Functional Proteomic Analysis of Nonalcoholic FattyLiver Disease in Rat Models: Enoyl–Coenzyme

A Hydratase Down-Regulation ExacerbatesHepatic Steatosis

Xuequn Zhang,1,2 Juntao Yang,1 Yuanbiao Guo,1 Hua Ye,2 Chaohui Yu,2 Chengfu Xu,2 Lei Xu,2 Songfeng Wu,1 Wei Sun,1

Hangdong Wei,1 Xue Gao,1 Yunping Zhu,1 Xiaohong Qian,1 Ying Jiang,1 Youming Li,2 and Fuchu He1,3

Nonalcoholic fatty liver disease (NAFLD) has emerged as a common public health problem that canprogress to end-stage liver disease. A high-fat diet (HFD) may promote the development of NAFLDthrough a mechanism that is poorly understood. We adopted a proteomic approach to examine theeffectofHFDonthe liverproteomeduringtheprogressionofNAFLD.MaleSprague-Dawleyrats fedan HFD for 4, 12, and 24 weeks replicated the progression of human NAFLD: steatosis, nonspecificinflammation, and steatohepatitis. Using two-dimensional difference gel electrophoresis (DIGE)combined with matrix-assisted laser desorption ionization time of flight/time of flight analysis, 95proteins exhibiting significant changes (ratio > 1.5 or <�1.5, P < 0.05) during the development ofNAFLD were identified. Biological functions for these proteins reflected phase-specific characteristicsduring theprogressionof thedisease.Thepotential roleof enoyl–coenzymeAhydratase (ECHS1), anenzyme that catalyzes the second step of mitochondrial fatty acid beta-oxidation, received furtherinvestigation. First, the reduced protein level of ECHS1 was validated both in rat models and inpatients with biopsy-proven hepatic simple steatosis via immunoblotting or immunohistochemicalanalysis. Then the small interfering RNA (siRNA)–mediated knockdown of ECHS1 in the murinehepatocyte cell line alpha mouse liver 12 (AML12) demonstrated increased cellular lipid accumula-tion induced by free fatty acid (FFA) overload. Furthermore, using a hydradynamic transfectionmethod, the in vivo silencing effect of siRNA duplexes targeting ECHS1 was further investigated inmice. Administering ECHS1 siRNA specifically reduced the expression of ECHS1 protein in miceliver, which significantly exacerbated the hepatic steatosis induced by an HFD. Conclusion: Ourresults revealed that ECHS1 down-regulation contributed to HFD-induced hepatic steatosis, whichmay help clarify the pathogenesis of NAFLD and point to potential targets for therapeutic interven-tions. (HEPATOLOGY 2010;51:000-000.)

Nonalcoholic fatty liver disease (NAFLD) hasemerged as one of the most common chronicliver diseases over the past decade, with a prev-

alence of 20% to 30% in the general population.1-3 It

encompasses a spectrum of liver damage ranging fromsteatosis to steatohepatitis, cirrhosis, and even hepatocel-lular carcinoma.4,5 Although the “two-hit” hypothesis hasbecome a widely accepted framework to guide current

Abbreviations: C/EBP, CCAAT/enhancer binding protein; DIGE, difference gel electrophoresis; ECHS1, enoyl-CoA hydratase; FABPL, liver fatty acid binding protein; FFA,free fatty acid; HFD, high-fat diet; IPA, ingenuity pathway analysis; NAFLD, non-alcoholic fatty liver disease; NASH, non-alcoholic steatohepatitis; PBS, phosphate-buffered saline;PDI, protein disulfide isomerase; PPAR�, peroxisome proliferator activated receptor; siRNA, small interfering RNA; TG, triglyceride.

From the 1State Key Laboratory of Proteomics, Beijing Proteome Research Center, Beijing Institute of Radiation Medicine, Beijing, China; 2Digestive Department, theFirst Affiliated Hospital, College of Medicine, Zhejiang University, Hangzhou, China; and 3Institutes of Biomedical Sciences, Fudan University, Shanghai, China.

Received July 2, 2009; accepted November 16, 2009.Supported in part by the Chinese State High-tech Program (863) (2006AA02A308) Chinese State Key Projects for Basic Research (973) (nos. 2006CB910401,

2006CB910801, 2006CB910602, and 2010CB912700), National Natural Science Foundation of China (30700356, 30900677, 30700988, and 30972909),National Natural Science Foundation of China for Creative Research Groups (30621063), Chinese State Key Project Specialized for Infectious Diseases (2008ZX10002-016, 2009ZX10004-103, 2009ZX09301-002), and International Scientific Collaboration Program (2009DFB33070).

Address reprint requests to: Fuchu He, State Key Laboratory of Proteomics, Beijing Proteome Research Center, Beijing Institute of Radiation Medicine, Beijing 102206,China. E-mail: hefc@nic.bmi.ac.cn; fax: 8610-68177417; or Youming Li at: zlym@zju.edu.cn; or Ying Jang at: iangying304@hotmail.com.

Copyright © 2009 by the American Association for the Study of Liver Diseases.Published online in Wiley InterScience (www.interscience.wiley.com).DOI 10.1002/hep.23486Potential conflict of interest: Nothing to report.Additional Supporting Information may be found in the online version of this article.

1

studies in this area,6 the pathogenesis of NAFLD remainslargely unknown. There is an urgent need to better un-derstand how NAFLD occurs and to develop treatmentstrategies for afflicted patients.

