Regulated Interaction of Protein Kinase Cδ with the Heterogeneous Nuclear Ribonucleoprotein K Protein*

The heterogeneous nuclear ribonucleoprotein (hnRNP) K protein recruits a diversity of molecular partners that are involved in signal transduction, transcription, RNA processing, and translation. K protein is phosphorylated in vivo andin vitro by inducible kinase(s) and contains several potential sites for protein kinase C (PKC) phosphorylation. In this study we show that K protein is phosphorylated in vitro by PKCδ and by other PKCs. Deletion analysis and site-directed mutagenesis revealed that Ser302 is a major K protein site phosphorylated by PKCδ in vitro. This residue is located in the middle of a short amino acid fragment that divides the two clusters of SH3-binding domains. Mutation of Ser302decreased the level of phosphorylation of exogenously expressed K protein in phorbol 12-myristate 13-acetate-treated COS cells, suggesting that Ser302 is also a site for PKC-mediated phosphorylation in vivo. In vitro, PKCδ binds K protein via the highly interactive KI domain, an interaction that is blocked by poly(C) RNA. Mutation of Ser302 did not alter the K protein-PKCδ interaction in vitro, suggesting that phosphorylation of this residue alone is not sufficient to alter this interaction. Instead, binding of PKCδ to K protein in vitro and in vivo was greatly increased by K protein phosphorylation on tyrosine residues. The ability of PKCδ to bind and phosphorylate K protein may serve not only to alter the activity of K protein itself, but K protein may also bridge PKCδ to other K protein molecular partners and thus facilitate molecular cross-talk. The regulated nature of the PKCδ-K protein interaction may serve to meet cellular needs at sites of active transcription, RNA processing and translation in response to changing extracellular environment.

The hnRNP 1 K protein has a diverse repertoire of molecular partners that are involved in signal transduction and gene expression. K protein binds with RNA, single-stranded, and double-stranded DNA, and it associates with a number of transcriptional activators and repressors, including TATA-binding protein (1). K protein also interacts with tyrosine (2) and serine/threonine kinases (3)(4)(5) as well as the proto-oncoprotein Vav (3,6,7). The diverse molecular interaction of K protein may account for the observations that K protein can both increase (8,9) and decrease (10 -12) gene transcription. For example, on one hand K protein synergies with TATA-binding protein to increase transcription from the c-myc promoter CT element (8), whereas on the other it represses C/EBP␤-mediated transcription of the agp gene (11) and inhibits Sp-1-mediated activation of the neuronal nicotinic acetylcholine receptor promoter (12). As one of the constituents of the hnRNP particle, K protein may be involved in the processing of pre-mRNA (13,14). K protein shuttles between the nucleus and cytoplasm and therefore could serve as a vehicle that is involved in RNA transport (15). K protein-mediated silencing of 15-lipoxygenase mRNA (16) represents an example of K protein involvement in the regulation of translation. Involvement of K protein in translational processes is further supported by its association with the elongation factor 1␣ (1). The association of K protein with tyrosine kinases (2,3) and with Vav (6, 7) may reflect involvement of K protein in signal transduction. Alternatively, the Vav-and/or tyrosine kinases-K protein interaction may regulate K protein transcriptional and/or translational activity. Considering the diversity of K protein molecular interactions, it is not surprising that new reports are emerging implicating K protein involvement in viral processes. For example, K protein has been shown to functionally interact with hepatitis C virus core protein (17) and to regulate translation of the human papillomavirus type 16 L2 mRNA (18).
K protein is made up of modular domains that bind different molecular partners. For example, the three KH domains are thought to mediate nucleic acid binding (19,20), whereas the two clusters of SH3-binding sites recruit the proto-oncoprotein Vav (3) and the Src-class of tyrosine kinases (2). The latter sites are contained within the same region that binds several transcriptional repressors, such as Zik1 (21), as well as the global regulator of anterior-posterior patterning, Eed (22). This module is adjacent to a nuclear shuttling domain, KNS (15), that may mediate direct coupling of K protein to a bona fide nuclear/ cytoplasmic shuttling transporter(s). Finally, a domain near the C terminus recruits an interleukin-1-responsive kinase that phosphorylates K protein in a nucleic acid-dependent fashion (3,4). Although K protein is already known to contain a number of binding domains, it seems that there are potentially others that remain to be identified. The abundant and ubiquitous expression of K protein, its multimodular structure, its potential to oligomerize via its two different dimerization do-mains, and its apparent involvement in a wide range of processes responsible for transcription, translation, and signal transduction suggest that K protein may act as a scaffold or docking platform. Alternatively, K protein may be a multifunctional factor that is involved in processes that are not directly related. In either scenario the function of K protein is likely to be regulated by post-translational modification and by cognate nucleic acid motifs.
