“Toxic Memory” via Chaperone Modification Is a Potential Mechanism for Rapid Mallory-Denk Body Reinduction

Abstract

The cytoplasmic hepatocyte inclusions, Mallory-Denk bodies (MDBs), are characteristic of several liver disorders, including alcoholic and nonalcoholic steatohepatitis. In mice, MDBs can be induced by long-term feeding with 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC) for 3 to 4 months or rapidly reformed in DDC-induced then recovered mice by DDC refeeding or exposure to a wide range of toxins for only 5 to 7 days. The molecular basis for such a rapid reinduction of MDBs is unknown. We hypothesized that protein changes retained after DDC priming contribute to the rapid MDB reappearance and associate with MDB formation in general terms. Two-dimensional differential-in-gel-electrophoresis coupled with mass spectrometry were used to characterize protein changes in livers from the various treatment groups. The alterations were assessed by real-time reverse-transcription polymerase chain reaction and confirmed by immunoblotting. DDC treatment led to pronounced charged isoform changes in several chaperone families, including Hsp25, 60, 70, GRP58, GRP75, and GRP78, which lasted at least for 1 month after discontinuation of DDC feeding, whereas changes in other proteins normalized during recovery. DDC feeding also resulted in altered expression of Hsp72, GRP75, and Hsp25 and in functional impairment of Hsp60 and Hsp70 as determined using a protein complex formation and release assay. The priming toward rapid MDB reinduction lasts for at least 3 months after DDC discontinuation, but becomes weaker after prolonged recovery. MDB reinduction parallels the rapid increase in p62 and Hsp25 levels as well as keratin 8 cross-linking that is normally associated with MDB formation.

Conclusion:

Persistent posttranslational modifications in chaperone proteins, coupled with protein cross-linking and altered chaperone expression and function likely contribute to the “toxic memory” of DDC-primed mice. We hypothesize that similar changes are important contributors to inclusion body formation in several diseases.

Intermediate filaments (IFs), together with actin microfilaments and microtubules, are the three filamentous systems of the eukaryotic cytoskeleton.^1,2 IFs make up the largest member of the cytoskeleton protein families and are expressed in a cell-specific manner (e.g., desmin in muscle, neurofilaments in neurons, and keratins in epithelial cells).^3,4 The functional importance of IFs is reflected by the growing list of human diseases that are caused by, or predisposed to, IF mutations or variants, respectively.⁵ In addition, IFs are relatively insoluble proteins that comprise the major constituents of protein inclusions/aggregates that are characteristic of several hepatic, neurodegenerative, and muscular disorders.^5–9 These IF-enriched inclusions consist of misfolded and ubiquitinated proteins together with chaperones and the ubiquitin-binding protein p62.^7,10,11

Mallory-Denk-bodies (MDBs) are hepatocyte-specific aggregates and are among the most common inclusion bodies because of the relative high prevalence of associated liver diseases.⁹ MDBs are characteristic of alcoholic steatohepatitis but are also found in other diseases, including nonalcoholic steatohepatitis, primary biliary cirrhosis, and Wilson disease.^9,12,13 MDBs consist mainly of the IFs keratin 8 and 18 (K8/K18) that have undergone several posttranslational modifications, including hyperphosphorylation and transamidation.^9,13 MDB formation requires an alteration in the normal equimolar K8:K18 ratio (under basal conditions) to disproportional K8>K18 levels that ultimately lead to K8 cross-linking and insolubility.^9,13

Our understanding of MDB pathogenesis has been facilitated by mouse models whereby feeding mice the hepatotoxic drugs griseofulvin or 3,5-diethoxycarbonyl-1,4-dihydrocollidin (DDC) for 2 to 5 months leads to MDB formation. Similar to the human situation, murine MDB formation is reversible, and MDBs largely disappear after switching the animals to a normal diet for 1 month.⁹ Interestingly, DDC-primed mice (i.e., mice fed DDC long-term, then allowed to recover by feeding a normal diet) posses a remarkable ability to rapidly reform MDBs within 5 to 7 days after exposure to a variety of stress-inducing agents.^9,12 This phenomenon is referred to as “toxic memory.”¹⁴

