Interferon signaling and treatment outcome in chronic hepatitis C

Abstract

Hepatitis C virus (HCV) infection is a major cause of chronic liver disease worldwide. The current standard therapy for chronic hepatitis C (CHC) consists of a combination of pegylated IFN alpha (pegIFNα) and ribavirin. It achieves a sustained viral clearance in only 50–60% of patients. To learn more about molecular mechanisms underlying treatment failure, we investigated IFN-induced signaling in paired liver biopsies collected from CHC patients before and after administration of pegIFNα. In patients with a rapid virological response to treatment, pegIFNα induced a strong up-regulation of IFN-stimulated genes (ISGs). As shown previously, nonresponders had high expression levels of ISGs before therapy. Analysis of posttreatment biopsies of these patients revealed that pegIFNα did not induce expression of ISGs above the pretreatment levels. In accordance with ISG expression data, phosphorylation, DNA binding, and nuclear localization of STAT1 indicated that the IFN signaling pathway in nonresponsive patients is preactivated and refractory to further stimulation. Some features characteristic of nonresponders were more accentuated in patients infected with HCV genotypes 1 and 4 compared with genotypes 2 and 3, providing a possible explanation for the poor response of the former group to therapy. Taken together with previous findings, our data support the concept that activation of the endogenous IFN system in CHC not only is ineffective in clearing the infection but also may impede the response to therapy, most likely by inducing a refractory state of the IFN signaling pathway.

Hepatitis C virus (HCV) infection is a major cause of chronic liver disease worldwide. An important and striking feature of hepatitis C is its tendency toward chronicity. In >70% of infected individuals, HCV establishes a persistent infection over decades that may lead to cirrhosis and hepatocellular carcinoma. An interesting hypothesis in HCV biology proposes that the viral NS3-4A protease not only processes the viral proteins but also cleaves and inactivates components of the intracellular sensory pathways, TRIF and Cardif, that detect viral infection and induce the transcriptional activation of type I IFN. Two RNA helicases, RIG-I and MDA5, identified as intracellular sensors of dsRNA, act through Cardif to induce IFNβ production (1). The ability of HCV to inhibit the activation of the endogenous type I IFN system could underlie its success in establishing a chronic infection (2).

Type I IFNs not only are crucial factors in the innate immune system but also are the most important components of current therapies against CHC. The current standard therapy consists of pegylated IFNα (pegIFNα) and the antiviral agent ribavirin (3). The treatment achieves a sustained virological response (SVR) in ≈55% of patients, with significant differences between genotypes (4). An SVR is defined as the loss of detectable HCV RNA during treatment and its continued absence for at least 6 months after stopping therapy. Studies of long-term followup on SVR patients demonstrate that this response is durable in >95% of patients. The probability of an SVR strongly depends on the early response to treatment. Patients who do not show an early virological response (EVR), defined as a decline of the viral load by >2 log₁₀ after 12 weeks of therapy, are highly unlikely to develop an SVR (5–7). Patients with an EVR have a good chance of being cured, with 65% of them achieving SVR (5, 7). The prognosis is even better for patients who have a rapid virological response (RVR), defined as serum HCV RNA undetectable after 4 weeks of treatment. Over 85% of them will achieve SVR (6, 8). Unfortunately, <20% of patients with genotype 1 and ≈60% of patients with genotypes 2 or 3 show an RVR (6, 8). The host factors that are important for an early response to therapy are currently unknown.

