Liver Protein Profiling in Chronic Hepatitis C - American Chemical

Nov 21, 2011 - 'INTRODUCTION. Hepatitis C virus (HCV) is one of the leading causes of chronic liver diseases in Western countries.1 The World Health ...
0 downloads 0 Views 3MB Size
ARTICLE pubs.acs.org/jpr

Liver Protein Profiling in Chronic Hepatitis C: Identification of Potential Predictive Markers for Interferon Therapy Outcome Ariel Basulto Perdomo,† Fabiola Ciccosanti,† Oreste Lo Iacono,‡ Claudio Angeletti,† Marco Corazzari,† Nicola Daniele,† Angela Testa,† Roberto Pisa,§ Giuseppe Ippolito,† Giorgio Antonucci,† Gian Maria Fimia,*,†,^ and Mauro Piacentini*,†,||,^ †

National Institute for Infectious Diseases IRCCS ‘L. Spallanzani’, Rome, Italy Gastroenterology Unit, Hospital del Tajo Aranjuez, Madrid, Spain § Anatomy Pathology Service, Azienda Ospedaliera S. Camillo-Forlanini, Rome, Italy Department of Biology, University of Rome ‘Tor Vergata’, Rome, Italy

)



bS Supporting Information ABSTRACT: The current anti-hepatitis C virus (HCV) therapy, based on pegylated-interferon alpha and ribavirin, has limited success rate and is accompanied by several side effects. The aim of this study was to identify protein profiles in pretreatment liver biopsies of HCV patients correlating with the outcome of antiviral therapy. Cytosolic or membrane/organelle-enriched protein extracts from liver biopsies of eight HCV patients were analyzed by two-dimensional fluorescence difference gel electrophoresis and mass spectrometry. Overall, this analysis identified 21 proteins whose expression levels correlate with therapy response. These factors are involved in interferon-mediated antiviral activity, stress response, and energy metabolism. Moreover, we found that post-translational modifications of dihydroxyacetone kinase were also associated with therapy outcome. Differential expression of the five best performing markers (STAT1, Mx1, DD4, DAK, and PD-ECGF) was confirmed by immunoblotting assays in an independent group of HCV patients. Finally, we showed that a prediction model based on the expression levels of these markers classifies responder and nonresponder patients with an accuracy of 85.7%. These results provide evidence that the analysis of pretreatment liver protein profiles is valuable for discriminating between responder and nonresponder HCV patients, and may contribute to reduce the number of nonresponder patients exposed to therapy-associated risks. KEYWORDS: HCV, liver biopsy, 2-D DIGE, antiviral therapy, prognostic markers

’ INTRODUCTION Hepatitis C virus (HCV) is one of the leading causes of chronic liver diseases in Western countries.1 The World Health Organization estimates that 3% of the world’s population is chronically infected with HCV.2 Chronic HCV infection leads to a wide spectrum of liver diseases ranging from mild chronic hepatitis to end-stage cirrhosis and hepatocellular carcinoma.3 The current treatment of choice for chronic hepatitis C is based on a combination of pegylated alpha-interferon (PEG-IFN) and ribavirin (RBV). However, the rate of therapeutic success is far from being optimal with approximately 50% of HCV genotype 1-infected patients achieving a sustained virological response (SVR).4 6 Several viral and host factors have been correlated to a successful therapy outcome, such as pretreatment viral load, age, gender, race, and body weight,7,8 but their accuracy in predicting the virological response is fairly poor and cannot be used to deny the therapy. To date, only viral kinetics after few weeks of therapy is considered clinically relevant to decide whether or not to continue the therapy in individual patients.8 r 2011 American Chemical Society

Recently, genomic and postgenomic studies have started to shed light on the important contribution of specific host factors to the outcome of the antiviral response. Polymorphisms in interleukin-28B, a gene encoding the interferon-lambda-3, have been associated with both spontaneous and treatment-induced HCV clearance.9 12 Moreover, DNA microarray analyses of liver biopsies from HCV patients have revealed the prognostic potential of interferon-related genes, with patients not responding to therapy (NR) having higher basal expression compared to those with SVR.13 16 These results have emphasized the importance of high-throughput analyses for the identification of predictive markers. However, a complete view of liver alterations in HCV patients also requires an analysis of the protein profiles in order to assess changes at post-transcriptional level, such as protein stability or post-translational modifications, which may provide an additional source of prognostic markers. To this end, we Received: July 9, 2011 Published: November 21, 2011 717

dx.doi.org/10.1021/pr2006445 | J. Proteome Res. 2012, 11, 717–727

Journal of Proteome Research

ARTICLE

applied a proteomic approach to identify protein alterations associated with therapy responsiveness in pretreatment liver biopsies from SVR and NR patients.

OxPhos complex II, histone H4, and tubulin (Supporting Information Figure S1). This fractionation procedure allowed the recovery of an adequate amount of material for proteomic analysis only for the cytosolic and the membrane liver fractions. Proteins were precipitated using the 2-D cleanup kit (GE-Healthcare) and resuspended in sample buffer (7 M urea, 2 M thiourea, 2% CHAPS, 1% sulfobetaine 3 10, 1% amidosulfobetaine 14, and 10 mM Tris-HCL pH 8.5). Liver biopsies used for immunoblotting were homogenized in CelLytic Mammalian Tissue extraction reagent (Sigma-Aldrich).

’ MATERIALS AND METHODS HCV Patients

A retrospective review of medical records of HCV patients treated with PEG-IFN+RBV attending one of the outpatients unit of the INMI between 2004 and 2007 was performed to select patients’ biopsies analyzed in this study. Liver biopsies were obtained from INMI Biobank collection where they were stored for research purpose upon the approval of the local ethical committee. Patients were treated in compliance with the ethical guidelines of the 1975 Declaration of Helsinki and informed consent was obtained from each patient before biopsy. Diagnosis of chronic HCV hepatitis were based on serological and qualitative RT-PCR assays, as well as elevated levels of alanine aminotransferase. Inclusion criteria for this study were as follows: to be na€ive to antiviral treatment; to be negative to HIVantibodies and hepatitis B surface antigen. Liver biopsies were performed within 1 year before treatment. Baseline data collected for each patient included demographics, clinical and laboratory data, use of alcohol within the 6 months preceding the initiation of therapy, type of peginterferon α (-2a/-2b), dosage of ribavirin (Supporting Information Table S1). SVR to antiviral treatment was defined as an undetectable serum HCV RNA measured with a qualitative method at month 6 after the completion of therapy. Patients who failed to clear HCV RNA from serum after 24 weeks of therapy were considered as NR.8 The Early Virologic Response (EVR) was also defined according to guidelines.8 Patients infected with HCV genotype 1 who did not achieve an EVR discontinued treatment and were considered as NR. Fifty-six adult patients, 31 classified as SVR and 25 as NR based on treatment outcome, were selected for this study and divided into four groups. Eight patients, HCV genotype 1, participated to the two-dimensional difference gel electrophoresis (2-D DIGE) analysis (2-D DIGE group). Twenty-three consecutive patients starting treatment in 2006 were included in the second study group (Validation and Training Group). Their liver specimens were used to validate differentially expressed proteins between SVR and NR patients by immunoblotting analysis. The data set resulting from this analysis was used as a training data set for building a prediction model to classify a third group of 14 patients infected exclusively with genotype 1 (Testing Group). Moreover, 11 consecutive patients were included in another study group in order to validate proteomic results by Real-Time PCR (Real-Time PCR Validation Group Patients).

