IFN-λ4 potently blocks IFN-α signalling by ISG15 and USP18 in hepatitis C virus infection

Genetic polymorphisms in IFNL4 have been shown to predict responses to IFN-α-based therapy in hepatitis C virus (HCV)-infected patients. The IFNL4-ΔG genotype, which encodes functional IFN-λ4 protein, is associated with a poor treatment response. In the present study, we investigated the induction and biological effects of IFN-λ4 in HCV-infected hepatocytes and their association with responsiveness to IFN-α. We also studied the effects of direct-acting antiviral (DAA) treatment on IFN-λ4 expression and IFN-α responsiveness. HCV infection induced IFN-λ4 expression at mRNA and protein levels in primary human hepatocytes (PHHs). In hepatoma cells, IFNL4 gene transfection or recombinant IFN-λ4 protein treatment robustly increased the protein levels of ISG15 and USP18 in an IFNLR1-dependent manner and potently blocked IFN-α signalling. The ISG15/USP18-mediated IFN-α unresponsiveness was demonstrated by transfection of siRNAs targeting ISG15 and/or USP18. This potent IFN-λ4 effect was related to prolonged ISG expression after IFNL4 gene transfection. DAA treatment of HCV-infected PHHs reduced the expression of IFN-λs, including IFN-λ4, and restored IFN-α responsiveness. These results demonstrate that virus-induced IFN-λ4 potently blocks IFN-α signalling by inducing high protein levels of ISG15 and USP18. Moreover, the data clearly demonstrate that DAA therapy restores IFN-α responsiveness in HCV-infected cells.

is the IFNL gene polymorphism that is primarily responsible for the treatment response to peg-IFN-α-based therapy 15 . Chronic hepatitis C patients with the rs368234815-∆G allele (∆G/TT or ∆G/∆G), which encodes the functional IFN-λ4 protein, respond poorly to peg-IFN-α-based therapy 14,15 .
In vitro experiments using the recombinant IFN-λ4 protein showed that IFN-λ4 activates the Janus kinase (JAK)-signal transducer and activator of transcription (STAT) signalling pathway by binding to the IFN-λ receptor 16 and induces the expression of ISGs 17 . As expected, the hepatic levels of ISGs in HCV-infected livers are associated with functional IFN-λ4 expression 18 , and a functionally impaired variant of IFN-λ4 is associated with weaker induction of ISGs in HCV-infected livers 19 . These genotype-phenotype correlation studies demonstrate that functional IFN-λ4 protein is the driver of high hepatic ISG expression as well as the cause of poor treatment response. However, there have been no mechanistic studies that could explain why the rs368234815-∆G allele encoding the IFN-λ4 protein is associated with a poor response to peg-IFN-α-based treatment.
The present study investigated the expression of IFN-λ4 in HCV-infected hepatocytes and the relationship of the biological effects of IFN-λ4 with IFN-α responsiveness. We also studied the effects of DAA treatment on the expression of IFN-λ4 and IFN-α responsiveness in HCV-infected hepatocytes.

Results
HCV infection results in IFN-λ4 expression and IFN-α unresponsiveness. First, we examined the expression of type I and type III IFNs after HCVcc infection in primary human hepatocytes (PHHs). In line with the previous studies 7-10 , we confirmed the induction of IFN-λs, rather than IFN-β ( Supplementary  Fig. 1). Next, using PHHs from four different donors with IFNL4 ΔG/ΔG or TT/ΔG genotypes, we observed that both poly(I:C) transfection (Fig. 1A) and cell culture-derived HCV (HCVcc) infection (Fig. 1B) induced IFN-λ4 mRNA expression, although the mRNA level of IFN-λ4 was much lower than that of IFN-λ1 (Fig. 1A,B). We also examined time kinetics of IFN-λ4 gene expression after HCVcc infection ( Supplementary Fig. 2). We confirmed the expression of IFN-λ4 after HCVcc infection at the protein level in PHHs with IFNL4 ΔG allele (Fig. 1C,D) whereas PHHs with IFNL4 TT/TT genotype did not produce IFN-λ4 protein after HCVcc infection (Fig. 1D). Furthermore, we detected IFN-λ4 protein in culture supernatant of PHHs with IFNL4 ΔG/ΔG genotype after HCVcc infection (Fig. 1E). We reported previously that prolonged stimulation with IFN-λ3 blocks the response to exogenous IFN-α 10 . To investigate whether the endogenous IFN-λ proteins that are produced in response to HCV infection, including the IFN-λ4 protein, also render cells nonresponsive to exogenous IFN-α, we infected PHHs with HCVcc and examined the response to exogenous IFN-α. Both STAT1 phosphorylation (Fig. 1F) and OAS1 upregulation (Fig. 1G) in response to exogenous IFN-α were attenuated in HCV-infected PHHs.
