The Scaffold MyD88 Acts to Couple Protein Kinase Cϵ to Toll-like Receptors*

Mice lacking protein kinase Cϵ (PKCϵ) are hypersensitive to both Gram-positive and Gram-negative bacterial infections; however, the mechanism of PKCϵ coupling to the Toll-like receptors (TLRs), responsible for pathogen detection, is poorly understood. Here we sought to investigate the mechanism of PKCϵ involvement in TLR signaling and found that PKCϵ is recruited to TLR4 and phosphorylated on two recently identified sites in response to lipopolysaccharide (LPS) stimulation. Phosphorylation at both of these sites (Ser-346 and Ser-368) resulted in PKCϵ binding to 14-3-3β. LPS-induced PKCϵ phosphorylation, 14-3-3β binding, and recruitment to TLR4 were all dependent on expression of the scaffold protein MyD88. In mouse embryo fibroblasts and activated macrophages from MyD88 knock-out mice, LPS-stimulated PKCϵ phosphorylation was reduced compared with wild type cells. Acute knockdown of MyD88 in LPS-responsive 293 cells also resulted in complete loss of Ser-346 phosphorylation and TLR4/PKCϵ association. By contrast, MyD88 overexpression in 293 cells resulted in constitutive phosphorylation of PKCϵ. A general role for MyD88 was evidenced by the finding that phosphorylation of PKCϵ was induced by the activation of all TLRs tested that signal through MyD88 (i.e. all except TLR3) both in RAW cells and in primary human macrophages. Functionally, it is established that phosphorylation of PKCϵ at these two sites is required for TLR4- and TLR2-induced NFκB reporter activation and IκB degradation in reconstituted PKCϵ–/– cells. This study therefore identifies the scaffold protein MyD88 as the link coupling TLRs to PKCϵ recruitment, phosphorylation, and downstream signaling.

Toll-like receptors (TLRs) 2 recognize microbial pathogenassociated molecular patterns and initiate common signaling pathways leading to specific inflammatory responses through activation of transcription factors such as nuclear factor B (NFB) and interferon regulatory factors (IRFs) (1). TLRs signal through Toll-IL1-R (TIR) domain-containing adaptor proteins that are recruited to receptor TIR domains upon ligand binding (2). Of five TIR domain-containing adaptors identified in humans, MyD88 has been shown to be involved in signal transduction for all TLRs except TLR3 (3,4). MyD88 deficiency in macrophages and dendritic cells leads to loss of MAPK activation, NFB activation, and proinflammatory cytokine production in response to various TLR ligands (5)(6)(7). However, some responses downstream of TLR4 are either only delayed (NFB and MAPK activation) or not affected (INF␤ production) in MyD88-deficient cells (8). These constitute MyD88-independent pathways and have led to the identification of other adaptor proteins. MyD88 adaptor-like (Mal) and Toll-IL-1R domaincontaining adaptor inducing interferon-␤ (TRIF)-related adaptor molecule (TRAM) work as bridging adaptors for MyD88 and TRIF to activate NFB and IRF3, respectively. Mal/ MyD88 signal from TLR2/TLR4 to regulate NFB activation (MyD88-dependent), whereas TRAM/TRIF signal to IRF3 in response to TLR4 activation (MyD88-independent) (9). The specificity in the activation of transcription factors by different TLRs using common signaling pathways is therefore achieved by differential use of adaptor proteins.
Protein kinase C (PKC) is a family of closely related serine/ threonine kinases that regulate diverse cellular processes such as proliferation, survival, immunity, and apoptosis (10,11). Based on the cofactor requirements, the PKC family is classified into three subfamilies as follows: conventional PKCs (␣, ␤I, ␤I⌱, and ␥) regulated by diacylglycerol, phosphatidylserine, and calcium; novel PKCs (␦, ⑀, , and ) regulated by diacylglycerol and phosphatidylserine; and atypical PKCs ( and /) that do not require diacylglycerol for activation (12). Studies in mice lacking different PKC isoforms have established an important role for PKCs in intracellular immune signaling (reviewed in Ref. 13). PKC␤ knock-out mice have an immunodeficiency because of defective B cell activation (14), whereas PKC is required for T cell receptor-mediated T cell activation (15). Mice deficient in PKC␦ and PKC have defective B cell anergy and NFB signaling, respectively (16,17). PKC⑀ Ϫ/Ϫ mice have impaired innate immunity and fail to clear Gram-positive and Gramnegative bacterial infection (18). LPS and INF␥-induced nitric oxide, tumor necrosis factor-␣, IL1␤, and prostaglandin E 2 production in PKC⑀ Ϫ/Ϫ macrophages is reduced, and this is attributed to the impaired NFB activation upstream of IB kinase ␤. Other PKC isoforms have also been implicated in TLR signal-ing. PKC␣ is involved in LPS-and poly(I-C)-induced NF-IL6 (19) and IRF3 (20) activation, respectively. LPS-induced MAPK activation, tumor necrosis factor-␣ production, and NFB activation were shown to be PKC-dependent in different cell types (21)(22)(23). Similarly, involvement of PKC⑀ in MKP-1 and IL12 induction in response to LPS stimulation in macrophages and dendritic cells, respectively, has been demonstrated (24,25). Despite many studies implicating PKC isoforms in TLR signaling, there is little evidence on the mechanism of their involvement. Recently, TRAM has been identified as a substrate of PKC⑀, and its phosphorylation has been shown to regulate RANTES production through IRF3 activation (26). However, the mechanism of PKC⑀ activation in response to LPS remains obscure, and although TRAM phosphorylation by PKC⑀ was required for it to signal, the exact function of this phosphorylation remains elusive. Similarly, Kubo-Murai et al. (27) have shown recently that PKC␦ binds to Mal, and this binding promotes TLR2 and TLR4 signaling to p38 MAPK and IB.
