Requirements of Protein Kinase Cδ for Catalytic Function

Recently, we reported that, in contrast to protein kinase C (PKC)α and βII, PKCδ does not require phosphorylation of a specific threonine (Thr505) in the activation loop for catalytic competence (Stempka et al. (1997) J. Biol. Chem. 272, 6805–6811). Here, we show that the acidic residue glutamic acid 500 (Glu500) in the activation loop is important for the catalytic function of PKCδ. A Glu500 to valine mutant shows 76 and 73% reduced kinase activity toward autophosphorylation and substrate phosphorylation, respectively. With regard to thermal stability and inhibition by the inhibitors Gö6976 and Gö6983 the mutant does not differ from the wild type, indicating that the general conformation of the molecule is not altered by the site-directed mutagenesis. Thus, Glu500 in the activation loop of PKCδ might take over at least part of the role of the phosphate groups on Thr497and Thr500 of PKCα and βII, respectively. Accordingly, PKCδ exhibits kinase activity and is able to autophosphorylate probably without posttranslational modification. Autophosphorylation of PKCδ in vitro occurs on Ser643, as demonstrated by matrix-assisted laser desorption ionization mass spectrometry of tryptic peptides of autophosphorylated PKCδ wild type and mutants. A peptide containing this site is phosphorylated also in vivo, i.e. in recombinant PKCδ purified from baculovirus-infected insect cells. A Ser643 to alanine mutation indicates that autophosphorylation of Ser643 is not essential for the kinase activity of PKCδ. Probably additional (auto)phosphorylation site(s) exist that have not yet been identified.

PKC␦ is one of the most thoroughly studied members of the novel PKC subfamily. After the discovery of the enzyme in 1986 (4), its cloning in 1987 (5), and its first purification to homoge-neity in 1990 (6), several groups have focused their interest on this PKC isoenzyme and have reported on its structural and enzymatic properties (7)(8)(9)(10)(11)(12)(13), regulation of expression (14 -18), interaction with binding proteins (19 -24), specific substrate phosphorylation (25)(26)(27), and cellular functions (28 -36). Recently, the role of phosphorylation of PKC␦, i.e. either autophosphorylation or phosphorylation by an exogenous protein kinase, for the regulation of its enzymatic activity could, to some extent, be elucidated and compared with that of other PKC isoforms. It was shown that, in contrast to PKC␣ (37) and PKC␤ II (38), PKC␦ does not require phosphorylation of a specific threonine (Thr 505 corresponding to Thr 497 and Thr 500 of PKC␣ and ␤ II , respectively) in the activation loop for catalytic competence (39). PKC␦ wild type as well as a Thr 505 to alanine mutant were expressed in bacteria in a catalytically competent form. Specific activities of these enzymes were comparable with that of native PKC␦ from porcine spleen (39). Recent studies with a Ser 643 to alanine mutant of PKC␦ indicated that phosphorylation of this residue might be required for enzymatic activity (40). Ser 643 was not unequivocally identified as an autophosphorylation site of PKC␦ but corresponds to one of the in vivo autophosphorylation sites of PKC␤ II , i.e. Thr 641 (41,42). Finally, it was reported by several groups that tyrosine phosphorylation affects the kinase activity of PKC␦ and also other PKC isoforms in vitro and in vivo (10,(43)(44)(45)(46)(47).
Here we demonstrate that glutamic acid 500 is important for the catalytic function of PKC␦ and thus might take over at least part of the role of the phosphate groups of Thr 497 and Thr 500 of PKC␣ and ␤ II , respectively. By MALDI mass spectrometry of tryptic PKC␦ peptides we provide evidence that Ser 643 is an autophosphorylation site of PKC␦ in vitro and that a PKC␦ peptide containing Ser 643 is phosphorylated also in vivo.
length cDNA with an NdeI restriction site at the initiation signal ATG and an EcoRI restriction site behind the stop codon TGA was amplified by polymerase chain reaction and cloned into the expression vector pET28. The plasmid was termed pET28␦wt.
