Features of Calmodulin That Are Important in the Activation of the Catalytic Subunit of Phosphorylase Kinase*

Calmodulin (CaM) is an integral subunit, called 6, of the phosphorylase kinase hexadecamer, and the activity of the isolated catalytic y-subunit of the kinase is stimulated by CaM. We report here the first analysis of functionally important features of CaM for activa- tion of the y-subunit. A set of genetically engineered CaMs, in which acidic residues in each of the four E-helices of the “EF-hands” were changed to basic lysine residues, was used to probe the relative importance of charge features in each domain of CaM. The maximal activation of the isolated y-subunit was diminished by all of the charge reversal mutations. The y-subunit was especially sensitive to reversals in the second and third E-helix of CaM (residues 45-47 and 82-84), the latter being present in the central helix. The results suggest the functional importance of electrostatics in the interactions between the &subunit (CaM) and the catalytic y-subunit of phosphorylase kinase, which is similar to results obtained with CaM-dependent myosin light chain kinase (MLCK) from chicken gizzard and CaM-dependent protein kinase I1 (CaMPK-11). However, novel features of the interaction between CaM and the y-subunit of phosphorylase kinase are the significant contribution of electrostatics throughout the CaM mol- ecule, including residues in both halves and on more than one face of CaM, and the lack of a major effect of

Calmodulin (CaM) is an integral subunit, called 6, of the phosphorylase kinase hexadecamer, and the activity of the isolated catalytic y-subunit of the kinase is stimulated by CaM. We report here the first analysis of functionally important features of CaM for activation of the y-subunit. A set of genetically engineered CaMs, in which acidic residues in each of the four Ehelices of the "EF-hands" were changed to basic lysine residues, was used to probe the relative importance of charge features in each domain of CaM. The maximal activation of the isolated y-subunit was diminished by all of the charge reversal mutations. The y-subunit was especially sensitive to reversals in the second and third E-helix of CaM (residues 45-47 and 82-84), the latter being present in the central helix. The results suggest the functional importance of electrostatics in the interactions between the &subunit (CaM) and the catalytic y-subunit of phosphorylase kinase, which is similar to results obtained with CaM-dependent myosin light chain kinase (MLCK) from chicken gizzard and CaMdependent protein kinase I1 (CaMPK-11). However, novel features of the interaction between CaM and the y-subunit of phosphorylase kinase are the significant contribution of electrostatics throughout the CaM molecule, including residues in both halves and on more than one face of CaM, and the lack of a major effect of the CaM mutations on substrate kinetic parameters, unlike the effects observed with MLCK and CaMPK-11. These results are consistent with a model in which the &subunit (CaM) of phosphorylase kinase interacts with an extended region or multiple regions of the ysubunit and suggest that the mechanism of CaM activation of the y-subunit may have features that are distinct from those of MLCK and CaMPK-11. Skeletal muscle phosphorylase kinase is a calmodulin (CaM)'-regulated protein kinase in which CaM is a tightly  and DK-32953 (to G. M. C.). The costs of * This work was supported by National Institutes of Health Grants publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Current address: Dept. of Neurology, Veteran's Administration Medical Center, Lexington,KY 40511. 11 To whom correspondence should be addressed: Dept. of Biochemistry, College of Medicine, University of Tennessee, 800 Madison Ave., Memphis, T N 38163.
The abbreviations used are: CaM, calmodulin; BB, bovine brain; BSA, bovine serum albumin; CaMPK-11, CaM-dependent protein kinase 11; MLCK, myosin light chain kinase; VU, Vanderbilt University genetically altered calmodulin(s). bound intrinsic subunit, termed the b-subunit, of the hexadecameric holoenzyme, (c~(3yS)~ (Cohen et al., 1978). Unlike other CaM-regulated protein kinases, such as the CaM-dependent myosin light chain kinase (MLCK), in which the CaM subunit can be readily dissociated by Ca2+ chelators and chromatographic fractionation, the &subunit of phosphorylase kinase remains associated with the holoenzyme even in the presence of high concentrations of chelators and column chromatographic fractionation (Cohen et al., 1978). Within the holoenzyme, the b-subunit has been shown to interact with the catalytic y-subunit (Picton et al., 1980), and the activity of the free isolated y-subunit is stimulated by CaM and made Ca2+-sensitive (Kee and Graves, 1986).