2-D DIGE Analysis

Forty micrograms of cytosolic proteins and 30 μg of proteins from membrane fraction of 4 SVR and 4 NR patients were labeled with 150 and 120 pmol, respectively, of the N-hydroxy succinimidyl ester derivatives of the cyanine dyes Cy3 and Cy5 accordingly to manufacturer’s instructions (GE-Healthcare). Difference in the amount of labeled proteins between cytosol and membrane was due to the different protein yields from fractionation of liver biopsies. The same amount of internal standard was labeled with Cy2. Differentially labeled samples were mixed together, supplemented with equal amount of sample buffer, containing DTT (20 mg/mL) and ampholines, pH 3 10 (2% (v/v). A dye-swapping scheme was used to ensure that samples from both groups were labeled either with Cy3 or Cy5 and each gel contained one sample from each group. Proteins were subjected to IEF using nonlinear 3 10 pH range dry strips on a IPGphor II (GE-Healthcare) as described.19 Second dimension electrophoresis were performed on 12% polyacrylamide gels at 8 mA/gel for 16 h in Tris-glycine SDS-PAGE running buffer (Bio-Rad). Image Analysis

The Cy2, Cy3 and Cy5-labeled images were acquired on a Typhoon 9410 scanner (GE-Healthcare). Gels were then fixed in 10% (v/v) methanol and 7% (v/v) acetic acid and stained with SyproRuby (Bio-Rad). Images were imported in DeCyder v6.5, 2-D Differential Analysis software (GE-Healthcare). The DeCyder Differential In-gel Analysis (DIA) software was used for spot codetection and normalized volume ratio calculation. DeCyder Biological Variation Analysis (BVA) software was then used to simultaneously match all spot maps from 2-D gels and to identify spots with statistical variations using the Student’s t test. The Extended Data Analysis (EDA) module was used for Principal Component Analysis (PCA) to assess underlying sources of variation within DIGE data set. To assess the capacity of identified spots in separating SVR and NR patients, unsupervised hierarchical clustering was also performed. MALDI TOF/TOF Analysis

Gels were stained with Sypro Ruby according to manufacturer’s instructions and selected spots were excised from the gels using a software-driven ProPic Spot picker (Genomic Solutions) and subjected to in-gel tryptic digestion. Briefly, gel pieces were washed twice in 100 mM ammonium bicarbonate and 50% (v/v) ACN, dehydrated by incubation in 100% (v/v) ACN for 5 min and rehydrated in 50 mM ammonium bicarbonate containing 4 ng/uL of trypsin. Rehydration was allowed for 30 min, the excess of trypsin was discharged, and 50 mM ammonium bicarbonate was added following digestion overnight at 37 C. Peptides in solution were conserved at 20 C until analysis. Tryptic peptides were concentrated with ZipTip mC18 pipet tips (Millipore) and coeluted onto the MALDI target in 1 μL of α-cyano-4-hydroxycinnamic acid matrix (5 mg/mL in 50% ACN,

Liver Biopsies

Liver biopsies were frozen within 1 h after collection and stored at 80 C. A sample of the biopsy tissue was placed in formalin for a histological evaluation. Liver biopsies included in the proteomic analysis were accurately selected in order to have the most similar histological characteristics. Liver fibrosis scores were evaluated accordingly to Ishak et al.17,18 Liver Fractionation and Sample Preparation

Liver biopsies were fractionated using the ProteoExtract Subcellular Proteome Extraction Kit (Calbiochem) in order to obtain extracts enriched in cytosolic, membrane/organelle, nuclear and cytoskeletal proteins. Fractionation efficiency was assessed by immunoblotting using antibodies to GAPDH, mitochondrial 718

dx.doi.org/10.1021/pr2006445 |J. Proteome Res. 2012, 11, 717–727

Journal of Proteome Research

ARTICLE

0.1% TFA). MALDI-MS and MALDI-MS/MS were performed on an Applied Biosystems 4700 proteomics analyzer with TOF/ TOF ion optics. The spectra were acquired in the positive reflector mode by 40 subspectral accumulations (each consisting of 50 laser shots) in a 800 4000 mass range, focus mass 2100 Da, using a 355 nm Nb:YAG laser with a 20 kV acceleration voltage. Peak labeling was automatically done by using a 4000 series explorer software version 3.0 (Applied Biosystems) without any kind of smoothing of peaks or baseline, considering only peaks that exceeded a S/N of 10 (local noise window 200 m/z) and a half maximal width of 2.9 bins. Calibration was performed using peptides resulting from autoproteolysis of trypsin (m/z 842.510, 1045.564, 2211.105, 2239.136, and 2807.300). In addition to peptide mass finger spectra, the five most abundant precursor ions masses having a S/N higher than 50 were chosen for MS/ MS fragmentation. MS/MS spectra were integrated over 1500 laser shots in the 1 kV positive ion mode with the metastable suppressor turned on. Air at the medium gas pressure setting (1.2561026 Torr) was used as collision gas in the CID off mode. The interpretation of both the MS and MS/MS data were carried out with the GPS Explorer software (version 3.6, Applied Biosystems) which acts as an interface between the Oracle database containing raw spectra and a local copy of the MASCOT search engine (Version 2.1, Matrix Science). Peaks were extracted from raw spectra by the GPS software using the following setting: MS peak filtering, mass range: 800 4000 Da; minimum S/N: 10; peak density filter: 50 peaks per 200 Da; maximum number of peaks: 65. MS/MS peak filtering, mass range: 60 20 Da below precursor mass; minimum S/N: 8; peak density filter: 20 peaks per 200 Da; maximum number of peaks: 65. An exclusion list of known contaminant ion masses of keratin and trypsin (842.510, 906.505, 917.300, 947.500, 1045.564, 1794.810, 2211.105, 2239.136, 2283.181, 2284.184, 2300.178) was used. A combined MS peptide fingerprint and MS/MS peptide sequencing search was performed against the NCBI nonredundant database (number of protein sequences: 2 095 284; number of amino acid residues: 703 926 160; released on 2007-1-27) without taxonomy restriction using the MASCOT search algorithm. These searches specified trypsin as the digestion enzyme, carbamidomethylation of cysteine as fixed modification, partial oxidation of methionine, and phosphorylation of serine, threonine, and tyrosine as variable modifications, and allowed for one missed trypsin cleavage. The monoisotopic precursor ion tolerance was set to 30 50 ppm and the MS/MS ion tolerance to 0.3 Da. MS/MS peptide spectra with a minimum ion score confidence interval greater than 95% were accepted; this was equivalent to a median ion score cut off of approximately 35 in the data set. Protein identifications were accepted with an statistically significant MASCOT protein search score greater than 76 that corresponded to an error probability of p < 0.05 in our data set.

