key: cord-0983677-607px154
authors: Hamming, Ole J; Terczyńska-Dyla, Ewa; Vieyres, Gabrielle; Dijkman, Ronald; Jørgensen, Sanne E; Akhtar, Hashaam; Siupka, Piotr; Pietschmann, Thomas; Thiel, Volker; Hartmann, Rune
title: Interferon lambda 4 signals via the IFNλ receptor to regulate antiviral activity against HCV and coronaviruses
date: 2013-10-29
journal: The EMBO Journal
DOI: 10.1038/emboj.2013.232
sha: aca9ae47fe5c97687239c1d3df2a905afdae2604
doc_id: 983677
cord_uid: 607px154

The IFNL4 gene is a recently discovered type III interferon, which in a significant fraction of the human population harbours a frameshift mutation abolishing the IFNλ4 ORF. The expression of IFNλ4 is correlated with both poor spontaneous clearance of hepatitis C virus (HCV) and poor response to treatment with type I interferon. Here, we show that the IFNL4 gene encodes an active type III interferon, named IFNλ4, which signals through the IFNλR1 and IL-10R2 receptor chains. Recombinant IFNλ4 is antiviral against both HCV and coronaviruses at levels comparable to IFNλ3. However, the secretion of IFNλ4 is impaired compared to that of IFNλ3, and this impairment is not due to a weak signal peptide, which was previously believed. We found that IFNλ4 gets N-linked glycosylated and that this glycosylation is required for secretion. Nevertheless, this glycosylation is not required for activity. Together, these findings result in the paradox that IFNλ4 is strongly antiviral but a disadvantage during HCV infection.

Type III interferon or interferon lambda (IFNl) is a recently discovered group of interferons (Dumoutier et al, 2003; Kotenko et al, 2003; Sheppard et al, 2003) . Although IFNls are clearly interferons (Ank et al, 2006; Doyle et al, 2006; Zhou et al, 2007) , they signal via a complex consisting of the IFNlR1 and IL-10R2 receptor chains and share both structural features and the IL-10R2 chain with the IL-10 family of cytokines . Type III interferons distinguish themselves in being highly tissue specific. The IFNlR1 receptor chain is expressed on cells of epithelial origin and a yet not clearly defined small subset of haematopoietic cells (Mennechet and Uze, 2006; Mordstein et al, 2010; Pott et al, 2011) . The liver is of particular interest to this report. In humans, hepatocytes express IFNlR1, and thus respond to IFNl Wang et al, 2013) . Humans possess four IFNl genes (IFNL1, -L2, -L3 and -L4) as well as a pseudogene (IFNL3P1) (Lasfar et al, 2006; Fox et al, 2009) . Whereas the IFNL1, -L2 and -L3 genes were described in 2003 (Kotenko et al, 2003; Sheppard et al, 2003) , the IFNL4 gene was described recently and the IFNL4 gene has been inactivated in large part of the human population by a frameshift mutation (Prokunina-Olsson et al, 2013) . Phase 2 of clinical trials using pegylated IFNl1 against hepatitis C virus (HCV) infection has recently been completed (Ramos, 2010) , and it has now entered the phase 3 trials. IFNls are interesting pharmaceuticals, as the rather specific expression pattern of the IFNlR1 receptor should reduce the adverse effects compared to the type I IFN treatment.

The responses to the current standard treatment for HCV infection, which consists of pegylated interferon-a2 combined with ribavirin (pegIFN-a2 RBV), depend both on the viral genotype and on the genetics of the patient. Rather unexpectedly, single-nucleotide polymorphisms (SNPs) located within and around the IFNl3 gene were discovered as powerful predictors of treatment outcome as well as the likelihood for spontaneous clearance of the virus Thomas et al, 2009) . Extensive studies of the genetic region around the IFNL3 gene revealed the existence of a novel gene, the IFNL4 gene, which harbours a dinucleotide variant (ss469415590, TT or DG), where the TT allele leads to a frameshift thus inactivating the gene, and the DG allele results in a functional IFNL4 gene (Prokunina-Olsson et al, 2013) . In humans, the TT allele is strongly positively associated with HCV clearance as well as with positive treatment outcome (Bibert et al, 2013; Prokunina-Olsson et al, 2013) . Thus, disruption of the IFNL4 gene is beneficial for humans in the context of HCV infection, though the reason for this remains unclear.

The transfection of cells with an expression plasmid encoding IFNl4 induced STAT1 and STAT2 phosphorylation, but the authors were unable to detect any significant secretion of the IFNl4 protein, which was ascribed to a very weak signal peptide (SP) in IFNl4 (Prokunina-Olsson et al, 2013) . In addition, the authors produced recombinant IFNl4 inactive protein using insect cells. However, this protein was purified from cell lysates and not from the media as it is normally done with secreted proteins, and it appears likely that the protein was not properly folded. The lack of IFNl4 secretion together with the clear observation of intracellular IFNl4 protein led to the suggestion that IFNl4 could signal via an intracellular receptor (Booth and George, 2013; Lupberger et al, 2013; Ray, 2013) . Furthermore, the sequence of IFNl4 is similar to other IFNls within the first and last helices, which bind IFNlR1, while the IL-10R2 binding region is poorly conserved. Thus, the authors questioned whether IFNl4 actually signals through IL-10R2.

We have expressed, purified and refolded IFNl4 from E. coli and show that this recombinant protein is active and signals via IFNlR1 and IL-10R2, as do the other members of the type III interferon family. Furthermore, we show that IFNl4 has antiviral activity in human hepatocytes against HCV and in primary human airway epithelia (HAE) cells against human coronavirus strain 229E (HCoV-229E) as well as the novel coronavirus MERS-CoV. We demonstrated that IFNl4 gets secreted from mammalian cells, but with a substantially lower efficiency than what is seen for IFNl3. Our data suggest that the poor secretion of IFNl4 is not just a consequence of the weak IFNl4 SP, but it might be connected with the glycosylation of IFNl4.

