IRAK1b, a Novel Alternative Splice Variant of Interleukin-1 Receptor-associated Kinase (IRAK), Mediates Interleukin-1 Signaling and Has Prolonged Stability*

Interleukin-1 (IL-1) is a pleiotropic cytokine essential for initiation of the immune response to infections and stress. IL-1 interacts with its type I receptor (IL-1RI) and triggers a number of intracellular signaling cascades leading to activation of transcription factors, transcrip-tional up-regulation of target genes, and mRNA stabilization. IL-1RI-associated kinase-1 (IRAK1) is a membrane proximal serine-threonine kinase involved in IL-1 signaling that becomes phosphorylated and progressively degraded in response to IL-1 induction. We have identified a novel variant of IRAK1, which we have named IRAK1b, that arises from the use of an alternative 5 * -acceptor splice site defined by sequence within exon 12 of IRAK1 . IRAK1b mRNA exhibits wide tissue expression and is evolutionarily conserved in both mouse and human. IRAK1b can activate the transcription factor nuclear factor k B and interacts with the IL-1 signaling factors Toll-interacting protein and tumor necrosis factor receptor-associated factor 6. It forms homodimers and heterodimers with the previously described isoform of IRAK1. We show that the IRAK1b protein is kinase-inactive and that, unlike IRAK1, its levels remain constant after IL-1 induction. The presence of an alternative splice variant of IRAK1, which is functionally active and highly stable following IL-1 stimulation, adds further complexity to the control mechanisms that govern IL-1 signaling. and Expression Vectors— The Renilla -luciferase and NF- k B-responsive luciferase reporter constructs are described elsewhere (29). cDNAs were amplified using the GC-RICH PCR system according to the manufacturer’s instructions. Primers for amplification of the coding sequence of IRAK1 were derived from GenBank y accession number L76191. The coding sequences of IRAK1 variants were cloned into the mammalian expression vectors pCI-Neo (Promega), pcDNA4/HisMax (Invitrogen, Carlsbad, CA), and p3XFLAG-CMV-14 (Sigma Chemical Co.). A kinase-inactive IRAK1a (kiIRAK1a) clone was generated con-taining two point mutations (E248A and I326V) in the kinase domain. IRAK1a mutants IRAK1a(S536A), IRAK1a(S541A), IRAK1a(S536A, S541A), IRAK1a(S536A,S541A,Y515F), IRAK1a(K520A), and IRAK1a-(K520A,S536A,S541A) were generated using overlapping primers con- taining the appropriate point mutations and PCR amplification of fragments of the IRAK1a coding sequence. PCR products were cloned into previously generated IRAK1a expression constructs. Primers for amplification of Tollip and TRAF6 coding sequences were derived from GenBank y accession numbers AJ242972 and U78798, respectively. cDNAs were cloned into pcDNA4/HisMax. The expression vector p3xFLAG-CMV-BAP encoding FLAG-tagged bacterial alkaline phosphatase (BAP) was obtained from Sigma. Western Blotting and Immunoprecipitation— Total proteins were extracted on ice in 50 m M HEPES, 150 m M NaCl, 20 m M b -glycerophos- phate, 1 m M EDTA, 1 m M benzamidine, 50 m M NaF, and 1 m M Na 3 VO 4 , 5 m M para-nitrophenyl

Infections, tissue injury, and/or stress trigger monocytes and macrophages to produce interleukin-1 (IL-1), 1 the cytokine that orchestrates much of the systemic acute phase response, the net effect of which is to neutralize the underlying physiological challenge (reviewed in Ref. 1). Systemically, IL-1 stimulates fever, vasodilation, and muscle contractions; at the cellular level it induces cells to adopt an enhanced "host defense" phenotype by eliciting radical changes in their protein expression profiles by modulating mRNA processing and stability, protein translation, and gene transcription (1). The latter is achieved by activating transcription factors, e.g. nuclear factor B (NF-B) and activator protein-1 (AP-1). Signaling by IL-1 depends on its engagement of the transmembrane IL-1 receptor type I (IL-1RI) and IL-1RI accessory protein (IL-1RAcP) (Ref. 2 and references therein). Binding of IL-1 to the extracellular domains of IL-1RI and IL-1RAcP is followed by recruitment of the intracellular adapter protein myeloid differentiation factor (MyD88) (3)(4)(5)(6) and a number of kinases to the evolving IL-1⅐IL-1RI⅐IL-1RAcP complex. Depending on cell type, these kinases may include IL-1RI-associated kinase-1 (IRAK1), which is expressed in all tissues (7), IRAK2, which has a narrower cellular distribution (3), and IRAK-M, which is mainly restricted to cells of myeloid origin (8). Phosphatidylinositol 3-kinase has also been implicated as an early component of the IL-1 signaling cascade (9). Once established, the cascade progresses through the stepwise activation/recruitment of several additional intermediates, including tumor necrosis factor (TNF) receptor-associated factor 6 (TRAF6), TRAF6 binding protein, transforming growth factor-␤-activated kinase 1 (TAK1), two TAK1 binding proteins (TAB1 and TAB2), and NF-B-inducing kinase (10 -15). Finally, the inhibitor of NF-B (IB) is phosphorylated by the resulting IB kinase complex and degraded. This allows NF-B to translocate to the nucleus where it activates transcription of a wide range of genes that are important for immune function and inflammation (16). The branches of the IL-1 signaling pathway that leads to activation of AP-1 and mRNA stabilization may diverge from the NF-B-activating branch of the cascade at TAK1/TAB1 (Ref. 17 and references therein). However, other studies suggest that the pathways diverge at IRAK1 or even earlier (18,19).
