A Novel Splice Variant of Interleukin-1 Receptor (IL-1R)-Associated Kinase 1 Plays a Negative Regulatory Role in Toll/IL-1R-Induced Inflammatory Signaling

Abstract

The interleukin-1 (IL-1) receptor-associated kinase 1 (IRAK1) is a member of the IRAK kinase family that plays a pivotal role in the Toll/IL-1 receptor (TIR) family signaling cascade. We have identified a novel splice variant, IRAK1c, which lacks a region encoded by exon 11 of the IRAK1 gene. IRAK1c expression was confirmed by both RNA and protein detection. Although both IRAK1 and IRAK1c are expressed in most tissues tested, IRAK1c is the predominant form of IRAK1 expressed in the brain. Unlike IRAK1, IRAK1c lacks kinase activity and cannot be phosphorylated by IRAK4. However, IRAK1c retains the ability to strongly interact with IRAK2, MyD88, Tollip, and TRAF6. Overexpression of IRAK1c suppressed NF-κB activation and blocked IL-1β-induced IL-6 as well as lipopolysaccharide- and CpG-induced tumor necrosis factor alpha production in multiple cellular systems. Mechanistically, we provide evidence that IRAK1c functions as a dominant negative by failing to be phosphorylated by IRAK4, thus remaining associated with Tollip and blocking NF-κB activation. The presence of a regulated, alternative splice variant of IRAK1 that functions as a kinase-dead, dominant-negative protein adds further complexity to the variety of mechanisms that regulate TIR signaling and the subsequent inflammatory response.

The Toll/interleukin-1 (IL-1) receptor (TIR) family, consisting of the Toll receptors, the IL-1 receptor and the IL-18 receptor, plays a crucial role in innate immunity (1). The TIR family of proteins is defined by the presence of an intracellular TIR domain that is required for signal transduction upon receptor activation. After ligand activation, TIR family members form multimeric receptor complexes and via cytoplasmic TIR domains recruit adapter proteins such as MyD88, TIRAP (TIR domain-containing adapter protein), and TRIF (TIR domain-containing adapter inducing β interferon) (24). This results in the sequential activation of a conserved signaling module that includes the IL-1 receptor-associated kinases (IRAKs) and TRAF6, followed by activation of nuclear factor-κB (NF-κB), p38 kinase, and c-Jun N-terminal kinase (JNK), eventually leading to gene transcription and induction of an inflammatory response (1).

The IRAK family of serine-threonine kinases consists of four members: IRAK1, IRAK2, IRAKM, and IRAK4 (1, 6). Despite significant structural homology between family members, they also have distinct functional roles in signal transduction. Both IRAK1 and IRAK4 exhibit kinase activity, while IRAKM and IRAK2 lack this activity (6). IRAKM functions as an induced negative regulator (13), and IRAK2 appears to be partially redundant for IRAK1 (21). Both human and mouse IRAK4 deficiency results in nonresponsiveness to a broad panel of TIR family ligands (19, 22). Although the importance of kinase activity remains controversial, IRAK1 deficiency results in a partial defect in TIR activation, with substantial decreases in IL-1, IL-18, and lipopolysaccharide (LPS) responsiveness (11, 12).

Upon ligand activation of TIR family members, IRAK4 and IRAK1 are recruited to the receptor complex (1). At the receptor, IRAK1 associates with Tollip, MyD88, and TRAF6, and phosphorylation by IRAK4 triggers IRAK1 autophosphorylation (10, 15, 17). IRAK1 hyperphosphorylation results in disassociation from Tollip and release from the receptor-MyD88 complex (2, 10). This leads to the formation of a new protein complex consisting of hyperphosphorylated IRAK1 and TRAF6, a prerequisite for TRAF6-mediated NF-κB activation and induction of an inflammatory response (14).

TIR signal transduction is regulated by multiple mechanisms to prevent tissue damage that arises from sustained inflammation. IRAK1 is one of the most receptor-proximal kinases and thus is regulated by multiple mechanisms. IRAKM is an induced regulator that is proposed to block IRAK4 activation and subsequent IRAK1 phosphorylation (13). IRAK1 hyperphosphorylation results in a decrease in protein stability, providing a potential mechanism to regulate IRAK1 activity (27). Alternative splicing is another mechanism by which a single gene can generate multiple, functionally distinct protein products. A splice variant of a key adapter protein in the TIR pathway, MyD88, can regulate signaling by serving as a dominant negative (8). Furthermore, several IRAK1 splice variants have been described in both humans and mice. IRAK1b, a splice variant found in humans, lacks 90 bp arising from the use of an alternative 5′ acceptor splice site in exon 12 of the gene (9). IRAK1b lacks kinase activity and exhibits a prolonged half-life (9). IRAK1s, a splice variant found in the mouse, is generated by a novel splice acceptor site in exon 12 of the murine gene resulting in a frameshift and generation of a premature stop codon (28). Although IRAK1s is kinase dead, it constitutively activated the NFκB and JNK pathway.

Here, we report the identification of a novel IRAK1 splice variant, IRAK1c, which lacks a region encoded by exon 11 of the gene. IRAK1c can be detected in different tissues and cell lines and is the exclusive form of IRAK1 expressed in brain. IRAK1c lacks kinase activity and potently suppresses activation by both IL-1β and Toll receptor ligands. IRAK1c can be recruited to the IL-1 receptor upon activation, but IRAK1c is not phosphorylated by IRAK4 and thus cannot disassociate from the adapter proteins MyD88 and Tollip, resulting in a block in NF-κB activation and cytokine production. Furthermore, unlike IRAK1 or the point mutant IRAK1-KD (where KD indicates a kinase-dead point mutation), IRAK1c is unable to reconstitute the IL-1β-driven IL-6 production in IRAK1^−/− fibroblasts. The identification of an inducible, dominant-negative splice variant of IRAK1 highlights a novel mechanism by which TIR receptor signaling and the inflammatory response can be regulated.

MATERIALS AND METHODS

Biological reagents and cell culture.

Human and murine IL-1β, granulocyte-macrophage colony-stimulating factor and IL-4 were obtained from R & D Systems, and LPS was from Sigma. Human embryonic kidney cells (293), human monocytes (THP-1), and human osteosarcoma cells (G292) were obtained from the American Type Culture Collection. IRAK1^−/− and wild-type murine embryonic fibroblasts were maintained as previously described (12). 293 cells were maintained in Dulbecco's modified Eagle's medium containing 10% fetal calf serum, 20 mM HEPES, 1 mM sodium pyruvate, 1 mM nonessential amino acids, 100 U of penicillin per ml, and 100 U of streptomycin per ml (HyClone). THP-1 cells were maintained in endotoxin-free RPMI (Sigma) medium containing 10% endotoxin-free fetal calf serum (Invitrogen) and endotoxin-free penicillin-streptomycin (Sigma). G292 cells were maintained in McCoy's medium containing 10% fetal calf serum and penicillin-streptomycin (HyClone). Monocytes were purified to 99% purity from peripheral human blood using CD14 microbeads and separated on an Auto Macs (Miltenyi Biotech) instrument. Monocyte-derived dendritic cells were prepared as previously described (23) by culturing in the presence of granulocyte-macrophage colony-stimulating factor and IL-4 in endotoxin-free medium and reagents.

