The protein phosphatases of Drosophila melanogaster and their inhibitors

Transcription

1 Eur. J. Biochem. 164,31-38 (1987) 0 FEBS 1987 The protein phosphatases of Drosophila melanogaster and their inhibitors Sara ORGAD Yadin DUDAI and Philip COHEN Department of Neurobiology, The Weizmann Institute of Science, Rehovot Department of Biochemistry, University of Dundee (Received October 22/December 1, 1986) - EJB Protein phosphatases-1, 2A and 2B have been identified in membrane and soluble fractions of Drosophila melanogaster heads. Similarities between Drosophila and mammalian protein phosphatase-1 included specificity for the fi subunit of phosphorylase kinase, sensitivity to inhibitor-1 and inkbitor-2, inhibition by protamine, retention by heparin-sepharose and selective interaction with membranes. In addition, an inactive form of protein phosphatase-1, termed protein phosphatase-l,, was detected in the soluble fraction that could be activated by preincubation with MgATP and mammalian glycogen synthase kinase-3. Inhibitor-2 partially purified from Drosophila had an identical molecular mass to its mammalian counterpart, and recombined with mammalian protein phosphatase-1 to form a hybrid protein phosphatase-1,. Similarities between Drosophila and mammalian protein phosphatase-2a included preferential dephosphorylation of the a subunit of phosphorylase kinase, insensitivity to inhibitors-1 and -2, activation by protamine, exclusion from heparin-sepharose and apparent molecular mass. A Ca2 +-dependent calmodulin-stimulated protein phosphatase (protein phosphatase-2b) that was inhibited by trifluoperazine was identified in the soluble fraction. The remarkable similarities between Drosophila protein phosphatases and their mammalian counterparts are indicative of strict phylogenetic conservation and demonstrate that the procedures used to classify mammalian protein phosphatases have a wider application. Characterisation of the Drosophila phosphatases will facilitate genetic analysis of dephosphorylation systems and their possible roles in neuronal and behavioural plasticity in Drosophila. The control of cellular activity by protein phosphorylation and dephosphorylation is a ubiquitous regulatory mechanism (reviewed in [l - 41) that involves two classes of enzyme termed protein kinases and protein phosphatases. The molecular structures and functions of these enzymes have been studied extensively, but information on their genetic control and specific roles in development and behaviour is still very limited. Application of genetics and molecular biology to the study of protein kinases and phosphatases is therefore highly desirable and for this purpose, the fruit fly, Drosophila melanogaster, offers great advantages. Some information on phosphorylation processes in Drosophila have recently become available [5-81. In addition, mutations already exist which affect enzymes that regulate protein phosphorylation and exert intriguing effects on development and behaviour, e. g. learning and memory (reviewed in [9, 101. However, to facilitate characterisation of existing mutants and the systematic search for new ones, it is essential to identify the protein kinases, protein phosphatases and their substrates in Drosophila using biochemical tech- Correspondence to P. Cohen, Department of Biochemistry, Medical Sciences Institute, University of Dundee, Dundee, Scotland DDI 4HN Abbreviations. SDS, sodium dodecyl sulphate; PhMeS02F, phenylmethylsulphonyl fluoride. Enzymes. Glycogen phosphorylase (EC l); phosphorylase kinase (EC ); glycogen synthase kinase-3 and CAMP-dependent protein kinase (EC ); protein phosphatase (EC ). niques. The present work contributes to this goal by describing the properties of protein phosphatases in Drosophila. There is increasing evidence that relatively few protein phosphatases are involved in many aspects of cellular regulation in mammalian tissues (reviewed in [l, 21). They have been classified into two types, depending on whether they dephosphorylate the fi subunit of phosphorylase kinase and are sensitive to inhibition by two thermostable proteins inhibitor-1 and inhibitor-2 (type l), or whether they dephosphorylate the M subunit of phosphorylase kinase preferentially and are insensitive to the inhibitors (type 2). The type-2 phosphatases are further classified into sub-types on the basis of substrate specificity and dependence on divalent cations. Molecular forms of protein phosphatase-1, protein phosphatase-2a and protein phosphatase-2c account for almost all the phosphatase activity towards enzymes that regulate the major biodegradative and biosynthetic pathways in liver and muscle cells, although in most cases the contribution of protein phosphatase-2c (a Mg2 +-dependent) enzyme is minor. However, protein phosphatases-1, 2A and 2C clearly have even wider roles, since they are present at similar levels in other tissues, such as brain [ll]. By contrast, the substrate specificity of the Ca2 +-dependent calmodulin-stimulated protein phosphatase (protein phosphatase-2b) is more restricted, and this enzyme is especially prominent in skeletal muscle and in the brain (where it has been termed calcineurin) (reviewed in [l, 2, 111). In the present work, we have employed the characteristic properties of mammalian protein phosphatases to identify

