Comparison of adenosine 3':5'-monophosphate-dependent protein kinases from rabbit skeletal and bovine heart muscle.

Homogeneous preparations of adenosine 3':5'-monophosphate (cyclic AMP)-dependent protein kinase from rabbit skeletal (Peak I) and bovine heart muscle have been compared. Each enzyme has an S20,w value of 7.0. Each enzyme binds 2 mol of cyclic AMP per mol of enzyme and is dissociated in the presence of saturating concentrations of cyclic AMP into a demeric regulatory subunit-cyclic AMP complex and two catalytic subunits. The isolated subunits recombine, resulting in the formation of the original holoenzyme in each case. Several differences between the two enzymes were found. Different salt concentrations are necessary for elution of the respective enzyme from DEAE-cellulose. Their regulatory subunits differ with respect to their sedimentation constants and mobility on sodium dodecyl sulfate gel electrophoresis. The regulatory subunit of the heart enzyme is rapidly phosphorylated by MgATP but this does not occur with the skeletal muscle enzyme. MgATP is bound with high affinity only to the skeletal muscle enzyme. The enzymes have different apparent dissociation constants and Hill coefficients for cyclic AMP binding. With the skeletal muscle enzyme MgATP increases the dissociation constants for cyclic AMP about 10-fold and decreases the Hill coefficient, while with the heart enzyme phosphorylation decreases the cissociation constant for cyclic AMP 5- to 6-fold and increases the Hill coefficient. Different concentrations of cyclic AMP are required to dissociate the skeletal and heart muscle enzymes. The presence of MgATP increases the concentration of cyclic AMP required to dissociate the skeletal muscle enzyme but decreases the concentration necessary to dissociate the heart enzyme.


Homogeneous preparations
of adenosine 3':5'-monophosphate (cyclic AMP)-dependent protein kinase from rabbit skeletal (Peak I) and bovine heart muscle have been compared. Each enzyme has an sZO,,, value of 7.0. Each enzyme binds 2 mol of cyclic AMP per mol of enzyme and is dissociated in the presence of saturating coricentrations of cyclic AMP into a dimeric regulatory subunit-cyclic AMP complex and two catalytic subunits. The isolated subunits recombine, resulting in the formation of the original holoenzyme in each case. Several differences between the two enzymes were found. Different salt concentrations are necessary for elution of the respective enzyme from DEAE-cellulose. Their regulatory subunits differ with respect to their sedimentation constants and mobility on sodium dodecyl sulfate gel electrophoresis.
The regulatory subunit of the heart enzyme is rapidly phosphorylated by MgATP but this does not occur with the skeletal muscle enzyme. MgATP is bound with high affinity only to the skeletal muscle enzyme. The enzymes have different apparent dissociation constants and Hill coefficients for cyclic AMP binding. With the skeletal muscle enzyme MgATP increases the dissociation constant for cyclic AMP about lo-fold and decreases the Hill coefficient, while with the heart enzyme phosphorylation decreases the dissociation constant for cyclic AMP 5-to 6-fold and increases the Hill coefficient. Different concentrations of cyclic AMP are required to dissociate the skeletal and heart muscle enzymes. The presence of MgATP increases the concentration of cyclic AMP required to dissociate the skeletal muscle enzyme but decreases the concentration necessary to dissociate the heart enzyme.
Cyclic AMP' exerts many diverse biological effects, a number of which are known to be due to the activation of protein kinase(s) (1). Evidence has been obtained that the mechanism through which cyclic AMP activates the enzyme involves the dissociation of an inactive holoenzyme form into a regulatory subunit-cyclic AMP complex and active catalytic subunits (2)(3)(4)(5). The following equilibrium expression of the reversible dissociation of the skeletal muscle holoenzyme (R,C,) into a dimeric regulatory subunit-cyclic AMP complex (R,.cyclic AMP,) and two free catalytic subunits (C) has been established (6,7). R,C, + 2 cyclic AMP * R,.(cyclic AMP), Early studies showed that cyclic AMP-dependent protein kinase activity in many tissue extracts can be separated into several fractions by ion exchange chromatography (8-13). Two major fractions have been identified and referred to as Peaks I and II. Corbin et al. (14) focused attention on the fact that these two types of enzyme can be distinguished by the ability of histones or high salt concentrations to alter their dependence on cyclic AMP. Presumably these effects are due to differences in the dissociability of the two kinds of protein kinase. The interpretation of some of these results has been difficult, however, because of the possibility that some of the observed peaks may have been due to partial proteolysis or because of ether complications that can arise in dealing with enzymes in crude fractions.