The following antibodies were used: anti-FLAG tag and mouse monoclonal anti-actin (Sigma); anti-myc tag (Covance); anti-Tollip mouse monoclonal (Alexis); rabbit polyclonal anti-IRAK1, mouse monoclonal anti-TRAF6, rabbit polyclonal anti-IL-1R, rabbit polyclonal anti-IκBα, and rabbit polyclonal anti-p-cJUN (Santa Cruz Biotechnology); rabbit polyclonal IRAK4 (Upstate Biotechnology); and rabbit polyclonal anti-IRAK1 as previously described (12).

RNA and RT-PCR.

Human tissue RNA and cDNA were from BD Biosciences. RNA from tissue culture cells was extracted using RNeasy (QIAGEN) according to the manufacturer's instructions including a DNase digestion (QIAGEN). cDNA was prepared using Superscript II (Invitrogen). Reverse transcription-PCR (RT-PCR) primers were designed to specifically recognize the IRAK1 splice variant by bridging exons 10 and 12. The specific forward primers for IRAK1c (GACCAAGTATCTGGTGTACGAGAG) and IRAK1 (GACCAAGTATCTGAAAGACCTGGTG) were used with a common reverse primer (TCAGCTCTGAAATTCATCACTTTC) from a 26-cycle RT-PCR. Control amplification primers for GAPDH (glyceraldehyde-3-phosphate dehydrogenase) were from Promega. PCR products were separated on agarose gels and visualized after ethidium bromide staining.

Expression constructs and stable transfection.

cDNAs were amplified using a GC-Rich PCRx system (Invitrogen). Amplification primers for IRAK1, IRAK2, IRAK4, TRAF6, MyD88, and Tollip were derived from the GenBank database. Myc, hemagglutinin (HA), and FLAG tags were introduced by PCR, and coding sequences were cloned into pCDNA3.1 or pFASTBac (Invitrogen). Purified recombinant HIS-IRAK4 for in vitro kinase assays was made using insect Sf9 cells and purified using affinity chromatography. Kinase-inactive IRAK1 and IRAK1c clones were generated by PCR containing the point mutation K239S. Glutathione transferase (GST)-IRAK1 exon 11 was generated by PCR subcloning into pGEX4T3, and GST protein was purified using glutathione beads (Amersham) after a 6-h induction with IPTG (isopropyl-β-d-thiogalactopyranoside). For retroviral vectors, coding sequences were cloned into MscvPuro (BD Biosciences), and retrovirus was produced in the provided 293 packaging cell line. Stable transfectants of G292, THP-1, and IRAK1^−/− fibroblasts cells were derived by continuous selection in puromycin (Sigma), with tagged protein expression verified by immunoblotting and intracellular staining.

Immunoprecipitation and immunoblotting.

Cells were untreated or treated with IL-1β and either lysed in a Triton-containing lysis buffer (0.5% Triton X-100, 50 mM Tris, pH 7.5, 150 mM NaCl, 1 mM NaF, 1 mM Na₃VO₄, 20 μM aprotinin, 1 mM phenylmethylsulfonyl fluoride [PMSF]) or a sodium dodecyl sulfate (SDS)-containing RIPA lysis buffer (1% TX-100, 0.5% deoxycholate, 0.1% SDS, 50 mM Tris, pH 7.5, 150 mM NaCl, 1 mM NaF, 1 mM Na₃VO₄, 20 μM aprotinin, 1 mM PMSF). Lysates were clarified by centrifugation at 14,000 × g for 10 min at 4°C. For immunoprecipitations, cell extracts were incubated with 5 μg of antibody for 2 h, followed by a 1-h incubation with 20 μl of protein G-Sepharose beads (Amersham Biosciences). After incubation, the beads were washed four times with lysis buffer, separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to Immobilon-P membranes (Millipore), and analyzed by immunoblotting as previously described (20).

In vitro kinase assay.

Cell lysis and immunoprecipitations were done in kinase lysis buffer (50 mM HEPES, pH 7.9, 20 mM MgCl₂, 1% TX-100, 1 mM NaF, 1 mM Na₃VO₄, 20 mM gylcerol-2-phosphate, 5 mM p-nitrophenyl-phosphate, 1 mM dithiothreitol, 1 mM PMSF). Immunoprecipitates were incubated in kinase buffer (20 mM Tris, pH 7.5, 20 mM MgCl₂, 1 mM EDTA, 1 mM Na₃VO₄, 20 mM gylcerol-2-phosphate, 20 mM p-nitrophenyl-phosphate, 25 nM IRAK4, 1 mM dithiothreitol, 2 μg of maltose binding protein [MBP] substrate, 20 μM ATP, and 10 μCi of [γ³²-P]ATP) for 20 min at 32°C. For GST substrate reactions, 3 μg of each GST substrate was included. Reactions were stopped by the addition of 3× SDS-PAGE sample buffer. Samples were subjected to SDS-PAGE and transferred to Immobilon-P membranes, and the proteins were visualized and quantified by autoradiography using a Typhoon 8600 Variable Mode Imager (Amersham Biosciences).

Reporter gene assays and cytokine ELISAs.

For the NF-κB reporter assay, 293 or G292 cells were seeded in 24-well plates and transfected with the indicated reporter and expression plasmid using Fugene-6 (Roche). pHTS-NF-κB-luciferase plasmid was used to measure NF-κB-dependent gene activation (Biomyx Technology). The cells were lysed in Glo lysis buffer and analyzed using the Bright Glo Luciferase assay system (Promega). For detection of secreted tumor necrosis factor alpha (TNF-α) and IL-6 protein, THP-1 cells (1 × 10⁴ cells/well), G292 cells (3 × 10⁴ cells/well) or murine fibroblasts (1 × 10⁴ cells/well) were incubated overnight in 96-well plates and stimulated for 4 h with LPS or for 6 h with IL-1β as indicated. For TLR9 stimulations, cells were stimulated with 1 μM GpC control (gggggACGATCGTCggggg) or the CpG 2216 stimulatory oligodeoxynucleotide (gggggAGCATGCTggggg) synthesized by GenBase (uppercase and lowercase letters indicate phosphodiester-linked and phosphorothioate-linked nucleotides, respectively). Supernatants were collected and analyzed for TNF-α or IL-6 using enzyme-linked immunosorbent assay (ELISA) kits (R&D Systems).

