Hepatic Metallothionein Expression in Chronic Hepatitis C Virus Infection Is IFNL3 Genotype-Dependent

ORIGINAL ARTICLE Hepatic metallothionein expression in chronic hepatitis C virus infection is IFNL3 genotype-dependent

KS O’Connor1, G Parnell1, E Patrick2, G Ahlenstiel3, V Suppiah1,3, D van der Poorten3, SA Read3,4, R Leung1,3, MW Douglas3,4, JYH Yang2, GJ Stewart1, C Liddle3, J George3 and DR Booth1

The IFNL3 genotype predicts the clearance of hepatitis C virus (HCV), spontaneously and with interferon (IFN)-based therapy. The responder genotype is associated with lower expression of interferon stimulated genes (ISGs) in liver biopsies from chronic hepatitis C patients. However, ISGs represent many interacting molecular pathways, and we hypothesised that the IFNL3 genotype may produce a characteristic pattern of ISG expression explaining the effect of genotype on viral clearance. For the first time, we identified an association between a cluster of ISGs, the metallothioneins (MTs) and IFNL3 genotype. Importantly, MTs were significantly upregulated (in contrast to most other ISGs) in HCV-infected liver biopsies of rs8099917 responders. An association between lower fibrosis scores and higher MT levels was demonstrated underlying clinical relevance of this association. As expected, overall ISGs were significantly downregulated in biopsies from subjects with the IFNL3 rs8099917 responder genotype (P ¼ 2.38 Â 10 À 7). Peripheral blood analysis revealed paradoxical and not previously described findings with upregulation of ISGs seen in the responder genotype (P ¼ 1.00 Â 10 À 4). The higher MT expression in responders may contribute to their improved viral clearance and MT-inducing agents may be useful adjuncts to therapy for HCV. Upregulation of immune cell ISGs in responders may also contribute to the IFNL3 genotype effect.

Genes and Immunity (2014) 15, 88–94; doi:10.1038/gene.2013.66; published online 16 January 2014 Keywords: IFNL3; metallothioneins; hepatitis C virus; interferon-stimulated genes

INTRODUCTION Type I and type III IFNs upregulate the same set of ISGs but 15,16 Approximately 3% of the world’s population has been infected signal through different receptor complexes. However, the with hepatitis C virus (HCV).1 With spontaneous clearance rates of level and timing of upregulation of immediate, early and late ISGs 9,15 only 20–30%, a substantial proportion of these individuals are at appears to be affected by the type of IFN. Although signalling risk of becoming chronic carriers of the virus and developing long- through the same receptor, the various Type 1 IFNs can have very term sequelae, including cirrhosis and hepatocellular carcinoma.2 different outcomes on ISG response in vivo and in vitro. This has The aim of HCV therapy is to achieve a sustained virological been attributed to the different avidity and kinetics of their binding response (SVR), defined as undetectable 6-month post-treatment to the receptor, interacting with different receptor densities and 17 HCV RNA in plasma. A significant number of individuals will fail to other microenvironmental factors. Factors controlling ISG achieve an SVR or develop major side effects from therapy. patterns in response to Type III IFNs have yet to be established, Genome-wide association studies have identified single- as well as how these patterns are affected by context. nucleotide polymorphisms in the vicinity of IFNL3 that are We sought to determine a hepatic ISG signature in HCV-infected predictive of response to pegylated interferon (IFN) alpha and liver biopsies and peripheral blood that is associated with IFNL3 ribavirin therapy in patients with HCV genotype 13–5 and haplotypes. We used the comparison of two methods, gene genotype 4.6 expression microarray and RNA high-throughput sequencing IFNL3 encodes IFN-lambda-3, a member of the type III IFN family (RNA-Seq), to improve specificity in detecting the most signifi- (INFL1, IFNL2 and IFNL3) that are functionally closely related to cantly associated ISGs and confirmed these findings by quantita- Type I IFNs.7,8 Both Type I (for example, IFNA) and Type III IFNs tive PCR (qPCR). induce interferon stimulated genes (ISGs) in HCV-infected cells.9 Before genome-wide association studies, studies had demonstrated that individuals with a non-SVR (NSVR) have RESULTS higher pre-treatment hepatic ISG expression.10,11 Notably, it has Association of IFNL3 genotype and ISG expression in whole blood subsequently been reported that there is a strong association and in liver biopsies between the IFNL3 unfavourable genotype and higher pre- In blood 182 ISGs were detected and in infected liver biopsies 338 treatment hepatic ISG levels.12–14 There is even evidence ISGs were detected. suggesting that hepatic ISG expression may be a better Expression microarray analysis of 32 whole-blood samples predictor of SVR than IFNL3 genotype.12,13 (Cohort 1) revealed a higher expression of ISGs from the 17