To investigate the properties of IFNl4, we cloned a codonoptimised cDNA encoding the mature form of human IFNl4 with an N-terminal 6 Â His tag followed by a tobacco etch virus (TEV) protease cleavage site into a pET-15b vector. This recombinant form of IFNl4 was expressed in E. coli and purified from inclusion bodies under denaturing conditions by metal-ion affinity chromatography. The protein was then refolded in vitro and purified to homogeneity by cation exchange chromatography ( Figure 1A ) followed by size-exclusion chromatography ( Figure 1B ) . IFNl4 was eluted from the size-exclusion chromatography column at B75 ml, consistent with the expected monomeric size of IFNl4. The purified protein has a size of 17 kDa ( Figure 1C ) corresponding to IFNl4 without the SP (residues 23-179 of IFNl4) (NCBI accession code AFQ38559).

The effect of the recombinant IFNl4 was tested in HL-116 cells. These cells were stably transfected with IFNlR1 and a luciferase reporter under the control of the IFI6 promoter (Uze and Monneron, 2007) . Recombinant IFNl4 is highly active and activates the IFI6 promoter in a concentrationdependent manner comparable to IFNl3 (Figure 2A ). Further, we verified the activity of IFNl4 in HepG2 cells, which express IFNlR1 naturally. HepG2 cells were treated with IFNa2, IFNl3 or IFNl4, and the induction of the wellknown interferon-stimulated genes (ISGs) MX1, IFIT1 and OASL was monitored by qPCR ( Figure 2B ). All three interferons clearly induced all three genes. In fact, we observed comparable induction by IFNl3 and IFNl4. Recombinant IFNl4 is thus a highly active interferon.

To determine the receptor complex utilised by IFNl4, we used HEK293 cells. These cells express low levels of IFNlR1 and normally respond very poorly to type III interferon (Meager et al, 2005) . They do, however, express IL-10R2. In our assay, we introduced a luciferase reporter under the control of the Mx1 promoter in order to measure the interferon activity, and at the same time, we introduced IFNlR1 by transfection and/or knocked down IL-10R2, using specific siRNA ( Figure 2C ). The expression of IFNlR1 by transfection renders them highly responsive to both IFNl4 and IFNl3. However, this signal is largely lost when IL-10R2 is knocked down using siRNA ( Figure 2C ). The IFNamediated signalling was not significantly affected by either overexpression of IFNlR1 or knock-down of IL-10R2. To confirm these results, we repeated the experiment now blocking IL-10R2 with a specific antibody which has previously been shown to block IL-10R2 signalling in relation to IFNl (Sheppard et al, 2003) . The IL-10R2 antibody did not result in any activation of the reporter gene on its own, but both IFNl3 and IFNl4 signalling were sensitive to the IL-10R2 antibody, whereas IFNa signalling was unaffected. These results conclusively demonstrate that IFNl4, like the other members of the type III interferon family, signals via a heterodimeric receptor complex consisting of IFNlR1 and IL-10R2.

Evaluating IFN k4 binding to IL-10R2 using structural modelling Since IFNl3 and IFNl4 interact with the same receptor complex, we made a sequence alignment in Clustal W ( Figure 3A) and generated a homology model of IFNl4 (homIFNl4) using the SWISS-MODEL Workspace with IFNl3 as a model ( Figure 3B ). The overall structure of homIFNl4 is similar to that of IFNl3 ( Figure 3C ), as is expected for a homology model. The following observations validate the accuracy of the model. Cys76 and Cys178, which are not present in IFNl3, are in close proximity in homIFNl4, and with minor local rearrangements, they could form a disulphide bridge ( Figure 3B ). Furthermore, two conserved disulphide links are expected to exist in IFNl4, connecting Cys27 to Cys122 and Cys62 to Cys152. In both cases, homIFNl4 is compatible with the formation of these disulphide links. Moreover, superimposing homIFNl4 onto the structure of IFNl1 bound to IFNlR1 clearly shows that the homIFNl4 structure is compatible with the IFNlR1 binding.

As noted by Prokunina-Olson et al, the residues in helices A and F, which bind IFNlR1, are well conserved between IFNl3 and IFNl4, whereas the D-helix, which is expected to bind IL-10R2, is quite different (Figures 3A and D ). Yet our data clearly show that both IFNl3 and IFNl4 use IFNlR1 and IL-10R2 for signalling. The model of IFNl4 suggests a conservation of the helical structure and the way this is presented to IL-10R2. The conserved residues in helix D are primarily hydrophobic residues, which dock helix D to the rest of the structure and thus, determine the steric conformation of this helix. This conservation is most likely crucial for the activation as both receptor chains need to be engaged simultaneously. It is, however, important to remember that IL-10R2 is a shared chain that is capable of binding several different cytokines (IL-10, IL-22 and IL-26 and the IFNls). The chain is thus, promiscuous, allowing itself to interact with different ligands (Logsdon et al, 2012) .

To evaluate whether the structure of homIFNl4 is compatible with binding to IFNlR1, we superimposed the structure of homIFNl4 onto the structure of IFNl1 in the IFNl1: IFNlR1 complex (PDB entry code: 3OG6). Figure 3E shows that the overall structure of homIFNl4 is very similar to IFNl1 in the receptor-bound conformation and there are thus no obvious reasons why IFNl4 would not bind IFNlR1. The glycosylation site N61 in IFNl4 is equivalent to W47 in IFNl1. W47 interacts weakly with IFNlR1, but is located at the periphery of the interaction site away from the membrane and is situated in a loop between the A-and B-helixes (Miknis et al, 2010) . We believe that this position offers sufficient flexibility to allow for simultaneous glycosylation of N61 and receptor binding.

As the ss469415590 SNP DG leading to the expression of IFNl4 is associated with poor spontaneous HCV clearance and a negative response to pegIFN-a/RBV treatment, we decided to test the effect of recombinant IFNl4 against HCV infection. Huh7-Lunet hCD81-Fluc cells were transfected with a HCV genome (JcR2a, encoding luciferase as a reporter), and the 4-h post-transfected cells were treated with IFNa, IFNl3 or IFNl4 for 72 h. All interferon treatments resulted in a concentration-dependent decline in HCV replication ( Figures 4A and B ). In the Huh7-lunet cells, IFNl4 is slightly weaker than IFNa but at the same level as IFNl3. The experiment was repeated in HepG2 cells, which were treated with the indicated interferons for 48 h. In HepG2 cells, the antiviral activity of all three interferons is at the same level. Thus, using two different liver cell lines, we do not see any measurable difference between IFNl3 and IFNl4.

