On the Mechanism of Regulation of Type I Phosphoprotein Phosphatase from Bovine Heart REGULATION BY A NOVEL INTRACYCLIC ACTIVATION-DEACTIVATION MECHANISM VIA TRANSIENT PHOSPHORYLATION OF THE REGULATORY SUBUNIT BY PHOSPHATASE-1 KINASE (FA)*

Adenosine 5’-(y-thio)triphosphate (ATPyS) can substitute for ATP in the activation of the ATP.MgZ+- dependent form of bovine heart type I protein phosphatase (M, = 75,000) catalyzed by phosphatase-1 ki- nase (FA). ATPrS activates the enzyme to a lower level than ATP, but it phosphorylates the regulatory (R)- subunit to a much higher extent. An [36S]phosphatase-1 ([36S]E-P) has been isolated, identified, and shown to be a key intermediate in the activation reaction. Treatment of [36S]E-P with dimethyl suberimidate results in cross-linking of the M, = 34,000 [36S]R-s~b~nit with the M, = 40,000 catalytic (C)-subunit to form a M, = 75,000 species, indicating that phosphorylation is not accompanied by dissociation of the holoenzyme. The catalytically active form (E,) is not the phosphorylated enzyme intermediate. Instead, E, is directly produced from the intermediate by a Mg2+-dependent, intramo- lecular autodephosphorylation reaction. The isolated E, derived from [36S]E-P or from ATP-activated phos-phatase-1 has the same half-life (23 min

effect. Limited trypsinization selectively digests the Rsubunit and the resulting C-subunit is Mg2+-dependent. Based on the present data, a novel intracyclic activation-deactivation mechanism via transient phosphorylation of the R-subunit is proposed for regulation of phosphatase-1.

Active
Type I phosphoprotein phosphatase (phosphatase-1) was first designated by Cohen (1) as an operational term for a rabbit skeletal muscle phosphatase activity that is sensitive * This work was supported by United States Public Health Service, National Institutes of Health Grant HL-22962. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ To whom correspondence should be addressed.
to inhibition by heat-stable proteins, inhibitor-1 (1-1)' and inhibitor-2 (1-2), and exhibits high specificity towards the psubunit of phosphorylase kinase (1-3). It turns out that the holoenzyme form of phosphatase-1 requires the simultaneous presence of M$+, ATP, and a protein activator, termed FA, for expression of its activity (4-11). The holoenzyme form was discovered and designated by Merlevede and co-workers (10) as ATP. Mg*+-dependent phosphoprotein phosphatase or Fc.M, where FC and M represents catalytic component and modulator protein (i.e. inhibitor-2), respectively (4-10). F A , which was found to exhibit a protein kinase activity towards glycogen synthase (4,5), was subsequently shown to be similar to glycogen synthase kinase 3 (12).
Electrophoresis-Electrophoresis on 10 or 12.5% polyacrylamide gel in the presence of sodium dodecyl sulfate (SDS-PAGE) was carried out according to Laemmli (27). Proteins in samples were precipitated with 12.5% trichloroacetic acid prior to electrophoresis. Gels were stained with Coomassie Brilliant Blue R-250.
The subunits of phosphatase-1 were quantitated by densitometric analysis of Coomassie Blue-stained gels, using an EC-980 Densitometer with a Hewlett-Packard 3380 Reporting Integrator. The color values of different subunits were assumed to be proportional to their molecular weights as determined by SDS-PAGE.
Radioactive labeled proteins on gels were located by autoradiography with the aid of Kodak lanex intensifying screens. For quantitative determination of 32P and 35S incorporation into the R-subunit of phosphatase-1, the portion of gels containing the R-subunit was cut out and placed in a vial containing 0.5 ml of 30% HZ02, and incubated at 60 "C for 16 h to dissolve polyacrylamide for determining the radioactivity by liquid scintillation counting.
Preparation of Phosphatase-1 and Other Proteins-Phosphatase-1 was purified from bovine cardiac muscle by a procedure involving (NH&SO, fractionation, DEAE-cellulose, Sephacryl S-200, and polylysine-Sepharose chromatographies (28) similar to that described by Yang et al. (4). SDS-PAGE analysis of several preparations indicated that the sum of the R-subunit (34K) and the C-subunit (40K) represented about 10-35% of the total proteins in these preparations, as estimated by densitometry scanning of the gels. Phosphatase-1 kinase (FA) was purified from the same tissue by a procedure involving DEAE-cellulose, Sephacryl S-200, and Red A Matrex Gel (Amicon) chromatographies (29). [32P]Phosphorylase a was prepared by phosphorylation of crystalline phosphorylase b from rabbit skeletal muscle with phosphorylase kinase and [y-32P]ATP as previously described (30,31). Inhibitors-1 (32) and -2 (33) were purified from rabbit skeletal muscle.