Despite the diagnostic importance of MDBs, the precise mechanisms underlying their formation are not completely understood.^9,13 Similar to other nonhepatocyte protein aggregates, MDB formation is thought to be facilitated by altered protein conformation that leads to exposure of hydrophobic amino acids.^7,15 A change in protein conformation might be caused by a genetic defect or by protein misfolding as a consequence of oxidative stress. In support of the latter possibility, MDB-forming diseases are associated with increased levels of oxidative stress, and rapid MDB reinduction in DDC-primed mice can be reversed by pretreatment with the antioxidant S-adenosylmethionine.^16–19

An important mechanism that cells use to prevent protein misfolding is the up-regulation and utilization of chaperones that associate with newly synthesized or damaged proteins and facilitate their correct folding.¹⁵ Heat shock proteins (Hsp) are the best-studied chaperones and are distributed in different subcellular compartments. For example, Hsp60 is found in mitochondria, whereas Hsp70 is a major cytoplasmic chaperone, while other organelles/compartments express different or related chaperone family members.²⁰ However, the protective capacity of chaperones might be depleted or compromised, thereby leading to aggregation of misfolded proteins.^7,11,15 We hypothesized that retained molecular changes in DDC-recovered mice are a component of the “toxic memory” that may play a role in the rapid MDB reformation upon rechallenge. To identify such changes, we used an unbiased proteomic approach that revealed that MDB formation is associated with persistent chaperone-charged isoform changes that lead to functional impairment in several chaperones.

Materials and Methods

Animal Experiments.

MDBs were induced in 2- to 3-month-old male C3H mice (Taconic Farms, German-town, NY) by feeding a powdered Formulab 5008 diet ad libitum for 3.5 months (Dean’s Animal Feeds, Redwood City, CA) containing 0.1% DDC (Aldrich, St. Louis, MO). To analyze MDB reformation, mice were recovered on a DDC-free Formulab 5008 diet (control diet) for 1 or 3 months followed by refeeding with the 0.1% DDC-containing diet for 6 to 7 days (Fig. 1A/Fig.5A schematics). Mice were sacrificed by CO₂ inhalation followed by blood collection using cardiac puncture, then removal of the liver. Serum was used to measure alanine aminotransferase (IU/L), alkaline phosphatase (IU/L), and total bilirubin (mg/dL). Livers were weighed, cut into pieces, and flash-frozen in liquid nitrogen for biochemical and two-dimensional differential-in-gel-electrophoresis (2D-DIGE). Alternatively, freshly isolated liver pieces were embedded in optimal cutting temperature compound for immunofluorescence staining or submerged in RNAlater Stabilization Reagent (Qiagen, Valencia, CA) and stored (−80°C) for subsequent real-time reverse-transcription polymerase chain reaction (RT-PCR) analysis. The animal experiments protocol was approved by the Institutional Animal Care Committee.

Fig. 1.

DDC treatment causes pronounced molecular changes. (A) Schematic of the mouse treatment protocol. Note that the “primed” mice do not exhibit any MDBs, but are rapidly predisposed to their formation. (B) 2D-DIGE of mouse liver homogenates (labeled with Cy3/Cy5) from control mice (Untreated, Cy3) or mice fed DDC for 3.5 months (Cy5-DDC 3.5M). Altered charged chaperone and cytoskeletal protein isoform changes upon DDC feeding are highlighted. (C) Recovery of DDC-fed mice for 1 month on control diet (Rec 1M) leads to resolution of most but not all of the DDC-induced molecular changes, whereas alterations in chaperones (boxed areas) are retained (see also Supplemental Fig. 1). (D) DDC readministration for 6 days to recovered mice (R1M+DDC6d) causes re-emergence of the molecular alterations seen in DDC 3.5 M mice. Three pairs of mice each for panels B-D were compared, and one representative example is shown.

Fig. 5.

The “toxic memory” is attenuated after 3 months of recovery. (A) Schematic of the protocol used to compare 1-month with 3-month recovery after DDC feeding for 3.5 months. (B) Refeeding of recovered mice with DDC leads to rapid MDB formation as noted with double immunofluorescence staining using antibodies to K8 and ubiquitin. MDBs were quantified using immunostained livers from three animals per condition and were found to be nearly 3-fold more abundant in the 1-month (R1M+DDC7d) as compared with the 3-month (R3M+DDC7d) recovery period that was followed by 7 days of rechallenge. Note that DDC administration to previously untreated mice for 7 days (DDC7d) does not lead to MDB formation. Bar = 50 μm.

