Protein Kinase C δ Associates with the Interleukin-6 Receptor Subunit Glycoprotein (gp) 130 via Stat3 and Enhances Stat3-gp130 Interaction*

The transcriptional regulation of Stat proteins is controlled through their C-terminal domains, which harbor both a tyrosine phosphorylation site, required for dimerization and subsequent nuclear translocation, and a serine phosphorylation site, required for maximum transcriptional activity. Previously, we reported that protein kinase Cδ (PKCδ) phosphorylates and interacts with Stat3 in an interleukin (IL)-6-dependent manner. In this study, we further characterized this interaction, and investigated the potential role of such an interaction. We show here that the catalytic domain of PKCδ interacts with the Src homology 2 domain and part of the adjacent C-terminal transactivation domain of Stat3. This interaction, which does not seem to involve a classical phosphotyrosine SH2-mediated binding, however, significantly enhances the interaction of Stat3 and the IL-6 receptor subunit glycoprotein (gp) 130, which is the initial step for Stat3 activation by IL-6. Expression of a dominant negative PKCδ or depletion of the endogenous PKCδ by phorbol 12-myristate 3-acetate treatment abrogates the association of Stat3 with gp130. At the same time, PKCδ is recruited to gp130 via association with Stat3, which may facilitate its phosphorylation on the gp130 receptor. Finally, we identified Thr-890, a putative PKC phosphorylation site on gp130, to be critical for the effect of PKCδ. Our data indicate that PKCδ plays important regulatory roles in IL-6 signaling.

The interleukin-6 (IL-6) 1 -type cytokines play pleiotropic roles in immunity, hematopoiesis, inflammation, liver regeneration, hepatocyte maturation, and neural development (1,2). This family of cytokines consists of IL-6, IL-11, leukemia inhibitory factor (LIF), ciliary neurotrophic factor, oncostatin M (OSM), cardiotrophin (CT-1), and neurotrophin-1/B-cell stimulatory factor-3/cardiotrophin-like cytokine (NNT-1/BSF3/CLC), which share glycoprotein (gp) 130 as a common receptor for signal transduction. Binding of the IL-6 family cytokines to their specific receptor subunits leads to homodimerization of gp130 or heterodimerization of gp130 with gp130-related receptors including LIF-receptor ␤, OSM receptor ␣, or CT-1 receptor ␣, which results in the activation of the gp130associated Janus kinases (JAKs) (3)(4)(5). JAKs subsequently phosphorylate six tyrosine residues located in the cytoplasmic region of gp130 that serve as docking sites for signaling molecules containing the Src homology 2 (SH2) domains. The second membrane-proximal tyrosine residue of gp130 (Tyr-756) is required for recruitment of the SH2-containing phosphatase-2 leading to the activation of the mitogen-activated protein kinase (MAPK) pathway, whereas any of the four most distal tyrosine residues residing within a YXXQ motif can be used for the Stat3 binding (6 -8). Stat3 was cloned as an acute-phase response factor activated by IL-6 in mouse liver and has been identified to be a major molecule for IL-6 signaling (9). Upon IL-6 stimulation, Stat3 is transiently associated with gp130 via its SH2 domain and is subsequently phosphorylated by JAKs on the tyrosine residue 705 at its C terminus. Phosphorylated Stat3 forms dimers via reciprocal interactions between the SH2 domain and the phosphorylated Tyr-705, enabling it to translocate to the nucleus where it binds to IL-6 response elements in the regulatory region of many acute-phase protein genes, thereby regulating their expression (6,8,10,11). In addition to the Tyr phosphorylation, a serine phosphorylation on Ser-727 has been shown to enhance Stat3 transcriptional activity (12). Upon exposure to different stimuli, various serine/threonine kinases have been shown to mediate Ser-727 phosphorylation of Stat3. The MAPK/extracellular signal-regulated kinase (Erk), c-Jun N-terminal kinase, and its upstream kinase MAPK/Erk kinase kinase-1 (MEKK-1) have been demonstrated to be responsible for the Ser phosphorylation of Stat3 upon epidermal growth factor (EGF) and/or stress such as UV radiation (13)(14)(15). In the case of IL-6 induction, we have shown that protein kinase C␦ (PKC␦) phosphorylates Ser-727 of Stat3 (16).
