IFN-κ suppresses the replication of influenza A viruses through the IFNAR-MAPK-Fos-CHD6 axis

See allHide authors and affiliations

Science Signaling  07 Apr 2020:
Vol. 13, Issue 626, eaaz3381
DOI: 10.1126/scisignal.aaz3381

IFN-κ protects against influenza A

Virus-infected cells produce type I interferons (IFNs), which protect neighboring cells from infection by stimulating innate defenses. He et al. found that IFN-κ was produced in the lungs of mice early after infection with various influenza A viruses (IAVs). In cultured human lung epithelial cells, IFN-κ inhibited IAV replication by stimulating the IFN receptors IFNAR1 and IFNAR2, and chromodomain helicase DNA binding protein 6 (CHD6) was a key effector of IFN-κ antiviral activity. IFN-κ–mediated induction of CHD6 expression required the kinase p38 MAPK and the transcription factor c-Fos. Pretreating mice with IFN-κ protected them from subsequent lethal influenza challenge, suggesting that IFN-κ therapy may be useful for preventing or treating IAV infection in humans.


Type I interferons (IFNs) are the first line of defense against viral infection. Using a mouse model of influenza A virus infection, we found that IFN-κ was one of the earliest responding type I IFNs after infection with H9N2, a low-pathogenic avian influenza A virus, whereas this early induction did not occur upon infection with the epidemic-causing H7N9 virus. IFN-κ efficiently suppressed the replication of various influenza viruses in cultured human lung cells, and chromodomain helicase DNA binding protein 6 (CHD6) was the major effector for the antiviral activity of IFN-κ, but not for that of IFN-α or IFN-β. The induction of CHD6 required both of the type I IFN receptor subunits IFNAR1 and IFNAR2, the mitogen-activated protein kinase (MAPK) p38, and the transcription factor c-Fos but was independent of signal transducer and activator of transcription 1 (STAT1) activity. In addition, we showed that pretreatment with IFN-κ protected mice from lethal influenza viral challenge. Together, our findings identify an IFN-κ–specific pathway that constrains influenza A virus and provide evidence that IFN-κ may have potential as a preventative and therapeutic agent against influenza A virus.


Influenza A viruses (IAVs) are a severe threat to human health. Antigenic drift and shift bring great challenges to the development of universal prophylactic vaccines and antiviral therapeutics (1). Compared to seasonal IAVs, the emerging potentially pandemic H5N1 and H7N9 IAVs caused more severe lung injury and much greater mortality in infected patients (2, 3). An outstanding feature of H5N1 and H7N9 human infections distinct from seasonal IAV infections is an increase in inflammatory cytokines and chemokines accompanying reduced production of type I interferons (IFNs) (4). Type I IFNs are primary antiviral cytokines produced by infected cells. The engagement of type I IFNs and their cell surface receptors, IFN-α receptors (IFNARs), activates Janus kinase (JAK)–signal transducer and activator of transcription (STAT) signaling, which promotes the transcription of a large array of IFN-simulated genes (ISGs) to exert antiviral activities (5).

Most of our current knowledge of type I IFN signaling is derived from extensive studies on IFN-α and IFN-β. In contrast, the antiviral activities of the other two members of type I IFN family, IFN-ω and IFN-κ (6), are less well understood. Although all type I IFNs can bind to IFNARs, their antiviral activities appear to be distinct. For example, IFN-ω exhibits increased anti-influenza activity in cultured cells compared to IFN-α2, albeit less than IFN-β1a (7). First identified in human keratinocytes and then in dendritic cells and monocytes (8), IFN-κ is primarily viewed as a keratinocyte-specific IFN dedicated to skin immune responses. However, IFN-κ can also induce antiviral responses in other human cell types (6). The actual antiviral activities of different type I IFNs during infection are part of the host-pathogen arms race between IFN production and virus replication and are influenced by the dynamics of IFN production. IFN-β is induced earlier than IFN-α in cultured cells in response to virus infection (9), and many virus-encoded proteins interfere with the production of IFN through various mechanisms (10). For example, the NS1 proteins of some IAVs are capable of inhibiting the 3′-end processing of cellular pre-mRNAs by binding to cleavage and polyadenylation specificity factor (CPSF30) and accordingly blocking the production of mature mRNAs, including those of IFN-α and IFN-β (11). Thus, the production of IFN is a battle between host and virus that profoundly affects the subsequent replication and spread of the virus and its pathogenicity.

In this study, we analyzed the expression of genes encoding different type I IFNs during infection of two avian-originating viruses, the epidemic-causing H7N9 virus and an H9N2 virus, in a mouse model and identified Ifnk, the gene that encodes IFN-κ, as the most differentially expressed type I IFN gene in the early phase of infection, being the only one increased after H9N2 infection but decreased after H7N9 infection. Consequently, we used cultured human cells to study the action of IFN-κ against IAV infection. On the basis of the identification of a mutant IFNK gene in human lung epithelial A549 cells and subsequent demonstration that wild-type IFN-κ—but not the mutant—failed to contain IAV in cultured human cells, we pinned down chromodomain helicase DNA binding protein 6 (CHD6) as the major effector molecule mediating the anti-influenza activity of IFN-κ. Compared to its induction by IFN-κ, CHD6 was less induced by IFN-α and IFN-β (IFN-α/β) and dispensable for IFN-α/β–mediated inhibition of IAV replication. We also identified the upstream signaling required by IFN-κ to stimulate CHD6 expression. Unlike IFN-α/β, which transduce antiviral signal preferentially through IFNAR1, IFN-κ required the engagement of both IFNAR1 and IFNAR2. IFN-κ was also distinct in that it induced CHD6 through a p38–c-Fos axis rather than the canonical JAK-STAT pathway. Last, we showed that pre-application of IFN-κ afforded protected mice against lethal IAV infection. In summary, we demonstrated the anti-influenza activity of IFN-κ and found that the underlying program was distinct from the anti-influenza activities of IFN-α/β. Our results suggest that IFN-κ may represent a new avenue toward improved IFN-based antiviral prevention and therapeutics.