Type I IFNs achieve their potent antiviral effects through the regulation of hundreds of IFN-stimulated genes (ISGs). ISG-encoded proteins establish a general antiviral state within the cell (9). IFNs induce ISG transcription by activating the Jak-STAT pathway (10). Type I IFNs bind to the same cell surface receptor (IFNAR) and activate the receptor-associated tyrosine kinases Jak1 and Tyk2. The kinases then phosphorylate and activate STAT1 and STAT2. The activated STATs translocate to the nucleus, where they bind specific DNA elements in promoters of ISGs. Many ISGs have antiviral activity, but some are involved in other processes such as lipid metabolism, apoptosis, protein degradation, and inflammatory cell responses (11). HCV interferes with the IFN system probably at multiple levels. IFN-induced Jak-STAT signaling is inhibited in cells and transgenic mice that express HCV proteins (12, 13) and in liver biopsies of patients with CHC (14). In vitro, HCV proteins NS5A and E2 bind and inactivate protein kinase R, an important antiviral protein (15). However, the molecular mechanisms that are important for the response to pegIFNα in patients with CHC remain unknown.

The capacity of HCV to interfere with the IFN pathway at many different levels is a likely mechanism underlying HCV success to establish a chronic infection (2). However, quite paradoxically, in chimpanzees acutely or chronically infected with HCV, hundreds of ISGs are induced in the liver (16, 17). Nevertheless, despite the activation of the endogenous IFN system, the virus is not cleared from chronically infected animals (18). The results obtained with chimpanzees are difficult to extrapolate to humans, because there are important differences in the pathobiology of HCV infection between these species. Whereas most chimpanzees acutely infected with HCV clear the virus spontaneously, infections in humans mostly become chronic. However, chronically infected chimpanzees can rarely be cured with IFN, whereas more than half of patients with CHC are successfully treated (19).

Induction of ISGs was also found in pretreatment liver biopsies of many patients with CHC, again demonstrating that HCV infection can lead to activation of the endogenous IFN system (20). Notably, patients with preelevated expression of ISGs tended to respond poorly to therapy when compared with patients with low initial expression (20). The cause of this differential response to therapy is not understood. Are patients with elevated initial expression refractory to further stimulation of ISGs by exogenous IFN? Does the administration of IFN to patients with low initial ISG values lead to ISG expression levels exceeding those found in the other group, possibly explaining a success of therapy in low-ISG patients? Are there specific ISGs important for viral clearance that are not activated in nonresponders? To circumvent limitations in the procurement of liver biopsies, several groups have assessed whether peripheral blood mononuclear cells (PBMCs) could serve as a surrogate tissue to evaluate the response to IFNα (21–23). However, no conclusive data are available for humans, because a direct comparison of IFN-induced ISGs between liver and PBMCs has not been done.

To approach focal questions related to the pathophysiology of HCV infection, we investigated IFN-induced signaling and ISG expression in paired liver biopsies and PBMCs collected from patients with CHC before and after the first injection of peg-IFNα. We correlated the biochemical and molecular data with the response to treatment, and we compared the response to IFNα in liver and PBMCs.

Results

Patients and Response to Treatment.

Sixteen patients included in this study, 6 women and 10 men, were treated with a weight-adjusted combination of s.c.-injected pegIFNα2b once weekly and oral ribavirin twice daily. All had two liver biopsies, the pretreatment biopsy (B-1) and the second biopsy (B-2), obtained 4 h after the first injection of pegIFNα2b. We have chosen to analyze gene expression 4 h after pegIFNα2b injection, because kinetics of the induction of ISGs by pegIFNα in liver of chimpanzees was maximal at this time and was followed by rapid down-regulation of many genes (22). We realize that we probably missed the up-regulation of some late-induced ISGs, but because of rapid down-regulation, we would have missed more ISGs when using later time points.

Seven of the patients were infected with HCV genotype (GT) 1, two with GT 4, four with GT 3, and three with GT 2. Eight patients who had negative serum HCV RNA after 4 weeks of treatment (RVR) and two patients with >3 log₁₀ drop of viral titer within the first 4 weeks were classified as rapid responders (RRs), whereas six patients showed a viral load reduction of <1.5 log₁₀ and were classified as non-RRs (Table 1).

Table 1.

