Methionine to glutamine substitutions in the C-terminal domain of calmodulin impair the activation of three protein kinases.

The 9 methionine residues of vertebrate calmodulin (CaM) were individually changed to glutamine residues in order to investigate their roles in enzyme binding and activation. The mutant proteins showed three classes of effect on the activation of smooth muscle myosin light chain kinase, CaM-dependent protein kinase IIα, and CaM-dependent protein kinase IV. First, some mutations had no appreciable effect on the ability of CaM to activate the three protein kinases. Included in this category were glutamine substitutions at residues 36 and 51 in the N-terminal domain, at residue 76 in the domain linker sequence, and at residues 144 and 145 in the C-terminal domain. Second, glutamine substitutions in the N-terminal domain of CaM, particularly those at positions 71 and 72, lowered the maximal activity of smooth muscle myosin light chain kinase while having no effect on the other two enzymes. Finally the affinity of CaM for all three enzymes was lowered by glutamine mutations at the neighboring methionines 109 and 124, located on a solvent-accessible surface of the C-terminal domain of Ca2+/CaM. This last result provides the first demonstration of the involvement of the same hydrophobic groups in the high affinity binding of CaM to three different enzymes.

The 9 methionine residues of vertebrate calmodulin (CaM) were individually changed to glutamine residues in order to investigate their roles in enzyme binding and activation. The mutant proteins showed three classes of effect on the activation of smooth muscle myosin light chain kinase, CaM-dependent protein kinase II␣, and CaM-dependent protein kinase IV. First, some mutations had no appreciable effect on the ability of CaM to activate the three protein kinases. Included in this category were glutamine substitutions at residues 36  Vertebrate calmodulin (CaM) 1 is a protein consisting of 148 amino acids, which belongs to the E-F hand class of Ca 2ϩbinding proteins. The crystal structure of Ca 2ϩ -bound CaM shows a primarily ␣-helical protein resembling a dumbbell, with two globular domains separated by an extended helix (1). Two Ca 2ϩ -binding sites are located in each domain, with the C-terminal pair having a higher affinity for Ca 2ϩ than the N-terminal pair. The recently solved NMR structures of Ca 2ϩfree CaM demonstrate how the binding of Ca 2ϩ induces the exposure of hydrophobic surfaces in the two separate domains (2). This conformational change enables Ca 2ϩ /CaM to bind in a 1:1 complex with target proteins. Thus CaM is capable of activating a diversity of functions such as protein phosphorylation/ dephosphorylation via kinases and phosphatases, regulation of cAMP levels via adenylyl cyclases and phosphodiesterases, maintenance of calcium homeostasis by modulating membrane Ca 2ϩ -ATPase pumps, and cellular integrity by interactions with cytoskeletal components. Since these target proteins have different CaM-binding sequences, the molecular mechanisms of regulation by CaM have been of intense interest.
Many studies have investigated the involvement of hydrophobic groups of CaM in its function. Affinity labeling of CaM with hydrophobic phenothiazine inhibitors targeted to either one or both domains selectively converted CaM to a partial or complete antagonist depending on the enzyme tested (3). The x-ray and NMR structures of CaM in complex with peptides, which correspond to the CaM binding domains of target enzymes, have provided detailed insights into the hydrophobic binding surfaces of both sets of proteins (4 -6). These structures clearly reveal direct van der Waals contacts between the peptides and many hydrophobic residues of CaM, including most of its 9 methionines. Furthermore, chemical studies have demonstrated that the oxidation of an unidentified number of Met residues of CaM to methionine sulfoxides decreased the ability of CaM to activate target enzymes (7,8). Yet attempts at defining the specific function of individual Mets of CaM by site-directed mutagenesis have provided few insights, since single substitutions of Met to Leu in mammalian CaM had little effect on the function of the protein (9). The apparent discrepancy between the results from structural analysis of CaM-peptide complexes and those of mutagenesis experiments might, in part, be explained by the conservative strategy of replacing one nonpolar side chain of CaM with another nonpolar residue. Since the substitution of Leu affected the size rather than the nonpolar nature of Met, then the functional significance of these hydrophobic residues of CaM remains an open question.