IRAK1 is a protein of 714 amino acids that has two known functional domains: an N-terminal death domain, which is involved in protein⅐protein interactions with MyD88 and Tollinteracting protein (Tollip), and a centrally positioned Ser/Thr kinase domain (3,(5)(6)(7)20). Several studies have shown that the kinase activity is not necessary for IRAK1 to be functional (21)(22)(23). In un-stimulated cells IRAK1 is associated with Tollip; however, following stimulation with IL-1, its recruitment to the IL-1⅐IL-1RI⅐IL-1RAcP complex is facilitated by the interaction between Tollip and IL-1RAcP (24). IRAK1 becomes phosphorylated, dissociates from Tollip, and is degraded by proteasomes (24,25). Studies to date indicate that the phosphorylation of IRAK1 is mediated by IRAK1 itself (7,20), although it remains a possibility that additional kinases are also involved * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF346607.
Alternative splicing is one of the mechanisms whereby an increased complexity in the number of functionally distinct protein products may be generated from a fixed pool of genes in the genome (27). We have identified a novel alternatively spliced variant of IRAK1 mRNA. It encodes an IRAK1 protein that has functional activity similar to that of the previously described IRAK1 isoform but is biologically distinct in that it is relatively much more stable after IL-1 activation. The existence of alternative forms of IRAK1 with very different capacities to support IL-1 signaling over time has important implications for the initiation and maintenance of inflammatory processes.

EXPERIMENTAL PROCEDURES
Cell Lines and Cultures-Human hepatoma cells (HepG2) and human embryonic kidney cells (293) were obtained from the American Type Culture Collection and maintained in Dulbecco's modified Eagle's medium with 25 mM HEPES and glutamax-1 (L-alanyl-Lglutamine) supplemented with 10% (v/v) fetal calf serum, 1 mM sodium pyruvate, 0.01 mM nonessential amino acids, and 50 g/ml gentamicin (Life Technologies, Inc., Grand Island, NY). For protein harvesting 10 6 cells were used per data point. For transfection experiments 2.5 ϫ 10 5 cells were used per data point. Treatments were performed in duplicate or triplicate. All experiments were repeated at least twice. For transfection experiments, cells were grown to ϳ50% confluence and transfected with 2-3 g of total DNA using FuGene6 (Roche Molecular Biochemicals, Indianapolis, IN) according to the manufacturer's instructions. In experiments where cells were transfected with varying amounts of expression vectors, the total DNA used in each transfection was held constant by co-transfecting appropriate amounts of empty vector. Cells were treated with medium (control) or IL-1␤ (10 ng/ml, National Cancer Institute, Frederick, MD) for 4 h. For luciferase reporter assays, cells were lysed and lysates were assayed for luciferase and Renilla-luciferase activity according to the manufacturer's instructions (Dual-Luciferase Reporter Assay System, Promega, Madison, WI).
RNA and Proportional Quantitative RT-PCR-Human tissue total RNA was obtained from CLONTECH Laboratories (Palo Alto, CA). Mouse total RNA was extracted using RNeasy (Qiagen, Santa Clarita, CA) according to the manufacturer's instructions. RNA from tissue culture cells was extracted as described elsewhere (28). Reverse transcription of 1 g of total RNA was performed at 42°C using avian myeloblastosis virus reverse transcriptase (Promega, Madison, WI), oligo-(dN) 6 primer (Amersham Pharmacia Biotech, Piscataway, NJ) and ANTI-RNase RNase inhibitor (Ambion Inc., Austin, TX). Proportional quantitative RT-PCR was performed as described elsewhere (28) using the GC-RICH PCR system (Roche Molecular Biochemicals) with forward (5Ј-AAAGGAGGCCTCCTATGACC-3Ј) and reverse (5Ј-ATGAT-GCAGAGCTG-3Ј) primers based on sequence from GenBank accession number L76191. PCR products were separated in agarose gels and visualized after ethidium-bromide staining. Mouse cDNAs were amplified with forward (5Ј-AGAAGAGGCCCCCCATGACC-3Ј) and reverse (5Ј-CAGGGATGAGCTGCCCGTGG-3Ј) primers based on sequence from GenBank accession number U56773.
