Interleukin 1/Toll-like Receptor-induced Autophosphorylation Activates Interleukin 1 Receptor-associated Kinase 4 and Controls Cytokine Induction in a Cell Type-specific Manner

Background: IRAK4 is a central kinase in IL-1R/TLR signaling.

Results: IRAK4 is activated by autophosphorylation, and its inhibition reduces cytokine induction in human monocytes but not dermal fibroblasts.

Conclusion: IL-1R/TLR-induced autophosphorylation activates IRAK4 and controls cytokine induction in a cell type-specific manner.

Significance: Our data provide the mechanism of IRAK4 activation and role in cytokine induction in human cells.

Abstract

IRAK4 is a central kinase in innate immunity, but the role of its kinase activity is controversial. The mechanism of activation for IRAK4 is currently unknown, and little is known about the role of IRAK4 kinase in cytokine production, particularly in different human cell types. We show IRAK4 autophosphorylation occurs by an intermolecular reaction and that autophosphorylation is required for full catalytic activity of the kinase. Phosphorylation of any two of the residues Thr-342, Thr-345, and Ser-346 is required for full activity, and the death domain regulates the activation of IRAK4. Using antibodies against activated IRAK4, we demonstrate that IRAK4 becomes phosphorylated in human cells following stimulation by IL-1R and Toll-like receptor agonists, which can be blocked pharmacologically by a dual inhibitor of IRAK4 and IRAK1. Interestingly, in dermal fibroblasts, although complete inhibition of IRAK4 kinase activity does not inhibit IL-1-induced IL-6 production, NF-κB, or MAPK activation, there is complete ablation of these processes in IRAK4-deficient cells. In contrast, the inhibition of IRAK kinase activity in primary human monocytes reduces R848-induced IL-6 production with minimal effect on NF-κB or MAPK activation. Taken together, these studies define the mechanism of IRAK4 activation and highlight the differential role of IRAK4 kinase activity in different human cell types as well as the distinct roles IRAK4 scaffolding and kinase functions play.

Introduction

IRAK4 is the central kinase in the Toll-like receptor pathway of the innate immune system. Toll-like receptors constitute the first line of defense against pathogenic micro-organisms such as bacteria, viruses, and yeast. The TLR⁴ family also includes three cytokine receptors as follows: the IL-1 receptor, the IL-18 receptor, and the IL-33 receptor. In addition to the innate immune system, IRAK4 is also expressed in T and B lymphocytes and nonhematopoietic/structural cells such as fibroblasts. In addition, IRAK4 regulates adaptive immunity and inflammatory responses mediated by TLR and IL-1 receptors.

The TLR/IL-1 receptors all contain a cytoplasmic TIR domain that complexes with the bifunctional adaptor protein MyD88 that also contains a death domain adaptor region. The death domain of MyD88 then recruits the kinases IRAK1 and IRAK4 into a signaling complex and associates with them via their death domains. It has been demonstrated that mice with a targeted deletion or a kinase-inactive version of IRAK4 are protected from disease in models of rheumatoid arthritis, atherosclerosis, multiple sclerosis, and Alzheimer disease (1–5). Thus, IRAK4 is an enticing candidate for drug discovery of therapies for these diseases (1–5).

The mechanism by which IRAK4 becomes activated is unclear. It is well known that certain kinases, e.g. members of the MAPK family, require inducible phosphorylation by an upstream kinase on their activation loop to activate (6, 7). However, other kinases such as PKA and cGMP protein kinase require autophosphorylation to become fully activated (8, 9). It has been shown that IRAK4 can undergo autophosphorylation in the presence of Mn²⁺-ATP (10). The sites within the activation loop that were identified by tandem mass spectroscopy in that study were Thr-342, Thr-345, and Ser-346. The authors hypothesized that IRAK4 autophosphorylation is proceeded by an intramolecular mechanism because IRAK4 was not able to intermolecularly autophosphorylate in vitro in the presence of Mn²⁺-ATP.

In this study, we examined the autophosphorylation of IRAK4 in the cell and in the presence of Mg²⁺, the physiological cation for ATP, in cell-free enzymatic assays. In addition to confirming the autophosphorylation of Thr-342, Thr-345, and Ser-346, we identified a fourth phosphorylation site, Thr-352. We found that mutation of single residues to alanine did not significantly affect the catalytic activity of the protein but that mutations of dual combinations of residues Thr-342, Thr-345, and Ser-346 completely abolished activity. These data suggest autophosphorylation of IRAK4 leads to the activation of its kinase activity. We show that autophosphorylation of these activation loop residues are inducible upon treatment with R848 in primary human monocytes and IL-1β in human dermal fibroblasts and that this autophosphorylation proceeds via an intermolecular mechanism both in the enzymatic and in the cellular context.

Additionally, we demonstrate that the kinase domain of IRAK4 is constitutively phosphorylated in the cell, but the full-length kinase only becomes phosphorylated following stimulation. This demonstrates the role of the death domain both in maintaining the kinase in an inactive state and in the induction of the kinase activity.

Importantly, we also show that pharmacological inhibition of IRAK4 by an IRAK4/IRAK1 dual inhibitor completely blocks IRAK4 autophosphorylation but, surprisingly, has minimal effects on activation of NF-κB, p38, JNK, and ERK in both human dermal fibroblasts and primary human monocytes. We find, as reported previously, that human dermal fibroblasts from patients with autosomal recessive IRAK4 deficiency do not activate NF-κB, p38, JNK, and ERK and do not produce cytokines in response to IL-1β (11–15). Interestingly, we observed that the inhibition of IRAK4 autophosphorylation blocks cytokine production in primary monocytes but not in dermal fibroblasts. These data clearly demonstrate that there are different roles of IRAK4 kinase activity and scaffolding activity in different human cell types.

