Regulatory Segments of Ca2+/Calmodulin-dependent Protein Kinases*

Catalytic cores of skeletal and smooth muscle myosin light chain kinases and Ca2+/calmodulin-dependent protein kinase II are regulated intrasterically by different regulatory segments containing autoinhibitory and calmodulin-binding sequences. The functional properties of these regulatory segments were examined in chimeric kinases containing either the catalytic core of skeletal muscle myosin light chain kinase or Ca2+/calmodulin-dependent protein kinase II with different regulatory segments. Recognition of protein substrates by the catalytic core of skeletal muscle myosin light chain kinase was altered with the regulatory segment of protein kinase II but not with smooth muscle myosin light chain kinase. Similarly, the catalytic properties of the protein kinase II were altered with regulatory segments from either myosin light chain kinase. All chimeric kinases were dependent on Ca2+/calmodulin for activity. The apparent Ca2+/calmodulin activation constant was similarly low with all chimeras containing the skeletal muscle catalytic core. The activation constant was greater with chimeric kinases containing the catalytic core of Ca2+/calmodulin-dependent protein kinase II with its endogenous or myosin light chain kinase regulatory segments. Thus, heterologous regulatory segments affect substrate recognition and kinase activity. Furthermore, the sensitivity to calmodulin activation is determined primarily by the respective catalytic cores, not the calmodulin-binding sequences.

linker region between the catalytic core and calmodulin-binding sequence contributes to inhibition. However, the boundary between the autoinhibitory and calmodulin-binding sequences remains controversial.
Recently, the crystal structures of Ca 2ϩ /calmodulin-dependent protein kinase I (12) and twitchin kinase (13,14) presented structural insights into the autoinhibitory mechanism. The regulatory segments have extensive molecular contacts with residues on the surface of their catalytic cores, with binding energy provided by both electrostatic and hydrophobic interactions. Twitchin, a protein kinase related to myosin light chain kinases, does not have a calmodulin-binding sequence involved in regulation. However, the structure of Ca 2ϩ /calmodulin-dependent protein kinase I shows multiple contacts between residues in the catalytic core and the calmodulin-binding sequence. It seems reasonable to assume a similar structural arrangement for myosin light chain kinases.
Regulatory segments of protein kinases may be involved in different functions. In the case of Ca 2ϩ /calmodulin-dependent protein kinases, in addition to calmodulin binding and autoinhibition, the regulatory segment may affect enzyme stabilization (15,16) or increase Ca 2ϩ /calmodulin-independent activity through autophosphorylation (17). Dekker et al. (18) reported that protein kinase C-was not able to phosphorylate histone; however, after deletion or substitution of some residues in the pseudosubstrate region of the enzyme, histone became a good substrate. Exchange of all or part of the pseudosubstrate sequence between protein kinase C-␣ and protein kinase Cresulted in changed substrate selectivity. Fujise et al. (19) found that deletion of the regulatory domain in protein kinase C decreased efficiency of substrate peptide phosphorylation.
The apparent diverse functions of some protein kinase regulatory segments stimulated our investigation on two types of Ca 2ϩ /calmodulin-dependent protein kinases. Questions are raised on the respective roles of the catalytic cores and calmodulin-binding sequences in determining sensitivity to activation by Ca 2ϩ /calmodulin and whether putative interactions between a regulatory segment and a catalytic core affects substrate recognition. In the present approach, the functional properties of regulatory segments were investigated by expressing skeletal muscle myosin light chain kinase and CaMKII catalytic cores containing chimeric regulatory segments. The results show that heterologous regulatory domains had effects on phosphorylation properties of protein substrates. Surprisingly, the sensitivity to calmodulin activation was determined primarily by the respective catalytic cores, not the calmodulin-binding sequences.

