Interferons (IFNs) elicit multifaceted effects in host innate defence. Accumulating evidence revealed that not only the first identified Jak-Stat pathway but also other newly found signalling pathways are required for the induction of versatile responses by IFNs. In particular, type I IFNs are inducible by viral infection through the recognition of pathogen-associated molecules by pattern recognition receptors, and the induction of multiple IFN-stimulated genes through the activation of type I IFN signalling confers antiviral and immunomodulatory activities. Any step in this process is often targeted by viruses for their immuno-evasion. The regulatory function of constitutive IFN-α/β signalling has been recognized in terms of its boosting effect on cellular responsiveness in host defence systems. Further comprehensive understanding of IFN signalling may offer a better direction to unravelling the complex signalling networks in the host defence system, and may contribute to their more effective therapeutic applications.
The transcriptional regulation of thousands of effector genes downstream of the Jak-Stat or other IFN-regulated pathways (Der et al., 1998) contributes to pleiotropic responses elicited by IFNs. Thus, the IFN system, which triggers the induction of numerous antiviral genes during viral infection, is a formidable barrier against viral multiplication in the infected host. On the other hand, it has become clearer that there are multiple strategies that viruses evolve for the evasion from the IFN system (Goodbourn et al., 2000; Levy and Garcia-Sastre, 2001; Garcia-Sastre, 2002; Katze et al., 2002; Weber et al., 2004; Johnson and Gale, 2005). Such a better understanding of virus–host interactions is also emerging with more obvious recognition of the vital part of the IFN system in the host defence system.
In this review, we focus mainly on type I IFN-mediated signalling and their regulatory functions in host defence systems. Type I IFNs, IFN-α/β, are massively produced in most cell types in response to viral and other microbial infections, and play a vital role in innate resistance to a wide variety of viruses through the induction of antiviral effects, both directly and indirectly manner. As for their indirect role in antiviral activities, IFN-α/β also have an immunomodulatory effect of activating natural killer (NK) cells, macrophages and dendritic cells (DCs), all of which are essential effector cells in the innate immune system (Pestka et al., 1987; De Maeyer and De Maeyer-Guignard, 1988; Vilcek and Sen, 1996; Biron et al., 1999; Theofilopoulos et al., 2005). By virtue of their potentiating effect on DC maturation, type I IFNs are currently recognized as pivotal cytokines bridging two aspect of host defence, innate and adaptive immune systems. It is noteworthy that a unique subset of DCs, termed plasmacytoid DCs (pDCs), was identified, a hallmark function of which is the capacity to produce prodigious amounts of type I IFNs, consisting of IFN-α, IFN-β and IFN-ω, upon viral exposure or the activation of certain Toll-like receptors (TLRs) (Liu, 2005).
On the other hand, with accumulating reports about the potential role of type I IFNs in tumour suppression, we also found a new linkage between IFN-α/β signalling and the tumour suppressor p53, which provided an insight into a possible role of p53 not only in IFN-α/β-mediated antitumour activity but also in antiviral defence (Takaoka et al., 2003). Furthermore, Schreiber’s group recently reported that type I IFNs are important components for cancer immunoediting process wherein protective antitumour responses are developed through their effect on immune systems (Dunn et al., 2005). In this regard, it is interesting that constitutively produced type I IFNs play a role in this process.
As also indicated above, recent advances elucidating the biology of these key cytokines also contribute to a better understanding of the important role of ‘a weak signalling’ by constitutively produced IFN-α/β in the absence of viral infection in modifying cellular responsiveness in the immune and other biological systems (Taniguchi and Takaoka, 2001; Takaoka and Taniguchi, 2003). Indeed, there are accumulating reports regarding low levels of IFN-α/β production observed in the absence of viral infection, both in vitro and in vivo (Bocci, 1985; Tovey et al., 1987; De Maeyer and De Maeyer-Guignard, 1988). Our recent studies revealed that this constitutive, weak IFN-α/β signalling, transmitted independently of viral infection, is critical for the enhancement of other cytokine signalling and their induction by virus, as well as for positive regulatory functions in adaptive immune responses such as CD8+ T cell activation (Hida et al., 2000). Furthermore, this constitutive signalling has been shown to negatively regulate the differentiation of the CD8α+ myeloid DC subgroup (Honda et al., 2004a; Ichikawa et al., 2004).
In this article, we outline the signalling events induced by both type I and type II IFNs as well as recently identified IFN family members, with the main focus on the overall signalling profile of type I IFN responses during viral infection from the offensive and defensive aspects of both the host and viruses. Furthermore, we also review the regulatory mechanism and the critical roles of constitutive type I IFN-mediated signalling in several facets of host defence systems.
