IFN-λ4 attenuates antiviral responses by enhancing negative regulation of interferon signaling

Abstract

Type-III interferons (IFNs) are important mediators of antiviral immunity. IFN-λ4 is a unique type-III IFN because it is produced only in individuals who carry a dG allele of a genetic variant rs368234815-dG/TT. Counterintuitively, those individuals who can produce IFN-λ4, an antiviral cytokine, are also less likely to clear HCV infection. Here, we searched for unique functional properties of IFN-λ4 that might explain its negative effect on HCV clearance. We used fresh primary human hepatocytes (PHH) treated with recombinant type-III IFNs or infected with Sendai virus (SeV) to model acute viral infection, and subsequently validated our findings in HepG2 cell line models. Endogenous IFN-λ4 protein was detectable only in SeV-infected PHH from individuals with the dG allele, where it was poorly secreted but highly functional even at concentrations below 50 pg/ml. IFN-λ4 acted faster than other type-III IFNs in inducing antiviral genes as well as negative regulators of IFN response, such as USP18 and SOCS1. Transient treatment of PHH with IFN-λ4 but not IFN-λ3 caused a strong and sustained induction of SOCS1, and refractoriness to further stimulation with IFN-λ3. Our results suggest unique functional properties of IFN-λ4 that can be important in viral clearance and other clinical conditions.

INTRODUCTION

IFN-λ4 is a novel type-III interferon (IFN) that was discovered as a factor interfering with clearance of hepatitis C virus (HCV) infection (1). Decreased ability to clear HCV is associated with the dG allele of a genetic variant (rs368234815-dG/TT); this allele creates an open reading frame for IFN-λ4 protein, while the TT allele introduces a frame-shift and eliminates this IFN (1). The dG allele is very common in individuals of African ancestry (up to 78% allele frequency), but less common in Europeans (~30%) and Asians (0–10%) (1). This strong variation between individuals in the ability to produce IFN-λ4 might be important for health disparity in immune response to viral infections, and other relevant conditions.

Before the discovery of rs368234815 (1), its association was captured by either of the two single nucleotide polymorphisms (SNPs) used in genome-wide association studies (GWAS) - rs12979860 (2, 3) or rs8099917 (4, 5). SNP rs12979860 is located within the first intron of IFNL4, with the dG and TT alleles of rs368234815 corresponding to the T and C alleles of rs12979860, respectively (1). SNP rs8099917 is located upstream of IFNL4, and captures the difference additionally contributed by a missense IFNL4 variant, rs117648444-A/G (Pro70Ser) that affects the activity of IFN-λ4 (1, 6). Thus, rs8099917-G allele tags IFNL4 haplotype (rs368234815-dG/rs117648444-G) that produces a more active protein (IFN-λ4-70Pro), while rs8099917-T allele captures two other IFNL4 haplotypes corresponding to lack of IFN-λ4 (rs368234815-TT/rs117648444-G), or a less active IFN-λ4-70Ser protein (rs368234815-dG/rs117648444-A) (1, 4–6).

The ability to produce IFN-λ4, with or without further modification by P70S, was linked with differential pre-treatment blood HCV load and hepatic ISGs expression (1, 6–10), variable clearance of HCV infection spontaneously or after treatment with IFNα-based therapies (1, 2, 9, 11) and slower kinetics of HCV clearance in patients treated with IFN-free direct acting antivirals (DAAs) (10). Growing literature demonstrates genetic associations between rs368234815 or its linked variants with other clinical phenotypes, such as relapse on DAA treatment of HCV (12), liver fibrosis (13, 14), hepatic metallothionein expression (15), post-partum immune activation (16, 17), risk of mucinous ovarian cancer (18), etc. Although other, invariantly expressed type-III IFNs might be also important for these phenotypes, only IFN-λ4 is directly and most dramatically affected by the associated variant, rs368234815, that either creates or eliminates IFN-λ4. Thus, IFN-λ4 might be the primary cause of further functional effects, including on other type-III IFNs, that have been attributed to these genetic associations.

Although genetic association studies provide strong support for the role of IFN-λ4 in diverse clinical phenotypes, functional studies on IFN-λ4 are still limited. Mouse models have been successfully used to study other type-III IFNs (19–21), but IFN-λ4 is missing in the mouse genome, making it difficult to perform comprehensive comparisons between type-III IFNs in the same experimental models.

All IFN-λs, including IFN-λ4, activate the JAK-STAT signaling pathway through a receptor complex consisting of IFNLR1 and IL10R2 (22, 23), leading to induction of IFN-stimulated genes (ISGs) and antiviral response (1, 22, 23). Although IFN-λ4 and its closest family member, IFN-λ3, share only 29% amino acid identity (1), they induce the same set of ISGs (1, 6, 24–26), making the specific contribution of IFN-λ4 to the antiviral response difficult to evaluate.

Here, we searched for functional properties of IFN-λ4 that differentiate it from other type-III IFNs. We used fresh primary human hepatocytes (PHH) that were treated with recombinant type-III IFNs or infected with Sendai virus (SeV) to model acute viral infection. We extended these findings by performing comparative studies between type-III IFNs using several HepG2 cell line models. Our results uncover important differences in the regulation of the antiviral response by IFN-λ4 and other type-III IFNs; these findings might be important for understanding the diverse clinical phenotypes that have been directly or indirectly linked with the genetic ability to produce IFN-λ4.

MATERIALS AND METHODS

Human hepatic cells and cell lines

Fresh primary human hepatocytes (PHH) from 14 donors (Supplementary Table) were purchased from BioreclamationIVT. Samples were received within 3 days after donors’ death; no samples were obtained from executed prisoners or other institutionalized persons. PHH were received attached in 24-well plates and maintained in InVitroGRO HI culture media with Torpedo antibiotic mix (BioreclamationIVT). DNA from PHH was extracted with DNeasy Blood and Tissue kit (Qiagen) and used for genotyping of IFNL4 rs368234815-dG/TT with a custom TaqMan assay as previously described (1). Paired pre- and post-treatment liver biopsies were obtained from 17 patients with chronic genotype-1 HCV treated with DAA – eight patients were treated for 24 weeks with sofosbuvir and ribavirin (clinical trial NCT01441180), and nine were treated for 12 weeks with sofosbuvir and ledipasvir, or for 6 weeks with sofosbuvir and ledipasvir in combination with an additional investigational DAA (clinical trial NCT01805882) (9, 10). Two of 17 patients experienced treatment relapse, but post-treatment biopsies were obtained prior to virologic relapse. Both studies were approved by the NIAID/NIH Institutional Review Board and all patients provided written informed consent (9, 10). The liver biopsies were initially screened with microarray analysis in a subset of samples to identify transcripts with largest changes in expression caused by DAA treatment (9, 10). Expression of select transcripts was validated with individual TaqMan assays and analyzed in groups of patients stratified by their ability to produce IFN-λ4 protein, based on the presence of rs368234815-dG allele (9, 10).

