A casein kinase I activity is constitutively associated with Nck.

Nck is a 47-kDa cytosolic protein devoid of intrinsic catalytic activity and consisting of Src homology 2 and 3 (SH2 and SH3) domains organized as follows: SH3-SH3-SH3-SH2. Nck is believed to act as an adaptor protein mediating signal transduction initiated by receptor tyrosine kinases (RTKs). Through its SH2 domain, Nck recognizes a specific phosphotyrosine residue on RTKs or on protein substrates of RTKs like insulin receptor substrate-1, the major substrate of the insulin receptor, and through its SH3 domains it interacts with poorly characterized effector molecules. To identify novel proteins that might interact with Nck, we have used the amino-terminal segment of Nck encompassing its three SH3 domains in the yeast two-hybrid system. Among the polypeptides that associate with Nck, we have identified the γ2 isoform of the serine/threonine casein kinase I (CKI-γ2). In transformed rat hepatocytes overexpressing the insulin receptor (HTC-IR cells), serine/threonine protein kinase activity coimmunoprecipitates with Nck, an interaction mediated mainly by the third SH3 domain of Nck. This kinase activity is not apparently modulated by insulin, nor is it sensitive to staurosporine or heparin, and it does not use GTP as a phosphate donor. However the kinase activity coimmunoprecipitated with Nck is completely abolished by N-(2-aminoethyl)-5-chloroisoquinoline-8-sulfonamide, a specific inhibitor of casein kinase I. In an in vitro renaturation gel kinase assay, a protein kinase of 70-75 kDa was detected associated with the SH3 domains of Nck. Far Western analysis demonstrated that the SH3 domains of Nck bound directly to a cytosolic protein of 70-75 kDa. A rabbit polyclonal antibody raised against the C-terminal region of CKI-γ2 protein kinase immunoprecipitated a single specific protein of 70-75 kDa from HTC-IR cell lysates and detected CKI-γ2 among the proteins coimmunoprecipitated with Nck. These results support an in vivo interaction between Nck and CKI-γ2 and suggest that CKI-γ2 could be involved in signaling pathways downstream of RTKs.

Common events triggered by membrane receptor tyrosine kinases (RTKs) 1 involve activation of the intrinsic receptor tyrosine kinase activity, autophosphorylation, and tyrosine phosphorylation of various intracellular proteins (1). Phosphorylation of specific tyrosine residues creates high affinity binding sites for a variety of cytoplasmic SH2 domain-containing proteins (2), which are recruited to propagate signals to downstream effector molecules (3). SH2 domains mediate proteinprotein interactions by direct recognition of phosphotyrosinecontaining motifs (4,5), and their role in the assembly of protein complexes has been firmly established (6). Many of the SH2 domain-containing proteins possess one or several SH3 domains, which mediate protein interactions through the recognition of proline-rich sequences in their target molecules (7,8). The importance of SH3 domain-mediated interactions was first revealed with Grb2, an adaptor molecule composed almost exclusively of SH2 and SH3 domains. Grb2 has been implicated in a highly conserved mechanism for the control of the Ras-dependent activation of mitogen-activated protein kinase by RTKs. Mutations in the SH3 domains of Sem5, the Caenorhabditis elegans homologue of Grb2 (9), have been reported to impair its ability to transmit a biological signal (10). Interestingly, the finding that an SH3 domain and a proline-rich sequence also mediate a direct interaction between two components of the phagocyte NADPH oxidase complex (11), suggests that the role of SH3-mediated protein interactions could be expanded to regulatory molecules involved in cell responses unrelated to cell division. These observations stress the importance of identifying proteins that bind SH3 domains.