Study patient characteristics

Patient no.	4-week response	Sex	Age	HCV GT	Viral load, log10 international units/ml			12-week response	Follow-up	Metavir	Weight, kg
					Baseline	4 weeks	12 weeks
1	RR	M	52	3a	7.14	Neg.	Neg.	EVR	SVR	A2/F2	75
2	RR	M	37	3a	4.90	Neg.	Neg.	EVR	SVR	A1/F2	73
3	RR	M	38	1a	6.91	Neg.	Neg.	EVR	EoTR	A2/F1	85
4	RR	M	33	2b	6.27	Neg.	Neg.	EVR	EoTR	A1/F2	57
5	RR	M	48	2b	6.67	Neg.	Neg.	EVR	SVR	A3/F4	110
6	RR	F	53	2a/c	4.95	Neg.	Neg.	EVR	SVR	A3/F3	74
7	RR	M	56	3	5.25	Neg.	Neg.	EVR	EoTR	A3/F4	61
8	RR	M	38	4	4.08	Neg.	Neg.	EVR	Ongoing	A2/F2	69
9	RR	F	50	1b	7.22	3.52	Neg.	EVR	Ongoing	A1/F2	47
10	RR	F	48	1	6.49	3.31	Neg.	EVR	Ongoing	A3/F4	60
11	Non-RR	F	54	3a	4.52	4.08	1.3	EVR	No EoTR	A3/F4	69
12	Non-RR	M	64	1b	6.24	4.83	3.46	EVR	No EoTR	A3/F4	74
13	Non-RR	M	49	4	6.91	5.87	5.22	PNR	-	A3/F4	102
14	Non-RR	M	56	1b	6.89	6.76	6.01	PNR	-	A2/F3	60
15	Non-RR	F	50	1a	7.11	6.58	6.35	PNR	-	A1/F2	77
16	Non-RR	F	47	1a	6.16	5.99	5.52	PNR	-	A2/F2	81

M, male; F, female; GT, genotype.

Serum IFNα concentrations were below the limit of detection in all patients before treatment and, in accordance with previously published pharmacokinetic data (24), between 34 and 360 pg/ml in samples obtained at 4 h after the pegIFNα2b injection (data not shown). There was no significant correlation between the virological response at week 4 and the serum IFNα concentration at 4 h postinjection.

IFN-Induced Regulation of Target Genes.

Gene expression was analyzed with Affymetrix U133plus2.0 arrays in B-1 and B-2 samples and also in PBMCs isolated from blood obtained before (PBMC-1) and 4 h after the first pegIFNα2b injection (PBMC-2). We identified 252 genes significantly [paired t test, P < 0.05, see supporting information (SI) Materials and Methods] changed >2-fold between B-1 and B-2 in >50% of the 10 RR biopsy samples, whereas only 36 genes passed the same criteria when analyzing the 6 non-RR patients (Fig. 1A). To compare the number of significantly regulated genes between RR and non-RR patients in groups with an equal number of patients, 15 groups of 6 patients randomly selected from the 10 RR patients were generated. In each group, the genes significantly (P < 0.05 or P < 0.01, paired t test) changed in >50% of patients were identified and counted. In liver biopsies of the 15 RR groups, the mean number (±SD) of regulated genes was 178.6 (±58.8) and 225.9 (±61.5) at significance levels P < 0.01 and P < 0.05, respectively. In the non-RR group, there were 23 and 36 genes significantly regulated at P < 0.01 and P < 0.05, respectively. The differences between the RR and non-RR samples were statistically significant (Fig. 1C).

Fig. 1.

PegIFNα2b induced regulation of gene expression in liver and PBMCs. (A) Venn diagram of genes significantly (paired t test, P < 0.05) up- or down-regulated >2-fold in response to pegIFN-α2b in >50% of the 10 RR and 6 non-RR biopsy samples. (B) Venn diagram of genes significantly (paired t test, P < 0.05) up- or down-regulated >2-fold in response to pegIFN-α2b in biopsy and PBMC samples of >50% of RR patients. (C) RR up- or down-regulate significantly more genes in the liver in response to pegIFN-α2b than non-RR patients. Shown are the mean (+SEM) number of genes changed >2-fold at significance levels P < 0.01 (lanes 1, 2, 5, and 6) and P < 0.05 (lanes 3, 4, 7, and 8) in >50% of patients within each response group in liver biopsies and PBMCs.