It is increasingly recognized that both genetic and ac-quired factors contribute to the pathogenesis of NAFLD.Among the acquired determinants, diet and nutrition (inparticular the amount and type of fat intake) can directlyaffect hepatic fatty infiltration and oxidative damage indifferent types of liver disease.7 Higher intake of saturatedfat and a lower intake of the antioxidant vitamins C and Ewere characteristics of patients with nonalcoholic steato-hepatitis (NASH) compared with obese people presentingno evidence of liver disease.8 Feeding Sprague-Dawleyrats a high-fat diet (HFD) resulted in the reproduction ofmany key features of human NAFLD, including obesity,liver steatosis, inflammation, insulin resistance, abnormalmitochondria, oxidative stress, and increased collagencontent.9 However, the exact role of saturated fat in thedevelopment of NAFLD remains poorly understood.

Proteomics may constitute an essential approach forelucidating the complex pathogenesis of NAFLD.10 Pro-teomic tools, such as two-dimensional difference gel elec-trophoresis (DIGE), provide a global approach to thediscovery and quantification of the functional dynamicsof the cell.11 As a modified form of two-dimensional elec-trophoresis, two-dimensional DIGE is based on labelingtwo to three samples with different fluorescent cyaninedyes that are then co-separated on the same two-dimen-sional gel, which minimizes spot pattern variability.Moreover, by incorporation of the same internal standardon every two-dimensional gel, DIGE has reduced theproblem of inter-gel variation and improved quantifica-tion accuracy. With high reproducibility and reliability,two-dimensional DIGE technology has been appliedwidely to comparative proteomics.12,13

In the current study, we attempted to define the liverproteins that contribute to the pathogenesis and progres-sion of NAFLD in animal models. Male Sprague-Dawleyrats fed with an HFD for 4, 12, and 24 weeks replicatedthe natural history of human NAFLD: simple steatosis,nonspecific inflammation, and then NASH. High-throughput proteomic tools, two-dimensional DIGEcombined with tandem mass spectrometry, revealed acluster of proteins exhibiting phase-specific changes ateach stage of the disease. Among them, enoyl–coenzymeA hydratase (ECHS1), an enzyme that catalyzes the sec-ond step of mitochondrial fatty acid beta-oxidation, un-derwent further investigation. The reduced level ofECHS1 was validated both in rat models and in patientswith simple steatosis. We further demonstrated that thedown-regulation of ECHS1 by small interfering RNA

(siRNA) significantly exacerbated lipid accumulation inhepatocytes induced by free fatty acid (FFA) overloadboth in vitro and in vivo.

Materials and Methods

Animals Models of NAFLD. Male 12-week-oldSprague-Dawley rats weighting 160 to 170 g were pur-chased from the Medical Science Institution of ZhejiangProvince (Hangzhou, China). All rats were maintained ina 12-hour day/night cycle at controlled room temperatureand provided free access to food and water. The rats wereacclimatized to laboratory conditions for 1 week beforethe study and then randomly divided into two groups: (1)control group (n � 30): animals treated with the standardchow diet (8% rice bran, 51% maize, 30% soybean pow-der, 3% bone powder, 1.3% multivitamin, and 6.7%mineral); (2) HFD group (n � 30): animals given anHFD (80.5% SCD, 2% cholesterol, 7% lard, 10% yolkpowder, and 0.5% bile salt). Ten rats of each group weresacrificed at weeks 4, 12, and 24 after being placed on therespective diet. Blood was collected just before sacrificefor serum biochemical analysis. Livers were quickly ex-cised, cleaned completely with ice-cold phosphate-buff-ered saline (PBS), and preserved in liquid nitrogen untiluse. Pieces of the remaining liver tissues were processedfor histology. All animal studies were approved by theAnimal Care and Use Committee of Zhejiang Universityin accordance with the Chinese guidelines for the care anduse of laboratory animals.

Patients and Samples. Access to human tissues com-plied with the guidelines of the Ethics Committee of Zhe-jiang Province. Twenty patients with biopsy-provenNAFLD were enrolled from Medical Treatment Centerof Lihuili Hospital for further validation. All of the pa-tients were defined as simple steatosis according to thehistological manifestation. Patients were excluded if alco-hol consumption was more than 40 g/day for men andmore than 20 g/day for women and if other liver diseaseswere detected by serological testing and imaging studies.For comparison purpose, 14 healthy controls undergoinghemangioma surgery at Lihuili Hospital without clinicalsigns or symptoms of other cause of liver disease, and nohistory of chronic illnesses, were analyzed (SupportingTable 1).

Histological Examination. See Supporting Material.Two-Dimensional DIGE and Matrix-Assisted La-

ser Desorption Ionization Time of Flight/Time ofFlight Analysis. The prepared liver tissue lysates werelabeled with Cy2, Cy3, and Cy5 following the protocolsdescribed in the Ettan DIGE User Manual SupportingTable 2). Gel electrophoresis in the first dimension was

2 ZHANG ET AL. HEPATOLOGY, Month 2010

performed using an immobilized pH gradient gel (Immo-biline Dry Strip gel, pH 3-10 linear, 18 cm; GE Health-care, Uppsala, Sweden). The two-dimensional separationwas performed on 12%, 20 � 24 cm2 sodium dodecylsulfate polyacrylamide gels. Cy2-labeled, Cy3-labeled,and Cy5-labeled images were acquired on a Typhoon9410 scanner (GE Healthcare) at the excitation/emissionvalues (nm) of 488/520, 532/580 and 633/670, respec-tively. The resulting DIGE images were analyzed usingDeCyder version 5.02 software (GE Healthcare). Statis-tically significant spots (P � 0.05) as compared by Stu-dent test and with an average volume ratio over 1.5-foldbetween control and HFD groups were defined as “differ-entially-expressed proteins” and excised from preparativegels for in-gel digestion and matrix-assisted laser desorp-tion ionization time of flight/time of flight identification(see Supporting Material).