Nck is a widely expressed 47-kDa cytoplasmic protein (12) consisting of SH2 and SH3 domains organized in the order SH3-SH3-SH3-SH2 (13). Nck has been shown to bind via its SH2 domain to the platelet-derived growth factor and epidermal growth factor receptors (12,14,15) and intracellular tyrosine-phosphorylated proteins, such as IRS-1 (16), and c-Src (15). Activation of growth factor RTKs induces Nck phosphorylation on tyrosine and serine residues (12,14,15,17). In fibroblasts, overexpression of Nck leads to cell transformation (12,15). Despite these suggestions of its involvement in mitogenesis and oncogenesis, studies examining the role of Nck in mediating these processes are sparse. By analogy to the crucial role played by Sos, which associates with the Grb2 SH3 domains, in the activation of Ras-dependent signaling (18,19), effector proteins interacting with the SH3 domains of Nck may also trigger cellular signaling events leading to biological responses. In fact, Nck, through its SH3 domains, has been shown to interact with effectors involved together with small GTPases in the regulation of the cytoskeleton (20), which include Sos (21), an activator of Ras, and mPAK3 (22), a serine/ threonine kinase activated by Rac1 and Cdc42, members of the Rho-GTPase family. Interestingly, Dock, a Drosophila adaptor protein homologous to human Nck, has been reported to be required for normal photoreceptor R cell axon guidance and targeting (23), providing further support for Nck involvement in regulating cytoskeletal organization. In addition, c-Cbl (24), Nck-associated kinase (NAK; a serine/threonine kinase) (25), and the Wiskott-Aldrich syndrome protein (WASP) (26) also interact with the SH3 domains of Nck; however, the functional significance of these interactions is currently unknown.
To identify novel proteins that might interact with Nck, we have used the amino-terminal domain of Nck encompassing its three SH3 domains, in the yeast two-hybrid system (27). Among the polypeptides we found associating Nck, was a protein very similar to the ␥2 isoform of the rat serine/threonine casein kinase I (CKI-␥2) (28). In the present study, we have examined the relation between Nck and CKI-␥2 in an insulinresponsive cell line and found that in transformed rat hepatocytes overexpressing the insulin receptor, CKI-␥2 was constitutively associated with Nck.

EXPERIMENTAL PROCEDURES
Cell Cultures-Transformed rat hepatocytes overexpressing the human insulin receptor (HTC-IR cells) (30) were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) containing 10% fetal calf serum (Life Technologies) and maintained in selection medium containing 40 g/ml of Geneticin (G418) (Life Technologies). Subconfluent HTC-IR cells were serum-starved in Dulbecco's modified Eagle's medium containing 0.1% bovine serum albumin (Sigma), 24 -48 h prior to the experiment. In experiments involving insulin stimulation (porcine insulin, Connaught-Novo Laboratories, Willowdale, Canada), insulin was added to cells in culture (final concentration 100 nM) for the times indicated in the figure legends.
Antibodies-A polyclonal Nck antibody was raised in rabbits, by immunizing with a GST fusion protein containing the SH3 domains of human Nck (residues 1-251). A polyclonal CKI-␥2 antibody was produced by immunizing rabbits with a GST fusion protein containing the C-terminal portion of the CKI-␥2 (residues 192-414). This was obtained by subcloning the partial CKI-␥2 cDNA isolated from the two-hybrid screen into pGEX4T2 (Pharmacia Biotech Inc.).
Nck Immunoprecipitation and in Vitro Kinase Assays-HTC-IR cells were lysed in the following buffer: 50 mM HEPES, pH 7.5, 150 mM sodium chloride, 10% (v/v) Triton X-100, 1.5 mM magnesium chloride, 1 mM EGTA, 10 g/ml aprotinin, 10 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, 10 mM sodium fluoride, 10 mM sodium pyrophosphate, 200 M sodium orthovanadate, and 40 mM ␤-glycerophosphate. Clarified lysates were adjusted with lysis buffer to contain a final protein concentration of 1 mg/ml and were then submitted to Nck immunoprecipitation at 4°C, using our polyclonal rabbit antibody (see above) and protein A-Sepharose (Sigma). After 90 min of gentle agitation, the protein A-Sepharose beads were washed three times with lysis buffer and once with kinase reaction buffer (20 mM HEPES, pH 7.5, 1 mM dithiothreitol, 5 mM MgCl 2 , 10 mM ␤-glycerophosphate) and were resuspended in 25 l of kinase buffer. In experiments where kinase activity was assayed on exogenous substrates, 15 g of dephosphorylated ␣-casein (Sigma), 5 g of myelin basic protein (Sigma), 15 g of ␤-casein (Sigma), 5 g of histone HI (Sigma), 15 g of phosvitin (Sigma), or 15 g of PolyGluTyr (Sigma) was added to the assay. After preincubation at 30°C for 5 min, the phosphorylation assay was initiated by adding [␥-P 32 ]ATP (20 M, 10 Ci) (DuPont NEN), and after 20 min at 30°C, the reaction was stopped by the addition of 6 l of 6 ϫ Laemmli buffer (31) followed by boiling for 2 min before subjecting to SDS-PAGE. Phosphorylation of exogenous substrates or proteins coimmunoprecipitated with Nck were analyzed following SDS-PAGE and autoradiography.