2D-DIGE Analysis.

Tissue lysates were prepared by homogenizing the preweighed liver pieces in 2D lysis buffer (30 mM Tris-HCl [pH 8.8]; 7 M urea; 2 M thiourea and 4% CHAPS) to achieve a final protein concentration of 4 to 8 mg/mL. The two liver samples to be compared are labeled with Cy3 or Cy5 dyes and a 1:1 mix is also labeled with Cy2 which was included with the test samples as an internal labeling control. Samples were then analyzed using an Amersham BioSciences 2D-gel system (Amersham BioSciences; Piscataway, NJ). Images were scanned using Typhoon TRIO and analyzed by Image Quant version 5.0 software (Amersham BioScience). An in-gel DeCyder software was used to determine the protein level ratios. Selected spots were collected with an Ettan Spot Picker (Amersham BioSciences), subjected to in-gel trypsinization, peptide extraction, and desalting, followed by MALDI-TOF/TOF (Applied Biosystems, Foster City, CA) analysis to determine the protein identity.

Antibodies.

The utilized antibodies are directed toward: K8/K18 (Ab-8592) and K18 (Ab-4668 or Troma-2);²¹ Hsp60 (LK2), transglutaminase-2 (Ab-4), α-tubulin (DM1A), and β-actin (ACTN05) (Labvision, Fremont, CA); Hsp25 (sc-1049), Hsp/c70 (sc-1060), ubiquitin (sc-8017), and p62 (sc-10117) (Santa Cruz Biotechnology, Santa Cruz, CA); K8 (Troma-1) and K19 (Troma-3) (Developmental Studies Hybridoma Bank, Iowa City, IA); and GRP75 (SPS-825), GRP78 (SPA-827), and Hsp72 (SPA-810) (Stressgen Bioreagents, Ann Arbor, MI). Because sc-1060 recognizes both inducible and constitutively-expressed Hsp70 family members, the proteins recognized by this antibody are termed Hsp/c70.

Immunofluorescence Staining and Quantitative Real-Time RT-PCR.

Acetone-fixed liver sections were stained as described,²¹ then analyzed using a Zeiss LSM510 confocal microscope (Zeiss MicroImaging; Thornwood, NY). Quantitative real-time RT-PCR analysis was performed as described²² using liver tissues preserved in RNAlater Stabilization Reagent (Qiagen). RNA translation was performed with Oligo-dT primers and target gene-specific primers (Supplemental Table 1). Three individual mice were tested for each treatment condition.

Immunoblotting and Sucrose Gradient Separation.

Total tissue homogenates were generated using a Dounce to solubilize liver pieces in 3% sodium dodecyl sulfate (SDS) containing sample buffer.²¹ For sucrose gradient sedimentation, livers were homogenized in 1% Triton X100-containing buffer followed by pelleting.²¹ The supernatant was incubated (30 minutes) in the presence or absence of 5 mM MgATP (22°C), then immediately layered over a discontinuous sucrose gradient consisting of 6%, 12%, 18%, 24%, 30% sucrose (2 mL each) and centrifuged (38,000 rpm for 18 hours). The bottom of the tube was then punctured with a 22-gauge needle, and 0.6 mL fractions (#1–21) were collected and diluted 1:1 using 4× reducing SDS-containing sample buffer.

In order to quantify the amount of chaperones in different fractions, solubilized liver homogenates (in 1% Triton X-100) were incubated with or without 5 mM MgATP (30 minutes), then pelleted. The pellet was homogenized in equal volume of 1% Triton X-100/4× reducing SDS-containing sample buffer. The distribution of chaperones within the fractions was examined by immunoblotting. Protein bands were visualized with an ECLplus kit (PerkinElmer, Boston, MA).²¹

Results

DDC Treatment Leads to Persistent Charged Isoform Changes in Several Chaperone Proteins.