The human hepatoma HepG2 cell line was purchased from the American Type Culture Collection (ATCC). HepG2 and all derived cell lines were maintained in DMEM (Cellgro) supplemented with 10% fetal bovine serum (FBS, Invitrogen), penicillin and streptomycin. A stable HepG2 cell line carrying the ISRE-Luc reporter (HepG2-ISRE-Luc cells) (1) was maintained in media with 1 μg/ml puromycin. A stable HepG2 cell line expressing doxycycline-inducible GFP-tagged IFN-λ3 protein (HepG2-IFN-λ3-GFP) was developed as described for HepG2-IFN-λ4-GFP cells (25); these cell lines were maintained in media with 5 μg/ml blasticidin and 1 mg/ml neomycin. A stable HepG2 cell line expressing both IFN-λ4-GFP and the ISRE-Luc reporter was generated by transducing HepG2-IFN-λ4-GFP cells with the Luciferase Cignal Lenti ISRE reporter construct (Qiagen) and selecting for positive clones as previously described (1); these cells were maintained in media with 5 μg/ml blasticidin, 1 mg/ml neomycin and 1 μg/ml puromycin.

Expression of IFN-λ3-GFP and IFN-λ4-GFP was induced by 0.5 μg/ml doxycycline for indicated time points, and monitored by flow cytometry analysis on FACS Aria III (BD Biosciences). Intracellular IFN-λ3-GFP was detectable only after blocking protein secretion with 6 μM GolgiStop (BD Biosciences) for 4 hrs, while IFN-λ4-GFP was detectable even without GolgiStop. All cell lines were regularly tested for mycoplasma using the MycoAlert Mycoplasma Detection kit (Lonza).

Recombinant IFNs

Commercially available IFN-λ1, IFN-λ2 and IFN-λ4 (all from R&D Systems) were generated in mouse myeloma cells, human HEK293 cells and E. Coli, respectively. Custom IFN-λ3 was generated in a baculoviral system (1), and custom IFN-λ4 was generated in E.coli, as described below. The custom IFN-λ4 represents wild-type protein that lacks the natural non-synonymous polymorphisms in the IFNL4 gene (rs73555604-C17Y, rs142981501-R60P and rs117648444-P70S), that may affect the activity of IFN-λ4 (1, 6). Open reading frame for IFN-λ4 was cloned in the pET28a vector (Novagen) introducing N-terminal His-tag, and expressed in E.coli after induction with 0.5 mM IPTG for 4 hrs at 37 ° C. Bacterial culture was centrifuged and the cells were resuspended in 5 ml of lysis buffer (50 mM Tris, pH 7.9, 0.5 M NaCl, 2% Triton X100 and 1 mM PMSF), and sonicated 3 times (15 sec on/15 sec off) at 4°C. Solution was centrifuged at 15000 rpm for 15 min at 4°C and the pellet (inclusion bodies) was resuspended in 6 M guanidine with rotation at 4°C for 1 hr. The protein was purified on a nickel sepharose bead column (GE Lifesciences) with 40 ml of 6 M guanidine, and eluted with a buffer containing 8 M urea and 300 mM imidazole. Protein refolding was done for 12 hrs in an optimized buffer containing 440 mM arginine, 300 mM NaCl, 1 mM reduced glutathione, 0.2 mM oxidized glutathione, 0.05% PEG3350, 2 mM MgCl₂ and 50 mM MES-buffer (pH 6.0). Salts and buffers were removed by multiple rounds of dialysis and IFN-λ4 was concentrated to 0.2 mg/ml. Endotoxin level in custom IFN-λ4 preparation was measured by Limulus Amebocyte Lysate test (Pierce) and found to be < 0.2 EU per 1 ug, which is comparable to the level in the commercial IFN-λ4 (<0.1 EU per 1 ug, based on product data sheet). Identity of IFN-λ4 was verified by Western blotting with primary monoclonal α-IFN-λ4 antibodies – rabbit (1:1500, ab196984, Abcam) and mouse (1:1000, MABF227, Millipore), and secondary HRP-linked antibodies - α-rabbit (1:5000, # 7074, Cell Signaling) or α-mouse (1:5000, sc-2031, Santa Cruz Biotechnology). Protein purity was evaluated by quantitative densitometry based on Coomassie staining and was determined as > 90% for custom IFN-λ4, compared to > 85% for commercial IFN-λ4 (based on product data sheet). The IFN-λ4 activity was initially confirmed by its ability to induce STAT1 phosphorylation that was attenuated by 10 μg/mL of blocking α-IL-10R2 antibodies - mouse monoclonal (MAB874) or goat polyclonal (AF874, both from R&D Systems). Protein concentrations were estimated using BCA protein quantification kit (Pierce). Recombinant proteins from the same batch were used at indicated concentrations for all experiments in this project.

ISRE-Luc assays

HepG2-ISRE-Luc cells (1) were seeded in 96-well plates with 1×10⁵ cells/well and treated with IFNs. Stable HepG2 cells were induced with 0.5 μg/ml doxycycline and then lysed and assayed for ISRE-Luc at indicated time points with Glomax-Multi detection system (Promega). All ISRE-Luc experiments were independently conducted at least three times with six biological replicates per sample each time, unless otherwise specified.

Co-culture assays

For the first co-culture model, 5×10³ HepG2-IFN-λ3-GFP or HepG2-IFN-λ4-GFP cells were seeded in a 1:1 ratio with 5×10³ HepG2-ISRE-Luc cells in a 96-well plate for a total of 1×10⁴ cells/well. For the second model, 5×10³ HepG2-IFN-λ4-GFP-ISRE-Luc cells were seeded in a 1:1 ratio with 5×10³ HepG2 cells; HepG2 cells were included to achieve a total of 1×10⁴ cells/well while keeping the counts of the IFN-λ4-producing and ISRE-Luc reporter cells (5×10³ cells) comparable to the first model. Cells were incubated for 24 hrs and then induced with 0.5 μg/ml doxycycline, lysed and assayed for ISRE-Luc at indicated time points. For blocking experiments, 20 μg/ml of rabbit antibodies - monoclonal α-IFN-λ4 (ab196984, Abcam), or polyclonal IgG control (R&D systems) were added immediately after induction with doxycycline.

Western blotting

PHH or HepG2 cells were lysed with RIPA buffer (Sigma) supplemented with protease inhibitor cocktail (Promega) and PhosSTOP (Roche). Equal amounts of proteins were resolved on 4–12% Bis-Tris Bolt gels and transferred using the iBlot2 (ThermoFisher). Primary antibodies were rabbit α-SOCS1 (1:200 dilution, Abcam, #ab62584), and a panel of antibodies from Cell Signaling - rabbit α-USP18 (1:500 dilution, #4813), α-STAT1 (1:1000 dilution, #9172), α-phospho-STAT1 (1:1000, Tyr701, #7649), α-STAT2 (1:1000, #4594) and α-phospho-STAT2 (1:100, Tyr690, #88410); the secondary antibody was a goat α-rabbit HRP-linked (1:5000, #7074, Cell Signaling). The results were detected with HyGlo Quick Spray (Denville Scientific) and viewed on the ChemiDoc Touch Imager with ImageLab 5.2 software (BioRad). Due to the rapid SOCS1 turnover (27), 50 μM of proteasome inhibitor MG132 (Sigma) was added to PHH 2 hrs before harvesting.