Phosphoamino Acid Analysis-Phosphoamino acid analysis of ␣-casein band was performed following SDS-PAGE, transfer of the proteins to Immobilon-P, and excision of the radioactive ␣-casein band. The latter was incubated in 6 N HCl at 110°C for 1 h. The hydrolysates were separated by two-dimensional electrophoresis (32). The 32 P-labeled phosphoamino acids were detected by autoradiography and compared with ninhydrin-stained phosphoamino acid standards (Sigma).
GST Fusion Proteins-GST fusion proteins containing various domains of wild type human Nck (GST-SH2, residues 282-377; GST-SH3, residues 1-251; GST-Nck, full-length cDNA) were produced by polymerase chain reaction using specific oligonucleotide primers containing the appropriate restriction sites, subcloned into pGEX4T2 or pGEX2TK plasmids (Pharmacia) and sequenced. For SH3 mutants, a point mutation was performed in individual SH3 domains by extension overlapping polymerase chain reaction (33). In each SH3 domain, the first tryptophan residue (Trp 38 , Trp 143 , Trp 229 ) of the tryptophan doublet, well conserved among several SH3 domains, was mutated to arginine. The mutations were confirmed by dideoxynucleotide sequencing (Pharmacia), and the proteins were expressed and immobilized on glutathione-agarose beads. Equal amounts of each fusion protein (2 g) were incubated with HTC-IR cell lysates for 90 min at 4°C, after which proteins bound to GST fusion proteins immobilized on beads were assayed for associated kinase(s).
In Vitro Renaturation Kinase Assay-␣-Casein (50 g/ml) was dissolved in the polyacrylamide gel solution just prior to polymerization. Protein kinases from HTC-IR cell lysates interacting with the GST fusion proteins of various domains of Nck or coimmunoprecipitating with Nck were detected directly in the gel, by their ability to phosphorylate the ␣-casein substrate in the gel (34). Samples were subjected to SDS-PAGE, and SDS was removed by washing with 20% (v/v) isopropyl alcohol in 50 mM imidazole, 28 mM iminodiacetic acid, pH 8.0, twice for 60 min at room temperature. The gels were then washed with 50 mM imidazole, 28 mM iminodiacetic acid, pH 8.0, containing 10 mM mercaptoethanol for 60 min at room temperature. Proteins in the gels were then denaturated with 8.0 M guanidine HCl in 50 mM imidazole, 25 mM iminodiacetic acid, pH 8.0, containing 50 mM mercaptoethanol for 90 min at room temperature. Protein renaturation was achieved by successive washes at 4°C (2 ϫ 90 min, 200 ml; overnight, 400 ml; 1 ϫ 60 min, 200 ml) in 25 mM imidazole, 14 mM iminodiacetic acid, pH 8.0, containing 20 mM KCl, 10% sucrose, 10 mM mercaptoethanol, 1% bovine serum albumin, and 0.04% Tween 20. The gels were equilibrated in a solution containing 10 mM HEPES (pH 8.0), 10 mM mercaptoethanol, 5 mM MgCl 2 , and 10 mM ␤-glycerophosphate for 60 min at room temperature and incubated for an additional 120 min following the addition of [␥-32 P]ATP (20 Ci/ml). Finally, the gels were extensively washed with 5.0% (w/v) trichloroacetic acid containing 1.0% (w/v) sodium pyrophosphate and 1.0% sodium phosphate. After fixing and drying, the gels were exposed for autoradiography.