EXPERIMENTAL PROCEDURES

Cloning and Expression of IRAK4

The full-length ORF of IRAK4 (GenBank^TM number AF445802) was obtained from Invitrogen. Both the full-length and the kinase domain (residues 154–460) were cloned with the addition of either C-terminal FLAG tags or His₆ tags via PCR into the Gateway entry vector pDONR 201 (Invitrogen) according to the manufacturer's instructions. For eukaryotic cell expression, the C-terminal FLAG-tagged constructs were recombined into the Gateway expression vector pcDNA3-DEST40 (Invitrogen). Mutagenesis was performed via PCR using KOD polymerase as described previously (16). Human dermal fibroblasts or HEK 293T cells were transfected with Lipofectamine 2000 (Invitrogen) according to the manufacturer's directions. For baculovirus expression, C-terminal His₆-tagged constructs were recombined into the baculovirus expression vector pDEST8, and baculovirus was produced via the Invitrogen baculovirus system. For expression, Sf21 cells were infected with virus at a multiplicity of infection of 1:1 and grown for 48 h in Invitrogen InsectGro media at 27 °C.

Protein Purification

Sf21 cells expressing IRAK4 full-length protein or kinase domain were lysed via Paar Bomb at 4 °C in lysis buffer (50 mm phosphate, pH 8, 300 mm NaCl, 25 mm glycerophosphate, 10 mm NaF, 1 mm orthovanadate, 5 mm β-mercaptoethanol, and EDTA-free protease inhibitors (Merck)). The kinase domain was initially purified via nickel-affinity chromatography (Ni-NTA, Qiagen), and contaminates were removed by flow-through cation exchange chromatography (Poros HS, Applied Biosystems). The protein was captured via anion exchange chromatography (Poros HQ, Applied Biosystems) and eluted with a 15-column volume of salt gradient (buffer A: 20 mm HEPES, pH 7.5, 5 mm DTT, 50 mm NaCl; buffer B: 20 mm HEPES, pH 7.5, 5 mm DTT, 1 m NaCl). The protein was then brought to greater than 99% purity via gel filtration chromatography (Tosoh G3000SW column).

Separation and Characterization of IRAK4 Protein Phosphorylation States

The different phosphorylation forms of the protein were separated via high affinity strong anion exchange chromatography on a Mono Q column (GE Healthcare) using a very shallow gradient of 40 column volumes (buffer A: 20 mm HEPES, pH 7.5, 5 mm DTT, 50 mm NaCl; buffer B: 20 mm HEPES, pH 7.5, 5 mm DTT, 600 mm NaCl).

Autophosphorylation of IRAK4 kinase domain was shown using purified kinase domain with and without dephosphorylation by λ phosphatase (New England Biolabs). Purified kinase domain was allowed to react with 10 mm MgCl₂ and 1 mm ATP for 1 h at RT. The varied phosphorylation states were then purified via high affinity strong anion exchange chromatography and analyzed by LCMS.

Measurement of Initial Rates

The enzymatic rate was determined at 25 °C using the coupled pyruvate kinase (PK)/lactate dehydrogenase assay at 340 nm on a Molecular Devices plate reader as described previously (17). The reaction was carried out in a final volume of 80 μl, in 25 mm HEPES, pH 7.5, 10 mm MgCl₂, 2 mm DTT, 0.008% Triton X-100, 100 mm NaCl, 20 units of PK, 30 units of lactate dehydrogenase, 0.025 mm NAD, 2 mm phosphoenolpyruvate, and phosphorylated or dephosphorylated forms of IRAK4 kinase domain or the full-length construct at 50 nm (17).

In Vitro Assay for IRAK4 Activity of Mutants

FLAG- tagged IRAK4 kinase domain mutant proteins were isolated from HEK 293T cells following transfection with Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Cells were lysed in Triton lysis buffer (1% Triton X-100, 20 mm HEPES, pH 7.4, 100 mm NaCl) containing protease inhibitors (Sigma) at 4 °C. Cell lysates were immunoprecipitated with anti-FLAG M2 beads (Sigma) washed extensively in lysis buffer and eluted with 200 μg/ml FLAG peptide (Sigma) in 10 mm HEPES, pH 7.4, with 150 mm NaCl. Eluates were dialyzed at 1:1000 against 20 mm Tris-HCl, pH 7.5, 150 mm NaCl to remove FLAG peptide. Protein was diluted to 10 ng/ml stock and assayed for activity by a DELFIA assay. Briefly, 10 ng/ml IRAK4 protein was preincubated with 2 mm ATP in a kinase reaction buffer of 20 mm HEPES, pH 7.5, 5 mm MgCl₂, 0.0025% Brij for 1 h at 25 °C. This protein was diluted to 5 nm and incubated with 200 nm peptide (biotinylated AGAGRDKYKTLRQIR, Tufts University) with 2 mm ATP for 1 h in reaction buffer. The reaction was stopped by the addition of 20 mm EDTA. 100 μl of reactions were transferred to a 96-well streptavidin-coated plate (R&D Systems) and allowed to bind for 30 min followed by extensive washing with 1× TBS containing 0.02% Tween. Plates were then incubated with anti-phospho-ezrin/radixin/moesin antibody (Cell Signaling Technology) diluted 1:1000 in blocking buffer (10 mm MOPS, pH 7.5, 150 mm NaCl, 3 mm NaN₃, 0.025% Tween 20, 0.2% gelatin, 2% BSA) for 1 h followed by extensive washing with TBS/Tween. Plates were developed by incubation for 30 min with an anti-rabbit IgG-europium conjugate (PerkinElmer Life Sciences) diluted 1:5000 in blocking buffer followed by extensive washing in TBS/Tween and finally the addition of DELFIA enhancement solution (PerkinElmer Life Sciences). Plates were read in an Envision reader (PerkinElmer Life Sciences) on a europium setting.

Antibodies

Polyclonal IRAK4 antibodies were made by a contract with BIOSOURCE against a C-terminal peptide of human IRAK4 (CEKTIEDYIDKKMNDADSC) or were purchased from Cell Signaling Technology. Monoclonal antibodies to IRAK4 were obtained from Abnova Corp. Antibodies to total and phospho-p38, total and phospho-JNK, MyD88, and human IRAK1 were purchased from Cell Signaling Technology. Phosphospecific antibodies to sites on IRAK4 (Thr(P)-342, Thr(P)-345, Ser(P)-346, Thr(P)-352, Thr(P)-345/Ser(P)-346, Thr(P)-342/Thr(P)-345, and Thr(P)-342/Thr(P)-346) were made by a contract with 21st Century Biochemicals (Marlborough, MA). Antibodies were immunodepleted against nonphosphorylated peptides and, in the case of antibodies against dual phosphorylated sites, immunodepleted against both monophosphorylated peptides to ensure specificity.

