Purification and characterization of rabbit liver calmodulin-dependent glycogen synthase kinase.

A rabbit liver cAMP-independent glycogen synthase kinase has been purified 4500-fold to a specific activity of 2.23 mumol of 32P incorporated per min per mg of protein using ion exchange chromatography on DEAE-Sephacel and phosphocellulose, gel filtration chromatography on Sepharose 6B, and affinity chromatography on calmodulin-Sepharose. This synthase kinase, which was completely dependent on the presence of calmodulin (apparent K0.5 = 0.1 microM) and calcium for activity, also catalyzed the phosphorylation of purified smooth muscle myosin light chain but not of smooth muscle myosin. Using 0.5 mM ATP, a maximal rate of phosphorylation of glycogen synthase was achieved in the presence of 10 mM magnesium acetate with a pH optimum of 7.8. Gel filtration experiments indicated a Stokes radius of about 70 A and sucrose density gradient centrifugation data gave a sedimentation coefficient of 10.6 S. A molecular weight of approximately 300,000 was calculated. A definitive subunit structure was not determined, but major bands observed after polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate corresponded to a doublet at 50,000 to 53,000. The calmodulin-dependent glycogen synthase kinase incorporated about 1 mol of 32P per mol of synthase subunit into sites 2 and 1b associated with a decrease in the synthase activity ratio from 0.8 to about 0.4. The calmodulin-dependent glycogen synthase kinase may mediate the effects of alpha-adrenergic agonists, vasopressin, and/or angiotensin II on glycogen synthase in liver.

A rabbit liver CAMP-independent glycogen synthase kinase has been purified 4500-fold to a specific activity of 2.23 pmol of 32P incorporated per min per mg of protein using ion exchange chromatography on DEAE-Sephacel and phosphocellulose, gel filtration chromatography on Sepharose 6B, and affinity chromatography on calmodulin-Sepharose. This synthase kinase, which was completely dependent on the presence o f calmodulin (apparent KO, = 0.1 p~) and calcium for activity, also catalyzed the phosphorylation of purified smooth muscle myosin light chain but not of smooth muscle myosin. Using 0.5 mM ATP, a maximal rate of phosphorylation of glycogen synthase was achieved in the presence of 10 M M magnesium acetate with a pH optimum of 7. 8

. Gel filtration experiments indicated a
Stokes radius of about 70 A and sucrose density gradient centrifugation data gave a sedimentation coefficient of 10.6 S. A molecular weight of approximately 300,000 was calculated. A definitive subunit structure was not determined, but major bands observed after polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate corresponded to a doublet at 50,000 to 53,000. The calmodulin-dependent glycogen synthase kinase incorporated about 1 mol of "P per mol of synthase subunit into sites 2 and l b associated with a decrease in the synthase activity ratio from 0.8 to about 0.4.
The calmodulin-dependent glycogen synthase kinase may mediate the effects of a-adrenergic agonists, vasopressin, and/or angiotensin 11 on glycogen synthase in liver.
The activity of glycogen synthase, which catalyzes the ratelimiting step in the biosynthesis of glycogen, is regulated in the cell both by allosteric modifiers and by covalent phosphorylation-dephosphorylation. Each 85,000-to 90,000-dalton subunit of glycogen synthase contains several phosphorylation sites which give rise to multiple forms of glycogen synthase. Phosphorylation and inactivation of the physiologically active a form of glycogen synthase to the inactive b form is a complex * This investigation was supported by National Institutes of Health Grant AM 17808. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore he hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. +Most of this work was submitted to Vanderbilt University in partial fulfilment of the requirements for the Ph.D. degree in Physiology. Present address, Laboratory of Molecular Cardiology, National Heart, Lung, and Blood Institute, National Institutes of Health, Building 10, Room 7B09, Bethesda, MD 20205. reaction involving at least three different classes of protein kinases: cyclic nucleotide-dependent, calcium-dependent, and cyclic nucleotide-and calcium-independent synthase kinase (for reviews see Refs. 1