For detection of endogenous IFN-λ4, ~3.5×10⁵ PHH per liver donor were infected with SeV for 6 or 24 hrs. At harvesting, culture media was collected separately, then cells were lysed with RIPA buffer and pulse-sonicated for 30 sec, with 10 sec burst-cooling cycles, at 4 ° C. Lysates were centrifuged at 10,000 × g for 5 min to separate lysates from insoluble debris. All fractions (media, lysates and insoluble debris) were immediately boiled in reducing sample buffer for 5 min at 100 ° C, resolved on a gel as described above and transferred using the XCell II Blot module (ThermoFisher). The blot was blocked with 0.5% milk for 1 hr and then incubated with either rabbit α-IFN-λ4 (1:500, ab196984, Abcam) or mouse α-IFN-λ4 (1:200, MABF227, EMD Millipore) in 0.5% milk at 4 ° C overnight. Secondary antibodies were HRP-linked goat α-rabbit (1:5000, #7074, Cell Signaling) or donkey α-mouse (1:5000, sc2314, Santa Cruz), and detection was done with SuperSignal West Femto Maximum Sensitivity Substrate (ThermoFisher). Glycosylation of IFN-λ4 was tested by treating PHH lysates with detectable levels of IFN-λ4 with peptide-N-glycosidase F (PNgaseF, NEB) for 1 hr at 37 ° C, followed by Western blotting as described above.

Enzyme-linked immunosorbent assay (ELISA)

IFN-λ4-GFP and IFN-λ3-GFP were detected by a GFP ELISA kit (Cell Biolabs) using purified GFP as a standard, with a lower limit of detection of 50 pg/ml.

Sendai virus infection

Stocks of Sendai virus (SeV) Cantell strain were purchased from Charles River Laboratories. PHH or HepG2 cells were infected in triplicates with SeV (7.5×10⁵ CEID⁵⁰/ml) for 1 hr, washed with PBS and collected at indicated time points.

RNA extraction and cDNA generation

Total RNA from PHH and HepG2 cells was extracted using RNeasy kit with on-column DNase I digestion (Qiagen). RNA quantity and quality were evaluated by NanoDrop 8000 (ThermoFisher). cDNA was generated starting from 200 ng of total RNA from PHH or 500 ng RNA from HepG2, using the RT² First Strand kit with an additional DNA removal step (Qiagen).

Gene expression analysis

Expression of ISGs was measured in duplicates with an RT² Profiler qRT-PCR Human Antiviral Response Array (Qiagen) that included 88 expression assays for antiviral genes, as well as endogenous controls (ACTB, B2M, GAPDH, HPRT1, RPLP0), and positive and negative controls. Expression of select genes was measured with individual TaqMan expression assays, in quadruplicates (Supplementary Table), with GAPDH and ACTB used as endogenous controls. SeV loads were measured with an expression assay for SeV defective-interfering (SeV-DI) RNA, in quadruplicates (Supplementary Table), with GAPDH and ACTB used as endogenous controls. Expression was measured on the QuantStudio 7 (ThermoFisher) using SYBR Green qPCR Master Mix (Qiagen) for the Antiviral Response Array and SeV load analysis, and TaqMan expression buffer (ThermoFisher) for individual assays. Total RNA input for TaqMan assays was 2.5 ng/reaction for PHH and 5 ng/reaction for HepG2; RNA input for Antiviral Response Arrays was 2.5 ng/reaction.

Expression was measured in Ct values (PCR cycle at detection threshold), which are distributed on log2 scale. Expression of target genes was normalized by geometric means of corresponding endogenous controls as ΔCt (target) = Ct (control) – Ct (target). Differences in expression between groups of samples were calculated using relative quantification method as ΔΔCt = ΔCt (group1) – Ct (group2). Fold differences can be calculated as fold = 2^ΔΔCt.

siRNA knock-down of SOCS1 expression

Fresh PHH cultured in a 24-well plate were transfected in triplicates with 50 nM SOCS1 siRNA (Silencer Select, ThermoFisher, #s16470) or scrambled negative control siRNA (Silencer Select, ThermoFisher, #4390843) using Lipofectamine RNAiMAX transfection reagent (ThermoFisher). After 48 hrs of transfection, media was replaced and PHH were treated for 24 hrs with either 20 ng/ml of IFN-λ3 or 50 ng/ml of IFN-λ4. To test for refractoriness, transfected PHH were then treated with 20 ng/ml of IFN-λ3 for an additional 8 hrs before harvesting. Based on analysis of mRNA expression in PHH transfected with SOCS1 and scrambled siRNA, and normalized by expression of endogenous control (GAPDH), siRNA knock-down resulted in a decrease of SOCS1 expression by 2.3-fold. The siRNA effect was also visualized by Western blotting. Due to fast turnover of SOCS1 (27), 50 nM of proteasome inhibitor MG132 was added to PHH 2 hrs before harvesting. Positive and negative Western blot controls (OriGene) were represented by lysates of HEK293 cells either transfected with a SOCS1 plasmid or untransfected, respectively.

Statistical analysis

Unless specified, data plotting and statistical analyses were performed with Prism 7 (GraphPad). The P-values are for the two-sided unpaired Student’s T-tests, with P<0.05 being considered significant. Means are presented with standard errors.

RESULTS

SeV infection induces expression of endogenous IFN-λ4 in human hepatic cells

Expression of endogenous IFNL4 mRNA in virally-infected samples has been reported by several studies (1, 8, 28, 29), however, expression of endogenous IFN-λ4 protein has been clearly demonstrated only by confocal microscopy in PHH treated with PolyI:C (1). To further explore this, we infected PHH from 11 liver donors with Sendai virus (SeV), as this virus has been shown to induce IFNL4 mRNA expression in hepatic cell lines (28). We detected strong induction of all type-III IFNs, including IFNL4 (Fig. 1A), with similar kinetic profiles (Supplementary Fig. 1A).

Fig. 1. IFN-λ4 is produced in PHH in response to SeV infection.

(A) Expression of IFNL4 and other type-III IFNs in PHH, 6 or 24 hrs post-SeV infection. Expression is presented in relation to IFNL4 genotypes - with dG allele (dG/dG, n=2 and dG/TT, n=5) vs. TT/TT, n=4; results are plotted as ΔΔCt values (log2 scale) after normalization to endogenous controls (GAPDH and ACTB) and to uninfected samples (0 hrs). Shown individual values and group means (black bars). (B) Western blotting of lysates of PHH (n=4) infected with SeV for 6 or 24 hrs, showing a band corresponding in size to glycosylated IFN-λ4. Recombinant IFN-λ4 (0.1 and 1 ng), produced in E.coli and therefore non-glycosylated, is used for comparison. Deglycosylation treatment with PNGaseF resulted in a band of ~19 kDa. M - protein size marker. Donor IDs correspond to Supplementary Table. (C) Expression of IFNL4 and SeV RNA in samples shown in (B). Gene expression is presented as ΔCt values (log2 scale) after normalization to endogenous controls (GAPDH and ACTB).