PKC␦ is a member of a heterogeneous multifamily of lipidregulated serine/threonine kinases. The PKC isoenzymes are classified into four groups according to their activator/cofactor requirements. The conventional PKCs (cPKCs: ␣, ␤, and ␥) require both Ca 2ϩ and diacylglycerol or phorbol esters such as phorbol 12-myristate 13-acetate (PMA) as cofactors. Novel PKCs (␦, ⑀, , and ) can be stimulated by diacylglycerol or PMA alone, independently of Ca 2ϩ , and atypical PKCs (, , and ) and PKC-related kinases (PKC/PKD) are independent of both stimuli, and the latter is even structurally different, containing a pleckstrin homology domain (reviewed in Ref. 17). Activated PKC␦ was found to be associated with the p60 tumor necrosis factor receptor (18), the insulin receptor (19), and the insulin-like growth factor I receptor (20), playing important roles in downstream signaling, for instance by down-regu-lation of surface receptor expression or by mediating receptor redistribution (18,19).
Previously, we have shown that PKC␦ not only phosphorylates Stat3 on Ser-727 but also specifically associates with Stat3 in an IL-6-inducible manner in HepG2 cells (16). To understand further the regulatory role of PKC␦ on Stat3, we characterized the interaction of these two proteins. We demonstrate that the catalytic domain of PKC␦ interacts with the C-terminal portion of Stat3 including the SH2 domain and the adjacent 27 amino acids of the transactivation domain. The interaction of PKC␦ and Stat3 not only strongly enhances Stat3 binding to the IL-6 receptor gp130 but also leads to an association of PKC␦ and gp130. Stat3-gp130 interaction is abrogated by a dominant negative mutant of PKC␦ or by depletion of PKC␦. Furthermore, PKC␦ phosphorylates gp130 in vitro. These data suggest that PKC␦ plays important regulatory roles in IL-6 signaling.

EXPERIMENTAL PROCEDURES
Construction of Plasmids-The expression plasmids of murine PKC␦ and its mutant K376R (PKC␦ Ϫ ) in the pCR3.1 vector (Invitrogen) were obtained from W. Li (see Ref. 21). They were digested with BamHI and XhoI and reinserted into the pXJ40-GST vector, a mammalian expression plasmid provided by E. Manser (see Ref. 22). Point mutations of GST-PKC␦ (termed Y52F, Y155F, and Y187F) were constructed using the QuikChange site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions and were checked by standard dideoxy termination sequencing. Most Stat3 constructs have been described previously (23,24). The remaining constructs were generated by PCR using primers containing a 5Ј BamHI site and a 3Ј XhoI site, and murine Stat3␣ as a template. The PCR products were subsequently cloned into the pXJ40-FLAG vector (22) or the pXJ40-HA (hemagglutinin) epitope-tagged vector (25) and checked by sequencing. The Epo-gp130 chimera (Eg) was a kind gift of F. Horn (see Ref. 26). Point mutations at Thr-890 and Thr-909 were introduced with the QuikChange kit as mentioned above. To construct gp130-ID (extending from aa 605 to 918 of the human gp130 receptor subunit, which contains 15 amino acids of the extracellular domain, the transmembrane region, and the complete cytoplasmic domain) for bacterial fusion-protein expression, a PCR was performed using the Epo-gp130-Flag plasmid as template with primers containing a 5Ј BamHI site and a 3Ј XhoI site. The PCR product was inserted into the pET32a vector (Novagen) via the BamHI and XhoI restriction sites to give rise to a fusion protein containing an N-terminal thioredoxin tag, S tag, and His tag plus an additional C-terminal His tag. The integrity of the construct was subsequently checked by sequencing.
Cell Culture and DNA Transfections-The human hepatoma cell line HepG2 and the African green monkey kidney fibroblast COS-1 cells were maintained in Dulbecco's modified Eagle's medium with 1000 mg/liter glucose containing 10% FBS, 100 units/ml penicillin, and 100 ng/ml streptomycin, 2 mM L-glutamine (all from Invitrogen), and 10 mM HEPES buffer, pH 7.3. COS-1 and HepG2 cells were transfected using LipofectAMINE (Invitrogen) and FuGENE 6 (Roche Diagnostics), respectively, following the manufacturer's instructions.
Antibodies, Immunoprecipitation, GST Pull-down, and Immunoblotting-Antibodies against Stat3 were purchased from either Transduction Laboratories (BD Biosciences) or Santa Cruz Biotechnology. Antisera against phospho-Tyr-705 and phospho-Ser-727 of Stat3 were from Cell Signaling and QCB (Biosource), respectively. The antibodies against the glutathione S-transferase (GST) were purchased from BD Biosciences (mouse monoclonal) or Santa Cruz Biotechnology (rabbit polyclonal, Z5). Monoclonal antisera against the FLAG tag (FLAG-M2) and PKC␦ were from Sigma and Transduction Laboratories, respectively. Antibodies against the gp130 subunit of the IL-6 receptor as well as the HA tag were from Santa Cruz Biotechnology. If not stated otherwise, cells were lysed 48 h after transfection, and cell lysates containing 500 g to 2.5 mg of proteins were subjected to the immunoprecipitation and immunoblotting experiments performed as described previously (27). In the case of GST pull-down, the total cell lysate containing 500 g of protein per sample was incubated for 2 h with glutathione-Sepharose 4B beads (Amersham Biosciences) before washing. SDS-PAGE, transfer, and immunoblotting were carried out as described previously (24).