Not surprisingly, many of the regulated genes represent known ISGs. However, contrary to our expectations, expression levels of most of these ISGs were not higher in post-pegIFNα2b treatment biopsies from RR patients compared with non-RRs. Only 7.5% of the 252 genes significantly regulated in >50% of 10 RR patients were higher in the B-2 samples of RRs (see Table S1). Rather, non-RR patient samples had a higher level of ISG expression already in B-1, and the fold change in the B-2 samples was therefore only minor. This is illustrated in Fig. 2A at the example of four ISGs. The genes show a very low expression in biopsies from individuals without hepatitis C (controls) and in B-1s of RR patients. The six non-RR patients had a high expression of these genes before treatment, and pegIFNα2b administration did not increase or only minimally increased their expression. There were very few exceptions to this rule (an example is shown in Fig. 2B). These genes had low expression in the pretreatment biopsies, and pegIFNα2b induced them in all patients. Nevertheless, the predominant pattern of gene expression resembled this shown in Fig. 2A. A complete list and a heat map of the expression of 252 genes significantly (P < 0.05) changed >2-fold between B-1 and B-2 in the RR group are shown for all biopsy samples in Table S1 and Fig. S1.

Fig. 2.

PegIFNα2b induced gene regulation in HCV-infected patients shows major differences between livers of RR and non-RR patients. (A) Four ISGs (Viperin, Mda5/helicard, OAS1, USP18) were chosen from the list of genes significantly regulated >2-fold between B-1 and B-2 in RR patients. In the liver of non-RR patients, expression of these genes is already high before treatment (lanes 25–30) and does not further increase after pegIFNα (lanes 31–36). In RR patients, pretreatment expression (lanes 5–14) is similar to controls (lanes 1–4), and pegIFNα induces a strong up-regulation (lanes 15–24). The y axes display absolute expression values. (B) An example of a gene (CCL8) up-regulated in liver in response to pegIFN-α2b in both RR and non-RR patients. The x axis represents individual biopsy samples. Lanes 5–14 (B-1) and 15–24 (B-2) correspond to RR patients number 1–10 in Table 1 (in the same order), and lanes 25–30 (B-1) and 31–36 (B-2) correspond to non-RR patients numbers 11–16.

There was a considerable overlap of pegIFNα2b-regulated genes in liver and PBMCs (Fig. 1B). Interestingly, in all patients, pegIFNα2b regulated more genes in PBMCs than in liver. However, the difference in the up-regulation of ISGs was more pronounced in biopsies compared with the PBMCs (Fig. 1C). In PBMCs, no preactivation of ISGs was found, and pegIFNα2b treatment induced ISGs in both RR and non-RR patients (Fig. 1C and Fig. S2). This indicates that chronic HCV infection has strong local effects on the IFN system in liver but little effect in PBMCs.

A Subset of Genes That Predicts Response to Treatment.

Supervised classifier analysis of array data allows the identification of a subset of genes that best predicts the outcome, in our case rapid response vs. nonresponse at week 4. All liver biopsy and PBMC datasets were subjected to supervised classifier prediction using the response at 4 weeks of treatment as grouping criteria. For PBMC samples, the analysis did not identify a subset of genes that could predict the treatment outcome (Fig. S3 C and D). In contrast, a subset of 16 genes was identified in the liver B-2 samples that predicted response to treatment with an error rate of 16.1% using the K Nearest Neighbors test (Fig. S3B). An even better prediction was possible with a subset of 29 genes in the pretreatment biopsies B-1, where the error rate was 4.3% (Fig. S3A). In this set, there were 22 genes up-regulated by pegIFNα2b (Table S2). Therefore, 76% best predictor genes represent ISGs.

