Human Type 2 Phosphatidic Acid Phosphohydrolases

Phosphatidic acid (PA), lysophosphatidic acid, ceramide 1-phosphate (C1P), and sphingosine 1-phosphate (S1P) are lipid mediators generated by phospholipases, sphingomyelinases, and lipid kinases. The major pathway for degradation of these lipids is dephosphorylation catalyzed by members of two classes (types 1 and 2) of phosphohydrolase activities (PAPs). cDNAs encoding two type 2 PAPs, PAP-2a and -2b, have been expressed by transient transfection and shown to catalyze hydrolysis of PA, C1P, and S1P (Kai, M., Wada, I., Imai, S., Sakane, F. and Kanoh, H. (1997) J. Biol. Chem. 272, 24572–24578). We report the cloning and expression of a third type 2 PAP enzyme (288 amino acids, predicted molecular mass of 32.6 kDa), PAP-2c, which exhibits 54 and 43% sequence homology to PAPs 2a and 2b. Expression of HA epitope-tagged PAP-2a, -2b, and 2c in HEK293 cells produced immunoreactive proteins and increased membrane-associated PAP activity. Sf9 insect cells contain very low endogenous PAP activity. Recombinant expression of the three PAP enzymes using baculovirus vectors produces dramatic increases in membrane-associated Mg2+-independent,N-ethylmaleimide-insensitive PAP activity. Expression of PAP-2a but not PAP-2b or -2c resulted in high levels of cell surface PAP activity in intact insect cells. Kinetic analysis of PAP-2a, -2b, and -2c activity against PA, lysophosphatidic acid, C1P, and S1P presented in mixed micelles of Triton X-100 revealed differences in substrate specificity and susceptibility to inhibition by sphingosine, Zn2+, and propranol.

Phosphatidic acid (PA), lysophosphatidic acid, ceramide 1-phosphate (C1P), and sphingosine 1-phosphate (S1P) are lipid mediators generated by phospholipases, sphingomyelinases, and lipid kinases. The major pathway for degradation of these lipids is dephosphorylation catalyzed by members of two classes (types 1 and 2) of phosphohydrolase activities (PAPs). cDNAs encoding two type 2 PAPs, PAP-2a and -2b, have been expressed by transient transfection and shown to catalyze hydrolysis of PA, C1P, and S1P (Kai, M., Wada, I., Imai, S., Sakane, F. and Kanoh, H. (1997) J. Biol. Chem. 272, 24572-24578). We report the cloning and expression of a third type 2 PAP enzyme (288 amino acids, predicted molecular mass of 32.6 kDa), PAP-2c, which exhibits 54 and 43% sequence homology to PAPs 2a and 2b. Expression of HA epitope-tagged PAP-2a, -2b, and 2c in HEK293 cells produced immunoreactive proteins and increased membrane-associated PAP activity. Sf9 insect cells contain very low endogenous PAP activity. Recombinant expression of the three PAP enzymes using baculovirus vectors produces dramatic increases in membrane-associated Mg 2؉ -independent, N-ethylmaleimide-insensitive PAP activity. Expression of PAP-2a but not PAP-2b or -2c resulted in high levels of cell surface PAP activity in intact insect cells. Kinetic analysis of PAP-2a, -2b, and -2c activity against PA, lysophosphatidic acid, C1P, and S1P presented in mixed micelles of Triton X-100 revealed differences in substrate specificity and susceptibility to inhibition by sphingosine, Zn 2؉ , and propranol.
Phosphatidate phosphohydrolase (PAP) 1 catalyzes the dephosphorylation of phosphatidic acid (PA) to form diacylglycerol (DG). This enzyme was first recognized as a pivotal component of metabolic pathways controlling the synthesis of glycerophospholipids and triacylglycerols (1). More recent advances have focused attention on PA as a receptor-generated intracellular mediator formed by the phospholipase D (PLD)-catalyzed hydrolysis of phosphatidylcholine (PC). PA generated in this manner appears to act on a variety of cell-specific target proteins that include atypical protein kinase C isoforms, the Raf protein kinases, the Ras GTPase-activating protein, several proteins involved in cytoskeletal organization, and the neutrophil respiratory burst oxidase. In this case, PAP-catalyzed hydrolysis of PA serves to terminate the signaling functions of PA and concurrently generates diacylglycerol for activation of conventional Ca 2ϩ and phospholipid-dependent protein kinase C enzymes. DG kinase generates PA by ATPdependent phosphorylation of DG. Presumably the subcellular localization, catalytic, and regulatory properties of these different lipid-metabolizing enzymes will dictate the source, fate, and signaling functions of PA and DG (2,3).
