Another detour on the Toll road to the interferon antiviral response
John Hiscott
The author is at the Lady Davis Institute for Medical Research-Jewish General Hospital, Departments of Microbiology & Immunology, Medicine and Oncology, McGill University, Montreal, Canada, H3T 1E2. john.hiscott@Mcgill.ca
Recent characterization of two distinct signals leading to IRF-3 activation in response to double-stranded RNA recognition by Toll-like receptor 3 provides new mechanistic information on the regulatory events that link the detection of viral invasion by the cell with the development of the antiviral response.
Infection by viral or bacterial pathogens is one of the great stresses in the life of a cell. How the invading pathogen is initially recognized and how the host cell responds within the first minutes to hours can dictate the subsequent development of immune protection, with far-reaching implications for the pathogenesis of infectious, immune and allergic diseases. The innate immune response—a first line of defense against viral pathogens—is active long before adaptive responses such as neutralizing antibodies or cytotoxic T lymphocytes. An integral component of innate antiviral defense is the production of type 1 interferons (IFN), a family of antiviral cytokines composed of IFN- and several IFN- species1,
2. These proteins induce a cascade of events through the activation of signaling mediated by the JAK-STAT pathway, resulting in the production of hundreds of proteins that function to limit viral replication and signal-adaptive immune responses3.
The transcription factor interferon regulatory factor 3 (IRF-3) is a key player in this initial triggering of interferon gene transcription. After virus infection, latent IRF-3 is phosphorylated on clustered, C-terminal serine residues, leading to protein dimer-ization, translocation to the nucleus, association with coactivator molecules, binding to discrete DNA elements in the IFN- and/or IFN- promoters and the activation of gene transcription4,
5,
6,
7. In the case of IFN-, activation is accomplished in synergy with NF-B and AP-1 family members8. The absence of IRF-3 or the closely related IRF-7 in murine knockout models results in markedly reduced IFN production in response to virus infection9, thus supporting a pivotal role for the activation of IRF-3 and IRF-7 in the development of the cellular antiviral response.
On page 1060 of this issue of Nature Structural & Molecular Biology, Sarkar et al.10 provide convincing evidence that, in response to double-stranded RNA (dsRNA)—long considered a product of virus replication and an early sentinel of virus infection—tyrosine phosphorylation of Toll-like receptor 3 (TLR-3) is required for dsRNA-dependent signaling, leading to IRF-3 activation. Recognition of dsRNA by TLR-3 results in the phosphorylation of two specific tyrosines (Tyr759 and Tyr858) within the cytoplasmic tail of TLR-3, recruitment of phosphatidyl-inositol-3 kinase (PI3K) to the receptor and the initiation of two distinct signaling pathways, mediated by PI3K and the noncanonical IKK-related kinases TANK-binding kinase (TBK-1) and IKK- (Fig. 1). Specifically, dsRNA activation of interferon-stimulated gene 56 (ISG56) was blocked under conditions that interfered with the PI3K pathway: treatment with inhibitors, expression of a catalytically inactive p110 subunit of PI3K or the use of a dominant-negative version of the downstream kinase Akt. Specific tyrosine point mutations abolished PI3K recruitment, indicating that the TLR-3−PI3K interaction was dependent on receptor phosphorylation.
Figure 1. Double-stranded RNA (dsRNA) and virus-mediated signals leading to interferon gene activation.
Engagement of TLR3 by dsRNA leads to recruitment of the TRIF adaptor molecule and the activation of TBK-1 and IKK- kinases that phosphorylate IRF-3 and IRF-7 transcription factors. TRIF can also mediate signals for NF-B activation, through a TRAF-6− or RIP-1−dependent mechanism, via the IKK-/ complex that phosphorylates the inhibitory subunit IB, resulting in release of NF-B DNA-binding subunits. After phosphorylation by IKK, IB is degraded (striped yellow oval). Recruitment of dsRNA also causes tyrosine phosphorylation of the cytoplasmic tail of TLRs at multiple residues and the recruitment of PI3 kinase. In the absence of PI3K activation, IRF-3 is incompletely phosphorylated and fails to stimulate IFN gene transcription. It remains to be determined at which level PI3K-Akt functions. Virus infection represents a distinct stress to the cell that may use TLR-3 independent mechanisms. RIG-I has been shown to stimulate NF-B− and IRF-dependent pathways and may be the sensor molecule that initially recognizes incoming cytoplasmic viral ribonucleoprotein complexes.
