molecular mechanisms involved in the interaction effects

INTRODUCTIONAlcoholic liver disease (ALD) and

chronic hepatitis C virus (HCV) infectionare the most frequent chronic liver dis-eases in the Western world. In addition,ALD and HCV frequently coexist in thesame individual. Although both diseasesalone have a similar progression se-quence leading to cirrhosis in approxi-mately 15% of patients within 10–20years, their coexistence dramatically ac-celerates disease progression in a syner-gistic manner (1). This synergism affectsboth fibrosis progression and the devel-opment of hepatocellular carcinoma. The

relationship between ALD and HCV wasfirst described in populations of alcoholicindividuals with liver disease (2) andconfirmed later by investigators whocompared the prevalence of HCV mark-ers in alcoholic individuals with andwithout liver disease and found a signifi-cantly higher prevalence in those withliver disease (2–4).

Even though the coexistence of HCVand alcoholism has been associated withaccelerated hepatic injury, the pathogene-sis is not fully understood but is sus-pected to be multifactorial (1–4). Liverbiopsy samples in HCV-infected patients

with reported alcohol consumption char-acteristically show a pattern of hepaticinjury consistent with chronic viral hepa-titis, indicating that the alcohol plays arole in potentiating the effects of HCV,rather than causing traditional alcohol-related liver injury (5,6). The use of alco-hol has been found to be associated withimmune control of HCV and to affectHCV replication and/or viral clearance.This final effect may have an impact onthe evolution of HCV quasispecies, pre-sumably through its effects on the im-mune system (7,8).

DNA microarray studies offer a robustmethod for unbiased analysis of whole-genome messenger RNA (mRNA) ex-pression patterns. Expression profilingwith DNA microarrays has been used toidentify molecular network responses toethanol in cell culture (10,11), animalmodels (9–12), and humans (13,14). Ex-pression profiling has led to the identifi-cation of novel gene networks that re-

M O L M E D 1 6 ( 7 - 8 ) 2 8 7 - 2 9 7 , J U L Y - A U G U S T 2 0 1 0 | M A S E T A L . | 2 8 7

Molecular Mechanisms Involved in the Interaction Effects ofAlcohol and Hepatitis C Virus in Liver Cirrhosis

Valeria R Mas,1 Ryan Fassnacht,1 Kellie J Archer,1,2 and Daniel Maluf1

1Hume-Lee Transplant Center, Division of Transplant, Department of Surgery and 2Department of Biostatistics, Virginia Commonwealth University Health System, Richmond, Virginia, United States of America

The mechanisms by which alcohol consumption accelerates liver disease in patients with chronic hepatitis C virus (HCV) arenot well understood. To identify the characteristics of molecular pathways affected by alcohol in HCV patients, we fit probe-setlevel linear models that included the additive effects as well as the interaction between alcohol and HCV. The study included livertissue samples from 78 patients, 23 (29.5%) with HCV-cirrhosis, 13 (16.7%) with alcohol-cirrhosis, 23 (29.5%) with HCV/alcohol cirrho-sis and 19 (24.4%) with no liver disease (no HCV/no alcohol group). We performed gene-expression profiling by using microarrays.Probe-set expression summaries were calculated by using the robust multiarray average. Probe-set level linear models were fitwhere probe-set expression was modeled by HCV status, alcohol status, and the interaction between HCV and alcohol. We foundthat 2172 probe sets (1895 genes) were differentially expressed between HCV cirrhosis versus alcoholic cirrhosis groups. Genes in-volved in the virus response and the immune response were the more important upregulated genes in HCV cirrhosis. Genes in-volved in apoptosis regulation were also overexpressed in HCV cirrhosis. Genes of the cytochrome P450 superfamily of enzymeswere upregulated in alcoholic cirrhosis, and 1230 probe sets (1051 genes) had a significant interaction estimate. Cell death andcellular growth and proliferation were affected by the interaction between HCV and alcohol. Immune response and response tothe virus genes were downregulated in HCV-alcohol interaction (interaction term alcohol*HCV). Alcohol*HCV in the cirrhotic tis-sues resulted in a strong negative regulation of the apoptosis pattern with concomitant positive regulation of cellular division andproliferation.© 2010 The Feinstein Institute for Medical Research, www.feinsteininstitute.orgOnline address: http://www.molmed.orgdoi: 10.2119/molmed.2009.00165

This work has been presented in part at the American Transplant Congress meeting, Bos-

ton, Massachusetts, 2009.

Address correspondence and reprint requests to Valeria R Mas, Division of Transplant, De-

partment of Surgery, West Hospital Ninth Floor, 1200 East Broad Street, PO Box 980057, Rich-

mond, VA 23298-0057. Phone: (804) 828-2364; E-mail: [email protected].

Submitted November 11, 2009; Accepted for publication March 25, 2010; Epub

(www.molmed.org) ahead of print March 26, 2010.

spond to ethanol or differ across animalstrains with differing responses to etha-nol. In similar research, many studieshave been performed to evaluate thegene expression patterns associated withchronic HCV infection (15–17).

In the present study, to identify mo-lecular pathways affected by the addi-tion of alcohol in HCV, we modeledgene expression, including alcoholic cir-rhosis and HCV cirrhosis as separateconditions and their interaction term(alcohol*HCV). A better understandingof the underlying molecular mecha-nisms of alcohol-HCV interaction couldhelp investigators to develop novel tar-geted treatment options.

PATIENTS AND METHODS

Patients and SamplesIn this study we evaluated 78 liver

samples, including 23 (29.5%) from pa-tients with cirrhosis due to HCV infec-tion, 13 (16.7%) from patients with cir-rhosis due to alcohol, and 23 (29.5%)from patients with cirrhosis due to bothHCV and alcohol. In addition, 19 (24.4%)of the liver samples were from donorswith normal livers and were included inthe study as a group with neither HCV-nor alcohol-associated liver damage.Liver function and histopathology for thesamples from the healthy liver donorswere shown to be normal. All 19 of thesedonors were seronegative for HCV anti-bodies. Characteristics of the patientswho donated liver tissue are shown inSupplemental Table 1.

None of the HCV-infected patients inthis study were receiving antiviral treat-ment at the time the liver sample was ob-tained. Pretransplantation alcohol intakewas assessed by the attending physicianat the time of evaluation for liver trans-plantation. Significant alcohol consump-tion was defined as >50 g/d for a periodof >5 years or >80 g/d for a shorter pe-riod. This information was obtained fromchart review and patient questionnaire.

