 4. Biochemical effects of DDC feeding on mouse livers. Total liver homogenates were prepared from DDC-fed mice from the five strains or their respective normal diet controls. The tissue lysates were blotted with the indicated antibodies. Each lane represents a homogenate from a single liver. Note the formation of keratin/Ub high molecular weight cross-linked species that are too large to enter the stacking gel (arrows); and the induction of TG2, p62, and keratin expression in the livers of the DDC-fed mice. S, stacking gel; R, resolving gel. An Hsp60 blot is included as a loading control.

Download figure to PowerPoint

We also assessed how these markers compared with MDB scores that were obtained either by Fl analysis or by histochemical HE staining. C3H mice, which had the lowest number of large MDBs (as determined by HE [Table 2] and Fl staining [Table 3]), also had the lowest level of both p62 induction and cross-link formation (Fig. 5). There was a general but incomplete concordance between the presence of the MDB-related biochemical markers and the presence of large MDBs observed by Fl staining, whereas the correlation with other MDB scores (total MDBs by Fl or HE staining) was less evident (Fig. 5). For example, FVB livers displayed low amounts of Ub-containing K8–cross-linked species together with limited p62 induction, which contrasts with the high MDB scores obtained by Fl and histological staining. Finally, the extent of TG2 protein elevation in the different mouse strains after DDC feeding correlated to some extent with the increase in liver-to-total body weight ratio. The most elevated TG2 levels were found in FVB mice (Fig. 5); these mice also had the greatest increase in liver size (Table 1), which supports the previously suggested role of TG2 in organ hypertrophy.11, 22

thumbnail image

Figure 5. Correlation of MDB formation with p62 and TG2 induction and keratin cross-linking. Total liver homogenates from DDC-fed mice were blotted with antibodies to the indicated proteins. Each lane represents an independent liver homogenate from the indicated genotype such that results from two mice per genotype are displayed. For each lane, the MDB scores (determined by HE and immunofluorescence staining) for the biochemically-analyzed liver lysates are also shown. S, stacking gel; R, resolving gel. Hsp60 is used as a loading control.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

The Genetic Background Affects MDB Development and DDC-Associated Liver Injury.

Our present study shows that different mouse strains exhibit differences in DDC-induced liver injury and MDB development, and thereby demonstrates the impact of genetic background on DDC-induced liver disease in general and MDB formation in particular. These findings are supported by the reported formation of mouse MDBs in hepatocellular tumors after dieldrin (a chlorinated hydrocarbon) treatment of two mouse strains for 85 weeks.23 In the dieldrin study, MDBs (assessed by HE staining) were reported in only 53% of C3H/He tumors but in 95% of C57BL/6J tumors (P < 0.01).23 Hence, the genetic differences of MDB formation between C3H and C57BL involve DDC-induced liver injury and dieldrin-induced tumors. Our findings also support available biochemical data1, 2 that MDB formation represents a multifactorial response (for example, keratin and p62 induction, keratin cross-linking). The findings herein provide a genetic basis as to why MDBs occur in some but not other patients with an MDB-associated liver disease, particularly ASH or NASH, and in cholestatic liver diseases. Identification of mouse strains with a low predisposition versus those with a high predisposition to the types of injury and responses that result in MDB formation provides the necessary model systems to identify additional molecular events that are involved in MDB formation as well as in DDC-mediated liver injury.

Several modifier genes are already known to be important in MDB formation and several can be implicated, although detailed expression profiling and proteomic studies supplemented by validation analysis will be required. Among the known gene modifiers are K8, p62, and TG2.1, 2 Gene modifiers that can be implicated include those involved in autophagy or in proteasome function given that autophagy activation prevents inclusion formation whereas proteasome inhibition promotes MDB formation in cell culture24 and in MDB mouse models.12 The relevance of the proteasome may extend to heat shock proteins and their major role in protein repair mechanisms,25 and the mouse strain differences in their expressions26 may affect aggregate development. In addition, the cytochrome P450 system exhibits strain-dependent variations27, 28 and CYP2E1 levels have been shown to affect DDC-mediated MDB formation.29

Aside from the regulation of MDB formation via specific gene products, another layer of regulation pertains to posttranslational modifications, particularly phosphorylation. For example, keratins become hyperphosphorylated at multiple sites within MDBs30 and prevention of human K8 Ser73 phosphorylation by a phospho-mutant K8 S73A expression in transgenic mice inhibits MDB formation in vivo (unpublished data). Therefore, differences in keratin phosphorylation may affect MDB formation, although this is likely to be site-specific because K18 phosphorylation at two sites (Ser33 and Ser52) does not appear to be involved in MDB formation.7 Strain differences in keratin phosphorylation during MDB formation have been reported previously, but not the extent of MDB formation. For example, phosphorylation of K8 Ser79, the murine analog of human K8 Ser73, was observed in griseofulvin-induced MDBs in C3H mice, but not in FVB mice.31

Partial Concordance Between Different Methods of MDB Detection: Does MDB Size Matter?