We employed an unbiased proteomic approach to identify protein changes that associate with MDB formation and with the “toxic memory” phenomenon that is noted in DDC-primed livers. DDC feeding for 3.5 months resulted in abundant MDB formation (not shown but well described by Zatloukal et al.⁹) and pronounced charged isoform changes in multiple protein categories including chaperones, detoxifying, and cytoskeletal proteins (Fig. 1A,B) (for quantification, see Table 1). The recovery of DDC-exposed mice after switching to control diet for 1 month led to MDB disappearance, and many of the molecular alterations returned to normal (Fig. 1C). However, prominent among the charged isoform changes that did not revert to baseline are those involving chaperones Hsp60/Hsc71 and GRP58/75/78 (Fig. 1; Table 1; Supplemental Fig. 1). To study the rapid MDB reappearance upon DDC rechallenge, recovered mice were refed DDC for 6 days and, as already well established,⁹ DDC readministration caused abundant MDB reformation (data not shown). Notably, the accompanying molecular alterations were similar to those induced by long-term DDC feeding (Table 1 and compare Fig. 1B with Fig. 1D and left and right panels in Supplemental Fig. 1).

Table 1.

DDC Administration Leads to Pronounced Isoform Changes in Multiple Proteins

	DDC 3.5M/Untreated (fold change)					Rec 1M/Untreated (fold change)					R1M+DDC6d/Untreated (fold change)
Isoform #	1	2	3	4	5	1	2	3	4	5	1	2	3	4	5
Cytoskeletal proteins
β-Actin	2.1	2.1	4.4	3.9		1.4	1.2	2.4	2.4		1.9	1.9	3.2	2.8
Keratin 8	1.3	−1.6	4.4	2.9	4.7	2.0	2.1	2.9	1.8		1.2	−1.7	3.5	2.8	3.7
Keratin 18	3.0	3.3	5.1	3.2		1.4	1.6	1.8	1.5		3.5	3.1	3.9	2.2
Chaperones
GRP58						1.7	1.4	2.3
GRP75	−1.2	1.4	1.3			−1.4	2.1	2.3			−1.1	1.5	1.2
GRP78	1.1	1.9				1.0					1.1	1.6
Hsp60	−1.2	1.3	1.4	1.7		−1.5	2.6	3.6	3.5		−1.0	1.4	1.5	1.6
Hsc71	−1.1	−1.1	1.7	1.6		−1.5	2.2	4.1	2.2		1.1	−1.0	1.8	1.8
Hsp 84						1.4					1.6
Detoxification
GST mu	3.7	6.2	1.6			−1.3	2.6	1.5			4.5	6.5	1.5
GST pi	−1.5	−1.4	−1.2								−2.0	−1.6	−1.3
Others:
Aconitate hydralase	−1.7	1.1	1.1			1.3	2.5	1.7			−1.5	1.1	1.1
Albumin	1.4	1.4	2.1	3.0		−1.0	2.4	4.6	4.8		1.4	1.4	2.0	2.8
ALDH4A1	−1.2	−1.7	−1.2	−1.1	−1.5	−1.5	1.0	2.2	3.3	2.1	−1.6	−2.3	−1.5	−1.4	−1.4
Annexin 5	3.6	4.6	7.1			1.4	2.0	2.2			3.0	3.8	4.8
Antioxidant protein 2	1.7					2.7					1.7
Carbonic anhydrase III	−4.7	−3.6	−1.7			−1.3	1.4	1.9			−3.6	−2.7	−1.5
Demethylglycine Dehydrogenase	−1.3	−2.5	−2.6	−2.0	−2.2	−1.5	−1.5	−1.1	1.4	1.5	1.2	−2.1	−2.4	−2.0	−2.4
10-formyltetrahydrofolate dehydrogenase	−1.7	−1.9	−1.5	−1.5	−2.7	−1.9	1.1	1.6	1.3	−1.2	−2.0	−2.1	−1.8	−1.9	−3.3
Heterogenous nuclear ribunucleoprotein	−2.3										−1.7
3-Hydroxy-3-methylgluta-ryl-CoA synthase 2	−3.3	−1.4	−1.4			−2.2	−1.3	−1.4			−3.4	−1.2	−1.2
Major urinary protein	−67	−61	−39	−19		−2.4	−2.2	−1.5	−1.3		−46	−30	−46	−27
Ornithine aminotransferase	−2.0					−2.1					−2.5
Transferrin	2.1	3.1	3.0	3.1		1.0	1.7	2.3	3.9		2.2	2.8	2.6	2.8

Individual spots that showed increased or decreased levels were cut and subjected to proteolysis and characterization as described in Materials and Methods. Three pairs of mice were used for each paired comparison as described in Fig. 1. Values of a representative experiment are shown.

DDC Treatment Results in Altered Chaperone Levels.

