Functional Significance of the Central Helix in Calmodulin*

The 3-A crystal structure of calmodulin indicates that it has a polarized tertiary arrangement in which calcium binding domains I and I1 are separated from domains I11 and IV by a long central helix consisting of residues 66-92. To investigate the functional significance of the central helix, mutated calmodulins were engineered with alterations in this region. Using oligonucleotide-primed site-directed mutagenesis, Thr-79 was converted to Pro-79 to generate CaMPM. CaMPM was further mutated by insertion of Pro-Ser- Thr-Asp between Asp-78 and Pro-79 to yield CaMIM. Calmodulin, CaMPM, and CaMIM were indistinguish- able in their ability to activate calcineurin and Ca2+-ATPase. All mutated calmodulins would also maxi- mally activate cGMP-phosphodiesterase and myosin light chain kinase, however, the concentrations of CaMPM and CaMIM necessary for half-maximal activation (ICaet) were 2- and 9-fold greater, respectively, than CaM23. Conversion of the 2 Pro residues in CaMIM to amino acids that predict retention of helical secondary structure did not restore normal calmodulin activity. To investigate the nature of the interaction between mutated calmodulins and target enzymes, syn- thetic peptides modeled after the calmodulin binding region of smooth and skeletal muscle myosin light chain kinase were prepared and used as inhibitors of calmodulin-dependent cGMP-phosphodiesterase. The data suggest that the

calcium binding proteins which includes troponin C, parvalbumin, and calbindins. Some of these calcium binding proteins, such as CaM and troponin C, serve as tranducers of calcium signals. Both of these proteins modulate the function or activity of target proteins via calcium-dependent alterations in protein-protein interactions. At the molecular level, information required for transmission of the calcium signal is encoded by the spatial arrangement and dynamic properties of complementary recognition domains in the calcium-binding protein and its target protein.
The analogous mechanism of action of CaM and troponin C may have provided the selective pressure to maintain the very similar tertiary structure shared by these homologous proteins (1)(2)(3). Each protein consists of four calcium-binding domains that conform to the helix-loop-helix or EF-hand motif originally observed in parvalbumin (4). In both proteins, calcium binding domains I and I1 are separated from domains I11 and IV by a long a-helix located in the central region of the proteins. Two other calcium binding proteins with known crystal structures, parvalbumin and the 7.5-kDa form of calbindin, also contain EF-hand calcium binding domains but do not have the elongated dumbell shape of CaM and troponin C (4, 5). This suggests a basic difference in the mechanisms of action of these two types of calcium binding proteins. Indeed, parvalbumin and the 7.5-kDa calbindin have not been demonstrated to have activator activity. The similar tertiary motif of CaM and troponin C may be typical of a class of calcium "switch" proteins in which the specificity of the switch is defined by variations in protein recognition domains within this general structural framework. For example, the amino-terminal helix in troponin C, which is absent in CaM, may contribute to functional divergence.
The structure of CaM and how it encodes functional information is of particular interest since CaM regulates a variety of proteins and enzymes. A series of studies using biochemical (6-9) and protein engineering (10-13) techniques have shown that CaM contains multiple functional domains that selectively interact with target enzymes. In this report we have mutated the central helix of CaM to investigate its role as a structural element and a protein-protein interactive site and show that its length and not composition appears more important for the activation of selected enzymes. The data are also consistent with our previous classification of CaM-dependent enzymes based on their interactions with bacterially synthesized CaM-like proteins (10, 11).