We then performed Western blotting for IFN-λ4 in SeV-infected PHH from 4 donors (2 with dG/dG, and 1 each with dG/TT and TT/TT genotypes). In samples with dG/dG genotype we detected a distinct band both in cell lysates and insoluble cell debris, but not in media (Fig. 1B, Supplementary Fig. 1B–C). In PHH from the donor with dG/TT genotype a faint band could be observed, but only after overexposing the original blot (Supplementary Fig. 1B). As expected, IFN-λ4 was not detected in PHH from the donor with TT/TT genotype (Fig. 1B). Expression of IFN-λ4 protein followed the pattern of IFNL4 mRNA expression (Fig. 1C). The efficiency of SeV infection was comparable in all samples, based on viral RNA loads, and could not explain the IFN-λ4 expression pattern (Fig. 1C).

As IFN-λ4 was shown to be N-glycosylated at amino acid N61 (26), we hypothesized that the larger than expected size of IFN-λ4 detected by Western blotting could correspond to the glycosylated form. To test this, we treated one of the PHH lysates where IFN-λ4 expression was detectable (dG/dG genotype), with peptide-N-glycosidase F (PNGaseF), and observed a shift from the initial band to a new ~19 kDa band, which matches the expected size of IFN-λ4 without any protein modifications (Fig. 1B). These results demonstrate that IFN-λ4 is expressed in PHH in response to SeV infection, is glycosylated, and by Western blotting can be detected in cell lysates and in insoluble cell debris but not in the media.

IFN-λ4 activity requires secretion but at very low levels

The amount of IFN-λ4 presumably secreted by PHH into media in response to SeV infection was below the detection limits of Western blotting. This could be because the activity of IFN-λ4 does not require secretion, or that even the low amounts of IFN-λ4 secreted into media are sufficient for this function. To distinguish between these possibilities, we developed stable doxycycline-inducible HepG2 cell lines expressing either GFP-tagged IFN-λ4 (HepG2-IFN-λ4-GFP) (25), or IFN-λ3 (HepG2-IFN-λ3-GFP). As reported previously (25), a significant proportion of IFN-λ4-GFP was retained intracellularly and was readily detectable, while most of IFN-λ3-GFP was secreted and could be detected intracellularly only after blocking protein secretion with GolgiStop for 4 hrs (Fig. 2A).

Fig. 2. IFN-λ4 activity requires secretion but at very low levels.

(A) Representative flow cytometry plots for expression of IFN-λ3-GFP and IFN-λ4-GFP induced with doxycycline (Dox) for 24 hrs in corresponding stable HepG2 cell lines; expression of IFN-λ4-GFP was readily detectable, while IFN-λ3-GFP was detectable only after treatment with GolgiStop for 4 hrs to block protein secretion. (B) Kinetics of ISRE-Luc activation in stable HepG2 cells expressing IFN-λ4-GFP or IFN-λ3-GFP co-cultured in a 1:1 ratio with HepG2-ISRE-Luc cells. (C) Concentration of GFP measured by ELISA (n=2) in culture media of the experiment described in (B); shown one of two independent experiments. (D) Western blotting for GFP in culture media of stable inducible HepG2 cells expressing IFN-λ4-GFP (not detectable) or IFN-λ3-GFP (detectable, marked by arrow) from experiment described in (B). M - protein size marker. (E) The effect of secreted IFN-λ4 is tested in stable HepG2 cells expressing IFN-λ4-GFP co-cultured in a 1:1 ratio with HepG2-ISRE-Luc cells. The effect of both secreted and intracellular IFN-λ4 is tested in stable HepG2 cells expressing IFN-λ4-GFP and ISRE-Luc reporter. Cells were induced with doxycycline (Dox) for 24 hrs to express IFN-λ4-GFP in the presence of 20 μg/ml of rabbit antibodies - blocking anti-IFN-λ4 or IgG control, shown one of two independent experiments.

To test the role of IFN-λ4 secretion in its activity, we co-cultured HepG2-IFN-λ4-GFP or HepG2-IFN-λ3-GFP cells, that would produce these IFNs upon induction, in a 1:1 ratio with HepG2-ISRE-Luc cells, that lack these IFNs but carry the reporter that would detect the signal only when these IFNs are secreted from the producing cells. Time-course analysis of ISRE-Luc activation in the co-culture system showed response as early as 6 hrs after inducing IFN-λ4-GFP expression, while the response to IFN-λ3-GFP was detectable only after 10 hrs (Fig. 2B). This observation was surprising because it contrasted the concentrations of these IFNs in media, measured by GFP tag (Fig. 2C, D). With the limit of detection of ~50 pg/ml, IFN-λ4-GFP was undetectable in media until 8 hrs (0.2 ng/ml), while IFN-λ3-GFP was measurable by 6 hrs after induction (4 ng/ml). After 14 hrs, the amount of secreted IFN-λ4-GFP was still very low (1 ng/ml, Fig. 2C), yet ISRE-Luc activation had peaked. These results suggest that IFN-λ4 is an extremely potent IFN despite its poor secretion and may be biologically active at concentrations below 50 pg/ml, which are not detectable with most assays, including Western blotting.

Next, we evaluated the possibility that IFN activity could be contributed by intracellular accumulation of IFN-λ4 due to its poor secretion. In the first model, HepG2-IFN-λ4-GFP cells were co-cultured in a 1:1 ratio with HepG2-ISRE-Luc cells - in this model IFN-λ4 must be secreted to activate ISRE-Luc in the reporter cells. For the second model, we developed a stable HepG2 cell line expressing both IFN-λ4-GFP and ISRE-Luc (HepG2-IFN-λ4-GFP-ISRE-Luc), thereby combining the production and detection of IFN-λ4 in the same cells. This model would detect the signal induced by IFN-λ4 secreted from the producing and the surrounding cells, as well as intracellularly retained IFN-λ4.

IFN-λ4 was induced in both models for 24 hrs, in the presence or absence of α-IFN-λ4 blocking antibody (25), which is expected to block signaling induced by secreted but not intracellular IFN-λ4, and ISRE-Luc activation was measured. We observed that α-IFN-λ4 blocking antibody was equally effective in blocking ISRE-Luc activation in both models (Fig. 2E), indicating that to induce IFN signaling, IFN-λ4 must be secreted and the intracellular retention of IFN-λ4 does not significantly contribute to its antiviral activity.