K protein is phosphorylated in vivo and in vitro on serine and threonine residues (5,23). At least in part, this phosphorylation is mediated by an associated kinase(s) that can respond to treatment of cells with interleukin-1 and other agents (4,5). Thus, K protein phosphorylation is likely to play a key role in the regulation of its activity. This postulate is supported by the observation that the binding of K protein to poly(C) in vitro is diminished by phosphorylation (23). Moreover, in hepatocytes following systemic administration of lipopolysaccharide into rats, there was a complete dissociation of the transcriptional factor C/EBP␤ from K protein (11), a process that may reflect phosphorylation of K protein. Analysis of the K protein amino acid sequence reveals a number of potential phosphorylation sites by casein kinase II, protein kinase C (PKC), and tyrosine kinases. Indeed, in vitro, K protein is an excellent substrate for casein kinase II (24). However, casein kinase II is typically a constitutively active enzyme (25), making its role in the inducible phosphorylation of K protein in vivo less likely than the role of inducible serine/threonine kinases such as the PKC family of enzymes.
The PKC family of enzymes transduce intracellular signals that regulate many different intracellular processes (26). This heterogenous family of enzymes is divided into three classes based on their Ca 2ϩ and lipid requirements (27). The conventional PKCs (␣, ␤, and ␥) require phosphatidylserine, diacylglycerol, or phorbol 12-myristate 13-acetate (PMA), and Ca 2ϩ ; the novel subgroup of PKCs (␦, ⑀, , , and ) are Ca 2ϩ -independent but require the other two co-factors; and finally the least understood subgroup of PKCs ( and ) are both Ca 2ϩ -and diacylglycerol-independent. Among the PKC isoenzymes, PKC␦ has unique properties that suggest a functional connection to K protein. First, PKC␦ phosphorylates elongation factor 1␣ (28), a factor that binds K protein (29). Second, K protein interacts with Src-tyrosine kinases, a class of enzymes that phosphorylate PKC␦ (30). Third, K protein contains several potential PKC phosphorylation sites. Because K protein contains sites for PKC phosphorylation and because PKC␦ and K protein share molecular partners, in this study we explored the possibility that K protein is a PKC␦ substrate.
Reagents-The bacterial expression vector pGEX-KT was provided by Dr. J. Dixon (University of Michigan). Glutathione-agarose beads, reduced glutathione, and PMA were obtained from Sigma. Fast Flow Protein A-Sepharose was obtained from Amersham Pharmacia Biotech. Polyclonal anti-K protein antibody 54 was made in rabbits as described previously (4). Monoclonal antiphosphotyrosine and polyclonal anti-PKC␦ antibody was purchased from Santa Cruz and anti-PKC␦ monoclonal antibody was purchased from Transduction Laboratories. Baculovirus expressed and spleen PKC␦ were produced as described previously (33,34).
Extraction of Cytoplasmic and Nuclear Proteins-Nuclear and cytoplasmic extracts were prepared by a modified version of the method of Dignam et al. (35) as described previously (5). In addition to 0.5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, and 10 g/ml leupeptin, lysis, extraction, and dilution buffers contained the following phosphatase inhibitors (all from Sigma): 30 mM p-nitrophenyl phosphate, 10 mM NaF, 0.1 mM Na 3 VO 4 , 0.1 mM Na 2 MoO 4 , and 10 mM ␤-glycerophosphate. Protein content was measured using the DC method (Pierce).
K Protein Plasmids Constructs-GST-K, GST-K3, GST-K, GST-K10, GST-K12, GST-K13, and GST-K14 were constructed as described (3). For the GST-K31 deletion mutant, polymerase chain reaction was used to introduce a BamHI site immediately upstream of Met 240 and to introduce a stop codon followed by an EcoRI site downstream of Val 337 . GST-K31 was created by cloning this excised fragment into pGEX-KT. For the GST-K⌬PB deletion mutant, pGEX-K was cut with PpuMI and BglII, then filled in with Klenow fragment, and religated to create the GST-K⌬PB construct. For Flag-K for mammalian expression, K protein was excised from pM1-K (22) with EcoRI and SalI and then ligated into p18Flag that had been cut with EcoRI and XhoI, creating the recombinant Flag-K. The point mutations of Ser 302 to Ala or Glu in GST-K, GST-K31, and Flag-K were generated using the QuickChange Site-Directed Mutagenesis kit (Stratagene). All plasmids were purified by CsCl gradient before use in transient transfections, and the mutations were confirmed by automated sequencing using the DyeDeoxy Terminator Cycle Sequencing kit (PE Applied Biosystems, Foster City, CA).
Synthesis and Purification of GST-K Constructs-GST fusion proteins were expressed in either BRL or BL21 (DE3) pLysS cells (Novagen, Madison, WI) using a modified manufacturer's protocol. Transformed cells were grown until they reached A 600 ϭ 0.6 and then were treated for 3 h with 1.0 mM isopropyl-␤-D-thiogalactoside. Following freezing and thawing, pellets were resuspended in E. coli extraction buffer and sonicated. Fusion proteins were recovered in the supernatant after centrifugation at 14,000 rpm for 30 min (4°C). GST-K and GST-K deletion mutants were then purified on a glutathione column as described previously (3).
