Purification and Characterization of a High Molecular Weight Type 1 Phosphoprotein Phosphatase from the Human Erythrocyte*

The major Mn2+-activated phosphoprotein phosphatase of the human erythrocyte has been purified to homogeneity from the cell hemolysate. It is sensitive to both inhibitors 1 and 2 of rabbit skeletal muscle, preferentially dephosphorylates the @ subunit of the phosphorylase kinase, and dephosphorylates a broad range of substrates including phosphorylase a, p-nitro-phenyl phosphate, phosphocasein, the regulatory sub- unit of cyclic AMP-dependent protein kinase, and both spectrin (K,,, = 10 PM) and pyruvate kinase (K, = 18 MM) purified from the human erythrocyte. The purified enzyme is stimulated by Mn2+ and to a lesser extent by higher concentrations of M$+. The purification procedure was selected to avoid any change in molecular weight, hence subunit composi- tion, between the crude and purified enzyme. Mainte-nance of the original structure is demonstrated by non- denaturing gel electrophoresis and gel filtration chromatography. Gel filtration of the purified holoenzyme shows 5 single active component with a Stokes radius of 58 A at a molecular weight position of 180,000. Sedimentation velocity in a glycerol gradient gives a value of 6.1 for s20,w. Together these data indicate a molecular weight of about 135,000. Two bands of equal intensity appear on sodium dodecyl sulfate-gel electrophoresis at molecular weights of 61,700 and 36,300, suggesting a subunit composition of two 36,000 and one 62,000 Sedimentation Velocity Measurements-These measurements were which had the salt and buffer compositions of buffer A. Mixtures of made in linear glycerol gradients with a density range of 1.01-1.07 phosphatase with standards were layered on the gradients which were then centrifuged from between 14 and 16 h at 40,000 or 50,000 rpm Beckman SW56 or SW60.1 rotors at 4 "C. Gradients were tapped by upward displacement and the 20-26 fractions were analyzed by enzyme and by absorbance at 420 for hemoglobin. Con- taken into account. molecular standards of 180,000.


Purification and Characterization of a High Molecular Weight Type 1 Phosphoprotein Phosphatase from the Human Erythrocyte*
(Received for publication, July 15, 1986) Peter A. Kienerl, Dennis Carroll, Bernard J. Roth, and Edward W. Westhead From the Department of Biochemistry, University of Massachusetts, Amherst, Massachusetts 01003 The major Mn2+-activated phosphoprotein phosphatase of the human erythrocyte has been purified to homogeneity from the cell hemolysate. It is sensitive to both inhibitors 1 and 2 of rabbit skeletal muscle, preferentially dephosphorylates the @ subunit of the phosphorylase kinase, and dephosphorylates a broad range of substrates including phosphorylase a, p-nitrophenyl phosphate, phosphocasein, the regulatory subunit of cyclic AMP-dependent protein kinase, and both spectrin (K,,, = 10 PM) and pyruvate kinase ( K , = 18 MM) purified from the human erythrocyte. The purified enzyme is stimulated by Mn2+ and to a lesser extent by higher concentrations of M$+.
The purification procedure was selected to avoid any change in molecular weight, hence subunit composition, between the crude and purified enzyme. Maintenance of the original structure is demonstrated by nondenaturing gel electrophoresis and gel filtration chromatography. Gel filtration of the purified holoenzyme shows 5 single active component with a Stokes radius of 58 A at a molecular weight position of 180,000. Sedimentation velocity in a glycerol gradient gives a value of 6.1 for s20,w. Together these data indicate a molecular weight of about 135,000. Two bands of equal intensity appear on sodium dodecyl sulfate-gel electrophoresis at molecular weights of 61,700 and 36,300, suggesting a subunit composition of two 36,000 and one 62,000 subunits.
The 36-kDa catalytic subunit can be isolated by freezing and thawing the holoenzyme or by hydrophobic chromatography of the holoenzyme. The catalytic subunit shows unchanged substrate and inhibitor specificity but altered metal ion activation.
It has been shown that several cytosolic proteins in the human erythrocyte can be phosphorylated either in vitro or in the intact cell (1, 2) and we have proposed that this may be a mechanism for regulating pyruvate kinase activity in vivo (2). There are also many reports of phosphorylation of the membrane proteins of the cell but the function of these reactions is still not clear (3). If phosphorylation is a mechanism for the regulation of either the function or metabolism of the erythrocyte, there must be a mechanism for reversing the phosphorylation of these proteins.
Preliminary reports have described cytosolic phosphopro-* This work was supported by National Science Foundation Grant DMB 8309306 and by United States Public Health Service Grant HL36704. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ Current address: Pharmaceutical Research and Development, Immunology Dept., Bristol Myers Co., Wallingford,CT 06492. tein phosphatase activity capable of dephosphorylating spectrin (4,5) as well as pyruvate kinase (6). A membrane-bound phosphoprotein phosphatase, capable of dephosphorylating membrane-associated proteins has been described by Fischer and co-workers (7). The relationship among these activities is not known. Usui and co-workers (8) have found several forms of phosphoprotein phosphatase in human erythrocyte cytosol. The phosphatases differ in their activity toward various protein substrates and in their molecular weight. Usui and co-workers have purified 3000-fold, to apparent homogeneity, a 104 kDa phosphatase from the erythrocyte. That enzyme differs in every measured characteristic from the enzyme to be described here (see "Discussion").