MATERIALS AND METHODS
Plasmid Construction-Mutation of CaM was accomplished using oligonucleotide-primed site-directed mutagenesis of the CaM expression plasmid pCaM23 (10) as outlined in Fig. 1. In step 1 a 274-base pair EcoRIIPstI fragment from pCaM23 was subcloned into the bacteriophage M13mp18 to obtain a single stranded DNA template molecule, phTemplate-1. In step 2, primer 1 was used to convert Ser-79 to Pro and also generate a BamHI for screening and as a site for 11242 insertion mutation. Site-directed mutagenesis was performed essentially as described by Zoller and Smith (14), 0.5 pmol of single stranded DNA template and 10 pmol of the mutagenic oligonucleotide were heated to 55 "C for 10 min and cooled slowly to room temperature in 10 p1 of annealing buffer containing 20 mM Tris, pH 7.5, 10 mM MgC12, 50 mM NaC1,l mM DTT. The annealed DNA was then diluted with 10 pl of 20 mM Tris, pH 7.5, 10 mM MgC12, 10 mM DTT, 1 mM ATP, 0.8 mM dNTPs, 1 unit of Klenow fragment DNA, 10 units of T4 DNA ligase. The reaction was incubated at 14-15 "C for 5 h, after which 1 r l was used to transform JM103 cells by the procedure of Hanahan (15). To screen for the desired mutation, step 3, the transformed population of bacteria was diluted to 5 ml with L-broth and incubated overnight at 37 "C. Bacteria from 1.5 ml of culture were collected by centrifugation and used to isolate replicative form DNA by the alkaline minilysate procedure (16). An aliquot of the mixed population of phage DNA was digested with BamHI, run on a 1% low gelling agarose gel (Bio-Rad), and stained with ethidium bromide, 0.5 pg/ml. Linearized DNA, which migrated faster than circular phage DNA, was identified and excised from the gel. The gel piece was diluted 5-fold with 10 mM Tris, pH 7.5, 1 mM EDTA, and heated at 70 "C with occasional vigorous mixing until melted. DNA in 2 r l of the melted agarose solution was circularized by ligation and used to transform JM103. Individual plaques were picked for the isolation of double and single-stranded DNA. After confirming the presence of predicted restriction endonuclease sites in double-stranded DNA the corresponding single stranded DNA was used as template for DNA sequencing by the dideoxy chain termination method (17). Recombinant phage containing the desired point mutation was called ph-CaMPM. In step 5, a 142-base pair AccIIPstI fragment from ph-CaMPM was subcloned into the corresponding sites of pCaM23 to yield phtermediate-1. To complete the amino acid coding region of CaMPM a 700-base pair PstI fragment from pCaM23 was subcloned into the unique PstI of phtermediate-1 to yield pCaMPM.
A CaM insert mutation was generated using cassette mutagenesis of pCaMPM as outlined in steps 5 and 6 of Fig. 1. pCaMPM, which has two BamHI sites, was partially digested with BamHI, dephosphorylated with bacterial alkaline phosphatase, and ligated with a BamHI fragment from pUC4K (Pharmacia LKB Biotechnology Inc.) that contains a kanamycin resistance marker within an inverted cloning cassette. Ligation products were used to transform JM109, and kanamycin-resistant colonies were picked for restriction endonuclease analysis of their plasmid DNA. The desired plasmid, phtermediate-2, was digested with SalI, re-ligated, and used to transform JM103 cells. The resulting plasmid called pCaMIM is identical to pCaMPM with the exception of an additional 12 nucleotides that encode 4 amino acids between Asp-78 and Pro-79 of CaMPM.
Steps 8-12 of Fig. 1 outline procedures that convert the 2 Pro residues in CaMIM to helix-forming amino acids. An EcoRI/PstI fragment from pCaMIM was first subcloned into M13mp18 to yield template DNA, phTemplate-2. Mutation of phTemplate-2 was performed in step 9 as described above using primer 2. In step 10, replicative form DNA was isolated from a mixed population of transformed bacteria, digested with BamHI, and used directly to transform JM103. Since replicative form DNA that have mutations in both Pro codons will be resistant to BarnHI digestion, they will remain circular and have a much higher efficiency of transformation. Phage ph-CaMIM-TQ, phCaMIM-KQ, and phCaMIM-QQ were identified first by restriction endonuclease analysis and then by DNA sequencing. In step 11 EcoRIIPstI fragments from the replicative form DNA of these three phage were subcloned into pCaM23 that had been digested with EcoRI and partially digested with PstI. In step 12 the CaMcoding region from phtermediates-3, -4 and -5 were isolated after digestion with EcoRI and partial digestion with PstI. These fragments were subcloned into pCaMPL which had been digested with PstI and partially digested with EcoRI. pCaMPL is a derivative of pCaM23N (18) in which the tac promoter has been replaced with a heat-sensitive P L promoter (19).
Protein Isolation and Enzyme Assays-Bacterially synthesized CaMs were isolated by phenyl-Sepharose chromatography as described previously (10, 18). As an additional purification step the isolated proteins were bound to a Waters DEAE-5PW high performance liquid chromatography column and eluted with a NaCl gradient in a buffer of 50 mM Tris, pH 7.5, 0.2 mM EDTA. For those experiments where a change of buffer was required, the proteins were either desalted into the appropriate buffer by gel filtration using Sephadex G-25, or subjected to 6-8 successive rounds of concentration and dilution using Amicon Centricon P-10 microconcentraters.