In Vitro Transcription and Translation-In vitro transcription and translation was performed using the TNT T7 Quick Coupled Transcription/Translation System as per the manufacturer's protocol (Promega, Madison, WI) Binding to Beads Bearing GST-K Proteins-Binding to beads was carried out by mixing glutathione beads with the standard binding buffer containing a given protein or RNA at 4°C. After extensive washes, proteins were eluted from the beads by boiling in 1ϫ SDSloading buffer (36) and were then loaded on a 10% SDS gel and autoradiographed.
Transient Transfections-COS cells were grown in Dulbecco's minimal essential medium supplemented with 10% fetal calf serum to approximately 60 -75% confluency in 100-mm diameter dishes and were transfected using SuperFect Transfection Reagent as per the manufacturer's protocol (Qiagen Inc., Santa Clarita, CA).

RESULTS
K Protein Is Phosphorylated in Vitro by PKC␦-K protein is phosphorylated in vivo and in vitro on serine residues by a kinase(s) with which it forms a complex (5,23). Analysis of K protein amino acid sequence reveals a number of potential sites for phosphorylation by PKC. To test whether K protein is a substrate for this class of enzymes, purified bacterially expressed full-length GST-K fusion protein was phosphorylated in solution by purified porcine spleen PKC␦ in the presence of [␥-32 P]ATP. The kinase reaction was terminated by boiling the sample with loading buffer and then separated by SDS-PAGE. Phosphorylation of GST-K protein was assessed by scintillation counting of the 32 P-labeled GST-K bands cut from the gel (Fig.  1). The results showed that PKC␦ phosphorylates K protein in vitro and that the rate of phosphate incorporation was dependent on GST-K protein concentration. Under these conditions, phosphorylation of GST-K had a V max of 286 pmol phosphate/ min/mg PKC␦ and a K m of 1.9 nM (Fig. 1B). These results show that under these conditions the velocity of phosphorylation of GST-K by PKC␦ was relatively slow, but the low K m suggests that K protein may bind PKC␦ with a high affinity. As expected, phosphorylation of K protein by PKC␦ is dependent on phosphatidylserine and PMA and does not require Ca 2ϩ (data not shown).
Mapping of K Protein Site That Is Phosphorylated in Vitro by PKC-A series of GST-K deletion mutants ( Fig. 2A) (3) was used to identify a domain that is phosphorylated by PKC␦. Glutathione beads bearing either full-length GST-K or one of the deletion mutants were phosphorylated by baculovirus expressed PKC␦ (28,37). The kinase reaction was terminated by washing the beads with HKMT buffer, proteins were eluted from the beads by boiling in SDS loading buffer, and the level of phosphorylation of GST-K fusion proteins was assessed by SDS-PAGE and autoradiography (Fig. 2B). In addition to phosphorylating the full-length GST-K, PKC␦ also phosphorylated GST-K13 (a.a. 1-337), GST-K3 (a.a. 171-337), and GST-K31 (a.a. 240 -337) deletion mutants. In contrast, the deletion mutants GST-K12 (a.a. 1-209) and GST-K7 (a.a. 318 -464) were not phosphorylated at all. These results indicated that the PKC␦ site(s) is contained within the GST-K31 deletion mutant (a.a. 240 -337). This domain contains three serines, Ser 276 , Ser 284 , and Ser 302 , and no threonines. Ser 302 is located within a good consensus for PKC-mediated phosphorylation, Arg-Gly-Gly-Ser-Arg-Ala-Arg (24,28). To test whether Ser 302 is the site that is phosphorylated in vitro by PKC␦, site-directed mutagenesis was used to mutate this residue to either Ala (GST-K S302A and GST-K31 S302A ) or Glu (GST-K S302E and GST-K31 S302E ) ( Fig. 2A). These mutants were tested as substrates for phosphorylation by PKC␦. As before, the phosphorylation reactions were carried out on beads bearing GST-K fusion proteins (Fig.  2). Autoradiographs from these phosphorylation reactions are shown in Fig. 2C. In sharp contrast to the full-length GST-K, the point mutants GST-K S302A and GST-K S302E were poorly or not at all phosphorylated by PKC␦. Although the GST-K31 S302A and GST-K31 S302E mutants were phosphorylated, the level of phosphorylation of these Ser 302 mutants was much lower than that observed with the wild-type GST-K31. These results indicate that Ser 302 is the major site for phosphorylation of GST-K protein by PKC␦ in vitro, whereas Ser 277 and/or Ser 284 are only minor site(s). This is supported by the observation that an internal deletion mutant that lacks the prolinerich SH3-binding domains, GST-K⌬PB (a.a. 288 -321 deleted), was a very poor substrate for PKC␦ (data not shown).
We used the above approach to determine whether K protein is a substrate for other PKC isoforms. Beads bearing different GST-K protein mutants (Fig. 3A) were phosphorylated by a partially purified mixture of PKC␣, ␤, and ␥. As with PKC␦, GST-K and GST-K31 were phosphorylated by PKC␣, ␤, and ␥, whereas GST-K7 (a.a. 318 -464) was not phosphorylated at all, and the level of phosphorylation of the internal deletion mutant that lacks the two clusters of the proline-rich SH3-binding domains (3), GST-K⌬PB (a.a. 288 -321 deleted) was very low (Fig. 3B). Because the deleted a.a. 288 -321 region contains no other phosphorylation sites, the markedly decreased level of phosphorylation of the GST-K⌬PB mutant suggests that Ser 302 is also a major K protein site for phosphorylation by the calciumdependent PKC␣, ␤, and ␥. The level of PKC␣-, ␤-, and ␥mediated phosphorylation of GST-K S302A , GST-K S302E , GST-K31 S302A , and GST-K31 S302E mutants was lower compared with their respective wild-type GST-K fusion proteins (Fig. 3C), providing further evidence that Ser 302 is a major site of phosphorylation by these PKC isoforms in vitro.