RESULTS

IRAK1c is widely expressed and the exclusive form of IRAK1 in the brain.

In the course of studying the expression of IRAK1 in human cell lines and tissues, we cloned a novel IRAK1 splice variant from a human spleen cDNA library. Sequence analyses indicated that this splice variant lacked all 237 bp comprising exon 11 of the IRAK1 gene (Fig. 1A; see also Fig. S1 in the supplemental material). Unlike two other described IRAK1 splice variants, IRAK1b and IRAK1s (9, 28), the IRAK1c splice variant does not result in a frameshift or premature termination of the translated protein (Fig. 1A). IRAK1c was also cloned, and the sequence was verified from human peripheral blood mononuclear cells as well as from Jurkat, THP-1, and HeLa cell cDNA libraries (data not shown). The existence of IRAK1c as a bona fide transcript was supported by the presence of IRAK1c sequences among the reported human IRAK1 expressed sequence tag (EST) clones. Two of these EST clones (BG1719188 and BG423016) were obtained, and the sequence was verified to be IRAK1c and not IRAK1, IRAK1b, or IRAK1s (data not shown). Furthermore, a full-length mRNA sequence (BC014963) corresponding to IRAK1c was also identified in the NCBI database.

FIG. 1.

Alternative splicing of IRAK1 and expression of the variant IRAK1c. (A) IRAK1c generated from alternative splicing of the IRAK1 gene and the encoded protein lacking all 237 bp (amino acids 435 to 514) of exon 11 (filled box) in the C-terminal region of the kinase domain are illustrated. (B) IRAK1 and IRAK1c were amplified by RT-PCR from the indicated human tissues and cell lines as described in Materials and Methods. GAPDH amplification was used as a control. (C) 293 cells were transfected with 0.25 μg of myc-IRAK1c or vector control, and lysates were prepared 48 h posttransfection. A total of 100 μg of lysate was immunoprecipitated with anti-myc and anti-IRAK1 antibodies and immunoblotted with an anti-IRAK1 antibody (top panel). Endogenous expression of IRAK1c protein was detected in two independent human brain lysate samples, THP-1 cell lysate as well as IRAK1c-transfected 293 cell lysate, by anti-IRAK1 immunoblot analyses of total protein extracts (bottom panel).

To further analyze the expression profile of IRAK1c, we designed PCR primers spanning the junctions of exons 10 and 11 that would recognize wild-type IRAK1 cDNA and primers that span the junctions of exons 10 and 12 that would specifically amplify IRAK1c cDNA. The specificity of these PCR primers was validated using cDNA prepared from 293 cells transfected with IRAK1 or IRAK1c expression vectors (data not shown). Using these primers, we further characterized the expression profile of IRAK1c in a variety of human tissues and cell lines (Fig. 1B). IRAK1c was widely expressed in both lymphoid and nonlymphoid tissues and cell lines, including lymph node, spleen, heart, brain, and liver. Interestingly, IRAK1c was the predominant form of IRAK1 detected in the brain sample (Fig. 1B, lane 6).

Transfection of 293 cells with expression vectors encoding myc-tagged IRAK1c followed by immunoprecipitation with anti-IRAK1 and anti-myc tag antibodies revealed that the splice variant was expressed as a protein of approximately 68 kDa (Fig. 1C, top panel). To confirm the expression of IRAK1c at the protein level, we analyzed lysates obtained from two independent human brain samples. When compared to 293 cells transfected with IRAK1c, lysate prepared from whole brain exhibited strong expression of IRAK1c with minute or undetectable IRAK1 expression (Fig. 1C, bottom panel). In contrast, the monocytic human cell line THP-1 predominantly expressed IRAK1 with minimal IRAK1c detected. Taken together, these results indicate that IRAK1c is an alternatively spliced variant of IRAK1 that is broadly expressed in many tissues and is the major form of IRAK1 expressed in the brain.

IRAK1c lacks kinase activity and cannot be phosphorylated by IRAK4.

Exon 11 of the IRAK1 gene (Fig. 1A), which is missing in IRAK1c, encodes a 79-amino-acid region corresponding to the C terminus of the kinase domain (see Fig. S1 in the supplemental material). To directly determine whether IRAK1c retained kinase activity, 293 cells were transfected with expression vectors encoding various myc-tagged IRAK1 proteins including a K239S kinase-dead IRAK1 point mutant (IRAK1-KD), IRAK1c, a K239S kinase-dead IRAK1c point mutant (IRAK1c-KD), or the empty vector. An in vitro kinase assay was performed with the anti-myc immunoprecipitates from transfected cell lysates using myelin basic protein as a substrate. IRAK1 overexpressed in 293 cells was constitutively active, as seen by autophosphorylation as well as phosphorylation of the MBP substrate (Fig. 2A, lane 1). In contrast, IRAK1c exhibited no kinase activity, as did the two kinase-dead point mutants IRAK1-KD and IRAK1c-KD. These data indicated that the region encoded by exon 11 of the IRAK1 gene, which is absent in IRAK1c, is required for kinase activity.

FIG. 2.

IRAK1c is kinase dead and cannot be phosphorylated by IRAK4. (A) The indicated myc-IRAK1 constructs were transfected into 293 cells, and proteins were immunoprecipitated with anti-myc antibody from 400 μg of lysate and subjected to an in vitro kinase assay as described. The upper panel shows IRAK1 autophosphorylation (pIRAK1), and the lower panel depicts phosphorylation of an MBP substrate. (B) Two sets of anti-myc immunoprecipitates from the transfected cell lysates shown in panel A were subjected to an in vitro kinase assay in the presence of recombinant IRAK4. One set of immunoprecipitates was analyzed by autoradiography. (C) The second set was washed with lysis buffer and immunoblotted with an anti-IRAK4 antibody (top panel). Expression of transfected IRAK-1 protein was verified by anti-myc immunoblotting of whole-cell lysate (middle panel). The associated IRAK4 (top) was quantified by densitometry, and association was corrected for IRAK1 protein expression (middle panel). (D) A GST fusion protein (GST-ex11) consisting of the missing exon 11 of IRAK1c was used as a substrate in an in vitro kinase assay using recombinant IRAK4 kinase. WT, wild type; KD, IRAK1-KD; c, IRAK1c; c-KD, IRAK1c-KD.

Upon TIR activation, IRAK1 becomes phosphorylated on multiple residues, resulting in a pronounced change in molecular weight as visualized by immunoblot analyses (14). IRAK4 is required to initiate the inflammatory response upon TIR activation and has been shown to phosphorylate and activate IRAK1 (15, 22). This leads to IRAK1 autophosphorylation resulting in hyperphosphorylated IRAK1 (14).