The failure to activate PI3K in cells expressing TLR-3 mutants resulted in partial IRF-3 activation—phosphorylation, dimerization and nuclear translocation still occurred, but IRF-3 failed to associate with CBP coactivator or induce transcription. Despite incomplete activation, IRF-3 was phosphorylated on at least one of its important C-terminal residues, Ser396. Subsequent two-dimensional analysis revealed multiple IRF-3−phosphorylated forms in dsRNA-treated cells as compared with untreated cells. In contrast, dsRNA-activated, nuclear IRF-3 isolated from cells treated with a PI3K inhibitor or cells expressing the TLR-3 Y759F mutant had intermediate isoelectric points, demonstrating that incomplete activation was accompanied by incomplete phosphorylation. Together, the results emphasize a crucial role for the PI3K-Akt pathway in the functional activation of IRF-3 and downstream interferon antiviral immunity, following TLR-3 engagement by dsRNA.
The study by Sarkar et al.10 clarifies a long-standing paradox in the IFN signaling field. Weaver et al.6 originally recognized that inhibitors of tyrosine phosphorylation blocked dsRNA-dependent signaling and activation of IRF-3, even though the actual targets on IRF-3 were serine residues. The demonstration that tyrosine phosphorylation of cytoplasmic domain of TLR-3 is necessary for recruitment of PI3K to TLR-3 provides a probable explanation for the requirement for tyrosine phosphorylation.
This study comes at an important juncture in the field of virus-host interactions. The questions at the heart of this research—what are the molecular mechanisms that 'sense' incoming virus and how are these sensors linked to the immediate antiviral response—have been the subject of intensive research. Both TLR-dependent and TLR-independent mechanisms of IFN gene activation have been documented11. Viral proteins and byproducts of the virus life cycle are recognized by evolutionarily conserved TLRs that trigger immune activation (reviewed in ref. 12). Of the 11 mammalian TLRs identified, TLR-2, TLR-3, TLR-4, TLR-7/8 and TLR-9 have been implicated in the response to different components of viruses. TLRs recognize biochemically conserved molecules that are exclusive to foreign pathogens. Collectively called pathogen-associated molecular patterns (PAMPs), Gram-negative bacterial lipolysaccharide, bacterial flagellin and viral single- or double-stranded RNA represent some of the ligands that initiate responses via TLR signaling. The complexity of TLR signaling has been comprehensively reviewed by Akira and Takeda13.
With respect to dsRNA-mediated signal-ing through TLR-3, it is known that TBK-1 and IKK- directly phosphorylate IRF-3 (refs. 14,15). These kinases are activated as a component of a MyD88-independent pathway regulated by the TLR-associated adaptor molecule TRIF/TICAM-1 (ref. 15). The TRIF adaptor directly interacts with TBK1. Notably, the N-terminal domain of TRIF that associates with TBK-1 is in close proximity to the TRAF6-interaction domain, a region of TRIF required to activate the classical IKK−NF-B signaling downstream of dsRNA−TLR-3 engagement16,
17. Thus, MyD88-independent signaling seems to bifurcate into two distinct pathways downstream of the TRIF/TICAM-1 adaptor: a TRAF6-dependent cascade that leads to IKK-/ activation and transcriptional upregulation of pro-inflammatory genes such as IL-6, IL-1 and TNF, as well as a TRAF-6−independent TBK-1/IKK- cascade that leads directly to IRF activation and production of TLR-3-specific genes such as the type I IFN, RANTES and IP-10 cytokines (Fig. 1). In light of the important role for TRIF in dsRNA−TLR-3 signaling, Sarkar et al.10 demonstrated that overexpression of the TRIF adaptor was able to overcome blockade of the PI3K pathway, suggesting that the activity of TRIF was either below or independent of PI3K activity.
Despite rapid progress in understanding the molecular details of TLR signaling, several observations challenge the idea that detection of virus occurs exclusively through TLR recognition of viral PAMPs11. Although TLR-3 engagement is capable of stimulating IFN production, gene-targeting experiments have demonstrated that TLR-3 and TRIF are not required for IFN production in virally infected cells13. Another complexity is raised by the subcellular localization and restricted tissue distribution of distinct TLRs. Some TLRs are transmembrane proteins whereas others exist in intracellular vesicles13. The ligand recognition moiety is topologically extracellular, whereas viruses often replicate in the cytoplasm and dsRNA structures may not be available for TLR binding. In fact, whether free dsRNA ever exists inside cells during virus infection is also unclear, given that the viral nucleocapsid proteins tightly wrap viral RNA into ribonucleoprotein complexes that limit intracellular accumulation of viral RNA. TLRs may be relevant for the systemic response to virus infection, but do not seem to provide an explanation for the immediate recognition of viral infection that induces the first wave of IFN production11.