and 2).
Cyclic AMP-dependent protein kinase can catalyze the incorporation of up to 3 mol of phosphate per mol of synthase subunit. Two mol are located in the COOH-terminal region (sites l a and lb) and 1 mol in the NH2-terminal domain or site 2 (3)(4)(5). Several investigators have reported the partial purification, primarily from skeletal muscle, of a class of protein kinases which phosphorylate and inactivate glycogen synthase in a manner independent of CAMP (6-9). These kinases are also not affected by added calcium, EGTA,' or calmodulin. Recent studies suggest that there may be several different CAMP-independent glycogen synthase kinases; however, their physiological function(s) remain(s) unknown. In general, these kinases seem to phosphorylate either site 2 or 3 (3, 5). Phosphorylase kinase, a calmodulin-containing enzyme, from liver or skeletal muscle catalyzes the incorporation into site 2 of 0.5-0.7 mol of P per mol of synthase subunit with partial inactivation of the synthase (10,11).
In rat liver, a-adrenergic agonists, vasopressin, and angiotensin I1 have been shown to promote the inactivation of glycogen synthase, presumably by a calcium-mediated process (12). Furthermore, Garrison et al. (13) reported that treatment of isolated hepatocytes with vasopressin or angiotensin I1 led to a calcium-dependent increase in the phosphorylation of glycogen synthase. We therefore became interested in purifying calcium-dependent synthase kinases from liver. We reported the existence in liver of a calcium, calmodulindependent protein kinase distinct from phosphorylase kinase which appears to be specific for glycogen synthase (14)(15)). This paper describes the extensive purification and some of the characteristics of this enzyme.

Methods
The glycogen synthase activity ratio (-glucose-6-P/+glucose-6-P) is defined as the ratio of activities determined in the absence and presence of glucose-6-P.
EGTA was included in the buffer during the initial steps in order to dissociate the calmodulin-dependent glycogen synthase kinase from calmodulin and to prevent calcium-dependent proteolysis. Calmodulin-dependent glycogen synthase kinase was purified from rabbit liver as described in the Miniprint. The enzyme eluted from the calmodulin-Sepharose column with EGTA was utilized for the following studies unless indicated otherwise.
Chemical a n d Physical Properties-The effect of pH on the phosphorylation of glycogen synthase by calmodulin-dependent glycogen synthase kinase was studied using several different buffers: Mes (pK, = 6.15), Pipes (pK,, = 6.801, Hepes (pK, = 7.55), and glycylglycine (pK, = 8.40) (Fig. 5). The pH optimum for the phosphorylation was around 7.8. In many instances, individual buffers rather dramatically affected the glycogen synthase kinase activity. For example, at pH 7.0, calmodulin-dependent synthase kinase activity was over 3fold higher using Hepes buffer rather than Pipes buffer. Moreover, the kinase activity at pH 8.0 was about 2-fold higher using Hepes rather than Tris (not shown). Consequently, a buffer combination of Mes, Hepes, and glycylglycine was used. The inset of Fig. 5 illustrates the effect of pH (using the buffer combination) on phosphorylation of glycogen synthase by the calmodulin-dependent kinase confirming that the pH optimum was about pH 7.8. The maximal rate of phosphorylation was achieved using a final concentration of 10 mM magnesium acetate (and 0.5 mM ATP). Preliminary studies suggest a strong dependence on Mg'+ for activity and that neither 10 mM MnCL nor 10 mM CaC12 could substitute for magnesium (Fig. 6).

Liver Catmodulin-dependent Glycogen Synthase Kinase
Upon sucrose density gradient centrifugation, the calmodulin-dependent glycogen synthase kinase migrated as a single, symmetrical peak with a sedimentation coefficient of 10.6 S * 0.1 (Fig. 7). The kinase activity toward glycogen synthase was completely dependent on calmodulin. Data obtained from gel filtration experiments on a calibrated Sepharose 6B column indicated a Stokes radius of 70 8, (Fig. 3). Based upon the sedimentation coefficient, the Stokes radius, and an assumed partial specific volume of 0.725 cm"/g (30), an approximate molecular weight of the calmodulin-dependent glycogen synthase kinase was calculated to be 300,000. A value of 1.57 was determined for the frictional ratio from the calculated molecular weight, the sedimentation coefficient, and the assumed partial specific volume.
The subunit composition of this enzyme is currently being investigated. Two major protein bands were resolved by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate with molecular weights of about 53,000 and 50,000 (Fig. 8). In some preparations, a faint band was observed at 19,000.
Phosphorylation of Myosin Light Chain-The CaM-dependent synthase kinase does not phosphorylate myosin light chain (M, = 20,000) isolated from either cardiac or skeletal muscle (14). Because the myosin light chain from smooth muscle has a different sequence around the phosphorylation site (Table 11) and is a substrate for other kinases (31), we tested it as a substrate for our kinase. Indeed, the purified  smooth muscle light chain was quite a good substrate with initial rates about 40-50% of those obtained with synthase as substrate (Table 111). Polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate showed that all the radioactivity was in the M, = 20,000 myosin light chain (not shown). However, when smooth muscle myosin was tested as substrate, there was no phosphorylation of the endogenous light chain (Table 111). In these experiments, we again confirmed that the myosin light chains from cardiac and skeletal muscle were not phosphorylated at a significant rate (not shown). Also, smooth muscle light chain kinase did not phosphorylate synthase in confirmation of earlier work (32).
Glycogen Synthase Phosphorylation Site Specificity-We have previously reported that the calmodulin-dependent synthase kinase can rapidly phosphorylate site 2 in glycogen synthase (14, 15). Synthase was phosphorylated to 0.9 mol of "'P/mol of subunit and subjected to CNBr cleavage followed by disc gel electrophoresis in the presence of sodium dodecyl sulfate. Approximately 40% of the '"P was in cyanogen bromide peptide 1 and 60% in cyanogen bromide peptide 2. In other experiments, synthase containing about 1 mol of :"P/ subunit was subjected to tryptic digestion. The tryptic '"Ppeptides were analyzed by peptide mapping on reverse phase high performance liquid chromatography (Fig. 9). The peak eluting at 70 min corresponds to site l b and the peak at 116 min corresponds to site 2 (5). The distribution of ""P between sites Ib and 2 was 34 and 66%, respectively. synthase was phosphorylated to about 1 mol of ,''P/mol of subunit using the calmodulin-dependent synthase kinase. The ."P-synthase was digested with trypsin (I mg/ml for 5 h) and the peptides were mapped by reverse phase high performance liquid chromatography using a gradient of 1-propanol (0-25% in 100 min, 25-50% in 10 min) in 0.1%) trifluoroacetic acid.