Based upon recent reports of the sequential activation of IRAK1 (14), we wanted to test whether IRAK4 could mediate the initial phosphorylation of IRAK1c. To test this, we performed in vitro kinase assays using immunoprecipitates from cells transfected with IRAK1, IRAK1-KD, IRAK1c, and IRAK1c-KD expression vectors that were incubated in the presence of recombinant purified IRAK4 (Fig. 2B). When these immunoprecipitates were incubated with recombinant IRAK4, we were able to detect a phosphorylated band that corresponded to IRAK1-KD (Fig. 2B, lane 2). Interestingly, neither IRAK1c nor IRAK1c-KD showed any phosphorylation in the presence of IRAK4 (Fig. 2B, lane 3 and 4). Based upon these data, we hypothesized that the IRAK1c splice variant is not phosphorylated by IRAK4 and thus is unable to become fully activated.

It remains possible that this finding could be due to the inability of IRAK1c to associate with IRAK4. To address this hypothesis, we tested the ability of IRAK4 to associate with the various IRAK1 proteins in this assay. A second set of anti-myc immunoprecipitates, which were subjected to an in vitro kinase assay in the presence of recombinant IRAK4, were washed with lysis buffer and immunoblotted with an anit-IRAK4 antibody. When the IRAK4 associated with IRAK1 (Fig. 2C, top panel) was quantified and corrected for IRAK1 protein expression (Fig. 2C, middle panel), we found that both IRAK1 and IRAK1-KD had similar levels of association with IRAK4 (Fig. 2C, bottom panel). However, both IRAK1c and IRAK1c-KD showed a 45% decrease in association with IRAK4. Since activated IRAK1 is targeted for degradation (27), IRAK1 expressed in 293 cells, which is kinase active, exhibits a pronounced gel shift and lower protein levels when compared to IRAK1-KD, IRAK1c, and IRAK1c-KD (Fig. 2C, middle panel).

Although the known IRAK4 phosphorylation sites in IRAK1 are not contained in the 79-amino-acid stretch comprising the missing exon 11 in IRAK1c (14), we wanted to test this experimentally. We generated a GST fusion protein consisting of exon 11 (GST-ex11) and subjected it to an in vitro kinase assay with recombinant human IRAK4. Although IRAK4 was able to mediate robust phosphorylation of a MBP substrate, no phosphorylation was detected with a GST control protein or GST-ex11 (Fig. 2D). These data suggested that the missing exon 11 of IRAK1c did not contain a previously undescribed IRAK4 phosphorylation site.

IRAK1c strongly associates with TIR family signaling proteins.

Previous studies have demonstrated that IRAK1 forms homodimers as well as heterodimers with other IRAK family members (25). To test whether IRAK1c could form a homodimer with IRAK1 or a heterodimer with IRAK2, 293 cells were transfected with HA-tagged IRAK1c and myc-tagged IRAK1 or IRAK2. Immunoblot analyses of coimmunoprecipitated proteins revealed that IRAK1c formed homodimers with IRAK1 as well as heterodimers with IRAK2 (Fig. 3A).

FIG. 3.

IRAK1c forms dimers with IRAK1 and IRAK2 and associates with MyD88, Tollip, and TRAF6. (A) 293 cells were transfected with 1 μg of the indicated myc-tagged plasmids and either cotransfected with a vector control or HA-IRAK1c. Lysates prepared in TX-100 lysis buffer 48 h posttransfection were subjected to anti-myc immunoprecipitation followed by anti-HA immunoblot analysis (top panel). Whole-cell lysates were also immunoblotted with anti-myc (middle panel) and anti-HA (bottom panel) antibodies. (B) 293 cells were transfected with 3 μg of the indicated FLAG-tagged MyD88, Tollip, and TRAF6 plasmids and 3 μg of the indicated myc-tagged IRAK1 plasmids. Lysates prepared in TX-100 lysis buffer 48 h posttransfection were subjected to anti-FLAG immunoprecipitation followed by anti-myc immunoblot analysis (top and second panels). Whole cell lysates were also immunoblotted with anti-myc (third panel) and anti-FLAG (bottom panel) antibodies to confirm protein expression.

Next, we wanted to examine if there were any biochemical differences in the associations with other signaling proteins in this pathway. In TIR signaling, IRAK1 is preassociated with Tollip in the cytosol (2). Activation of TIRs recruits the IRAK1-Tollip complex to the receptor, resulting in association with the MyD88 adapter protein. IRAK1 also associates with TRAF6, a key molecule in this pathway that activates JNK and NF-κB (4). Myc-tagged IRAK1 proteins were expressed alone or with FLAG-tagged MyD88, Tollip, and TRAF6 in 293 cells. FLAG-tagged proteins were immunoprecipitated, separated by SDS-PAGE, and immunoblotted with anti-myc antibody to detect coimmunoprecipitated IRAK1 proteins (Fig. 3B). As previously reported (15), IRAK1 expressed in 293 cells is autophosphorylated and rapidly degraded (Fig. 3B, third panel). Due to this lower level of IRAK1 protein expression, a short exposure indicated that kinase-dead IRAK1 associated strongly with the adapter proteins MyD88 and Tollip, as well as TRAF6, compared to wild-type IRAK1 (Fig. 3B, top panel). However, a longer exposure revealed that IRAK1, expressed at lower levels in these cells, did indeed associate with MyD88, Tollip, and TRAF6 (second panel). Interestingly, IRAK1c was found to behave similarly to IRAK1-KD and strongly associated with MyD88, Tollip, and TRAF6 (Fig. 3B, compare lanes 4 to 6 with lanes 7 to 10). Tollip is the adapter protein that is preassociated with IRAK1 and is responsible for recruiting IRAK1 from the cytosol to the receptor (2). The ability of IRAK1c to strongly associate with Tollip suggested that IRAK1c is recruited to the receptor complex upon ligand activation. It also indicated that the missing exon 11-encoded region in IRAK1c does not perturb the association of IRAK1 with MyD88, Tollip, and TRAF6.

IRAK1c is recruited to the IL-1R and associates with endogenous MyD88, Tollip and TRAF6 after TIR activation.

Although our previous experiment suggested IRAK1c can be recruited to the IL-1R, we wanted to experimentally test this hypothesis. To examine ligand-induced receptor association, we used G292 cells, a human osteosarcoma cell line that is strongly IL-1β responsive. To study the effects of IRAK1c, we generated stable G292 cell lines overexpressing myc-IRAK1, myc-IRAK1-KD, or myc-IRAK1c by retroviral transduction. Cells were either left unstimulated or stimulated with IL-1β, and lysates were prepared. Immunoprecipitation of IL-1R followed by anti-myc immunoblotting revealed that IRAK1 was recruited to the receptor in an activation-dependent manner (Fig. 4A). As previously reported, IRAK1-KD associated with the receptor in a manner similar to IRAK1 (10), and IRAK1c was also recruited to the IL-1R upon treatment with IL-1β.