PAP activity is widely expressed, and two classes of these enzymes can be distinguished in mammalian cells on the basis of their subcellular distribution and catalytic properties (4). Type 1 activity (PAP-1) is associated with the cytosol and endoplasmic reticulum and appears to redistribute from the soluble to membrane compartment upon treatment of hepatocytes with glucagon and glucocorticoids (4,5). PAP-1 activity is sensitive to inhibition by sulfhydryl reagents (most notably N-ethylmaleimide (NEM)) and displays an absolute requirement for Mg 2ϩ . PAP-2 activity has been localized to the plasma membrane and, in contrast to PAP-1, is independent of Mg 2ϩ and insensitive to inhibition by NEM (4 -8). PAP-2 activity is presumably directed toward the inner leaflet of the plasma membrane, although several reports also describe a cell surface enzyme with PAP-2-like properties that may serve to terminate the receptor-directed signaling functions of lyso-PA (LPA) and related compounds (9,10). Sphingosine 1-phosphate (S1P) and ceramide 1-phosphate (C1P) phosphatase activities have also been described in mammalian cells, and purified preparations of rat liver PAP-2 also catalyze the hydrolysis of these lipid phosphomonoesters (11). C1P, S1P, and their hydrolysis products exhibit a number of interesting biological activities and may function as intra-and possibly extracellular signaling molecules (12). The inference drawn from these studies is that PAP-1 most likely functions in lipid synthesis, while PAP-2 may have an important role to play in modulating the signaling functions of PA and LPA as well as lipid phosphomonoesters derived from sphingomyelin.
The tight association of PAP with membranes and the low abundance of the proteins has hampered purification attempts. PAP-1 has not yet been isolated. Several groups have prepared highly enriched preparations of PAP-2 from a number of tissue sources (6 -8). Rat liver PAP-2 was identified as a glycosylated protein of about 50 kDa that could be converted to a 28-kDa protein by N-glycanase treatment (6). A mouse cDNA (mPAP-2a) encoding a protein that is highly similar to a heat-inducible gene product (Hic53) has recently been isolated. When ex-pressed in HEK 293 cells, this protein localizes to the plasma membrane and is accompanied by a Mg 2ϩ -independent PAP activity (13). A human cDNA encoding a second mammalian PAP-2 enzyme, hPAP-2b (the human homolog of a previously described rat intestinal epithelial cell endoplasmic reticulumresident protein, Dri42), has been cloned (14,15). Transient expression of PAP-2a and -2b in HEK-293 cells produced increases in membrane-associated PA and C1P phosphatase activities. Hydrophobicity analysis suggests that PAP-2a and -2b are integral membrane proteins with six hydrophobic membrane-spanning regions (13)(14)(15)(16). Comparison of the sequence of mPAP-2a with other phosphatases defines a protein motif composed of three regions of conserved sequence. PAP-2 homologs have been identified in yeast and Drosophila (16 -20).
We sought to identify further human PAP homologs, and a search of expressed sequence tag data bases identified a number of candidate sequences. Full-length cDNAs encoding three of these (PAP-2a, PAP-2b, and a novel PAP-2 isoform, PAP-2c) have been cloned. These enzymes have been expressed as HA epitope-tagged proteins by transient transfection of HEK-293 cells and in Sf9 insect cells using baculovirus vectors. These systems have been used to investigate the cell surface activity of the three PAP enzymes, to examine their specificity for lipid phosphomonoester substrates presented as components of nonionic detergent micelles and susceptibilities to inhibition by a number of agents that have been widely used as modulators of PAP-2 activity in intact and broken cell systems.

EXPERIMENTAL PROCEDURES
Isolation and Analysis of hPAP-2a, -2b, and -2c cDNAs-We conducted blast searches of the GenBank TM data base of expressed sequence tags (ESTs) using regions of sequence conserved among mouse PAP-2 and a variety of previously recognized homologs including the product of the Drosophila wunen gene and four Saccharomyces cerevisiae open reading frames. We identified a number of candidate human sequences in this manner. Sequences encoding human homologs of mouse PAP-2a and rat Dri42 and a third apparently novel sequence were selected for further study. These cDNAs and the proteins they encode are hereafter referred to as PAP-2a, -2b, and -2c. An open reading frame containing the complete cDNA sequence of hPAP-2a was constructed from overlapping ESTs, and a complete cDNA was amplified from reverse-transcribed HL-60 cell cDNA by polymerase chain reaction using primers with the sequences 5Ј-GCTCTAGAACCATGT-TTGACAAGACGCGG-3Ј (forward) and 5Ј-CAGCCCGGGCTCGGCAC-CCTGCTG-3Ј (reverse) using standard methodology (21). The cDNA was ligated into pGEM-7ZF. One of the ESTs identified (accession number U79294) appeared to contain a complete open reading frame encoding hPAP-2b. This EST was obtained from the IMAGE Consortium, and the hPAP-2b cDNA was sequenced by a combination of manual sequencing using Sequenase 2.0 and automated sequencing performed in the center for analysis of macromolecules at Stony Brook. An open reading frame containing the complete cDNA sequence of hPAP-2c was constructed from overlapping ESTs, and a complete cDNA was amplified from reverse-transcribed HeLa cDNA using primers with the sequences 5Ј-GGTTCTAGAACCATGCAGCGGAGGTGGGTC-3Ј (forward) and 5Ј-GGGGGATCCCCTCAGGAGGAGGAGTGCGGG-3Ј (reverse) using standard methodologies (21). The cDNA was ligated into pGEM7ZF and sequenced by a combination of manual sequencing using Sequenase 2.0 and automated sequencing performed in the center for analysis of macromolecules at Stony Brook.