Site-directed mutagenesis was performed using the "overlap extension" method as described previously (39). A rat PKC␦ full-length clone of 3000 bp served as a template. The following pairs of mutagenic oligonucleotides were used (only the sense oligonucleotides are given, and changed bases are underlined): (E/V)500, 5Ј-GAAT ATA TTT GGG GTG AAC CGG GCT* AGC ACA TTC-3Ј; (S/A)643, 5Ј-GAG AAA CCC CAA CTT GCC TTC AGT GAC AAG AAC C-3Ј; and (S/A)643(S/A)645, 5Ј-G AAA CCC CAA CTT GCC TTC GCT GAC AAG AAC CTC-3Ј (synthesized by Dr. W. Weinig, German Cancer Research Center). Successful mutation was confirmed by sequencing and, in the case of (E/V)500, in addition by introducing a new NheI restriction site that was created by a silent point mutation (see base with asterix). The mutant cDNAs were cloned into pET28 and were termed pET28␦Val 500 , pET28␦Ala 643 , and pET28␦Ala 643/5 .
Bacterial Expression and Partial Purification of Recombinant Histagged PKC␦-E. coli BL21(DE3)pLysS cells were used for expression of the recombinant PKC␦ wt and mutants. Cells were grown under conditions described previously (39). Washed and sedimented cells were cracked by the method of freezing and thawing and resuspended in ice-cold buffer (50 mM sodium phosphate, pH 8.0, 150 mM NaCl, 5 mM imidazole, 1% Triton X-100, 10% glycerol, and protease inhibitors: phenylmethylsulfonyl fluoride, aprotinin, leupeptin, and pepstatin). After sonication with a Branson sonifier and centrifugation at 100,000 ϫ g for 30 min at 4°C, the supernatant was applied to an nickel-nitrilotriacetic acid column following the manufacturer's recommendation. Bound proteins, eluted with 50 mM and 100 mM imidazole, were pooled, diluted 1:5 with buffer (10 mM Tris-HCl, pH 7.5, and protease inhibitors), and further purified by chromatography on a Mono-Q column. Elution of bound proteins was achieved with NaCl (steps of 100, 200, 300, and 500 mM). The 200 mM NaCl fraction was selected, because it contained PKC␦ with the highest specific activity. According to Coomassie Blue staining a 75% purity of PKC␦ was estimated. Eluted PKC␦ was stabilized by addition of 10% glycerol and 10 mM ␤-mercaptoethanol and stored at Ϫ70°C.
Expression of Recombinant PKC␦ Containing a C-terminal His Tag in the Baculovirus-Insect Cell System-For the cloning of PKC␦ fulllength cDNA into the pBac1 baculovirus transfer plasmid (Novagen), a PKC␦ cDNA with an EcoRI restriction site at the initiation signal ATG and an XhoI restriction site following the removed stop codon TGA was amplified by polymerase chain reaction as described for the pET28␦wtplasmid (39). The oligonucleotides 5Ј-GAC GAA TTC ATG GCA CCG TTC-3Ј and 5Ј-GGA CCC TCG AGT TCC AGG AAT TGC-3Ј were used as 5Ј and 3Ј primers, respectively. The construction and amplification of recombinant baculovirus were performed using the Bac Vector 2000 transfection kit (Novagen) following the manufacturer's recommendation. Sf9 cells were harvested after 65 h of infection with the recombinant baculovirus, and recombinant PKC␦ was partially purified as described above for the bacterially expressed enzyme. According to Coomassie Blue staining, 80% purity of PKC␦ was estimated. The kinase activities of partially purified PKC␦ wild type (A) and Val 500 mutant (B) were determined as described under "Experimental Procedures." Five g of the pseudosubstrate-related peptide ␦ were phosphorylated by PKC␦ wild type and mutant (0.3 g each of protein) with [ 32 P]ATP (concentrations as indicated) in the presence of PS and TPA at 30°C for 9 min. The reciprocal values of phosphate incorporation (1/V) were plotted as a function of the reciprocal ATP concentrations. The intercepts of the double reciprocal plots with the x-and y-axis give the K m and V max values, respectively (see Table I).

TABLE I Kinase activities and K m and V max values of PKC␦ wild type and mutants
Phosphorylation of the pseudosubstrate-related peptide ␦ (peptide ␦) with partially purified PKC␦ wild type (wt) and mutants (0.3 g each of protein) in the presence of PS and TPA was performed as described under "Experimental Procedures." ND, not determined. V max of wt (100%), 384 units/mg; kinase activity of wt (100%), 259,700 cpm. The values of kinase activities are the mean of two determinations of two independent experiments Ϯ S.E. 113 Ϯ 12 ND ND ND a According to slightly different amounts of PKC␦ wild type and mutants in the enzyme preparations (e.g. see immunoblots of Figs. 1 and 8), kinase activities and V max of mutants were normalized in comparison with that of wild type (see text).