PMA-inducible Phosphorylation of K Protein in Vivo-K pro- tein is phosphorylated in vivo on serine residues (5). Thus, to determine whether Ser 302 can also be phosphorylated in vivo, we compared the levels of phosphorylation of wild-type Flag-K and of point mutant Flag-K S302A (Ser 302 3 Ala) that were co-expressed with HA-PKC␦ in COS cells. Transfected cells were metabolically labeled with [ 32 P]orthophosphate and were then treated with or without 10 Ϫ7 M PMA for 1 h. Following treatment, cells were harvested, and cytoplasmic and nuclear extracts were prepared as described previously (5). Equal amounts of extracts were precipitated with either pre-immune or immune anti-K protein serum (antibody 54). The immunoprecipitates were separated by SDS-PAGE and were then electrotransferred to Immobilon-P membrane (Millipore, Bedford, MA) and were analyzed by autoradiography (Fig. 4, 32 P) and by Western blotting using anti-K protein serum (Fig. 4, ␣K). The autoradiograph revealed that in the cytoplasm the constitutive level of phosphorylation of Flag-K was higher that of Flag-K S302A . In the nuclear fraction, phosphorylation of Flag-K was PMA-inducible, whereas phosphorylation of Flag-K S302A was not. In both the nucleus and the cytoplasm, the level of Flag-K PMA-inducible phosphorylation was higher than the level of Flag-K S302A phosphorylation in PMA-treated cells. Unlike the exogenously expressed Flag-K proteins, the levels of PMAinducible phosphorylation of endogenous K protein (Fig. 4, lower band in the autoradiograph marked K), in the Flag-K and Flag-K S302A transfected cells were similar. These experiments provide evidence that Ser 302 can be phosphorylated in vivo, a reaction that may, in part, be mediated by PKC␦ and/or other PKC isoenzymes.
K Protein Binds PKC␦ through Its Highly Interactive Domain-The low K m for the phosphorylation of GST-K by PKC␦ ( Fig. 1) suggests that the two proteins may bind one another with a high enough affinity to form detectable complexes. To test such a possibility, beads bearing GST or GST-K fusion proteins (Fig. 5) were mixed with baculovirus expressed PKC␦, and after binding and washing, bound proteins were eluted from the beads by boiling in loading buffer. Proteins eluted from the beads were separated by SDS-PAGE and electrotransferred to Immobilon-P membrane for immunostaining with anti-PKC␦ antibody. To ensure that beads contained similar levels of GST fusion proteins, another gel was stained with Coomassie. These results showed that beads bearing GST-K protein, but not beads bearing GST alone, pulled down PKC␦, indicating that K protein binds PKC␦ in vitro. Several GST-K deletion mutants were used to map the K protein domain that binds PKC␦ in vitro. GST 5 and 7-10). After IP, beads were washed for 5 min with 0.2 ml of HKMT-PI buffer, followed by 5 min with 0.2 ml of wash buffer. Proteins were eluted by boiling in SDS-loading buffer, separated by SDS-PAGE, and then electrotransferred onto Immobilon-P membrane. Membranes were immunostained with anti-K protein antibody 54 (1:10,000 dilution) (lower panel, ␣K), and were then autoradiographed (upper panel, 32 P). The exogenous Flag-K and Flag-K S302A and the endogenous K protein are indicated. B, COS cells were transfected with either Flag or Flag-K, extracts were prepared and K and Flag-K proteins were immunoprecipitated with anti-K antibody as above. Immunoprecipitated proteins were resolved by SDS-PAGE, and Western blotting was done with either anti-K protein (lanes 1 and 2) or anti-Flag (lanes 3 and 4) antibodies. The endogenous K protein (K) and Flag-K fusion protein (Flag-K) are marked.
PKC␦ in vitro is contained within the a.a. 240 -337 region and that the proline-rich region may be important in this interaction. The a.a. 240 -337 K protein domain interacts with a number of known K protein partners. In addition to recruiting PKC␦, this domain also mediates the binding of the transcriptional repressors Zik1 (21) and Eed (22), several tyrosine kinases (3), and the proto-oncoprotein Vav (6). It is likely that this domain recruits many other factors that are involved in signal transduction and gene expression. We designate this region as the KI domain, for K protein interactive domain.
