“toxic memory” via chaperone modification is a potential mechanism for rapid mallory-denk body...

“Toxic Memory” via Chaperone Modification Is aPotential Mechanism for Rapid Mallory-Denk Body

ReinductionPavel Strnad,1,2 Guo-Zhong Tao,1 Phillip So,1 Kenneth Lau,3 Jim Schilling,3 Yuquan Wei,4 Jian Liao,4 and

M. Bishr Omary1

The cytoplasmic hepatocyte inclusions, Mallory-Denk bodies (MDBs), are characteristic of sev-eral liver disorders, including alcoholic and nonalcoholic steatohepatitis. In mice, MDBs can beinduced by long-term feeding with 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC) for 3 to 4months or rapidly reformed in DDC-induced then recovered mice by DDC refeeding or expo-sure to a wide range of toxins for only 5 to 7 days. The molecular basis for such a rapidreinduction of MDBs is unknown. We hypothesized that protein changes retained after DDCpriming contribute to the rapid MDB reappearance and associate with MDB formation ingeneral terms. Two-dimensional differential-in-gel-electrophoresis coupled with mass spectrom-etry were used to characterize protein changes in livers from the various treatment groups. Thealterations were assessed by real-time reverse-transcription polymerase chain reaction and con-firmed by immunoblotting. DDC treatment led to pronounced charged isoform changes inseveral chaperone families, including Hsp25, 60, 70, GRP58, GRP75, and GRP78, which lastedat least for 1 month after discontinuation of DDC feeding, whereas changes in other proteinsnormalized 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 proteincomplex formation and release assay. The priming toward rapid MDB reinduction lasts for atleast 3 months after DDC discontinuation, but becomes weaker after prolonged recovery. MDBreinduction parallels the rapid increase in p62 and Hsp25 levels as well as keratin 8 cross-linkingthat is normally associated with MDB formation. Conclusion: Persistent posttranslational mod-ifications in chaperone proteins, coupled with protein cross-linking and altered chaperone ex-pression and function likely contribute to the “toxic memory” of DDC-primed mice. Wehypothesize that similar changes are important contributors to inclusion body formation inseveral diseases. (HEPATOLOGY 2008;48:931-942.)

Intermediate filaments (IFs), together with actin mi-crofilaments and microtubules, are the three filamen-tous 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 kera-tins in epithelial cells).3,4 The functional importance ofIFs is reflected by the growing list of human diseases that

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.

From the 1Department of Medicine, Palo Alto VA Medical Center, Palo Alto, CA, and Stanford University Digestive Disease Center; the 2Department of InternalMedicine I, University of Ulm, Ulm, Germany; the 3Department of Pediatrics, Stanford University School of Medicine; and the 4State Key Laboratory of Biotherapy,Sichuan University, Chengdu, China.

Received February 5, 2008; accepted May 9, 2008.Supported by National Institutes of Health Grant DK52951 and the Department of Veterans Affairs (M. B. O.); National Institutes of Health Grant DK56339; and

in part by an EMBO postdoctoral fellowship as well as German Research Foundation grant STR 1095/1-1 (P. S.).Address correspondence to: Bishr Omary, Department of Molecular and Integrative Physiology, 7744 Medical Science Building II, 1301 E. Catherine, Ann Arbor, MI

48109-0622. tel: 734-764-4376; fax: 734-936-8813.Copyright © 2008 by the American Association for the Study of Liver Diseases.Published online in Wiley InterScience (www.interscience.wiley.com).DOI 10.1002/hep.22430Potential conflict of interest: Dr. Omary owns stock in and advises Applied.Supplementary material for this article can be found on the Hepatology Web site (http:interscience.wiley.com/jpages/0270-9139/suppmat/index.html).

931

are caused by, or predisposed to, IF mutations or variants,respectively.5 In addition, IFs are relatively insoluble pro-teins that comprise the major constituents of protein in-clusions/aggregates that are characteristic of severalhepatic, neurodegenerative, and muscular disorders.5-9

These IF-enriched inclusions consist of misfolded andubiquitinated proteins together with chaperones and theubiquitin-binding protein p62.7,10,11

Mallory-Denk-bodies (MDBs) are hepatocyte-specificaggregates and are among the most common inclusionbodies because of the relative high prevalence of associ-ated liver diseases.9 MDBs are characteristic of alcoholicsteatohepatitis but are also found in other diseases, includ-ing nonalcoholic steatohepatitis, primary biliary cirrhosis,and Wilson disease.9,12,13 MDBs consist mainly of the IFskeratin 8 and 18 (K8/K18) that have undergone severalposttranslational modifications, including hyperphos-phorylation and transamidation.9,13 MDB formation re-quires an alteration in the normal equimolar K8:K18 ratio(under basal conditions) to disproportional K8�K18 lev-els that ultimately lead to K8 cross-linking and insolubil-ity.9,13

Our understanding of MDB pathogenesis has beenfacilitated by mouse models whereby feeding mice thehepatotoxic drugs griseofulvin or 3,5-diethoxycarbonyl-1,4-dihydrocollidin (DDC) for 2 to 5 months leads toMDB formation. Similar to the human situation, murineMDB formation is reversible, and MDBs largely disap-pear after switching the animals to a normal diet for 1month.9 Interestingly, DDC-primed mice (i.e., mice fedDDC long-term, then allowed to recover by feeding anormal diet) posses a remarkable ability to rapidly reformMDBs within 5 to 7 days after exposure to a variety ofstress-inducing agents.9,12 This phenomenon is referredto as “toxic memory.”14

