Genomic response to interferon-α in chimpanzees: Implications of rapid downregulation for hepatitis C kinetics

Authors

  • Robert E. Lanford,

    Corresponding author
    1. Department of Virology and Immunology, Southwest Foundation for Biomedical Research, Southwest National Primate Research Center, San Antonio, TX
    • Department of Virology and Immunology, Southwest National Primate Research Center and Southwest Foundation for Biomedical Research, 7620 NW Loop 410, San Antonio, TX 78227
    Search for more papers by this author
    • fax: 210-670-3329

  • Bernadette Guerra,

    1. Department of Virology and Immunology, Southwest Foundation for Biomedical Research, Southwest National Primate Research Center, San Antonio, TX
    Search for more papers by this author
  • Helen Lee,

    1. Department of Virology and Immunology, Southwest Foundation for Biomedical Research, Southwest National Primate Research Center, San Antonio, TX
    Search for more papers by this author
  • Deborah Chavez,

    1. Department of Virology and Immunology, Southwest Foundation for Biomedical Research, Southwest National Primate Research Center, San Antonio, TX
    Search for more papers by this author
  • Kathleen M. Brasky,

    1. Department of Comparative Medicine, Southwest Foundation for Biomedical Research, Southwest National Primate Research Center, San Antonio, TX
    Search for more papers by this author
  • Catherine B. Bigger

    1. Department of Virology and Immunology, Southwest Foundation for Biomedical Research, Southwest National Primate Research Center, San Antonio, TX
    Current affiliation:
    1. Children's Research Institute, Room WA 4104, 700 Children's Drive, Columbus, OH
    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.

Abstract

The mechanism of the interferon-alpha (IFN-α)-induced antiviral response during hepatitis C virus (HCV) therapy is not completely understood. In this study, we examined the transcriptional response to IFN-α in uninfected chimpanzees after single doses of chimpanzee, human, or human-pegylated IFN-α. Liver and peripheral blood mononuclear cell (PBMC) samples were used for total genome microarray analysis. Most induced genes achieved maximal response within 4 hours, began to decline by 8 hours, and were at baseline levels by 24 hours post-inoculation, a time when high levels of circulating pegylated IFN-α were still present. The rapid downregulation of the IFN-α response may be involved in the transition between the observed phase I and phase II viral kinetics during IFN-α therapy in HCV-infected patients. The response to all three forms of IFN-α was similar; thus, the reasons for previous failures in antiviral treatment of chimpanzees with human IFN-α were not due to species specificity of IFN-α. The response to IFN-α was partially tissue-specific. A total of 1,778 genes were altered in expression by twofold or more by IFN-α, with 538 and 950 being unique to the liver or PBMC, respectively. Analysis of the IFN-α and IFN-γ responses in primary chimpanzee and human hepatocytes were compared as well. IFN-α and IFN-γ induced partially overlapping sets of genes in hepatocytes. In conclusion, the response to IFN-α is largely tissue-specific, and the response is rapidly downregulated in vivo, which may have a significant influence on the kinetics of antiviral response. (HEPATOLOGY 2006;43:961–972.)

Worldwide, approximately 170 million people are chronically infected with hepatitis C virus (HCV), which frequently progresses to serious liver disease, including cirrhosis and hepatocellular carcinoma.1, 2 The current therapy involves the combination of pegylated (peg)-interferon alpha (IFN-α) and ribavirin (reviewed in Feld and Hoofnagle3) and has response rates for sustained viral clearance of approximately 40% to 50% and 80% to 90% for genotypes 1 and 2/3, respectively.4, 5 However, a significant proportion of the population still develops serious disease as a consequence of HCV infection. HCV infection is the leading cause for liver transplantation in the United States1, 2 and liver cancer due to HCV infection is one of the most rapidly increasing types of cancer in the United States.6 Little is understood regarding the factors leading to successful or failed viral clearance during IFN-α therapy. The early kinetics of viral RNA loss from the circulation during IFN-α therapy suggests the presence of two phases. Phase I occurs during the first 24 to 48 hours and is presumed to be due to the decrease in secretion of new virions, whereas phase II kinetics vary between individuals, are predictive of the outcome of therapy, and are thought to be a measurement of loss of infected cells.7–9

We10, 11 and others12 have previously performed gene expression analyses on liver from chimpanzees that experienced acute-resolving or chronic HCV infections. The most notable changes in gene expression occurred in the IFN-stimulated genes (ISGs), suggesting that an ongoing type I IFN or double-stranded RNA (dsRNA) response is occurring in the liver during both acute and chronic infections. Similar findings have been observed in human liver and peripheral blood mononuclear cells (PBMC) during chronic infection.13–16 One potential interpretation of these findings consistent with observations in several systems is that an ongoing IFN-α/β response limits virus replication and spread in the liver.10, 11 The importance of IFN-α in HCV clearance has been demonstrated in several studies: (1) the high rate of sustained viral clearance of chronic infections after combined therapy with peg-IFN-α and ribavirin,4, 5 (2) the near 100% viral clearance rate using IFN-α monotherapy in acutely infected individuals,17 and (3) the sensitivity of HCV-replicons to IFN-α.18–20

Previous studies in chimpanzees using traditional IFN-α therapy with and without ribavirin, as well as adenovirus-based gene therapy to induce high-level expression of IFN-α in the liver, failed to induce a reduction in viral load despite high levels of circulating IFN-α for extended times after adenovirus gene therapy.21, 22 One concern was whether human IFN-α was effective in the chimpanzee. Because the chimpanzee is the only experimental animal model for HCV, resolution of this question was deemed essential.