FIG. 4.

IRAK1c associates with the IL-1 receptor and binds to Tollip, MyD88, and TRAF6 upon IL-1β stimulation. G292 stable transfectants expressing myc-IRAK1, myc-IRAK1-KD or myc-IRAK1c were either unstimulated or stimulated with 5 ng/ml of IL-1β for 3 min. (A) Cell lysates were prepared in TX-100 buffer, and 1,000 μg of lysate was subjected to anti-IL-1R immunoprecipitations followed by anti-myc immunoblotting (top panel). Whole-cell lysates were also immunoblotted with anti-myc (bottom panel) and anti-IL-1R (middle panel) antibodies to confirm protein expression. (B) Lysates from G292 cells were prepared as described for panel A. Lysates were subjected to anti-Tollip immunoprecipitations followed by anti-myc immunoblotting (top panel). Whole-cell lysates were also immunoblotted with anti-Tollip (bottom panel). (C) Lysates from panel B were subjected to anti-myc immunoprecipitations followed by anti-TRAF6 (top panel) and anti-MyD88 (third panel) immunoblotting. Whole-cell lysates were also immunoblotted with anti-TRAF6 (second panel) and anti-MyD88 (bottom panel).

Previous studies have demonstrated that Tollip associates with IRAK1 in the cytosol and, upon TIR activation, recruits IRAK1 to the membrane (2). Subsequent IRAK1 phosphorylation and autophosphorylation result in disassociation from Tollip and activation of downstream pathways through TRAF6 (14). Our overexpression data in 293 cells indicated that the IRAK1c splice variant had the ability to associate with Tollip, MyD88, and TRAF6 (Fig. 3B) yet lacked kinase activity and the ability to be phosphorylated by IRAK4 (Fig. 2). Based upon our data and the recent model proposed by Kollewe et al. (14), we hypothesized that IRAK1c might function as a naturally occurring, kinase-dead IRAK1 protein that is unable to disengage from Tollip and activate downstream pathways through TRAF6. To test whether this is the mechanism by which IRAK1c functions as a negative regulator, we used IL-1-responsive G292 cells. Anti-Tollip immunoprecipitates were separated by SDS-PAGE and immunoblotted with anti-myc antibody to visualize associated IRAK1, IRAK1-KD, or IRAK1c (Fig. 4B). As previously reported (2, 14), IRAK1 is associated with Tollip prior to activation (lane 1). Upon activation with IL-1β, IRAK1 becomes phosphorylated, exhibits kinase activity (data not shown), and undergoes a shift in molecular weight (Fig. 4A, lane 2, bottom panel). Concomitantly, IRAK1 no longer associates with Tollip (Fig. 4B, lane 2, upper panel). Both IRAK1-KD and the IRAK1c splice variant associate with Tollip in unstimulated cells (Fig. 4B, lanes 3 and 5, top panel). However, this association remains even after IL-1β stimulation (Fig. 4B, lanes 4 and 6, top panel). Similar results were obtained using THP-1 cells transfected with various IRAK1 expression constructs and stimulated with LPS (data not shown). These data suggested that IRAK1c was unable to disengage from the IL-1R complex.

Since our overexpression studies in 293 cells had indicated that IRAK1c could associate with MyD88 and TRAF6, we also tested the ability of IRAK1c to associate with these key proteins in this activation-driven cellular system. Anti-myc immunoprecipitates were immunoblotted with either an anti-MyD88 or an anti-TRAF6 antibody (Fig. 4C). Like wild-type IRAK1, both IRAK1-KD and IRAK1c associated with MyD88 and TRAF6, supporting previous data that the lack of IRAK1-KD and IRAK1c kinase activity does not affect association with MyD88 and TRAF6 at the receptor complex (10). Based upon these results, we hypothesized that because IRAK1c cannot be phosphorylated by IRAK4 and lacks kinase activity, it therefore cannot disengage from the receptor complex containing Tollip, MyD88, and TRAF6 and thus would fail to activate downstream signaling that results in NF-κB activation.

IRAK1c expression blocks NF-κB activation.

To initially test the effect of IRAK1c on TIR signaling, we used a luciferase reporter to assay NF-κB activation in 293 cells triggered by the inflammatory cytokine IL-1β. To confirm the validity of the assay, a MyD88 expression vector was transfected resulting in high levels of NF-κB activation, even in the absence of IL-1β stimulation (Fig. 5A). Transfection of IRAK1 or IRAK1-KD in 293 cells activated a NF-κB-luciferase reporter, and this activity was enhanced upon IL-1β stimulation (Fig. 5A). These results are consistent with previous reports (16) and suggest that kinase activity of IRAK1 may not be crucial for activation of downstream signaling pathways such as NF-κB. However, no NF-κB-luciferase activity was detected in IRAK1c-transfected cells (Fig. 5A); furthermore, NF-κB activity induced by IL-1β treatment was suppressed by IRAK1c overexpression. Similar effects were observed with the IRAK1c-KD-transfected cells. This negative effect of IRAK1c was dose dependent, as demonstrated in transfection studies with a titration of IRAK1c (Fig. 5C). To ensure that these observed effects were not cell line specific, the same experiment was repeated using G292 cells, an osteosarcoma cell line that is highly IL-1 responsive. Unlike 293 cells (data not shown), G292 cells are fully competent and respond to IL-1β stimulation by producing proinflammatory cytokines like IL-6. In G292 cells, IRAK1c also functioned as a potent negative regulator that blocked IL-1β-induced NF-κB activation in a dose-dependent manner (Fig. 5C). Interestingly, when we compared the ability of IRAK1, IRAK1-KD, IRAK1c, and IRAK1c-KD to modulate NF-κB-luciferase activity in G292 cells, we found that an IL-1-responsive cell line like G292 did not behave like 293 cells. Unlike the 293 cell system, overexpression of IRAK1 or IRAK1-KD did not result in NF-κB activation in the absence of IL-1 stimulation (Fig. 5A and B). Upon IL-1 stimulation, G292 cells transfected with IRAK1 showed enhanced NF-κB activity, but those transfected with IRAK1-KD did not show an overall increase compared to vector-transfected cells (Fig. 5B). These results suggested that the role of IRAK1 and IRAK1-KD can vary depending on the cell type used. However, in the case of IRAK1c, both cell types indicated that the distinct functional role of IRAK1c as a negative regulator is not merely due to the lack of kinase activity.

FIG. 5.