IFN-κ inhibits various IAVs in different cells

To gain a better understanding of the mechanisms behind the pathogenicity of IAVs, we established a mouse infection model to compare the outcome of virus challenge between two subtypes of avian-originating IAVs: the A/Shanghai/4664T/2013 strain, an H7N9 virus causing hundreds of human infections in 2013, and the A/Chicken/Shanghai/F/98 strain, an H9N2 virus. H7N9- and H9N2-infected animals experienced essentially the same continuous weight loss during the first 2 days after infection, with an about 10 to 15% loss at day 2. However, whereas the H9N2-infected group regained normal weights afterward and fully recovered 1 week after infection, the H7N9-infected group continued to lose weight (Fig. 1A), and consequently, all died on day 7 after infection (Fig. 1B). Thus, the mouse model study substantiated our initial speculation that, for the two strains examined, the H7N9 virus was more virulent than the H9N2 virus.

Fig. 1 IFN-κ broadly inhibits IAV replication.

(A) Body weight change in mice challenged with H7N9 IAV, H9N2 IAV, or PBS expressed as means ± SEM. n = 15 mice for each group. (B) Survival data for mice challenged with H7N9 IAV, H9N2 IAV, or PBS pooled from three independent experiments, each with five mice per group. (C) Expression of Ifn genes in H7N9- and H9N2-infected mice from microarray-based transcriptome analysis of RNA pooled from three lungs for each group. (D) Protein sequence comparison between IFN-κ cloned from A549 cells (IFN-κDual) and the published human IFN-κ sequence, highlighting the two amino acid changes at positions 133 and 164. (E) Western blotting analysis of IAV NP and IFN-κ in A549 cells transfected with empty vector (control) or the indicated IFN-κ construct for 24 hours and then infected with H9N2, H7N9, PR8, or H1N1 for 48 hours. Nontransfected, noninfected A549 cells served as mock control. β-Actin is a loading control. Blots are representative of three independent experiments. Quantitation of the blots in (E) is shown in the Supplementary Materials (fig. S7A).

To identify key cellular events that might be involved in the pathogenicity of IAV, we performed transcriptome analysis on lung tissues from H7N9- and H9N2-infected mice with a special focus on transcripts for type I IFNs. The microarray data revealed that Ifnk was the most differentially expressed type I IFN gene at early stages of infection (6 hours after infection), being increased in H9N2 infection and suppressed in H7N9 infection (Fig. 1C). We further examined the transcriptional induction of IFNs in human A549 cells in response to infection with three different IAVs, including the two viruses described above as well as the A/PR/8/34 strain of H1N1 (abbreviated as PR8). H9N2 virus induced human IFNK most efficiently with fold increases comparable to those of IFNA and IFNB, particularly at the early phase of infection (6 hours after infection). In comparison, there was substantially less IFNK induced in PR8- and H7N9-infected cells despite a similar degree of induction of IFNA and IFNB (fig. S1). Thus, IFN-κ is an early type I IFN induced in response to IAV infection, and different IAVs vary in their ability to induce IFN-κ.

We next used A549 cells to examine the potential antiviral functions of IFN-κ. We first cloned the IFNK gene from A549 cells. Unexpectedly, we identified polymorphisms resulting in two missense amino acid substitutions, K133E and R164K, in IFNK from A549 cells as compared to the published human IFNK sequence (Fig. 1D). To test whether IFN-κ had an anti-influenza activity and whether the polymorphism identified in A549 cells affected such activity, we expressed IFN-κ in A549 cells. Plasmids encoding either wild-type IFN-κ or mutants carrying K133E and R164K mutations, individually or in combination (IFN-κDual), were transfected into A549 cells, which were subsequently infected with H9N2, H7N9, PR8, or the seasonal A/California/07/2009 strain of H1N1. The introduction of wild-type IFN-κ resulted in effective inhibition of the replication of all the tested IAVs as indicated by reduced abundance of the influenza virus nucleoprotein (NP). In contrast, all of the three IFN-κ mutants showed little, if any, anti-influenza activity (Fig. 1E and fig. S7A). This was further confirmed by a reduction in the virus titer present in the cultural medium at the same time NP expression was examined in PR8-infected cells (fig. S2). Thus, both Lys133 and Arg164 residues are needed for IFN-κ to achieve effective inhibition of IAV.

CHD6 is the primary effector mediating the anti-influenza activity of IFN-κ

We hypothesized that dissecting the differences in downstream target genes between wild-type IFN-κ and IFN-κDual might provide clues to the molecular mechanism of antiviral action of IFN-κ. Accordingly, we analyzed the transcriptome profiles of A549 cells transfected with plasmids encoding either wild-type IFN-κ or IFN-κDual or empty vector by microchip analysis. Unexpectedly, only a few genes showed higher induction in cells expressing wild-type IFN-κ compared to cells expressing IFN-κDual (Fig. 2A). Among these differentially expressed genes, FOS, FOSB, and CHD6 drew our special attention because of their known functions in IFN-induced antiviral responses. CHD6 was previously identified as a suppressor of IAV replication (12). We verified our microchip data with quantitative reverse transcription polymerase chain reaction (qRT-PCR) analyses, which showed that CHD6 was induced by wild-type IFN-κ much more efficiently than by IFN-κDual, whereas the two IFN-κs showed no significant difference in their induction of several well-known anti-IAV ISGs including OAS1, OAS2, PKR, IFITM2, IFITM3, MX1, and MX2 (Fig. 2B). We further found that the IFN-κ–induced increase in CHD6 mRNA also occurred upon subsequent IAV infection (Fig. 2C).

Fig. 2 CHD6 is the key effector molecule mediating the anti-influenza activity of IFN-κ.