IFN-λ4 is associated with earlier antiviral response

Analysis of the acute antiviral response in relation to IFNL4 genotypes, which correspond to the ability to produce IFN-λ4, may provide additional insights into IFN-λ4 function. To explore this, we measured the antiviral response to SeV in PHH from 11 donors, 7 of whom carry the dG allele and can produce IFN-λ4 (2 with dG/dG and 5 with dG/TT genotypes) and 4 with the TT/TT genotype (do not produce IFN-λ4). We measured the expression of three important ISGs - MX1, ISG15 and OAS1, 6 and 24 hrs post-SeV infection. Both genotype groups expressed similar levels of type-I, II and III IFNs (Fig. 1A, Fig. 3A), and had similar levels of viral infection, monitored by measuring SeV RNA (Fig. 3B). PHH from donors with dG allele expressed significantly higher levels of MX1, ISG15 and OAS1 compared to donors with TT/TT genotype 6 hrs post-SeV infection (Fig. 3A). However, by 24 hrs, the expression levels of these ISGs were similar between the genotype groups. These results suggest that the presence of dG allele, that is the ability to produce IFN-λ4, is associated with a faster acute antiviral response.

Fig. 3. IFN-λ4 is associated with earlier antiviral response SeV infection.

(A) Expression of select ISGs and IFNs in PHH, 6 or 24 hrs post-SeV infection, presented in relation to IFNL4 genotypes, with dG allele (dG/dG, n=2 and dG/TT), n=5 vs. TT/TT, n=4. (B) Expression of SeV in corresponding samples from (A). For (A) gene expression is presented as ΔΔCt values after normalization to endogenous controls (GAPDH and ACTB) and uninfected samples (0 hrs), while for (B) expression is presented as ΔCt values after normalizing to endogenous controls only. Both ΔCt and ΔΔCt values are presented on log2 scale, shown individual values and group means (black bars). *P<0.05. LLOD, lower limit of detection.

Activity of IFN-λ4 is rapid but transient compared to other type-III IFNs

To further evaluate the implications of IFN-λ4 activity during viral infection, we generated biologically active recombinant IFN-λ4 that was expressed in E. coli, refolded and purified. Recombinant IFN-λ4 was characterized by Coomassie staining and Western blotting (Fig. 4A, B), and by detecting IL10R2 - dependent STAT1 phosphorylation in HepG2 cells (Fig. 4C, D).

Fig. 4. Activity of IFN-λ4 is rapid but transient compared to other type-III IFNs.

Detection of the custom recombinant IFN-λ4 (350 ng) with (A) Coomassie staining and (B) Western blotting using the monoclonal mouse and rabbit α-IFN-λ4 antibodies. All methods detected a single band at ~19 KDa, which corresponds to the estimated size of IFN-λ4. M indicates protein size marker. (C) Western blotting for pY701-STAT1 and STAT1 in HepG2 cells treated for 30 min with different concentrations of IFN-λ4. (D) Western blotting as in (C) but in the presence of 10 μg/ml of the blocking α-IL-10R2 mouse monoclonal (mAb) or goat polyclonal (pAb) antibodies that are expected to block IFN-λ4 signaling and decrease pY701-STAT1 expression. (E). ISRE-Luc activity in HepG2-ISRE-Luc cells treated with 11 concentrations of all four type-III IFNs in a 24-hour time course. Vertical dashed lines indicate the peak ISRE-Luc induction observed at 10 hrs. (F) The ratio of average activity at 4 hrs and (G) at 24 hrs to the peak at 10 hrs, shown for top 6 concentrations of each type-III IFN. (H). Similar stability of IFN-λ3 and IFN-λ4 demonstrated by preincubation of recombinant IFNs at 37 °C for indicated time points followed by treatment HepG2-ISRE-Luc cells for 8 hrs. ***P<0.001.

We evaluated the comparative kinetics of all recombinant type-III IFNs, including IFN-λ4, by testing the range of 11 concentrations (0.02 – 1000 ng/ml) for each IFN in HepG2-ISRE-Luc cells in a 24-hr time course. The induction of ISRE-Luc by IFN-λ4 was rapid, reaching about 50% of its peak activity (determined at 10 hrs) within the first 4 hrs of treatment, while other IFNs at 4 hrs were at about 20% of peak induction (Fig. 4E, F). By 24 hrs, the activity of IFN-λ4 had dropped to 20% of its peak (Fig. 4E, G), while activity of other IFNs was still above 60% of their peaks. These results indicate that IFN-λ4 induces a rapid but transient response, while other type-III IFNs demonstrated slower but more sustained IFN activity. Similar rapid but transient activity was observed for a commercially sourced IFN-λ4 (Supplementary Fig. 2). Pre-treatment incubation of IFN-λ4 and IFN-λ3 for 12 hrs at 37 °C resulted in similar declines in their activities (Fig. 4H), thus arguing against the selective loss of IFN-λ4 stability as the reason for its shorter activity time.

IFN-λ4 shows faster antiviral activity than IFN-λ3 in a SeV infection model

It is unclear whether differences in IFN kinetics reflect their antiviral properties. To evaluate this, we compared the antiviral activities of IFN-λ4 and IFN-λ3. We selected IFN-λ3 for comparison because IFN-λ3 is the most potent of IFN-λ1-3 – these proteins share over 95% of protein sequence similarity and antiviral profiles (30); with 29% of protein identity, IFN-λ3 is also most similar to IFN-λ4 (1). We pretreated HepG2 cells with different concentrations of IFN-λ3 and IFN-λ4 for 10 hrs (estimated peak antiviral activity, Fig. 4E), infected the cells with SeV, and measured SeV RNA after 12 hrs (Fig. 5A). Maximum inhibition of SeV infection was achieved by 50 ng/ml of IFN-λ3 and 200 ng/ml of IFN-λ4 (Fig. 5B), since further increases in protein concentrations did not affect SeV RNA levels. This maximum inhibition resulted in a 13-fold reduction in SeV RNA levels (Fig. 5B), giving an estimate that 50% inhibition corresponded to a 6.5-fold reduction in SeV RNA levels, which was observed at 20 ng/ml of IFN-λ3 and 50 ng/ml of IFN-λ4 (Fig. 5B). In HepG2-ISRE-Luc cells these protein concentrations also provided similar peak antiviral activity by 8 hrs, but IFN-λ4 showed significantly higher activity at 4 hrs and much lower activity by 24 hrs, compared to IFN-λ3 (Fig. 5C). The concentrations that provided similar peak IFN activity, were considered equipotent and used to evaluate the antiviral kinetics of IFN-λ3 and IFN-λ4 in subsequent experiments. These equipotent concentrations need to be experimentally estimated for specific conditions, assays and batches of protein preparations.

Fig. 5. IFN-λ4 shows faster activity than IFN-λ3 in a SeV infection model.