Peptide Binding Assay-The biotinylated decapeptides of the IL-6 receptor subunit gp130 (pY 3 and Y 3 ) were purchased from piCHEM R&D. The peptide binding assay was performed as described previously (23).
Production and Purification of Bacterial Fusion Protein-The constructs (pET32a vector and gp130-ID) were transformed into the BL21 Escherichia coli strain. At A 600 ϭ 0.5 cells were induced with 1 mM isopropyl-␤-D-thiogalactoside for 2.5 h at 37°C. Subsequently, the bacterial pellets were resuspended in cold PBS containing 0.2 mg/ml lysozyme and incubated for 1 h at 4°C and 15 min at 30°C. After sonication for a total of 2 min (6 times for a 20-s sonication with 15-s intervals in between), cells were pelleted and resuspended in 3 ml of PBS containing 1 tablet of protease inhibitor mixture per 10 ml. To purify the protein, the supernatant was incubated with 300 l of 50% nickel-nitrilotriacetic acid-agarose bead slurry (Qiagen) at 4°C overnight. The beads were subsequently washed 4 times with cold PBS and analyzed by SDS-PAGE.
PKC␦ Kinase Assay-The kinase assay was performed as described previously (16,28). In brief, about 1 g of purified bacterial recombinant protein per reaction (corresponding to 30 l of bead slurry) was washed 3 times with cold PBS and 3 times with PKC␦ kinase buffer (25 mM Tris-HCl, pH 7.4, 5 mM MgCl 2 , 0.5 mM EGTA, 1 mM dithiothreitol). The kinase reaction was performed in a 20-l total volume of PKC␦ kinase buffer containing 20 g of phosphatidylserine, 5 Ci of [␥-32 P]ATP (Amersham Biosciences), and 1 g of recombinant PKC␦ (Panvera) for 10 min at 30°C. The reaction was stopped by adding 10 l of 3ϫ reducing Laemmli sample buffer, subjected to SDS-PAGE, fixed in 10% acetic acid and 40% methanol for 20 min, stained with Coomassie Blue, dried, and exposed to Hyperfilm for 15 min.

PKC␦ Interacts with the SH2 Domain and the Adjacent 27
Amino Acids in the C-terminal Domain of Stat3-Stat proteins consist of several functional domains including an N-terminal domain, a coiled-coil domain, a DNA binding domain, and a linker domain, followed by an SH2 domain and a C-terminal transactivation domain (29,30). In order to determine the exact region of Stat3 binding to PKC␦, we generated a series of FLAG-tagged mutants, which either sequentially delete the functional domains from the N terminus or are truncation mutants containing one or more functional domains, as illustrated in Fig. 1A (a-n). COS-1 cells, which express low levels of endogenous Stat3 and PKC␦, were cotransfected with plasmids expressing Stat3 mutants and/or GST-PKC␦ fusion protein and subjected to GST pull-down experiments. As shown in Fig. 1B, all N-terminal deletion constructs of Stat3 harboring a complete SH2 domain were binding well to GST-PKC␦, suggesting that the domains N-terminal to the SH2 domain are dispensable for this interaction (upper left panel, constructs a-f). Indeed, the mutant constructs containing either the N-terminal portion including the N-terminal domain and the coiled-coil domain (construct g) or the middle region consisting of the DNA binding and the linker domains (construct h) of Stat3 did not bind to PKC␦, in contrast to the construct covering the DNA binding to the SH2 domain (construct i). Furthermore, the C-terminal domain (construct j) was also unable to interact with PKC␦ (upper middle panel). To further map the binding site, we found that the SH2 domain alone seems to be required but not sufficient for PKC␦ binding by using constructs containing the SH2 domain and the C-terminally truncated mutants (constructs k-n). With these mutants, the minimal binding region was determined to reside within aa 600 -715 (construct m) including the major part of SH2 domain (aa 585-688) and its adjacent C-terminal 27 aa (Fig 1B, upper right panel). The lower panels of Fig. 1B show the amounts of GST-PKC␦ that were pulled down by GST beads. The expression of the various Stat3 constructs and GST-PKC␦ in the different sets of experiments was found to be comparable, except for construct m, which showed a lower expression level (Fig. 1C, upper and lower panels, respectively). The various Stat3 constructs did not non-specifically interact with the GST protein alone (data not shown).