1Institute for Immunology and Allergy Research, Westmead Millennium Institute, University of Sydney, Sydney, New South Wales, Australia; 2Department of Mathematics, University of Sydney, Sydney, New South Wales, Australia; 3Storr Liver Unit, Westmead Millennium Institute and Westmead Hospital, University of Sydney, Sydney, New South Wales, Australia and 4Centre for Infectious Diseases and Microbiology, Sydney Emerging infections and Biosecurity Institute, University of Sydney and Westmead Hospital, Sydney, New South Wales, Australia. Correspondence: Dr KS O’Connor, Institute for Immunology and Allergy Research, Westmead Millennium Institute, University of Sydney, Sydney, 2145 New South Wales, Australia. E-mail: [email protected] Received 3 October 2013; revised 11 November 2013; accepted 12 November 2013; published online 16 January 2014 Hepatic MT expression in chronic HCV infection KS O’Connor et al 89 patients who had failed to clear HCV after treatment than 15 who trend towards higher expression in the TG/GG non-responder had cleared virus with pegylated IFN alpha and ribavirin (Table 1; genotype (Figure 1d). In contrast to the liver biopsy samples, P ¼ 1.15 Â 10 À 8 by the Sign test). The HCV-infected blood samples MT1E, MT1G, MT1H or MT1M were not detected by gene (n ¼ 17) exhibited higher ISGs in the favourable rs8099917 expression microarray analysis in peripheral blood. genotype (Table 1; P ¼ 1.00 Â 10 À 4). For rs12979860 a non- significant trend in the same direction was also observed with a greater number of upregulated ISGs seen in the responder Confirmation of metallothionein significance by RNA-Seq and qPCR genotype (CC n ¼ 100) versus the non-responder genotype (TC/TT n ¼ 82). The converse was seen on microarray expression Figure 2 demonstrates t-test values for ISGs from the microarray analysis of the 22 HCV-infected liver biopsies (Cohort 2) examined, cohort plotted against the RNA-Seq subjects (Cohort 3) for rs8099917 TT responders. As expected, the majority of ISGs with highly significant differences demonstrated (Table 1; P ¼ 2.38 À 13 Â 10 À 7). Thus, a greater number of upregulated ISGs were (Po7.8 Â 10 ) plot into the left lower quadrant consistent with detected in the unfavourable genotypes for rs8099917. However, downregulation of ISGs seen in the responder genotype. The right no significant difference was observed for rs12979860 (upregu- upper quadrant represents ISGs upregulated in the responder lated ISGs CC n ¼ 177 and TC/TT n ¼ 161). genotype and a cluster of five MTs are demonstrated (Figure 2). To validate the association between MT expression and rs809917 genotype, we employed qPCR, as a third method, in a Metallothioneins are associated with IFNL3 rs8099917 genotype new cohort of liver biopsy samples (Cohort 4, n ¼ 37) and peripheral blood (Cohort 5, n ¼ 24) samples. MT1F and MT1G Expression microarray analysis of 22 HCV-infected liver biopsies were selected as candidate MTs for qPCR analysis based on earlier (Cohort 2) revealed a statistically significant difference (Po0.05) in findings. Results correlated with RNA-Seq and microarray findings ISG expression between rs8099917 responder (TT) and non- with higher expression of both MT1F and MT1G in liver biopsy responder (TG/GG) genotypes for 22 of the ISGs analysed samples of TT responder genotype (Figure 3) although these (Table 2). Twelve of these showed higher expression and 10 results did not reach statistical significance. There was, however, revealed lower expression in the TT responder genotype. The three most significantly upregulated genes were metallothioneins (MTs), isoforms MT1F, MT1G and MT1H. MT1X and MT1M also featured amongst these 12 genes (Table 2). Given this cluster of Table 2. Correlation of IFNL3 rs8099917 genotype and level of MTs, we extended our analysis to all MTs including those that are expression of ISGs showing significant differences by microarray not in the original ISG list. In liver tissue seven MT1 isoforms were analysis detectable as well as MT2A. All MTs demonstrated higher Fold change TT P-value expression levels in the rs8099917 TT responder genotype and versus TG and GG this was statistically significant in all but two MT genes (Figure1a; Po0.05 for MT1A, MT1H and MT1X and Po0.005 in MT1F and Upregulated rs8099917 MT1G). For rs12979860 a non-significant trend in the same MT1F 1.43 0.0002 direction was observed for all MTs examined with higher MT1G 1.78 0.0005 expression levels seen in the CC responder genotype (data not MT1H 1.69 0.0009 shown). Higher MT levels were also seen for SVRs compared with APOL3 1.086 0.0016 NSVRs, although this only reached statistical significance for MT2A MASTL 1.54 0.0041 MT1X 1.47 0.0047 (Figure 1b; Po0.05). MT1M 1.62 0.0069 MT1A, MT1F, MT1X and MT2A microarray expression data were SLC25A30 1.99 0.0109 available for analysis from peripheral blood microarray data SAA1 1.85 0.0218 (Cohort 1). Comparing HCV-infected subjects (NSVRs) versus ABTB2 1.42 0.0267 those that had cleared virus (SVRs) we detected higher levels SERPINB9 2.70 0.0300 in the infected subjects for all MT1 isoforms and a significantly MAX 1.01 0.0309 higher level for MT2A (Figure 1c; Po0.05). Examining MTs in the HCV-infected samples (that is, NSVRs) revealed a non-significant Downregulated in rs8099917 TT DUSP5 À 1.30 0.0104 TXNIP À 1.47 0.0137 PRAME À 1.15 0.0206 Table 1. Expression microarray analysis demonstrating the number of FAM46A À 1.19 0.0224 upregulated ISGs in peripheral blood (n ¼ 32) and HCV-infected liver GZMB À 1.56 0.028 biopsies (n ¼ 22) IRF7 À 1.10 0.0309 CD274 À 1.06 0.0344 Genotype Number of P-value MAP3K14 À 1.43 0.0380 upregulated ATF3 À 1.08 0.0418 ISGs CES1 À 1.16 0.0427