system (Invitrogen). Proteins were electrotransferred onto nitrocellulose or PVDF membranes for 60 min at 25 V in NuPAGE transfer buffer using a Trans-Blot SD Semi-Dry transfer system (Bio-Rad). Membranes were incubated with primary and secondary antibodies as described.19 Antibodies were revealed using ECL Plus Detection Reagents (GE-Healthcare). For normalization, membranes were reprobed with anti-GAPDH antibodies. Intensity of bands from individual patients was quantified using Quantity One software (BioRad) and normalized with the respective GAPDH intensity. The following antibodies were used: anti-Mx1 (sc-34128, Santa-Cruz Biotech.), anti-β-enolase (sc-100811, Santa-Cruz Biotech.), anti-STAT1 (9172, Cell Signaling Tecnologies), antiGAPDH (Calbiochem), anti-ALDH1L1 (sc-100497, Santa-Cruz Biotech.), anti-vimentin (sc-6260, Santa-Cruz Biotech.), antiPD-ECGF (sc-47702, Santa-Cruz Biotech.), anti-DD4 (sc-100526, Santa-Cruz Biotech.), anti-FTCD (sc-53128, Santa-Cruz Biotech.), anti-DAK (H00026007-B01P, Abnova), anti-ABAT (H000000018B01, Abnova), anti-GAPDH (Sigma-Aldrich), anti-mitochondrial OxPhos complex II (Invitrogen), anti-histone H4 (Santa Cruz), and anti-tubulin (Sigma-Aldrich). Anti-mouse, anti-rabbit, and anti-goat peroxidase conjugated antibodies were from Jackson ImmunoResearch. Real-Time PCR

RNA levels were analyzed by real-time PCR as described.19 AKR1C4 forward: ACACAGTGGATCTCTCTGCCACAT; AKR1C4 reverse: AGCTGCCTGCAGT TGAAGTTTGAC; DAK forward: AGCCCAGGAGCTGATCTGTTACAA; DAK reverse: TCCGGCTCCAGCTTCCATATTCTT; MX1 forward: TAATGTGGACATCGCCACCACAGA; MX1 reverse: TCC TTTGTCCACCAGATCAGGCTT; PD-ECGF forward: GGA TTCAATGTCATCCAGAGCCCA; PD-ECGF reverse: TGAG AATGGAG GCTGTGATGAGTG; STAT1 forward: CAATT GTGCCAGCCTGGTTTGGTA; STAT1 reverse: TGCACA TGGTGGAGTCAGGAAGAA. L34 mRNA levels were used as an internal controls. Statistical Analysis

Patients’ clinical data were compared using the Student’s t test for continuous variables, while categorical data were compared using the Fisher’s exact test. Expression level of proteins in liver biopsies of SVR and NR patients analyzed by immunoblotting were compared by nonparametric Mann Whitney test. Pearson correlation coefficient was used to calculate the association between therapy outcome and the level of proteins analyzed by immunoblotting either individually or with a linear combination of the n 1 different Y measures (Yc) chosen such that the correlation between the Y1 variable and the combined Yc variable is maximal. Discriminant analysis was used to categorize HCV patients into 2 groups, SVR and NR, based on immunoblotting data. Classification was performed with a minimum allowed tolerance of 0.001. Discriminant functions (Z) computed in the training group were used to predict the therapy outcome in the testing group. Discrimination capacity of the discriminant functions was evaluated in the training and testing data sets by analyzing both the proportion of correctly classified subjects and the c-index, that is, the area under the Receiver Operating Characteristic curve (AUC). AUC ranges from zero to one and measures the capacity of the model to predict the outcome of interest. Probabilities below 1.5, p-value < 0.05) are indicated. Red: proteins with increased levels in NR. White: proteins with increased levels in SVR. MW, molecular weight; kDa, kilodaltons. (B E) Representative results of protein spots differentially expressed between SVR and NR. Bidimensional images from one SVR and one NR patient are accompanied by the graphs indicating the changes in protein levels in all analyzed HCV patients.

electrophoresis (2-D DIGE)20,21 using protein extracts enriched in cytosolic or membrane/organelle proteins. From a total of 3309 and 2332 spots detected respectively in cytosolic and membrane/organelle 2-D maps, 24 in the first and 10 in the second fraction showed a significant difference between SVR and NR patients (Figure 1A, Supporting Information Figures S1, S2; see average ratio and t test values in Table 1). Interestingly, most of the observed changes correspond to proteins with increased levels in the liver of NR patients (Table 1, see also Figure 1B E). In the cytosolic fraction, PCA based on differentially expressed spots revealed a first source of variability (PC1) that explains 74.5% of variance and separated samples into two groups: SVR and NR (Figure 2A). In the membrane/organelle fraction, PCA

Version 1.8 released on December 31, 2007) and with Stata, Release 10 (Stata Corp, College Station, Tex).

’ RESULTS Identification of Differentially Expressed Proteins between Responder and Nonresponder Patients by 2-D DIGE

The proteomic analysis was performed on liver biopsies from 4 SVR and 4 NR patients infected with HCV genotype 1. Patients were well balanced for most demographic, virologic, and clinical characteristics with the exception of age (p < 0.05) (Supporting Information Table S1; 2-D DIGE group). Two independent screenings were carried out by two-dimensional difference gel 720

dx.doi.org/10.1021/pr2006445 |J. Proteome Res. 2012, 11, 717–727

Journal of Proteome Research

ARTICLE

Table 1. Proteins Differentially Expressed in Liver Biopsies of SVR and NR HCV patients average

master no.a

ratiob

t test (p)

symbols

protein name

protein

protein

peptide

sequence

GIc

scored

countse

coverage

proteins differentially expressed in cytosolic fraction 1457

0.0127

1.82

ABAT

4-aminobutyrate aminotransferase

4261876

249

18

47%

773

0.0381

1.67

ACON

aconitase precursor

3600098

170

10

16%

1482

0.0181

1.69

AGAT

L-arginine:

791049

404

22

56%

601 1728

0.0170 0.0440

2.26 1.67

ALDH1L1 (FTHFD) ALDOB

10-formyl tetrahydrofolate dehydrogenase chain R. fructose 1.6-bisphosphate Aldolase

