Introduction
Toll-like receptors (TLRs) sense the invasion of microbes in the body by recognizing their structural components and activate intracellular signaling pathways leading to the expression of genes responsible for inflammatory and immune responses1, 2, 3. Studies have identified specific components detected by each TLR, such as diacyl lipoprotein (TLR6 and TLR2), triacyl lipoprotein (TLR1 and TLR2), double-stranded RNA (TLR3), lipopolysaccharide (LPS; TLR4), flagellin (TLR5), single-stranded RNA (TLR7 and TLR8) and CpG DNA (TLR9). Stimulation with TLR ligands induces the formation of homo- or heterodimers of TLRs for the recruitment of adaptor molecules containing the Toll–interleukin 1 receptor (IL-1R) homology domain (TIR domain). MyD88, an adaptor containing a TIR domain and a death domain, is essential for signaling 'downstream' of various TLRs, except TLR3 (refs. 4,5). MyD88 interacts with members of the IL-1R-associated kinase (IRAK) family, which dissociate from MyD88 and interact with tumor necrosis factor (TNF) receptor–associated factor 6 (TRAF6). TRAF6 acts as an ubiquitin protein ligase to catalyze the formation of a Lys63-linked polyubiquitin chain on TRAF6 itself and on the transcription factor NF-
B modulator NEMO. Transforming growth factor-
–activated kinase 1 (TAK1) is also recruited to TRAF6 and then phosphorylates inhibitor of NF-B (IB) kinase- (IKK) and mitogen-activated protein (MAP) kinase kinase 6. Subsequently, the IKK complex, composed of IKK
, IKK and NEMO, is formed. NF-B binds to IB in resting cells, where it is sequestered in the cytoplasm. Phosphorylation of IB by the IKK complex leads to degradation of IB by the ubiquitin-proteasome system, freeing NF-B to translocate into the nucleus, where it activates the expression of genes encoding proinflammatory cytokines. Activation of the MAP kinase cascade is responsible for gene expression induced by the transcription activator AP-1.
There are three additional TIR domain–containing adaptors involved in TLR signaling. TIRAP (also called Mal) has been shown to be critical for TLR2- and TLR4-induced NF-B activation6, 7, 8, 9. TLR3 and TLR4 trigger the signaling cascade leading to the production of type I interferons through another adaptor, TRIF10, 11, 12. The fourth adaptor, TRAM, bridges TLR4 and TRIF13 and is critical for TLR4-induced interferon-inducible gene expression14, 15.
The IRAK family has four members: IRAK1 (A001277), IRAK2 (A001278), IRAKM and IRAK4 (A003450)16. IRAK family members are composed of an amino-terminal death domain and a serine-threonine kinase domain. IRAK4 is known to be essential for TLR–IL-1R-mediated cellular responses17, 18, 19. The kinase activity of IRAK4 is essential for its function in TLR-induced cytokine production20, 21, 22. IRAK4 deficiency in humans also leads to impaired TLR responses together with recurrent infections with pyrogenic bacteria, particularly Streptococcus pneumoniae23, 24, 25, 26. In contrast, cells lacking IRAKM showed increased cytokine production in response to TLR stimuli, and IRAKM functions to inhibit formation of the IRAK1-TRAF6 complex27. IRAK1, the first member of the IRAK family to be discovered, is also involved in TLR–IL-1R signaling pathways28. IRAK1-deficient mice are more resistant to LPS-induced shock than are wild-type control mice29, 30, although Irak4–/– mice and Myd88–/– mice are more resistant to LPS inoculation5, 17, 22. IRAK1-deficient fibroblasts and macrophages reportedly show impaired cytokine production in response to IL-1 or LPS4, 17. However, IRAK1-deficient cells are still able to produce cytokines after TLR or IL-1R stimulation, and it has been hypothesized that other IRAK family members or other proteins could take over the function of IRAK1. Instead, given that IRAK1 is critical for the actions of TLR9-mediated type I interferons but not for the actions of proinflammatory cytokines in plasmacytoid dendritic cells, it is most likely that the main contribution of IRAK1 to the TLR responses of plasmacytoid dendritic cells is in the regulation of interferon-regulatory factors but not of NF-B31.
IRAK2 is suggested to be involved in signaling through TIRAP, the adaptor protein responsible for TLR2 and TLR4 responses6. Overexpression of IRAK2 in cells results in NF-B activation32, 33. IRAK2 is also involved in TLR3 and TLR8 signaling34. However, the function of IRAK2 in vivo and its relation to other IRAK family members are not yet understood. It has remained unclear whether IRAK2 functions as an active kinase, as a critical aspartate residue in the IRAK catalytic domain is an asparagine residue in IRAK2, and IRAK2 fails to undergo autophosphorylation, unlike IRAK1 and IRAK4 (ref. 35).