In order to address the importance of Met in CaM function, we decided to systematically survey the effects of mutating individual Mets to less conserved, more polar residues. Each of the 9 Met residues of CaM was changed to a polar Gln by site-directed mutagenesis to introduce an oxygen atom at the same position in the side chain as the harmful sulfoxide. We have tested the mutant proteins with three different CaMdependent protein kinases. An important advantage of this approach is the fact that the CaM binding properties of two of these enzymes, smooth muscle myosin light chain kinase (smMLCK) and Ca 2ϩ /CaM-dependent protein kinase II␣ (CaMKII␣), have been extensively characterized. This has allowed greater insight at the molecular level concerning the roles of individual Mets of CaM in the separate functions of enzyme recognition and activation.

Materials
Restriction enzymes were obtained from Boehringer Mannheim and New England Biolabs. PCR reagents including Taq polymerase and buffers were supplied as a kit from Boehringer Mannheim. Double-* These experiments were supported by National Institutes of Health Research Grant GM-33976 (to A. R. M.) and by a grant from the KECK Foundation (to Duke University Medical Center). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed: Dept. of Pharmacology, Duke University Medical Center, P. O. Box 3813, Durham, NC 27710. Tel.: 919-681-6209; Fax: 919-681-7767. 1 The abbreviations used are: CaM, calmodulin; smMLCK, smooth muscle myosin light chain kinase; CaMKII␣, Ca 2ϩ /calmodulin-dependent protein kinase II␣; CaMKIV, Ca 2ϩ /calmodulin-dependent protein kinase IV; PCR, polymerase chain reaction; MOPS, (3-(N-morpholino)propanesulfonic acid); DTT, dithiothreitol; PAGE, polyacrylamide gel electrophoresis; skMLCK; skeletal muscle myosin light chain kinase; CaMKI, Ca 2ϩ /calmodulin-dependent protein kinase I. stranded DNA sequencing was performed with reagents from the Sequenase kit supplied by Amersham. Seakem agarose (ME) was from FMC Bioproducts, and acrylamide was purchased from Serva. Bacterial reagents were from Difco and media for maintaining Sf9 cells were from Life Technologies, Inc. Peptide substrates for kinase assays were provided courtesy of Dr. Bruce E. Kemp from the St. Vincents Institute, Holt Laboratory, Melbourne, Australia. Radiochemical reagents were purchased from Amersham. Buffers for enzyme assays and column chromatography were made with Milli-Q purified water. All other reagents and chemicals were of the highest available quality.

Methods
Construction of Calmodulin Methionine to Glutamine Mutants-Mutations from Met to Gln were introduced into the CaM expression plasmid pCaMpl (10) by the method of polymerase chain reaction (PCR) mutagenesis (11). Pairs of complementary oligonucleotides (ϳ20-mers), both sense (Ms) and antisense (Ma), where the appropriate ATG of Met was changed to CAG of Gln in conjunction with another pair of outer oligonucleotide sense and antisense primers (Os and Oa), were used in two rounds of PCR with pCaMpl as a template to generate DNA containing the mutation of choice. In the first round of PCR, 30 ng of pCaMpl were combined with primers Ms and Oa (30 g each), 0.2 mM dNTPs, 1 ϫ reaction buffer, 0.5 units of Taq polymerase in a total vol of 50 l and subjected to the following regimen: 92°C for 1 min, 55°C for 2 min, 72°C for 2 min, repeated 20 times. In a separate tube the same reaction was carried out under parallel conditions except using Ma and Os as the PCR primers. The resulting two PCR products were separated from the parent plasmid by agarose gel (1.5%) electrophoresis, purified by centrifugation through a Millipore MC microconcentrator tube (0.45 M), combined, and concentrated by isopropanol precipitation in preparation for an intervening 3Ј-extension step. The PCR products were subjected to 3Ј-extension in the presence of 0.5 mM dNTP, 1 unit of Taq polymerase, 2 ϫ reaction buffer in a total volume of 30 l under the following conditions: 90°C for 1 min, 37°C for 1 min, 72°C for 2 min, repeated 10 times. The resulting mixture was diluted 100-fold and 15 l used as DNA template in the final round of PCR containing 30 g each of outer primers Os and Oa, 0.2 mM dNTP, 1 ϫ reaction buffer, 0.5 units of Taq polymerase in a total volume of 50 l under the same conditions as the first PCR reaction. An aliquot of the resulting product was analyzed by agarose gel electrophoresis, and the remainder concentrated by alcohol precipitation prior to restriction digestion.