Protein Kinase Assay and Autophosphorylation-Phosphorylation reactions were carried out at 30°C for 5 min as described previously (39). The pseudosubstrate-related peptide ␦ was used as a substrate. 32 Plabeled phosphoproteins were visualized and quantitated by measuring the intensity of photo-stimulated luminescence using a Bio-Imaging Analyzer (Fuji Bas 1500).
Immunoblotting-Proteins were separated by SDS-polyacrylamide gel electrophoresis (7.5%). PKC␦ was detected and quantitated by immunoblotting using the monoclonal anti-PKC␦ antibody P36520 as the first antibody and an alkaline phosphatase-conjugated second antibody. The blots were scanned using a scanner (MacIntosh), and the values were expressed as arbitrary units relative to the background.
Enzymatic Digestion and MALDI Mass Spectrometry-Autophosphorylation of 10 g of partially purified PKC␦ wild type and mutants was performed as described above, but [␥-32 P]ATP was omitted, and the reaction was terminated by precipitation of the proteins with methanolchloroform (49). Proteins were separated by SDS-polyacrylamide gel electrophoresis (7.5%, 0.75 mm). After staining with Coomassie R250 (Bio-Rad), the PKC␦ bands were excised, cut into small pieces (1 ϫ 1 mm), washed, dehydrated (2 ϫ 30 min with H 2 O, 3 ϫ 15 min with 50% acetonitrile, and 1 ϫ 15 min with acetonitrile), and incubated with 2 g of trypsin in 50 l of digest buffer (50 mM NH 4 HCO 3 , pH 8.0) at 37°C for 16 h. The digest was sonicated and centrifuged, and the supernatant was subsequently analyzed by MALDI mass spectrometry using the thin film preparation technique (50). Aliquots of 0.3 l of a saturated solution of ␣-cyano-4-hydroxycinnamic acid in acetone containing nitrocellulose were deposited onto individual spots on the target. Subsequently, 1 l of 10% formic acid and 0.5 l of the digest were loaded on top of the thin film spots and allowed to dry slowly at ambient temperature. The spots were washed with 10% formic acid and deionized water to remove salts.
MALDI mass spectra were recorded in the positive ion mode on a Reflex II time-of-flight instrument (Bruker-Franzen, Bremen, Germany) equipped with a SCOUT multiprobe inlet and a 337-nm nitrogen laser. Ion acceleration voltage was set to 25 kV, and the reflector voltage was 26.5 kV. When using delayed extraction the first extraction plate was set to 18.5 kV. Mass spectra were obtained by averaging 20 -50 individual laser shots. Calibration of the spectra was performed externally by a two-point linear fit using angiotensin I and oxidized insulin ␤-chain.

RESULTS AND DISCUSSION
Phosphorylation of Thr 497 and Thr 500 in the activation loop of PKC␣ (37) and PKC␤ II (38), respectively, is known to be required for catalytic competence of the enzymes. Recently, we demonstrated that PKC␦ exhibits full enzymatic activity without phosphorylation of the corresponding Thr 505 (39). As previously discussed, a structural difference in the activation loops of the isoenzymes could possibly explain the differential behavior of PKC␦ to PKC␣ and ␤ II . In contrast to PKC␣ and PKC␤ II , PKC␦ contains a glutamic acid in position 500 (Glu 500 ) that might take over the role of the phosphorylated threonines of PKC␣ and ␤ II .