21614513 11514046

488 244

29 11

30% 36%

1114

0.0122

1.56

DAK

dihydroxyacetone kinase/FAD-AMP lyase

20149621

167

10

16%

1115

0.0076

1.49

DAK

dihydroxyacetone kinase/FAD-AMP lyase

20149621

92

5

14%

1866

0.0125

1.57

DD4/AKR1C4

3 alpha-hydroxysteroid/dihydrodiol

556518

323

13

52%

556518

128

9

32%

glycine amidinotransferase

dehydrogenase 4 1870

0.0012

1.62

DD4/AKR1C4

3 alpha-hydroxysteroid/dihydrodiol dehydrogenase 4

1457 1162

0.0127 0.0079

1.82 1.74

ENO3 FTCD

enolase 3 formimidoyltransferase-cyclodeaminase

16878083 46255035

216 566

14 27

38% 48% 45%

1168

0.0122

1.79

FTCD

formimidoyltransferase-cyclodeaminase

46255035

562

25

578

0.0305

1.65

GANAB

glucosidase II

2274968

309

22

24%

812

0.0135

1.94

MX1

interferon-induced GTP-binding protein Mx1

4505291

102

11

17%

821

0.0104

2.51

MX1

interferon-induced GTP-binding protein Mx1

4505291

90

10

16%

831

0.0054

2.15

MX1

interferon-induced GTP-binding protein Mx1

4505291

108

11

16%

1285

0.0120

1.71

PD-ECGF

endothelial cell growth factor 1. platelet-derived

4503445

410

21

40%

994 704

0.0213 0.0004

1.69 2.43

SDHA STAT1

succinate dehydrogenase Fp signal transducer and activator of transcription

54607098 6274552

245 483

11 25

18% 34%

3307

0.0033

1.49

STAT1

signal transducer and activator of transcription

6274552

150

17

23%

6274552

228

16

23%

145

14

20%

1-alpha/beta 1-alpha/beta 3308

0.0058

1.88

STAT1

signal transducer and activator of transcription 1-alpha/beta

3309

0.0049

1.6

STAT1

signal transducer and activator of transcription

6274552 4507895

86

12

34%

10863945

194

21

36%

1140

0.0145

2.04

VIM

1-alpha/beta vimentin

3309

0.0049

1.6

XRCC5

X-ray repair cross-complementing protein 5

696

0.0006

1.8

not identifiedf

1141

0.0015

2.17

not identifiedf proteins differentially expressed in membrane/organelle fraction

799

0.0101

2.08

ABAT

4-aminobutyrate aminotransferase

7434172

516

27

800

0.0250

1.97

ABAT

4-aminobutyrate aminotransferase

7434172

493

29

. 70%

917 2320

0.0238 0.0305

1.56 1.62

ACADM AGAT

medium-chain specific acyl-CoA dehydrogenase mitochondrial a L-arginine: glycine amidinotransferase

4557231 6730020

127 77

11 5

26% 11%

628

0.0023

1.56

CAT

catalase

4557014

326

18

36%

371

0.0086

1.98

MX1

interferon-induced GTP-binding protein Mx1

4505291

129

11

20%

388

0.0127

2.05

MX1

interferon-induced GTP-binding protein Mx1

4505291

251

13

24%

375

0.0218

1.54

POR

cytochrome P450 reductase

247307

188

20

32%

917

0.0238

1.56

OAT

ornithine aminotransferase. mitochondrial isoform 1

4557809

270

14

26%

858

0.0180

1.61

TUFM

elongation factor Tu. mitochondrial

833999

115

11

24%

365

0.0148

1.68

not identifiedf

365

0.0148

1.68

a

Master numbers refer to spots in Figure 1. b Average fold increase (+) or decrease ( ) in NR patients. c Sequence identification numbers (Protein GI) from NCBI database. d Mascot scores indicate the confidence of protein identification (Mascot score > 75 = p < 0.05). e Number of peptides matching the identified proteins. f Mascot scores were below the statistical significant threshold (Mascot score < 75).

identified a PC1 which explains 80.1% of variance; however, SVR patients were less efficiently separated from NR patients in this data set (Figure 2B). Conversely, the second principal component

did not account for a significant variance in both protein fractions (PC2cytosol = 12.6%, PC2memb./organelle = 9.5%). Moreover, PCA performed using all spots of 2-D maps, independently of their 721

dx.doi.org/10.1021/pr2006445 |J. Proteome Res. 2012, 11, 717–727

Journal of Proteome Research

ARTICLE

Figure 2. Unsupervised principal component and hierarchical clustering analyses of protein spots differentially expressed in liver biopsies from SVR and NR. (A, B) Score and loading plots of PC1 and PC2 from PCA of cytosol (A) and membrane/organelle (B) fractions. (C, D) The hierarchical clustering generated an ordered treelike structure with samples segregated according to the response of patients to the therapy except for one SVR sample assigned to the group of NR in the membrane/organelle data set. Each colored bar represents a protein up-regulated (red) or down-regulated (green). Protein spot numbers correspond to DeCyder master numeration used in Table 1.

average ratio variation and t test probability, did not account for a greater source of variability in these data sets able to discriminate more efficiently SVR and NR (data not shown). Unsupervised hierarchical clustering analysis of these data sets produced an ordered treelike structure, with samples segregated accordingly

to the therapy response with the exception of one SVR patient which was miss-assigned to the group of NR when proteins of the membrane/organelle fraction were considered (Figure 2C,D). Spots selected by 2-D DIGE were then analyzed by mass spectrometry in order to establish their identity (Table 1, Supporting 722

dx.doi.org/10.1021/pr2006445 |J. Proteome Res. 2012, 11, 717–727

Journal of Proteome Research

ARTICLE

Information Figure S1B). Overall, 34 spots corresponded to 21 unique proteins: 16 in the cytosolic fraction and 8 in the membrane/organelle fraction, with 3 proteins present in both subcellular fractions. Three spots were not identified, possibly because of their relative low abundance (see Supporting Information Figure S2), while the identification of 3 spots (2 in cytosolic and 1 in the membrane fraction) was not unambiguous since two proteins were detected in each spot (Table1, see protein identifications with the same master number). Six proteins were identified in more than one spot, indicating that different isoforms of the same protein were expressed in the liver. Predominant categories of differentially expressed proteins, according to biological function, included proteins related to interferon-mediated antiviral activity, stress response, and energy metabolism (Supporting Information Table S2). The largest variations in protein levels with the greatest statistical confidence between SVR and NR corresponded to 4 spots identified as signal transducer and activator of transcription 1-alpha/beta (STAT1) and 5 spots (three in the cytosol and two in the membrane/organelle fraction) identified as interferon-induced GTP-binding protein (Mx1), two important regulators of the antiviral response mediated by interferon (Table 1 and Supporting Information Table S2). The levels of all these spots appeared to be increased in NR patients, suggesting that a change in protein levels rather than differential post-translational modifications occurs in NR patients (Figure 1B and Supporting Information Figure S3). On the other hand, two specific isoforms of dihydroxyacetone kinase (DAK) were found to differently correlate with therapy outcome, with one isoform of DAK increased while the other decreased in SVR patients when compared to NR patients (Figure 1D). The proteomic analysis revealed also significant changes in enzymes involved in different detoxifying pathways, such as DD4 (Figure 1C), catalase, cytochrome P450 reductase and ALDH1L1, as well as proteins playing a role in lipid, aminoacid or energy metabolism, with most of the identified proteins having a mitochondrial localization (Table 1 and Supporting Information Table S2). Taken together, the 2-D DIGE analysis show that specific protein expression profiles in the liver of HCV patients, prior to treatment, are associated with therapy responsiveness

changes of 4.36 (p < 0.05), 2.10 (p < 0.05), 0.7 (p < 0.01), and 1.5 (p < 0.05), respectively. Notably, these markers were found to be differentially expressed also when their levels were analyzed by real-time PCR (Supporting Information Figure S5). Moreover, we found that, even though immunoblotting on monodimensional gel did not allow to visualize differentially expressed DAK isoforms, its whole protein level also correlated with therapy outcome with an average fold change of 1.17 (p < 0.05) between NR and SVR. Although statistically significant differences were observed in the average levels of these proteins between SVR and NR patients, their expression showed a relatively high variation among individual patients of the same group (Supporting Information Figure S4). This observation suggests that the prospective accuracy in the prediction of the therapy outcome in individual patients may not rely on a single marker but rather on the combined use of different proteins. To address this point, we analyzed the correlation between the therapy outcome and the levels of proteins either individually or in combination. As shown in Figure 3B, there is a relatively poor contribution of single proteins to this correlation, while a stronger correlation was obtained when the expression levels of all five proteins were combined in multivariable correlation analysis which achieved a maximum correlation coefficient of 0.740 (p = 0.0012). The correlation analysis was also performed by combining clinical data of patients and protein expression levels. Interestingly, we found that among a number of demographic, virologic, and clinical characteristics (Supporting Information Table S3) the addition of the HCV genotype better improved this correlation reaching a coefficient of 0.778 (Figure 3B). Accuracy-Based Assessment of Validated Proteins for Therapy Outcome Prediction in an Independent Patient Cohort

