Basic Residues Are Important for Ca2+/Calmodulin Binding and Activation but Not Autoinhibition of Rabbit Skeletal Muscle Myosin Light Chain Kinase*

Several allosterically modulated protein kinases have been shown to be regulated by an autoinhibitory domain located within the kinase molecules. The inhibitory domain has been proposed to act as a “pseudosub- strate” inhibitor binding to the substrate binding site of the kinase, thereby blocking the binding of the en- zyme’s true substrate. In this report, site-directed mutagenesis has been used to further investigate the mechanism of activation of the inhibitory domain of rabbit skeletal muscle myosin light chain kinase. Basic residues within the pseudosubstrate domain (572-573, 577-579, 580-581), which are analogous to the important substrate determinants of the myosin light chain, were found not to be required in order to maintain the kinase in an inhibited state. Two groups of these residues (577-579 and 581-582) were, however, found to be important for high affinity calmodulin binding to the kinase. These data suggest that the autoinhibitory domain of myosin light chain kinase may not function by directly mimicking the light chain substrate.

myosin light chain kinase results in the formation of a 35-kDa fragment (residues 256-584) which is catalytically active (Edelman et al., 1985), thus confirming the location of the catalytic domain. This chymotryptic fragment no longer requires calmodulin for kinase activity and does not bind to a calmodulin affinity column. In contrast, a tryptic fragment comprising residues 236-594 remains dependent on calmodulin for enzyme activity. These data together with synthetic peptide studies  have led to residues 577-593 being assigned as the calmodulin binding domain of rabbit skeletal muscle myosin light chain kinase. The constitutive activity of the chymotryptic fragment suggests that an inhibitory region is also located in the carboxyl terminus of the kinase. This inhibitory domain is similar to the region of the light chain substrate containing the phosphorylatable serine residue. It has been proposed that when the kinase is inactive the inhibitory region binds to the active site of the enzyme preventing it from interacting with the light chain substrate (Kennelly et al., 1987). When Ca2+/ calmodulin binds to the kinase a conformational change occurs in which the inhibitory region is removed from the active site, thus reversing its inhibition of enzyme activity. This mechanism of regulation by a "pseudosubstrate" inhibitory domain has been proposed to be a general mechanism employed by all myosin light chain kinases and several other protein kinases (Pearson et al., 1988;Soderling, 1990). Recent studies on the chicken nonmuscle myosin light chain kinase have, however, suggested that the inhibitory region of that enzyme may not simply function as a pseudosubstrate inhibitor (Shoemaker et al., 1990). In the current study the mechanism of action of the inhibitory domain of the rabbit skeletal muscle myosin light chain kinase has been investigated. Using site-directed mutagenesis I have shown that basic residues within the proposed inhibitory domain are not required to maintain the kinase in an inhibited state in the absence of Ca*+/calmodulin. In contrast, these residues are important for the high affinity binding of calmodulin to the kinase.

Protein Purification and Kinase
Assays-Rabbit skeletal muscle myosin light chain kinase was prepared as described previously  except that the Affi-Gel Blue column was replaced by a phenyl-Sepharose column (50 ml equilibrated with 10 mM MOPS,' pH 7.0, 0.5 mM EDTA, 250 mM NaCl, 1 mM dithiothreitol, 0.1 mM phenylmethylsulfonyl fluoride). The kinase eluted in the flowthrough fraction of this column. Rabbit skeletal muscle myosin light chains were purified according to Blumenthal and Stull (1980). Calmodulin was purified from bovine testes (Blumenthal and Stull, 1982). COS cell lysates were prepared by detergent lysis (1% Nonidet The abbreviations used are: MOPS, 4-morpholinepropanesulfonic acid; EGTA, [ethylenebis(oxyethylenenitrilo)]tetraacetic acid. P-40, 20 mM MOPS, 0.5 mM EGTA, 2 mM MgCI?, 1 mM dithiothreitol). Kinase concentrations in the lysates were determined from immunoblots using purified rabbit skeletal muscle myosin light chain kinase as a standard (Herring et al., 1990a). Ca'+/calmodulin activation assays were performed as described by Blumenthal and Stull (1980). Assays were performed a t a fixed high concentration of calmodulin (1.2 PM) and various concentrations of free Ca" as determined by CaZ+/EGTA buffers (Potter and Gergely, 1975). The concentrations of other reactants are indicated in the figure and table legends.
Expression of Wild Type and Mutant Myosin Light Chain Kinases-Wild type and mutant cDNAs were subcloned into a pCMV 2 expression vector (Anderson et al., 1989) and expressed in COS cells as described previously (Herring et al., 1990a).
Immunoblot and Calmodulin Overlay Assays-Immunoblotting using polyclonal or monoclonal antibodies directed against rabbit skeletal muscle myosin light chain kinase was performed as described by Herring et al. (1990a). The same procedure was employed to generate blots which were reacted with biotinylated calmodulin (23 pg/ml; Billingsley et al., 1985) and developed using phosphataseconjugated avidin, except that 10 mM CaCI' was added to all solutions.