Peripheral blood (Cohort 1) Abbreviations: ABTB2, ankyrin repeat and BTB (POZ) domain containing 2; APOL3, apolipoprotein L 3; ATF3, activating transcription factor 3; CD274, HCV infected (NSVRs) — 130 1.15 Â 10 À 8 CD274 molecule; CES1, carboxylesterase 1; DUSP5, dual speciﬁcity Cleared HCV (SVRs) — 52 phosphatase 5; FAM46A, family with sequence similarity 46 member A; rs8099917 (NSVRs only) TT 117 1.00 Â 10 À 4 IRF7, interferon regulatory factor 7; ISGs, interferon stimulated genes; TG/GG 65 GZMB, granzyme B; MASTL, microtubule-associated serine/threonine HCV-infected liver biopsies (Cohort 2) kinase-like; MAP3K14, mitogen-activated protein kinase 14; MAX, MYC- associated factor X member 9; MT1F, metallothionein 1F; MT1G, rs8099917 TT 121 2.38 Â 10 À 7 TG/GG 217 metallothionein 1G; MT1H, metallothionein 1H; MT1X, metallothionein 1X; MT1M, metallothionein 1M; PRAME, preferentially expressed antigen in Abbreviations: HCV, hepatitis C virus; ISGs, interferon stimulated genes; melanoma; SLC25A30, solute carrier family 25 member 30; SAA1, serum NSVRs, non-sustained virological responses; SVRs, sustained virological amyloid A1; SERPINB9, serpin peptidase inhibitor clade B (ovalbumin); responses. TXNIP, thioredoxin-interacting protein.

& 2014 Macmillan Publishers Limited Genes and Immunity (2014) 88 – 94 Hepatic MT expression in chronic HCV infection KS O’Connor et al 90

Figure 1. Expression levels by microarray for MTs in liver biopsy samples and peripheral blood from HCV-infected subjects. (a) Analysed by rs8099917 genotype for liver biopsies (n ¼ 22). All MT1 isoforms and MT2A exhibit higher expression in the rs8099917 TT responder genotype. This is significant for MT1A, MT1H and MT1X (Po0.05) and MT1F and MT1G (Po0.005). (b) NSVRs versus SVRs for liver biopsies. SVRs demonstrate higher pre-treatment levels of all MT1 isoforms and MT2A (Po0.05). (c) Peripheral blood analysis for HCV-positive subjects (n ¼ 17) versus subjects that have cleared HCV (n ¼ 15). A trend towards higher expression is seen in the HCV-positive subjects for MT1 isoforms (not significant) and MT2A (Po0.05). (d) Peripheral blood analysis for HCV-positive subjects (n ¼ 18) for differences in rs8099917 genotype. Results show a non-significant trend towards higher expression in the TG/GG non-responder genotype. By gene expression microarray we were unable to detect MT1E, MT1G, MT1H or MT1M in peripheral blood.

trend test. Results were signiﬁcant for liver biopsies (MT1F P ¼ 0.0068 and MT1G P ¼ 0.0062). However, combined results for peripheral blood samples (microarray and qPCR) did not reach statistical signiﬁcance.

MT expression and liver fibrosis We next examined for an association between liver fibrosis and MT expression. All liver biopsy samples (Cohorts 2–4) were divided into high (METAVIR or Scheuer fibrosis score X2) or low fibrosis (METAVIR or Sheuer fibrosis score p1). qPCR data demonstrated a significant association between liver fibrosis and MT expression with higher MT expression observed in the low-fibrosis samples (MT1F Po0.005 and MT1Go0.005) (Figure 4a). A non-significant trend for higher MT1 isoforms in the low-fibrosis samples was also observed for the microarray and RNA-Seq liver biopsy samples Figure 2. Independent comparisons RNA-Seq and microarray of (Figures 4b and c). Importantly, there were no significant expression of ISGs (open circles) in HCV-infected liver biopsies. The differences in fibrosis between IFNL3 groups to explain the upper right quadrant represents genes high in IFNL3 responder genotype effect detected for MTs (Supplementary Table 2). haplotypes (rs8099917 TT) in both cohorts, lower left quadrant represents genes low in both. Most genes are in the left lower quadrant. Five MTs are upregulated in TT genotypes (red circles). Axes are t-test values for a microarray cohort plotted against values DISCUSSION in an RNA-Seq cohort. A Wilcoxon rank sum test was performed to In contrast to previous studies, we identified a set of ISGs, MTs, determine if the set of ISG genes were differentially expressed as a which are upregulated, not downregulated like most other ISGs, in set. The two-sided test gave a P-value of 7.78 Â 10 À 13. The one- the IFNL3 responder genotype and among patients who achieve an sided test where the expression of TT is greater that GT is À 13 SVR. This significant association between upregulated hepatic MTs 3.89 Â 10 . and rs8099917 IFNL3 favourable genotype, has not been previously reported. Three methods (microarray, RNA-Seq and qPCR) and strong correlation between MT1F and MT1G expression different HCV patient cohorts were used to confirm these findings, (r ¼ 0.7135; Po0.0001). We did not detect MT1G in peripheral suggesting that the results are biologically valid and likely to be blood by qPCR but, MT1F expression clearly correlated with pathogenically important. The weaker associations with rs12979860 microarray findings with higher expression in the TG/GG (non- single-nucleotide polymorphism were expected, as half of the responder) genotype (Figure 3). ‘non-responder T’ haplotypes are actually of neutral effect.18,19 We combined the statistical associations between the three In line with previously published work12–14 we have individual cohorts (microarray, RNA-Seq and qPCR) using Fisher’s demonstrated, using expression microarrays and RNA-Seq of