Far Western Analysis-Total cell lysates (20 g of proteins) from control or insulin-stimulated HTC-IR cells were resolved by SDS-PAGE and transferred to nitrocellulose. Membranes were blocked in 20 mM HEPES, pH 7.5, 5 mM MgCl 2 , 1 mM KCl, 5 mM dithiothreitol, 5 mM NaF, 0.02% sodium azide, 5% nonfat dry milk for 24 h at 4°C and probed overnight at room temperature with 32 P-labeled GST or 32 P-labeled GST-SH3 Nck (specific activity 1 ϫ 10 6 cpm/ml in blocking buffer) prepared according to the manufacturer's recommendations (pGEX2TK; Pharmacia). At the end of the incubation period, the membranes were extensively washed in Tris-buffered saline containing 0.1% Triton X-100 and exposed for autoradiography.
Nck and CKI-␥2 Coimmunoprecipitation-HTC-IR cells were harvested in buffer A (5 mM Tris-HCl, pH 7.4, 1 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride, 2 mM NaF, 2 mM Na 3 VO 4 , 1 mM MgCl 2 , and 0.25 M sucrose and homogenized using a Teflon-glass homogenizer. After centrifugation at 15,000 rpm for 10 min, the supernatant was gently agitated with anti-Nck or anti-CKI-␥2 rabbit polyclonal antibodies and protein A-Sepharose beads at 4°C. After 90 min, the beads were washed twice with cold phosphate-buffered saline, resuspended in Laemmli buffer, boiled for 2 min, and loaded on a 7.5% acrylamide gel. Proteins were transferred onto nitrocellulose, and the membrane was blocked in 5% milk. CKI-␥2 was detected using our rabbit polyclonal CKI-␥2 antibody, the horseradish peroxidase-coupled sheep anti-rabbit antibody (Bio-Rad), and the enhanced chemiluminescence reagent (ECL; Amersham Corp.).

RESULTS
A Serine/threonine Protein Kinase Coimmunoprecipitated with Nck-Using the yeast two-hybrid system, we have isolated a kinase-related cDNA encoding a protein that interacts with the three SH3 domains of Nck and whose amino acid sequence is identical to the C-terminal domain of rat CKI-␥2, which contains a unique proline-rich sequence (28). To verify an interaction between Nck and CKI-␥2 in intact cells, we immunoprecipitated Nck from unstimulated HTC-IR cell lysates and assayed for in vitro protein kinase activity using a variety of proteins as exogenous substrates. As shown in Fig. 1, a protein kinase activity was coimmunoprecipitated with Nck and preferentially phosphorylated myelin basic protein, ␣-casein, histone H1, and phosvitin as compared to ␤-casein and PolyGluTyr. The kinase activity toward ␣-casein was magnesium-dependent ( Fig. 2A). When magnesium and manganese were combined in the same assay, a lesser degree of kinase activity was observed (data not shown). Phosphoamino acid analysis of phosphorylated ␣-casein revealed specific phosphorylation of serine and threonine residues, demonstrating that the Nck-associated kinase is a serine/threonine protein kinase (Fig. 2B).
Effect of Insulin and Protein Kinase Inhibitors-We next The kinase activity coimmunoprecipitated with Nck is a magnesium-dependent serine/threonine protein kinase. A, unstimulated HTC-IR cells were lysed in 1% Triton lysis buffer and subjected to immunoprecipitation with antibodies to Nck. Immunoprecipitated proteins were assayed for in vitro kinase activity in the presence of 10 Ci of [␥ 32 P]-ATP, 20 M cold ATP, and 15 g of ␣-casein. In the assay, MgCl 2 or MnCl 2 were added at the concentrations indicated. Proteins were fractionated by 12% SDS-PAGE and visualized by autoradiography. B, phosphoamino acid analysis of radioactive ␣-casein was performed following SDS-PAGE and transfer of the proteins on Immobilon-P. The region of the membrane corresponding to radioactive ␣-casein was excised and incubated in 6 N HCl for 1 h. The hydrolysate was separated by two-dimensional electrophoresis. The 32 P-labeled phosphoamino acids were detected by autoradiography and compared with ninhydrin-stained phosphoamino acid standards.