Despite the diagnostic importance of MDBs, the pre-cise mechanisms underlying their formation are not com-pletely understood.9,13 Similar to other nonhepatocyteprotein aggregates, MDB formation is thought to be fa-cilitated by altered protein conformation that leads toexposure of hydrophobic amino acids.7,15 A change inprotein conformation might be caused by a genetic defector by protein misfolding as a consequence of oxidativestress. In support of the latter possibility, MDB-formingdiseases are associated with increased levels of oxidativestress, and rapid MDB reinduction in DDC-primed micecan be reversed by pretreatment with the antioxidantS-adenosylmethionine.16-19

An important mechanism that cells use to prevent pro-tein misfolding is the up-regulation and utilization ofchaperones that associate with newly synthesized or dam-aged proteins and facilitate their correct folding.15 Heat

shock proteins (Hsp) are the best-studied chaperones andare distributed in different subcellular compartments. Forexample, Hsp60 is found in mitochondria, whereasHsp70 is a major cytoplasmic chaperone, while other or-ganelles/compartments express different or related chap-erone family members.20 However, the protectivecapacity of chaperones might be depleted or compro-mised, thereby leading to aggregation of misfolded pro-teins.7,11,15 We hypothesized that retained molecularchanges in DDC-recovered mice are a component of the“toxic memory” that may play a role in the rapid MDBreformation upon rechallenge. To identify such changes,we used an unbiased proteomic approach that revealedthat MDB formation is associated with persistent chaper-one-charged isoform changes that lead to functional im-pairment in several chaperones.

Materials and Methods

Animal Experiments. MDBs were induced in 2- to3-month-old male C3H mice (Taconic Farms, German-town, NY) by feeding a powdered Formulab 5008 diet adlibitum for 3.5 months (Dean’s Animal Feeds, RedwoodCity, CA) containing 0.1% DDC (Aldrich, St. Louis,MO). To analyze MDB reformation, mice were recoveredon a DDC-free Formulab 5008 diet (control diet) for 1 or3 months followed by refeeding with the 0.1% DDC-containing diet for 6 to 7 days (Fig. 1A/Fig.5A schemat-ics). Mice were sacrificed by CO2 inhalation followed byblood collection using cardiac puncture, then removal ofthe liver. Serum was used to measure alanine aminotrans-ferase (IU/L), alkaline phosphatase (IU/L), and total bil-irubin (mg/dL). Livers were weighed, cut into pieces, andflash-frozen in liquid nitrogen for biochemical and two-dimensional differential-in-gel-electrophoresis (2D-DIGE). Alternatively, freshly isolated liver pieces wereembedded in optimal cutting temperature compound forimmunofluorescence staining or submerged in RNAlaterStabilization Reagent (Qiagen, Valencia, CA) and stored(�80°C) for subsequent real-time reverse-transcriptionpolymerase chain reaction (RT-PCR) analysis. The ani-mal experiments protocol was approved by the Institu-tional Animal Care Committee.

2D-DIGE Analysis. Tissue lysates were prepared byhomogenizing the preweighed liver pieces in 2D lysisbuffer (30 mM Tris-HCl [pH 8.8]; 7 M urea; 2 M thio-urea and 4% CHAPS) to achieve a final protein concen-tration of 4 to 8 mg/mL. The two liver samples to becompared are labeled with Cy3 or Cy5 dyes and a 1:1 mixis also labeled with Cy2 which was included with the testsamples as an internal labeling control. Samples were thenanalyzed using an Amersham BioSciences 2D-gel system

932 STRNAD ET AL. HEPATOLOGY, September 2008

(Amersham BioSciences; Piscataway, NJ). Images werescanned using Typhoon TRIO and analyzed by ImageQuant version 5.0 software (Amersham BioScience). Anin-gel DeCyder software was used to determine the pro-tein level ratios. Selected spots were collected with anEttan Spot Picker (Amersham BioSciences), subjected toin-gel trypsinization, peptide extraction, and desalting,followed by MALDI-TOF/TOF (Applied Biosystems,

Foster City, CA) analysis to determine the protein iden-tity.

Antibodies. The utilized antibodies are directed to-ward: K8/K18 (Ab-8592) and K18 (Ab-4668 or Troma-2);21 Hsp60 (LK2), transglutaminase-2 (Ab-4), �-tubulin(DM1A), and �-actin (ACTN05) (Labvision, Fremont,CA); Hsp25 (sc-1049), Hsp/c70 (sc-1060), ubiquitin (sc-8017), and p62 (sc-10117) (Santa Cruz Biotechnology,

Fig. 1. DDC treatment causes pronounced molecular changes. (A) Schematic of the mouse treatment protocol. Note that the “primed” mice donot exhibit any MDBs, but are rapidly predisposed to their formation. (B) 2D-DIGE of mouse liver homogenates (labeled with Cy3/Cy5) from controlmice (Untreated, Cy3) or mice fed DDC for 3.5 months (Cy5-DDC 3.5M). Altered charged chaperone and cytoskeletal protein isoform changes uponDDC 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 theDDC-induced molecular changes, whereas alterations in chaperones (boxed areas) are retained (see also Supplemental Fig. 1). (D) DDCreadministration for 6 days to recovered mice (R1M�DDC6d) causes re-emergence of the molecular alterations seen in DDC 3.5 M mice. Three pairsof mice each for panels B-D were compared, and one representative example is shown.