Multiple forms of phosphatase activity have been purified from other tissues but it is not clear whether all the forms exist in the tissues or result from breakdown during the purification procedure. In many cases an approximately 30,000-Da subunit can be generated from a larger species by treating the enzyme with ethanol, acetone, or trypsin, or by freezing (9). This appears to be a catalytic subunit common to many of the phosphoprotein phosphatases (9). In the red blood cell, 3 of the 4 reported phosphatases can be converted to an active subunit of 35,000 Da. However, it remains unclear what the relationship is between this catalytic subunit and the noncatalytic peptides associated with the more complex forms of the enzyme. Several reports have suggested that the noncatalytic peptides of phosphoprotein phosphatases function as inhibitors of the enzyme (10, 11) and that appears well established in at least one case (12). However, the extent of purification has made it uncertain as to whether the inhibitors were truly subunits of the phosphatase or co-purifying proteins.
In this paper we describe the purification and properties of the major Mn2+-activated phosphoprotein phosphatase in the human erythrocyte which is also the highest molecular weight phosphatase in that cell. The purification procedures reported here were developed to maintain the molecular weight form of the enzyme found in the initial hemolysate as a prerequisite for more detailed studies on the subunit structure and functions.

MATERIALS AND METHODS
Ultrogel AcA 34 and Ampholine PAG plates were obtained from LKB, Sephadex G-200, DEAE A-50, and Sephacryls were obtained from Pharmacia P-L Biochemicals, and CM Bio-Gel and Chelex 100 were obtained from Bio-Rad. All other materials were obtained from sources described previously (6).
Spectrin was prepared essentially as described by Gratzer (14) and stored at 4 "C in 0.3 mM phosphate, 0.1 mM EDTA, 0.1 mM EGTA, 10 p~ sodium azide, pH 8.0. Prior to phosphorylation the solution was dialyzed against a buffer containing 20 mM Tris, 50 mM KC1, 1 mM MgC12, pH 7.4. SDS-gel electrophoresis was routinely used to ensure that there was no breakdown of the two spectrin subunits into smaller fragments.
Pyruvate kinase (5-10 mg/ml), spectrin (5-10 mg/ml), and casein (about 10 mg/ml) were phosphorylated by incubating the proteins with 250-500 pCi of [y3'P]ATP (25-50 p~) , 600 pM units of protein kinase, and 50 p~ CAMP in buffer A for 3-5 h at room temperature. Excess label was removed by precipitating casein with 10% trichloroacetic acid and pyruvate kinase and spectrin with 50% ammonium sulfate; the proteins were redissolved and dialyzed against buffer B until no radioactivity could be detected in the dialysate. Casein and pyruvate kinase were stored frozen, spectrin was stored at 4 ' C in buffer containing 10 p~ sodium azide.
Phosphorylase kinase (15), inhibitor 1 (16), and inhibitor 2 (17) were purified from rabbit skeletal muscle. An initial sample of inhibitor 2 was kindly donated by Dr. David Brautigan (Brown University). Phosphorylase b was purchased from Sigma, 32P-labeled phosphorylase a was prepared from phosphorylase busing Sigma grade phosphorylase kinase and [32P]ATP (18,19). Protein phosphatase 1 was purified from rabbit skeletal muscle (20). 3ZP-Labeled phosphorylase kinase was prepared using cyclic AMP-dependent protein kinase purchased from Sigma and [y-32P]ATP (15,19). The product contained approximately equal 3zP label in the CY and @ subunits as determined by densiometric scans of SDS-gel autoradiographs.
Phosphatase Assays-p-Nitrophenylphosphatase activity was routinely measured in buffer C by following the increase in absorbance at 400 nm at 30 "C. The substrate concentration was 10 mM; any variations from these conditions are detailed in the text. One unit of activity is defined as the enzyme which will hydrolyze 1 nmol of pnitrophenyl phosphate/min.
Unless otherwise stated, all phosphoprotein phosphatase activity was measured by following release of [32P]phosphate from the proteins. The labeled proteins (50,000-200,000 cpm, about 0.2 mg of protein in 0.5 ml) were incubated with the phosphatase in buffer C (except where stated) at 30 "C and samples withdrawn and pipetted into 200 pl of 20 mg/ml bovine serum albumin; this mixture was then precipitated by the addition of 750 pl of 10% trichloroacetic acid; 750 pl of the supernatant was then added to 5 ml of ScintiVerse and counted in a Beckman LSlOO scintillation counter.