Interferon family members
Although in this review we mainly focus on the signalling activated by IFNs, particularly type I IFNs, a brief explanation will be initially made about their ligands, which can be divided into three types on the basis of their structural and functional properties (Table 1).
The type I IFNs (Roberts et al., 1998) consist of IFN-α, -β, -ω, -ɛ (Langer et al., 2004; Pestka et al., 2004) and -κ (LaFleur et al., 2001) (Table 1). In addition, IFN-δ (Lefevre et al., 1998), -τ (Roberts et al., 1999) and -ζ (limitin) (Oritani et al., 2000) are included in this group, although they are only detected in pigs/cattle, ruminants and mice respectively (Table 1). All these members are induced in virally infected cells to confer an antiviral state on uninfected cells. Most of the intensive studies have focused mainly on IFN-α/β. Recent rapid progress in studies about pathogen recognition by the host revealed that the induction of IFN-α/β genes are mediated by the activation of pattern recognition receptors (PRRs) including TLRs (Takeda et al., 2003; Akira and Takeda, 2004; Takeda and Akira, 2005; Kawai and Akira, 2006) and retinoic acid-inducible gene-I (RIG-I) and melanoma differentiation-associated gene 5 (Mda5) (Rothenfusser et al., 2005; Yoneyama et al., 2005). IFN-α/β also show a variety of important immunomodulatory roles in not only innate immune responses but also adaptive immune responses. Additionally, a direct or an indirect tumour suppression is one of the major therapeutical activities of IFN-α/β (Parmar and Platanias, 2003). All of the members transmit signals through a receptor complex composed of two subunits, IFNAR-1 and IFNAR-2, although it seems that there are some differences in both quality and efficiency among them; e.g. a significant difference in the activity of boosting human NK cells was found among IFN-α subtypes (Ortaldo et al., 1984). Interestingly, some studies showed that although IFNAR-2 is phosphorylated by treatment with both IFN-α and IFN-β, the association of IFNAR-2 with IFNAR-1 is only detected upon IFN-β stimulation (Abramovich et al., 1994), and that Tyk2-deficient cells lose their responsiveness to IFN-α but still respond to IFN-β (Velazquez et al., 1992). These studies suggest that the interaction of IFN-β and IFN-α with the receptor might lead to a different conformational assembly of the receptor complex, resulting in a distinct activation of downstream signalling pathways.
Type II IFN
Type II IFN comprises solely IFN-γ (Bach et al., 1997; Pestka, 1997; Ikeda et al., 2002) (Table 1), which similarly has antiviral activity. However, this cytokine is strongly produced by activated T cells or NK cells but not by virus-infected cells. IFN-γ signalling is essential for the activation of macrophages to constitute the effective form of innate immunity to intracellular microorganisms such as Mycobacteria and Listeria, and also contributes to the promotion of the development of CD4+ Th1 cells and cytotoxic CD8+ T cells (Ikeda et al., 2002). IFN-γ signals through a pair of receptor subunits, IFNGR-1 and IFNGR-2.
Type III IFNs
Newly identified IFN members, IFN-λs or IL-28/29, have been identified by two different research groups (Kotenko et al., 2003; Sheppard et al., 2003; Vilcek, 2003) (Table 1). In humans, this group includes three homologous proteins, IFN-λ1-3 (IL-28A, IL-28B and IL-29). Similar to type I IFNs, they are induced upon viral infection and have their antiviral activity by inducing so-called IFN-stimulated genes such as those encoding OAS (2′,5′-oligoadenylate synthetase), PKR (protein kinase, double-stranded RNA-dependent) and MxA, through the activation of unidentified Jak kinase(s) and subsequent formation of the IFN-stimulated gene factor 3 (ISGF3) complex (Fig. 1). However, the major differences are that they are structurally distinct from type I IFNs and that they utilize their specific receptor subunit, IFN-λR1 or IL-28Rα, together with IL-10R2 that is known to be a shared receptor subunit among IL-10, IL-22 and IL-26. In this regard, they might be separated into the third group (type III IFNs). Tyk2, which was already shown to physically interact with IL-10R2 (Kotenko et al., 1996), is likely a candidate kinase, whereas the Jak kinase(s) associated with the IFN-λR1 subunit have not yet been identified. Downstream signalling pathways activated by IFN-λs remain to be clarified in further detail.
These receptors that are utilized by all types of IFNs belong to the class II cytokine receptor family (CRF2) (Langer et al., 2004), other members of which include receptors for the IL-10 family members (IL-10, IL-19, IL-20, IL-22, IL-24 and IL-26). In all cases, ligand-binding to these receptors leads to the activation of the Jak-Stat pathway (described below).