For cGMP-phosphodiesterase assays, CaM23 concentration was determined by amino acid analysis, and the concentration of CaM mutants was determined by the method of Bradford (20) using CaM23 as a standard. The concentration of CaM binding peptides was determined by absorption at 278 nm using molar extinction coefficients of 5556 and 5554 for skeletal and smooth muscle isoforms of myosin light-chain kinase, respectively. CaM-deficient phosphodiesterase was purified from bovine brain by a modification of the method of Sharma et al. (21) up to the CaM-Sepharose 4B chromatography step. The eluant from the CaM-Sepharose 4B column was dialyzed overnight against 20 mM Tris-HCl, pH 7.5, 1 mM magnesium acetate, 10 mM 2-mercaptoethanol, 100 mM NaCl. Bovine serum albumin was added to a final concentration of 1 mg/ml, and the enzyme preparation was frozen in liquid nitrogen until use. The purified phosphodiesterase was stimulated 10-fold by saturating concentrations of CaM.
Phosphodiesterase was assayed by a modification of the procedure of Thompson et al. (22). The reaction was performed in a volume of 0.25 ml containing 50 PM phosphodiesterase and varying concentrations of CaM in a buffer of 40 mM Tris-HCI, pH 8.0,5 mM magnesium acetate, 1 mM CaC12,0.03 mM cGMP, 0.15 pCi of [8-3H]cGMP, 1 mM DTT, 640 pg/ml bovine serum albumin. All reactions were carried out in polypropylene tubes. The reaction was initiated by addition of cGMP, incubated at 30 'C for 40-60 min, and terminated by boiling for 3 min followed by the addition of 25 p1 of 1 mg/ml snake venom (Crotalus atrox, Sigma) and an additional incubation at 30 "C for 10 min. After addition of 25 pl of 10 mM guanosine, unreacted cGMP was adsorbed by the addition of 0.5 ml of a 50% (v/v) AG 2-X8 resin in 30% ethanol. The slurry was separated by centrifugation (850 X g for 10 rnin), and 200 p1 of the supernatant was counted in 10% Beckman BioSolve-Spectrofluor. Percent activation was calculated as described previously (23), and Kapp was calculated from a Hill plot of the data.
ICso values for the CaM-binding peptides were determined by a competitive enzyme assay using phosphodiesterase. The reaction mixture contained appropriate concentrations of the binding peptide, CaM (either 1.7 nM CaM23; 3.4 nM CaMPM, or 16.5 nM CaMIM), 20% glycerol, and the phosphodiesterase reaction mixture as described above.
Chicken gizzard myosin light chains and myosin light chain kinase were isolated as described previously (24,25). Myosin light chains were chromatographed on phenyl-Sepharose CL-4B to reduce contaminating CaM (26). The assay was performed in a 0.1-ml volume containing 50 mM HEPES, pH 7.6, 100 mM KCl, 10 mM MgC12, 1 mM CaC12,l mM DTT, 0.2 mM ATP (6-8 X lo6 cpm of [y3'P]ATP), 0.05 mg/ml bovine serum albumin, 0.02 mM myosin light chains, 8 X lo-' mM myosin light chain kinase, and the indicated amount of activator. The mixture was incubated at 30 "C for 20 min, the reaction was terminated and incorporation of 32P into myosin light chains was determined as described previously (27).
Calcineurin was isolated and assayed as described previously (6,28). The assay was conducted at 28 "C and contained 20 mM Tris, pH 8.0, 100 mM NaCl, 6 mM MgCl,, 0.5 mM CaCl,, 0.5 mM DTT, 5 mM p-nitrophenyl phosphate, 0.1 mg/ml bovine serum albumin, lo-' M calcineurin, and the indicated amount of activator. Maximal stimulation of enzyme activity in the presence of saturating amounts of activator was 18-fold over basal activity in the presence of EGTA and 8-fold over the activity in the presence of Ca2+ but no activator. Erythrocyte ATPase was isolated and assayed as described by Niggli et al. (29).
