Substrate Recognition by Ca2+/Calmodulin-dependent Protein Kinase Kinase

Mammalian Ca2+/CaM-dependent protein kinase kinase (CaM-KK) has been identified and cloned as an activator for two kinases, CaM kinase I (CaM-KI) and CaM kinase IV (CaM-KIV), and a recent report (Yano, S., Tokumitsu, H., and Soderling, T. R. (1998) Nature 396, 584–587) demonstrates that CaM-KK can also activate and phosphorylate protein kinase B (PKB). In this study, we identify a CaM-KK from Caenorhabditis elegans, and comparison of its sequence with the mammalian CaM-KK α and β shows a unique Arg-Pro (RP)-rich insert in their catalytic domains relative to other protein kinases. Deletion of the RP-domain resulted in complete loss of CaM-KIV activation activity and physical interaction of CaM-KK with glutathioneS-transferase-CaM-KIV (T196A). However, CaM-KK autophosphorylation and phosphorylation of a synthetic peptide substrate were normal in the RP-domain mutant. Site-directed mutagenesis of three conserved Arg in the RP- domain of CaM-KK confirmed that these positive charges are important for CaM-KIV activation. The RP- domain deletion mutant also failed to fully activate and phosphorylate CaM-KI, but this mutant was indistinguishable from wild-type CaM-KK for the phosphorylation and activation of PKB. These results indicate that the RP-domain in CaM-KK is critical for recognition of downstream CaM-kinases but not for its catalytic activity (i.e. autophosphorylation) and PKB activation.

Ca 2ϩ /calmodulin-dependent protein kinases (CaM-Ks) 1 constitute a diverse group of enzymes which are involved in many aspects of calcium signaling such as neurotransmitter release, excitation-contraction coupling in muscle, and gene expression (1)(2)(3)(4). Recent studies have demonstrated that two CaM kinases, CaM-KI and -IV, are activated through phosphorylation by an upstream CaM kinase kinase (CaM-KK) (5)(6)(7)(8)(9)(10)(11)(12), analogous to other kinase cascades such as PKA/phosphorylase ki-nase (13), MAP kinase (14), and AMP kinase (15). CaM-KK is a recently cloned protein kinase that phosphorylates and activates CaM-KI and CaM-KIV, constituting the CaM-K cascade (11,39,47). Like other CaM-Ks, CaM-KK is negatively regulated by an intrasteric mechanism through its autoinhibitory domain (residue 436 -441) and activated by the Ca 2ϩ ⅐CaM complex (16). Ca 2ϩ /CaM binding to both CaM-KK and its downstream target CaM-Ks are required to activate the CaM kinase cascade (6,7,12). This CaM-K cascade has been functionally demonstrated for CaM-KIV activation in response to Ca 2ϩ mobilization using transfected COS-7 cells (12), Jurkat cells (17), and cultured hippocampal neurons (18) and for CaM-KI activation in PC-12 cells upon membrane depolarization (19). The CaM-KK/CaM-KIV cascade can stimulate gene transcription through phosphorylation of Ser 133 in cAMP response elementbinding protein, and this may play an important physiological role in learning and memory (11, 18, 20 -23). It has also been demonstrated that the MAP kinase pathways, especially JNK and p38, may be indirectly activated by the CaM-KK/CaM-KIV cascade (24). Recently, evidence has been provided that CaM-KK may mediate the anti-apoptotic effect of modest elevations of Ca 2ϩ through phosphorylation and activation of protein kinase B (PKB) (25). This result also indicates that multiple protein kinases might be phosphorylated and activated by CaM-KK, resulting in regulation of a wide variety of functions.