IRAK1c inhibits IL-1β-induced NF-κB activation. (A) 293 cells were transfected with 250 ng of the indicated myc-IRAK1 and FLAG-MyD88 plasmids and 1 μg of pHTS-NF-κB-luciferase plasmid. Cells were nonstimulated (−) or stimulated (+) with IL-1β (5 ng/ml) for 12 h, and cell extracts were analyzed for luciferase activity in replicates of four. The means and standard deviations are plotted. (B) G292 cells were transfected with the indicated myc-IRAK1 constructs and 2 μg of pHTS-NF-κB-luciferase plasmid. Cells were nonstimulated (−) or stimulated (+) with IL-1β (5 ng/ml) for 12 h, and cell extracts were analyzed for luciferase activity in replicates of five. (C) 293 cells (solid line) or G292 cells (dashed line) were transfected with increasing amounts of myc-IRAK1c, and NF-κB-luciferase activity was analyzed as described above. Percent activation compared to vector-transfected cells was calculated and plotted. (D) G292 cells were transfected with 250 ng of myc-IRAK1c and the indicated amounts of FLAG-TRAF6 plasmid. Cells were stimulated and analyzed as described above. (E) IRAK1^−/− fibroblasts reconstituted with IRAK1, IRAK1-KD, and IRAK1c were stimulated with 5 ng/ml of IL-1β for 30 min. Whole-cell lysates were prepared directly in sample buffer and subjected to immunoblot analysis with anti-IκBα (top panel), p-cJUN (middle panel), and anti-actin antibodies (bottom panel).

To determine if IRAK1c blocked the activation of the downstream effector TRAF6, we coexpressed TRAF6 with IRAK1c in 293 cells. Overexpression of TRAF6 attenuated the inhibitory effect of IRAK1c on NF-κB activation in a dose-dependent manner (Fig. 5D). These data suggested that IRAK1c functioned as an inhibitor of TIR signaling upstream of TRAF6 and did not affect signaling events downstream of TRAF6.

IRAK1c blocks IL-1β-induced NF-κB and MAPK activation.

Our previous findings using transfected 293 or G292 cells and a NF-κB-luciferase assay (Fig. 5A to D) suggested that IRAK1c blocked NF-κB activation. In order to specifically examine signaling events upstream of NF-κB activation, we stimulated IRAK1^−/− murine embryonic fibroblasts that had been reconstituted with IRAK1, IRAK1-KD, and IRAK1c and prepared lysates directly in sample buffer. Immunoblot analysis with an anti-IκBα antibody indicated that while both IRAK1 and IRAK1-KD could activate the NF-κB pathway as demonstrated by IκBα degradation, IRAK1c-reconstituted cells exhibited a block in this pathway (Fig. 5E, top panel). Similar results were obtained when we examined the mitogen-activated protein kinase (MAPK) pathway by assaying JNK activation using a phospho-specific antibody to the JNK substrate cJUN. While IRAK1 and IRAK1-KD cells stimulated with IL-1β showed robust p-cJUN, IRAK1c cells showed a dramatic block in cJUN phosphorylation (Fig. 5E, middle panel). Together, these data indicated that cells expressing IRAK1c could not activate the NF-κB or MAPK pathways, suggesting that IRAK1c might function as an inhibitor of TIR receptor-triggered inflammation.

IRAK1c blocks IL-1β and LPS-induced IL-6 and TNF-α production.

The most potent biological effect triggered upon TIR receptor activation and signaling through IRAK1 is the release of mediators of the inflammatory response such as IL-6 and TNF-α (1). To study the effects of IRAK1c on cytokine production, we tested the ability of IRAK1c to block the production of inflammatory cytokines. 293 cells, although commonly used to study the early biochemical events of TIR receptor and IRAK signal transduction, are not TIR responsive. Extensive analysis of 293 cells indicated that IL-1β stimulation of 293 cells did not lead to the production of inflammatory cytokines such as TNF-α or IL-6 or chemokines such as IL-8 (data not shown). However, G292 cells are sensitive to TIR receptor stimulation and produce both cytokines and chemokines upon IL-1β treatment. Using the stably transduced G292 cell lines, we examined the ability of IRAK1, IRAK1-KD, and IRAK1c to modulate production of the proinflammatory cytokine IL-6 upon IL-1β stimulation. Overexpression of IRAK1 resulted in a statistically significant increase in IL-6 production (Fig. 6A). Consistent with our NF-κB-luciferase data (Fig. 5B), IRAK1-KD overexpression did not enhance IL-6 production (Fig. 6A). Furthermore, IRAK1c functioned as an inhibitor that significantly decreased IL-6 production when compared to vector-transduced cells (P < 0.05).

FIG. 6.

IRAK1c inhibits IL-1β-induced IL-6 in G292 and reconstituted IRAK1^−/− cells and LPS- and CpG-induced TNF-α secretion in THP-1 cells. (A) G292 cells expressing the indicated myc-IRAK1 constructs were generated by retroviral transfection and either left unstimulated (−) or were stimulated (+) with 1 ng/ml of IL-1β for 6 h. Supernatants were analyzed for IL-6 by ELISA. The mean values and standard deviations of five replicates are plotted, and statistical significance was calculated based upon vector control values using a Student t test; an asterisk indicates a P value of <0.05. (B) The indicated THP-1-transfected cells were stimulated with 50 ng/ml of LPS for 4 h. TNF-α secretion was measured by ELISA and reported as described above. (C) The indicated THP-1-transfected cells were stimulated with 1 μM GpC control (unfilled bar) and stimulatory CpG oligodeoxynucleotides (filled bars) for 6 h. TNF-α secretion was measured by ELISA and reported as above. (D) IRAK1^−/− murine embryonic fibroblasts (KO) or stable cell lines reconstituted with IRAK1, IRAK1-KD (KO-KD), or IRAK1c (KO-1c) were stimulated together with wild-type (WT) cells using 1 ng/ml of IL-1β for 6 h. IL-6 secretion was measured by ELISA and reported as above. MEF, murine embryonic fibroblast.

Since IRAK1 functions as a signaling molecule downstream of all TIR family members, we also wanted to test the ability of IRAK1c to modulate Toll-like receptor (TLR) signaling. We therefore generated stably transduced cell lines using THP-1 monocytic cells to study the effect of IRAK1c in TLR signaling (6). Upon TLR4 stimulation with LPS, THP-1 cells secrete the inflammatory mediator TNF-α (Fig. 6B); TNF-α production is significantly decreased in THP-1 cells that overexpress myc-IRAK1c. This effect was also observed in THP-1 cells overexpressing an HA-tagged or untagged IRAK1c (data not shown). Similar results were obtained when THP-1 cells were stimulated with the TLR9 ligand, CpG (Fig. 6C). These data indicated that the IRAK1c splice variant not only blocked NF-κB activation but also significantly decreased production of inflammatory cytokines like TNF-α upon activation of several TIR family members. Consistent with our findings using G292 cells, IRAK1 enhanced LPS- and CpG-induced TNF-α production, while IRAK1-KD showed no change compared to vector-transfected cells (Fig. 6 B and C).

