Peptide Specificity Determinants at P−7 and P−6 Enhance the Catalytic Efficiency of Ca2+/Calmodulin-dependent Protein Kinase I in the Absence of Activation Loop Phosphorylation*

Phosphorylation of Ca2+/calmodulin-dependent protein kinase I (CaM KI) at Thr-177 by recombinant rat Ca2+/calmodulin-dependent kinase kinase B (CaM KKB) modulates the kinetics of synapsin-(4–13) peptide phosphorylation by reducing the K m 44-fold and decreasing theK CaM 4-fold. There is also a slight decrease inK m for ATP and increase in enzymeV max. A synthetic peptide substrate from the yeast transcription factor, ADR1-(222–234)G233 is a 15-fold better substrate for the Thr-177 dephospho-form of CaM KI than synapsin-(4–13). The Thr-177 dephospho-enzyme has aK m and V max for ADR1-(222–234)G233 similar to the values with synapsin-(4–13) using the Thr-177 phosphorylated enzyme. Likewise, with ADR1-(222–234)G233 as substrate, phosphorylation of Thr-177 or substitution of T177A had very little effect on the kinetic values. Using chimeric peptides between synapsin-(4–13) and ADR1-(222–234)G233 we found that N-terminal basic residues at P−7 and P−6 positions were sufficient to allow efficient phosphorylation by the Thr-177 dephospho-form of CaM KI. Phosphorylation of Thr-177 expands the substrate specificity of CaM KI and is not merely an “on-off” switch for kinase activity.

Protein phosphorylation plays a regulatory role in signal transduction for many physiological processes in eukaryotic cells. Frequently, both receptor and nonreceptor kinases not only phosphorylate cellular proteins to elicit biological responses, but are also phosphoproteins themselves. Phosphorylation of kinases by either autophosphorylation in response to ligand binding, in the case of receptor kinases, or by an upstream kinase, in the case of nonreceptor kinases, is an important step in the signal transduction cascade. The regulatory phosphorylation event(s) usually occur on a single (or multiple) residue(s) located in a region termed the activation loop between kinase subdomains VII and VIII.
Virtually every kinase family, including the mitogen-activated protein kinases (1)(2)(3)(4), CDKs 1 (5), protein kinase Cs (6 -10), Janus kinases (11)(12)(13), nonreceptor tyrosine kinases such as the Src kinases (14), and receptor tyrosine kinases such as the insulin (15,16) and Trk (17)(18)(19)(20) receptors, contains members regulated by activation loop phosphorylation. In all of these cases, activation loop phosphorylation markedly activates the protein kinase, and the phosphorylated form of the kinase is thought to be physiologically relevant. For example, mutation of Tyr-1007 to Ala in the Janus 2 kinase eliminated activity, and the mutant could not restore erythropoietin signaling in cells lacking Janus 2 kinase (12). A second example of the biological significance of activation loop phosphorylation is the phosphorylation of I-B kinase ␣ by the NF-B-inducing kinase. A mutated I-B kinase ␣ in which the activating Ser-176 was changed to Ala acts as a dominant negative inhibitor of interleukin-1 and tumor necrosis factor-induced NF-B activation (21). These data are illustrative of the biological importance of phosphorylatable residues that reside in the activation loop of kinases.
The most extensively characterized kinase that is regulated by activation loop phosphorylation is the cAMP-dependent protein kinase, PKA. In the crystal structure of PKA, Thr-197 is phosphorylated, and the phosphate makes electrostatic contacts with the small and large lobes of the kinase. These contacts were proposed to facilitate both ATP and peptide substrate binding (22)(23)(24). Biochemical analysis of PKA confirmed that the phosphate of Thr-197 is required for both ATP binding as well as efficient phosphate transfer to the peptide substrate (25,26). When PKA is expressed in bacteria, Thr-197 phosphorylation occurs by autophosphorylation, but in mammalian cells Thr-197 may be phosphorylated by another protein kinase. Phosphoinositide-dependent protein kinase 1 can phosphorylate and activate PKA in vitro (27) favoring the idea that PKA is a downstream target of a protein kinase cascade.