Here we generated Irak2–/– mice and investigated the function of IRAK2 in TLR responses. Irak2–/– mice were resistant to TLR4- and TLR9-mediated shock responses and Irak2–/– cells showed impaired production of proinflammatory cytokines. IRAK2 was essential for sustaining TLR-mediated expression of genes encoding proinflammatory cytokines. Recruitment of NF-B to the Il6 promoter within 8 h of TLR2 stimulation was impaired in Irak2–/– macrophages, which suggested that IRAK2 is critical for sustaining IL-6 promoter activation. Macrophages lacking both IRAK1 and IRAK2 showed substantial defects in the induction of genes encoding proinflammatory cytokines in response to TLR ligands relative to that of cells lacking either IRAK alone, which indicated that IRAK1 and IRAK2 function redundantly in the TLR signaling pathway. An intact kinase domain of IRAK2 was essential for TLR-induced cytokine production, and the kinase activity of IRAK2 was sustained for longer than that of IRAK1 after TLR stimulation. Our results indicate that IRAK2 is critical in late-phase TLR responses and that IRAK1 and IRAK2 function redundantly in the initial response.
Results
Essential function for IRAK2 in TLR-induced responses
To investigate the function of IRAK2 in vivo, we generated Irak2–/– mice by homologous recombination of embryonic stem cells. We targeted exons 4, 5 and 6 of mouse Irak2 with a neomycin-resistance cassette in embryonic stem cells and established Irak2–/– mice (Supplementary Fig. 1a online). We confirmed homologous recombination of the Irak2 locus by Southern blot analysis (Supplementary Fig. 1b). Expression of IRAK2 mRNA and protein was abrogated in Irak2–/– cells (Supplementary Fig. 1c,d). Irak2–/– mice grew normally and did not show any gross abnormalities until the age of 24 weeks. Flow cytometry showed that the spleens and lymph nodes of wild-type and Irak2–/– mice did not have a different composition of T cells, B cells, macrophages and dendritic cells (data not shown).
We first examined the function of IRAK2 in response to stimulation with TLR ligands in vivo. After challenge with LPS or CpG DNA together with D-galactosamine, all wild-type mice succumbed to shock and died. In contrast, Irak2–/– mice were more resistant to such challenge; 80% and 40% of mice survived after challenge with LPS and CpG DNA, respectively (Fig. 1a,b). These results indicate that IRAK2 is critical for TLR-induced shock in vivo. We next examined cytokine production by macrophages in response to TLR ligands, including MALP-2 (TLR6-TLR2), polyinosinic-polycytidylic acid (poly(I:C); TLR3), LPS (TLR4), resiquimod (R-848; TLR7) and CpG DNA (TLR9). We stimulated thioglycollate-elicited peritoneal macrophages with each TLR ligand and measured the production of proinflammatory cytokines by enzyme-linked immunoassay (ELISA). The production of IL-6, TNF and the inflammatory chemokine KC in response to these TLR ligands, except poly(I:C), was considerably impaired in Irak2–/– macrophages relative to that in wild-type cells (Fig. 1c–e). In contrast, TNF production in response to poly(I:C) was not different for wild-type versus Irak2–/– macrophages. Conventional dendritic cells from Irak2–/– mice also showed defective cytokine production in response to these TLR ligands (data not shown). Thus, IRAK2 is important for eliciting cytokine production in response to various TLR ligands, except for a TLR3 ligand.
Figure 1: Critical function for IRAK2 in TLR-mediated responses in vivo.
(a,b) Survival of age-matched wild-type mice (WT; n = 5) and Irak2–/– mice (n = 5) challenged with LPS (3 mg; a) or CpG DNA (20 nmol) together with D-galactosamine (20 mg; b) and monitored for 5 d. (c–e) ELISA of IL-6 (c), TNF (d) and KC (e) in culture supernatants of thioglycollate-elicited wild-type and Irak2–/– peritoneal macrophages stimulated for 24 h with MALP-2 (10 ng/ml), poly(I:C) (100 
g/ml), LPS (100 ng/ml), R-848 (10 nM) or CpG DNA (1 M). ND, not detectable. Data represent the mean and s.d. of triplicates. Similar results were obtained in three independent experiments.
TLR-mediated gene expression in Irak2–/– macrophages
We next determined by RNA blot analysis whether the impaired cytokine production resulting from IRAK2 deficiency occurred at the 'message' level. We chose MALP-2, a TLR2 ligand, as the stimulant, because TLR2 signals only through the MyD88-dependent pathway. In response to stimulation with MALP-2, wild-type macrophages expressed Il6, Tnf (called 'Tnfa' here) and Ptgs2 (called 'Cox2' here; Fig. 2a). In contrast, Irak2–/– macrophages failed to express Il6 in response to stimulation with MALP-2. However, expression of Tnfa and Cox2 was induced normally at 1 h after stimulation, even in the absence of IRAK2, and then was attenuated at 4 h after stimulation. Thus, we hypothesized that IRAK2 signaling was involved in regulating the mRNA stability of TLR-inducible genes. To examine the possible regulation of mRNA stability by IRAK2, we stimulated peritoneal macrophages from wild-type and Irak2–/– mice with MALP-2 for 1.5 h, then treated them with actinomycin D for an additional 0.5–8 h (Fig. 2b). The abundance of mRNA of TLR-inducible genes such as Cxcl1, Tnfa, Cox2 and Nfkbiz gradually decreased after actinomycin D treatment with a similar time course in wild-type and Irak2–/– macrophages, which indicated that the impaired cytokine gene expression in Irak2–/– cells was not due to mRNA instability.
Figure 2: Impaired MALP-2-induced gene expression in Irak2–/– cells.