2 32 protein phosphatases in soluble and membrane fractions from Drosophilu heads. The choice of heads, which contain about 30% brain tissue, was dictated by our interest in the role of protein phosphatases in neuronal function in general and neuronal plasticity in particular. *m MATERIALS AND METHODS Flies Drosophilu melunogaster of the Canton-S (C-S) strain were used. All stocks were cultured under standard conditions [12]. For all experiments day-old flies were used. Chemiculs [Y-~~PIATP was from Amersham (Bucks, UK). Benzamidine, phenylmethylsulphonyl fluoride (PhMeS02F) and protamine sulphate (salmon) were from Sigma (Poole, UK). Bovine serum albumin (fraction V) was from BDH Chemicals (Poole, UK) and blue-sepharose CL6B and heparin- Sepharose from Pharinacia GB (Hounslow, UK). Trifluoperazine was a gift from Smith, Kline and French (Welwyn Garden City, UK). All other chemicals were of analytical grade. Buffers Buffer A: 1 tnm EGTA, 0.1% (v/v) 2-mercaptoethanol, 1 mm benzamidine 0.1 mm PhMeS02F and 5% (v/v) glycerol in 20 mm Tris/HCl, ph 7.0. Buffer B: 1 mm EGTA, 0.1% (v/v) 2-mercaptoethanol, 1 mm benzamidine. 0.1 mm PhMeSOzF and 5% (v/v) glycerol in 20 mm triethanolamine chloride, ph 7.0. Protein prepurutions The following proteins were prepared from rabbit skeletal muscle, as described in the indicated references: glycogen phosphorylase [ 131, phosphorylase kinase [14], glycogen synthase kinase , the 37-kDa catalytic subunit of protein phosphatase-i [ 161. inhibitor-2 and the active phosphorylated form of inhibitor-1 [17], protein phosphatase-2b [18] the catalytic subunit of CAMP-dependent protein kinase [19], and calmodulin [20]. Preparation qf 32 P-lubelledprotein substrates 32P-labelled phosphorylase a (1 mol phosphate/97-kda subunit) was prepared by phosphorylation with phosphorylase kinase [21]; 32P-labelled phosphorylase kinase (1.4 mol phosphate/aj3$ unit, containing equal amounts of radioactivity in the r and fl subunits [22], and 32P-labelled inhibitor-1 (1 mol phosphate/l9-kda protein) were prepared by phosphorylation with CAMP-dependent protein kinase [22], all to a specific activity of 106 cpm/nmol. Preparation of'niemhrune and soluhle,fractions,from Drosophila heads Heads were separated from bodies by freezing in liquid nitrogen followed by shaking and straining through nylon sieves. Homogenisation, using 20 strokes of a glass/glass homogeniser, was performed in ice-cold 10 mm Tris/HCl, ph 7.0, containing 350 mm sucrose, 2 mm EGTA, 2 mm EDTA, I I I I I I I I I I Inhibitor-', nm inhibitor. nm Fig. 1. Effect of inhibitor proteins on phosphorylase phosphatase (PP) activity in the membrane and soluble fractions,from Drosophila. Fractions were diluted 20-fold (cytosol), 10-fold (membranes) or 5- fold (salt-extracted membranes) to 0.4 U/ml; ml aliquots were preincubated with 0.01 ml mammalian inhibitor-1 or inhibitor-2 before assay. (A) Effect of inhibitor-i on various Drosophila fractions: (0) soluble fraction; (0) membrane fraction; (A) fraction obtained by extracting the membranes with 0.5 M NaCI. (B) Effect of inhibitor-i (0) and inhibitor-2 (0), assayed in parallel, on the Drosophila membrane fraction 1 mm benzamidine, 0.1 mm PhMeS02F, and 0.1Y0 (v/v) 2- mercaptoethanol using 1 ml buffer/l6 mg wet weight of heads. The homogenates were centrifuged for 3 0 min at 1000 x g, and the supernatants collected and centrifuged (unless otherwise indicated) for 1 h at I00000 x g. The supernatants thus obtained (termed the soluble fraction) were kept on ice and the pellets resuspended in the original volume of the same buffer and recentrifuged as above. The washed pellets were resuspended in a quarter of the original volume in 50mM Tris/HCl, ph 7.0, containing 0.1 mm EGTA, 0.1% (v/v) 2- mercaptoethanol, 0.1 mm PhMeS02F, and 1 mm benzamidine, and termed the membrane fraction. Extraction of membrane proteins was performed with 0.5 M NaCl or 0.2% Triton X-100 as indicated under Results. Protein phosphatase assays Assays of protein phosphatases-1 and 2A ( ) contained 0.1 mm EGTA, 0.1% (v/v) 2-mercaptoethanol, 0.6 mg/ml bovine serum albumin, protein phosphatase, and 32P-labelled substrate in 50 mm Tris/HCl, ph 7.0. Reactions were initiated by addition of 32P-labelled substrate, carried out for min at 30"C, and terminated and analysed for released 32P as in [23]. Caffeine (5 mm) was present in all phosphorylase phosphatase assays. Substrate concentrations were 10 pm for phosphorylase a, 2.5 pm for phosphorylase kinase, and 5 pm for inhibitor-1. Release of radioactivity was restricted to < 30% to ensure linear reaction rates. Enzyme was omitted from control reactions and 32P released under these conditions was subtracted as a blank. When inhibitor-i, inhibitor-2 or protamine were present in the assays, they were preincubated with the enzyme for 10 min at 30 C before addition of the substrate. Protein phosphatase-1 was defined as the proportion of the activity that could be inhibited by inhibitor-1 or inhibitor-2. Protein phosphatase-2a was the activity detected in the presence of 30 nm inhibitor-2 and 15 pg/ ml protamine. Assays of protein phosphatase-li also included 1.25 mm magnesium acetate, mm ATP, and glycogen synthase kinase-3. The mixture was preincubated for 15 min at 30 C before addition of the substrate, to activate PP-lI maximally.