Functional Annotation and Pathway Analysis.Differentially expressed proteins during the development ofNAFLD were characterized in more detail using various an-notation tools, including the “Panther Classification Sys-tem” (www.pantherdb.org) for functional annotation andthe “Ingenuity Pathway Analysis” (IPA) (www.ingenuity.com) for pathway and network-oriented analysis.

Immunoblotting and Immunohistochemistry. SeeSupporting Material.

Small Interfering RNA. For RNA silencing, four dif-ferent sequences of siRNA targeting mouse ECHS1 weredesigned and synthesized by GenePharma. The siRNA 1was raised against the sequence beginning at nt 344 (5�-GAACATTTCAGGACTGTTA-3�), siRNA 2 at 413(5�-cggtcatcgcagctgtcaa-3�), siRNA 3 at 480 (5�-catctat-gctggcgagaaa-3�), and siRNA 4 at 766 (5�-gcctttgagat-gacgttaa-3�). Nonsilencing siRNA was a commerciallyavailable duplex (GenePharma) that was used as a nega-tive control siRNA. The siRNA 4 with 2�-O-methylmodification was used for in vivo studies.

Cell Tranfection and In Vitro Model of CellularSteatosis. For gene silencing, the mouse hepatocyte celllines, AML12 hepatocytes were transfected with 200pmol siRNA using the Lipofectamine 2000 (Invitrogen)as instructed by the manufacturer. Seventy-two hours af-ter transfection, protein extracts were analyzed by immu-noblotting to confirm protein knockdown.

Fat overloading induction of cells was done mainlyaccording to previously established methods14 in whichAML12 cells were exposed to a mixture of FFAs (oleateand palmitate) at a final ratio of 2:1 and final concentra-tion of 1 mM. To assess the influence of ECHS1 knock-down on cellular steatosis, cells were treated with FFAsafter 72 hours of siRNA transfection and harvested after24 hours incubation with FFAs.

Hydrodynamic Tail Vein Injection and In VivoModel of Hepatic Steatosis. Male Balb/c mice, 8 weeksof age, were purchased from Vital River and acclimatedfor 1 week after arrival before they were used for experi-ments. Synthetic ECHS1 siRNA 4 with 2�-OMe modifi-cation were delivered in vivo using a modified“hydrodynamic transfection method”15 by which 60 �gsiRNA dissolved in 1.8 mL PBS was rapidly injected intothe tail vein. The control mice were injected with an equalvolume of negative control siRNA. The animals werethen treated with control or HFD diet after 24 hours ofsiRNA injection and sacrificed at the fourth week afterbeing placed on the diet. The siRNA injection was re-peated once per week during on the diet.

Oil Red O Staining. Cells were washed twice withPBS and fixed with 7.5% formaldehyde in PBS for 30minutes. After two washes in PBS, cells were stained for15 minutes in freshly diluted Oil Red O solution. Thedishes were then rinsed in water and counterstained withhematoxylin for 10 seconds. To determine hepatic lipidaccumulation, frozen sections of liver (10 �m) werestained with Oil Red O for 10 minutes, washed, andcounterstained with hematoxylin for 20 seconds. Repre-sentative photomicrographs were captured at 200� mag-nification using a system incorporated in the microscope.

Triglyceride and FFA Assay. Intracellular and livertriglycerides were assayed using a triglyceride assay kit(GPO-POD; Applygen Technologies Inc., Beijing,China). Intracellular and hepatic free fatty acids were es-timated using an ultrasensitive assay kit for free fatty acids(Applygen Technologies Inc.) according to the manufac-turer’s recommended protocol.

Statistical Analysis. Results are expressed as means �standard error. The significance of the difference in meanswas determined by two-tailed Student t test.

Results

Effects of High-Fat Diet on Sprague-Dawley Rats.The HFD successfully induced hepatic steatosis and ste-atohepatitis in rats (Fig. 1). After 4 weeks of HFD treat-ment, approximately 30%-60% of all hepaticparenchymal cells were filled with multivesicular fat(mainly microvesicular), but no necroinflammatory le-sion was observed (Fig. 1B). At 12 weeks, besides amarked increase in fat droplets, focal hepatocyte necrosiswas notable in the centrilobular region with mild to mod-erate infiltration of inflammatory cells (Fig. 1C). By 24weeks, most of the hepatic lobules lost their normal archi-tecture, which was associated with ballooned hepatocytesin centrilobular parenchyma and plenty of macrovesicularfat (Fig. 1D). No histological abnormalities were ob-

HEPATOLOGY, Vol. 51, No. 2, 2010 ZHANG ET AL. 3

served in the livers of the control group (Fig. 1A). TheHFD resulted in a significant (P � 0.01) increase in thecirculating levels of alanine aminotransferase, aspartateaminotransferase, triglycerides (TG), and cholesterol.There was also an increase in the levels of insulin (P �0.05) and a decrease in leptin (P � 0.05). However, therewere no significant effects of HFD on body weight, serumTG, and glucose levels (Table 1).

Quantitative Proteomic Analysis. The liver proteinprofiles of Sprague-Dawley rats fed with HFD and con-trol diet (Chow) for 4, 12, or 24 weeks were determinedby a two-dimensional DIGE approach. Based on the re-sults of the image analysis with DeCyder software (GEHealthcare), an average of 1600-1800 protein spots weredetected per gel. Differentially expressed spots were pro-duced by the two-dimensional pattern comparison be-tween control and HFD groups at each time point.According to the Student t tests, 111, 63, and 68 protein

spots showed more than 1.5-fold difference with statisti-cal significance (P � 0.05) by 4, 12, and 24 weeks, respec-tively. These protein spots were picked from thepreparative gels and subject to in-gel digestion and massspectometry/mass spectometry analysis. For each timepoint, 53, 42, and 49 differentially expressed proteinswere identified (Fig. 2), with 95 unique proteins observedaltogether (Supporting Table 3).

