Purification and characterization of glycogen synthase from a glycogen-deficient strain of Saccharomyces cerevisiae.

Chromatography of wild-type yeast extracts on DEAE-cellulose columns resolves two populations of glycogen synthase I (glucose-6-P-independent) and D (glucose-6-P-dependent) (Huang, K. P., Cabib, E. (1974) J. Biol. Chem. 249, 3851-3857). Extracts from a glycogen-deficient mutant strain, 22R1 (glc7), yielded only the D form of glycogen synthase. Glycogen synthase D purified from either wild-type yeast or from this glycogen-deficient mutant displayed two polypeptides with molecular masses of 76 and 83 kDa on sodium dodecyl sulfate-gel electrophoresis in a protein ratio of about 4:1. Phosphate analysis showed that glycogen synthase D from either strain of yeast contained approximately 3 phosphates/subunit. The 76- and 83-kDa bands of the mutant strain copurified through a variety of procedures including nondenaturing gel electrophoresis. These two polypeptides showed immunological cross-reactivity and similar peptide maps indicating that they are structurally related. The relative amounts of these two forms remained constant during purification and storage of the enzyme and after treatment with cAMP-dependent protein kinase or with protein phosphatases. The two polypeptides were phosphorylated to similar extent in vitro by the catalytic subunit of mammalian cyclic AMP-dependent protein kinase. Phosphorylation of the enzyme in the presence of labeled ATP followed by tryptic digestion and reversed phase high performance liquid chromatography yielded two labeled peptides from each of the 76- and 83-kDa subunits. Treatment of wild-type yeast with Li+ increased the glycogen synthase activity, measured in the absence of glucose-6-P, by approximately 2-fold, whereas similar treatment of the glc7 mutant had no effect. The results of this study indicate that the GLC7 gene is involved in a pathway that regulates the phosphorylation state of glycogen synthase.


Chromatography
of wild-type yeast extracts on DEAE-cellulose columns resolves two populations of glycogen synthase I (glucose-6-P-independent) and D (glucose-6-P-dependent) ( Glycogen synthase (EC 2.4.1.11), the rate-limiting enzyme for controlling glycogen biosynthesis, is regulated by covalent phosphorylation and dephosphorylation (Cohen, 1982(Cohen, , 1986Roach, 1986 whereas the dephosphorylated form is active in the absence of this allosteric activator (active, I form).' The enzyme from rabbit skeletal muscle can be phosphorylated on at least 10 different serine residues by a number of protein kinases. The effect of phosphorylation on glycogen synthase activity varies from no effect to marked inactivation, depending on the residue that is phosphorylated.
The glycogen synthases from liver (Roach, 1986), heart (Wolleben et al., 1987, brain (Inoue et al., 1987), and adipose tissue (Lawrence et al., 1986) also have multiple phosphorylation sites. In yeast, interconversion between two forms (D and I forms) of glycogen synthase has been found in vitro and in vivo (Rothman-Denes and Cabib, 1970;Huang and Cabib, 1972). Huang and Cabib (1974a) developed a method for purification of the D form of glycogen synthase from Saccharomyces cerevisiae using four chromatographic steps. They reported a molecular mass of 77 kDa and a minor proteolysis band of 71 kDa Cabib, 1974a, 1974b). Many other details of the structure and regulation of yeast glycogen synthase remain to be established, partly because of the difficulty in preparing sufficient quantities of the enzyme. Of additional interest is the existence of strains of yeast that are deficient in glycogen synthesis (Cannon et al., 1986;Rothman-Denes and Cabib, 1970).* Glycogen-deficient mutants with characterized defects have been shown to have increased activity of the CAMP-dependent protein kinase, either by inactivation of the regulatory subunit (bcyl) (Cannon and Tatchell, 1987;Toda et al., 1987a) or by increased CAMP levels (id) (Tanaka et al., 1989). However, mechanisms independent of CAMP have also been shown to he important in regulating glycogen accumulation in yeast (Cameron et al., 1988), raising the possibility that some glycogen-deficient mutants may have defects not related to the CAMP pathway.