Work with purified enzymes has also supported the idea that two major types of cyclic AMP-dependent protein kinases exist. The most intensively studied enzymes have been the rabbit skeletal muscle protein kinase isolated from Peak I separated on DEAE-cellulose (6,7,15) and the bovine heart protein kinase (16)(17)(18)(19)(20). The major distinguishing characteristics that have been reported for the purified heart and skeletal muscle enzymes are as follows. Rabbit skeletal muscle protein kinase' is easily dissociated into its subunits by cyclic AMP 7796 (13), whereas the bovine heart enzyme is readily dissociated only after its regulatory subunits have been phosphorylated (19). No phosphorylation of the regulatory subunits of rabbit skeletal muscle protein kinase has been reported (21). Rabbit skeletal muscle protein kinase binds 2 mol of cyclic AMP for each dimeric regulatory subunit (7,15), whereas the bovine heart enzyme has been reported to bind only 1 mol of this nucleotide for each regulatory subunit dimer (18,20). 3 The molecular weight of the bovine heart protein kinase appears to be slightly higher than that of the rabbit skeletal muscle enzyme (6,7,17). The individual subunits of the rabbit muscle protein kinase isolated from the second DEAE-cellulose peak have also been purified and many of their properties determined (21). Preliminary studies carried out with the regulatory subunit of this enzyme show that it has many features in common with the regulatory subunit of the heart enzyme (21).
Because of the current interest in cyclic AMP-dependent protein kinases, and in view of the fact that rabbit skeletal muscle cyclic AMP-dependent protein kinase and the enzyme from bovine heart are both available as homogeneous proteins, it appeared worthwhile to do a simultaneous comparative study of the two enzymes. The possible benefits to be derived from such a study were increased by indications that these enzymes may serve as prototypes for two general isozymic forms of cyclic AMP-dependent protein kinase (21 (26). Enzyme Preparations-Rabbit skeletal muscle cyclic AMP-dependent protein kinase was isolated from the first peak separated by DEAE-cellulose chromatography as described previously (13).* Cyclic AMP-dependent protein kinase from bovine heart' was prepared utilizing some steps taken from the method for the skeletal muscle enzyme (13) and one step taken from the procedure of Rubin et al. (16). Initial steps through the first DEAE-cellulose column were carried out according to the first of these procedures. Fractions corresponding to the major peak of activity (see Fig. 1B) were pooled and solid ammonium sulfate was added (16.0 g/100 ml). The pH was kept between 7.0 and 7.5 by dropwise addition of ammonium hydroxide.

7797
(not illustrated). This change was seen only when the Trisglycine and not when the sodium phosphate buffer system was used.

Phosphorylation
of Regulatory Subunit-In confirmation of the results of Erlichman et al. (19) it was found that the regulatory subunit of bovine heart cyclic AMP-dependent protein kinase was rapidly phosphorylated when the holoenzyme was incubated with [y-?']ATP. In the presence of cyclic AMP a maximal plateau value of 1.8 to 2.1 mol of 32P incorporated per mol of heart holoenzyme was reached within 1 min and maintained for at least 1 hour. However, essentially no radioactive phosphate was incorporated into the skeletal muscle enzyme at the early time points and less than 0.5 mol/mol of holoenzyme was found after 3 hours of incubation.
In order to exclude the possibility that the regulatory subunit of the heart enzyme was phosphorylated by a kinase missing from the skeletal muscle preparation, both enzymes were incubated together in the presence of [T-~?']ATP and cyclic AMP. The reaction was terminated by addition of sodium dodecyl sulfate and the complete reaction mixture was applied to sodium dodecyl sulfate gels. Under these conditions radioactive phosphate was detected only in the slowest moving band corresponding to regulatory subunit of the heart enzyme ( Fig. 3). This result would indicate that only the regulatory subunit of the heart enzyme is a substrate for the phosphotransferase reaction, and that the lack of phosphorylation of the regulatory subunit of the skeletal muscle enzyme was not due to the lack of an appropriate kinase in the preparation.