The research protocol was approvedby the institutional review board, and in-formed consent was obtained from all

patients. Normal livers and cirrhoticliver samples were obtained through theLiver Tissue Cell Distribution System,Richmond, VA, USA, which was fundedby NIH Contract #N01-DK-7-0004 /HHSN267200700004C.

Sample Collection and PathologicalData

At the time of collection, liver tissuesamples were placed in liquid nitrogenor RNA-later solution (Ambion, Austin,TX, USA) and stored at –80°C until use.The cirrhotic tissues from explanted liv-ers were classified by using Knodellscore and Ishak grade (18).

RNA Isolation, cDNA Synthesis and invitro Transcription for Labeled cRNAProbe

The sample preparation protocol wasperformed according to the AffymetrixGeneChip® Expression Analysis Manual(Santa Clara, CA, USA). Briefly, totalRNA was reverse transcribed by usingT7-polydT primer and converted intodouble-stranded cDNA (One-Cycle Tar-get Labeling and Control Reagents,Affymetrix, Santa Clara, CA, USA), withtemplates being used for an in vitro tran-scription reaction to yield biotin-labeledantisense cRNA. According to theAffymetrix GeneChip protocol, the la-beled cRNA was chemically fragmentedand made into the hybridization cocktail,which was then hybridized to HG-U133A2.0 GeneChips. The array image was gen-erated by the high-resolution GeneChip®

Scanner 3000 by Affymetrix. The datawere analyzed by using Affymetrix Mi-croarray Suite Software, Version 5.0, andBioconductor packages (19) available inthe R programming environment (20).

Microarray Data AnalysisAfter image analysis, probe-set expres-

sion summaries were calculated by usingthe robust multiarray average.

We conducted an unsupervised analy-sis by first filtering the data set by retain-ing only the most variable probe sets, forwhich the range of each probe set wascalculated and probe sets having a range

among the top 1% were retained. There-after, using the Ward method we appliedagglomerative hierarchical clustering tothe filtered data set using the 1-Pearsoncorrelation as the dissimilarity measure.

For each probe set, we comparedHCV cirrhotic liver tissue versus normalliver tissue and alcohol-cirrhotic tissueversus normal liver tissue using a two-sample t test with the Satterwaite ap-proximation to the degrees of freedomto account for unequal variances. There-after, probe-set level linear models werefit such that probe-set expression wasmodeled by HCV status, alcohol (EtOH)status and the interaction between HCVand alcohol. Specifically, the model wastaken to be:

,

where yij represents the log2 intensity ofprobe set j for subject i, μj represents thebaseline log2 intensity for probe set j,HCVi is a dichotomous variable repre-senting the HCV status of subject i and

represents the effect due to HCV

status on probe set j. EtOHi is a dichoto-mous variable representing the alcoholstatus of subject i and represents the

effect due to alcohol status on probe set j;represents the effect of the combination

of HCV and alcohol on probe set j, and eij

represents the random error term. Ofparticular interest were those probe setsfor which the term was significant,

indicating that the effect of the combina-tion of HCV and alcohol on gene expres-sion was not simply additive, but eithersynergistic or antagonistic. To control formultiple hypothesis tests, a Bonferronicorrection was used for all comparisons.

Interaction Networks and FunctionalAnalysis

Gene ontology and gene interactionanalyses were executed by using Inge-nuity Pathways Analysis tools 7.0(http://www.ingenuity.com) (IngenuitySystems, Redwood City, CA, USA).

β1j

β3j

β3j

β2j

y HCV EtOH

HCV EtOH

ij j i i

i i ij

j j

j

= + + +

( )( )+

μ β β

β ε

1 2

3

2 8 8 | M A S E T A L . | M O L M E D 1 6 ( 7 - 8 ) 2 8 7 - 2 9 7 , J U L Y - A U G U S T 2 0 1 0

H C V - A L C O H O L I N T E R A C T I O N I N C I R R H O T I C T I S S U E S

Validation of Microarray ResultsWe carried out a quantitative reverse-

transcriptase–real-time PCR (QPCR) forcaspase 1, apoptosis-related cysteinepeptidase (CASP1), interleukin 8 (IL-8),IL-15, platelet derived growth factor C(PDGFC), myxovirus (influenza virus)resistance 1 (Mx1), and (transforminggrowth factor [TGF]-β1) mRNAs fromthe same RNA samples that were sub-jected to microarray study. We per-formed reverse transcription of totalRNA using TaqMan® Reverse Transcrip-tion Reagents (Applied Biosystems, Fos-ter City, CA) according to the manufac-turer’s protocol. Real-time PCR reactionsthen were carried out in an ABI Prism7700 Sequence Detection System, usingTaqMan® Gene Expression Assays (Ap-plied Biosystems). Data was analyzed ac-cording to the comparative cycle thresh-old method and was normalized with ahousekeeping gene (β2-microglobulin).The Pearson correlation coefficient (r)was calculated to examine the relationbetween microarray and real-time PCRresults. P values < 0.05 were consideredsignificant.

All supplementary materials are availableonline at www.molmed.org.

RESULTS

Normal Liver Selection and AnalysisThe selection of the normal liver group

was critical for the analysis. We studiedgene expression profiles using high- density oligonucleotide arrays in 19donor liver samples. To identify probesets significantly correlated with donorage, we fit probe-set–specific linear mod-els that included age as the independentvariable and obtained the significance ofthe estimated slope parameter. Probe setswith a raw P value < 0.001 were consid-ered significant. The false-discovery rate(FDR) method of Benjamini andHochberg was used to adjust for multi-ple comparisons. Donor liver groupswere classified by age as <25, 25–50 and>50 years, and a Jonckheere-Terpstra testfor trend was applied to identify probe