We have developed an improved, more rapid, purification method that can be applied to either I or D forms of yeast glycogen synthase. We have used this method to purify glycogen synthase from wild-type and from a glycogen-deficient mutant, 22Rl (glc7). We report here that purified glycogen synthase from either strain displays two polypeptides with molecular masses of 76 and 83 kDa and provide evidence that the glc7 mutation lies in a gene regulating the phosphorylation state of glycogen synthase.
1 Classically, the terms I and D refer to glucose 6-phosphateindependent and -dependent forms, respectively. In this report we use the terms I and D to describe two chromatographically distinct populations that differ in relative dependence on glucose-6-P. This is not meant to imply the existence of only two discrete forms. A, mutant and wild-type yeast were grown in YEPD medium (1% yeast extract (w/v), 2% Bacto-peptone (w/v), 2% glucose) at 30 "C with shaking, collected by centrifugation, washed, and resuspended in YEPD medium. The cells were then plated onto YEPD plates and incubated at 30 "C for 3 days before being stained with iodine solution (0.025% iodine in 0.2% KI). M, the mutant strain 22Rl (glc7); IV, the wild-type strain C276. B, the crude extracts were made as described under "Experimental Procedures" from the mutant (M) and the wild-type (w) and activity was determined in the absence (open bars) and presence (solid bars) of glucose-6-P. Results are mean + S.E. of five independent experiments.

RESULTS
Glycogen and Glycogen Synthase in Wild-type Yeast and in Yeast Containing the glc7 Mutation-A collection of yeast mutants unable to accumulate glycogen was obtained from Dr. John R. Pringle (University of Michigan).
We assayed glycogen synthase activity in crude extracts from the mutants, looking for altered regulation of activity. In general these strains showed little change in total glycogen synthase activity but in some strains the glycogen synthase activity ratio was decreased. The glc7 mutant strain showed the greatest difference from wild-type and was chosen for further study. The glc7 mutant strain contains much less glycogen than the wildtype as shown by staining of similarly sized colonies of each with Ip (Fig. lA). In the presence of glucose-6-P the glycogen synthase activity in cell extracts of mutant and wild-type are similar, but in the absence of glucose-6-P the mutant has much less activity than the wild-type (Fig. 1B).
Chromatography of yeast cell extracts (Fig. 2) showed that two populations of glycogen synthase can be separated on DEAE columns as noted earlier by Huang and Cabib (1974a).
The later fraction, eluting at 0.25 M NaCl, has a lower activity ratio and is referred to as the D form. The early fraction (I form) has a higher activity ratio and frequently is not retained on DEAE columns. In some preparations the I form bound to 3 Portions of this paper (including "Experimental Procedures" and Figs. 2 and 3) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press. Polyacrylamide gel electrophoresis in 7.5% acrylamide under nondenaturing conditions was carried out at 4 "C essentially according to the procedure of Hedrick and Smith (1968). Followingelectrophoresis, one lane (7 rgof protein) was stained with Coomassie Blue and destained (A). One adjacent lane was sliced into 1.5-mm sections which were then incubated with 0.1 ml of the dilution buffer used for the assay of glycogen synthase activity (see "Experimental Procedures") at 4 "C overnight to extract proteins from the gel. Aliquots of these extracts were used directly for measuring glycogen synthase activity (B). Electrophoretic mobility in A and B is from left to right (toward the anode). The broad protein band stained by Coomassie Blue was sliced into three sections (lower, middle, and upper mobility regions, shown from left to right), subjected to SDS-PAGE and stained with Coomassie Blue (C). the column and was only partially separated from the D form. As can be seen in Fig. 2, extracts of mutant yeast contain