Expression and Purification of Recombinant Myosin Light Chain and
Calmodulin-Recombinant human smooth and rabbit skeletal muscle myosin light chains were expressed and purified as described previ-ously (20). Recombinant bovine calmodulin was subcloned into Escherichia coli SMH 174 (DE3) (21). Cells were harvested at 8000 ϫ g for 5 min, and the pellet was resuspended in 50 mM Tris/HCl, pH 7.5, 2 mM EDTA, 1 mM dithiothreitol, 200 g/ml lysozyme, and 0.1% Nonidet P-40. The lysate was sonicated for 30 s, three times. After centrifugation, the lysate supernatant fraction was removed, and CaCl 2 was added to a final concentration of 5 mM. The protein solution was applied to a phenyl-Sepharose column equilibrated in 50 mM Tris/HCl, pH 7.5, 1 mM dithiothreitol, and 0.2 mM CaCl 2 . The column was washed with 0.5 M NaCl 2 , and calmodulin was eluted with 50 mM Tris/HCl, pH 7.5, 0.5 mM dithiothreitol, and 5 mM EGTA. The concentration of calmodulin was determined by measuring the absorbance at 280 nm and using an extinction coefficient of 0.21 for a 0.1% concentration.
Construction and Expression of Recombinant Protein Kinases and Chimeric Kinases-cDNAs for wild-type rabbit skeletal and smooth muscle myosin light chain kinase, ␣-CaMKII, and neuronal nitric oxide synthase were described previously (22)(23)(24)(25). The polymerase chain reaction was used to construct truncations and mutant proteins. tSkM-LCK (residues 257-607) containing the catalytic core and regulatory segment was made by deleting the N-terminal 256 amino acid residues. tCaMKII (residues 1-330) was devoid of the C-terminal association domain. A restriction site, NcoI, was created between the catalytic core and the regulatory segment of skeletal and smooth muscle myosin light chain kinase, CaMKII, and then N terminus of calmodulin-binding sequence from neuronal nitric oxide synthase so that chimeras could be constructed by exchange of heterologous regulatory segments (Fig. 1). All cDNA constructs of truncated and chimeric kinases were confirmed by DNA sequencing. Wild-type, truncated, and chimeric protein kinase cDNAs were subcloned into pCMV5 expression vector and transfected into COS cells (26). The cells were collected after 48 h, and lysates were prepared as described previously (22) in a buffer containing 10% glycerol, 1% Nonidet P-40, 20 mM MOPS, pH 7.5, 1 mM EGTA, 2 mM MgCl 2 , 2 mM dithiothreitol, 10 g/ml pepstatin, 4 g/ml aprotinin, 10 g/ml leupeptin, 100 g/ml phenylmethylsulfonyl fluoride, and 100 g/ml N-tosyl-L-phenylalanine chloromethyl ketone. The quantity of kinases in COS cell lysates was determined by calmodulin overlay with purified kinases as standards (5).
Immunoblots and Calmodulin Overlays-Immunoblots of tSkMLCK and chimeras were performed with monoclonal antibody 12a raised to skeletal muscle myosin light chain kinase (23). Biotinylated calmodulin overlay was performed as described previously (5) with biotinylated calmodulin prepared according to Billingsley et al. (27).
Kinase Assays and Calmodulin Activation-The activities of wildtype, truncated, and chimeric kinases were measured by 32 P incorporation into myosin regulatory light chains as described by Zhi et al. (20). The cell lysates containing expressed kinases were diluted 1:25 to 1:1000 into the kinase reaction mixture. A 1:10 dilution of mock-transfected COS cells had no detectable kinase activity in the presence or absence of Ca 2ϩ /calmodulin. The Ca 2ϩ /calmodulin-independent activity of kinases was measured in the presence of 5 mM EGTA. K m and V max values were determined from Lineweaver-Burke plots after measuring rates of 32 P incorporation with varying concentrations of myosin regulatory light chain.
Calmodulin activation properties of kinases were assessed in the following reaction mixture: 50 mM MOPS, pH 7.2, 1 mM dithiothreitol, 1 M calmodulin, 1 mM [␥-32 P]ATP (180 -300 cpm/pmol), 10 mM magnesium acetate, and 10 M light chain at 30°C and various concentrations of free Ca 2ϩ as determined by Ca 2ϩ /EGTA buffers (10,26). Quantitative changes in the concentration of Ca 2ϩ /calmodulin required for half-maximal activation (K CaM ) of truncated and chimeric kinases relative to wild-type skeletal muscle myosin light chain kinase were determined by calculating the ratio of activities at Ca 2ϩ concentrations that result in less than maximal activity to maximal activity measured at 100 M Ca 2ϩ (7,10,28). The activity at a specific Ca 2ϩ concentration that is less than that required for maximal activity decreases quantitatively as the K CaM value for a truncated or chimeric kinase increases relative to the wild-type enzyme. Although the ratio of activities does not allow determination of the absolute value of K CaM , it may be used to calculate differences relative to wild-type kinase based upon the quantitative relationship described previously (7,10,28,29). It is proposed that a decrease or increase in the K CaM value reflects a weakening or strengthening, respectively, of binding interactions between the regulatory segment and catalytic core.