The IFNlR1 chain is primarily expressed on cells of epithelial origin, and it is thus here that IFNl mostly exerts its effect. We therefore decided to investigate the effect of IFNl4 in an epithelial cell system. For this study, we used primary HAE cultures. This system is based on primary human bronchial epithelial cells grown in air-liquid interface to obtain fully differentiated pseudostratified HAE layers, and it reflects many characteristics of the conducting human airways, such as the presence of basal, secretory, columnar and ciliated cell populations and a physical barrier, that is, the mucus (Kindler et al, 2013) . The HAE represents the entry port of human respiratory virus infection and is especially well suited for investigating the role of IFNls. HAE cultures derived from three separate donors were treated with IFNa2, IFNl3 or IFNl4 prior to exposure to a human coronavirus 229E expressing luciferase upon replication (HCoV-229E-luc, 4000 plaque-forming units (PFUs)) (van den Worm et al, 2012). As can be seen in Figure 5A , treating the HAE culture with IFNa, IFNl3 or IFNl4 reduces replication of HCoV-229Eluc. IFNa is the strongest interferon, whereas IFNl3 and IFNl4 are equally strong. In addition, we observed a concentration-dependent effect of IFNl4.

We then performed an experiment testing the effect of IFNa2, IFNl3 and IFNl4 against the novel and highly pathogenic coronavirus MERS-CoV (4000 PFUs). Again, we observed a concentration-dependent effect of IFNl4. To further investigate this effect, we looked at the induction of Mx1, OASL and IFIT1 by qPCR in the HAE cells treated with IFNa, IFNl3 or IFNl4. All three interferons induced all three genes, and the induction by IFNl3 and IFNl4 is at the same level, whereas IFNa is slightly higher. Thus, there is a good agreement between the antiviral activity measured and the induction of ISGs.

Poor secretion of IFN k4 is not due to a weak SP It was hypothesised by Prokunina-Olson et al that poor secretion of IFNl4 was due to a non-functional SP. Thus, we made chimaeric proteins of IFNl3 with the SP of IFNl4 and vice versa. HEK293 cells were then transfected with these constructs, and the protein secretion was evaluated by western blots of both the media and the cells ( Figure 6A ). For both the MYC-and FLAG-tagged constructs, IFNl3 is present in the media regardless of whether it has its own or the IFNl4 SP. Contrary to this, IFNl4 is not detectable by western blot in the media regardless of the SP. Thus, the poor secretion of IFNl4 cannot solely be ascribed to the SP.

In the case of IFNl4, we observed two bands in the transfected cells at around 18-19 and 20-22 kDa, respectively ( Figure 6A , left panels). The bottom band corresponds to the expected size of IFNl4 with the MYC or FLAG tags. As IFNl4 is predicted to contain a single N-linked glycosylation site at Asn61 (marked with a square in Figure 3A and labelled in Figure 3E ), the upper band could be due to glycosylation. To test this, we treated the cell lysates from the IFNl4transfected cells with PNGase F that cleaves N-linked glycosylation between the asparagine and the innermost N-acetylchondrosamine of high mannose, hybrid and complex oligosaccharides. As can be seen on the right in Figure 6A , treatment with PNGase F resulted in a single band at 18-19 kDa, showing that IFNl4 gets glycosylated.

To test whether active IFNl was secreted from the transfected cells, we added the supernatant from the transfected cells in Figure 6A to HEK293 transfected with an interferoninducible luciferase reporter system, with and without the expression of IFNlR1. This resulted in a clear signal from both IFNl3 and IFNl4, which was dependent upon IFNlR1 ( Figure 6B ). In order to estimate how much IFNl4 is secreted, we titrated the supernatants from IFNl3-and IFNl4-transfected cells ( Figure 6C ), and here we observed a substantially lower activity in the supernatant from IFNl4-transfected cells as compared to IFNl3-transfected cells (5-to 6-fold difference in EC50 values). Thus, IFNl4 is secreted at substantially lower levels. Swapping the SPs made no difference for IFNl3, which was equally well produced with its own or the SP of IFNl4. In the case of IFNl4, adding the SP of IFNl3 lead to lower levels of secreted interferon activity.

Glycosylation of IFN k4 is required for secretion but does not influence activity As described above, IFNl4 contains a potential N-linked glycosylation site, and we observed a fraction of the intracellular protein which had a size suggesting post-translational modifications. Thus, we wanted to address the glycosylation state of the secreted IFNl4. As IFNl4 levels were too low to be detected using standard western blotting, we first refined the detection of IFNl4 in the media of transfected cells using acetone precipitation ( Figure 7A ). The western blot revealed IFNl4 of a size consistent with glycosylation, and this result was confirmed with PNGase F treatment. Next, we made a mutant of IFNl4 where the glycosylated asparagine residue N61 was mutated to aspartate (IFNl4 N61D). HEK293 cells were transfected with empty vector, IFNl4, IFNl3 or IFNl4 N61D, and the intracellular and extracellular fractions were analysed by western blotting ( Figure 7B ). IFNl4 N61D only gives one band on the western blot of the intracellular fractions corresponding to the unmodified IFNl4. Neither IFNl4 nor IFNl4 N61D is detectable in the extracellular fraction by standard western blotting. However, when we carry out acetone precipitation on the media before western blotting, we see a clear band for IFNl4, but not for IFNl4 N61D, showing that this mutation further impairs the secretion of IFNl4. This is also reflected in the IFNl activity (performed as in Figure 6C ), where the activity in the supernatant of cells transfected with IFNl4 N61D is greatly decreased compared to that from cells transfected with IFNl4 ( Figure 7C ).