Role of Ser 302 on the in Vitro Interaction of K Protein with PKC␦ in the Presence or Absence of RNA-Ser 302 is located within the highly interactive KI domain, in the middle of a short amino acid stretch that splits the two clusters of SH3binding domains. The location of Ser 302 suggests that this residue might play a role in the regulation of K protein interaction with PKC␦ and/or other partners. To test this possibility, beads bearing either wild-type GST-K protein or GST-K protein with mutated Ser 302 , GST-K S302A or GST-K S302E , were mixed with baculovirus expressed PKC␦, and the binding was assessed as before by SDS-PAGE and Western blotting with an anti-PKC␦ antibody. Results illustrated in Fig. 6A show that mutating Ser 302 to either Ala or Glu did not alter the ability of K protein to recruit PKC␦. The in vitro interaction of K protein with many of its molecular partners can be regulated by cognate RNA, such as poly(C) RNA, or a cognate DNA, such as the B, motif (4, 21). Thus, we also tested whether mutation of Ser 302 alters the affinity of K protein-PKC␦ complex when K protein is bound to poly(C). These results showed that poly(C) RNA abrogated the in vitro interaction of K protein with PKC␦ independently of Ser 302 and suggests that the ability of K protein to bind poly(C) is not altered by the mutation of Ser 302 . Abrogation of the in vitro association between K protein and PKC␦ by poly(C) is similar to the observations of other K protein partners that are recruited to K protein by the KI domain (21).
Although under present conditions mutation of Ser 302 had no detectable effect on the in vitro interaction of K protein with PKC␦ or on its ability to bind poly(C), this residue may play a role in the engagement of other K protein partners. For example, the protein Eed is recruited to K protein by the KI domain (22), an interaction that could be modulated by Ser 302 . 35 S-Eed was mixed with GST-K, GST-K S302A , or GST-K S302E beads that had been pre-equilibrated with or without RNA. The K proteinbound 35 S-Eed was analyzed by SDS-PAGE and autoradiography (Fig. 6B). These results revealed that mutation of Ser 302 to Ala diminished the ability of 35 S-Eed to be recruited by K protein, and the Ser 302 to Glu 302 mutation had even more of a blocking effect on this association. These results suggest that the recruitment of some of the K protein partners may be regulated by phosphorylation of Ser 302 .
Tyrosine Phosphorylation of K Protein Modulates Its Interaction with PKC␦ in Vitro and in Vivo-Phosphorylation of K protein by the tyrosine kinase Lck regulates K protein inter- action with a number of its molecular partners. 2 We tested whether tyrosine phosphorylation also regulates the association of PKC␦ with K protein. Glutathione beads bearing GST-K protein were incubated in Lck phosphorylation buffer with or without baculovirus Lck. After extensive washes, the beads were incubated with 35 S-PKC␦ synthesized in the cell-free system (Promega). The beads were washed again, and the bound proteins were eluted by boiling, separated by SDS-PAGE, and transferred to Immobilon-P membrane. Fig. 7A illustrates re-sults from this experiment. Lck-mediated tyrosine phosphorylation of K protein (upper blot) greatly enhanced the in vitro binding of 35 S-PKC␦ to K protein (Fig. 7A, lanes 1 and 2, middle  blot). The gels also showed that tyrosine phosphorylation of K protein shifted its electrophoretic mobility and the protein band became wider (Fig. 7A, compare lanes 1 and 2, bottom  blot). The marked changes in the electrophoretic mobility indicate a significant change in K protein structure that coincides with strong recruitment of PKC␦ to K protein. Not unexpectedly, the Lck-mediated phosphorylation of K protein has different effects on the interaction of K protein with different molecular partners. Although the K protein binding of PKC␦  1-3) or immune (lanes 4 -6) anti-K protein antibody 54 (␣K) rabbit serum. After sonication, the samples were centrifuged for 5 min at 13,000 ϫ g (4°C), and the supernatants were added to 20 l of protein A/G beads (Santa Cruz Biotechnology). The suspensions were mixed for 30 min (4°C), then the beads were washed four times with 1 ml of IP buffer, and the proteins were eluted by boiling in 50 l of loading buffer and resolved by SDS-PAGE. After electrotransfer to Immobilon-P membrane, immunostaining was done either with (lower insert) anti-K protein serum 54 (1:5000 dilution), alkaline phosphatase-conjugated anti-rabbit antibody (1:3000 dilution, Bio-Rad), and 5-bromo-4-chloro-indolyl-phosphatase/nitroblue tetrazolium phosphatase substrate (Kirkegaard & Perry Laboratories) or with (upper insert) anti-phosphotyrosine monoclonal (PY99) as in A (middle blot). C, 200 l of lysates from given time points of H 2 O 2 /Na 3 VO 4 -treated Jurkat cells were sonicated for 2 h (4°C) with 50 l of either anti-PKC␦ monoclonal antibody (␣PKC␦, lanes 1-3) (Transduction Laboratories) or 50 l of anti-Flag monoclonal antibody (␣Flag, lanes 4 -6) (Santa Cruz). After sonication, 20 l of protein A/G beads were added to each sample. The suspensions were mixed for 2 