The PCR products were treated with a pair of the appropriate restriction enzymes (NcoI/AccI or AccI/XbaI) to yield DNA cassettes, which were subsequently ligated into the parent pCaMpl vector via the same two unique restriction sites, thus replacing the wild type DNA sequence with the mutant. Ligation reactions were then used to transform competent bacteria (strain MM 294 Cl ϩ ). The mutation was verified by doublestranded DNA sequencing of plasmids isolated from the resulting bacterial colonies. In each case sequencing confirmed the presence of only one amino acid substitution from Met (ATG) to Gln (CAG).
Expression and Purification of Calmodulin-CaM mutants were expressed in bacteria by heat induction and purified to homogeneity by conventional column chromatography. Individual pCaMpl plasmids (ϳ0.1 g) were used to transform a strain of Escherichia coli (N5151) suitable for expressing the mutant proteins. Several colonies were added to 1 liter of Luria-Bertoni (LB) liquid medium and grown at 30°C with moderate shaking overnight to an A 595 nm of ϳ1.0. The liquid cultures were induced to express protein by the addition of an extra 0.5 L LB at 68°C and allowed to incubate with mixing for another 1.5 h at 42°C. Bacteria were then harvested by centrifugation and lysed by first suspending them in 20 ml of a buffered sucrose solution: 2.4 M sucrose, 40 mM Tris-HCl (pH 8.0), 10 mM EDTA for 2-4 h at 4°C, followed by the addition of 80 ml lysis buffer: 50 mM MOPS (pH 7.5), 0.1 M KCl, 1 mM EDTA, 1 mM DTT, 15-20 mg of lysozyme, and overnight incubation at 4°C. The lysed bacteria were clarified by centrifugation at 100,000 ϫ g for 2 h at 4°C, and the supernatant containing the expressed CaM was stored at Ϫ70°C for further purification.
In preparation for column chromatography, the bacterial supernatant was diluted with an equal volume of buffer A: 10 mM Tris-HCl (pH 7.5), 1 mM CaCl 2 , 1 mM DTT followed by the addition of 1 M CaCl 2 to a final concentration of 5 mM CaCl 2 . The solution was then slowly loaded onto a phenyl-Sepharose column (1 ϫ 16 cm), which had been previously equilibrated in buffer A. After washing the column extensively with buffer A, a solution of buffer B (buffer A containing 0.5 M NaCl) was passed through the column to remove contaminating protein, followed by washing with five additional volumes of buffer A. Highly purified CaM was eluted with buffer E, which contained 10 mM Tris-HCl (pH 7.5), 0.15 M NaCl, 1 mM DTT, 1 mM EDTA. In a final step low M r contaminants were separated from CaM by chromatography on a Superdex 30 column (0.6 ϫ 50 cm) equilibrated and eluted in buffer E. The final yield of protein was generally between 10 and 20 mg/liter of culture with the protein purity verified at Ͼ99% by sodium dodecyl sulfate polyacrylamide gel (12.5%) electrophoresis (SDS-PAGE). Concentrations of purified CaM mutants were determined by UV absorbance, E 253 nm 1% ϭ 0.95 and E 277 nm 1% ϭ 1.9 (12). The concentration of CaM was confirmed by the Bradford dye binding method (13) (14) inserted into the baculovirus expression vector pVL1393 was transfected into Sf9 cells (100 -200 million cells at Ͼ97% viability) at a multiplicity of infection of approximately 5:1 and incubated in media with stirring for 3 days at 27°C. Cells were harvested by centrifugation at 10,000 ϫ g for 10 min at 4°C. The cells were then lysed by sonication and the cell extract subjected to precipitation by addition of ammonium sulfate to 40% saturation. The supernatant was concentrated by the addition of ammonium sulfate to 60% saturation, and the resulting precipitate was dissolved and passed over a Superdex G-200 gel filtration column. The peak activity of smMLCK was then pooled and subjected to affinity chromatography on a CaM-Sepharose column with the final EDTA eluate stored in 40% glycerol at Ϫ70°C. The purity of the Sf9 expressed protein was confirmed by SDS-PAGE analysis. The recombinant smMLCK demonstrated similar kinetic properties to those of the chicken gizzard enzyme.