DISCUSSION
Calmodulin-dependent glycogen synthase kinase was isolated from rabbit liver and purified 4500-fold to a specific activity of 2.23pmol of'"'P incorporated into glycogen synthase per min per mg of protein at 30 "C. Most of the calmodulinsensitive glycogen synthase kinase recovered was eluted from the calmodulin-Sepharose with EGTA and NaC1; consequently, the studies detailed in this paper pertain to that fraction (unless otherwise stated). The calmodulin-sensitive kinase which did not adsorb to calmodulin-Sepharose (see Fig. 4), about 20% of the total activity, was not due to overloading the column since it did not adsorb when reapplied to different affinity columns varying the concentration of calcium and the pH. This fraction has not yet been studied extensively. Based upon data obtained utilizing other calmodulin-dependent enzymes, some possible explanations for the resolution of two glycogen synthase kinases with differing affinities for calmodulin by chromatography on calmodulin-Sepharose are as follows: the existence of isozymes of the enzyme (33), the existence of phospho-and dephospho-forms of the enzyme (34), or the partial degradation by proteolysis of the native enzyme (35). Only fresh, unfrozen livers were used for the preparation of the calmodulin-dependent glycogen synthase kinase. In addition, the livers were homogenized in the cold in the presence of 0.25 M sucrose, 4 mM EGTA, 2 mM EDTA, and several protease inhibitors. Also, several protease inhibitors and either EDTA and/or EGTA were included in the buffers throughout the early purification stages. Consequently, although it cannot be conclusively ruled out, it is unlikely that the calmodulin-dependent activity which did not bind to the affinity column is a proteolytic breakdown product. The relationship between the two calmoduliq-dependent glycogen synthase kinases must await further studies on the structural and enzymatic characterization of the two forms.
We previously reported (Fig. 1 of Ref. 14) that chromatography on Sepharose 6B resulted in the resolution of the calmodulin-dependent kinase from a calmodulin-independent synthase kinase. In the current purification scheme, the calmodulin-independent synthase kinase was removed by the 40 mM NaCl wash of the DEAE-Sephacel column step. This glycogen synthase kinase, which is unaffected by either CAMP or calmodulin, has not been studied sufficiently to comment further.
Based upon a Stokes radius of about 70 A, a sedimentation coefficient of 10.6 S, and an assumed partial specific volume of 0.725 cm"/g, a M,of about 300,000 and a frictional ratio of about 1.6 were calculated for calmodulin-dependent glycogen synthase kinase. An unequivocal subunit structure has not been defined however, the major bands observed after polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate corresponded to molecular weights of approximately 50,000 to 53,000. Presently, it has not been established whether these represent two different subunits or result from phosphorylation or proteolysis of a single subunit.
Liver calmodulin-dependent glycogen synthase kinase appears to be distinct from other reported kinases. This enzyme is active toward skeletal muscle glycogen synthase, liver glycogen synthase, and purified smooth muscle myosin light chain, but inactive toward skeletal muscle phosphorylase b, liver phosphorylase b, myosin light chain from skeletal and cardiac muscle, native smooth muscle myosin, liver pyruvate kinase, liver phosphofructokinase, histone IIA, casein, and the regulatory subunit of type I1 CAMP-dependent protein kinase (14,15). 5 The ability of this kinase to phosphorylate isolated light chains of smooth muscle but not intact myosin is similar to the cyclic AMP-dependent protein kinase (31) and indicates that the reaction is probably not of physiological significance. Both liver calmodulin-dependent glycogen synthase kinase and skeletal muscle phosphorylase kinase can phosphorylate the serine which is located 7 residues from the NH2 terminus of skeletal muscle glycogen synthase (site 2) (11). In addition to site 2, the calmodulin-dependent glycogen synthase kinase also phosphorylated site l b (Fig. 9). The decrease in the synthase activity ratio can probably be attributed to the phosphorylation of site 2. Only phosphorylase kinase can phosphorylate the serine 14 residues from the NH2 terminus of phosphorylase 6. There is considerable homology between residues 4-15 from the NH2 terminus of glycogen synthase (11) and residues 11-22 in phosphorylase b (36). Therefore, it is interesting to note that the calmodulin-dependent glycogen synthase kinase will phosphorylate glycogen synthase but not the homologous sequence in phosphorylase b. An interesting common feature of all the known substrates of liver CaMdependent synthase kinase is the sequence -R-X-X-S(P)as shown in Table 11. Studies using synthetic peptide substrates may provide an answer to these specificity differences.