sets with a significant monotonic trendwith age. For the trend tests, the FDRwas estimated by using the q-valuemethod (Storey and Tibshirani). For eachprobe set a two-sample t test assumingunequal variances was used to comparenormal livers with respect to race (whiteversus African American), sex and causeof death (stroke/cerebrovascular acci-dent [CVA] versus trauma). The FDRwas estimated by using the q-valuemethod (Storey and Tibshirani). Donorcharacteristics included median age 41(range 9–66) years, 68.4% male and 73%white. Causes of donor death werestroke/CVA (n = 7), trauma (n = 9) andother (n = 3). With the use of a rawP value of < 0.001, 21 probe sets weresignificantly associated with age. Whenwe examined probe-set expression val-ues with a significant monotonic increas-ing and decreasing trend with age (P <0.001), 11 probe sets were significantlyincreasing and 11 probe sets were signifi-cantly decreasing. For the tests of signifi-cance that compared cause of death dueto stroke/CVA to trauma, only 10 probesets were significant at the 0.001 level.Fifty-six probe sets were significantly dif-ferent between males and females, withthe probe sets with very small P valuesand FDR being located on the x or ychromosome as expected. However, 288probe sets were significantly different forrace comparison (white versus AfricanAmerican) (P < 0.001). After performingthese analyses and recognizing a highFDR was associated with the probe setsidentified in most of these analyses, weconcluded that race was the more criticalvariable for which to control, with re-spect to normal tissues. Therefore wematched the race distribution betweenour control and study groups.

Cluster AnalysisFor the unsupervised analysis, after

we applied our variance-based filter thedata set consisted of 223 probe sets. Theheatmap resulting from the agglomera-tive hierarchical clustering procedure isshown in Figure 1. The unsupervisedanalysis showed that all the normal liver

samples clustered together. Moreover, wealso observed that the samples with alco-holic cirrhosis tended to cluster togetherin the same group whereas the tissuesamples with HCV-cirrhosis and thosewith the combination of HCV and alco-hol clustered together in a differentgroup. These findings might indicatethat the dominance of the HCV-cirrhoticgene expression pattern in those cirrhotictissues was associated with the interac-tion of HCV and alcohol.

Molecular Pathways Involved inAlcohol-Cirrhosis and HCV-Cirrhosis

By comparing signatures of normalliver tissue and tissue affected by otherliver diseases, we were able to exploit themolecular profiles of particular diseaseprocesses. Specifically, 2565 probe setswere significantly differentially ex-pressed when we compared alcoholic cir-rhotic versus normal liver tissues. Fur-thermore, 3370 probe sets weresignificantly differentially expressedwhen we compared HCV versus normalliver, with 916 probe sets being signifi-cant in both comparisons (Figure 2). Be-cause main effects cannot be directly in-terpreted when interaction is present,only the 1406 alcohol-specific probe sets(corresponding to 1272 unique genes)and the 2222 HCV-specific probe sets(corresponding to 1,841 unique genes)were further examined by using Ingenu-ity Pathway Analysis.

Core analysis was performed to inter-pret the data set in the context of biolog-ical processes, pathways and molecularnetworks. From the analysis of the dif-ferentially expressed genes in alcoholiccirrhotic samples compared with normallivers, 100 networks were identified (46networks with a score higher than 15).The top canonical pathways associatedwith these genes included pyrimidinemetabolism, NF-kB signaling, CD28 sig-naling in T-helper cells, and glycolisis/gluconeogenesis. The most overex-pressed genes included genes involvedin antigen presentation (IGKC, HLA-Cand IGHM) and inflammatory response(IL8). IL-8, a cytokine produced by a

R E S E A R C H A R T I C L E


host of cells, including monocytes,macrophages, Kupffer cells and hepato-cytes, can activate neutrophils. Periph-eral neutrophilia and liver neutrophil in-filtration are frequently noted in patientswith alcoholic liver disease. However,the relationship between IL-8 and differ-ent stages of alcoholic liver disease isuncertain. Genes involved in the im-mune response were downregulated inalcoholic cirrhosis (IRAK1, IL17RA,IL13RA1, LRIG1 and IKBKG, among oth-ers). In a recent study, Lemmers et al.(22) studied the IL-17 pathway in alco-holic cirrhosis and alcoholic hepatitis. Inthe present study, IL-17B gene expres-sion was upregulated in alcoholic cir-rhotic tissues. The protein encoded bythis gene is a T-cell–derived cytokinethat shares sequence similarity with

IL-17. This cytokine was reported tostimulate the release of tumor necrosisfactor-α (TNF-α) and IL-1 β from amonocytic cell line.

Growth factors were overexpressed inalcoholic cirrhotic tissues compared withnormal liver tissues. Also overexpressedin alcoholic cirrhosis were CD96 (the atype I membrane protein encoded by thisgene belongs to the immunoglobulin su-perfamily and may play a role in the ad-hesive interactions of activated T and NKcells during the late phase of the immuneresponse) and CD7 (the protein encodedby this gene is found on thymocytes andmature T cells).

Moreover, ALDH3B1 (aldehyde dehy-drogenase 3 family, member B1), wasalso overexpressed in alcoholic cirrhrotictissues. The aldehyde dehydrogenases

are a family of isozymes that may play amajor role in the detoxification of alde-hydes generated by alcohol metabolismand lipid peroxidation.

From the analysis of the differentiallyexpressed genes in HCV cirrhotic tissuescompared with normal livers, we identi-fied 94 networks (45 with a score higherthan 15). The top-scoring network isshown in Figure 3. Cell-to-cell signalingand interaction and cellular developmentwere identified as the principal molecu-lar and cellular functions associated withthese genes.

A characteristic pattern was observedin HCV cirrhotic tissues showing overex-pression of interferon-inducible genes(IFITM1, IFI27 and ISG20, among others)and genes involved in viral response(MX1, PRKRA and IL18, among other).Also, genes related to immune responseemerged as important in HCV cirrhotic



Figure 1. Unsupervised cluster analysis. An unsupervised analysis was conducted by first fil-tering the data set by retaining only the most variable probe sets, for which the range ofeach probe set was calculated and probe sets having a range among the top 1% wereretained. Thereafter, agglomerative hierarchical clustering performed using the Wardmethod was applied to the filtered data set with the 1-Pearson correlation as the dissimi-larity measure. Blue light boxes, normal liver tissues; black boxes, alcohol-cirrhotic liver tis-sues; pink boxes, HCV-cirrhotic liver tissues; green boxes, alcohol-HCV cirrhotic liver tissues.