We have used DNA microarray analysis to characterize changes in liver and PBMC gene expression in chimpanzees after a single dose of either chimpanzee, human, or human peg-IFN-α2. These studies have allowed the simultaneous comparison of transcriptional changes of up to 47,000 genes and have demonstrated a consistent pattern of gene expression among the different animals. Peak expression for most genes occurred either by 4 hours and was declining or at baseline by 8 hours. We hypothesized that the rapid downregulation of the IFN-α response in vivo may play a role in the difference in kinetics of viral decline observed over time during IFN-α therapy of HCV infection (see Discussion). The response to IFN-α in vivo was largely tissue specific, with significant differences in the response in liver and PBMC. Primary chimpanzee and human hepatocytes were also examined for the hepatocyte-specific response to IFN-α and IFN-γ. In hepatocytes, these cytokines induce a partially overlapping set of genes.

Abbreviations

HCV, hepatitis C virus; IFN, interferon; peg, pegylated; ISG, interferon stimulated gene; dsRNA, double-stranded RNA; PBMC, peripheral blood mononuclear cells; RT-PCR, reverse transcription polymerase chain reaction; ELISA, enzyme-linked immunosorbent assay; IRG, interferon regulated gene; NK, natural killer; TLR, Toll-like receptor; IRF, interferon regulatory factor; SOCS, suppressor of cytokine signaling.

Materials and Methods

Chimpanzees and Hepatocytes.

Chimpanzees were housed 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. Chimpanzee hepatocytes were isolated and cultivated in a modified serum-free medium formulation23 as previously described.24 Primary human hepatocytes were purchased from BD Gentest (Woburn, MA) and cultivated under the same conditions.

Microarray and TaqMan Analyses.

Total RNA prepared from liver and PBMC was used to perform microarray analyses10, 11 and quantitative reverse transcription polymerase chain reaction (RT-PCR; TaqMan)19 as described in Supplemental Methods (Supplementary materials is available at the HEPATOLOGY website: http://interscience.wiley.com/jpages/0270-9139/suppmat/index.html). Uninfected chimpanzees were inoculated subcutaneously with 5 million IU IFN-α; blood and liver samples were obtained at 0, 4, 8, and 24 hours post-inoculation, followed by 30 days of rest before a second exposure to a different IFN-α (Fig. 1).The in vivo analysis was performed on 36 microarrays. Analysis of primary hepatocytes was performed on 18 arrays from three independent experiments, two with human hepatocytes and one with chimpanzee hepatocytes (6 baselines, 6 IFN-α treated, 6 IFN-γ treated). In each experiment, cultures were treated with 1,000 IU/mL IFN-α or IFN-γ, and samples were harvested at 4 and 8 hours posttreatment. Each sample comprised a pool from three cultures. A more comprehensive data set is available at http://www.sfbr.org/virology/lanford, including Excel versions of the Supplemental Tables with gene ontology descriptions.

Figure 1.

Cross-over design of chimpanzee studies on IFN-α. Three uninfected chimpanzees (4x0363, 4x0386, 4x0291) were inoculated subcutaneously with 5 million IU chimpanzee (Ch), human (Hu), or human pegylated (Hu peg) IFN-α. Blood and liver samples were obtained at 0, 4, 8, and 24 hours post-inoculation for microarray analysis. The animals were rested for 30 days, and the procedure was repeated with a different species of IFN-α. Each animal received chimpanzee IFN-α and either human or human peg-.

Cloning and Expression of IFN-α.

Chimpanzee and human IFN-α2 genes were amplified from total liver RNA using RT-PCR with primers based on the human IFN-α2 gene. Alignment of the sequences showed only two amino acid differences between chimpanzee and human IFN-α2, none of which corresponded to the known polymorphisms in human IFN-α2, subtypes a-c (Supplemental Fig. 1). Details on expression of IFN-α are available in Supplement Methods. Peg-IFN-α2b (Schering Plough, Kenilworth, NJ) was purchased from a pharmacy. IFN-γ was purchased from R&D Research (Minneapolis, MN).

Results

The IFNα Response Is Rapidly Downregulated In Vivo.

In the current study, we cloned both human and chimpanzee IFN-α2 for a direct comparison of the two IFNs produced under the same conditions. Three uninfected chimpanzees were used in a crossover study in which each animal received 5 million IU of either chimpanzee or human IFN-α2 followed 30 days later by the opposite species of IFN-α (Fig. 1). Liver and blood samples were obtained at 0, 4, 8, and 24 hours after injection. The data from the first phase of the crossover study indicated that human and chimpanzee IFN-α were equivalent with regard to induction of ISGs in the liver and PBMC. Thus, we chose to use peg-IFN-α2 as the human IFN-α during the cross-over phase, because the use of peg-IFN-α would be more practical in future long-term antiviral studies.

Circulating IFN-α levels were determined by enzyme-linked immunosorbent assay (ELISA) at each time point (Fig. 2A). Chimpanzee and human IFN-α levels ranged from 50 to 150 pg/mL at 4 hours post-inoculation, whereas human peg-IFN-α levels were 200 to 260 pg/mL. By 24 hours, the non-peg IFNs had declined to less than 20 pg/mL, whereas peg-IFN-α persisted at levels above 150 pg/mL. Transcripts for a panel of ISGs (IP-10, ISG12, and ISG15) known to be highly expressed in the liver of HCV-infected chimpanzees from our previous microarray studies10, 11 were examined by quantitative RT-PCR (TaqMan). Surprisingly, the levels of ISG transcripts rapidly returned to baseline in all animals in both the liver and PBMC. At 4 hours post-IFN, IP-10 transcripts were increased by as much as 400- and 670-fold in the liver (Fig. 2C) and PBMC (Fig. 2D), respectively, only to return to near baseline levels by 8 hours post-IFN, a time at which peg-IFN-α levels were still rising in the circulation. Serum levels of IP-10 followed a similar pattern, reaching peak levels of 4,700 pg/mL followed by return to near baseline levels by 24 hours (Fig. 2B). These data indicate that parts of the IFN signal transduction-transcriptional activation pathway were rapidly suppressed in vivo after high doses of IFN-α. The time required to regain responsiveness was not determined in these studies, but clearly exceeded 24 hours.

Figure 2.