(A) Volcano plots of differentially expressed genes in A549 cells transfected with IFN-κ versus those transfected with IFN-κDual or pSV1.0 empty vector. Dashed vertical lines mark twofold change. Data were generated from n = 3 biological replicates. (B) qRT-PCR analyses of CHD6 expression and some known anti-influenza ISGs in A549 cells that were mock-transfected or transfected with either the IFN-κ or IFN-κDual construct for 24 hours. (C) qRT-PCR analysis of CHD6 transcripts in A549 cells transfected with empty vector (EV) or the IFN-κ construct for 24 hours and then infected with the indicated IAV for 48 hours. (D) qRT-PCR analysis of CHD6 transcripts in CHD6-KO A549 cells. (E) Western blotting analysis of IAV NP, IFN-κ, and CHD6 in wild-type (WT) and CHD6-KO A549 cells either EV-transfected or transfected with IFN-κ construct for 24 hours and then infected with the indicated IAV virus for 48 hours. β-Actin is a loading control. Data in (B) to (D) are means ± SD from three biological replicates. Blots are representative of three independent experiments. Quantitation of the blots (E) is shown in the Supplementary Materials (fig. S7B). *P < 0.05, **P < 0.01, and ***P < 0.001 by Bonferroni post hoc test (B) or unpaired t test (C and D).

We next determined the role of CHD6 in the anti-IAV activity of IFN-κ. To this end, we generated CHD6-knockout (CHD6-KO) A549 cells by using the CRISPR-Cas9 system (Fig. 2D) and tested IFN-κ–mediated suppression of IAV in these cells. The anti-IAV activity of IFN-κ was substantially compromised against subsequent infections of H9N2, H7N9, or PR8 in these knockout cells (Fig. 2E and fig. S7B). Thus, CHD6 appears to be a critical effector molecule mediating the anti-IAV activity of IFN-κ.

To understand the role of CHD6 in the antiviral activities of other type I IFNs, we also used CHD6-KO A549 cells to examine whether CHD6 was also required by IFN-α and IFN-β to exert their anti-influenza functions. Notably, IFN-α and IFN-β remained capable of effectively inhibiting IAV replication in these cells (fig. S3A). Thus, unlike IFN-κ, IFN-α and IFN-β did not appear to require CHD6 for their anti-influenza activities. To determine whether CHD6 was an IFN-α/β–inducible gene, we analyzed the induction of CHD6 along with three well-characterized ISGs (OAS1, MX1, and IFITM1) by IFN-α and IFN-β in comparison to IFN-κ in the absence or presence of subsequent challenge with IAV. The results showed that both IFN-α and IFN-β induced CHD6 less than did IFN-κ (fig. S3B). In contrast, IFN-α and IFN-β were more potent than IFN-κ in inducing OAS1, MX1, and IFITM1. The inductions of OAS1, MX1, and IFITM1 were similar in the presence or absence of CHD6 (fig. S3C), arguing against the possibility that CHD6 might indirectly inhibit IAV by modulating the expression of other ISGs. Thus, CHD6 is a new bona fide ISG preferentially induced by IFN-κ to directly counter the IAV.

IFN-κ stimulates CHD6 expression through IFNAR–p38–c-Fos axis

We further determined the signaling pathway by which IFN-κ stimulates CHD6 expression. Previous studies reported that different type I IFNs recognize the two subunits of IFNAR, IFNAR1 and IFNAR2, with different affinity and that IFN-β can transduce signals through IFNAR1 independently of IFNAR2 (13). To determine whether one or both IFNARs is or are the receptor(s) for IFN-κ, we used the CRISPR-Cas9 system to individually knock out IFNAR1 or IFNAR2 in A549 cells (Fig. 3A). CHD6 mRNA failed to be induced by IFN-κ in either of the two knockout cell lines after H9N2 infection (Fig. 3B). Consequently, both cell lines lost response to IFN-κ in terms of the inhibition of IAV infection (Fig. 3C). These results showed that the stimulation of CHD6 expression by IFN-κ required the engagement of the IFNAR1-IFNAR2 dimeric receptor.

Fig. 3 IFN-κ requires both IFNAR1 and IFNAR2 to transduce signals for CHD6 induction.

(A) Western blotting analysis of IFNAR1 and IFNAR2 in IFNAR1-KO and IFNAR2-KO A549 cells generated by CRISPR-Cas9 genomic editing. WT A549 cells served as a positive control. β-Actin is a loading control. (B) qRT-PCR analysis of CHD6 transcripts in WT, IFNAR1-KO, and IFNAR2-KO A549 cells that were either mock-transfected or transfected with the IFN-κ construct for 24 hours and then infected with H9N2 IAV for 48 hours. (C) Western blotting analysis of IAV NP, IFN-κ, and CHD6 in WT, IFNAR1-KO, and IFNAR2-KO A549 cells transfected with the IFN-κ construct as indicated for 24 hours and then infected with H9N2, H7N9, or PR8 virus for 48 hours. Blots are representative of three independent experiments. Data are means ± SD of three biological replicates. *P < 0.05, **P < 0.01, and ***P < 0.001 by unpaired t test.

Next, we sought to identify the intracellular signal transduction pathway through which IFN-κ stimulated CHD6 expression. As described above, the IFN-κDual mutant retained the capability to stimulate several ISGs known to be downstream targets of the canonical JAK-STAT1 pathway (Fig. 2B). This was in line with the observation that, when both wild-type IFN-κ and the IFN-κDual mutant were expressed in A549 cells, they promoted similar amounts of phosphorylated STAT1 (p-STAT1) (fig. S4). The observation that IFN-α/β stimulated OAS1, MX1, and IFITM1 more effectively than did IFN-κ but acted as a weaker inducer of CHD6 (fig. S3, B and C) further strengthened our belief that there exists an axis alternative to the JAK-STAT1 pathway for transducing the IFN-κ signal to promote CHD6 expression.