Immunoblot analyses of several proteins with commer-cially available antibodies were performed to confirm theDIGE result. Among them are enzymes involved in lipidmetabolism (liver fatty acid binding protein [FABPL],ECHS1) and amino acid metabolism (carbamoyl-phos-phate synthase, argininosuccinate synthase), endoplasmicreticulum chaperones (protein disulfide isomerase [PDI],protein disulfide isomerase A3), and proteins in otherprocesses (14-3-3 epsilon, cofilin 1, and macrophage mi-gration inhibitory factor). As shown in Fig. 3, the resultsof western blot were in accordance with that of proteomicanalysis.

Functional Annotation of Differentially ExpressedProteins. Differentially expressed proteins at each timepoint were classified into several functional categories us-ing the tools at www.pantherdb.org, which showedphase-specific biological characteristics during the devel-opment of NAFLD (Supporting Fig. 1). At the simplesteatosis stage, almost half of the identified proteins weremetabolic enzymes and transporters, involved in lipid,amino acid, carbohydrate, and other metabolic processes.By the second stage (nonspecific inflammation), mito-chondrial proteins were mostly affected (approximately25%), including four enzymes involved in mitochondrialelectron transduction and oxidative phosphorylation. Atthe last stage (NASH), nearly 20% were cytoskeleton pro-teins, which showed significantly increased expressionlevels during the development of NASH.

Biological functions and signaling networks were fur-ther identified by IPA, a robust tool to develop biological

Fig. 1. Hepatic pathology. Liver sections stained with hematoxylin-eosin from rats fed with SCD for 12 weeks (A) or HFD for 4 weeks (B),12 weeks (C), or 24 weeks (D). Original magnification, 200�.

Table 1. Serum Biochemical Parameters of Animal Models

Parameter

4w (n � 10) 12w (n � 10) 24w (n � 10)

Chow HFD Chow HFD Chow HFD

Weight (g) 376.5 � 23.3 385.7 � 25.8 453.7 � 34.9 483.4 � 36.0 553.0 � 45.3 554.1 � 65.0Triglycerides (mmol/L) 0.33 � 0.14 0.40 � 0.21 0.47 � 0.09 0.51 � 0.18 0.50 � 0.14 0.54 � 0.26Cholesterol (mmol/L) 1.42 � 0.35 2.05 � 0.46** 1.71 � 0.18 2.57 � 0.49** 1.95 � 0.32 2.79 � 0.48**ALT (U/L) 37.9 � 7.8 64.2 � 27.6** 44.9 � 4.5 217.5 � 31.2** 46.6 � 6.79 140.1 � 58.0**AST (U/L) 148.4 � 54.3 179.6 � 42.6 127.7 � 16.9 328.6 � 45.2** 136.6 � 21.7 215.9 � 56.5**Glucose (mg/dL) 5.59 � 1.87 4.45 � 1.06 5.33 � 0.46 4.83 � 0.54 5.67 � 0.75 5.34 � 0.62Insulin (ng/dL) 24.79 � 1.47 27.25 � 3.05* 27.05 � 8.92 29.84 � 5.04 27.82 � 3.46 28.45 � 4.78Leptin (ng/dL) 0.59 � 0.25 0.39 � 0.07* 0.74 � 0.26 0.55 � 0.17 0.91 � 0.42 0.56 � 0.34

Data are the mean � SEM. ALT, alanine aminotransferase; AST, aspartate aminotransferase.*P � 0.05, Chow versus HFD.**P � 0.01 Chow versus HFD.

4 ZHANG ET AL. HEPATOLOGY, Month 2010

networks based on the Ingenuity Pathways KnowledgeBase. Based on the 95 unique proteins identified byDIGE, three highly ranked networks were identified, andthe top ranked network incorporated 16 focus proteinsidentified by DIGE (Fig. 4A). This network identifiedperoxisome proliferator activated receptor � (PPAR�),CCAAT/enhancer binding protein � (C/EBP�), and ste-rol regulatory element binding proteins as the key signal-ing “nodes,” which interacted with many differentiallyexpressed proteins involved in urea cycle (carbamoyl-phosphate synthase, arginase-1, and argininosuccinatesynthase), fatty acid �-oxidation (peroxisomal multifunc-tional enzyme type 2 [HSD17B4] and fatty acid-bindingprotein, liver [FABpL]), ketogenesis (hydroxymethylglu-taryl-CoA synthase [HMGCS2]), and lipid transport (ad-

ipose differentiation-related protein [ADRP]), and otherprocesses. In view of their central position in the network,the expression of PPAR� and C/EBP� were confirmed bysemiquantitative western analysis. PPAR� belongs to theperoxisome proliferators activated receptor superfamilyand regulates the transcription of many genes encodingenzymes involved in fatty acid �-oxidation. Its expressionlevel was decreased at all the three stages of NAFLD (Fig.4B, 4C), in accordance with the decreased expression ofenzymes involved in fatty acid �-oxidation in our results.The abundance of C/EBP� was increased at the early twostages of NAFLD and decreased at the latest stage (Fig.4B, 4D).