Two members of the Ca 2ϩ /calmodulin (CaM)-dependent family of protein kinases, CaM KI and IV, are activated by upstream kinases (28,29). Activation loop phosphorylation of CaM KI and CaM KIV by CaM kinase kinase A and CaM KKB, both of which are also members of the CaM kinase family, increases enzyme activity 10 -50-fold (30). Both CaM KKs contain phosphorylatable residues within their activation loops and also may be targets of upstream kinases, although to date, no kinases have been identified. Activation loop phosphorylation of CaM KI and IV has been shown to occur in intact cells. Aletta et al. (31) has shown that elevated intracellular Ca 2ϩ concentrations, resulting from either ionomycin or KCl treatment of PC12 cells, lead to phosphorylation of CaM KI on Thr-177. CaM KI immunoprecipitated from cells treated with either agent was significantly more active than CaM KI immunoprecipitated from unstimulated cells and showed decreased responsiveness to purified CaM KK in vitro. In addition, anti-IgM treatment of the B cell line, BJAB (32), as well as T cell receptor stimulation of Jurkat cells (33) have been shown to activate CaM KIV. CaM KIV induction of AP-1 (34) and cAMP response element-binding protein-mediated transcription (35) is inhibited by a T200A mutation in human CaM KIV. These data provide compelling evidence that phosphorylation of CaM KI and IV is a physiological response to signals that increase intracellular Ca 2ϩ with these kinases serving as intermediates in signal transduction cascades.
Recent data from Chin et al. (36) revealed that Thr-177dephosphorylated CaM KI has one of the highest specific activities reported for any protein kinase, and peptide substrates could show greater than 400-fold differences in specificities. These surprising observations raised the possibility that activation loop phosphorylation of CaM KI may not be essential for kinase activity but could modulate substrate specificity. Here we demonstrate that phosphorylation is not an absolute requirement for CaM KI activity but, indeed, serves to expand the peptide substrate specificity of the enzyme. We have also found that CaM KIV is regulated in a similar manner. CaM KI and IV, therefore, represent the first kinases whose peptide substrate specificity has been shown to be regulated by activation loop phosphorylation and raises the possibility that such control may be extended to members of other protein kinase families.
Protein Expression and Purification-The human CaM KI clone was isolated from a differentiated HL-60 cDNA library as described previously (28). The T177A mutant was generated by PCR using a 5Јoligonucleotide primer containing the appropriate base change. The PCR product was digested with SmaI and EcoRI, gel purified, and used to replace the wild-type sequence in the CaM KI clone. The entire PCR fragment was sequenced to ensure no secondary mutations had occurred (28). BL21 (DE3) Escherichia coli strains were transformed with the appropriate CaM KI clones. Overnight starter cultures were used to inoculate 1-liter large scale cultures. These were grown to an A 600 of 0.7 at 37°C; isopropyl-1-thio-␤-D-galactopyranoside was then added to a final concentration of 0.4 mM, and the cultures were allowed to grow for another 2-3 h to induce CaM KI expression. Cells were then harvested by centrifugation at low speed at 4°C for 10 min. Cells were resuspended in phosphate-buffered saline containing 20 g/ml trypsin inhibitor, 10 g/ml aprotinin, 10 g/ml leupeptin, 2 g/ml pepstatin, 100 g/ml Pefabloc, and 100 M phenylmethylsulfonyl fluoride and lysed by mild sonication at 4°C. The cell debris was collected by centrifugation at 10,000 ϫ g for 30 min. The supernatants were loaded onto glutathione-Sepharose columns, extensively washed with phosphate-buffered saline, and eluted with 10 mM glutathione, 50 mM Tris-HCl buffer, pH 8.0. Purity of enzyme preparations was analyzed by SDS-polyacrylamide gel electrophoresis. Enzymes were aliquoted in 40% glycerol, snap-frozen in liquid nitrogen, and stored at Ϫ70°C. Enzymes were thawed and used once only.
The rat CaM KKB gene was isolated and subcloned into the pMAL-c2X vector (New England Biolabs) as described (35). The KKB 1-487 mutant was generated by by PCR by placing a stop codon immediately after amino acid residue 487. The PCR product was subcloned into the pMAL-c2X vector and sequenced in its entirety. Protein was induced as above. Cells were resuspended in 10 mM phosphate, 30 mM NaCl, 0.25% Tween 20, 10 mM 2-mercaptoethanol, 10 mM EDTA, and 10 mM EGTA. Lysates were loaded onto an amylose column and washed with 12-15 column volumes of buffer containing 10 mM phosphate, 0.5 M NaCl, 0.25% Tween 20, 10 mM 2-mercaptoethanol, 1 mM EGTA. The fusion protein was eluted with 10 mM maltose in 50 mM Tris, pH 7.6. The MBP-KKB was analyzed and handled as described for CaM KI.
Protein Determinations-The concentrations of KKB and CaM KI were determined using the Lowry assay with some modifications (38). Bovine serum albumin was used as the standard. Bovine CaM concentrations were determined by amino acid analysis.
Peptide Synthesis-Synapsin- (4 -13) and the alcohol dehydrogenase repressor peptides (ADR1)-(222-234)G233 peptides were synthesized by Dr. David Klapper at the University of North Carolina, Chapel Hill and quantified by amino acid analysis. (Dr. William Abrams, University of Pennsylvania School of Dental Medicine). All other peptides were quantified by phenylalanine absorption at 257 nm and standardized to the known concentrations of synapsin-(4-13) and ADR1-(222-234)G233.