A Functional SH2 Domain Is Not Required for the PKC␦-Stat3 Interaction-SH2 domains usually serve as a proteinprotein interaction domain by interacting with phosphoty-rosine residues (31). Because PKC␦ is known to be phosphorylated on tyrosine residues in its regulatory domain upon various stimuli including IL-6 (16,17,32), and because of the importance of the Stat3 SH2 domain in the binding of PKC␦, we next examined whether the association depended on an interaction between the SH2 domain of Stat3 and phosphotyrosine residue(s) of PKC␦. To this end, a point mutation was first introduced in FLAG-tagged Stat3 by substitution of Arg-609 with Leu (R609L), which destroys the binding pocket of the SH2 domain and was shown to be unable to bind to the receptor gp130 peptide (23). The wild-type (wt) or the mutant Stat3 were cotransfected with GST-PKC␦, and GST pull-down experiments were performed. In the case of overexpressed PKC␦, we found that there is already a high level of tyrosine phosphoryl- GST pull-down assays were performed, and the expression of the proteins was analyzed with Western blotting of total cell lysate using the indicated antibodies as described in Fig. 1B. C, COS-1 cells were cotransfected with SH2-720 and different GST constructs of PKC␦ (wt or point mutants Y52F, Y155F, or Y187F). GST pull-down assays and Western blots of total cell lysate were performed in the same way as described in Fig. 1B. Numbers given indicate the corresponding amino acids of the start/end of the constructs. The full-length (FL) Stat3 starts at aa 1 and ends at aa 770, and constructs k-n start at aa 600 and extend to the aa given in their names. B, COS-1 cells were cotransfected with the Stat3 constructs (a-n) or FLAG-vector (Vec) and GST-PKC␦. Cells were lysed after 48 h, and 500 g of total cell lysate (TCL) was used in a GST pull-down assay and analyzed on 10 or 15% SDS-PAGE gel (upper panels) by Western blot using Flag-M2 antibody as described under "Experimental Procedures." Molecular mass markers are indicated in kDa on the left. Blots were stripped and reprobed with a GST-antibody (lower panels). C, 50 g of total cell lysate was analyzed by Western blot using the antibodies indicated on the left. ation without further induction (data not shown), so we used cell lysates from non-induced cells. We observed that in comparison with the wt Stat3, the amount of PKC␦ bound to R609L was reduced but not eliminated ( Fig. 2A, upper panel). Comparable amounts of GST-PKC␦ protein have been pulled down, and the expressions of wild-type or mutant Stat3, as well as GST-PKC␦, in each sample were shown to be equal ( Fig. 2A, lower panels). To confirm further the role of SH2 domain, we also introduced a point mutation in SH2-720 construct (SH2-720R609L) that contains the minimum interaction region of Stat3. This mutant displayed an interaction with PKC␦ in a similar affinity compared with its wt counterpart (SH2-720) in the GST pull-down experiment (Fig. 2B, top panel). These results suggest that a functional SH2 domain may not be required for Stat3 interaction with PKC␦.
On the other hand, the tyrosine phosphorylations of PKC␦ upon stimulation with growth factors/cytokines are known to occur on tyrosine residues 52, 155, and 187 within its regulatory domain (33,34). In the following experiments, we created three point mutations (Y52F, Y155F, and Y187F) in GST-PKC␦ where each of the tyrosine residues was replaced by phenylalanine. Their interaction with the Stat3-SH2-720 was tested by GST pull-down assays. The results showed comparable amounts of SH2-720 interacting with either the wild-type or the mutants (Fig. 2C, top panel), indicating that the mutations of PKC␦ did not affect the interaction with Stat3. The expression levels were also shown to be equal (Fig. 2C, 3rd and bottom panels). These data suggested that the binding of Stat3 to PKC␦ may not involve these tyrosine residues. Alternatively, although the interaction between Stat3 and PKC␦ involves the SH2 domain of Stat3, the binding may not occur via a classical SH2-phosphotyrosine interaction.