FIG. 3. Effects of insulin and various agents on the in vitro kinase activity coimmunoprecipitated with Nck. Unstimulated or insulin-stimulated (100 nM, 5 min) HTC-IR cells were lysed in 1% Triton lysis buffer and immunoprecipitated with antibodies to Nck. A, in vitro kinase assays were performed (as described in Fig. 2) on immunoprecipitated proteins in the presence of CKI-7, GTP, heparin, and staurosporine at the concentrations indicated. Reaction products were fractionated by SDS-PAGE and visualized by autoradiography. B, dose response of CKI-7 inhibition of the kinase activity coimmunoprecipitated with Nck, as evaluated by in vitro phosphorylation of ␣-casein (15 g). studied the effect of treating cells with insulin and various drugs on in vitro protein phosphorylation in Nck immunoprecipitates (Fig. 3A). Insulin, when added for 5 min to the cells in culture, did not affect the kinase activity coimmunoprecipitated with Nck. Staurosporine, an inhibitor of the protein kinase C family, cAMP-and cGMP-dependent kinases, Ca 2ϩ / calmodulin-dependent protein kinase, and myosin light chain kinase (35), had no effect in vitro on the kinase activity, thus excluding these kinases as likely candidates. To evaluate the presence of casein kinases in Nck immunoprecipitates, we tested the effect of heparin (a known inhibitor of casein kinase II (CKII)) (36), GTP (which could be used as a phosphate donor by CKII but not by casein kinase I (CKI)) (37), and CKI-7 (a specific inhibitor of CKI activity) (38). Neither heparin nor GTP had an effect on protein phosphorylation in Nck immunoprecipitates, suggesting that CKII is not involved. However, CKI-7 strongly inhibited protein phosphorylation in Nck immunoprecipitates (Fig. 3A) and also prevented in vitro phosphorylation of ␣-casein (Fig. 3B) with an IC 50 of 20 M, which is in the range of concentrations for the inhibitory effect of CKI-7 on purified (38) or recombinant CKI activity (28,39). Taken together, these results are highly suggestive of an interaction between Nck and a member of the CKI family.
To determine whether the Nck-associated protein kinase activity could be modulated by insulin over a longer time course, insulin was added to the cells up to 30 min prior to the preparation of cell lysates. The effects of insulin on Nck-asso-ciated kinase activity were evaluated by in vitro phosphorylation of the overall proteins coimmunoprecipitated with Nck or by in vitro phosphorylation of ␣-casein added in the assay. When ␣-casein was used as an exogenous substrate, insulin stimulation did not seem to affect the kinase activity associated with Nck (Fig. 4, bottom). On the other hand, when the overall in vitro phosphorylation of proteins coimmunoprecipitated with Nck was analyzed, some proteins appeared less phosphorylated when the cells were stimulated 15 and 30 min by insulin (Fig. 4, top), suggesting that the kinase activity toward specific protein substrates might be regulated by insulin. These experiments did not discriminate between a change in the specific activity of the Nck-associated kinase and the amount of proteins associated with Nck upon insulin stimulation. A decrease in either could produce an apparent reduced phosphorylation of target proteins The Third SH3 Domain of Nck Mediates the Interaction with the Kinase-To determine which domain of Nck mediates the interaction with the CKI-like protein kinase, GST fusion proteins containing various Nck domains were incubated with HTC-IR cell lysates, and kinases bound to these GST fusion proteins were assayed using ␣-casein as substrate. Fig. 5 demonstrates that the SH3 domains of Nck associate with protein(s) containing kinase activity. In fact, the protein(s) with a kinase activity bound equally well to the GST fusion protein encoding the three SH3 domains of Nck (GST-SH3(3)) as it did to the full-length Nck (GST-Nck), but neither the fusion protein expressing the SH2 domain of Nck nor GST itself could interact with a kinase mediating in vitro phosphorylation of ␣-casein. The fusion proteins GST-SH3(3) and GST-Nck were also phosphorylated in this assay (data not shown), suggesting that Nck was a substrate of this kinase.