Since IRAK1 can homodimerize, the preceding experiments are performed in the presence of endogenous IRAK1. In order to obtain a clear understanding of the functional role of IRAK1c, we decided to use murine embryonic fibroblasts derived from wild-type and IRAK1^−/− mice (KO). Using KO cells reconstituted with IRAK1, IRAK1-KD, and IRAK1c, we examined the role of the various IRAK1 proteins by assaying IL-1β-induced IL-6. The genetic ablation of IRAK1 (KO) results in a significant decrease in IL-1β-induced IL-6 when compared to control wild-type cells (Fig. 6D). Reconstitution of the KO cells with either IRAK1 or IRAK1-KD rescues this defect, suggesting that kinase activity itself is not required for the role of IRAK1. However, reconstitution of these cells with IRAK1c did not rescue the defect in IL-6 production. These data suggested that although IRAK1c, like a point mutant IRAK1 (IRAK1-KD), lacks kinase activity, the missing 79 amino acids comprising exon 11 play a pivotal role in the function of IRAK1.

IRAK1c expression can be induced in human macrophages and dendritic cells.

Previous studies of TIR family signaling molecules have indicated that expression of regulatory proteins can be induced. A negative regulator of the IRAK kinase family, IRAKM, is specifically induced in macrophages upon LPS stimulation (13). A splice variant of the adapter protein MyD88 was also induced by LPS treatment in human monocytes (7). To test whether IRAK1c is inducible, we purified monocytes and prepared dendritic cells from peripheral human blood. Cells were then stimulated with LPS, and RNA was prepared for RT-PCR analyses of IRAK1 versus IRAK1c expression. Although dendritic cells express lower levels of IRAK1 when compared to macrophages, both macrophages and dendritic cells exhibited inducible expression of IRAK1c upon LPS stimulation (Fig. 7A). As has been previously reported, we also observed a slight decrease in IRAK1 expression upon LPS stimulation in macrophages (29). In order to confirm that these RNA changes were reflected by changes in protein levels, lysates were prepared from monocytes and dendritic cells stimulated for 12 h with LPS. IRAK1 immunoblot analysis indicated that IRAK1 protein is degraded upon activation, while IRAK1c protein is induced in both monocytes and dendritic cells (Fig. 7B). These data suggest that, together with inducible regulators like IRAKM, the splice variant IRAK1c can function as a regulator to suppress or fine-tune signaling downstream of TIRs.

FIG. 7.

IRAK1c is upregulated upon LPS treatment of human dendritic cells and monocytes. Human monocytes and dendritic cells were prepared from whole blood as described. (A) Cells were either unstimulated or stimulated with 100 ng/ml of LPS for 6 h. RNA was prepared, and IRAK1 cDNA was amplified as described in Materials and Methods. GAPDH amplification was used as a control. (B) Human monocytes and dendritic cells were stimulated for 16 h with 100 ng/ml of LPS, and cell lysates were prepared. Whole-cell lysates were subjected to anti-IRAK1 and anti-actin immunoblot analysis. DCs, dendritic cells.

DISCUSSION

The receptor-proximal IRAK family members, IRAK1 and IRAK4, are known to play a pivotal role in inflammation mediated by IL-1/IL-18/TLR ligands (1, 6). Due to the importance of these kinases in both the initiation and perpetuation of the inflammatory response, multiple mechanisms exist to regulate their functions. In this study, we report the identification of a novel IRAK1 splice variant, IRAK1c. IRAK1c expression, at both the RNA and protein levels, was confirmed in normal human tissues, primary cells, and cell lines. The presence of IRAK1c as a bona fide expressed transcript was supported by the existence of multiple IRAK1 ESTs and a full-length human mRNA clone, which were reported to be IRAK1 clones but whose sequences actually correspond to IRAK1c (data not shown). The IRAK1c expression profile is distinct from IRAK1, in that it is inducible in human monocytes and dendritic cells upon inflammatory stimuli such as LPS. Furthermore, IRAK1c is the predominant form of IRAK1 in the brain, suggesting that it might play a specific functional role in the central nervous system. In our studies, we demonstrate that IRAK1c functions as a dominant-negative protein in TIR-mediated signaling and inflammation. IRAKM, another member of the IRAK family, is known to be a negative modulator and is induced specifically in hematopoietic cells (13, 25). IRAK1c may function similarly to IRAKM but exerts a more general regulatory role in many cell types, including those of the hematopoietic lineage. Specifically in the brain, IRAK1c is the major IRAK1 isoform and potentially plays a protective role by modulating inflammatory responses. Since the brain is known to be an immune-privileged site, it remains of interest to determine if IRAK1c is also preferentially expressed in other immune-privileged sites such as the anterior chamber of the eye and the placenta.

Unlike two other reported IRAK1 splice variants that utilize the alternative splice acceptor sites of exon 12 (9, 28), the IRAK1c transcript that we identified and characterized lacks a region of 79 amino acids that are encoded by exon 11. However, this splicing does not cause any frameshift changes in the region encoded by the downstream exons 12, 13, and 14. The missing portion of IRAK1c is in the C-terminal region of the IRAK1 kinase domain. This region is highly enriched with charged amino acid residues and could potentially form an accessible binding interface. Although all of these IRAK1 variants lack kinase activity (9, 28), IRAK1c is distinct from the other variants in that it functions as an inhibitor of TIR-induced inflammation. Similar differential roles of splice variants have also been recently reported for murine IRAK2 (5).

In order to functionally characterize IRAK1c, we transfected IRAK1c in multiple cellular systems and studied the resulting effects on different TIR ligand-mediated signaling events and inflammatory cytokine induction. The 293 human embryonic kidney epithelial cell line is a convenient system utilized by many research groups to study the early biochemical events of TIR signal transduction. Using 293 cells transfected with a NF-κB-luciferase reporter, we showed that IRAK1c had a distinct inhibitory effect on NF-κB activation compared to IRAK1 and the kinase-dead point mutant IRAK1-KD. However, 293 cells are poorly responsive to TIR ligands and do not produce proinflammatory cytokines. We therefore turned to three other cell types that are competent in responding to TIR ligands. Using IL-1β-stimulated G292 osteosarcoma cells and LPS- or CpG-stimulated THP-1 monocytic cells, we observed that IRAK1c expression significantly suppressed NF-κB activation and the induction of inflammatory cytokines such as IL-6 and TNF-α. The dominant-negative effect of IRAK1c was upstream of TRAF6 as it could be rescued by TRAF6 overexpression. Furthermore, we generated fibroblast cell lines from IRAK1^−/− mice and wild-type controls. IRAK1^−/− fibroblasts are defective in IL-1-induced IL-6 production (12). Transfection of IRAK1^−/− cells with IRAK1 or IRAK1-KD could reconstitute the IL-1 response of these cells with activation of NF-κB and MAPK pathways and induction of IL-6. However, the splice variant IRAK1c could not rescue this defect, either in signaling or in cytokine production. These findings suggest that the naturally occurring IRAK1c splice variant, which lacks 79 amino acids and has no kinase activity, functions as an inhibitory protein and is distinct from the kinase-dead point mutant IRAK1-KD or an IRAK1 whole kinase domain deletion mutant that has been reported to behave similarly to IRAK1-KD (10). This result indicates that the lack of kinase activity alone is not the key factor for the negative regulatory role of IRAK1c. Other functional roles are apparently missing in the IRAK1c splice variant due to the deletion of the 79-amino-acid region.