Expression and Purification of Calmodulin-dependent Protein Kinase II␣-Rat Ca 2ϩ /CaM-dependent protein kinase II␣ (CaMKII␣) was obtained after overexpression and purification from a baculovirus expression vector BlueBac II using methods based on a previously established procedure (15). Briefly, 3 days after viral infection (multiplicity of infection Ͼ 5:1) of the Sf9 cells and incubation with stirring at 27°C, the cells were harvested by centrifugation. The cells were lysed by sonication, the supernatant collected and the 60% saturated ammonium sulfate precipitate fraction subjected to gel filtration on Sephadex G-200 prior to CaM-Sepharose affinity chromatography.
Expression and Purification of Calmodulin-dependent Protein Kinase IV-The rat calmodulin-dependent protein kinase IV (CaMKIV) gene under the control of a baculovirus expression vector pVL1393 was expressed in Sf9 cells as described previously (15). The protein kinase was purified from cell extracts by a two-step procedure, first by ion exchange chromatography on DEAE-cellulose and then by CaM-Sepharose chromatography. This yielded a highly pure fraction of CaMKIV as judged by SDS-PAGE (10%). In purifications of the CaM-dependent kinases, protein concentrations were determined by the method of Bradford (13) using commercially supplied IgG as protein standard.
Assay of Calmodulin-dependent Activation of Myosin Light Chain Kinase-smMLCK was assayed for light chain phosphorylation in the presence of increasing amounts of CaM as was described previously (16). Assays were initiated by the addition of enzyme (2 nM final concentration) in a total volume of 50 l and performed for 10 min at 30°C in the following solution: 50 mM Hepes (pH 7.5), 5 mM MgCl 2 , 1 mM CaCl 2 , 1 mM DTT, 0.1% Tween 80, 0.5 mg/ml bovine serum albumin, 0.1 mM ATP (0.2 Ci), and 50 M myosin light chains, which were prepared from a bacterial expression vector as described elsewhere (17). Aliquots of 40 l were loaded onto a Whatman No. 3MM filter and washed with four to five changes in a solution of 10% trichloroacetic acid and 2% sodium pyrophosphate. Assays were also conducted under similar conditions except using 200 M myosin light chain peptide MLC 11-23 as substrate. In this case aliquots were loaded onto Whatman P-81 filters and washed in 75 mM phosphoric acid. Filters were counted on a Beckman LS 6000 scintillation counter.
Assay of Calmodulin-dependent Activation of Calmodulin Kinase II␣-Ca 2ϩ /CaM-dependent autophosphorylation of CaMKII␣ was measured at 1 mM CaCl 2 , by a modified version of a previously described assay (16) with the exception that reactions were spotted on Whatman No. 3MM paper and washed four or five times in solutions of 20% trichloroacetic acid, 2% sodium pyrophosphate before scintillation counting as described previously. This technique gave identical results for CaM activation constants when compared to those derived from the more conventional analysis of CaMKII autophosphorylation by SDS-PAGE.
Assay of Calmodulin-dependent Activation of Calmodulin Kinase IV-The conditions for determining the activity of CaMKIV for phosphorylation of the peptide GS-10 in the presence of CaM at 1 mM CaCl 2 were based on an earlier established procedure (15).  (18,19), and Gln does not change the backbone structure of the protein but rather the distal portion of the amino acid side chain. Both CaM and mutant proteins were produced to similar levels in a heatinducible bacterial expression system and were purified to homogeneity by the identical methods of column chromatography. Furthermore, gross structural analysis of the CaM point mutants by gel filtration chromatography, SDS-PAGE (with or without Ca 2ϩ ), and scanning UV-spectroscopy did not reveal any significant differences, indicating no detectable changes in the mutant proteins when compared to authentic CaM.