Contrary to the predominance of ISGs in the best predictor set from pretreatment biopsies, only 3 (19%) of the 16 best predictor genes derived from an analysis of the B-2 biopsies were ISGs (Table S3). These results support the findings shown in Fig. 2 that expression levels of most ISGs in B-2 do not differ between RR and non-RR samples and therefore are not suited for the discrimination of responders from nonresponders. Among the non-ISGs present in the B-1 and B-2 liver biopsy lists discussed above are genes having functions in signal transduction, cell cycle regulation, apoptosis, and amino acid metabolism.

RT-qPCR Analysis of ISG Expression in Liver Biopsies.

Array analysis of the paired liver biopsies emphasized the importance of ISG expression in B-1 biopsies for the outcome of therapy. To confirm these data, we measured by real-time quantitative PCR (RT-qPCR) the expression of selected ISGs in 16 patients with B-1 and B-2 biopsies and in pretreatment biopsies of 96 additional patients with CHC (Tables S4 and S5). In the 16 patients with paired biopsies, the RT-qPCR values matched well the array expression, validating the quality of the array data (Fig. S4A and data not shown). The expression of all four ISGs in pretherapy biopsies was significantly different between the RR and non-RR groups (Fig. 3A), further supporting the conclusion that there is an inverse correlation between the pretreatment expression of ISGs in liver and the response to IFNα therapy. A significant up-regulation of ISGs correlated also with nonresponse at week 12 and with final treatment outcome (Fig. S4 B–D).

Fig. 3.

RT-qPCR analysis of selected ISGs. (A) The expression level of selected ISGs in pretreatment biopsies is lower in RR than in non-RR patients. (B) Expression levels of USP18 and IFI27 in pretreatment biopsies are significantly higher in patients infected with genotype 1 (GT1) compared with GT 3. (C) Expression levels of USP18 and IFI27 are higher in non-RR patients also after stratification of the patients in a GT1/4 (“difficult-to-treat”) and a GT2/3 group. In A–C, the y axis shows expression relative to that of GAPDH. The P values were obtained with the Mann–Whitney test. N = number of patients in each group.

Pretreatment ISG Expression Levels Correlate with HCV Genotype.

We also analyzed the expression of ISGs with regard to the HCV genotype (GT) (Fig. 3B). Interestingly, the investigated ISGs showed significantly higher expression in patients infected with the “difficult-to-treat” GTs 1 and 4 than with GTs 2 and 3, which can be successfully treated in >80% of patients. Importantly, the expression levels of ISGs were higher in non-RR than RR patients independently from the HCV GT (Fig. 3C). Therefore, the increased ISG expression level in non-RR patients (Fig. 3A) cannot simply be explained by the fact that GT 1 is overrepresented in the non-RR group. Rather, that patients with HCV GT 1 and 4 more frequently have an increased expression of ISGs in their liver provides a plausible explanation for the poor response of these patients to IFN therapy.

IFN-Induced Jak-STAT Signaling.

The injected pegIFNα2b binds to IFN receptors and activates the Jak-STAT pathway. A central event in this activation is the phosphorylation of STAT1 on tyrosine 701 (25). We analyzed extracts from all B-1 and B-2 biopsies by Western blot using a phospho-specific STAT1 antibody (Fig. 4A). A semiquantitative analysis of the phospho-STAT1 bands revealed a median induction of 3.6-fold in RR patients and 1.6-fold in non-RR patients (P = 0.03).

Fig. 4.