Phosphatase Assay-The activity of phosphatase-1 was measured by the release of 32Pi from [32P]phosphorylase a. Two different assay methods were employed. Method I (Direct Assay) was for measuring the spontaneous activity: phosphatase-1 (0.5-5 milliunits) was incubated at 30 "C for 10 min in a volume of 50 pl containing 50 mM Tris.HC1, pH 7.4, 0.2 mg/ml BSA, 1 mM theophylline, and 10 p~ [32P]phosphorylase a. 32Pi released was determined as previously described (30). Method I1 (Preincubation Assay) was for measuring the FA. ATP. Me-activated activity: phosphatase-1 was preincubated at 30 "C for 20 min in a volume of 40 pl containing 62.5 mM Tris.HCI, pH 7.4, 0.25 mg/ml BSA, 1.25 mM theophylline, 6.25 mM MgCl,, 62.5 p~ ATP, and 0.25 milliunits/ml FA to convert the enzyme into its activated form. The phosphatase reaction was then initiated by the addition of 10 pl of 50 p~ [32P]phosphorylase a. The reaction was carried out at 30 "C for 10 min. One unit of phosphatase-1 activity is defined as 1 nmol of 32Pi released per min.
Inhibitor-2Assay-For determining the inhibitory activity, 1-2 was included in the preincubation mixture as described in the Preincubation Assay. Phosphatase-1 of 5 milliunits was used. For measuring the 1-2 activity associated with the R-subunit, phosphatase-1 was first heated at 95 "C for 3 min to inactivate the C-subunit. To locate 1-2 activity on gels after SDS-PAGE of phosphatase-1, gels were sliced into 1-mm sections. Proteins in each section were extracted with solution A (20 mM Tris.HC1, 0.2 mg/ml BSA, pH 7.4) and used as an 1-2 source.

The concentration of [35S]E-P was expressed in terms of its 35S
content. SDS-PAGE analysis revealed that the protein staining patterns of [35S]E-P were identical to those of the native enzyme, indicating that no significant proteolysis occurred during the preparative process.
Autodethwphosphorylation A~say-[~~S]Phosphatase-l was incubated at 30 "C with 50 mM Tris. HCl, pH 7.4, 0.2 mg/ml albumin, and 5 mM MgC1, for an appropriate period of time. Aliquots were then withdrawn and mixed with cold trichloroacetic acid to 16%. The denatured proteins were precipitated by centrifugation and 35S released in the supernatant was determined by liquid scintillation counting.
Isolation of the Activated Form of Phosphatase-1 -[35S]Phosphatase-1 (120 nM hound "S, 60 pg/ml protein) was incubated at 30 "C in a volume of 0.2 ml containing 50 mM Tris. HCI, pH 7.4,0.2 mg/ml BSA, and 5 mM MgC12. After a 60-min incubation, about 50-60% bound %S was released accompanied with the generation of spontaneous phosphatase activity. The enzyme was immediately separated from the activator, M e , and unbound 35S by a rapid gel filtration procedure as follows. Sephadex G-50 superfine presaturated with solution A was packed to a bed volume of 2 ml in a 3-ml plastic syringe and precentrifuged at 1000 x g for 4 min at 4 "C to remove excess buffer. Then 0.15 ml of the reaction mixture was applied onto the column, centrifuged at 1000 x g for 3 min, and the desalted enzyme was collected. The enzyme solution so obtained was free from unbound 35S and stored at 4 "C prior to use for deactivation studies.
To prepare the activated form derived from native phosphatase-1, the enzyme (5-50 pg/ml) was incubated at 30 "C for 30 min in a volume of 0.2 ml containing 50 mM Tris. HCl, pH 7.4, 0.2 mg/ml BSA, 5 mM MgC12, 50 p~ ATP, and 0.2 units/ml FA. The enzyme was immediately separated from its low molecular weight activators, M e and ATP, by centrifuging through a Sephadex G-50 column as described above. The procedure removed more than 99.5% of ATP as determined by [-y-32P]ATP.
Other Methods-Cross-linking of the [35S]R-subunit to the Csubunit of phosphatase-1 by dimethyl suberimidate was carried out by the method of Davies and Stark (34). Protein concentration was determined by the method of Lowry et al. (35), following precipitation by trichloroacetic acid. Bovine serum albumin was used as a standard.

RESULTS
Time Course of Activation by ATP and ATPyS-As shown in Fig. 1, the activation by ATP does not parallel the incorporation of 32P into the R-subunit ( Fig. la) while that by ATPyS does (Fig. lb). Furthermore, ATPyS activates the phosphatase activity a t a much slower rate (about 5-7% that of ATP) and to a much lower extent (above 20-25% that of ATP) but is accompanied with a much higher level of phosphorylation (incorporation of 35S into the R-subunit is at least 30-fold higher than "P). Addition of excess EDTA to block the kinase reaction results in a rapid decrease of both the ATPyS-and the ATP-activated activities without affecting the level of either the bound 35S or the bound 32P.