Expression and Ca2+/Calmodulin Affinity of Mutant
Kinases-Immunoblot analysis of COS cell lysates using antirabbit skeletal muscle myosin light chain kinase polyclonal antibodies demonstrates that all mutant myosin light chain kinases were expressed at levels similar to that of the wild type kinase with the exception of the double mutant PS2/3 which was expressed a t 4-5-fold lower levels (data not shown). Changes in the Ca'+/calmodulin affinity of the mutant kinases relative to the wild type enzyme was estimated by use of biotinylated calmodulin overlays of the kinases following FIG. 1. Domain organization and location of PS mutations on rabbit skeletal muscle myosin light chain kinase. A schematic representation of the linear amino acid sequence of rabbit skeletal muscle myosin light chain kinase is shown (top). The domain organization is as defined by Herring et al. (1990b). The amino acid sequence of the calmodulin binding region is shown (middle); the proposed calmodulin binding domain  is underlined. The amino acid changes made in each of the mutant kinases together with their nomenclature are indicated. The aminoterminal sequence of the rabbit skeletal muscle myosin light chain is shown for comparison (bottom). On this sequence the important basic substrate determinants are boxed and the phosphorylatable serine residue is circled.
transfer of the protein to nitrocellulose. Mutant kinases in which basic residues within the proposed calmodulin binding domain had been altered (PS2, -3, -4, and -2/3) no longer bound biotinylated calmodulin. In contrast, PS1, wild type kinase, and mutants of the proposed light chain binding region (LCB1, LCB2, and LCB3, Herring et al., 1990b) were readily detectable by this procedure (Fig. 2).
Kinetic Properties of PS Mutants-All the myosin light chain kinases produced by the PS series of mutations were completely dependent on Ca2+/calmodulin for activity. Under standard assay conditions (i.e. at least a hundred-fold dilution of lysate into an assay) no kinase activity could be detected in the presence of 2 mM EGTA. When assayed a t low dilution (10-20-fold) a small amount of "P incorporation could be detected in the absence of calcium. The calcium-independent activity was less than 5% of total activity for all mutants except the double mutant PS2/3 (wild type, <0.1%; PS1, 0.3%; PS2, 2%; PS3, <0.1%; PS4, 2.9%; PS2/3, 6%). Assays performed under these conditions exhibit a significant background of "P incorporation (approximately 0.1% of the total activity of the wild type enzyme) which was seen even in lysates produced from expression vectors in which the cDNA was oriented in the antisense direction (Herring et al., 1990a). The Ca2+/calmodulin activation properties of wild type and mutant kinases were assessed by performing assays a t a fixed high concentration of calmodulin and by using Ca2+/EGTA buffers to alter the free calcium levels. Hence the Ca2+/ calmodulin concentration would be determined by the free Ca'+ concentration ("Materials and Methods"). Examples of the Ca2+ activation curves for the wild type kinase and the PS mutants are shown in Fig. 3. From these curves the concentrations of Ca2+ required to produce half-maximal activation of the enzymes were determined; these data are summarized in Table I. Alteration of basic residues aminoterminal of the proposed calmodulin-binding domain of myosin light chain kinase did not significantly change the concentration of Ca2+ required to half-maximally activate the enzyme (wild type [Ca2+]0.Smax 0.12 p~; PS1 0.15 pM; Fig. 3, Table I). In contrast, altering basic amino acids within the proposed calmodulin binding domain of the kinase resulted by immunoblotting using monoclonal antibody 14a (Nunnally et al., 1987)) were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The proteins were transferred to nitrocellulose, and the blot was treated as described previously for monoclonal antibody (19a) immunoblots except that all solutions contained 10 mM CaCI? (Herring et al., 1990a). In place of the first-step antibody 23 pg/ml biotinylated calmodulin was used, and avidin conjugated to alkaline phosphatase was used in place of a second-step antibody. Mutant kinases LCB1, LCB2, and LCB3 were described previously (Herring et al., 1990b). All other mutations are as described under Fig. 1 and under "Materials and Methods."

K, values of mutants PS3 and PS4 for myosin light chain
and ATP were similar to those of the wild type enzyme (Table   I).