We examined the DDC-induced changes in chaperones by immunoblotting and real-time RT-PCR. As described previously, DDC administration increases the levels of cytoskeletal proteins.⁹ However, most chaperone protein levels were either unaltered by DDC treatment (Hsp60, Hsp/c70, GRP78) or decreased (Hsp72, GRP75) (Fig.2), with the caveat that this assessment is based on antibody reactivity as contrasted with 2D-DIGE (Fig. 1) based on protein identification by mass spectrometry. Of the tested chaperones, only Hsp25 was DDC-inducible and was most significantly up-regulated after DDC refeeding (Fig. 2). Quantitative real-time RT-PCR revealed that the lowered GRP75 and Hsp72 levels were at least in part due to transcriptional down-regulation. HSF1/2 levels were not significantly different (Table 2). The constitutive Hsc71 messenger RNA expression was slightly up-regulated after DDC rechallenge when compared with recovered mice, whereas the remaining tested chaperones did not exhibit significant changes (Table 2).

Fig. 2.

DDC treatment leads to altered protein levels of chaperones and cytoskeletal proteins. Immunoblot analysis of total mouse liver lysates depicts alterations in several proteins after DDC administration. Mice were fed DDC for 3.5 months (DDC 3.5M) then recovered for 1 month on a control diet (Rec 1M) followed by DDC refeeding for 6 days (R1M+DDC6d). Long-term DDC feeding or short refeeding causes up-regulation of the cytoskeletal proteins K8, K18, actin, and, to a lesser extent, tubulin. In contrast, Hsp72 and GRP75 become significantly down-regulated after long-term DDC feeding. Two independent mice per condition are shown, but similar findings were noted using at least three mice per condition.

Table 2.

DDC Treatment Results in Altered Chaperone Messenger RNA Expression

	Untreated	DDC 3.5M	Rec 1M	R1M+DDC6d
HSF1	1 ± 0.17	1.05 ± 0.12	1.33 ± 0.12	1.20 ± 0.16
HSF2	1 ± 0.24	0.78 ± 0.07	0.80 ± 0.09	0.78 ± 0.15
GRP75	1 ± 0.13*	0.71 ± 0.03*	0.81 ± 0.11	0.75 ± 0.08
GRP78	1 ± 0.17	0.86 ± 0.09	0.95 ± 0.13	0.75 ± 0.16
Hsp72	1 ± 0.29*	0.15 ± 0.10*	0.59 ± 0.25	0.28 ± 0.11
Hsc71	1 ± 0.20	0.92 ± 0.21	0.86 ± 0.04†	1.18 ± 0.08†
Hsp60	1 ± 0.17	0.70 ± 0.05	0.82 ± 0.02	0.82 ± 0.04

Messenger RNA levels were estimated as described in Materials and Methods. Data were averaged from the amplification of four separate livers and normalized to the levels of untreated mice, which were set as 1. The values are expressed as the mean ± standard deviation.

^*P < 0.05 (when comparing the two marked conditions).

^†P < 0.01 (when comparing the two marked conditions).

DDC Administration Impairs Chaperone–Substrate Association.

To test the effect of DDC treatment on the ability of chaperones to associate with their substrates, we devised a sedimentation-based assay that examined Hsp/c70 and Hsp60 chaperone complex formation and relative size in nonionic detergent-soluble liver lysates. This assay demonstrated that DDC exposure interferes with the ability of Hsp/c70 and Hsp60 to form large-sized complexes as demonstrated by a right-to-left shift in the presence of these chaperones from large to smaller-sized complexes (see boxed areas, Fig. 3A,C). No significant shift was noted in the distribution of Hsp/c70 and Hsp60 after 1 month recovery (Fig. 3A,C).

Fig. 3.

DDC treatment results in altered chaperone function. Livers were removed from age- and sex-matched control (Untreated), 3.5-month DDC-fed, DDC-fed then 1- or 3-month recovered (Rec 1M/3M) mice. Tissues pieces were homogenized in 1% Triton X-100 containing buffer. The supernatants were incubated for 30 minutes in the presence or absence of 5 mM MgATP (+ATP and −ATP) followed by sucrose gradient (0%−30%) sedimentation. Fractions (#1–21) were then collected followed by immunoblotting using antibodies to (A,B) Hsp/c70 and (C,D) Hsp60. Note that DDC feeding causes a decrease in Hsp association with high molecular weight complexes as evidenced by a minimal signal in fractions 1 through 5 (boxes in −ATP panels). In contrast, similar chaperone distribution is noted in livers from 1 month of recovery and the untreated mice (panels A, C). However, ATP treatment induces a more prominent release of chaperone complexes from their substrates in “Rec 1M” as compared with “untreated” livers (based on loss of signal in fractions 5 through 8; ellipses in panels B, D).