DAA treatment restores IFN-α unresponsiveness caused by HCV infection. Next, we examined the effect of DAAs on IFN-λ4 expression in HCV-infected PHHs. Treatment with telaprevir and sofosbuvir reduced the intracellular HCV RNA titre, resulting in greatly attenuated IFN-λ1 and IFN-λ4 gene expression ( Fig. 2A). These data indicate that viral replication is critical for the induction of IFN-λ expression in HCV-infected PHHs. Moreover, treatment with telaprevir and sofosbuvir restored IFN-α responsiveness, as demonstrated by restored STAT1 phosphorylation (Fig. 2B) and OAS1 upregulation (Fig. 2C). Taken together, these data demonstrate that DAA treatment reduces the expression of IFN-λs and restores IFN-α responsiveness in HCV-infected cells.

IFN-λ4 induces ISG expression.
To examine the effect of IFN-λ4, we first transfected the IFN-λ4-expressing plasmid into Huh-7 hepatoma cells of which IFNL4 genotype is TT/TT. We also transfected the cells with IFN-β-or IFN-λ1-expressing plasmids for comparative purposes. As results, we detected IFN-λ4 protein in cell lysates after transfection with the IFN-λ4-expressing plasmid (Fig. 3A). Moreover, we detected IFN-λ4 protein in culture supernatant by immunoprecipitation and immunoblotting ( Fig. 3B) although IFN-λ4 protein is known to be poorly secreted when it is expressed in mammalian cells 14,20,21 . Transfection with the IFN-λ4-expressing plasmid induced STAT1 phosphorylation (Fig. 3C) and ISG expression (Fig. 3D) in both Huh-7 cells and in PHHs ( Supplementary Fig. 3). Our previous report showed that HCV-infected hepatocytes produce IFN-λs that increase the protein levels of STAT1, STAT2, and IRF9, and that increased levels of STAT1, STAT2, and IRF9 prolong the upregulation of some ISGs 10 . Transfection with the IFN-λ4-expressing plasmid also caused sustained upregulation of these ISGs, including Mx1, ISG15, OAS1, OAS2, OAS3, and OASL (Fig. 3E). In fact, the intracellular HCV RNA titre was decreased by transfection with the IFN-λ4-expressing plasmid (Fig. 3F). Furthermore, we used the IFNL4-TT plasmid, which cannot generate IFN-λ4 protein due to a premature stop codon 14 and found that it did not cause STAT1 phosphorylation (Fig. 3G) nor did it induce ISG expression (Fig. 3H). STAT1 phosphorylation and induction of ISG expression were also observed by treatment with recombinant IFN-λ4 protein ( Supplementary Fig. 4). Silencing Cells that overexpress IFN-λ4 or treated with IFN-λ4 show weaker responses to IFN-α treatment. We reported previously that IFN-λ3-pretreated cells show an attenuated response to exogenous IFN-α treatment, as evidenced by reduced STAT1 phosphorylation and ISG expression compared to non-pretreated cells 10 . Similarly, IFN-λ4-transfected cells also showed weaker STAT1 phosphorylation (Fig. 4A), weaker STAT2 phosphorylation ( Supplementary Fig. 6A) and lower ISG expression (Fig. 4B) in response to exogenous IFN-α treatment compared to mock-transfected control cells. However, this unresponsiveness was shown only in response to IFN-α treatment, not to IFN-β or -λ treatment (Fig. 4A). IFN-λ4-transfected cells were also unresponsive to peg-IFN-α-2b treatment ( Supplementary Fig. 6B). IFN-λ1 pre-treatment also resulted in an attenuated response to exogenous IFN-α, and the response to IFN-α was almost completely blocked by the combination of IFN-λ1 pre-treatment and IFN-λ4 gene transfection (Fig. 4C). This indicates the importance of additional IFN-λ4 protein in IFNL4-∆G/TT or IFNL4-∆G/∆G hosts. We also confirmed the unresponsiveness to exogenous IFN-α by IFN-λs using recombinant IFN-λ proteins including IFN-λ4 (Fig. 4D).

IFN-α unresponsiveness depends on ISG15 and USP18 in cells that overexpress IFN-λ4 or treated with IFN-λ4.