We have recently identified novel phosphorylation sites (Ser-346 and Ser-368) in the V3 region of PKC⑀ that regulate its association with 14-3-3␤. 3 Phosphorylation at these sites occurs sequentially through p38 (Ser-350) followed by GSK-3␤ (Ser-346) and auto-phosphorylation or classic PKC trans-phosphorylation (Ser-368) (43). The subsequent binding of 14-3-3␤ is required for efficient separation of cells at the end of cytokinesis. Here we investigated the role of PKC⑀ in TLR signaling and discovered that LPS induced both recruitment of PKC⑀ to TLR4 and its phosphorylation at Ser-346 and Ser-368, resulting in its association with 14-3-3␤. PKC⑀ recruitment to TLR4, phosphorylation, and binding to 14-3-3␤ were all dependent on MyD88 expression. We therefore propose that MyD88 represents the missing link that couples PKC⑀ to TLR4 in response to LPS.

EXPERIMENTAL PROCEDURES
Reagents-Lipopolysaccharide (L7261) and lipoteichoic acid (L2515) were purchased from Sigma, and other TLR ligands were from Invivogen. All the inhibitors except BIRB 796 (a gift from Dr. Ana Cuenda, Dundee, Scotland, UK) were from Calbiochem. Horseradish peroxidase-conjugated anti-mouse and anti-rabbit antibodies, ECL reagent, glutathione-Sepharose, and protein G-Sepharose were from Amersham Biosciences. The dual luciferase reporter system was purchased from Promega. Rabbit polyclonal antibodies against PKC⑀ (C-15) and MyD88 (HFL-296) were from Santa Cruz Biotechnology. Mouse monoclonal anti-FLAG-M2 antibodies were obtained from Sigma. Mouse monoclonal Anti-GFP antibodies 3E1 (for Western blot) and 4E12/8 (for immunoprecipitation) were from the London Research Institute monoclonal facility. Rabbit polyclonal anti-phospho-p38 antibodies were from Cell Signaling. Generation of phospho-specific antibodies to serine 346 and serine 368 were carried out essentially as described previously (28) using the immunogens DRSKS(P)APTS and KIT-NS(P)GQRR, respectively. All other reagents were from Sigma.
Generation of Stable Cell Lines-ϳ70% confluent RAW cells in 10-cm plates were transfected with 10 g/plate of the plasmid DNA (GFP-PKC⑀ WT, Ser-346/S368A, S346A, S368A, or vector control) using Lipofectamine 2000. Cells were split into 15-cm plates the next day and selected with 500 g/ml Zeocin (Invitrogen) or 1 mg/ml G418 (depending on the constructs). Single clones were picked, and the rest were pooled and analyzed for GFP-PKC⑀ expression by Western blot. PKC⑀ MEFs stably expressing different GFP-PKC⑀ constructs were generated by transfection and Zeocin selection of polyclonal populations. 3 siRNA Knockdown-MyD88-N siRNA duplex (Qiagen) targeted the N-terminal region (nucleotides 181-201) of human MyD88 and had the following sequence 5-CCGGCAACUG-GAGACACAAdTdT-3 and 5-UUGUGUCUCCAGUUGCCG-GdAdT-3. 2 ϫ 10 5 293/hTLR4 cells in 6-well plates were transfected with 50 nM siRNA using 5 l of Lipofectamine RNAiMAX (Invitrogen) according to the manufacturer's instructions. 48 h after transfection with siRNA, cells were retransfected with GFP-PKC⑀. For experiments investigating PKC⑀-TLR4 interaction after MyD88 knockdown, 60-mm plates were used, and 48 h after MyD88 siRNA transfection cells were co-transfected with GFP-PKC⑀ and FLAG-TLR4. Cells were analyzed after a further 24 h. Immunoprecipitation, Pulldown, and Western Blot Analysis-Cell lysis was carried out on ice with lysis buffer containing 1% Nonidet P-40, 50 mM Tris-HCl, pH 7.4, 120 mM NaCl, 5 mM sodium vanadate, 50 mM sodium fluoride, and EDTA-free protease inhibitor tablet from Roche Applied Science. Protein concentration was measured, and equal amounts of protein were immunoprecipitated by incubation at 4°C with anti-GFP and anti-FLAG antibodies as indicated for 2 h followed by a further 1-h incubation with protein G-Sepharose. Beads were washed twice with TNET (TNE ϩ 1% Triton X-100) and once with TNE (50 mM Tris-HCl, pH 7, 140 mM NaCl, and 5 mM EDTA) and boiled in LDS sample buffer. Proteins were resolved by 4 -12% NuPAGE gels, transferred to polyvinylidene difluoride membranes, and immunoblotted with specific antibodies. GST-14-3-3␤ pulldown assays were performed using bacterially expressed GST-14-3-3␤-loaded glutathione beads. Native extracts were tumbled overnight at 4°C prior to washing twice with TNET (TNE ϩ 1% Triton X-100) and once with TNE (50 mM Tris-HCl, pH 7, 140 mM NaCl, and 5 mM EDTA) and elution in LDS buffer.