The microchip analyses described above suggested that the activity of c-Fos, encoded by FOS, might be regulated by IFN-κ (Fig. 2A). The primary known function of c-Fos is to interact with Jun to form the transcription factor complex AP-1, and the transactivating potential of c-Fos is enhanced by its phosphorylation (14). Our bioinformatics analyses predicted the presence of two potential AP-1 binding sites in the promoter of CHD6. Thus, we wanted to determine whether c-Fos plays a role in the induction of CHD6 by IFN-κ. To this end, we constructed a chimeric IFN-κ construct (IFN-κ-Fc) that contains a human immunoglobulin G2a (IgG2a) Fc fragment fused to the C terminus of IFN-κ as an affinity tag for purification and a signal peptide replacement from mouse interleukin-2 (IL-2) protein for improving secretion (fig. S5A). This recombinant IFN-κ would allow virtually all the cells to be stimulated and is suitable for time-course analysis of cellular signaling upon IFN stimulation. We first compared the anti-influenza activity of nonmodified IFN-κ versus the chimeric version when both were introduced into cells by transfection. Both IFN-κ constructs effectively suppressed subsequent IAV infection in A549 cells, demonstrating that the protein engineering did not adversely affect the activity of IFN-κ (fig. S5B). We further tested the IFNAR dependency of IFN-κ-Fc along with its capability to induce CHD6, using recombinant IFN-α2 as a control. In contrast to essentially no anti-IAV response elicited by IFN-κ-Fc pretreatment in either IFNAR1-KO or IFNAR2-KO cells, the IFN-α2–mediated inhibition of IAV was only eliminated in IFNAR1-KO cells (fig. S5C). This result further validated our above conclusion that IFN-κ signaling is potentially transduced through IFNAR1-IFNAR2 dimeric receptor. To analyze the CHD6 induction, A549 cells were exposed to equal amounts of IFN-α2, IFN-κ-Fc, or IFN-λ for a period ranging from 30 to 120 min, followed by immunoblotting for CHD6. IFN-λ, a type III IFN, was tested because it mainly acts on epithelial cells. We found that, in agreement with our above analysis of IFN-expressing cells (fig. S3B), IFN-α2 only weakly stimulated the production of CHD6 protein with a degree of induction significantly lower than that of IFN-κ-Fc. IFN-λ appeared to behave similarly to IFN-α2 to be a weak CHD6 inducer (fig. S6).

We subsequently compared the time courses of c-Fos activation in parallel with CHD6 induction in A549 cells responding to IFN-κ-Fc stimulation. The amount of phosphorylated c-Fos (p-c-Fos) was significantly increased at 10 min after the stimulation and maintained throughout the 120-min detection period, whereas the enhanced accumulation of CHD6 became evident 30 min after stimulation (Fig. 4A and fig. S7C). This result was consistent with the idea that IFN-κ stimulates the transcription of CHD6 by stimulating the phosphorylation, and therefore the activity, of c-Fos.

Fig. 4 IFN-κ stimulates CHD6 expression through the p38-Fos axis.

(A) Western blotting analysis of phosphorylated c-Fos (p-c-Fos), total c-Fos, and CHD6 in lysates of A549 cells exposed to purified hIFN-κ-Fc protein for the indicated time. β-Actin is a loading control. (B) Western blotting analysis of p-c-Fos and total c-Fos in A549 cells pretreated with inhibitors of STAT1 (iSTAT1, fludarabine), JNK (iJNK, SP600125), ERK (iERK, SCH772984), p38 (ip38, SB203580), p65 (ip65, curcumenol and PG490), or phosphorylated AKT (ipAKT, triciribine), as indicated for 6 hours, and then stimulated with purified hIFN-κ-Fc protein for 30 min. (C and D) Western blotting analysis of phosphorylated p38 (p-p38), total p38, p-c-Fos, total c-Fos, and CHD6 in A549 cells (C) and BEAS-2B cells (D) pretreated with the p38 inhibitor SB203580, as indicated for 6 hours before exposure to purified hIFN-κ-Fc protein for 30 min. (E) Western blotting analysis of IAV NP, p-p38, and total p38 in A549 cells pretreated with SB203580 for 6 hours, as indicated before exposure to either IFN-κ-Fc or IFN-α2 for 24 hours, and subsequently infected with PR8 virus for 48 hours. (F) Western blotting analysis of phosphorylated STAT1 (p-STAT1), total STAT1, and CHD6 in A549 cells pretreated with iSTAT1 (fludarabine), as indicated for 6 hours before hIFN-κ-Fc exposure for 30 min. (G) Western blotting analysis of IAV NP, p-STAT1, and total STAT1 in A549 cells pretreated with fludarabine, as indicated for 6 hours before exposure to hIFN-κ-Fc for 12 hours, and then infected with PR8 virus for 48 hours. Blots are representative of three independent experiments. Quantitation of the blots in (A), (B), (F), and (G) is shown in the Supplementary Materials (fig. S7, C to F).

To pin down the IFN-κ signaling upstream of c-Fos activation, we evaluated IFN-κ-Fc–induced c-Fos phosphorylation in the presence of various small-molecule inhibitors targeting key signaling proteins known to be activated by type I IFNs. The inhibitor panel included fludarabine, SB203580, curcumenol and PG490, SCH772984, triciribine, and SP600125, which inhibit STAT1, the mitogen-activated protein kinase (MAPK) p38, the nuclear factor κB (NF-κB) subunit p65, extracellular signal–regulated kinase (ERK), the kinase AKT, and c-Jun N-terminal kinase (JNK), respectively. Only SB203580, the specific inhibitor of p38, substantially reduced the phosphorylation of c-Fos, whereas all other inhibitors had no or negligible effects (Fig. 4B and fig. S7D). We then studied the effect of SB203580 on IFN-κ–mediated CHD6 induction and IAV inhibition. When SB203580-pretreated A549 cells were exposed to IFN-κ-Fc, they showed essentially no increase in CHD6 (Fig. 4C). A549 cells are lung cancer cells and may transduce IFN-κ signaling differently from normal lung cells. To exclude this possibility, we repeated the SB203580 treatment experiment in the BEAS-2B cell line, which is a normal human bronchus cell line, and observed a similar inhibitory effect of SB203580 on c-Fos phosphorylation and CHD6 induction (Fig. 4D). As expected, the elimination of IFN-κ–mediated induction of CHD6 by SB203580 pretreatment was concomitant with the loss of inhibition of IAV infection. In contrast, SB203580 had little or no impact on the IFN-α2–induced suppression of IAV, consistent with our conception that CHD6 is dispensable for the anti-influenza action of IFN-α/β. Notably, we found that the amount of phosphorylated p38 (p-p38) induced by IFN-κ was much higher than that induced by IFN-α2, which might explain the preferential induction of CHD6 by IFN-κ as compared with IFN-α/β (Fig. 4E).