![Figure 2 : Impaired MALP-2-induced gene expression in Irak2|[ndash]|/|[ndash]| cells.](/ni/journal/v9/n6/thumbs/ni.1606-F2.jpg)
(a) RNA blot analysis of the expression of Il6, Tnfa and Cox2 in wild-type and Irak2–/– peritoneal macrophages stimulated for various times (above lanes) with MALP-2 (10 ng/ml). Bottom, rehybridization of same membrane with an Actb probe. (b) RNA blot analysis of the expression of Cxcl1, Tnfa and Nfkbiz in peritoneal macrophages left untreated (0) or treated for 1.5 h with MALP-2 (10 ng/ml) and then treated for various times (above lanes; ActD) with actinomycin D (5 g/ml) and MALP-2 (10 ng/ml). Bottom, rehybridization of same membrane with an Actb probe. Similar results were obtained in three independent experiments.
Next we examined the activation of signaling molecules, including MAP kinases and NF-B, in Irak2–/– macrophages. Activation of the MAP kinases Jnk, p38 and Erk by MALP-2 was not impaired in Irak2–/– macrophages (Fig. 3a), which indicated that IRAK2 is dispensable for TLR-induced MAP kinase activation. We then analyzed NF-B activation. In response to stimulation with MALP-2, IB was degraded within 10 min (Fig. 3b) and NF-B DNA-binding activity was induced (Fig. 3c) in wild-type macrophages. It is well known that NF-B induces the expression of IB proteins, leading to the restoration of IB protein abundance to that seen before stimulation within 40 min. The degradation and recovery of IB protein, as well as NF-B DNA-binding activity, were not impaired in Irak2–/– macrophages. Next we examined whether IRAK2 controls NF-B activation later after stimulation with MALP-2. NF-B DNA-binding activity was sustained until 16 h after stimulation in wild-type macrophages (Fig. 3d). In contrast, activation beyond 4 h after stimulation was impaired in Irak2–/– cells. Gel mobility 'supershift' assays showed that the p65 and p50 subunits of NF-B were activated in both wild-type and Irak2–/– cells (Supplementary Fig. 2 online). Antibodies to other NF-B subunits such as c-Rel and p52 failed to produce 'supershifting' of NF-B in wild-type or Irak2–/– macrophages, which suggested that the composition of NF-B was not altered in wild-type versus Irak2–/– cells. These results indicate that IRAK2 is critical for sustaining NF-B activation after TLR stimulation.
Figure 3: Activation of NF-
B and MAP kinases in Irak2–/– macrophages in response to MALP-2.
![Figure 3 : Activation of NF-|[kappa]|B and MAP kinases in Irak2|[ndash]|/|[ndash]| macrophages in response to MALP-2.](/ni/journal/v9/n6/thumbs/ni.1606-F3.jpg)
(a,b) Immunoblot of whole-cell lysates of wild-type and Irak2–/– peritoneal macrophages stimulated for various times (above lanes) with MALP-2, analyzed with antibody to phosphorylated (p-) Erk, p38 or Jnk (a) or anti-IB (b). Total Erk, p38 and Jnk serve as a loading control. (c,d) EMSA of NF-B DNA-binding activity in nuclear extracts of wild-type and Irak2–/– macrophages stimulated for various times (above lanes) with MALP-2 (10 ng/ml), assessed with an NF-B-specific probe. Arrows indicate the induced NF-B complex. Results are representative of three independent experiments.
Requirement for IRAK2 kinase activity in the TLR signaling
It has been reported that IRAK2 overexpressed in IRAK1-deficient human embryonic kidney 293 (I1A) cells is unable to autophosphorylate because of the substitution in its kinase domain35. Nevertheless, all IRAK family members contain a functional ATP-binding pocket with an invariant lysine residue in the protein kinase subdomain16. Expression of IRAK1 and IRAK2 together leads to phosphorylation of IRAK2, whereas the expression of IRAK2 with a substitution (KK237AA) in its ATP-binding pocket fails to induce IRAK2 phosphorylation, which suggests that IRAK2 may function as an active kinase when phosphorylated by IRAK1 (ref. 35). To determine whether IRAK2 has intrinsic kinase activity in response to TLR stimulation in vivo, we immunoprecipitated IRAK2 from MALP-2-stimulated macrophages and did an in vitro kinase assay. IRAK2 phosphorylation was induced after 20 min in wild-type but not Irak2–/– macrophages in response to TLR2 stimulation (Fig. 4a). In contrast, IRAK2 deficiency did not affect MALP-2-induced IRAK1 autophosphorylation (Fig. 4b). The extent of IRAK2 phosphorylation was not different in wild-type versus IRAK1-deficient (Irak1–/–) macrophages (Fig. 4c), which indicated that IRAK2 was not phosphorylated by coprecipitated IRAK1. Furthermore, Irak4–/– macrophages failed to induce IRAK2 phosphorylation (Fig. 4d), which indicated that IRAK4 is essential for the activation of IRAK2.
Figure 4: Requirement for IRAK2 kinase activity in the response to stimulation with MALP-2.