Isolation of Thioglycollate-elicited Peritoneal Macrophages-Peritoneal macrophages were isolated as described previously (18). Briefly, mice (MyD88 knock-out and C57 black/6 WT) were injected with 3% thioglycollate intraperitoneally and sacrificed after 4 days. The peritoneum was flushed with sterilized phosphate-buffered saline (containing 5 mM EDTA), and the peritoneal suspension containing macrophages was carefully removed and centrifuged at 200 ϫ g for 10 min. Cells were resuspended and seeded in RPMI 1640 medium with 10% FCS in 10-cm plates at 37°C with 10% CO 2 . After 1 h of incubation, nonadherent cells were removed by extensive washing with phosphate-buffered saline. Cells were used for experiments on the following day.

LPS Triggers
Phosphorylation of PKC⑀-The defective innate immunity of PKC⑀ knock-out mice formally defines PKC⑀ as a key regulator of TLR responses. For Gram-negative bacteria the relevant signaling paradigm involves LPS stimulation of TLR4; however, there is little direct evidence that TLR4 activation engages PKC⑀. Recent studies have identified two phosphorylation sites within the V3 region of PKC⑀ that enable its binding to 14-3-3␤ and that are associated with the engagement of PKC⑀ in signaling processes. 3 To determine whether LPS activation of TLR4 triggers these responses, RAW 264.7 cells (here referred to as RAW cells) stably expressing GFP-PKC⑀ were treated with LPS, and phosphorylation of PKC⑀ was determined by Western blot using phospho-specific antibodies. Phosphorylation of both the Ser-346 and Ser-368 sites was found to be induced in response to LPS (Fig. 1A); a similar induction was also seen in GFP-PKC⑀-transfected 293 cells stably expressing the human proteins TLR4, MD2, and CD14 (Fig.  1B). There is a degree of constitutive phosphorylation of PKC⑀ at the Ser-368 site in both cell types; however, there is clear LPS-dependent induction. LPS also induced Ser-346 phosphorylation of GFP-PKC⑀ in HEK 293 cells transiently expressing TLR4/MD2; however, no phosphorylation was observed in vector control transfected cells (data not shown). To test the ng/ml LPS and analyzed as described above. C, 14-3-3␤ binding of endogenous PKC⑀. PKC⑀ from LPS-stimulated RAW GFP-PKC⑀ cells was pulled down with GST-14-3-3␤ (lower panel) and analyzed by Western blot using anti-PKC⑀ antibodies. Membranes from pulldown were stained with Coomassie for GST-14-3-3␤ as loading controls. TCL, total cell lysate. D, PKC⑀ phosphorylation in primary human macrophages. Human monocytes were differentiated into macrophages as described under "Experimental Procedures" and stimulated with 200 ng/ml LPS for times indicated. Cell lysates were pulled down with GST-14-3-3␤ and blotted with PKC⑀ antibodies. E, phosphorylation of both Ser-346 and Ser-368 is required for 14-3-3␤ binding. RAW cells stably expressing different mutants of GFP-PKC⑀ were stimulated with 200 ng/ml LPS for 50 min, and a GST-14-3-3␤ pulldown assay was performed. Total cell lysates were probed with PKC⑀ antibodies for equal expression in different clones. AA, S346A/A368A. F and G, time and concentration dependence of PKC⑀ phosphorylation. GFP-PKC⑀ expressing RAW cells were either stimulated with 200 ng/ml of LPS for different times (F) or with different concentrations of LPS for 30 min (G). Cells lysates were immunoprecipitated and analyzed for Ser-346 and Ser-368 phosphorylation using site-specific antibodies.
response of the endogenous PKC⑀ in terms of phosphorylation at Ser-346 and Ser-368, use was made of the 14-3-3␤ interaction of Ser-346/368 doubly phosphorylated PKC⑀ (see below). Although this assay does not distinguish whether both sites are induced for the endogenous protein, in light of the observations above it is anticipated that both are increased (serum titers for the site-specific antisera were found to be inadequate for Western blotting the endogenous protein). Pulldown with GST-14-3-3␤ demonstrated that LPS induced an increased recovery of PKC⑀ for both ectopic GFP-PKC⑀ and the endogenous PKC⑀ (Fig. 1C). A similar pulldown assay was used to demonstrate LPS-induced PKC⑀ phosphorylation in primary human macrophages (Fig. 1D). The absolute requirement for dual 346/368 phosphorylation for the 14-3-3␤ interaction was confirmed through the use of single and double Ser 3 Ala mutants (Fig. 1E).