Stability Studies on Chimeric Kinases-For limited proteolysis, COS cell lysates containing kinases expressed at approximately 5-10 ng/l were diluted 1:5 into the following 50-l reaction mixture: 6 mM MOPS, pH 7.5, 0.6 mM dithiothreitol, and 1 M calmodulin in the presence of 1 mM CaCl 2 or 5 mM EGTA. Digestion at room temperature was initiated by the addition of varying amounts of trypsin (type I bovine pancreas) and stopped after 10 min by the addition of soybean trypsin inhibitor in a 2-fold excess relative to the amount of trypsin. An aliquot of each digest was then diluted into SDS sample buffer, and the resultant proteolytic digestion pattern examined following SDS-polyacrylamide gel electrophoresis and immunoblotting (5). In another aliquot, myosin light chain kinase activity was measured at room temperature following 4 -12-fold dilution of the proteolyzed extracts into reaction mixtures containing either Ca 2ϩ or EGTA as described above.
For thermal stability studies, COS cell lysates were diluted 1:5 in 1 mM dithiothreitol and 1 mg/ml bovine serum albumin in the absence of Ca 2ϩ /calmodulin. The lysates were incubated at 46°C, and aliquots were removed at the indicated times were cooled on ice. Kinase activities were measured in the presence of Ca 2ϩ /calmodulin as described above. Immunoblotting following SDS-polyacrylamide gel electrophoresis was also performed as described above.

Constructs of Wild-type, Truncated, and Chimeric Kinases-
All of the wild-type, truncated, and chimeric kinases were constructed as shown in Fig. 1. tSkMLCK was made as de- scribed previously (9) by removing the N-terminal 256 amino acid residues, which have no known functional role. tCaMKII was constructed by deleting the C-terminal association domain responsible for oligomerization of the holoenzyme (30, 31) The calmodulin-binding sequence and additional N-terminal residues of neuronal nitric oxide synthase were used to construct the chimera tSkMLCK[nNOS]. To keep a consistent length of the flanker region between the calmodulin-binding sequence and the catalytic core, additional residues were added to make a regulatory segment similar in length to myosin light chain kinase and CaMKII (Fig. 2). A NcoI restriction site was created between the catalytic core and the regulatory segment of the myosin light chain kinases, CaMKII, and the N termini of the calmodulin-binding sequence of nitric oxide synthase. The chimeras were constructed by ligating different catalytic cores and regulatory segments (Figs. 1 and 2).
All constructs were expressed in COS cells after transfection, and biotinylated calmodulin overlays were performed to visualize the calmodulin-binding properties and to quantify the amount of kinase in cell lysates. Results showed that all of the chimeric and truncated kinases bound calmodulin (Fig. 3).
Kinetic Properties of Wild-type, Truncated, and Chimeric Kinase-tSkMLCK has kinetic properties similar to wild-type skeletal muscle myosin light chain kinase in regards to K m and V max values for both skeletal and smooth muscle regulatory light chains (9) (Table I). Additionally, tSkMLCK and chimeric tSkMLCK[SmMLCK] have similar activities toward regulatory light chains from smooth and skeletal muscle; however, tSkM-LCK[CaMKII] had an 8-fold increase in the K m value and a 7-fold decrease in the V max :K m ratio values for SmRLC. Surprisingly, the rates of skeletal muscle regulatory light chain phosphorylation were too low to measure K m and V max values for this chimera. The rate of phosphorylation at 40 M skeletal muscle regulatory light chain was at least 1000-fold less than tSkMLCK with 32 P incorporation only 2-fold (or less) that observed in a mock-transfected COS cell lysate. tSkMLCK[n-NOS] had no activity toward either light chain, although it was expressed ( Fig. 3 and Table I).