High Affinity Binding of MgATP-Skeletal muscle cyclic AMP-dependent protein kinase is known to bind MgATP with high affinity (6,7,30). The catalytic subunit derived from the holoenzyme also binds MgATP, as evidenced by the fact that it utilizes the nucleotide-metal complex as a substrate in the phosphotransferase reaction, but the K, in this reaction is considerably higher than the apparent K, for the binding of MgATP to the holoenzyme (7,8). In the present comparative study it was confirmed that MgATP binds to the skeletal muscle protein kinase (K, S 35 nM) but no comparable high affinity site was found with the bovine heart enzyme (Fig. 4).

RESULTS
Ion Exchange Chromatography-Crude extracts from rabbit skeletal or bovine heart muscle were applied to separate but otherwise identical DEAE-cellulose columns and eluted with a linear NaCl gradient. Under these conditions cyclic AMPdependent protein kinase activity from each extract was resolved into two peaks (Fig. 1) which eluted at conductivities corresponding to about 20 mM (Peak I) and 165 mM NaCl (Peak II). However, 80% of the activity present in the rabbit skeletal muscle extract emerged at the lower salt concentration, whereas 90% of the activity present in the bovine heart extract eluted at the higher salt concentration, with the remaining activity in each case present in the other peak. The degree of asymmetry noted in the second peak of activity with the skeletal muscle extract was variable from one experiment to another. When homogeneous preparations of rabbit skeletal and bovine heart muscle cyclic AMP-dependent protein kinase were rechromatographed using this same type of column, they eluted in their original positions.2 This would indicate that the elution profile obtained with the crude extracts was probably not affected by nonspecific protein-protein interactions.
Sodium Dodecyl Sulfate Gel Electrophoresis-Purified rabbit skeletal muscle cyclic AMP-dependent protein kinase and the bovine heart enzyme were essentially homogeneous as judged by sodium dodecyl sulfate gel electrophoresis (Fig. 2). The catalytic subunits of both enzymes had the same relative mobility corresponding to a molecular weight of about 40,000. However, different mobilities were found for the regulatory subunits, corresponding to molecular weights of 48,000 and 55,000 for the skeletal muscle and heart enzymes, respectively. This difference in mobility was found when either the holoenzymes or the isolated regulatory subunits were used. Phosphorylation of the regulatory subunit of the heart enzyme resulted in a small but reproducible decrease in mobility to a position that corresponded to a molecular weight of 59,000 FIG. 1. DEAE-cellulose column profiles of rabbit skeletal muscle (A) and beef heart muscle @I) protein kinase activity. Extracts were prepared as previously described (13) from 40 g of muscle. Separate but otherwise identical columns (1.6 x 30 cm) were eluted at the same flow rate with a common NaCl gradient (10 to 400 mM; total volume 260 ml). Fraction volumes were 2.5 ml. -, absorbance at 280 nm; ---, conductivity; protein kinase activity in the absence (0) and presence (0) of 2 PM cyclic AMP, using 5-~1 aliquots of the corresponding fractions.