Overview of cardinal signalling pathways activated by type I and II IFNs
The binding of both types of IFNs to IFNAR or IFNGR results in the cross-activation of these Jak PTKs, which then phosphorylate their downstream substrates, which are two members of the family of signal transducers and activators of transcription (Stats), namely, Stat1 and Stat2 (Darnell et al., 1994; Ihle and Kerr, 1995; Schindler and Darnell, 1995; Stark et al., 1998). The tyrosine phosphorylation of these Stats leads to the formation of two transcriptional activator complexes, IFN-α-activated factor [AAF; also termed IFN-γ-activated factor (GAF); Decker et al., 1991a] and IFN-stimulated gene factor 3 (ISGF3; Darnell et al., 1994; Haque and Williams, 1994; Bluyssen et al., 1996) (Fig. 1). AAF/GAF is a homodimer of tyrosine-phosphorylated Stat1, whereas ISGF3 is a heterotrimeric complex of tyrosine-phosphorylated Stat1, Stat2 and another transcription factor member, IRF-9/p48/ISGF3γ (Bluyssen et al., 1996) (Fig. 1). There is a clear difference in the activation level of ISGF3 or GAF/AAF between type I and type II IFN signalling (Fig. 1; the thicknesses of arrows for ISGF3 and GAF/AAF represent the extents of their involvement). Type I IFNs more strongly activate the formation of ISGF3 than type II IFN, whereas type II IFN mainly activates GAF/AAF. These complexes translocate into the nucleus; subsequently, AAF and ISGF3 bind to their specific DNA sequences containing each of the common motifs, namely, the IFN-γ-activated sequence (GAS; Decker et al., 1991b; Lew et al., 1991) and the IFN-stimulated regulatory element (ISRE; Kessler et al., 1990; Williams, 1991) respectively. The IFN stimulation of promoters containing ISRE and GAS results in the transcriptional induction of a large number of target genes (IFN-stimulated genes; ISGs) to evoke versatile biological activities.
Recent identified signalling pathways downstream of type I IFN receptor
In addition to the well-established ISGF3 complex, type I IFNs have been shown to induce other types of Stat complex (Fig. 1), such as Stat3 or Stat5 homodimer and Stat1-Stat3 or Stat5-CrkL (Fish et al., 1999) (v-crk sarcoma virus CT10 oncogene homologue) heterodimers. Stat5 constitutively interacts with Tyk2 and undergoes phosphorylation in response to stimulation by IFN, and then CrkL binds to the tyrosine-phosphorylated Stat5 via its SH2 domain. This Stat5-CrkL heterodimeric complex translocates to the nucleus and binds to the GAS element (TTCTAGGAA) (Fish et al., 1999).
Several reports further indicate that type I IFNs can activate pathways that are distinct from the canonical Jak-Stat pathway. In response to type I IFN stimulation, two of the members of insulin receptor substrate (IRS)-1 and IRS-2 become tyrosine-phosphorylated, allowing for the N- and C-terminal SH2 domain-mediated binding of the p85 subunit of phosphatidylinositol (PI) 3-kinase (Uddin et al., 1995; Platanias et al., 1996; Burfoot et al., 1997) (Fig. 1). This binding results in the activation of the p110 catalytic subunit of PI3-kinase (Uddin et al., 1995). PI3-kinase activity is not required for the induction of the ISGF3 complex or any IFN-inducible genes, whereas the association of PI3-kinase with the IFNAR-1 subunit can occur in a Stat3-dependent manner (Pfeffer et al., 1997). Presumably PI3-kinase activation may be linked to an activation pathway of the proto-oncogene product Akt (Nguyen et al., 2001) and PKCδ (Uddin et al., 2002).
The overexpression of Stat3 in IFN-resistant B cell lymphoma cells showed that Stat3 is possibly involved in the induction of antiproliferative effects by type I IFNs (Yang et al., 1998), besides the well-characterized signalling pathways involving the activation of Stat1 and Stat2. However, this finding is contradicted by the observation that Stat3 appears to be also required for the activation of PI3-kinase to promote IFN-α/β-mediated cell survival, although the direct involvement of Stat3 was not shown (Yang et al., 2001). In this regard, there is evidence showing that PI3-kinase is activated by type I IFN in a Stat3-independent manner. Although the relationship between the Stat3-dependent pathway and PI3-kinase-dependent pathway has not been fully clarified, it has been shown that the PI3-kinase-dependent pathway can mediate both proapoptotic and survival activities in IFN-mediated signalling: IFN-α/β-activated PI3-kinase signalling mediates survival signals through the activation of its downstream effector, Akt or PKCδ, in a cell-type specific manner; e.g. primary astrocytes (Barca et al., 2003) or neutrophils (Wang et al., 2003b). By contrast, the activation of PI3-kinase and its downstream molecule, mammalian target of rapamycin (mTOR) serine-threonine protein kinase, is also essentially involved in the IFN-α-induced apoptosis of multiple myeloma cells (Thyrell et al., 2004) Thus, IFN-activated PI3-kinase signalling seems to regulate apoptotic responses in a context-dependent manner.