The oligomerization of CaMKII depends upon the association domain located C-terminal of the catalytic core and calmodulinbinding sequence (24). Deletion of the association domain does not change the kinetic properties of the enzyme in vitro (32). In the present paper, tCaMKII was constructed by removing the association domain, and the activity assays showed that the K m value was 6 M and that the V max value was 475 pmol of 32 P incorporated/min/pmol of kinase with smooth muscle regulatory light chain as a substrate. Although the skeletal muscle light chain could be phosphorylated, the rates were too low to accurately determine K m and V max values, similar to results obtained with tSkMLCK[CamKII] with skeletal muscle regulatory light chain. The K m value of chimeric tCaMKII[SkM-LCK] was increased 4-fold, whereas the V max value was decreased 2-fold. Similar results were obtained with tCaMKII [SmMLCK]. Kinetic values with skeletal muscle light chain as a substrate were undeterminable for all chimeras containing the tCaMKII catalytic core (Table I).
We also constructed tSkMLCK, tSmMLCK, and tCaMKII, in which their respective regulatory domains were deleted, keeping only the catalytic core sequences. In all cases insignificant kinase activities were detected (data not shown).
Ca 2ϩ /Calmodulin Activation Properties of Wild-type, Truncated, and Chimeric Kinases-The Ca 2ϩ /calmodulin activation properties were determined by performing assays at a high calmodulin concentration (1 M) with a Ca 2ϩ /EGTA buffer used to vary the free Ca 2ϩ concentration and hence the Ca 2ϩ /calmodulin concentration (10,28). All of the truncated and chimeric kinases were dependent on Ca 2ϩ /calmodulin for activity. The concentration of Ca 2ϩ /calmodulin for half-maximal activation of myosin light chain kinases is about 1 nM (33). In comparison, wild-type and tSkMLCK had similar K CaM values with both skeletal and smooth muscle light chain substrates (Table  II). The relative values for tSkMLCK[SmMLCK] and tSkML-CK[CaMKII] were less but in the same range.
In contrast to myosin light chain kinase, tCaMKII has an apparent lower affinity for Ca 2ϩ /calmodulin, with half-maximal activity attained between 25-100 nM (34). In this study, tCaMKII had a K CaM value of 50 nM relative to tSkMLCK (Table II). The chimeric tCaMKII[SkMLCK] and tCaMKI-I[SmMLCK] also had relatively high K CaM values of 27 and 28 nM, respectively (Table II).
Stability Studies on Chimeric SkMLCK-The extent of structural identity among the regulatory segments of tSkMLCK, tSmMLCK, and tCaMKII is low (Fig. 2), and thus it could be expected that the stability properties of the chimeras would be significantly altered. Alterations in the structure of the catalytic core with different regulatory segments may lead to significant increases in susceptibility to digestion by proteases (5). Therefore, limited trypsin digestion was used to assess structural and activation properties of the chimeras (Fig. 4). tSkM-LCK, tSkMLCK[SmMLCK], and tSkMLCK[CaMKII] were resistant to trypsin digestion in the presence of EGTA. In the presence of Ca 2ϩ /calmodulin, tSkMLCK and tSkML-CK[SmMLCK] were more sensitive to digestion by trypsin (Fig.  4). However, the pattern of digestion of tSkMLCK [CaMKII] was similar in the absence and presence of Ca 2ϩ /calmodulin and different from tSkMLCK and tSkMLCK [SmMLCK].
When the trypsin digestion was performed in the presence of Ca 2ϩ /calmodulin, the Ca 2ϩ /calmodulin-dependent kinase activities of the chimeras decreased with increasing trypsin concentrations with partial conversion to Ca 2ϩ /calmodulin-independent activities for tSkMLCK, tSkMLCK[SmMLCK], and tSkMLCK [CaMKII]. The decrease in activity for tSkMLCK-[CaMKII] is associated with a modest reduction in the apparent mass of the chimera compared with tSkMLCK and tSkM-LCK[SmMLCK] (Fig. 4).
The patterns of tSkMLCK[nNOS] digestion in the presence of Ca 2ϩ /calmodulin or EGTA were similar to the patterns of those that were boiled before digestion (data not shown). Additionally, tSkMLCK[nNOS] showed no kinase activity, even after limited digestion (data not shown). Thus, this chimera may not fold correctly for activity.