The data are interpreted to mean that both the rate and extent of activation are determined by the turnover rate of the phospho(or thiophospho) group on the R-subunit but not by the extent of its phosphorylation (or thiophosphorylation) as detected at selected time points during the course of the activation reactions. In other words, it is the dephosphorylation step, rather than the phosphorylation step which is responsible for generation of the phosphorylase phosphatase activity. On the basis of this concept, the differences between the activation-phosphorylation curves presented in Fig. 1, a and b, can be readily explained by assuming that ATPyS is a much poorer substrate for the FA kinase activity. Therefore, the rate of phosphorylation of the R-subunit is much faster (20,000 cpm/pmol). At the indicated time, aliquots of 0.1 ml were removed and the R-subunit was separated by SDS-PAGE for determining its 3zP content (0) as described in the text. A separate incubation mixture, identical except that nonlabeled ATP replaced (y-32P]ATP was used to monitor the spontaneous activity (m) by the Direct Assay as described in the text. At the indicated time, aliquots of 10 pl were removed, mixed with 0.2 ml of solution A, and IO pl of the diluted enzyme were used for assay. The arrow indicates the time at which EDTA was added to the preincubation mixtures so that its concentration was 1.8 mM in excess of Mg+. b, activation by ATPyS.
The exueriments were identical to those described in a except that [35S]ATPyS replaced [y-32P]ATP and ATPyS replace2 ATP.
than that of thiophosphorylation. In the case of ATP activation, the rate of dephosphorylation is faster than that of phosphorylation and the reverse is true for the ATPyS activation reaction. The observed EDTA effects suggest that the presence of Mg2+ is essential for dephosphorylation of the phosphoenzyme intermediate and/or for maintenance of the phosphatase activity generated. The fact that the ATPySand the ATP-activated activities exhibit identical response to EDTA indicates that they are similar to each other. The time course of activation-phospho~lation shown in Fig. l a is similar to those obtained with reconstituted (13,14) or native (17) phosphatase-1 from skeletal muscle. , however, found that activation was closely parallel to phosphorylation and, therefore, proposed that the phosphorylated enzyme was the active species.
Isolation of [35S]Phosphatase-I -The observations that: ( a ) the level of 35S incorporation is high and ( b ) the thiophosphorylation is stable in the presence of EDTA (Fig. l b ) indicate that it is feasible to isolate the thiophosphoenzyme intermediate for detail analysis of its properties. To do this, phosphatase-1 was incubated for a prolonged period of time with ["'SSJATPyS, Mg2+, and FA to insure maximum thiophosphorylation. The resultant [35S]E-P was then isolated by repetitive (NHJ2SOa precipitation and extensive dialysis in the presence of 0.1 mM EDTA (details are presented under "Experimental Procedures"). No significant release of 35S was observed after stored at -20 "C for at least 6 months. Fig. 2a, the isolated [35S]E-P is eluted from Sephacryl S-200 as a single peak of phosphorylase phosphatase activity corresponding to an apparent M, = 89,000, which is identical to that of the native phosphatase-1. The elution profile of the phosphatase activity coincides with that of: (a) 35S-protein and ( b ) the heat-stable inhibitory activity associated with phosphatase-1. In order to identify the thiop~osphorylated component(s), SDS-PAGE of the active fractions was carried out. As shown in Fig. 2b (right panel), protein staining bands corresponding to M, = 40,000 and 34,000, respectively, are enriched in the active peak fractions. Autoradiography demonstrates that only the 34K staining band is radioactive (Fig.  2b, left panel). An identical, unstained gel was sliced into 1mm sections and protein in each section was extracted and assayed for inhibitory activity towards the FA.ATP.M$+activated native phosphatase-1. The results indicate that the inhibitory activity is associated with the M, = 34,000 species (data not shown). Densitometry tracing of the Coomassie Blue-stained gels reveals that the molar ratio of the 40K and the 34K species is about 1:l. In a separate study, we have purified the phosphatase to near homogeneity by an improved purificaton procedure.' SDS-PAGE analysis of the enzyme preparation shows only two major protein staining bands of M , = 40,000-41,000 and 34,000-35,000, respectively, in a molar ratio of about 1:1. The two species account for more than 85% of the proteins detectable in the enzyme preparation? These and results by others (16) have led US to conclude that the 40K protein represents the C-subunit of the phosphatase.

Molecular Property of the Isolated f5S]E-P-As shown in
If thiophospho~lation of the 34K R-subunit does not result in dissociation from the 40K C-subunit, then, these two components should be able to be cross-linked by dimethyl suberimidate to form a 75K species. As shown in Fig. 3 (lower panel), incubation of [35S]phosphatase-l with increasing concentrations of dimethyl suberimidate causes a progressive decrease in the 34K and the 40K staining bands with a concomitant increase of a 75K species on the SDS-PAGE. Autoradiography of the same gel shows a decrease of a 34K radioactive band and appearance of a major doublet radioactive band corresponding to M, = 75,000 (Fig. 3, upper panel).  A minor radioactive band corresponding to M, = 100,000 is also detected.