As first suggested by Reimann et al. (1984), the region of the y-subunit that interacts with CaM has been localized to its COOH-terminal quarter by proteolysis, cross-linking, and peptide analog studies. Proteolytic removal of the COOHterminal quarter of its sequence results in a truncated ysubunit that is active but is no longer sensitive to Ca2+/CaM (Harris et al., 1990). Moreover, a derivatized CaM has been shown to chemically cross-link within the COOH-terminal three-eighths of the y-subunit (James et al., 1991). Screening of the COOH-terminal quarter of the y-subunit for potential high affinity CaM binding sequences through the use of overlapping peptides revealed two distinct foci of CaM binding activity (Dasgupta et al., 1989). In summary, although the effects of site-specific mutageneis of the proposed CaM-binding region of the y-subunit have not been reported yet, there is an extensive body of literature indicating that the physiologically relevant CaM-binding domain of this subunit occurs within its COOH-terminal region. The two foci within this region that have been shown to interact most avidly with CaM are both enriched in positive charges (Dasgupta et al., 1989), and one is highly homologous to the basic inhibitory peptide of troponin I that interacts with the relatively acidic proteins actin and troponin C (Buschmeier et al., 1987;Dasgupta et al., 1989;Paudel and Carlson, 1990). These results suggest that positive charge features of the y-subunit may be functionally important in its interaction with CaM.
In contrast to the accumulating information regarding the interacting region(s) of the y-subunit, there have been no previous reports concerning the regions of the &subunit (endogenous CaM) that may be important in its interaction with the y-subunit. One approach that has been used to probe the functional importance of protein electrostatic properties in CaM-enzyme recognition for two other classes of protein kinases, the CaM-dependent myosin light chain kinases and protein kinase 11, is the use of site-specific mutagenesis and protein engineering combined with enzymological studies of purified proteins (for examples see Craig et al. (1987), Weber et al. (1989), Shoemaker et al. (1990, Herring (1991), Van-Berkum and Means (1991)). Therefore, we have used this approach to study activation of the y-subunit by genetically engineered CaMs. The results summarized here represent the first report of an analysis of the features of CaM functioning as the &subunit and are focused on assessing the relative functional importance of CaM's electrostatic properties in activation of the y-subunit. The results are consistent with a model (Craig et al., 1987;Weber et al., 1989) in which the electrostatic properties of CaM are important in protein kinase recognition and regulation. However, the results demonstrate that the functionally important electrostatic features of CaM for regulation of the y-subunit are spread throughout both lobes and more than one face of CaM, suggesting either more than one CaM binding segment or an extended CaM binding segment in the y-subunit of phosphorylase kinase.

Enzymes, Proteins, and
Peptides-Phosphorylase kinase from white skeletal muscle of rabbits, its free catalytic y-subunit, and phsophorylase b were prepared and treated exactly as previously described (Farrar and Carlson, 1991). The synthetic tetradecapeptide substrate SDQEKRKQISVRGL corresponding to the convertible region of phosphorylase b was purified as before (Farrar and Carlson, 1991). Myosin light chain kinase was purified from chicken gizzards following the general procedures described previously (Adelstein and Klee, 1982;Lukas et al., 1986).
Mutant CaMs were constructed using the cassette mutagenesis technology (Roberts et al., 1985) and expressed and purified from Escherichia coli cells as described previously (Craig et al., 1987). The DNA sequence of all CaM constructs was verified. The purified proteins gave the expected amino acid compositions, and in selected cases the amino terminus of the protein was verified by amino acid sequence analysis. BB-CaM (P 2277) was from Sigma.
The concentrations of phosphorylase kinase, its catalytic y-subunit, and phosphorylase b were determined as before (Farrar and Carlson, 1991). The concentrations of the tetradecapeptide substrate, BB-CaM, and the VU-CaMs were determined by amino acid analyses.