The LPS-induced responses in RAW cells expressing GFP-PKC⑀ were time-and dose-dependent (Fig. 1, F and G). For both the Ser-346 and Ser-368 sites the responses peak at 20 -30 min, and both are effectively at basal levels by 2 h. As illustrated (Fig. 1F), the induction of Ser-346 and Ser-368 was Ͼ7and Ͼ2-fold, respectively. Over a series of experiments the mean induction for Ser-346 was 7 Ϯ 1.01-and 2.5 Ϯ 0.16-fold for Ser-368. Consistent with the mechanism of Ser-346 phosphorylation in fibroblasts (Ser-350 phosphorylation by p38 primes Ser-346 for phosphorylation by GSK3-␤; see below), activation site phosphorylation of p38 MAPK was induced and peaked just ahead of PKC⑀ Ser-346 phosphorylation. After 30 min of stimulation, phosphorylation of Ser-346 was induced by as little as 10 ng/ml of LPS and was maximum at 25 ng/ml. LPS Induces PKC⑀ Phosphorylation via p38␥/␦ and GSK-3-Previous studies in fibroblasts have identified Ser-346 as a target for GSK-3, primed by p38␣/␤ phosphorylation at Ser-350. To determine whether these same pathways account for the LPS-induced response in macrophages, cells were stimulated with LPS and various inhibitors assessed for their effects on Ser-346 phosphorylation. Three different GSK-3 inhibitors were found to block LPS-induced Ser-346 phosphorylation ( Fig. 2A and data not shown for SB216763). Inhibition of all four p38 MAPK isoforms (with BIRB 796 (32)) also blocked Ser-346 phosphorylation indicative of a priming role (Fig. 2B). However, the p38␣/␤-selective inhibitors failed to block LPS-induced PKC⑀ Ser-346 phosphorylation (see further below). Previously, p38␣/␤ have been implicated in UV-induced PKC⑀ Ser-346 phosphorylation in fibroblasts. 3 Hence the effect of LPS was investigated in MEFs expressing GFP-PKC⑀. As in RAW cells, pan-p38 MAPK inhibition blocked LPS-induced Ser-346 phosphorylation, whereas the p38␣/␤-selective inhibitors (SB203580 and SB202190) were not inhibitory (Fig. 2C).
To confirm the specificity of this behavior, the UV response was re-assessed in MEFs in parallel to the LPS response. As shown in Fig. 2D, BIRB inhibited both the LPS-and UV-induced phosphorylation of PKC⑀ at the Ser-346 site. However, whereas SB203580 inhibited the UV-induced response, it did not inhibit the LPS-induced response. These distinct patterns of behavior are indicative of selective activation and/or targeting of specific p38 MAPKs to the same priming phosphorylation of PKC⑀ in response to distinct agonists.
To assess the expected PKC dependence for Ser-368 phosphorylation, LPS-or LTA (TLR2 ligand; see further below)-stimulated cells were pretreated with BIMI (classic PKC ϩ novel PKC inhibitor). Unexpectedly, no inhibition of Ser-368 phosphorylation was observed even at high BIMI concentration (Fig.  2E), although there was an inhibition by Go6976, a novel PKC inhibitor (data not shown) indicating that a distinct basophilic protein kinase is involved in the LPS response. Thus, although LPS triggers the phosphorylation of the 14-3-3␤binding sites of PKC⑀ in both macrophages and fibroblasts, it does so via p38␥/␦ ϩ GSK-3 alongside an unknown basophilic kinase. This contrasts with the p38␣/␤ ϩ GSK-3 and PKC-dependent phosphorylation of these sites under other conditions. 3 Notwithstanding this notable distinction, it is evident that LPS/TLR4 induce phosphorylation of PKC⑀ in line with the established in vivo requirement for this kinase.