PKC␦ Binds to Stat3 via Its Catalytic Region-To address the above questions further, we next tried to delineate the interaction region on PKC␦. PKC␦ consists of two structurally well defined domains as follows: the N-terminal regulatory domain, which consists of a C2-like domain (phospholipid binding domain), a pseudo-substrate domain, and two cysteine-rich domains; and a C-terminal catalytic domain. The latter contains the ATP-binding domain (C3) and the substrate-binding active kinase site (C4), which maintains the enzyme in its inactive form by binding to the pseudo-substrate domain in the absence of substrate. These well conserved regions are interspaced and followed by less conserved (therefore termed variable) regions V1, V2, V3, V4, and V5 (17) as illustrated in Fig. 3A. To determine the domain of PKC␦ that interacts with Stat3, we initially constructed three GST-tagged deletion mutants of PKC␦ (named F1, F2, and F3) and cotransfected them with the FLAG-tagged wt Stat3 into COS-1 cells. We found that both the N-terminal section (F1), comprising most of the regulatory domain of PKC␦, and the middle portion (F2), containing the second Cys-rich and the ATP-binding domain, cannot bind to Stat3. However, the C-terminal region of PKC␦, spanning from aa 381 to 735 of the catalytic domain, exhibited very strong binding (Fig. 3B, top left panel). The amounts of the constructs pulled down (Fig. 3B, 2nd panel), as well as the expression of the PKC␦ fragments or Stat3 (3rd and bottom panels), were comparable. To further delineate the binding site to either the active kinase site (C4 domain) or the C-terminally adjacent variable domain (V5), we constructed two more GSTtagged fragments of PKC␦, called C4 and V5 (Fig. 3A). Both regions, although not overlapping, seemed to bind Stat3 to a similar extent (Fig. 3B, top right panel). Together these results indicate the interaction occurring between the SH2 domain of Stat3 and the C-terminal catalytic domain of PKC␦, which also supports our previous findings that Stat3 is a substrate of PKC␦ by showing its binding to the catalytic domain.
PKC␦ Enhances Stat3 Binding to the gp130 Receptor-Stat3 is recruited to the gp130 upon IL-6 stimulation via its SH2 domain. Because PKC␦ also interacts with the SH2 domain of Stat3, we wondered whether PKC␦ could affect Stat3 binding to the receptor. To address this issue, we transfected Stat3 and PKC␦ into HepG2 cells, and examined their interaction with the endogenous gp130 with or without IL-6 treatment by immunoprecipitation/blotting experiments. A weak association of Stat3 with the endogenous gp130 was detected in the absence of IL-6 induction, which was up-regulated upon IL-6 treatment. Interestingly, the presence of PKC␦ strongly enhanced the binding of Stat3 to the gp130 after IL-6 stimulation (Fig. 4A,  top panel, compare lane 2 and lane 6). On the other hand, PKC␦ alone did not interact with gp130 but associated with gp130 in the presence of Stat3 upon IL-6 stimulation (2nd panel, lanes 4  and 6). The quantity of immunoprecipitated gp130 and the expression of Stat3 and PKC␦ in each sample are shown in the 3rd and 4th and bottom panels.
To confirm these data, we also used a heterologous system by transfecting COS-1 cells with a chimeric receptor, consisting of the extracellular domain of the murine erythropoietin (Epo) receptor and the transmembrane and intracellular domains of the human gp130 receptor tagged with the FLAG epitope at its C terminus (Epo-gp130-Flag), together with a HA-tagged Stat3 and the GST-tagged PKC␦. The obtained results confirmed our findings in HepG2 cells, showing a strong increase in the amount of Stat3 binding to the gp130 receptor in the presence of PKC␦ after stimulation with Epo (Fig. 4B, top panel, compare lane 2 and lane 6), whereas PKC␦ itself was unable to interact with the receptor and could only do so in the presence of Stat3 in response to Epo stimulation (2nd panel). Next, we investigated whether phosphorylation of Stat3 on Ser-727 would affect the role of PKC␦ on the interaction. We observed that PKC␦ was able to enhance the interaction between the Ser-727 mutant (S727A) and gp130 (Fig. 4C, top panel, lanes 1  and 2). Furthermore, PKC␦ was also able to associate with gp130 in the presence of S727A (2nd panel), suggesting that phosphorylation on this site may not be critical for the PKC␦ effect. In conclusion, these findings indicate that PKC␦ enhances the ability of Stat3 to bind to gp130, which is independent of Ser-727 phosphorylation.
PKC␦ Binds to gp130 Receptor via Stat3-The results described above suggest that the association of PKC␦ and gp130 in response to IL-6 occurs via Stat3. To explore this possibility further, we utilized the R609L mutant that cannot bind to the gp130 receptor peptide (23) but binds to PKC␦ (Fig. 2, A and B). FLAG-tagged mutant R609L or the wt Stat3 were transfected into HepG2 cells in the absence or the presence of GST-PKC␦, and their interactions with gp130-derived peptides were tested. Stat3 can bind to any of the four most C-terminal phosphotyrosine residues (Y 3 -Y 6 ) within the cytoplasmic tail of the gp130 (35). The peptide binding assays were performed using phos- Cells were either treated with murine erythropoietin (Epo, 7 units/ ml) for 15 min or left unstimulated. Immunoprecipitation and Western blot analysis were performed as described in A using the antibodies indicated on the left. C, the same experiment as described for B was performed except using HA-tagged S727A, in which Ser-727 was substituted by Ala, to replace the wt HA-Stat3. phorylated (pY 3 ) and non-phosphorylated (Y 3 ) decapeptides surrounding the third membrane proximal tyrosine residue on gp130 receptor. Stat3 bound to pY 3 strongly but not to Y 3 in the absence or the presence of PKC␦, whereas Stat3 mutant R609L as expected did not bind to pY 3 in any case (Fig. 5, 2nd panel). On the other hand, PKC␦ only associated with pY 3 in the presence of the wt Stat3 (top panel, lane 10) but not of R609L mutant (lane 12). The expression of the wt Stat3 and the R609L mutant was similar (bottom panel). These data support the hypothesis that PKC␦ is recruited to the receptor gp130 by association with Stat3.