Thermal instability was also used to assess structural perturbations in chimeric tSkMLCKs. The decrease in sensitivity to trypsin digestion in tSkMLCK [CamKII] in the presence of Ca 2ϩ /calmodulin could be due to the loss of a proteolytic cleavage site in the chimera. Incubation of tSkMLCK, tSkML-CK[SmMLCK], and tSkMLCK[CaMKII] at 46°C in the presence of EGTA resulted in similar time-dependent losses of kinase activities (Fig. 5). Immunoblots show that the loss of activities was most likely due to thermal denaturation, because there was no evidence of proteolysis (Fig. 5). DISCUSSION Myosin light chain kinases and Ca 2ϩ /calmodulin-dependent protein kinase II have similar but distinct biochemical properties. Myosin light chain kinase is a dedicated protein kinase with high substrate specificity toward myosin regulatory light chain (35), whereas Ca 2ϩ /calmodulin-dependent protein kinase II is a multifunctional protein kinase with diverse substrate recognition (24, 36 -38). Proteolysis studies supported the hypothesis that both types of kinases contain an autoinhibitory sequence in the C-terminal region close to the catalytic core (3, 39 -47). The current structural model is that the kinases are inactive in the absence of Ca 2ϩ /calmodulin due to an autoinhibitory sequence folding back onto the catalytic core and blocking substrate binding. This type of regulation is referred to as intrasteric regulation (48).
The recently determined structure of twitchin kinase provides some insight into how an autoinhibitory segment may bind to the catalytic core (13,14). The autoinhibitory segment has multiple interactions with the catalytic core of enzyme, with 47 hydrogen bonds and as many as 351 van der Waal contacts. The structure of Ca 2ϩ /calmodulin-dependent protein kinase I (12) shows that the position of its autoinhibitory sequence differs dramatically from twitchin kinase. The twitchin autoinhibitory sequence enters the active site in the cleft between the two lobes of the catalytic core passing over the activation loop of the enzyme. The autoinhibitory and calmodulin-binding sequences of Ca 2ϩ /calmodulin-dependent protein kinase I also traverse the surface of the large lobe of the catalytic core, but they turn in an opposite direction and avoid the activation loop entirely by traversing the small lobe. However, like twitchin kinase, the regulatory segment of Ca 2ϩ / calmodulin-dependent protein kinase I has multiple contacts with the catalytic core.
In the present work, we constructed chimeric protein kinases with heterologous regulatory segments and tested their autoinhibitory function. The activity of all chimeric kinases containing kinase regulatory segments were Ca 2ϩ /calmodulin-dependent, with no significant activity in the presence of EGTA. Thus, the heterologous regulatory segments had a similar autoinhibitory function that was reversed by Ca 2ϩ /calmodulin. However, there was some apparent specificity because the chimera tSk-MLCK[nNOS] was not active in the presence or absence of Ca 2ϩ /calmodulin, even though it bound calmodulin. The calmodulin-binding sequence of neuronal nitric oxide synthase contains basic residues at its N terminus in a sequence arrangement similar to the putative pseudosubstrate sequence in the calmodulin-binding sequence of myosin light chain kinases (2,49). This structure alone does not provide sufficient stability to the catalytic core, which is consistent with the hypothesis that the linker region between the catalytic core and the calmodulin-binding sequence is important in autoinhibition and stability of the catalytic core (11). The other heterologous regulatory segments, however, appear to provide similar structural stability, as revealed by limited proteolysis and thermal denaturation measurements. Faux et al. (15) found that a 61-kDa tryptic fragment of SmMLCK in which the regulatory segment was truncated underwent rapid inactivation. However, a synthetic peptide with a similar primary structure of the autoinhibitory sequence protected the 61-kDa fragment from thermal inactivation. Kennelly et al. (50) found calmodulin binding to skeletal muscle myosin light chain kinase resulted in a timeand temperature-dependent inactivation in the absence of substrates. Separation of the components of the inactive complex yielded functional calmodulin but catalytically inert kinase. Similarly, Ishida and Fujisawa (16) removed the autoinhibitory sequence of Ca 2ϩ /calmodulin-dependent protein kinase II and found that the enzyme was more labile than the original enzyme.