The expression levels of the 5 validated proteins in the 23 liver biopsies of the Validation group were used as a training data set to build a prediction model based on supervised discriminant analysis. Nineteen out of 23 samples (82.6%) resulted correctly classified [85.7% (12 out of 14) and 77.8% (7 out of 9) for SVR and NR, respectively] (Table 2, Supporting Information Tables S4 and S5). In order to estimate the prognostic potential of these proteins, the model was used to predict the therapy outcome in an independent group of liver biopsies, considered as a blind test cohort. Since patients infected with HCV genotype 1 are most unlikely to respond to the therapy, this analysis was restricted to a group of 14 liver samples from patients infected exclusively with this genotype (Supporting Information Table S1; Testing group). As shown in Table 2, this analysis correctly classified all but two patients, which represents a predictive accuracy of 85.7% [SVR = 87.5% (7/8); NR = 83.3% (5/6)]. The discriminating ability of the model was further assessed by receiver operating characteristic (ROC) curve analysis. The areas under the curve (AUC) for the model based on the 5 selected proteins were 0.960 and 0.854 in the training and testing data sets, respectively (Supporting Information Table S6). Moreover, in both data sets, the AUC based on all proteins showed an improved discrimination capacity, although not statistically significant in every case, when compared to the AUC based on individual proteins (Supporting Information Table S6).

Validation of Liver Proteins Discriminating between Responder and Nonresponder Patients

In order to validate the results obtained from the proteomic screening, expression levels of 2-D DIGE identified proteins were verified by immunoblotting analyses on whole extracts from liver biopsies of 23 consecutively enrolled patients (Table S1; Validation and Training Study group). Fourteen out of 23 patients were SVR, 13 were infected with HCV genotype 1, 6 with genotype 2, and 4 with genotype 3. Immunoblotting analysis of HCV liver biopsies was carried out with 10 out of 21 proteins identified in the proteomic screening. This subset of proteins was selected on the basis of the average ratio variation and t test probability, with 5 chosen among the up-regulated proteins (STAT1, Mx1, PDECGF, vimentin, and FTCD), 4 among the down-regulated (ALDH1L1, DD4, enolase 3, and ABAT) in NR patients and DAK, which showed different variation depending on the isoform analyzed (see Table 1 and figure 1D). As illustrated in Figure 3A (see also Supporting Information Figure S4), this validation approach confirmed the differential expression of Mx1, STAT1, DD4, and PD-ECGF between NR and SVR with average fold

’ DISCUSSION The identification of prognostic markers of therapy outcome is an important issue for the management of HCV patients, with the current therapy being relatively inefficient and often associated 723

dx.doi.org/10.1021/pr2006445 |J. Proteome Res. 2012, 11, 717–727

Journal of Proteome Research

ARTICLE

Figure 3. Validation of proteomic results by immunoblotting analysis in an independent group of SVR and NR. (A) Immunoblotting analysis of STAT1, Mx1, DD4, DAK, and PD-ECGF in liver extracts. Differences are represented as scatter plots showing the mean and the standard deviation of protein levels normalized to the expression of GAPDH; P-values correspond to the overall comparison between SVR and NR using the Mann Whitney test. (B) Correlation between the therapy outcome and protein expression levels in liver biopsies from SVR and NR. Analysis was performed using single Pearson and multiple correlation analysis to calculate the correlation of single or combined variables, respectively. Correlation coefficients (r) are given together with statistical significance (p) and combination of variables used in the analysis.

Table 2. Supervised Discriminant Analysis Using STAT1, Mx1, DD4, DAK, and PD-ECGF Proteins Quantified by Immunoblottinga NR predictive accuracy

SVR predictive accuracy

overall predictive accuracy

training set (n = 23)

77.8% (7/9)

85.7% (12/14)

82.6% (19/23)

test set (n = 14)

83.3% (5/6)

87.5% (7/8)

85.7% (12/14)

a

Predicted group membership for SVR and NR along with the calculated overall classification rate for the training (n = 23) and testing data set (n = 14) are shown.

with serious side effects. Recently, different proteomic approaches have been applied to identify serum predictive markers for response to interferon therapy in HCV patients.22 28 Here we describe a comparative proteomic analysis of liver biopsies collected before therapy initiation aimed at identifying protein profiles discriminating SVR and NR patients. In order to have

an in-depth view of the proteome changes occurring in the liver of HCV patients, the analysis was performed following subcellular fractionation to better visualize low abundant proteins. This approach allowed the identification of 21 proteins whose expression levels correlate with the therapy outcome, including many low-level regulatory proteins such as signal transduction or 724

dx.doi.org/10.1021/pr2006445 |J. Proteome Res. 2012, 11, 717–727

Journal of Proteome Research transcription factors. Due to the limited number of patients initially analyzed in the proteomic analysis, the expression of the best performing identified proteins was further validated on a larger cohort of patients using immunoblotting assays. The larger variations observed between SVR and NR patients correspond to two proteins related to the IFN pathway, Mx1 and STAT1, both up-regulated in NR patients. STAT1 is a member of the signal transducers and activators of transcription family, involved in the up-regulation of genes activated by type I and type II interferons.29 Mx1 is an IFN inducible gene, playing an important role in the antiviral activity of IFN as well as in the induction of apoptosis of infected cells.30 Mx1 and STAT1 have been extensively characterized as part of the innate immune response to HCV in the liver.31 36 The identification of STAT1 and Mx1 in our proteomic study is consistent with results obtained in the analysis of liver biopsies of HCV patients using DNA microarray and immunohistochemical approaches13 16,31 and supports the evidence that pretreated HCV NR patients have increased basal levels of IFN-stimulated genes. Importantly, differential intrahepatic expression levels of IFN-inducible genes have been described to correlate with genetic variants of IL28B associated with increased probability of spontaneous HCV clearance and positive IFN therapy outcome.37 Our screening also identified liver proteins not directly associated with the IFN response, that is, DAK, DD4, and PDECGF. DAK is a member of the evolutionarily conserved family of dihydroxyacetone kinases.38 Interestingly, besides its role in different metabolic processes,39 DAK has been proposed to play a role in the innate immunity pathway. DAK interacts and negatively regulates the activity of the MDA5, a cellular RNA helicase that, together with RIG-I, functions as a viral sensor and is responsible for the activation of the IFN-mediated antiviral response.38 Our data showed increased levels of DAK in the liver of NR patients and, in particular, of specific protein isoforms, which likely represents different posttranslational modifications of DAK since only the net charge of this protein was changed (see Figure 1D). The characterization of these posttranslational modifications and the enzyme carrying them out would possibly contribute to better elucidate pathways relevant for antiviral response in HCV patients. DD4 belongs to the aldo/keto-reductase superfamily, a class of enzymes that catalyze the conversion of aldehydes and ketones to their corresponding alcohols.40 DD4 is involved in the metabolism of prostaglandins and various steroids, including bile acids, as well as of xenobiotics.41 Although the role of DD4 in liver metabolism remains to be elucidated, decreased levels of DD4 in NR patients might account for a reduced liver detoxifying activity, which has been previously associated with IFN therapy failure.7 Interestingly, an inverse correlation between bile acid levels and therapy response in patients infected with HCV genotype 2 and 3 has been recently reported.42 The endothelial cell growth factor PD-ECGF, together with other angiogenic factors of the vascular endothelial growth factors family, has been previously proposed as a noninvasive hepatocellular carcinoma marker for monitoring hepatitis C virus cirrhotic patients.43 46 Notably, reduced levels of VEGF and angiopoietin-2, and increased soluble Tie-2 have been also associated with the clearance of HCV.47 Apart from proteins involved in the IFN pathway, most of the factors identified in our screening have not been reported to be differentially expressed in DNA microarray studies.13 16 The presence of changes detectable only at protein levels, as confirmed in