Rapid downregulation of IFN-α response. (A) Serum IFN-α2 levels. Chimpanzees were inoculated with IFN-α as described in Fig. 1, and circulating IFN-α levels were measured by ELISA. (B) Serum levels of IP-10 were determined by ELISA. (C and D) Total cellular RNA was purified from liver (C) and PBMCs (D) at 0, 4, 8, and 24 hours post-inoculation of IFN-α. The levels of IP-10 (CXCL10) transcripts were determined by quantitative, real-time RT-PCR and were expressed as the fold-change in comparison with time 0, preinoculation. Despite persistent elevated levels of peg-IFN-α (A), IP-10 transcript (C-D) and protein (B) levels decreased rapidly.

Tissue-Specific Responses to IFN-α.

Microarray analysis of liver and PBMC RNA indicated a large number of genes consistently altered in expression after inoculation of IFN-α. The term ISG will be used for genes increased in expression by IFN, and the term interferon-regulated gene (IRG) will include both upregulated and downregulated. Genes were first selected to be significantly altered with a minimum of a twofold change in comparison with the baseline sample, and were then selected only if they were consistently altered in the majority of samples at a given time across all six experiments. A total of 1,778 IRGs were identified when liver and PBMC were combined. Some genes were counted more than once, because they are represented by different probe sets representing splice variants that may significantly affect function. The response to IFN-α was largely tissue specific, with 538 genes unique to the liver, 950 unique to PBMC, and 290 expressed in both tissues (Fig. 3; Tables 1-3; Supplement Tables 1-3). Although initially not anticipated, these results are not surprising considering that the total transcriptomes of these two tissues are highly specialized and divergent. Of the 1,778 IRGs, 1,044 were ISGs, with 236 unique to the liver, 544 unique to PBMC, and 264 common to both tissues. Many of the negatively regulated genes were down by twofold to threefold and may represent a decrease in transcription due to the diversion of transcriptional resources to the production of ISGs, as well as the degradation of housekeeping transcripts during a programmed stress response. However, a number of downregulated genes were consistently decreased by 10- to 40-fold in multiple animals at multiple time points and thus have the appearance of being specifically decreased by IFN. Although a mechanism for this regulation is not currently known, IFN induces a number of negative regulators. One striking example is the 25- to 38-fold decrease in FOS at 4, 8, and 24 hours in the liver (Table 2), but not in PBMCs (Table 3), and the down-regulation of numerous lymphocytic receptors in PBMCs, including killer cell lectin-like and natural killer (NK) cell receptors (Table 3).

Figure 3.

Tissue-specific response to IFN-α in chimpanzees. Microarray analysis was performed on total cellular RNA from liver and PBMC from chimpanzees inoculated with IFN-α as described in Fig. 1. A total of 1,778 IRGs were identified in combined liver and PBMC samples. The diagram illustrates the common IRGs detected in both tissues, the total IRGs detected in liver and PBMC, and the tissue-specific or unique IRGs detected in each tissue. PBMC, peripheral blood mononuclear cells; IRG, interferon-regulated gene.

Table 1. IFN Regulated Genes Common to Liver and PBMC*
Gene DescriptionGene SymbolAliasAvg. Fold Change of Six Experiments
LiverPBMC
4 hr8 hr24 hr4 hr8 hr24 hr
  • *

    Selected from a total of 290 genes altered in expression by twofold or more in both liver and PBMC. The complete table with gene ontology is available in supplementary data, and more inclusive data are available at http://www.sfbr.org/virology/lanford. The average fold change was calculated from six experiments with three chimpanzees at 0, 4, 8, and 24 hours after IFNα inoculation and was expressed as the fold change from the 0-hour point.

  • Aliases are provided for some genes, especially those that are better known by a name no longer used by the HUGO Gene Nomenclature Committee.

IFN-induced protein with tetratricopeptide repeats 2IFIT2ISG54477 1885
2′,5′-oligoadenylate synthetase 3, 100 kdOAS3 29243112114
IFN stimulated gene 20kdISG20 2845 85 
IFN-induced protein with tetratricopeptide repeats 3IFIT3RIG-G2613111296
IFN-induced protein 44-likeIFI44L 24311019136
Chemokine (C-X-C motif) ligand 10CXCL10IP-10246 315 
Radical S-adenosyl methionine domain 2RSAD2viperin202451995
IFN, alpha-inducible protein (clone IFI-15K)G1P2ISG1520231414134
Myxovirus (influenza virus) resistance 1MX1 1918161274
IFN-induced protein with tetratricopeptide repeats 1IFIT1ISG56181371696
Chemokine (C-C motif) ligand 8CCL8MCP-2167 146 
IFN regulatory factor 7IRF7 151014763
Suppressor of cytokine signaling 3SOCS3 1422 4  
IFN induced with helicase C domain 1IFIH1MDA5138 83 
DEAD (Asp-Glu-Ala-Asp) box polypeptide 58DDX58RIG-I135 104 
2′,5′-Oligoadenylate synthetase 2, 69/71 kdOAS2 11115753
2′,5′-Oligoadenylate synthetase 1, 40/46 kdOAS1 1196643
Ubiquitin-specific protease 18USP18 1044239 
28 kd IFN responsive proteinIFRG28 93463 
Signal transducer and activator of transcription 1STAT1 75464 
Phospholipid scramblase 1PLSCR1 74 53 
Apo B mRNA editing enzyme 3GAPOBEC3GCEM157  3  
Likely ortholog of mouse D11lgp2LGP2 67 1095
GTP binding protein 2GTPBP2 66 3  
Tripartite motif-containing 22TRIM22STAF5066 33 
G binding protein 4GBP4Mpa253 4  
Transporter 1TAP1RING453 3  
TNF (ligand) superfamily, member 10TNFSF10TRAIL5  64 
IFN, alpha-inducible protein (clone IFI-6-16)G1P3IFI616454894
Ubiquitin-conjugating enzyme E2L 6UBE2L6UBCH844353 
IFN regulatory factor 1IRF1 4  2  
Tripartite motif-containing 5TRIM5RNF8834 3  
Proteasome subunit, beta type, 9PSMB9RING1234 2  
Signal transducer and activator of transcription 3STAT3 33  2 
TNF (ligand) superfamily, member 13bTNFSF13B 33 53 
Myxovirus (influenza virus) resistance 2 (mouse)MX2 32 753
Tripartite motif-containing 21TRIM21 3  3  
Adenosine deaminase, RNA-specificADAR 23 32 
Signal transducer and activator of transcription 2STAT2 23 33 
Myeloid differentiation primary response gene (88)MYD88 2  22 
IFN regulatory factor 2IRF2 2  3  
IFN gamma receptor 1IFNGR1  3  3 
v-fos FBJ murine osteosarcoma viral oncogene homologFOS −38−25−2632 
Transmembrane protein 14ATMEM14APTD011−3−5  −2 
Nuclear receptor subfamily 1, group D, member 2NR1D2 −3−3  −2 
Table 2. IFN Regulated Genes Unique to Liver*
Gene DescriptionGene SymbolAliasAvg. Fold Change of Six Experiments
LiverPBMC
4 hr8 hr24 hr4 hr8 hr24 hr
  • *