Peptide Kinase Assays-CaM KI or IV activity was assayed in a buffer containing 50 mM Tris, pH 7.6, 10 mM MgCl 2 , 2 mM CaCl 2 , 0.1% Tween 20, 1 mM dithiothreitol, 0.5 mg/ml bovine serum albumin, and varying concentrations of CaM, ATP, and peptide substrate in a volume of 40 -50 l. Assays were initiated by the addition of enzyme or, in the cases of pre-incubation, by the addition of [␥-32 P]ATP (10 mCi/mmol) and substrate. Pre-incubation and assay times varied and are described in the figure legends. Reactions were terminated by spotting the mixture on P81 phosphocellulose filter papers (39). P81 filters were washed 3 times for 1 h in either 75 mM phosphoric acid for peptide kinase assays or in 20% trichloroacidic acid ( Fig. 1), rinsed in 100% acetone, and air dried. Incorporation of 32 P i into peptide or enzyme was quantified using a Beckman LS 6000 scintillation spectrometer. For kinetic analyses the 5 or 6 ATP concentrations ranged from 400 to 5000 M. ADR1-(222-234)G233, peptides containing variations of ADR1, and the LKK and LRR-synapsin-(4 -13) peptides were assayed at 5 concentrations ranging from 50 to 1000 M for 10 min with 1 ng of Thr-177 dephospho-CaM KI or for 5 min with 1 ng of Thr-177 phospho-CaM KI. The synapsin-(4 -13) peptide and variants other than the LKK and LRR peptides were assayed at 5 concentrations ranging from 50 to 1200 M for 15 min with 1 ng of Thr-177 dephospho-CaM KI and for 5 min with 1 ng of Thr-177 phospho-CaM KI. The CaM concentration for all kinetic studies was 1 M. KKB did not phosphorylate any of the peptides used for these studies. Kinetic constants were determined by using the KaleidaGraph program. The general curve fit option of this program was used to apply Michaelis-Menten equations to directly fit the raw data using the Levenberg-Marquardt algorithm.
Gel Quantitation of Phosphorylated Peptides-Peptides (200 M) were phosphorylated in the standard assay buffer containing 500 M ATP for 5 min using 5 ng of Thr-177 dephospho-CaM KI. Reactions were stopped by the addition of protein sample buffer to a final concentration of 2X 100 mM Tris, pH 6.8, 4% SDS, 0.2% bromphenol blue, and 20% glycerol. An eighth of the reaction mixture was electrophoresed on 20% Laemmli gels in a buffer containing 25 mM Tris, 200 mM tricine, and 0.1% SDS. After electrophoresis, gels were dried and exposed to film. The radioactive bands were carefully excised and counted by liquid scintillation spectrometry.
Anion Exchange Chromatography-The procedure was similar to Kemp et al. (40). Bio-Rad anion exchange resin AG1 ϫ 8 (acetate form) was equilibrated in water, and the buffer was then changed to 30% acetic acid. Prior to the experiments, the ATP binding capacity of the resin was monitored by absorbance at 259 nm so Ͼ95% of the ATP bound. Chimeric peptides were phosphorylated by 5 ng of CaM KI for 5 min in the presence of 500 M ATP and 5 Ci of [␥-32 P]ATP in a volume of 40 l. Reactions were stopped by adding 360 l of 30% acetic acid. The reactions were loaded onto individual AG1 ϫ 8 columns at 4°C. The flow through and 6-ml 30% acetic acid washes were counted by a Beckman LS 6000 scintillation spectrometer in the 3 H channel (Cherenkov radiation).

Mechanism of Activation of CaM KI by KKB-CaM
KI has been shown to be activated by phosphorylation on Thr-177, a residue residing in the activation loop (28), by two distinct Ca 2ϩ /CaM-dependent protein kinases. The smaller kinase (64 kDa), designated CaM kinase kinase A, was identical to the reported CaM KIV kinase (41), whereas CaM KKB (68 kDa) was a novel CaM KK (30,42). We have isolated a cDNA encod-ing CaM KKB from both human and rat brain libraries, expressed the rat kinase as a maltose-binding fusion protein, and examined its properties (35). Here we asked whether recombinant CaM KKB can phosphorylate CaM KI as does its tissuepurified counterpart (28,30). Recombinant CaM KKB catalyzes a time-dependent phosphorylation of CaM KI to approximately 0.8 mol of phosphate/mol of enzyme (Fig. 1). The extent of phosphate incorporation at Thr-177 directly correlates with the degree of CaM KI kinase activity. 2 In the presence of Ca 2ϩ / CaM, KKB also autophosphorylates to a stoichiometry of 0.2 mol of phosphate/mol of enzyme ( Fig. 1), but this autophosphorylation does not alter KKB activity (35).