Given the central role of the JAK-STAT1 pathway in type I IFN signaling, we determined the relationship between this pathway and p38–c-Fos–CHD6 axis in the IFN-κ–induced anti-influenza response by using fludarabine to block STAT1 phosphorylation. Fludarabine pretreatment did not block the IFN-κ–mediated induction of CHD6, reinforcing the notion that the JAK-STAT1 pathway plays no role or only a minor role in the regulation of CHD6 (Fig. 4F and fig. S7E). However, pre-exposure to fludarabine compromised the inhibitory effect of IFN-κ on IAV infection (Fig. 4G and fig. S7F), indicating an important role of JAK-STAT1 pathway in IFN-κ–mediated anti-influenza cellular response. Together, these results demonstrated that IFN-κ stimulated CHD6 expression and accumulation of CHD6 protein through a p38–c-Fos pathway rather than the JAK-STAT1 pathway, but both pathways were required by IFN-κ to suppress IAV infection.

Pretreatment of IFN-κ afforded effective protection against IAV challenge in mice

We conducted a proof-of-principle experiment in mice to test the prophylactic efficacy of IFN-κ against IAV infection. Either purified mouse IFN-κ protein (IFN-κ-Fc) or phosphate-buffered saline (PBS) was administered intranasally to mice 6 hours before challenge with PR8 IAV. Compared to the PBS-pretreated group, the IFN-κ–pretreated group showed greatly reduced susceptibility to PR8 challenge as indicated by weight recovery and survival rates (Fig. 5A). Histological examination of infected lung tissues showed that IFN-κ pretreatment ameliorated PR8-induced pulmonary pathologies such as edema, alveolar hemorrhaging, increased alveolar wall thickness, and neutrophil infiltration on day 7 after infection (Fig. 5B). For a better appraisal of lung physiology as a whole, we adopted a semiquantitative method wherein each individual pathological parameter was integrated to generate an overall score (Table 1). The resulting score curves (Fig. 5B) revealed that, although the pathological score of the IFN-κ–pretreated group was higher than that of PBS-pretreated controls (6 versus 3) through day 2, it decreased afterward, reaching approximately the value of 4 on day 7. In contrast, the pathological score of the PBS group continued to rise, reaching a value of 9 on day 7. We also determined whether the p38–c-Fos–CHD6 axis was activated by IFN-κ in infected lung tissues. The IFN-κ–pretreated group showed a significantly greater increase in p38 phosphorylation in lung on day 1 after infection, the earliest time point examined, as compared with the control group (Fig. 5C). Phosphorylation tapered off thereafter, possibly due to the discontinuation of IFN-κ stimulation. The detection of CHD6 transcripts using RNA in situ hybridization indicated that the amount of CHD6 mRNA in IFN-κ–pretreated mice was substantially higher than in the PBS control group throughout the course of infection (Fig. 5D). These animal studies provided further evidences in support of our conclusion that IFN-κ stimulated signaling that activated p38 and consequently promoted the production of CHD6, which acts as an effector molecule to mediate the suppression of IAV replication.

Fig. 5 IFN-κ is protective against IAV infection in mice.

Mice were injected intranasally with IFN-κ-Fc protein or PBS 6 hours before intranasal challenge with PR8 virus. (A) Body weight change and survival. n = 12 mice for each group. Body weight change data are means ± SEM. (B) Representative images of H&E-stained lung sections from IFN-κ-Fc and PBS treatment groups at day 1, 2, 3, and 7 after infection (D1, D2, D3, and D7). Quantified pathological scores were calculated on the basis of lung histology data. n = 3 mice for each group. Scale bar, 200 μm. Results are expressed as mean ± SD. (C) Quantification of histochemical staining for p-p38 in lung tissues from the IFN-κ-Fc and PBS treatment groups and representative images. n = 3 mice for each group. Scale bar, 100 μm. Quantification data are expressed as mean ± SD. (D) Quantification of and representative images showing in situ hybridization for CHD6 mRNA in lung tissues from mice treated with PBS or IFN-κ-Fc. Scale bar, 50 μm. Quantification data are expressed as means ± SD (n = 3 for each group). *P < 0.05, **P < 0.01, and ***P < 0.001 by unpaired t test.

Table 1 Definitions of scores for the inflammatory response in lung tissue.

The grading of histology sections incorporated evaluations from three categories, including (A) degree of bronchial infiltrate (<25%, 25 to 50%, 50 to 75%, >75%), (B) frequency of inflammatory cells (none, <10%, 10 to 20%,>20%), and (C) degree of parenchymal pneumonia/infiltrate (<25%, 25 to 50%, 50 to 75%,>75%). The total score (ranged from 0 to 9) was set as sum of subscores from (A) to (C). Total score <3, mild; total score 3 to <6, moderate; total score 6 to ≤9, intense.

View this table:


Although biochemical and structural studies have indicated that all type I IFNs signal through the same IFNARs, it has been speculated that different type I IFNs have different physiological functions (15). In this study, we found that IFN-κ was induced in respiratory epithelial cells in response to IAV infection and was capable of inhibiting IAV by inducing the production of CHD6 protein. Moreover, we showed that IFN-κ treatment afforded substantial protection against lethal PR8 challenges in mice. Collectively, our data support that IFN-κ functions in lung epithelium to promote the innate anti-influenza response, rather than playing a specialized role in skin immunity, as suggested by its constitutive expression in keratinocytes (16).