    Selected from a total of 538 genes altered in expression by twofold or more in liver but not in PBMC. The complete table with gene ontology is available in supplementary data, and more inclusive data are available at http://www.sfbr.org/virology/lanford. Analyses were performed as described in Table 1.

  • Aliases are provided for some genes, especially those that are better known by a name no longer used by the HUGO Gene Nomenclature Committee.

Chemokine (C-X-C motif) ligand 11CXCL11I-TAC2125    
Chemokine (C-C motif) ligand 2CCL2MCP-120     
SPRY domain-SOCS box proteinSSB1 67    
Nucleotide-binding oligomerization domains 27NOD27 63    
Tripartite motif-containing 38TRIM38 6     
ATP-binding cassette,MDR/TAP 4ABCB4MDR2/353    
IL 15 receptor, alphaIL15RA 5     
Tryptophanyl-tRNA synthetaseWARSTrpRS410    
Serum amyloid A1/// serum amyloid A2SAA1/2 489   
C-reactive protein, pentraxin-relatedCRP 466   
Toll-like receptor 3TLR3 4     
IFN, alpha-inducible protein 27IFI27ISG12364   
Leptin receptorLEPR 343   
MAP kinase kinase kinase 5MAP3K5MAPKKK533    
SAM domain and HD domain 1SAMHD1 32    
Myeloid cell leukemia seq 1 (BCL2-related)MCL1 32    
Tripartite motif-containing 56TRIM56 32    
Apolipoprotein L, 3APOL3 3     
IL28 receptor, alpha (IFN, lambda receptor)IL28RA 3     
Caspase 7, apoptosis-relatedCASP7 3     
IL 1 receptor, type IIL1R1 23    
DEAD box polypeptide 3, X-linkedDDX3XDDX142     
Serine (or cysteine) proteinase inhibitor BSERPINB9 2     
IL 15IL15 2     
Hepatocellular carcinoma-assoc gene TD26LOC55908 −16−5    
Visinin-like 1VSNL1 −10−24    
G protein-coupled receptor 30GPR30GPCR-Br−10−9    
Thyroid hormone responsiveTHRSPSPOT14−5−33−5   
Fatty acid synthaseFASN −3−5    
Stearoyl-CoA desaturase (delta-9-desaturase)SCD  −7−9   
Table 3. IFN Regulated Genes Unique to PBMC*
Gene DescriptionGene SymbolAliasAvg. Fold Change of Six Experiments
LiverPBMC
4 hr8 hr24 hr4 hr8 hr24 hr
  • *

    Selected from a total of 950 genes altered in expression by twofold or more in PBMC but not in liver. The complete table with gene ontology is available in supplementary data, and more inclusive information is available at http://www.sfbr.org/virology/lanford. Analyses were performed as described in Table 1.

  • Aliases are provided for some genes, especially those that are better known by a name no longer used by the HUGO Gene Nomenclature Committee.

PGDF receptor-likePDGFRLPRLTS   28  
Lysosomal-assoc membrane prot 3LAMP3    2011 
Regulator of G-protein signalling 1RGS1    20  
Defensin, beta 1DEFB1    1513 
Ras and Rab interactor 2RIN2    107 
TNF, alpha-induced protein 6TNFAIP6    9  
Serine (cysteine) proteinase inhibSERPING1    795
Toll-like receptor 7TLR7    67 
Lectin, galactoside-binding s9LGALS9galectin 9   65 
ApoB mRNA editingAPOBEC3B    65 
C-type lectin domain family 4 ECLEC4E    54 
C-type lectin domain family 4 DCLEC4D    45 
Suppressor of cytokine signaling 1SOCS1CIS1   4  
Chemokine (C-C motif) receptor 1CCR1MIP1aR   35 
Toll-like receptor 2TLR2    34 
Lectin, galactoside-binding, s3LGALS3BP    34 
Colony-stimulating factor 3 receptorCSF3R    34 
Lectin, galactoside-binding, s2LGALS2    33 
Toll-like receptor 4TLR4    33 
Janus kinase 2JAK2    33 
Fas (TNF receptor superfamily 6)FASAPO-1   3  
Ubiquitin specific protease 25USP25    3  
IFN gamma receptor 2IFNGR2     3 
TGF beta-induced, 68 kDaTGFBI     3 
Toll-like receptor 8TLR8     3 
Toll-like receptor 5TLR5     3 
IFN regulatory factor 5IRF5     2 
Tetratricopeptide repeat domain 3TTC3    −17−5 
G protein-coupled receptor 133GPR133    −12−6 
T-cell receptor beta chain     −8 
C-type lectin domain family 2BCLEC2B     −7 
TNF receptor superfamilyTNFRSF25     −6 
Killer cell lectin-like receptorKLRC4/KLRK1     −4 
Killer cell lectin-like receptorKLRF1NKp80    −4 
Killer cell lectin-like receptorKLRC1/KLRC2NKG2A/B    −4 
Killer cell Ig-like receptorKIR3DL2NKAT4    −4 
T cell receptor alpha locusTRAa     −4 
T cell receptor alpha/delta locusTRAα/Y     −4 
T cell receptor beta constant 1TRBC1     −4 
TNF receptor superfamily 7TNFRSF7     −3 
Killer cell lectin-like receptorKLRK1NKG2D    −3 
Killer cell lectin-like receptorKLRB1CD161    −3 
Lymphocyte antigen 9LY9CD229    −3 
Chemokine (C-C) receptor 7CCR7BLR2    −3 
CD244 NK cell receptor 2B4CD244NKR2B4    −3 
Killer cell Ig-like receptorKIR2DL4CD158D    −2 