Figure 2. Venn diagram. For each sample,RNA was extracted and after being con-verted to biotin-labeled cRNA was hy-bridized to an Affymetrix HG-U133Av2GeneChip. After image analysis, probe-setexpression summaries were calculated byusing the robust multiarray average. There-after, probe-set–level linear models were fitsuch that probe-set expression was mod-eled by HCV status, alcohol status, andthe interaction between HCV and alco-hol. Of particular interest were thoseprobe sets for which all three terms (HCV,Alcohol, and HCV*Alcohol) were signifi-cant. To control for multiple hypothesistests, probe sets were significant at an αlevel of 0.0001.

tissues (IL15, IL8, IL10RA, IL18, FCGR2Aand CD48, among others). Specifically,genes related to natural killer cells wereupregulated, including IL15, CCL5,KLRC4, KLRB1, KIR2DL3 and CD44. Theprotein encoded by the IL15 gene is a cy-tokine that regulates activation and pro-liferation of T cells and natural killercells. Moreover, IL-15 signaling wasidentified as one of the top canonicalpathways (P = 0.027).

We also observed overexpression ofthe gene that encodes IL-18, an inter-feron γ– inducing factor that plays a criti-cal role in the T-helper type 1 response

required for host defense againstviruses. Antibodies to IL-18 have beenfound to prevent liver damage in amurine model (23).

Another important finding was thatchemokine genes were upregulated inthis group (CCL21, CCL4, CCL5, CXCL9and CXCL10). Moreover, genes involvedin the cell-mediated immune response asgranzyme A and granzyme B were up-regulated. The proteins encoded by thesegenes are crucial for the rapid inductionof target-cell apoptosis by cytolyticT lymphocytes (CTL) in cell-mediatedimmune responses.

Genes Differentially Expressed inAlcoholic Cirrhosis versus HCVCirrhosis

We found 2172 probe sets (corre-sponding to 1896 unique genes) thatwere differentially expressed. This find-ing was revealed by the comparisonanalysis of 23 tissues with HCV cirrho-sis versus 13 tissues with alcoholic cir-rhosis performed by using a two-samplet test with a Bonferroni correction (P <0.05/ 22,277 considered significant).From this analysis, 89 networks wereidentified (39 with a score higher than15). RNA posttranscriptional modifica-tions (P = 6.7E-08 to 3.9E-03, 42 genes);cell death (P = 1.2E-06 to 5.2E-03, 348genes); cellular development (P = 8.5E-06 to 5.5E-03), gene expression, and cel-lular growth and proliferation were theprincipally associated molecular andcellular functions.

Interferon and interferon-induciblegenes were upregulated in HCV cirrhotictissues compared with alcoholic cirrhotictissues (MX1, IFI27, PRKRA, IFI 44,ISG20 and IFI16, among others). Thegene expression levels of the growth fac-tors TGF-β1, VEGFA and PDGFC werealso upregulated in HCV cirrhotic tis-sues. Moreover, the expression of IL-8and IL-15 was increased in these tissues.

Gene symbols for all Affymetrix probesets were obtained using the annotateBioconductor package. Thereafter, allprobe sets annotated to interrogate spe-cific genes in the pathways of interest (re-sponse to virus, immune and inflamma-tory response, synthesis of cholesterolsteroids and other lipids, solute trans-porters, growth factors and growth factorreceptors and apoptosis) were retained,and agglomerative hierarchical clusteringusing the Ward method with 1-Pearson(1 minus Pearson correlation) as the dis-similarity measure was applied to eachpathway. The resulting heatmap and den-drograms for the specific pathway relatedwith response to the virus are shown inFigure 4. Once again, the gene expressionpattern of HCV cirrhotic tissues is moresimilar to the profiles of tissues with cir-rhosis due to both HCV and alcohol.



Figure 3. The top-scoring network of interactions among the 3370 probe sets identified assignificantly differentially expressed when we compared HCV with normal samples. Theprobe sets were subsequently analyzed by using the Ingenuity pathway analysis software(https://analysis.ingenuity.com). This software is designed to identify dynamically gener-ated biological networks, global canonical pathways and global functions. At intercon-nections of significant functional networks, protein nodes appeared in different shades ofred and green or white depending on being upregulated and downregulated or no-change, respectively, in HCV-cirrhotic samples.

Interaction between HCV and EthanolIt was of great interest to study the in-

teraction between ethanol and HCV inthe cirrhotic tissues. Probe-set–level lin-ear models were fit such that probe-setexpression was modeled by HCV status,alcohol status, and the interaction be-tween HCV and alcohol. Among the22,277 probe-set–specific linear models,1230 probe sets (corresponding to 1051unique genes) had a significant interac-tion. The estimates of the interactionterm, which provides information on theexpected change in gene expression whenboth HCV and alcohol are causative fac-tors of cirrhosis, are reported in Supple-mental Table 2.

The top molecular and cellular func-tions affected by the interaction betweenHCV and alcohol were cell death andcellular growth and proliferation. Thetop canonical pathways for those geneswere the antigen-presentation pathway(P = 0.0079), IL-4 signaling (P = 0.0069),hepatic fibrosis/hepatic stellate cell acti-vation (P = 0.013) and the coagulationsystem (P = 0.02).

A substantial number of genes in-volved in apoptosis were present in theinteraction analysis (TRADD, CASP1,AVEN, BAD, TNFRSF21, BCL2 andBAK1, among others). For some of thegenes involved in apoptosis that wereaffected by the interaction betweenHCV and alcohol, interaction plots wereconstructed to enable visualization ofthe expression changes based on theHCV status (HCV positive versus nega-tive) or alcohol status (yes versus no)(Figure 5).

Among the genes having a significantinteraction, genes involved in immuneresponse had a negative estimate for theinteraction term. These genes includedgenes associated with immunoglobulins(IGBP1, IGL, IGKC), chemokines(CXCL12, CXCL6, CXCR4, CXCR6), inte-grins (ITGBL1) and immune cell recep-tors (TRA@, TRBC1, TICAM2). Also, anegative estimate for the interactionterm was associated with matrix metal-loproteinase (MMP) genes (MMP7,MMP2), genes involved in cell-to-cell

and cell-to-matrix interactions (THBS1,TIMP3, BGN, DCN, COL6A3, COL1A2,among others).

Interestingly, IFNAR1 and IRF2 geneshad a significant positive interaction. Theprotein encoded by this gene is a type Imembrane protein that forms one of thetwo chains of a receptor for interferons αand β. Binding and activation of the re-ceptor stimulates Janus protein kinases,which in turn phosphorylate several pro-teins, including STAT1 and STAT2. Theencoded protein also functions as an an-tiviral factor. IRF2 encodes interferonregulatory factor 2, a member of the in-

terferon regulatory transcription factor(IRF) family. IRF2 competitively inhibitsthe IRF1-mediated transcriptional activa-tion of interferons (α and β), and pre-sumably other genes that employ IRF1for transcription activation. Table 1 sum-marizes the more important findings foreach of the described tissue comparisonanalysis.