Enzymatic Activity Assays-As previously described (Farrar and Carlson, 1991), the isolated y-subunit in 8 M urea, 0.1 M H3P04, and 0.1 mM EDTA at pH 3.3 was renatured overnight a t 4 "C in the presence of the various CaMs. The standard renaturation mixture contained y-subunit (10 pglml), CaM (varied), BSA (1.66 mg/ml), urea (0.8 M), H,PO, (10 mM), EDTA (0.1 mM), Ca*' (0.5 mM), dithiothreitol (0.3 mM), and Hepes (90 mM, pH 8.0) in a total volume of 100 pl. Any variations to this standard renaturation mixture, usually in the type and concentration of CaM, are noted in the appropriate figure legends. Before assay, the renatured y-subunit was diluted 10-fold with buffer (100 mM Hepes, 0.2 mM dithiothreitol, 0.5 mM Ca2+, 0.1 mM EDTA, 1 mg/ml BSA, pH 7.0). One part of this diluted mixture was added to three parts of assay mixture to start the reactions. The final concentrations in the assays were: y-subunit The above conditions were used for generating Figs. 1 and 3 and Table I; however, for the initial velocity kinetic studies of Table 11, the conditions were slightly different. The y-subunit was renatured as above except that the amounts of y-subunit and CaM in the renaturation mixture were fixed a t 20 and 200 pg/ml, respectively. T h e renatured y-CaM complex was diluted before assay and assayed according to the standard protocol described above, except that the concentration of the y-subunit was doubled over the standard assay and the concentrations of [y-32P]ATP and tetradecapeptide substrate were varied from 6-100 p M and 14-228 pM, respectively. The assay reaction mixture was preincubated for 2 min a t 30 "C before the reaction was started with the diluted y-CaM complex and allowed to incubate for 6 min a t 30 "C. Product formation was linear with respect t o time and the amount of enzyme, and less than 2% of the limiting substrate was consumed during the reactions.
CaM activation of myosin light chain kinase was performed as previously reported (Haiech et al., 1991). Briefly, the kinase substrate was the synthetic peptide KKRPQRATSNVFAM, an analogue of the phosphorylatable light chain of chicken gizzard myosin, at a concentration of 50 p~. The concentration of [y-'*P]ATP (Amersham Corp., PB10218) was 200 p~ (200-400 cpm/pmol). Reactions were carried out a t a kinase concentration of 1.4 nM in 50 mM Hepes, pH 7.5, containing 1 mM dithiothreitol, 5 mM MgC12, 0.1 mM CaCl,, and 0.1 mg/mI BSA.
Determination of Kinetic Constants-The Vmax and K, (amount of CaM required to achieve 50% VmaX) for activation of the y-subunit by the various CaMs listed in Table I were determined by nonlinear regression using the Enzfitter program (Elsevier-Biosoft). Because it was necessary to have the denaturant urea as a carryover component (0.8 M ) in the renaturation mixtures used in generating the values for Table I, it is possible that the measured KO values may he an overestimate. Nevertheless, their relative values allow comparison of the overall effectiveness of the various mutant CaMs to produce a functional y-CaM complex. The K, values for VU-12 and VU-18 were determined by their ability to inhibit y-subunit activated by BB-CaM using the method of Bhatnagar et al. (1983), assuming that VU-12 and VU-18 were competitive inhibitors of BB-CaM. The Michaelis constants for MgATP and tetradecapeptide listed in Table I1 were obtained using the FORTRAN program described by Cleland (1979) that makes a weighted least squares fit to the equation for a Bi Bi sequential mechanism (Farrar and Carlson, 1991).

RESULTS
The y-subunit was activated in parallel by BB-CaM and VU-1 ( Fig. l A ) , the parent chimeric CaM from which the other altered CaMs used in this study were derived. VU-1 (Roberts et al., 1985), also referred to as SYNCAM CaM (Weber et al., 1989), is a chimera of vertebrate (Watterson et al., 1980) and higher plant (Lukas et al., 1984) CaM sequences.