In a separate analysis, the raw signal values were subjected to paired t tests for selection of genes with a P value of <.002. This analysis resulted in a partially overlapping set of genes compared with those selected by the first method (see http://www.sfbr.org/virology/lanford). A total of 2,143 IRGs were identified in the liver (1,705) and PBMC (438) by this approach. Genes lacking from the first analysis and present in this one were for the most part altered by less than 2.0-fold. Genes present in the first but not the second analysis often exhibited significant differences in baseline values between animals, but were nonetheless highly induced. Many methods have been developed for microarray analysis, and each has unique strengths and weaknesses.

Antiviral Genes.

The classical IFN antiviral genes were expressed by both liver and PBMC; such as MX1, MX2, and OAS1-3. Other genes with known antiviral activity included viperin (RSAD2),14 phospholipid scramblase, and ADAR. Many genes involved in the IFN response pathway were commonly upregulated in both tissues as well, including STAT1, STAT3, interferon regulatory factor 2 (IRF2), IRF5, IRF7, and IRF9 (ISGF3G) (Table 1). ISG15 was commonly upregulated and encodes a ubiquitin-like protein that is conjugated to many proteins (ISGylation), is involved in increasing the stability of activated STAT1,25 and may act as a cytokine enhancing the activity of NK cells.26

RIG-I is the dsRNA binding helicase that activates IRF3 and nuclear factor-kappaB through a caspase recruitment domain and results in the induction of type 1 IFN.27 Along with TLR3, RIG-I is one of the main mechanisms for dsRNA induction of type 1 IFN. The HCV NS3/4a protease28, 29 blocks activation of IRF3 by this pathway via cleavage of a newly identified adapter protein (CARDIF, IPS-1, MAVS, VISA).30–33 RIG-I was upregulated in both liver and PBMC (Table 1). Mda5 is a related dsRNA binding helicase involved in this pathway28, 34 that was also upregulated in both tissues. Recently, LGP2 has been described as a dsRNA binding helicase similar to RIG-I and Mda5, which lacks the activation domain and appears to function as a negative regulator of this pathway.34, 35 LGP2 was up-regulated in both liver and PBMC (Table 1) and was also induced in hepatocytes by IFN-α) but not by IFN-γ.

Toll-like Receptors.

A number of toll-like receptors (TLRs) were induced by IFN-α, and each was specific to either the liver or PBMC. TLR3 was specific to the liver (Table 2), whereas 2, 4, 5, 7, and 8 were induced in PBMC (Table 3). MyD88, an adapter protein involved in TLR signaling, was induced in both liver and PBMCs (Table 1). Clearly, exposure to IFN-α induces a heightened state of detection of pathogen-associated molecular patterns via TLRs. Because the ligand for TLR3 is dsRNA, the increased expression of TLR3 in the liver may play a role in maintaining the high ISG expression level during HCV infection. However, in infected cells, TLR3 signaling is suppressed by NS3/4a protease cleavage of its essential adapter protein TRIF.29, 36

Chemokines.

Of particular interest is the expression of IP-10 (CXCL10) and ITAC (CXCL11), both of which are CXCR3, Th1 chemokines and are important for recruitment of activated T and NK cells to sites of inflammation. Both are upregulated in the liver during acute and chronic HCV infection.10, 11, 37 Whereas ITAC was the most highly upregulated liver-specific gene (25-fold; Table 2), IP-10 was upregulated in both liver and PBMC (24- to 31-fold; Table 1). The expression of IP-10 by PBMC and high serum levels of IP-10 during chronic infection suggests this chemokine does not function in the homing of T cells to the liver during hepatitis. Thus, ITAC may be the more important CXCR3 chemokine for this function during HCV infection. Although in the current context the increase in ITAC expression was specific to the liver, its expression is known to be involved in the recruitment of T and NK cells to other sites of inflammation. MCP-1 (CCL2), a CC chemokine important for recruitment of macrophages and NK cells to the liver during viral infection,38 was also highly up-regulated and liver specific (Table 2).

Differential Response of Hepatocytes to IFN-α and IFN-γ.

Primary chimpanzee and human hepatocytes were examined for the hepatocyte-specific IFN response (a response lacking the contribution of nonparenchymal cells of the liver). Primary hepatocytes were treated with both IFN-α and IFN-γ. Three experiments were performed, two on human hepatocytes and one on chimpanzee hepatocytes. In each experiment, samples were harvested after 4 and 8 hours of IFN induction. A total of 563 IRGs were consistently altered across different experiments and times, and of these, 292 were unique to IFN-α, 84 were unique to IFN-γ, and 187 were altered by both IFNs (Fig. 4; Tables 4-6; Supplement Tables 4-6). Of the 563 IRGs, 443 were ISGs, with 219 unique to IFN-α, 59 unique to IFN-γ, and 165 altered in expression by both IFNs.

Figure 4.