Our dissection of the IFN-κ signaling pathway revealed a clear divergence from the IFN-α/β paradigm. This divergence started with receptor engagement, because both IFNAR1 and IFNAR2 were essential for IFN-κ to induce CHD6 and eliminate influenza in A549 cells, whereas IFN-α/β can act through a single IFNAR. This is in line with a biochemical study showing that the in vitro binding affinities of IFN-κ to IFNARs, especially IFNAR2, are significantly lower than those of IFN-α/β (17). The importance of IFNAR recognition in imparting signaling specificity was further highlighted by our identification of the IFN-κDual mutant from human A549 cells. IFN-κDual failed to inhibit IAV infection, and comparison of it with wild-type IFN-κ in the activation of downstream genes led us to discover CHD6 as the pivotal effector molecule mediating the IAV-inhibitory effect of IFN-κ. Despite its inability to stimulate CHD6 expression, IFN-κDual retained the ability to stimulate expression of some other ISGs, consistent with largely normal STAT1 phosphorylation. Future biochemical and structural characterization of IFNAR recognition by wild-type IFN-κ versus IFN-κDual might provide new insight into what governs the biological consequences of different IFNs—the binding affinity to the receptor, the binding mode, or both.

By discovering the intracellular IFN-κ signaling axis responsible for CHD6 induction, we further elucidate the divergence between IFN-κ and IFN-α/β. The IFN-κ–mediated induction of CHD6 was unexpectedly independent of the JAK-STAT1 pathway, which is considered to be the default pathway through which type I IFNs regulate transcription, but rather through a p38–c-Fos axis. p38 MAPK was previously reported to participate in transcriptional modulation imparted by IFN-α/β (18); however, the actual mechanism was not identified. We observed an increase in p38 phosphorylation in IFN-κ–treated cells relative to IFN-α–treated cells, providing a plausible explanation for the enhanced potency of IFN-κ in stimulating CHD6 expression compared to IFN-α/β. In contrast, for all the known ISGs we tested, the stimulatory effect of IFN-κ appeared to be weaker than that of IFN-α or IFN-β, especially IFN-β. Thus, CHD6 is a bona fide ISG preferentially induced by IFN-κ.

Our functional analyses highlighted the importance of CHD6 in the IFN-κ–mediated anti-IAV innate response. In the absence of CHD6, the capacity of IFN-κ to inhibit IAVs was substantially compromised and, in some cases, close to a complete loss, whereas there was negligible effect on the IAV inhibition afforded by IFN-α/β. This result, coupled with our finding that IFN-κ promotes expression of only a small portion of ISGs induced by IFN-α/β (Fig. 2B and fig. S3, B and C) and the previously published finding that IFN-κ binds to individual IFNARs weaker than does IFN-α/β (17), suggested different modes of actions between IFN-α/β and IFN-κ. The former use multiple signaling pathways to activate a diverse array of ISGs, exerting profound effects on both virus and cells, whereas the latter instead exhibited a selective use of downstream signaling arsenals, the p38–c-Fos axis in particular, resulting in a relatively narrower spectrum of downstream targets among which some effector genes are preferentially simulated, as seen for CHD6. Such focused strategy, underlying the observed dominance of a single effector molecule in the antiviral activity of IFN-κ—CHD6 for influenza as shown here and Sp100 for human papillomavirus (HPV) as shown in a previous study (19)—would bestow a benefit on the host by constraining the responding cells from overreacting. It is thus tantalizing to speculate that the temporal regulation of different type I IFNs may balance viral containment with the control of host tissue damage.

The lack of early IFN-κ induction during H7N9 infection may result from the action of one or more viral proteins. NS1A is widely regarded as the primary influenza viral protein countering innate immunity. One major function of NS1A is binding to CPSF30 and consequently blocking the pre-mRNA processing. The NS1 protein of H7N9 virus does not have the two residues (Phe103 and Met106) that are critical for CPSF30 binding. Among type I IFNs, IFN-κ is the only member containing an intron, which is located in its 3′-untranslated region. Some publications have suggested that NS1A is able to regulate mRNA splicing (20). However, this capability is mainly mediated by the RNA binding domain, which is highly conserved between different NS1As (21, 22). A major difference between the H9N2 and H7N9 strains used in this study lies in their PB2 proteins, with H9N2 PB2 having an avian signature of glutamic acid at position 627 and H7N9 PB2 containing lysine at the same position, which accounts for the increased viral polymerase activity in mammalian cells. It will be of interest for future studies to determine whether such PB2 polymorphisms also underlie the differential activation of IFN-κ by the two viruses.

Our animal study demonstrated the prophylactic use of recombinant IFN-κ protein against influenza infection, validating our findings that IFN-κ activated p38 kinase and subsequently CHD6 transcription in cultured human cells. It will be exciting to examine in the future whether IFN-κ exhibits inhibitory activity against other important viruses, which could substantiate the usage of IFN-κ as a broad-spectrum reagent to treat human viral diseases.


Animal studies

The mice used in this study were 8- to 10-week-old = C57BL/6 mice, which were purchased from the B&K Universal Group Limited (Shanghai, China) and maintained under specific pathogen–free conditions at the animal facilities of Shanghai Public Health Clinical Center (Shanghai, China). For each experimental group, there were at least five mice analyzed at each collection time point. For IAV infection experiments, mice were intranasally inoculated with either 3.5 × 105 TCID50 (tissue culture infectious dose) of H7N9 virus, or 1.7 × 107 EID50 (50% egg infectious dose) of H9N2 virus, or PBS (control group) in a volume of 50 μl. After infection, mice were monitored daily for clinical symptoms, survival, and body weight during a 14-day observation period. Mice losing over 30% of their initial body weight were considered dead and humanely euthanized. For microarray-based transcriptome analysis, infected mice were sacrificed at various time points ranging from 6 hours to day 14 after virus challenge for collection of lung tissues from which RNA was extracted and analyzed. For evaluation of prophylactic efficacy of IFN-κ, mice were subjected to intranasal injection of IFN-κ protein (2 mg/kg) or PBS (control group), followed 6 hours later by infection with 500 TCID50 PR8 given intranasally. All animal experiments were performed in accordance with Home Office guidelines and were approved by the Shanghai Public Health Center Local Ethical Committee.