Another important indication of appropriate chaperone function is their ability to release their bound substrates via chaperone ATPase activity in the presence of MgATP.23 This chaperone property was examined using our sedimentation assay in the presence of adenosine triphosphate (ATP). Under basal conditions, the distribution profile of Hsp/c70 and Hsp60 switches to lower-sized complexes after ATP exposure, although the pattern of shifting differs for Hsp/c70 (compare Fig. 3A with Fig. 3B) versus Hsp60 (compare Fig. 3C with Fig. 3D). This ATP-induced release is even more pronounced in DDC-fed mice (Fig. 3B,D) which implies that DDC results in unstable chaperone-substrate binding. Notably, 1-month recovery normalizes the complex distribution profile in the absence of ATP (Fig. 3A,C) but not in the presence of ATP (Fig. 3B,D). A prolonged 3-month recovery of DDC-fed mice further normalizes the spread of Hsp/c70 distribution after ATP treatment (relative complex stability) to make it more similar to the non–DDC-treated spread (Fig. 3B) and similar findings were observed for Hsp60 (not shown).

We also examined the effect of DDC and ATP on the distribution of Hsp/c70 and Hsp60 in the detergent-soluble (the fraction we used for the complex formation sedimentation assay) and the detergent-insoluble (the remaining pellet) fractions. Both Hsp60 and Hsp/c70 are predominantly soluble proteins and their distribution is not affected by DDC or ATP, although DDC and ATP do change the overall protein profile based on Coomassie blue staining of the entire fraction as may be expected (Fig. 4). DDC-feeding generates an additional Hsp/c70 antibody-reactive species of slightly slower migration (Fig. 4, arrowhead), which was seen only in Triton-insoluble fractions and likely reflects an uncharacterized insoluble form of Hsp/c70.

Fig. 4.

Hsp60 and Hsp/c70 are primarily Triton-soluble proteins and their solubility is not altered after ATP treatment. Liver homogenates from control mice and mice fed DDC for 3.5 months were prepared using 1% Triton X-100 – containing buffer and incubated with ±5 mM ATP (30 minutes) followed by pelleting. The supernatant (S) was collected and the pellet (P) was suspended in reducing Laemmli sample buffer. Equal fractions were separated by SDS–polyacrylamide gel electrophoresis, then analyzed by Commassie staining and immunoblotting with antibodies to Hsp/c70 and Hsp60.

Priming Toward Rapid MDB Reinduction Becomes Weaker with Time, but Lasts for at Least 3 Months.

We examined the persistence of the “toxic memory” by feeding mice DDC for 3.5 months followed by recovery on a normal diet for 1 or 3 months (Fig. 5A). In both cases, DDC challenge for 7 days led to rapid MDB reformation, albeit to nearly 3-fold less in the 3-month versus the 1-month recovery duration (Fig. 5Bd–i). In contrast, no MDBs were noted in naïve mice fed DDC for 7 days (Fig. 5Ba–c). Therefore, priming toward rapid MDB reinduction lasts for at least 3 months but becomes weaker with time. This supports prior reports showing that rapid reinduction can occur 2 months after recovery.¹⁴ The 3-month recovery also leads to less liver injury upon DDC readministration as determined by the decreased rise in alanine aminotransferase levels (Table 3). Aside from the ability to form MDBs rapidly, primed mice retain slightly elevated alkaline phosphatase levels and repeated DDC exposure results in higher alkaline phosphatase levels when compared with first-time DDC treatment (Table 3). Also, the liver size does not normalize even after 3 months of recovery on control diet and repeated DDC exposure induces pronounced liver hypertrophy (Table 3).

Table 3.