The activation sites in CaM-KI (Thr 177 ) (7), CaM-KIV (Thr 196 ) (6,12), and PKB (Thr 308 ) (25) which are phosphorylated by CaM-KK are located in their "activation loops" analogous to Thr 197 in PKA (26) and Thr 160 in cdk2 (27). For most protein kinases the substrate recognition determinants are located just NH 2 -and COOH-terminal of the phosphorylated Ser/Thr. For example, many substrates of PKA have basic residues at position 2 and 3 NH 2 -terminal of the phosphorylated Ser, and these basic residues are recognized by Glu 127 and Glu 170 in PKA (33)(34)(35)(36)(37). However, there is little sequence similarities in the NH 2 -terminal sequences of the activation sites of CaM-KI, CaM-KIV, or PKB. There is considerable COOHterminal sequence identity, but these sequences are also conserved in many other kinases that have activation loops but are not phosphorylated by CaM-KK. This suggests that CaM-KK has specific mechanisms to recognize its target kinases other than the primary sequence surrounding the phosphorylated Thr. Although CaM-KK can phosphorylate a synthetic peptide corresponding to the sequence in the CaM-KIV activation loop (KKKEHQVLMKT 196 VCGTPGY), very high concentrations of the peptide are required (28). In this report, we explored the mechanisms of selective substrate recognition of CaM-KK by identifying a unique kinase insert domain which is involved in its interaction and activation of CaM-KI and IV but not PKB.

EXPERIMENTAL PROCEDURES
Materials-CaM-KK cDNA (GenBank accession number L42810) was from a rat brain cDNA library (11). Recombinant CaM-KIV was expressed in Sf9 cells and purified as described previously (21). Recombinant CaM-KKs were expressed in Escherichia coli BL-21 (DE3) pLys(S) with pET16b vector and purified by CaM-Sepharose as described previously (12). CaM was purified from bovine brain (29). CaM-KIV peptide (KKKEHQVLMKT 196 VCGTPGY (28)) was synthesized by the Bio-Synthesis, Inc. Caenorhabditis elegans CaM-KK cDNA was cloned by RT-PCR by using a sense oligonucleotide (5Ј-TTACTC-GAGATGTACACATTTCAGTCGGTCTCACAGCAG-3Ј), and an antisense oligonucleotide (5Ј-AGACTAGTTTACCTGAATGGATTGCCAAA-CCGCTTGCGA-3Ј) based on the sequence of C. elegans cosmid (C05H8.1) and reverse transcribed DNA from mRNA of N2 stage C. elegans as a template, which was kindly provided from Dr. Shouhei Mitani (Tokyo Woman's Medical College). A 1.1-kilobase pair of PCR fragment was digested with XhoI and SpeI and then inserted into pME18s vector. Nucleotide sequence of cloned C. elegans CaM-KK cDNA (GenBank accession number AB016838) was confirmed and the deduced amino acid sequence (357 amino acids) was completely matched with predicted coding sequence using the program Genefinder from C05H8.1 (30). The 5Ј-untranslated region of C. elegans CaM-KK cDNA was also cloned by RT-PCR using a sense primer (5Ј-AAACTT-CTGTAGTATTTACA-3Ј) and antisense primer (5Ј-AAATAGTTGTCA-TTTGGATCG-3Ј) based on the cosmid sequence and reverse transcribed DNA from mRNA of N2 stage C. elegans as a template (see Fig.  1A) and completely sequenced. Rat CaM-KI cDNA was cloned by RT-PCR using a sense oligonucleotide (5Ј-ATGCCAGGGGCAGTGGAAGG-3Ј) and an antisense-oligonucleotide (5Ј-TCAGTCCATGGCCCTAGAG-C-3Ј) based on the published sequence (31) and reverse transcribed cDNA from rat brain mRNA (CLONTECH) as a template. The PCR product (1.1 kilobase pairs) was subcloned into pT7Blue plasmid and then sequenced. CaM-KI cDNA was digested with EcoRI and SalI, and then subcloned into pGEX-4T3 and transformed into E. coli JM109. Expressed GST-CaM-KI was purified on glutathione-Sepharose followed by CaM-Sepharose. GST-CaM-KIV (17-469, T196A) was constructed into pGEX-KG as follows: His-tagged CaM-KIV (T196A) cDNA in pME18s vector (12) was digested with BstEII at the position of Val 17 and then filled in and digested with XbaI. CaM-KIV fragment was ligated into SmaI-XbaI digested pGEX-KG and then transformed into E. coli BL-21 (DE3). Nucleotide sequence of this construct was confirmed. Expressed GST-CaM-KIV was purified by CaM-Sepharose. All other chemicals were from standard commercial sources.