Transient Expression of hPAP-2a, -2b, and -2c in HEK-293 Cells-hPAP-2a, -2b, and -2c cDNAs were subcloned into pCGN, which is a cytomegalovirus-based vector for expression of proteins with an NH 2terminal HA epitope tag. HEK-293 cells were cultured in Dulbecco's modified Eagle's medium containing 10% fetal calf serum. 35-mm diameter lysine-coated dishes of 50% confluent cells were transfected with 1 g of pCGN hPAP-2a, pCGN hPAP-2b, or pCGN hPAP-2c using lipofectamine in Opti-MEM (Life Technologies, Inc.). The transfection medium was removed after 24 h and replaced with complete Dulbecco's modified Eagle's medium. The cells were harvested 24 h later by washing in phosphate-buffered saline followed by scraping into ice-cold lysis buffer containing 20 mM Tris, pH 7.5, 5 mM EGTA, 0.1 mM benzamidine, 0.1 mM phenylmethylsulfonyl fluoride. The lysate was disrupted by sonication on ice with a probe-type sonicator, and the material was used in assays within 24 -48 h as described below. In some cases, the lysate obtained was fractionated into total membrane and cytosolic fractions by centrifugation at 30,000 ϫ g at 4°C. The membrane fraction was resuspended in ice-cold lysis buffer by vortexing.
Baculovirus Expression of PAP-2a, -2b, and -2c-The PAP-2a, -2b, and -2c cDNAs were subcloned into pFastBac (Life Technologies, Inc.), and recombinant baculoviruses were generated by sequential transformation of DH10Bac cells, isolation of recombinant bacmids, and transfection of Sf9 cells using Cellfectin reagent. Recombinant baculoviruses were selected and propagated using standard procedures to generate high titer stocks (22). Monolayers of exponentially growing insect cells cultured in complete Grace's medium containing antibiotics, antimycotics, 10% fetal bovine serum and supplemented with lactalbumin and Yeastolate (generally 20 ϫ 10 6 cells in a 225-cm 2 flask) were infected with recombinant baculoviruses for expression of the PAP enzymes or of a control protein (PLD2) at a multiplicity of 10. The cells were cultured for 48 h at 27°C. For assays using intact insect cells, the monolayer of cells was carefully washed with unsupplemented Grace's medium, and the cells were dislodged by gentle pipetting and transferred to a 50-ml conical centrifuge tube. The intact cells were sedimented by centrifugation at 100 ϫ g for 10 min at room temperature and resuspended in unsupplemented Grace's medium. The cells were kept at 27°C and used within 6 h of isolation. Where indicated, for determinations of total lactate dehydrogenase or PAP activity, these cells were disrupted by sonication on ice (see below), and the sonicated material was kept on ice before use.
For studies using Sf9 cell membranes or detergent-extracted membrane proteins, the monolayers of infected cells were washed gently with phosphate-buffered saline and lysed by the addition of 4 ml of ice-cold lysis buffer and scraping. The cell suspension was transferred to a 15-ml conical tube, and the cells were disrupted by sonication (Vertis Systems Sonifier), 10 10-s pulses on ice. The disrupted cells were centrifuged at 20,000 ϫ g at 4°C for 20 min. The cytosolic fraction was removed, and the membrane fraction was resuspended in ice-cold lysis buffer. Detergent extracts were prepared from the membranes by the addition of Triton X-100 and ␤-D-octylglucopyranoside to final concentrations of 1% followed by incubation at 4°C with constant rocking for 1 h. The solubilized material was centrifuged at 26,000 ϫ g at 4°C for 30 min, and the supernatant was removed.