HEPATOLOGY, Vol. 48, No. 3, 2008 STRNAD ET AL. 933

Santa Cruz, CA); K8 (Troma-1) and K19 (Troma-3)(Developmental Studies Hybridoma Bank, Iowa City,IA); and GRP75 (SPS-825), GRP78 (SPA-827), andHsp72 (SPA-810) (Stressgen Bioreagents, Ann Arbor,MI). Because sc-1060 recognizes both inducible and con-stitutively-expressed Hsp70 family members, the proteinsrecognized by this antibody are termed Hsp/c70.

Immunofluorescence Staining and Quantitative Real-Time RT-PCR. Acetone-fixed liver sections were stainedas described,21 then analyzed using a Zeiss LSM510 con-focal microscope (Zeiss MicroImaging; Thornwood,NY). Quantitative real-time RT-PCR analysis was per-formed as described22 using liver tissues preserved inRNAlater Stabilization Reagent (Qiagen). RNA transla-tion was performed with Oligo-dT primers and targetgene-specific primers (Supplemental Table 1). Three in-dividual mice were tested for each treatment condition.

Immunoblotting and Sucrose Gradient Separation.Total tissue homogenates were generated using a Dounceto solubilize liver pieces in 3% sodium dodecyl sulfate

(SDS) containing sample buffer.21 For sucrose gradientsedimentation, livers were homogenized in 1% TritonX100-containing buffer followed by pelleting.21 The su-pernatant was incubated (30 minutes) in the presence orabsence of 5 mM MgATP (22°C), then immediately lay-ered over a discontinuous sucrose gradient consisting of6%, 12%, 18%, 24%, 30% sucrose (2 mL each) andcentrifuged (38,000 rpm for 18 hours). The bottom ofthe tube was then punctured with a 22-gauge needle, and0.6 mL fractions (#1-21) were collected and diluted 1:1using 4� reducing SDS-containing sample buffer.

In order to quantify the amount of chaperones in dif-ferent fractions, solubilized liver homogenates (in 1%Triton X-100) were incubated with or without 5 mMMgATP (30 minutes), then pelleted. The pellet was ho-mogenized in equal volume of 1% Triton X-100/4� re-ducing SDS-containing sample buffer. The distributionof chaperones within the fractions was examined by im-munoblotting. Protein bands were visualized with anECLplus kit (PerkinElmer, Boston, MA).21

Table 1. DDC Administration Leads to Pronounced Isoform Changes in Multiple Proteins

Isoform #

DDC 3.5M/Untreated (fold change) Rec 1M/Untreated (fold change) R1M�DDC6d/Untreated (fold change)

1 2 3 4 5 1 2 3 4 5 1 2 3 4 5

Cytoskeletal proteins�-Actin 2.1 2.1 4.4 3.9 1.4 1.2 2.4 2.4 1.9 1.9 3.2 2.8Keratin 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 3.7Keratin 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

ChaperonesGRP58 1.7 1.4 2.3GRP75 �1.2 1.4 1.3 �1.4 2.1 2.3 �1.1 1.5 1.2GRP78 1.1 1.9 1.0 1.1 1.6Hsp60 �1.2 1.3 1.4 1.7 �1.5 2.6 3.6 3.5 �1.0 1.4 1.5 1.6Hsc71 �1.1 �1.1 1.7 1.6 �1.5 2.2 4.1 2.2 1.1 �1.0 1.8 1.8Hsp 84 1.4 1.6

DetoxificationGST mu 3.7 6.2 1.6 �1.3 2.6 1.5 4.5 6.5 1.5GST 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 1.1Albumin 1.4 1.4 2.1 3.0 �1.0 2.4 4.6 4.8 1.4 1.4 2.0 2.8ALDH4A1 �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 �1.4Annexin 5 3.6 4.6 7.1 1.4 2.0 2.2 3.0 3.8 4.8Antioxidant protein 2 1.7 2.7 1.7Carbonic anhydrase III �4.7 �3.6 �1.7 �1.3 1.4 1.9 �3.6 �2.7 �1.5DemethylglycineDehydrogenase �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-formyltetrahydrofolatedehydrogenase �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 nuclearribunucleoprotein �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 �27Ornithine aminotransferase �2.0 �2.1 �2.5Transferrin 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. Threepairs of mice were used for each paired comparison as described in Fig. 1. Values of a representative experiment are shown.


Results

DDC Treatment Leads to Persistent Charged Iso-form Changes in Several Chaperone Proteins. We em-ployed an unbiased proteomic approach to identifyprotein changes that associate with MDB formation andwith the “toxic memory” phenomenon that is noted inDDC-primed livers. DDC feeding for 3.5 months re-sulted in abundant MDB formation (not shown but well

described by Zatloukal et al.9) and pronounced chargedisoform changes in multiple protein categories includingchaperones, detoxifying, and cytoskeletal proteins (Fig.1A,B) (for quantification, see Table 1). The recovery ofDDC-exposed mice after switching to control diet for 1month led to MDB disappearance, and many of the mo-lecular alterations returned to normal (Fig. 1C). How-ever, prominent among the charged isoform changes thatdid not revert to baseline are those involving chaperonesHsp60/Hsc71 and GRP58/75/78 (Fig. 1; Table 1; Sup-plemental Fig. 1). To study the rapid MDB reappearanceupon DDC rechallenge, recovered mice were refed DDCfor 6 days and, as already well established,9 DDC read-ministration caused abundant MDB reformation (datanot shown). Notably, the accompanying molecular alter-ations were similar to those induced by long-term DDCfeeding (Table 1 and compare Fig. 1B with Fig. 1D andleft and right panels in Supplemental Fig. 1).