The argument in favor of a TLR-independent recognition pathway received a strong boost with the discovery of a retinoic acid−inducible gene, RIG-I, as an essential component of viral and dsRNA detection18. The RIG-I protein contains an RNA helicase domain and a caspase recruitment domain (CARD). The CARD domain alone is capable of stimulating IRF-3, NF-B and IFN- production, whereas the helicase activity seems to function in a regulatory capacity. RIG-I has attractive properties as a 'sensing' molecule for cytoplasmic viruses. For example, interaction of the helicase domain with viral RNA may induce a conformational change in RIG-I and promote protein-protein interactions between the CARD domain and other adaptor proteins that lead to downstream activation of TBK and IKK- kinases. The relationship between the RIG-I adaptor-mediated response and TLR-3/TRIF-mediated signaling, as well as how these two independent pathways converge on IRF-3 phosphorylation and IFN activation, remains to be clarified18 (Fig. 1).
Any model of IRF-3 activation involving PI3K—whether TLR3-dependent or not—must consider the relationship to the TBK-1/IKK- kinases that phosphorylate the C-terminal serines in IRF-3 and IRF-7 in response to virus or dsRNA14,
15. It is not yet clear whether PI3K phosphorylates additional IRF-3 residues directly or rather influences the activity of TBK-1/IKK- (Fig. 1). The two-step model proposed by Sarkar et al.10 asserts that TBK-1 phosphorylates Ser396 and potentially other serines in the cluster of residues 396−405, while PI3K may phosphorylate—directly or indirectly—two essential residues at Ser385 and Ser386. Ultimately the above new physiological information must be interpreted in light of recent X-ray crystallography of the C-terminal domain of IRF-3 (refs. 19,20). In this structure, the N- and C-terminal portions of the peptide form -helical bundles that flank a -sandwich core and bury several key residues. Qin et al.19 interpreted the structure in light of a autoinhibitory mechanism in which virus-inducible, C-terminal phosphorylation is expected to abolish auto-inhibitory interactions by introducing charge repulsions that unmask the active site and realign the DNA-binding domain to form the transcriptionally active protein. In this scenario, multiple serine/threonine phosphorylation events within two serine clusters are required for relief of latent autoinhibition19. In contrast, Takahashi et al.20 concluded that phosphorylation of the C terminus directly precipitated IRF-3 dimerization, and generated an acidic pocket responsible for p300/CBP association. In this case, the serines at positions 385 and 386 are the initial targets for virus-inducible, as well as IKK-− and TBK-1− mediated, IRF-3 phosphorylation, because these residues are the most accessible amino acids. Future studies will reveal the exact mechanism of IRF-3 activation.
Recognition of dsRNA and signaling via TLR-3 do not necessarily reflect virus infection. It is clear that dsRNA-mediated signaling is distinct from virus-mediated activation of the IFN response and that only a subset of genes induced by virus infection are also activated by dsRNA treatment. In fact, IRF-3 induction and IFN gene expression occur in virus-infected cells in the presence of PI3 kinase inhibitors and in cells lacking TLR-3 or the TRIF/TICAM-1 adaptor, indicating that the host response to virus infection must in some way bypass the requirement for PI3K or intersect the pathway downstream of the molecular interactions described in this study. Although dsRNA-induced TLR-3 signaling via TBK-1 and IKK- represents a fundamental mechanism for IRF-3 activation, TLR-3 engagement by dsRNA is not the sole trigger for TBK-1/IKK- activation. The viral ribo-nucleoprotein (RNP) complex is an attractive candidate as the viral structural component that could trigger TBK-1 activation, possibly through a functional association with RIG-I (Fig. 1). RNP complexes isolated from whole vesicular stomatitis virus (VSV) were shown recently to activate TBK-1 and induce IRF-3 phosphorylation with kinetics similar to those observed with VSV infection, but distinct from the kinetics of dsRNA activa-tion21. Undoubtedly, future investigations will unravel the signals and signposts on the road to an antiviral immune response.
and several IFN-
species
B and AP-1 family members
(