Spectral Measurements-Tyrosine fluorescence measurements were made using an Aminco SPF-BOO Ratio Spectrofluorimeter. Proteins were desalted into a buffer of 50 mM Tris, pH 7.5, 150 mM NaC1, 0.2 mM EGTA and adjusted to a protein concentation of 0.2 mg/ml. Calcium standard solutions of 100, 20, 5, and 2 mM CaC12 were prepared from a 100 mM CaCl, standard (Orion). Calcium from the standards was added sequentially in 2-pl aliquots to 2 ml of protein solution. Addition of the four standards was selected to achieve a uniform increase in tyrosine fluorescence. After completion of the titration, the volume change due to calcium addition was less than 3%. Free calcium concentrations were computed based on the total calcium added and the calcium dissociation constants of EGTA (30).
CD measurements were taken on a JASCO-500 Spectropolarimeter. Proteins were dialyzed against 50 mM HEPES, pH 7.5, 0.1 mM EGTA and adjusted to an A278 of 0.11. Samples were made 100 mM in KC1 and scanned from 260 to 185 nm in the presence or absence of 1 mM CaCI2.

RESULTS
The amino acid changes introduced into CaM by the procedures outlined in Fig. 1 are summarized in Fig. 2, panel A. Bacterially synthesized CaM23 has the sequence of vertebrate CaM and is expressed from a plasmid containing the chicken calmodulin cDNA. Despite the absence of an acetylated amino terminus and trimethylation of Lys-115, CaM23 has been shown to be physically and functionally identical to naturally occurring calmodulin by all criteria tested thus far (10, 11) and is used as a control in this study. CaMPM contains a single point mutation in which Thr-79 is changed to a Pro residue. CaMIM is a derivative of CaMPM and has Pro-Ser-Thr-Asp inserted between Asp-78 and Pro-79. CaMIM-T (Thr), CaMIM-TQ (Thr and Gln), and CaMIM-KQ (Lys and Gln) are all derivatives of CaMIM in which one or both Pro residues are converted to the indicated amino acids. Multiple substitution of the Pro residues was accomplished using a mixture of oligonucleotide primers. The oligonucleotide was designed to replace the second Pro with Thr, Gln or Lys; however, only Gln was obtained. This probably reflects a Although necessarily imprecise, secondary structure in the mutant proteins was predicted by the method of Gamier et al. (31) and is summarized in Fig. 2, panel B. This calculation assigns a numerical value for the probability that a given amino acid will be in either a random coil (C), turn (T), 0sheet (S), or a-helix ( H ) secondary structure. The program accurately predicts the non-helical regions of CaM23 which constitute the four Ca2+ binding loops and the non-helical region that connects domains I and 11. The central region in CaM23 is assigned a helical configuration; however, the numerical value for Thr-79 indicates that this helix is not strongly favored. Analysis of the central region of CaMIM indicates that Lys-77 through Asp-80, including the 4-amino acid insertion, probably assume a non-helical conformation in the protein. Conversion of the Pro residues to either polar or charged amino acids predicts a more mild disruption of the a-helix with CaMIM-KQ predicted to retain an a-helical secondary structure across the insertion mutation. Similar to

FIG. 3. Electrophoretic mobility of calmodulin.
Purified proteins were resolved on 15% sodium dodecyl sulfate-polyacrylamide gel electrophoresis gels (29.2% acrylamide to 0.8% bisacrylamide) using buffers described by Laemmli (32). Proteins were solubilized in sodium dodecyl sulfate sample buffer containing either 5 mM EDTA (-Calcium) or 5 mM CaC12 (+Cafciurn). The gel and running buffers contained no added EDTA or CaC12. The relative molecular mass of the standard proteins (Std) are listed on the left in kilodaltons.
CaM23, helical conformation of the central region in CaMIM-KQ is not strongly favored.
The purity and electrophoretic mobility of the mutant CaMs is shown in Fig. 3. All proteins exhibited an equivalent Ca2+-dependent decrease in apparent molecular weight. Proteins with insert mutations migrated slightly slower than CaM23 and CaMPM in the presence and absence of Ca2+, consistent with an additional 4 amino acids. CaMIM-T, CaMIM-TQ, and CaMIM-KQ all co-migrated with CaMIM in the presence and absence of calcium (data not shown).