Multiple TLR Ligands Trigger PKC⑀ Phosphorylation-PKC⑀ knockout mice are defective in the clearance of both Gram-positive and Gram-negative bacteria. To determine whether model Gram-positive ligands acting via TLR2 promote PKC⑀ phosphorylation as determined above for LPS-TLR4, RAW cells were stimulated with either LTA or FSL-1 (TLR2/TLR6). Both ligands were found to increase PKC⑀ Ser-346 and Ser-368 phosphorylation (Fig. 3A), although the Ser-368 site was delayed relative to the Ser-346 site and induced only an ϳ2-fold increase. In view of the conservation of this response, ligands engaging TLRs 1-9 were tested (Fig.  3B). All but the TLR3 ligand poly(I-C) stimulated PKC⑀ Ser-346 phosphorylation. To ensure that poly(I-C) was acting via TLR3, 293/ hTLR4 cells were transfected with TLR3 or TLR4, and responses to poly(I-C) and LPS were monitored. LPS-induced phosphorylation of (1 g/ml), FSL-1(100 ng/ml), poly(I-C) (10 g/ml), flagellin (10 g/ml), single strand RNA (10 g/ml), or ODN1826 (10 g/ml). Cells were lysed, and cell lysates were immunoprecipitated (IP) with anti-GFP antibodies. Immunoprecipitates were analyzed by immunoblotting with anti-Ser-346 and GFP antibodies. C, TLR4 but not TLR3 induces Ser-346 phosphorylation in 293 cells. FLAG-TLR4 and FLAG-TLR3 were overexpressed in 293/hTLR4 cells. Cells were stimulated with LPS (200 ng/ml) or poly(I-C) (10 g/ml) for 60 min and lysed, and cell lysates were analyzed for Ser-346 phosphorylation and TLR expression by immunoblotting. D, 293hTLR4 cells overexpressing F-TLR3 respond to poly(I-C). 293/hTLR4 cells were co-transfected with F-TLR4 or F-TLR3 along with an IRF-3-dependent reporter (pGL3-561) and Renilla control overnight. Cells were then stimulated with LPS (200 ng/ml) or poly(I-C) (10 g/ml) for 6 h. IRF-3 reporter activation was analyzed by the dual luciferase assay system according to the manufacturer's instructions. E, PKC⑀ phosphorylation by different TLR ligands in primary human macrophages. Human macrophages, differentiated from monocytes as described under "Experimental Procedures," were treated with 200 ng/ml LPS, 10 g/ml LTA, 20 g/ml poly(I-C), or 100 ng/ml flagellin for 50 min. Cells were lysed, and cell lysates were pulled down with GST-14-3-3␤ overnight and blotted with anti-PKC⑀ antibodies. Total cell lysates were analyzed for protein loading, and the membrane from pulldown was stained with Coomassie for GST-14-3-3␤ loading.
Ser-346 in TLR4 expressing cells was observed, but no such response to poly(I-C) in TLR3-expressing cells was seen (Fig. 3C). However, poly(I-C) did induce an IRF-3-dependent reporter response in TLR3-expressing cells but not in cells expressing TLR4 (Fig.  3D). Thus a functionally linked TLR3 receptor is not linked to PKC⑀ phosphorylation.
A similar PKC⑀ response was observed for primary human macrophages. In freshly isolated macrophages challenged with LPS, LTA, poly(I-C), or flagellin, an increase in PKC⑀ phosphorylation at both Ser-346 and Ser-368 was observed with all ligands except poly(I-C) (TLR3) as indicated by PKC⑀ binding to GST-14-3-3␤ (Fig. 3E). The induction observed for LTA was ϳ2-fold compared with the 4 -6-fold induction for LPS and flagellin. As noted above these responses are indicative of ligand-induced PKC⑀ phosphorylation at 346/368 sites, although we cannot distinguish the changes in the individual sites by this procedure.
MyD88 Links PKC⑀ to TLRs-The pattern of PKC⑀ responses to these TLRs parallels their engagement of MyD88, i.e. all but TLR3. To test whether MyD88 was responsible for linking PKC⑀ to TLRs, cells from MyD88 knock-out mice were tested for responses to LPS using capture on GST-14-3-3␤ beads to monitor endogenous PKC⑀ Ser-346/Ser-368 phosphorylation as evidenced by the increased recovery of PKC⑀ complexed to 14-3-3␤. By contrast to MyD88-replete MEFs, no response to LPS was observed in MyD88 knock-out cells (Fig. 4A). Peritoneally elicited macrophages from WT mice also responded to LPS with a substantial increase in PKC⑀ recovered in a 14-3-3␤ pulldown, equivalent to that observed with the potent combination of TPA/calyculin. By contrast there was no LPS-induced recovery of PKC⑀ from the MyD88 knock-out macrophages, despite a "WT" response to TPA/calyculin (Fig. 4B).
The evidence from the knock-out model demonstrates a requirement for MyD88 in the TLR4-triggered PKC⑀ response. To confirm this in an acute model, siRNAs to MyD88 were employed to knock down MyD88 expression. As illustrated in Fig. 4C, siRNA knockdown of MyD88 also abrogated the LPS-induced PKC⑀ Ser-346 phosphorylation. The specificity of the effect of MyD88 knockdown on Ser-346 phosphorylation was confirmed by rescue experiments with mouse MyD88. As shown in Fig. 4D, reexpression of mouse MyD88 in cells with knockdown of endogenous MyD88 by human-specific siRNA rescued the Ser-346 phosphorylation. Interestingly, when FLAG-tagged MyD88 or untagged MyD88 is overexpressed in 293/hTLR4 cells, constitutively high Ser-346 phosphorylation is observed with no further increase in LPS stimulation (Figs. 4, E and F). This reflects elevated p38 phosphorylation in these MyD88 overexpressing cells (data not shown).