We attempted to identify which of the three SH3 domains of Nck interacts with the protein kinase by site-directed mutagenesis. Thus, we created single point mutations in each SH3 domain by substituting for arginine the first tryptophan residue (Trp 38 , Trp 143 , Trp 229 ) of the well conserved tryptophan doublet. This particular point mutation of SH3 domains has been reported to reduce dramatically the ability of SH3 domains to interact with specific proteins (40,41). Mutation in the first SH3 domain of Nck had no effect on the kinase activity recruited from HTC-IR cell lysate; however, mutations of the second and the third SH3 domain were accompanied by a sharp  6. The third SH3 domain of Nck associates a protein kinase. Purified GST fusion proteins containing wild type or individually mutated SH3 domains of Nck (W38R, W143R, and W229R) were incubated with cell lysates prepared from unstimulated HTC-IR cells (lysed in 1% Triton lysis buffer). Proteins associated to the GST fusion proteins were subjected to in vitro kinase assays as described in the legend to Fig. 2. Reaction products were fractionated on 12% acrylamide gel, and the results were visualized by autoradiography. decrease in Nck-associated kinase activity. This effect was greater with a mutation in the third, rather than the second SH3 domain (Fig. 6).
Identification of the Nck-associated Protein Kinase-To identify the protein kinase associated with Nck, an in vitro renaturation kinase assay in acrylamide gel was performed (Fig.  7A). This procedure allows detection of protein kinase in the gel by their ability to phosphorylate an exogenous substrate embedded throughout the gel. GST fusion proteins encoding the SH3 domains of Nck as well as the whole Nck protein associated with a protein kinase of 70 -75 kDa, which phosphorylates ␣-casein present in the gel (Fig. 7A). A similar protein, which phosphorylates ␣-casein, was also detected in Nck immunoprecipitates (Fig. 7A). In a direct binding assay (Far Western) on total cellular proteins immobilized on nitrocellulose, the 32 Plabeled GST fusion protein encoding the three SH3 domains of Nck strongly associated with a protein of 70 -75 kDa, and this association was not modulated by prior stimulation of cells with insulin (Fig. 7B). Other protein species of different molecular mass (55,120,190, and 230 kDa) also interacted directly with the three SH3 domains of Nck; however, in the in vitro renaturation kinase assay, none possessed kinase activity. Interestingly, association of the three SH3 domains of Nck with the 120-kDa protein (Fig. 7B) seems to be strongly reduced by prior cell stimulation with insulin, suggesting that insulin could modulate association of specific effector proteins with the SH3 domains of Nck. When 32 P-labeled GST was used to probe an identical membrane, no protein association was detected (data not shown). A rabbit polyclonal antibody against the C-terminal domain of casein kinase I-␥2 (cDNA obtained from the yeast two-hybrid screen) recognized, in HTC-IR cell lysates, a specific protein of 70 -75 kDa (inhibited by an excess of antigen; data not shown) (Fig. 7C), which is identical to the molecular mass of the Nck-associated protein kinase detected in the in vitro renaturation gel kinase assay and with one of the proteins associated with the SH3 domains of Nck in the direct binding assay. CKI-␥2 was identified in Nck immunoprecipitates from HTC-IR cells when cytosolic fractions were prepared in the absence of Triton X-100 (Fig. 7D). When 1% Triton X-100 was added to the cell extract before the immunoprecipitation of Nck, it reduced the specific interaction between Nck and CKI-␥2 to below the level of protein detection that could be FIG. 7. CKI-␥2, a serine/threonine protein kinase of 70 -75 kDa is constitutively associated with Nck. A, in vitro gel kinase assay.