DDC Treatment Results in Altered Chaperone Lev-els. We examined the DDC-induced changes in chaper-ones by immunoblotting and real-time RT-PCR. Asdescribed previously, DDC administration increases thelevels of cytoskeletal proteins.9 However, most chaperoneprotein levels were either unaltered by DDC treatment(Hsp60, Hsp/c70, GRP78) or decreased (Hsp72,GRP75) (Fig.2), with the caveat that this assessment isbased on antibody reactivity as contrasted with 2D-DIGE(Fig. 1) based on protein identification by mass spectrom-etry. Of the tested chaperones, only Hsp25 was DDC-inducible and was most significantly up-regulated afterDDC refeeding (Fig. 2). Quantitative real-time RT-PCRrevealed that the lowered GRP75 and Hsp72 levels wereat least in part due to transcriptional down-regulation.HSF1/2 levels were not significantly different (Table 2).The constitutive Hsc71 messenger RNA expression wasslightly up-regulated after DDC rechallenge when com-

Fig. 2. DDC treatment leads to altered protein levels of chaperonesand cytoskeletal proteins. Immunoblot analysis of total mouse liverlysates depicts alterations in several proteins after DDC administration.Mice were fed DDC for 3.5 months (DDC 3.5M) then recovered for 1month 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 lesserextent, tubulin. In contrast, Hsp72 and GRP75 become significantlydown-regulated after long-term DDC feeding. Two independent mice percondition are shown, but similar findings were noted using at least threemice per condition.

Table 2. DDC Treatment Results in Altered ChaperoneMessenger RNA Expression

Untreated DDC 3.5M Rec 1M R1M�DDC6d

HSF1 1 � 0.17 1.05 � 0.12 1.33 � 0.12 1.20 � 0.16HSF2 1 � 0.24 0.78 � 0.07 0.80 � 0.09 0.78 � 0.15GRP75 1 � 0.13* 0.71 � 0.03* 0.81 � 0.11 0.75 � 0.08GRP78 1 � 0.17 0.86 � 0.09 0.95 � 0.13 0.75 � 0.16Hsp72 1 � 0.29* 0.15 � 0.10* 0.59 � 0.25 0.28 � 0.11Hsc71 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 normalizedto the levels of untreated mice, which were set as 1. The values are expressed asthe mean � standard deviation.

*P � 0.05 (when comparing the two marked conditions).†P � 0.01 (when comparing the two marked conditions).


pared with recovered mice, whereas the remaining testedchaperones did not exhibit significant changes (Table 2).

DDC Administration Impairs Chaperone–Sub-strate Association. To test the effect of DDC treatmenton the ability of chaperones to associate with their sub-strates, we devised a sedimentation-based assay that exam-ined Hsp/c70 and Hsp60 chaperone complex formationand relative size in nonionic detergent-soluble liver ly-sates. This assay demonstrated that DDC exposure inter-feres with the ability of Hsp/c70 and Hsp60 to formlarge-sized complexes as demonstrated by a right-to-leftshift in the presence of these chaperones from large tosmaller-sized complexes (see boxed areas, Fig. 3A,C). Nosignificant shift was noted in the distribution of Hsp/c70and Hsp60 after 1 month recovery (Fig. 3A,C).

Another important indication of appropriate chaper-one function is their ability to release their bound sub-strates via chaperone ATPase activity in the presence ofMgATP.23 This chaperone property was examined usingour sedimentation assay in the presence of adenosinetriphosphate (ATP). Under basal conditions, the distri-

bution profile of Hsp/c70 and Hsp60 switches to lower-sized complexes after ATP exposure, although the patternof shifting differs for Hsp/c70 (compare Fig. 3A with Fig.3B) versus Hsp60 (compare Fig. 3C with Fig. 3D). ThisATP-induced release is even more pronounced in DDC-fed mice (Fig. 3B,D) which implies that DDC results inunstable chaperone-substrate binding. Notably, 1-monthrecovery normalizes the complex distribution profile inthe absence of ATP (Fig. 3A,C) but not in the presence ofATP (Fig. 3B,D). A prolonged 3-month recovery ofDDC-fed mice further normalizes the spread of Hsp/c70distribution after ATP treatment (relative complex stabil-ity) to make it more similar to the non–DDC-treatedspread (Fig. 3B) and similar findings were observed forHsp60 (not shown).