It has been also reported that the mitogen-activated protein (MAP) kinases, extracellular-signal-regulated kinase 2 (ERK2) and p38, are activated by IFN-α/β (David et al., 1995; Goh et al., 1999) (Fig. 1). In addition, it was shown that the inhibition of p38 activity by SB203580, a pharmacological inhibitor of p38, blocks IFN-α-induced gene transcription via GAS elements as well as ISRE elements (Uddin et al., 1999; 2000a). On the other hand, the tyrosine phosphorylation of Stat1 or Stat2 is not affected by the inhibition of p38 activity, nor the serine phosphorylation of Stat1 or Stat3 (Uddin et al., 2000a). These results indicate that the suppression of the type I IFN-mediated gene transcription is independent of Stat1 activation (Uddin et al., 1999). Further studies have revealed that p38 is activated in a type I IFN-dependent manner in breakpoint cluster region-v-abl Abelson murine leukaemia viral oncogene homologue 1 (BCR-ABL)-expressing cells and in cells from patients with chronic myelogenous leukaemia (CML) (Mayer et al., 2001). In addition, p38 inhibitors abrogated the suppressive effects of type I IFN on leukaemic progenitors from the bone marrow of CML patients. These findings suggest that p38-mediated signalling is essential for the antileukaemic effects of type I IFN (Mayer et al., 2001).
Amplifying effect of type I IFN signalling on IFN-α/β gene induction by viral infection
Prior to going into the details of the regulatory function of IFN-α/β signalling in their induction by viral infection, we would like to describe the current view of the mechanism underlying IFN-α/β gene induction in conjunction with PRRs and the IFN-regulatory factor (IRF) system.
The induction of IFN-α/β genes upon viral infection is transcriptionally controlled. A gene disruption study (Sato et al., 2000) revealed that among nine hitherto identified members (IRF-1 to IRF-9) of the IRF family, IRF-3 and IRF-7 are essential transcriptional factors for IFN-α/β gene induction by viruses. Consistently, no induction of IFN-α/β mRNA was detected in IRF-3- and IRF-7-doubly deficient mice (Honda et al., 2005a). In fibroblasts or conventional DCs (Fig. 2), infection by viruses such as Newcastle disease virus (NDV), Sendai virus (SeV) and vesicular stomatitis virus (VSV), is sensed by RIG-I and Mda5, which results in the stimulation of IPS-1/MAVS/VISA/Cardif-mediated downstream signallings, partly leading to the activation of IRF kinases [TANK-binding kinase 1 (TBK1) and IκB kinase ∈/i (IKK∈/i)]. These activated kinases in turn phosphorylate the serine/threonine residues of IRF-3 and IRF-7 at their carboxyl terminal region to become an active form. The activated IRF-3 and IRF-7 undergo nuclear translocation, and subsequently bind to IRF-binding elements (IRF-Es) [i.e. positive regulatory domains (PRDs) I and III, and PRD-like elements (PRD-LEs)] in the IFN-α/β promoter.
This positive feedback regulation depends on the IFN-inducible expression of IRF-7 in an ISGF3-dependent manner (Marie et al., 1998; Sato et al., 1998). Due to the unique expression profile of IRF-7, which is expressed by the initial induction of IFN-α/β, it augments its intracellular level, leading to the amplification of IFN-α/β gene induction (Fig. 2, right panel). Therefore, massive IFN-α/β production can be achieved through this positive feedback mechanism, whereby the signalling triggered by de novo synthesized IFN-α/β amplifies IFN-α/β production through the increase in the intracellular expression level of IRF-7 (Marie et al., 1998; Sato et al., 1998; 2000).