To test this possibility, we mutated Glu 500 to the neutral amino acid valine and determined the enzymatic activity of the bacterially expressed PKC␦Val 500 mutant. All phosphorylation assays were performed in the presence of PS and TPA. Autophosphorylation of partially purified wild type and PKC␦Val 500 mutant is shown in Fig. 1. Approximately equal amounts of the enzymes were applied to the assay. The values determined by autoradiography were normalized according to the slight difference in the amount of wild type and mutant enzyme (factor, 1.2) that was observed by immunoblotting with an anti-PKC␦ antibody and scanning the immunoblots (Fig. 1). Using different preparations of the enzymes in two independent experi- ments, one of which is shown in Fig. 1, a 76 Ϯ 4% reduction in PKC␦Val 500 autophosphorylation activity, compared with the wild type, was observed. A similar decrease in enzymatic activity after mutation of Glu 500 to Val was seen when substrate phosphorylation by wild type and mutant were compared. Phosphorylation of the pseudosubstrate-related peptide ␦ by PKC␦Val 500 was reduced by 73 Ϯ 2% compared with that by the wild type (Table I). The low incorporation of phosphate into the peptide by PKC␦Val 500 was essentially attributable to a slower velocity of the phosphorylation reaction. V max of the PKC␦Val 500 catalyzed reaction was reduced by 72% compared with V max of the wild type, whereas the K m values for the peptide substrate did not differ significantly (Table I). The K m value for ATP, however, was approximately three times higher with PKC␦Val 500 (116 M) than with the wild type (42 M; Fig.  2 and Table I). Thus, binding of ATP appeared to be impeded with mutation of Glu 500 to Val. For comparative purposes, some kinetic data of the partially purified PKC␦Ala 505 and PKC␦Ala 504/505 mutants are also shown in Table I. The kinase activities of these mutants did not differ from the wild type, as reported previously (39). Taking the autophosphorylation and substrate phosphorylation data together, mutation of Glu 500 to Val causes an ϳ75% reduction in PKC␦ kinase activity. These results clearly support the idea that the negatively charged carboxylate of Glu 500 in the activation loop is important for the kinase activity of PKC␦ and might function in a similar way as the phosphate groups of Thr 497 and Thr 500 in PKC␣ and ␤ II , respectively, in correctly aligning residues involved in catalysis by interaction with positively charged residues of the catalytic core. However, the residual activity of PKC␦Val 500 (ϳ25%) indicates that, in addition to Glu 500 , some as yet unknown structural features of PKC␦ might be required for its catalytic function. Several kinases are known that contain acidic residues such as Glu, instead of phosphorylated residues, in the activation loop (51). Moreover, catalytic competence of PKC␤ II could be attained by mutation of Thr 500 to Glu (38). No other PKC isoenyzme contains a glutamic acid residue in a position corresponding to position 500 of PKC␦; however, aspartic acid is located in this position in some isoforms (PKC, /, and ). It is not yet known whether these aspartic acid residues function similarly to the glutamic acid 500 of PKC␦. According to Orr and Newton (38), aspartic acid may be less suited than glutamic acid for the electrostatic interactions that cause correct alignment of catalytic residues.
Mutation of Glu 500 to Val affected neither the thermal stability of PKC␦ nor its inhibition by the specific PKC inhibitors Gö6983 and Gö6976. The kinase activity of the mutant, like wild type PKC␦, remained relatively stable at room tempera- ture for at least 40 min (Fig. 3). Inhibition of mutant and wild type PKC␦ was almost identical, either by Gö6983 in the nM range or by Gö6976 in the M range (Fig. 4). These data indicate that site-directed mutagenesis did not alter the general conformation of the molecule.
PKC␤ II autophosphorylates residues Thr 641 and Ser 660 (41,42). Phosphorylation of Thr 641 appears to be essential for maintaining catalytic competence of the enzyme (38). The same holds true for Thr 642 of PKC␤ I (52). Phosphorylation of the corresponding Thr 638 of PKC␣, however, is not required for the catalytic function of the enzyme, as reported by Bornacin and Parker (53). To elucidate a putative role of autophosphorylation in the enzymatic activity of PKC␦, as a first step we attempted to identify autophosphorylation sites of in vitro phosphorylated PKC␦.
