Calcium/Calmodulin-dependent Protein Kinase I1 CHARACTERIZATION OF DISTINCT CALMODULIN BINDING AND INHIBITORY DOMAINS*

the Ca2’/cal- modulin-dependent protein kinase dithiothreitol, 20% glycerol, 0.64 mg/ml bovine serum albumin, 2.38 CaM, 50 p~ CaM- deficient phosphodiesterase, and various concentrations of Ca2+/ CaM-dependent protein kinase I1 synthetic peptides. Followingprein- cubation of the reaction mixture at 30 "C, the reaction was initiated by the addition of substrate. After 50 min the reaction was terminated by boiling. Conditions were selected in which the reactions were linear with respect to time.

8 To whom correspondence should be addressed.
The abbreviations used are: CaM, calmodulin; HEPES, 4-(2-hydroxyethy1)-1-piperazineethanesulfonic acid EGTA, [ethylenebis(oxyethylenenitri1o)ltetracetic acid HPLC, high pressure liquid chromatography. intracellular calcium. Initial studies on the interactions between CaM and proteins utilized naturally occurring model peptides such as melittin (Comte et al., 1983;Barnette et al., 1983;Seeholzer et al., 19861, mastoparans (Malencik and Anderson, 1983;Barnette et al., 1983), and P-endorphins (Weiss et al., 1980;Giedroc et al., 1983). Within the last 2 years, however, CaM-binding domains have been identified within specific target enzymes. The first such study was by BIumenthaI et al. (1985) who IocaIized the CaM-binding domain of rabbit skeletal muscle myosin light chain kinase near the carboxyl terminus. The synthetic 26-residue peptide analog of this sequence inhibited the CaM-dependent activation of skeletal muscle myosin light chain kinase and bound CaM in a calcium-dependent manner with a molar stoichiometry of 1:l (Klevit et al., 1985). Similar studies have been carried out using chicken gizzard smooth muscle myosin light chain kinase (Lukas et al., 1986;. A synthetic peptide corresponding to residues 480-501 (deduced from cDNA sequence; Guerriero et al., 1986) inhibited the CaMdependent activation of myosin light chain kinase = 46 nM). In addition, this peptide was also a substrate antagonist with IC5,, values of 2.7 and 0.9 p~ for myosin light chains and synthetic peptide substrate, respectively . When a CaM-independent tryptic fragment of myosin light chain kinase was used to eliminate the possibility of the peptide interacting with CaM, peptide 480-501 inhibited kinase activity with an ICso of 33 nM (assayed using 2 p M light chains) (Ikebe et al., 1987). These observations on smooth and skeletal muscle myosin light chain kinase suggest that there is an inhibitory domain, which prevents substrate recognition in the absence of CaM, closely associated with the CaM-binding domain .
Amino acid sequence has recently been derived from cDNA clones of both the 50-kDa (Hanley et al., 1987;Lin et al., 1987) and 60-kDa (Bennett and Kennedy, 1987) subunits of rat brain Ca2+/CaM-dependent protein kinase 11. Hanley et al. (1987) proposed that the CaM-binding domain of Ca2+/ CaM-dependent protein kinase I T was located within residues 290-314 of the rat brain BO-kDa subunit and showed that a synthetic peptide corresponding to this region inhibited activation of Ca2+/CaM-dependent phosphodiesterase and formed a calcium-dependent complex with CaM. Investigation of the Ca2+/CaM-binding domain of Ca2+/CaM-dependent protein kinase I1 is particularly interesting since the kinase can become independent of Ca2+/CaM following an intramolecular autophosphorylation (Saitoh and Schwartz, 1985;Miller and Kennedy, 1986;Schworer et al., 1986;Lai et al., 1986;Lou et al., 1986) on threonine residues (Lai et al., 1987). Furthermore, autophosphorylation at additional sites appears to block the interaction of Ca2+/CaM with Ca2+/ CaM-dependent protein kinase I1 (Hashimoto et ai., 1987).