In Vitro Intermolecular Phosphorylation Assay

FLAG-tagged kinase-inactive form (D329A) of the kinase domain of IRAK4 was expressed and purified as described above under “In Vitro Assay for IRAK4 Activity of Mutants.” Immunoprecipitated kinase-inactive proteins were incubated with 20 nm of fully phosphorylated IRAK4 full-length or kinase domain proteins purified from baculovirus in kinase reaction buffer containing 2 mm ATP for 1 h. Reactions were analyzed by Western blotting with total IRAK4 and anti-Thr(P)345/Ser(P)-346.

Transfection of Human Dermal Fibroblasts and Immunoprecipitation of Mutant Proteins

SV40-transformed dermal fibroblasts from an IRAK4-deficient patient 15-IRAK4-SV40 fibroblast, 1–1096_40 + 23del/1–1096_40 + 23del (IRAK4^−/−) and a healthy control (wild type) provided by Jean-Laurent Casanova were used (12). Cells were grown to 80% confluence in DMEM + 10% FBS and transfected with wild-type and mutant IRAK4 constructs using Lipofectamine 2000 (Invitrogen) according to the manufacturer's directions. Generally, 5 μg of construct in pcDNA3-DEST40 vector (Invitrogen) was used for a 10-cm plate. Transfected cells were placed in DMEM + 0.5% FBS 2 h prior to stimulation. Cells were stimulated with 10 ng/ml IL-1β (R&D Systems) for 30 min and then lysed in 1 ml of Triton lysis buffer (above) containing protease inhibitors, 1 mm sodium vanadate, and 5 mm β-glycerophosphate. Recombinant IRAK4 proteins were immunoprecipitated using anti-FLAG M2-Sepharose beads (Sigma)

Monocyte Enrichment and Treatment

Buffy coats (from MGH) were incubated with RosetteSep Mixture (Stem Cell Technologies) and purified according to the manufacturer's directions. Monocytes were resuspended for culture in RPMI + 0.5% FBS. Monocytes were then pretreated for 30 min with 10 μm compound 1 and then stimulated with 1 μg/ml TLR7/8 agonist R848 for 15 min (phosphoprotein) and 4 h (cytokines).

Dermal Fibroblast Treatment

Wild-type and IRAK4-null human dermal fibroblast cells (11) were grown to 80% confluence and then pretreated for 30 min with 10 μm compound 1 followed by stimulation with 10 ng/ml IL-1β and collected post-stimulation at 15 min (phosphoprotein) and 4 h (cytokines).

Western Blotting

Samples were washed with ice-cold PBS, lysed in 200 μl of protein lysis buffer (Novex), sonicated, heated to 95 °C for 2 min, separated by SDS-PAGE, and transferred to nitrocellulose membranes. Membranes were then blocked in Odyssey Blocking Buffer (Li-Cor) and incubated overnight in primary antibody at 4 °C. Membranes were then washed three times for 5 min each in 1× PBS-T and incubated in secondary antibody for 60 min at RT. Membranes were again washed three times for 5 min each in PBS-T and imaged on the Odyssey CLx (Li-Cor).

ELISA

Culture supernatants were collected 1 and 4 h after stimulation, aliquoted to 4-plex pro-inflammatory II plates (MSD N45025B-1), and developed according to the manufacturer's directions.

ActivX Analysis of Human Monocytes

50 million human monocytes were incubated with 10 μm compound 1 for 30 min and 50 million with DMSO. Both groups were then spun down at 1500 rpm in a Sorvall Legend RT swinging bucket centrifuge for 10 min and frozen on dry ice. The cell pellet was then sent to ActivX Biosciences (La Jolla, CA) for analysis in the KiNativ in situ kinase profiling panel.

RESULTS

Expression of IRAK4 and Separation of Distinct Phosphorylation States by Ion Exchange Chromatography

The C-terminal His₆-tagged kinase domain (amino acids 154–460) and the full-length construct of IRAK4 were expressed in Sf21 cells. Chromatography via size exclusion showed a single peak at 35 kDa for the kinase domain construct. We observed that Ni-NTA chromatography of the full-length protein gave two species of IRAK4, the full-length protein and a smaller 34-kDa fragment (data not shown). This smaller fragment could be readily removed by size exclusion chromatography to give a homogeneous preparation of the full-length protein.

Five milligrams of IRAK4 kinase domain protein from Ni-NTA were bound to a column of Q-Sepharose and eluted via a salt gradient from 50 mm to 1 m NaCl over 40 column volumes. Two distinct peaks were observed that could be interconverted by incubation with 1 mm ATP or with γ-phosphatase demonstrating that the mobility shift is due to phosphorylation (data not shown).

ESI-LC MS on the individual peaks revealed that the faster migrating peak had a molecular mass of 35,383 Da. This differs from the expected molecular mass for the IRAK4-kinase domain construct of 35,330 Da by 50 Da indicating the possibility of N-terminal acetylation. Similar analysis of the slower migrating peak showed a molecular mass of 35,622, which correlates to the presence of an N-terminal acetylation and three phosphates (data not shown). A much smaller amount of the phosphoprotein from this peak had a molecular mass of 35,700 which contained four phosphates with the N-terminal acetylation.