Recombinant PKC␦ partially purified from bacterial extracts was phosphorylated in the presence of PS and TPA. Phosphorylated and nonphosphorylated PKC␦ were applied to SDSpolyacrylamide gel electrophoresis. The PKC␦ bands were cut out of the gels, and the gel slices were washed and incubated with trypsin. Aliquots of the tryptic digests were analyzed by MALDI mass spectrometry. Expanded views of the obtained mass spectra from m/z 2760 to m/z 2910 are given in Fig. 5. The upper panel shows ion signals of tryptic peptides from nonphosphorylated PKC␦ wild type, whereas the lower panel shows the corresponding tryptic peptides from the phosphorylated PKC␦ wild type. The signal at m/z 2807 present in all mass spectra (Figs. 5-7) was observed previously when tryptic digests from gel slices were applied. Its nature is not known. The ion signal at m/z 2791 in the mass spectrum of nonphosphorylated PKC␦ (Fig. 5, upper panel) could be assigned to the PKC␦ peptide 624 SPSDYSNFDPEFLNEKPQLSFSDK 647 . In the mass spectrum of phosphorylated PKC␦ (Fig. 5, lower  panel) the ion signal at m/z 2791 had disappeared almost completely. Instead a signal at m/z 2871 was observed. Thus, the digest of in vitro phosphorylated PKC␦ contained predominantly the phosphorylated peptide 624 -647 (m/z 2791 ϩ 80, i.e. the mass of one phosphate group). A signal at m/z 2871, which was rather weak compared with that at m/z 2791, could be observed also in the mass spectrum of the nonphosphorylated PKC␦ (Fig. 5, upper panel), indicating that peptide 624 -647 of bacterially expressed PKC␦ was to some degree phosphorylated in vivo. However, a quantitative evaluation and comparison of signals in MALDI mass spectra is possible only to a very limited extent. Precursor ion selection was applied to confirm that the ion signal at m/z 2871 indeed represented a phosphopeptide. When the precursor ion selector was set to m/z 2791, no fragmentation occurred (Fig. 5, upper panel,  inset). However, when the molecular ion at m/z 2871 was selected as precursor ion, one additional signal was observed at m/z 2783, which is attributable to the loss of H 3 PO 4 (Fig. 5,  lower panel, inset). The signal at m/z 2783 was seen also in the original spectrum of the phosphorylated PKC␦ (Fig. 5, lower  panel). This metastable fragmentation is characteristic for pep- tides containing phosphoserine and phosphothreonine (46,54). The ion signal for the peptide fragment produced by metastable fragmentation did not appear at its correct m/z value because the mass scale was not calibrated for fragment ions. However, previous studies with phosphopeptides have shown that under the tuning parameters used in this experiment the observed mass deviation is in fact characteristic for the loss of H 3 PO 4 (46).
The phosphorylated tryptic peptide 624 SPSDYSNFDPEFL-NEKPQLSFSDK 647 contains 5 serine residues, each of which could have been phosphorylated. However, Ser 643 corresponds to the in vivo autophosphorylation site Thr 641 of PKC␤ II (41,42) and thus could be assumed to be one of the autophosphorylated residues of PKC␦. Furthermore, comparison of phosphopeptide maps of in vivo phosphorylated PKC␦ wild type and Ser 643 to alanine mutant indicated that Ser 643 might be an in vivo phosphorylation site of PKC␦ (40). However, autophosphorylation was not proven, and a phosphopeptide containing Ser 643 was not identified. Therefore, we mutated Ser 643 to alanine (PKC␦Ala 643 ) and also produced the double mutant Ser 643/645 -Ala (PKC␦Ala 643/5 ) by site-directed mutagenesis. Both mutants were phosphorylated and then treated as described above for the PKC␦ wild type. Expanded views of the MALDI mass spectra obtained from the tryptic peptides of the phosphorylated mutants PKC␦Ala 643/5 (upper panel) and PKC␦Ala 643 (lower panel) are shown in Fig. 6. The expected mass signals of the nonphosphorylated peptides corresponding to the peptide m/z 2791 of the PKC␦ wild type with masses of 2759 (2791 Ϫ 32, i.e. the mass of two oxygens) for PKC␦Ala 643/5 and 2775 (2791 Ϫ 16, i.e. the mass of one oxygen) for PKC␦Ala 643 were found. However, we failed to detect signals of significant intensity at m/z 2839 (2759 ϩ 80) or m/z 2855 (2775 ϩ 80), corresponding to the phosphorylated forms of the mutants (see Fig. 6, arrows). Thus, the mass spectra of the phosphorylated mutants are distinctly different from the mass spectrum of the phosphorylated wild type, which almost completely lacked the signal of the nonphosphorylated peptide and instead showed the signal of the phosphorylated peptide (compare Fig. 5). This result clearly demonstrates that Ser 643 is one of the in vitro autophosphorylated residues of PKC␦. A small, not clearly recognizable, signal at m/z 2855 in the mass spectrum of PKC␦Ala 643 (Fig. 6, lower panel, arrow) may indicate a very weak phosphorylation of Ser 645 , because the corresponding signal at m/z 2839 is completely missing in the mass spectrum of the double mutant PKC␦Ala 643/5 (Fig. 6, upper  panel, arrow). The slight phosphorylation of Ser 645 possibly occurs only with mutation of the major phosphorylation site Ser 643 . The tryptic peptide 624 -647 containing Ser 643 was found to be phosphorylated also in the digest of recombinant PKC␦ partially purified from baculovirus-infected insect cells. The MALDI mass spectrum indicated complete phosphoryla- FIG. 8. Autophosphorylation of PKC␦ wild type (wt) and PKC␦ Ala 643 mutant (Ala643). Partially purified PKC␦ wild type and mutant (0.3 g of each protein) were phosphorylated at 30°C for 5 min and quantitated with a scanner after immunoblotting (a.u., arbitrary units) as described in Fig. 1 and under "Experimental Procedures." Radiolabeled PKC␦ was visualized by autoradiography and quantitated using a phosphoimager (PSL, photostimulated luminescence; see Fig. 1).
