Isolation and characterization of two 3-phosphatases that hydrolyze both phosphatidylinositol 3-phosphate and inositol 1,3-bisphosphate.

Inositol-polyphosphate 3-phosphatase catalyzes the hydrolysis of the 3-position phosphate bond of inositol 1,3-bisphosphate (Ins(1,3)P2) to form inositol 1-monophosphate and inorganic phosphate (Bansal, V.S., Inhorn, R.C., and Majerus, P.W. (1987) J. Biol. Chem. 262, 9444-9447). Phosphatidylinositol 3-phosphatase catalyzes the analogous reaction utilizing phosphatidylinositol 3-phosphate (PtdIns(3)P) as substrate to form phosphatidylinositol and inorganic phosphate (Lips, D.L., and Majerus, P.W. (1989) J. Biol. Chem. 264, 19911-19915). We now demonstrate that these enzyme activities are identical. Two forms of the enzyme, designated Type I and II 3-phosphatases, were isolated from rat brain. The Type I 3-phosphatase consisted of a protein doublet that migrated at a relative Mr of 65,000 upon sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis. The Mr of this isoform upon size-exclusion chromatography was 110,000, suggesting that the native enzyme is a dimer. The Type II enzyme consisted of equal amounts of an Mr = 65,000 doublet and an Mr = 78,000 band upon SDS-polyacrylamide gel electrophoresis. This isoform displayed an Mr upon size-exclusion chromatography of 147,000, indicating that it is a heterodimer. The Type II 3-phosphatase catalyzed the hydrolysis of Ins(1,3)P2 with a catalytic efficiency of one-nineteenth of that measured for the Type I enzyme, whereas PtdIns(3)P was hydrolyzed by the Type II 3-phosphatase at three times the rate measured for the Type I 3-phosphatase. The Mr = 65,000 subunits of the two forms of 3-phosphatase appear to be the same based on co-migration on SDS-polyacrylamide gels and peptide maps generated with Staphylococcus aureus protease V8 and trypsin. The peptide map of the Mr = 78,000 subunit was different from that of the Mr = 65,000 subunits. Thus, we propose that the differing relative specificities of the Type I and II 3-phosphatases for Ins(1,3)P2 and PtdIns(3)P are due to the presence of the Mr = 78,000 subunit of the Type II enzyme.


Isolation and Characterization of Two 3-Phosphatases That Hydrolyze Both Phosphatidylinositol 3-Phosphate and Inositol 1,3-Bisphosphate*
(Received for publication, March 19, 1991) Kevin K. CaldwellS 264,[19911][19912][19913][19914][19915]. We now demonstrate that these enzyme activities are identical. Two forms of the enzyme, designated Type I and I1 3-phosphatases, were isolated from rat brain. The Type I 3-phosphatase consisted of a protein doublet that migrated a t a relative M, of 65,000 upon sodium dodecyl sulfate (SDS)polyacrylamide gel electrophoresis. The M, of this isoform upon size-exclusion chromatography was 110,000, suggesting that the native enzyme is a dimer. The Type I1 enyzme consisted of equal amounts of an M, = 65,000 doublet and an M, = 78,000 band upon SDS-polyacrylamide gel electrophoresis. This isoform displayed an M, upon size-exclusion chromatography of 147,000, indicating that it is a heterodimer. The Type I1 3-phosphatase catalyzed the hydrolysis of Ins( 1 ,3)Pz with a catalytic efficiency of one-nineteenth of that measured for the Type I enzyme, whereas PtdIns(3)P was hydrolyzed by the Type I1 3-phosphatase at three times the rate measured for the Type 13phosphatase. The M, = 65,000 subunits of the two forms of 3-phosphatase appear to be the same based on co-migration on SDS-polyacrylamide gels and peptide maps generated with Staphylococcus aureus protease VS and trypsin. The peptide map of the M, = 78,000 subunit was different from that of the M, = 65,000 subunits. Thus, we propose that the differing relative specificities of the Type I and I1 3 tidylinositol (PtdIns)' have been shown to act as signals coupling various extracellular stimuli to intracellular responses (1)(2)(3). Five phosphorylated forms of the parent molecule, PtdIns, have been identified: phosphatidylinositol 4-phosphate (PtdIns(4)P), phosphatidylinositol 4,5-bisphosphate (PtdIns(4,5)P2), phosphatidylinositol 3-phosphate (PtdIns(3)P), phosphatidylinositol 3,4-bisphosphate (PtdIns(3,4)P2), and phosphatidylinositol 3,4,5-trisphosphate (PtdIns(3,4,5)P3) (4, 5). Of these lipids, PtdIns, PtdIns(4)P, and PtdIns(4,5)P2 are hydrolyzed by PtdIns-specific phospholipase C to yield inositol phosphates and 1,2-diacylglycerol (6,7). In contrast, the 3-phosphate-containing lipids are resistant to the action of PtdIns-specific phospholipase C (8,9). Rather, these lipids are metabolized by a group of relatively poorly described phosphatases and kinases (4, 5). We recently identified one of these enzymes, phosphatidylinositol 3-phosphatase, in NIH 3T3 cell extracts (10). This enzyme catalyzes the hydrolysis of the phosphate bond in the 3-position of the inositol ring of PtdIns(3)P to form PtdIns and inorganic phosphate.