Protein concentration was determined by Coomassie Blue dye binding (21). SDS-gel electrophoresis was done according to the method of Laemmli (22). Discontinuous acrylamide gels (5-15% gradient) were run and calibrated with standards: phosphorylase, bovine serum albumin, ovalbumin, carbonic anhydrase, trypsin inhibitor, and lysozyme. Gel isoelectric focusing was done with prepoured Ampholine PAG plates on an LKl3 Multiphor. Nondenaturing polyacrylamide gel electrophoresis was carried out essentially as described by Shuster (23); 5% stacking gels, pH 6.8, and 6% gels, pH 8.8, were used with a running buffer of pH 8.3. Apparent molecular weights (and Stokes radii) were determined by column gel filtration (1.7 X 100 cm) using Ultrogel AcA 34 or Sephacryl S200 or by thin layer gel chromatography (Pharmacia P-L Biochemicals) using a Sephadex G-200 superfine bed 1-mm thick. Ferritin, pyruvate kinase, y-globulin, hemoglobin, and carbonic anhydrase were used as standard molecular weight and Stokes radius markers.
Rabbit Muscle Inhibitor Assay-Activity of holoenzyme and catalytic subunit was assayed in the presence or absence of rabbit muscle inhibitor 1 or 2 in buffer C. Assays were against either [32P]phosphorylase a or [32P]phosphocasein (100,000 cpm/l0 pl); incubations were at 30 "C for 30 min. Phosphatase activity was measured by following the increase in acid soluble radioactivity.
Phosphorylase Kinase: a and / 3 Specificity-Specificity of the holoenzyme and the 36,300-Da catalytic subunit for the a and @ subunits of [3ZP]phosphorylase kinase was determined by densitometric scans of SDS gel autoradiographs. The reaction mixture (200 pl) included [32P]phosphorylase kinase, 150 pl of buffer C, and either the holoenzyme or the catalytic subunit. Incubations were at 30 "C for 30 min and stopped by acid precipitation as previously described.
Preparation of Cytosol-For the survey of the phosphatase population in erythrocyte cytosol, 200 ml of blood was drawn from human male adults into a Na+:heparin anticoagulant. All subsequent steps were carried out at 4 'C in the presence of 100 qM phenylmethylsulfonyl fluoride and 14 mM mercaptoethanol. The whole blood was centrifuged at 2,000 X g for 10 min at 4 "C and washed 3 X with isotonic NaCl at pH 7.0. The plasma and buffy layer were carefully removed by aspiration. The cells were then lysed in about 4 times their packed volume of 10 mM Tris buffer at pH 7.0. The membranes were immediately sedimented for 45 min at 24,000 X g, resuspended in the same volume of fresh lysis buffer, and resedimented.
For purification and characterization of the major Mn-activated protein phosphatase, cytosol was prepared as above except that the starting material was 1500 ml of blood obtained from a local hospital. During the early stages of development of the purification procedure it was found that enzymatic activity was sensitive to both oxidation and proteolysis, especially in the absence of Mn2+, so dithioerythritol or mercaptoethanol was added (to about 10 mM) at all stages of the purification and 100 p~ phenylmethylsulfonyl fluoride was added to the hemolysate. Initial attempts at purification in the presence of M%+ or without metal, resulted in lower yields of enzymatic activity: Mn2+ protected the enzyme and caused no change in its molecular weight properties so MnCll was included in all except the lysis buffers. The purified enzyme in solution tends to spontaneously degrade to the 36.3-kDa active form unless stabilized with glycerol (this will be discussed later), so purified samples of enzyme contained 10% glycerol.
CM-Cellulose and Ultragel Chromatography-This procedure was used specifically in the gel filtration survey of the phosphatase activities in the cytosol. The isolated cytosol was adjusted to pH 5.5 and its conductivity lowered to less than 1 mmho via dialysis against a 10-fold excess solution of 20 mM Tris succinate, pH 5.5, containing either 1 mM Mn2+ or 5.7 mM M$+, and passed through a preequilibrated column of carboxymethyl cellulose (CM-cellulose) to remove the hemoglobin. The eluant was concentrated to 2 ml on an Amicon ultrafiltration cell equipped with a YM-10 filter membrane and dialzyed against buffer A or buffer C in preparation for gel filtration chromatography. Details of the gel filtration are given in the legend to Fig. 1.
Ammonium Sulfate Fractionation-The lysate was brought to 50% saturation of ammonium sulfate by the addition of the solid salt; the pH was maintained at 7.5. After 60 min at 4 "C the suspension was centrifuged for 30 min at 24,000 X g and the pellet redissolved in 200 ml of buffer C. This solution was made 20% saturated with ammonium sulfate, adjusted by conductivity measurement, left for 1 h at 4 "C and then centrifuged as before. The supernatant was removed and made 45% saturated by the addition of solid ammonium sulfate; the pH was maintained at 7.4. The suspension was left overnight at 4 "C.
Zon-exchange Chromatography-In these stages all the buffers were 20 mM succinate containing 1 mM MnC12, adjusted to the appropriate pH by addition of solid Tris base. The 45% ammonium sulfate precipitate was dissolved in succinate buffer, pH 5.5, and dialyzed against this buffer until the conductivity of the solution was less than 1 mmho. The suspension was centrifuged for 20 min at 30,000 X g to remove the precipitated protein and the supernatant passed through a CM Bio-Gel A column (2.5 X 10 cm) equilibrated with succinate buffer, pH 5.5. The percolate from this column together with the first washing of 50 ml of succinate buffer, pH 5.5, was raised to pH 8.0 by the addition of solid Tris, and this solution passed through a DEAE-A50 column (2.5 X 10 cm), previously equilibrated with 20 mM Tris succinate, pH 8.0. The column was washed with 50 ml of Tris succinate, pH 8.0, followed by 500 ml of Tris succinate, pH 8.0, containing 0.1 M KC1. The phosphatase was eluted by applying a 500ml salt gradient, 0.1-1.0 M KCl, 3-ml fractions were collected and the pooled active fractions brought to 50% saturation of ammonium sulfate.