Pulse Chase-Transfected cell monolayers were rinsed twice with PBS and incubated with methionine-deficient Dulbecco's modified Eagle's medium for 1 h prior to addition of 15 Ci of [ 35 S]methionine (Amersham Pharmacia Biotech). After labeling for 1 h, medium was removed and cells were rinsed twice with PBS prior to the addition of methionine-sufficient Dulbecco's modified Eagle's medium with or without 10 ng/ml IL-1␤. At appropriate times cells were lysed in the presence of 50 mM 4-hydroxy-5-iodo-3-nitrophenylacetyl-Leu-Leu-Leu-vinylsulfone (NLVS, proteasome inhibitor, Calbiochem, San Diego, CA), and proteins were immunoprecipitated with ANTI-FLAG M2 as described above. Following SDS-PAGE and transfer to nitrocellulose, 35 S-labeled proteins were quantified using STORM and ImageQuaNT technologies (Molecular Dynamics Inc., Sunnyvale, CA).
Computer Analyses-Computer analyses were performed through the Baylor College of Medicine search launcher available at searchlauncher.bcm.tmc.edu. The programs HSPL (prediction of splice sites in human DNA sequences) and FEXH (prediction of internal, 5Јand 3Ј-exons in human DNA sequences) were used to predict splice sites (30) in the IRAK1 sequence.

Identification of Potential Alternatively Spliced cDNA
Clones-IRAK1 is an essential component of the IL-1 signaling cascade. It has been suggested that degradation of IRAK1 is responsible for desensitization of cellular targets to IL-1 and lipopolysaccharide after prolonged exposure to these agents (25,26). To test this hypothesis we generated a range of IRAK1 expression vectors. Primers for amplification of IRAK1 sequences were designed from GenBank accession number L76191, and the entire coding sequence was amplified by PCR using proofreading DNA polymerases. The re-sulting cDNA was cloned into the mammalian expression vector pCI-Neo. Sequencing of four individual clones revealed that two contained an identical deletion of 90 bp (sequence was submitted to GenBank under accession number AF346607). To determine whether these clones represent an alternatively spliced variant of IRAK1 mRNA, we aligned their sequences with that of the known IRAK1 gene sequence (GenBank accession number U52112). The 90 bp, which were missing in two of our clones, aligned perfectly with the first 90 bp of IRAK1 exon 12 (Fig. 1A). We then performed computer analysis to predict the likely locations of the exon splice sites involving IRAK1 exons 11-13. The program HSPL (prediction of splice sites in human DNA sequences) predicted one 5Ј-acceptor site at position ϩ1 of exon 12 and a second 5Ј-acceptor site at position ϩ91 of exon 12 (Fig. 1A). The FEXH (prediction of internal, 5Ј-and 3Ј-exons in human DNA sequences) program only predicted the 5Ј-acceptor site at position ϩ91.
These data strongly support the possibility of alternative splicing in exon 12 of the IRAK1 gene. Alternative splicing of exon 12 at position ϩ1 or ϩ91 would result in the previously described form of IRAK1 (hereafter referred to as IRAK1a for clarity) and a shorter form, respectively. We have named this shorter putative splice variant IRAK1b. IRAK1 will be used below as a generic term covering both splice variants. The IRAK1b sequence predicts an in-frame deletion of 30 amino acids (residues 514 -543) at the C-terminal end of the kinase domain (Fig. 1B).