h (4°C), the beads were washed four times with 1 ml of IP buffer, and the proteins were eluted by boiling in loading buffer. Eluted proteins were resolved on SDS-PAGE, and after electrotransfer to Immobilon-P membrane, immunostaining of K protein was carried out with anti-K protein antibody 54 (␣K) using the sandwich technique (upper insert). To assess the levels of PKC␦ in the immunoprecipitates (middle insert), Immobilon-P membranes were immunostained with a monoclonal anti-PKC␦ antibody (1:200 dilution) (Transduction Laboratories), alkaline phosphatase-conjugated goat anti-mouse antibody (1:1000 dilution) (Santa Cruz), and 5-bromo-4-chloro-indolyl-phosphatase/nitroblue tetrazolium. To assess tyrosine phosphorylation of PKC␦ (lower insert) membranes were immunostained with anti-phosphotyrosine monoclonal antibody (1:200 dilution) (PY99) and horseradish peroxidase-conjugated goat anti-mouse secondary antibody (1:2000 dilution) (Amersham Pharmacia Biotech) and developed with ECL (Amersham Pharmacia Biotech). (Fig. 7A), as well as the binding of tyrosine kinases and Vav, is increased by tyrosine phosphorylation of K protein, the K protein binding of Zik-1 and Eed is blocked by tyrosine phosphorylation. Moreover, the K protein binding of elongation factor 1␣ and TATA-binding protein is not affected by K protein tyrosine phosphorylation. 2 The combination of H 2 O 2 /Na 3 VO 4 stimulates tyrosine kinases in a myriad of cell types (38). To test the effects of tyrosine phosphorylation of K protein on in vivo binding of PKC␦, Jurkat cells were treated with H 2 O 2 /Na 3 VO 4 . At given time points shown in Fig. 7, cell lysates were prepared, and immunoprecipitations were carried out with either pre-immune or immune anti-K protein serum. After SDS-PAGE and electrotransfer, Immobilon-P membranes were immunostained with either anti-phosphotyrosine (Fig. 7B, lower blot) or anti-K protein (Fig. 7B, upper blot). Anti-phosphotyrosine immunostaining showed that treatment of Jurkat cells with the combination of these agents induced a transient increase in tyrosine phosphorylation of K protein; there was a low constitutive level of K protein tyrosine phosphorylation, an easily detectable increase after 15 min of treatment, and a decrease after 60 min (Fig. 7B, upper blot). As in vitro, the electrophoretic mobility of in vivo tyrosine phosphorylated K protein was slower (Fig. 7B, compare lanes 4 and 5, upper blot).
Next we tested whether PKC␦ exists in a complex with K protein in vivo and, if so, whether this association is regulated by tyrosine phosphorylation. Immunoprecipitations were carried on cell lysates from Jurkat cells treated with H 2 O 2 / Na 3 VO 4 using a monoclonal anti-PKC␦ antibody. A monoclonal anti-Flag antibody was used as a control. Fig. 7C (upper blot) shows that K protein co-immunoprecipitated with PKC␦ from cell lysates; a very low level of K protein co-immunoprecipitated with PKC␦ from cell lysates of untreated cells (lane 1); there was a large increase in co-immunoprecipitated K protein after 15 min of treatment (lane 2) and a significant decrease at 60 min (lane 3). These results suggest that in vivo there is a low constitutive level of PKC␦ binding to K protein, an association that is greatly enhanced by treatment of cells with H 2 O 2 / Na 3 VO 4 . To assess the levels of immunoprecipitated PKC␦ and to determine whether there is tyrosine phosphorylation of PKC␦, anti-PKC␦ immunoprecipitates from Jurkat cell lysates were analyzed by SDS-PAGE and anti-PKC␦ (Fig. 7C, middle blot) and anti-phosphotyrosine (Fig. 7C, bottom blot) immunostaining. These results showed that the amount of PKC␦ immunoprecipitated from cell lysates progressively decreased after the treatment and that the electrophoretic mobility of PKC␦ was slower (compare lanes 1-3). The amount and the electrophoretic mobility of the immunoprecipitated PKC␦ reflect accurately the levels of PKC␦ found in these cell lysates by Western blotting (data not shown). Although the decrease in PKC␦ levels in cell lysates may reflect translocation of this enzyme to the particulate fraction and/or proteolytic cleavage, the slower electrophoretic mobility may result from tyrosine phosphorylation of PKC␦. Because the in vitro binding of PKC␦ to K protein is increased by tyrosine phosphorylation of K protein (Fig. 7A) and because treatment of Jurkat cells with H 2 O 2 /Na 3 VO 4 stimulates tyrosine phosphorylation of K protein (Fig. 7B), the enhanced association of PKC␦ with K protein in vivo is likely the result, at least in part, of tyrosine phosphorylation of K protein. Although in vitro the increased binding of PKC␦ to tyrosine phosphorylated K protein does not require tyrosine phosphorylation of PKC␦ (Fig. 7A), treatment of cells with H 2 O 2 /Na 3 VO 4 stimulates tyrosine phosphorylation of this enzyme (Fig. 7C, lower blot), a modification that may contribute to the enhanced PKC␦-K protein association. Regardless of the specific mechanisms responsible for the enhanced associa-tion, these results illustrate that the binding of PKC␦ to K protein is regulated in response to changes in the extracellular environment.
