The full names of the genes indicated by gene symbols in the tables and figures can be found at www.genecards.org.
Potential conflict of interest: Nothing to report.
The mechanism of the interferon-alpha (IFNα)–induced antiviral response is not completely understood. We recently examined the transcriptional response to IFNα in uninfected chimpanzees. The transcriptional response to IFNα in the liver and peripheral blood mononuclear cells (PBMCs) was rapidly induced but was also rapidly down-regulated, with most interferon-alpha–stimulated genes (ISGs) returning to the baseline within 24 hours. We have extended these observations to include chimpanzees chronically infected with hepatitis C virus (HCV). Remarkably, using total genome microarray analysis, we observed almost no induction of ISG transcripts in the livers of chronically infected animals following IFNα dosing, whereas the response in PBMCs was similar to that in uninfected animals. In agreement with this finding, no decrease in the viral load occurred with up to 12 weeks of pegylated IFNα therapy. The block in the response to exogenous IFNα appeared to be HCV-specific because the response in a hepatitis B virus–infected animal was similar to that of uninfected animals. The lack of a response to exogenous IFNα may be due to an already maximally induced ISG response because chronically HCV-infected chimpanzees already have a highly up-regulated hepatic ISG response. Alternatively, negative regulation may block the response to exogenous IFNα, yet it does not prevent the continued response to endogenous ISG stimuli. The IFNα response in chronically HCV-infected chimpanzees may be mechanistically similar to the null response in the human population. Conclusion: In chimpanzees infected with HCV, the highly elevated hepatic ISG expression may prevent the further induction of ISGs and antiviral efficacy following an IFNα treatment. (HEPATOLOGY 2007.)
Approximately 170 million people worldwide are chronically infected with hepatitis C virus (HCV), which frequently progresses to serious liver disease, including cirrhosis and hepatocellular carcinoma.1, 2 A combination therapy with pegylated (PEG) interferon-alpha (IFNα) and ribavirin results in sustained viral clearance for approximately 40%-50% and 80%-90% of patients infected with genotypes 1 and 2/3, respectively.3–5 Little is understood regarding the factors leading to successful or failed viral clearance during IFNα therapy, but studies on gene expression are beginning to illustrate differences in these populations. Gene expression studies in chimpanzees with acute resolving and chronic HCV infections have revealed elevated interferon-stimulated gene (ISG) expression in the liver, which is indicative of a response to double-stranded RNA (dsRNA) and/or IFNα,6–8 and similar observations have been made of chronically infected humans.9, 10 More recently, an inverse correlation has been observed between the pretreatment hepatic levels for some ISG transcripts and the virologic response to therapy,11 whereas a positive correlation has been observed between the magnitude of the ISG response in IFNα-treated peripheral blood mononuclear cells (PBMCs) and the virologic response to therapy.12
The early kinetics of viral RNA loss from the circulation during IFNα therapy suggest the presence of 2 phases. Phase 1 occurs during the first 24-48 hours and is presumably due to the decrease in the secretion of new virions, whereas phase 2 kinetics vary between individuals, are predictive of the outcome of therapy, and are thought to be a measurement of the loss of infected cells.13–15 Our recent data on gene expression in uninfected chimpanzees dosed with IFNα suggests that the rapid down-regulation of the IFNα-induced ISG response may be partially responsible for the rapid change in the kinetics from phase 1 to phase 2.16 The peak expression for most genes occurred within 4 hours after IFNα dosing and was declining or at the baseline within 8 hours. The transcriptional response to IFNα in vivo was largely tissue-specific, with significant differences in the response in the liver and PBMCs.
Previous studies with IFNα in chimpanzees have failed to demonstrate a reduction in the viral load through the use of either traditional IFNα therapy with ribavirin or adenovirus-based gene therapy to induce high-level expression of IFNα in the liver17, 18 (R.E.L., unpublished data). Our recent studies comparing human and chimpanzee IFNα in uninfected animals indicated that IFNα from both species was equally effective with respect to ISG induction in the liver and PBMCs. Thus, the reason for the lack of an antiviral effect of IFNα in the chimpanzee remains unresolved. Here we have examined human PEG IFNα in 3 chronically HCV-infected chimpanzees and have performed a total genome DNA microarray analysis to characterize changes in the liver and PBMC gene expression. No decline in the viral load was observed with up to 12 weeks of therapy. The induction of ISGs in PBMCs from infected animals was similar to that observed in uninfected animals, whereas the livers of HCV-infected animals were resistant to exogenous IFNα. The implications of these findings for HCV-infected humans and possible mechanisms of resistance to exogenous IFNα are discussed.