Nondenaturing Gel Electrophoresis-The ammonium sulfate suspension of enzyme from the DEAE pH 8 column was centrifuged and dissolved in 1 ml of nondenaturing sample buffer (50 mM Tris-C1,20 mM mercaptoethanol, 10% glycerol pH 6.8), centrifuged to remove any undissolved particles, and then loaded onto two 10 X 14-cm nondenaturing polyacrylamide gels (6%). The gels were run at a constant current of 30 mA/gel until the bromphenol blue dye reached the bottom; the gels were stained for activity against p-nitrophenyl phosphate; within 1 min a major active band was visible and this was cut out of the gel. The enzyme was electroeluted from the gel into 20 Erythrocyte Phosphoprotein Phosphatase mM Tris, 50 mM KCl, 20 mM mercaptoethanol, 5% glycerol, 250 pM MnC12, pH 7.4, at 4 "C, at 150 V for 10 h (until no activity could be detected in the compartment containing the sliced gel). The enzyme solution was made 50% saturated in ammonium sulfate by adding the solid salt.
Gel Filtration-The ammonium sulfate suspension of protein from the nondenaturing gel was centrifuged and dissolved in 0.5 ml of buffer D. This solution was applied to an Ultrogel AcA 34 column (1.5 X 90 cm) and the protein eluted with buffer D; 1.2-ml samples were collected. Fractions possessing phosphatase activity of greater than 50% of the peak activity were pooled and made 50% saturated in ammonium sulfate.
Freezing and Thawing of t k Phosphatase-The ammonium sulfate suspension of the phosphatase from the gel filtration column was centrifuged and the pellet redissolved at a concentration of about 1 mg/ml protein in buffer B containing 15 mM mercaptoethanol. The solution was frozen on dry ice for 5 min and the sample, still frozen, centrifuged at 12,000 X g for 1 min at room temperature (Eppendorf Microfuge); during this time the samples melted. The supernatant was removed and assayed for activity and protein concentration and the freeze-thaw cycle repeated.
Octyl-Sephcrrose Chrumutography-The 50% ammonium sulfate suspension of the phosphoprotein phosphatase from the gel filtration column was diluted with buffer C to a find concentration of 1 M ammonium sulfate. The solution was passed through an octyl-Sepharose column (1 X 12 cm) previously equilibrated with buffer C containing 1 M ammonium sulfate, pH 7.4, at 4 "C. The column was eluted with an ammonium sulfate gradient (2 X 200 ml; 1-0 M) in 5 mM Hepes, 1 mM Mn2+, pH 7.4; 3-ml fractions were collected and assayed for phosphatase activity.
Sedimentation Velocity Measurements-These measurements were which had the salt and buffer compositions of buffer A. Mixtures of made in linear glycerol gradients with a density range of 1.01-1.07 phosphatase with standards were layered on the gradients which were then centrifuged from between 14 and 16 h at 40,000 or 50,000 rpm in Beckman SW56 or SW60.1 rotors at 4 "C. Gradients were tapped by upward displacement and the 20-26 fractions were analyzed by enzyme activity and by absorbance at 420 nm for hemoglobin. Considerations cited by Martin and Ames (24) were taken into account. As internal standards we used hemoglobin, lactic dehydrogenase, yeast alcohol dehydrogenase, and catalase with s~,,,~ values of 4.3,7.3, 10.6, and 11.3, respectively. Three different preparations of phosphatase were examined and on the last occasion an aliquot of the same phosphatase was simultaneously chromatographed on an Ultrogel AcA 34 column to establish that its gel permeation behavior was exactly as expected from previous measurements which had given a molecular weight relative to standards of 180,000.

Metal Actiuation Profile of Cytosolic Phosphatase-A survey
of the different populations of phosphatase found in the human red blood cell cytosol is shown in Fig. 1; the forms can be distinguished by apparent molecular weight and differences in metal activation. The broken curue represents the activation of the different molecular weight forms by Mn2+ and the solid curue represents Mg2+-stimulated activity. The major Mn2+-activated peak is found at an equivalent sphere position of 180,000 Da and constitutes 60% of all phosphatase activity: in the presence of Mg2+ this enzyme constitutes only 20% of the total activity. Overall Mn2+ activation is seen to be more than 30-fold greater than M$+ activation, The phosphatase we discuss in this paper is the form labeled 180,000 Da in Fig.  L2 Characterization of the Purified Holoenzyme and Its Catalytic Subunit-The purpose of the adopted procedures was to purify the major phosphoprotein phosphatase activity without changing the molecular weight of the enzyme. For convenience, enzymatic activity was routinely followed by assaying Mn2+-dependent hydrolysis of p-nitrophenyl phosphate but Although the data in this paper demonstrate that the true molecular weight is near 135,000 we have retained the gel permeation molecular weight in the figure for comparison with the elution data of others, especially those of Usui et al. (8). at all of the states of purification Mn2+-dependent phosphospectrin, phosphocasein, and phosphopyruvate kinase phosphatase activities were coincident with Mn2+-dependent pnitrophenylphosphatase activity. Activity against p-nitrophenyl phosphate and phosphospectrin was proportional to the amount of enzyme added over a 10-fold range of concentration, and the ratio of p-nitrophenylphosphatase activity to spectrin phosphatase activity remained constant. Activity against casein and phosphorylase was not so simple; at higher concentrations of enzyme, enzymatic activity was not directly proportional to enzyme concentration but showed a relative increase in activity as the protein was diluted.