FIG. 2. Electrophoresis of purified proteins on sodium dodecyl sulfate polyacrylamide gels. Proteins were preincubated with 2% sodium dodecyl sulfate and 5% 2-mercaptoethanol for 60 min at 30'. The following proteins were applied (from left to right): skeletal muscle holoenzyme (8 rg); skeletal muscle and heart holoenzyme (8 pg each). heart holoenzyme (8 rg), regulatory subunit of the skeletal muscle enzyme (R,) (4.2 rg), regulatory subunit of the skeletal muscle (RI) and heart enzyme (R,,) (4.2 pg each) and regulatory subunit of the heart enzyme (R,,) (4.2 pg). C, catalytic subunit. After gel electrophoresis of the reaction mixture in the presence of sodium dodecyl sulfate, the gel was stained with Coomassie brilliant blue dye, destained, and intensity of stain was recorded at 550 nm. The gel was next cut into 2-mm slices. Slices were digested with H,O, and counted for radioactivity. The solid line is Asso; the shaded area is counts per min of JT per slice. Bands corresponded to (from left to right): regulatory subunit of the heart protein kinase, regulatory subunit of skeletal muscle protein kinase, and catalytic subunits of the skeletal muscle and heart protein kinase. Cyclic AMP Binding-In experiments carried out at protein kinase concentrations similar to those found in muscle (15,27) it was found that the rabbit skeletal and heart muscle protein kinase each bound 2 mol of cyclic AMP per mol of holoenzyme ( Fig. 5) with apparent K,' of 0.1 jtM for the skeletal muscle enzyme and 2.8 PM for the heart enzyme. The presence of 'K, is arbitrarily defined as that concentration of free cyclic AMP required for the binding of 1 mol of cyclic AMP per mol of holoenzyme. This number varies with enzyme concentration and may be influenced by interactions between the regulatory and catalytic subunits. MgATP did not change the total amount of cyclic AMP bound to either enzyme; however, the apparent K, for cyclic AMP were increased lo-fold to 1.0 KM for the skeletal muscle enzyme but decreased about 6-fold to 0.5 PM for the heart enzyme by MgATP. The same values were obtained when incubations were performed for 5, 15, or 30 min. Using phosphorylated heart enzyme that had been extensively dialyzed to remove MgATP (see "Methods"), an apparent Kb for cyclic AMP of 0.4 PM was obtained, showing that phosphorylation of the regulatory subunit and not the presence of MgATP was responsible for the decrease in the apparent dissociation constant.
Scatchard or Lineweaver-Burk plots of the cyclic AMP binding data were nonlinear in the absence of MgATP but became nearly linear in its presence. Hill plots of the binding data gave n values of 1.2 and 0.85 for the skeletal muscle and heart enzyme, respectively, in the absence of MgATP (Fig. 6). In the presence of MgATP an n value of 1.0 was obtained for both enzymes. The same value was found when phosphorylated heart enzyme was used in the absence of MgATP. These studies would indicate that in the absence of MgATP or phosphorylation, cyclic AMP binds in an apparent positively cooperative manner to the skeletal muscle protein kinase (n value > 1) and in an apparent negatively cooperative manner to the nonphosphorylated heart enzyme (n value <l). These apparent cooperative properties were no longer found when MgATP was present.

Sedimentation
of Protein Kinases and Their Purified Subunits in Sucrose Density Gradients-The holoenzyme forms of the rabbit skeletal and heart muscle protein kinases sedimented as single species in the absence of cyclic AMP with an slO,W of 7.0 in each case as shown in Fig. 7, A  skeletal muscle protein kinase and the bovine heart enzyme dissociated into their respective dimeric regulatory subunitcyclic AMP complexes and catalytic subunits in the presence of high concentrations of cyclic AMP (Fig. 7. C and D). However, the concentration of cyclic AMP needed to effect dissociation was different for the two enzymes (Table I). In the absence of MgATP the skeletal muscle enzyme was completely dissociated at 0.15 pM cyclic AMP, whereas 10 j&M cyclic AMP was required to dissociate the heart enzyme. In the experiment of Table I it was assumed that complete dissociation of the holoenzymes had occurred when the same s20,w value was obtained for the cyclic AMP binding peaks as for the isolated regulatory subunit-cyclic AMP complexes of the respective enzymes. At intermediate concentrations of cyclic AMP single symmetrical binding peaks having intermediate s20,w values were seen. In the presence of MgATP the concentration of cyclic AhlP that was required to bring about complete dissociation of the skeletal muscle enzyme was increased to around 1.0 MM, but for the heart enzyme MgATP decreased the cyclic AMP requirement to about this same value. It will be noted that the concentrations of cyclic AMP required to dissociate the two protein kinases with or without MgATP were approximately the same as those required for the binding of 2 mol of cyclic AMP to the holoenzymes under the same conditions (refer to Fig. 5).
As in the cyclic AMP binding studies (see above), it was shown that the effect of MgATP on the dissociation of the heart protein kinase was due to phosphorylation of the enzyme, and not simply to the presence of MgATP. This was demonstrated by the experiment of Fig. 8 in which the apparent sZO,w value of the cyclic AMP binding peaks were plotted as a function of the cyclic AMP concentration for nonphosphorylated enzyme (upper curve) and phosphorylated enzyme from which MgATP had been removed by dialysis (lower curve).