tion of PKC␦ at this site, because exclusively the signal of the phosphorylated peptide at m/z 2871 (m/z 2791 ϩ 80) was observed (Fig. 7). Precursor ion selection (Fig. 7, inset) was applied to confirm that the ion signal at m/z 2871 indeed represented a phosphopeptide. As with the in vitro phosphorylated PKC␦ (see above, Fig. 5), one additional signal was observed at m/z 2783 which is attributable to the loss of H 3 PO 4 . This signal was distinct also in the original spectrum. Even though PKC␦ can be expected to contain more than one (auto)phosphorylation site (see below), neither in vitro nor in vivo additional phosphopeptides could be detected by MALDI mass spectrometry as yet. In two independent experiments using different preparations of the enzymes, autophosphorylation of the PKC␦Ala 643 mutant was reduced to 54 Ϯ 11% of that of the wild type. One of the two experiments is shown in Fig. 8. Values determined by autoradiography were normalized according to the slightly different amount of PKC␦ wild type and mutant (see the immunoblot in Fig. 8), as described above. A similar result was obtained with the PKC␦Ala 643/5 mutant (data not shown). The impeded autophosphorylation of the mutants supports the result of the mass spectrometric study showing that Ser 643 is an autophosphorylation site of PKC␦. It also indicates, however, that additional phosphorylation site(s) exist, because the mutation of Ser 643 did not completely abolish autophosphorylation. Mutation of Ser 643 or Ser 643/5 to alanine affected neither kinase activity, and K m and V max values of PKC␦ (Table I), nor its thermal stability (Fig. 3). The lack of requirement of Ser 643 autophosphorylation for kinase activity of PKC␦ is in agreement with the entirely intact kinase activity of untreated PKC␦ wild type expressed in bacteria, which, according to the mass spectrum (Fig. 5), contains predominantly nonphosphorylated Ser 643 . It is also in agreement with the finding of Bornacin and Parker (53) showing that the corresponding Thr 638 of PKC␣ is not required for the catalytic function of the enzyme. However, it is in contrast to a previous report by Li et al. (40) that indicated a reduced kinase activity after Ser 643 to alanine mutation of murine PKC␦ expressed in myeloid 32D cells. One possible explanation for these contradictory results might be the different sources and purities of PKC␦ used in the two in vitro studies. For example, it is conceivable that the Ser 643 to alanine mutation causes an increased sensitivity of the enzyme to proteolysis or dephosphorylation and that the PKC␦ Ala 643 mutant from myeloid 32D cells is partially inactivated because of the action of contaminating proteases and/or phosphatases.
Taken together, we have shown that the catalytic function of PKC␦ depends in part on the acidic residue Glu 500 in the activation loop and, as reported earlier (39), not on phosphorylated Thr 505 . Thus, PKC␦ exhibits kinase activity and is able to autophosphorylate without the posttranslational modification that is required for catalytic competence of PKC␣ and ␤ II . Autophosphorylation in vitro, and most likely also in vivo, occurs on Ser 643 , which corresponds to the autophosphorylation site Thr 641 of PKC␤ II . Probably other (auto)phosphorylation site(s) of PKC␦ exist that have not yet been identified.