We recently observed that prolonged IFN-λ3 stimulation induces robust and sustained upregulation of ISG15, which stabilizes the USP18 protein and diminishes the response to exogenous IFN-α treatment 10 . Here we found that the USP18 and ISG15 protein levels were upregulated in HCV-infected PHHs (Fig. 5A) and in IFN-λ4-transfected cells (Fig. 5B), which was expected in light of the previous observations. We next performed experiments to confirm the roles of ISG15 and USP18 in IFN-λ4-induced IFN-α unresponsiveness. As expected from the previous study 10 , silencing ISG15 expression decreased the protein level HCVcc at 10 MOI. After 48 hours, the cells were harvested, and gene expression was analysed by real-time qPCR. (C) PHHs with IFNL4 ΔG/TT genotype were infected with JFH1 HCVcc at 10 MOI. After 72 hours, the cell lysate was harvested, and protein expression were analysed by immunoblotting. (D) PHHs from two different donors (one with IFNL4 TT/TT genotype and the other with IFNL4 ΔG/ΔG genotype) were infected with JFH1 HCVcc at 10 MOI. After 72 hours, the cell lysates were harvested, and protein expression were analysed by immunoblotting. (E) PHHs with IFNL4 ΔG/TT genotype were infected with JFH1 HCVcc at 10 MOI. After 72 hours, the culture supernatant was harvested, and IFN-λ4 protein expression was analysed by immunoblotting. The data are representative of two independent experiments. (F,G) PHHs with the IFNL4 ΔG/ΔG genotype were infected with JFH1 HCVcc at 5 MOI. After 72 hours, the cells were treated with 100 IU/ ml IFN-α for 30 minutes (F) or 6 hours (G). The cells were harvested, and gene and protein expression were analysed by immunoblotting (F) or real-time qPCR (G). Data are presented as means ± S.E.M. ***P ≤ 0.001 (Student's t-test). All the data are representative of two independent experiments. Full-length blots of PY-STAT1 (F) with multiple exposure times are included in the Supplementary Fig. 9A.
We also confirmed that ISG15 maintains USP18 protein levels independently of its ISGylating activity. ISG15 silencing decreased the amount of USP18 protein in recombinant IFN-λ4-treated cells, and the protein level of USP18 was restored not only by transfection of wild type (WT) ISG15 gene but also by transfection of conjugation-defective ISG15 AA mutant gene ( Supplementary Fig. 8A). As a result, WT ISG15-or AA mutant-transfected cells showed refractoriness against exogenous IFN-α treatment ( Supplementary Fig. 8A). Furthermore, we confirmed that USP18 protein interferes with IFN-α signaling irrespective of its enzymatic activity. USP18 silencing restored phosphorylation of STAT1 after IFN-α treatment, and forced expression of WT USP18 or enzymatically inactive C64S USP18 mutant blocked phosphorylation of STAT1 ( Supplementary  Fig. 8B).
Finally, we examined relative importance of IFN-λ4 among the members of the IFN-λ family in the induction of ISG15 during HCV infection using a mixture of anti-IFN-λ1 and anti-IFN-λ2/3 which neutralize the effect of IFN-λ1, -λ2, and -λ3. As a result, ISG15 expression was highly maintained even after addition of a mixture of neutralizing anti-IFN-λ1 and anti-IFN-λ2/3 in HCV-infected cells. This result indicates that ISG15 upregulation is derived and maintained by IFN-λ4 produced during HCV infection.

Discussion
In this study, we show that HCV infection induces IFN-λ4 expression and that IFN-λ4 potently attenuates the response to exogenous IFN-α treatment by the action of ISG15 and USP18. DAA treatment of HCV-infected cells reduced the expression of IFN-λs, including IFN-λ4, and restored IFN-α responsiveness.
We and other groups have reported robust production of IFN-λs, rather than IFN-β, at both the mRNA and protein levels in HCV-infected PHHs [7][8][9][10] . It was demonstrated previously that IFN-λ4 expression is induced by poly(I:C) transfection into PHHs 14, 21 , but it has not been known whether IFN-λ4 expression is induced by HCV infection in PHHs. We have demonstrated for the first time that IFN-λ4 expression is induced by HCV infection in PHHs.
A recent study reported that IFNL4-ΔG/TT is the primary IFNL4 gene polymorphism determining the treatment response to peg-IFN-α-based therapy 15 . In HCV-infected patients, the IFNL4-TT genotype, which does not produce IFN-λ4 protein, is associated with a favourable response to IFN-α-based therapy. The IFNL4-ΔG genotype, which encodes functional IFN-λ4 protein, is associated with a poor treatment response. However, the mechanisms underlying this paradoxical association remained unclear. Our data explain why the presence of functional IFN-λ4 protein in IFNL4-ΔG genotype-carrying hosts is associated with a poor response to IFN-α-based HCV treatment. Moreover, our data suggest that DAA therapy can restore IFN-α responsiveness in HCV-infected patients.