In this paper the properties of the CaM-binding domain of Ca2'/CaM-dependent protein kinase I1 are more precisely defined by the use of several synthetic peptides. It was also of interest to ascertain whether these peptides exhibited substrate-directed inhibition of the kinase. In addition, putative autophosphorylation sites flanking the CaM-binding and inhibitory domains are identified.

EXPERIMENTAL PROCEDURES
Materials and Method~-[y~~P]ATP was either purchased from ICN Radiochemicals or prepared using the method of Walseth and Johnson (1979) using carrier-free 32P04 from the same company. Phosphocellulose (P8l) paper was purchased from Whatman. Sequanol-grade trifluoroacetic acid was from Pierce, and HPLC-grade acetonitrile was from Burdick and Jackson.
Protein and Peptide Purification-Rat forebrain Ca2'/CaM-dependent protein kinase I1 was purified as described previously (Hashimoto et al., 1987). CaM was purified from bovine brain (Gopalakrishna and Anderson, 1982) or bacterial lysates (Putkey et al., 1985). Bovine brain Ca2+/CaM-dependent phosphodiesterase was purified using the method of Sharma et al. (1980). The Bradford (1976) method was used to determine the protein concentration of the kinase using bovine serum albumin as standard, and the concentration of CaM was assessed either by absorbance (E276nm. mg,ml = 0.18) or amino acid analysis.
Syntide-2, PLARTLSVAGLPGKK, a peptide analog similar to the sequence surrounding phosphorylation site 2 of glycogen synthase, was synthesized as described previously . A synthetic peptide derived from the phosphorylation site of myosin light chains (KKRPQRATSNVFS) was purchased from Peninsula Laboratories Inc. Solid phase synthesis of the peptides derived from Ca*+/CaM-dependent protein kinase I1 was accomplished with either Beckman 990 automatic synthesizer or an Applied Biosystems Model 430 machine by the procedures of Hodges and Merrifield (1975). Using benzbydrylamine resin, the peptides were prepared as the carboxyl-terminal amide form. Simultaneous deprotection and cleavage from the resin was achieved with anhydrous HF in the presence of 10% anisole at 0 "C for 60 min (Stewart and Young, 1966) using a Protein Research Foundation HF apparatus (Osaka, Japan). The peptides were then purified by gel filtration chromatography in the presence of 5.5 M acetic acid followed by either chromatography on SP-Sephadex (Kemp, 1979) or preparative reverse phase HPLC on a C18 Amicon resin (250-A pore size) in the presence of 0.1% (v/v) trifluoroacetic acid using a gradient of acetonitrile. Peptide concentrations were determined using a Waters Associates Pico-tag amino acid analyzer.
Protein Kinase Assays-Phosphorylation of synthetic peptides and protein substrates was determined using the phosphocellulose paper method of Roskowski (1985). Two procedures were used, depending on the aims of the experiment. Control reactions were performed in which CaZ'/CaM-dependent protein kinase 11, substrate, or Ca2+/ CaM-dependent protein kinase I1 synthetic peptides were omitted. Method 1: for the determination of the CaM antagonistic effects of the Ca2'/CaM-dependent protein kinase I1 synthetic peptides, assays were performed in 50 mM HEPES, pH 7.5,lO mM magnesium acetate, 0.5 mM calcium chloride, 0.4-0.5 mM [ T -~~P ] A T P (300-600 cpm/ pmol), 250 pM syntide-2 substrate plus the indicated concentrations of CaM and Ca2'/CaM-dependent protein kinase I1 synthetic peptide (25-pl total volume). The reactants were incubated on ice for 10 min before the reaction was initiated by the addition of Ca2'/CaMdependent protein kinase 11. The kinase had been diluted appropriately in dilution buffer (50 mM HEPES, pH 7.5, containing 10% (v/ v) ethylene glycol, 1 mg/ml bovine serum albumin). The assay was at 30 "C for 1-1.5 min, and the reaction was terminated by spotting aliquots onto phosphocellulose papers. Method 2: for the determination