Dominant Negative PKC␦ Blocks Stat3 Binding to Receptor gp130 -In the following experiments, we wanted to determine whether a dominant negative PKC␦ affects the binding ability of Stat3 to gp130. Mutation of lysine 376 to arginine (K376R) in PKC␦ destroys its ability to bind ATP but not to its substrate, thus acting as a dominant negative form of PKC␦ (21). Stat3 was cotransfected with either wt PKC␦ or PKC␦ Ϫ into COS-1 (Fig. 6A) or HepG2 cells (Fig. 6B), and the effect of this mutant (PKC␦ Ϫ ) was tested in the peptide binding assays. The results in both cell lines showed that Stat3 alone bound to pY 3 , but not Y 3 , in the absence of wt PKC␦ or PKC␦ Ϫ . Whereas the wt PKC␦ increased Stat3 binding to pY 3 , the dominant negative mutant PKC␦ Ϫ abolished Stat3 binding almost completely (Fig. 6, A  and B, upper panels), although the expression level of the mutant was very low (lower panels). We also evaluated the effect of PKC␦ Ϫ in vivo with an immunoprecipitation/Blot experiment using the same lysates as described for Fig. 6B. We showed that PKC␦ Ϫ abrogated the interaction of Stat3 and the endogenous gp130 (Fig. 6C, upper panel, compare lane 2 and lane 3), although itself seems to be able to associate with gp130 weakly (middle panel, lane 3). Together, these data demonstrate that the dominant negative PKC␦ impairs the recruitment of Stat3 to the receptor gp130 in IL-6 signaling.
Depletion of Cellular PKC␦ Diminishes Stat3 Binding to the Receptor gp130 -It has been described that classical and novel PKCs can be depleted from cells by a prolonged treatment with 100 -200 nM PMA (36 -38). To examine whether the regulation of Stat3 binding to receptor by PKC␦ occurs under physiological conditions, we tested the binding of the cellular Stat3 to gp130 when the endogenous PKC␦ is depleted. We observed that IL-6-induced Stat3 interaction with gp130 was diminished with PMA treatment of the cells for 24 h and totally abolished at 48 h (Fig. 7A, upper panel). The efficiency of the PMA depletion was shown to be sufficient as examined by Western blotting of the total cell lysate (Fig. 7B, middle panel). To further verify the effect was because of PKC␦, GST-PKC␦ was introduced into PMA-depleted cells. We observed that the reintroduced GST-PKC␦ restored the interaction of the endogenous Stat3 and the gp130 (Fig. 7C, 3rd lane, top panel). These results further confirm a physiological role of PKC␦ in the binding of Stat3 to the receptor gp130.
PKC␦ Phosphorylates the Intracellular Domain of gp130 in Vitro-Ser/Thr phosphorylation has been shown to play an important role in regulating several aspects of growth factor or cytokine receptor signaling, such as down-regulation of endocytosis, and cell surface expression (18,39,40). We were interested to know whether PKC␦ also phosphorylates the cytoplasmic domain of the gp130 receptor. Potential PKC phosphorylation sites are located around Thr-909 according to the TXRX motif described by Pearson and Kemp (41) and Falquet et al. (42) and Thr-890 in the cytoplasmic tail of fulllength human gp130, identified in analogy to the PKC phosphorylation site (RXXT) in the insulin receptor (43). We used recombinant bacterially expressed cytoplasmic domain of gp130 (named as gp130-ID) in an in vitro kinase assay with purified recombinant PKC␦. The result showed that gp130-ID was phosphorylated by PKC␦, suggesting that gp130 is a potential substrate of PKC␦ (Fig. 8A).
Identification of Thr-890 of gp130 as an Essential Site for the PKC␦ Effect-Finally, we attempted to determine whether the putative PKC phosphorylation sites on gp130 would affect the role of PKC␦ in the Stat3-gp130 interaction. Thr-890 and Thr-909 of gp130 were replaced individually by alanine, and their interaction with Stat3 in the presence of PKC␦ was tested. The results indicated that Stat3 interaction with wt gp130 or the mutant T909A was enhanced by PKC␦ (Fig. 8B, top panel,  lanes 5 and 7), and PKC␦ was also recruited to the receptor via Stat3 (lanes 5 and 7, 2nd panel). However, we could not detect an obvious change in the interaction between Stat3 and the mutant gp130 T890A without PKC␦ (not shown), and PKC␦ could not enhance or stabilize this interaction (lane 6, top and 2nd panel). The results imply that one of the putative PKC phosphorylation sites is crucial for the stabilization of Stat3-gp130 interaction by PKC␦. Collectively, these data suggest that PKC␦ is recruited to the receptor gp130 via Stat3, which not only stabilizes the Stat3 and gp130 receptor interaction, but may also phosphorylate gp130. Thr-890 seems to be a crucial site for the effect of PKC␦.