Poly(C) RNA Disrupts the Native K Protein⅐PKC␦ Complex-The above results demonstrate that the in vitro complex formation between recombinant K protein and PKC␦ is blocked by poly(C) RNA (Fig. 6), which tenaciously binds K protein. Next we tested whether cognate RNA can disrupt the native PKC␦-K protein complex. Proteins from cytoplasmic extracts were precipitated with either pre-immune or immune anti-K protein serum and protein A/G beads. After a round of washing, proteins were eluted from the beads with either PKC buffer alone or PKC buffer containing either poly(A) or poly(C) RNA. Eluates were then analyzed by SDS-PAGE followed by Western blotting with a monoclonal anti-PKC␦ antibody. Immunostaining revealed that PKC␦ was eluted from beads bearing anti-K serum but not from the pre-immune beads and that the amount of eluted PKC␦ was the highest with poly(C) RNA (Fig.  8). This finding suggests that in vivo the cognate nucleic acids have the potential to regulate the PKC␦-K protein association. Along with the observation that tyrosine phosphorylation modulates the association between K protein and PKC␦ (Fig. 7), the effect of poly(C) RNA on this interaction (Figs. 6 and 8) provides further evidence that the association between the two proteins is regulated. DISCUSSION K protein has previously been shown to be phosphorylated in vivo, both constitutively and inducibly, by serine/threonine kinases (5,23). Until now, casein kinase II was the only kinase known to phosphorylate K protein (24,39). In the present study, we demonstrate that K protein forms a complex with PKC␦ (Figs. 6 -8) and can serve as its substrate (Figs. 1 and 2). K protein is not a specific PKC␦ substrate because it can also be phosphorylated by PKC␣, ␤, and ␥ (Fig. 3). In addition to the conventional (␣, ␤, and ␥) and novel PKCs (␦, ⑀, , and ), K protein may also be a substrate for the atypical PKCs ( and ). In that regard, it is notable that hnRNP A1 is a substrate for, and binds to, PKC (40).
We have identified Ser 302 as a major site of phosphorylation by PKC␦ (Fig. 2) and by PKC (␣, ␤, and ␥) (Fig. 3) in vitro. This site is also phosphorylated in vivo in response to treatment of cells with PMA (Fig. 4), suggesting that the in vivo phospho- FIG. 8. Poly(C) RNA disrupts native K protein-PKC␦ complex. 2.0 mg of EL-4 cytoplasmic extracts containing 1% Triton X-100 were pre-cleared by centrifugation for 30 min (4°C) at 15,000 ϫ g. Supernatants were sonicated for 1 h (4°C) with either 10 l of pre-immune or anti-K protein antibody 54 rabbit serum. The samples were centrifuged again, and the supernatants were mixed with 40 l of bovine serum albumin-blocked protein A beads for 30 min (4°C). After precipitation, beads were washed three times with 1.0 ml of HKMT buffer and were then mixed for 30 min (4°C) with 50 l of PKC binding buffer containing 2 g of bovine serum albumin without RNA (lanes 1 and 4) or with 1 g of either poly(A) (lanes 2 and 5) or poly(C) (lanes 3 and 6). After elution, the suspension was centrifuged, and the eluates were boiled in SDS-loading buffer. Eluted proteins were separated by SDS-PAGE, electrotransferred to Immobilon-P membrane, and then immunostained with anti-PKC␦ monoclonal antibody (50 g/10 ml TBST) (Transduction Laboratory). 20% of the total amount of PKC␦ directly immunoprecipitated from 2.0 mg of EL-4 cytoplasmic extracts with anti-PKC␦ monoclonal antibody was run in lane 7.
rylation of K protein is, at least in part, mediated by PKC␦ and/or other PKCs. Ser 302 is located exactly in the middle of a stretch that separates the two proline-rich domains that exhibit SH3 binding activity (3). This strategic location suggests that Ser 302 plays a role in the regulation of K protein interaction with some of its partners. In support of this hypothesis is the observation that mutation of Ser 302 to Glu diminished the affinity of K protein-Eed interaction in vitro (Fig. 6). Although the mutation of Ser 302 had no effect on K protein binding to PKC␦ or RNA in vitro (Fig. 6), there may be other partners in addition to Eed whose association with K protein are regulated by this residue. It is also likely that phosphorylation of other sites in conjunction with Ser 302 may play a role in modulating the binding of PKC␦ to K protein. Moreover, phosphorylation of Ser 302 may regulate cross-talk among factors that are simultaneously engaged by K protein. For example, Vav binds K protein and is tyrosine phosphorylated in response to cytokines (41). Phosphorylation of Vav may occur in the context of K protein simultaneously engaging Vav and a tyrosine kinase via the two clusters of SH3-binding domains (3). If so, it is conceivable that the juxtaposition of Vav and its tyrosine kinases may be altered by PKC␦-mediated phosphorylation of K protein Ser 302 . Such a model could be applied to many of the K protein molecular partners.