We used two different approaches to quantify MDBs: the standard HE staining that is used by pathologists worldwide, and the more labor-intensive Fl staining that is far less commonly used but lends itself to more quantitative and qualitative evaluations. Initially, we were surprised to find significant differences (summarized in Fig. 6) when we compared the extent of MDB formation as measured by HE with Fl staining. For example, C3H and 129X had the fewest numbers of MDBs by HE staining, yet had the highest total MDBs by Fl staining (Fig. 6). A closer inspection showed that most of the MDBs that were measured by Fl were small, and indeed the ratio of small to large MDBs in the five strains we studied ranged between 27-fold (FVB) to 91-fold (C3H) (Table 3). Hence, when only large MDBs were counted after Fl staining, the concordance with HE assessment was much better (for example, the C3H mouse has the least MDBs using both methods; Fig. 6). The differences that remain are likely due to the subjective nature of our assignment of “small” versus “large” MDBs, as well as the subjective nature of MDB assignment by HE evaluation. This finding emphasizes the fact that different methods likely relate to the actual definition of the microscopically observed lesion known as MDB. Furthermore, a small intracytoplasmic inclusion may represent the “tip” of an MDB in a particular plane of section that by HE stain one is unwilling to diagnose, but by a specific antigenic marker, one can easily recognize. Other hepatocellular intracytoplasmic inclusions visualizable by HE may encompass not only MDBs but also intracytoplasmic hyaline bodies and megamitochondria among others.1

thumbnail image

Figure 6. Overview of mouse genetic variation effects on the responses to DDC feeding. Scores of hepatocyte ballooning and MDB formation (determined either by HE or immunofluorescence staining), steatosis, or acidophil bodies are stratified from “most” to “least” for the five genotypes that were used. The solid bars that link 129X and C3H (left two columns) imply that the C3H and 129X strains had a similar extent of hepatocyte ballooning.

Download figure to PowerPoint

The clinical significance of small MDBs (also termed pre-MDBs9) is unclear except for the likelihood that they serve as precursors to bona fide mature MDBs.32, 33 The observation that the ratio of small-to-large MDBs does not reflect the extent of large MDB presence suggests that the presence of small “pre-MDBs” may be necessary but not sufficient by itself to generate the large MDBs and this may in part be related to genetic differences in the ability to mount responses that promote MDB turnover such as autophagy.12 However, there is a reasonably good correlation between the extent of steatosis and the total number of MDBs estimated by Fl staining (Fig. 6). Although this correlation needs to be verified in subsequent studies (for example, mouse MDB models fed a high-fat diet), it fits well with the clinical and histopathological (via HE staining) presence of the more classical MDBs in livers from patients with NASH and ASH.

The prognostic significance of MDBs is currently unknown. Likewise, the possibility remains that MDBs are epiphenomena of specific forms of hepatic injuries. Most studies that have assessed this question were performed using HE-based scores. Such studies may not reflect the significance of MDB formation, because aggregate precursors (that is, small MDBs), rather than mature inclusion bodies are likely to be pathologically relevant in neural diseases.34, 35 Therefore, the development of more sensitive methods to analyze the levels of MDBs and their precursors and aggregate intermediates is needed in order to better appreciate the importance of MDBs in liver disease prognosis.

The Method of MDB Characterization Provides Unique Correlations with Specific Histological Features.

Previous studies implicated a relationship between hepatocyte ballooning (that is, hypertrophy or enlargement which may be accompanied by degenerative changes) and MDB development.1, 7, 11, 36 In the study herein, hepatocyte ballooning also had a strong correlation with the presence of large MDBs (measured by HE or Fl staining) but not with the total number of MDBs (measured by Fl) nor with steatosis (Fig. 6). In addition, ballooned cells frequently contained MDBs and vice versa, thereby suggesting a causal relationship. An alternative explanation is that MDBs are more readily observed in enlarged hepatocytes with relatively cleared out cytoplasm, as seen in ballooned hepatocytes. In fact, in human alcoholic hepatitis, many MDB-containing hepatocytes are not ballooned, but are shrunken and deeply eosinophilic. Likewise, in chronic cholestatic liver diseases, the hepatocytes that contain MDBs are not necessarily enlarged or ballooned. Many hepatocytes with MDBs typically also have disruption of their keratin filament network,37 but not all cell ballooning (which occurs in the context of most liver diseases) is associated with keratin filament disruption in human liver disease. For example, K8/K18 immune staining is either absent or markedly reduced in ballooned hepatocytes of patients with NASH, ASH, and chronic cholestatic conditions but not in cases of acute hepatitis, chronic hepatitis B, or autoimmune hepatitis.36

Biochemical detection of MDB surrogates such as high molecular weight keratin-Ub cross-links may be another modality to identify MDBs, but it is likely to be of value primarily in livers that have a significant “large MDB” content. For example, keratin-Ub cross-links were barely measurable in C3H livers, which had the lowest number of large MDBs but the second highest total MDBs (Fig. 6). However, our current ability to detect MDB-containing cross-links is somewhat crude because it depends on retained reactivity of the high molecular weight species to keratin/Ub antibodies and their ability to enter the polyacrylamide gels. More refined reagents such as those that detect specific cross-links are likely to improve the detection sensitivity and specificity.

The method of MDB detection (HE or Fl staining) and the categorization of MDB as large (reflected by standard HE staining or by Fl) or small pre-MDBs (which make up the majority of total MDBs estimated after Fl staining) affords other potential associations with histological features such as acidophil bodies (which may reflect apoptosis in the liver21). For example, there is partial correlation of large MDBs with the presence of acidophil bodies (for the FVB and C57BL strains [Fig. 6, Table 2]) and inverse correlation of total MDBs with acidophil bodies (for the 129X strain). This may ultimately prove biologically relevant because keratins (the bulk constituents of MDBs) are essential in protecting hepatocytes from apoptosis,2, 4 but it is presently unknown whether MDBs shunt hepatocytes toward or away from apoptosis. As our understanding of the factors that are important in MDBs formation improves, the ability to assess MDBs in a more specific and sensitive fashion offers more selective molecular links to commonly encountered histological features of MDB-associated liver injury.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

We are grateful to Evelyn Resurreccion for assistance with immune staining and Kris Morrow for help with figure preparation.

References

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References