The indistinguishable effects of VU-1 and BB-CaM suggest that variations in the amino acid sequence among most naturally occurring CaMs would not be expected to dramatically alter their ability to activate the y-subunit. VU-1 has been observed previously to also be isofunctional with vertebrate CaM in activating MLCK (Craig et al., 1987) and CaMPK-I1 (Weber et al., 1989). Because phosphorylase b can bind CaM (Villar-Palasi et al., 1983;Slaughter and Means, 1985), throughout this study we used the synthetic tetradecapeptide SDQEKRKQISVRGL as substrate when comparing the activating abilities of the various altered CaMs. This peptide, which corresponds to the region surrounding the site of phosphorylation in phosphorylase b, has been shown previously to be a valid alternative substrate that mimics phosphorylase b in its phosphorylation by the CaM-activated, catalytic ysubunit of phosphorylase kinase (Farrar and Carlson, 1991). Moreover, mobility shift and phosphorylation assays failed to detect interactions between this peptide and CaM.' In this study we have used a charge perturbation approach to determine how selected mutations in CaM affect the activation of the y-subunit of phosphorylase kinase. This approach has been used previously to elucidate the importance of electrostatics in the CaM activation of other protein kinases (Weber et al., 1989;Shoemaker et al., 1990) and CaM-dependent enzymes (Craig et al., 1987;Kosk-Kosicka and Bzdega, 1991), and in the binding of basic drugs to CaM (Massom et al., 1991). Simple charge cluster changes have been shown not to have a dramatic effect on the overall secondary structure of CaM (Craig et al., 1987). Moreover, calculations of the asymmetric electrostatic potential on the surface of CaM suggest that only localized, rather than global, effects are caused by these charge reversals (Weber et al., 1989), which is consistent with the crystallographic findings of Alber et al. that charge perturbation mutations of CaM are selective in their functional effects, i.e. only certain CaM-dependent enzymes are affected by specific mutations, which further indicates that the structural effects of such mutations are localized (Craig et al., 1987;Weber et al., 1989;Kosk-Kosicka and Bzdega, 1991). Therefore, based on these positive precedents, the initial screening of CaM mutants for effects on the ysubunit of phosphorylase kinase focused on a set of charge cluster mutations in all four domains of the molecule. To facilitate comparisons among the various mutant CaMs, the activation results with the y-subunit have been grouped in Fig. 1 into three sets corresponding to alterations in the different parts of the CaM molecule.
Maximal activation of the y-subunit was diminished when acidic to basic substitutions were made in any of the four domains. In the first domain, VU-24 (DEQ6-8KKK) retained only 40% of the maximal activating ability of VU-1 CaM or BB-CaM ( Fig. 1A and Table I). In the second domain, VU-43 (EAE45-47KKK) retained only 12% of the activating capacity. Nearly the same low extent of activation was observed when only Glu-47 was replaced with Lys, VU-44 (E47K). In the third domain, VU-8 (EEE82-84KKK) showed only 15% of maximal activation (Fig. l B ) , with the change of the middle of the three glutamic acids (E83K) making the largest contribution to the diminished activation; the single substitutions of E82K, E83K, and E84K resulted in maximal activations of 35, 20, and 39%, respectively ( Fig. 1B and Table I). In the fourth domain, the single substitution of E120K (VU-29) caused an approximate 50% decrease in the maximal activation, and further substitution of the entire acidic DEE triad at residues 118-120 (VU-12A) did not result in any further loss of activation ability ( Fig. 1C and Table I). When the two acidic triads in the third and fourth domains were simultaneously substituted (EEE82-84KKK plus DEE118-120KKK), the approximate 6% maximal activation retained by the resultant VU-18A CaM was arithmetically predicted by the results obtained with the two separate component triad substitutions of VU-8 and VU-12A, i.e. 15% of 56% ( Fig. 1C and Table I). Thus, there was no apparent synergism in the interactions with the y-subunit of the two regions of CaM in question.
Even though the various mutated CaMs activated the ysubunit to quite different extents, the Michaelis constants for peptide substrate and for MgATP were relatively similar regardless of the CaM variant used for activation (Table 11). Overall, there was less than a %fold variation in the K,,, for either substrate when various CaMs were analyzed as activators of the y-subunit (Table I1 and data not shown). These data suggest that modulation of substrate binding does not account for the different extents of activation of the y-subunit by the altered CaMs, unlike the situation with another CaMregulated protein kinase, MLCK (discussed below). It must be noted, however, that the relatively low activity after renaturation of the y-subunit with some of the altered CaMs prevented determination of kinetic parameters for those complexes. In addition to giving rise to similar substrate K,,, values, the apparent K, values for half-maximal activation of the y-subunit by the various CaMs were also relatively similar ( Table I). As described under "Discussion," however, the K, for activation of the y-subunit by CaM is a complex term that may contain as many as three components; and as will become clear in the cases of VU-12 and VU-18, it is possible that an altered CaM may interact as tightly with the y-subunit as the fully activating native CaM does, but only induce partial activation. It is this last factor that we suspect is largely responsible for the differing extents of activation of the ysubunit by the various CaMs.