3 33 l.7 ~ t t Time, rnin Fig. 2. Dephosphorylation of the GI and b subunits of mammalian phosphorylase kinase by the Drosophila membrane fraction. Assays were carried out as described in the text, and phosphate remaining in the GI subunit (0) and the p subunit (0) was quantified at various times as described in [22]. The membrane fraction was assayed at a final dilution of 15-fold. (A) Membrane fraction; (B) membrane fraction extracted with 0.5 M NaCl PP-lI was the activity measured after preincubation with MgATP and glycogen synthase kinase-3 after correction for activity measured in the presence of MgATP alone. Phosphdtase-2B activity was measured by monitoring the release of 32P from 32P-labelled inhibitor-1. Unless otherwise indicated, activity was measured in the presence of 3 mm EGTA and 2.64 mm + CaZ (to yield a final free + CaZ concentration of 3 pm) and compared to the activity in the presence of 3 mm EGTA with no exogenous Ca2+ added. The reaction was initiated by addition of phosphatase rather than substrate, due to the lability of protein phosphatase-2b observed in the presence of CaZf (probably due to Ca'+-dependent proteinases) [ll, 181. One unit of protein phosphatase-1 or 2A activity was defined as that amount of enzyme which catalysed the release of 1 nmol phosphate/min from phosphorylase a. One unit of protein phosphatase-2b was that amount of enzyme which catalysed the release of 1 nmol phosphate/min from inhibitor-1. Preparation of u phosphutuse inhibitor fraction from Drosophila A soluble fraction was prepared from Drosophila heads as described above, supplemented with 0.05 vol. 1 M sodium phosphate, ph 7.0, and heated in a boiling water bath for 20min. The preparation was then cooled on ice and centrifuged for 20 rnin at x g. The supernatant was collected and trichloroacetic acid added to a final concentration of 10%. After standing on ice for 90 min, the precipitate was collected by centrifugation at x g for 30 min, and either resuspended in 0.1 mm EGTA, 0.5 M Tris/HCl ph 8.5 and dialysed against 0.1 mm EGTA, 20 mm Tris/HCl ph 8.5, or washed three times with ether to remove trichloroacetic acid, followed by resuspension in 0.1 mm EGTA, 20 mm Tris/HCl ph 8.5 (after evaporation of the ether). In both cases the preparations were chromatographed on a Mono Q anionexchange column, equilibrated with buffer A. The column was developed with a 40-ml linear gradient of mm NaC1. Individual fractions were dialysed against 0.1 mm EGTA, 0.1% (v/v) 2-mercaptoethanol, 20 mm Tris/HCl, ph I I I I I I Protomine, pq/ml Fig. 3. Effect of protamine on the phosphorylase phosphatase (PP) activity in membrane and soluble fractions prepared from Drosophila heads. Membrane (0) and soluble (0) fractions were obtained after centrifugation at x g for 60 min. The soluble fraction was then further centrifuged at I00000 x g for 60 min, to yield a high-speed pellet (W) and a high-speed supernatant (A). Fractions were diluted to 0.5 U/ml and aliquots (0.01 ml) preincubated with protamine (0.01 ml) before assaying for phosphorylase phosphatase activity as described under Methods 7.0, and assayed for their ability to inhibit the catalytic subunit of protein phosphatase-1 from mammalian muscle. One unit of inhibitor activity was defined as that amount which inhibited 0.01 U protein phosphatase-1 by 50% under the standard assay conditions. Phosphorylution and dephosphorylution of the Drosophila inhibitor fraction The inhibitor fraction prepared as above (but not chromatographed on Mono Q) was incubated in a reaction containing 200 pm ATP, 2 mm magnesium acetate, the catalytic subunit of CAMP-dependent protein kinase, and 50 mm Tris/HCl, ph 7.0. The reaction was carried out at 30 C for 40 min and terminated by addition of EDTA (final concentration 4 mm) and heating for 5 min at 100 C, followed by centrifugation for 2min at 1OOOOxg to remove denatured proteins. Dephosphorylation of the inhibitor fraction was carried out in a reaction mixture containing 1 mm MnCIz, 25 U/ml of the catalytic subunit of protein phosphatase-2a, 0.1 % (v/v) 2-mercaptoethanol, and 50 mm Tris/HCl, ph 7.0. The reaction was carried out for 40 min at 30 C and terminated by addition of EGTA (final concentration 4 mm) followed by immersion for 2 min in a boiling water bath and centrifugation for 2 min at x g to remove denatured proteins. Protein determination Protein was routinely determined as described by Bradford [24], using bovine serum albumin (A;&,,,,, = 6.5) as a standard. The concentrations of phosphorylase and phosphorylase kinase were determined from absorbance measurements taking their values as 13.1 [25] and 12.4 [14], respectively.