Recombination of Isolated Subunits-In previous communications it was shown (6, 7) that MgATP facilitates recombination of skeletal muscle protein kinase subunits resulting in formation of the original cyclic AMP-dependent protein kinase. In the experiments shown in Fig. 7, E and F, the effect of MgATP on recombination of skeletal muscle and heart enzyme subunits was compared. Isolated subunits of skeletal muscle protein kinase mixed in a 1:l ratio (based on the molecular weights of the monomer subunits) recombined in the presence of MgATP to yield a cyclic AMP-dependent enzyme with an sZO.w value of 7.0 (Fig. 7E). When the skeletal muscle enzyme subunits were mixed in the same ratio but in the absence of MgATP no recombination occurred (data not shown). However, when the isolated regulatory and catalytic subunits of the heart protein kinase were incubated together in a 1:l ratio complete recombination occurred in the absence (Fig. 7F)  The skeletal muscle and heart protein kinases also differ in a number of ways. In the absence of MgATP, the apparent K, 7800 values and Hill coefficients for cyclic AMP binding are markedly different. In the presence of MgATP the apparent K, for cyclic AMP binding increases for the skeletal muscle enzyme but decreases for the heart kinase. The concentration of cyclic AMP required for dissociation of the respective holoenzymes is different and is also shifted in the opposite direction by MgATP. Finally, the heart enzyme does not require MgATP for recombination in contrast to the skeletal muscle enzyme. The differences noted between the rabbit skeletal muscle and heart holoenzymes appear to be due largely to differences in their regulatory subunits as shown by differences in sl,,+ values, mobilities on sodium dodecyl sulfate gel electrophoresis, and their ability to serve as phosphate acceptors. It would appear that the catalytic subunits present in both enzymes are similar in all properties tested to date (21,27).
The effect of MgATP on the apparent K, for cyclic AMP binding when the skeletal muscle protein kinase was used is in good agreement with previously reported results (15,30). However, the effects of the nucleotide on the dissociability of the heart enzyme reported in this study differ somewhat from those reported by Erlichman et al. (19). These investigators reported that phosphorylation was required for dissociation of the enzyme; whereas, the present study indicates that both the phospho and dephospho forms of the enzyme can be dissociated by cyclic AMP but that the dephospho form requires higher concentrations for dissociation to occur. The reasons for the discrepancy between these observations and those reported in the other study (19) are not known at present, but may reflect differences in the conditions employed or the preparations and concentrations of enzymes used. In particular, the higher temperatures (20") employed in these studies might be expected to affect the binding of cyclic AMP and the interaction between the regulatory and catalytic subunits.
The physiological significance, if any, of the differences found between the heart and skeletal muscle enzymes is not clear. However, it seems possible that the difference in the apparent K, values for cyclic AMP binding may be important in vivo. For the heart enzyme this property can be modified by phosphorylation, and one might expect that the state of the enzyme with respect to this parameter may be important in the control of its activity by cyclic AMP, especially if it is found that a mechanism exists for regulating phosphorylationdephosphorylation.
Changes in the activation properties of the skeletal muscle enzyme by cyclic AMP in viuo due to MgATP binding would seem less likely since MgATP is present in most tissues at concentrations greatly exceeding the dissociation constant for its binding. In any event, the results suggest that the activation of both enzymes is closely linked to the binding of cyclic AMP by the regulatory subunit, resulting in the release of an active catalytic subunit.
The dissociation constants' for cyclic AMP reported in this study are higher than those reported previously (15,18,20). This is probably due to the fact that both enzymes were studied at relatively high concentrations close to those determined for the enzymes in uivo (15,27). An almost identical activation constant5 for cyclic AMP has been reported (15) when physiological concentrations of the skeletal muscle enzyme were used. These higher values strongly suggest that in each tissue only a small part of the protein kinase is dissociated and activated when basal concentrations of cyclic AMP are present.
It is of interest that the differences between the heart and skeletal muscle protein kinases described in this communication may hold true for cyclic AMP-dependent protein kinase isoenzymes present in other tissues which have similar elution parameters on DEAE-cellulose (21).