(a–d) In vitro kinase assay of the activation of IRAK1 and IRAK2 in wild-type and Irak2–/– (a,b), Irak1–/– (c) or Irak4–/– (d) peritoneal macrophages stimulated for various times (above lanes) with MALP-2 (10 ng/ml); cell lysates were immunoprecipitated (IP) with anti-IRAK2 (a,c,d) or anti-IRAK1 (b). kDa, kilodaltons. Data are representative of three independent experiments. (e–h) Requirement for IRAK2's kinase activity in the response to TLR stimulation. (e) Mutant IRAK2 construct, containing a point substitution of the lysine at position 237 with an alanine residue (K237A). (f,g) ELISA of IL-6 (f) and TNF (g) in culture supernatants of macrophages generated from Irak2–/– bone marrow cells retrovirally transduced with wild-type or K237A IRAK2; macrophages were stimulated for 24 h with MALP-2 (10 ng/ml). Data represent the mean and s.d.; similar results were obtained in three independent experiments. (h) In vitro kinase assay of the kinase activity of IRAK2 (top) and immunoblot (IB) analysis of IRAK2 expression (middle) in retrovirally transduced bone marrow cells stimulated for 30 min with MALP-2 (10 ng/ml); lysates were immunoprecipitated with anti-IRAK2 for the kinase assay. Bottom, immunoblot analysis of -tubulin (loading control). Data are representative of three independent experiments.
We next determined whether IRAK2 kinase activity is required for its function. We retrovirally expressed wild-type IRAK2 or a kinase-defective IRAK2 mutant with a K237A substitution (Fig. 4e) in Irak2–/– macrophages and assessed the responses of the cells to TLR stimulation. The populations of CD11b+ cells expressing retroviral constructs were similar for cells transduced with wild-type or K237A IRAK2 (Supplementary Fig. 3a online). Immunoblot analysis confirmed the expression of both wild-type and K237A IRAK2 (Supplementary Fig. 3b). Reconstitution of Irak2–/– macrophages with wild-type IRAK2 increased the production of IL-6 and TNF in response to stimulation with MALP-2. In contrast, K237A IRAK2 failed to restore MALP-2 responsiveness (Fig. 4f,g). Furthermore, reconstitution with wild-type IRAK2 restored the phosphorylation of IRAK2 in response to stimulation with MALP-2, but reconstitution with K237A IRAK2 did not (Fig. 4h), which confirmed that an intact kinase domain is essential for IRAK2 phosphorylation. These results indicate that IRAK2 kinase activity is indispensable for its function in vivo.
Sustained activation of IRAK2 after TLR2 stimulation
IRAK1 is known to be ubiquitinated and degraded after IL-1 stimulation36, 37. However, it remains unknown how expression of IRAK2 and IRAK4 is regulated in response to TLR stimulation. In response to stimulation with MALP-2, there was a decrease in IRAK1 protein within 30 min of stimulation, and it remained suppressed for up to 8 h after stimulation (Fig. 5a). We noted a similar decrease in IRAK1 expression in Irak2–/– macrophages. In contrast, the abundance of IRAK2 or IRAK4 protein was not altered in response to stimulation with MALP-2 (Fig. 5b,c). The activation of IRAK1 autophosphorylation was induced rapidly and peaked at 0.5 h after stimulation with MALP-2 (Fig. 5d). Subsequently, this autophosphorylation decreased coincidentally with the diminished expression of IRAK1 protein. Notably, MALP-2-induced IRAK2 phosphorylation was sustained up to 8 h after stimulation (Fig. 5e). We further examined association between IRAK4 and IRAK1 or IRAK2 in response to stimulation with MALP-2. The association between IRAK4 and IRAK1 was induced 5 min after stimulation and decreased at 1 h after stimulation (Fig. 5f). In contrast, IRAK2 was precipitated together with IRAK4 at later time points and this was sustained until 8 h after stimulation (Fig. 5g). These results suggest that IRAK1 and IRAK2 are activated in response to TLR stimulation at early and late time points, respectively.
Figure 5: Sequential activation of IRAK1 and IRAK2 kinase activity after stimulation with MALP-2.
(a–c) Immunoblot analysis of the expression of IRAK1 (a), IRAK2 (b) and IRAK4 (c) in response to stimulation with MALP-2. Bottom, immunoblot analysis of -tubulin (loading control). Similar results were obtained in two independent experiments. (d,e) In vitro kinase assay of the activation of IRAK1 and IRAK2 in peritoneal macrophages stimulated for various times (above lanes) with MALP-2 (10 ng/ml); lysates were immunoprecipitated with anti-IRAK1 (d) or anti-IRAK2 (e). Data are representative of three independent experiments. (f,g) Immunoassay of wild-type macrophages stimulated for various times (above lanes) with MALP-2 (10 ng/ml); lysates were immunoprecipitated with anti-IRAK4, followed by immunoblot analysis with anti-IRAK1 (f) or anti-IRAK2 (g). Below, immunoblot analyses of IRAK4 serve as a loading control. Similar results were obtained in three independent experiments.
Relationship between IRAK1 and IRAK2 in TLR responses
The differential activation of IRAK1 and IRAK2 suggested that these kinases function redundantly in TLR signaling and responses. The production of TNF, IL-6 and KC in response to TLR ligands was modestly impaired in Irak1–/– macrophages (Supplementary Fig. 4 online). Therefore, we generated mice doubly deficient in IRAK1 and IRAK2 (Irak1–/–Irak2–/– mice) and examined their responses to TLR simulation. When we inoculated wild-type and Irak1–/–Irak2–/– mice with LPS or CpG DNA, Irak1–/–Irak2–/– mice were completely resistant to septic shock induced by LPS or CpG DNA (Fig. 6a,b). The production of IL-6, TNF and KC in response to these TLR ligands, except poly(I:C), was abrogated in Irak1–/–Irak2–/– macrophages (Fig. 6c–e).