Conversion of the Isolated p'S]E-P to a Catalytically Active Form (Ed-As predicted by the data shown in Fig. 1   Aliquots were removed for determining "S released (El) and spontaneous phosphatase activity (0) by the Direct Assay as described in the text.

Mechanism of Regulation of Type I P h o s p~~r o t e i n Phosphatase
indicates that the catalytic site responsible for autodethiophosphorylation (designated as Site A) differs from that (designated as Site S) for the genuine substrate, phospho~lase a.
Therefore, phosphorylase a fails to act as a competitive inhibitor with respect to the thiophospho group on the R-subunit. Table I shows that Mn2+ and Mg2+ generate similar levels of phosphorylase phosphatase activity but Mn2+ is more effective than Mg2' in stimulating 35S release.
As shown in Fig. 4 The arrow indicates the time at which EDTA was added to the preincubation mixtures so that its concentration was 1.8 mM in excess of M$+.  Fig. 5. The data indicate that the autodephosphorylation-activation process is an intramolecular rather than an intermolecular event. The inset in Fig. 6 shows that 35S release follows first order kinetics. The apparent rate constant, kobs, is estimated as 0.013 min" which is equivalent to a half-life (tth) of 53 min.

Isolation of the Activated Form (Ed Derived from [35SJE-P
and from Native Phosphatase-1 Activated by ATP-[35S]E-P and native phosphatase-1 were separately converted to catalytically active forms by incubating at 30 "C with Mg2+ alone and with FA, ATP plus M$+, respectively. The reaction mixtures were quickly centrifuged through Sephadex G-50 at 4 "C to separate proteins from Mg2+ along with other low molecular components as described under "Experimental Procedures." More than 90% of the phosphatase activity was recovered after the separation process. The enzymatic activity was relatively stable at 4 "C despite that it deactivated much faster at 30 "C as discussed in the following sections. The data indicated that, once converted to activated form, the presence of free M$+ is not required for the catalytic activity. Con#ersion of E, to Resting State (E,)-As shown in Fig. 7, incubation of t.he activated form derived from either f35S]E-P or from native phosphatase-1 at 30 "C results in a timedependent loss of activity, The loss represents a deactivation rather than a denaturation process, since the enzyme can be fully reactivated by incubating with F A . ATP. Mg2+. Thus, the ?e I Phosphoprotein Phosphatase 6421 deactivation is a reversion of E, to its resting state, E,. Fig. 7 shows that the phosphatase activity derived from [35S]E-P exhibits an identical deactivation rate as that from its normally phospho~lated counterpart consistent with the data shown in Fig. 1. The inset in Fig. 7 shows that deactivation follows first order kinetics. The kObs is estimated as 0.03 min" which corresponds to t% = 23 min. We have measured the rate of deactivation with a wide range of enzyme concentrations and found they are the same. Thus, the deactivation of E,, similar to aut~ethiophospho~lation-activation of [3%]E-P, is also an intramolecular process. It should be noted that the deactivation rate of isolated E, (Fig. 7) appears to be much slower than those observed in the complete activation systems (Fig. 1). The difference is not likely to be due to acceleration by EDTA, since the deactivation rate shown in Fig. 5 is identical to that shown in Fig. 7. The reason for the difference is not yet understood.
We have asked the question of whether E. can catalyze dethiophosphorylation of ["SIE-P in the absence of M e . If it can, then, one should observe a continued release of 35S from [35S]E-P after removing MS2' from the activation mixture by Sephadex G-50. Instead, we have found that, in a mixture containing both E, and [35S]E-P in the absence of M$+, the bound 35S is stable at 4 or 30 "C, consistent with the results presented in both Figs. 1 and 5, which demonstrate that phosphorylation is stable after addition of EDTA. The data indicate that E, is specific for phosphorylase a and does not catalyze the hydrolysis of the phospho-group on the Rsubunit of another enzyme molecule.

Effects of Chelators on the Activated Forms Derived from p5S]E-P and Native P h o s p~t a s e -1 Activated by ATP-As
shown in Table 11, at 1 mM concentration EDTA and EGTA show little effects on accelerating the deactivation of E. derived from either [35S]E-P or the native enzyme. By contrast, PPi is more effective than ATP. The accelerating effects are not due to irreversible inactivation of the enzyme, since both the ATPor the PPi-treated enzymes can be fully reactivated by incubating with FA. ATP, M$+. Table I1 also shows that the E, derived from the native enzyme is slightly more sensitive to both ATP and PPr than that derived from [35S] E-P. The reason for this difference is not understood. Fig. 8 shows that the accelerating effect of PPi on deactivation of E, derived from native phosphatase-1 is dependent on PPi concentration. Moreover, the effect of PPi can be cancelled by the presence of excess Mg2+ (5 mM). The data are consistent with the interpretation that E, contains an essential cofactor, Mg"+, in the catalytic site. PPi and ATP, due to their structural similarity to the phosphoprotein substrate and their chelating ability, can readily enter catalytic sites to remove Mg2+ and, thus, can accelerate the deactivation of E,. The presence of excess Mg2' abolishes the PPi chelating function and, therefore, can cancel its effect. The ineffectiveness of EDTA and EGTA may be due to the fact that they bear no structural similarity to the substrate and, thus, cannot readily enter catalytic sites to exert their chelating function.