In addition to the different response in their K,,, values to the altered CaMs, the pattern of activation of the y-subunit by the charge mutant CaMs was also distinctly different from that observed with MLCK. In contrast to their effect on the y-subunit ( Fig. 1 and Table I), CaMs with charge cluster mutations in the A and C helices (VU-24 (DEQ6-8KKK) and VU-43 (EAE45-47KKK)) had little influence on the ability of these proteins to activate MLCK (Fig. 2). Similarly, the point charge mutant CaMs VU-39 (E82K) and VU-45 (E83K) were nearly indistinguishable from the parent VU-1 CaM in activating MLCK (Fig. 2). Thus, this set of charge mutants does not represent "denatured" CaMs that are incapable of interacting with target enzymes. Like the y-subunit, MLCK is sensitive to charge cluster mutations in residues 82-84 and 118-120 of CaM (Craig et al., 1987;Weber et al., 1989). More recently, it has been demonstrated that the single point mutants VU-28 (E84K) and VU-29 (E120K) perturb the ability of the altered CaMs to activate MLCK by increasing the apparent K,,, for peptide substrate by the CaM.MLCK complex (Shoemaker et al., 1990 and Table 11); these same mutants, however, do not dramatically change the K,,, for MgATP, which was found to be in the same range as that previously reported for this enzyme when a peptide was used as substrate (Table 11; Foster et al., 1986). The strong dependence on electrostatic properties in both halves of CaM for activation of the y-subunit and the lack of major effects on substrate kinetics by the charge mutant CaMs are features that distinguish its interaction with CaM from that of MLCK's.
The possibility of altered CaMs acting as antagonists toward the y-subunit was suggested by the low extent of acti-  Refers to the domain structure proposed by Babu et al. (1988) for the four E F hands with the central helix (D) being in both domains I1 and 111.
' Refers to the a-helix boundaries proposed by Babu et al. (1988): A, 5-19;B, 29-38;C, 45-55;D, 65-92;E, 102-111;F, 118-128 G, 138-147. 'Maximal activation as a percent of the maximal activation observed with BB-CaM. Unless stated otherwise this was determined by nonlinear regression using the Enzfitter program as described under "Materials and Methods." CaM concentration that gives half-maximal activation of the y-subunit. The K, , was determined using the Enzfitter program as described under "Materials and Methods." The concentration of CaM refers to its concentration in the renaturation mixture, as opposed to in the activity assay.
Inhibition constant, determined as described under "Materials and Methods." 'Because of the low extent of activation this was estimated directly from the saturation curve rather than determined by the Enzfitter program.
Ind, indeterminable, because of the low extent of activation. ND, not determined. " T h e renaturation and assays of the y-subunit with peptide as substrate were carried out as described under "Materials and Methods" for the protocol specific for initial velocity kinetic studies.
'The apparent K,,, values for MgATP and myosin light chain peptide analogue were determined by varying one substrate in the presence of a fixed concentration of the other. The fixed concentrations were the same as those used in the standard assay conditions as described under "Materials and Methods." From Shoemaker et al. (1990). ND, not determined. vation by VU-12 (Table I), which differs from its parent VU-12A only at position 124, where a Met has been changed to a n Ile. The resultant large decrease in the extent of activation from 56 to 6% going from VU-12A to VU-12 was especially noteworthy because VU-12 was the most effective activator of the charge mutant CaMs tested. The antagonist3 properties a For ease of discussion, we prefer to categorize as "antagonists" those essentially ineffective, altered CaMs that otherwise might be categorized as "weak partial agonists," with a partial agonist having a n efficacy between 0 and 1 when compared with wild type. Because CaMs. Assays were done as described under "Materials and Methods." The data are expressed as the percent of maximum activation observed with the wild-type CaM. Each point represents the average of simultaneous duplicates in the assay; bars represent the range of the data, with no bars indicating that the range was within the size of the symbol.