Preparation of Substrates-[ 32 P]PA, C1P, and LPA were prepared by phosphorylation of DG, C8 ceramide, or monooleoylglycerol (Avanti Polar Lipids) using Escherichia coli DG kinase (Calbiochem) and [␥-32 P]ATP (ICN) (11). The reactions were terminated by extraction with acidified CHCl 3 and MeOH, and the dried organic phases obtained were resuspended in 0.4 ml of 20:9:1 CHCl 3 /MeOH/H 2 O (solvent A) and neutralized by the addition of a small volume of 20% NH 4 OH in MeOH. This material was applied to an Econosil NH 2 5U high pressure liquid chromatography column (250 ϫ 4.2 mm) (Alltech Associates). The column was washed with 20 ml of solvent A and then eluted with a 40-ml linear gradient of 0 -1 M ammonium acetate in solvent A. 0.5-ml fractions of the eluant were collected, and associated radioactivity was determined by liquid scintillation counting. 32 P-Labeled products were pooled and extracted from the eluant by the addition of 3 M HCl and CHCl 3 to give two phases. [ 32 P]PA and [ 32 P]C1P prepared in this way were dried and resuspended in a small volume of solvent A. The lipids were stored at Ϫ20°C until use. [ 32 P]S1P was prepared by acid hydrolysis of [ 32 P]C1P and purified by thin layer chromatography on silica gel plates (11) or by phosphorylation of sphingosine using Swiss 3T3 cells as a source of sphingosine kinase (23). Radiolabeled material was identified by autoradiography, the silica was scraped from the plate, and lipids were eluted by the addition of acidified CHCl 3 /MeOH. The eluted material was stored at Ϫ20°C. Dipalmitoyl PA was obtained from Avanti Polar Lipids. Unlabeled C8 C1P and S1P were obtained from Biomol Inc. Sphingosine was from Avanti Polar lipids.
PAP Assays-The assay procedures used were adapted from those described by others (6,7). In brief, assays were performed in medium containing 20 mM Tris, pH 7.5, 1 mM EGTA, and 2 mM EDTA, and in some cases the MgCl 2 concentration of the assay medium was varied by the addition of MgCl 2 as indicated. C1P and S1P were stored as 1 mg/ml solutions in 1 mM Triton X-100. 32 P-Labeled lipid substrates were dried under vacuum and resuspended in 6.4 mM Triton X-100. Unlabeled PA was mixed with the radiolabeled substrate before drying and resuspension, while C1P and S1P were added to appropriate dried radiolabeled substrates from stock solutions in Triton X-100. In some experiments, dodecyl ␤-D-maltoside was substituted for Triton X-100. The substrates were dispersed by bath sonication and vortexing. Unless otherwise noted, the assay volume was 100 l, and each assay contained final concentrations of 3.2 mM Triton X-100 and 100 M 32 P-labeled lipid substrate. For investigations of the dependence of enzyme activity on the surface concentration of substrate, the substrate concentration was 100 M, and the detergent concentration was varied as indicated. Transfected cell fractions, Sf9 cell fractions, or detergent-extracted membrane proteins (generally 0.1-5 g of protein) were added directly to these incubations. In some cases, the source of PAP activity was incubated with 5 mM NEM at 37°C before addition to the incubations. Assays were performed at 37°C and were terminated by the addition of ice-cold 10 mg/ml bovine serum albumin and 10% trichloroacetic acid. The samples were centrifuged for 5 min in a microcentrifuge, and [ 32 P]PO 4 2Ϫ released into the supernatant was quantitated by liquid scintillation counting. This assay was validated by demonstrating that the water-soluble radioactivity released from the substrates was [ 32 P]PO 4 2Ϫ by quantitative extraction with ammonium molybdate. For assays of PAP activity using intact Sf9 cells, [ 32 P]dipalmitoyl-PA (approximately 10,000 dpm) was resuspended in unsupplemented Grace's medium containing 2 mg/ml bovine serum albumin to a final concentration of 20 mM by bath sonication. Assays were initiated by adding 1 ml of this substrate preparation to 1 ml of cells (generally 400 ϫ 10 3 cells) and incubation at 37°C with constant shaking. Aliquots of the suspension were removed at various times for determination of PAP activity by measurement of [ 32 P]PO 4 2Ϫ release as described above. In some cases, samples were removed for determination of lactate dehydrogenase activity.
Other Methods-SDS-polyacrylamide gel electrophoresis, Western blotting, and protein determinations were performed as described previously (24). Lactate dehydrogenase activity was determined as described (25). Inorganic [ 32 P]PO 4 2Ϫ was determined by extraction with ammonium molybdate (26). Fig. 1 shows an alignment of the deduced amino acid sequences of hPAP-2a, -2b, and -2c. PAP-2a comprises 289 amino acids corresponding to a protein of 32,788 kDa. PAP-2b comprises 311 amino acids corresponding to a protein of 35,120 kDa. PAP-2c comprises 288 amino acids corresponding to a protein of 32,577 kDa. PAP-2c is 54% identical to PAP-2a and 43% identical to PAP-2b. These values increase to 68 and 54%, respectively, when conservative amino acid substitutions are accounted for. It is notable that in comparison with PAP-2a and -2c, PAP-2b contains an extended N terminus with an enrichment of basic amino acid residues. The C-terminal 40 amino acids of the three proteins are also highly divergent. All three proteins contain a single consensus site for N-linked glycosylation (residue 140 in PAP-2c). Hydropathy analysis of all three sequences suggests that PAP-2a, -2b, and -2c are integral membrane proteins with six membrane-spanning regions of 17-25 hydrophobic amino acid residues. These sequences are underlined in Fig. 1 and denoted as regions I-VI. Mutagenesis of rat PAP-2b supports this putative transmembrane topology (15).