It is of interest to determine whether this positive feedback mechanism is observed in pDCs, a specialized subset of DCs, which is characterized by the capability of producing large amounts of IFN-α/β upon viral infection (Asselin-Paturel and Trinchieri, 2005; Liu, 2005). IFN induction in pDCs is dependent on MyD88 and IRF-7, but not IRF-3, which is different from that in fibroblasts or conventional DCs (both IRF-7- and IRF-3-dependent, but MyD88-independent) (Honda et al., 2005a). Recent data demonstrated the existence of a similar positive feedback mechanism in pDCs (Honda et al., 2005a). In fact, robust production of IFN-α in response to stimulation by CpG-A, a TLR9 ligand, is abolished in IFNAR-1-deficient or IRF-9-deficient pDCs, both of which are defective in the IFN-signal-dependent gene induction of IRF-7. This positive feedback mechanism is also regarded as an essential regulation for a robust IFN induction in pDCs, similarly in fibroblasts. In relation to this finding, our recent study demonstrated a unique spatiotemporal regulation of the signalling process for high-levels IFN induction in pDCs (Honda et al., 2005b).
Regulatory roles of weak signalling by spontaneously produced IFN-α/β
Evidence has been provided that this weak signalling by the constitutively produced IFN-α/β is critical for eliciting strong responses of cells to IFN-γ and interleukin (IL)-6 (Takaoka et al., 2000; Mitani et al., 2001; Fig. 3). In fact, the IFN-γ-induced DNA-binding activity of activated Stat1 was found to be severely diminished in cells from mice deficient in IFNAR-1 (Takaoka et al., 2000). Similarly, the full activation of Stat1and Stat3 by IL-6 did not occur in this mutant cells (Mitani et al., 2001). Further analyses revealed that this weak IFN-α/β signalling is critical for maintaining the intracellular tyrosine residues of IFNAR-1 in a phosphorylated form so as to provide niches where these Stats can dimerize efficiently upon stimulation by IFN-α/β or IL-6 (Takaoka et al., 2000; Mitani et al., 2001).
In relation to the above-mentioned positive feedback system for type I IFN induction, a constitutive, weak IFN-α/β signalling was also found to contribute to the mechanism underlying the amplification of type I IFN induction by viral infection. Our recent study using IRF-7-deficient mice revealed that the viral induction of IFN-α/β genes is severely impaired in IRF-7-deficient MEFs (Honda et al., 2005a), which indicates that IRF-7 plays a major role in the induction of IFN-α/β genes upon viral infection. As supported by this finding, the expression level of IRF-7 before infection is shown to be essential for activating the positive feedback loop (Fig. 2, right panel) for the efficient induction of IFN-α/β genes by viruses. In splenocytes, which express a relatively high level of IRF-7 even before infection, this induction mechanism is more effective than in fibroblasts with much lower levels of IRF-7 expression (Hata et al., 2001). Interestingly, the expression level of IRF-7 was found to correlate with that of constitutive IFN-α/β (Hata et al., 2001). Therefore, prior to viral invasion, the constitutive, weak IFN-α/β signalling renders cells in a state of ‘revving up’ for the robust and efficient production of IFN-α/β upon viral infection (Taniguchi and Takaoka, 2001; 2002; Takaoka and Taniguchi, 2003).
Furthermore, there is also an observation that the expression levels of IFN-α/β only marginally increase upon the stimulation of CD8+ T cells by a mixed lymphocyte reaction (Ogasawara et al., 2002), suggesting that this weak IFN-α/β signalling may also plays a role in the regulation of adaptive immune responses. The level of IFN-α/β mRNA expression induced by TCR stimulation is less than 1/200 that of induction by viral infection (Buller et al., 1987). It can be speculated that the increase in local IFN-α/β concentration at the site of T cell activation, by the concomitant, massive production of IFN-α/β in response to viral infection (Hou et al., 1995; Ridge et al., 1998), may contribute to a more efficient operation of this mechanism for the activation of TCR signalling in CD8+ T cells. Consistent with this hypothesis, CD8+ T cells lacking IFNAR-1 cannot respond efficiently to antigen stimulation, and the exogenous addition of recombinant IFN-β markedly enhances the proliferation of CD8+ T cells (Ogasawara et al., 2002). These observations additionally indicate the regulatory role of weak IFN-α/β signalling in the efficient activation of CD8+ T cell upon TCR engagement.
In this context, a study with IRF-2-deficient mice demonstrated a possible outcome of a dysregulated, weak IFN-α/β signalling. IRF-2 is constitutively expressed in most cell types, and functions as an attenuator at the ISRE to suppress ISGF3-mediated signalling (Taniguchi et al., 2001; Taki, 2002). Therefore, one can postulate that IRF-2 is a negative regulator of the constitutive, weak IFN-α/β signalling. IRF-2-deficient mice were found to spontaneously develop inflammatory skin lesions accompanied by a marked upregulation of IFN-inducible genes (Hida et al., 2000). The polyclonal activation of CD8+ T cells seems to be involved in the pathogenesis of this condition (Hida et al., 2000). Furthermore, both the skin lesions and the hyperresponsiveness of CD8+ T cells are suppressed in IRF-2- and IFNAR-1-doubly deficient mice. These results suggest the importance of the IRF-2-mediated attenuating regulation of the constitutive, weak IFN-α/β signalling (Matsuyama et al., 1993). Intriguingly, additional abnormalities, such as the suppression of haematopoiesis and predisposition to pancreatitis, are also observed in IRF-2 null mice (unpublished observation).