(A) SeV RNA quantified by qRT-PCR and presented as Ct values (log 2 scale) in HepG2 cells, 3, 6, 12 and 24 hrs post-SeV infection (B) Change in SeV RNA in HepG2 cells infected with SeV after pretreatment with different concentrations of IFN-λ3 or IFN-λ4 for 10 hrs. SeV RNA values are normalized to expression of endogenous controls (GAPDH and ACTB) and presented as % of SeV RNA levels in untreated cells. Shown one of two independent experiments, each with three biological replicates. (C) ISRE-Luc induction in HepG2-ISRE-Luc cells treated with IFN-λ3 or IFN-λ4 in a 24-hrs time course. (D) Schematic representation of SeV infection experiment that includes pretreatment with IFN-λ3 or IFN-λ4. (E-F) SeV RNA in PHH (E) or stable HepG2 cells (F) pretreated or induced to express IFN-λ4-GFP or IFN-λ3-GFP, and then infected with SeV. Results are presented as in (B). *P<0.05, **P<0.01, ***P<0.001.

To compare the effects of these IFNs on the kinetics of SeV infection, PHH were pre-treated with IFN-λ3 or IFN-λ4 for different periods of time and then infected with SeV; cells were harvested after 12 hrs, and analyzed for SeV RNA expression (Fig. 5D). Pretreatment for 12 hrs with either IFN inhibited SeV, but only IFN-λ4 caused viral inhibition at earlier time points (4 hrs, Fig. 5E). Stable HepG2 cells induced to express IFN-λ4-GFP also showed an earlier antiviral response compared to cells expressing IFN-λ3-GFP (Fig. 5F). Overall, IFN-λ4-GFP was more potent than IFN-λ3-GFP in inhibiting SeV infection at all time points, showing that despite its poor secretion (Fig. 2C), IFN-λ4 induced an earlier and more potent antiviral response compared to IFN-λ3.

Transcriptional analysis identifies ISGs differentially induced by IFN-λ3 and IFN-λ4

To determine which antiviral genes are differentially induced by IFN-λ3 and IFN-λ4 at early and late time points, we used antiviral qRT-PCR arrays to measure mRNA expression of a panel of 88 ISGs in HepG2 cells treated with IFN-λ4 and IFN-λ3 (Fig. 6A). Within the first 1–2 hrs of treatment, we observed a strong increase in expression of ISGs such as MX1, ISG15, OAS2, DDX58 (RIG-I), DHX58 and STAT1 in cells treated with IFN-λ4 compared to IFN-λ3 (Fig. 6A). However, at 8 hrs both IFNs induced these ISGs to a similar level (Fig. 6A); expression of most ISGs induced by IFN-λ4 declined by 24 hrs, while the effect of IFN-λ3 was more sustained and lasted beyond 24 hrs (Fig. 6A). These expression profiles for individual ISGs recapitulated our results for ISRE-Luc reporter assays, including the determination of equipotent concentrations for these IFNs at 8 hrs (Fig. 5C).

Fig. 6. Transcriptional profiling identifies ISGs differentially induced by IFN-λ3 and IFN-λ4.

(A) Heatmap for expression of select ISGs measured by an antiviral qRT-PCR array in HepG2 cells treated with IFN-λ4 or IFN-λ3 for indicated time points. Red and green colors indicate higher expression of ISG induced by IFN-λ4 treatment, or IFN-λ3 treatment, respectively. (B) Expression of select ISGs in PHH obtained from three liver donors and treated with IFN-λ3 or IFN-λ4. Donor IDs correspond to Supplementary Table. (C). Expression of select ISGs in stable HepG2 cells induced to express IFN-λ3-GFP or IFN-λ4-GFP. Expression is presented as ΔΔCt values (log 2 scale) that are normalized to endogenous controls (GAPDH and ACTB) and uninfected samples (0 hrs). Plots show individual biological replicates; *P<0.05, **P<0.01, ***P<0.001.

Similarly, treatment of PHH from 3 liver donors showed stronger effects of IFN-λ4 compared to IFN-λ3 on expression of MX1, ISG15, and OAS1 at 2 hrs, and similar effects at 8 hrs of treatment; by 24 hrs IFN-λ4 activity had declined significantly, while IFN-λ3 activity was still sustained or even increased (Fig. 6B). We then monitored the kinetics of ISG expression in stable HepG2 cell lines induced to express IFN-λ4-GFP or IFN-λ3-GFP. Consistent with our results in HepG2 and PHH treated with recombinant IFNs exogenously, we observed that endogenously expressed IFN-λ4-GFP also induced higher levels of OAS1, ISG15 and MX1 compared to IFN-λ3-GFP at all time points (Fig. 6C). However, by 24 hrs there was no decline in response induced by endogenously produced IFN-λ4 compared to cells treated with IFN-λ4 exogenously (Fig. 6C), highlighting possible differences in response to continuous (chronic) vs. transient exposure to IFN-λ4. Overall, we show that IFN-λ4 induces an earlier activation of key antiviral effectors compared to IFN-λ3.

IFN-λ4 is a potent inducer of negative regulators of IFN response

To understand the mechanisms for the faster but more transient activity of IFN-λ4, we explored the kinetics of STAT phosphorylation (pSTAT). Both in HepG2 and PHH, IFN-λ4 induced strong pSTAT1 and pSTAT2 phosphorylation within first 15 min to 1hr of treatment, but the effect started to diminish by 24 hrs, while IFN-λ3 caused a more gradual increase of pSTAT1 and pSTAT2 that was sustained beyond 24 hrs (Fig. 7A and Supplementary Fig. 3A for an additional PHH donor).

Fig. 7. IFN-λ4 is a potent inducer of negative regulators of IFN response.

(A). Western blot analysis for pY701-STAT1, STAT1, pY690-STAT2 and STAT2 in HepG2 cells and PHH treated with IFN-λ3 and IFN-λ4, shown for one of three independent experiments. (B). Expression of select transcripts in HepG2 cells and PHH treated with IFN-λ3 or IFN-λ4, and stable HepG2 cells induced to produce IFN-λ3-GFP or IFN-λ4-GFP. Gene expression is presented as ΔΔCt values (log 2 scale) that are normalized to endogenous controls (GAPDH and ACTB) and untreated samples. (C) Western blot analysis of USP18 and SOCS1 after treatment of PHH with IFN-λ3 or IFN-λ4 in a time course. Positive (+) control represents lysate of HEK293 cells transfected with SOCS1 plasmid, negative (–) control represents untransfected HEK293 cells. (D) Expression of SOCS1 in PHH treated with different concentrations of all four type-III IFNs for 8 hrs. Gene expression analyzed as described in (B). (E) Expression of select transcripts in PHH obtained from one liver donor and treated with IFN-λ3 or IFN-λ4 in different combinations. X and Y represent PHH pretreated with IFN-λ3 or IFN-λ4, respectively, before restimulating with IFN-λ3. Gene expression is analyzed as described in (B). Results represent one of three independent experiments, each in biological duplicates. (F) Western blot analysis of SOCS1 in PHH transfected with either scrambled siRNA (Scr siRNA) or SOCS1 siRNA for 48 hrs and then stimulated with IFN-λ4 for 24 hrs; 50 nM or proteasome inhibitor MG132 was added to PHH 2 hrs prior to harvesting. (G) Expression of SOCS1 and select ISGs in PHH transfected with scrambled or SOCS1 siRNA. 48 hrs after transfection, cells were treated with IFN-λ3 (20 ng/ml) or IFN-λ4 (50 ng/ml) for 24 hrs. After 24 hrs, all the samples were restimulated with IFN-λ3 (20 ng/ml) for 8 hrs, and analyzed for gene expression. Expression is presented as ΔΔCt values (log 2 scale) that are normalized to endogenous control (GAPDH) and uninfected samples (0 hrs). siRNA knock-down resulted in a decrease of SOCS1 expression by a ΔΔCt value of 1.2, which corresponds to a 2.3-fold decrease. Plots show individual biological replicates from one of two independent experiments; *P<0.05, **P<0.01, ***P<0.001.