Sequence Comparison of PAP-2a, -2b, and -2c-
Expression of PAP-2a, -2b, and -2c in HEK293 Cells-We expressed PAP-2a, -2b, and -2c as HA epitope-tagged proteins by transient transfection of HEK293 cells using a cytomegalovirus promoter-based vector. Cells were harvested 48 h posttransfection. PAP activity was determined in lysates of control and transfected cells using PA as substrate. The cells expressing PAP-2a, -2b, and -2c exhibited 7-, 15-, and 7-fold increases in membrane-associated PAP activity, respectively ( Fig. 2A). Total protein from the transfected cells was separated by SDSpolyacrylamide gel electrophoresis on a 10% gel and analyzed by Western blotting using the HA epitope-specific 12CA-5 monoclonal antibody. In comparison with samples from untransfected cells, major immunoreactive proteins with estimated molecular masses of 33 and 37 kDa, 34 kDa, and 34 and 33 kDa were detected in cells expressing PAP-2a, -2b, and 2c, respectively (Fig. 2B). The immunoreactive material of higher molecular weight observed in the case of PAP-2a and -2c presumably represents the glycosylated forms of the proteins. We found that immunoreactive species of lower mobility could be partially converted to the faster migrating species upon treatment of membrane extracts with N-glycanase prior to SDSpolyacrylamide gel electrophoresis. PAP-2b also contains a consensus site for N-linked glycosylation, and we occasionally observed a minor immunoreactive species of higher molecular weight upon expression of this cDNA in COS-7 and HEK293 cells (not shown). It is unclear why comparatively higher levels of activity were observed in the pCGN PAP-2b transfected cells. It is possible that PAP-2b has an intrinsic higher specific activity than the other enzymes under the assay conditions we used or that this enzyme is differentially regulated in a manner that produces a specific increase in activity relative to the other enzymes.
Expression of PAP-2a, -2b, and -2c in Insect Cells Using Baculovirus Vectors-Recombinant baculovirus-infected Sf9 insect cells have proved to be an effective system for studying the regulation of integral membrane signaling proteins including adenylyl cylase isoforms and G-protein-coupled receptors (27,28). In comparison with the transient transfection studies described above, this approach also provides a simple and consistent means to produce recombinant protein that is necessary for detailed investigations of the catalytic properties of these enzymes. Recombinant baculoviruses for expression of PAP-2a, -2b, and -2c were prepared and monolayer cultures of insect cells infected with these viruses and a virus expressing a control protein (murine PLD2) as described under "Experimental Procedures." Cells were harvested 48 h post-transfection and fractionated into cytosolic and membrane fractions. PAP activity in these fractions was determined using PA as substrate. In comparison with control cells, cells expressing PAP-2a, -2b, and -2c exhibited dramatic (1140-, 540-, and 460-fold) increases in membrane-associated PAP activity (Table I). Cytosolic PAP activity was unaltered in cells expressing PAPs-2a, -2b, and -2c. PAP activity could be effectively extracted from the Sf9 cell membranes by a combination of 1% Triton X-100 and 1% ␤-D-octylglucopyranoside. This procedure routinely solubilized 70 -80% of PAP-2a, -2b, and -2c activity and approximately 30% of total membrane protein. Type 2 PAP activities are characteristically independent of Mg 2ϩ and insensitive to inhibition by NEM. Our standard assay medium contains 2 mM EDTA and no added Mg 2ϩ . Activity of the extracted PAP-2a, -2b, and -2c was unchanged by the addition of MgCl 2 to the assays to give a Mg 2ϩ concentration of 5 mM. Similarly, preincubation of the extracted material with 5 mM NEM produced very modest decreases in PAP-2a, -2b, and -2c activity (Table I).
Baculovirus-infected Sf9 cells are therefore an excellent model system for investigating the catalytic properties of recombinantly expressed PAP-2 enzymes.