Analysis of Jak-STAT signaling in liver biopsies. (A) STAT1 phosphorylation in extracts of liver biopsies collected before (B-1) and after (B-2) pegIFN-α2b injection. Extracts were analyzed by Western blot analysis by using antibodies specific for PY (701)-STAT1. Signals were quantified by using Odyssey Imaging Software to calculate the integrated intensity (kilo counts × mm²). The values represent the fold increase of phosphorylation in B-2 samples. RR patient numbers are shown in blue, non-RR patients in red. Blots were stripped and reprobed for total STAT1 used as a loading control for each pair of samples. (B) Representative examples of B-1 and B-2 of RR and non-RR patients. No nuclear staining is evident in pretreatment biopsies of RR patients (Pat. 4). The light-blue color of the nuclei originates from the counterstaining with hematoxylin. Four hours after pegIFNα, most hepatocytes show strong nuclear staining. In non-RR patients (Pat. 12), weak nuclear staining is already present in pretreatment biopsies, and pegIFNα induces little change in hepatocytes. The visible increased nuclear staining is confined to Kupffer cells.

Phosphorylated STAT1 translocates into the nucleus and binds as a dimer to specific response elements of ISG promoters (25). Immunohistochemical analysis of phospho-STAT1 localization in paired biopsies of RR patients revealed a minimal nuclear staining in B-1 samples and a strong staining in most hepatocyte nuclei in B-2 samples (Fig. 4B and Fig. S5A). In contrast, all but one (number 11) non-RR patients showed a remarkably different staining pattern. In the pretreatment biopsies, a large proportion of hepatocytes already had an appreciable nuclear staining, which did not increase in B-2 samples. Interestingly, there was a visible increase of phospho-STAT1 staining in nuclei of Kupffer cells (liver macrophages) in B-2 samples of non-RR patients (Fig. 4B).

STAT1 DNA binding was assessed in extracts of B-1 and B-2 biopsies by EMSAs. All RRs showed a marked increase in STAT1 DNA binding in the B-2 samples. In contrast, most non-RR patients showed a minimal or no increase of the gel-shift signal upon pegIFNα application (Fig. S5B). Taken together, the data demonstrate substantial differences in the IFN-induced Jak-STAT signaling between RR and non-RR patients.

Discussion

To learn more about possible mechanisms underlying differential response of HCV-infected patients to IFN therapy, we investigated the IFN-induced signaling and ISG induction in paired liver biopsies collected from patients with CHC before and during therapy with pegIFNα. Comparison of IFN signaling in two liver samples obtained from the same patient and comparison with the ISG induction in matching PBMC samples originating from the same patient allowed us to obtain unequivocal evidence that patients who respond poorly to therapy show preactivation of their IFN system, and that the preactivation is confined to the liver and is not evident in PBMCs. Importantly, in patients with low initial ISG expression, representing future responders to therapy, expression of ISGs in response to pegIFNα did not exceed that seen in nonresponders, either before or after therapy. This could suggest that patients with the initial preactivation of the IFN system, future nonresponders, have some defects at steps downstream of ISG expression, making them refractory to both endogenous IFN and IFN therapy. These findings in human patients are in accordance with observations in chimpanzees chronically infected with HCV (19). A human study reported a blunted response to IFNα in PBMCs of patients with a relative lack of viral response to treatment (26). In our group of 16 patients, we also detected a difference in the number of genes significantly changed at P < 0.05 in PBMCs between RR and non-RR patients (Fig. 1C), but this difference was less pronounced compared to the liver biopsies.

IFNα-induced STAT1 phosphorylation was stronger in RR than in non-RR patients, but there was still a clear activation of STAT1 in four of the six non-RR patients (Fig. 4A). The immunohistochemical analysis revealed a more pronounced difference. In non-RR samples, pegIFNα strongly induced nuclear STAT1 translocation in Kupffer cells, contrary to RR samples, where nuclear STAT1 accumulation was induced predominantly in hepatocytes. Interestingly, most non-RR patients had nuclear phospho-STAT1 already present in pretreatment biopsies. This is consistent with the observation that ISG transcripts are up-regulated in pretreatment biopsies of later nonresponders. How this preactivation of the Jak-STAT pathway is connected to the refractoriness of the IFN system in non-RR patients requires further investigation.