Blocking of STAT phosphorylation and inhibition of the JAK/STAT pathway is modulated on several levels, including by negative regulators of IFN signaling, such as USP18 and SOCS proteins (31–35). Increased expression of USP18 and SOCS1 has also been associated with reduced HCV clearance (36–39). We observed that both in HepG2 and PHH, USP18 had an expression profile of a typical ISG (Fig. 6), with earlier but more transient induction by IFN-λ4 compared to IFN-λ3 (Fig. 7B). Expression profile of SOCS1 was more distinct – in HepG2, and even more so in PHH - it was induced earlier and lasted beyond 24 hrs following IFN-λ4 treatment compared to IFN-λ3 (Fig. 7B). In stable inducible HepG2 cells the expression profiles of both USP18 and SOCS1 (Fig. 7B) were comparable to that observed for other ISGs (Fig. 6C), however, there was no attenuation of response by 24 hrs, possibly due to continuous exposure to IFNs in these cells. In PHH treated with IFN-λ4, USP18 protein was induced already at 8 hrs, but at 24 hrs USP18 was similarly induced by IFN-λ3 and IFN-λ4 (Fig. 7C). Expression of SOCS1 protein was similar at 2 hrs, but a sustained induction at 8 and 24 hrs was observed only in PHH treated with IFN-λ4 (Fig. 7C).

Of all recombinant type-III IFNs tested at different concentrations for treating PHH for 8 hrs, IFN-λ4 induced the highest levels of SOCS1 (Fig. 7D); the effects were comparable between our custom and commercial IFN-λ4 proteins (Supplementary Figure 3B). We hypothesized that strong and sustained induction of SOCS1 in PHH treated with IFN-λ4 might induce refractory state to additional IFN-λ3 treatment. To test this, we pretreated PHH with either IFN-λ4 or IFN-λ3 for 24 hrs (the timing was determined based on SOCS1 expression profiles, Fig. 7C), and then additionally treated with IFN-λ3 for another 8 hrs. Pretreatment with IFN-λ4 significantly attenuated the subsequent response to IFN-λ3 (Fig. 7E), while pretreatment with IFN-λ3 had no effect. Since USP18 protein levels were similar at 24 hrs of treatment with either IFN-λ3 or IFN-λ4, we reasoned that increase of SOCS1 rather than USP18 expression may be responsible for the refractoriness of IFN-λ4-treated PHH to subsequent IFN-λ3 stimulation. To test this, we monitored refractoriness of IFN-λ4-treated PHH in the presence or absence of siRNA knock-down of SOCS1 (Fig. 7F). The 2.3-fold knock-down of SOCS1 was sufficient to cause a significant increase in ISG expression (Fig. 7G). This indicates that SOCS1 induction was contributing to the refractoriness of IFN-λ4-treated PHH to additional IFN-λ3 stimulation.

IFN-λ4 is associated with higher expression of USP18 and SOCS1 during viral infection

Our findings suggest that the antiviral response could be differentially negatively regulated based on the genetic ability to produce IFN-λ4, which is defined by the presence of the dG allele. We tested this hypothesis in the set of SeV-infected PHH used in our previous analysis (Fig. 3) that included 11 samples – 7 with dG allele (2 with dG/dG and 5 with dG/TT genotypes) vs. 4 with TT/TT genotype. Expression of SOCS1 was significantly higher in carriers of the dG allele both 6 and 24 hrs post-SeV infection; the increase in expression of USP18 was more transient – it was higher only 6 hrs after infection, while at 24 hrs expression decreased and was similar in both genotype groups (Fig. 8A).

Fig. 8. IFN-λ4 is associated with higher expression of USP18 and SOCS1 during viral infection.

(A) Expression of SOCS1 and USP18 in PHH, 6 or 24 hrs post-SeV infection, presented in relation to IFNL4 genotypes, with dG allele (dG/dG, n=2 and dG/TT, n=5) vs. TT/TT, n=4, and plotted as ΔΔCt values (log2 scale) that are normalized to endogenous controls (GAPDH and ACTB) and to uninfected samples (0 hrs). *P<0.05. (B) Pre- and post- treatment liver biopsies from 17 HCV patients were analyzed for expression of MX1, ISG15, OAS2 and USP18. The results are presented as ΔCt values (log 2 scale) normalized by expression of endogenous control (GAPDH) and presented according to IFNL4 genotype groups. P-values for pre- and post-treatment expression levels are based on paired T-tests within genotype groups, and unpaired T-tests between genotype groups. Shown individual values and group means (black bars).

We also analyzed the available data for expression of USP18 and several other ISGs (MX1, ISG15 and OAS2) in paired (pre- and post-treatment) liver biopsies from 17 HCV patients treated with DAA - either with sofosbuvir and ribavirin for 24 weeks (n=8), or with sofosbuvir combined with 1–2 additional DAA for 6–12 weeks (n=9) (9, 10). It is known that expression of many ISGs, including USP18, is increased in pre-treatment liver biopsies of carriers of the dG allele (9); this conclusion could be made based on our liver biopsies as well (Fig. 8B). However, by analyzing changes in gene expression in paired liver biopsies, we could draw additional conclusions. We observed that sustained high expression of USP18 and other ISGs in pre-treatment liver biopsies occurred mainly in carriers of the dG allele, in line with what we observed in stable HepG2 cells continuously exposed to IFN-λ4 (Fig 6C, Fig 7B). In our HepG2 cell line models IFN-λ4 expression is induced in-vitro, while in liver biopsies it is induced by HCV infection. DAA treatment resulted in significant decrease of expression of USP18 and other ISGs, but only in carriers of the dG allele (Fig. 8B). These results suggest that viral clearance, even after DAA treatment, depends on reduced expression of negative regulators of IFN signaling, such as USP18, in carriers of the dG allele. Individuals who are genetically unable to produce IFN-λ4 (TT/TT genotype) may be less compromised by expression of negative regulators, leading to faster viral clearance, in line with a previous report (10).