As the E. coli-produced IFNl4 is fully active and contain no glycosylation, this cannot be a prerequisite for activity. However, the question arose whether the glycosylated IFNl4 is active or whether low levels of unglycosylated protein that is undetectable by western blotting even after acetone precipitation mediate the activity. To exclude that non-glycosylated IFNl4 could be the source of the interferon activity, we incubated media from IFNl4-transfected cells with Concanavalin A (Con A) beads. Con A is a lectin that binds terminal a-D mannose and a-D glucose found on high mannose and hybrid N-linked glycans. Media from cells transfected with IFNl4 or empty vector were incubated with Con A beads. In the IFNl4-transfected cells, there was interferon activity in the input before addition of the Con A beads and this activity was removed after incubation with the beads ( Figure 7E ). This shows that the glycosylated IFNl4 is the source of the measured interferon activity. We attempted to elute IFNl4 from the beads using standard elution buffer but without success, as seen by the lack of activity ( Figure 7D ) and protein ( Figure 7E ) in the eluate. Nevertheless, we were able to confirm that IFNl4 was bound to the Con A beads by boiling these beads in SDS page buffer and performing a western blotting ( Figure 7E ). Figure 6 Secretion of human interferon lambda 4. (A) Constructs of IFNl3 and IFNl4 containing either the IFNl3 or IFNl4 signal peptide were transfected into HEK293 cells using pcDNA3.1 as a control. Western blots were performed on the intracellular and extracellular fractions as shown, using antibodies against the MYC or FLAG tag. For the cell fraction, the samples of IFNl4 were also subjected to treatment with PNGase F to confirm the presence of N-linked glycosylation. (B, C) The extracellular supernatants from HEK293 cells transfected with the FLAG-tagged constructs from (A) were added to HEK293 cells transfected with the pEF2 vector containing IFNlR1 as well as the interferon inducible luciferase reporter system (see Figure 2C ), in order to measure IFNl activity present in the extracellular supernatants. In (B), the extracellular fraction was undiluted whereas in (C) a serial dilution was performed.

IFNk4 signals through the IFN kR1:IL-10R2 receptor complex We produced recombinant IFNl4 protein in E. coli and did not observe any substantial difference in the behaviour of IFNl4 compared to the other isoforms of IFNl during purification. First, we tested the activity of IFNl4 in a standard reporter gene assay, utilising a luciferase gene under the control of the IFI6 gene promoter (Uze and Monneron, 2007) . The resulting dose response curves were comparable to IFNl3 and IFNl4. Furthermore, we tested induction of individual ISGs by both IFNl3 and IFNl4, and again we observed comparable levels of induction by both isoforms. Thus, we conclude that IFNl3 and IFNl4 are equally strong in inducing ISGs.

Based upon the low sequence similarity between IFNl4 and other isoforms of IFNl in the region known to bind IL-10R2, Prokunina-Olsson et al (2013) understandably questioned whether IFNl4 uses this receptor chain for signalling. First, we confirmed the use of the IFNlR1 receptor chain by IFNl4, as IFNl4 signalling was restored in HEK293 cells upon transfection with IFNlR1. Next, we demonstrated the involvement of IL-10R2 both by siRNA knockdown and by blocking the IL-10R2 chain by a specific antibody that has been used to define the receptor usage of the other IFNls (Sheppard et al, 2003) . Thus, IFNl4 leads to the activation of an interferon response and mediates antiviral effects through the canonical IFNl receptor complex composed of IFNlR1 and IL-10R2. However, these results do not exclude the possibility that IFNl4 can signal through other types of cytokine receptors, but it would indicate that if such a signalling existed it would not involve the regulation of classical ISGs.

We measured the antiviral activity of IFNl4 against HCV, HCoV-229E and MERS-CoV and compared it to the antiviral activity of IFNl3 and IFNa2. To our surprise, the antiviral activity of IFNl3 and IFNl4, respectively, was indistinguishable in all viral infection models tested. For HCV, we tested two different hepatic cell lines, Huh7 and HepG2, and in neither case did we observe any difference between IFNl3 and IFNl4. Likewise, using primary HAE cells for the infection with either HCoV-229E or MERS-CoV, we did not observe any difference between IFNl3 and IFNl4. This is remarkable as the sequence identity between the two isoforms is only 29% (Prokunina-Olsson et al, 2013), and our preliminary bioinformatics studies reveal that the protein sequence of IFNl4 is well conserved among mammals (data not shown). Thus, there must have been an evolutionary pressure to keep IFNl4 as a functional protein throughout the mammalian evolution until the sudden introduction of a frameshift mutation in humans. Since the inactivation of the IFNL4 gene is strongly correlated with increased likelihood of spontaneous clearance of HCV as well as with a positive response to the treatment with type I IFN, it appears that the production of IFNl4 protein is actually a disadvantage during HCV infection. Furthermore, there appears to be a positive selection in humans for the frameshift mutation abolishing IFNl4 production (Prokunina-Olsson et al, 2013) . Whether this selection is solely driven by HCV is currently not known. IFNl4 production could even be beneficial in the context of other viral infections. It was thus recently reported that the SNPs that have been shown to be favourable for the treatment outcome as well as the likelihood for spontaneous clearance of HCV are associated with poor recovery from hepatitis B virus infection (Kim et al, 2013) . Figure 2C ). The cells were subsequently treated with the media from (B) and the luciferase activities were measured. (D) Measurement of IFNl activity was performed as in (C). Input: media before addition of Con A beads, Flow through: flow through from the Con A beads and Eluate: eluate from the Con A beads. (E) Western blot of the samples from (D) as well as the boiled Con A beads. The arrows show the position of IFNl4.

How a functional interferon suddenly becomes a liability during HCV infection is a paradox that we are currently unable to explain. As discussed above, the induction of ISGs occurs through the canonical IFNl receptor complex, but we cannot exclude that IFNl4 has activities outside the induction of ISGs, which could be mediated through an as yet unidentified receptor. However, our data suggests that IFNl4 is highly active against HCV despite the fact that it has been shown to be a predictor of poor response to HCV. The current data cannot exclude indirect genetic effects, and thus it is not firmly proven that the IFNl4 protein is the causal agent for the poor prognosis of HCV patients with a functional IFNL4 gene. Furthermore, no evidence for the presence of the IFNl4 protein in HCV patients exists to date. However, if one assumes that the IFNl4 protein is the causal agent, this would suggest a complicated relationship between IFNl4 and HCV in humans, where IFNl4 somehow impairs a full immune response towards HCV. We have produced fully functional IFNl4 protein which should be used for further studies of IFNl4 on hepatic and immune cells. Furthermore, it will be important to address whether HCV is driving the selection of the TT allele (non-functional IFNl4) in humans, and if the introduction of the TT allele changes susceptibility towards other viral infections.