DDC Induces Cholestatic and Hepatic Damage and Liver Hypertrophy

	Untreated	DDC 7d	DDC 3.5M	Rec 1M	R1M + DDC7d	Rec 3M	R3M + DDC7d
Bilirubin (mg/dL)	0.5 ± 0.4	1.3 ± 1.4	0.6 ± 0.7	0.4 ± 0.2	1.1 ± 1.1	ND	0.5 ± 0.5
ALT (IU/L)	49 ± 24	1,342 ± 1,237	1,170 ± 90	51 ± 13	2,006 ± 452*	ND	987 ± 462*
ALP (IU/L)	90 ± 12*	251 ± 8†	654 ± 77	115 ± 6*	479 ± 50†	ND	496 ± 54†
Liver weight/body weight (%)	4.4 ± 0.2	5.4 ± 1.3	16.6 ± 0.2	7.1 ± 0.4	12.0 ± 0.9	6.1 ± 0.1	8.1 ± 0.4

Sex- and age-matched mice were used (4 to 6 mice per condition). Values are expressed as the mean ± standard deviation.

Abbreviations: ALT, alanine aminotransferase; ALP, alkaline phosphatase; ND, not determined.

^*P = 0.05.

^†P < 0.02 (DDC7d versus DDC refeeding).

Repeated DDC Exposure Elicits Molecular Changes Different to Those Induced by One-Time DDC Feeding.

In addition to comparing long-term DDC with DDC rechallenge as described above, we also analyzed the molecular differences in response to short-term (7-day) single administration versus short-term DDC refeeding. Repeat DDC exposure resembles the single short-term course of DDC feeding in terms of the induction of K8 and K18 and to a lesser extent tubulin, as well as the induction of transglutaminase-2 (Fig. 6). However, one distinct difference is that DDC refeeding leads to prominent K8 cross-linking (based on the presence of K8 species in the stacking [S] gel) which is not seen in control livers,24 and to elevated Hsp25, K19, and p62 (Figs. 6 and 7). Although most of K8 cross-linking disappears during recovery, remnant cross-linked K8-containing species do remain up to 3 months after recovery. Similar findings, in terms of increased K19 (Fig. 7) and Hsp25 (Supplemental Fig. 2) were noted by immunofluorescence staining. For example, prominent K19-expressing cells persist 1 or 3 months after recovery and increase significantly after rechallenge (Fig. 7c–f). Interestingly, the chaperone Hsp25 is present at minimal levels in control mice but its levels increase slightly after one-time DDC feeding as compared with the dramatic increase and colocalization with MDBs after DDC rechallenge (Supplemental Fig. 2).

Fig. 6.

DDC refeeding results in a different molecular response when compared with first-time DDC feeding. The levels of the indicated proteins were compared by immunoblotting of liver homogenates obtained from untreated mice (Untreated); mice fed DDC for 7 days (DDC7d); mice fed DDC for 3.5 months, then recovered for 1 or 3 months (Rec 1M and Rec 3M, respectively); and mice refed DDC for 7 days after a 1-month (R1M+DDC7d) or 3-month (R3M+DDC7d) recovery.

Fig. 7.

Double immunofluorescence staining of K8/K18 (red) and K19 (green) highlights the increased number of K19-positive cells in Rec 1M, R1M+DDC7d and R3M+DDC7d mice when compared with the other treatment conditions. Bar = 100 μm.

Discussion

The major findings of our study are as follows. (1) DDC treatment leads to pronounced charged isoform changes in several chaperones including Hsp25/60/GRP58/GRP75/GRP78/Hsc71 which are retained at least 1 month after discontinuation of DDC feeding, whereas changes in numerous other proteins become less prominent upon recovery. (2) DDC feeding results in altered expression of Hsp72, GRP75, and Hsp25 and in functional impairment of Hsp60 and Hsp/c70 chaperone function. Hence, the well-described phenomenon of rapid MDB reappearance⁹ that was termed “toxic memory” by Denk and colleagues¹⁴ corresponds to, at least in part, specific changes in chaperones and their functions. (3) The “toxic memory” phenotype lasts for at least 3 months after DDC discontinuation but becomes weaker as recovery becomes longer. (4) Long-term followed by short-term DDC readministration differs from a single short-term DDC feeding in the extent of K19, Hsp25, and p62 induction as well as K8 cross-linking.