Genes and Immunity (2014) 88 – 94 & 2014 Macmillan Publishers Limited Hepatic MT expression in chronic HCV infection KS O’Connor et al 91

Figure 3. qPCR for MT expression for HCV liver biopsies and peripheral blood. Panel (a) shows expression of MT1F and MT1G for liver biopsy samples (n ¼ 37). Results correlate with microarray and RNA-Seq data showing higher expression of MTs in the TT (responder genotype), Panel (b) shows linear regression analysis Figure 4. Analysis of MT expression level in low versus high fibrosis between MT1F and MT1G showing a strong correlation (r ¼ 0.7135) liver biopsy samples. Panel (a) performed by qPCR (n ¼ 37), panel and Panel (c) displays peripheral blood HCV samples (n ¼ 24) for (b) by microarray (n ¼ 22) and panel (c) RNA-Seq (n ¼ 26). Low MT1F showing much lower expression than in biopsies. 2 À DCT fibrosis samples exhibit higher MT expression by all three methods. represents: 2 to the power of À delta CT (CT Target gene À CT This is significant for MT1F and MT1G performed by qPCR. À DCT GAPDH). 2 represents: 2 to the power of À delta CT (CT Target gene À CT GAPDH).

HCV-infected liver biopsies, an association between the IFNL3 rapid virological response to treatment, where the innate immune unfavourable genotype (TG/GG) and higher pre-treatment hepatic system is pivotal.22 An association between IFNL3 responder ISG levels. However, we show for the first time that peripheral genotype and spontaneous clearance is well established,23 blood analysis of HCV subjects demonstrates the reverse pattern suggesting that the responder genotypes produce an ISG profile with the highly significant association between the unfavourable capable of inhibiting viral replication in the absence of exogenous genotype and lower ISG levels. We have thus demonstrated IFNA. Thus, despite lower hepatocyte ISGs in the responder paradoxical findings in ISG upregulation between circulating genotypes, there are higher levels of peripheral blood ISGs, immune cells and liver tissue (in which hepatocytes predominate), ongoing immune cell activation and hepatocyte cell death. in HCV-infected patients. Previous work has highlighted a MTs are low-molecular weight, intracellular proteins24 and were potential dichotomy between immune cells and hepatocytes, one of the earliest identified IFN response genes.25 The two major with two studies demonstrating downregulated ISGs by isoforms of MT, (MT1 and MT2) are found in all types of tissues, immunohistochemistry in Kupffer cells (liver macrophages) from whereas the minor isoforms (MT3 and MT4) display restricted individuals who fail to respond to treatment with IFNA.20,21 The tissue expression. MT2 is encoded by a single gene MT2A, whereas elevated peripheral blood ISG response seen with the favourable MT1 comprises multiple subsets encoded by a number of genes IFNL3 genotype in our study is in keeping with its association with including MT1A, MT1B, MT1E, MT1F, MT1G, MT1H, MT1J, MT1K, MT1l,