The inability of the IFNl4 protein to be properly secreted by cells was previously reported, and the authors speculated that this might be due to a weak SP (Prokunina-Olsson et al, 2013) . We produced chimaeric cDNAs where we had swapped the SPs between IFNl3 and IFNl4. Here, we observed that the IFNl4 was retained within the cells regardless of which SP was used, and likewise the secretion of the mature IFNl3 protein was not significantly affected by the SP used. By both immunoprecipitation and acetone precipitation, we were able to show that IFNl4 get secreted, but with much lower efficiency than what seen for IFNl3, which is also reflected by the reduced activity of media from IFNl4-transfected cells compared to media from IFNl3-transfected cells.

We tested for the presence of intracellular IFNl4 by western blots of cell lysates and observed two isoforms of IFNl4. Digestion with PNGase F, which removes N-linked glycans, revealed that this was due to incomplete glycosylation of IFNl4. By using acetone precipitation to concentrate the protein in the media, we were able to show that all secreted IFNl4 protein appeared to contain the N-linked glycosylation. This is in agreement with the current dogma that proteins need to complete their glycosylation before being exported to the extracellular media. It is not clear to us how the cell senses the difference between proteins, which are glycosylated like IFNl4 and proteins that are not glycosylated like IFNl3. We produced a glycosylation-deficient mutant of IFNl4 (IFNl4 N61D), and observed that the secretion of this mutant was greatly impaired, confirming that the N-linked glycosylation is needed for proper secretion. These results also suggested that IFNl3 and IFNl4 use different pathways for secretion, and that removing the N-linked glycosylation site is not sufficient to make IFNl4 shift to the secretion pathway used by the non-glycosylated IFNl3.

The question whether the glycosylation impairs activity was also raised. As the E. coli-produced protein is fully active, it is obvious that the glycosylation is not required for activity, but could it interfere with receptor binding? Our structure modelling suggested that the sugars were attached outside the receptor-binding site, and the activity that we recovered from the supernatant of IFNl4-transfected cells, which appeared only to contain glycosylated IFNl4, suggested that the sugars did not interfere with activity. However, to confirm this result, we used Con A beads to deplete the media from glycosylated IFNl4. As this led to an almost complete loss of activity, we conclude that the IFNl4 secreted from HEK293 cells is both glycosylated and active.

The poor processing and secretion of the IFNl4 protein are currently what makes it stand out in comparison to the other IFNl proteins, and our data suggest that the block in secretion takes place after the translocation to the Golgi, as the SP appears to be efficiently cleaved off. The lack of secretion of IFNl4 led several news and views papers to suggest the presence of an intracellular receptor. Our data clearly demonstrate that IFNl4 does use the normal receptor situated at the cellular membrane, although we cannot formally exclude the presence of an intracellular receptor. However, we consider it likely that the activation of the interferon pathway, which was observed after transfection of HepG2 cells with IFNl4 expressing plasmids (Prokunina-Olsson et al, 2013) , is due to low levels of secreted IFNl4.

IFNl4 (NM_001276254, amino acids 23-179) preceded by a 6 Â His tag followed by a TEV protease cleavage site was codon optimised for E. coli and purchased from Invitrogen. This construct was cloned into the pET-15b vector using Fastdigest KpnI (Thermo Scientific, catalogue number FD0524) and Fastdigest XhoI (Thermo Scientific, catalogue number FD0694). BL21 (DE3) E. coli cells transformed with the plasmids were grown at 371C in Luria Bertani medium containing 100 mg/ml ampicillin and 100 ml antifoam A concentrate (Sigma-Aldrich, catalogue number A5633) under continuous shaking until an OD 600 of 0.8-1. Protein expression was induced by adding 1 mM isopropyl-b-D-thiogalactopyranoside and incubated for another 4 h at 371C. Refolding and purification were performed as previously described ).

The pEF2-IFNl3 and the pEF2-IFNlR1 vectors were kind gifts from Professor Sergei Kotenko (UMDNJ-New Jersey Medical School, Newark, USA). The human IFNL4 gene (NM_001276254), including the SPSP, was purchased from Invitrogen. The following constructs were generated using Accupol (Amplicon, catalogue number 210302) following the manufacturer's instructions: IFNl4_FLAG (Template: IFNl4, forward primer: gcttggtaccatgcggccgagtgtctgg, reverse primer: agttctagatcacttgtcatcgtcatccttgtaatccgatccgaggcaagg ccc), IFNl3_FLAG (Template: pEF2-IFNl3, forward primer: gcttggta ccatgaccggggactgc, reverse primer: agttctagatcacttgtcatcgtcatccttgta atcacttccgacacacaggtccccactggc), IFNl3SP_IFNl4_FLAG (Template: IFNl4, forward primer: gcttggtaccatgaccggggactgcatgccagtgctggtgct gatggccgcagtgctgaccgtgactggagcagccccccggcgctgcctgctctcgc, reverse primer: agttctagatcacttgtcatcgtcatccttgtaatccgatccgaggcaaggccc), and IFNl4SP_IFNl3_FLAG (Template: pEF2-IFNl3, forward primer: gcttggtaccatgcggccgagtgtctgggccgcagtggccgcggggctgtgggtcctgtgcacgg tgatcgcagaggttcctgtcgccaggctccgcgggg, reverse primer: agttctagatc acttgtcatcgtcatccttgtaatcacttccgacacacaggtccccactggc), as well as IFNl4_MYC (Template: IFNl4, forward primer: gcttggtaccatgcgg ccgagtgtctgg, reverse primer: agttctagatcacagatcctcctcactaatcagtttc tgctccgatccgaggcaaggccc), IFNl3_MYC (Template pEF2-IFNl3, forward primer: gcttggtaccatgaccggggactgc, reverse primer: agtt ctagatcacagatcctcctcactaatcagtttctgctcacttccgacacacaggtccccactggc), IFNl3SP_IFNl4_MYC (Template: IFNl4, forward primer: gcttggtacc atgaccggggactgcatgccagtgctggtgctgatggccgcagtgctgaccgtgactggagcag ccccccggcgctgcctgctctcgc, reverse primer: agttctagatcacagatcctcctca ctaatcagtttctgctccgatccgaggcaaggccc), and IFNl4SP_IFNl3_MYC (Template: pEF2-IFNl3, forward primer: gcttggtaccatgcggccgagtgtct gggccgcagtggccgcggggctgtgggtcctgtgcacggtgatcgcagaggttcctgtcgccag gctccgcgggg, reverse primer: agttctagatcacagatcctcctcactaatcagtttctg ctcacttccgacacacaggtccccactggc). The following PCR programme was used 1: 951C for 5 min 2: 30 cycles of 951C for 1 min, 591C for 1 min and 721C for 1 min and 45 s 3: 721C for 7 min. All the constructs were cloned into the pEF2 vector using Fastdigest Kpn I (Thermo Scientific, catalogue number FD0524) and Fastdigest XbaI (Thermo Scientific, catalogue number FD0684) following the manufacturer's instructions.