Cell Surface Activity of PAP-2a, -2b, and -2c in Baculovirusinfected Insect Cells-Cell surface PAP activity has been detected in many different cell types, but the identity of the enzyme responsible has not been determined. To investigate the cell-surface expression of PAPs-2a, -2b, and -2c, we measured PAP activity of intact baculovirus-infected insect cells expressed using [ 32 P]dipalmitoyl-PA presented in 1 mg/ml bovine serum albumin. Fig. 3 shows time courses of PAP activity determined as 32 P release measured using intact cells or equiv- alent quantities of cells that have first been disrupted by sonication. Cell integrity was determined by measurement of lactate dehydrogenase release in parallel incubations. As expected, sonicated dispersions of PAP-2a-, PAP-2b-, and PAP-2c-expressing Sf9 cells exhibited high levels of PAP activity. Intact Sf9 cells expressing PAP-2a but not PAP-2b or -2c exhibited significantly elevated levels of PAP activity. In comparison with the sonicated cells, 60% of total Sf9 cell PAP activity could be measured in intact cells expressing PAP-2a, while only 8 and 10% of total PAP activity was detected in intact cells expressing PAPs -2b and -2c. Lactate dehydrogenase release after these 30-min incubations varied between 2 and 5%, indicating that the majority of the cells were intact and that the observations of PAP activity could not be accounted for by cell lysis during the incubation period. One possible explanation for these results is that the PA substrate is accumulated by the cells and metabolized at an intracellular site. This did not appear to be the case, since no appreciable radioactivity became associated with the cells upon incubation with [ 32 P]PA, and the kinetics of substrate hydrolysis measured as decrease in [ 32 P]PA paralleled the release of [ 32 P]PO 4 2Ϫ (data not shown).

Kinetic Analysis of PAP-2a, -2b, and -2c Activity against Substrates Presented in Nonionic Detergent Micelles-Purified
preparations of rat liver type 2 PAP activity have been reported to hydrolyze C1P and S1P in addition to PA, while transient expression of PAPs-2a and -2b resulted in an increased hydrolysis of both PA and C1P with the 2b isoenzyme exhibiting an apparent selectivity for S1P. Measurements of the affinities and maximal rates of substrate hydrolysis by PAPs-2a, -2b, and -2c would provide useful insight into their likely biological functions. Kinetic analyses of lipolytic or lipid-modifying en-zymes are best made using substrates presented in micelles of detergents with high aggregation numbers, where the surface concentration of substrate can be varied without compromising the physical state of the micelles. Other workers have usefully employed a surface dilution kinetic model (29) in which substrates are presented as Triton X-100 micelles for analysis of yeast and liver type 2 PAP activities (11,30). Clearly, these types of studies would best be performed using purified enzymes, but we were encouraged that the extremely high levels of expression obtained with the baculovirus system suggested that the use of detergent-extracted recombinant PAP enzymes in these assays would minimize interference from contaminating activities. We first examined the activity of PAP-2a, -2b, and -2c against PA, C1P, and S1P presented as components of Triton X-100 or dodecyl ␤-D-maltoside micelles. The substrate concentration was 100 mM, and the detergent concentration was 3 mM. The results obtained are shown in Table II. All three PAP enzymes were highly active against PA presented in both detergents. PAP-2a, -2b, and -2c also hydrolyzed C1P with high activity. For PAP-2a and -2b, activity with C1P as substrate was 53 and 72% of that observed with PA, whereas for PAP-2c, activity was 130% higher with C1P as substrate. The comparable activities of the enzymes against substrates presented in micelles of two structurally different detergents suggested that the apparent differences in substrate utilization reflected true enzyme specificity rather than a selective effect of the detergent on substrate presentation to the enzyme.
To calculate kinetic constants, we determined the activity of PAP-2a, -2b, and -2c using mixed micelles of Triton X-100 and PA, LPA, C1P, or S1P substrates. Initial rates of hydrolysis were measured as the surface concentration of each substrate  was increased, and the data obtained are shown in Fig. 4. For all three enzymes, activity increased with increasing substrate concentration in an apparently saturable manner, suggesting that a surface dilution kinetic model was appropriate for analysis of these enzymes. The data were analyzed by weighted nonlinear regression fit to the equation where V is the initial rate and A is the substrate concentration in mol % to calculate V max and K m values. R values for this analysis ranged from 0.84 to 0.97. The results obtained are presented in Table III.