Assay of ["2P]Ptdlns(3)P3 Hydrolysis-PtdIns(3)P hydrolysis was measured as follows. ['2P]PtdIns(3)P in chloroform/methanol and PtdIns (carrier lipid) were dried under a stream of nitrogen; suspended in 500 mM KCl, 100 mM MES, p H 6.5, 12.5 mbf EDTA, and 1.5% (w/v) octyl glucoside; and sonicated for 30 s in a bath sonicator. Reaction mixtures contained 1 nM ["'PP]PtdIns(Y)P, 20 p M PtdIns, 100 mM KC1, 20 mM MES, pH 6.5, 2.5 mM EDTA, 0.2 mg of cytochrome c/ml, and 0.3% (w/v) octyl glucoside in 20 pl. Reactions were started by addition of enzyme, incubated a t 37 "C for 3-10 min, and terminated by addition of 500 pl of 10% (w/v) trichloroacetic acid followed by 50 pl of 20% (v/v) Triton X-100 as described (15). The mixture was centrifuged, and the aqueous and organic phases were counted in a liquid scintillation counter with ScintiVerse I (Fisher). Radioactivity in the product (""PO4) was 10-50% of the total. The limited quantities of [,"'P]PtdIns(3)P that could be synthe- was lyophilized to remove ammonium formate, dissolved in deionized water, and stored a t -90 "C. The phosphate content of the preparation was determined as described by Ames and Dubin (17).
Assay of ['HlIns(l,3)P2 Hydrolysis-Throughout the purification procedure, inositol-polyphosphate 3-phosphatase activity was assayed (Assay 1) by incubating ["H]Ins(l,3)P2 (1300-2800 cpm/pmol, 2000 cpm) with the enyzme preparation in 50 mM MES, pH 6.5, 5 mM EDTA, 0.2 mg of bovine serum albumin/ml, and 8 mM 2-mercaptoethanol for 10 min a t 37 "C in 25 pl. Reactions were stopped by addition of 1 ml of water and applied to a 0.5-ml Dowex A G1-X8 column previously equilibrated with 50 mM ammonium formate. The product, [:'H]Ins(l)P, was eluted in two 2.5-ml fractions with 0.2 M ammonium formate, 0.1 M formic acid and then counted in a liquid scintillation counter. Under these assay conditions of low (30-60 nM) substrate concentrations, enzyme activity follows first-order kinetics and is expressed as counts/minute of product formed. For comparisons of inositol-polyphosphate 3-phosphatase activity with The K , and VmaX values were derived from reactions assayed (Assay 2) in the presence of 20 mM MES, pH 6.5, 2.5 mM EDTA, 3.5 mM MgC12, 0.3% (w/v) octyl glucoside, and 0.2 mg of cytochrome c/ml using ["H]Ins(l,3)PZ of a lower specific activity (170 cpm/pmol). In these assays, <20% of the substrate was consumed.
Heat Inactiuation-Samples of the purified Type I and I1 3-phosphatases (14 and 100 pg of protein/ml, respectively) were diluted 1:lOO and 1:200, respectively, into 20 mM MES, p H 6.5, 2.5 mM EDTA, 100 mM KC1, 0.2 mg of cytochrome c/ml, and 0.3% (w/v) octyl glucoside. Samples (10 p l ) were warmed to room temperature and then treated for various times a t 45 "C for the Type I enzyme and at 49 "C for the Type I1 enzyme.