Analysis of Kinetic Constants-Kinetic
In contrast to the similarities in their structures, some CaM mutants were impaired in their ability to activate three different CaM-dependent kinases when compared to wild type CaM. Tests for the ability of the CaM point mutants to activate autophosphorylation of the CaMKII␣ revealed differing effects (Fig. 1). With the exception of M124Q in the C-terminal domain of CaM, all mutants were capable of maximally activating the kinase. The four mutants in the N-terminal domain, M36Q, M51Q, M71Q, and M72Q (Fig. 1A), were quite similar to the wild type protein in their respective activation profiles although M36Q and M72Q exhibited approximately 4-fold increases in their activation constants (K CaM ). Similarly, M76Q in the domain linker behaved almost identically to the wild type CaM in activating CaMKII␣ (Fig. 1A). The effects due to mutations in the C-terminal domain (Fig. 1B) can be divided into two classes with the first class represented by the adjacent mutants M144Q and M145Q, which have activation constants approximately 4-fold greater than wild type and closely resemble the activation profiles of M36Q and M72Q in the N-terminal domain. The second class of effects is exhibited by mutants M109Q and M124Q, which show the largest increases in activation constants of all the Gln point mutants tested, at 50-and 25-fold, respectively, over the wild type.
In comparison to CaMKII␣, the effects of the Gln point mutants on the activation of smMLCK showed a slightly different pattern (Fig. 2). Similar to CaMKII␣, the mutants in the N-terminal domain of CaM produced small increases in the activation constants for smMLCK with M36Q exhibiting the largest, approximately 3-fold, increase ( Fig. 2A). However unlike CaMKII␣ the adjacent mutants M71Q and M72Q produced 40% and 65% losses in maximal activity respectively. Mutation of the nearby residue 76, located in the domain linker ( Fig. 2A), behaved the same as the wild type in activating smMLCK. The effects of the mutants in the C-terminal domain of CaM (Fig.  2B) were reminiscent of CaMKII␣ since M144Q and M145Q behaved very similar to wild type, while the largest increases in K CaM were detected for M109Q and M124Q, which exhibited greater than 7-and 35-fold increases, respectively. Similar to CaMKII␣, M124Q produced an accompanying 30% decrease in maximal activity, whereas M109Q produced a smaller 15% loss in activity.
The effects of individual Gln mutants on the activation of CaMKIV were not as pronounced as in the two previous cases. An examination of the activation profiles (Fig. 3) revealed that all activation constants were within an order of magnitude of the wild type. Mutants in the N-terminal domain (Fig. 3A) all achieved maximal activity and had at the most a 2.5-fold increase in K CaM relative to wild type. Again M76Q behaved very similar to the wild type protein (Fig. 3A). By contrast Gln mutants in the C-terminal domain (Fig. 3B) all showed higher activation constants for this enzyme, with M124Q achieving the greatest effect, a 7-fold increase, followed by M109Q with a 4-fold increase in K CaM . The CaM mutants M144Q and M145Q behaved similar to those in the N-terminal domain with at most a 2.5-fold increase in K CaM . Similar to the other two kinases, M124Q resulted in a 20% loss in maximal activity, whereas in contrast to the other two enzymes, M144Q produced a 15% decrease in the maximal activity of only CaMKIV.
The effects of the individual Met to Gln mutants of CaM are summarized in Table I, which shows the activation constants (K CaM ) and maximal activity for the three kinases tested. Some general observations can be drawn from a comparison of the effects of these mutants. Focusing on the separate domains of CaM reveals that, in general, the Gln mutants of the N-terminal domain at residues 36, 51, 71, and 72 produced maximal enzyme activity accompanied by small changes in the affinity of CaM for all three enzymes tested, as evidenced by at most a 4-fold increase in their activation constants. The two major exceptions to this rule were M71Q and M72Q, both of which were unable to activate smMLCK to maximal velocity but behaved normally relative to the other two kinases. The sole Gln mutant in the domain linker at residue 76 behaved almost identically to the wild type protein for all three kinases. Mutations in the C-terminal domain of CaM resulted in one of two effects. M144Q and M145Q produced at most a 4-fold increase in K CaM with all three kinases, which, with the exception of the 85% activation of CaMKIV by M144Q, were maximally activated. The other class of effect is exhibited by M109Q and M124Q, which produced large increases in K CaM for all three enzymes and was accompanied by a 20 -35% loss in maximal activity.