Expression of Alternatively Spliced Variants of Human IRAK1-IRAK1a has previously been shown to be expressed in a wide range of cell types as an ϳ3.5-kb mRNA (7,20). The similarity in size between the IRAK1a and IRAK1b mRNAs precludes their resolution by Northern blot analysis. We therefore chose a RT-PCR approach to identify cell types that express the IRAK1b mRNA. We designed a forward primer corresponding to sequence within IRAK1 exon 11 and a reverse primer corresponding to sequence within IRAK1 exon 12 downstream of the putative alternative splice site. To test the relative amplification efficiencies of the two IRAK1 cDNAs derived from the alternatively spliced IRAK1 mRNA species, plasmids carrying IRAK1a and IRAK1b coding sequence were co-amplified by PCR, and multiple aliquots were taken for analysis after each of five cycles. Equivalent amplification kinetics during the exponential phase of IRAK1a cDNA and IRAK1b cDNA amplification was confirmed; the ratio of products was also maintained after the plateau phase of amplification had been reached (not shown). Reverse-transcribed total RNA from a range of human tissues was subjected to the above proportional quantitative PCR method. A strong band of 441 bp corresponding to the predicted size of IRAK1a cDNA and a weaker band of ϳ350 bp were detected in all samples ( Fig. 2A). Because the expected size of the IRAK1b cDNA is 351 bp, we concluded that IRAK1b mRNA is co-expressed with IRAK1a mRNA in all tissues tested. The ratio of cDNA products derived from the IRAK1a to IRAK1b mRNAs was similar in all samples.
We next wished to establish whether the putative IRAK1b protein could be detected in protein extracts from HepG2 cells. Cells were transfected with pCI-Neo expression plasmids encoding either IRAK1a or IRAK1b. Because the former becomes hyperphosphorylated when overexpressed and consequently undergoes a shift in electrophoretic mobility (5,7,8,20) (see below), we used a kinase-inactive variant (kiIRAK1a) for this transfection series. Proteins were extracted from both transfected and untransfected HepG2 cells and IRAK1 proteins detected by immunoblotting. Single bands of 76 and 73 kDa were detected in cell extracts from the kiIRAK1a-and IRAK1b-transfected cells, respectively (Fig. 2B). These sizes are similar to those predicted for the two isoforms based on their polypeptide sequences. Two bands with mobilities identical to those observed in cells transfected with kiIRAK1a and IRAK1b were also observed in extracts from untransfected cells (Fig. 2B), suggesting that HepG2 cells endogenously synthesize both IRAK1a (upper band) and IRAK1b (lower band) proteins. The above results indicate that IRAK1b is an alternatively spliced variant of IRAK1 that is expressed at the mRNA level, and probably at the protein level, in a wide range of tissues.
Evolutionary Conservation of IRAK1b-The mouse IRAK1a mRNA sequence (IRAK1/mPLK, GenBank accession number FIG. 1. Exon 12 of IRAK1 and position of alternative splice sites. A, exon 12 of the human IRAK1 gene (GenBank accession number U52112) is shown in uppercase letters and segments of surrounding intron sequence are shown in lowercase letters. The nucleotide sequence, which is not present in IRAK1b mRNA, is underlined. B, schematic representation of IRAK1a polypeptide. The amino acid sequence, which is not present in IRAK1b protein, is shown above the diagram of the IRAK1 domain organization and its relative location is indicated. The positions of the two serine residues (ٙ), and the single lysine and tyrosine residues (*) within this sequence are indicated. The tripeptide that is absent from mouse IRAK1a is underlined. U56773) was aligned with that of the human IRAK1 mRNA. The two splice sites for IRAK1a and IRAK1b are perfectly conserved in the mouse sequence (not shown). Intriguingly, three amino acid residues, which are present in the region of the human IRAK1 that is spliced out of human IRAK1b, are absent from the mouse gene (Fig. 1B). A proportional quantitative RT-PCR method similar to the one described above for human IRAK1a and IRAK1b mRNA splice variants was developed for the mouse putative IRAK1 mRNA splice variants. Both cDNA products could be amplified from total RNA extracted from mouse liver, kidney, and testis (Fig. 2C). The retention of the IRAK1b alternative splice variant in two evolutionarily distant mammalian species suggests that it has an important physiological function. PELLE, the Drosophila homologue of IRAK1, contains death and kinase domains that are similar to those present in IRAK1; however, the PELLE peptide sequence ends immediately after the kinase domain and the full-length protein is only 501 amino acid residues long (31). PELLE therefore does not contain the C-terminal region, which is subject to alternative splicing in mammalian IRAK1. The presence of this domain in both human and mouse IRAK1 suggests that it serves functions, as yet undefined, that are specific to mammals. In this context it is intriguing that the C-terminal domain incorporates potential further structural diversity via the use of alternative splicing to yield distinct protein products that may themselves have differentiated functions.