ECHS1 Level in Patients with Simple Steatosis.One of the differentially expressed proteins, ECHS1, was

Fig. 2. Differentially expressed protein spots displayed in two-dimensional DIGE images. Representative two-dimensional DIGE images from thecontrol and HFD-treated rats for 4 weeks (A), 12 weeks (B), and 24 weeks (C) are presented. Numbered spots correspond to proteins exhibiting amodified expression level on HFD treatment. Spot numbers correspond to those in Supporting Table 2.

HEPATOLOGY, Vol. 51, No. 2, 2010 ZHANG ET AL. 5

further validated in patients with simple steatosis. ECHS1functions in the second step of the mitochondrial fattyacid beta-oxidation pathway. The relationship betweenECHS1 and NAFLD has never been reported. Immuno-histochemistry staining of ECHS1 was performed on par-affin sections that contained both steatotic and normalliver tissue from 20 patients with simple steatosis and 14controls. Results showed that the positive staining rate ofECHS1 was 75% to 100% in control tissues. In steatoticliver tissues, only a minor portion of hepatocytes (approx-imately 25%-50%) were positively stained (Table 2 andSupporting Fig. 2). The ECHS1 was stained in the cyto-plasm of hepatocyte (Fig. 5).

Silencing ECHS1 Expression Exacerbates Hepato-cellular Steatosis In Vitro Models. To determine therole of ECHS1 on hepatic steatosis, four pairs of siRNAduplexes were constructed for down-regulation of ECHS1in the mouse hepatocyte cell lines, the AML12 hepatocytes.Immunoblot analysis showed that the inhibitory efficiencyof the fourth pair of siRNAs was more significant than that ofthe other three pairs of siRNAs and they were therefore usedin the following experiment (Fig. 6A, B).

To determine whether knockdown of ECHS1 has a cru-cial role on lipid accumulation in hepatocytes, AML12 cellstransfected with ECHS1 siRNA or negative siRNA werecultured in full media with or without FFA mixtures. As

depicted in Fig. 6C, Oil Red O staining demonstrated thatincreased lipid stores were present in hepatocytes when cellswere treated with FFAs for 24 hours. The increases weremarkedly augmented in cells transfected with ECHS1siRNA. As expected, the siRNA knockdown of ECHS1 re-sulted in a significant 60% to 70% increase in cellular tri-glyceride (P � 0.05) (Fig. 6D). Because of the role ofECHS1 in fatty acid oxidation, the cellular FFAs levels wereassessed. Figure 6E showed a 30% to 40% increase in FFAlevels with ECHS1 siRNAs as compared with negativesiRNA after FFAs exposure (P � 0.05).

Silencing ECHS1 Expression Exacerbated HepaticSteatosis In Vivo Models. On the basis of in vitro re-sults, we next examined whether ECHS1 siRNA treat-ment also exacerbated HFD-induced hepatic steatosis inmice. We adopted a modified “hydrodynamic tail veininjection” to delivered synthetic siRNAs into mousehepatocytes in vivo. ECHS1 protein expression was mea-sured by immunoblotting after 72 hours of injection. Theresult showed that ECHS1 siRNA 4 reduced ECHS1protein expression by approximately 80% as comparedwith negative siRNA injection (Fig. 7A, B).

To evaluate further the effect of ECHS1 on HFD-induced hepatic steatosis, mice treated with ECHS1siRNA 4 or negative siRNA were fed with either controlor high-fat diet for 4 weeks. Because siRNA-mediated

Fig. 3. Western blot validation of nine se-lected proteins. Hepatic expression levels ofargininosuccinate synthase, carbamoyl-phos-phate synthase, FABPL, ECHS1, protein disul-fide isomerase (PDI), protein disulfideisomerase A3, 14-3-3 epsilon, cofilin 1, andmacrophage migration inhibitory factor in HFDrats and control rats. Liver extracts from a poolof six animals in each group were analyzed bywestern blotting using specific antibodies. Rel-ative expression levels were normalized to glyc-eraldehyde 3-phosphate dehydrogenase. Theresults correlated well with the quantification ofDIGE, which is also shown.

6 ZHANG ET AL. HEPATOLOGY, Month 2010

suppression of genes in vivo in mouse liver was main-tained for approximately 10 days,15,16 the siRNA injec-tion was repeated once per week for 4 consecutive weeksduring the diet. As shown in Fig. 7C, Oil Red O stainingdemonstrated that an HFD resulted in increased hepaticlipid stores in mice. Silencing ECHS1 expression bysiRNA further exacerbated steatotic changes in the liver(Fig. 7C). In accordance with morphological findings,hepatic triglyceride content increased by 35% in miceinjected with ECHS1 siRNAs than with negative siRNAsafter 4 weeks of HFD diet (P � 0.05) (Fig. 7D). Therewas also a significant increase in hepatic FFAs withECHS1 siRNAs (P � 0.05) (Fig. 7E).

DiscussionDespite the high prevalence and the potential to progress

to end-stage liver disease, the molecular mechanisms under-lying NAFLD initiation and progression remains poorly un-derstood. The current study systematically analyzed the liverproteomes during different stages of NAFLD in an HFD-induced rat models, which may help elucidate the mecha-nism involved in the progression of the disease.

Fig. 4. Pathway analysis and validation of key molecules. (A) The top ranked network of differentially expressed proteins based on Ingenuity PathwayAnalysis. Focus proteins were indicated by gray icons. As described in the legend provided, nodes are displayed using various shapes that represent thefunctional class of the gene product. Edges are displayed with various labels that describe the nature of the relationship between the nodes. The proteinsdirected in red were further validated by western blotting. (B, C, D) Hepatic expression levels of PPAR� and C/EBP� in HFD and control rats. Liver extractsfrom a pool of six animals in each group were analyzed by western blotting using specific antibodies. Relative expression levels were normalized toglyceraldehyde 3-phosphate dehydrogenase. Representative blots from three independent experiments are shown. *P � 0.05, **P � 0.01.