Limited Trypsinization of Phosphatase-1-As shown in Fig.  9, native phosphatase-1 shows little activity when measured with M g + alone. Incubation of the enzyme with 2 rg/ml of trypsin at 4 "C in the absence of added divalent cation results in a rapid increase of a ~g"+-activated activity to the level of the activity measured with the simultaneous presence of FA. ATP"$+. This is accompanied with the disappearance of the 34K R-subunit (Fig. 9, inset). Mn2+ (1 mM) or Co2+ (1 mM) can substitute for M P in the activation of the trypsinized enzyme and Co2+ is slightly more effective. The activity measured with MnZ* is about 1-fold higher than that of M$' . Effects of chelators on the activated form of phosphatase-1 derived from native or 35S-enzyme The activated phosphatase-1 preparations, which had been isolated from low molecular weight effectors, were incubated at 30 "C in a mixture containing 50 mM Tris. HCl, pH 7.4,0.2 mg/ml BSA without and with the addition of 1 m M EDTA, EGTA, ATP, and PPi as indicated. After 10 min, aliquots were removed for measuring the spontaneous and the Fa.ATP. Me-activated activities by Direct and Preincubation Assay, respectively. The parenthesis indicates the percent activity of those without the addition of a chelator.  The activated phosphatase-1 derived from native enzyme, which had been separated from ATP and Mg2+ as described in the text, was incubated at 30 "C in a mixture containing 50 mM Tris. HCI, pH 7.4, and 0.2 mg/ml BSA with 5 mM MgCI2 (A) or with various concentrations of PP, as indicated, in the absence (0, B) and presence (0) of 5 mM MgC1,. After 10 min, aliquots were removed for measuring the spontaneous activity (0, 0, A) by Direct Assay and the FA.ATP. Mgz+-activated activity (W) by Preincubation Assay as described in the text.
The M$+-activated activity is not due to a contaminating phosphatase since it exhibits similar sensitivity as the native enzyme to inhibition by 1-1 and 1-2. Chromatography of the trypsin digest (10 min incubation) on a Sephacryl S-200 column revealed that the M$+-activated activity had an apparent M , of 35,000-40,000 in contrast to M , = 89,000 observed for native phosphatase-1 (not shown). The data are interpreted to mean that the R-subunit imposes an inhibitory effect on the C-subunit and confers the sensitivity of the holoenzyme to activation by F A . ATP. Mg+. Selective digestion of the R-subunit by trypsin mimics the FA.ATP . M c activation by removal of the constraint imposed on the Csubunit without insertion of the essential cofactor, M$+, into the catalytic site. Therefore, the free C-subunit remains dependent on Mg2+ for activity despite that it is already in an activated conformation. It should be noted that in this particular preparation used for trypsinization studies, the molar ratio between the Rand C-subunit is lower than 1. We have observed that the R-subunit is much more sensitive to proteolytic attack than the C-subunit in purification and storage, and this is accompanied by a decrease in sensitivity to activation by FA. ATP. M$+. The inset in Fig. 9 also shows that incubation of the enzyme with trypsin longer than 10 min results in degradation of the 40 K C-subunit to 30-35K fragments. The fragments are likely to possess catalytic activity since no significant loss of the Me-activated activity is found after 10 min incubation (Fig. 9). Brautigan et al. (24) reported that trypsinization of a M , = 83,000 skeletal muscle enzyme (presumably phosphatase-1) in the presence of Mn2+ or M$+ resulted in a M , = 35,000 species with concomitant activation of the phosphatase activity. M$+ was less effective than Mn2+ in activation (24). Recent studies from the same laboratory, however, indicated that the free C-subunit (38K), isolated by chromatography of a purified holoenzyme (70K) on polyanion SI resin at pH 6.0, could be activated by the presence of Mn2+ or Co2+ alone but not by M$+ or Ca2+ (16).