Activation of NF-B by IRAK1 Splice Variants-To determine if IRAK1b is capable of activating NF-B, HepG2 cells were co-transfected with an NF-B luciferase reporter, a control luciferase-Renilla reporter, and increasing amounts of IRAK1a or IRAK1b expression constructs. The level of NF-B activation was evaluated after 24 h by assessment of luciferase/ Renilla reporter ratios. Both IRAK1a and IRAK1b result in a concentration-dependent activation of NF-B (Fig. 3). IRAK1a appears to be more efficient at activating NF-B than IRAK1b, because ϳ20 times less IRAK1a than IRAK1b expression construct is needed to induce equivalent levels of NF-B activation. The proportional difference in NF-B activation between IRAK1a and IRAK1b is similar to that previously reported between IRAK1a and either IRAK2 or IRAK-M (8), i.e. IRAK1b appears to be as efficient as IRAK2 and IRAK-M at activating NF-B.
Protein⅐Protein Interaction with IRAK1 Splice Variants-The lower efficiency of NF-B activation by IRAK1b relative to that effected by IRAK1a prompted us to investigate if there are measurable differences in the interactions of IRAK1a and IRAK1b with the downstream signaling component TRAF6. His-tagged IRAK1 proteins were expressed either alone or with His-tagged TRAF6 in 293 cells. TRAF6 was immunoprecipitated from cellular extracts with a TRAF6-specific antibody, and co-immunoprecipitated proteins were separated by SDS-PAGE and analyzed by Western blotting. All three IRAK1 proteins, IRAK1b, kiIRAK1a, and IRAK1a co-precipitated equally well with TRAF6 (Fig. 4A). Neither of the IRAK1 proteins could be immunoprecipitated with anti-TRAF6 in the absence of co-transfected TRAF6 (not shown) verifying that direct interaction with TRAF6 is necessary for co-immunoprecipitation. Furthermore, His-tagged Tollip co-expressed with TRAF6 could not be co-immunoprecipitated with TRAF6, confirming that the IRAK1⅐TRAF6 interaction is specific (Fig. 4A).
It has been shown that IRAK1a forms homodimers and heterodimers with IRAK2 and IRAK-M (8). We therefore next examined if IRAK1b would form homo-and heterodimers with itself and IRAK1a, respectively. Cells were transfected with expression vectors for the FLAG-tagged IRAK1 proteins IRAK1b, kiIRAK1a, or IRAK1a or the FLAG-tagged BAP protein (a control that is not involved in IL-1 signaling). In addition all cells were transfected with an expression vector encoding His-tagged IRAK1b. FLAG-tagged proteins were immunoprecipitated using anti-FLAG antibodies. Immunoprecipitated and co-immunoprecipitated proteins were separated by SDS-PAGE and analyzed by Western blotting. His-tagged IRAK1b was detected using anti-His antibodies and was found to co-immunoprecipitate equally well with all the FLAG-tagged IRAK1 proteins, i.e. IRAK1b, kiIRAK1a, and IRAK1a (Fig. 4B), establishing that IRAK1b can form both homodimers and heterodimers with IRAK1a. His-tagged IRAK1b did not co-immunoprecipitate with FLAG-tagged BAP (Fig. 4B), demonstrating that the generation of all of the above classes of dimers is specific. FLAG-tagged immunoprecipitated proteins were detected with anti-FLAG to verify efficient precipitation (Fig. 4B).
To investigate if IRAK1b interacts with signaling components upstream of IRAK1 in the IL-1 signaling cascade, we expressed His-tagged Tollip with FLAG-BAP, FLAG-IRAK1b, FLAG-kiIRAK1a, or FLAG-IRAK1a. FLAG-tagged proteins were immunoprecipitated, and proteins were analyzed as described above. His-Tollip co-immunoprecipitated only with FLAG-tagged IRAK1b and kiIRAK1a and not with FLAGtagged BAP or IRAK1a. The latter is consistent with the previous observation by Burns et al. (24) that IRAK1a dissociates from Tollip when phosphorylated.
Tollip is the factor that is associated with IRAK1 before IL-1 treatment and is responsible for recruiting IRAK1 to the IL-1⅐IL-1RI⅐IL-1RAcP complex. The ability of Tollip to specifically co-immunoprecipitate with IRAK1b suggests that IRAK1b is recruited to the IL-1⅐IL-1RI⅐IL-1RAcP complex in response to IL-1 stimulation and provides further support for our conclusion that IRAK1b is involved in IL-1 signaling.