substantially more D form and virtually no I form, a finding that is consistent with the activity ratio differences noted in cell extracts. Phosphate measurements (Buss and Stull, 1983) showed that glycogen synthase D from wild-type yeast contained 3.3 f 0.8 (mean f S.D., n = 2) phosphates/subunit and that glycogen synthase D from the glc7 mutant strain of yeast contained 3.1 f 0.1 (n = 3) phosphates/subunit. Characterization of Yeast Glycogen Synthme-Unless otherwise noted, the experiments described below used the D form isolated from the glc7 mutant. The purified enzyme exhibited two bands (76 and 83 kDa) when analyzed by SDS-PAGE4 (Fig. 3). Under nondenaturingconditions, an apparent molecular mass of about 300 kDa was obtained from gel filtration on TSK-G4000 HPLC (data not shown), consistent with earlier conclusions that yeast glycogen synthase is a tetramer (Huang and Cabib, 1974a). Analysis of the fractions from the TSK-G4000 column by SDS-PAGE revealed that the 76-and 83-kDa bands did not resolve and co-eluted with enzyme activity during gel filtration (not shown). The ratio of 76:83 kDa from most preparations of yeast glycogen synthase was about 4:l based on densitometry scanning after staining with Coomassie Blue. When the purified enzyme preparation was analyzed by gel electrophoresis in the absence of SDS a single broad protein band was seen ( Fig. 4-4). Measurement of glycogen synthase activity in slices from an adjacent lane showed that enzyme activity coincided with the protein band (Fig. 4B). When this protein band was sliced ' The abbreviations used are: SDS-PAGE, polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate; HPLC, high performance liquid chromatography.
into three sections and analyzed by SDS-PAGE, the 76-and 83-kDa polypeptides were present in similar ratios in each section (Fig. 4C).
In order to determine if the two polypeptides from yeast glycogen synthase are homologous, antiserum was raised against the 76-kDa protein and the antibody was affinitypurified as described under "Experimental Procedures." The affinity-purified antibody against the 76-kDa protein inhibited the glycogen synthase activity in both the crude extract and the pure enzyme (data not shown) and cross-reacted with the 83-kDa protein (Fig. 5). Antibody against the 83-kDa form purified with a similar affinity purification procedure also reacted with the 76-kDa form (not shown). Peptide maps obtained by using the Cleveland et al. (1977) technique and digestion with chymotrypsin followed by staining with Coomassie Blue (Fig. 6A) or by immunoblotting as described under "Experimental Procedures" (Fig. 6B) showed that the 76-and 83-kDa proteins exhibited a number of distinct peptides although many peptides were identical. The two proteins also yielded similar peptide maps when Staphylococcal V8 proteinase was used instead of chymotrypsin (data not shown). The relative amounts of the two bands did not change during purification (Fig. 5) suggesting that the 76-kDa form is not generated by proteolysis of the 83-kDa form. When yeast cells were harvested and extracted after different growth times the two proteins were found in the same protein ratio at all time points by Western blot analysis. These data show that yeast glycogen synthase is comprised of two related polypeptides which are both present at different growth phases of the yeast cell. Both bands were present in extracts of wild-type as well as the glc7 mutant showing that the mutation did not affect the distribution between these two forms. I form purified from the wild-type displayed the same two protein bands with a similar protein ratio (Fig. 3,  thuse-Purified yeast glycogen synthase D was phosphorylated in vitro by the catalytic subunit of CAMP-dependent protein kinase from bovine heart. After phosphorylation the samples were analyzed by SDS-PAGE and the two individual protein bands were cut out of the gel to determine radioactivity. The results showed that during a time course of phosphorylation, the radioactivity ratio for the two proteins agreed