The overall purification is given in Table I. A purification of about 300,000-fold with a yield of up to 53% can be achieved routinely. Often during the preparation procedure there was a significant increase in overall phosphatase activity compared with that seen in the 20-45% ammonium sulfate fraction. This did not consistently occur at a particular point of the purification procedure and there was no apparent change in the subunit structure of the enzyme. Molecular Weight-Gel filtration of the 25-45% ammonium sulfate fraction of the hemolysate and the purified enzyme showed one major peak corresponding to a relative molecular weight of 180,000 based on simple comparison to standard proteins (Fig. 2, A and B). Sedimentation rates in a glycerol gradient, however, indicated a much lower molecular weight. Four samples of three preparations of the phosphatase showed that the phosphatase sedimented behind lactic dehydrogenase (Mr = 140,000). Based on interpolation on a curve of sedimentation distance versus sz0,,,, for the standard proteins, the s20,w for the phosphatase is 6.1 f 0.2. If we calculate from  the gel filtration elution volume compared with the standard proteins and assume a v of 0.725 we get a molecular weight estimate of 135,000, substantially lower than the 180,000 based on gel filtration alone. The elution of the phosphatase at the position of a higher molecular weight equivalent sphere suggests appreciable asymmetry in the holoenzyme. The Stokes radius of the phosphatase holoenzyme calculated from the gel filtration data is 58 A.
If the holoenzyme was treated with acetone or ethanol, a procedure frequently used to generate active subunits of protein phosphatases, much activity was lost. This was true both after the ammonium sulfate fraction and after the pH 8 DEAE column (Fig. 3). After ethanol treatment of the ammonium sulfate fraction, 20% of the original activity remained; this had a molecular weight of 36,300 by SDS-gel electrophoresis. Freezing and thawing of the purified holoenzyme in the presence of mercaptoethanol produced several different molecular weight forms of the enzyme as detected by gel filtration (Fig.  2C). All show Mn2'-simulated activity against both phosphocasein andp-nitrophenyl phosphate. After three or four cycles of freezing and thawing, the protein concentration of the sample dropped due to precipitation. Gel filtration of the supernatant still showed a major peak of activity at an equivalent sphere molecular weight of about 180,000 but a minor band of lower molecular weight was also visible (Fig. 2C). If these freeze-thaw cycles were continued more protein was removed; gel filtration of the enzyme after these freezing and thawing cycles showed that nearly all the activity eluted in a peak corresponding to an apparent molecular weight of about 36,000 (Fig. 20). The transformation to the lower molecular weight form was greatly inhibited by the presence of 10% glycerol. The effect of freezing and thawing on the enzyme suggested that hydrophobic interactions may be involved in the holoenzyme organization so the enzyme was chromatographed on octyl-Sepharose. Gel filtration of the enzyme eluted from the octyl-Sepharose column showed that the majority of the enzyme was converted to the 36.3 kDa form. This enzyme is identical (by gel filtration) to the lower molecular weight form of the frozen and thawed enzyme, which suggests that they are the same small subunit of the enzyme.
After several weeks, the pure holoenzyme, kept as an ammonium sulfate suspension in 10% glycerol at 4 "C, degrades into lower molecular weight forms. This may be due to pro-Erythrocyte Phosphoprotein Phosphatase teolysis from trace contaminants or to slow spontaneous alteration of the subunit structure. In the absence of glycerol, the enzyme degrades to the 36.3-kDa form over 24-36 h at 4 "C.
Gel Electrophoresis-From the densitometric scans of purified holoenzyme run on nondenaturing gel and stained with Coomassie Blue, it is possible to estimate that the active phosphatase band represents at least 90% of the protein from the Ultrogel column (Fig. 4A).
If the phosphatase preparation is either frozen and thawed or chromatographed on an octyl-Sepharose column, the active band of R F value 0.49 disappears and is replaced by one major active band of higher mobility, R F 0.68 (Fig. 4B).