Phosphorylation and Degradation of IRAK1 Splice Vari- ants-IRAK1a undergoes autophosphorylation in response to IL-1, and when overexpressed in the absence of cytokine (5,7,8,20,25). Phosphorylated IRAK1a exhibits slower mobility in PAGE. It also has been shown that phosphorylated IRAK1a is degraded by the ubiquitin-proteasome pathway. We observed that, when overexpressed, IRAK1a migrates with an apparent molecular mass of ϳ110 -120 kDa (Fig. 5A), whereas both the endogenous unphosphorylated IRAK1a and the transfected kiIRAK1a mutant migrate at 76 kDa, as would be expected of the unmodified polypeptide. Unlike IRAK1a, IRAK1b migrates at the expected 73-kDa position when overexpressed in HepG2 cells (Fig. 5A), indicating that IRAK1b does not undergo autophosphorylation. To investigate the consequences of IL-1 induction on the endogenous IRAK1 isoforms, HepG2 cells were treated with IL-1 for various lengths of time prior to Western blot analysis of IRAK1 proteins. As previously described in 293 and MRC-5 cells (7, 25) the 76-kDa IRAK1a band gradually disappeared as bands of slower mobility appeared (Fig. 5B). At least four phosphorylated forms of IRAK1a with apparent molecular masses of 79, 83, 90, and 110 -120 kDa appear. No changes in intensity of the 73-kDa band were observed (Fig.   5B), suggesting that, unlike IRAK1a, IRAK1b does not become phosphorylated and degraded in response to IL-1. In unstimulated cells, and at the earliest time points after cytokine induction, IRAK1a is the dominant IRAK1 isoform; however, at later time points post-stimulus the levels of the two unphosphorylated isoforms are almost equimolar. After prolonged exposure (24 h) of cells to IL-1, levels of IRAK1b were as high as, and in some experiments higher than, those of IRAK1a (Fig. 5C).
Half-lives of IRAK1 Proteins-It has been reported that in MRC-5 cells the 76-kDa non-phosphorylated IRAK1a band has a t1 ⁄2 of only about 2 min in response to IL-1 due to phosphorylation and degradation (25). We used ImageQuaNT technology applied to Western blots (Fig. 5B) to quantify the amount of non-phosphorylated 76-kDa IRAK1a protein remaining after IL-1 treatment. The turnover of non-phosphorylated IRAK1a appears to be biphasic. Immediately following IL-1 treatment it has a t1 ⁄2 of ϳ15 min (Fig. 6A). Shortly thereafter there is an apparent transition to a t1 ⁄2 value of a least 2 h (Fig. 6A). The longer half-life of IRAK1a in the period immediately after IL-1 treatment in HepG2 compared with that observed in MRC-5 cells may reflect tissue-specific differences in the rate of signal factor recruitment and/or differences in levels of catalytic enzymes.
To further investigate the half-lives of IRAK1a and IRAK1b, we performed pulse-chase experiments. FLAG-IRAK1a, FLAG-kiIRAK1a, or FLAG-IRAK1b were expressed in HepG2 cells and labeled with [ 35 S]methionine. Cells were treated with medium only or IL-1 and the 35 S-labeled pro- FIG. 4. Co-immunoprecipitation of IRAK1 isoforms with IL-1 signaling factors. A, His-tagged Tollip, IRAK1b, kiIRAK1a, or IRAK1a were co-expressed with His-TRAF6 in 293 cells (see "Experimental Procedures"). Total cellular lysates were harvested after 24 h, and proteins immunoprecipitated (IP) with anti-TRAF6 antibodies. Immunoprecipitates were analyzed using Western blotting (WB). Histagged proteins were detected using anti-His antibodies. Two protein bands not related to IL-1 signaling were detected in total cellular extracts (*). B, FLAG-tagged bacterial alkaline phosphatase (BAP), IRAK1b, kiIRAK1a, or IRAK1a were co-expressed with His-tagged IRAK1b in 293 cells. Immunoprecipitation was performed using anti-FLAG antibodies. His-tagged IRAK1b and FLAG-tagged proteins were detected in Western blots using anti-His and anti-FLAG antibodies, respectively. C, FLAG-tagged BAP, IRAK1b, kiIRAK1a, or IRAK1a were co-expressed with His-tagged Tollip in 293 cells. Immunoprecipitation and Western blotting was performed as in B. Tollip was detected using anti-His antibodies.

FIG. 5. Phosphorylation of alternatively spliced variants of IRAK1.
A, IRAK1a, kiIRAK1a, or IRAK1b were expressed in HepG2 cells. Total cellular extracts were harvested, and electrophoretic mobility of IRAK1 proteins was analyzed using Western blotting and anti-IRAK1 antibodies. IRAK1 protein extracts were run next to an extract from mock transfected cells. B and C, HepG2 cells were treated with 10 ng/ml IL-1, and total cellular protein extracts were made at the indicated time points. Levels of IRAK1 protein were determined using Western blotting. teins chased for up to 8 h. Immunoprecipitation of the FLAGtagged proteins allowed us to examine one isoform at a time. IRAK1a, which when overexpressed migrates at 110 -120 kDa (i.e. equivalent to the largest hyperphosphorylated IRAK1 endogenous species observed following IL-1 treatment of non-transfected cells as depicted in Fig. 5B), has a t1 ⁄2 of ϳ2.5 h in the absence of IL-1 (PP-IRAK1a/Med, Fig. 6A) and a t1 ⁄2 of ϳ6 min (PP-IRAK1a/IL1, Fig. 6A) in the presence of IL-1. These data are in agreement with the observed high levels of hyperphosphorylated IRAK1a when overexpressed in the absence of IL-1 (Fig. 5A) and the much lower level of accumulation of the endogenous hyperphosphorylated IRAK1a at later time points following IL-1 treatment (Fig. 5B).