DISCUSSION
Altering residues involved in maintaining myosin light chain kinase in an inactive state should result in a kinase which is more easily activatable or constitutively active. It has been shown that two groups of basic residues amino-  Table I. 0, wild type kinase; W, PS1; +, PS2; 0, PS3; A, PS4; +, PS2/3. terminal of the phosphorylatable serine of rabbit skeletal muscle myosin light chain are important substrate determinants for the kinase (Michnoff et al., 1986; Fig. 1). Thus, the pseudosubstrate hypothesis predicts that altering basic residues within the inhibitory domain would produce a less inhibited or constitutively active enzyme. In contrast, the results presented in this report demonstrate that alteration of these basic residues (577-579 and 581-582) produced mutant kinases which not only remained dependent on Ca2+/calmodulin for activity but also required higher concentrations of Ca2+/calmodulin to produce half-maximal activation than the wild type enzyme. In addition, the inability of mutant kinases PS3, PS4, and PS2/3 to be fully activated, even at very high Ca2+/calmodulin concentrations, may suggest that the basic residues are important for the activation rather than the inhibition of kinase activity.
The increased Ca*+/calmodulin concentrations required to activate mutant kinases PS2, PS3, PS4, and PS2/3 was due, at least in part, to a decreased affinity of the mutant kinases for Ca2'/calmodulin. This demonstrates the importance of these residues in the binding of the kinase to Ca2+/calmodulin and is consistent with their location within the previously defined calmodulin binding domain of the enzyme. The greater concentration of Ca*+/calmodulin required to activate PS4 (KRR + EEE) relative to PS2 (KRR + ETL) suggests that the introduced negative charges may be interacting with similarly charged residues in calmodulin. This proposal is in agreement with previous data in which similar mutations in an analogous position of the nonmuscle myosin light chain kinase were able to compliment reciprocal charge mutations in calmodulin (E84K, E120K;Shoemaker et al., 1990). Thus, it is reasonable to propose that the basic residues, at positions 577-579 and probably also those at positions 581-582, of the rabbit skeletal muscle myosin light chain kinase are interacting with acidic residues in calmodulin. It would, therefore, be difficult to envision that these same residues could be involved in the interaction of the inhibitory domain with the substrate binding site of the kinase. Nevertheless, it may be argued that mutation of the basic residues, within the inhibitory domain weakens, but does not obliterate, the interaction of the inhibitory region with the substrate binding site. Hence, the mutated kinases would remain Ca2+/calmodulin-dependent and a potential decrease in the concentrations of Ca*+/calmodulin required to half-maximally activate the kinases may be offset

TABLE I
Kinetic properties of mutant and wild type myosin light chain kinases K, and Vmax values were obtained from double-reciprocal plots. All assays were performed in duplicate, values obtained from more than two independent assays are given as the means f standard deviation, and the number of assays performed is given in brackets. Assays used to determine K , values for myosin light chains were performed a t 1 mM ATP, 1.2 p~ calmodulin and 600 p~ CaCI2. Assays used to determine the K , values for ATP were performed at the same CaC12 and calmodulin concentrations and at 50 WM myosin light chain. Ca2+/calmodulin Ca'+ concentrations as determined by Ca2+ .EGTA buffers (Blumenthal and Stull, 1980;Potter and Gergely, 1975 by their decreased affinity for Ca2+/calmodulin. In support of this supposition a Ca2'/calmodulin-independent form of the chicken nonmuscle myosin light chain kinase was produced by substitution of 6 basic residues, within the inhibitory region, with acidic glutamic acid residues (Shoemaker et al., 1990). However, the synthetic peptide studies of Michnoff et al. (1986) demonstrated a 40-50-fold increase in the K, for peptides modeled after the myosin light chain, in which either group of important basic residues were replaced with the neutral amino acid alanine. If similar changes were to occur in the pseudosubstrate region mutant PS3 (KK + QQ 582-583) would be expected to exhibit a 50-fold decrease in affinity, and mutants PS2, PS4, and PS2/3 would exhibit a much greater decrease due to the the incorporation of acidic residues. Under the assay conditions used in this study, in which the light chain substrate is at 10-fold higher concentrations than its K,,, value, one may have predicted that the substrate would be able to competitively overcome the inhibition exerted by the inhibitory domain. Further evidence contradicting the importance of basic amino acids for the function of the inhibitory region can be obtained from proteolysis studies. A 35-kDa chymotryptic fragment of the rabbit skeletal muscle myosin light chain kinase (residues 256-584) is constitutively active (Edelman et al., 1985). This fragment still retains the basic residues analogous to the substrate determinants in the myosin light chain and would, therefore, according to the pseudosubstrate model, have been predicted to be inactive. In addition the inhibitory potency of synthetic peptides modeled after the inhibitory domain of smooth muscle myosin light chain kinase did not simply correlate with the number of basic residues in the peptide (Lukas et al., 1988). These peptides have also been reported to inhibit enzyme activity competitively with respect to both ATP and myosin light chain (Ikebe, 1990). Together these data suggest that the inhibitory region of myosin light chain kinase does not function by simply mimicking the light chain substrate. The data may be more consistent with a recently proposed "flip-flop" model describing the activation of calmodulin-dependent protein kinases (Jarrett and Madhaven, 1990). In this model the calmodulin binding domain is proposed to bind to a part of the kinase which resembles calmodulin (residues 492-522) rather than to the substrate binding site. This model, however, also evokes charge interactions between the calmodulin binding domain and the "calmodulin like-binding site" and would, therefore, also not be consistent with the current data.
In summary, the experiments presented here demonstrate that basic amino acids within the calmodulin binding domain of myosin light chain kinase are important for high affinity binding to Ca2+/calmodulin. These residues may also be involved in the subsequent activation of the kinase. In contrast, no evidence was obtained to suggest that the basic residues are essential for the function of the inhibitory domain. This may indicate that the inhibitory region does not function by directly mimicking the myosin light chain substrate.