We also examined the effect of DDC and ATP on thedistribution of Hsp/c70 and Hsp60 in the detergent-sol-uble (the fraction we used for the complex formationsedimentation assay) and the detergent-insoluble (the re-maining pellet) fractions. Both Hsp60 and Hsp/c70 arepredominantly soluble proteins and their distribution is

Fig. 3. DDC treatment results in altered chaperone function. Livers were removed from age- and sex-matched control (Untreated), 3.5-monthDDC-fed, DDC-fed then 1- or 3-month recovered (Rec 1M/3M) mice. Tissues pieces were homogenized in 1% Triton X-100 containing buffer. Thesupernatants 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 thatDDC feeding causes a decrease in Hsp association with high molecular weight complexes as evidenced by a minimal signal in fractions 1 through5 (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).


not affected by DDC or ATP, although DDC and ATPdo change the overall protein profile based on Coomassieblue staining of the entire fraction as may be expected(Fig. 4). DDC-feeding generates an additional Hsp/c70antibody-reactive species of slightly slower migration

(Fig. 4, arrowhead), which was seen only in Triton-insol-uble fractions and likely reflects an uncharacterized insol-uble form of Hsp/c70.

Priming Toward Rapid MDB Reinduction BecomesWeaker with Time, but Lasts for at Least 3 Months. Weexamined the persistence of the “toxic memory” by feed-ing mice DDC for 3.5 months followed by recovery on anormal 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-monthrecovery duration (Fig. 5Bd-i). In contrast, no MDBs werenoted in naıve mice fed DDC for 7 days (Fig. 5Ba-c). There-fore, priming toward rapid MDB reinduction lasts for atleast 3 months but becomes weaker with time. This supportsprior reports showing that rapid reinduction can occur 2months after recovery.14 The 3-month recovery also leads toless liver injury upon DDC readministration as determinedby the decreased rise in alanine aminotransferase levels (Ta-ble 3). Aside from the ability to form MDBs rapidly, primedmice retain slightly elevated alkaline phosphatase levels andrepeated DDC exposure results in higher alkaline phospha-tase levels when compared with first-time DDC treatment(Table 3). Also, the liver size does not normalize even after 3months of recovery on control diet and repeated DDC ex-posure induces pronounced liver hypertrophy (Table 3).

Repeated DDC Exposure Elicits Molecular ChangesDifferent to Those Induced by One-Time DDC Feed-ing. In addition to comparing long-term DDC with DDCrechallenge as described above, we also analyzed the molec-ular differences in response to short-term (7-day) single ad-ministration versus short-term DDC refeeding. RepeatDDC exposure resembles the single short-term course ofDDC feeding in terms of the induction of K8 and K18 andto a lesser extent tubulin, as well as the induction of transglu-taminase-2 (Fig. 6). However, one distinct difference is thatDDC refeeding leads to prominent K8 cross-linking (basedon the presence of K8 species in the stacking [S] gel) which isnot seen in control livers,24 and to elevated Hsp25, K19, andp62 (Figs. 6 and 7). Although most of K8 cross-linking dis-appears during recovery, remnant cross-linked K8-contain-

Fig. 4. Hsp60 and Hsp/c70 are primarily Triton-soluble proteins andtheir solubility is not altered after ATP treatment. Liver homogenates fromcontrol mice and mice fed DDC for 3.5 months were prepared using 1%Triton X-100–containing buffer and incubated with �5 mM ATP (30minutes) followed by pelleting. The supernatant (S) was collected andthe pellet (P) was suspended in reducing Laemmli sample buffer. Equalfractions were separated by SDS–polyacrylamide gel electrophoresis,then analyzed by Commassie staining and immunoblotting with antibod-ies to Hsp/c70 and Hsp60.

Table 3. DDC Induces Cholestatic and Hepatic Damage and Liver Hypertrophy

Untreated DDC 7d DDC 3.5M Rec 1MR1M �

DDC7d Rec 3MR3M�

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 � 0.5ALT (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).


ing species do remain up to 3 months after recovery. Similarfindings, in terms of increased K19 (Fig. 7) and Hsp25 (Sup-plemental Fig. 2) were noted by immunofluorescence stain-ing. For example, prominent K19-expressing cells persist 1or 3 months after recovery and increase significantly afterrechallenge (Fig. 7c-f). Interestingly, the chaperone Hsp25 ispresent at minimal levels in control mice but its levels in-crease slightly after one-time DDC feeding as comparedwith the dramatic increase and colocalization with MDBsafter DDC rechallenge (Supplemental Fig. 2).

DiscussionThe major findings of our study are as follows. (1)

DDC treatment leads to pronounced charged isoformchanges in several chaperones including Hsp25/60/GRP58/GRP75/GRP78/Hsc71 which are retained atleast 1 month after discontinuation of DDC feeding,whereas changes in numerous other proteins becomeless prominent upon recovery. (2) DDC feeding resultsin altered expression of Hsp72, GRP75, and Hsp25

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 afterDDC feeding for 3.5 months. (B) Refeeding of recovered mice with DDC leads to rapid MDB formation as noted with double immunofluorescence stainingusing antibodies to K8 and ubiquitin. MDBs were quantified using immunostained livers from three animals per condition and were found to be nearly 3-foldmore 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 �m.


and in functional impairment of Hsp60 and Hsp/c70chaperone function. Hence, the well-described phe-nomenon of rapid MDB reappearance9 that wastermed “toxic memory” by Denk and colleagues14 cor-responds to, at least in part, specific changes in chap-erones and their functions. (3) The “toxic memory”phenotype lasts for at least 3 months after DDC dis-continuation but becomes weaker as recovery becomeslonger. (4) Long-term followed by short-term DDCreadministration differs from a single short-term DDCfeeding in the extent of K19, Hsp25, and p62 induc-tion as well as K8 cross-linking.