PKC⑀ Is Complexed with TLR4 via MyD88-The adaptor role of MyD88 and its requirement for linking PKC⑀ to TLR4 suggested that PKC⑀ may be physically associated with the (active) receptor. By employing FLAG-tagged TLR4, it was found that a fraction of co-expressed GFP-PKC⑀ or myc-PKC⑀ (Fig. 5A) could be recovered in TLR4 immunoprecipitates in an LPSinducible manner; much lower levels of PKC⑀ were recovered in anti-FLAG control immunoprecipitates (data not shown and see Fig. 5B). The complex formation between TLR4 and PKC⑀ was further supported by co-immunoprecipitation of YFP-TLR4 and myc-PKC⑀ using both anti-GFP (YFP-TLR4) and anti-Myc antibodies (data not shown). To map the PKC⑀ domain required for its recruitment to TLR4, GFP-PKC⑀ regulatory and GST-PKC⑀ catalytic domains were used. GFP-PKC⑀ regulatory domain was recovered in FLAG-TLR4 immunoprecipitates from 293/hTLR4 cells (Fig. 5B), whereas GST-PKC⑀ catalytic domain could not be recovered with FLAG-TLR4 (data not shown). The constitutive basal recovery of PKC⑀ in TLR4 immunocomplexes was enhanced by co-expression of MyD88 (Fig. 5C). siRNAs to MyD88 were employed to determine its requirement in the PKC⑀-TLR4 interaction. FLAG-TLR4 interaction with GFP-PKC⑀, determined by immunoprecipitation as described above, was reduced in 293/hTLR4 cells transfected with MyD88 siRNA as compared with the control siRNA (Fig. 5D). We further confirmed this by a rescue experiment, in which endogenous MyD88 was knocked down in 293/ hTLR4 cells and myc-PKC⑀ and YFP-TLR4 interaction was determined by immunoprecipitation in the presence of vector control or mouse MyD88 (Fig. 5E). The effect of MyD88 knockdown on PKC⑀-TLR4 interaction was completely recovered by expression of ectopic mouse MyD88.
Phosphorylation of PKC⑀ at the 14-3-3␤-Binding Sites Is Required for NFB Transcriptional Activation-To assess the role of PKC⑀ phosphorylation downstream of TLRs, stable cell lines expressing matched amounts of WT GFP-PKC⑀ and an S346A/ S368A PKC⑀ mutant were tested for activation of a luciferase NFB reporter. WT PKC⑀ expression enhanced LTA-and LPSinduced luciferase expression. By contrast, expression of the S346A/S368A PKC⑀ mutant failed to facilitate luciferase expression (Fig. 6A). To ensure that this distinction was not an artifact of the clonal isolates, pools of stably expressing cells were also tested. As observed for the clonal isolates, the WT protein supported induction of luciferase, whereas the mutant did not (Fig. 6B).
To determine whether the effects of PKC WT expression were exerted through the control of IB degradation, WT and mutant PKC⑀ were compared for their ability to support LTA-induced IB degradation. Although the WT protein was effective in supporting IB degradation in response to LTA, the mutant was not (Fig. 6C).

DISCUSSION
Activated macrophages play an integral part in launching an effective immune response to microbial infections (33). Studies using specific inhibitors and dominant interfering mutants of FIGURE 5. TLR4-PKC⑀ association. A, PKC⑀ co-immunoprecipitation with TLR4. FLAG-TLR4 was co-transfected with GFP-PKC⑀ or myc-PKC⑀ in 293/ hTRL4 cells. 24 h after transfection, cells were stimulated with LPS (200 ng/ml) for 50 min, and cells were lysed. Cell lysates were immunoprecipitated (IP) with anti-FLAG antibodies, and immunoprecipitates were analyzed by immunoblotting with anti-PKC⑀ or anti-FLAG antibodies. B, PKC⑀ regulatory domain association with F-TLR4. 293/hTLR4 cells were co-transfected with F-TLR4 and GFP-PKC⑀ and stimulated with LPS (200 ng/ml) after 24 h. Cells were lysed, and lysates were analyzed for PKC⑀-TLR4 association by co-immunoprecipitation as described above. TCL, total cell lysate. C, MyD88 overexpression enhances TLR4-PKC⑀ association. MyD88 or empty vector was co-transfected with GFP-PKC⑀ and F-TLR4. 24 h after transfection cells were stimulated with LPS (200 ng/ml) for 50 min and lysed, and cell lysates were analyzed for TLR4-PKC⑀ association as described above. D, MyD88 knockdown suppresses TLR4-PKC⑀ association. 293/hTLR4 cells were transfected with MyD88-N siRNA, followed 48 h later by co-transfection of FLAG-TLR4 and GFP-PKC⑀. Cells were stimulated 24 h later with LPS, lysed, and immunoprecipitated with anti-FLAG antibodies. Immunoprecipitates were analyzed by immunoblotting using GFP antibodies. E, rescue of TLR4-PKC⑀ association in MyD88 knockdown cells by re-expression of MyD88. MyD88 was knocked down in 293/hTLR4 cells by MyD88-N siRNA transfection, and cells were co-transfected after 48 h with YFP-TLR4 and Myc-PKC⑀Ϯ FLAG-tagged mouse MyD88. Cells were analyzed 24 h later for TLR4-PKC⑀ association by immunoprecipitation with anti-GFP antibodies and immunoblotting with PKC⑀, GFP, and FLAG antibodies. different isoforms have identified PKCs as important mediators of endotoxin signaling in macrophages (19,23,25,34). Previously we showed that targeted disruption of the PKC⑀ gene in mice leads to compromised innate immunity (18). PKC⑀ Ϫ/Ϫ mice had defects in clearance of both Gram-positive (TLR2) and Gram-negative (TLR4) bacterial infections. LPS-induced signaling in macrophages from these mice was severely attenuated resulting in reduced production of pro-inflammatory cytokines. Here we sought to investigate the mechanism of PKC⑀ coupling to the TLR and found it to be phosphorylated and in a complex with TLR4 upon LPS stimulation. PKC⑀ recruitment to TLR4 and the phosphorylation that resulted in its ability to bind 14-3-3␤ were both MyD88-dependent. The requirement for the recruitment of PKC⑀ and its subsequent phosphorylation downstream of TLRs were evidenced by the effects of the WT and mutant PKC⑀ in supporting NFB-dependent luciferase expression.