Over the last few years, important insights into the interference of HCV with the innate immune system revealed the ability of HCV to inhibit both TLR3-TRIF-IRF3 and the RIG-I/MDA5-Cardif signaling pathways of IFNβ induction (27, 28). This capacity of HCV could help explain why the virus often establishes a chronic infection. However, our data and previously published results (20) demonstrate that the endogenous IFN system is constantly activated in many patients. Moreover, patients with a preactivated IFN system seem to respond poorly to IFN therapy. This finding is counterintuitive (one would expect that an active innate immune system would help eliminate the virus during IFNα therapy), but it is largely supported by other published data from chimpanzees and human patients (16, 17, 20). From the analyses of ISG expression in liver biopsies, it is apparent that in some patients, HCV induces (or at least does not block) the endogenous IFN system, whereas in others it successfully represses it, possibly by cleaving TRIF and/or Cardif. Paradoxically, this difference has no apparent impact on the ability of HCV to maintain a chronic infection.

In patients without a preactivated IFN system, pegIFNα2b induced a robust up-regulation of many ISGs in the liver within 4 h. Similar high ISG expression was already present in the pretreatment biopsies of patients who later did not show a RR at week 4. It is somewhat perplexing why the latter patients do not resolve the chronic HCV infection spontaneously despite the strong activation of the IFN system. One possibility is that ISG proteins that are up-regulated in both cases possess different posttranscriptional modifications. In an alternative scenario, nonresponse to both endogenous and exogenous IFNα may be caused by the lack of induction of a few critical ISGs that are specifically required for the elimination of HCV. We cannot exclude this possibility, but an array analysis performed on paired liver samples did not reveal ISGs that were specifically up-regulated in RRs. Furthermore, this model cannot explain why preactivation of the endogenous IFN system is so closely linked to later nonresponse to treatment.

Alternatively, the kinetics of induction of the IFN response could be decisive. In patients without a preactivated IFN system, the injection of exogenous IFNα during treatment should induce an antiviral state very rapidly in most liver cells, and HCV would not have “enough” time to escape from the IFN-induced defense. However, the buildup of the antiviral state could be slow in the other group of patients, which would give HCV enough time to adapt to and evade the intracellular antiviral defense system, making it also resistant to the subsequent IFN therapy.

How could the induction of the endogenous IFN system compromise the success of IFNα therapy? Clearly, the activation of negative feedback loops that inhibit IFN signaling could play a role. Prominent candidates among the negative regulators are suppressors of cytokine signaling 1 (SOCS1) and SOCS3 (29), two IFN-induced proteins that bind to the IFN receptor and inhibit the activity of Jak1 and Tyk2, and the more recently described regulator Ubp43 (protein of the USP18 gene), an IFN-stimulated protein that binds to IFNα receptor 2 (IFNAR2) and blocks the access of Jak1 to it (30). However, we could not find a significant difference in the expression levels of these negative regulators in the pegIFNα2b-stimulated liver biopsies of RR compared with non-RR patients (data not shown). Moreover, a general up-regulation of negative regulators such as SOCSs and Ubp43 is not compatible with the observed strong constitutive expression of a large number of ISGs in the subset of patients that poorly respond to IFN therapy. If IFNα signaling were indeed inhibited by the induction of SOCSs and Ubp43 in the majority of liver cells, then one should not observe such a pronounced preactivation of ISGs in pretreatment livers.

Notably, although the preactivation of tested ISGs occurred more frequently in liver biopsies of patients infected with HCV GT 1 and 4 than with GT 2 or 3, it did occur in both groups. Preactivation of ISGs was found in non-RR patients with GT 2/3, whereas RR patients with GT 1/4 had no induction in pretreatment biopsies (Fig. 3C). Therefore, preactivation of the endogenous IFN system is strongly linked to the later response to treatment independent of the HCV GT. Our finding that the frequency and degree of preactivation of the endogenous IFN system depend on the HCV GT could provide an explanation for the observation that GT 2 and 3 infections can be cured in >80% of patients, compared with <50% of infections with GT 1 (4). Perhaps HCV GTs 2 and 3 are more successful in preventing the activation of innate immunity in the liver. The success of the virus in preventing the induction of the endogenous IFN system would, however, come at the cost of being more susceptible to IFNα therapies. Of note, a single chimpanzee infected with GT 3 has been shown to have lower ISG expression levels than animals infected with GT 1 (17).