CXCL10, chemokine (C-X-C motif) ligand 10; dsRNA, double-stranded RNA; HBV, hepatitis B virus; HCV, hepatitis C virus; IFN, interferon; IFNα, interferon-alpha; IP-10, interferon-gamma–inducible protein-10; IRF, interferon regulatory factor; ISG, interferon-stimulated gene; PBMC, peripheral blood mononuclear cell; PEG, pegylated; RT-PCR, reverse-transcription polymerase chain reaction; SOCS1, suppressor of cytokine signaling 1; SOCS3, suppressor of cytokine signaling 3; STAT1β, signal transducer and activator of transcription 1 beta; TLR3, toll-like receptor 3.
Materials and Methods
Chimpanzees were housed at the Southwest National Primate Research Center at the Southwest Foundation for Biomedical Research. The animals were cared for in accordance with the Guide for the Care and Use of Laboratory Animals, and all protocols were approved by the Institutional Animal Care and Use Committee. Biopsies were obtained in the morning from fasting animals to avoid postprandial and diurnal changes in liver gene expression. The chimpanzees used in this study had been chronically infected for 16-28 years, had been monitored on a regular basis in recent years, and had stable levels of viremia. The 3 HCV-infected animals were 4x0081 (genotype 1b; infected in 1983 with non A, non B [NANB] hepatitis), 4x0119 (genotype 3a; infected in 1981 with non A, non B [NANB] hepatitis contaminated factor VIII), and 4x0341 (genotype 1a; infected in 1991 with strain HCV1). The hepatitis B virus (HBV)–infected animal was 4x0139 (infected in 1979).
Microarray and Reverse-Transcription Polymerase Chain Reaction (RT-PCR) Analyses.
The total RNA prepared from the liver and PBMCs was used to perform microarray analyses6, 7 and to monitor viral and cellular transcripts by quantitative RT-PCR (TaqMan),19 as described in the Supplementary Array Methods. A microarray analysis was performed on 71 microarrays, which included a previous data set from 36 arrays from uninfected animals.16 A more comprehensive data set is available at http://www.sfbr.org/pages/virology/lanford, which includes Excel versions of the supplementary tables that can be searched, sorted, and downloaded.
PEG IFNα2a (Roche, Nutley, NJ) was partially a gift of Dr. David Thomas (Johns Hopkins University) and was partially purchased from a pharmacy. Chimpanzees were inoculated subcutaneously with 5 million IU of IFNα, and blood and liver samples were obtained 2 and 4 weeks before dosing and 4, 8, and 24 hours and 7 days after inoculation. The genotype 1 animals were continued on a weekly administration of PEG IFNα for 8 or 12 weeks as indicated.
Lack of an Antiviral Response in Chimpanzees to PEG IFNα.
In this study, we examined the relationship of the antiviral response to PEG IFNα in HCV-infected chimpanzees and the induction of gene expression in the liver and PBMCs. Three chimpanzees chronically infected with HCV were dosed with PEG IFNα weekly for 1-12 weeks, and the viral load was determined by quantitative RT-PCR (TaqMan). The genotype 1a chimpanzee (4x0341) was dosed for 12 weeks, whereas the dosing of the genotype 1b chimpanzee (4x0081) was stopped after 8 weeks because of unrelated health issues, and the genotype 3a chimpanzee (4x0119) was given a single dose of IFNα to examine ISG induction in the liver and was followed only for 14 days. No significant decline in viral RNA was observed in the serum (Fig. 1) or liver (data not shown).
Lack of an Interferon-Gamma–Inducible Protein-10 (IP-10) Transcriptional Response to IFNα in the Livers of HCV-Infected Chimpanzees.