Since M109Q and M124Q impaired the activation of the protein kinases in a similar manner, their calcium-bound structures were compared to that of wild type Ca 2ϩ /CaM by CD and fluorescence spectroscopy. Whereas CD measures overall secondary structural content, absorption and emission scanning fluorescence spectroscopy provides information on the microenvironments of Tyr-99 and Tyr-138 in Ca 2ϩ -binding The results showed that both mutants and wild type CaM exhibited identical CD spectra with a maximum at 192 nm and minima at 207 and 222 nm. The change in ellipticity (⌬E) was also the same between the mutant and wild type CaMs as measured by their difference spectra. The results from the fluorescence studies were the same for both Ca 2ϩ -bound mutants and wild type protein with the excitation maxima at 278 nm and the emission maxima at 307 nM. The magnitude of the fluorescence excitation and emission was also identical (Ϯ 1%) for the proteins as judged from difference spectra. 2

DISCUSSION
Previous studies have implicated hydrophobicity as a significant component in the interaction of Ca 2ϩ /CaM with peptides and enzymes (20 -22). To directly address the specific contributions of Met side chains in CaM, we individually changed each of its 9 Met residues to a Gln by site-directed mutagenesis. Gln substitutions would affect CaM function by replacing the thioether group of the hydrophobic Met with a polar amide. This apparently small change in the side chain may have significant functional consequences, however, since the sulfur of Met is thought to promote sequence-independent nonpolar interactions between proteins (23).
The enzymes chosen to evaluate the function of the CaM mutants were the protein kinases smMLCK, CaMKII␣, and CaMKIV. The mechanism of activation of these enzymes by Ca 2ϩ /CaM is different and the sequence of the CaM binding domains is distinct in each case (see Fig. 4a). For CaMKII␣ and smMLCK the structure of each CaM binding domain complexed with Ca 2ϩ /CaM has been solved by x-ray crystallography (4, 6). As summarized in Table II, all 9 Met residues of CaM interact with the smMLCK peptide structure and 6 of the 9 make contact in the CaMKII␣ peptide complex. Some of the methionines, such as Met-144, make three or four contacts with both peptides. Therefore, it was surprising to find that the substitutions of Met-36, -51, -76, -144, or -145 had little effect on the activation of any of the three protein kinases by Ca 2ϩbound CaM. Results from an earlier systematic mutagenesis study on the Mets of CaM in general support these results since, with the exception of Met-36, Leu substitutions at the other 8 Mets of vertebrate CaM had little effect on the activation of the Ca 2ϩ /CaM-dependent phosphodiesterase (9). This is also compatible with results from a study in yeast in which each of the 8 Phe residues of CaM was changed individually to Ala (24).
The differences in the way CaM interacts with smMLCK and CaMKII␣ account for the differing capabilities of the M71Q and M72Q to activate these two enzymes. The 40 -65% loss in the maximal velocity of smMLCK with no change in the activation constants due to the adjacent N-terminal domain mutants M71Q and M72Q demonstrated that these residues of CaM participated in enzyme activation without affecting affinity. In comparison, the activation of CaMKII␣ with the same two mutants closely resembled the wild type protein. These results 2 D. Chin and A. R. Means, unpublished observations.

FIG. 4.