Histopathological examination and RNA in situ hybridization

Lung tissues were excised from sacrificed mice, fixed with formalin, and subsequently embedded in paraffin, followed by transverse sectioning at 5-μm thickness with a microtome. The resulting slides were stained with hematoxylin and eosin (H&E) and scanned by TissueFAXS Confocal Plus 200 (TissueGnostics). The acquired images were analyzed by Strata Quest 6.0X software to assess the number of bronchi, inflammatory cells, and inflammatory infiltration areas. Histopathology severities were further graded by a reported scoring system, which was originally used for evaluation of histopathology associated with infection of respiratory syncytial virus (23). The slides for in situ detection of CHD6 transcript were prepared following the same procedure, and the RNA in situ hybridization was performed according to the manufacturer’s instruction (RNAscope, Advanced Cell Diagnostics).

Cells and viruses

Human embryonic kidney (HEK) 293T (CRL-11268), A549 (CCL-185), BEAR-2B (CRL-9609), and Madin-Darby canine kidney (MDCK; CCL-34) cells were obtained from the American Type Culture Collection; U-251MG was purchased from BeNa Culture Collection. All the cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM; Mediatech Inc.) supplemented with 10% fetal bovine serum and 1% PSG [penicillin (100 U/ml), streptomycin (100 μg/ml), and 2 mM l-glutamine] at 37°C in a humidified atmosphere containing 5% CO2. The influenza A/Puerto Rico/8/1934 (PR8) virus was grown on MDCK cells. The influenza A/Chicken/Shanghai/F/98 (H9N2) virus (GenBank accession numbers: AY253750.1, AY253751.1, AY253752.1, AY253753.1, AY253754.1, AY743215.1, AY253756.1, and AY743216.1), A/California/07/2009 (H1N1) virus (GenBank accession numbers: KU933482.1, KU933483.1, KU933484.1 KU933485.1 KU933486.1, KU933487.1, KU933488.1, and KU933489.1), and A/Shanghai/4664T/2013 (H7N9) virus (GenBank accession numbers: KC853225.1, KC853226.1, KC853227.1, KC853228.1, KC853229.1, KC853230.1 KC853231.1, and KC853232.1) were propagated in 10-day-old embryonated chicken eggs. TCID50 or EID50 of virus stock was determined on MDCK cells or embryonated eggs, respectively.

Virus infection of cultured cells

The indicated cells were incubated with indicated IAV in serum-free DMEM for 2 hours at 37°C. The multiplicity of infection (MOI) for H9N2, PR8, and H7N9 is 2, 1, and 0.5, respectively. After incubation, the inoculum was removed and prewarmed 2% fetal bovine serum–containing DMEM was added to the cells after washing three times with PBS. Mock- and virus-infected cells were harvested 48 hours after infection, and the resulting whole-cell lysates were analyzed by immunoblotting using indicated antibody. In case of viral growth determination, the supernatants were assessed for TCID50 titer using MDCK cells.

IFN-κ expression constructs

The original attempt to clone IFN-κ gene from human A549 cells using RT-PCR led to the discovery of a mutant IFN-κ gene, named IFN-κDual, because it differs from published gene sequence by a lysine to aspartic acid substitution at position 133 (K133E) and an arginine to lysine at position 164 (R164K). The correct IFN-κ gene as well as two mutants carrying single-point K133E or R164K substitution were generated by PCR-mediated mutagenesis using IFN-κDual as the template. The three IFN-κ coding sequences, along with that of IFN-κDual, were individually cloned into pSV1.0 vector to generate expression vector for transient transfection experiments.

CRISPR-Cas9–mediated gene knockout

IFNAR1, IFNAR2, and CHD6-targeting single-guide RNAs (sgRNAs) were designed according to the Zhang laboratory CRISPR design tool ( and individually cloned into the LentiCRISPR v2 vector (Addgene, 52961) to construct knockout plasmids that were subsequently used for lentivirus production in HEK293T cells. The sequences for IFNAR1, IFNAR2, and CHD6-specific sgRNAs were as follows: GCTCGTCGCCGTGGCGCCAT (IFNAR1), GTGTATATCAGCCTCGTGTT (IFNAR2), and GCGAAGCCCAATGGCGCGAC (CHD6). The lentivirus-mediated introduction of sgRNA into A549 cells was performed following published protocol (24). The identification of IFNAR1 and IFNAR2 knockout clones was facilitated by flow cytometry analysis using phycoerythrin-conjugated IFNAR1 antibody (PBL Assay Sci, 21370-3) or IFNAR2 antibody (PBL Assay Sci, 21385-3) to detect the cell surface expression of IFNAR1 or IFNAR2. The CHD6 knockout clones were identified by CHD6 immunoblotting of whole-cell lysates.

Reverse transcription polymerase chain reaction

RNA was extracted using RNEasy RNA isolation kit (Qiagen) according to the manufacturer’s instructions. The reverse transcription of RNA sample was carried out by GoScript reverse transcriptase (Promega, Charbonnieres, France), using oligo(dT) primer with 1 μg of RNA per reaction. SYBR green–based RT-PCR analysis of the resulting complementary DNA was performed using the GoTaq qPCR Master Mix (Promega, Charbonnieres, France) on an Applied Biosystems 7300 RT-PCR cycler. The qRT-PCR data were analyzed using SDS software (Applied Biosystems) to determine the relative transcript abundance of gene of interest after normalization with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) internal control (GAPDH primers). The primer sequences used in the study are listed in table S1.