In an effort to detect potential differences in secondary structure, calcium-dependent changes in tyrosine fluorescence and CD spectra were compared for CaM23, CaMPM, and CaMIM. Fig. 4 shows that intrinsic tyrosine fluorescence is not affected by the amino acid changes in the central helix of CaM. This was not unexpected since the 2 tyrosine residues in CaM are located in domains I11 and IV at positions 99 and 138. Fig. 5, panels A and B, show the CD spectra of CaM23 and CaMPM to be indistinguishable in the presence and absence of calcium. Panels C and D show the magnitude of the mimima in the spectrum for CaMIM is about 10% greater relative to CaM23. This suggests that the sum of secondary structures between these two proteins is different but does not identify the nature of the difference. A recent paper by Hennessey et al. (33) suggests that, at high ionic strength, calcium-dependent changes in the secondary structure of CaM are due to a reorientation of helices rather than an increase in helical content. The differences in spectra between CaM23 and CaMIM may reflect general differences in the organization of secondary structures. Fig. 6 shows the activation of calcineurin and Ca2+-ATPase by CaM23, CaMPM, and CaMIM. Activation characteristics of both enzymes were unaffected by mutations in the central helix of CaM. In contrast, Fig. 7 shows that the activation characteristics of cGMP-phosphodiesterase and myosin light chain kinase are both influenced by alteration in the central helix of CaM. A summary of multiple experiments with cGMP-phosphodiesterase and myosin light chain kinase is shown in Table I. Although both enzymes are maximally stimulated by all bacterially synthesized CaMs, the concentrations of CaMPM and CaMIM necessary for half-maximal activation (Kaa) are approximately 2-and 9-fold greater, respectively, than CaM23. Although the apparent ICad values from two experiments with myosin light chain kinase differ, the relative values shown in brackets in Table I are very similar. Inter-assay variability is most likely due to effects of storage on enzyme and substrate.
The aberrant ability of CaMIM to activate cGMP phosphodiesterase and myosin light chain kinase could be due to a disruption of secondary structure or a lengthening of the central region in CaM by 4 amino acids. To approach this   Activation constants for activation of cGMP-phosphodiesterase and myosin light chain kinase by CaM23, CaMPM, and CaMZM The following numbers represent the apparent K,,, (nanomolar) for enzyme activation by the indicated CaM. K,,, is defined as the amount of CaM required for half-maximal activation under standard assay conditions. The values for phosphodiesterase are the average f S.D. of n determinations (in parentheses). The values for myosin light chain kinase represent two separate experiments and the error values are derived from computer fits of the data. The numbers in brackets are the -fold increase in Kact for CaMPM and CaMIM relative to CaM23.  9. Inhibition of cGMP-phosphodiesterase activity by the CaM-binding peptide (SrnK) from smooth muscle myosin light chain kinase. The assay was performed as described under "Materials and Methods" but in the presence of varying concentrations of inhibitor peptide. Activator, enzyme, peptide, and calcium were added together prior to the addition of substrate. The amount of activator in each assay is shown in Table I1 and was adjusted to yield equivalent initial enzyme activities. The data shown is for inhibition by smooth muscle kinase. ICs0 values for both smooth and skeletal muscle kinase (SkK) are given in Table 11.
question, a series of mutants were generated in which 1 or both of the introduced Pro residues were changed to charged or polar amino acids (Figs. 1 and 2) in an attempt to retain helical secondary structure. Fig. 8 shows that cGMP-phosphodiesterase is activated identically by CaMIM, CaMIM-T, CaMIM-TQ, and CaMIM-KQ. Although secondary structure in the mutated CaMs may differ from computer predictions, the data demonstrate that the functional abnormalities of CaMIM are not due to the disruptive influence of Pro residues.  The activation of target enzymes by CaM involves complex macromolecular interactions including the binding of CaM to the enzyme. To investigate the nature of the interaction between mutated calmodulins and myosin light chain kinase, peptides modeled after the CaM-binding sites in both smooth and skeletal muscle isoforms of myosin light chain kinase (34,35) were synthesized and used as inhibitors of activation of cGMP-phosphodiesterase. Dose-response curves for the inhibition of cGMP-phosphodiesterase by smooth muscle kinase are shown in Fig. 9 and IC5o values for both smooth and skeletal muscle kinase are summarized in Table  11. In all experiments the concentrations of CaM23, CaMPM, and CaMIM were adjusted to achieve equivalent initial enzyme activities. Although the ICso values in Table I1 differ for the three CaMs, the numbers are proportional to the concentration of activator present in the assay and the K,,, values shown in Table I. When corrected for varying CaM levels, the relative ICso values show no apparent difference. We would interpret these data to indicate that the synthetic peptides interact similarly with the three CaM proteins.