Unstimulated HTC-IR cells lysed in 1% Triton lysis buffer were immunoprecipitated with antibodies to Nck or mixed with GST fusion proteins containing various domains of Nck. Bound proteins were fractionated by SDS-PAGE and subjected to an in vitro gel kinase assay as described under "Experimental Procedures." B, Far Western analysis. HTC-IR cells, unstimulated or stimulated with 100 nM insulin for 5 min, were lysed in 1% Triton lysis buffer. Proteins (20 g) were resolved by SDS-PAGE and transferred to nitrocellulose. Membranes were blocked 24 h at 4°C in 20 mM Hepes, pH 7.5, 5 mM MgCl 2 , 1 mM KCl, 5 mM dithiothreitol, 5 mM NaF, 0.02% NaN 3 , and 5% nonfat dry milk and then probed overnight at room temperature with 32 P-labeled GST-SH3 domains of Nck (10 6 cpm/ml) in the same buffer. Membranes were extensively washed in Tris-buffered saline containing 0.1% Triton and exposed for autoradiography. C, CKI-␥2 antibodies. Unstimulated HTC-IR cell lysate was immunoprecipitated with antibodies to CKI-␥2 (described under "Experimental Procedures") or normal serum. Immunoprecipitated proteins and proteins from total cell lysate (20 g) were resolved by SDS-PAGE and transferred to nitrocellulose. Membranes were blocked, and CKI-␥2 was revealed using antibodies to CKI-␥2, protein A-horseradish peroxidase, and ECL. D, coimmunoprecipitation of Nck-CKI-␥2. Unstimulated HTC-IR cells were harvested in 50 mM Tris-HCl, pH 7.4, 1 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride, 2 mM NaF, 1 mM MgCl 2 , and 0.25 M sucrose and homogenized using a Teflon-glass homogenizer. Lysates were immunoprecipitated with antibodies to Nck, CKI-␥2, or normal serum. Immunoprecipitated proteins were fractionated by SDS-PAGE, transferred to nitrocellulose, and immunoblotted with antibodies to CKI-␥2. Detection was achieved using goat-anti rabbit-HRP and ECL. observed in Western blot analysis of protein-protein interactions (data not shown). DISCUSSION Nck is a substrate of various cell surface tyrosine kinase receptors such as the epidermal growth factor and the plateletderived growth factor receptors (12,14,17). It also appears to be a substrate for a variety of serine/threonine kinases activated by several membrane receptors, like protein kinase C and protein kinase A (cAMP-dependent protein kinase) (12,14,17) and has been reported to be highly phosphorylated on serine residues in vivo in resting cells (14). Recently, two independent groups have demonstrated a constitutive physical association between serine/threonine protein kinases and the SH3 domains of Nck, providing a potential explanation for such a high level of serine phosphorylation of Nck in resting cells. First, NAK a 65-kDa serine/threonine kinase, has been reported to interact with the second SH3 domain of Nck (25). However, neither its biochemical characterization nor its exact role in mediating Nck signaling was described. Second, a serine/threonine kinase, mPAK-3 (p21 (Cdc42/Rac1)-activated kinase) was shown to specifically bind to at least one SH3 domain of Nck (22). The kinase activity of mPAK-3, as observed for other members of the PAK family of serine/threonine protein kinases, is increased by activated Rac1 and Cdc42, the small GTP-binding proteins involved in the characteristic cytoskeletal rearrangements associated with cell motility (20,42). This raises the possibility that Nck is involved in cytoskeletal interactions through its interaction with mPAK-3.
In this study, we have demonstrated that in HTC-IR cells, Nck is constitutively associated with a serine/threonine protein kinase whose activity is magnesium-dependent. This association is mediated particularly by the third SH3 domain, although the second SH3 domain in the linear sequence might be involved in stabilizing the interaction. These results contrast with those for NAK and mPAK3, which have been shown to bind exclusively to the second SH3 domain of Nck (25,43). This supports our conclusion that in HTC-IR cells the serine/threonine protein kinase coimmunoprecipitated with Nck differs from NAK and mPAK3. In addition, it is possible that NAK and PAK3 could be the same protein kinase, because both are 65 kDa and are found associated to the same SH3 domain of Nck.
Several lines of evidence support the identity of the Nckassociated kinase in HTC-IR cells as a member of the CKI family of protein kinases. First, CKI-7 a specific inhibitor for CKI enzymes completely inhibited the activity of the protein kinase coimmunoprecipitated with Nck. Second, the Nck-associated kinase could be distinguished from CKII by its inability to use GTP as a phosphate donor and by its weak sensitivity to heparin. Furthermore, numerous kinases were ruled out by the absence of effect of staurosporine. Although staurosporine is a broad spectrum inhibitor of serine/threonine and tyrosine kinase activities, it has not been reported as an inhibitor of the casein kinase family of protein kinases (35). Moreover, the fact that the kinase activity coimmunoprecipitated with Nck was not (at least apparently) affected by prior treatment of cells with insulin is in agreement with the fact that casein kinase family members are second messenger-independent (37). Finally, we have demonstrated that CKI-␥2, a member of the CKI family of protein kinases (28), coimmunoprecipitates with Nck in vivo. Among casein kinase I-␥ isoforms, CKI-␥2 is the only one that contains a proline-rich sequence (28). Moreover, this unique proline-rich sequence in the C-terminal region of CKI-␥2 is composed of two juxtaposed PXXP motifs (VHPD-VPSQPPHR), which are recognized as consensus motifs for SH3-ligand interactions and could support the constitutive association of CKI-␥2 with Nck.