IFN-α and IFN-γ regulated genes in primary hepatocytes. Microarray analysis was performed on total cellular RNA extracted from primary chimpanzee and human hepatocytes (3 experiments) at 4 and 8 hours after treatment with IFN-α or IFN-γ. A total of 563 IRGs were detected in combined IFN-α and IFN-γ treated samples. The diagram illustrates the IRGs common to hepatocytes treated with IFN-α or IFN-γ, the total IRGs detected in cells treated with either IFN-α or IFN-γ, and the IRGs specific or unique to hepatocytes treated with each IFN. IFN-α, interferon alpha; IFN-γm interferon gamma; IRG, interferon-regulated gene.

Table 4. IFN Regulated Genes Common to IFN-α and IFN-γ in Hepatocytes*
Gene DescriptionGene SymbolAliasAvg. Fold Change
IFN-αIFN-γ
4 hr8 hr4 hr8 hr
  • *

    Selected from a total of 187 genes altered in expression by twofold or more by both IFN-α and IFN-γ in human and chimpanzee primary hepatocyte cultures. The complete table including gene ontology is available in supplementary data, and more inclusive data are available at http://www.sfbr.org/virology/lanford. The average fold change was calculated from three experiments with samples at 0, 4, and 8 hours after IFN treatment and was expressed as the fold-change from the 0-hour point.

  • Aliases are provided for some genes, especially those that are better known by a name no longer used by the HUGO Gene Nomenclature Committee.

Radical S-adenosyl methionine domain 2RSAD2viperin1048477
IFN-induced protein 44-likeIFI44L 991191221
Chemokine (C-X-C motif) ligand 11CXCL11I-TAC698247104
IFN-induced prot; tetratricopeptide repeats 2IFIT2ISG54481588
IFN-induced prot; tetratricopeptide repeats 1IFIT1ISG56453753
Chemokine (C-X-C motif) ligand 10CXCL10IP-1042514357
G binding protein 5GBP5 363657179
Chemokine (C-C motif) ligand 8CCL8MCP-2219125
DEAD (Asp-Glu-Ala-Asp) box polypeptide 58DDX58RIG-I161333
IFN induced with helicase C domain 1IFIH1MDA5141344
Suppressor of cytokine signaling 1SOCS1 14 77
IFN stimulated gene 20kDaISG20 1320610
Ubiquitin specific protease 18USP18 1214 2
IFN, alpha-inducible protein (clone IFI-15K)G1P2ISG15111333
Myxovirus (influenza virus) resistance 1MX1 101333
G binding protein 1, IFN-inducible, 67kDaGBP1 991216
TNF (ligand) superfamily, member 10TNFSF10TRAIL9445
Chemokine (C-X3-C motif) ligand 1CX3CL1 9 1314
IFN-induced protein 44IFI44MTAP44812 3
2′-5′-oligoadenylate synthetase 3, 100kDaOAS3 81124
2′,5′-oligoadenylate synthetase 1, 40/46kDaOAS1 810 3
Chemokine (C-X-C motif) ligand 9CXCL9MIG854249
Apo B mRNA editing enzyme 3GAPOBEC3G 79 3
G binding protein 4GBP4Mpa275816
2′,5′-oligoadenylate synthetase 2, 69/71kDaOAS2 6923
Signal transducer and activator of transcription 1STAT1 6847
Tripartite motif-containing 22TRIM22STAF506734
Signal transducer and activator of transcription 1STAT1 5635
Transporter 1 (MDR/TAP)TAP1RING45568
TNF (ligand) superfamily, member 13bTNFSF13B 54 3
TNF, alpha-induced protein 3TNFAIP3 525 
TNF, alpha-induced protein 3TNFAIP3 5242
TNF receptor superfamily, member 10aTNFRSF10A 464 
Janus kinase 2JAK2 4445
IL 15IL15 4436
Transporter 2 (MDR/TAP)TAP2RING114434
IFN regulatory factor 1IRF1 4 1011
Phospholipid scramblase 1PLSCR1MMTRA1B34 2
MHC, class I, EHLA-E 33 3
Chemokine (C-C motif) ligand 2CCL2 2 36
Glycine-N-acyltransferaseGLYATGAT−29−26−22−23
Complement factor HCFH −11−4−5−10
Alcohol dehydrogenase 1ABC, gammaADH1A 1B 1C −7−32−4−14
Alcohol dehydrogenase IB (class I), betaADH1B −5−6−6−7
Alcohol dehydrogenase 4 (class II), piADH4 −3−2 −9
Table 5. IFN Regulated Genes Unique to IFN-α in Hepatocytes*
Gene DescriptionGene SymbolAliasAvg. Fold Change
4 hr8 hr
  • *

    Selected from a total of 292 genes altered in expression by twofold or more by IFN-α but not IFN-γ in human and chimpanzee primary hepatocyte cultures. The complete table is available in supplementary data, and more complete data are available at http://www.sfbr.org/virology/lanford, including the gene ontology data. Analyses were performed as described in Table 4.

  • Aliases are provided for some genes, especially those that are better known by a name no longer used by the HUGO Gene Nomenclature Committee.