Table 2. Immunohistochemistry Results for ECHS1 inPatients with Simple Steatosis and Controls

� � �� ��� ���� Total

Control 0 0 0 5 9 14NAFLD 0 8 7 3 2 20

Specimens with 0, �25, �50, �75, and �100% positively stained hepato-cytes were scored as –, �, ��, ���, and ����, respectively.

Fig. 5. Immunohistochemistry results of ECHS1. Liver sections fromsimple steatosis patients and controls were used for immunohistochem-istry staining. The positive staining rate of ECHS1 was 75% to 100% incontrol tissues (A). In steatotic liver tissues, only a minor portion ofhepatocytes (approximately 25%-50%) were positively stained (B). TheECHS1 was stained in the cytoplasm of the hepatocyte.

HEPATOLOGY, Vol. 51, No. 2, 2010 ZHANG ET AL. 7

Ninety-five unique proteins associated with the devel-opment of NAFLD were identified through a compara-tive proteomics approach. Several of the proteins,including ATP5B, FABPL, and prohibitin, have alreadybeen indicated in published studies addressing proteinexpressions in hepatic steatosis or NASH.17-19 Neverthe-less, functional annotation of the identified proteins indi-cated phase-specific biological characteristics duringdifferent stages of NAFLD, which were not reported in pre-vious proteomic studies. We found that enzymes involved infatty acid oxidation (ECHS1, short chain 3-hydroxyacyl-CoA dehydrogenase [HADH], HSD17B4, and FABPL)and mitochondrial respiration chain (ubiquinol-cytochromec reductase core protein [IUQCRC1], cytochrome c oxidasepolypeptide Va [COX5A], electron-transfer-flavoprotein,beta polypeptide [ETFB], and ATP synthase beta chain[ATP5B]) were down-regulated during simple steatosisand nonspecific inflammation stage, respectively, whereascytoskeleton proteins were up-regulated in NASH. Thesefindings indicated that impairment of fatty acid oxidationmay contribute to the initiation of hepatic steatosis,whereas mitochondrial dysfunction and cytoskeleton dis-turbance were associated with disease progression. Thefindings also provide a molecular explanation for the role

of oxidative stress and mitochondrial dysfunction in thepathogenesis of NASH.20,21 Those cytoskeleton proteinswith increased expression level in the last stage wouldbe potential markers of NASH. One of these proteins,cytokeratin 18, has already been considered as a novelbiomarker of NASH. The level of caspase 3–generatedcytokeratin 18 fragment was reported to be correlatedwith disease severity in NAFLD patients.22

IPA showed that several identified proteins involved inlipid metabolism were regulated by PPAR�, a transcrip-tional activator of fatty acid oxidation genes.23,24 In ourstudy, a reduced protein level of PPAR� was found inrat models of NAFLD, in accordance with the down-regulation of its downstream genes involved in fatty acidoxidation (HSD17B4 and FABPL). These findings dem-onstrated that impairment of fatty acid �-oxidation oc-curred in our model, which may contribute to thepathogenesis of NAFLD. Another transcription factoridentified by IPA was C/EBP�, which is generally consid-ered a master regulator of adipose tissue development.25

More recently, the C/EBP family has emerged as an im-portant group of regulators of hepatic lipogenesis.26,27

The up-regulation of C/EBP� and its target gene ADRPwere detected in the current study, indicating an enhance-

Fig. 6. SiRNA-mediated knockdown of ECHS1 increased lipid accumulation in AML12 cells. (A) ECHS1 siRNA markedly reduced the ECHS1 proteinlevel in AML12 cells. (B) Relative protein level of ECHS1 after siRNA transfection. The ECHS1-s4 has the most significant inhibitory efficiency. (C)Oil Red O staining of cells treated with or without FFAs after transfection of ECHS1-s4 or negative siRNA. Original magnification, 200�. (D) Influenceof ECHS1 knockdown on intracellular triglyceride and FFA (E) level induced by FFA overload. Representative results from three independentexperiments are shown. Data are mean � standard deviation in B, D, and E. *P � 0.05 versus negative siRNA-transfected cells. **P � 0.01 versusnegative siRNA-transfected cells.

8 ZHANG ET AL. HEPATOLOGY, Month 2010

ment of lipogenesis in response to HFD treatment. Takentogether, the underlying cause of fat accumulation inNAFLD is mostly attributable to the inhibition of fattyacid oxidation and enhancement of lipid synthesis.

These results emphasized the power of proteomic ap-proaches in providing high-throughput information toelucidate the complex pathogenesis of NAFLD. To give amore profound insight into the mechanism of NAFLD,one of the identified proteins, ECHS1, underwent furtherinvestigation. ECHS1 catalyzes the second step in thephysiologically important �-oxidation pathway of fattyacid metabolism. Oxidation of fatty acids occurs in threesubcellular organelles, with �-oxidation confined to mi-tochondria and peroxisomes and �-oxidation occurringin the endoplasmic reticulum.23,24 Although impairmentof mitochondrial �-oxidation of fatty acid was consideredto account for excess lipid storage in several animal mod-els of hepatic steatosis,28-30 the expression level and thepotential role of ECHS1 in the pathogenesis of NAFLDhave not been fully elucidated. In the current study, wefound that the expression of ECHS1 protein was signifi-cantly down-regulated in rats fed with HFD. The resultswere further confirmed in patients with simple steatosis.Moreover, siRNA-mediated knockdown of ECHS1 sig-nificantly exacerbates lipid accumulation in hepatocytes

in in vitro and in vivo models of NAFLD. The increases ofFFA levels by ECHS1 silencing were in parallel with thatof TG levels. These findings indicated that ECHS1down-regulation contributed to the defect of mitochon-drial fatty acid �-oxidation, which subsequently leads totriglyceride storage in the liver. However, whether de-creased mitochondrial fatty acid oxidation induced byECHS1 down-regulation is regulated by PPAR� orwhether it plays a primary role in causing the insulinresistance,28,31remains to be determined.