of the inhibitory properties of Ca2'/CaM-dependent protein kinase I1 synthetic peptides the autophosphorylated, Ca*+/CaM-independent form of CaZ+/CaM-dependent protein kinase I1 was used in order to eliminate potential complications due to CaM-peptide interactions. Autophosphorylation was achieved by incubation of Ca2+/CaM-dependent protein kinase I1 (5.3 p~ subunit concentration) on ice in 50 mM HEPES, pH 7.5, containing 10 mM magnesium acetate, 1 mM calcium chloride, 6 pM CaM, and 0.5 mM ATP for 5-8 min. The reaction was terminated by the addition of dilution buffer containing 5 mM EDTA to an appropriate final kinase dilution . The subsequent assay was performed as described in Method 1, except that 0.5 mM EGTA was substituted for calcium chloride, CaM was omitted, and various concentrations of syntide-2 were used as substrate. The potential phosphorylation of the Ca*'/CaM-dependent protein kinase I1 synthetic peptides was also investigated by this procedure. In these experiments the indicated concentration of Ca*+/CaM-dependent protein kinase I1 synthetic peptides was substituted for syntide-2.
Phosphodiesterase Assay-Ca"/CaM-dependent phosphodiesterase was assayed at 30 ' C using a modification of a described method (Thompson et al., 1979). The reaction mixture contained 40 mM Tris-HCl, pH 8.0, 5 mM magnesium acetate, 1 mM calcium chloride, 30 p~ cGMP, 0.15 pCi of [3H)cGMP, 1 mM dithiothreitol, 20% glycerol, 0.64 mg/ml bovine serum albumin, 2.38 nM CaM, 50 p~ CaMdeficient phosphodiesterase, and various concentrations of Ca2+/ CaM-dependent protein kinase I1 synthetic peptides. Followingpreincubation of the reaction mixture at 30 "C, the reaction was initiated by the addition of substrate. After 50 min the reaction was terminated by boiling. Conditions were selected in which the reactions were linear with respect to time.

CaM-Binding Domain in the Ca2+/CaM-dependent Protein
Kinase II-When the amino acid sequences, deduced from the cDNA sequences, of the 50- (Hanley et al., 1987;Lin et al., 1987) and 60 (Bennett and Kennedy, 1987)-kDa subunits of rat brain Ca*+/CaM-dependent protein kinase I1 are compared to the sequences of other protein kinases, certain regions, e.g. the ATP-binding site, are highly conserved. When the sequence comparison is confined to the myosin light chain kinases and phosphorylase kinase y-subunit, residues 291 to 315 of the 50-kDa subunit and residues 292 to 316 of the 60-kDa subunit show homology to the CaM-binding domains of these other CaM-dependent protein kinases. Hanley et al. (1987) showed that a 25-residue peptide (290-314 of the 50-kDa subunit) did inhibit the CaM-dependent activation of brain cyclic nucleotide phosphodiesterase and produced a Ca2+-dependent shift in the mobility of CaM in nondenaturing gel electrophoresis. To define further the CaM-binding domain of Ca2+/CaM-dependent protein kinase 11, six peptides were synthesized which contained different portions of the 290-309 50-kDa sequence as well as extensions in both the amino-and carboxyl-terminal directions (see Table I).
Peptide 290-309 was slightly more potent in terms of CaM binding than any of the other peptides. In the presence of a large excess of substrate (250 p M syntide-2) and limiting CaM (100 nM), it inhibited Ca2+/CaM-dependent protein kinase I1 with an ICso of approximately 52 nM ( Fig. 1 and Table I). Peptides 296-309 and 294-319 also strongly inhibited the kinase under these conditions with IC,, values of approximately 57 and 77 nM, respectively ( Fig. 1 and Table I). In all three cases the inhibition was reversed by addition of excess CaM, indicating that the mechanism of peptide inhibition was through interaction with CaM and not by a direct effect on the kinase. No inhibition was observed using peptides 281-290, 290-302, or 295-304 (Fig. 1). Similar results were obtained using glycogen synthase (1 mg/ml) as the substrate (not shown).