& 2014 Macmillan Publishers Limited Genes and Immunity (2014) 88 – 94 Hepatic MT expression in chronic HCV infection KS O’Connor et al 92 MT1M and MT1X. Different cell types express different MT isoforms MATERIALS AND METHODS and levels suggesting varied functions.26,27 MTs are induced by Study cohort cytokines (including IFNA), metal ions, growth factors and All samples were collected from HCV genotype 1-infected patients. 28 oxidative stress. MT involvement in gene expression and Samples included: (1) 32 PAXgene Blood RNA tubes from subjects transcription regulation, apoptosis and cell proliferation has following the completion of treatment with pegylated IFN alpha and been previously described.29,30 Functionally, MTs regulate ribavirin (15 of these had achieved SVRs and 17 remained HCV positive intracellular concentrations of zinc and other metal ions31 and (NSVRs)), (2) liver biopsies from 22 individuals that were used for have a role in oxidative stress through their ability to scavenge a microarray analysis, (3) liver biopsies from 26 subjects for RNA-Seq, range of reactive oxygen species. Oxidative stress has been (4) liver biopsies from 37 patients for qPCR and (5) 24 PAXgene Blood RNA tubes from individuals for qPCR. Cohorts 2–5 were made up of subjects postulated to have a role in HCV infection and the development of 32 who were untreated at the time of collection of the samples and each of chronic hepatitis, cirrhosis and HCC. the five cohorts consisted of different individuals. Data were collected on There is some evidence that MT expression may promote liver clinical response to pegylated IFN alpha and ribavirin as well as liver 33 cell apoptosis through NF-kB inhibition in HCV-infected livers, fibrosis scores. Baseline characteristics of the five study cohorts are and thus may counteract the anti-apoptotic effect of the HCV summarised in Supplementary Table 1. protein.34,35 For example, the HCV NS5A protein, with a role in IFN Ethics approval was obtained from the Human Research Ethics resistance, has been shown to activate NF-kB.36 Thus, individuals Committees of Sydney West Area Health Service and the University of carrying the IFNL3 responder genotype, with the higher pre- Sydney. All subjects gave written informed consent (HREC2002/12/ treatment hepatic MT levels identified here, may display enhanced 4.9(1564). hepatocyte apoptosis. In keeping with this hypothesis is the evidence of an association between IFNL3 responder genotype and increased hepatic necroinflammation.37–39 Disruption of NF-kB IFNL3 genotyping activation has been shown to sensitize cells to the pro-apoptotic All subjects were genotyped for the rs8099917 single-nucleotide poly- 4 effects of both IFNA and IFNB.40 Thus, HCV liver biopsies with morphism and rs12979860 by methods previously reported or by Taqman higher MT levels may reflect an enhanced response to IFN-based genotyping (Applied Biosystems, Carlsbad, CA, USA). Genotyping results therapy, as infected hepatocytes may be eliminated more easily via are summarised in Table 3. enhanced apoptosis. Again, this is in keeping with our findings of higher pre-treatment MT levels in the IFNL3 responder genotype, known to have increased SVR rates with IFNA-based therapy. Microarray methods Also consistent with our findings, previous work has demon- For microarrays, messenger RNA (mRNA) was isolated from PAXgene strated a relationship between higher pre-treatment MT liver whole-blood samples according to the manufacturer’s instructions protein levels detected by radioimmunoassay and SVR.41 This (Qiagen, Valencia, CA, USA). Liver mRNA was purified from liver biopsies study also demonstrated a strong inverse correlation between MT using QiagenRNeasy (Qiagen). mRNA was amplified, labelled and 41 hybridised to Illumina Human HT-12_V3 arrays as per the manufacturer’s expression and fibrosis score. A subsequent report revealed instructions (Illumina, San Diego, CA, USA). The average signal intensity for higher MT protein levels (by immunohistochemistry) in liver each gene was measured using Beadstudio v3. The sample signals were biopsies of SVRs compared with NSVRs 12 months after treatment normalized with cubic spline in order to minimize variation owing to non- with IFNA further suggesting that MT may have a role in the biological factors. Expression levels of IFN sensitive genes (ISGs, as defined therapeutic effects of IFNA for HCV.42 In addition, studies have in Schoggins et al.50) with detection P-value less than 0.01 for at least one shown that zinc, a potent MT inducer, increases the therapeutic of all samples were used (the detection P-value measures the probability of response to IFNA in HCV subjects.43,44 More recently, a study observing signals without specific probe-target hybridization). identified the upregulation MT2A 12 h post IFNA as a strong predictor of treatment response.45 In conclusion, we show for the first time that MTs are RNA-Seq methods upregulated in HCV-infected livers from patients with favourable A replication cohort consisting of 26 biopsy samples were assayed using IFNL3 polymorphisms, in an opposite direction to most other ISGs. next-generation RNA-Seq. RNA was isolated as above and prepared for This suggests a potential for a therapeutic role for MT induction in sequencing on Illumina HiSeq 2000 using the Illumina TruSeq RNA sample combination with HCV therapy. The benefits of genetically or preparation kit V1 (Illumina). Samples were indexed and six samples were pharmacologically enhanced MT expression is currently being run per lane, generating at least 10 million 50-bp single-end reads per sample. The raw reads were aligned to hg19 human genome using explored in a wide variety of other disease states including Bowtie.51 These alignments were then summarised as gene counts by diabetes, diabetic complications,46 wound healing,47 traumatic 48 49 counting the number of aligned reads that lay within the exon segments of brain injuries and Alzheimer’s disease. This report supports the annotated genes that did not overlap with the exons of other annotated inclusion of HCV infection in future clinical exploration of the role genes. A cyclic loess strategy was used to normalize differences in library of novel, MT-inducing therapeutic agents. sizes, GC content bias and other variations observed in the data.

Table 3. HCV genotype 1 patient cohorts studied and the results of IFNL3 rs8099917 and rs12979860 genotyping

Cohort n Sample type Assay Treated or untreateda rs8099917b rs12979860c

TT TG/GG CC TC/TT

1 32 Peripheral blood Microarray Treated 14 18 12 20 2 22 Liver Microarray Untreated 13 9 10 12 3 26 Liver RNA-Seq Untreated 12 14 8 18 4 37 Liver qPCR Untreated 22 15 15 22 5 24 Peripheral blood qPCR Untreated 10 14 6 18 Abbreviations: HCV, hepatitis C virus; qPCR, quantitative PCR; RNA-Seq, RNA-sequencing. aTreatment status before collection of sample. bFor rs8099917 TTs represent ‘responders’ and TG/GGs ‘non-responders’: cFor rs12979860 CCs represent ‘responders’ and TC/TT ‘non-responders’.