Most of the studies implicating PKC⑀ in TLR signaling have analyzed its function in response to LPS (TLR4) apart from in vivo studies with Gram-positive (TLR2) and Gram-negative (TLR4) bacteria (18). Recently, McGettrick et al. (26) have demonstrated that IRF3 activation and RANTES production in response to LPS and not poly(I-C) is regulated by PKC⑀ phosphorylation of TRAM. Because TRAM works as a bridging adaptor for TRIF in the MyD88-independent pathway downstream of TLR4, this might imply a restricted role for PKC⑀ in MyD88-independent TLR4 signaling (35,36). The evidence here, however, indicates that PKC⑀ gets phosphorylated in response to activation of TLRs that signal through MyD88 in RAW cells and primary human macrophages (Fig. 3, B and E), thereby implying a broader role in MyD88-dependent pathways. Ligands for TLR4 and TLR2/6 induced Ser-346 and Ser-368 phosphorylation that resulted in PKC⑀ association with 14-3-3␤ in vitro (Figs. 1, A and B, and 3A). PKC⑀ phosphorylation and the subsequent 14-3-3␤ binding capacity elicited by TLR4 activation were MyD88-dependent, and the evidence for this dependence was provided by three different approaches. (a) PKC⑀ binding to GST-14-3-3␤ in MyD88 knock-out MEFs and thioglycollate-elicited macrophages in response to LPS was compromised compared with their WT counterparts (Fig. 4, A  and B). (b) Knockdown of MyD88 significantly reduced LPSinduced PKC⑀ phosphorylation at Ser-346 in 293/hTLR4 cells (Fig. 4, C and D). (c) Overexpression of MyD88 in 293/hTLR4 cells induced ligand-independent phosphorylation of PKC⑀ and recovery on 14-3-3␤ beads (Fig. 4, E and F, and data not shown). Because PKC⑀ binding to 14-3-3␤ requires its phosphorylation at both Ser-346 and Ser-368 (Fig. 1E), LPS-induced recovery of PKC⑀ by 14-3-3␤ from RAW cells, primary human macrophages, mouse embryo fibroblasts, and thioglycollateelicited macrophages (Figs. 1, 3, and 4) demonstrated that the endogenous protein behaved like the ectopic PKC⑀ and was therefore phosphorylated at both of these required sites and that one or both sites were induced in response to ligand. So LPS-induced PKC⑀ phosphorylation in a variety of cell lines (including immortalized and primary cells) suggests a conserved role of PKC⑀ in TLR4 signaling. LPS also induced recruitment of a small fraction of PKC⑀ to TLR4. PKC⑀ complexed with TLR4 was phosphorylated on Ser-346 (data not shown). Similar to its phosphorylation, PKC⑀ interaction with TLR4 was also dependent on MyD88, which we therefore can conclude recruits the kinase to TLR4. Consistent with this conclusion, overexpression of MyD88-induced ligand-independent binding of PKC⑀ to TLR4 and MyD88 knockdown reduced this association in a manner rescued by re-expression of an siRNA-resistant MyD88 (Fig. 5, C, D, and E).