(DEQG-SKKK), (0); VU-43 (EAE45-47KKK), (A); VU-39
of VU-12 for the y-subunit became evident when activation experiments were performed a t a fixed subsaturating concentration of BB-CaM and with increasing amounts of mutant CaMs. The positive control for this series of experiments is VU-12 and VU-18 induce only extremely low amounts of activation, appear to bind to but not to renature and/or activate the y-subunit, and share a common structural feature (M1241), we think that they are most appropriately categorized as antagonists.
shown in Fig. 3A in which VU-1 and BB-CaM are compared. At the fixed concentration of BB-CaM chosen, activation of the y-subunit was 20% of maximum; when this same concentration of BB-CaM was included with different concentrations of VU-1, the activation was about 20% greater than that observed with VU-1 alone (compare VU-1 with VU-1 + BB in Fig. 3A). Similar additive activations were observed with the VU-CaMs 8, 28, 29, 39, 43, 44, and 45, i.e. they behaved as partial agonists (data not shown). When the effect of the essentially nonactivating VU-12 was compared with that of its activating VU-1ZA parent in the same type of experiment, contrasting results were obtained. Whereas VU-12A behaved as an agonist in additively augmenting the activation achieved with BB-CaM, VU-12 with the additional change of Met-124 t o Ile acted as an antagonist as illustrated by the decrease in activation observed with increases in its concentration (Fig.  2B). The competition between VU-12 and BB-CaM was characterized by a Ki for VU-12 of 1 ~L M (Table I). Analogous agonist uersus antagonist behavior was also observed with the VU-18A/VU-18 pair (Fig. 3C), with the latter also having the additional Met-124 to Ile change. The screening of additional CaM mutants for specific antagonist behavior toward the ysubunit is an ongoing effort to provide a tool for specific in uiuo inhibition of phosphorylase kinase; thus, VU-12 and VU-18 CaMs are initial examples of the feasibility of finding such species.

DISCUSSION
We conclude from the effects of CaM charge reversal mutations on the activation of the y-subunit (this report), MLCK (Craig et al., 1987;Weber et al., 1989;Shoemaker et al., 1990; this report), and CaMPK-I1 (Weber et al., 1989) that the electrostatic properties of CaM are functionally important in the regulation of CaM-dependent protein kinases but that there is an electrostatically variable component to CaM's interaction with each kinase. Further, the conversion of a partial agonist CaM mutant to an effective antagonist by a single hydrophobic amino acid change (this report) demonstrates that combinations of electrostatic and small hydrophobic surface changes can have rather dramatic effects on the interaction between CaM and the y-subunit. Maximal activation of the y-subunit was significantly diminished by alterations throughout the CaM molecule (Table I), suggestive of extensive areas of interaction between the two proteins, including the possibility of multiple sites of interaction. Therefore, the results reported here are consistent with the proposal of Dasgupta et al. (1989) that there are two adjacent, but discrete, CaM-binding regions within the COOH-terminal region of the y-subunit. Extensive interactions between the two proteins would also be consistent with the tight binding of the d-subunit (CaM) to the holoenzyme that occurs even in the absence of Ca2+ ions (Cohen et al., 1978). Because the bacterially expressed parent CaM, VU-1, is isofunctional with vertebrate CaM in activating the y-subunit but lacks the posttranslational modifications found in vertebrate CaM (Roberts et al., 1985), it is concluded that NHp-terminal acetylation and trimethylation of Lys-115 of CaM are not required for the interaction between the y-and &subunits.
Although the maximal activation of the y-subunit was diminished by negative to positive charge reversals throughout the CaM molecule (Table I), it was most sensitive to those alterations in domains I1 and I11 of CaM. For example, the cluster charge reversals at positions 45-47 in domain I1 and 82-84 in domain I11 diminished the maximal activation to only 12 and 15% of the control, respectively. Previous studies have shown (Weber et al., 1989) that CaM's calculated electrostatic potential surface is asymmetrically distributed, with concentrations of uncompensated negative charge localized in several regions of the molecule, and that pertubation of this surface by cluster charge reversal mutagenesis can alter CaM interactions with MLCK and CaMPK-11. Relevant to the studies reported here, two of the regions of uncompensated surface charge on CaM are near residues 47-54 and 83-85.