Construction of Plasmids-Mutagenesis of CaM-KK using pME-CaM-KK (wild-type) plasmid as a template was carried out by using either the GeneEditor™ in vitro Site-Directed Mutagenesis System PKB Activation by CaM-KK-GST-rat PKB (␣ isoform) was transiently expressed in COS-7 cells with pME18s vector as described above and purified on glutathione-Sepharose as described previously (25). Purified GST-PKB (0.1 g) was incubated for the indicated times with 50 mM HEPES (pH 7.5), 10 mM Mg(Ac) 2 , 1 mM DTT, 1 mM CaCl 2 , 3 M CaM, 0.2 mg/ml histone H2B, 400 M [␥-32 P]ATP (1000 cpm/pmol), and approximately 100 ng of wild-type, mutant CaM-KK, or buffer. After terminating the reaction by addition of SDS-PAGE sample buffer, 32 P incorporation into GST-PKB and histone H2B was analyzed by SDS-15% PAGE followed by autoradiography and then quantitated by densitometric scanning. The data of GST-PKB activity are expressed as fold of control value of 32  Autophosphorylation of CaM-KK-Autophosphorylation reaction was essentially the same as CaM-KIV peptide phosphorylation assay as described above except for 100 M [␥-32 P]ATP (1000 -2000 cpm/pmol) and omitting peptide substrate in a reaction solution. After terminating the reaction by addition of 5 l of SDS-PAGE sample buffer, samples were loaded onto SDS-10% PAGE and then subjected to autoradiograph.
Binding of CaM-KK with GST-CaM-KIV (T196A)-After 25 l of glutathione-Sepharose (50 l of 50% slurry) was loaded with either buffer or 0.5 g of GST-CaM-KIV (T196A) which was purified by CaM-Sepharose, the resin was washed three times with 1 ml of phosphatebuffered saline and then washed three times with 1 ml of 50 mM HEPES (pH 7.5), 2 mM CaCl 2 , and 10 mM Mg(Ac) 2 . Equal amounts (approximately 200 ng) of either wild-type or mutant CaM-KK were applied to the resin in a solution (100 l) containing 50 mM HEPES (pH 7.5), 2 mM CaCl 2 , 10 mM Mg(Ac) 2 , 5 M CaM, and 1 mM ATP and then incubated for 1 h at room temperature. These resins were washed three times with 500 l of 50 mM HEPES (pH 7.5), 2 mM CaCl 2 , 10 mM Mg(Ac) 2 , 1 M CaM at 4°C and then GST-CaM-KIV was eluted with 80 l of 50 mM Tris-HCl (pH 8.0) and 10 mM glutathione. Equal volumes of samples were subjected to SDS-7.5% PAGE followed by Western blotting by using both anti-CaM-KIV antibody or anti-CaM-KK antibody.
Others-Western blotting was carried out using antiserum (1/1000 dilution) against a peptide corresponding to a conserved protein kinase motif (residues 132-146 of CaM-KII), anti-CaM-KIV antibody (Transduction Laboratories), or anti-CaM-KK antibody (Santa Cruz Biotechnology, Transduction Laboratories), and the biotinylated-CaM overlay was done as described previously (10). Detection was performed by using chemiluminesence reagent (NEN Life Science Products Inc.). Protein concentration was estimated by Coomassie dye binding (Bio-Rad) using bovine serum albumin as a standard.