To determine the effect of Thr-177 phosphorylation on the kinetics of peptide phosphorylation, we stoichiometrically phosphorylated CaM KI using recombinant KKB and examined changes in its ability to phosphorylate the most extensively studied CaM KI peptide substrate, synapsin-(4 -13). This peptide, LRRRLS 9 DANF, is derived from the sequence surrounding site 1, LRRRLS 9 DSNF, of synapsin. We determined that Thr-177 dephospho-CaM KI has a K m for the synapsin-(4 -13) peptide of 209 M and a V max of 26.1 mol/min/mg (Table I). After CaM KI phosphorylation by recombinant KKB, the K m for synapsin-(4 -13) decreases 44-fold (209 to 4.7 M) and the V max increases to 65.5 mol/min/mg. After Thr-177 phosphorylation, the catalytic efficiency of the enzyme, as indicated by V max /K m , increases 107-fold.
Phosphorylation within the autoinhibitory and CaM binding domains of Ca 2ϩ /calmodulin-dependent protein kinase II (43) and myosin light chain kinase (MLCK) (44) is known to affect enzyme activation by CaM. To address whether activation loop phosphorylation of CaM KI affects its sensitivity to CaM, we determined the concentration of CaM required to produce 50% maximal enzyme activity (K CaM ) with Thr-177 dephospho-and phospho-CaM KI. CaM KI (10 ng) was phosphorylated for 20 min by 400 ng of 1-487 KKB in the presence of 20 nM CaM. The 1-487 KKB is a constitutively active truncation mutant from which the C-terminal autoinhibitory and CaM binding domains were removed. Thus, the CaM included in the preincubation mixture served only to bind CaM KI and expose Thr-177 to KKB (28). After the 20 min preincubation, the reaction mixture was diluted 200-fold in the presence of EGTA. This effectively decreased the concentration of CaM. The CaM activation curve was then generated by assaying CaM KI activity toward synapsin- (4 -13) in the presence of 2 mM CaCl 2 and CaM concentrations ranging from 100 pM to 1 M (Fig. 2). In this experiment, representative of five, the K CaM for Thr-177 dephospho-CaM KI was 14.0 Ϯ 0.95 nM. The CaM curve for Thr-177 phospho-CaM KI was left-shifted from that of the Thr-177 dephospho-CaM KI and the K CaM 4-fold lower or 3.7 Ϯ 0.39 nM. Thus, Thr-177 phosphorylation causes a decrease not only in the peptide K m but also in the K CaM of CaM KI.
Substrate Specificity of Dephospho-CaM KI-ADR1 is phosphorylated by PKA in Saccharomyces cerevisiae (45). A peptide surrounding the PKA phosphorylation site has also been shown to be a good substrate for Ca 2ϩ /calmodulin-dependent protein kinase II, CaM KIV (45), and CaM KI (36). We have tested 10 variants of ADR1 and found that substitutions at positions within the synapsin-(4 -13) peptide, which were shown to be critical or dispensable for phosphorylated CaM KI by Lee et al. (46), showed similar importance in the context of ADR1 using unphosphorylated CaM KI. 2 For example, changing Arg at Ϫ3 to Leu completely abolished the ability of CaM KI to phosphorylate the ADR1-(222-234) peptide, a result consistent with the observation of Lee et al. (46) that substitution of Arg at Ϫ3 in synapsin-(4 -13) caused a 240-fold decrease in V max /K m . On the other hand, changes in PϪ2 and PϪ1 had only very modest effects on CaM KI activity. The best variant of ADR1-(222-234), the ADR1-(222-234)G233 (substitution of Ala with Gly at the Pϩ3 position) was an approximately 14-fold better substrate at a concentration of 100 M than was the synapsin-(4 -13) peptide. Amino acid requirements at the Pϩ3 position have not been analyzed for CaM KI previously, and because substitutions at this position can change CaM KI activity at least 17-fold, it appears to be very important in substrate recognition. 2 The best peptide, ADR1-(222-234)G233, was chosen for further studies.