To evaluate the possible reasons for the failed antiviral response to IFNα in chimpanzees, the induction of IP-10 [chemokine (C-X-C motif) ligand 10 (CXCL10)] transcription was examined in the liver and PBMCs by quantitative RT-PCR. We have previously demonstrated that this chemokine is highly elevated during acute7 and chronic6 HCV infection and following IFNα dosing in uninfected chimpanzees,16 primary hepatocytes,16 and Huh7 cells.19 Four hours after inoculation with PEG IFNα in an uninfected chimpanzee (4x0363), IP-10 transcripts were elevated 394-fold and 667-fold in the PBMCs and liver, respectively. Similarly, in the PBMCs of an HCV-infected animal (4x0081), IP-10 transcripts were increased 107-fold, but surprisingly, no induction was observed in the liver of this animal (Fig. 2). As observed in uninfected chimpanzees,16 the transcriptional response to IFNα was rapidly down-regulated. IP-10 transcripts were reduced nearly to the baseline in the livers and PBMCs of uninfected animals and in PBMCs of HCV-infected animals within 8-24 hours after inoculation. The response in the other HCV-infected animals was similar to that of 4x0081, and an examination of several other ISGs by quantitative RT-PCR revealed patterns of expression similar to that of IP-10. To determine if the lack of an ISG response in the liver was unique to HCV or common to other chronic viral infections, a chimpanzee chronically infected with HBV (4x0139) was dosed with PEG IFNα with an identical protocol. IP-10 transcripts were elevated 107-fold and 24-fold 4 hours after inoculation in the PBMCs and liver, respectively (Fig. 2). We also examined the serum levels of the IP-10 protein by an enzyme-linked immunosorbent assay. In uninfected animals, the levels are highly induced and mimic the induction and down-regulation of the transcript observed in the liver and PBMCs. In the HCV chronically-infected animals, there was no significant increase in the serum IP10 levels despite the transient increase in transcripts in PBMCs. This suggests that because of its size, the liver may represent the primary source of IP-10 in the serum when induced by an HCV infection or IFNα treatment. In the HBV-infected animal, the serum levels of IP-10 did not increase as much as observed in the uninfected animals. This correlates with the lower induction of IP-10 transcripts in both the liver and PBMCs in comparison with the uninfected animals and is likely due to animal-to-animal variation, although HBV infection may influence IP-10 induction.
Lack of an ISG Response to Exogenous IFNα in the Livers of HCV-Infected Chimpanzees.
To evaluate the extent of nonresponsiveness to IFNα in the livers of HCV-infected chimpanzees, a total genome microarray analysis was performed on the livers and PBMCs 2 times before dosing and 4, 8, and 24 hours after dosing. A nearly total lack of an ISG response to IFNα was observed in the livers of all 3 HCV-infected animals, whereas the PBMC fractions of infected and uninfected animals exhibited very similar ISG induction profiles. Table 1 illustrates the lack of an ISG response in the liver by selecting 28 genes that were highly induced in the livers of uninfected animals for comparison with HCV-infected livers. The genes were selected from 500 hepatic ISGs in uninfected animals on the basis of the magnitude of the fold induction in the liver and PBMCs and/or name recognition in various aspects of the interferon (IFN) antiviral pathway16 (Supplementary Table 1). The values shown are the average fold changes 4 hours after dosing in relation to the pretreatment baselines for 3 uninfected animals (2 independent experiments each) and 3 HCV-infected animals. Only 5 of the ISGs that were highly induced in uninfected livers were induced in the livers of HCV-infected chimpanzees, and for these 5 ISGs, the fold induction was lower than that observed in uninfected animals. All but 5 of the ISGs in Table 1 were induced in the liver of the HBV-infected chimpanzee, and the magnitude of the induction was similar to that observed in the uninfected animals. A microarray analysis of the liver and PBMCs from 1 HCV-infected animal (4x0081) over 8 weeks of PEG IFNα therapy revealed that the ISGs in PBMCs remained down-regulated after the initial burst of induction and that ISG transcripts in the liver were not induced even after repeated exposure to IFNα (Fig. 3).
Table 1. Lack of ISG Induction in HCV-Infected Livers
Average Fold Change 4 Hours After PEG IFNα
The values are the average fold changes between 0 and 4 hours after IFNα for 3 uninfected chimpanzees (each with 2 independent dosing studies), 3 HCV-infected chimpanzees, and 1 HBV-infected chimpanzee. The ISGs were selected from the 500 most highly induced genes after 4 hours in uninfected animals.