RESULTS
Identification and Characterization of C. elegans CaM-KK-A protein kinase gene was recently identified in the C. elegans genome data base that is highly homologous to rat CaM-KK (30,47). To determine whether this clone encodes CaM-KK, cDNA of 1074 bp was cloned by RT-PCR from C. elegans mRNA (Fig. 1A). In order to ensure that the methionine at position 1 is the point of translation initiation, we also cloned the 5Ј-untranslated region by RT-PCR using two nucleotide sequences, 200 bp upstream in C. elegans genome sequence and 353 bp downstream from the first Met (ATG), as PCR primers (Fig. 1A, underlined). The nucleotide sequence of the PCR product (553 bp) includes 200 bp of the 5Ј-untranslated region containing three in-frame stop codons and the remaining 353-bp region is completely matched with its open reading frame from the first ATG. Since the antisense primer is located in the third exon, this PCR product is derived from C. elegans mRNA and the Met at position 1 is likely the translation initiation. The full-length clone encoded a protein of 357 amino acids with a calculated molecular mass of 40,701 (Fig. 1A). This protein is likely the C. elegans homologue of CaM-KK as it contains the unique RP-rich insert (residues 60-81) in the catalytic domain highly homologous to similar inserts in the ␣ (11) and ␤ (39, 47) isoforms of mammalian CaM-KK. The C. elegans clone was transiently expressed in COS-7 cells, and a CaM overlay in the presence of Ca 2ϩ (Fig. 1B, inset panel) identified a 45-kDa protein which is smaller than the mammalian ␣CaM-KK. COS-7 cell extracts from mock transfected or transfections with vectors encoding mammalian ␣CaM-KK or C. elegans protein were tested for their abilities to activate recombinant mammalian CaM-KIV. The C. elegans protein gave activation of CaM-KIV which was about 2-3-fold less efficient than the mammalian CaM-KK in the initial rate of activation (Fig. 1B). Next we tested Ca 2ϩ /CaM dependence of the C. elegans enzyme by using both wild-type CaM-KIV (Fig. 1C, left panel) and constitutively active mutant of CaM-KIV ( 316 FN-DD, Fig. 1C, right panel) (10) which has shown to be activated by the con-stitutively active form of rat ␣CaM-KK (1-434) in a Ca 2ϩ /CaMindependent manner (16). As shown in Fig. 1C, both wild-type and mutant CaM-KIVs are activated by the C. elegans enzyme in a complete Ca 2ϩ /CaM-dependent manner suggesting that C. elegans enzyme requires Ca 2ϩ /CaM for its activity as well as the mammalian CaM-KK (11). Furthermore, the COOH-terminal end of the C. elegans clone (residues 331-356, Fig. 1A) is 54% identical to the CaM-binding domain of ␣CaM-KK (16). We have also detected Ca 2ϩ -dependent CaM binding of the syn- thetic peptide corresponding to residues 331-356 in the C. elegans clone (data not shown) similar to that of a CaM-binding peptide of ␣CaM-KK (16). These biochemical criteria confirm that the protein is the C. elegans homologue of CaM-KK.