Preparation of Rat Brain Homogenate and Cytosolic Fraction-All procedures were performed a t 4 "C. Rat brains (2950 brains, wet weight = 5160 g), obtained from Pel-Freez Biologicals (Rogers, AR), were shredded with a commercial vegetable shredder and suspended (1:3, w/v) in homogenization buffer (50 mM MES, pH 6.5, 5 mM EDTA, 225 mM sucrose, 20 mM 2-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride (PMSF), 0.5 mM benzamidine, 10 pg of aprotinin/ml, 10 pg of leupeptin/ml, and 1 pg of pepstatin A/ml). The tissue was homogenized with a Polytron (Brinkmann Instruments) and centrifuged (20,300 X g for 30 min). The supernatant fraction was decanted, and the pellet was resuspended in buffer a t 1:2 (v/v) and rehomogenized and centrifuged as described above. The supernatants were combined and mixed with phosphocellulose.
Phosphocellulose Batch Chromatography-Six liters (packed volume) of phosphocellulose, equilibrated with wash buffer (50 mM MES, p H 6.5, 5 mM EDTA, 20 mM 2-mercaptoethanol, 1 mM PMSF, and 0.5 mM benzamidine) at 4 "C, was mixed with one-half of the brain cytosolic fraction in a 150-liter plastic container. The mixture was allowed to stand for 90 min, and the supernatant was aspirated. The second half of the cytosolic preparation was then mixed with the phosphocellulose and processed as described above. The resin was washed twice with 24 liters of wash buffer and then suspended in 18 liters of 667 mM NaCl (to give a 500 mM NaCl final concentration) in elution buffer (same as wash buffer with 10 pg of leupeptin/ml). The suspension was stirred slowly for 20 min and then allowed to settle for 90 min. The supernatant was decanted, and the phosphocellulose was mixed with 18 liters of 500 mM NaCl in elution buffer. The phosphocellulose was allowed to settle as described above, and the supernatant was decanted and combined with the first eluate. The eluates were centrifuged briefly at 1000 rpm in an IEC Model Pr-2 centrifuge to remove residual phosphocellulose. Solid ammonium sulfate (351 g/liter) was added over 10 min with stirring to give a 55% saturated solution. The suspension was mixed for an additional 20 min. The mixture settled for 30 min and then was centrifuged (20,300 X g for 15 min). The supernatant was decanted, and the pellet was dissolved in 600 ml of DEAE equilibration buffer (see below). The sample was dialyzed overnight against 2 X 25 liters of the same buffer. The dialyzed sample was centrifuged (41,400 X g for 30 min), and the resulting supernatant was filtered through a 0.45-pm filter. The filtrate was adjusted to pH 8 with 1 M Tris, and the solution was diluted with DEAE equilibration buffer to lower the conductivity to 1.5 mmho (at 4 "C).
DEAE HPLC-The sample was loaded a t 20 ml/min onto a Bio-Rad Bio-Gel TSK DEAE-5PW industrial preparative column (55 X 200 mm) equilibrated with 20 mM Tris-C1, pH 8.0 (room temperature), 20 mM 2-mercaptoethanol, 4 pg each of calpain inhibitors I and 11/ ml, and 0.5 mM benzamidine. The column was washed with 1.15 liters of equilibration buffer a t 20 ml/min and then eluted with a 7.2-liter gradient of 0-250 mM NaCl at 20 ml/min. Fractions (20 ml) were collected and immediately adjusted to contain 1 mM EDTA.
The two peaks of enzyme activity from the DEAE column (see Fig.  1) were pooled separately and precipitated with ammonium sulfate as described above. The pellets were resuspended in 75 ml of 20 mM sodium phosphate, pH 7.0, 20 mM 2-mercaptoethanol, 4 pg each of calpain inhibitors I and II/ml, 0.5 mM benzamidine, and 1 mM PMSF. The conductivity of the sample was measured, and solid ammonium sulfate was added (as described above) to increase the conductivity to that of a 25% saturated solution of ammonium sulfate. The supernatant after centrifugation (41,400 X g for 20 min) was filtered through a 0.45-pm filter and chromatographed on a hydrophobic interaction column.
Hydrophobic Interaction Chromatography-Each sample was applied a t 8 ml/min to a Bio-Rad Bio-Gel phenyl-5PW column (21.5 X 150 mm) equilibrated with 20 mM sodium phosphate, pH 7.0, 20 mM 2-mercaptoethanol, and 25% saturated ammonium sulfate. The column was washed with 165 ml of equilibration buffer and eluted with a 1156-ml gradient from 25% saturated to 0% ammonium sulfate in 20 mM sodium phosphate, pH 7.0, and 20 mM 2-mercaptoethanol a t 8 ml/min. Fractions (20 ml) were collected, and EDTA was added to a 2.5 mM final concentration.