The genes denoted by − did not meet our criteria for inclusion: average fold change ≥ 2.0 and P ≤ 0.02.
In contrast to the selection of the genes most highly induced in uninfected livers for comparison with infected livers, when all genes significantly up-regulated in the livers of HCV-infected animals in response to IFNα were selected, only 24 ISGs (from 500 in the uninfected liver) met the criteria for inclusion (fold change ≥ 2.0, P ≤ 0.02; Table 2). Of particular interest was the induction in the HCV-infected liver of both suppressor of cytokine signaling 1 (SOCS1) and suppressor of cytokine signaling 3 (SOCS3), which are inhibitors of the JAK-STAT (Janus kinase–signal transducer and activator of transcription) pathways. SOCS1 was not detected in the uninfected liver, but SOCS3 was similarly induced in the livers of all animals. Despite the remarkable differences in the hepatic response to IFNα, the response in the PBMCs was similar in uninfected, HCV-infected, and HBV-infected animals. All 25 of the ISGs shown in Table 1 that were induced in the PBMCs of uninfected animals were also induced in the PBMCs of HCV-infected animals. Thus, the lack of a hepatic response to exogenous IFNα appeared to involve nearly all ISGs in HCV-infected animals but was not observed in the PBMCs of HCV-infected animals or the HBV-infected liver and PBMCs.
Table 2. Average Fold Change 4 Hours After PEG IFNα
In contrast to Table 1, this table contains all ISGs that were up-regulated in the livers of HCV-infected chimpanzees by PEG IFNα. Of the 500 genes up-regulated in uninfected livers, only 24 were up-regulated in HCV chronically-infected livers. The values are the average fold changes between 0 and 4 hours after IFNα for 3 uninfected chimpanzees (each with 2 independent dosing studies), 3 HCV-infected chimpanzees, and 1 HBV-infected chimpanzee.
The genes denoted by − did not meet our criteria for inclusion: average fold change ≥ 2.0 and P ≤ 0.02.
Baseline ISG Expression in Chronically HCV-Infected Animals.
Although IFNα did not induce an increase in ISG expression in the livers of HCV-infected animals, the ISG response was already highly induced because of HCV infection. We have previously characterized the persistent ISG induction in chronically HCV-infected animals by a microarray analysis.6 These results were confirmed during this study by the analysis of the baseline infected and uninfected liver samples before dosing. An examination of the IP-10 RT-PCR data for all the animals in this study, using copies of the transcript rather than the fold change from the baseline, revealed that IP-10 transcripts were elevated at the baseline in the livers of HCV-infected animals in comparison with the uninfected animals and that the hepatic IP-10 baseline values for chronically HCV-infected animals were similar to the IFNα-induced levels in uninfected and HBV-infected animals (Fig. 4, black bars). The PBMC IP-10 baseline levels were similar for infected and uninfected animals and were highly induced by IFNα in both infected and uninfected animals (Fig. 4, gray bars).
To better illustrate the elevated baseline levels of ISGs in the livers of chronically HCV-infected chimpanzees, the microarray data were examined as the signal intensity rather than the fold change from the baseline. A set of 822 genes were identified as significantly induced by IFNα with an ANOVA (signal log ratio ≥ 1 = fold change of 2, P ≤ 0.02) and PAM (partitioning around medoids clustering) (Supplementary Table 2). This analysis revealed that the total ISG response followed the same trend as IP-10. In Fig. 5, a heat map illustrates the results from this analysis with 1-dimensional hierarchical clustering. In uninfected animals, most genes are induced 4 and 8 hours after IFNα (a shift from green to red/black) and are down-regulated within 24 hours (a shift back to green). In contrast, most genes were already up-regulated in the chronically infected animals and were either not further induced by IFNα or were minimally induced. The trend for elevated baseline values for ISG transcripts in the livers of HCV-infected animals was further illustrated when the selected set of the most highly induced ISGs from Table 1 was subjected to the same type of analysis (Fig. 6). Although only 1 HBV-infected animal was available for this study, a comparison of the baseline values for ISG transcripts in the livers of uninfected animals and the HBV-infected animal failed to detect a significant increase in ISG expression due to HBV infection (data not shown), and this is consistent with previous studies on HBV infection in chimpanzees.20
The relationship of the total gene profiles from each sample type can best be visualized with a hierarchical cluster analysis of the entire gene set (Fig. 7). As expected, liver and PBMC samples cluster separately, independently of HCV infection and IFNα dosing, because of their divergent transcriptomes. The uninfected (shaded area) and chronically infected liver samples cluster separately from each other because of baseline ISG induction in the chronically infected liver, whereas this is not true for PBMC samples, for which the chronically infected and uninfected pretreatment samples cluster together. For the uninfected liver (shaded area), the 0-hour and 24-hour samples cluster separately from the 4-hour and 8-hour samples after IFNα dosing because of the induced ISGs after 4 and 8 hours and the down-regulation to the baseline levels within 24 hours. For the HCV-infected liver, little difference is observed in the pretreatment and posttreatment samples; 4 and 8 hours after IFNα, the samples do not cluster together.