Hexokinase 2HK2 307
Promyelocytic leukemiaPML 2912
3′ Repair exonuclease 1TREX1DRN32019
Angiopoietin-like 1ANGPTL1ANG31742
Lysosomal-associated membrane protein 3LAMP3 1727
2′,5′-Oligoadenylate synthetase-likeOASL 1310
IL 28 receptor, alpha (IFN, lambda receptor)IL28RAIFNλR11 
Lymphocyte-activation gene 3LAG3 9 
Likely ortholog of mouse D11lgp2LGP2 88
IFN regulatory factor 7IRF7 78
Guanosine monophosphate reductaseGMPR 65
Myxovirus (influenza virus) resistance 2MX2 64
SMAD, mothers against DPP homolog 3SMAD3 5 
IFN, alpha-inducible protein 27IFI27ISG12410
IFN, alpha-inducible protein (clone IFI-6-16)G1P3IFI61646
Polyribonucleotide nucleotidyltransferase 1PNPT1 45
Apo B mRNA editing enzyme like 3G-3FAPOBEC3G-F 44
G protein-coupled receptor 37GPR37 4 
RAS and EF hand domain containingRASEF 4 
Toll-like receptor 3TLR3 33
Cyclin D3CCND3 33
Serine (or cysteine) proteinase inhibitor B1SERPINB1 33
Myeloid differentiation gene (88)MYD88 33
Tripartite motif-containing 25TRIM25 3 
BCL2-like 13 (apoptosis facilitator)BCL2L13 3 
Caspase 1, (IL 1, beta, convertase)CASP1 23
IFN-stimulated transcription factor 3, gammaISGF3GIRF923
MHC, class I, FHLA-F 2 
Signal transducer and activator of transcription 2STAT2 2 
Lectin, galactoside-binding, soluble, 9 (galectin 9)LGALS9  4
Signal transducer and activator of transcription 2STAT2  4
Ubiquitin-activating enzyme E1-likeUBE1LD8 3
Endothelial protein C receptorPROCREPCR 2
G binding protein 3GBP3  2
Hepatocellular carcinoma-associated gene TD26LOC55908  2
GTP binding protein 1GTPBP1  2
Ubiquitination factor E4B (UFD2 homolog, yeast)UBE4BUBOX3−6−6
RAB40B, member RAS oncogene familyRAB40BSEC4L−6−3
IL-6 signal transducerIL6ST  −3
TNF receptor superfamily 10dTNFRSF10D  −2
Table 6. IFN Regulated Genes Unique to IFN-γ in Hepatocytes*
Gene DescriptionGene SymbolAliasAvg. Fold Change
4 hr8 hr
  • *

    Selected from a total of 84 genes altered in expression by twofold or more by IFN-γ but not IFN-α in human and chimpanzee primary hepatocyte cultures. The complete table is available in supplementary data, and more complete data are available at http://www.sfbr.org/virology/lanford, including the gene ontology data. Analyses were performed as described in Table 4.

  • Aliases are provided for some genes, especially those that are better known by a name no longer used by the HUGO Gene Nomenclature Committee.

Apolipoprotein L, 4APOL4 819
Tripartite motif-containing 31TRIM31HCG159
Nuclear localized factor 1NLF1 417
Intercellular adhesion molecule 1ICAM1CD5445
Ubiquitin DUBD 35
IFN regulatory factor 8IRF8ICSBP133
Regulator of G-protein signalling 14RGS14 22
TNF alpha-induced protein 2TNFAIP2 22
Indoleamine-pyrrole 2,3 dioxygenaseINDOCD107B 19
Visinin-like 1VSNL1  5
CD40 (TNF receptor superfamily member 5)CD40TNFRSF5 5
Opioid growth factor receptorOGFR  4
Proteasome subunit, beta type, 10PSMB10LMP10 3
MHC class II, DMAHLA-DMA  3
Tripartite motif-containing 31TRIM31HCG1 3
Cathepsin SCTSS  3
MHC, class II, DRAHLA-DRA  2
Interleukin 8IL8CXCL8 −2
C-reactive protein, pentraxin-relatedCRP  −2

Most of the genes of interest in IFN-α–treated hepatocytes were discussed in the context of the in vivo experiments. Of interest in the in vitro studies was the common induction of the chemokines ITAC (CXCL11), IP-10 (CXCL10), MIG (CXCL9), and MCP2 (CCL8) by IFN-α and IFN-γ (Table 4). Although IP-10 was originally identified as an IFN-γ-specific transcript, its induction by IFN-α is well documented.19 In contrast, MIG is still considered to be an IFN-γ-specific transcript and a sensitive marker for IFN-γ function.39 The current analysis demonstrated that MIG can be induced in hepatocytes by IFN-α, although induction by IFN-γ was 5- to 10-fold greater. The dsRNA binding proteins involved in the induction of type 1 IFN, RIG-I, MDA5, and TLR3 were detected in hepatocytes, with TLR3 being unique to IFN-α (Table 5). RIG-I and MDA5 were induced in hepatocytes by both IFN-α and IFN-γ, whereas the negative regulator of this pathway,LGP2, was induced only by IFN-α (Tables 4 and 5). Although TLR2 was not increased in vivo in the liver by the twofold cutoff, it was detected in IFN-α–treated hepatocytes. The IFN lambda receptor or IL28RA was also induced by IFN-α in hepatocytes. IFN-γ–specific transcripts included ICAM1, IRF8, and the HLA class II molecules (DMA, DRA) (Table 6). HLA-E and F are nonclassical class I molecules that were induced by both IFN-α and IFN-γ and are induced during acute and chronic HCV infections.10, 11

A number of the transcripts detected in vivo in the liver were noticeably absent from the hepatocyte analyses. Some of these may be expressed by nonparenchymal cells in the liver, whereas others may be induced in hepatocytes by a cytokine secreted by nonparenchymal cells in response to IFN-α. For example, C-reactive protein (Table 2) is an acute-phase reactant that was induced in the liver in vivo by IFN-α but not in hepatocytes. C-reactive protein is also induced in the liver during acute and chronic HCV infection.10, 11In vivo, IFN-α likely induces nonparenchymal liver cell types to produce cytokines known to induce C-reactive protein in hepatocytes (e.g., IL-1b or IL-6). The same may be true for serum amyloid protein (SAA1-2; Table 2). A total of 171 ISGs were shared by both the liver and hepatocytes, showing a remarkable consistency for the induction of many of the genes detected in both analyses.

Discussion

In this study, we have demonstrated (1) that the in vivo response to IFN-α is rapidly downregulated in the liver and PBMC; (2) that the response to IFN-α is largely tissue specific, with a total of 1,778 IRGs altered in expression by twofold or more; (3) that the chimpanzee response to chimpanzee and human IFN-α are indistinguishable; and (4) that the hepatocyte-specific responses to IFN-α and IFN-γ overlap considerably but are also highly specialized. Previous microarray studies have characterized the genomic response to IFN-α, IFN-γ, and dsRNA.16, 40–42 The ISGs detected in this study were compared with those seen in PBMC from HCV-infected patients treated with IFN-α. Because the studies employed different arrays with different gene sets, the data are not directly comparable. Ji and coworkers16 identified 1,035 IRGs, of which only 366 were increased by twofold or more; the cutoff set for our studies. We identified 199 genes common to the two data sets (marked by * in Supplemental Tables 1-3), suggesting a good concordance between the studies. Our studies significantly expand the known IFN-responsive genes by analyzing the response both in vivo and in vitro, in liver, PBMC, and primary hepatocytes, and by using a total genome microarray. This data set will serve as an essential reference source for future studies on IFN function.