Gene Expression Quantitation byUsing QPCR

We used two criteria to select thegenes for the QPCR microarray assay:the genes were in the top of the differ-



Figure 4. Supervised cluster analysis for response to virus genes. Gene symbols for allAffymetrix probe sets were obtained by using the Bioconductor annotation package.Thereafter, all probe sets annotated to interrogate specific genes in the pathways of inter-est (response to virus) were retained, and agglomerative hierarchical clustering was per-formed by using the Ward method with 1-Pearson as the dissimilarity measure was appliedto the pathway. Blue light boxes, normal liver tissues; black boxes, alcohol-cirrhotic liver tis-sues; pink boxes, HCV-cirrhotic liver tissues; green boxes, alcohol-HCV cirrhotic liver tissues.

entially expressed genes and/or thegenes were among the top scoring networks and pathways when we usedinteraction networks and functionalanalysis for the different comparisonanalyses.

Expression levels of the CASP1, IL-8,IL-15, PDGFC, MX1, and TGF-β1mRNAs were further confirmed byusing QPCR. These genes were identi-fied as significantly differentially ex-pressed between different tissues types.The results from the microarray were re-produced by QPCR (r = 0.82, P < 0.001; r = 0.71, P = 0.002; r = 0.65, P < 0.001; r =0.7, P < 0.001; r = 0.73, P < 0.001; and r =0.61 P < 0.001 respectively).

DISCUSSIONThe underlying molecular mechanisms

of alcohol and HCV-mediated liver dis-ease are complex and they are still in-completely understood despite decadesof intensive investigative efforts. Both al-cohol and HCV can reproduce the foursequential hallmarks of liver disease:steatosis, steatohepatitis, fibrosis and he-patocellular carcinoma (1).

In the present study, characteristicgene expression patterns were observedin HCV cirrhosis and alcoholic cirrhosis.Comparisons between normal liver tis-sues and HCV-cirrhotic and alcohol- cirrhotic liver tissues showed the princi-pal pathways and molecular and cellular

functions affected in each condition.Also, a direct comparison analysis wasperformed between the cirrhotic tissueswith the two different etiologies. More-over, the direct effect of the interactionbetween alcohol and HCV in the livertissue permitted the identification ofgenes synergistically and antagonisticallyaffected by the combination of alcoholand HCV.

The liver is the primary site of alcoholmetabolism, and therefore is vulnerableto severe, chronic injury. Chronic heavyalcohol consumption damages the liver(as manifested by fatty liver, steatosis, fi-brosis, cirrhosis and hepatic cancer) andimpairs the liver’s ability to regenerate(1-4). Fibrosis of the liver is a wound-healing process that occurs in responseto persistent hepatocellular injury. Al-though the major mechanisms of fibroge-nesis are independent of the origin ofliver injury, alcoholic liver fibrosis fea-tures several special characteristics. Animportant aspect in alcoholic liver dis-ease is the pronounced inflammatory re-sponse of Kupffer cells and other typesof leukocytes (macrophages, neutrophils,lymphocytes).

All mature liver cells, including hepa-tocytes, Kupffer cells, hepatic stellatecells and endothelial cells, as well asoval stem cells, have been implicated inalcohol-induced liver damage and im-paired regeneration. Acetaldehyde, animmediate metabolite of ethanol, is pri-marily produced in hepatocytes and canactivate hepatic stellate cells (HSC) in aparacrine manner. Activation of HSC isthe primary event that triggers the pro-cess of fibrogenesis. Kupffer cells havebeen implicated as mediators of alco-holic liver injury through the release offree radicals as well as generation of in-flammatory and fibrogenic mediators inresponse to alcohol and lipopolysaccha-ride (1–4). In the present study, the topoverexpressed genes that were identi-fied as associated with alcoholic cirrho-sis compared with normal livers in-cluded genes involved in antigenpresentation, inflammatory responseand growth factors.



Figure 5. Significant interactions. For CASP1, BCL2, TRADD and BAD genes, interaction plotswere constructed to enable visualization of the expression changes based on the HCVstatus (HCV positive versus negative) or alcohol status (yes versus no). In each plot oneline represents alcohol = yes (solid line) and the other line represents alcohol = no(dashed line). Each line is formed by connecting two points: the group mean expressionfor HCV-negative patients to the group mean expression for HCV-positive patients.

Chronic ethanol intake is also associ-ated with accelerated liver damage whenit coexists with conditions such as HCVinfection. Reported data suggest thatHCV is not directly cytopathic to the he-patocyte, and the main focus of pathol-ogy of HCV infection has been on in-

flammation from the immune responseas the main contributor of reactive oxy-gen species from macrophages and neu-trophils (24). Alcohol is considered an ac-celerant of illness in individuals infectedwith HCV (25,26). There is a higher rateof HCV infection among alcoholics than

the general population and there is a40% increased risk of morality in alcoholabusers with HCV (27).

Global gene expression profiling ofliver tissue revealed characteristic por-traits in HCV and ethanol-induced cir-rhosis. In general, gene expression



Table 1. Summary of the more important findings for each analysis.a

Alcohol cirrhotic tissues HCV cirrhotic tissues HCV versus Significant versus normal livers versus normal livers alcohol cirrhotic tissues interaction analysis

Number of differentially 1406/1272 2222/1841 2172/1896 1230/1051expressed probe sets/genes

Associated Network Functions (five top-scored networks)

Top molecular and cellular functions

Continued

1. Protein degradation,protein synthesis,immunological disease(score 37);

2. Gene expression,cancer, infectiousdisease (score 37);

3. Cell morphology,tissue morphology,RNA damage andrepair (score 35);

4. Cell cycle, cancer,gastrointestinaldisease (score 35);

5. Cell cycle, cellulardevelopment, skeletaland muscular systemdevelopment andfunction (score 35)

Cellular growth andproliferation (P 1.47E-04to 4.24E-02), 314 genes;

Gene expression (P 1.92E-04 to 3.81E-02), 236 genes;