4 34 ' 0 O 210 L 20 L. 30 L. 40 l Time, min Fig. 4. efject ofprotumine on dephosphorylation of the CL and fl subunits of phosphoryluse kinuse by membrane and soluble fractions from Drosophila. Aliquots of the membrane (A) or soluble (B) fractions were diluted five-fold and aliquots (0.05 ml) preincubated in the absence (0, m) or presence (0, Kl) of 15 pg/ml protamine (0.05 ml). Following the addition of phosphorylase kinase, phosphate remaining in the a subunit ([I, W) or /l subunit (0, 0) was quantified at various times as described in [22] RESULTS Protein phosphuluse activity associated with the Drosophila membrane fraction Protein phosphatase-i and protein phosphatase-2a are the only enzymes in mammalian tissues with significant phosphorylase phosphatase activity [2,1 I] ; initial experiments were therefore carried out using phosphorylase a as substrate. Phosphorylase phosphatase activity of the membrane preparations increased with dilution, reaching 160 U/g wet weight heads and a specific activity of 4 U/mg protein at final dilutions of 500-fold or more. Activity was 4-fold lower if the membranes were assayed at only a 5-fold final dilution of the membrane fractions. Similar observations have been made in mammalian tissues [ll]. Approximately 90% of the phosphorylase phosphatase activity could be extracted from the membrane fraction with either 0.2% Triton or 0.5 M NaCl, and this activity was inhibited SO-90% by either inhibitor-1 or inhibitor-2 from mammalian muscle (Fig. 1). Furthermore, the concentration of either inhibitor required for 50% inhibition was 3 nm (Fig. l), which is similar to the Js0 values obtained using mammalian protein phosphatase-1 (e. g. [26]). Since the membrane fraction also dephosphorylated the p subunit of phosphorylase kinase specifically, either before (Fig. 2A) or after (Fig. 2B) extraction of the membranes with 0.5 M NaCl, the results indicated that nearly all of the directly measurable phosphatase activity towards phosphorylase a and phosphorylase kinase was catalysed by an enzyme(s) that closely resembled mammalian protein phosphatase-i, The dephosphorylation of phosphorylase a and phosphorylase kinase by mammalian protein phosphatase-i is inhibited very potently by protamine, whereas this polybasic protein is an activator of protein phosphatase-2a at low con- Fraction Fig. 5. Separation of Drosophila protein phosphatase-1 and protein phosphatase-2a activities on heparin-sepharose (3 x 2 cm). An aliquot (4 ml) of a 0.2% Triton X-100 extract of head membranes (A) and a further aliquot (10 ml) of the soluble fraction (B) were made 0.1 M in NaCl and applied to heparin-sepharose pre-equilibrated with buffer A M NaCI. Arrows indicate the points at which the columns were eluted with buffer A M NaCI. Fractions (3 ml) were assayed for phosphorylase phosphatase (PP) activity with no additions (O), 30 nm inhibitor-2 (A) or 15 pg/ml protamine (0) centrations and only an inhibitor at higher concentration [27, 281. Protamine can therefore be used like inhibitor-1 and inhibitor-2 to distinguish protein phosphatases-1 and 2A. The phosphorylase phosphatase activity associated with the Drosophila membrane fraction was inhibited by protamine (Fig. 3) and dephosphorylation of the p subunit of phosphorylase kinase was abolished at 15 pg/ml protamine (Fig. 4), demonstrating further similarity between Drosophila and mammalian protein phosphatase-i. However, dephosphorylation of the a subunit of phosphorylase kinase by the membrane fraction was enhanced by protamine (Fig. 4), suggesting that the membranes contained a latent form of protein phosphatase-2a whose activity was only revealed in the presence of this protein. Dephosphorylation of the a subunit also became detectable after extraction of the membrane with Triton (not shown), indicating that the latent form of protein phosphatase-2a was slightly activated in the presence of detergent. These observations were confirmed by chromatography of the Triton extract on heparin-sepharose as described below. It has been shown previously that mammalian type-1 protein phosphatases are retained by heparin-sepharose whereas the type-2a enzymes are excluded [29]. Identical results were obtained with the Drosophila membrane fraction (Fig. 5). The flow-through fractions contained small amounts of phosphorylase phosphatase activity that was unaffected by 30 nm inhibitor-2 and stimulated about 4-fold by 15 pg/ml protamine, whereas the 0.5 M NaCl eluate contained a much larger amount of phosphorylase phosphatase activity that was inactivated > 90% by 30 nm inhibitor-2 and 65-70% by 15 pg/ml protamine (Fig. 5). Protein phosphatase activity associated with the soluble fraction Phosphorylase phosphatase activity in the soluble fraction also increased with dilution, reaching 635 U/g wet weight and