DISCUSSION
We have found that ATPyS can substitute for ATP in the activation of phosphatase-1. A 35S-labeled thiophosphorylated enzyme has been isolated, identified, and shown to be a key intermediate in the activation processes. We further demonstrate that the enzymatically active form of phosphatase-1 is not the phosphorylated enzyme intermediate. Instead, it is directly produced from the intermediate by a Mg2+-dependent autodephosphorylation reaction. The data do not support the hypothesis that the phosphorylated enzyme is the catalytically active form as suggested by several laboratories (14-18). The present studies demonstrate that the autodephosphorylation of the intermediate is an intramolecular event rather than a process catalyzed by a separate activated phosphatase as suggestedby others (13,14,16,17). The availability of isolated 35S-labeled thiophosphorylated enzyme in nondenatured form has provided a powerful means for investigating the molecular properties of phosphatase-1. These experiments have provided clear cut evidence that: ( a ) phosphatase-1 is a heterodimer of 75K consisting of a C-subunit of 40K and an inhibitory R-subunit of 34K and ( 6 ) in the process of activation, the phosphorylated R-subunit remains in association with the C-subunit. The data are in agreement with those by Villa-Moruzzi et al. (16) and Jurgensen et al. (17), but differ from those by Hemmings et al. (13,14) who reported that phosphorylation of the 1-2 component of a reconstituted phosphatase-1 resulted in its dissociation from the catalytic component, leading to expression of the enzymatic activity. The dissociation mechanism was further supported by the observation that cross-linking of the 1-2 with the catalytic component prevented the enzyme from being activated by FA. ATP. M%+ (14).
In contrast to the present results, several studies indicated that ATP-yS could not replace ATP in the activation of either native (4, 7,lO) or reconstituted (13, 14) phosphatase-1 from skeletal muscle, despite the fact that it could thiophosphorylate the 1-2 component (13,14). Furthermore, Hemmings et al. (13) reported that thiophosphorylation prevented the enzyme from being activated by FA. ATP. Mg+. The reason for these discrepancies is not clear.
The R-subunit of phosphatase-1 has generally been referred to as 1-2 (1-5, 12-16, 18-20, 23) or modulator protein (6-10, 17), reflecting the current concept that it may be a free regulator protein in uiuo, similar to 1-1. This concept has stemmed from historical ground. Both 1-1 and 1-2 had been discovered (37)(38)(39)(40)(41), extensively studied (1-3, 10, 23), and considered as free regulatory proteins in tissue long before any knowledge concerning the molecular composition of the phosphatase-1 holoenzyme was known. However, there is no experimental evidence indicating the 1-2 exists as a free form in uiuo. On the other hand, several lines of evidence indicate that it does exist as an integral subunit of the ATP.M$+dependent phosphatase. ( a ) We have found that 1-2 co-purifies with the ATP. Mf-dependent phosphatase from bovine heart throughtout our purification procedures (28).* ( b ) Yang et al. (9) have reported that it co-purifies with the phosphatase from rabbit skeletal muscle throughout their purification procedures. ( c ) Ballou et al. (15) have found that the molar ratio between 1-2 and C-subunit in the purified phosphatase-1 is 1:l. ( d ) Resink et al. (14) have estimated the in vivo concentrations of 1-2 and C-subunit in rabbit skeletal muscle and concluded that the molar ratio between them is approximately 1:l. The fact that the 1-2, but not C-subunit, is extremely sensitive to proteolytic attack, also works against the free regulatory protein concept. We propose that i n uiuo, 1-2 exists as an integral subunit of phosphatase-1 holoenzyme. Therefore, we designate it as "R-subunit," replacing the conventional terms "1-2" and "modulator protein." Based on the present data, a model for activation of phosphatase-1 from cardiac muscle is presented in Fig. 10. In the resting state, phosphatase-1 is inactive (E,). Upon phosphorylation of the R-subunit, the enzyme changes its conformation, leading to the formation of active Site A, which is specific for the phospho group on the R-subunit of the same enzyme molecule (E-P). Site A requires M%+ as a cofactor for expressing its catalytic activity (E; re). The M$+ could be derived directly from the M%+ .ATP complex or in rapid equilibrium with free M%+ in the medium. Intramolecular autodephosphorylation of the R-subunit with concomitant insertion of M%+ results in the transformation of Site A into Site S on the C-subunit, which is specific for the extramolecular substrate, phosphorylase a. The activated enzyme (E,) has a finite life span. It spontaneously reverts to E,. During this intramolecular event, Mg2+ is released and E, is then ready for the next round of activation. In this model, M%+ plays a central role: ( a ) it forms a M%+.ATP complex to serve as a substrate for phosphatase-1 kinase and ( b ) it serves as a cofactor for Sites A and S . The model indicates that continued hydrolysis of ATP is required for the maintenance of the activated form, consistent with the observation by Jurgensen et al. (17) that, in activation of phosphatase-1, F A exhibits an ATPase activity.  The relative rates among the three steps in the activationdeactivation cycle, E, -E* MgZ+ " + Eo " + E,, are estimated as follows. For the ATP-and A T P~S -s u p~~d processes, the orders would be kz L k, 2 k3 (0.030 min") and k, > ks (0.030 min") > k2 (0.013 min"), respectively, where k, (ATP) >> kl (ATPrS), k2 (ATP) >> 122 (ATP+), and k3 (ATP) = k3 (ATP+). These are consistent with the characteristic time courses for the ATP-and A T P~S -s u p p o~d activations shown in Fig. 1, a and b, respectively.