DISCUSSION
A long central helix has been observed in both CaM and fast skeletal muscle troponin C from structural analysis or the crystal strucutes (1)(2)(3). Although thermodynamic arguments do not favor the stability of an eight-turn a-helix that is exposed to solvent, experimental evidence not only support the existence of the central helix but also calcium-dependent conformational changes in this region. It is proposed that solvent-exposed helices in troponin C are stabilized by intrahelical electrostatic and salt-bridge interactions between the basic and acidic amino acid side chains (36). Vacuum-UV CD measurements in the presence of calcium predict a helical content for CaM that is in good agreement with the crystal structure (33), and they predict that binding of calcium by CaM a t physiologic ionic strength induces a reorganization of the helices rather than a change in helical content. Fluorescence anisotropy measurements on the dityrosine derivative of CaM in the presence of high calcium concentrations suggest that protein has a length equivalent to that predicted by the crystal structure (37) and that in the absence of calcium the protein appears more compact and exhibits segmental motion.
Dynamic changes in the central helical region of CaM is an attractive model that would explain not only spectral data but also calcium-dependent differential accessibility of this region to both proteases (38) and acetic anhydride (39). Dynamic changes could involve a collapse of the amino-and carboxyl-terminal lobes to shield the central helix as might be predicted from the crystal structure of troponin C and/or an increase in the conformational flexibility of this region. These are attractive models; however, it must be appreciated that structural studies of isolated CaM may not fully represent its biologically active conformation when complexed with its target enzymes. For example, binding of melittin and peptides derived from myosin light chain kinase induce conformational changes in both halves of CaM as determined by NMR analysis (40, 41) and binding of CaM to target proteins increases its affinity for Ca2+ (42). This presents the possibility that interactions of CaM with target enzymes may stabilize conformations that do not predominate in solution. Crystallization of CaM may approximate interactions with target enzymes.
Assuming that the central helix in CaM does exist either in the Ca2+-bound form of the isolated protein or when complexed with a target protein, at least two possible functions for this structure can be hypothesized. First, the helix may maintain a proper linear and/or rotational orientation between the amino-terminal and carboxyl-terminal lobes such that recognition sites in these regions can functionally interact with complementary sites in target proteins. Alternatively, the central helix may encode recognition sites that are rendered accessible to target enzymes by Ca2+. We have attempted to investigate these mechanisms by generating a series of mutated proteins in which the length and composition of the central helix in CaM is altered.
CaMIM, CaMIM-T, CaMIM-TQ, and CaMIM-KQ all have an insertion in the central helix of 4 amino acids but with varying degrees of predicted disruption of the helix. All four mutants exhibit identical functional abnormalities with respect to activation of phosphodiesterase. Therefore, functional characteristics of these activators are not due to the Pro residues. Aberrant activation of myosin light chain kinase by CaMIM does not appear to result from a disruption of a recognition site due to insertion of amino acids since CaMbinding peptides derived from both the smooth and skeletal kinases have similar affinities for CaM and CaMIM. Therefore, the length and not composition of the central helix appears more important for the activation of certain target enzymes. A similar conclusion was also reached for skeletal muscle troponin C by Reinach and Karlsson (43) who showed that conversion of Gly-92 to Pro did not alter the properties of troponin C in a reconstituted actomyosin ATPase assay. A requirement for a specified length of the central helix in troponin C has yet to be investigated by mutagenesis.
Using an analogous approach, Craig et al. (13) have reported that conversion of Glu 82-84 in the central helix to lysines effectively inhibits the ability of the mutated CaM to activate myosin light chain kinase and plant NAD kinase while the activation of phosphodiesterase is minimally affected. Although the Ca2+-dependent CD spectra of control and Lyssubstituted CaM are quite different, suggesting that this considerable change in local charge density has distal effects on protein secondary structure, this acidic cluster may represent a recognition site that is more important for activation of the former two enzymes. If so, this recognition site does not appear to overlap Thr-79 since CaM-binding peptides from smooth and skeletal muscle myosin light chain kinase show the same relative affinity for CaM23 and CaMIM. Together, these results suggest that the central helix contributes to the functional characteristics of CaM by both providing sites of protein recognition and maintaining either a proper orientation or linear relationship between the lobes in CaM, and that the requirement for these structural features vary for different CaM-dependent enzymes.