In this study, we have demonstrated that the livers of HCV-infected chimpanzees are nonresponsive to exogenous IFNα. Previous studies with human IFNα in chronically HCV-infected chimpanzees failed to demonstrate an antiviral response18 (R.E.L., unpublished data). These studies included conventional IFNα with ribavirin for 28 days and the use of adenovirus-based gene therapy.17, 18 The lack of an antiviral response was assumed to be due to the use of human IFNα. However, our recent studies comparing chimpanzee and human IFNα in uninfected chimpanzees demonstrated that the 2 forms of IFNα were essentially equivalent with respect to the induction of ISGs in the liver and PBMCs.16 Thus, additional studies using PEG IFNα were warranted in the hope of developing a chimpanzee model for the examination of combined therapies with newly developed antivirals. Although the PEG IFNα treatment of HCV-infected chimpanzees provided no antiviral efficacy, important aspects of the reason for the lack of an antiviral response were revealed.
The most notable change in hepatic gene expression during acute and chronic HCV infection in chimpanzees is the up-regulation of ISGs,6–8 which is suggestive of an ongoing type I IFN or dsRNA response. Similar findings have been observed in the human liver during chronic HCV infection.9–11, 21 During an acute infection in chimpanzees, the level of ISG transcript induction in the liver clearly parallels the increase and decrease of viral RNA in the serum and liver.7, 8, 22 However, in an analysis of 10 chronically HCV-infected chimpanzees with viral loads differing over 1000-fold (104-107 genome equivalents/mL), no correlation was observed in the hepatic ISG transcript levels and the viral load, and this suggested that the response may be maximally induced even at relatively low levels of viral replication.6 The stimulus for ISG induction during HCV infection has not been unambiguously defined and may be multifactorial. An increase in type 1 IFN transcripts was not detectable by a microarray; however, if a small fraction of cells are producing IFN, the dilution of the transcript by the total liver RNA may render the detection of an increase in IFN transcripts unlikely. Viral RNA can interact with a number of cellular proteins to trigger the innate response, most notably toll-like receptor 3 (TLR3), RIGI (retinoic acid inducible gene I), and MDA5 (melanoma differentiation-associated gene 5). The elevated hepatic ISG levels during chronic HCV infection occur despite the capacity of the NS3/4a protease to block the activation of interferon regulatory factor 3 (IRF3) and the secretion of type I IFN by the RIGI, MDA5, and TLR3 pathways.23–25 Newly infected hepatocyte cells may produce type I IFN until sufficient viral protein is available to block these pathways. In addition, the response of dendritic cells and macrophages to dsRNA and the resulting production of type I IFN presumably would not be subject to NS3 inhibition.
Recently, a correlation was observed between the sustained virologic response to IFNα and the reduced pretreatment levels for some hepatic ISG transcripts.11 Of the 8 genes showing more than 90% correlation with the treatment outcome, 5 were ISGs. Lower pretreatment serum levels of IP-10 have also been found to correlate with the treatment outcome.26–29 The levels of IP-10 decrease further during therapy,27, 28 and this suggests that the liver and PBMCs are no longer responding to the exogenous IFNα; this is consistent with our observation that the transcriptional response to IFNα is down-regulated after the initial burst of transcription despite the continued presence of high levels of circulating IFNα (Fig. 1).16 The decrease in the IP-10 levels during IFNα therapy in responders also implies that the original stimulus for IP-10 induction, presumably a dsRNA response, is declining.