5 35 a specific activity of 5 U/mg protein at final dilutions of 100- fold or greater. Activity was 5-fold lower if the membranes were assayed at final dilutions of only 5-fold. The level of phosphorylase phosphatase activity in Drosophilu heads is at least as high as in mammalian brain. Identical experiments to those described above were carried out using the soluble fraction ( x g supernatant). The results obtained were similar to those obtained with the membrane fraction, except that the proportion of protein phosphatase-2a was higher. Only about 50% of the phosphorylase phosphatase activity in the soluble fraction was inhibited at 300 nm inhibitor-i, compared to 80% in the membrane fraction (Fig. 1). The fi subunit of phosphorylase kinase was dephosphorylated more rapidly than the a subunit, but dephosphorylation of the latter was significant even in the absence of protamine (Fig. 4). Protamine stimulated dephosphorylation of the a subunit and abolished dephosphorylation of the p subunit (Fig. 4). Phosphorylase phosphatase activity was slightly stimulated at low concentrations of protamine and inhibited at higher concentrations (Fig. 3). This suggested that inhibition of protein phosphatase-1 was counterbalanced by activation of protein phosphatase-2a at low concentrations of protamine. This interpretation was confirmed by chromatography on heparin-sepharose, which resolved an inhibitor-2-insensitive phosphorylase phosphatase activity stimulated 6-7-fold by protamine that was present in the flow-through, from an inhibitor-2-sensitive activity in the 0.5 M NaCl eluate (Fig. 5B). The proportion of the inhibitor- 2-insensitive activity was higher than in the membrane fraction (Fig. 5B). It should be noted that following chromatography of the soluble fraction on heparin-sepharose, the 0.5 M NaCl eluate also contained some protein phosphatase-2a since inhibitor- 2 decreased activity by only 70%, while protamine only inhbited slightly (Fig. 5B). Small amounts of protein phosphatase-2a were also present in the 0.5 M NaCl eluate after chromatography of the membrane fraction on heparin- Sepharose (Fig. 5 A). Protein phosphatase-2a does not appear to be removed from heparin-sepharose completely by washing with 0.1 M NaCI. Identfication of protein phosphatase-1, and inhibitor-2 in the soluble fraction Mammalian tissues contain an inactive form of protein phosphatase-1, termed protein phosphatase-1 I, that consists of a 1 : 1 complex between the phosphatase catalytic subunit and inhibitor-2. This species has also been termed the MgATPdependent protein phosphatase because its activation, which involves the reversible phosphorylation of inhibitor-2 on a threonine residue, requires preincubation with MgATP and the enzyme glycogen synthase kinase-3 (reviewed in [l]). Protein phosphatase-1, is located in the soluble fraction of mammalian tissues [30] and is not retained by blue-sepharose [16], in contrast to the active forms of protein phosphatases- 1 and 2A. The soluble fraction prepared from Drosophila heads was therefore chromatographed on blue-sepharose and the fractions assayed before and after preincubation with MgATP and mammalian glycogen synthase kinase-3. As shown in Fig. 6, the flow-through fractions contained a phosphatase whose activity was dependent on preincubation with MgATP and glycogen synthase kinase-3, whereas phosphatase activity eluted with 0.5 M NaCl was not increased by this treatment. After preincubation with MgATP and glycogen synthase kinase-3, the activity in the flow- Fraction Fig. 6. Identification of protein phosphatase-li in the soluble fraction of Drosophila by chromatography on blue-sepharose (2 x 1 cm). An aliquot of the soluble fraction (5 ml) was applied to blue-sepharose equilibrated with 0.1 % (v/v) 2-mercaptoethanol, 10% (v/v) glycerol, 20mM Tris/HCl, ph 7.0, and the column washed with the same buffer. At the point indicated by an arrow, the buffer was supplemented with 0.5 M NaCl. Fractions (1 ml) were assayed for phosphorylase phosphatase (PP) after preincubation with 1.25 mm Mg2+ and mm ATP, in the presence (0) or absence (0) of GSK3, as described under Methods through fractions could be blocked (80-90%) by inhibitor-2 (not shown), confirming that this activity was indeed a form of protein phosphatase-1. The presence of protein phosphatase-ll in Drosophila suggested that inhibitor-2 must be present in the soluble fraction. A partially purified inhibitor fraction was therefore prepared by heat treatment and trichloroacetic acid precipitation as described under Methods. This fraction inhibited the catalytic subunit of mammalian protein phosphatase-1, but not protein phosphatase-2a (not shown), suggesting a relationship to either inhibitor-1 or inhibitor-2. Following chromatography on Mono Q, the inhibitor was recovered as a single peak which eluted at about 0.3 M NaCl. Preincubation of the fractions from Mono Q with MgATP and the catalytic subunit of CAMP-dependent protein kinase did not increase the amount of inhibitor, nor did it lead to the appearance of a new peak of inhbitor activity. In addition, the inhibitor activity was not decreased by preincubation with the catalytic subunit of protein phosphatase-2a in the presence of Mn2+ (see Methods) under conditions that caused 100% inactivation of mammalian inhibitor-1 (not shown). These results indicated that inhibitor-1 may not exist in Drosophilu heads, at least not in a heat- and acid-stable form. The following evidence demonstrates that the Drosophilu inhibitor detected by chromatography on Mono Q is inhibitor-2. Mammalian inhibitor-2 is not inactivated by sodium dodecyl sulphate and its mobility on SDS/polyacrylamide gels can therefore be determined even in crude preparations [31, 321. The apparent molecular mass of the inhibitor was found to be 31 kda (Fig. 7) identical to mammalian inhbitor-2 [23, 321. Furthermore, preincubation of the Drosophilu inhibitor with the catalytic subunit of mammalian protein phosphatase- 1 led to the formation of protein phosphatase-l,, whose activity depended on preincubation with MgATP and glycogen synthase kinase-3 (Fig. 8). The Drosophilu inhibitor was