Inhibition of PAP-2a, -2b, and -2c Activity by Sphingosine, Zn 2ϩ , and Propranolol-The product of S1P hydrolysis by PAP, sphingosine, has been reported to be a potent inhibitor of several type 2 PAP activities. PAP-2a, -2b, and -2c activity was determined using PA as substrate in the presence of increasing concentrations of sphingosine. All three enzymes were inhibited by this compound. Half-maximal inhibition of the 2a and 2b isoenzymes was observed at approximately 350 mM sphingosine, while the 2c isoform was appreciably more sensitive to this compound with half-maximal inhibition observed at approximately 100 mM (Fig. 5A). Several other inhibitors of type 2 PAP activities have been described and used to modulate PAP activity both in vitro and in intact cells. Zn 2ϩ was an effective inhibitor of all three enzymes with half-maximal inhibition observed at 30 mM (Fig. 5B). By contrast, propranolol, which is an effective inhibitor of type 1 PAP activities was only modestly effective as an inhibitor of PAP-2a, -2b, and -2c, with the -2b isoform appearing somewhat more sensitive to inhibition by this agent (Fig. 5C). DISCUSSION PAP-2c is a third member of the family of PAP-2 isoenzymes. This enzyme shares the putative transmembrane topology of the other PAP-2 isoenzymes, having six regions of predominantly hydrophobic amino acids linked by extramembrane regions. This proposed transmembrane structure is similar to that of the membrane-spanning portions of membrane-bound adenylylcyclases and transport proteins of the P-glycoprotein superfamily, which consist of a short N terminus and two transmembrane regions consisting of six hydrophobic spans that link globular cytoplasmic domains (28). Comparison of the sequences of hPAP-2a and hPAP-2b with previously recognized homologs identifies three regions of conserved sequence (denoted as A, B, and C in Fig. 1), which contain invariant amino acids that define a signature sequence motif shared among several proven or putative lipid phosphatases, the mammalian glucose 6-phosphatases, and some bacterial nonspecific and acid phosphatases. This group also contains yeast and Drosophila PAP-2 homologs (13)(14)(15)(16)(17)(18)(19)(20). These conserved regions lie predominantly within the hydrophilic regions of the proteins, and homologous sequences are also found in a soluble globular proteins that include bacterial acid phosphatase, mammalian glucose 6-phosphatase, and a fungal vanadium-dependent chloroperoxidase (29). The structure of this latter enzyme has been recently determined (30). The pentacoordinate vanadate cofactor resembles the transition state structure of phosphate. This suggests a two-step reaction in which a charge relay system involving conserved His and Asp residues in region C establishes a histidine-phosphate bond and the catalytic histidine residue in region B acts as an acid to cleave this intermediate, releasing the dephosphorylated substrate. This residue then acts as a base to facilitate nucleophilic attack on the phosphohistidine intermediate by a water molecule. Catalytically inactive PAP-2 mutants would be valuable tools for in vivo studies. We found that substitution of Ser 174 with alanine in PAP-2a produces a catalytically inactive protein. 2 This residue is postulated to form a hydrogen bond to the phosphate atom of the substrate, stabilizing the pentacoordinate transition state structure during catalysis.
Like PAP-2a and PAP-2b, PAP-2c is exclusively membranebound when expressed in mammalian or insect cells. As with many integral membrane proteins, purification of the type 2 PAP enzymes has been difficult. We made baculoviruses for expression of PAP-2a, -2b, and -2c and found that baculovirusinfected insect cells were an excellent model system for generating recombinant PAP enzymes. These cells exhibited extremely low levels of endogenous PAP activity. Infection with appropriate viruses resulted in dramatic increases in PAP activity. Membranes or detergent-extracted membrane proteins from these cells provided an abundant source of activity for kinetic studies. The Sf9 cell system should be effective for investigating the regulation of the type 2 PAP enzymes and may also provide an appropriate source for purification of recombinant proteins.
We used Triton X-100 mixed micelles to investigate the substrate specificity of PAP-2a, -2b, and -2c. The activity of all three enzymes was dependent on the surface concentration of substrates, suggesting that a surface dilution kinetic model was appropriate for analyzing the kinetic behavior of the enzymes in this model system. According to this model, which was originally developed for studies of phospholipase A 2 , enzyme activity depends on both bulk and surface concentrations of substrate, because enzyme binding to the micelle interface precedes substrate binding and catalysis (31,32). Since the PAP-2 enzymes are clearly integral membrane proteins, their association with detergent micelles would be expected to be effectively irreversible, so this type of analysis can be simplified to measurements of the dependence of enzyme activity on the surface concentration of substrate. The major limitation of this approach is that substrate or enzyme molecules can exchange between detergent micelles, and for highly active enzymes this rate of substrate exchange can be limiting (33). It is also possible that the detergent employed for these measurements may have selective affects on enzyme activity, although our finding that the enzymes displayed similar profiles of activity with 2 R. Roberts and A. J. Morris, unpublished results.     (11). Our results are in broad agreement with studies that used transiently transfected HEK293 cells as a source of enzyme activity to demonstrate that PAP-2a and -2b could hydrolyze PA, C1P, and S1P, although we did not observe the selectivity of PAP-2b for S1P reported by these investigators (13,14). The diversity of PAP-2 enzymes revealed by cDNA cloning was somewhat unexpected, since purification studies suggested that this enzyme behaved as a single activity from a number of different mammalian tissues (4 -8). PAP-2 activity has been studied extensively in rat liver, where a highly purified enzyme preparation has been reported to hydrolyze PA, C1P, and S1P with comparable avidity. The finding that each substrate inhibited the hydrolysis of the other in a competitive manner further supports the contention that this PAP-2 enzyme hydrolyzes all three substrates (11). The relationship of the rat liver PAP-2 enzyme to the cloned isoenzymes is not known, and this issue clearly requires further investigation. The S. cerevisiae genome contains four open reading frames encoding proteins with extensive homologies to the type 2 PAPs. Recent work has identified a member of one class as a PA and diacylglycerol pyrophosphate phosphatase and the members of the remaining class as S1P phosphatases (18 -20). PA and C1P did not inhibit the hydrolysis of S1P by these yeast enzymes (19), suggesting that they are highly specific for S1P. In comparison with mammalian PAP-2a, -2b, and -2c, the yeast S1P phosphatases have several unique structural features including an extended C terminus and several nonconservative substitutions of amino acid residues around the presumed active site. We have identified a cDNA encoding a fourth human type 2 PAP enzyme, which has a number of provocative homologies to the yeast S1Pases. 2 This may prove to be a mammalian S1P-selective PAP-2 isoenzyme.