Cell Cycle (P 4.81E-04to 4.42E-02) 148 genes;

Amino acid metabolism(P 7.68E-04 to 4.51E- 02),54 genes;

Posttranslationalmodification (P 7.68E-04to 4.38E-02), 135 genes

1. Molecular transport,connective tissuedisorders,dermatologicaldiseases andconditions (score 43);

2. Cell death, moleculartransport, cellularassembly andorganization (score 40);

3. Lymphoid tissuestructure anddevelopment, organdevelopment, cellulardevelopment (score37);

4. Gene expression, lipidmetabolism, moleculartransport (score 37);

5. Cancer,immunologicaldisease, organismaldevelopment (score 35)


Cell death (P 2.80E-13to 3.90E-03), 361 genes;

Cellular movement (P 2.88E-12 to 3.83E-03), 211 genes;

Cellular development (P 1.76E- 09 to 3.76E-03),260 genes;

Cell-to-cell signaling andinteraction (P 6.38E-08to 4.26E-03), 248 genes;

1. Lipid metabolism,small moleculebiochemistry,molecular transport(score 42);

2. Protein synthesis,infection mechanism,nervous systemdevelopment andfunction (score 39);

3. Gene expression,connective tissuedevelopment andfunction, skeletal andmuscular systemdevelopment andfunction (score 37);

4. Gene expression, lipidmetabolism, smallmolecule biochemistry(score 37);

5. Molecular transport,RNA posttranscriptionalmodification, RNAtrafficking (score 34)

Gene expression (P 1.49E-09 to 1.21E-02),285 genes;

RNA posttranscriptionalmodification; (P 2.50E-09 to 7.19E-05), 51 genes;

Cell death (P 1.35E-07to 1.21E-02), 326 genes;


Protein synthesis (P 2.47E-06 to 5.74E- 03),107 genes

1. Genetic disorder,hematologicaldisease, cell cycle(score 45);

2. Cancer, cell death,reproductive systemdisease (score 40);

3. Lipid metabolism,small-moleculebiochemistry, cellularmovement (score 38);

4. Hematologicaldisease, geneticdisorder, cellularassembly andorganization (score 38);

5. Protein trafficking,developmentaldisorder, geneticdisorder (score 38)

Cell death (P 1.00E-08to 2.19E-02), 228genes;


Cellular compromise (P 2.88E-05 to 1.17E-02),14 genes;

Cell morphology (P 3.36E-05 to 2.21E-02),144 genes;

Cell-to-cell signalingand interaction (P 6.79E-05 to 1.63E-02)96 genes

changes in cirrhotic livers were more rel-evant in livers with HCV infection thanin those with alcohol-induced diseasewhen both were compared with normallivers. Lederer et al. (28), in a recentstudy, performed global transcriptionalprofiling using oligonucleotide microar-rays on liver biopsy samples from pa-tients with cirrhosis caused by eitherchronic alcohol consumption or HCV in-fection. As the authors discussed in thearticle, this was a small cross-sectionalstudy, but the gene expression profiles incirrhotic livers from alcoholic liver dis-ease were distinct from the gene expres-sion profiles of HCV-induced cirrhosis.These results are in concordance with

our findings. Moreover, in HCV-cirrhotictissues, we observed upregulation ofgenes that might reflect the antiviral re-sponse of liver hepatocytes induced bychronic HCV infection. Previous studies(15,17) showed overexpression of severalof these genes, including IFI27, ISG15,MX1, IL8 and IFIT1.

In the present study we included animportant number of samples per groupand moreover, we studied the interactionof HCV and alcohol in cirrhosis induc-tion. Little is known about the combinedeffects of ethanol and HCV on the patho-genesis of liver disease. Proposed mecha-nisms of interaction include enhance-ment of viral replication and increased

oxidative stress and cytotoxicity, as wellas impairment of immune response (29).

Both HCV and alcohol are knownstimuli of hepatic oxidative stress andlipid peroxidation, suggesting that coex-istence of these factors might enhancethese pathways (30), thereby leading toincreased activation of liver fibrogeniccells and subsequent acceleration of fi-brogenesis. Thus, chronic administrationof alcohol to HCV-core transgenic miceresulted in additive hepatic lipid peroxi-dation, synergistic induction of the profi-brogenic cytokine TGF-β1 and activationof HSC (31). Experimental data suggestthat alcohol-induced impairment of im-munity may account for the high rates of



Table 1. Continued.

Top canonical pathways

Top overexpressed genes (fold changes)

Top underexpressed genes (fold changes)

aAnalysis performed by using Ingenuity Pathways Analysis tools 7.0 (http://www.ingenuity.com).

JAK/Stat signaling (P =5.39E-04),

Protein ubiquitinationpathway (P = 8.22E-04),

VEGF signaling (P =1.06E-03)

NTN3 (19.62)IGKC (11.95)IGL@ (10.56)HLA-C (7.08)IGHM (6.8)HBB (includes EG:3043)(6.65)

ZCWPW2 (includesEG:152098) (5.73)

MAFF (5.51)IL8 (5.4)HBA1 (5.26)S100A8 (–5.57)FCN2 (–3.58)RFC5 (–3.54)SAA2 (–3.47)TFPI (–3.30)ST6GAL1 (–3.1)KCNN2 (–3.1)ABCA1 (–2.99)TGM2 (–2.97)LMAN1 (–2.95)

Hepatic fibrosis/hepaticstellate cell activation(P = 9.17E-06),

Phospholipase Csignaling (P = 1.62E- 05),

G-protein–coupledreceptor signaling (P =3.46E-05),

Thrombin signaling (P =4.21E-05)

IFI27 (21.72)NTN3 (19.62)IGKC (16.04)CXCL10 (10.89)IGL@ (10.56)HLA-DQB1 (9.98)IFIT1 (9.07)AKR1B10 (8.09)HBB (includes EG:3043)(8.01)

IGHM (7.68)

LPA (–4.5)ATF5 (–4.49)STEAP1 (–3.8)PZP (–3.63)SAA2 (–3.47)MT1M (–3.45)TFPI (–3.3)DHODH (–3.3)SLCO4C1 (–3.2)KCNN2 (–3.1)

Death-receptorsignaling (P = 1.12E-04),

ILK signaling (P = 1.82E-04),

B-cell–receptor signaling(P = 6.3E-04)

IFI27 (11.7)ISG15 (5.73)CXCL10 (5.68)IFIT1 (5.13)MX1 (4.09)HSP90B1 (4.06)ID1 (3.73)ISG20 (3.66)IFI44 (3.33)MCL1 (3.27)

BBOX1 (–4.75)ATP8B1 (–4.0)SLC35E1 (–2.55)PRG2 (includesEG:79948) (–2.46)

RBM25 (includesEG:58517) (–2.41)

DBP (–2.39)ZC3H7Ba (–2.37)POLR1B (–2.26)UBB (–2.21)FGFR1 (–2.20)

Antigen-presentationpathway (P = 7.92E-07),

IL-4 signaling (P =6.92E-03),

Riboflavin metabolism(P = 1.27E-02),

Hepatic fibrosis/hepaticstellate cell activation(P = 1.32E-02 )coagulation system (P = 2.19E-02)

Not applicable

Not applicable

persistent viral infection reported in ex-cessive drinkers (32).