Considerations concerning the possible conformational changes of the two subunits during the activation-deactivation cycle are presented as follows. It is assumed that the C-and R-subunit in E, are in the T (tight) and R (relaxed) states, respectively (Fig. 10). The effects of phosphorylation would be in transforming the R-subunit into a partially T state which allows the formation of Site A on the C-subunit. The energy released from autodephosphoryiat~on reaction drives the R-subunit into a transient-stable T state and allows the transformation of Site A to Site S. The R-subunit spontaneously reverts to its R state which forces the C-subunit back to the inactive T state with concomitant release of M$+. The transition of the R-subunit from the R to the T state would result in a decrease of its physical interaction with the Csubunit and vice versa. The postulation is consistent with the observation by  that the phosphorylated form of phosphatase-1 is more susceptible than the nonphosphorylated form to dissociation by polyanion SI chromatography at pH 7.0. It also offers an explanation for the observation that phosphorylation caused dissociation of reconstituted phosphatase-1 (13,14). An alternate model is that the conformation state of Site A is identical to that of Site S. The failure of phosphorylase a to inhibit autodephosphorylation of E-P may be explained by assuming that the active site has been occupied by the phospho-group on the R-subunit and, therefore, is unaccessible to extramolecular phosphoprotein substrate. The third model is that the C-subunit exists only in the R state and does not change its conformation during the activation-deactivation cycle. In such a model, the active site in E, is covered by the R-subunit in its R state.
The energy released from the phosphorylation-autodephosphorylation cycle drives the R-subunit into a transient-stable T state and the active site is thus uncovered (E,,). The spontaneous reversion of the R-subunit from the T to the R state results in again covering up the active site (E,). We favor the first model (Fig. lo), since it represents the most general case. It offers a plausible explanation for the relationship between the phosphorylat~on-dephosphorylation cycle of the R-subunit and activation-deactivation of the phosphata~ activity. It also explains the facts that: ( a ) ATP hydrolysis is required for each activation-deactivation cycle, (b) the free C-subunit is in an activated state, and (c) the free R-subunit is spontaneous inhibitory. The present model indicates that formation of the active site is a consequence of R-subunit dislocation. Therefore, added free R-subunit (1-2) could readily occupy the active site. The postulation offers an explanation for the 1-2 inhibition on the FA. ATP. MP-activated activity and is consistent with the kinetic studies that 1-2 acts as a competitive inhibitor with respect to phosphorylase a (42).
The present data (Fig. 9) and results by others (6, 10, 14-15) indicate that the R-, but not the C-subunit, is extremely sensitive to proteolysis. Therefore, proteolytic damage of the R-subunit is likely to occur during puri~cation man~pulation. Minor damage on the phosphorylation site or other functional parts of the R-subunit may not result in dissociation of the holoenzyme or reduction of its apparent size but may impair the regulatory function of the R-subunit. If proteolysis occurs in E,, the resulting enzyme would exhibit an increase in sensitivity to activation by divalent cation alone, since the Csubunit has been transformed to the R state without incorporation of the cofactor. These explain the general observations that the sensitivities of phosphatase-1 holoenzyme to activation by Mn2+ aIone/FA.ATP.M~+ may vary from one preparation to another. BaHou et aZ. (15) reported that the sensitivity of phosphatase-1 towards Mn2+ increased progressively during purification. Addition of protease inhibitors throughout preparation of phosphatase-1 allowed the recovery of an enzyme preparation that was poorly activated by Mn2+ (15). A similar phenomenon has also been observed by others (6, lo).
The R-subunit in Ea is responsible for transformation of the C-subunit from the R to T state and for expelling M$+ from the active site. Therefore, proteolytic damage of the Rsubunit in E, will result in locking the enzyme in its catalytically active form indefinitely. Various molecular forms of phosphatase-1 ( M , = 30,000-260,000) which are spontaneously active and insensitive to stimulation by Mn2+, have been isolated from skeletal muscle by several laboratories (1-3, 10, 19, 20, 23). These may be derived from E. in which the R-subunit has been either dissociated or damaged by proteolytic attack. Villa-Moruzzi et al. (16) reported that the free Csubunit dissociated from the FA. ATP. Mg2f-activated phosphatase-1 (E,) by polyanion SI chromatography was spontaneously active and slightly inhibited by Mn2+ while that from the native, inactive enzyme (Er) was Mnz+-dependent, consistent with the present interpretations. The C-subunit derived from E, is much more stable than that from E, (16). Thus, it is reasonable to assume that the C-subunit (or its degradative products) derived from E, is more resistant than that from E, to harsh dissociation processes, such as precipitation by 80% ethanol at room temperature. Therefore, the low molecular forms of phosphatase-1 (Mr = 30,00~35,000) purified by procedures involving ethanol treatment may be derivatives of E, rather than E,. Indeed, these enzyme preparations are generally spontaneously active and insensitive to stimulation by 13,14,18,19).