One of the initial impetuses for conducting this study with both chimpanzee and human IFN-α were the previous failures of human IFN-α treatment of HCV-infected chimpanzees, including a study involving high-level expression of IFN-α in the liver by adenovirus-based gene therapy.21, 22 Because the response in the liver and PBMC of chimpanzees was similar for human and chimpanzee IFN-α in this study, the previous failures in therapy were not likely due to a lack of response of chimpanzees to human IFN-α. Rather, at least for the small group of animals adequately studied, HCV-infected chimpanzees may represent null responders. Further evaluation of this phenomenon in chronically infected chimpanzees using newer methodologies will be required to approach the apparent defect involved in lack of response.

Downregulation of the IFN transcriptional response was recognized soon after the discovery of transcripts induced by IFN-α.43, 44 The rapid down-regulation of ISG transcripts in the liver and PBMC after IFN treatment might be facilitated by several mechanisms. Early studies indicated that depletion of the receptor is not the primary mechanism of downregulation, but rather the induction of factors that perpetuate the desensitization as long as IFN-α remains present.43 A number of transcriptional inhibitors have been implicated in downregulation of the IFN response: IRF2, ICSAT, IRF8 (ICSBP), and STAT1B45–47 IRF2 was upregulated in liver, PBMC, and primary hepatocytes. STAT1β is an alternate splice variant of STAT1 that is a dominant-negative form of the protein that binds DNA but cannot activate transcription. STAT1β was upregulated in this study and was preferentially upregulated over STAT1α in acute HCV infection.10 In addition, the proteins involved in signal transduction are modulated posttranslationally by ubiquitination (suppressor of cytokine signaling; SOCS), sumoylation (protein inhibitor of IFN alpha signaling), and tyrosine-dephosphorylation. JAK and STAT proteins are ubiquitinated and degraded.48 SOCS-induced ubiquitination has been implicated in degradation of JAKs. After IFN-α treatment, SOCS3 was upregulated in both liver and PBMC, and SOCS1 was induced in PBMC. Recent studies have demonstrated that SOCS1 associates directly with IFNAR1, a component of the type 1 IFN receptor, thereby blocking activation of STAT1.49

The magnitude and synchrony of the in vivo downregulation of ISG transcription, especially in the presence of circulating peg-IFN-α, was not anticipated, but clearly must be evaluated for its role in the kinetics of viral clearance of HCV during IFN-α therapy. The duration of the downregulation was not determined in this study but must exceed 24 hours. Previous in vitro studies have suggested that 72 hours was required to regain full responsiveness after removal of IFN-α43 and that return to complete responsiveness required the removal of IFN-α. It seems likely that this is not readily accomplished while peg-IFN persists in the circulation, and repeated dosing of peg-IFN-α during therapy may prevent restoration of full responsiveness, such that the liver is not capable of the high-level ISG response observed immediately following the first dose of IFN-α. However, whether a difference exists between chimpanzees and humans in the kinetics or duration of the IFN nonresponsive period or whether an increase in the downregulation of the IFN pathways in chimpanzees contributes to the lack of virological response to IFN-α previously observed in HCV-infected chimpanzees is not known.

The current studies were conducted in uninfected animals, and during chronic HCV infection, the ISG transcript levels in the liver are already highly induced.10–15 We have proposed a model11 that reconciles the in vitro observations that HCV proteins block the dsRNA-type I IFN response with the in vivo observations that HCV-infected livers have high levels of ISG transcripts. Based on the level of viral RNA in the liver, we concluded that only a small percentage of hepatocytes are infected, and that turnover of infected hepatocytes results in a constant supply of newly infected cells that secrete IFN until sufficient viral protein is available to inhibit this pathway. Presumably, the secreted IFN induces zones of resistant cells, resulting in a loss of available replication space in the liver. The secretion of IFN from a minority of the cells may be sufficient to induce an antiviral response that protects most of the hepatocytes from infection, thus maintaining a low level of infection in the liver. Several other scenarios are possible, including the infection of a limited number of hepatocytes due to differences in differentiation status in zones 1 through 3. Recent observations suggest the pretreatment levels of ISG transcripts in the liver are predicators of therapy outcome, with lower pretreatment levels favoring sustained viral clearance.13 The ISG response induced by IFN-α administration in HCV chronically infected livers has not been examined; thus, whether low pretreatment ISG levels provide an opportunity for greater response to exogenously provided IFN is not known.

Whether downregulation of the ISG response due to inoculation with high levels of exogenous IFN-α plays a significant role in viral kinetics during therapy remains to be determined, but one could envision a scenario whereby this lack of responsiveness fits current data on viral kinetics. As previously hypothesized, the rapid decline of virus during phase I would be due to the initial inhibition of viral secretion brought on by synchronous induction of ISGs by the first dose of IFN-α.7–9 After phase I, a brief rebound of virus is often observed,9 which may occur during the downregulation period observed in the current study. The decreased slope in phase II may be due to the cumulative effects of reduction in the number of infected cells as previously hypothesized,7 as well as a reduction in efficacy of IFN, with efficacy being negatively impacted by a decreased responsiveness to IFN in cells with partially or completely downregulated IFN-α pathways. The inability to quantify infected cells in the liver and the number of HCV genomes per infected cell makes this a particularly challenging distinction to resolve.

Acknowledgements

The authors acknowledge the excellent technical assistance of Sharon Kuss and biostatistics advice by Dr. Mark Sharp.

Ancillary