The precise nature of the chaperone charge-alteringisoform-generating alterations remain to be defined, butdeamidation is one likely candidate modification that wasnoted in several proteins (not shown). Within chaper-ones, deamidation of crystallin, a member of the Hsp27family, occurs during aging and increases during cataractformation.25,26 Interestingly, crystallin deamidation is inpart catalyzed by tissue transglutaminase, an enzymeknown to be involved in mouse MDB formation.24,27

Notably, protein deamidation may lead to protein aggre-gation as noted for the spontaneous deamidation of theamylin peptide that induces the formation of amyloid-like aggregates.28 Chaperone alterations have been ob-

served after exposure to alcohol and fatty diet,29-31 whichin humans can lead to alcoholic steatohepatitis and non-alcoholic steatohepatitis, respectively.

Another striking feature of the observed chaperonechanges is their longevity. Chaperone alterations areretained for at least one month after discontinuation ofDDC-feeding, while isoform changes in many otherproteins normalize. Further studies are needed to dis-tinguish whether the observed persistent changes aredue to an irreversible alteration coupled with a remark-ably slow chaperone turnover or due to an ongoingchaperone modification despite withdrawal of thedrug. In addition to chaperone modification, DDC-primed and DDC-refed mice posses numerous otheralterations in gene expression and protein levels(present study),32 which may influence MDB forma-tion in DDC-primed livers.

In addition to chaperone function, chaperone ex-pression levels are also important. Chronic DDC-feed-ing diminishes the levels of Hsp72 and GRP75 at theprotein and messenger RNA levels (Fig. 2 and Table 2).Although HSF1/HSF2 messenger RNA levels were un-altered in DDC-fed mice (Table 2), impaired HSFtranscriptional activity might be responsible, becauseheat shock factor (HSF) transcriptional activity is reg-ulated at multiple posttranslational levels.33 Similar toDDC, the human MDB inducer ethanol leads to no orlimited chaperone induction,34,35 whereas liver chaper-one levels are regularly elevated in multiple acute andchronic stress conditions.36-38 Therefore, the low chap-erone levels in DDC-fed or chronic alcohol-consumingindividuals may facilitate aggregate formation by de-creasing chaperone function (Fig. 3) and levels (Fig. 2).Similarly, depleted chaperone levels were observed inaggregate-forming mouse models of Huntington dis-ease39 and amyotrophic lateral sclerosis,40 and chaper-one overexpression successfully suppressed aggregategeneration.41,42 In contrast, Hsp25 was up-regulated,particularly in DDC-refed mice. Although both chap-erones potentially protect cells from protein aggrega-tion,43 Hsp70 may be more important than Hsp25 inMDB formation given its interaction with K8/1844 andits easier detectability in MDBs.10,45

We propose an important role for chaperones dur-ing the stepwise process of MDB formation (Fig.8).Initially, several early changes (such as keratin overex-pression with an increased K8�K18 ratio) occur thatare not sufficient for MDB formation.24 In parallel,elevated oxidative stress16,17,46 is initially “absorbed”via chaperone stabilization of misfolded proteins. Aschaperone function becomes compromised and insuf-ficient (findings herein), progressive keratin misfolding

Fig. 6. DDC refeeding results in a different molecular response whencompared with first-time DDC feeding. The levels of the indicated proteinswere compared by immunoblotting of liver homogenates obtained fromuntreated mice (Untreated); mice fed DDC for 7 days (DDC7d); mice fedDDC for 3.5 months, then recovered for 1 or 3 months (Rec 1M and Rec3M, respectively); and mice refed DDC for 7 days after a 1-month(R1M�DDC7d) or 3-month (R3M�DDC7d) recovery.


occurs47,48 together with K8 cross-linking24,49 that ul-timately leads to keratin aggregation and MDB forma-tion (Fig. 8). Upon mouse recovery after switching to anormal diet, keratin expression normalizes and K8cross-linking disappears except for some remnants(Fig. 6) together with MDB resolution. However, theretained chaperone dysfunction predisposes to rapidand relatively unprotected keratin misfolding, keratincross-linking, p62 accumulation, and MDB reforma-tion upon repeated DDC exposure (Fig. 6). Persistentchaperone damage may therefore provide one explana-tion for why rapid MDB reformation can be induced

by a variety of stresses in primed mice that would oth-erwise not induce MDBs in unprimed mice. Furtherstudies using animals with altered chaperone levels/function are needed to clarify their role in MDB for-mation. While our study focused on Hsp60/70, ourfindings do not exclude the possibility that other chap-erones may play similar or even more important roles.

Acknowledgment: We are grateful to Evelyn Resur-rection for assistance with immune staining, Kris Morrowfor help with figure preparation, and Dr. Rainer Muche(University of Ulm) for advice in statistical analysis.

Fig. 7. Double immunofluores-cence staining of K8/K18 (red) andK19 (green) highlights the increasednumber of K19-positive cells in Rec1M, R1M�DDC7d and R3M�DDC7dmice when compared with the othertreatment conditions. Bar � 100 �m.


References1. Bershadsky AD, Vasiliev JM. Cytoskeleton. New York: Plenum Press,

1988:133-154.2. Ku N-O, Zhou X, Toivola DM, Omary MB. The cytoskeleton of

digestive epithelia in health and disease. Am J Physiol 1999;277:G1108-G1137.