ARTICLE

the case of the DAK protein (Figure 3A and Supporting Information Figure S5), suggests that alterations occurring posttranscriptionally could account for part of the variation observed between SVR and NR. Further studies focusing on the direct comparison of RNA and protein levels of the identified markers will be required to fully elucidate this point. It is interesting to note that, among identified proteins, MX1, ABAT, and AGAT were present in both cytosolic and membrane fractions (Table 1). Although this could be due to the fact that the fractionation was not entirely successful, two proteins, ABAT and AGAT, have mitochondrial localization signals and their presence in the cytosolic fraction could reflect an increased mitochondrial permeability which is known to occur in cells under stress, such as viral infection, as an initial step of the induction of apoptosis.48 On the other hand, MX1 has been shown to have different subcellular localization which could be modulated, for instance, by its interaction with membrane channels.49 An important observation arising from our analysis is that the assessment of single protein markers is not sufficient to discriminate individual patients according to treatment response, while their combined evaluation, together with HCV genotype, results in an efficient classification (r = 0.778, p = 0.011). On the basis of these results, we developed a prediction model based on supervised machine learning-algorithms that classified SVR and NR patients with an accuracy of 85.7%. This model remains to be tested on a larger cohort of patients, with particular attention to the influence of HCV genotypes and comorbidities of the patients on the predictive potential of these markers. On the other hand, it should be underlined that the usage of multiple markers in routine diagnostics could represent a limitation since this would lead to a significant increase of their cost. In conclusion, our study allowed the identification and preliminary validation of a small group of proteins able to predict the therapy outcome. Our results may be useful to develop simple and efficient assays to identify HCV patients, in particular genotype 1-infected patients, with the highest chance of response to treatment. Although the predictive capability of these markers does not allow to deny treatment to HCV patients and does not eliminate the need for viral kinetics after a few weeks of therapy, a decision process could be implemented for those na€ive patients with factors related to a decreased likelihood of SVR, such as high viral load or older age, but with a favorable proteomic pattern of therapy outcome. Our findings further underline the importance of integrating molecular, virological, and clinical data in order to identify those patients with better chances of responding to the therapy. This is particularly relevant now that the availability of large-scale expression profiling data is rapidly increasing, thanks to the recent development of accurate quantitative high-troughput technologies.

’ ASSOCIATED CONTENT

bS

Supporting Information Additional experimental results, tables, and figures. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*(G.M.F.) Mailing address: INMI Spallanzani, Via Portuense 292, 00149 Rome, Italy. E-mail: gianmaria.fi[email protected]. Telephone: +390655170908. Fax: +39065582825. (M.P.) Mailing address: 725

dx.doi.org/10.1021/pr2006445 |J. Proteome Res. 2012, 11, 717–727

Journal of Proteome Research

ARTICLE

University of Tor Vergata, Department of Biology, Via della Ricerca Scientifica, 00133 Roma, Italia. Telephone: +390672594370. Fax: +3906 2023500. E-mail: [email protected].

Yano, K.; Masaki, N.; Sugauchi, F.; Izumi, N.; Tokunaga, K.; Mizokami, M. Genome-wide association of IL28B with response to pegylated interferon-alpha and ribavirin therapy for chronic hepatitis C. Nat. Genet. 2009, 41, 1105–1109. (12) Thomas, D. L.; Thio, C. L.; Martin, M. P.; Qi, Y.; Ge, D.; O’Huigin, C.; Kidd, J.; Kidd, K.; Khakoo, S. I.; Alexander, G.; Goedert, J. J.; Kirk, G. D.; Donfield, S. M.; Rosen, H. R.; Tobler, L. H.; Busch, M. P.; McHutchison, J. G.; Goldstein, D. B.; Carrington, M. Genetic variation in IL28B and spontaneous clearance of hepatitis C virus. Nature 2009, 461, 798–801. (13) Chen, L.; Borozan, I.; Feld, J.; Sun, J.; Tannis, L. L.; Coltescu, C.; Heathcote, J.; Edwards, A. M.; McGilvray, I. D. Hepatic gene expression discriminates responders and nonresponders in treatment of chronic hepatitis C viral infection. Gastroenterology 2005, 128, 1437–1444. (14) Feld, J. J.; Nanda, S.; Huang, Y.; Chen, W.; Cam, M.; Pusek, S. N.; Schweigler, L. M.; Theodore, D.; Zacks, S. L.; Liang, T. J.; Fried, M. W. Hepatic gene expression during treatment with peginterferon and ribavirin: Identifying molecular pathways for treatment response. Hepatology 2007, 46, 1548–1563. (15) Sarasin-Filipowicz, M.; Oakeley, E. J.; Duong, F. H.; Christen, V.; Terracciano, L.; Filipowicz, W.; Heim, M. H. Interferon signaling and treatment outcome in chronic hepatitis C. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 7034–7039. (16) Asselah, T.; Bieche, I.; Narguet, S.; Sabbagh, A.; Laurendeau, I.; Ripault, M. P.; Boyer, N.; Martinot-Peignoux, M.; Valla, D.; Vidaud, M.; Marcellin, P. Liver gene expression signature to predict response to pegylated interferon plus ribavirin combination therapy in patients with chronic hepatitis C. Gut 2008, 57, 516–524. (17) Ishak, K.; Baptista, A.; Bianchi, L.; Callea, F.; De Groote, J.; Gudat, F.; Denk, H.; Desmet, V.; Korb, G.; MacSween, R. N. Histological grading and staging of chronic hepatitis. J. Hepatol. 1995, 22, 696–699. (18) Scheuer, P. J. Classification of chronic viral hepatitis: a need for reassessment. J. Hepatol. 1991, 13, 372–374. (19) Ciccosanti, F.; Corazzari, M.; Soldani, F.; Matarrese, P.; Pagliarini, V.; Iadevaia, V.; Tinari, A.; Zaccarelli, M.; Perfettini, J. L.; Malorni, W.; Kroemer, G.; Antinori, A.; Fimia, G. M.; Piacentini, M. Proteomic analysis identifies prohibitin down-regulation as a crucial event in the mitochondrial damage observed in HIV-infected patients. Antiviral Ther. 2010, 15, 377–390. (20) Unlu, M.; Morgan, M. E.; Minden, J. S. Difference gel electrophoresis: a single gel method for detecting changes in protein extracts. Electrophoresis 1997, 18, 2071–2077. (21) Van den Bergh, G.; Arckens, L. Recent advances in 2D electrophoresis: an array of possibilities. Expert Rev. Proteomics 2005, 2, 243–252. (22) Patel, K.; Lucas, J. E.; Thompson, J. W.; Dubois, L. G.; Tillmann, H. L.; Thompson, A. J.; Uzarski, D.; Califf, R. M.; Moseley, M. A.; Ginsburg, G. S.; McHutchison, J. G.; McCarthy, J. J. MURDOCK Horizon 1 Study Team High predictive accuracy of an unbiased proteomic profile for sustained virologic response in chronic hepatitis C patients. Hepatology 2011, 53, 1809–1818. (23) Younossi, Z. M.; Limongi, D.; Stepanova, M.; Pierobon, M.; Afendy, A.; Mehta, R.; Baranova, A.; Liotta, L.; Petricoin, E. Protein pathway activation associated with sustained virologic response in patients with chronic hepatitis C treated with pegylated interferon (PEG-IFN) and ribavirin (RBV). J. Proteome Res. 2011, 10, 774–779. (24) Devitt, E. J.; Power, K. A.; Lawless, M. W.; Browne, J. A.; Gaora, P. O.; Gallagher, W. M.; Crowe, J. Early proteomic analysis may allow noninvasive identification of hepatitis C response to treatment with pegylated interferon alpha-2b and ribavirin. Eur. J. Gastroenterol. Hepatol. 2011, 23, 177–183. (25) Fujita, N.; Nakanishi, M.; Mukai, J.; Naito, Y.; Ichida, T.; Kaito, M.; Yoshikawa, T.; Takei, Y. Identification of treatment efficacy-related host factors in chronic hepatitis C by ProteinChip serum analysis. Mol. Med. 2011, 17, 70–78.