Characterization of Phosphorylation Sites on IRAK4 Kinase Domain via HPLC ESI-MS-MS

To unambiguously determine the major sites of phosphorylation on the fully phosphorylated form of the IRAK4 kinase domain, the IRAK4 kinase domain autophosphorylated protein was subjected to tandem mass spectrometric analysis following trypsin digestion. Four distinct peptides were observed as follows: T3 (residues 165–174 (positions denoted from the N-terminal of the full-length protein)), T4 (residues 175–191), T19 (residues 339–347), and T20 (residues 348–362) (supplemental Fig. S1). Ion chromatograms of these peptides showed that the T3, T19, and T20 peptides are all phosphorylated. Further separation and analysis showed that T3 was weakly phosphorylated on Ser-167 and that T20 was phosphorylated on Thr-352. Both peptides contained only one phosphorylation site. The T19 peptide had phosphorylation sites at Thr-342, Thr-345, and Ser-346 (supplemental Fig. S1). Although this peptide appeared as three distinct singly phosphorylated peptides, it also appeared as three distinct dual phosphorylated peptides. The dual phosphorylated peptides eluted were of three possible combinations: Thr(P)-342/Thr(P)-345, Thr(P)-342/Ser(P)-346, or Thr(P)-345/Ser(P)-346 (supplemental Fig. S2). Therefore, our results indicate that the majority of triphosphorylated species consist of at least one of the dual phosphorylated species on peptide T19 in combination with Thr-352. The identified phosphorylation sites in the kinase domain are shown within the context of the full protein sequence in Fig. 1A and the IRAK4 structure with phosphorylation sites in Fig. 1B.

FIGURE 1.

A, results of ESI-MS/MS analysis of phosphorylated IRAK4 kinase domain. Positively identified peptides are highlighted in red with underscored arrows delineating subfragments of these peptides. Designation of the peptides (supplemental Fig. S2) is denoted in red below the corresponding arrow. Positively identified phosphorylated residues are highlighted in blue with the residue number designated above. The activation loop is denoted by a green bracket. B, IRAK4 structure derived from modified Protein Data Bank structure file 2NRU. Residues Thr-342, Thr-345, Ser-346, and Thr-352 are modified with phosphates, and the C-helix and ATP binding pocket are indicated for clarity.

Cheng et al. (10) observed phosphorylation in the active site on Thr-342, Thr-345, and Ser-346 but not on Thr-352. This difference could be due to the fact Cheng et al. (10) used Mn²⁺ as the cationic metal to coordinate the ATP, whereas we used Mg²⁺, which is the physiological metal. The difference could also be due to the fact that the phosphorylation sites that we identified came from a constitutively autophosphorylated kinase domain, whereas Cheng et al. (10) used the full-length protein.

Activity of Phosphorylated and Nonphosphorylated Forms of IRAK4

It is well known that autophosphorylation of certain kinases can increase their activity (9, 18). To qualitatively determine whether there is a difference in activity between the phosphorylated and nonphosphorylated forms of full-length IRAK4 and its kinase domain, we observed the initial rates of conversion of a substrate peptide via a coupled enzyme assay. As shown in Fig. 2A, the phosphorylated form of both the kinase domain and the full-length protein shows a consistent and constant linear rate of ATP conversion. However, the nonphosphorylated forms of the protein demonstrate a lag phase, where the initial rate is slow but gradually increases over time and eventually reaches the same linear rate as the phosphorylated form. These results suggest that the fully phosphorylated form of the protein is more catalytically active than the nonphosphorylated form. They also suggest that the change in rate over time of the nonphosphorylated form is due to autophosphorylation, which increases the catalytic activity. In addition, the length of the “lag” phase was reduced as the concentration of the nonphosphorylated full-length protein increased (data not shown). The concentration dependence of the lag phase and of the rate indicates a second order reaction and demonstrates that the phosphorylation step is most likely intermolecular.

FIGURE 2.

A, kinetic analysis of phosphorylated and unphosphorylated forms of the kinase domain and full-length IRAK4. The kinetics were determined as described under “Experimental Procedures.” Initial rates from the progress curve show that the unphosphorylated forms of IRAK4 show an initial lag compared with the phosphorylated forms. B, enzymatic activity of activation loop mutations in the kinase domain of IRAK4. The mutant FLAG-tagged kinase domain construct of IRAK4 was purified from 293T cells and assayed by DELFIA in vitro kinase assay as described under “Experimental Procedures.” Inset, Coomassie Blue gel of purified mutant IRAK4 proteins. The 1st lane is the molecular weight marker, with the 49-kDa marker indicated, followed sequentially by IRAK4 mutants in the same order as depicted in the bar graph.

We also observed that the presence or absence of the death domain did not affect the rate of autophosphorylation, the rate of reaction, or the length of the lag phase as the rates of ATP conversion are similar (Fig. 2A). These data demonstrate that IRAK4 kinase is constitutively active in the presence and absence of the death domain in a cell-free reaction.

Activity of Activation Loop Mutants in the IRAK4 Kinase Domain

To determine the contribution of the putative phosphorylation sites to IRAK4 activity, single mutations to alanine at positions 342, 345, 346, and 352 were made in the C-terminal FLAG-tagged kinase domain construct and expressed in HEK 293T cells. Several combinations of mutations at these residues were also made and evaluated. The protein was purified to >70% purity, and the enzymatic activity of the purified mutants was determined at 1 mm ATP, 10 mm MgCl₂ following a 1-h preincubation in this concentration of ATP/MgCl₂ to maximize the level of autophosphorylation to ensure a constant initial rate of kinase activity. The activity was determined via fluorescent ELISA (DELFIA) assay as described under “Experimental Procedures.” The activity of the mutants was in the linear range of a standard curve determined with wild-type kinase and normalized to the protein concentration by Western blot and Coomassie Blue staining. As shown in Fig. 2B, we observed that the single mutations of any of these residues to alanine did not substantially inhibit the enzymatic activity. However, dual and triple combinations of mutations at Thr-342, Thr-345, and Ser-436 substantially inhibited activity, indicating that at least two of these residues must be phosphorylated for activity. This finding is in contrast to the findings of Cheng et al. (10) who found that single mutations to alanine at Thr-342, Thr-345, or Ser-346 substantially inhibited the activity of the kinase.