DISCUSSION

We demonstrate here that the kinetics of IFN-λ4 differs significantly from other type-III IFNs, and describe potential consequences of IFN-λ4 expression during viral infection that could affect viral clearance in individuals with different genotypes of rs368234815. Antiviral activity of IFN-λ4 was suggested to be comparable to the activities of other type-III IFNs (24, 26, 28). We now show that IFN-λ4 acts faster than other type-III IFNs in inducing acute antiviral response, but at the same time induces expression of negative regulators of IFN response as well. Strong and prolonged induction of SOCS1 was achieved even after transient exposure of PHH to IFN-λ4, while the prolonged induction of USP18 required continuous exposure to IFN-λ4. Hepatocytes from HCV-infected patients may present another example of this regulation – even low but continuous expression of IFN-λ4 induced by HCV in carriers of the dG allele, might sustain expression of ISGs as well as negative regulators of IFN response, making hepatocytes refractory to additional IFN stimulation and resistant to spontaneous or treatment-induced viral clearance.

IFN-λ4 is unique among IFNs because it is poorly secreted and undetectable in media of virally-infected cells (1, 25, 26, 28). We show IFN-λ4 detection with Western blotting in lysates from SeV-infected PHH, and specifically from donors with the dG allele, consistent with our previous observations showing IFN-λ4 in PHH treated with Poly I:C (1). Although we did not detect IFN-λ4 in media of SeV-infected PHH, we show that IFN-λ4 acts as a secreted protein at media concentrations below the limit of detection of our assays (<50 pg/ml), indicating that IFN-λ4 is an extremely potent IFN. Interestingly, the antiviral response in stable HepG2 cells expressing IFN-λ4 was significantly higher than in PHH treated with recombinant IFN-λ4. This could be due to biological differences between immortalized cancer cells and fresh primary hepatocytes, and the availability of biologically active IFN-λ4. In stable HepG2 cells, induction provides continuous production of endogenous IFN-λ4, while PHH are exposed to a single-dose of recombinant IFN-λ4. Furthermore, even the limited amount of IFN-λ4 secreted from mammalian cells is expected to be properly folded and biologically active, while only a fraction of recombinant IFN-λ4 produced in E. Coli is properly refolded (usually estimated to be less than 25% of total protein refolded from E. Coli inclusion bodies (40)).

IFN-λ4 has only 29% amino acid similarity with IFN-λ3 (1, 41), with the most conserved regions corresponding to the protein sequences that interact with IFNLR1, and the least conserved region corresponding to the putative binding site of IL10R2 (1, 41). Thus, higher potency and faster antiviral kinetics of IFN-λ4 might be attributed to its stronger receptor affinity compared to other type-III IFNs (42). Faster, but more transient antiviral kinetics has been demonstrated in PHH treated with IFNα when compared to type-III IFNs (43, 44). IFNα and IFN-λs activate the same JAK/STAT pathway, but IFNα induces stronger STAT1 phosphorylation compared to similar amounts of IFN-λs (45). In our experiments, IFN-λ4 also had faster and more transient effects compared to equipotent concentrations of IFN-λ3. This suggests that IFN-λ4 and IFNα may be utilizing similar mechanisms to induce negative regulators of IFN signaling, which may contribute to the failure of IFNα-based therapies in managing HCV infection in patients expressing IFN-λ4 (1, 46, 47).

Higher levels of ISGs in pre-treatment liver biopsies were reported in HCV patients who can produce IFN-λ4 (6, 8–10, 48), and we observed the same pattern for several ISGs, including USP18, in pre-treatment biopsies of 17 HCV patients. USP18 was proposed as a key inhibitor of anti-HCV activity of IFNα (37, 49, 50), with attenuation of USP18 expression correlating with restoration of intrahepatic IFNα signaling and success of IFN-free HCV therapy (6, 9). Our analysis of ISG expression in liver biopsies pre- and post-treatment with DAA, showed that HCV-associated increase in USP18 expression occurred only in individuals that can produce IFN-λ4. This suggests that increased negative regulation of IFN signaling may be contributing to higher risk of development of chronic HCV in individuals with the dG allele, while individuals not producing IFN-λ4 are less affected by negative regulation and have more robust viral clearance. In PHH treated with IFN-λ4 compared to IFN-λ3, USP18 protein expression was induced faster (already at 8 hrs), but consequences of this faster induction on the IFN response are unclear. Treatment of a hepatic cell line (Huh7) with IFN-λ4 led to a sustained induction of USP18, making cells refractory to IFNα stimulation, but kinetics of USP18 expression induced by IFN-λ4 vs. other IFNs was not explored (49).

Increased expression of SOCS1 has been shown to block antiviral responses induced by type-I and type III-IFNs (32, 33, 38, 39). In HCV infection, downregulation of SOCS1 expression increases the anti-HCV activity of IFNs (39). Robust type-III IFN signaling is important during the acute phase of HCV infection in PHH in-vitro (51, 52), and in chimpanzees (51, 53), implying that a defective type-III IFN signaling could contribute to development of chronic HCV infection. We show that IFN-λ4 induces a strong and sustained induction of SOCS1 expression, which may make PHH refractory to subsequent type-III IFN signaling, although other negative regulators of IFN responses cannot be ruled out. Increased expression of SOCS1 induced by IFN-λ4 may also have an effect on the adaptive immune response to viral infections, since SOCS1 plays an important role in regulating the activity of IL-6 and IFNγ (35), key cytokines in the adaptive immune response.

Type-III IFNs are considered first responders to pathogens that enter the body through epithelial surfaces (54, 55). If efficient, an early innate antiviral response modulated by type-III IFNs at the pathogen entry sites might limit the infection and prevent the necessity of other defense mechanisms (56) and inflammation (21). However, an earlier antiviral response can be detrimental if it is inefficient and impairs the activity of other IFNs (57), or inhibits the adaptive immune response (58). Rapid antiviral kinetics of IFN-λ4 might be beneficial for some infections that require immediate albeit short-lived immune response while being a liability in chronic infections such as HCV.

Type-III IFNs have been shown to control a range of clinically important infections such as flu (21, 59), respiratory syncytial virus (60), norovirus (19, 20), rotavirus (61), Zika virus (62), West Nile virus (63) and Dengue (64). Expression of type-III IFNs in cells of epithelial origin was also induced by clinically relevant bacterial pathogens such as S. aureus and L. monocytogenes as well as M. tuberculosis (65). It is unknown if IFN-λ4 can be induced endogenously by these viral and bacterial pathogens and the role it plays during these infections, and this question warrants investigation. Although our studies demonstrate differential kinetics and potency of IFN-λ4 compared to other type-III IFNs in PHH using SeV infection as a model, additional studies are necessary before extrapolating these findings to other cell types and diseases, including HCV.

ABBREVIATIONS

IFN

Interferon

HCV

Hepatitis C Virus

ISG

Interferon stimulated genes

PHH

Primary human hepatocytes

ISRE

Interferon-stimulated response element

SeV

Sendai virus

SOCS1

Suppressor of cytokine signaling 1

DAA

Direct-acting antiviral