We have shown that HCV inhibits the IFNα-induced signaling via the Jak-STAT pathway by up-regulating a protein phosphatase PP2A (12, 14, 31, 32). Inhibition of IFNα signaling by HCV by this or other mechanisms could explain why the strong preactivation of the endogenous IFN system does not lead to a spontaneous elimination of HCV. If one assumes that not all hepatocytes are infected by HCV, but rather a minority, then the induction of ISGs observed in pretreatment biopsies of non-RR patients could occur predominantly in noninfected hepatocytes. In infected cells, IFN would be ineffective because of the inhibition of the Jak-STAT signaling pathway. The IFN responsible for the preactivation of the system would be secreted by hepatocytes infected with a virus that is not successful in preventing IFN production. Because of the HCV-induced inhibition of the Jak-STAT pathway, the secreted IFNβ would not induce an antiviral state in the infected hepatocytes but rather in noninfected neighbor cells. To gain further insights into the pathobiology of CHC, future studies should focus on analysis at the single-cell level. Unfortunately, the detection of HCV-infected hepatocytes in liver biopsies is still unsatisfactory, making such studies difficult.

Although the precise mechanism of the HCV escape from the immune defense system still remains to be elucidated, the impairment of hepatitis C therapy by preactivation of the endogenous IFN system is now well established. It would be interesting to investigate whether this preactivation is a reversible process. The injection of neutralizing anti-IFNα/β antibodies or other factors blocking the IFN response before treatment could return the endogenous IFN system to a “naive” state and potentially enhance the response to IFNα-based therapies.

Materials and Methods

Patient Samples and Treatment.

From January 2006 to April 2007, patients with CHC referred to the outpatient liver clinic of University Hospital Basel were asked for permission to use part of their diagnostic liver biopsy (B-1) for research purposes. Patients who then were treated with pegylated-IFNα2b (PegIntron) and ribavirin (Rebetol, both from Essex Chemie) were asked to participate in this study. Sixteen patients agreed to undergo a second liver biopsy (B-2) 4 h after the first injection of 1.5 μg/kg body weight pegIFNα2b (PegIntron). All were Caucasians. The first dose of ribavirin was given after this second biopsy to avoid further confounding factors. The protocol was approved by the Ethics Committee of the University Hospital Basel. Written informed consent was obtained from all patients. Blood for PBMC isolation was collected before treatment and 4 h after the first pegIFNα2b injection. Patients were treated with pegIFNα2b (1.5 μg/kg body weight) and ribavirin (weight-based dosing: <65 kg: 800 mg/d; 65–85 kg: 1 g/d; >85 kg: 1.2 g/d). HCV RNA was quantified before treatment initiation and at weeks 4 and 12 of the treatment (Table 1). Treatment duration was 24 weeks for patients with genotypes 2/3 and 48 weeks for genotypes 1/4. As non-CHC controls, four patients who underwent ultrasound-guided liver biopsies of focal lesions gave informed consent for a biopsy from normal liver tissue outside the focal lesion. Pretreatment liver biopsies from 96 additional patients (all but one were Caucasians) with CHC were used for RT-qPCR for selected ISGs (patient data shown in Table S4).

Measurement of IFNα in Serum.

Pretreatment IFNα levels and the concentration of pegIFNα2b 4 h after the first injection were measured in serum using the human IFNα ELISA kit from PBL Biomedical Laboratories according to the manufacturer's instructions. This kit has been shown to recognize both unpegylated and pegylated human IFNα (22).

RNA Isolation, Western Blots, EMSA, Immunohistochemistry, and Microarray and RT-qPCR Analyses.

The procedures are described in detail in SI Materials and Methods. All original array data are being deposited in the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) database.