In insulin signal transduction, the importance of IRS-1 has been firmly established (44 -46). Following insulin receptor activation, tyrosine-phosphorylated IRS-1 has been shown to regulate PI-3 kinase activity by its direct association with p85, the regulatory subunit of PI-3 kinase. Binding of p85 to IRS-1 leads to activation of the p110 catalytic domain of PI-3 kinase (47,48). Similarly, IRS-1 has been reported to associate with Grb2 upon insulin stimulation (49). The binding of Grb2 to IRS-1 results in recruitment of Sos to IRS-1, thus providing a link between the activation of the Ras signaling pathway and insulin receptor activation (50). Nck is found associated with IRS-1 upon insulin stimulation (16), suggesting that this complex could regulate specific effector molecules bound to Nck. Unlike Grb2 and p85, the associated catalytic components of Nck are not yet well characterized, and its biological function in insulin signaling remains uncertain. Our finding that Nck is constitutively bound to CKI-␥2 in HTC-IR cells could implicate CKI-␥2 in insulin signaling. It is possible that following insulin receptor activation, the binding of Nck to IRS-1 may regulate the catalytic activity of CKI-␥2 or result in its translocation to specific subcellular sites, thus bringing it in proximity to substrate molecules. Interestingly, the ␤-subunit of the insulin receptor has been reported to be phosphorylated by casein kinase I in vitro (29), suggesting that if this happens in vivo, casein kinase I might participate in regulating insulin receptor activity. Alternatively, IRS-1 could be a target of casein kinase I, since several consensus motifs for phosphorylation by casein kinase I (XEXXS*X, where X is any residue, E is glutamic acid, and S* is the phosphoserine) (51) are present in the amino acid sequence of IRS-1. Of course, the presence of consensus motifs for phosphorylation does not ensure that IRS-1 is phosphorylated by CKI-␥2 in vivo. However, the serine/threonine phosphorylation of IRS-1 by an unidentified kinase was reported to abrogate insulin-stimulated glucose transport (52), suggesting a mechanism of negative feedback regulation triggered by activation or translocation of serine/threonine kinases following insulin stimulation. Casein kinase I reported to be associated to the p75 tumor necrosis factor receptor, plays an important role in the negative regulation of tumor necrosis factor receptor signaling (53). Hypothetically, CKI, by phosphorylating the insulin receptor and/or IRS-1, could exert a similar effect on insulin-activated signaling pathways.
In insulin-stimulated cells, since Nck forms a stable complex with IRS-1 (16), it is suggested that the insulin receptor could regulate the function of Nck and consequently the activity of effector molecules associated with the SH3 domains of Nck. The level of Nck was increased in epididymal fat and in liver from obese insulin-resistant diabetic mice (KKA y ) (54) and significantly decreased in liver of streptozotocin-treated rats (55). These findings suggest that changes in the level of Nck expression might contribute to altered insulin responsiveness involving modification of the Nck downstream signaling components.
Insulin is well known to induce increases in the rates of RNA and protein synthesis in many tissues (56). These effects require the insulin receptor and IRS-1 (57). More specifically, insulin appears to regulate the serine/threonine phosphorylation of several elongation and initiation factors involved in controlling protein synthesis (58 -60), and in vitro studies have demonstrated that these factors are phosphorylated by several kinases, including CKI (61)(62)(63). Thus, CKI could be required for insulin regulation of protein synthesis. The finding that Nck associates with CKI-␥2 demonstrates a new mechanism by which the insulin receptor can couple to unidentified signaling molecules. Further work is needed to characterize the interac-tion between Nck and CKI-␥2 and to determine the downstream elements regulated by this association.