Denaturing polyacrylamide electrophoresis of the purified enzyme shows two bands of equal intensity with molecular weights corresponding to 61,700 and 36,300 (Fig. 5A). If the two subunits stain equally with Coomassie Blue, then the MIGRATION -   C, casein (prepared as described in the text). A, with no phosphatase added; 0, with phosphatase + 5 mM EDTA; 0, with phosphatase + 5.7 mM MgC12; 0, with phosphatase + 1 mM MnC12. Assays were performed as described under "Materials and Methods"; about 5 units of phosphatase was Chelex treated and then incubated in the appropriate buffer for 5 min prior to assay. holoenzyme structure might be two subunits of 36,300 and one of 61,700, for a molecular weight of 134,000, in agreement with calculations from sedimentation velocity and gel filtration chromatography. When the purified enzyme is frozen and thawed five times, then centrifuged to remove the insoluble material prior to SDS electrophoresis, only the 36-kDa form of the enzyme is observed (Fig. 5B). The dissociated 62-kDa noncatalytic subunit precipitates out of solution under these conditions. Metal Dependence of Holoenzyme Activity-Enzymatic activity of the holoenzyme againstp-nitrophenyl phosphate and the phosphorylated proteins is very dependent on the presence of divalent metal ions. In the absence of metal ions or in the presence of 5 mM EDTA there is a low but significant level of activity (Fig. 6). Activity is increased in the presence of M e and even more in the presence of Mn2+. With pyruvate kinase and casein as substrates, Mn2+ is a far better activator than M e . p-Nitrophenylphosphatase activity shows simple saturation kinetics with increasing concentrations of Mn2+; a K, for Mn2+ of 80 PM was obtained (Fig. 7A). With phosphorylated casein as substrate the saturation curve does not show simple Michelis-Menten kinetics (Fig. 7B), from the curve a After treatment of the holoenzyme with 1 mM Mn2+ and subsequent removal of metal with EDTA, the kinetic properties of the enzyme, against all the substrates, are the same as those of enzyme purified without the addition of Mn2+. The effect of Mn2+ was found to be fully reversible. This is in contrast to the behavior reported on phosphatases from the rabbit muscle (13,27). Although addition of Mn2+ to the lysate of freshly drawn blood does not cause a significant stimulation of activity, upon treatment of the lysate with Dowex 2, the phosphatase does become characteristically activated by Mn2+. Presumably Mn2+ activation prior to Dowex 2 treatment of the lysate is obscured by the high concentration of chelating ions found in the red blood cell cytosol. In the absence of Mn2+ the enzymatic activity is very sensitive to oxidized glutathione and air and is also labile to trypsin; in the presence of MnZ+ the enzyme is much more stable (data not shown).
The stimulation by M%+ does not approach saturation and it is not possible t o obtain a value for K,. Compared with Mn2+, the affinity of the enzyme for M%+ is much lower. Cobalt stimulates phosphatase activity to about the same extent as Mn2+, but this is only detectable after removal of protective sulfhydryl reagents. Calcium (0.5-3.5 mM) or cal-cium and calmodulin (150 pg/ml) show no activation: millimolar concentrations of zinc are inhibitory, but micromolar concentrations show neither activation nor inhibition. Lack of inhibition by micromolar Zn2+ as well as sensitivity to fluoride ( Table 11) is characteristic of phosphoseryl phosphatases in contrast to phosphotyrosyl phosphatase (28).
Metal Dependence of the Catalytic Subunit-As in the case of holoenzyme, the 36,300-Da catalytic subunit shows activity in the absence of added metal ion and in the presence of 5 mM EDTA, but is markedly stimulated by divalent cations (Fig. 8). The catalytic subunit differs from the holoenzyme in several respects. Upon conversion of the enzyme to the 36,300-Da subunit, by freezing and thawing, there is an increase in

TABLE I1
Inhibition of erythrocyte phosphoprotein phosphatase Assays were carried out as described under "Materials and Methods,'' in buffer C; 6 units of phosphatase (holoenzyme) were used in each assay. The pH of the inhibitors was adjusted to 7.4 prior to assay. The hydrolysis is expressed as percent of the hydrolysis of the uninhibited substrate in 10 min. total Mn2+-stimulated activity but there is no marked change in the for Mn2+. In this respect the larger subunit would appear inhibitory. The activity of the catalytic subunit is also more effectively stimulated by MgZ+ than is the activity of the holoenzyme; activity approaches that achieved upon Mn2+ stimulation (Fig. 8). This change may arise both from a change in the affinity of M e for the enzyme and an increase in maximum velocity upon removal of the 61,700 noncatalytic subunit. Under normal assay conditions, in the presence of either 5.7 mM MgC1, or 1 mM MnCl,, there is an &fold increase in Mg+-stimulated activity and a 2-3-fold increase in Mn2+-stimulated activity when the holoenzyme is converted to the catalytic subunit.
Substrate Profile-The high molecular weight phosphatase shows activity against a broad range of substrates. Those hydrolyzed include: p-nitrophenyl phosphate, phosphospectrin, phosphorylase, phosphorylase kinase (@ subunit), the regulatory subunit of CAMP-dependent protein kinase, phosphocasein, and phosphorylated pyruvate kinase.