Efficient Peptide Phosphorylation in the Absence of Activation Loop Phosphorylation-Because ADR1 was such an effective substrate for Thr-177 dephospho-CaM KI, we were interested in determining whether phosphorylation of CaM KI on Thr-177 would further increase enzyme activity. Our control substrate for these studies was the synapsin-(4 -13) peptide. Phosphorylation of CaM KI by KKB causes a large increase in specific activity toward synapsin-(4 -13) (Fig. 3A). The activity of CaM KI T177A (a mutant with the activation loop Thr changed to Ala) is not enhanced by KKB. Thus, the activity of CaM KI toward the synapsin-(4 -13) peptide is dependent on Thr-177 phosphorylation. When using the ADR1-(222-234)G233 peptide as the substrate, however, phosphorylation of wild type CaM KI by KKB results in only a 25% increase in specific activity (Fig. 3A). In addition, the T177A mutant, in the presence or absence of KKB, shows similar specific activities to the native CaM KI. Hence, KKB causes only a small increase in CaM KI activity, which is independent of Thr-177 phosphorylation. This was not due to phosphorylation of ADR1-(222-234)G233 by KKB because no peptide tested with recombinant KKB was appreciably phosphorylated. 2 Thus, these data suggest that the activity of CaM KI toward synapsin-(4-13) depends on Thr-177 phosphorylation, whereas the activity toward ADR1-(222-234)G233 is largely Thr-177 phosphorylation-independent.
Because CaM KIV is also regulated by activation loop phosphorylation, we tested if this kinase, like CaM KI, could efficiently phosphorylate ADR1-(222-234)G233 without Thr-196 being phosphorylated. In Fig. 3B synthase sequence, which has previously been used to assay CaM KIV activation (29). CaM KIV has very low activity toward GS-(1-10)A9,10 that is enhanced 7-fold by KKB. When using ADR1-(222-234)G233, however, the specific activity was 13 times that of GS-(1-10)A9,10, and phosphorylation of CaM KIV by KKB causes only a 50% increase in activity. Thus, CaM KIV is a second example where activation loop phosphorylation serves to expand its substrate specificity range.
Previously, Lee et al. (46) had shown for phospho-CaM KI that substitutions at PϪ2 and PϪ4 had minimal effects on CaM KI activity. Accordingly, we chose to focus on the N terminus of the peptides and added LKK at positions PϪ8 through PϪ6 onto synapsin-(4 -13). As seen in Fig. 4, either LKK or LRR at these positions transforms synapsin-(4 -13) into a substrate as good as ADR1-(222-234)G233. Moreover, Thr-177 phosphorylation no longer has a dramatic effect on CaM KI kinetics if the LKK or LRR-synapsin-(4 -13) peptides are used (Table I).
These data indicate that a peptide containing an aliphatic hydrophobic amino acid at PϪ8 with two basic residues at PϪ7 and PϪ6 is an excellent substrate for Thr-177 dephospho-CaM KI.
Kinetic analysis with the synapsin-(4 -13) peptide revealed that activation involves primarily a decrease in the peptide K m . We determined that the K m peptide values for ADR1-(222-234)G233 and the LKK and LRR-synapsin-(4 -13) chimeric peptides using dephospho-CaM KI were similar to values seen with Thr-177 phospho-CaM KI using the synapsin-(4 -13) substrate (Table I). Thr-177 dephospho-CaM KI has a K m for ADR1-(222-234)G233 of 17.4 M and it decreases to 7.1 M after activation. As seen with synapsin-(4 -13), the V max for ADR1-(222-234)G233 increases a similar magnitude from 39.4 mol/min/mg for the dephospho-kinase to 87 mol/min/mg for the Thr-177 phospho-CaM KI. The K m for ATP when using either synapsin-(4 -13) or ADR1-(222-234)G233 decreases approximately 40% after activation (110 to 68 M). Thus, the V max and K m ATP is increased or decreased, respectively, independent of the peptide substrate. These results indicate that the primary effect of CaM KI activation is to lower the K m for peptide substrate. ADR1-(222-234)G233 is a much better substrate for Thr-177 dephospho-CaM KI than synapsin-(4 -13) because of its low K m . The kinetic data obtained using the chimeric peptides also support this conclusion. The addition of LKK at PϪ8 through PϪ6 in synapsin-(4 -13) creates a substrate with a K m of 6.7 and V max of 40.4 mol/min/mg, a K m lower and V max even higher than ADR1-(222-234)G233. Remarkably, the catalytic efficiency (V max /K m ) of dephospho-CaM KI for synapsin-(4 -13) increases over 45-fold by the addition of LKK at PϪ8 through PϪ6. Phosphorylation of CaM KI results in an increase in V max /K m using synapsin-(4 -13) of 107-fold, whereas the V max /K m of LKK-synapsin-(4 -13) increases a mere 3.6-fold. Thus, the addition of LKK onto the synapsin-(4 -13) peptide creates a substrate that is largely independent of acti-vation loop phosphorylation in that it is 30-fold less responsive to CaM KI Thr-177 phosphorylation.