6 ~ 36 Table 1. Effect of Ca2' and calmodulin on inhibitor-1 phosphatase activity in Drosophila fractions Solid ammonium sulphate was added to the soluble fraction to bring the degree of saturation to 45%, and after standing in ice for 45 min the solution was centrifuged for 45 min at 4200 x g. The supernatant was discarded and the precipitate redissolved in 10 mm Tris/HCl, ph 7.0, 0.1% (v/v) 2-mercaptoethanol, 1 mm benzamidine, 0.1 mm PhMeSOzF (solution A) and reprecipitated at 45% ammonium sulphate. The precipitated protein was collected by centrifugation as described above, and redissolved in solution A containing 5 mm EGTA. Denatured protein was removed by centrifugation and the supernatant dialysed against solution A + 3 mm EGTA. It was then diluted and assayed in the presence of either 3 mm EGTA or 3 pm free Ca2+, or 3 pm Ca2+ plus 3 pm calmodulin, as described under Methods Slice No Fig. I. Electrophori~sls of the Drosophila inhibitor fraction on SDS/ polyucrylumide gel. An aliquot of the inhibitor fraction (150 U) that had been partially purified through chromatography on Mono Q was subjected to electrophoresis on 12% polyacrylamide gel according to Laemmli [45]. The gel was sliced into 2-mm sections and each section extracted for 16 h at 20 C with 80 pl of 1 mm EGTA in 50 mm triethanolamine chloride. ph 7.0. The extracts were diluted fivefold and assayed for inhibition of the catalytic subunit of mammalian protein phosphatase- 1 (0.3 Ujml), as described under Methods. Marker proteins were bovine serum albumin (68 kda), ovalbumin (44 kda), mammalian inhibitor-2 (31 kda) and carbonic anhydrase (29.5 kda). Inhibitor activity was only detected in the three slices indicated Fraction Activity with Activation EGTA Caz+ CaZ+ + by calmodulin calmodulin U/ml -fold Cytosol st 45% ammonium sulphate precipitate 2nd 45% ammonium sulphate precipitate I "Oh I L-. I I I I I I I Inhibitor, U/ml Fig. 8. Reconstitution of protein phosphatase-li,from the catalytic subunit oj mammaliun protein phosphatase-1 and either mammalian inhibitor-2 (0) or the Drosophila inhibitorfraction (0). The catalytic subunit was mixed at 1 U/ml with the indicated amounts of inhibitor in a total volume of 50 pl. After incubation for 90 min at 30"C, phosphatase activity was measured after a preincubation with I.25 mm +. MgZ mm ATP and GSK-3. Activity observed after preincubation with MgATP in the absence of GSK-3 was subtracted to give protein phosphatase-1,. Protein phosphatase-li is plotted as a percentage of the activity obtained after incubating the catalytic subunit in the absence of inhibitor-2 equally as effective as mammalian inhibitor-2 in promoting the Formation of protein phosphatase-i, (Fig. 8). The amount of inhibitor-2 present in Drosophilu heads was 1200 U/g wet weight, at least as high as the amount present in mammalian muscle and liver [32] (and unpublished experiments). Identification of protein phosphatase-2b in Drosophila The experiments described above were carried out in the presence of EGTA (see Methods) and would not therefore TFP, pg/ml Fig. 9. Effect of trijluoperazine (TFP) on the activity of Drosophila protein phosphatase-2b (PP). The Drosophilu enzyme was prepared by fractionation with ammonium sulphate as described in the legend to Table 1. Activity was assayed at 0.03 U/ml using inhibitor-i as substrate in the presence of 3 pm free Ca2+ and 3 pm calmodulin (see Methods) have detected protein phosphatase-2b, a Ca2 +-dependent, calmodulin-stimulated enzyme. However, when the soluble fraction was assayed using 32P-labelled inhibitor-i (the most effective in vitro substrate for protein phosphatase-2b [IS]) phosphatase activity in the soluble fraction of Drosophilu was stimulated 4-%fold by Ca2+ (Table 1). No stimulation by calmodulin was observable in the soluble fraction, however, presumably due to the high concentrations of endogenous calmodulin. In order to demonstrate activation by calmodulin, it was necessary to precipitate protein phosphatase-2b with 45% ammonium sulphate and to wash the pellet with EGTA. After this treatment activity was stimulated nearly 4- Fold by calmodulin (Table 1). The Ca2'-dependent, calmodulin-stimulated activity could be inhibited almost completely by trifluoperazine (Fig. 9). The Is0 (65 pm) was

7 31 similar to the concentration that inhibits mammalian protein phosphatase-2b [18]. The activities of protein phosphatases- 1, 2A and 2C are unaffected by this drug [18]. The total protein phosphatase-2b activity present in the soluble fraction of Drosophilu heads was 196 U/g wet weight. All the 32P radioactivity released from inhbitor-1 in the presence of Ca2+ and calmodulin could be complexed with molybdate and extracted into isobutanol/benzene (1 : l), demonstrating that the product was inorganic phosphate and not a phosphopeptide released by the action of Ca'+-dependent proteinase(s). No Ca2+-dependent, calmodulin-stimulated protein phosphatase activity was detected in the Drosophilu membrane fraction. DISCUSSION The results described in this paper have shown that protein phosphatases-l,2a and 2B are present in Drosophilu, and that their properties are remarkably similar to the corresponding enzymes in mammalian cells. Thus Drosophilu protein phosphatase-i dephosphorylated the p subunit of phosphorylase kinase and was inactivated by mammalian inhibitor-1 and inhibitor-2 at concentrations similar to those that inhibit mammalian protein phosphatase-1 (Figs 1, 2 and 4). Like mammalian protein phosphatase-1, the Drosophilu enzyme was retained by heparin-sepharose (Fig. 5) and dephosphorylation of phosphorylase u and phosphorylase kinase by this enzyme was prevented by protamine (Figs 3 and 4). The identification of an inactive form of protein phosphatase-1, in Drosophila that can be activated by mammalian glycogen synthase kinase-3 and MgATP (Fig. 6) implies that the threonine residue on inhibitor-2 that is phosphorylated by glycogen snythase kinase-3 is conserved in Drosophila. Drosophilu inhibitor-2 has the same molecular mass as its mammalian counterpart (Fig. 7) and combines with the catalytic subunit of mammalian protein phosphatase- 1 to form a 'hybrid' protein phosphatase-1,. This implies a striking conservation of the overall structure of inhibitor- 2 and the regions involved in interaction between protein phosphatase-1 and inhibitor-2. Finally, the observation that a significant proportion of Drosophilu protein phosphatase-1 is associated with membranes is consistent with recent work, which indicates that much of the protein phosphatase-1 activity in mammalian tissues is bound to organelles and membranes (reviewed in [l, 331). The properties of Drosophila protein phosphatase-2a also closely resemble those of the mammalian enzyme, e. g. insensitivity to inhibitor-1 and inhibitor-2 (Fig. 5), preferential dephosphorylation of the a subunit of phosphorylase kinase (Fig. 4), activation by protamine (Figs 4 and 5) and lack of retention by heparin-sepharose. In mammalian cells, the catalytic (C) subunits of protein phosphatase-1 and protein phosphatase-2a have molecular masses of 37 kda and 36 kda, respectively, and peptide mapping studies have established that they are the products of distinct genes However, the free subunits do not exist in vivo, but are complexed to other proteins that have a regulatory function or are involved in targetting the catalytic subunits to particular subcellular locations (reviewed in 111). The two major forms of protein phosphatase-2a in mammalian cells, termed protein phosphatase-2ao and protein phosphatase-2al, have molecular masses of almost 200 kda [28]. Similarly, Drosophilu protein phosphatase-2a in either the membrane (after extraction with 0.5 M NaCl) or soluble fraction yielded an apparent molecular mass of about 200 kda when subjected to gel-filtration on Superose 6 (not shown). The directly measurable protein phosphatase-1 activity (i.e. not protein phosphatase-lj was found by us to be larger, the apparent molecular masses of the membrane and soluble forms being about 700 kda (after solubilisation with 0.5 M NaCl) and about 500 kda, respectively (not shown). After purification of protein phosphatase-1 from the membrane fraction by chromatography on heparin-sepharose and mono-q, the apparent molecular mass decreased to about 35 kda, suggesting that the free catalytic subunit had been released, either by dissociation or degradation of other subunits (not shown). Protein phosphatase-2b was also detected in Drosophilu, and its concentration in Drosophila heads was of the same order of magnitude as the corresponding enzyme in mammalian tissues 111, 181. We have not yet addressed ourselves to the presence of protein phosphatase-2c in Drosophila: this form, however, accounts for only a very small proportion of the activity towards phosphorylase a, phosphorylase kinase and inhibitor-i in mammalian tissues. Mutations are already known in Drosophilu in genes that code for proteins involved in phosphorylation cascades, and these mutations lead to relatively specific lesions in behavioural plasticity (reviewed in [9, 101). For example, mutations in the dunce gene that codes for an isozyme of camp phosphodiesterase result in fleeting memory 135, 361. Products of the same gene may also play a role in development [35]. Mutations in the rutabaga gene cause loss of Ca2+/ calmodulin modulation and of other regulatory properties of adenylate cyclase, and also to fleeting memory Mutations in the turnip gene affect the activity of protein kinase C and impair learning and memory [40]. Mutations in the Ddc gene that codes for dopa decarboxylase, an enzyme that participates in the synthesis of dopamine and serotonin, lead to a lesion in acquisition [41]. These two neurotransmitters function, at least partially, via activation of adenylate cyclase Taken together, the data support the notion that phosphorylation processes play an import role in neuronal and behavioural plasticity. They also indicate that subpopulations of these enzymes may play relatively specific roles in the above-mentioned processes. Studies of protein phosphatases and their inhibitors in Drosophilu heads should facilitate genetic analysis of the structure, function and regulation of dephosphorylation systems, taking advantage of the powerful genetic tools available for the fruit fly. Mutations that affect Drosophila acid phosphatase and alkaline phosphatase are already well known 1431, and also appear to spare protein phosphatase activity [6]. Although only a few protein phosphatase catalytic subunits have been identified to date, many subpopulations of these ubiquitous enzymes are likely to be present, since as described above, the catalytic subunits form complexes with a variety of regulatory and/or targetting subunits. Subspecies of the catalytic subunits may also exist, produced for example by differential gene splicing, as recently suggested for camp phosphodiesterase in the dunce gene [44]. Identification of specific physiological roles for subpopulations of enzymes that have broad and overlapping specificities, such as protein phosphatase-1 and protein phosphatase-2a, may be facilitated by identification of mutated phenotypes. Studies of the level of protein phosphatase-2b in various mutants will be of special interest, since ths enzyme is a prominent protein phosphatase in the mammalian nervous system and plays a role in Ca2 +-regulation of phosphorylation cascades.