A model of calmodulin-dependent activation of a target enzyme. a, the CaM-binding amino acid sequences of five protein kinases (single-letter code) are aligned, with the conserved target residue indicated by an asterisk. The identity of the enzymes is given in the text. b, a model of the initial interaction (ϭϭ) between CaM, represented by the dumbbell (The N-terminal domain is identified by an N) and an enzyme, represented by the larger hemisphere. The small Ca 2ϩ represents resting concentrations of intracellular free [Ca 2ϩ ]. The active site of the enzyme is inhibited by the CaM-binding, autoinhibitory sequence, which is represented by a striped bar. A large, exposed, hydrophobic residue at the N-terminal end of the autoinhibitor is represented by an asterisk. The exposed hydrophobic surfaces in the two domains of CaM are represented by striped circles. c, the hydrophobic surface of the C-terminal domain of CaM interacts with the N-terminal portion of the enzyme autoinhibitor. d, a transient increase in Ca 2ϩ concentration, represented by a large Ca 2ϩ leads to activation of the enzyme by enabling the Ca 2ϩ -activated N-terminal domain of CaM to interact with the CaM-binding autoinhibitor sequence. This is aided by the bending of the CaM domain linker.  (16,17,25). In addition affinity labeling of CaM at the N-terminal domain Lys-75 with the hydrophobic phenothiazine analog norchlorpromazine isothiocyanate changed the protein to a high affinity antagonist of smMLCK without affecting activation of CaMKII␣ (3). The lowered maximal velocities of the protein kinases due to Gln mutants located in the C-terminal domain is attributed to the autoinhibitory function of the CaM binding domain of these enzymes. The effect of the CaM mutant M124Q was to lower the maximal velocity of all three enzymes between 20 and 35%, whereas M109Q and M144Q had lesser effects on the maximal activity of smMLCK and CaMKIV, respectively (Table I). A review of the crystal structures of the peptides of smMLCK or CaMKII␣ bound to CaM (Table II) reveals that Met-109 and Met-124 make multiple contacts with regions of the two protein kinases that are involved in enzyme autoinhibition (residues 796 -803 in smMLCK and residues 293-300 in CaMKII␣). The autoinhibitory sequences of the protein kinases are responsible for repressing the activity of these enzymes by blocking access to substrates (26). The role of CaM is to displace the autoinhibitor, allowing substrate binding and catalysis. If M109Q, M124Q, and M144Q (in the case of CaMKIV) partially interfered with the normal interaction between CaM and the autoinhibitory region of these kinases, this might explain their decreased efficiency in promoting catalysis. For example, Met-124 contacts Ala-796 of smMLCK and Phe-293 of CaMKII␣, which have both been implicated in separate deletion studies to play important roles in repressing the activities of their respective enzymes (27,28). In addition several other studies have demonstrated the deleterious effects of mutations in the Cterminal domain of CaM on the maximal activity of target enzymes (29,30).
The large increases in the activation constants (K CaM ) of the CaM mutants M109Q and M124Q on the three protein kinases suggest that individually Met-109 and Met-124 are involved in generating a high affinity interaction between the CaM Cterminal domain and different target enzymes. Both Met-109 and Met-124 are spatially adjacent residues located on a solvent-accessible surface of the hydrophobic pocket in the Cterminal domain of Ca 2ϩ /CaM (2). The magnitude of the changes in K CaM for M109Q and M124Q is consistent with a local effect, as opposed to extensive changes in the secondary and tertiary structures of the proteins, since the largest differences in free energy calculated between the wild type and mutant proteins (⌬⌬G ϭ 2.00 -2.38 kcal/mol) is in the range of the calculated difference in the free energy of transfer/solvation between Met to Gln (⌬⌬G ϭ 1.98 -2.62 kcal/mol) (31,32). This agrees with the lack of effect on the remainder of the protein structure by these mutants when compared to wild type CaM by CD and fluorescence spectroscopy. Several other lines of evidence support a primary role for the C-terminal domain of CaM in the high affinity binding of target enzymes. Studies on proteolytic fragments comprising either the N-or C-terminal domains of CaM (residues 1-77 or 78 -148) showed that the C-terminal domain has a higher affinity for target enzymes (33) and is the only domain capable of activating some enzymes (34,35). Also, chimeras generated by domain swaps between CaM and other E-F hand proteins implicated the C-terminal domain of CaM in promoting a high affinity interaction with target enzymes (25,36).
The interaction of CaM Met-124 with a conserved, large hydrophobic residue in the CaM binding domains of different enzymes suggests these two complementary hydrophobic residues help to define the specificity of CaM for CaM-binding proteins. Since changing the nonpolar nature of CaM Met-124 led to a loss in affinity for the three enzymes, then the CaM binding domains of these enzymes might contain a complementary hydrophobic residue that interacts with Met-124. Although CaM-binding sequences are notoriously divergent, a conserved bulky hydrophobic side chain, which is absolutely required for CaM binding and activation (37,38), is present at the same position in the N-terminal half of the CaM binding domains of five CaM-dependent kinases (Fig. 4a) and all known Ca 2ϩ /CaM-binding proteins (39). It is noteworthy that the results from x-ray crystallography and NMR spectroscopy show that Met-124 interacts with this conserved, large hydrophobic side chain in three CaM-dependent kinases: Leu-299 in CaMKII␣, Trp-800 in smMLCK (Table II), and Trp-580 in skeletal muscle myosin light chain kinase (skMLCK) (5).