Because the substitution of LKK at positions PϪ8 through PϪ6 in the synapsin-(4 -13) peptide markedly increased catalytic efficiency in the absence of Thr-177 phosphorylation, we tested the effect of Ala substitutions at these positions. We found that the AAA and LAA variants did not bind to P81 phosphocellulose filters (Fig. 5A). This result was surprising because previous work had demonstrated that two basic residues with a free amino group within a peptide were sufficient for P81 phosphocellulose filter binding (48,49). CaM KI activity toward the peptides was quantified following electrophoresis in 20% Laemmli gels, which were immediately dried and visualized by autoradiography. The bands were excised and counted by liquid scintillation spectrometry. The results are shown in Fig. 5B. The AAALTRRASFSGQ peptide phosphorylation is 50% greater than the synapsin-(4 -13) peptide. The LAALTRRASFSGQ peptide is only a slightly better substrate than the AAALTRRASFSGQ peptide, and the AKKLTRRASF-SGQ peptide is phosphorylated as well as the ADR1-(222-234)G233 peptide. These results indicate that Leu at PϪ8 is not a critical determinant for dephospho-CaM KI activity. Individual mutation of each Lys indicates that both PϪ7 and PϪ6 contribute to the efficiency of phosphorylation by dephospho-CaM KI, with Lys at PϪ7 having the dominant effect. The results obtained in these experiments were also confirmed by anion exchange chromatography (40). In this assay, the peptides are passed over a positively charged resin and the free ATP binds, whereas the phosphorylated peptides are not retained. The results of this assay are consistent with those of the gel assay. 2 Thus, the most important determinates for efficient peptide phosphorylation by dephospho-CaM KI are the Lys residues at PϪ7 and PϪ6, with the Lys at PϪ7 being most critical. DISCUSSION A large number of protein kinases have Ser/Thr or Tyr within their activation loops, and many of these kinases are phosphorylated on these residues. In all cases examined thus far, phosphorylation is thought to be required for appreciable kinase activity (24). It was previously thought that CaM KI also required phosphorylation on Thr-177 in order for the enzyme to be active (28, 30, 39, 50 -52). However, in analyzing the substrate specificity of Thr-177 dephospho-CaM KI, data that agree with studies done with the phosphorylated kinase (46), we were surprised to find that ADR1-(222-234)G233 was phosphorylated 15-fold better than the best CaM KI substrate, synapsin-(4 -13). We then asked whether the specific activity toward ADR1-(222-234)G233 could further be enhanced by Thr-177 phosphorylation. Because phosphorylation had only modest effects on CaM KI activity toward ADR1-(222-234)G233, kinetic analysis was undertaken to determine the mechanism of activation and why certain substrates could apparently diminish the need for Thr-177 phosphorylation. Using the synapsin-(4 -13) substrate, we found that Thr-177 phosphorylation modulates virtually every enzymatic parameter, V max , K m for substrate, K m for ATP, and K CaM . The largest effect is a 44-fold decrease in K m for substrate. These data are in contrast to Inoue et al. (51) who used an inefficient substrate for CaM KI, syntide-2, to show that K m for peptide decreases 7-fold and K m ATP decreases 3-fold after CaM KI phosphorylation.
Our finding that phospho-CaM KI has an increased affinity for CaM as compared with dephospho-CaM KI may have important physiological implications. Previously, Meyer et al. (43) have shown that autophosphorylation of another CaM-dependent kinase, Ca 2ϩ /calmodulin-dependent protein kinase II, markedly decreases the K CaM of the enzyme, resulting in a sensitization to repetitive Ca 2ϩ spikes. Sensitization of Ca 2ϩ / calmodulin-dependent protein kinase II to changes in intracellular Ca 2ϩ concentration is important in such a neuronal process as long term potentiation and long term depression. CaM KI is also expressed in neurons (53) and becomes phosphorylated on Thr-177 in response to KCl depolarization of PC12 cells (31). CaM KI phosphorylation and the accompanying left-shift in the CaM activation curve may also be important in regulation of the neuronal processes of long term potentiation and long term depression.