TABLE 1 Effects of synthetic peptide analogs of Ca2'/CaM-dependent protein kinase 11
Determinations of each parameter were made as described under "Experimental Procedures." For measurement of CaM antagonistic activity using Ca2+/CaM-dependent protein kinase 11, 100 n M CaM and 250 pM syntide-2 substrate were used. In the substrate inhibition experiments 20 pM syntide-2 was used as substrate. CaM antagonistic activity using Ca'+/CaM-dependent phosphodiesterase was determined with 2.38 nM CaM. *, no inhibition observed. a, not measurable due to Substrate-directed Inhibition by Peptides-Smooth muscle and skeletal muscle myosin light chain kinase contain regions associated with their CaM-binding domains that show competitive inhibition against substrates . It was therefore of interest to determine whether any of the six peptides shown in Table I exhibited substrate-directed inhibition. To avoid any complications due to CaM binding of the peptides, the autophosphorylated form of Ca2+/CaM-dependent kinase 11, which is active in the presence of excess EGTA , was used. When assayed in the presence of excess EGTA, the calciumindependent form of the kinase has a K, of about 20 pM for the substrate syntide-2 (Hashimoto et al., 1987). Using these assay conditions, peptides 290-309 and 290-302 inhibited the phosphorylation of syntide-2 with IC5, values of approxi-mately 25 and 50 p~, respectively (Table I). Peptide 294-319 gave an IC,, of 115 PM, and peptides 295-304 and 296-309 had ICso values of approximately 200 p~. Peptide 281-290 was not tested since it was phosphorylated by the kinase (see below). Under these assay conditions, peptide 294-319 was not significantly phosphorylated.
The kinetic mechanism of the inhibition was next examined as a function of syntide-2 concentration (Fig. 2). With peptides 290-309, 290-302, and 296-309, the intercepts of the Lineweaver-Burk plots (Fig. 2, A-C) with increasing concentrations of peptides intersected close to or on the ordinate, indicative of competitive inhibition. In order to examine the nature of the competitive inhibition, Dixon plots (Fig. 2, D-F ) were constructed and found to be nonlinear, indicating partially competitive inhibition for all three peptides. Likewise, slope replots from the Lineweaver-Burk data were also not linear (not shown).
Several control experiments were performed to assess the specificity of the substrate-directed inhibition. When another peptide substrate, K-K-R-P-Q-R-A-T-S-N-V-F-S (modeled after the sequence in smooth muscle myosin light chain), was used in the kinase assay, similar inhibition was obtained (not shown). When the phosphorylation of syntide-2 (20 p~) by CAMP-dependent protein kinase was measured, no inhibition by peptides 290-309 and 290-302 (up to 100 WM) was observed (not shown). However, when a protein substrate such as glycogen synthase (0.15 mglml) or casein (1 mglml) was used in the assay for Ca2+/CaM-dependent protein kinase 11, no inhibition by peptide 290-309 (up to 158 p M ) or peptide 290-302 (up to 280 p~) was obtained.
Phosphorylation of Peptides by Ca2'/CaM-dependent Protein Kinase 11-Examination of the amino acid sequences surrounding serine or threonine residues phosphorylated by Ca2+/CaM-dependent protein kinase 11 suggests a consensus sequence of -R-X-X-SIT- (Payne et al., 1983). The essential nature of the arginine three residues amino-terminal of the phosphorylated serine or threonine has been confirmed using synthetic peptides (Pearson et al., 1985;Soderlinget al., 1986). which contain these sites, were phosphorylated at rates of 2.50 and 0.05 pmol/min/mg, respectively, compared to a value of 6.0 gmol/min/mg for syntide-2.