The precise nature of the chaperone charge-altering isoform-generating alterations remain to be defined, but deamidation is one likely candidate modification that was noted in several proteins (not shown). Within chaperones, deamidation of crystallin, a member of the Hsp27 family, occurs during aging and increases during cataract formation.^25,26 Interestingly, crystallin deamidation is in part catalyzed by tissue transglutaminase, an enzyme known to be involved in mouse MDB formation.^24,27 Notably, protein deamidation may lead to protein aggregation as noted for the spontaneous deamidation of the amylin peptide that induces the formation of amyloid-like aggregates.²⁸ Chaperone alterations have been observed after exposure to alcohol and fatty diet,^29–31 which in humans can lead to alcoholic steatohepatitis and nonalcoholic steatohepatitis, respectively.

Another striking feature of the observed chaperone changes is their longevity. Chaperone alterations are retained for at least one month after discontinuation of DDC-feeding, while isoform changes in many other proteins normalize. Further studies are needed to distinguish whether the observed persistent changes are due to an irreversible alteration coupled with a remarkably slow chaperone turnover or due to an ongoing chaperone modification despite withdrawal of the drug. In addition to chaperone modification, DDC-primed and DDC-refed mice posses numerous other alterations in gene expression and protein levels (present study),³² which may influence MDB formation in DDC-primed livers.

In addition to chaperone function, chaperone expression levels are also important. Chronic DDC-feeding diminishes the levels of Hsp72 and GRP75 at the protein and messenger RNA levels (Fig. 2 and Table 2). Although HSF1/HSF2 messenger RNA levels were unaltered in DDC-fed mice (Table 2), impaired HSF transcriptional activity might be responsible, because heat shock factor (HSF) transcriptional activity is regulated at multiple posttranslational levels.³³ Similar to DDC, the human MDB inducer ethanol leads to no or limited chaperone induction,^34,35 whereas liver chaperone levels are regularly elevated in multiple acute and chronic stress conditions.^36–38 Therefore, the low chaperone levels in DDC-fed or chronic alcohol-consuming individuals may facilitate aggregate formation by decreasing chaperone function (Fig. 3) and levels (Fig. 2). Similarly, depleted chaperone levels were observed in aggregate-forming mouse models of Huntington disease³⁹ and amyotrophic lateral sclerosis,⁴⁰ and chaperone overexpression successfully suppressed aggregate generation.^41,42 In contrast, Hsp25 was up-regulated, particularly in DDC-refed mice. Although both chaperones potentially protect cells from protein aggregation,⁴³ Hsp70 may be more important than Hsp25 in MDB formation given its interaction with K8/18⁴⁴ and its easier detectability in MDBs.^10,45

We propose an important role for chaperones during the stepwise process of MDB formation (Fig.8). Initially, several early changes (such as keratin overexpression with an increased K8>K18 ratio) occur that are not sufficient for MDB formation.²⁴ In parallel, elevated oxidative stress^16,17,46 is initially “absorbed” via chaperone stabilization of misfolded proteins. As chaperone function becomes compromised and insufficient (findings herein), progressive keratin misfolding occurs^47,48 together with K8 cross-linking^24,49 that ultimately leads to keratin aggregation and MDB formation (Fig. 8). Upon mouse recovery after switching to a normal diet, keratin expression normalizes and K8 cross-linking disappears except for some remnants (Fig. 6) together with MDB resolution. However, the retained chaperone dysfunction predisposes to rapid and relatively unprotected keratin misfolding, keratin cross-linking, p62 accumulation, and MDB reformation upon repeated DDC exposure (Fig. 6). Persistent chaperone damage may therefore provide one explanation for why rapid MDB reformation can be induced by a variety of stresses in primed mice that would otherwise not induce MDBs in unprimed mice. Further studies using animals with altered chaperone levels/function are needed to clarify their role in MDB formation. While our study focused on Hsp60/70, our findings do not exclude the possibility that other chaperones may play similar or even more important roles.

Fig. 8.

Persistent chaperone dysfunction associates with rapid MDB reformation upon DDC refeeding. The schematic highlights the major molecular events that occur during MDB formation, recovery, and rapid MDB reformation.

Abbreviations:

ATP

adenosine triphosphate

DDC

3,5-diethoxycarbonyl-1,4-dihydrocollidine

2D-DIGE

two-dimensional differential-in-gel-electrophoresis

HSF

heat shock factor

Hsp

heat shock protein

IF

intermediate filament

K

keratin

MDB

Mallory-Denk body

RT-PCR

reverse-transcription polymerase chain reaction

SDS

sodium dodecyl sulfate