Acidic Residues Are Not Essential for CaM-KK Substrate Recognition-The mechanism of substrate recognition by CaM-KK has not been studied. To determine critical residues or domains for substrate recognition in CaM-KK, we first aligned and compared the amino acid sequences of the catalytic domains of mammalian ␣ and ␤ CaM-KK and C. elegans CaM-KK with mammalian CaM-KII and cAMP-dependent protein kinase (PKA), the best protein kinase for enzyme-substrate recognition mechanisms ( Fig. 2A). According to the studies of: 1) the crystal structure of PKA with PKI (33, 34); 2) charged to alanine mutagenesis of the yeast homologue of PKA (35); and 3) labeling with a water-soluble carbodiimide (36), acidic residues Glu 127 and Glu 170 interact with the Arg in the P-2 and P-3 positions of PKA substrates. It has been shown that mutation of these Glu residues in PKA affect its K m with the peptide substrate Kemptide (35,37). Residues equivalent to Glu 127 and Glu 170 are also conserved in Arg/Lys-directed CaM kinases (myosin light chain kinase, CaM-KI, -II, -IV, and phosphorylase kinase). Residues Pro 237 and Ser 279 in rat CaM-KK are analogous to Glu 127 and Glu 170 in PKA ( Fig. 2A). Therefore, we initially mutated Pro 237 to Glu, Val, or Ala and Ser 279 to Glu, Asp, or Ala to test whether these residues are involved in CaM-KK activity. Mutant rat constructs, including wild-type ␣CaM-KK, were transfected into COS-7 cells, and the expression level of CaM-KK in the cells were quantitated by immunoblotting (Fig. 2, B and C, inset panels). All of the Pro 237 and Ser 279 mutants were expressed to an extent similar to wildtype enzyme. The same amount of CaM-KK mutants as wildtype enzyme were used for CaM-KIV activating assays in the presence of Ca 2ϩ /CaM. The time course of CaM-KIV activation by all the Pro 237 and Ser 279 CaM-KK mutants were similar to wild-type CaM-KK (t1 ⁄2 ϭ 1 min) (Fig. 2, B and C). Our results indicate that Pro 237 and Ser 279 are not essential for specific substrate recognition of CaM-KK.
The RP-domain Is Essential for CaM-KIV Activation by CaM-KK-The function of the unique RP-domain insert in all cloned CaM-KKs ( Fig. 2A) is completely unknown. To analyze its function, the RP-domain (residues 167-189) was deleted from ␣CaM-KK, and this mutant was expressed in COS-7 cells (Fig. 3A). As expected, deletion of the RP-domain from CaM-KK gave a slightly faster migrating species than wild-type enzyme (68 kDa) on SDS-PAGE. Surprisingly, the RP-deletion mutant was completely unable to activate CaM-KIV (Fig. 3A). The critical role of the RP-domain was further analyzed by sitedirected mutation of Arg 172 and Arg 177 , which are conserved among all three CaM-KKs. Mutation of both Arg 172 and Arg 177 to Glu resulted in significant reduction of CaM-KK activity. Ala mutation was less defective (t1 ⁄2 ϭ 4 min) compared with either RP-domain deletion or Glu mutants (Fig. 3A). Single Glu mutation of either Arg 172 and Arg 177 revealed that both residues are required for the maximum activity of CaM-KK (Fig. 3B). Although mutation of Arg 172 has a more significantly affect on the catalysis than that of Arg 177 , both residues seem to synergistically contribute to CaM-KK activity. Another Arg residue at 173 is also conserved in ␣ and ␤ isofoms of CaM-KK, but replaced by Gln 66 in C.elegans enzyme ( Fig. 2A). Since C. elegans CaM-KK is a 2-3-fold less efficient activator than rat CaM-KK toward mammalian CaM-KIV (Fig. 1B), we tested the possibility that Arg 173 contributes to the catalytic efficiency (Fig. 3C). Glu mutation on Arg 173 gave a significant reduction of the initial rate of CaM-KIV activation. However, when we mutated Gln 66 in C. elegans CaM-KK to Arg, we could not detect a significant increase in CaM-KK activity (data not shown). This is consistent with the mutation of Arg 173 in rat CaM-KK to a neutral amino acid residue, such as Ala, which caused little effect on CaM-KK activity.