with the protein ratio (not shown), indicating that the kinase does not preferentially phosphorylate one of these two polypeptides. There was no evidence for conversion of one band to the other during phosphorylation or when the preparation was treated with mammalian type 1 or yeast type 2A phosphatases under conditions where the activity ratio increased from 19% to about 35% (data not shown). When the two phosphorylated subunits were cut out from the gel and digested separately with chymotrypsin in 18% SDS-PAGE as described under "Experimental Procedures," the two subunits yielded identical phosphopeptide maps with four phosphopeptides ranging from 15 to 30 kDa (data not shown). Thus the two subunits gave identical phosphopeptide maps, whereas protein staining demonstrated similar but distinct peptide maps (Fig. 6B). Under the phosphorylation conditions used for these experiments the only amino acid residue that was phosphorylated was serine (data not shown). In uitro phosphorylation of the I form of yeast glycogen synthase by the catalytic subunit of CAMP-dependent protein kinase resulted in 32P incorporation to the extent of about 0.6 mol/mol of subunit and most of the label was incorporated into the 76 kDa band, as noted above for the D form. The phosphorylation of yeast glycogen synthase (I form) was closely correlated with a decrease in the activity ratio (Fig.  7A) alytic subunit of CAMP-dependent protein kinase, tryptic phosphopeptides of the I form were analyzed by reversed phase HPLC (Fig. 8A). Two phosphopeptides, Tl and T2, eluted at 34 and 36% acetonitrile, respectively. The phosphorylation level on peptide T2 was approximately three times higher than that of peptide Tl. The D form of glycogen synthase from wild-type (Fig. 7B)  kinase, but without further inactivation of the enzyme. Analysis of the tryptic phosphopeptides showed that the peptide Tl was also labeled in the D-form enzyme but T2 was barely detectable (Fig. 8B). Separation of the 76-and 83-kDa bands by SDS-PAGE, followed by tryptic digestion and HPLC, showed that the 76-and 83-kDa bands each gave rise to Tl and T2. Effects of Li* on Yeast Glycogen Synthase-Previous studies in mammalian systems have shown that Li+ treatment of cells causes activation of glycogen synthase (Bosch et cd., 1986;Cheng et al., 1983;Haugaard et al., 1974), but a similar effect has not been described for yeast glycogen synthase. We therefore incubated yeast cells with 20 mM LiCl at 30 "C for 30 min before preparing extracts. Treatment of wild-type yeast in stationary phase with Li+ increased the activity ratio of yeast glycogen synthase from a control level of 40% to a level of 82% (Fig. 9). In contrast, Li' treatment of the mutant had no effect on the enzyme activity, indicating that the mutant is defective with respect to conversion of the D form to the I form. When NaCl was used in place of LiCl, no activation of glycogen synthase was observed. The increase in activity ratio was also observed after gel filtration of the crude extracts, indicating that the activation by Li' is not due to changes in a low molecular weight activator such as glucose-6-P (data not shown). The activity ratio for glycogen synthase is low during the logarithmic phase and rises sharply at the beginning of the stationary phase ( Fig. 9) in accordance with the observations of Rothman-Denes and Cabib (1970). Interestingly activation by Li+ was not seen in cells that were in the logarithmic phase of growth (Fig. 9).
Phosphtase Activity in Wild-type and Mutant Yeast-The failure of the glc7 mutant to activate glycogen synthase upon reaching stationary phase or upon Li+ treatment suggests a possible defect in phosphatase activity. In support of this hypothesis we have very recently determined by DNA sequence analysis5 that the GLC7 gene is the same as the DIS2 gene reported last year by Ohkura et al. (1989). The amino acid sequence of the predicted gene product shows 81% identity with mammalian protein phosphatase la, but enzymatic activity of the gene product has not been reported.