Genes and Immunity (2014) 88 – 94 & 2014 Macmillan Publishers Limited Hepatic MT expression in chronic HCV infection KS O’Connor et al 93 Gene expression by qPCR 7 Kotenko SV, Gallagher G, Baurin VV, Lewis-Antes A, Shen M, Shah NK et al. IFN- RNA was extracted from liver biopsies and PAX blood using the RNeasy kit lambdas mediate antiviral protection through a distinct class II cytokine receptor (Qiagen) according to the manufacturer’s instructions. RNA quality and complex. Nat Immunol 2003; 4: 69–77. concentration was assessed using the Agilent 2100 Bioanalyser (Agilent, 8 Sheppard P, Kindsvogel W, Xu W, Henderson K, Schlutsmeyer S, Whitmore TE et al. Waldbronn, Germany). cDNA was prepared using qscript (Quanta IL-28, IL-29 and their class II cytokine receptor IL-28R. Nat Immunol 2003; 4: 63–68. Biosciences, Gaithersburg, MD, USA) in a Mastercycler gradient 5331 9 Marcello T, Grakoui A, Barba-Spaeth G, Machlin ES, Kotenko SV, MacDonald MR (Eppendorf AG, Hamburg, Germany). Gene expression was measured by et al. Interferons alpha and lambda inhibit hepatitis C virus replication with qPCR using TaqMan Gene Expression Assays for MT1F (00744661_sH) and distinct signal transduction and gene regulation kinetics. Gastroenterology 2006; MT1G (002578922_gH). GAPDH (4352934) was used as the house keeping 131: 1887–1898. gene (Applied Biosystems). Expression was measured using CT values, 10 Asselah T, Bieche I, Narguet S, Sabbagh A, Laurendeau I, Ripault MP et al. Liver normalized to that of GAPDH (DCT ¼ CT (GAPDH) À CT (target) and then gene expression signature to predict response to pegylated interferon plus À DCT expressed as 2 . CT values were less than 30, and all amplifications ribavirin combination therapy in patients with chronic hepatitis C. Gut 2008; 57: were done in duplicate. 516–524. 11 Sarasin-Filipowicz M, Oakeley EJ, Duong FH, Christen V, Terracciano L, Filipowicz W et al. Interferon signaling and treatment outcome in chronic hepa- Statistics titis C. Proc Natl Acad Sci USA 2008; 105: 7034–7039. GraphPad Prism version 5.0 (GraphPad Software, Inc., La Jolla, San Diego, 12 Dill MT, Duong FH, Vogt JE, Bibert S, Bochud PY, Terracciano L et al. Interferon- CA, USA) was used to perform statistical analysis including: (1) unpaired induced gene expression is a stronger predictor of treatment response than t-test for comparison of microarray ISG expression between IFNL3 IL28B genotype in patients with hepatitis C. Gastroenterology 2011; 140: genotypes and between SVRs/NSVRs, (2) Mann–Whitney test for analysis 1021–1031. of qPCR data and (3) linear regression analysis for comparison between 13 Honda M, Sakai A, Yamashita T, Nakamoto Y, Mizukoshi E, Sakai Y et al. ISGs. For microarray data, we interrogated the expression of the ISGs Hepatic ISG expression is associated with genetic variation in interleukin 28B and identified by Schoggins et al.50 The fold change between responder and the outcome of IFN therapy for chronic hepatitis C. Gastroenterology 2010; 139: non-responder genotypes were determined, the number of ISGs4 and o1 499–509. was tabulated and deviation from expected was calculated with the Sign 14 Urban TJ, Thompson AJ, Bradrick SS, Fellay J, Schuppan D, Cronin KD et al. test. To combine P-values between the different study cohorts we used IL28B genotype is associated with differential expression of intrahepatic inter- Fisher’s trend test.52 A two-sided P-value of less than 0.05 was considered feron-stimulated genes in patients with chronic hepatitis C. Hepatology 2010; 52: significant. 1888–1896. For RNA-Seq, a two-sample moderated t-test was used to compare the 15 Kohli A, Zhang X, Yang J, Russell RS, Donnelly RP, Sheikh F et al. Distinct and expression of ISGs between the IFNL3 genotypes for rs8099917. The gene overlapping genomic profiles and antiviral effects of Interferon-lambda and variances were moderated to be the maximum of their sample variance -alpha on HCV-infected and noninfected hepatoma cells. J Viral Hepat 2012; 19: and sample mean. This is similar, although less conservative, to the 843–853. approach used in differential expression Seq.53 Fisher’s exact test was used 16 Witte K, Gruetz G, Volk HD, Looman AC, Asadullah K, Sterry W et al. Despite to test whether the amount of mutually downregulated ISGs from each IFN-lambda receptor expression, blood immune cells, but not keratinocytes or platform was more than expected after acknowledging that the ISGs were melanocytes, have an impaired response to type III interferons: implications for downregulated as a set on both platforms. A one-sided version of therapeutic applications of these cytokines. Genes Immun 2009; 10: 702–714. Pearson’s test was used to combine the significance of differential 17 Moraga I, Harari D, Schreiber G, Uze G, Pellegrini S. Receptor density is key expression from the microarray and RNA-Seq. A one-sided Wilcoxon rank to the alpha2/beta interferon differential activities. Mol Cell Biol 2009; 29: sum test was used to test whether the expression of the ISGs were 4778–4787. downregulated as a set in relation to genotype. 