The IFNl4 mutant IFNl4 N61D was generated by site-directed mutagenesis using IFNl4_FLAG in the pEF2 vector as a template. The reaction was performed using PfuUltra II with the primers (gctg ggggcagcgcgactgctccttccgcccc and ggggcggaaggagcagtcgcgctgccccc agc) according to the manufacturer's instructions. The following PCR programme was used 1: 951C for 5 min 2: 30 cycles of 951C for 1 min, 591C for 1 min and 721C for 5 min and 45 s 3: 721C for 7 min.

Unless otherwise stated, all cells were grown in Dulbecco's modified Eagle's medium (DMEM), which was supplemented with 10% fetal bovine serum, 100 U/ml penicillin and 100 mg/ml streptomycin. Furthermore, the Huh-lunet cells were supplemented with 2 mM L-glutamine, and kept under blasticidin selection (5 mg/ml). The HL-116 cell line was supplemented with hypoxanthine, thymidine and aminopterin, and 400 mg/ml G418. Cells were maintained at 371C with 5% CO 2 . HAE cells were generated as previously described (Kindler et al, 2013) and were maintained for 2 months.

The activity of IFNl4 and IFNl3 was tested in HL-116 cells, an HT1080 derived cell line containing a luciferase reporter gene controlled by the interferon inducible IFI6 promoter. Furthermore, the HL-116 cells were stably transfected with IFNlR1 to render them responsive to IFNl (Uze and Monneron, 2007) . To measure the interferon activity, 1 Â10 4 HL-116 cells were plated in a 96-well plate and treated in triplicates for 3 h with 8 dilutions of IFNl3 or IFNl4 in a concentration range covering 0.0001-1000 ng/ml. The cells were lysed, and the luciferase activity was quantified using the DualGlo luciferase assay system (Promega).

HepG2 cells were seeded at a density of 2 Â10 5 cells per well in 12well plates and incubated for 24 h. Then, fresh media was added with the indicated interferons. The cells were incubated for 4 h and then lysed, and RNA was purified using an extraction kit (Omega) according to the manufacturer's instructions. cDNA synthesis and analysis by real-time quantitative PCR were performed as previously described ). This reference also lists primer sequences. The crossing points of the amplification curves were determined using the second derivate method on the Roche LightCycler software 3.5 (Roche). The data obtained from the Light Cycler were normalised using the mathematical model described by Pfaffl (2001) . The experiments were performed in quadruplicates. For the untreated control, the mean of the quadruplicates was used to calculate fold induction for the other samples.

For transfection experiments, 1 Â10 5 HEK293 cells per well were seeded in a 24-well plate in DMEM supplemented with 10% FBS and left to rest for 24 h. After 24 h, the cells were transfected with siRNA against IL-10R2 or control siRNA (ON-TARGETplus Pool, Thermo Scientific) using Lipofectamine 2000 (Invitrogen). Eighteen hours post transfection, media was changed to fresh media supplemented with 10% FBS; and 24 h post transfection, cells were transfected with the pEF2 plasmid encoding IFNlR1, Firefly luciferase under the control of the Mx1 promoter (Jorns et al, 2006) , Renilla luciferase under the control of the b-actin promoter and siRNA against IL-10R2 or control siRNA. After 6 h, the media was changed to fresh media supplemented with 10% FBS, and the cells were left to rest for the next 12 h. Eighteen hours post transfection cells were induced with 10 ng/ml of IFNl3, IFNl4 or 1000 U/ml of IFNa2 (Chemicon) for 24 h. After 24 h, the cells were washed with PBS and lysed with Passive Lysis Buffer (Promega). Lysates were spun down at 10 000 r.c.f. for 2 min at 41C, and the cleared lysates were used for the measurement of luciferase activities (Dual-Luciferase Reporter Assay System, Promega).

Neutralisation assay HEK293 cells were seeded in 48-well plates in a concentration of 4 Â 10 5 cells per ml in DMEM supplemented with 10% FBS. After 24 h, the cells were transfected using Lipofectamine 2000 (Invitrogen) with plasmids coding IFNlR1, Firefly Luciferase under the control of the Mx1 promoter and Renilla Luciferase under the control of the b-actin promoter. Twenty hours post transfection cells were incubated for 1 h in media containing IL-10R2 (R&D Systems) or control antibody in a concentration of 6 mg/ml, after which the cells were induced with 10 ng/ml of IFNl3 or IFNl4 or 1000 U/ml of IFNa2a. After 24 h of induction, the cells were lysed with Passive Lysis Buffer (Promega), and cleared lysates were used for the measurement of luciferase activities (Dual-Luciferase Reporter Assay System, Promega).

The proteins were run on 12% SDS-PAGE gels. Gel staining was done using Coomassie brilliant blue. Western blotting was performed using a PVDF STAR 0.45 mm transfer membrane (applichem) using SuperSignal West Dura extended duration Substrate (Thermo Scientific). The membrane was exposed to MG-SR plus medical film (Konica Minolta), which was developed on an AGFA CURIX 60 film processor. The antibodies used were Mouse MYC antibody (Mycl-9E10 mouse hybridoma), Mouse monoclonal anti-FLAG s M2 antibody (Sigma-Aldrich, catalogue number F3165), IL-10R2 antibody from goat (R&D Systems, catalogue number AF874) and rabbit polyclonal GAPDH antibody (Santa Cruz Biotechnology, catalogue number FL-335). Anti-mouse IgG, horseradish peroxidase-linked species-specific whole antibody from sheep (GE Healthcare, catalogue number NA931) and Polyclonal swine anti-rabbit immunoglobulins/HRP (Dako Cytomation, catalogue number P 0399).