Figure 6: Redundant functions of IRAK1 and IRAK2 in TLR responses.
(a,b) Survival of age-matched wild-type mice (n = 4) and Irak1–/–Irak2–/– mice (n = 4) challenged with LPS (3 mg; a) or CpG DNA (20 nmol) togetherwith D-galactosamine (20 mg; b) and monitored for 5 d. (c–e) ELISA of IL-6 (c), TNF (d) and KC (e) in culture supernatants of wild-type, and Irak1–/–Irak2–/– peritoneal macrophages stimulated for 24 h with MALP-2 (10 ng/ml), poly(I:C) (100 g/ml), LPS (100 ng/ml), R-848 (10 nM) or CpG DNA (1 M). Similar results were obtained in three independent experiments.
We next analyzed MALP-2-induced mRNA expression in peritoneal macrophages (Fig. 7a). In response to stimulation with MALP-2, Irak1–/–Irak2–/– macrophages failed to express Il6, Cxcl1, Cox2 or Nfkbiz, even at early time points. Although Tnfa was expressed in response to MALP-2 even in the absence of IRAK1 and IRAK2, its expression was lower in Irak1–/–Irak2–/– macrophages. Analysis of signaling molecules showed that IRAK1 and IRAK2 were essential for the activation of MAP kinases, including Jnk, p38 and Erk (Fig. 7b). Nevertheless, NF-B DNA-binding activity was induced after stimulation with MALP-2, even in Irak1–/–Irak2–/– macrophages, although the activity was lower (Fig. 7c).
Figure 7: Double deficiency in IRAK1 and IRAK2 causes impaired signaling and is similar to IRAK4 deficiency.
(a) RNA blot analysis of the expression of Il6, Cxcl1, Tnfa, Cox2 and Nfkbiz in peritoneal macrophages stimulated for various times (above lanes) with MALP-2 (10 ng/ml). Bottom, rehybridization of same membrane with an Actb probe. (b) Immunoblot analysis of phosphorylated Erk, p38 and Jnk in whole-cell lysates of wild-type and Irak1–/–Irak2–/– macrophages stimulated with MALP-2. Total Erk, p38 and Jnk serve as a loading control. (c) EMSA of nuclear extracts of MALP-2-stimulated wild-type and Irak1–/–Irak2–/– macrophages, analyzed with an NF-B-specific probe. Arrow indicates the induced NF-B complex. Data are representative of three independent experiments.
IRAK1 and IRAK2 have been linked to the response to IL-1. The production of IL-6 in response to IL-1 in Irak2–/– and Irak1–/– mouse embryonic fibroblasts was impaired relative to that in wild-type mouse embryonic fibroblasts and was abrogated in Irak1–/–Irak2–/– mouse embryonic fibroblasts (Supplementary Fig. 5 online), which indicated that IRAK1 and IRAK2 are involved in the IL-1R signaling.
These findings collectively suggest that IRAK1 and IRAK2 function redundantly in TLR and IL-1R signaling downstream of IRAK4. The defects in the MALP-2 responsiveness of Irak1–/–Irak2–/– macrophages are reminiscent of those of Irak4–/– macrophages22. Therefore, these results indicate that either IRAK1 or IRAK2 is required for IRAK4 signaling.
Gene expression in the absence of IRAK1 and IRAK2
Finally, we examined the expression of MALP-2-inducible genes in wild-type, Irak2–/– and Irak1–/–Irak2–/– macrophages by microarray analysis. In wild-type macrophages, in response to stimulation with MALP-2, 171 genes were upregulated more than tenfold at 2, 4 or 8 h after stimulation; we defined these as 'MALP-2-inducible genes' (Supplementary Table 1 online). As shown by a 'heat map' of MALP-2-inducible genes in wild-type, Irak2–/– and Irak1–/–Irak2–/– macrophages, the expression of various MALP-2-inducible genes at 2 h after stimulation was similar in wild-type and Irak2–/– macrophages (Fig. 8). Nevertheless, the expression of most genes in Irak2–/– cells was lower by 8 h after stimulation. This finding is consistent with impaired activation of NF-B at later time points. In the absence of both IRAK1 and IRAK2, the expression of MALP-2-inducible genes was impaired much more substantially. However, their expression was not completely abrogated, which suggests that signaling pathways independent of the IRAK family are involved in the regulation of TLR2-inducible genes.
Figure 8: Expression of MALP-2-inducible genes in IRAK1- and IRAK2-deficient macrophages.
Cluster images of microarray analysis (heat map and dendrogram) of MALP-2-inducible genes in wild-type, Irak2–/– and Irak1–/–Irak2–/– macrophages stimulated for various times (above columns) with MALP-2 (10 ng/ml). Genes upregulated more than tenfold in wild-type macrophages at 2, 4 or 8 h after stimulation were defined as 'MALP-2-inducible genes'.