The fractions containing enzyme activity were pooled and concentrated in a Micro-ProDiCon (Bio-Molecular Dynamics, Beaverton, OR) against 50 mM HEPES, pH 7.2, 5 mM EDTA, 20 mM 2mercaptoethanol, 10% (v/v) glycerol, 1 mM PMSF, 0.5 mM benzamidine, 10 pg of leupeptin/ml, and 4 pg of calpain inhibitor II/ml. Following concentration, the sample was stored at -90 "C. Both the Type I and I1 enzyme activities were stable for at least 6 weeks a t -90 "C. The next three chromatography steps, Mono S, hydroxylapatite, and Mono Q, were completed in a single day.
Mono S Chromatography-The samples from hydrophobic interaction chromatography were thawed at room temperature and then diluted 1:4 (v/v) with Mono S equilibration buffer (20 mM HEPES, p H 7.2, 5 mM EDTA, 20 mM 2-mercaptoethanol, 20% (v/v) glycerol, 0.3% (w/v) octyl glucoside, 1 mM PMSF, 0.5 mM benzamidine, 10 pg each of calpain inhibitors I and II/ml, 40 pg of leupeptin/ml, and 10 gg of aprotinin/ml). Each sample was applied to a Mono S HR5/5 (Pharmacia LKB Biotechnology Inc.) column (5 X 50 mm) a t 1 ml/ min. The column was washed with equilibration buffer and eluted at a flow rate of 1 ml/min with a 30-ml linear gradient of 0-640 mM NaCl in 50 mM HEPES, pH 7.2, and EDTA, 2-mercaptoethanol, glycerol, octyl glucoside, and protease inhibitors as described for equilibration buffer. Fractions (0.5 ml) were collected, and the enzyme activity was pooled and adjusted to contain 20 mM sodium phosphate. The sample was chromatographed on hydroxylapatite.
Hydroxylapatite Chromatography-Each sample was applied at 3 ml/min to a 10 x 100-mm column of hydroxylapatite equilibrated with 20 mM sodium phosphate, p H 7.0, 20 mM 2-mercaptoethanol, 20% (v/v) glycerol, 0.3% (w/v) octyl glucoside, 0.5 mM benzamidine, 10 pg each of calpain inhibitors I and II/ml, 40 pg of leupeptin/ml, and 10 pg of aprotinin/ml. The column was washed with equilibration buffer and eluted a t 3 ml/min with a 320-ml linear gradient of 20-500 mM sodium phosphate in the same buffer.
Fractions (8 ml) with the peak of enzyme activity were pooled, diluted 1:4 (v/v) with Mono Q equilibration buffer (see below), and applied to a Mono Q column.
Mono Q Chromatography I-Each sample was loaded at 1.5 ml/ min onto a Mono Q HR5/5 (Pharmacia LKB Biotechnology Inc.) column (5 X 50 mm) equilibrated with 20 mM BisTris, pH 7.0,20 mM 2-mercaptoethanol, 20% (v/v) glycerol, 0.3% (w/v) octyl glucoside, and the same protease inhibitors as described above. The column was washed with equilibration buffer and eluted with a 30-ml linear gradient of 0-640 mM NaCl in 50 mM BisTris, p H 7.0, and 2mercaptoethanol, glycerol, octyl glucoside, and protease inhibitors. Fractions (0.5 ml) were collected, and the enzyme activity was pooled and concentrated in a Micro-ProDiCon against 50 mM MES, pH 6.5, 5 mM EDTA, 50% (v/v) glycerol, 5 mM dithiothreitol, 0.3% (w/v) octyl glucoside, and 100 mM KC1. The concentrated enzyme preparations were frozen in liquid nitrogen and then stored at -90 "C. The enzyme activity was stable for a t least 1 month at -90 "C. However, for the procedure detailed in this report, the samples were left for 4 months at -90 "C and displayed significant losses of activity (see "Results" and Table I).
Mono Q Chromatography 2-Each sample was thawed a t room temperature, diluted 1:20 (v/v) with Mono Q equilibration buffer (see above), and rechromatographed on Mono Q as described above, except that the fraction volume was 0.25 ml and the flow rate was 1 ml/min.
The peak of enzyme activity was pooled and subjected to sizeexclusion chromatography. Size-exclusion Chromatography-Samples (1 ml) from Mono Q chromatography 2 were applied a t a flow rate of 0.5 ml/min to a series of three 600 X 7.5-mm Bio-Rad TSK size-exclusion columns equilibrated with 50 mM MES, pH 6.5, 2.5 mM EDTA, 200 mM NaCl, and 0.3% (w/v) octyl glucoside. Fractions (0.5 ml) were collected and assayed for enzyme activity, and portions of each were analyzed by The fractions containing enzyme activity were pooled and concentrated in a Micro-ProDiCon as described above and frozen a t -90 "C.