18 Smith KR, Suppiah V, O’Connor K, Berg T, Weltman M, Abate ML et al. Identification of improved IL28B SNPs and haplotypes for prediction of drug response in treatment of hepatitis C using massively parallel sequencing in a cross-sectional European cohort. Genome Med 2011; 3: 57. CONFLICT OF INTEREST 19 Fischer J, Bohm S, Scholz M, Muller T, Witt H, George J et al. Combined effects of The authors declare no conflict of interest. different interleukin-28B gene variants on the outcome of dual combination therapy in chronic hepatitis C virus type 1 infection. Hepatology 2012; 55: 1700–1710. 20 Chen L, Borozan I, Sun J, Guindi M, Fischer S, Feld J et al. Cell-type specific gene ACKNOWLEDGEMENTS expression signature in liver underlies response to interferon therapy in chronic KSO was supported by an Australian National Health and Medical Research hepatitis C infection. Gastroenterology 2010; 138: e1121–e1123. Scholarship. Work on this manuscript was in part supported by research grants 21 McGilvray I, Feld JJ, Chen L, Pattullo V, Guindi M, Fischer S et al. Hepatic cell-type from the National Health and Medical Research Council of Australia (APP1006759) specific gene expression better predicts HCV treatment outcome than IL28B and by an Australian ARC linkage grant (LP0990067). JG is supported by the Robert W genotype. Gastroenterology 2012; 142: 1122–1131. Storr bequest to the Sydney Medical Foundation, University of Sydney. DRB is 22 Naggie S, Osinusi A, Katsounas A, Lempicki R, Herrmann E, Thompson AJ et al. supported by the Hunt Family Senior Research Fellowship. Dysregulation of innate immunity in hepatitis C virus genotype 1 IL28B-unfa- vorable genotype patients: impaired viral kinetics and therapeutic response. Hepatology 2012; 56: 444–454. 23 Thomas DL, Thio CL, Martin MP, Qi Y, Ge D, O’Huigin C et al. Genetic variation in REFERENCES IL28B and spontaneous clearance of hepatitis C virus. Nature 2009; 461: 798–801. 1 National Institutes of Health Consensus Development Conference Statement: 24 Cherian MG, Jayasurya A, Bay BH. Metallothioneins in human tumors and Management of hepatitis C 2002 (June 10-12, 2002). Gastroenterology 2002; 123: potential roles in carcinogenesis. Mutat Res 2003; 533: 201–209. 2082–2099. 25 Friedman RL, Stark GR. Alpha-Interferon-induced transcription of HLA and 2 Hoofnagle JH. Course and outcome of hepatitis C. Hepatology 2002; 36: S21–S29. metallothionein genes containing homologous upstream sequences. Nature 3 Ge D, Fellay J, Thompson AJ, Simon JS, Shianna KV, Urban TJ et al. Genetic 1985; 314: 637–639. variation in IL28B predicts hepatitis C treatment-induced viral clearance. Nature 26 Varshney U, Jahroudi N, Foster R, Gedamu L. Structure, organization, and reg- 2009; 461: 399–401. ulation of human metallothionein IF gene: differential and cell-type-specific 4 Suppiah V, Moldovan M, Ahlenstiel G, Berg T, Weltman M, Abate ML et al. IL28B is expression in response to heavy metals and glucocorticoids. Mol Cell Biol 1986; 6: associated with response to chronic hepatitis C interferon-alpha and ribavirin 26–37. therapy. Nat Genet 2009; 41: 1100–1104. 27 Schmidt CJ, Hamer DH. Cell specificity and an effect of ras on human metal- 5 Tanaka Y, Nishida N, Sugiyama M, Kurosaki M, Matsuura K, Sakamoto N et al. lothionein gene expression. Proc Natl Acad Sci USA 1986; 83: 3346–3350. Genome-wide association of IL28B with response to pegylated interferon-alpha 28 Takahashi S. Molecular functions of metallothionein and its role in hematological and ribavirin therapy for chronic hepatitis C. Nat Genet 2009; 41: 1105–1109. malignancies. J Hematol Oncol 2012; 5:41. 6 Asselah T, De Muynck S, Broet P, Masliah-Planchon J, Blanluet M, Bieche I et al. 29 Krizkova S, Ryvolova M, Hrabeta J, Adam V, Stiborova M, Eckschlager T et al. IL28B polymorphism is associated with treatment response in patients with Metallothioneins and zinc in cancer diagnosis and therapy. Drug Metab Rev 2012; genotype 4 chronic hepatitis C. J Hepatol 2012; 56: 527–532. 44: 287–301.