Full size image (47 KB)Discussion
In this study we generated Irak2–/– mice and examined the function of IRAK2 in TLR signaling, relative to that of IRAK1. Irak2–/– mice were resistant to shock responses mediated by LPS and CpG DNA. Although Irak1–/– mice are also reported to be resistant to LPS shock, the difference in the mortality rates of wild-type and Irak1–/– mice is subtle38. Thus, IRAK2 seems to be more involved in the mortality caused by TLR stimulation in vivo than is IRAK1. IRAK2 deficiency attenuated responses to various TLR ligands, including MALP-2, LPS, R-848 and CpG DNA, with the lone exception of poly(I:C). IL-1-induced cytokine production was also impaired in Irak2–/– cells. Although previous results have suggested that IRAK2 acts downstream of TIRAP, an adaptor responsible for TLR2 and TLR4 signaling, our findings here have indicated that IRAK2 is critical not only for the TLR2 and TLR4 signaling pathways but also for the signaling pathways downstream of other TLRs with MyD88.
IRAK2 is critical for sustaining the expression of genes encoding proinflammatory cytokines in response to TLR stimulation. Although the initial upregulation of Tnfa and Cxcl1 in response to TLR2 stimulation was not different control versus Irak2–/– macrophages, their expression was lower in Irak2–/– macrophages 4 h after stimulation. Thus, the amount of TNF protein produced during the initial 24 h was much lower in Irak2–/– macrophages. Although we hypothesized that IRAK2 was essential for ensuring the stability of mRNA encoding various cytokines, there was no difference in the degradation of cytokine transcripts in wild-type versus Irak2–/– macrophages. Thus, it seems that IRAK2 is important for sustaining the transcription of genes encoding proinflammatory cytokines involved in TLR signaling. Indeed, NF-B DNA-binding activity was lower at later time points after TLR2 stimulation, which indicated that IRAK2-dependent signaling is critical for sustaining the activation of transcription factors such as NF-B.
It has been reported that IRAK2 is involved in the activation of a MyD88-independent signaling pathway by activating TIRAP6. Another report has shown that IRAK2 is involved in TLR3 signaling 'upstream' of TRIF34. However, we found that TLR4-induced interferon-inducible gene expression was not impaired in Irak2–/– macrophages (data not shown). In addition, cytokine production in response to poly(I:C) stimulation was not impaired in Irak2–/– or Irak1–/–Irak2–/– macrophages. Although the reason for discrepancy with previous reports is not clear, we believe that IRAK2 is involved in MyD88-dependent signaling pathways but not in the TRIF-dependent pathway emanating from TLR4.
Among IRAK family members, IRAK1 and IRAK4 have been shown to have intrinsic kinase activity39, 40. However, a critical aspartate residue in the IRAK catalytic domain is an asparagine or serine in IRAK2 or IRAKM, and their kinase domains are supposedly inactive. However, it has also been shown that coexpression of IRAK1 and IRAK2 leads to phosphorylation of IRAK2, whereas expression of IRAK2 with a substitution in its ATP-binding pocket (KK237AA) together with IRAK1 fails to induce IRAK2 phosphorylation35. That report points out the possibility that IRAK2 is phosphorylated by IRAK1, thereby activating the intrinsic kinase activity of IRAK2. We have also shown here, by an in vitro kinase assay, that TLR2 stimulation induced phosphorylation of IRAK2 when cell lysates were immunoprecipitated with an IRAK2-specific antibody. Although it has been hypothesized that IRAK1 activates IRAK2 (ref. 35), there was TLR2-induced IRAK2 phosphorylation even in Irak1–/– macrophages. These results indicate that IRAK1 is not responsible for the phosphorylation of IRAK2 in response to TLR stimulation. However, IRAK4 deficiency abrogated the phosphorylation of IRAK2, which suggests that activated IRAK4 phosphorylates IRAK1 and IRAK2, thereby inducing their autophosphorylation activity. Further, reconstitution of Irak2–/– macrophages with wild-type IRAK2 restored the phosphorylation of IRAK2 in response to MALP-2, but reconstitution with K237A IRAK2 did not, which indicates that the kinase activity of IRAK2 is critical for IRAK2 phosphorylation. It has also been reported that IRAK2 expression promotes the ubiquitination of TRAF6 (ref. 34). Thus, both IRAK1 and IRAK2 activate intracellular signaling cascades through TRAF6.
The requirement for the kinase activity of IRAK family members is still not well understood. Although we have shown here that IRAK4's kinase activity was critical for its function, IRAK1's kinase activity is reportedly dispensable for IL-1-mediated signaling37, 40, 41, 42, 43. Furthermore, a kinase-defective IRAK1 mutant potently induced NF-B activation. Reconstitution of Irak2–/– macrophages with wild-type IRAK2 restored IL-6 production in response to stimulation with MALP-2, but reconstitution with K237A IRAK2 did not, which suggests that IRAK2's kinase activity is essential for its function in regulating cytokine production.
It has been shown that IRAK1 is rapidly ubiquitinated and degraded by the ubiquitin-proteasome system after TLR stimulation36. IRAK1 is also reported to have 'PEST' sequences, which are involved in regulating proteolysis16, 44. IRAK1 protein abundance and autophosphorylation were lower after TLR2 stimulation. In contrast, IRAK2 expression was not altered in response to TLR2 stimulation. Notably, IRAK2 autophosphorylation began 20 min after TLR2 stimulation, peaked at 8 h after stimulation and was sustained as late as 16 h after stimulation. Thus, IRAK2 may function exclusively to sustain TLR-mediated signaling and cause robust proinflammatory cytokine production.