Iodination The gel slices were thawed and washed as described above; and the proteins were digested, or not (control), with S. aurew protease V8 or trypsin and electrophoresed as described by Cleveland et al. (19). Peptide digests were detected by autoradiography of the stained gel.

SDS-PAGE.
Enzyme Purification-Initial studies of phosphatidylinosi-to1 3-phosphatase disclosed similarities to inositol-polyphosphate 3-phosphatase, including the observations that both activities: 1) displayed similar subcellular distributions, 2) were maximal in the presence of EDTA, and 3) bound to and were eluted from DEAE and Mono Q columns under similar conditions. However, preliminary experiments with crude preparations of the PtdIns(3)P-hydrolyzing activity from NIH 3T3 cells (10) indicated that there was little inositolpolyphosphate 3-phosphatase activity in fractions enriched for the PtdIns(3)P phosphatase (see "Discussion"). Thus, to determine the relationship of these enzymatic activities to each other, we performed simultaneous purification of each from rat brain. At each stage of the procedure described below, the elution profiles of the Ins(l,3)Pz and PtdIns(3)P phosphatase activities were highly similar. For the sake of clarity, the purification procedure will be detailed for Ins(l,3)P2 phosphatase until the final step of the procedure, size-exclusion chromatography.
Approximately 70% of the total Ins(l,3)Pz phosphatase activity in the rat brain homogenate was in the 20,300 X g soluble fraction. Batch elution of the cytosolic enzyme from phosphocellulose resulted in a 15-fold purification as shown in Table I. The enzyme activity was resolved into two peaks on a DEAE HPLC column (Fig. 1). We designated these activities as Type I and I1 3-phosphatases based on their elution at conductivities of 2.4 and 3.3 mmho, respectively. We initially considered the possibility that the two peaks resulted from partial proteolysis. However, two pieces of evidence available at that time led us to conclude that they were two distinct forms of inositol-polyphosphate 3-phosphatase. First, inclusion of a wide spectrum of protease inhibitors in all buffers did not change significantly the relative distribution of the Type I versus I1 3-phosphatase. Second, fraction-   Conductivities of the fractions (---I were measured at 4 "C.
ation of calf brain homogenates by the same procedure yielded the same two peaks of activity, with the same relative distribution.' In light of this, we continued the purification of each type separately.
The Type I 3-phosphatase eluted from a hydrophobic interaction chromatography column a t 11% ammonium sulfate, whereas the Type I1 enyzme eluted a t -13% ammonium sulfate. The Type I 3-phosphatase had been purified 1900fold, and the Type I1 3-phosphatase had been purified 2200fold at this stage when both samples were frozen.
We next carried out chromatography on Mono S, hydroxylapatite, and Mono Q columns in a single day and refroze each preparation. Both enzyme preparations displayed cross-  contamination with the other form of the enzyme. The Type I and I1 3-phosphatases were best resolved on the Mono Q column (Fig. 2, upper and lower, respectively). The crosscontamination of the preparations accounts for the observed activity in fractions 21-24 in Fig. 2 (upper), which reflects Type I and 113-Phosphatases the Type I1 enzyme in the Type I preparation, and that in fractions 14-18 in Fig. 2 (lower), which reflects the Type I 3phosphatase in the Type I1 preparation. SDS-PAGE of fractions from the Mono Q columns indicated that the activity of the Type I preparation correlated with a polypeptide migrating at an M , of 65,000; however, the preparation was not homogeneous. The enzyme activity in the Type I1 preparation correlated with two protein bands present in approximately equal amounts, corresponding to M , = 65,000 and 78,000.
Several other proteins also were detected in these fractions. Therefore, we further subjected each preparation to repeat chromatography on Mono Q followed by size exclusion chromatography (Fig. 3, upper and center).
Both the Ins(l,3)P2 and PtdIns(3)P phosphatase activities in the Type I preparation eluted from the size-exclusion column with an M , of 110,000 (Fig. 3, upper). SDS-PAGE analysis of the column fractions (data not shown) showed that the activity correlated with a protein doublet at M , = 65,000. This gel was stained with silver to determine whether any M , = 78,000 polypeptide was present. None was detected in the peak fractions, which were pooled and concentrated. In these fractions, >95% of the protein migrated at M, = 65,000.