The alignment of human IFNl3 (NCBI accession code: NP_742151.2) and human IFNl4 (NCBI accession code: AFQ38559.1) was performed in Clustal W2 using the default settings (Larkin et al, 2007) . The full-length proteins including the SPs were used. The model of IFNl4 was generated in the SWISS-MODEL workspace (Bordoli et al, 2009 ) using the sequence of IFNl4 without the SP and the structure of human IFNl3 (PDB entry code: HHC3) as a model. Structural superimposition was performed in pymol (DeLano, 2008) .

Human bronchial epithelial cells were isolated from patients (418 years old), who underwent bronchoscopy and/or surgical lung resection in their diagnostic pathway for any pulmonary disease and that gave informed consent. This was done in accordance with the local regulation of the Kanton St. Gallen, Switzerland, as part of the St. Gallen Lung Biopsy Biobank (SGLBB) of the Kantonal Hospital, St. Gallen, which received approval by the ethics committee of the Kanton St. Gallen (EKSG 11/044, EKSG 11/103) . HAE cultures were prepared as previously described (Dijkman et al, 2009) . HAE cultures were used 28 days post exposure of the apical surface to air for infection studies. IFN aA/D (I4401, Sigma Aldrich), IFNl3 or IFNl4 was added to the basolateral medium 4-16 h prior to infection, after which the basolateral medium was replaced, and 20 000 PFUs of HCoV-229E-ren were applied apically. At 24 h post infection Renilla luciferase activity was determined from cell lysates infected with HCoV-229E-ren.

The MERS-CoV infection was performed as previously described (Kindler et al, 2013) .

The Huh7-Lunet N hCD81-FLuc cell line was generated from the Huh7-Lunet N hCD81 parental cell line (Bitzegeio et al, 2010 ) by lentiviral gene transfer as previously described (Gentzsch et al, 2011) . It constitutively expresses the Firefly luciferase gene (FLuc), which is used in our assay as a marker for cell viability.

In all, 4 Â 10 6 Huh7-Lunet N hCD81-FLuc cells or 6 Â10 6 HepG2-CD81/mi122 cells (Narbus et al, 2011) were electroporated with 5 mg of in vitro-transcribed JcR-2a RNA as previously described . The JcR-2a construct corresponds to the full-length infectious HCV Jc1 chimaeric clone (Pietschmann et al, 2006) , expressing a Renilla luciferase reporter gene (Reiss et al, 2011) . Electroporated cells were resuspended into 20 ml complete medium and seeded in 96-well dishes (100 ml/well). Four hours post electroporation, the cell medium was replaced by serially diluted IFN a2b (IntronA s , Essex Pharma), IFNl3 or IFNl4. For each dilution, triplicate wells were used. Cells were lysed 48 (HepG2 derived) or 72 h (Huh7-lunet) post electroporation in passive lysis buffer (Promega), and Renilla luciferase activity was measured to evaluate HCV replication (Vieyres and Pietschmann, 2013) .

For transfection experiments, HEK293 cells were seeded in 24-well plates (1.5 Â10 5 cells/well) or 6-well plates (7 Â10 5 cells/well) in DMEM supplemented with 10% FBS and left to rest for 24 h. After 24 h, cells were transfected using Lipofectamine 2000 (Invitrogen) either with plasmids coding IFNls (6-well format) or co-transfected with plasmids coding IFNlR1, Firefly Luciferase under the control of the Mx1 promoter and Renilla Luciferase under the control of the b-actin promoter (24-well format). Six hours post transfection, cells transfected with IFNls were given fresh media (DMEM, 10% FBS and 100 U/ml Penicillin and 100 mg/ml Streptomycin). Twenty hours post transfection, media from cells transfected with IFNls was harvested, spun down at 500 r.c.f. for 8 min and added to cells cotransfected with IFNlR1 and Luciferases in different dilutions. After 24 h, the cells were washed with PBS and lysed. Lysates were centrifuged at 10 000 r.c.f. for 2 min at 41C, and cleared lysates were used for the measurement of Firefly activity (Dual-Luciferase Reporter Assay System, Promega).

In all, 8 Â10 6 HEK293 cells were seeded in a 15-cm dish and transfected with IFNl3-FLAG, IFNl4-FLAG or empty vector (pcDNA3.1). After 5-6 h, the media was changed to media without serum, and the cells were transfected using 40 mg DNA per dish using polyethylenimine (PEI). In all, 40 mg DNA was mixed with media without antibiotics and serum to a concentration of 1.5 ml.

In all, 120 ml of PEI was mixed with media without antibiotics and serum to a concentration of 1.5 ml. The DNA and PEI were mixed and left for 15-20 min at RT before addition to the cells. The cells were incubated for 18 h, after which the media was isolated by centrifugation at 7000 r.p.m. for 10 min.

In all, 8 Â10 6 HEK293 cells were grown in 15 cm dishes using 20 ml of media and transfected as described. The supernatants were incubated with 100 ml of ANTI-FLAG s M2 Affinity Gel (SIGMA-Aldrich, catalogue number A2220) for 3 h. The beads were spun down by centrifugation at 8000 r.p.m. for 1 min. The supernatant was removed, and the beads were washed two times in 0.5 ml PBS containing 2% Triton X-100. The beads were then incubated in the elution buffer (PBS containing 2% Triton X-100 and 500 mg of FLAG peptide (SIGMA-Aldrich, catalogue number F3290) for 30 min. The beads were precipitated by centrifugation at 8000 r.p.m. for 1 min, and the supernatant was isolated and analysed by western blotting.

Deglycosylation was performed using Glycerol Free PNGase F (New England Biolabs, catalogue number P0705S). For deglycosylation, 9 ml of the cell lysate was mixed with 1 ml of 10x Glycoprotein denaturing buffer and denatured by heating at 1001C for 10 min. Then 5 ml H 2 O, 2 ml of G7 reaction buffer, 2 ml 10% NP-40 and 1 ml PNGase F was added. This mixture was incubated at 371C for 15 h and analysed by western blotting.