Therefore, a weak signal by constitutively produced IFN-α/β, renders cells ‘ready-to-go’ for the enhancement of cellular responses to rapid environmental changes, such as viral infection. Analogous to a car engine revving up for a thrust start and acceleration, we figuratively termed this machinery the ‘revving-up’ system. The ‘revving up’ of cellular engines by a weak signal is implicated in the elicitation of robust cellular responses against infections (Fig. 3; Taniguchi and Takaoka, 2001; 2002; Takaoka and Taniguchi, 2003). In this regard, this ‘revving-up’ system could be considered as a unique regulatory mechanism for amplifying cellular responsiveness in host defence.
We also reported that the IFN-β gene is induced in bone marrow macrophages upon stimulation by the RANK ligand (receptor activator of NF-κB ligand), a negative regulator of osteoclast differentiation, and that this induction is not dependent on IRF-3 or IRF-7, but on RANKL-induced c-Fos, one of the essential transcriptional factors for osteoclast differentiation (Takayanagi et al., 2002). In addition, the constitutive, weak IFN-α/β signalling has been shown to negatively regulate the differentiation of CD8α–DCs from fms-like tyrosine kinase ligand (Flt3L)-stimulated bone marrow cells (Honda et al., 2004a). Consistently, the differentiation of this DC population is impaired in IRF-2-deficient DCs, whereas this abnormality is rescued in IRF-2- and IFNAR-1-doubly deficient DCs. Thus, the weak signalling by constitutive type I IFN exhibits its regulatory role in not only host defence but also in other cellular processes such as cell differentiation.
Viral evasion from IFN system
Viruses evolutionarily have acquired their own mechanisms of evading host immune responses against themselves. Viruses frequently evolve various strategies to abolish the activities of key components of the IFN system, such as the signalling molecules of the Jak-Stat pathway, or PRR-mediated pathway for IFN production (Weber et al., 2004; Johnson and Gale, 2005). In this section, we will briefly explore potential mechanisms by which viruses evade antiviral defence systems, on the basis of previous reports about viral strategies counteracting two major aspects of the IFN system: (i) IFN production and (ii) IFN signalling. Because there have been many excellent reviews published on this subject (Goodbourn et al., 2000; Levy and Garcia-Sastre, 2001; Garcia-Sastre, 2002; Katze et al., 2002), we focus on several major viruses, hepatitis C virus (HCV), Paramyxoviruses and vaccinia virus (VV) (Fig. 2).
Hepatitis C virus
i. Recently, much attention has been focused on an HCV serine protease, NS3/4A, which cleaves IPS-1/VISA/MAVS/Cardif, an adaptor protein, leading to the inactivation of RIG-I- or Mda5-mediated signalling pathways including both NF-κB and IRF-3 pathways (Meylan et al., 2005) (Fig. 2). In addition, there is a report documenting that this NS3/4A protease also causes the proteolysis of TRIF (Fig. 2), which is the critical adaptor protein linking TLR3 to its downstream NF-κB and IRF-3 pathways for the double-stranded RNA (ds-RNA) response (Li et al., 2005).
ii. HCV NS5A and E2 are shown to interact with PKR, one of the IFN-inducible antiviral proteins. This interaction results in the inactivation of this kinase (He and Katze, 2002) (Fig. 2). NS5A also directly targets IFN signalling pathway by disrupting the crosstalk between the MAP kinase and Jak-Stat pathways (He and Katze, 2002).
Hepatitis C virus also exploits other strategies such as continual viral genetic variation to evade host responses. A complex combination of these evasion processes may contribute to the persistence of HCV infection (Gale and Foy, 2005).