The enzyme shows very little activity towards simple phosphate esters other than p-nitrophenyl phosphate. Eleven low molecular weight phosphate esters were tested as substrates at 30 "C in buffer C at initial substrate concentrations of 1 mM. The phosphate release after 2 h incubation with 100 units of purified phosphatase was measured and compared to a control without phosphatase. The most active substrates were threonine phosphate and phosphoenolpyruvate; 13% of their phosphate was released. That is 0.3% of the rate of hydrolysis of p-nitrophenyl phosphate. Fructose 1,g-bisphosphate and serine phosphate were hydrolyzed at about half that rate. Phosphoethanolamine, phosphoglycolate, and glucose 6-phosphate were hydrolyzed at about 0.08% of the rate of p-nitrophenyl phosphate and 2,3-bisphosphoglycerate, 2phosphoglycerate, ATP, and ADP were all hydrolyzed at half that rate or less. Because the degree of hydrolysis is low these rates should approximate initial rates of hydrolysis.
The substrate saturation curve for activity against phosphospectrin enzyme appears to follow simple Michelis-Menten kinetics. However, it was not possible to obtain phosphorylated spectrin in sufficient concentration to observe full saturation of the enzyme in this assay system. From data obtained using up to 11 p~ protein-bound phosphate we estimated values of 10 p~ (protein bound Pi) for K,,,.
Phosphorylation of spectrin yielded labeled protein containing 1-2 mol of radioactive phosphate/mole of spectrin @ subunit compared with the reported maximum of 4 phosphates/@ subunit. The less than maximal incorporation of 32P may be due to the fact that the spectrin used was not dephosphorylated prior to phosphorylation. SDS-gel electrophoresis showed that at least 80% of the radioactive label could be attributed to spectrin labeled in the @ subunit. If we assume that the phosphatase does not distinguish among the phosphate groups on the p subunit and assuming 4 mol of Pi/mol of @ subunit, then the K,,, based on spectrin @ subunits would be of the order of 6 pM.
In this laboratory pyruvate kinase has been phosphorylated to the level of 2-3 mol of Pi/mol of pyruvate kinase tetramer, with a reported maximum of 4 mol of PJmol of enzyme (6). Dephosphorylation of pyruvate kinase shows simple saturation kinetics, similar to those found for spectrin dephosphorylation. Based on a maximum value of 1 mol of Pi/mol of pyruvate kinase subunit, a K,,, of 18 IM (protein-bound phosphate) or 24 p~ (subunit pyruvate kinase), is obtained.
Dephosphorylation of casein by the holoenzyme also shows apparent Michaelis-Menten kinetics with an apparent K,,, of 63 p~ casein. However, casein prepared as described by Rei-mann and co-workers (13) was found to still contain at least 95% of the original phosphate groups leaving a maximum of 5% of the sites available for phosphorylation with radioactive label. Thus dephosphorylation of the labeled phosphate groups may not be representative of overall dephosphorylation. Assuming that all phosphate groups of casein are hydrolyzed at similar rates, the maximum velocity for casein dephosphorylation showed a specific activity between 0.4 and 1.7 pmol/min/mg of phosphatase. Subject to the uncertainties described above, we can summarize the maximum velocities of the 3 protein substrates in buffer C at 30 "C. Spectrin, pyruvate kinase, and casein show maximum velocities within a factor of two of each other at about 1.0-2.0 pmol/min/mg of phosphatase. This is 2-4% of the rate obtained with p-nitrophenyl phosphate which is 44 pmol/min/mg enzyme or 44 X lo3 units/mg of phosphatase.
The low molecular weight phosphate esters are hydrolyzed at one-tenth the rate of the protein substances at similar concentrations.
The catalytic subunit shows the same substrate specificity as the holoenzyme.
Relative Specificity for Phosphorylase Kinase a and p Subunits-Assays were carried out as described under "Materials and Methods." Densitometric scans of the autoradiogram of phosphorylase kinase showed there to be an approximate 1:1 labeling of the a and subunits (Fig. 9A). After 30 min the fraction incubated in the presence of the holoenzyme showed a 10% loss in label in the a subunit and a 79% loss in the @ subunit (Fig. 9B).
The catalytic subunit showed similar specificity. Under the same conditions it caused a 6% loss of label in the (Y subunit and a 72% loss in the @ subunit (Fig. 9C).
Inhibition of the Phosphatase-Since the enzyme is markedly sensitive to the presence of Mn2+, any chelating agent is likely to have a pronounced effect on the enzymatic activity. EDTA, ADP, and ATP, all quite strongly inhibit activity against phosphocasein (Table 11). In the presence of higher concentrations of Mn2+, inhibition can still be observed but to lesser extents; ADP and ATP show the most inhibition. It was not possible to measure inhibition by 2,3-bisphosphoglycerate in the presence of higher Mn2+ due to precipitation of the metal. Dephosphorylation of both pyruvate kinase and spectrin is markedly inhibited by ADP:Mn2+ and ATP:MnZ+; again KF inhibited but less strongly (Table 11). The catalytic subunit showed an identical inhibition profile.