The media was mixed in a 1:4 ratio of media to cold ( À 201C) acetone. The samples were vortexed and incubated at À 201C for 60 min. The protein was precipitated by centrifugation at 6000 r.p.m. for 45 min. The supernatant was decanted, and the protein pellet was resuspended in PBS.

In all, 2 ml glucose-free DMEM (Sigma) from IFNl4-FLAG and mock (pcDNA3.1) transfected cells was incubated for 30 min with 100 ml of Concanavalin A (Con A) beads. The beads and the media were added to a column, and the flow through was collected. The beads were washed with PBS before incubation with 1 ml of elution buffer (glucose-free DMEM supplemented with 500 mM glucose) for 30 min. The activity of the flow through and eluate was investigated in HEK293 cells co-transfected with plasmids coding IFNlR1, Firefly Luciferase under the control of the Mx1 promoter and Renilla Luciferase under the control of the b-actin promoter (24-well format) as previously described. The protein content was evaluated in the flow through, eluate and on the beads by western blotting. The beads were boiled in SDS loading buffer for 10 min before loading on the gel.

Lambda interferon (IFN-lambda), a type III IFN, is induced by viruses and IFNs and displays potent antiviral activity against select virus infections in vivo

IL28B expression depends on a novel TT/-G polymorphism which improves HCV clearance prediction

Adaptation of hepatitis C virus to mouse CD81 permits infection of mouse cells in the absence of human entry factors

Loss of function of the new interferon IFN-lambda4 may confer protection from hepatitis C

Protein structure homology modeling using SWISS-MODEL workspace

The PyMOL Molecular Graphics System

Recombinant IFNk4 is antiviral OJ Hamming et al

Human interferon-lambda3 is a potent member of the type III interferon family

Interferon-lambda (IFN-lambda) induces signal transduction and gene expression in human hepatocytes, but not in lymphocytes or monocytes

Human bocavirus can be cultured in differentiated human airway epithelial cells

Interleukin-29 uses a type 1 interferon-like program to promote antiviral responses in human hepatocytes

Cloning of a new type II cytokine receptor activating signal transducer and activator of transcription (STAT)1, STAT2 and STAT3

The role of genomic data in the discovery, annotation and evolutionary interpretation of the interferon-lambda family

Interferon-lambda is functionally an interferon but structurally related to the interleukin-10 family

Genetic variation in IL28B predicts hepatitis C treatment-induced viral clearance

Hepatitis C virus complete life cycle screen for identification of small molecules with pro-or antiviral activity

Mousespecific residues of claudin-1 limit hepatitis C virus genotype 2a infection in a human hepatocyte cell line

Rapid and simple detection of IFN-neutralizing antibodies in chronic hepatitis C non-responsive to IFN-alpha

Association between IL28B polymorphisms and spontaneous clearance of hepatitis B virus infection

Efficient replication of the novel human betacoronavirus EMC on primary human epithelium highlights its zoonotic potential

IFN-lambdas mediate antiviral protection through a distinct class II cytokine receptor complex

Clustal W and Clustal X version 2.0

Characterization of the mouse IFN-lambda ligand-receptor system: IFN-lambdas exhibit antitumor activity against B16 melanoma

Structural basis for receptor sharing and activation by interleukin-20 receptor-2 (IL-20R2) binding cytokines

Interferon-lambda polymorphisms and hepatitis C virus clearance revisited

Biological activity of interleukins-28 and -29: comparison with type I interferons

Differential regulation of the OASL and OAS1 genes in response to viral infections

Interferon-lambda-treated dendritic cells specifically induce proliferation of FOXP3-expressing suppressor T cells

Crystal structure of human interferon-lambda1 in complex with its high-affinity receptor interferon-lambdaR1

Lambda interferon renders epithelial cells of the respiratory and gastrointestinal tracts resistant to viral infections

HepG2 cells expressing microRNA miR-122 support the entire hepatitis C virus life cycle

A new mathematical model for relative quantification in real-time RT-PCR

Construction and characterization of infectious intragenotypic and intergenotypic hepatitis C virus chimeras

IFN-lambda determines the intestinal epithelial antiviral host defense

A variant upstream of IFNL3 (IL28B) creating a new interferon gene IFNL4 is associated with impaired clearance of hepatitis C virus

Preclinical and clinical development of pegylated interferon-lambda 1 in chronic hepatitis C

Hepatitis: New gene IFNL4 is associated with impaired clearance of HCV

Recruitment and activation of a lipid kinase by hepatitis C virus NS5A is essential for integrity of the membranous replication compartment

IL-28, IL-29 and their class II cytokine receptor IL-28R

Genetic variation in IL28B and spontaneous clearance of hepatitis C virus

Reverse genetics of SARS-related coronavirus using vaccinia virus-based recombination

Entry and replication of recombinant hepatitis C viruses in cell culture

Induction of interferon-lambda contributes to Toll-like receptor-3-activated hepatic stellate cell-mediated hepatitis C virus inhibition in hepatocytes

Type III interferon (IFN) induces a type I IFN-like response in a restricted subset of cells through signaling pathways involving both the Jak-STAT pathway and the mitogen-activated protein kinases

We wish to thank Dr Sergei Kotenko for the kind gift of the IFNlR1 and IFNl3 expression plasmids, Dr Georg Kochs for the Mx-Luc reporter plasmid and Dr Gilles Uzé for the gift of HL-116 cells expressing the IFNlR1. We are also in debt to Lisbeth Heilesen and Dr Hans Henrik Gad for critical reading of the manuscript. This work was funded by the Danish Cancer Society (grant: R20-A927; RH), and the Danish Council for Independent Research, Medical Research (grant 11-107588; RH); the Swiss National Science Foundation (project 31003A_132898; VT), the 3R Research Foundation Switzerland (project 128-11; VT and RD), and the Deutsche Forschungsgemeinschaft (Priority Programme (SPP) 1596; VT).Author contributions: OJH, ET-D, GV, RD, SEJ, PS and HA designed, performed and analysed the experiments; TP, VT and RH supervised research; OJH and RH conceived the project and prepared the manuscript; all authors commented on the manuscript.

The authors declare that they have no conflict of interest.