Our kinetic analysis of dephospho-and phospho-CaM KI using both the synapsin-(4 -13) and ADR1-(222-234)G233 peptide substrates revealed that CaM KI has an extremely high activity in the absence of Thr-177 phosphorylation toward substrates with very low K m values. Substrates that have high K m values are phosphorylated quite poorly unless Thr-177 is phosphorylated. Poor substrates for Thr-177 dephospho-CaM KI, however, can be efficiently phosphorylated by this form of the enzyme by introducing Lys or Arg residues at PϪ7 and PϪ6 relative to the phosphoacceptor site. The literature contains several examples of protein kinases that have substrate spec-ificity determinates at PϪ7 or PϪ6. Earlier studies on PKA showed that high affinity binding of the inhibitor peptide PKI depended on: 1) the presence of a basic residue at the PϪ6 position, 2) dibasic residues at PϪ2 and PϪ3, and 3) an aromatic residue at the PϪ11 position. The importance of the PϪ6 position is illustrated by the kinetics of PKA phosphorylation of the PKI(14 -22)S21 peptide, GRTGRRASAI, which has a K m of 0.11 M, 40-fold lower than Kemptide (K m 4.7 M), LRRASLG, a substrate that does not contain a basic residue at PϪ6 (54). Substrate specificity requirements seven residues away from the phosphoacceptor are also reported for smooth muscle MLCK (55,56). Recent work by Millward et al. (57) demonstrated that basic residues at PϪ7 and PϪ6 might also be important substrate determinants for the S100-regulated kinase, Ndr. The peptide sequence AARNRTLSVA in which the Lys residues at PϪ7 and PϪ6 were changed to Ala was not phosphorylated by this kinase. These authors used the P81 phosphocellulose binding assay but did not report whether the Ala analog peptides bound to the P81 paper quantitatively. We have shown here that the ADR1 peptide, AAALTRRASFSGQ, containing substitutions at PϪ8 through PϪ6, although having two positively charged residues and a free N terminus, does not bind quantitatively to P81 phosphocellulose filters.
The kinetic mechanism of CaM KI activation is similar to that of PKA in that K m for peptide is dramatically affected in both cases. Phosphorylation of PKA also significantly decreases its K m ATP, whereas phosphorylation of CaM KI only slightly affects K m ATP. Mutation of the equivalent Thr in PKA, Thr-197, to either Ala or Asp results in a 25-100-fold increase in K m substrate and K m ATP (26). In addition, Steinberg et al. (25) demonstrated that the unphosphorylated PKA also has elevated K m values for substrate and ATP similar to those seen with the mutant catalytic subunits (26). The increased K m for ATP was because of a decreased binding affinity, but the K m substrate increases were not because of peptide binding affinities, but rather a reduced rate of phosphoryl transfer.
The catalytic subunit of PKA was crystallized as an "active" kinase with Thr-197 phosphorylated and Mg 2ϩ /ATP present. The structures of CaM KI (58), CDK2 (59), and extracellular signal-regulated kinase 2 (60) all reflect inactive kinases, where the activating residue(s) within the activation loop were not phosphorylated. It is thought that the conformation of the activation loops of both dephosphorylated CDK2 and extracellular signal-regulated kinase 2 precludes substrate binding. However, mutagenesis data from PKA reveal that the loss of Thr-197 phosphate has no effect on peptide binding; the changes in K m peptide are because of phosphate transfer. Based on our data, the nonphosphorylated activation loop of CaM KI, like the nonphosphorylated activation loop of PKA, does not preclude substrate binding, as was proposed for CDK2 and extracellular signal-regulated kinase 2. The crystal structure of the CaM KI 1-320 enzyme contains a disordered activation loop providing us with no information about the conformation of the unphosphorylated loop. Our studies suggest that the change in peptide K m , in response to Thr-177 phosphorylation, whether because of peptide binding or rate of phosphoryl transfer, can be overcome by the sequence of the substrate. It seems reasonable to suggest that the unphosphorylated loop of CaM KI does not adopt a conformation that prevents substrate binding because unphosphorylated CaM KI efficiently phosphorylates the ADR1-(222-234)G233 substrate.
The crystal structures of CaM KI and PKA provide some insight as to how kinases recognize their substrate. CaM KI contains an autoinhibitory domain that, in the absence of Ca 2ϩ / CaM, lies within the substrate binding groove. When CaM KI is autoinhibited, its autoinhibitory domain is predicted to make similar contacts with the catalytic domain as would a substrate. Ca 2ϩ /CaM binding to the enzyme results in the removal of the autoinhibitory domain from the catalytic pocket and allows substrate binding. The CaM KI fragment (1-320) was crystallized in the absence of Ca 2ϩ /CaM and Thr-177 phosphorylation. Hence, the structure reflects the autoinhibited kinase. The catalytic subunit of PKA does not contain an autoinhibitory domain but the kinase was crystallized in the presence of the inhibitory protein, PKI. PKI binds to PKA in a manner competitive with substrate. In addition, a variant of PKI that contains a PKA autophosphorylation site in its C terminus is no longer an inhibitor but a substrate (54). Thus, by inspecting the contacts of PKI and the autoinhibitory domain of CaM KI with the catalytic domain of PKA or CaM KI, respectively, we can predict how CaM KI recognizes individual residues surrounding the substrate phosphoacceptor site. CaM KI has a strict requirement for a basic residue at PϪ3. Glu-102 of CaM KI is in the equivalent position of Glu-127 in PKA, a residue that makes contacts with the basic Arg at PϪ3 within the substrate (Fig. 6). Lys-300 within the autoinhibitory domain of CaM KI mimics Arg at PϪ3 by interacting with Glu-102. Mutation of Glu-102 alters the enzyme specificity at PϪ3 (58). Likewise, the preference of CaM KI for a hydrophobic residue at PϪ5 within the substrate is illustrated by the autoinhibitory domain residue Phe-298 being buried within a deep hydrophobic pocket formed by Phe-104, Ile-210, and Pro-216.