It has been reported that the regulatory segment affects substrate recognition. Dekker et al. (18,51) observed that the pseudosubstrate site mediated the low histone kinase activity of wild-type PKC-␣. Deletion of the pseudosubstrate region generates a cofactor-independent kinase that has high histone kinase activity, but deletion of the regulatory segment decreases the phosphorylation rate for a peptide substrate. Fujise et al. (19) exchanged the autoinhibitory pseudosubstrate sequence between PKC-␣ and PKC-, which changed substrate recognition properties. Our results show that regulatory segments of Ca 2ϩ /calmodulin-dependent protein kinases may also affect substrate recognition and hence may be a more general mode of regulation than previously recognized. Chimeric tSk-MLCK[CaMKII] had a 7-fold decrease and undeterminable V max :K m values for smooth and skeletal muscle light chains, respectively. Hence, the heterologous regulatory segment affected the recognition of the physiological substrate, the skeletal muscle light chain.
The substrate determinants for smooth and skeletal muscle light chains are different (33,49). An Arg at the P-3 position relative to the phosphorylatable serine (P-0 position) is impor-tant for substrate recognition for the smooth muscle light chain. This residue is replaced with Glu in the skeletal muscle light chain, and other residues are important for recognition. The Ca 2ϩ /calmodulin-dependent protein kinase II requires a basic residue at the P-3 position, which is why it readily phosphorylates smooth but not skeletal muscle light chain. tCaMKII[SkMLCK] and tCaMKII[SmMLCK] had 10 -20-fold lower V max :K m values for smooth muscle regulatory light chain. Thus, the heterologous regulatory segments in the chimeric kinases function differently in terms of affecting catalytic activity. In the case of tSkMLCK[CaMKII], the effect was dramatic, involving a loss of significant rates of phosphorylation of the skeletal muscle light chain. However, other chimeras primarily show changes in kinetic properties with decreases in the V max :K m ratio. Some possible mechanisms include 1) the regulatory segment, along with the complexed calmodulin, affects the structure of the catalytic core and hence kinase activity, or 2) the regulatory segment with bound calmodulin in the chimeric kinase sterically modifies substrate binding.
Recently, Krueger et al. (52) used small-angle x-ray and neutron scattering with contrast variation to obtain the first structural view of calmodulin complexed to a functional enzyme, tSkMLCK (52). The results show that calmodulin was near the large lobe of the catalytic core away from the catalytic cleft, implying a major displacement of the regulatory segment from the active site. However, the close position of calmodulin to the catalytic core suggests that the regulatory segment could still affect substrate recognition.
The mechanism by which calmodulin activates enzymes is intriguing. The crystal structure of Ca 2ϩ /calmodulin shows a dumbbell-shaped molecule with two globular regions connected by an extended central helix (53). The structure of calmodulin bound to peptides of the regulatory segments of myosin light chain kinases and CaMKII shows that the extended structure collapses to a cis orientation, and the central helix is disrupted by a long flexible loop (54 -56). Mutations, truncations, and chimeras of calmodulin reveal that different target proteins need different critical residues or domains of calmodulin to activate (57)(58)(59). Furthermore, N-terminal residues of calmodulin appear to participate in activation but not in binding (60,61).
The concentration of Ca 2ϩ /calmodulin required for half-maximal activation of myosin light chain kinases is approximately 1 nM, whereas half-maximal activity of Ca 2ϩ /calmodulin-dependent protein kinase II is attained at a Ca 2ϩ /calmodulin concentration of approximately 50 nM (24, 62). Herein, it was shown that chimeric tSkMLCK[SmMLCK] and tSkMLCK-[CaMKII] had similar or low K CaM values compared with wild-type myosin light chain kinase. However, the K CaM values for chimeric kinases containing the catalytic core of Ca 2ϩ / calmodulin-dependent protein kinase II were greater and more similar to wild-type kinase. These results indicate that the determinants of K CaM values for SkMLCK and CaMKII are primarily dependent upon the respective catalytic core, not the calmodulin-binding sequence. These results are consistent with the observations that activation is not necessarily associated with Ca 2ϩ /calmodulin binding but depends upon interactions between specific residues in Ca 2ϩ /calmodulin and the catalytic core (58 -61). The determination of the threedimensional structure of these Ca 2ϩ /calmodulin-dependent protein kinases with and without bound calmodulin will provide specific insights into the structural mechanisms involved.