The initial TLR2-induced expression of Cxcl1 and Tnfa mRNA was lower in Irak1–/– macrophages, although their expression was similar to the expression in wild-type cells at later time points. Those observations suggest that the presence of two kinases is beneficial for inducing strong initial cytokine responses to the invasion of pathogens. We speculate that the degradation of IRAK1 might be a prerequisite for preventing the overproduction of proinflammatory cytokines, which may cause harmful septic shock.
Among the IRAK family members, IRAK4 has been shown to be essential for the responses to TLR and IL-1R stimulation in Irak4–/– mice17, 19. In response to TLR stimulation, IRAK4 phosphorylates IRAK1, inducing IRAK1 activation and an association between IRAK1 and TRAF6. Although IRAK1 is critical for IL-1R-mediated signaling in human embryonic kidney 293 cells lacking IRAK1 expression35, analysis of Irak1–/– mice shows that IRAK1 contributes modestly to TLR-induced cytokine production in macrophages29, 30, and the presence of IRAK1-independent signaling pathways has been predicted. We have shown here that TLR-mediated cytokine production was abrogated in the absence of both IRAK1 and IRAK2. Activation of MAP kinases was also abrogated when both IRAK1 and IRAK2 were absent. These results indicate that IRAK1 and IRAK2 function redundantly in the production of cytokines and MAP kinase activation. The phenotype of Irak1–/–Irak2–/– macrophages is reminiscent of that of Irak4–/– macrophages. It has been shown that IRAK4 phosphorylates IRAK1, thereby activating its kinase activity. Thus, it is plausible that IRAK2 is also phosphorylated by IRAK4, leading to its activation. Indeed, phosphorylation of IRAK2 has not been reported in Irak4–/– macrophages22. These results indicate that IRAK4 acts upstream of both IRAK1 and IRAK2 to activate downstream signaling cascades.
Our microarray analysis showed that IRAK1 and IRAK2 were critical for the regulation of only some MALP-2-inducible genes, although cytokines and chemokines are regulated by IRAK1 and IRAK2. That result is consistent with the induction of NF-B even in the absence of IRAK1 and IRAK2. A reported MyD88-dependent IRAK4 independent pathway22 may be responsible for the expression of these genes. Identifying the signaling molecules responsible for the IRAK-independent signaling pathways will be essential for understanding complex TLR-induced gene expression mechanisms. However, the gene expression profiles of wild-type and Irak2–/– macrophages 2 h after stimulation with MALP-2 were similar. The expression of genes encoding cytokines was not sustained in Irak2–/– macrophages, and the expression of 38 genes was impaired 50% in Irak2–/– cells 8 h after stimulation. That result supports the idea that IRAK2 is involved in the expression of genes encoding cytokines later in response to the TLR stimulation. Further studies are needed to understand the mechanism underlying this signaling pathway.
Finally, our data have shown that IRAK1 and IRAK2 acted redundantly at early time points after TLR stimulation, whereas IRAK2 was critical for sustaining the responses at later time points. Both IRAK1 and IRAK2 seemed to be activated downstream of IRAK4. Moreover, the kinase activity of IRAK2 was essential for cytokine production in response to TLR stimulation in macrophages. Given that IRAK2 deficiency minimally affected TLR-induced gene expression, except for that of genes encoding proinflammatory cytokines, the development of a small molecule targeting IRAK2 kinase activity will be beneficial to therapies for septic shock by preventing a wide spectrum of immune suppression.
Methods
Generation of Irak2–/– mice.
Irak2 was isolated from genomic DNA extracted from embryonic stem cells by PCR. The targeting vector was constructed by replacement of a 2.9-kilobase fragment encoding the Irak2 open reading frame with a neomycin-resistance cassette and a herpes simplex virus thymidine kinase cassette driven by the promoter of the gene encoding phosphoglycerate kinase, which had been inserted into the genomic fragment for negative selection. After the targeting vector was transfected into embryonic stem cells, colonies doubly resistant to the aminoglycoside G418 and gancyclovir were selected and screened by PCR and their identities were further confirmed by Southern blot analysis. Homologous recombinants were microinjected into blastocysts from C57BL/6 female mice, and heterozygous F1 progeny were intercrossed to obtain Irak2–/– mice. Irak2–/– mice on the 129/Sv 
C57BL/6 background and their littermates (controls) were used. All animal experiments were with the approval of the Animal Research Committee of the Research Institute for Microbial Diseases (Osaka University).
Cells.
Peritoneal exudate cells were isolated from the peritoneal cavities of mice 3 d after mice were injected with 2 ml of 4.0% (wt/vol) thioglycollate medium (Sigma) by washing with ice-cold Hank's buffered-salt solution (Invitrogen).
Reagents.
MALP-2 was provided as described45. LPS from Salmonella minnesota Re-595 was from Sigma-Aldrich; poly(I:C) was from Amersham Biosciences; and R-848 was provided by the Pharmaceuticals and Biotechnology Laboratory of Japan Energy. CpG oligonucleotide was synthesized as described46. Polyclonal antibodies to phosphorylated Jnk (9251), p38 (9211) and Erk (9101) were from Cell Signaling. Polyclonal antibody to -tubulin (anti--tubulin; sc-5274), anti-IB (sc-371), anti–NF-B p50 (sc-1192) and anti–NF-B p65 (sc-109) were from Santa Cruz. Rabbit polyclonal anti-IRAK1 and anti-IRAK4 were made as described22, 47. Rabbit polyclonal anti-IRAK2 (3595) was from ProSci.