DISCUSSION
Stat3 plays important roles in regulating a multitude of different processes initiated by a variety of extracellular ligands including those of the IL-6 cytokine family and growth factors such as EGF. Its transcriptional activity is regulated by the tyrosine phosphorylation on Tyr-705, as well as an additional phosphorylation on Ser-727. We have previously shown PKC␦ as a Ser/Thr kinase responsible for the Ser-727 phosphorylation of Stat3 upon stimulation with IL-6 (16). Correspondingly, another study found that PKC␦ is activated downstream of Rac1 and SEK1/MKK4 to regulate Stat3 phosphorylation on Ser-727 (44). In addition, Uddin et al. (45) recently described PKC␦ as the serine kinase for Stat1 Ser-727 phosphorylation, and also as an interacting protein of Stat1 in an interferon ␣-inducible manner. These data suggest that PKC␦ might play a more widespread role as a Ser/Thr kinase for Stat proteins.
In addition to the phosphorylation, we previously observed an interaction between the endogenous Stat3 and PKC␦ induced by IL-6, OSM, and LIF, implying that the interaction has physiological effects on the IL-6 family. We therefore further characterized this interaction. Because the interaction is inducible by IL-6, and the time course of the interaction coincides with the kinetics of Stat3 Tyr-705 phosphorylation, we initially postulated that the interaction might depend on the Stat3 Tyr-705 phosphorylation. However, mutation on Tyr-705, as well as on Ser-727, had little effect on PKC␦ binding (data not shown). We then delineated the interacting regions in both proteins. We identified that the catalytic domain of PKC␦ binds to the SH2 domain of Stat3 (Figs. 1-3). The phosphotyrosine residues of PKC␦ (Tyr-55, Tyr-155, and Tyr-187), which are reported to be phosphorylated by cytokine or growth factor stimulation, are located outside the catalytic domain and are unlikely to be involved in the interaction with the Stat3 SH2 domain. On the other hand, there are also tyrosine residues that can be phosphorylated in the catalytic domain of PKC␦, but so far only oxidative stress (H 2 O 2 ) has been shown to induce their phosphorylations (46). Furthermore, mutation on the key residue Arg-609 of the SH2 pocket has little or no effect on PKC␦ association (Fig. 2). Hence, these data suggest that the interaction of Stat3 and PKC␦ may not be mediated by a typical interaction between SH2 domain and phosphotyrosine. In agreement with these data, it also has been demonstrated on several occasions that SH2 domains specifically bind to proteins in a phosphotyrosine-independent manner. For example, the binding region of the SH2 domain-interacting proteins in some cases contains phosphoserine or phosphothreonine instead (47)(48)(49). PKC␦ has been found to have three Ser/Thr phosphorylation sites in its C terminus including Thr-505 (possibly being phosphorylated by an upstream kinase), Ser-643, and Ser-662, which are autophosphorylated to contribute to the full activity of the enzyme (50,51). Because we detected constitutive phosphorylation on Ser-643 and Thr-505 after overexpressing PKC␦ (data not shown), these phosphorylations might contribute to the binding of PKC␦ V5 domain to the SH2 domain of Stat3.
Stat proteins are recruited to the cytokine receptors via their SH2 domain (6). In agreement with our data described above, we found that instead of competing with Stat3 binding to the IL-6 receptor subunit, gp130, PKC␦ significantly enhanced Stat3-gp130 interaction in heterogeneous systems (Fig. 4). Although the mechanism of this effect is not clear, a few possi-bilities can be considered. For instance, binding of PKC␦ might induce a conformational change in Stat3, which enhances the affinity of Stat3 binding to gp130. We recently reported that the binding of Stat3 to the receptor gp130 is regulated by an intramolecular interaction in which the coiled-coil domain of Stat3 interacts with its flexible C-terminal domain and thus retains Stat3 in a conformation with an accessible SH2 domain for receptor binding (24). Furthermore, the SH2 domain of Stat3 is known to be among the most divergent SH2 domains in amino acid sequence (52). In contrast to the SH2 domains of other SH2-containing proteins such as Src family kinases, which can bind to their phosphotyrosyl peptides on their own, the SH2 domain alone of Stat3 is not sufficient for receptor binding (23). Therefore, it is also possible that PKC␦ stabilizes the intramolecular interaction of Stat3 to promote its receptor binding. Alternatively, since the interaction between endogenous Stat3 and PKC␦ is inducible by IL-6, PKC␦ may stabilize Stat3-gp130 interaction at the receptor level.