We can structurally rationalize how the addition of basic residues at PϪ7 and PϪ6 onto a CaM KI substrate substantially decreases its K m . Our chimeric peptide studies indicated that the most crucial factor for efficient phosphorylation of ADR1-(222-234)G233 (a low K m value) by dephospho-CaM KI was Lys at PϪ7. It seems reasonable that Lys at PϪ7 is interacting with Glu-221, a residue residing at the end of the loop preceding the G ␣-helix within the catalytic core. Alternatively, depending on how the Lys side chain at PϪ7 lies, it may be interacting with Val-108, a residue that interacts with PϪ8 within the autoinhibitory domain. Interestingly, CaM KIV, another enzyme that displays similar substrate phosphorylation properties to CaM KI in the absence and presence of activation loop phosphorylation, has conserved Glu and Val residues at these same positions. In PKA, the PϪ6 site represented by the positively charged Arg-15 in PKI, interacts with the negatively charged Glu-203 within the catalytic core (22). The equivalent residue to Glu-203 in CaM KI is Gly-183. It seems reasonable that if Gly-183 were a negatively charged Glu, as in PKA, the positively charged Lys at PϪ6 in ADR1-(222-234)G233 peptide may contribute more significantly to the "independence" of ADR1-(222-234)G233 for Thr-177 phosphorylation. Other enzymes shown to be regulated by activation loop phosphorylation such as protein kinase C, PKA, and CDK2 have a Glu (in the case of protein kinase C) or an Asp (in the cases of PKA and CDK2) at the equivalent position as Glu-221 in CaM KI, raising the possibility that they may also have peptide substrates that are efficiently phosphorylated in the absence of activation loop phosphorylation.
Our data reveal that Thr-177 phosphorylation expands the substrate specificity of CaM KI. What structural role does Thr-177 phosphorylation play with substrates, like synapsin-(4 -13), whose V max /K m is dramatically influenced by activation loop phosphorylation? The answer to this question may be partially explained by small angle x-ray and neutron scattering with another CaM-dependent enzyme, skeletal (sk) muscle MLCK. Krueger et al. (61) have reported that CaM binding to skMLCK causes the autoinhibitory sequences to swing away from the catalytic core. The addition of either the nonhydrolyzable ATP analogue, AMPPNP, or a peptide substrate from the myosin light chain to this CaM⅐MLCK complex causes a compaction of the catalytic domain (62). The closure between the small and large lobes of the kinase and the accompanying shift in the CaM and skMLCK centers of mass toward one another, however, require both Mg 2ϩ /ATP and peptide substrate binding. Although the activity of skMLCK is not regulated by activation loop phosphorylation, its activity is stimulated by CaM in a manner similar to CaM KI. The crystal structure of CaM KI in the absence of CaM, Mg 2ϩ /ATP, peptide substrate, and Thr-177 phosphorylation is in the open conformation. If we extend the findings with skMLCK to CaM KI, we predict that the ability of peptide substrate to induce conformational changes may be influenced by Thr-177 phosphorylation. Peptide substrates that are phosphorylated efficiently in the absence of Thr-177 bind to dephospho-CaM KI with high affinity and induce the requisite conformational change associated with substrate binding, whereas other "activation-dependent" substrates would require Thr-177 phosphorylation in order for them to induce the conformational change. Alternatively, Thr-177 phosphorylation may induce conformational changes in the kinase that can be mimicked by binding of a high affinity substrate.
Regardless of the molecular mechanism, our results identify the possibility that substrates directly regulate kinase activity. Previously, Walsh et al. (63) have pointed out that different substrate affinities may be important in regulating the order of substrate phosphorylation events within cells to coordinate different metabolic events. Our results highlight the possibility that phosphorylation of protein kinase activation loops may add a further regulatory dimension to the timing of phosphorylation events within cells. FIG. 6. Alignment of the residues within the autoinhibitory domain of CaM KI with residues from the ADR1-(222-234)G233 substrate. The arrows indicate residues within the catalytic core of the enzyme that have been demonstrated to interact with either the autoinhibitory domain or the peptide substrate. The boxed amino acids are predicted to interact with Lys at PϪ7 in ADR1-(222-234)G233.