3. Herrmann H, Bar H, Kreplak L, Strelkov SV, Aebi U. Intermediate fila-ments: from cell architecture to nanomechanics. Nat Rev Mol Cell Biol2007;8:562-573.

4. Kim S, Coulombe PA. Intermediate filament scaffolds fulfill mechanical,organizational, and signaling functions in the cytoplasm. Genes Dev 2007;21:1581-1597.

5. Omary MB, Coulombe PA, McLean WH. Intermediate filament proteinsand their associated diseases. N Engl J Med 2004;351:2087-2100.

6. Goebel HH. Congenital myopathies with inclusion bodies: a brief review.Neuromuscul Disord 1998;8:162-168.

7. Ross CA, Poirier MA. Protein aggregation and neurodegenerative disease.Nat Med 2004;10:S10-S17.

8. Der Perng M, Su M, Wen SF, Li R, Gibbon T, Prescott AR, et al. TheAlexander disease-causing glial fibrillary acidic protein mutant, R416W,accumulates into Rosenthal fibers by a pathway that involves filamentaggregation and the association of alpha B-crystallin and HSP27. Am JHum Genet 2006;79:197-213.

9. Zatloukal K, French SW, Stumptner C, Strnad P, Harada M, ToivolaDM, et al. From Mallory to Mallory-Denk bodies: what, how and why?Exp Cell Res 2007;313:2033-2049.

10. Zatloukal K, Stumptner C, Fuchsbichler A, Heid H, Schnoelzer M, Ken-ner L, et al. p62 Is a common component of cytoplasmic inclusions inprotein aggregation diseases. Am J Pathol 2002;160:255-263.

11. Macario AJ, Conway de Macario E. Sick chaperones, cellular stress, anddisease. N Engl J Med 2005;353:1489-1501.

12. Jensen K, Gluud C. The Mallory body: morphological, clinical and experi-mental studies (Part 1 of a literature survey). HEPATOLOGY 1994;20:1061-1077.

13. Ku NO, Strnad P, Zhong BH, Tao GZ, Omary MB. Keratins let liver live:mutations predispose to liver disease and crosslinking generates Mallory-Denk bodies. HEPATOLOGY 2007;46:1639-1649.

14. Denk H, Stumptner C, Zatloukal K. Mallory bodies revisited. J Hepatol2000;32:689-702.

15. Bukau B, Weissman J, Horwich A. Molecular chaperones and proteinquality control. Cell 2006;125:443-451.

16. Mehta K, Van Thiel DH, Shah N, Mobartan S. Nonalcoholic fatty liverdisease: pathogenesis and the role of antioxidants. Nutr Rev 2002;60:289-293.

17. Dey A, Cederbaum AI. Alcohol and oxidative liver injury. HEPATOLOGY

2006;43(Suppl):S63-S74.18. Farrell GC, Larter CZ. Nonalcoholic fatty liver disease: from steatosis to

cirrhosis. HEPATOLOGY 2006;43(Suppl):S99-S112.19. Li J, Bardag-Gorce F, Dedes J, French BA, Amidi F, Oliva J, et al. S-

adenosylmethionine prevents mallory denk body formation in drug-primed mice by inhibiting the epigenetic memory. HEPATOLOGY 2008;47:613-624.

20. Georgopoulos C, Welch WJ. Role of the major heat shock proteins asmolecular chaperones. Annu Rev Cell Biol 1993;9:601-634.

21. Ku NO, Toivola DM, Zhou Q, Tao GZ, Zhong B, Omary MB. Studyingsimple epithelial keratins in cells and tissues. Methods Cell Biol 2004;8:489-517.

22. Nakamichi I, Toivola DM, Strnad P, Michie SA, Oshima RG, BaribaultH, et al. Keratin 8 overexpression promotes mouse Mallory body forma-tion. J Cell Biol 2005;171:931-937.

23. Gething MJ, Sambrook J. Protein folding in the cell. Nature 1992;355:33-45.

24. Strnad P, Harada M, Siegel M, Terkeltaub RA, Graham RM, Khosla C, etal. Transglutaminase-2 regulates Mallory body inclusion formation andliver size. Gastroenterology 2007;132:1515-1526.

25. Van Kleef FS, De Jong WW, Hoenders HJ. Stepwise degradations anddeamidation of the eye lens protein alpha-crystallin in ageing. Nature1975;258:264-266.

Fig. 8. Persistent chaperone dysfunction associates with rapid MDB reformation upon DDC refeeding. The schematic highlights the major molecularevents that occur during MDB formation, recovery, and rapid MDB reformation. [Color figure can be viewed in the online issue, which is availableat www.interscience.wiley.com]


26. Takemoto L, Boyle D. Increased deamidation of asparagine during humansenile cataractogenesis. Mol Vis 2000;6:164-168.

27. Boros S, Wilmarth PA, Kamps B, de Jong WW, Bloemendal H, Lampi K,et al. Tissue transglutaminase catalyzes the deamidation of glutamines inlens betaB(2)- and betaB(3)-crystallins. Exp Eye Res 2008;86:383-393.