The single mutation of Thr-352 to alanine did not significantly affect the activity nor did the mutation of Thr-352 in combination with single mutations of Thr-342, Thr-345, or Ser-346 (data not shown). Therefore, phosphorylation of Thr-352 is not essential for kinase activity. The physiological importance of phosphorylation of Thr-352 is unclear; however, it is observed in the crystal structure of IRAK4 that the hydroxyl side chain of Thr-352 is exposed to solvent and lines the substrate binding pocket as shown in Fig. 1B (19). It is possible that a phosphate at Thr-352 could induce a different binding position for the substrate and allow phosphorylation of different residues on the substrate or allow for phosphorylation of a different substrate protein. It should be noted that we were unable to show phosphorylation of Thr-352 in cells using our antibody to Thr(P)-352 (data not shown). Thus, the physiological significance of this phosphorylation remains to be determined.

To develop reagents to observe specific phosphorylated states of IRAK4 and determine their physiological significance, we made phosphospecific antibodies to Thr(P)-342, Thr(P)-345, Ser(P)-346, Thr(P)-352, and to the dual phosphorylated species of Thr(P)-342/Thr(P)-345, Thr(P)-342/Ser(P)-346, and Thr(P)-345/Ser(P)-346. The antibodies to the monophosphorylated sites were immunodepleted over the corresponding nonphosphorylated peptides and, in the case of the dual phosphorylated species, over each monophosphopeptide as well. This ensured that the antibodies will specifically recognize the requisite phosphorylated species. The antibodies did not recognize the purified unphosphorylated form of full-length IRAK4 but robustly recognize the purified autophosphorylated form of the full-length IRAK4 in Western blots (data not shown). These data recapitulated the results of the mass spectrometry and show that these residues are indeed autophosphorylated in the intact protein at all of the identified residues in vitro. The most robust signal was obtained with the antibody against Thr(P)-345/Ser(P)-346, henceforth referred to as the pIRAK4 antibody.

IRAK4 Can Be Autophosphorylated Intermolecularly on Activation Loop Residues in Vitro

Our observation that the initial lag phase in kinase activity is dependent on concentration strongly suggests that IRAK4 autophosphorylates by an intermolecular mechanism. Previously, Cheng et al. (10) showed that IRAK4 does not autophosphorylate via an intermolecular reaction. Their experiment monitored the ability of wild-type kinase to phosphorylate the kinase-inactive mutant at 30 μm ATP with 2 mm MgCl₂, and detection was monitored using a phosphospecific antibody to Thr(P)-345/Ser(P)-346. We performed a similar experiment at a more physiological condition using 2 mm ATP and 5 mm MgCl₂. We monitored the ability of the active full-length kinase to phosphorylate the kinase-inactive D329A kinase domain protein. As shown in Fig. 3, the kinase-inactive kinase domain was phosphorylated at Thr(P)-345/Ser(P)-346, clearly demonstrating that these residues can be phosphorylated via an intermolecular mechanism. Additionally, we observed a slight mobility shift upon autophosphorylation of the full-length kinase domain that was partially inhibited by the kinase-inactive D329A kinase domain (Fig. 3). Our results show that mammalian IRAK4 becomes activated by a similar mechanism as Pelle, the Drosophila homolog of IRAK4, which has been shown to be activated by intermolecular autophosphorylation (20).

FIGURE 3.

Full-length wild-type IRAK4 phosphorylates D329A kinase domain in vitro. Wild-type and kinase-inactive forms of IRAK4 were purified from HEK 293T cells and co-incubated in the presence of 2 mm ATP, 5 mm MgCl₂ for 1 h. Reactions were analyzed by Western blotting with antibodies to pIRAK4 (Thr(P)-345/Ser(P)-346) or anti-IRAK4. Arrows indicate full-length IRAK4 (FL) and D329A kinase domain (KD).

IL-1-induced Autophosphorylation of IRAK4 in Human Dermal Fibroblasts

We transfected with FLAG-tagged wild-type IRAK4 or kinase-inactive (D329A) mutant and treated with IL-1β for 30 min. Immunoprecipitation with anti-FLAG antibody was followed by Western blotting with the pIRAK4 (Thr(P)-345/Ser(P)-346) antibody. We observed robust IL-1β-induced phosphorylation at the dual site Thr-345/Ser-346 (Fig. 4A). We were unable to detect phosphorylation on monophosphorylated sites or other dual phosphorylated sites in the overexpressed protein. This could be due to low levels of phosphorylation at these sites by IL-1 activation or, alternatively, to low affinity of these antibodies for their epitopes. We repeated this experiment with the kinase-inactive mutant D329A, which is unable to autophosphorylate at Thr-345/Ser-346 in vitro. Surprisingly, after Western blotting of immunoprecipitated and overexpressed D329A from IL-1-treated dermal fibroblast cells with pIRAK4 Thr(P)-345/Ser(P)-346 antibody, we observed that the D329A mutant of IRAK4 was also inducibly phosphorylated on these residues similar to the wild-type protein (Fig. 4A). This demonstrates that IL-1-induced phosphorylation of the activation loop of IRAK4 can come from trans-phosphorylation by active endogenous IRAK4 or by an upstream kinase.

FIGURE 4.

A, phosphorylation of wild-type and kinase-inactive D329A IRAK4 at Thr-345/Ser-346 is inducible by IL-1β. Wild-type human dermal fibroblasts were transfected with FLAG-tagged constructs of IRAK4 and immunoprecipitated (IP) with anti-FLAG antibody and Western-blotted with antibodies to IRAK4 or anti-pIRAK4 (Thr(P)-345/Ser-346). B, phosphorylation of kinase-dead IRAK4 is ablated in IRAK4^−/− human dermal fibroblasts. Wild-type and IRAK4-deficient human dermal fibroblasts were transfected with full-length FLAG-tagged D329A stimulated for 30 min with 10 ng/ml IL-1β. Samples were FLAG-immunoprecipitated and Western-blotted with antibodies to pIRAK4 or total IRAK4. Whole cell lysate (WCL) blotted with anti-IRAK4 antibody is shown for comparison of endogenous IRAK4 levels.