Fractionation studies using rat hepatocytes indicate that PAP-2 activity is enriched in the plasma membrane, and recombinantly expressed mouse PAP-2a is also localized to this compartment in HEK 293 cells (4,13). Several cells have been reported to express a cell surface PAP activity (9, 10). We measured PAP activity in intact Sf9 cells expressing PAPs-2a, -2b, and -2c. The results obtained clearly indicate that when expressed in these cells, the active site of PAP-2a has effective access to extracellular substrates. The simplest explanation for these results is that this is an ectoenzyme that is oriented in the plasma membrane with its active site facing the extracellular space. Although we cannot rule out the possibility that a highly active "flippase" delivers PA substrate to this enzyme at an intracellular site, we consider this possibility unlikely because hydrolysis of PA by PAP-2a-expressing insect cells was rapid and proceeded without significant accumulation of radiolabeled substrate by the cells. PA, LPA, and S1P are receptoractive compounds, and it is probable that one function of PAP-2a is to terminate the signaling functions of these lipid agonists. Studies using fluorescent PA analogs reveal that dephosphorylation of PA to form DG precedes DG uptake into the cell. Cell surface PAP-2a may also provide a mechanism for formation of DG from extracellular substrates (34). Unfortunately, immunocytochemical analysis reveals that the HAtagged PAP-2a fails to exit the endoplasmic reticulum when expressed in HEK293 cells, so we have therefore not yet been able to examine cell surface PAP activity in mammalian cells expressing this enzyme. 3 As discussed above, the PAP-2 enzymes appear to be glycoproteins that share the same predicted transmembrane topology. Our results suggest that, in comparison with PAP-2a, PAP-2b and -2c are localized to different intracellular membrane compartments. Clearly, localization studies, preferably using PAP-2 isoenzyme-selective antibodies to study endogenous enzymes, are needed to further define the subcellular distribution of the PAP-2 isoenzymes. Rat PAP-2b is localized to the endoplasmic reticulum in cultured rat intestinal epithelial cells (15). A portion of type 1 PAP activity in rat liver co-localized with endoplasmic reticulum markers during cell fractionation studies (4,35). Despite this apparent co-localization, PAP-2b is clearly a type 2 PAP enzyme, and presumably the endoplasmic reticulum-localized PAP-1 activity reported in 3  rat liver represents the product of a separate gene. These findings are particularly interesting in light of the growing body of information that points to a role for PLD-generated PA in controlling vesicular transport between the endoplasmic reticulum and Golgi apparatus as well as the cis and trans compartments of the Golgi. PA has been proposed to play an essential role in recruitment of cytosolic coatomer complexes to the surfaces of the respective membrane compartments involved in these transport processes, which in turn initiates formation of coated transport vesicles (36 -38). Such a mechanism would require tight control of PA levels, and localization of PAP activity to this membrane compartment may therefore play an important role in this process.
Clearly, establishment of the roles played by the PAP-2 enzymes in cellular lipid metabolism remains an important priority. Our results suggest roles for the PAP-2 enzymes in the metabolism of both phospholipid-and sphingolipid-derived signaling molecules. Further work is required to establish the physiological role of these enzymes in metabolism of these bioactive lipid substrates. In this regard, it is noteworthy that PAP activities appear to act in series with PLD to generate and interconvert PA and diacylglycerol. PLD1 and PLD2, two recently identified mammalian PLD enzymes, are localized to different subcellular compartments in fibroblasts. PLD1 is found in a perinuclear compartment (most likely representing the endoplasmic reticulum or Golgi), while PLD2 is localized to the plasma membrane (39). These findings raise the possibility that the PAP-2 enzymes play selective roles in signaling processes controlled by these two PLDs, and future work will be directed toward investigating this possibility.