In addition to larger surface charge features such as that found with the computed electrostatic potential surface of CaM, point electrostatics can also contribute to specific interactions in a complex. The y-subunit is sensitive to perturbation of a set of discrete acidic residues that includes  In fact, the single charge reversals at positions 47 (VU-44) and 83 (VU-45) were nearly as effective in diminishing the activation of the y-subunit as the more extensive cluster charge reversals at the same and adjacent positions (VU-43 and VU-8). In contrast, E82K and E83K CaM mutations do not affect MLCK as single point mutants (Fig. 2), but when combined with the E84K CaM mutation (e.g. as in VU-8 CaM), MLCK activation is significantly poorer (Craig et al., 1987;Weber et al., 1989). Thus, the ysubunit of phosphorylase kinase and MLCK respond differently to the individual mutations of CaM, and the synergistic effects of charge mutations in a single domain of CaM observed with MLCK activation are not evident in the activation of the y-subunit.
Although much of the above discussion has focused on electrostatics, the contribution of hydrophobic effects to the protein-protein interaction and the synergism of electrostatics and hydrophobicity should not be overlooked. For example, the results of CaM covalent modification studies with basic drugs (Lukas et al., 1985) and peptides (O'Neil and DeGrado, 1989) have implicated methionine residues in both halves of the molecule as participants, or nearest neighbors of residues TWO of these methionine residues (Met-124 and Met-144), labeled by both drug and peptide ligands, are in the carboxyl-terminal lobe of CaM and border the "hydrophobic" cleft observed in the three dimensional structure of CaM derived from x-ray crystallography (Babu et al., 1988). The substitution of Met-124 with Ile in the context of the DEE118-12OKKK charge mutation (difference between VU-12A and VU-12) changed a partial agonist (VU-12A) to an antagonist (VU-12) of the y-subunit. These results suggest that an interplay of electrostatic and hydrophobic properties may be useful to exploit for selective antagonist development.
Even though the concentration dependence of CaM in the generation of activated y-subunit is hyperbolic (Farrar and Carlson, 1991), the activation observed may be the end result of a relatively complex process. It has been suggested that during renaturation/activation of the y-subunit, CaM may function in part as a folding template (Kee and Graves, 1986). Thus, the K, values listed in Table I may be the product of at least three distinct steps: binding, refolding, and activation. I t follows that any given mutation of CaM could affect independently any of the three steps. Concerning the mutations in this study, it is further possible that electrostatic interactions could predominantly affect some step(s) and hydrophobic interactions could affect other(s). There are undoubtedly even more steps at which CaM would participate during formation of the Ca2+-activatable holoenzyme because of the presence in it of the inhibitory a-and @-subunits (Paudel and Carlson, 1987), either of which is capable of binding CaM (James et al., 1991;Newsholme et al., 1992).
The lack of specific kinetic effects on the y-subunit (i.e. no significant change in substrate interactions) by the CaM mutations that severely diminished maximal activation suggests that the mechanism of CaM activation of the y-subunit is distinct from that found with MLCK and CaMPK-I1 (Lukas et al., 1986;Hanson et at., 1989;Waxham et a!., 1989;Weber et al., 1989;Shoemaker et al., 1990;Chabbert et al., 1991;Meador et al., 1992). For example, the CaM binding site analogue of the smooth muscle/nonmuscle form of MLCK does not have direct contact with the residues altered in VU-24 and VU-43 calmodulins (Meador et al., 1992). Consistent with this recently elucidated structural relationship, charge reversal mutations of these calmodulin residues do not alter MLCK activation (Weber et al., 1989;Shoemaker et al., 1990) but alter the activation of the y-subunit (this report). Conversely, similar charge reversal mutations at Glu-84 and Glu-120 of calmodulin, residues in close proximity to the bound MLCK peptide (Meador et al., 1992), alter MLCK activation properties (Craig et al., 1987;Weber et al., 1989;Shoemaker et al., 1990). These results are somewhat unexpected based on the prior observations by Shoemaker et al. (1990) that the CaM binding site of MLCK can be inverted in amino acid sequence or replaced with that of CaMPK-I1 with retention of autoinhibitory activity, of dependence on CaM and Ca2+ for activation, and of substrate specificity by the active site. Our results, when combined with these previous studies, dem-onstrate that the common theme found with MLCK and CaMPK-11, namely the functional importance of protein electrostatics in CaM-enzyme recognition, is also true for the ysubunit of phosphorylase kinase. Similarly, our results demonstrate that there are specific aspects of these electrostatic properties that provide selectivity to the interaction of CaM with its target enzymes.