The Ins(l,3)P2-and PtdIns(3)P-hydrolyzing Type I1 enzyme eluted from the size-exclusion column with an M , of 147,000 (Fig. 3, center). The enzyme activity correlated with polypeptides that migrated at M , = 65,000 and 78,000 upon SDS-PAGE (Fig. 3, lower). These two bands were present in approximately equal amounts, suggesting that the Type I1 enzyme is a heterodimer of these polypeptides. Samples of the pooled concentrated Type I and I1 enzymes were analyzed by SDS-PAGE as shown in Fig. 4. The M , = 65,000 proteins of the Type I and I1 enzymes co-migrate, suggesting that they are the same. This hypothesis was substantiated by the results of peptide mapping (see below).
The Type I enzyme had been purified 38,000-fold, and the Type I1 3-phosphatase had been purified 8000-fold. The total yield of enzyme activity was only 0.13%. When samples were thawed for the final Mono Q and size-exclusion chromatography steps, there was a substantial loss of activity (Table I).
In retrospect, a larger yield may have been obtained had the second Mono Q column been omitted and size-exclusion chromatography carried out directly after Mono Q chromatography 1. In a repeat of this procedure, we omitted the Mono Q and size-exclusion chromatography steps to achieve higher yields of protein for sequencing and only purified the Type I1 enzyme. In this preparation, we obtained -400 pg of the Type I1 enzyme.
The copurification of the Ins(l,3)P2 and PtdIns(3)P phosphatase activities is summarized in Table I. The purification was similar as monitored by either assay with two exceptions. After separation of the Type I and I1 3-phosphatases on DEAE, the -fold purification using Ins(1,3)P2 as substrate was 120-133 (Table I, phenyl load), whereas using PtdIns(3)P, the purification was only 30-38-fold. This apparent discrepancy may be explained by the fact that the PtdIns(3)P phosphatase assay is extremely sensitive to changes in octyl glucoside concentration (10). The assay conditions were optimized using crude supernatant fractions that contained undetermined amounts of lipid that may alter the optimal octyl glucoside concentration. Therefore, after DEAE chromatography, a low apparent yield of activity may be due to the use of less than optimal assay conditions compared to the crude preparation. The other discrepancy occurred in the Blue. The Type I preparation was derived from fractions 90-92, and the Type I1 enzyme preparation from fractions 86-88 of the respective size-exclusion column. The relative migration of protein standards (see Fig. 3 (lower) legend) is shown.

Type I and 113-Phosphatases
Type I1 preparation upon Mono Q column 1, where the yield was excellent as monitored by PtdIns(3)P phosphatase activity, whereas two-thirds of the Ins(l,3)Pp phosphatase activity was lost. We have no explanation for this result except to speculate that some alteration in the enzyme occurred that favored the lipid substrate.
Relative Enzyme Activities-The absolute and relative activities of the purified Type I and I1 3-phosphatases for both substrates are shown in Table 11. In terms of absolute activity differences, the Type I1 enzyme catalyzed the hydrolysis of PtdIns(3)P three times as well as did the Type I enzyme. This is in contrast to the &fold greater rate of Ins(l,3)PP metabolism by the Type I enzyme. This difference in the rate of Ins(l,3)P2 breakdown was even greater when more optimal assay conditions were used, in which case the Type I 3phosphatase had a 19-fold greater catalytic efficiency than the Type I1 enzyme (see below). A comparison of the relative specificities of the Type I and I1 3-phosphatases for the two substrates shows that the Type I1 enzyme is 16 times more selective for the lipid than is the Type I enzyme (ratio of relative activities = 68, cf. 4.2). These results imply that the Type I1 enzyme functions primarily to metabolize PtdIns(3)P and thus is most appropriately designated phosphatidylinositol 3-phosphatase. The greater catalytic efficiency of the Type I enzyme toward Ins(l,3)P2 suggests that it regulates Ins(l,3)P2 levels in cells; and thus, we designate it inositolpolyphosphate 3-phosphatase. These differences in specificity were documented repeatedly and are apparent from the activities in Table I and Fig. 3.
These differences in specificity may explain the apparent difference in -fold purification in Table I. Since the initial separation of Type I and I1 3-phosphatases is not complete, each is initially cross-contaminated with the other. Thus, when the Type I1 3-phosphatase is assayed with PtdIns(3)P as substrate, contamination with the Type I enzyme has little effect on the total activity measured. Thus, as the contaminating Type I enzyme is removed, there is little effect on the calculated yield and -fold purification for the Type I1 preparation. However, when Ins(l,3)P2 phosphatase activity is measured on the same preparation, the removal of the contaminating Type I enzyme has a major effect on the total yield of activity. Thus, the apparent -fold purification and yield of Type 11 3-phosphatase is less. The same argument can explain the fact that the apparent purification of the Type I enzyme assayed with Ins(l,3)Pp as substrate is much greater than that determined using PtdIns(3)P (Table I).