8 38 This study was carried out during tenure of an EMBO short-term Fellowship (to Sara Orgad), by a grant from the US-Israel Binational Science Foundation, Jerusalem (to Yadin Dudai) and by the Medical Research Council, London, and Royal Society (to Philip Cohen). The phosphorylase kinase, CAMP-dependent protein kinase, glycogen synthase kinase-3. protein phosphatase-i, protein phosphatase-2b, inhibitor-i, inhibitor-2 and calmodulin were purified by Alexander Chisholm, Charles Holmes, Clare MacGowan, Carl Smythe, Lex Stewart and Nicholas Tonks in Dundee. REFERENCES 1. Cohen, P. (1985) Eur. J. Biochem. 151, Ingebritsen, T, S. & Cohen, P. (1983) Science (Wash. DC) 221, Greengard, P. (1978) Science (Wash. DC) 199, Nestler, E. J. & Greengard, P. (1984) Proteinphosphorylation in the nervous svstem, J. Wiley, New York. 5. Buxbaum, J. & Dudai, Y. (1985) SOC. Neurosci. 11, 1090 (abstract). 6. Orgad, S. & Dudai, Y. (1985) Soc. Neurosci. 11, 1090 (abstract). 7. Devay, P., Solti. M., Kiss, I., Dombardi, V. & Friedrich, P. (1984) Int..I. Biochem. 16, Hesse, J. (1984) Adv. Cyclic. Nucleotide Protein Phosphorylation Res. 17A, Dudai, Y. (1985) Trends Neurosci. 8, Quinn, W. G. ( 1984) in Biology oflearning (Marler, P. & Terrace, H. S., eds) pp , Springer-Verlag, Berlin. 11. Ingebritsen, T. S., Stewart, A. A. & Cohen, P. (1983) Eur. J. Biochem. 132, Lewis, E. B. (1960) Drosophila Inf. Ser. 34, Fischer, E. H. & Krebs, E. G. (1958) J. Biol. Chem. 231, Cohen, P. (1973) Eur. J. Biochem. 34, Woodgett, J. K., & Cohen, P. (1984) Biochim. Biophys. Acta 788, Resink, T. J., Hemmings, B. A., Tung, H. Y. L. & Cohen, P. (1983) Eur..J. Biochem. 133, Tonks, N. K. & Cohen. P. (1984) Eur. J. Biochem. 145, Stewart, A. A., Ingrebritsen, T. S. & Cohen, P. (1983) Eur. J. Biochem. 132, Beavo, J. A.. Bechtel, P. J. & Krebs, E. G. (1974) Methods Enzymol38, Shenolikar, S., Cohen, P. T. W., Cohen, P., Nairn, A. C. &Perry, S. V. (1979) Lur. J. Biochem. 100, Antoniw, J. W., Nimmo, H. G., Yeaman, S. J. & Cohen, P. (1977) Biochem. J. 162, Stewart, A. A,, Hemmings, B. A,, Cohen, P., Goris, J. & Merlevede, W. (1981) Eur. J. Biochem. 115, Foulkes, J. G. & Cohen, P. (1980) Eur. J. Biochem. 105, Bradford, M. M. (1976) Anal. Biochem. 72, Cohen, P., Duewer, T. & Fischer, E. H. (1973) Biochemistry 10, Alemany, S., Pelech, S., Brierley, C. H. & Cohen, P. (1986) Eur. J. Biochem. 156, Pelech, S. & Cohen, P. (1985) Eur. J. Biochem. 148, Tung, H. Y. L., Alemany, S. & Cohen, P. (1985) Eur. J. Biochem. 148, Erdodi, F., Csortos, C., Bot, G. & Gergely, P. (1985) Biochem. Biophys. Res. Commun. 128, Tung, H. Y. L. & Cohen, P. (1984) Eur. J. Biochem. 145, Huang, F. L. & Glinsmann, W. H. (1976) Eur. J. Biochem. 70, Chisholm, A. A. K. & Cohen, P. (1985) Biochim. Biophys. Acta 847, Kuret, J., Bell, H. & Cohen, P. (1986) FEBSLett. 203, Tung, H. Y. L., Resink, T. J., Hemmings, B. A., Shenolikar, S. & Cohen, P. (1984) Eur. J. Biochem. 138, Byers, D., Davis, R. L. & Kiger, J. A. (1981) Nature (Lond.) 289, Dudai, Y. (1979) J. Comp. Physiol. 130, Dudai, Y., Zvi, S. & Segel, S. (1984) J. Comp. Physiol. 155, Livingstone, M. S., Sziber, P. P. & Quinn, W. G. (1984) Ce1137, Dudai, Y. (1985) FEBS Lett. 191, Smith, R. F., Choi, K.-W., Tully, T. & Quinn, W. G. (1986) Abstracts of Cold Spring Harbor Symposium on G-proteins, p. 16, Cold Spring Harbor, New York. 41. Tempel, B. L., Livingstone, M. S. & Quinn, W. G. (1984) Proc. Natl Acad. Sci. USA 81, Uzzan, A. & Dudai, Y. (1982) J. Neurochem. 38, Lindsley, D. L. & Grell, E. H. (1968) Genetic variations qf Drosophila, Carnegie Institution, Washington. 44. Davis, R. L. & Davidson, N. (1985) Mol. Cell Bid. 6, Laemmli, U. K. (1970) Nature (Lond.) 227, Note udded in proof(received February 5, 1987). Dombradi, V., Friedrich, P. and Bot, G (1987) Comp. Biochem. Physiol. in the press, have also used inhibitor-i and inhibitor-2 from mammalian muscle and chromatography on heparin-sepharose to identify protein phosphatase-i and protein phosphatase-2a in drosophila eggs and larvae, and in the whole body and heads of adult flies.