& 2014 Macmillan Publishers Limited Genes and Immunity (2014) 88 – 94 Hepatic MT expression in chronic HCV infection KS O’Connor et al 94 30 Wu C, Pot C, Apetoh L, Thalhamer T, Zhu B, Murugaiyan G et al. Metallothioneins 41 Carrera G, Paternain JL, Carrere N, Folch J, Courtade-Saidi M, Orfila C et al. Hepatic negatively regulate IL-27-induced type 1 regulatory T-cell differentiation. Proc Natl metallothionein in patients with chronic hepatitis C: relationship with severity of Acad Sci USA 2013; 110: 7802–7807. liver disease and response to treatment. Am J Gastroenterol 2003; 98: 1142–1149. 31 Davis SR, Cousins RJ. Metallothionein expression in animals: a physiological 42 Nagamine T, Suzuki K, Kondo T, Nakazato K, Kakizaki S, Takagi H et al. Interferon- perspective on function. J Nutr 2000; 130: 1085–1088. alpha-induced changes in metallothionein expression in liver biopsies from 32 Alter HJ, Seeff LB. Recovery, persistence, and sequelae in hepatitis C virus patients with chronic hepatitis C. Can J Gastroenterol 2005; 19: 481–486. infection: a perspective on long-term outcome. Semin Liver Dis 2000; 20: 17–35. 43 Nagamine T, Takagi H, Takayama H, Kojima A, Kakizaki S, Mori M et al. Preliminary 33 Mao J, Yu H, Wang C, Sun L, Jiang W, Zhang P et al. Metallothionein MT1M is a study of combination therapy with interferon-alpha and zinc in chronic hepatitis tumor suppressor of human hepatocellular carcinomas. Carcinogenesis 2012; 33: C patients with genotype 1b. Biol Trace Elem Res 2000; 75: 53–63. 2568–2577. 44 Takagi H, Nagamine T, Abe T, Takayama H, Sato K, Otsuka T et al. Zinc supple- 34 Marusawa H, Hijikata M, Chiba T, Shimotohno K. Hepatitis C virus core protein mentation enhances the response to interferon therapy in patients with chronic inhibits Fas- and tumor necrosis factor alpha-mediated apoptosis via NF-kappaB hepatitis C. J Viral Hepat 2001; 8: 367–371. activation. J Virol 1999; 73: 4713–4720. 45 Devitt E, Lawless MW, Sadlier D, Browne JA, Walsh C, Crowe J. Early viral and 35 Tai DI, Tsai SL, Chen YM, Chuang YL, Peng CY, Sheen IS et al. Activation of nuclear peripheral blood mononuclear cell responses to pegylated interferon and riba- factor kappaB in hepatitis C virus infection: implications for pathogenesis and virin treatment: the first 24 h. Eur J Gastroenterol Hepatol 2010; 22: 1211–1220. hepatocarcinogenesis. Hepatology 2000; 31: 656–664. 46 Li X, Cai L, Feng W. Diabetes and metallothionein. Mini Rev Med Chem 2007; 7: 36 Girard S, Vossman E, Misek DE, Podevin P, Hanash S, Brechot C et al. Hepatitis C 761–768. virus NS5A-regulated gene expression and signaling revealed via microarray and 47 Lansdown AB. Metallothioneins: potential therapeutic aids for wound healing in comparative promoter analyses. Hepatology 2004; 40: 708–718. the skin. Wound Repair Regen 2002; 10: 130–132. 37 Agundez JA, Garcia-Martin E, Maestro ML, Cuenca F, Martinez C, Ortega L et al. 48 Pedersen MO, Larsen A, Stoltenberg M, Penkowa M. Cell death in the injured Relation of IL28B gene polymorphism with biochemical and histological features brain: roles of metallothioneins. Prog Histochem Cytochem 2009; 44: 1–27. in hepatitis C virus-induced liver disease. PloS One 2012; 7: e37998. 49 Aschner M, Cherian MG, Klaassen CD, Palmiter RD, Erickson JC, Bush AI. Metal- 38 Bochud PY, Bibert S, Kutalik Z, Patin E, Guergnon J, Nalpas B et al. IL28B alleles lothioneins in brain—the role in physiology and pathology. Toxicol Appl Phar- associated with poor hepatitis C virus (HCV) clearance protect against inflam- macol 1997; 142: 229–242. mation and fibrosis in patients infected with non-1 HCV genotypes. Hepatology 50 Schoggins JW, Wilson SJ, Panis M, Murphy MY, Jones CT, Bieniasz P et al. A diverse 2012; 55: 384–394. range of gene products are effectors of the type I interferon antiviral response. 39 Thompson AJ, Clark PJ, Zhu M, Zhu Q, Ge D, Sulkowski MS et al. Genome-wide Nature 2011; 472: 481–485. association study identifies IL28B polymorphism to be associated with baseline 51 Langmead B, Trapnell C, Pop M, Salzberg SL. Ultrafast and memory-efficient ALT and hepatic necro-inflammatory activity in chronic hepatitis C patients alignment of short DNA sequences to the human genome. Genome Biol 2009; 10:R25. enrolled in the IDEAL study. Hepatology 2010; 52: 1220A–1221A. 52 Ge D. http://compute1.lsrc.duke.edu/softwares/MetaP/metap.php. accessed: 40 Yang CH, Murti A, Pfeffer SR, Basu L, Kim JG, Pfeffer LM. IFNalpha/beta promotes July 2013. cell survival by activating NF-kappa B. Proc Natl Acad Sci USA 2000; 97: 53 Anders S, Huber W. Differential expression analysis for sequence count data. 13631–13636. Genome Biol 2010; 11: R106.

Supplementary Information accompanies this paper on Genes and Immunity website (http://www.nature.com/gene)

Genes and Immunity (2014) 88 – 94 & 2014 Macmillan Publishers Limited