One limitation of the current study is that the HFDdietary animal model used here did not manifest obesity,high serum TG level, or insulin resistance, although thismodel developed hepatic steatosis and steatohepatitis thatwere morphologically similar to human NAFLD. Manypublished studies used HFD-induced NAFLD modelsthat differed in phenotypes. Some manifested obesity,9,32

and others did not.33,34 The reasons are implicated. Thespecies, strain, or sex of the animals, the amount and typeof dietary fat, and the diet duration are all important ininfluencing vulnerability to dietary effects. Because fewanimal models can entirely reflect the natural course andthe causative background of human NAFLD, it would bebetter to validate differentially expressed proteins ob-tained from animal models in human samples. The 2D-

Fig. 7. Silencing ECHS1 expression in vivo exacerbated hepatic steatosis induced by the HFD. (A) Immunoblot of liver lysate obtained from micetreated with modified ECHS1-s4 or negative siRNA 72 hours after injection. (B) Relative protein level of ECHS1. ECHS1-s4 significantly reduced ECHS1protein level in mice livers. (C) Oil Red O staining of liver sections obtained from mice treated with modified ECHS1-s4 or negative siRNA after chowor HFD for 4 weeks. Original magnification, 200�. (D) Influence of ECHS1 knockdown on hepatic TG and FFA (E) content in mice treated with HFD.Data are mean � standard deviation in B, D, and E; n � 5, *P � 0.05 versus negative siRNA-treated mice. **P � 0.01 versus negativesiRNA-treated mice.

HEPATOLOGY, Vol. 51, No. 2, 2010 ZHANG ET AL. 9

DIGE method we used in the current study also has itsown shortcomings. As a two-dimensional technique, itdoes not generally permit the resolution of proteins withhigh (�150 kDa) or low (�10 kDa) molecular masses, orvery basic or hydrophobic proteins, thus limiting the pro-teomic coverage of most biological sample types.

In conclusion, we comprehensively studied the nonalco-holic fatty liver proteome in rat models by quantitative globalprotein profiling using DIGE technology. We identifiedproteins that were differentially expressed during variousstages of NAFLD. Analysis of biological processes and regu-latory networks provided both global and specific informa-tion regarding the molecular events that cause this complexdisease. The impairment of liver fatty acid oxidation, as wellas mitochondrial dysfunction, and the changes of cytoskele-ton organization provide a molecular feature for the patho-genesis and progression of NAFLD. More importantly, thedown-regulation of ECHS1 was found to be associated withhepaticsteatosisafterexposuretoanHFD.However, theprecisemechanism by which ECHS1 promotes hepatic steatosis re-mains tobe elucidated, and further investigations areunderway.

References1. Bellentani S, Saccoccio G, Masutti F, Croce LS, Brandi G, Sasso F, et al.

Prevalence of and risk factors for hepatic steatosis in Northern Italy. AnnIntern Med 2000;132:112-117.

2. Nomura H, Kashiwagi S, Hayashi J, Kajiyama W, Tani S, Goto M. Prev-alence of fatty liver in a general population of Okinawa, Japan. Jpn J Med1988;27:142-149.

3. Fan JG, Zhu J, Li XJ, Chen L, Li L, Dai F, et al. Prevalence of and risk factors forfatty liver inageneralpopulationofShanghai,China. JHepatol2005;43:508-514.

4. Adams LA, Lymp JF, St Sauver J, Sanderson SO, Lindor KD, Feldstein A,et al. The natural history of nonalcoholic fatty liver disease: a population-based cohort study. Gastroenterology 2005;129:113-121.

5. Matteoni CA, Younossi ZM, Gramlich T, Boparai N, Liu YC, McCul-lough AJ. Nonalcoholic fatty liver disease: a spectrum of clinical and patho-logical severity. Gastroenterology 1999;116:1413-1419.

6. Day CP, James OF. Steatohepatitis: a tale of two “hits”? Gastroenterology1998;114: 842-845.

7. Mezey E. Dietary fat and alcoholic liver disease. HEPATOLOGY 1998;28:901-905.

8. Musso G, Gambino R, De Michieli F, Cassader M, Rizzetto M, Durazzo M,et al. Dietary habits and their relations to insulin resistance and postprandiallipemia in nonalcoholic steatohepatitis. HEPATOLOGY 2003;37:909-916.

9. Lieber CS, Leo MA, Mak KM, Xu Y, Cao Q, Ren C, et al. Model ofnonalcoholic steatohepatitis. Am J Clin Nutr 2004;79:502-509.

10. Hanash S. Disease proteomics. Nature 2003;422:226-232.11. Unlu M, Morgan ME, Minden JS. Difference gel electrophoresis: a single

gel method for detecting changes in protein extracts. Electrophoresis 1997;18:2071-2077.

12. Sun W, Xing B, Sun Y, Du X, Lu M, Hao C, et al. Proteome analysis ofhepatocellular carcinoma by two-dimensional difference gel electrophore-sis: novel protein markers in hepatocellular carcinoma tissues. Mol CellProteomics 2007;6:1798-1808.