28. Nilsson MR, Driscoll M, Raleigh DP. Low levels of asparagine deamidationcan have a dramatic effect on aggregation of amyloidogenic peptides: implica-tions for the study of amyloid formation. Protein Sci 2002;11:342-349.

29. Carbone DL, Doorn JA, Kiebler Z, Sampey BP, Petersen DR. Inhibitionof Hsp72-mediated protein refolding by 4-hydroxy-2-nonenal. Chem ResToxicol 2004;17:1459-1467.

30. Carbone DL, Doorn JA, Kiebler Z, Ickes BR, Petersen DR. Modificationof heat shock protein 90 by 4-hydroxynonenal in a rat model of chronicalcoholic liver disease. J Pharmacol Exp Ther 2005;315:8-15.

31. Moon KH, Hood BL, Kim BJ, Hardwick JP, Conrads TP, Veenstra TD,et al. Inactivation of oxidized and S-nitrosylated mitochondrial proteins inalcoholic fatty liver of rats. HEPATOLOGY 2006;44:1218-1230.

32. Bardag-Gorce F, Dedes J, French BA, Oliva JV, Li J, French SW. Mallorybody formation is associated with epigenetic phenotypic change in hepa-tocytes in vivo. Exp Mol Pathol 2007;83:160-168.

33. Voellmy R. Feedback regulation of the heat shock response. Handb ExpPharmacol 2006;172:43-68.

34. Calabrese V, Renis M, Calderone A, Russo A, Barcellona ML, Rizza V.Stress proteins and SH-groups in oxidant-induced cell damage after acuteethanol administration in rat. Free Radic Biol Med 1996;20:391-397.

35. Rakonczay Z Jr, Boros I, Jarmay K, Hegyi P, Lonovics J, Takacs T. Ethanoladministration generates oxidative stress in the pancreas and liver, but failsto induce heat-shock proteins in rats. J Gastroenterol Hepatol 2003;18:858-867.

36. Fujio N, Hatayama T, Kinoshita H, Yukioka M. Induction of mRNAs forheat shock proteins in livers of rats after ischemia and partial hepatectomy.Mol Cell Biochem 1987;77:173-177.

37. Broome U, Scheynius A, Hultcrantz R. Induced expression of heat-shockprotein on biliary epithelium in patients with primary sclerosing cholan-gitis and primary biliary cirrhosis. HEPATOLOGY 1993;18:298-303.

38. Lee KJ, Terada K, Oyadomari S, Inomata Y, Mori M, Gotoh T. Induction ofmolecular chaperones in carbon tetrachloride-treated rat liver: implications inprotection against liver damage. Cell Stress Chaperones 2004;9:58-68.

39. Hay DG, Sathasivam K, Tobaben S, Stahl B, Marber M, Mestril R, et al.Progressive decrease in chaperone protein levels in a mouse model of Hun-tington’s disease and induction of stress proteins as a therapeutic approach.Hum Mol Genet 2004;13:1389-1405.

40. Maatkamp A, Vlug A, Haasdijk E, Troost D, French PJ, Jaarsma D.Decrease of Hsp25 protein expression precedes degeneration of motoneu-rons in ALS-SOD1 mice. Eur J Neurosci 2004;20:14-28.

41. Cummings CJ, Mancini MA, Antalffy B, DeFranco DB, Orr HT, ZoghbiHY. Chaperone suppression of aggregation and altered subcellular protea-some localization imply protein misfolding in SCA1. Nature Genet 1998;19:148-154.

42. Vacher C, Garcia-Oroz L, Rubinsztein DC. Overexpression of yeasthsp104 reduces polyglutamine aggregation and prolongs survival of atransgenic mouse model of Huntington’s disease. Hum Mol Genet 2005;14:3425-3433.

43. Meriin AB, Sherman MY. Role of molecular chaperones in neurodegen-erative disorders. Int J Hyperthermia 2005;21:403-419.

44. Liao J, Lowthert LA, Ghori N, Omary MB. The 70-kDa heat shock pro-teins associate with glandular intermediate filaments in an ATP-dependentmanner. J Biol Chem 1995;270:915-922.

45. Janig E, Stumptner C, Fuchsbichler A, Denk H, Zatloukal K. Interac-tion of stress proteins with misfolded keratins. Eur J Cell Biol 2005;84:329-339.

46. Tephly TR, Coffman BL, Ingall G, Ziet-Har MS, Goff HM, TabbaHD, et al. Identification of N-methylprotoporphyrin IX in livers ofuntreated mice and mice treated with 3, 5-diethoxycarbonyl-1, 4-dihy-drocollidine: source of the methyl group. Arch Biochem Biophys 1981;212:120-126.

47. Cadrin M, French SW, Wong PT. Alteration in molecular structure of cy-toskeleton proteins in griseofulvin-treated mouse liver: a pressure tuning infra-red spectroscopy study. Exp Mol Pathol 1991;55:170-179.

48. Kachi K, Wong PT, French SW. Molecular structural changes in Mallorybody proteins in human and mouse livers: an infrared spectroscopy study.Exp Mol Pathol 1993;59:197-210.

49. Zatloukal K, Fesus L, Denk H, Tarcsa E, Spurej G, Bock G. High amountof epsilon-(gamma-glutamyl)lysine cross-links in Mallory bodies. Lab In-vest 1992;66:774-777.


“toxic memory” via chaperone modification is a potential mechanism for rapid mallory-denk body...

Documents