An important insight into how CaM is able to recognize this one residue has been provided by the first x-ray crystal structure of a CaM-dependent protein kinase, Ca 2ϩ /calmodulindependent protein kinase I (CaMKI) (40). In this structure the conserved hydrophobic target residue of CaMKI, Trp-303, (Fig.  4a) juts away from the enzyme and into the solvent. This presents an energetically inviting target to the solvent-exposed hydrophobic surface of the CaM C-terminal domain, including Met-124, and thus might allow an initial grip on the enzyme. Since the major effect of M124Q was on the affinity of CaM for all three kinases, we propose that the interaction between this conserved, hydrophobic "target" residue on the CaM-binding enzymes and the hydrophobic patch on CaM, which includes Met-124, is a significant factor in defining the specificity between CaM and CaM-binding proteins. Support for this proposal comes from the activation of a glutathione S-transferase fusion protein of CaMKI with the M124Q mutant of CaM, which has a greater than 40-fold higher activation constant than wild type CaM. 2 The x-ray crystallographic and NMR structures for CaM bound to the peptides of either smMLCK, skMLCK, or CaMKII␣ also reveal that the hydrophobic target residue (Trp or Leu) of these three enzymes each interact with Leu-105, Met-124, and Met-144 of CaM, respectively. This implies that CaM has maintained a complementary hydrophobic surface in order to recognize its enzyme partners. It is interesting to observe that whereas Met-144 can be conservatively substituted by either Leu or Val in CaM from yeast to mammals, the other two residues of CaM which interact with the proposed target residue, Leu-105 and Met-124, are evolutionarily invariant. Indeed both Met-124 and Met-144 of CaM cross-link with model CaM-binding peptides at sequence positions where a photoreactive p-benzoylphenylalanine was substituted in place of the corresponding Trp or Leu target residue (41).
Recent NMR structures of Ca 2ϩ free-CaM demonstrate that unlike the 6 other methionines located in the two domains, Ca 2ϩ binding does not alter the relative exposure of Met-144 to solvent and Met-124 still retains 40% of its Ca 2ϩ -dependent solvent accessibility (2). These results suggest the intriguing possibility of interactions at low levels of Ca 2ϩ occupancy, between this region of the CaM C-terminal domain and the exposed hydrophobic target residue of CaM-binding enzymes. Indeed studies on the Ca 2ϩ dependence of the interactions between CaM and target proteins suggest that at resting free Ca 2ϩ concentrations, CaM is still capable of binding to either the protein phosphatase calcineurin or smMLCK without activating these enzymes (42,43). Such conditions may be sufficient to promote interactions between CaM and enzymes, since at low concentrations of free Ca 2ϩ , the initial binding of CaMbinding peptides increases the affinity of the CaM C-terminal calcium-binding loops for Ca 2ϩ (44 -46). This hypothesis is currently under investigation using the individual Met mutants.
In conclusion the picture that emerges from an overall analysis of the effects of Met to Gln point mutants on CaM-dependent activation of protein kinases is presented in Fig. 4 (b-d). First, at low or resting concentrations of intracellular free Ca 2ϩ , we propose that a group of solvent-accessible nonpolar residues in the C-terminal domain of CaM, which includes Met-124, initially interacts with a large, solvent-exposed hydrophobic side chain located at the N-terminal end of the CaM binding domain of the target enzyme (Fig. 4b). This enables the remainder of the C-terminal hydrophobic domain of CaM to bind to the N-terminal end of the CaM binding domain (Fig.  4c). In some special cases, the interaction with the C-terminal domain of CaM may be sufficient to displace the enzyme autoinhibitory regions and allow enzyme activation to proceed. However, most CaM-dependent enzymes require both domains of CaM for full activity and would require an additional step involving an increase in free Ca 2ϩ concentration (Fig. 4d). The excess Ca 2ϩ would occupy both the C-and N-terminal Ca 2ϩbinding sites of CaM. Finally with the aid of the flexible domain linker of CaM (47,48), the N-terminal domain of CaM participates in additional interactions with the enzyme, thus completely exposing the active site, which leads to full enzyme activity.