Notes ^

Joint senior authors.

’ ACKNOWLEDGMENT The authors would like to thank the technicians of the Biological Bank of INMI for their technical assistance with biopsy specimens and Tonino Alonzi for discussions and critical reading of the manuscript. This study was supported by a grant from European Community (APO-SYS Health F4-2007200767) to M.P., Ministry for Health of Italy (“Ricerca Corrente” to M.P., G.M.F., and G.A., “Ricerca Finalizzata RF07.103” to G.M.F., and “Ricerca Oncologica n. ONC-ORD 35/07” to G.M.F.), Ministry of University and Research of Italy to M.P. The authors have declared no conflict of interest. ’ REFERENCES (1) Seeff, L. B.; Hoofnagle, J. H. Appendix: The National Institutes of Health Consensus Development Conference Management of Hepatitis C 2002. Clin. Liver Dis. 2003, 7, 261–287. (2) Lavanchy, D. The global burden of hepatitis C. Liver Int. 2009, 29 (Suppl 1), 74–81. (3) Lauer, G. M.; Walker, B. D. Hepatitis C virus infection. N. Engl. J. Med. 2001, 345, 41–52. (4) Manns, M. P.; McHutchison, J. G.; Gordon, S. C.; Rustgi, V. K.; Shiffman, M.; Reindollar, R.; Goodman, Z. D.; Koury, K.; Ling, M.; Albrecht, J. K. Peginterferon alfa-2b plus ribavirin compared with interferon alfa-2b plus ribavirin for initial treatment of chronic hepatitis C: a randomised trial. Lancet 2001, 358, 958–965. (5) Fried, M. W.; Shiffman, M. L.; Reddy, K. R.; Smith, C.; Marinos, G.; Goncales, F. L., Jr.; Haussinger, D.; Diago, M.; Carosi, G.; Dhumeaux, D.; Craxi, A.; Lin, A.; Hoffman, J.; Yu, J. Peginterferon alfa-2a plus ribavirin for chronic hepatitis C virus infection. N. Engl. J. Med. 2002, 347, 975–982. (6) Hadziyannis, S. J.; Sette, H., Jr.; Morgan, T. R.; Balan, V.; Diago, M.; Marcellin, P.; Ramadori, G.; Bodenheimer, H., Jr.; Bernstein, D.; Rizzetto, M.; Zeuzem, S.; Pockros, P. J.; Lin, A.; Ackrill, A. M. PEGASYS International Study Group Peginterferon-alpha2a and ribavirin combination therapy in chronic hepatitis C: a randomized study of treatment duration and ribavirin dose. Ann. Intern. Med. 2004, 140, 346–355. (7) Mihm, U.; Herrmann, E.; Sarrazin, C.; Zeuzem, S. Review article: predicting response in hepatitis C virus therapy. Aliment. Pharmacol. Ther. 2006, 23, 1043–1054. (8) Ghany, M. G.; Strader, D. B.; Thomas, D. L.; Seeff, L. B. American Association for the Study of Liver Diseases Diagnosis, management, and treatment of hepatitis C: an update. Hepatology 2009, 49, 1335–1374. (9) Ge, D.; Fellay, J.; Thompson, A. J.; Simon, J. S.; Shianna, K. V.; Urban, T. J.; Heinzen, E. L.; Qiu, P.; Bertelsen, A. H.; Muir, A. J.; Sulkowski, M.; McHutchison, J. G.; Goldstein, D. B. Genetic variation in IL28B predicts hepatitis C treatment-induced viral clearance. Nature 2009, 461, 399–401. (10) Suppiah, V.; Moldovan, M.; Ahlenstiel, G.; Berg, T.; Weltman, M.; Abate, M. L.; Bassendine, M.; Spengler, U.; Dore, G. J.; Powell, E.; Riordan, S.; Sheridan, D.; Smedile, A.; Fragomeli, V.; Muller, T.; Bahlo, M.; Stewart, G. J.; Booth, D. R.; George, J. IL28B is associated with response to chronic hepatitis C interferon-alpha and ribavirin therapy. Nat. Genet. 2009, 41, 1100–1104. (11) Tanaka, Y.; Nishida, N.; Sugiyama, M.; Kurosaki, M.; Matsuura, K.; Sakamoto, N.; Nakagawa, M.; Korenaga, M.; Hino, K.; Hige, S.; Ito, Y.; Mita, E.; Tanaka, E.; Mochida, S.; Murawaki, Y.; Honda, M.; Sakai, A.; Hiasa, Y.; Nishiguchi, S.; Koike, A.; Sakaida, I.; Imamura, M.; Ito, K.; 726