Next we tested whether the RP-deletion mutant could physically interact with CaM-KIV using GST-CaM-KIV (T196A). The rational for using the T196A mutant, which cannot be activated by CaM-KK, is that it may capture the intermediate complex form between CaM-KK and CaM-KIV. GST-CaM-KIV (T196A) loaded glutathione-Sepharose was mixed with either the wild-type or RP-deletion mutant of CaM-KK, and a pulldown experiment was performed in essentially the same condition as an activation reaction with low ionic strength. As shown in Fig. 3D, association of wild-type CaM-KK with GST-CaM-KIV (T196A) was readily detected which was approximately 10% of applied CaM-KK, and the interaction occurred in a Ca 2ϩ /CaM-dependent manner (data not shown). Deletion of the RP-domain resulted in a loss of interaction to CaM-KIV compared with the wild-type CaM-KK.

RP-domain Mutants of CaM-KK Are
Catalytically Active-To test whether these RP-domain mutants are catalytically defective, we measured their abilities to autophosphorylate. As shown in Fig. 4A, all of the RP-domain mutants showed Ca 2ϩ / calmodulin-dependent autophosphorylation similar to wildtype kinase although the R172A,R177A mutant gave a slightly weaker activity as compared with others. As a second test of catalysis, we determined their abilities to phosphorylate the CaM-KIV activation domain sequence (KKKEHQVLMKT-196 VCGTPGY) (28) which contains the phosphorylation-activation Thr 196 in CaM-KIV. Again, all of the mutants including the RP-domain deletion could phosphorylate CaM-KIV peptide in a Ca 2ϩ /CaM-dependent manner (Fig. 4B). In these experiments, we used recombinant mutant enzymes expressed in E. coli and partially purified by CaM-Sepharose instead of using transfected COS-7 extract. CaM-KIV activating and phosphorylating activity of both wild-type and RP-domain deletion mutants of E. coli expressed CaM-KK was shown to be essentially the same as COS-7 cell expressed enzyme (see Fig.  5B). We also checked CaM-KIV activating activity of other E. coli-expressed CaM-KK mutants with the same conditions as shown in Fig. 3A with triplicate experiments and obtained essentially the same results as that obtained with COS-7 cellexpressed enzymes (data not shown). These results indicate that deletion and mutations on the RP-domain do not affect catalytic activity with regard to autophosphorylation and peptide phosphorylation. Also, the regulatory mechanism of CaM-KK, such as autoinhibition and CaM binding, remained intact with these mutations and deletions.
Requirements of the RP-domain for Activation of CaM-Kinases but Not for PKB-Although the RP-domain mutants can catalyze Ca 2ϩ /calmodulin-dependent autophosphorylation and phosphorylation of a synthetic peptide, we wanted to test other physiological substrates. Two such substrates are CaM-KI and PKB. CaM-KI is phosphorylated on Thr 177 resulting in its activation (7). As shown in Fig. 5A, recombinant GST-CaM-KI was activated and phosphorylated by wild-type CaM-KK as described previously (11). However, activation and phosphorylation of GST-CaM-KI by the RP-deletion mutant of CaM-KK was strongly reduced compared with wild-type (Fig. 5A) in a manner similar to CaM-KIV (Fig. 5B). PKB can be activated through phosphorylation of its activation loop Thr 308 by PDK1 (40) and CaM-KK (25). In contrast to CaM-KI and CaM-KIV, PKB was activated and phosphorylated by the RP-deletion mutant of CaM-KK in a manner comparable to that obtained with wild-type CaM-KK (Fig. 5C) while the rate of activation of PKB by CaM-KK is lower than that of CaM-KI or CaM-KIV which is consistent with previous observations (25). This indicates that the RP-domain deletion mutant is catalytically active and recognizes PKB as its substrate. DISCUSSION In this study we demonstrated that the RP-domain of CaM-KK, which is conserved in all three known CaM-KKs, is required for selective substrate recognition. Thus, deletion of the RP-domain obviated CaM-KK activity toward its two classical substrates CaM-KI and CaM-KIV. Mutational analysis indicated that the three highly conserved Arg in rat CaM-KK, Arg 172 , Arg 173 , and Arg 177 are essential