Loss of Endogenous IRAK4 Prevents IL-1-inducible Phosphorylation of Ectopically Expressed IRAK4

To determine whether IRAK4 phosphorylation is mediated by intermolecular IRAK4 autophosphorylation or another independent kinase, we determined whether kinase-dead IRAK4 kinase can be phosphorylated in the IRAK4-deficient dermal fibroblast cells. As shown in Fig. 4B, the IL-1-induced autophosphorylation of the kinase-inactive IRAK4 is preserved in the wild-type cell line, but it does not occur in the IRAK4-deficient cell line. This result clearly shows that, within the cell, another molecule of IRAK4 is responsible for the phosphorylation and activation of IRAK4.

Death Domain Prevents Constitutive Activation of IRAK4 Kinase Activity

The death domain of IRAK4 is believed to be integral for the recruitment of the IRAK kinases to the receptor-MyD88 signaling complex. To assess whether autophosphorylation of Thr-345/Ser-346 is dependent on the presence of an intact death domain, we transfected dermal fibroblast cells with the FLAG-tagged wild-type kinase domain only, as well as a kinase-inactive form of IRAK4 (D329A) kinase domain. We observed robust IL-1-inducible phosphorylation of endogenous IRAK4 in the whole cell lysate (Fig. 5A).

FIGURE 5.

Wild-type kinase domain of IRAK4 is constitutively autophosphorylated, but D329A-containing kinase domain fails to be phosphorylated in human dermal fibroblasts. A, wild-type human dermal fibroblasts were transfected with wild-type and D329A mutations of FLAG-tagged kinase domain constructs. Cells were stimulated for 30 min with 10 ng/ml IL-1β, and the lysates were blotted with pIRAK4 (Thr(P)-345/Ser(P)-346) and a polyclonal antibody to the C terminus of IRAK4. Arrows indicate position of full-length endogenous IRAK4 (FL), a nonspecific band (NS), endogenous cleaved IRAK4, and transfected kinase domain (KD). B, lysates were immunoprecipitated with anti-FLAG antibody and Western-blotted with anti-pIRAK4 and anti-FLAG antibodies. Molecular weight markers are indicated on the left margin of each blot.

We observed a band in the pIRAK4 blot at the molecular weight of the transfected wild-type kinase domain (∼35 kDa) in the unstimulated cells, which was not present in the vector or kinase-inactive lanes. These data suggest that a significant amount of constitutive autophosphorylation occurs on the wild-type IRAK4 kinase domain in unstimulated cells. It is worth noting that the pIRAK4 blots also detected a 35-kDa band in the IL-1β-treated cells, likely due to the proteolytic fragment from the endogenous IRAK4, which was also observed by Hatao et al. (21) following stimulation of macrophages with TLR agonists. To directly look at the phosphorylation of the transfected IRAK4 kinase domain, we performed the immunoprecipitation with an anti-FLAG antibody followed by immunoblot with the pIRAK4 antibody. In this experiment, the constitutive autophosphorylation of the wild type but not the kinase-inactive form of the kinase domain becomes readily apparent. (Fig. 5B)

Unlike the endogenous full-length IRAK4, the wild-type kinase domain was constitutively phosphorylated in the absence of IL-1β, and the phosphorylation level did not increase after treatment with IL-1β. Furthermore, the D329A kinase domain mutant, unlike the full-length D329A mutant, did not show any phosphorylation at Thr-345/Ser-346 in either the presence or absence of IL-1β (Fig. 5B). It is clear from these results that the ability of IL-1 to induce phosphorylation of IRAK4 is dependent on an intact death domain. The death domain acts as a scaffold to recruit the kinase to the receptor-associated MyD88. Thus, the inability of the kinase-inactive kinase domain to be phosphorylated could be due to the inability of it to be recruited to the receptor-associated MyD88 and another molecule of endogenous IRAK4. Importantly, the fact that the active kinase domain is constitutively phosphorylated in the absence of the death domain also implies that the death domain plays a role in keeping the full-length protein inactive in the absence of activation by IL-1R or TLRs. Thus, the death domain may play a role as a negative regulator of IRAK4 kinase activity in the context of the unstimulated cell.

Pharmacological Inhibition of IRAK4 Kinase Blocks IRAK4 Phosphorylation in Both Human Monocytes and Dermal Fibroblasts

Having demonstrated that autophosphorylation of IRAK4 occurs physiologically, we were interested in using phosphorylated IRAK4 as a potential biomarker to determine the extent of inhibition of IRAK4 activation in cells using a pharmacological inhibitor. We wanted to see if pharmacological inhibition of IRAK4 kinase would inhibit phosphorylation of IRAK4 in the context of the TLR/IL-1R pathway and what functional consequences this inhibition may have in different human cell types. We utilized dermal fibroblasts due to their availability from wild-type and IRAK4-deficient patients. We also examined primary human monocytes from healthy volunteers, because we were interested in innate immune cells where IRAK4 activity plays an important role as an inflammatory mediator. We used compound 1, a previously characterized inhibitor of IRAK4 kinase for our studies on the pharmacology of IRAK4 in human cells (19). This compound is highly potent against IRAK4 (IC₅₀ = 2 nm at K_m = 300 μm ATP) (22) and to a lesser extent IRAK1 (IC₅₀ = 15 nm at K_m = 35 μm ATP). Although it possesses excellent selectivity, it has poor physical properties such as solubility and permeability, which necessitated use of micromolar concentrations for our studies. To test target occupancy in the context of the cell, we subjected cells treated with 10 μm compound 1 to analysis by ActivX technology in the KiNativ panel (23). We observed that the compound has excellent selectivity against 170 kinases in the panel with 90% occupancy against IRAK4 and only 55% occupancy against IRAK1. The only other kinases that were inhibited greater than 70% were PIP4K2C (76%), and Aurora A and B (71%), none of which are known to inhibit TLR/IL-1R signaling (supplemental Table S1).

We used the TLR7/8 agonist R848 to activate IRAK4 in human monocytes and IL-1β to activate IRAK4 in wild-type and IRAK4-deficient dermal fibroblasts. The choice of agonist is based on the ability of these agonists to induce cytokine production in these cell types. Pharmacological inhibition of IRAK4 kinase with 10 μm of compound 1 completely blocks IRAK4 autophosphorylation in dermal fibroblasts. Compound 1 also significantly but does not completely block ubiquitin modification of IRAK1, demonstrating that post-translational modification of IRAK1 is not entirely dependent on IRAK4 kinase activity. This is in contrast to IRAK4-deficient fibroblasts where IL-1β-induced modification of IRAK1 is completely absent (Fig. 6A). Compound 1 also completely blocks IRAK4 phosphorylation in human monocytes (Fig. 6B).