Several studies have shown phospholipase C activation and diacylglycerol production in response to LPS (37)(38)(39). Therefore, it follows that diacylglycerol production after LPS stimulation might be important in PKC⑀ recruitment and activation. A PIP 2 binding domain has been identified in the bridging adaptor Mal/TIRAP that recruits it to the membrane upon LPS NFB reporter and Renilla control plasmids were transfected in MEF clones at a 10:1 ratio in 24-well plates in duplicate overnight. Cells were stimulated with 25ng/ml LPS or 10 g/ml LTA for 7 h. NFB reporter activation was analyzed by the dual luciferase assay system according to the manufacturer's instructions. B, LPS-and LTA-induced NFB reporter activation in PKC⑀ Ϫ/Ϫ MEF clones and pool cells expressing vector, GFP-PKC⑀ WT, or GFP-PKC⑀ AA. Cells were transfected with NFB and Renilla control plasmids, stimulated, and analyzed for reporter activation as described above. C, PKC⑀ WT enhances IB degradation in PKC⑀ Ϫ/Ϫ MEFs in response to TLR2 activation. PKC⑀ Ϫ/Ϫ MEF clones expressing vector, GFP-PKC⑀ WT, or GFP-PKC⑀ AA were stimulated with 20 g/ml LTA for the indicated times. Cells were collected and analyzed for IB degradation, PKC⑀ expression, and loading by Western blot analysis using specific antibodies. stimulation (40). Mal then delivers MyD88 to TLR4 for further downstream signaling. As noted above, PKC⑀ is recruited to the activated TLR4 by MyD88 (Fig. 5). Because MyD88 recruitment to TLR4 itself is mediated by Mal and is PIP 2 -dependent, PKC⑀ recruitment to TLR4 might also be dependent on Mal and PIP 2 . However, PKC⑀ phosphorylation was observed in response to several TLRs, including TLR7 and TLR9, which are localized to intracellular compartments potentially less rich in PIP 2 , suggesting a Mal and PIP 2 -independent recruitment of PKC⑀. The implication is that MyD88 plays the dominant role in this response.
Two different phosphoserine/threonine-containing motifs have been identified in 14-3-3-binding proteins (41). The Ser-346 site (RSKSAP) is within an optimal 14-3-3 binding mode I (RSX(pS/T)XP) motif, whereas the sequence surrounding Ser-368 (RKALSFD) resembles a mode II (RXXX(pS/T)XP) motif, although lacking a proline at the ϩ2 position. Recently we have characterized the 14-3-3-PKC⑀ binding in detail and demonstrated that the 14-3-3␤ dimer binds to PKC⑀ phosphorylated at both Ser-346 and Ser-368 in a 1:1 dimer-PKC⑀ complex. 3 Phosphorylation of both sites is also required for 14-3-3␤ binding to PKC⑀ after LPS stimulation, and mutation of either or both of these sites to alanine abolishes this binding (Fig. 1). Therefore, the 14-3-3␤ binding resulting from the phosphorylation at Ser-346 and Ser-368 downstream of TLR4/MyD88 is tightly regulated by different inputs. Downstream of TLR4 these inputs come from p38␥/␦ (priming phosphorylation of Ser-350), GSK-3␤ (Ser-346), and an unknown basophilic kinase (Ser-368). LPS-induced phosphorylation of p38 precedes Ser-346 phosphorylation (Fig. 1F), supporting its priming role for Ser-346 phosphorylation by GSK-3␤. Using pan-p38 and p38␣/ ␤-specific inhibitors (BIRB 796 and SB203580, respectively), we demonstrated that, depending on the signal, different p38 isoforms can prime PKC⑀ for Ser-346 phosphorylation in the same cell model, indicative of selective activation and/or targeting of p38 family members by different agonists (Fig. 2). For LPSinduced priming, p38␥/␦ was involved in both RAW cells and MEFs. Ser-368 was expected to be an auto-phosphorylation site; however, BIM I did not inhibit the phosphorylation in response to LPS, although it was able to inhibit the response to UV. This ruled out auto-phosphorylation and/or trans-phosphorylation by classical or novel PKC isoforms. A classical PKC inhibitor Go6976 did, however, very efficiently inhibit LPS-induced Ser-368 phosphorylation (Fig. 2E). Considering no inhibition by BIM I, this effect of Go6976 indicates that a non-PKC Go6976-sensitive kinase is involved.
14-3-3 binding can regulate various properties of its target proteins, including localization, stability, activity, and/or interactions with other proteins (42). We therefore tested the possible role of 14-3-3 binding in PKC⑀ recruitment to TLR4. However, there was no difference in TLR4 recruitment of WT or the S346A/S368A mutant PKC⑀ in response to LPS, thereby excluding a targeting role for this complex (data not shown).14-3-3␤ binding locks PKC⑀ in an open conformation, 3 thereby regulating its lipid-independent activity. We propose that a similar mechanism of lipid-independent activation of PKC⑀ by 14-3-3␤ binding exists in response to TLR activation. However it remains to be determined whether a 14-3-3␤-binding defective mutant of PKC⑀ gets activated in response to LPS or not.
In conclusion, we have shown that PKC⑀ is linked to various TLRs through the adaptor protein MyD88. This serves both to recruit the kinase to the receptor and to enable its phosphorylation at previously defined 14-3-3␤-binding sites. These events are shown to be critical for the subsequent degradation of IB and activation of NFB. This novel mechanism of receptor coupling to PKC⑀ therefore underlies signaling downstream of TLRs, which explains the compromised innate immune response in PKC⑀ knock-out mice.