dx.doi.org/10.1021/pr2006445 |J. Proteome Res. 2012, 11, 717–727

Journal of Proteome Research

ARTICLE

metastasis in human hepatocellular carcinoma. J. Gastroenterol. 1998, 33, 376–382. (44) Morinaga, S.; Yamamoto, Y.; Noguchi, Y.; Imada, T.; Rino, Y.; Akaike, M.; Sugimasa, Y.; Takemiya, S.; Takanashi, Y. Platelet-derived endothelial cell growth factor (PD-ECGF) is up-regulated in human hepatocellular carcinoma (HCC) and the corresponding hepatitis liver. Hepatogastroenterology 2003, 50, 1521–1526. (45) Poon, R. T.; Lau, C. P.; Cheung, S. T.; Yu, W. C.; Fan, S. T. Quantitative correlation of serum levels and tumor expression of vascular endothelial growth factor in patients with hepatocellular carcinoma. Cancer Res. 2003, 63, 3121–3126. (46) Mas, V. R.; Maluf, D. G.; Archer, K. J.; Yanek, K. C.; Fisher, R. A. Angiogenesis soluble factors as hepatocellular carcinoma noninvasive markers for monitoring hepatitis C virus cirrhotic patients awaiting liver transplantation. Transplantation 2007, 84, 1262–1271. (47) Salcedo, X.; Medina, J.; Sanz-Cameno, P.; Garcia-Buey, L.; Martin-Vilchez, S.; Borque, M. J.; Lopez-Cabrera, M.; Moreno-Otero, R. The potential of angiogenesis soluble markers in chronic hepatitis C. Hepatology 2005, 42, 696–701. (48) Galluzzi, L.; Brenner, C.; Morselli, E.; Touat, Z.; Kroemer, G. Viral control of mitochondrial apoptosis. PLoS Pathog. 2008, 4, e1000018. (49) Lussier, M. P.; Cayouette, S.; Lepage, P. K.; Bernier, C. L.; Francoeur, N.; St-Hilaire, M.; Pinard, M.; Boulay, G. MxA, a member of the dynamin superfamily, interacts with the ankyrin-like repeat domain of TRPC. J. Biol. Chem. 2005, 280, 19393–19400.

(26) Molina, S.; Misse, D.; Roche, S.; Badiou, S.; Cristol, J. P.; Bonfils, C.; Dierick, J. F.; Veas, F.; Levayer, T.; Bonnefont-Rousselot, D.; Maurel, P.; Coste, J.; Fournier-Wirth, C. Identification of apolipoprotein C-III as a potential plasmatic biomarker associated with the resolution of hepatitis C virus infection. Proteomics Clin. Appl. 2008, 2, 751–761. (27) Asselah, T.; Bieche, I.; Paradis, V.; Bedossa, P.; Vidaud, M.; Marcellin, P. Genetics, genomics, and proteomics: implications for the diagnosis and the treatment of chronic hepatitis C. Semin. Liver Dis. 2007, 27, 13–27. (28) Paradis, V.; Asselah, T.; Dargere, D.; Ripault, M. P.; Martinot, M.; Boyer, N.; Valla, D.; Marcellin, P.; Bedossa, P. Serum proteome to predict virologic response in patients with hepatitis C treated by pegylated interferon plus ribavirin. Gastroenterology 2006, 130, 2189–2197. (29) Najjar, I.; Fagard, R. STAT1 and pathogens, not a friendly relationship. Biochimie 2010, 92, 425–444. (30) Haller, O.; Staeheli, P.; Kochs, G. Interferon-induced Mx proteins in antiviral host defense. Biochimie 2007, 89, 812–818. (31) MacQuillan, G. C.; de Boer, W. B.; Platten, M. A.; McCaul, K. A.; Reed, W. D.; Jeffrey, G. P.; Allan, J. E. Intrahepatic MxA and PKR protein expression in chronic hepatitis C virus infection. J. Med. Virol. 2002, 68, 197–205. (32) Giannelli, G.; Guadagnino, G.; Dentico, P.; Antonelli, G.; Antonaci, S. MxA and PKR expression in chronic hepatitis C. J. Interferon Cytokine Res. 2004, 24, 659–663. (33) Melen, K.; Fagerlund, R.; Nyqvist, M.; Keskinen, P.; Julkunen, I. Expression of hepatitis C virus core protein inhibits interferon-induced nuclear import of STATs. J. Med. Virol. 2004, 73, 536–547. (34) Yan, W.; Lee, H.; Yi, E. C.; Reiss, D.; Shannon, P.; Kwieciszewski, B. K.; Coito, C.; Li, X. J.; Keller, A.; Eng, J.; Galitski, T.; Goodlett, D. R.; Aebersold, R.; Katze, M. G. System-based proteomic analysis of the interferon response in human liver cells. Genome Biol. 2004, 5, R54. (35) Feld, J. J.; Hoofnagle, J. H. Mechanism of action of interferon and ribavirin in treatment of hepatitis C. Nature 2005, 436, 967–972. (36) Lan, K. H.; Lan, K. L.; Lee, W. P.; Sheu, M. L.; Chen, M. Y.; Lee, Y. L.; Yen, S. H.; Chang, F. Y.; Lee, S. D. HCV NS5A inhibits interferonalpha signaling through suppression of STAT1 phosphorylation in hepatocyte-derived cell lines. J. Hepatol. 2007, 46, 759–767. (37) Urban, T. J.; Thompson, A. J.; Bradrick, S. S.; Fellay, J.; Schuppan, D.; Cronin, K. D.; Hong, L.; McKenzie, A.; Patel, K.; Shianna, K. V.; McHutchison, J. G.; Goldstein, D. B.; Afdhal, N. IL28B genotype is associated with differential expression of intrahepatic interferonstimulated genes in patients with chronic hepatitis C. Hepatology 2010, 52, 1888–1896. (38) Diao, F.; Li, S.; Tian, Y.; Zhang, M.; Xu, L. G.; Zhang, Y.; Wang, R. P.; Chen, D.; Zhai, Z.; Zhong, B.; Tien, P.; Shu, H. B. Negative regulation of MDA5- but not RIG-I-mediated innate antiviral signaling by the dihydroxyacetone kinase. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 11706– 11711. (39) Bachler, C.; Flukiger-Bruhwiler, K.; Schneider, P.; Bahler, P.; Erni, B. From ATP as substrate to ADP as coenzyme: functional evolution of the nucleotide binding subunit of dihydroxyacetone kinases. J. Biol. Chem. 2005, 280, 18321–18325. (40) Barski, O. A.; Tipparaju, S. M.; Bhatnagar, A. The aldo-keto reductase superfamily and its role in drug metabolism and detoxification. Drug Metab. Rev. 2008, 40, 553–624. (41) Hara, A.; Matsuura, K.; Tamada, Y.; Sato, K.; Miyabe, Y.; Deyashiki, Y.; Ishida, N. Relationship of human liver dihydrodiol dehydrogenases to hepatic bile-acid-binding protein and an oxidoreductase of human colon cells. Biochem. J. 1996, 313 (Pt 2), 373–376. (42) Iwata, R.; Stieger, B.; Mertens, J. C.; Muller, T.; Baur, K.; Frei, P.; Braun, J.; Vergopoulos, A.; Martin, I. V.; Schmitt, J.; Goetze, O.; Bibert, S.; Bochud, P. Y.; Mullhaupt, B.; Berg, T.; Geier, A.; Swiss Hepatitis C Cohort Study Group The role of bile acid retention and a common polymorphism in the ABCB11 gene as host factors affecting antiviral treatment response in chronic hepatitis C. J. Viral Hepat. 2010. (43) Jinno, K.; Tanimizu, M.; Hyodo, I.; Nishikawa, Y.; Hosokawa, Y.; Doi, T.; Endo, H.; Yamashita, T.; Okada, Y. Circulating vascular endothelial growth factor (VEGF) is a possible tumor marker for 727

dx.doi.org/10.1021/pr2006445 |J. Proteome Res. 2012, 11, 717–727