It is well documented that most members of the Paramyxoviridae subfamily, such as simian virus 5 (SV5), NDV, SeV, measles virus, mumps virus and parainfluenza virus, circumvent the IFN system by abrogating both processes; IFN production and IFN signalling.
i. The C-terminal domain of the V protein of most paramyxoviruses binds selectively to Mda5, thereby blocking the activation of its downstream pathway for IFN production (Andrejeva et al., 2004; Yoneyama et al., 2005) (Fig. 2). There is no evidence that the V protein interacts with RIG-I, which is another caspase recruitment domain (CARD)-containing recognition receptor of intracellular ds-RNA (Yoneyama et al., 2004).
ii. The V protein of SV5 inhibits both type I and type II IFN signallings through the proteasomal degradation of Stat1 (Didcock et al., 1999) (Fig. 2). Furthermore, the V protein of other paramyxoviruses has been found to target other Stat proteins as well as Stat1 to inhibit Stat-mediated signallings through diverse mechanisms including proteasomal degradation, sequestration and blockade of nuclear translocation (Horvath, 2004) (Fig. 2).
Vaccinia virus, a member of the Poxviridae, has been shown to evolve multiple evasion strategies against the IFN system (Haga and Bowie, 2005).
i. It has been shown that A52R associates with both IRAK2 and TRAF6 to block the NF-κB activation pathway by various TLRs such as TLR4 (Bowie et al., 2000), whereas another VV protein, A46R, which contains a TIR domain, can target TLR adaptors such as MyD88, TRIF and TRAM, thereby inhibiting both MyD88- and TRIF-dependent pathways (Stack et al., 2005). Furthermore, Stack et al. showed that upon TLR3 activation, A52R more potently inhibits the activation of the NF-κB pathway than A46R, whereas A46R exhibits a more prominent inhibition of the IRF-3 activation pathway (Stack et al., 2005) (Fig. 2).
ii. It has been reported that poxviruses interfere with signallings by key cytokines that function in host immune systems, by expressing a variety of viral proteins such as soluble forms of receptors of cytokines, tumour necrosis factor, interleukin-1β, chemokines and IFNs (Alcami and Smith, 1996a; Haig, 1998). One of the VV proteins, vIFN-γR (viral IFN-γ receptor), which in the VV strain Western Reserve is encoded by the viral gene B8R (Alcami and Smith, 1995), binds to IFN-γ, thereby interfering with the binding of IFN-γ to the cellular receptor, and neutralizing the ligand activity. IFN-γ plays an important role in host defence against poxvirus infections, as demonstrated by a number of reports (Karupiah et al., 1990; 1993a,b; Kohonen-Corish et al., 1990; Huang et al., 1993; Melkova and Esteban, 1994; Harris et al., 1995), including a gene knockout analysis showing that IFNGR-1-deficient mice exhibit a severe susceptibility to VV (Huang et al., 1993). These observations may provide a plausible explanation for the viral strategy targeting the host IFN-γ system by expressing a soluble homologue of the cellular IFN-γ receptor. Furthermore, the expression of such a soluble IFN-γ receptor homologue and the encoding gene are highly conserved among most of other poxviruses, including the cowpox, ectromelia, variola, myxoma and Shope fibroma viruses (Alcami and Smith, 1996b), and their IFN-γ binding activity is presumably attributable to the sequence similarity of the viral soluble receptor to the extracellular binding domain of the IFN-γ receptor although these viral receptors show a ligand-binding activity in a species-specific manner (Alcami and Smith, 1995; 1996b). In this respect, because IFN-γ shows the activity of species-specific binding to its receptor (Pestka et al., 1987), it is also speculated that the ligand specificity of poxvirus-encoded IFN-γ receptors may reflect the spectrum of the host(s) where the virus has evolved (Alcami and Smith, 1996b). Consistently, the first described poxvirus IFN-γ receptor expressed by the myxoma virus, which causes myxomatosis specifically in rabbits, binds to rabbit IFN-γ but not mouse or human IFN-γ (Upton et al., 1992).
On the other hand, the type I IFN system is important for protection against poxvirus infection, which is supported by several studies showing that pretreatment with IFN-α/β abrogated the in vivo replication of the VV (Rodriguez et al., 1991) and IFNAR-1-deficient mice are more susceptible to VV infection (Muller et al., 1994). Therefore, it is not surprising that the type I IFN system is also targeted by poxviruses; the vIFN-α/β-BP, a VV protein encoded by B18R, binds to IFN-α/β with a broad species specificity, and prevents the binding of IFN-α/β to cellular receptors, consequently leading to the inhibition of the type I IFN response (Colamonici et al., 1995; Symons et al., 1995) (Fig. 2). Although the role of this type I IFN-binding protein in natural infection remains to be clarified, it was found to be expressed by other orthopoxviruses such as the cowpox, ectromelia and camelpox viruses (Colamonici et al., 1995; Symons et al., 1995). A mechanism underlying the binding activity of the B18R protein is not clear: Colamonici et al. showed that despite showing a very limited sequence similarity, the B18R protein has significant regions of homology with mouse, human and bovine IFNAR-1 subunits, and shows binding and neutralizing activities against several recombinant human type I IFNs (Colamonici et al., 1995). On the other hand, Symons et al. demonstrated that the B18R protein has three immunoglobulin-like domains, belonging to the immunoglobulin superfamily, which is different from cellular type I IFN receptors containing fibronectin type III domains (Symons et al., 1995).