in this substrate recognition. In contrast, the RP-deletion mutants showed normal autophosphorylation, phosphorylation of a synthetic substrate containing the activation site of CaM-KIV, and, more impor-tantly, in phosphorylation and activation of PKB. Since these reactions were Ca 2ϩ /CaM-dependent, these results indicate that the RP-domain is involved in recognition of CaM-KI and CaM-KIV as substrates. This conclusion is substantiated by the fact that the GST fusion of CaM-KIV could pull down wild-type CaM-KK whereas the RP-domain deletion mutant was not able to interact with the GST fusion of CaM-KIV. The fact that RP-domain mutants could phosphorylate CaM-KIV peptide but not CaM-KIV itself is perhaps consistent with the report (28) that the CaM-KIV peptide (K m ϭ 263 M) is a very poor substrate for CaM-KK as compared with CaM-KIV (K m ϭ 0.71 M). Therefore, the recognition determinants of CaM-KK toward this peptide substrate are likely to be quite different than for the full-length CaM-KIV. The RP-domain of CaM-KK is localized between subdomains II and III in the catalytic domain, placing it between ␣-helices B and C in PKA. It is conceivable that this domain may recruit and stabilize the downstream CaM-KI and -IV to maintain proper orientation toward the catalytic cleft, resulting in high affinity and specific interaction between CaM-KK and these substrates. Interestingly, we could not detect significant interaction between CaM-KIV and a synthetic peptide corresponding to the RP-domain by using a direct binding assay or an inhibition assay for CaM-KIV activation, suggesting that domain addition to the RD-domain might be required for the interaction (data not shown). Since the RP-domain of CaM-KK is not necessary for phosphorylation and activation of PKB (Fig. 5C), the RP-domain seems to be specifically utilized to operate the CaMkinase cascade. This is also consistent with a recent report that the PKB activating kinase PDK1, which does not possess the RP-domain, was unable to phosphorylate CaM-KIV (42). In vitro activation of PKB requires about 10-fold higher concentrations of CaM-KK than does activation of CaM-KIV or CaM-KI (25). Likewise, in NG108 cells stimulated by NMDA agonists, activation of PKB peaks at about 1 h (25) whereas activation of CaM-KIV is maximal at about 2-5 min. 2 These results suggest that different substrate specificity determinants may be utilized by CaM-KK for PKB activation, consistent with the difference in activation of PKB versus CaM-KIV/ CaM-KI by the RP-deletion mutant. Therefore, the RP-deletion mutant of CaM-KK may serve as a useful tool for evaluating the function of Ca 2ϩ -dependent PKB activation and regulation of apoptosis without concurrent activation of CaM-KI and -IV.
Crystallographic studies of PKA with PKI (33,34) and recent studies of several protein kinases using an oriented degenerate peptide library (41) clearly demonstrated that critical determinants for substrate recognition by those protein kinases were found at residues in the vicinity of the phosphorylation site. Two acidic residues in PKA (Glu 127 and Glu 170 ) are predicted to form interactions with the P-3 basic residue in substrates. However, there is no obvious consensus basic residues at those positions in the activation loops of CaM-KI, -IV, and PKB, and CaM-KK has a Pro 237 and Ser 279 in positions analogous to Glu 127 and Glu 170 of PKA. Not surprisingly, mutations of Pro 237 and Ser 279 did not affect the ability of CaM-KK to phosphorylate and activate CaM-KIV.
In conclusion, our data contribute to the understanding of how CaM-KK specifically recognizes and phosphorylates its downstream CaM kinases. CaM-KK contains a RP-domain which allows efficient targeting and subsequent activation of CaM-KI and -IV. Based on a limited number of mutants, we cannot rule out the possibility that other residues or regions are involved in substrate recognition by CaM-KK. In this regard, our results are consistent with a study showing that mutations in the activation loop of MAP kinases has little effect on their recognition as substrates for their upstream activating MAP kinase kinases (46).