To determine the half-life of the non-phosphorylated IRAK1a in the absence of IL-1, we used the kinase-inactive mutant kiIRAK1a as a surrogate to allow us to overexpress FLAGtagged protein without autophosphorylation (Fig. 5A). In pulsechase experiments kiIRAK1a has a t1 ⁄2 of ϳ7 h in the absence of IL-1 (kiIRAK1a/Med, Fig. 6B) and an essentially similar t1 ⁄2 of 6 h in the presence of IL-1 (kiIRAK1a/IL1, Fig. 6B).
In parallel experiments IRAK1b has similar half-lives of 7 and 6 h in the absence and presence of IL-1, respectively (Fig.  6C). These results establish that, although in the absence of IL-1 IRAK1a and IRAK1b have almost identical half-lives, the turnover rate of IRAK1b is essentially unchanged in response to IL-1 whereas IRAK1a becomes dramatically destabilized due to phosphorylation.
It is possible that the differential stability of the two IRAK1 isoforms in response to IL-1 stimulation reflects fundamental differences in "functional kinetics." Because IRAK1a is the dominant isoform in unstimulated cells it is likely to be the isoform that is predominantly recruited to the IL-1⅐IL-1RI⅐IL-1RAcP complex during the initial stages of assembly. The strong activity and relative abundance of IRAK1a would result in the rapid attainment of the high levels of NF-B mobilization that are desirable early in the acute phase response. However, to prevent chronic "overstimulation" of NF-kB-mediated gene expression in the medium to long term, it would be necessary to efficiently degrade IRAK1a. At such later stages of the acute phase response the replacement of IRAK1a with IRAK1b would facilitate a continued but less vigorous input into the IL-1 signaling cascade thereby allowing NF-B activation and other downstream components to be maintained at a more modest and sustainable level. At the terminal stages of the acute phase response, additional control mechanisms may be implemented to ensure that the IL-1 signaling pathway is finally shut down.
An alternative or additional possibility is that IRAK1a and IRAK1b play different roles in supporting the various downstream branches of the IL-1 signaling cascade, e.g. activation of NF-B, activation of AP-1, and mRNA stabilization. Their different half-lives following IL-1 activation may reflect different requirements for initiating and maintaining subpathways with particular subsets of target genes for specific amounts of time under a range of static and dynamically changing induction conditions.
Phosphorylation of IRAK1a Mutants-Two potential serine phosphorylation sites (Ser-536 and Ser-541), one potential tyrosine phosphorylation site (Tyr-515), and one potential ubiquitination site (Lys-520) are located within the peptide sequence that is present only in IRAK1a (Fig. 1B). To determine whether these sites are involved in hyperphosphorylation and hence destabilization of IRAK1a, we generated mutants that were deficient in one or more of these sites. To minimize steric changes serine residues were replaced with alanines, tyrosine with phenylalanine, and lysine with alanine. The IRAK1a mutants were overexpressed in cells following transient transfection, and protein extracts were analyzed by Western blotting. All mutant proteins (i.e. IRA-K1a(S536A), IRAK1a(S541A), IRAK1a(S536A,S541A), IRA-K1a(K520A), and IRAK1a(K520A,S536A,S541A)) migrated in SDS-PAGE at the same 110-to 120-kDa position as wild type-hyperphosphorylated (and possibly ubiquitinated) IRA-K1a, indicating that the mutant proteins are most likely post-translationally modified to the same extent as IRAK1a (Fig. 7).