The cellular type I IFN-receptor binds to only the N-terminal region of its ligand, which is proposed to determine a strict species specificity of IFN-binding (Liptakova et al., 1997), whereas the B18R protein was shown to interact with type I IFNs through their C-terminal region as well as their N-terminal region (Liptakova et al., 1997). This distinct feature of the B18R protein may explain a possible mechanism for its broad species specificity (Liptakova et al., 1997).
In addition, the VV E3L gene product codes for ds-RNA binding proteins, thus resulting in the inhibition of the activation of PKR (Fig. 2) and OAS (2′, 5′-oligoadenylate synthetase), both of which require ds-RNA for their activation (Chang et al., 1992). On the other hand, the K3L gene product acts as a decoy of eukaryotic initiation factor (eIF)-2α to disrupt the interaction of eIF-2α with PKR (Fig. 2). Thus, K3L competitively inhibits eIF-2α phosphorylation as well as the autophosphorylation of PKR (Davies et al., 1992) (Fig. 2). In conjunction with the soluble viral type I and type II IFN receptors, E3L and K3L contribute to viral resistance to IFN signalling.
Conclusion and future prospects
Since the first discovery of the cardinal Jak-Stat pathway in the IFN system, numerous studies have contributed to not only the elucidation of involvement of other signalling molecules in the positive or negative regulation of the Jak-Stat pathway, but also the identification of non-Stat pathways. Although there is accumulating evidence regarding the important role of the cooperation of several signalling pathways for IFN activities, the underlying molecular mechanism is not fully understood.
On the other hand, as described above, viral infection triggers the gene induction of many members of the type I IFN family, all of which activate the type I IFN receptor complex in a distinct manner on the basis of the activation property of each ligand. Among these ligands, it was reported that human IFN-α subtypes contain 12 distinct proteins (Pestka, 2000; Pestka et al., 2004). It is of note that they seem to show different relative activities although they similarly utilize the identical receptor complex. In this context, interesting data obtained by Pestka’s group in their studies with recombinant IFN-αs show that there is a marked difference (by more than 10 000-fold) in increasing NK activity among the IFN-α subtypes (Ortaldo et al., 1984). IFN-αJ, which is a recombinant protein corresponding to IFN-α7, exhibits virtually no capability of boosting NK activity, and can even act as an antagonist to the effects exerted by other IFN-α subtypes, whereas it has potent antiviral and antiproliferative activities (Ortaldo et al., 1984). In addition to quantitative differences between IFN-α1, -α2 and -α21 in their ability to activate Stat1–5 and ISGs, IFN-γ-inducible protein-10 (IP-10) is highly induced by IFN-α2 and -α21 but to a much lower extent by IFN-α1 in DCs (Hilkens et al., 2003).
These differences in biological activity profiles among the IFN-α subtypes may be caused by several different properties in the receptor-ligand interaction or affinity, which may result in different activations of intracellular signalling pathways in terms of intensity, quality and duration. As Pestka et al. suggested the importance of structure determination of more IFN subtypes to accurately define the molecular basis for their respective activities (Pestka et al., 2004), such analyses regarding extracellular events in relation to ligand properties and ligand–receptor interaction will be required for further clarification of regulatory mechanisms for diversified IFN responses, which could provide deep insights into the physiological roles of type I IFNs in various aspects of host defence systems. In addition, more integral studies using genomics and proteomics will be required to elucidate the molecular mechanism underlying the co-ordination and cooperation of the complex crosstalk networks of intracellular signalling pathways, which underlies diversity of biological activities induced by IFNs.
Interferon biology particularly has an important impact on the clinical field, where IFNs have been so far used in various clinical settings, including viral infections, malignant tumours, immunodeficiency and autoimmune diseases (Liang et al., 2000; Malik and Lee, 2000; Parmar and Platanias, 2003). Further comprehensive understanding of IFN-mediated signalling from both intracellular and extracellular mechanistic points of view may provide some clues to the development of new agents and efficient therapeutic strategies.
We would like to thank Prof. T. Taniguchi and our colleagues for their continuous support to the work from our laboratory described in this review, and the work in our laboratory was supported by Grant-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Culture, Sports, Science and Technology of Japan.