In the present study, we observed thatthe interaction of HCV and alcohol wascharacterized by downregulation ofmany genes involved in the immune re-sponse. Acute alcohol ingestion results ingeneralized suppression of the immunesystem (33). The antigen-specific CTLthat are produced in response to HCV in-fection migrate to the liver, where theycause apoptosis of infected hepatocytesvia interaction of Fas ligand expressedon activated CTL and Fas on hepatocytes(34,35), and this action may be enhancedby alcohol (36).

According to our analysis, the topcanonical pathways affected for interac-tion of HCV and ethanol were antigenpresentation, IL-4 signaling, hepatic fi-brosis/HSC activation, and coagulation.Evidently, fibrosis is exacerbated fromthe interaction of both etiologies. Specificanalysis of the hepatic fibrosis/HSC acti-vation canonical pathway revealed thatthe principal genes were BAX, BCL2,COL1A2, CTGF, FGFR1, FGR2, MMP2and PDGFRB, among others.

Connective tissue growth factor(CTGF = CCN2), has been recognized asan important player in fibrogenic path-ways on the basis of findings in non-hepatic tissues and emerging resultsfrom liver fibrosis. By changing the ac-tivity ratio of TGF-β to its antagonist,bone-morphogenetic protein 7, CTGFhas been proposed as a fibrogenic mas-ter switch for epithelial-mesenchymaltransition (37).

HSC are known to synthesize the ex-cess matrix that characterizes liver fibro-sis and cirrhosis. Activated HSC expressthe matrix-degrading MMP enzymes andthe tissue inhibitors of metalloproteinases(TIMPs). During spontaneous recoveryfrom experimental liver fibrosis, the ex-pression of TIMP-1 declines and hepaticcollagenolytic activity increases. This pro-cess is accompanied by HSC apoptosis. Ina recent publication, Hartland et al. (38)reported a role for both N-cadherin andMMP-2 in mediating HSC apoptosis, inwhich N-cadherin works to provide

cell-survival stimulus and MMP-2 pro-motes HSC apoptosis concomitant withN-cadherin degradation. In the presentstudy, MMP2 and CDH1 were negativelyaffected by the interaction of HCV and al-cohol as well as macrophage-derivedMMP-7 and TIMP3.

In a recent publication, Kendall et al.(39) suggested that cleavage of proNGFby MMP7 during the early phase of re-covery from liver fibrosis alters thepro/mature NGF balance to facilitate he-patic myofibroblast loss. Whereas fibro-sis develops in the absence of p75(NTR)signaling, the dominant effects of loss ofp75(NTR) ligand–mediated events arethe retardation of liver fibrosis resolutionvia regulation of hepatic myofibroblastproliferation and apoptosis, and the re-duction of hepatocyte- and oval-cell proliferation.

Apoptosis was also affected by theinteraction, because it was well reflectedfor the negative effect in CASP1 by theinteraction and it was overexpressed inHCV cirrhotic tissues compared withboth normal and alcoholic cirrhoticliver tissue. Similarly, TNFRSF21 andBCL2 were negatively affected for theinteraction.

We are aware of the limitation that cel-lular heterogeneity of tissue samples in-troduces to the analysis. Liver tissuesamples used for microarray analysiscontained a mixture of different cellsand cell infiltrates. Thus, with the excep-tion of cell-specific genes (for example,E-selectin), the source of mRNA was un-known, and this limited our ability to in-terpret the cellular signatures. However,in this study we aimed to evaluate theglobal differences that characterize cir-rhosis related to alcohol versus HCV,and moreover, how the interaction ofboth etiologies affects the overall molec-ular pathways.

In conclusion, according to our results,the characteristic profiles of the tissuesunder the additive/synergic effect ofHCV and alcohol were characterized byan increase in fibrosis development andapoptosis inhibition and a decrease inimmune response. These initial findings

will contribute to further studies to dis-tinguish which pathways would be bet-ter targets for new therapeutic alterna-tives for end-stage liver disease.

ACKNOWLEDGMENTSThis project was partially supported

by the Commonwealth Foundation forCancer Research as a pilot project of theMassey Cancer Center.

DISCLOSUREThe authors declare that they have no

competing interests as defined by Molecu-lar Medicine, or other interests that mightbe perceived to influence the results anddiscussion reported in this paper.

REFERENCES1. Mueller S, Millonig G, Seitz HK. (2009) Alcoholic

liver disease and hepatitis C: a frequently under-estimated combination. World J. Gastroenterol.15:3462–71.

2. Singal AK, Anand BS. Mechanisms of synergy be-tween alcohol and hepatitis C virus (2007). J Clin.Gastroenterol. 41:761–72.

3. Bhattacharya R, Shuhart MC. (2003) Hepatitis Cand alcohol: interactions, outcomes, and implica-tions. J. Clin. Gastroenterol. 36:242–52.

4. Gitto S, Micco L, Conti F, Andreone P, BernardiM (2009). Alcohol and viral hepatitis: a mini-re-view. Dig. Liver Dis. 41:67–70.

5. Anderson S, Nevins CL, Green LK, El-Zimaity H,Anand BS (2001). Assessment of liver histologyin chronic alcoholics with and without hepatitisC virus infection. Dig. Dis. Sci. 46:1393–98.

6. Tamai T, et al. (2000) Effects of alcohol consump-tion on histological changes in chronic hepatitisC: a clinicopathological study. Alcohol Clin. Exp.Res. 24:106S–11S.