DISCUSSION
Three important questions concerning mechanisms of regulation of Ca'+/CaM-dependent protein kinase I1 are the following: 1) why is the kinase inactive in the absence of Ca2+/CaM?; 2) how does Ca'+/CaM activate the kinase?; and 3) how does autophosphorylation generate a Ca2+-independent species of the kinase? Our understanding of each of these areas is enhanced by the results of this investigation which utilized synthetic peptides of the CaM-binding domain and flanking sequences. Although the peptides used in this study were synthesized from the 281-319 amino acid sequence of the 50-kDa subunit of the kinase (Lin et at., 1987), it should be emphasized that the corresponding sequence in the 60-kDa subunit (Bennett and Kennedy, 1987) differs by only a Asp to Glu substitution a t position 289 and several conservative substitutions at the extreme carboxyl terminus.
The 20-amino acid peptide 290-309 was the most potent of the six peptides tested in terms of both CaM-binding and in substrate-directed inhibition of the kinase (Table I). Thus, both of these regulatory properties reside in close proximity to each other in the primary sequence. Although there may be some overlap in amino acid determinants for these two properties, it is clear that different regions of peptide 290-309 are involved. For example, peptide 290-302 did not bind CaM but did show substrate-directed inhibition of the kinase that was almost as potent as peptide 290-309. Likewise, peptide 296-309 bound CaM with about the same affinity as peptide 290-309 but was relatively weak with regard to sub-strate-directed inhibition. These results indicate that the majority of determinants for CaM binding reside in the carboxyl-terminal portion, whereas the determinants of substrate-directed inhibition are localized predominantly in the amino-terminal portion of peptide 290-309 (see bottom diagram of Table I). The 10-residue peptide 295-304, comprising the middle portion of peptide 190-309, did not bind CaM and showed only weak substrate-directed inhibition. This result would be consistent with the above conclusion that the determinants for these two properties of peptide 290-309 are in the carboxyl-and amino-terminal sequences, respectively. Alternatively, it is possible that peptide 295-304 was too short to form a stable interaction with CaM.
The kinetic analysis of the substrate-directed inhibition indicated partial-competitive inhibition with respect to the peptide substrate syntide-2 (Fig. 2). The Dixon plots were not linear, indicating that the inhibition was not strictly competitive. This is not surprising since the sequence of peptide 290-309 does not contain the consensus sequence of -R-X-X-B/ -T-found in good substrates of Ca'+/CaM-dependent protein kinase 11. Thus, binding of the inhibitory peptide may not be exactly the same as occurs with a substrate. Inhibition was only observed using peptides as substrates, either syntide-2 or the peptide modeled after smooth muscle myosin light chain. With either skeletal muscle glycogen synthase or casein as substrates, peptides 290-302 and 290-309 did not give inhibition. The explanation for this is not readily apparent but may relate to binding of peptides by glycogen synthase or casein as well as steric problems or multiple binding interactions with the larger protein substrates. Lack of substratedirected inhibition was probably not due to binding of the peptides by glycogen synthase since CaM antagonism was observed with peptide 290-309 using glycogen synthase as substrate. On the other hand, substrate-directed inh.ibition was not due to some unusual interaction of the inhibitory peptides with the syntide-2 since phosphorylation of syntide-2 by CAMP-dependent protein kinase was not affected by peptides 290-302 and 290-309.
These results on substrate-directed inhibition by a sequence contiguous to the CaM-binding domain are very similar to analogous experiments using both skeletal muscle  and smooth muscle  myosin light chain kinases. Both of these enzymes have pseudosubstrate sites next to the CaM-binding sequences. Synthetic peptides containing these pseudosubstrate sites gave either strict competitive inhibition (skeletal muscle kinase) or nonlinear competitive inhibition (smooth muscle kinase). The peptides which give substrate-directed inhibition of Ca2+/ CaM-dependent protein kinase I1 do not contain any apparent pseudosubstrate sequences. The sequence -R-R-K-L-would probably not be a good pseudosubstrate site since the addition of the second arginine residue (R-R-X-$/T) appears to be a negative determinant for Ca'+/CaM-dependent protein kinase I1 . Additional studies with substitutions of certain residues in peptide 290-302 may further define specificity determinants involved in the substrate-directed inhibition of Ca'+/CaM-dependent protein kinase 11.