Given the negative regulatory role of the IRAK1c splice variant, we attempted to dissect the mechanism by which it functioned. We showed that IRAK1c could dimerize with IRAK1 or IRAK2. IRAK1c associated with Tollip in resting cells and was recruited to the receptor complex upon activation. IRAK1c interacted with MyD88 and the key NF-κB-activating protein TRAF6 but had a decreased association with IRAK4. Furthermore, IRAK1c is not phosphorylated by IRAK4. More importantly, IRAK1c cannot terminate its association with Tollip at the receptor complex. Since the kinase-dead point mutant IRAK1-KD had similar defects in disengaging from Tollip yet did not block NF-κB activation or cytokine induction, it seems that the prolonged Tollip association cannot fully explain the distinct dominant-negative effect observed for IRAK1c but not for IRAK1-KD. It remains possible that other unidentified signaling proteins may interact with IRAK1 via exon 11, which is missing in IRAK1c. Alternatively, loss of the charged exon 11 might mask binding sites in IRAK1c and prevent formation of an optimal protein complex at the receptor. We therefore hypothesize that a change in the formation of a receptor complex mediated by IRAK1c could account for the functional defects observed upon expression of IRAK1c.

One key event upon IRAK1 recruitment to the receptor complex is the phosphorylation of IRAK1 by IRAK4 (1). Our studies indicated that while both IRAK1 and IRAK1-KD could serve as substrates for recombinant human IRAK4, IRAK1c could not be phosphorylated by IRAK4. This could be due to a loss of IRAK4 phosphorylation sites that are located in the region missing in IRAK1c. This possibility was excluded since this region expressed as a recombinant protein did not serve as a substrate for IRAK4. However, it is still possible that the lack of this region destabilizes the conformation of the IRAK1c kinase domain and thus masks the IRAK4 Thr209 phosphorylation site. Furthermore, the decreased IRAK1 and IRAK4 association that we observed may also contribute to this defect. In conclusion, the striking feature of IRAK1c that we identified is that it can no longer be phosphorylated by IRAK4.

Previous reports have indicated that IRAK1 forms homodimers and heterodimers with other IRAK family members such as IRAK2 (25). IRAK1 dimers are associated with the adapter protein Tollip in the cytosol at resting state (2, 10). Upon activation, Tollip-IRAK1 is recruited to the TIR receptor complex and interacts with the key adapter protein MyD88 and the NF-κB-activating protein TRAF6 (10). At the receptor, IRAK1 dimers are phosphorylated at Thr209 by IRAK4, resulting in a conformational change of the kinase domain (14). This leads to phosphorylation of Thr387 in the activation loop of IRAK1 and triggers autophosphorylation on several residues in the ProST region located between the death domain and kinase domain (14). When IRAK1 is hyperphosphorylated, it disassociates from the receptor protein complex of MyD88 and Tollip (3, 10). IRAK1 continues to associate with TRAF6, eventually triggering activation of the NF-κB pathway. Based on our biochemical studies, we proposed a model to explain the mechanism by which IRAK1c exerts its dominant-negative effects on TIR signaling (Fig. 8). In resting cells, IRAK1c dimerizes with IRAK1 or IRAK2 and associates with Tollip. Upon TIR activation, IRAK1c is recruited to the receptor complex and interacts with MyD88 and TRAF6 in a manner similar to IRAK1 and IRAK1-KD. However, unlike IRAK1, IRAK1c cannot be phosphorylated by IRAK4 and does not have kinase activity to undergo autophosphorylation. Furthermore, IRAK1c does not terminate its association with Tollip at the receptor complex, a prerequisite for further activation of the downstream NF-κB and MAPK signaling cascades. Due to the functional defects of IRAK1c, it eventually leads to shutdown of TIR-mediated signaling and a decrease in inflammatory mediator production.

FIG. 8.

A model of IRAK1c regulation of inflammation. Upon TIR activation by the appropriate ligand, the Tollip-IRAK1 complex is recruited to the TIR receptor at the membrane and interacts with MyD88 and TRAF6. IRAK4 phosphorylates IRAK1 (1) resulting in IRAK1 autophosphorylation and activation (2). IRAK1 then disengages from the receptor complex of MyD88 and Tollip (3) but remains associated with TRAF6, resulting in activation of the NF-κB pathway and production of proinflammatory cytokines (4). The splice variant IRAK1c cannot be phosphorylated by IRAK4, does not undergo autophosphorylation, and thus does not activate downstream signaling events.

The hypothesis that cells could selectively deplete IRAK1 and replace it with a functionally defective IRAK1c is supported by our findings that both human macrophages and myeloid-derived dendritic cells showed a decrease in IRAK1 and a concurrent increase in IRAK1c message upon TIR activation. This strategy of regulated expression of splicing isoforms to produce either functional or defective protein kinase provides an elegant mechanism to attenuate activation signals emanating from a receptor-proximal kinase, a function most often performed by a protein phosphatase. As yet, no phosphatase has been described that dephosphorylates IRAK1 or IRAK4. Interestingly, a splice variant of p38α, another key kinase in the inflammatory pathway, was recently reported to down-regulate the NFκB pathway (26).

Our studies have highlighted that alternative splicing is one of the mechanisms utilized in cells to regulate the proinflammatory signaling cascade triggered by TIRs. The key adapter protein in TIR signaling, MyD88, has been reported to generate an alternative splice variant, MyD88s, which negatively regulated NF-κB activation (8). Expression of splice variants seems to be a strategy used by the IRAK family to regulate TIR response. Two splice variants of IRAK1 have been reported, although endogenous protein expression of these variants has yet to be demonstrated (9, 28). Furthermore, a recent paper reported four alternatively spliced variants of the murine IRAK2 gene (5). Interestingly, splice variants of the IRAK family seem to be species specific, probably reflecting a mechanism established late in evolution. All IRAK2 splice variants are found only in the mouse (5). IRAK1s is only found in the mouse, while IRAK1b is found both in mouse and human (9, 28). To date, we have been unable to clone an IRAK1c splice variant from mouse tissues or cell lines. These differences in splice variants between mouse and humans, together with other differences in expression of TLRs themselves, suggest that multiple differences in the TLR system, including both receptors and signaling molecules, contribute to the differences observed between the innate immune systems of mice and humans (18).