Immunoblotting and antibodies

Whole-cell lysates were prepared using minute total protein extraction kit (Invent, SD001/SN002) and separated by denaturing SDS gel electrophoresis, followed by transfer to nitrocellulose membranes. The resulting membranes, after blocking with either 5% dried skim milk or 5% bovine serum albumin (in the cases of detection of protein phosphorylation) in PBS supplemented with 0.1% Tween 20 (PBST) for 1 hour at room temperature, were incubated with indicated antibodies in PBST overnight at 4°C. After extensive washing with PBST, the membranes were incubated with horseradish peroxidase–conjugated secondary antibody either against mouse (1:3000, B2615, Santa Cruz Biotechnology) or rabbit (1:3000, B2615, Santa Cruz Biotechnology) for 1 hour at room temperature, developed with the SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Scientific), and then photographed on a Kodak Image Station. The quantification of protein band was performed using ImageJ software. The involved antibodies include anti-NP (1:1000, Bioss, BS-4976R), anti–IFN-κ (1:1000, Abnova, H00056832-M01), anti-STAT1 [1:1000, Cell Signaling Technology (CST), 9172s], anti–p-STAT1 (1:1000, CST, 7649s), anti–c-Fos (1:1000, CST, 2250s), anti–p-c-Fos (1:1000, CST, 5348s), anti-CHD6 (1:500, Bethyl, A301), anti–p-p38 (1:1000, CST, 8690s), anti-p38 (1:1000, CST, 9216s), and anti-actin.

Microchip assay

For preparation of RNA samples from lung tissues of infected mice, the tissues were flash-frozen in liquid nitrogen and pulverized with a mortar and pestle, and RNA was then extracted using RNAzol (Molecular Research Center Inc., USA) coupled with the Direct-zol RNA MiniPrep Kit (Zymo Research, Irvine, CA, USA). For preparation of RNA samples from IFN-κ–expressed A549 cells, the plasmids expressing wild-type IFN-κ or IFN-κDual, or empty vector (mock) were individually transfected into A549 by TurboFect (Thermo Scientific, R0531), and 24 hours later, the cells were lysed in RNAzol and the contained RNA was purified using the Direct-zol RNA MiniPrep Kit (Zymo Research, Irvine, CA, USA). Transfection of each plasmid was performed in three biological replicates. The quality of RNA was ensured by measurement on an Agilent Bioanalyzer 2100 using RNA NanoChip. Micro-chip assay was performed using Agilent SurePrint G3 Human Gene Expression v2. The transcript abundances of individual genes in wild-type IFN-κ–expressing cells were expressed as fold changes relative to mock group and IFN-κDual group, and a minimum of 1.5-fold change was used as a starting cutoff for identification of differentially expressed genes. T test criteria were also applied to separation of differentially expressed gene from nondifferentially expressed gene. The microchip data were deposited in Gene Expression Omnibus (GEO) under accession number GSE142709.

Expression and purification of hIFN-κ-Fc proteins

The hIFN-κ-Fc–expressing plasmid was generated by inserting into pSV1.0 vector an engineered hIFN sequence with the original signal peptide of IFN-κ being replaced by that of mice IL-2 and human IgG2a being fused to the C terminus of IFN-κ. For the production of hIFN-κ-Fc protein, the hIFN-κ-Fc–expressing plasmid was transfected into 293T cells, which were cultured in Expi293 medium for additional 72 to 96 hours. The medium was then collected and subjected to overnight selection with IgG beads (GE Healthcare) at 4°C. After extensive washing, the hIFN-κ-Fc protein bound to the beads was eluted with glycine buffer, pH 2.7, followed by neutralization with tris buffer (pH 9.0). The concentration of the yielded protein preparation was determined by NanoDrop, and its purity was confirmed with SDS–polyacrylamide gel electrophoresis and Coomassie blue staining. The concentration of hIFN-κ-Fc used for cell culture experiment was 1 μg/ml, the same as recombinant IFN-α2 (C005, novoprotein) and IFN-λ (30002L, PeproTech).

Statistical analysis

All statistical analyses were conducted using the GraphPad software (Prism 5; San Diego, CA). Data were expressed as mean ± SD or SEM. One-way analysis of variance (ANOVA) test was applied when comparing more than two groups, and the differences between groups were evaluated using the post hoc Bonferroni test. A P value of ≤0.05 was considered statistically significant.


Fig. S1. IAV-mediated induction of IFNs in human A549 cells.

Fig. S2. Inhibition of IAV by IFN-κ but not its A549-derived mutants.

Fig. S3. IFN-κ preferentially stimulates CHD6 expression upon subsequent influenza infection.

Fig. S4. IFN-κDual largely retains the capability to stimulate STAT1 phosphorylation.

Fig. S5. Design, purification, and functional analyses of a recombinant IFN-κ protein.

Fig. S6. Temporal induction of CHD6 in A549 cells responding to different recombinant IFNs.

Fig. S7. Densitometric quantification of immunoblots.

Table S1. Primer sequences.


Acknowledgments: We thank W. J. Liu (Chinese Center for Disease Control and Prevention) for providing the H1N1/A/California/07/2009 virus, T. Qiu (Shanghai Public Health Clinical Center, Fudan) for providing us a modeled structure of IFN-κ/IFNAR1/IFNAR2 complex, and X.-Y. Wang (Fudan University) for being an independent consultant of our manuscript on statistical data analysis. Funding: This work was supported by the National Science Foundation of China (81430030 and 81771704), Shanghai Outstanding Academic Leaders Plan (2016-035), Shanghai Pujiang Program (19PJ1409100), and National 13th Five-Year Grand Program on Key Infectious Disease Control and Prevention (2017ZX10304402-002-007 and 2018ZX10301403-003). Author contributions: J.X. conceived this study. J.X., X.Z., C. Zhao, and Y.H. designed the experiments. Y.H., W.F., K.C., Q.H., X.D., S.Y., J.C., T.C., Y.Y., and C. Zhu performed the experiments. L.D. prepared the virus. L.Z. analyzed the microchip data. Z.L. provided virus and reagents. Y.H. and C. Zhao prepared the manuscript. J.X., X.Z., and C. Zhao revised the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: The microchip data were deposited in GEO ( under accession number GSE142709. All other data needed to evaluate the conclusions in this paper are present in the paper or the Supplementary Materials.

Stay Connected to Science Signaling

Navigate This Article