The present data indicate that, if phosphorylation occurs at a single Thr residue, incorporation of either phosphate or thiophosphate into the R-subunit will never reach 1:l molar ratio, since the phosphorylation is transient.  reported that phosphorylation of the R-subunit is far below the 1:1 molar ratio, in agreement with our data. Hemmings et al. (13), however, reported that thiophosphorylation stoichiometry approached 1 mol of 35S incorporated per mol of reconstituted enzyme. If so, phosphorylation of the enzyme might occur at multiple Thr residues.
The present studies indicate that metabolites that affect the rate and extent of: ( a ) phosphorylation of E, by phosphatase-1 kinase; ( b ) the autodephosphorylation of E -P and ( c ) the deactivation of E, are of potential importance in regulating the extent as well as duration of phosphatase-1 activity.
Recently, DePaoli-Roach (18) reported that the R-subunit of phospha~se-1 could be phospho~iated at Ser residue by casein kinase 11. The action of casein kinase I1 alone did not result in activation of the phosphatase. However, it could potentiate the extent of phosphoryIation at Thr residue and activation of the phosphatase activity catalyzed by suboptimal concentrations of FA. The possibility that the requirement of M$+ for autodephosphorylation-activation reaction is particular to the thiophosphorylated enzyme intermediate is unlikely since removal of Mg2+ by EDTA results in stabilization of both phosphorylated and thiophosphorylated enzyme intermediates (Fig. 1). Table I indicates that the ratio between phosphatase activity generated and 35S released from [=S]E-P with M$+ as an activator is higher than that with Mn2+. The data are interpreted to mean that M$+, but not Mn2+, is the natural cofactor for phosphatase-1. Therefore, M$+ is efficiently incorporated into the active site at the correct position where it exerts the cofactor function more effectively than Mn2+. The interpretation is supported by the observations that Mn2+ is only about 50-60% as effective as M P in activation of phosphatase-1 holoenzyme catalyzed by F A with ATP as a substrate (6-8, 15, 16, 28, 26). The present results indicate that M$+ is less effective than MnZ+ in the activation of trypsinized phosphatase-1. The data may be explained by assuming that M e but not Mn2+ requires the action of Rsubunit for its incorporation into the active site in an efficient and correct way. The possibility that Mn2+ might exert its activating effect by a mechanism different from that of M$+, however, cannot be excluded. All the current published data indicate that MgZ+ alone is ineffective in activating phosphatase-1, either in the form of the free C-subunit or holoenzyme (1-3,10,15,16,23). The discrepancy may be due to differences in experimental conditions. It has been suggested that the activation of phosphatase-1 by Mn2+ may have resulted from a Mn2+-catalyzed covalent modification of the enzyme, perhaps involving sulfhydryl groups of the C-subunit (16). Such a reaction has been proposed by Yan and Graves (43). The present studies, however, favor the hypothesis that activation is accompanied by insertion of M$+ into the C-subunits. Evidence that supports the hypothesis includes: (a) M$' is required for autodephosphorylation-activation of E-P (Fig. 5), ( b ) deactivation of E. can be accelerated by a metal ion chelator, PPi, whose effect can be blocked by Mg2+ (Fig.  8), and ( c ) the free C-subunit produced by limited trypsinization can be activated by Mg2+ (Fig. 9). It has been suggested that most of the enzymes involved in transferring phosphate groups should have at least one cationic group suitably disposed near the active site in order to facilitate the complexation and orientation of the phosphate group (36, 44). It therefore seems reasonable to assume that MgZ* may serve as a cationic site. Recently, we have shown that Caz+ and calmodulin act as allosteric activators in transforming phosphatase-2 (calcineurin) into a relaxed conformation which required the presence of Mg2+ or a transition metal ion for expressing its catalytic activity (44)(45)(46) supporting the notion that M$+ may represent an essential cofactor in catalysis for phosphoprotein phosphatases in general.
The regulation of phosphatase-1 by intracyclic activationdeactivation reaction via transient phosphorylation of the Rsubunit represents a novel regulatory mechanism in covalent modification of enzymatic activity. In conventional covalent modification, phosphorylation results in locking the modified enzyme in either an active (e.g. phosphorylase a ) or an inactive (e.g, glycogen synthase b ) state and the reversal requires a separate phosphatase. By contrast, the intracyclic activation-deactivation mechanism involves the formation of a transient-activated form which requires continued utilization of ATP for enzyme phospho~lation to maintain the activated form. In such a mechanism, the activity of only t.he kinase in conjunction with the intrinsic activation-deactivation reactions determine the level and duration of the enzymatic activity. Such a mechanism, which involves a built-in automatic turn-off device, would be more sensitive than the conventional extracyclic phosphorylation-dephosphorylation mechanism in controlline the duration of the activated state