Certain structural features are thought to be important for interaction of proteins or peptides with CaM. Peptides with widely divergent functions such as melittin, the mastoparans, and the peptide analogs derived from both smooth and skeletal muscle myosin light chain kinases and phosphorylase kinase all bind CaM with high affinity and share certain structural features thought to be important for interaction with CaM. These include the ability to form an amphipathic a-helix (Barnette et al., 1983;Malencik and Anderson, 1983;Domains Comte et al., 1983;Maulet and Cox, 1983), clusters of basic residues, and a hydrophobic region adjacent to the basic residues (Malencik and Anderson, 1982;Cox et al., 1985;O'Neil et al., 1987). Protein secondary structure analysis predicted that peptide 290-309, the most potent at binding CaM, would exist as about 80% a-helix (Fig. 3). Extending the peptide from 281 to 319 would reduce the calculated helical content to 59%. The average hydropathy for the 290-309 sequence was calculated to be 0.0, while that of the 281 to 319 sequence was -0.6 (Garnier et al., 1978). The hydropathy analysis shown in Fig. 3 illustrates that the most hydrophobic region of the sequence 281-319 exists between residues 300 and 307, just carboxyl-terminal to the cluster of basic residues. The CaM-binding domain does not contain any acidic residues, which is important since CaM is an acidic protein. Furthermore, using the algorithm of Garnier et al. (1978), the secondary structure in this area would be predicted to be primarily e-helix (Fig. 3). Thus Ca2+/CaM-dependent protein kinase I1 contains a domain which conforms to the structural requirements thought to be important for CaM-binding proteins. Just amino-terminal to this CaM-binding domain there exists an inhibitory domain that may be responsible for inactivation of the kinase in the absence of Ca2+/CaM. Binding of Ca2+/CaM may induce a-helical structure, disrupting the inhibitory interaction with the catalytic site and thereby activating the kinase. This hypothesis would be consistent with the models proposed for myosin light chain kinase  and may represent a common mechanism Upper, hydropathy analysis of the sequence using the algorithm reported in Kyte and Doolittle (1982). Between the two panels, the peptide sequence is displayed using the one-letter code. Lower, probabilities for a-helix ( H ) , @-sheet ( S ) , reverse turn (T), and random coil (C) using the algorithm of Garnier et al. (1978). The bars represent the regions in which the respective conformations would be predicted to occur. for other CaM-activated enzymes.

Sequence M H R Q E T V D C L K K F N A R R K L K G A I L T T M L A T R N F S G G K S
A unique property of Ca2+/CaM-dependent protein kinase I1 is the formation of a substantially (50-80%) Ca2+-independent form upon Ca2+-dependent autophosphorylation (Saitoh and Schwartz, 1985;Miller and Kennedy, 1986;Schworer et al., 1986;Lai et al., 1986;Lou et al., 1986). This is associated with autophosphorylation of a threonine residue (Lai et al., 1987). Subsequent Ca2+-independent ( i e . in the presence of EGTA) autophosphorylation of the kinase results in loss of stimulation by Ca*'/CaM in the subsequent kinase assay (Hashimoto et al., 1987). The amino acid sequence (peptide 290-309) in the kinase which contains both the CaMbinding and substrate-directed inhibitory domains is flanked on both sides by potential autophosphorylation sequences, -R-Q-E-T-(peptide 283-286) and -R-N-F-S-(peptide 311-314). The peptides containing these sequences were both phosphorylated by Ca*+/CaM-dependent protein kinase 11. Studies are currently in progress to assess the effects of phosphorylation of these sites on the substrate-directed inhibition and CaM-binding properties of the adjacent sequences.