At this time, no data are available on the hepatic ISG transcript levels after IFNα dosing in humans because of the difficulty in obtaining multiple biopsies in humans. Our data on the rapid up-regulation and subsequent down-regulation of the ISG response in chimpanzees following IFNα dosing16 suggest that studies in humans will be further complicated by the difficulty in obtaining biopsies during the narrow time frame of maximum ISG induction. Currently, the postdosing ISG response in humans has been examined only in PBMCs. An increased ISG response was noted in responders using PMBCs exposed in vitro to IFNα12 and in PBMCs taken during the treatment30; however, unlike the liver, no relationship between the ISG baseline values and the response to therapy was observed. Some differences in the response of the liver and PBMCs might be expected because HCV primarily replicates in the liver. Influences of viral replication on the IFNα response may be confined to the liver, whereas the response in PBMCs may reflect individual genetic differences that precede HCV infection.
The mechanism for the lack of a hepatic response to IFNα in HCV-infected chimpanzees is not known, but it is possible that during chronic HCV infection, the hepatic ISG response is already maximally induced. In a study of 10 chronically infected chimpanzees with over a 1000-fold difference in the viral load, the level of the hepatic ISG response did not correlate with the viral load and was similar in all animals,6 suggesting a possible saturation of the response to the virus. The persistence of the virus in the presence of an elevated hepatic ISG response suggests that the endogenous ISG response may have limited antiviral activity. However, an alternative view would be that the endogenous ISG response is functioning appropriately in limiting the replication and spread of the virus. The determination of the percentage of infected hepatocytes has been problematic, and the results have varied. Strand-specific in situ hybridization has detected negative-strand RNA in up to 40% of cells,31 although the exact percentage of positive cells varied between biopsies and different fields of the same biopsy. We have used the level of viral RNA in the liver as an indirect measurement of the level of infection. On the basis of our assumptions, on average less than 10% of the cells were infected during acute and chronic infections, whereas the range of hepatic viral RNA in different animals suggests that 0.1%-30% of hepatocytes may be infected.6, 16 Thus, our estimations are not inconsistent with in situ hybridization data and would allow for up to 100% infection of hepatocytes in individuals with very high viremia. A low percentage of infected hepatocytes is also consistent with calculations of the total daily production of virus in humans. If we assume that the liver contains 5 × 1011 hepatocytes and produces 1012 HCV particles per day,13 only 2 particles per cell need to be produced per day. However, at best, the endogenous ISG response only suppresses replication, and thus there may be qualitative and quantitative differences between the ISG response to HCV infection (endogenous response) and that induced by exogenous IFNα in sustained responders. Presumably, humans that respond to therapy have a lower endogenous hepatic ISG response and an increased response to exogenous IFNα. The recent finding that nonresponders have higher baseline levels for some hepatic ISGs11 suggests a possible correlation between nonresponders in the human population and chimpanzees, both of which have elevated hepatic baseline ISG levels.
A number of factors are involved in the negative regulation of the IFNα response, and our studies of uninfected chimpanzees have demonstrated that negative regulation is rapidly invoked, with the ISG response being down-regulated within 8-24 hours of PEG IFNα dosing.16 The transcriptional inhibitors involved in the down-regulation of the IFN response include IRF2, IRF4, IRF8, and signal transducer and activator of transcription 1 beta (STAT1β). IRF2 was up-regulated in the uninfected chimpanzee liver, PBMCs, and primary hepatocytes16 following a treatment with IFNα. STAT1β was up-regulated in the uninfected chimpanzee liver following IFNα dosing and during acute and chronic HCV infection.6, 7 In addition, the proteins involved in signal transduction are subject to posttranslational regulation by ubiquitination (SOCS: suppressor of cytokine signaling), sumoylation (pIAS: protein inhibitor of IFNα signaling), ISGylation, and tyrosine dephosphorylation. Following the IFNα treatment, SOCS3 was up-regulated in both the infected and uninfected livers, but SOCS1 was significantly up-regulated only in the infected liver.
A precise understanding of the mechanism for the lack of a response to exogenous IFNα may require detailed studies in the chimpanzee model, and an analysis of the IFNα hepatic response in null responders in the human population will be required to determine if the chimpanzee actually mimics this response in humans.
The authors thank Dr. David Thomas, Johns Hopkins University, for the kind gift of PEG IFNα.