Effect of Rabbit Muscle Inhibitors 1 and 2 on Phosphatase Activity-Sensitivity to rabbit muscle inhibitors 1 and 2 is a distinguishing characteristic of type 1 phosphoprotein phosphatases. When measured with phosphorylase a as a substrate, both the holoenzyme and the catalytic subunit were inhibited 40-50% by the same concentrations of inhibitors 1 or 2 that inhibited rabbit muscle protein phosphatase 50% under equivalent circumstances. Similar degrees of inhibition were found with phosphocasein as substrate.
pH Profile and Isoelectric Point-The pH dependence of activity against phosphocasein and p-nitrophenyl phosphate shows rather a broad profile with an optimum around pH 7.4.
Isoelectric focusing of the purified enzyme proved to be quite difficult because often activity was lost during the focusing, or the protein precipitated out in the gel. However, preliminary data from wide pH range prepoured gels, pH 3.5-9.5, indicated that the enzyme has a PI of about 5.0. This is in agreement with the characteristics of the enzyme shown during ion-exchange chromatography.

DISCUSSION
Although 30,000-35,000 catalytic subunits of phosphoprotein phosphatases have been purified from a number of tissues, purification of the higher molecular weight forms found in crude extracts has proved difficult. Ingebritsen and Cohen (26) have proposed a useful classification of protein phosphatases based on substrate specificity, metal activation, and sensitivity to protein inhibitors 1 and 2. Very recently high molecular weight forms of three type 2 phosphatases have been purified (22,30,31). These enzymes act preferentially on the a! subunit of phosphorylase kinase and are not inhibited by protein inhibitors 1 and 2 (32). The enzyme we have purified is, by these same criteria, clearly a type 1 phosphatase. Purification of a type 1 phosphatase from liver glycogen particles has been reported by Stralfors et al. (33). That enzyme appears to be a dimer of 1 catalytic subunit (37,000 Da) and one other subunit (103,000 Da) which causes binding of the enzyme to glycogen particles. The noncatalytic subunit is thus much larger than the one we find but it may have a similar function. In the enzyme we have isolated, the 62-kDa subunit has an inhibitory effect when activities are measured with Mg2+ as activator (presumably the physiological condition). It also appears to mediate binding of the enzyme to the erythrocyte membrane. We have shown that the enzyme binds reversibly to the membrane and that the membrane-bound form is inactive (34,35). Inside-out red cell membrane vesicles will selectively remove only the holoenzyme form of the enzyme from crude preparations (37) but experiments on binding of purified holoenzyme and catalytic subunit to red cell membranes have given equivocal results (25,38).
Gruppuso et al. (36) have shown that in muscle extracts there is a 60-kDa protein that cross-reacts immunologically with inhibitor 2 and that in a crude fraction of phosphatase, tryptic digestion leads to phosphatase activation in step with destruction of the 60-kDa antigen. Samples of our holoenzyme have been examined by Western blotting in Dr. Brautigan's laboratory but no evidence of cross-reaction with inhibitor 2 antibody was found.
A high molecular weight protein phosphatase purified from human erythrocytes by Usui et al. (8) is particularly interesting for its relationship with the enzyme we have purified. The 104,000-Da enzyme purified by Usui et al. is the major M eactivated enzyme of the red cell and is composed of one 32,000-Da catalytic subunit and one 69,000-Da subunit. It was not tested for activity toward phosphorylase kinase a and @ subunits but was judged to be a type 2 enzyme because it was not inhibited by rabbit muscle inhibitor 2. That enzyme was also completely inhibited by Mn2+ concentration that fully activates the enzyme we purified. The enzyme that we purified is the enzyme designated phosphatase I in the survey by Usui et al.; the one they purified was their phosphatase IV. Using spectrin phosphate activity as the point of comparison of the two papers, we find that the papers are in agreement on the specific activity of the homogenate and on the relative amounts of the enzyme forms.
Two points need mention: Usui et al. (8) reported the same Stokes radius for the crude "phosphatase I" that we report for the purified enzyme, but they found an s20,w for that enzyme of 7.4 while we find 6.1 & 0.2. They therefore calculate a molecular weight of 180,000 for the crude enzyme in agreement with the gel permeation estimate. Our sedimentation results show a much more asymmetric molecule with a molecular weight of about 135,000. A second point of apparent disagreement is that we find that our enzyme is sensitive to inhibitors 1 and 2 of rabbit muscle, using inhibitors prepared in this laboratory and also a sample of inhibitor 2 generously provided by Dr. David Brautigan. Rabbit muscle protein phosphatase was used as a positive control and to assess the activity of the inhibitor preparations. Usui et al. reported that none of the erythrocyte phosphatases were sensitive to inhibitor 2 but did not report a positive control. The possibility exists that the purified enzyme which we studied shows properties different from the enzyme in the crude homogenate, but since substrate specificity, gel permeation behavior, and metal ion specificity do not change during purification that possibility is not high.
The size of the catalytic subunit and the noncatalytic subunits in the enzyme we have purified are very similar to those of the type 2 enzymes isolated from rabbit skeletal muscle (29), rabbit heart (30), and turkey gizzard (31). In none of these cases has the function of the large subunit been understood and despite the type 1-type 2 differences, it may be that these enzymes are much closer in structure and function than is apparent now.
The meaning of Mn2+ activation is also a problem for future work. Our current expectation is that Mn2+ elicits an activity which is cryptic under physiological conditions and that a physiological activator is to be discovered.