Our data also demonstrate that PKC␦ is recruited to the gp130 receptor via Stat3. What are the possible physiological consequences for this association? First, PKC␦ binding to Stat3 stabilizes the interaction between Stat3 and gp130, which may facilitate Stat3 Tyr phosphorylation to occur by JAKs preassociated with the box 1 region of gp130. Second, PKC␦ binding to gp130 may be required for the Ser phosphorylation of Stat3. We have mapped the binding site of PKC␦ on Stat3 within aa 600 -720, which is located immediately upstream of Ser-727, the phosphorylation site for PKC␦. Furthermore, Stat3 interacts with the substrate binding site of PKC␦, suggesting that Stat3 is a bona fide substrate for PKC␦. Abe and colleagues (53) recently have also mapped the C-terminal portion (aa 533-711) of Stat3 as the required binding region for an H7-dependent kinase which overlaps the binding site (aa 600 -715) that we have mapped in this study, although they could FIG. 8. PKC␦ phosphorylates gp130 in vitro. A, 1 g of the His-tagged recombinant protein (gp130-ID) or the tag alone (Vec) was used in in vitro kinase assay at 30°C for 10 min, together with 1 g of recombinant PKC␦ and 5 Ci of [␥Ϫ 32 P]ATP. Proteins were separated on a 7.5% SDS-PAGE gel, fixed, dried, and exposed to film to detect the phosphorylated proteins. The molecular mass markers in kDa are indicated on the left, and the position of gp130-ID is marked by an arrow on the right. The asterisk indicates the autophosphorylated PKC␦ (left panel). The right panel shows about 20 g of the bacterially expressed fusion protein, separated on a 7.5% SDS-PAGE and stained by Coomassie Blue. B, COS-1 cells were cotransfected with Epo-gp130-Flag or its mutants (as labeled on top) together with the GST-PKC␦ and HA-Stat3 or the respective vectors (Vec). The immunoprecipitation (IP)/Blot and expression analysis were performed as described in Fig. 4B. not show the interaction of Stat3 and PKC␦. Interestingly, they indicate that the YXXQ motif in gp130 is not only important for Stat3 binding and its subsequent Tyr phosphorylation but also crucial for the Ser-727 phosphorylation of Stat3 by an unknown mechanism. Our findings reveal that PKC␦ associates with gp130 at the YXXQ motif via Stat3, suggesting that this association is necessary for Stat3 Ser phosphorylation by PKC␦ in response to IL-6 stimulation, which may provide an explanation for this requirement. Third, PKC␦ may phosphorylate and therefore regulate gp130 receptor activity and/or the signaling molecules that are associated with gp130. It has been demonstrated in many instances that transmembrane receptors are regulated in various aspects, such as endocytosis, down-regulation, stabilization, tyrosine kinase activity, or degradation, by means of a Ser/Thr phosphorylation within their cytoplasmic domain (54,55). Besides Erk kinase which enhances LIFR degradation (39), and calmodulin-dependent kinase II which regulates the receptor kinase activity of EGFR (56), PKCs, and more specifically PKC␦, have been shown to be the kinase in several cases (18,36). In our experiments, we detected the phosphorylation of gp130 by PKC␦ in vitro (Fig. 8A), but the receptor turnover does not seem to be affected nor can we detect significant changes in the activity of JAK1 (data not shown). Interestingly, we have identified a putative PKC phosphorylation site (Thr-890) on gp130 that is necessary for the PKC␦ effect on the Stat3-gp130 interaction. However, whether Thr-890 is phosphorylated by PKC␦ in vivo, and the mechanism for this requirement remain to be determined.
In summary, our data suggest that there is an effect of PKC␦ not only as a Ser/Thr kinase phosphorylating Stat3 but also enhancing the binding of Stat3 to the receptor. This binding of Stat3 to the receptor subunit at the same time facilitates PKC␦ function by bringing the kinase into close proximity with the gp130 receptor subunit that further regulates gp130-mediated signaling. It is noteworthy that the complex formation of the endogenous Stat3 and PKC␦ is transient in response to IL-6 stimulation (16). It is possible that PKC␦ initially stabilizes Stat3-gp130 interaction to facilitate Tyr and Ser phosphorylation of Stat3, as well as the Thr phosphorylation of gp130. These phosphorylation events may lead to the conformational changes of these proteins and promote the dissociation of this complex to enable the dimer formation and subsequent nuclear translocation of Stat3.