K protein binds PKC␦ with high enough affinity (Figs. 1 and 6) to allow the two proteins to exist as detectable complexes in vivo (Figs. 7 and 8). What is the physiological relevance of the inducible nature of this association? First, the inducible binding is likely to ensure effective phosphorylation of K protein by PKC␦, especially in response to a changing extracellular environment. Second, the binding of PKC␦ to K protein may link this enzyme to its targets or effectors that are concurrently present in the K protein microenvironment. Third, because K protein shuttles between the nucleus and cytoplasm (15), PKC␦ could be co-transported to specific subcellular compartments by K protein. These three scenarios are not mutually exclusive, and there may be other physiological meanings of the PKC␦-K protein association. Whatever the physiological role(s) may be, the PKC␦-K protein binding could be modulated by changes in the extracellular environment that may include oxidative stress (Fig. 7, B and C) and acute phase reactions. The in vivo interaction between the two partners may be further regulated by cognate nucleic acids, such as specific RNA sequences. These findings, in conjunction with the previous observations that K protein phosphorylation is modulated by interleukin-1 (5), and the report that in intact organs K protein association with some of its molecular partners is modulated by an acute phase reaction (11), provide evidence that K protein function is regulated and responds to the needs of the cell in the face of a changing external environment.
Based on what is already known about K protein and its molecular partners, there are a number of specific scenarios that can be envisaged where PKC␦-K protein interaction may play a role in determining cellular response. For example, in response to cytokines, oxidative stress, or acute phase reaction, the Src family of tyrosine kinases are activated, phosphorylating and then binding to K protein. Tyrosine phosphorylation of K protein would then induce enhanced binding of PKC␦ to K protein (Fig. 7). The simultaneous binding of a tyrosine kinase and PKC␦ in the context of K protein would provide an opportunity for cross-talk between these enzymes. Tyrosine-phosphorylated and activated PKC␦ could then dissociate from K protein and target other factors. Alternatively, K protein may provide a platform that facilitates the ability of PKC␦ to target those substrates that are simultaneously recruited by K protein. In regard to these models, a number of PKCs, including PKC␦ (Fig. 7C), are tyrosine phosphorylated, a modification that renders them phospholipid-and Ca 2ϩ -independent (42). Because the recruitment of PKC␦ to K protein can be dramatically altered in vitro by cognate RNA (Figs. 6 and 8), the K protein-facilitated cross-talk between PKC␦ and its targets and/or its effectors may be regulated by RNA in vivo. Moreover, K protein serving as a nucleic acid-interacting docking platform would facilitate molecular cross-talk at sites of active transcription, translation, and other processes involving RNA and DNA. The recruitment of PKC␦ to K protein is likely to be just one example of a more general phenomenon involving the association of K protein with several members of the PKC family of enzymes. This is supported by the observations that PKC␣, ␤, and ␥ phosphorylate K protein (Fig. 3) and by a report that K protein binds PKC␣ in vitro (43). Moreover, because the yeast homologs of mammalian K protein and PKC have been shown to be functionally linked (44), the K protein-PKC interactions appear to be evolutionarily conserved in species as diverse as yeast and mammals.
PKC␦ binds to a site within the highly interactive K protein region hereby designated as the KI domain (Fig. 5). Besides PKC␦, the KI domain binds the Src family of kinases (3), the proto-oncoprotein Vav (7), the transcriptional repressor Zik1 (21), the Polycomb group protein Eed (22), and likely many other factors. Within the KI domain these K protein partners may bind to the same or different sites. However, not all K protein partners bind to the KI domain. For example, TATAbinding protein binds K protein very strongly (8) through a region that is different from the KI domain. 3 Other examples of KI domain-independent interaction includes the binding of RNA and DNA to the KH domains (19,45). Moreover, although the transcriptional factor C/EBP␤ binds to a domain contained in the N-terminal half of the molecule (11), an interleukin-1responsive kinase binds to a domain in the vicinity of the C terminus (3).
What is the significance of the observation that many factors that interact with K protein are recruited by the KI domain? At least two models can be construed. The two dimerization domains may allow K protein to form higher order structures. If so, oligomerized K protein would contain a number of KI domains that could engage the same or different partners permitting uni-or multi-lateral cross-talk. It has been suggested that K protein may be involved in the transport of mRNA (15). Because K protein has the ability to shuttle between the nucleus and cytoplasm (15), it may not only serve to transport RNA, but it may also shuttle proteins. In that case, the KI domain may serve as a docking site for transport of these factors between the nuclear and cytoplasmic compartments. Whatever the role of the KI domain may be, the activity of this region toward some of the partners is regulated by growth factors, cytokines, and other extracellular stimuli that could exert their effect through the activation of PKC␦, Src-tyrosine kinases, and other enzymes that phosphorylate sites located within the KI domain.
In summary, we have shown that PKC␦ binds and phosphorylates K protein. These observations broaden the range of K protein interactions. PKC␦ targets Ser 302 , which is located in the middle of what appears to be a highly interactive KI domain. The ability of PKC␦ to inducibly bind and phosphorylate K protein may serve not only to alter the activity of K protein itself, but K protein may also provide an avenue for PKC␦ to engage in a cross-talk with other K protein molecular partners in response to specific changes in the extracellular environment.