Kinase Activity of IRAK1 Splice Variants-The first 11 amino acid residues of the 30 residues that are spliced out from IRAK1b comprise the C-terminal end of the IRAK1a FIG. 6. Half-lives of IRAK1 isoforms. Non-transfected HepG2 cells were treated with IL-1. and proteins were harvested at indicated time points. The amount of endogenous IRAK1 was quantified using Western blotting and ImageQuaNT technology. For pulse-chase experiments cells were transfected with appropriate expression vectors encoding FLAG-tagged IRAK1 proteins. Proteins were labeled with [ 35 S]methionine for 1 h (pulse). Medium was replaced with fresh medium without [ 35 S]methionine (time point zero) with or without IL-1. Proteins were harvested at appropriate time points (chase). FLAG-tagged IRAK1 proteins were immunoprecipitated using anti-FLAG antibodies and quantified using ImageQuaNT technology following SDS-PAGE. Levels of IRAK1 protein remaining after medium only (Med) or IL-1 (IL1) treatment are expressed as percentage of amount present at time point zero (defined as 100%) on an exponential scale. A, non-transfected cells were treated with IL-1, and amounts of endogenous non-phosphorylated IRAK1a were determined as described above (boxes). Hyperphosphorylated FLAG-tagged IRAK1a (PP-IRAK1a) was expressed in HepG2 cells for pulse-chase experiments as outline above. Cells were treated with medium only (triangles) or IL-1 (circles). B, FLAG-tagged kiIRAK1a (kiIRAK1a) was expressed in HepG2 cells for pulse-chase experiments as outline above. Cells were treated with medium only (boxes) or IL-1 (triangles). C, FLAG-tagged IRAK1b (IRAK1b) was expressed in HepG2 cells for pulse-chase experiments as outline above. Cells were treated with medium only (boxes) or IL-1 (triangles). kinase domain (Fig. 1B). This led us to speculate that the activity of the truncated kinase domain (306 remaining amino acid residues) in IRAK1b may be substantially different from that of IRAK1a, thereby explaining the lack of autophosphorylating capacity in IRAK1b and the consequent differential stabilities of IRAK1a and IRAK1b following IL-1 induction. To investigate this further, we performed in vitro kinase assays. FLAG-tagged IRAK1a, kiIRAK1a, and IRAK1b were overexpressed in cells and immunoprecipitated with anti-IRAK or anti-FLAG antibodies (use of either antibody gave identical results). Washed immunoprecipitates were then used in kinase reactions with myelin basic protein (MBP) as an exogenous target. Although an excess of the kiIRAK1a and IRAK1b was used, only IRAK1a was labeled with 32 P (Fig. 8). This result suggests that IRAK1b is kinaseinactive. A trivial explanation may be that the shorter polypeptide sequence of IRAK1b mandates a structural change that prevents IRAK1b from autophosphorylating/becoming phosphorylated at the site (or sites) that become(s) phosphorylated in IRAK1a. However, only in reactions containing IRAK1a were there higher levels of 32 P incorporation into MBP (Fig. 8) than those seen in reactions from mock transfected cells (background), leading us to conclude that this is not the case and that IRAK1b is truly kinase-inactive.
The phosphorylation of IRAK1a is believed to be mediated by IRAK1a itself, but it is not clear whether one IRAK1a molecule phosphorylates another IRAK1a molecule or if IRAK1a undergoes autophosphorylation. Our observations that IRAK1a becomes phosphorylated in response to IL-1 whereas IRAK1b does not, in conjunction with the results of the co-immunoprecipitation experiments showing that IRAK1b can form dimers with IRAK1a, suggest that in vivo IRAK1a undergoes a genuine autophosphorylation and not an IRAK1a to IRAK1 transphosphorylation process.
Conclusion-We have described a novel alternatively spliced variant of IRAK1, which we have named IRAK1b. The splice variant is evolutionarily conserved in mammals and can activate NF-B, suggesting that it plays an important role in IL-1 signaling. IRAK1b distinguishes itself biologically from the previously described isoform of IRAK1, IRAK1a, by not being capable of autophosphorylation and by being resistant to IL-1-mediated degradation. This leads us to propose a model of membrane proximal IL-1 signaling in which (i) IRAK1a is the isoform that is initially recruited to the IL-1⅐IL-1RI⅐IL-1RAcP complex to rapidly initiate the intracellular IL-1 signaling cascade, and (ii) IRAK1a levels are subsequently dramatically reduced to blunt the acute phase response and to permit IRAK1b to have a more prominent role in mediating a slower but more sustained engagement of downstream components of the signaling cascade. The net effect in this model would be an orderly switch in the predominant IRAK1 isoform, which is recruited into the IL-1⅐IL-1RI⅐IL-1RAcP complex from IRAK1a to IRAK1b, that is accompanied by a transition in the kinetics of the downstream signaling events such that an early rapid vigorous cellular response to IL-1 is superceded by a sustained response characterized by a more stable modified phenotype.
Further studies of the role of IRAK1 in IL-1-dependent signal transduction in the context of the two additional kinases IRAK2 and IRAK-M and other membrane proximal factors will further enhance our overall understanding of the means whereby cytokines elicit changes in cellular phenotype.