FIGURE 6.

A, pharmacological inhibition of IRAK4 kinase blocks IRAK4 phosphorylation in human dermal fibroblasts. Human wild-type and IRAK4-deficient (P15) dermal fibroblasts were treated with 10 μm compound 1 for 30 min followed by stimulation with 10 ng/ml IL-1β for 30 min. Cell lysates were collected and analyzed by Western blot for pIRAK4 (Thr(P)-345/Ser(P)-346), IRAK4, and IRAK1. B, pharmacological inhibition of IRAK4 kinase blocks IRAK4 phosphorylation in human monocytes. Human monocytes were treated with 10 μm compound 1 for 30 min followed by stimulation with 1 μg/ml R848 for 15 min. Cell lysates were collected and analyzed by Western blot for pIRAK4 (Thr(P)-345/Ser(P)-346) and IRAK4.

IRAK4 Kinase Activity Is Required for Cytokine Production in Human Monocytes but Dispensable in Dermal Fibroblasts

We next wanted to investigate the downstream functional consequences of this block in IRAK4 phosphorylation. Using the pharmacological inhibitor of IRAK4 kinase in both human monocytes and human dermal fibroblasts, we measured IL-1, IL-6, and TNF-α cytokine induction. Fig. 7 shows that these cytokines are significantly reduced with the IRAK4 inhibitor in human monocytes; however, there is no effect on cytokine production in the wild-type dermal fibroblasts. As expected, there is no induction of cytokines in the IRAK4-deficient dermal fibroblasts.

FIGURE 7.

A, IRAK4 kinase activity is dispensable for cytokine production in human dermal fibroblasts. Human wild-type and IRAK4-deficient dermal fibroblasts were treated with 10 μm compound 1 for 30 min followed by stimulation with 10 ng/ml IL-1β. B, IRAK4 kinase activity is required for cytokine production in human monocytes. Human monocytes were treated with 10 μm compound 1 for 30 min followed by stimulation with 1 μg/ml R848. Cell supernatants were collected and analyzed by ELISA (MSD) for levels of IL-6 and TNF-α. Significance of inhibition was determined by a Student's t test in both experiments. NS, not significant; *, p < 0.03.

The data suggest that in monocytes, the kinase activity of IRAK4 is required for cytokine production. However, in dermal fibroblasts, although deficiency of IRAK4 abolishes cytokine production entirely, inhibition of kinase has no effect on these cytokines. This result is consistent with a previous report showing that reconstitution of IRAK4-deficient fibroblasts with kinase-inactive full-length IRAK4 can reconstitute IL-1-induced IL-6 production (24). Thus, IRAK4 protein scaffold activity is required but not the kinase activity in dermal fibroblasts, suggesting IRAK4 scaffolding activity allows cytokine production to continue in this cell type. Clearly, there are different roles for IRAK4 kinase activity and scaffolding activity in different cell types.

Minimal Effects on NF-κB and MAPK Are Observed with Complete Inhibition of IRAK4 Kinase in Dermal Fibroblasts and Monocytes

Normal TLR/IL-1R signaling can exert its effects on cytokines via activation of NF-κB and MAPKs, including the core MAPK pathways, JNK, ERK, and p38. Therefore, we investigated the activation status of these pathways. Surprisingly, our data revealed only minimal inhibition of pp38, pERK, pJNK, and pp65 in both cell types following complete inhibition of IRAK4 kinase (Fig. 8). Understanding the exact signaling events mediated by IRAK4 kinase activity that lead to cytokine production in monocytes, as well as the scaffold function of IRAK4 in dermal fibroblasts that is responsible for cytokine production, is the focus of our immediate future studies.

FIGURE 8.

A, minimal effects on NF-κB and MAPK are observed with inhibition of IRAK4 kinase in human dermal fibroblasts. Human wild-type and IRAK4-deficient dermal fibroblasts were treated with 10 μm compound 1 for 30 min followed by stimulation with 10 ng/ml IL-1β. B, minimal effects on NF-κB and MAPK are observed with inhibition of IRAK4 kinase in human monocytes. Human monocytes were treated with 10 μm compound 1 for 30 min followed by stimulation with 1 μg/ml R848 for 15 min. Cell lysates were collected and analyzed by Western blot for phospho-p65, phospho-ERK, phospho-JNK, and phospho-p38 (p-p65, p-ERK, p-JNK, and p-p38, respectively).

DISCUSSION

We have identified four phosphorylation sites within the activation loop that are the result of autophosphorylation: Thr-342, Thr-345, Ser-346, and Thr-352. Previously, Cheng et al. (10) identified Thr-342, Thr-345, and Ser-346 by MS analysis, and other groups identified these residues by crystallography (19, 22). In this report, we present several lines of evidence to support the hypothesis that autophosphorylation of IRAK4 is required for the full activation and subsequent signaling events in a cell type-specific manner. First, we showed purified unphosphorylated IRAK4 exhibited a lag in the kinase assay in contrast to the phosphorylated form, suggesting the requirement of autophosphorylation for the kinase activity. Second, we show that mutation of at least two residues at positions Thr-342, Thr-345, and Ser-346 reduce the kinase activity. These data suggest that phosphorylation at these sites are required for the full activation of IRAK4 kinase activity. Third, importantly, inhibition of IRAK4 phosphorylation can be achieved by an IRAK4/1 inhibitor and results in the reduction of cytokine production in a cell type-specific manner, i.e. reduction in human monocytes but no effect in dermal fibroblasts.