Heat Inactivation Studies-As further support for the conclusion that the enzyme activities that catalyze the hydrolysis of Ins(l,3)Pp and PtdIns(3)P were identical, we performed heat inactivation studies utilizing the purified Type I and I1 3-phosphatases. Both the Type I and I1 3-phosphatases were stable at 41 "C for at least 30 min (data not shown). However, the two enzyme activities displayed different sensitivities to temperatures >41 "C. Paired samples of the purified Type I

TABLE I1
Relative specificities of Type I and II3-phosphatases for PtdIns(3)P and Ins(l,3)P2 The ratios of the activities are expressed as first-order rate constants of the breakdown of PtdIns(3)P and Ins(l,3)P2 for the purified Type I and I1 enzymes. Data are the mean of triplicate assays. and I1 enzymes were heated for varying times and then assayed for both Ins(l,3)P2 and PtdIns(3)P phosphatase activities. The Type I 3-phosphatase was more heat-labile than the Type I1 3-phosphatase. Type I 3-phosphatase activity (Fig. 5, upper) had a t,,2 at 45 "C of 13 min for both Ins(l,3)P2 and PtdIns(3)P. The tlh for the Type I1 3-phosphatase (Fig.  5, lower) at 49 "C was 17 and 16 min for Ins(l,3)P2 and PtdIns(3)P, respectively. The tL,2 for the Type I 3-phosphatase at 49 "C was 3 min (data not shown).
Peptide Mapping of Type I and II 3-Phosphatases-To assess the structural similarity of the M , = 65,000 proteins of the Type I and I1 3-phosphatases and to compare each to the M , = 78,000 subunit of the Type I1 enzyme, mapping of peptides obtained from protease digests of '251-radiolabeled proteins was undertaken (Fig. 6). Cleveland digests (19) of the M , = 65,000 subunits of Type I and I1 3-phosphatases yielded the same pattern of peptides with both S. aureus protease V8 and trypsin. Shorter exposures of the film demonstrated that the low molecular weight peptides generated from the M , = 65,000 subunit were identical. Both the Type I and I1 M , = 65,000 subunits were slightly degraded in the control (no protease) lanes to the same peptides and to approximately the same extent. These observations demonstrate that the Type I and I1 M , = 65,000 subunits are the same, or very similar, and thus implicate this subunit as the catalytic subunit for Ins(l,3)P2 and PtdIns (3)   results indicate that any activity of Type I or I1 3-phosphatase with these inositol polyphosphates is (1% of that with Ins( 1,3)P2.

DISCUSSION
The ratio of the activity of the Type I 3-phosphatase to that of the Type I1 3-phosphatase for PtdIns(3)P hydrolysis was markedly different than that for Ins(l,3)P2 hydrolysis (Table 11). Whereas the rate of PtdIns(3)P hydrolysis by the Type I1 enzyme was -3-fold greater than that by the Type I enzyme, the Type I enzyme catalyzed the hydrolysis of Ins( 1,3)P2 approximately five times better than did the Type I1 enzyme. The latter value underestimates the differences between the two enzymes since the kinetic data in Fig. 7 indicate that the Type I enzyme hydrolyzes Ins(l,3)P2 19 times better than does the Type I1 enzyme (catalytic efficiency = V,,,,,/K,). The demonstration that peptide maps of protease digests of the putative catalytic (Mr = 65,000) subunits for the two enzymes were the same (Fig. 6) suggests that the substrate specificity differences between the Type I and I1 3phosphatases are due to the presence of the M , = 78,000 subunit in the Type I1 isoform. These results contradict our previous conclusion (10) that the PtdIns(3)P-and Ins(l,3)P2-hydrolyzing activities are distinct. Under the assay conditions used previously to measure PtdIns(3)P-hydrolyzing activity (100 mM KCl, 20 pM PtdIns), the breakdown of Ins(l,3)P2 by the Type I and I1 enzymes is inhibited by 95 and 80%, respectively (data not shown). Thus, little Ins(l,3)P2 phosphatase activity would have been detected. Indeed, we did note that Ins(l,3)P2 was broken down a t -6% of the rate measured for the hydrolysis of PtdIns(3)P (10).