Measurement of cytokine production.
Concentrations of cytokines in culture supernatants were measured by ELISA. ELISA kits for mouse TNF, IL-6 and KC were from R&D Systems.
In vitro kinase assay.
Peritoneal macrophages stimulated with MALP-2 (10 ng/ml) were lysed and immunoprecipitated with anti-IRAK1, anti-IRAK2 or anti-IRAK4, then the activity of IRAK1, IRAK2 and IRAK4 was measured by an in vitro kinase assay as described22.
RNA hybridization.
Peritoneal macrophages were treated for 0, 1, 2, 4 or 8 h with MALP-2 (10 ng/ml) and total RNA was extracted with TRIzol reagent (Invitrogen). RNA was separated by electrophoresis and transferred to nylon membranes and then hybridized with the appropriate cDNA probe. For the detection of Irak2 mRNA expression, a 314–base pair fragment (nucleotides 894–1208) was used as a probe. The same membrane was rehybridized with an Actb probe (encoding -actin) as a loading control.
Immunoblot analysis.
Peritoneal macrophages were treated for various times with MALP-2 (10 ng/ml). Cells were then lysed in lysis buffer containing 1.0% (vol/vol) Nonidet-P40, 150 mM NaCl, 20 mM Tris-HCl, pH 7.5, 1 mM EDTA and a protease inhibitor 'cocktail' (Roche). Cell lysates were separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes. Membranes were blotted with the appropriate specific antibodies and were visualized with an enhanced chemiluminescence system (NEN Life Science Products).
Electrophoretic mobility-shift assay.
Nuclear extracts prepared from 4 106 peritoneal macrophages stimulated with MALP-2 (10 ng/ml) as described12 were incubated with or without anti–NF-B p65 or anti–NF-B p50 and then were further incubated with a specific probe for NF-B DNA-binding sites, separated by electrophoresis and visualized by autoradiography.
Construction of IRAK2 expression plasmids.
Full-length IRAK2 cDNA was obtained by RT-PCR from a human cDNA library, and a point mutation resulting in a K237A substitution in the kinase domain was introduced by site-directed mutagenesis. Full-length or mutated IRAK2 cDNA was cloned into the 'LZR' vector.
Retroviral transduction.
Highly efficient retroviral transduction of macrophages was achieved by transduction of hematopoietic stem cells before differentiation into macrophages as described48. Bone marrow was isolated from Irak2–/– mice that had been injected intraperitoneally 4 d earlier with 5 mg of 5-fluorouracil (Nacalai Tesque). Cells were cultured in stem cell media (DMEM supplemented with 15% (vol/vol) FCS, sodium pyruvate (10 mM), L-glutamine (2 mM), -mercaptoethanol (50 M), penicillin (100 U/ml), streptomycin (100 g/ml), stem cell factor (100 ng/ml), IL-6 (10 ng/ml) and IL-3 (10 ng/ml)). Then, 48 h later, cells were transduced with retroviral supernatant (supplemented with stem cell factor, IL-6, IL-3 and 10 ng/ml of polybrene) on 2 successive days. Virus was produced with the PlatE retrovirus packaging cell line. After the second transduction, cells were washed and resuspended in macrophage growth media (RPMI 1640 medium supplemented with 10% (vol/vol) FCS, HEPES (10 mM), pH 7.0, sodium pyruvate (10 mM), L-glutamine (2 mM), -mercaptoethanol (50 M), penicillin (100 U/ml), streptomycin (100 g/ml) and macrophage colony–stimulating factor (40 ng/ml)). Cells were cultivated for 7 d and then analyzed.
Microarray.
Peritoneal macrophages were stimulated for 0, 2, 4 or 8 h with MALP-2 (10 ng/ml), then total RNA was extracted with TRIzol (Invitrogen Life Technologies) and further purified with an RNeasy kit (Qiagen). Biotin-labeled cDNA was synthesized from 100 ng total RNA with the Ovation Biotin RNA Amplification and Labeling System (Nugen) according to the manufacturer's protocol. Affymetrix mouse Genome 430 2.0 microarray chips were hybridized, stained, washed and scanned according to the manufacturer's instructions. Data were analyzed MicroArray Suite software (Affymetrix) and ArrayAssist software (Stratagene).
Accession codes.
UCSD-Nature Signaling Gateway (http://www.signaling-gateway.org): IRAK1 (A001277), IRAK2 (A001278), IRAKM and IRAK4 (A003450); GEO: microarray data, GSE10765.
Note: Supplementary information is available on the Nature Immunology website.
Author contributions
T. Kawagoe, O.T. and S.A. designed the research and analyzed data; T. Kawagoe did most of the experiments; S.S., K. Matsushita., H.K., K. Matsui., Y.K., T.S. and T.K. provided advice; and T. Kawagoe, O.T. and S.A. prepared the manuscript.
is IRAK-4 dependent and redundant for protective immunity to viruses