7. Tanaka T, et al. (2000) Contribution of hepatitis Cvirus to the progression of alcoholic liver disease.Alcohol Clin. Exp. Res. 24:112S–116S.

8. Anand BS, Velez M. (2000) Influence of chronicalcohol abuse on hepatitis C virus replication.Dig. Dis. 18:168–71.

9. Thibault C, et al. (2000) Expression profiling ofneural cells reveals specific patterns of ethanol-responsive gene expression. Mol. Pharmacol.58:1593–600.

10. Hassan S, Duong B, Kim KS, Miles MF. (2003)Pharmacogenomic analysis of mechanisms medi-ating ethanol regulation of dopamine beta- hydroxylase. J. Biol. Chem. 278:29960–9.

11. Rimondini R, Arlinde C, Sommer W, Heilig M.(2002) Long-lasting increase in voluntary ethanolconsumption and transcriptional regulation inthe rat brain after intermittent exposure to alco-hol. FASEB J. 16:27–35.



12. Saito M, Smiley J, Toth R, Vadas C. (2002) Mi-croarray analysis of gene expression in rat hippocampus after chronic ethanol treatment.Neurochem. Res. 27:1221–9.

13. Kerns RT, et al. (2005) Ethanol-responsive brainregion expression networks: implications for be-havioral responses to acute ethanol in DBA/2Jversus C57BL/6J mice. J. Neurosci. 25:2255–66.

14. Lewohl JM., et al. (2000) Gene expression inhuman alcoholism: microarray analysis of frontalcortex. Alcohol Clin. Exp. Res. 24:1873–82.

15. Smith MW, et al. (2003) Hepatitis C virus andliver disease: global transcriptional profiling andidentification of potential markers. Hepatology.38:1458–67.

16. Shackel NA, McGuinness PH, Abbott CA, Gor-rell MD, McCaughan GW. (2002) Insights into thepathobiology of hepatitis C virus-associated cir-rhosis: analysis of intrahepatic differential geneexpression. Am. J. Pathol. 160:641–54.

17. Bièche I, et al. (2005) Molecular profiling of earlystage liver fibrosis in patients with chronic hepa-titis C virus infection. Virology. 332:130–44.

18. Knodell RG, et al. (1981) Formulation and appli-cation of a numerical scoring system for assess-ing histological activity in asymptomatic chronicactive hepatitis. Hepatology. 1:431–5.

19. Gentleman RC, et al. (2004) Bioconductor: opensoftware development for computational biologyand bioinformatics. Genome Biol. 5:R30.

20. R Development Core Team. R: a language andenvironment for statistical computing [computerprogram]. Vienna (Austria): R Foundation forStatistical Computing; 2007.

21. Swiatkowska-Stodulska R, Bakowska A,Drobinska-Jurowiecka A (2006). Interleukin-8 inthe blood serum of patients with alcoholic liverdisease. Med. Sci. Monit. 12:CR215–20.

22. Lemmers A, et al. (2009) The interleukin-17 path-way is involved in human alcoholic liver disease.Hepatology. 49:646–57.

23. Sharma A, Chakraborti A, Das A, Dhiman RK,Chawla Y. (2009) Elevation of interleukin-18 inchronic hepatitis C: implications for hepatitis Cvirus pathogenesis. Immunology. 128:e514–22.

24. Loguercio C, Federico A. (2003) Oxidative stressin viral and alcoholic hepatitis. Free Radic. Biol.Med. 34:1–10.

25. Safdar K, Schiff ER. (2008) Alcohol and hepatitisC. Semin. Liver Dis. 24:305–15.

26. Mallat A, Hezode C, Lotersztajn S. (2008) Envi-ronmental factors as disease accelerators duringchronic hepatitis C. J. Hepatol. 48:657–65.

27. Lederer SL, et al. (2006) Distinct cellular re-sponses differentiating alcohol- and hepatitis Cvirus-induced liver cirrhosis. Virol. J. 3:98.

28. Jamal MM, Morgan TR (2003). Liver disease inalcohol and hepatitis C. Best Pract. Res. Clin. Gas-troenterol. 17:649–62.

29. Otani K, et al. (2005) Hepatitis C virus core pro-tein, cytochrome P450 2E1, and alcohol producecombined mitochondrial injury and cytotoxicityin hepatoma cells. Gastroenterology. 128:96–107.

30. Perlemuter G, et al. (2003) Alcohol and hepatitis Cvirus core protein additively increase lipid peroxi-dation and synergistically trigger hepatic cytokineexpression in a transgenic mouse model. J. Hepa-tol. 39:1020–27.

31. Piasecki BA, et al. (2004) Influence of alcohol use,race, and viral coinfections on spontaneous HCVclearance in a US veteran population, Hepatology.40:892–99.

32. Cook RT. (1998) Alcohol abuse, alcoholism anddamage to the immune system-a review. AlcoholClin. Exp. Res. 22:1927–42.

33. Cerny A, Chisari FV. (1999) Pathogenesis ofchronic hepatitis C: immunological features ofhepatic injury and viral persistence. Hepatology.30:595–601.

34. Piasecki BA, et al. (2004) Influence of alcohol use,race, and viral coinfections on spontaneous HCVclearance in a US veteran population, Hepatology.40:892–9.

35. Cook RT. (1998) Alcohol abuse, alcoholism anddamage to the immune system: a review. AlcoholClin. Exp. Res. 22:1927–42.

36. Pianko S, Patella S, Ostapowicz G. (2001) Fas me-diated hepatocyte apoptosis is increased by hep-atitis C virus infection and alcohol consumptionand may be associated with hepatic fibrosis:mechanisms of liver cell injury in chronic hepati-tis C virus infection. J Viral Hepat. 8:406–13.

37. Gressner OA, Gressner AM. (2008) Connectivetissue growth factor: a fibrogenic master switchin fibrotic liver diseases. Liver Int. 28:1065–79.

38. Hartland SN, et al. (2009) Active matrix metallo-proteinase-2 promotes apoptosis of hepatic stel-late cells via the cleavage of cellular N-cadherin.Liver Int. 29:966–78.

39. Kendall TJ, et al. (2009) Neurotrophin receptorsignaling regulates hepatic myofibroblast prolif-eration and apoptosis in recovery from rodentliver fibrosis. Hepatology. 49:901–10.



molecular mechanisms involved in the interaction effects

Documents