In view of these findings we compared the phosphatase activity of wildtype and glc7 mutant strains of yeast. We found that incubation of crude extracts from untreated wild-type or mutant yeast at 30 "C resulted in activation of glycogen synthase in both cases. Although accurate quantitative comparisons are not possible since the glycogen synthase substrate was not the same for the two assays, this experiment indicates that the mutant strain does not completely lack glycogen synthase phosphatase activity. Because of the difficulties in using glycogen synthase as a substrate in these experiments, yeast extracts were also assayed for phosphatase activity with different substrates under a variety of conditions. The substrates included [32P]phosphorylase from rabbit skeletal muscle, casein labelled in the presence of [T-~~P]ATP and CAMP-dependent protein kinase, and yeast glycogen synthase D labeled in the presence of [T-~~P]ATP and CAMP-dependent protein kinase. The selective inhibition of phosphatase 1 by inhibitor 2, the relatively selective inhibition of phosphatase 2A by okadaic acid, and the Mg2+ requirement of type 2C have been suggested as tools to differentiate these three phosphatases in assays of crude extracts of mammalian (Cohen et al., 1989a) and yeast (Cohen et al., 1989b) extracts. Using this approach we found no consistent difference between wild-type and glc7 mutant strains with respect to the activity of any of the phosphatases.
We also considered the possibility that the mutation resulted in formation of a form of glycogen synthase that is resistant to phosphatase action. However, treatment of purified glycogen synthase D from wild-type or mutant strains with a crude yeast phosphatase preparation from wildtype yeast increased the activity ratio from an initial value of 1560% in both cases.

DISCUSSION
Yeast glycogen synthase can be resolved chromatographitally into two forms, which are designated I and D based on their relative dependence on glucose-6-P (Huang and Cabib, 1972). This is in contrast to mammalian glycogen synthase for which the I and D forms cannot be resolved chromatographically.
The yeast D form elutes from DEAE-cellulose at 0.25 M KCl, whereas the I form elutes primarily in the flow through fraction, but in some preparations a significant amount could be eluted with a 0.2 M wash as reported previously by Huang and Cabib (1974a). We have modified the purification procedure they described by using ultracentrifugation to sediment glycogen synthase bound to high molecular weight glycogen. This allows the purification to be completed more rapidly so that proteolysis to a 71-kDa form is avoided.
Yeast glycogen synthase contains two subunits of 76 and 83 kDa which can be resolved by gel electrophoresis in the presence of SDS. This finding is a little different from that of an earlier publication of Huang and Cabib (1974a) in that they did not report the presence of the 83 kDa protein. The reason for this is not clear but could be due to inability to resolve the bands with the gel electrophoresis system (Weber and Osborn, 1969) they used. Using that system we noted only partial resolution of the 76-and 83-kDa forms. These two polypeptides comigrate during nondenaturing gel electrophoresis and show immunological cross-reactivity, suggesting that they are homologous. Peptide maps produced by proteolysis in SDS-PAGE reveal some differences in structure, although many fragments seem to be common to both the 83and 76-kDa subunits. We have not observed conversion of the 83-kDa form to the 76-kDa form during preparation or storage of the enzyme, suggesting they are not related by proteolysis. Alternatively the two bands may differ in degree of phosphorylation since DePaoli-Roach et al. (1983) and Camici et al. (1984) observed that phosphorylation of glycogen synthase can increase the apparent subunit molecular weights. However, we could not detect subunit molecular weight differences between the D form and the I form in yeast glycogen synthase and phosphorylation of yeast glycogen synthase by CAMP-dependent protein kinase from bovine heart did not result in the conversion of one protein band to the other, nor did treatment with protein phosphatases. Furthermore, the two proteins of the enzyme preparation can be phosphorylated to the same extent by the catalytic subunit of the mammalian CAMP-dependent protein kinase, which is homologous to the yeast CAMP-dependent kinases (Shoji et al., 1983;Showers and Maurer, 1986;Toda et al., 1987b). Although this argues against differences in phosphorylation state it is still possible that the two bands in yeast glycogen synthase represent differences in phosphorylation of one or more sites but that phosphorylation state at these sites is not altered under the conditions of these experiments. All attempts to separate these two bands under nondenaturing conditions have been unsuccessful. These conditions included gel filtration, electrophoresis, and chromatography on phosphocellulose, phenyl-Sepharose, concanavalin A-Sepharose, and heparin-sepharose (data not shown). This suggests that these two proteins are intimately associated and/or have similar native structures.