Unphosphorylated full-length IRAK4 can autophosphorylate in vitro in a cell-free system. In the cellular context, however, autophosphorylation is inducible by IL-1β. The mechanism of intermolecular autophosphorylation is inferred from the observations that wild-type IRAK4 can phosphorylate the kinase-inactive construct in an in vitro assay, that autophosphorylation in vitro is concentration-dependent, and that overexpressed kinase-inactive IRAK4 is inducibly autophosphorylated in the wild type but not in an IRAK4-deficient cell line. The inducibility of IRAK4 autophosphorylation in the cellular context is dependent on the death domain. This is demonstrated by the fact that both the full-length form of kinase-active and kinase-inactive IRAK4 are inducibly phosphorylated following stimulation with IL-1β in wild-type human dermal fibroblasts. In contrast, kinase domain constructs exhibit different behavior from the full-length constructs under the same conditions. Kinase-active constructs of the IRAK4 kinase domain constitutively autophosphorylate when expressed in wild-type human dermal fibroblasts in the absence of IL-1β stimulation, but the kinase-inactive construct of the kinase domain is not phosphorylated in either the presence or absence of IL-1β stimulation. These data clearly demonstrate the importance of the death domain in regulating the kinase activity of IRAK4 in the cellular context.

It is likely that IRAK4 becomes activated because ligand binding by the IL-1/TLR receptor causes the aggregation/multimerization of receptors, which recruits the adaptor protein MyD88 via the TIR domains of the receptor and MyD88. This complex in turn recruits molecules of IRAK4 via the death domains of IRAK4 and MyD88 into the signaling complex. The high local concentration of IRAK4 in this complex promotes intermolecular autophosphorylation and activation of the kinase. Indeed, it has been shown that the death domain of MyD88 and the death domain of IRAK4 form multimeric complexes in vitro (25, 26). Thus, based on this model, the kinase-inactive form of the kinase domain does not become autophosphorylated because it lacks the death domain and is unable to associate with MyD88 and undergo intermolecular phosphorylation by the endogenous IRAK4 as observed in Fig. 5B.

This mechanism of activation is reminiscent of that proposed for Pelle, the Drosophila analog of the IRAKs (20, 27). In this model, Pelle and the adaptor protein Tube are pre-associated with the receptor Toll. Upon binding of the ligand Spatzle, the Toll receptors dimerize causing intermolecular autophosphorylation of Pelle and the subsequent activation of signaling. The high degree of homology between Pelle and human IRAK proteins supports a similar activation mechanism (28).

Another possible model of activation is one in which the death domain of IRAK4 associates with a protein that keeps IRAK4 inactive until it is released by receptor activation. The fact that both the kinase domain and the full-length protein are able to autophosphorylate in in vitro assays with similar rates (Fig. 2A) indicates that the death domain alone is insufficient to keep the kinase inactive. This implies that the death domain or its interacting protein in the cellular context is keeping the protein inactive until it is released upon stimulation by IL-1R or TLR ligands. A similar function has been proposed for the protein Tollip, which has been found to have a negative regulatory effect on IRAK1 (29).

Importantly, our data using the IRAK4 inhibitor compound 1 also support that IRAK4 phosphorylation is an autophosphorylation event and is associated with cytokine induction in a cell type-specific manner. The inhibitor used has an IC₅₀ for IRAK4 of 2 nm and for IRAK1 of 15 nm (19), making it difficult to definitively assess the contribution of either kinase on cellular signaling. However, assessment of ATP-binding site occupancy by the inhibitor in the cell shows that at the concentrations tested (10 μm), IRAK4 is 90% occupied, whereas IRAK1 is only 55% occupied by the inhibitor, indicating that most of the effects are derived from inhibition of IRAK4 (see supplemental Table S1). However, a more selective inhibitor of IRAK4 over IRAK1 will be required to clearly define the role of IRAK4 kinase activity.

We show that the inhibitor completely blocks IRAK4 phosphorylation but has minimal effects on activation of NF-κB, p38, JNK, and ERK in IL-1R or TLR7/8-stimulated human dermal fibroblasts and monocytes. The inhibitor does not block IL-6 and TNF-α production in human dermal fibroblasts, whereas these cytokines are blocked in human monocytes, suggesting cell type-specific requirements for IRAK4 kinase activity in human cells. Our data are not consistent with that of Chiang et al. (30), who showed no effect of an IRAK4/IRAK1 dual inhibitor on cytokine production in human monocytes. It is possible that the inhibitor used by Chiang et al. (30) was not potent enough to inhibit IRAK4 kinase activity as there was no measurement of pIRAK4 to gauge the extent of inhibition in the monocytes. Indeed, the potency of the inhibitor described by Chiang et al. (30) was considerably less potent against IRAK4 and IRAK1 than the inhibitor used in this study, and we have been unable to document inhibition of pIRAK4 in either IL-1β-stimulated dermal fibroblasts or R848-stimulated primary human monocytes at the concentrations described in that paper (data not shown).

The contribution of IRAK4 catalytic versus scaffold activity in regulating cytokine production is shown to be cell type-specific in human cells. In dermal fibroblasts, our data suggest IRAK4 utilizes its scaffolding activity to allow signaling and cytokine production to go forward unhindered in the absence of kinase activity. However, in monocytes, the kinase activity is required for cytokine production. This suggests cell type-specific requirements for IRAK4 scaffolding versus kinase activity consistent with previous reports (24, 30, 31). More detailed studies need to be undertaken on cell type differences in the requirement for the IRAK4 kinase. The minimal effect of an IRAK4 inhibitor on MAPK and NF-κB activation in both cell types is consistent with the study using macrophages from IRAK4 kinase-inactive knock-in mice (32, 33). These data suggest that IRAK4 kinase activity may impact a novel, yet to be determined pathway in regulating cytokine production. Perhaps, as suggested, the kinase activity may impact cytokines via targeting the mRNA stability of genes containing AU-rich elements in their 3′UTRs (32) via an undetermined pathway independent of NF-κB and MAPKs. Future work is needed to understand the precise mechanism of how IRAK4 kinase activity mediates cytokine production. The requirement for IRAK4 kinase activity for cytokine production in human immune cells such as monocytes supports the concept that IRAK4 inhibitors can be developed as potential therapeutic agents for treating autoimmune and inflammatory diseases, including lupus and rheumatoid arthritis.