It has recently been shown that the subunit structure of liver phosphorylase kinase is similar if not identical with the skeletal muscle enzyme including the presence of calmodulin (37). While phosphorylation of glycogen synthase by calmodulin-dependent glycogen synthase kinase is completely dependent on calmodulin and inhibited by the phenothiazine trifluoperazine, this is not the case for liver phosphorylase kinase. The two enzymes also differ in the effect of excess magnesium. Magnesium in concentrations exceeding the concentration of ATP profoundly inhibited both nonactivated and activated liver phosphorylase kinase ( 3 7 , 3 8 ) . In contrast, the calmodulin-dependent glycogen synthase kinase was most active with the concentration of magnesium greatly exceeding that of the ATP. Calmodulin-dependent glycogen synthase kinase is also distinguished from liver phosphorylase kinase by their respective observed pH optima for phosphorylation and the reported holoenzyme molecular weights (37).
Furthermore, the calmodulin-dependent synthase kinase has been purified from the livers of the New Zealand strain of rat with a glycogen storage disease (gsd/gsd) characterized by the absence of liver phosphorylase kinase (39).
Calmodulin-dependent glycogen synthase kinase also differs from myosin light chain kinase from cardiac and skeletal muscle which, although dependent on calmodulin, is unable to phosphorylate glycogen synthase. A third calmodulin-stimulated protein kinase, which phosphorylates protein I, appears to be localized in neural tissue and was not detected in rat liver homogenates (40). Thus, calmodulin-dependent glycogen synthase kinase does not appear to be related to three of the known calmodulin-dependent kinases: phosphorylase kinase, myosin light chain kinase, and the calcium, calmodulin-stimulated protein I kinase.
Many investigators have shown that the effects in rat liver of a-agonists, vasopressin, and angiotensin I1 do not involve CAMP or the CAMP-dependent protein kinase (41, 42). Calcium, however, has been demonstrated to be involved in their actions on the liver. It is well established that in rat liver CYadrenergic agonists such as epinephrine or phenylephrine promote phosphorylase activation by a calcium-dependent mechanism(s) (12). While the breakdown of glycogen in the liver is stimulated by catecholamines, glycogen synthesis is inhibited. The natural catecholamine epinephrine, the synthetic catecholamine phenylephrine, and vasopressin have been shown to promote the inactivation of liver glycogen synthase as well as the activation of phosphorylase in a calcium-dependent manner (12, 42).
Studies by Garrison and co-workers (13) in isolated hepatocytes suggested that treatment with vasopressin or angiotensin I1 led to a calcium-dependent increase in the phosphorylation of phosphorylase b, glycogen synthase, and pyruvate kinase. Our data indicate that the calmodulin-dependent glycogen synthase kinase will not catalyze the phosphorylation of either liver phosphorylase b or liver pyruvate kinase. Although, to our knowledge, there is no report of studies in liver in which glycogen synthase inactivation was not accompanied by phosphorylase activation , Strickland et al. (12) recently demonstrated quantitative differences. In isolated hepatocytes stimulated by epinephrine, the activation of phosphorylase occurred more rapidly than the inactivation of glycogen synthase. In addition, vasopressin and epinephrine were both shown to be more potent in inactivating glycogen synthase than in activating phosphorylase.
In conclusion, a calcium, calmodulin-dependent protein kinase which phosphorylates and inactivates glycogen synthase has been purified from liver. This enzyme may mediate the actions of a-adrenergic agonists, vasopressin, and angiotensin I1 on glycogen synthase in liver. In one possible model, these agents interact with specific receptors located on the plasma membrane; this results in the mobilization of calcium from mitochondria (Fig. 10). Calcium then presumably binds to the ubiquitous calcium-binding protein, calmodulin, forming an active conformer. This active conformer subsequently binds to the inactive calmodulin-dependent glycogen synthase kinase. As a result, an active enzyme would be formed which would then phosphorylate and inactivate glycogen synthase.