When yeast glycogen synthase I was phosphorylated by CAMP-dependent protein kinase, the activity ratio decreased and peptide mapping experiments yielded two labeled peptides, Tl and T2, with most of the phosphate incorporated into T2, suggesting that this peptide contains the site responsible for inactivation.
Phosphorylation of the D form led to very little incorporation into T2 suggesting that this peptide was already phosphorylated, as one would expect if phosphorylation of a residue in this peptide is responsible for inactivation of the enzyme. Whether or not phosphorylation of Tl leads to changes in activity ratio cannot be determined from these experiments, since the activity ratio of the D form was already quite low before phosphorylation.
Several glycogen-deficient strains of yeast have been identified, and two have been characterized biochemically. One of these mutations, bcyl, produces a defective regulatory subunit of the CAMP-dependent protein kinase, resulting in overexpression of kinase activity (Cameron et al., 1988). Recently, Tanaka et al. (1989) cloned and sequenced a gene that may be an inhibitory regulator of the RAS-CAMP pathway (IRAl). IRAl, previously called PPDl (Matsumoto et al., 1985), appears to be allelic to GLCl, which was isolated as a glycogendeficient mutant (Rothman-Denes and Cabib, 1970). We have examined the properties of glycogen synthase in the glc7 mutant strain to try to identify the defect in this mutant. This glycogen-deficient mutant has low glycogen content but normal or slightly elevated levels of glycogen synthase when assayed in the presence of glucose-6-P.
However, glycogen synthase in extracts of the mutant strain has a decreased activity ratio in comparison to the wild-type and chromatography of the extracts shows that little or no I form is present in the mutant. The data obtained in this study indicate that the glc7 mutation increases the amount of glycogen synthase D without affecting other properties of glycogen synthase, suggesting a defect in glycogen synthase phosphatase activity. Furthermore, DNA sequence analysis' indicates that the GLC7 gene encodes a protein phosphatase homologous to the catalytic subunit of mammalian protein phosphatase la. We have thus far been unable to obtain any definitive evidence for a defect in glycogen synthase phosphatase activity. It is possible that the conditions for assay of phosphatase were not suitable for detecting the difference in phosphatase activity between wild-type and mutant yeast. For example, the difference may be revealed only when the substrate is glycogen synthase phosphorylated at specific residues or when certain inhibitors or activators of the phosphatase are included in the assay. It is further possible that this putative phosphatase does not have properties similar to mammalian protein phosphatase 1 and that it dephosphorylates another protein involved in the regulation of glycogen synthase, e.g. the defect may lead to abnormal activation of a kinase. Clearly, more studies are needed to determine the nature of the defect in the glc7 mutant.
Li' activates glycogen synthase in several mammalian tissues (Bosch et al., 1986;Cheng et al., 1983;Haugaard et aZ., 1974). The results presented in this paper demonstrate that addition of Li' to yeast cells also causes glycogen synthase activity ratios to increase. Cheng et al. (1983) reported that treatment of rat adipocytes with Li+ caused activation of glycogen synthase without affecting pyruvate dehydrogenase or glycogen phosphorylase activity, suggesting that Li+ acts by stimulating a phosphatase or by inhibiting a protein kinase that is specific for glycogen synthase. Since no glycogen synthase phosphatase or kinase sensitive to Li+ has been described so far, there appears to be an intermediate step between the initial action of Li+ and the alteration of phosphatase or kinase activity. Interestingly Li+ did not increase the activity ratio in the glc7 mutant, giving further evidence that the glc7 mutation causes a defect in the regulation of glycogen synthase phosphatase or kinase.