The large purification that was achieved indicates that Type I and I1 3-phosphatases are trace proteins in rat brain.
Furthermore, both Type I and I1 3-phosphatases are highly specific for Ins(1,3)P2 and PtdIns(3)P since neither hydrolyzed other 3-phosphate-containing inositol phosphates including Ins(3,4)P2, Ins(1,3,4)P3 and Ins(1,3,4,5)P4. Gee and co-workers (23) have identified a Mg2'-independent form of Ins( 1,3)P2 phosphatase in bovine brain cytosolic preparations. We have identified two forms of 3-phosphatase in bovine brain that resemble the rat brain enzymes based on chromatographic properties and substrate specificity. ' The regulation of the substrate specificity of the catalytic subunit of the 3-phosphatase by the M, = 78,000 subunit is similar to two previously described activities: the regulation of galactosyltransferase activity by a-lactalbumin and the regulation of thrombin activity by thrombomodulin. a-Lactalbumin decreases the apparent K,,, for galactosyltransferase for both glucose and N-acetylglucosamine while increasing the VmaX for glucose and decreasing the V,,, for N-acetylglucosamine (31). Binding of thrombomodulin to thrombin produces a complex that activates protein C rather than factors V and VI11 (32, 33).
PtdIns(3)P has been implicated as a mediator of cell growth and transformation based on the association of phosphatidylinositol 3-kinase with growth factor receptors (13, 39) and oncoproteins (40). However, cellular levels of PtdIns(3)P inferred from incorporation of 32P04 do not change dramatically with growth factor stimulations (35) or upon transformation by polyoma virus (a), although incorporation of 32P04 into PtdIns(3,4)P2 and PtdIns(3,4,5)P3 increases markedly (41). The flux through PtdIns(3)P as well as the actual mass remain to be determined, although they are presumably controlled by the relative activities of three or more enzymes: PtdIns 3kinase (42, 43), PtdIns(3)P phosphatase, and kinases that further phosphorylate the molecule (5, 16). PtdIns(3)P is a constituent of nonproliferating and nontransformed cells such as cultured fibroblasts (34), human vascular smooth muscle cells (35), blood platelets (16,36), and Saccharomyces cereuisiae (38). Thus, it likely plays a role in normal as well as transformed cellular function.
Interestingly, PtdIns 3-kinase and PtdIns(3)P phosphatase share several features. Both may exist either in a form containing only the catalytic subunit (Type I) or as a heterodimer (Type 11) that contains a regulatory subunit (Ref. 42 and this study). In the case of PtdIns 3-kinase, the heterodimeric Type I1 enzyme is five times less active than the monomeric catalytic subunit (42). The regulatory subunit is M , = 85,000 and is phosphorylated on tyrosine in response to cellular activation by PDGF (43) or insulin (44). In stimulated cells, formation of 3-phosphate-containing phosphatidylinositols is markedly enhanced transiently (41), implying increased PtdIns 3-kinase activity or reduced PtdIns(3)P phosphatase activity. PtdIns(3)P phosphatase also exists in two forms; and in this case, the heterodimeric form is more active than the homodimeric catalytic subunit. It is therefore possible that cellular activation by a growth factor such as PDGF stimulates modification of the regulatory subunits of both enzymes, which promotes dissociation into forms containing the catalytic subunit only, with active kinase and inactive phosphatase. Later, upon dephosphorylation, the heterodimeric enzymes could re-form, yielding a state where phosphatase activity predominates. Further characterization of PtdIns(3)P phosphatase and PtdIns 3-kinase will answer this and other questions regarding their role in cellular signaling.
The physiological actions of Ins(l,3)P2 are unknown. It is formed by inositol-polyphosphate 4-phosphatase-catalyzed hydrolysis of inositol 1,3,4-trisphosphate (11,12). Nahorski and co-workers (37, 45) have demonstrated that Ins(l,3)P2 levels increase severalfold in rat cerebral cortex slices following carbachol treatment of K+-dependent depolarization. The molecular basis and physiological consequences of these increases are unknown. However, the mechanism of this effect could be analogous to the dissociation mechanism postulated for the degradation of PtdIns(3)P, with the dissociated catalytic (Mr = 65,000) subunit more efficiently degrading Ins(l,3)Pp (since the Type I enzyme is more active toward Ins(l,3)P3). By such an interactive process, the signal(s) elicited by altered Ins(l,3)P2 and PtdIns(3)P metabolism may be integrated. Further insight into these mechanisms will be gained by the application of molecular cloning and immunological techniques to characterize the isoforms of 3-phosphatase.