13. Zhang X, Guo Y, Song Y, Sun W, Yu C, Zhao X, et al. Proteomic analysisof individual variation in normal livers of human beings using differencegel electrophoresis. Proteomics 2006;6:5260-5268.

14. Gomez-Lechon MJ, Donato MT, Martınez-Romero A, Jimenez N, CastellJV, O’Connor JE. A human hepatocellular in vitro model to investigatesteatosis. Chem Biol Interact 2007;165:106-116.

15. Song E, Lee SK, Wang J, Ince N, Ouyang N, Min J, et al. RNA interferencetargeting Fas protects mice from fulminant hepatitis. Nat Med 2003;9:347-351.

16. Lewis DL, Hagstrom JE, Loomis AG, Wolff JA, Herweijer H. Efficientdelivery of siRNA for inhibition of gene expression in postnatal mice. NatGenet 2002;32:107-108.

17. Santamaria E, Avila MA, Latasa MU, Rubio A, Martin-Duce A, Lu SC, etal. Functional proteomics of nonalcoholic steatohepatitis: mitochondrialproteins as targets of S-adenosylmethionine. Proc Natl Acad Sci U S A2003;100:3065-3070.

18. Van Greevenbroek MM, Vermeulen VM, De Bruin TW. Identification ofnovel molecular candidates for fatty liver in the hyperlipidemic mousemodel, HcB19. J Lipid Res 2004;45:1148-1154.

19. Charlton M, Viker K, Krishnan A, Sanderson S, Veldt B, Kaalsbeek AJ, etal. Differential expression of lumican and fatty acid binding protein-1: newinsights into the histologic spectrum of nonalcoholic fatty liver. disease.HEPATOLOGY 2009;49:1375-1384.

20. Begriche K, Igoudjil A, Pessayre D, Fromenty B. Mitochondrial dysfunc-tion in NASH: causes, consequences and possible means to prevent it.Mitochondrion. 2006;6:1-28.

21. Neuschwander-Tetri BA, Caldwell SH. Nonalcoholic steatohepatitis:summary of an AASLD Single Topic Conference. HEPATOLOGY 2003;37:1202-1219.

22. Wieckowska A, Zein NN, Yerian LM, Lopez AR, McCullough AJ, FeldsteinAE. In vivo assessment of liver cell apoptosis as a novel biomarker of diseaseseverity in nonalcoholic fatty liver disease. HEPATOLOGY 2006;44:27-33.

23. Rao MS, Reddy JK. PPARalpha in the pathogenesis of fatty liver disease.HEPATOLOGY 2004;40:783-786.

24. Reddy JK, Hashimoto T. Peroxisomal beta-oxidation and peroxisome pro-liferator-activated receptor alpha: an adaptive metabolic. Annu Rev Nutr.2001;21:193-230.

25. Linhart HG, Ishimura-Oka K, DeMayo F, Kibe T, Repka D, PoindexterB, et al. C/EBPalpha is required for differentiation of white, but notbrown, adipose tissue. Proc Natl Acad Sci U S A 2001;98:12532-12537.

26. Matsusue K, Gavrilova O, Lambert G, Brewer HB Jr, Ward JM, Inoue Y,et al. Hepatic CCAAT/enhancer binding protein alpha mediates inductionof lipogenesis and regulation of glucose homeostasis in leptin-deficientmice. Mol Endocrinol 2002;18:2751-2764.

27. Qiao L, MacLean PS, You H, Schaack J, Shao J. Knocking down liverccaat/enhancer-binding protein alpha by adenovirus-transduced silent in-terfering ribonucleic acid improves hepatic gluconeogenesis and lipid ho-meostasis in db/db mice. Endocrinology 2006;147:3060-3069.

28. Zhang D, Liu ZX, Choi CS, Tian L, Kibbey R, Dong J, et al. Mitochon-drial dysfunction due to long-chain Acyl-CoA dehydrogenase deficiencycauses hepatic steatosis and hepatic insulin resistance. Proc Natl Acad SciU S A 2007;104:17075-17080.

29. Tolwani RJ, Hamm DA, Tian L, Sharer JD, Vockley J, Rinaldo P, et al.Medium-chain acyl-CoA dehydrogenase deficiency in gene-targeted mice.PLoS Genet 2005;1:e23.

30. Ibdah JA, Perlegas P, Zhao Y, Angdisen J, Borgerink H, Shadoan MK, et al. Miceheterozygous for a defect in mitochondrial trifunctional protein develop hepaticsteatosis and insulin resistance. Gastroenterology 2005;128:1381-1390.

31. Shulman GI. Cellular mechanisms of insulin resistance. J Clin Invest 2000;106:171-176.

32. Svegliati-Baroni G, Candelaresi C, Saccomanno S, Ferretti G, Bachetti T, Marzi-oni M, et al. A model of insulin resistance and nonalcoholic steatohepatitis in rats:role of peroxisome proliferator-activated receptor-alpha and n-3 polyunsaturatedfatty acid treatment on liver injury. Am J Pathol 2006;169:846-860.

33. Safwat GM, Pisano S, D’Amore E, Borioni G, Napolitano M, Kamal AA,et al. Induction of non-alcoholic fatty liver disease and insulin resistance byfeeding a high-fat diet in rats: does coenzyme Q monomethyl ether have amodulatory effect? Nutrition 2009;25:1157-1168.

34. OmagariK,KatoS,TsuneyamaK,InoharaC,KurodaY,TsukudaH,etal.Effectsof a long-term high-fat diet and switching from a high-fat to low-fat, standard dieton hepatic fat accumulation in Sprague-Dawley rats. Dig Dis Sci 2008;53:3206-3212.

10 ZHANG ET AL. HEPATOLOGY, Month 2010