HCV-infected patients with IFNL4-∆G allele exhibit high baseline levels of ISGs in their livers 2 . Among the proteins enriched in HCV-infected liver, USP18 has been reported to confer unresponsiveness to IFN-α by interacting with IFNAR2 and inhibits downstream signals induced by IFN-α 22 . In fact, USP18 overexpression decreases the responsiveness to exogenous IFN-α 23 . Recent reports show that intracellular free ISG15 protein maintains the stability of the USP18 protein in human cells 24,25 , although ISG15 has direct antiviral properties against many viruses in mice 26 . ISG15 deficiency caused increased viral resistance in humans but not mice, since only human ISG15 interacted with USP18 and decreased the response to IFN-α 25 . More specifically, in human cells, free ISG15 stabilizes the USP18 protein by preventing its ubiquitination, leading to an attenuated response to IFN-α 24,25 . Importantly, the ISGylation activity of ISG15 and the deISGylation activity of USP18 do not have any role in this process. It was shown that ISGylation-defective ΔGG ISG15 also binds to USP18 and prevents its degradation 24 . Furthermore, previous studies reported that blocking of IFN-α signaling by USP18 does not depend on the enzymatic activity of USP18 27 . The IFN-α signal-blocking activity of USP18 is mediated by the binding of USP18 to the intracellular domain of IFNAR2, which prevents the binding of JAK1 to IFNAR2 27 .
Our group recently found that prolonged stimulation of IFN-λ3 induces the robust and sustained upregulation of ISG15 that stabilizes the USP18 protein and causes the unresponsiveness to exogenous IFN-α treatment 10 . We also reported that both ISG15 and USP18 protein levels are increased in HCV-infected livers 10 . In the current study, for the first time, we demonstrated that ISG15 and USP18 protein levels are increased in HCV-infected PHHs (Fig. 5A) and IFN-λ4-expressing (Fig. 5B) or -treated cells (Fig. 5F) and that ISG15 and USP18 proteins mediate IFN-α unresponsiveness in IFN-λ4-expressing or -treated cells (Fig. 5D-F). Recently, Fan et al. also demonstrated that IFN-λ4 inhibits the JAK-STAT signalling pathway by inducing USP18 expression 28 . However, our current study showed the expression of IFN-λ4 after HCV infection for the first time. In addition, we observed IFN-α unresponsiveness in a more natural system for HCV infection. We infected PHHs with HCVcc and observed IFN-α unresponsiveness which was caused by endogenously produced IFNs, including IFN-λ4. Moreover, we identified ISG15 as a critical regulator for increased USP18 protein levels in IFN-λ4-expressing or -treated cells and demonstrated for the first time that DAAs can restore IFN-α responsiveness by decreasing the production of endogenous IFN-λs, including IFN-λ4, in HCV-infected cells.
Taken together, our data clearly demonstrate that virus-induced IFN-λ4 potently blocked IFN-α signalling via upregulation of ISG15 and USP18 in HCV infection and that DAA therapy restored the IFN-α responsiveness The cells were harvested, and protein expression was analysed by immunoblotting. (C) Huh-7 cells were transfected with control, USP18-, or ISG15-targeting siRNAs. After 24 hours, the cells were transfected with control or IFN-λ4-expressing plasmids. After 48 hours, the cells were re-transfected with the respective siRNAs. The cells were harvested, and gene expression was analysed by immunoblotting 24 hours after the second siRNA transfection. (D,E) Huh-7 cells were transfected with control, USP18-, or ISG15-targeting siRNAs. After 24 hours, the cells were transfected with control or IFN-λ4-expressing plasmids. After 48 hours, the cells were re-transfected with the respective siRNAs. After 48 hours, the cells were treated with 100 IU/ml of IFN-α for 15 minutes (D) or for 6 hours (E). Then the cells were harvested, and protein and gene expression were analysed by immunoblotting (D) or real-time qPCR (E). All the data are representative of two independent experiments. (F) Huh-7 cells were treated with or without 100 ng/mL recombinant IFN-λ4. After 48 hours, the cells were transfected with control siRNAs, USP18-, or ISG15-targeting siRNAs. The USP18-, or ISG15-targeting siRNAs used in this experiment had different sequences from siRNAs used in the experiment for (D). After 48 hours, the cells were treated with 100 IU/ml of IFN-α for 15 minutes. Then the cells were harvested, and protein expression were analysed by immunoblotting. (G) PHHs with IFNL4 ΔG/ΔG genotype were infected with JFH1 HCVcc at 15 MOI. After 6 hours, the cells were treated with 10 μg/ml anti-IFN-λ1 and 10 μg/ml anti-IFN-λ2/3 or IgG for 72 hours. The cells were harvested, and ISG15 gene expression was analysed by real-time qPCR. Data are presented as means ± S.E.M. **P ≤ 0.01, ***P ≤ 0.001 (Student's t-test).