Purification and characterization of phosphatidylcholine phospholipase D from pig lung.

Phospholipase D, which mediates phosphatidylcholine hydrolysis in response to agonist stimulation, is an important component of signal transduction. We now report the purification of this enzyme to homogeneity from pig lung microsomes. The enzyme was solubilized with heptylthioglucoside and purified 2,200-fold by successive chromatography on sulfate-Cellulofine, ether-Toyopearl, chelate-Toyopearl, Q-Sepharose, heparin-Toyopearl, and hydroxyapatite. The final enzyme preparation gave a single protein band of M(r) = 190,000 on SDS-polyacrylamide gel electrophoresis. The enzyme hydrolyzed phosphatidylcholine but not lysophosphatidylcholine, phosphatidylethanolamine, and phosphatidylinositol. Optimum pH was 6.6. Half-maximal activity was obtained at 0.8 mM dipalmitoylglycerophosphocholine. The products were identified as phosphatidic acid and choline, but in the presence of ethanol, phosphatidylethanol was produced at the expense of phosphatidic acid. Ethanolamine and serine were not utilized as the phosphatidyl acceptor. Although not obligatory, Ca2+ and Mg2+ were stimulatory at high concentrations. The enzyme was markedly stimulated by unsaturated fatty acids in the presence of Mg2+ but not in its absence or by saturated fatty acids. N-Ethylmaleimide and detergents were inhibitory. Sucrose monolaurate had an aberrant effect on enzyme activity.

. DAG formation from PC causes a delayed, prolonged increase in DAG as compared with an early but transient increase in DAG produced by hydrolysis of inositol phospholipids (Wright et al., 1988;Motozaki and Williams, 1989;Nakashima et al., 1991;Leach et al., 1991;van Blitterswijk et al., 1991). Thus the agonist-induced stimulation of PLD elicits prolonged activation of specific isoforms of protein kinase C and thereby plays a role in long term cellular responses, such as proliferation and differentiation (Asaoka et al., 1992;. Furthermore, recent studies demonstrated that PA, the direct product of PLD, acts as an activator for protein kinase. Bocckino et al. (1991) demonstrated PA-dependent protein phosphorylation with rat tissue extracts and suggested the existence of a PA-dependent protein kinase(s). A DAG-independent isoform of protein kinase C, 6, was purified and shown to be activated by PA (Nakanishi and . More recently, Khan et al. (1994) partially purified and characterized a new protein kinase from human platelets, which was also activated by PA. The kinase was different from any of the currently identified protein kinase C isozymes since it did not cross-react with antibodies raised against them. The identification of PA-dependent protein kinases has suggested a new role for PLD in the protein kinase cascade of cell signaling.
Only limited information is available about the molecular properties of PLD because the enzyme has never been highly purified. Taki and Kanfer (1979) purified PLD 240-fold from freeze-dried rat brain after solubilization with Miranol H2M and cholate. The specific activity of their preparation was very low, 2 nmol min-' mg protein", partly because the activator of PLD was not known at that time. Their preparation utilized not only PC but also phosphatidylethanolamine (PE) with pH optimum at 6.0. Later, Chalifour and Kanfer (1982) identified unsaturated fatty acid as an activator of PLD in rat brain microsomes. Although the enzyme was located in the membrane fraction in rat tissues (Chalifour and Kanfer, 1982;Kobayashi and Kanfer, 1987) and Madin-Darby canine kidney cells (Huang et al., 19921, Wang et al. (1991) demonstrated that in bovine tissues a majority of enzyme activity was cytosolic. The cytosolic enzyme was purified 20-fold and shown to hydrolyze various phospholipids in the order of PE > PC > PI. The enzyme had a high K,,, whereas the membrane-bound enzyme was specific to PC with a low K,,,. The cytosolic and membranebound PLDs are believed to be different isoforms.
The present study was undertaken to obtain a highly purified preparation of membrane-bound PLD. We used pig lung as the enzyme source because lung contains the highest PLD activity (Chalifour and Kanfer, 1982;Kobayashi and Kanfer, 1987;Wang et al., 1991). "he purified PLD was a 190-kDa protein. It catalyzed not only hydrolysis but also transphosphatidylation and selectively utilized PC as substrate. Unsaturated fatty acids were a potent activator of the enzyme. Unlike base-exchange enzyme (Kanfer, 1972), PLD catalyzed transphosphatidylation in the absence of Ca". Fractions from Gigapite column chromatography were subjected to SDS-PAGE on a 10% gel under the reducing conditions, followed by silver staining. Lanes 13, neighboring fractions from Gigapite with activity peak a t lane 2. Lane 4, molecular mass markers (from top to hottom): myosin (205 kDa), 6-galactosidase (116 kDa), phosphorylase b (97.4 kDa), bovine serum albumin (66 kDa), egg albumin (45 kDa), and carbonic anhydrase (29 kDa).
PLD Assay-1,2-Dipalmitoyl-lmethyl-~H1GPC was mixed with 1,2dipalmitoyl-GPC, suspended in 0.05% Triton X-100 to 10 mM by sonication, and used as the substrate. The standard assay mixture contained 50 mM sodium dimethylglutarate, pH 6.6, 1 mM 1,2-dipalmitoyl-ImethyZ-:'H1GPC (3,700 dpm/nmol), 2 mu sodium oleate, 2 mM MgCl,, and enzyme in a total volume of 0.1 ml. Reaction was started by the addition of 1,2-dipalmotoyl-GPC, allowed to proceed for 20 or 40 min at 30 "C, and terminated by the addition of 3 ml of chlorofordmethanol (2:l) and 0.6 ml of saline. The mixture was shaken and centrifuged a t 600 x g for 5 min. The upper layer was quantitatively transferred to a tube, shaken with 2 ml of butyronitrile containing 30 mg/ml tetraphenylboron and 0.5 ml of 70 mM sodium phosphate, pH 7.2, and centrifuged a t 600 x g for 5 min. The upper phase containing the released choline was transferred to a vial, and radioactivity was counted in a toluene/ Triton X-100 scintillant with a Beckman LS7000 liquid scintillation spectrometer.
Purification of PLD from Pig Lung-All procedures were carried out a t 4 "C. Pig lung was minced and homogenized in 3 volumes of 0.25 M sucrose containing 5 mM Hepes-HC1, pH 7.2, 1 mM dithiothreitol, and 0.2 mu phenylmethylsulfonyl fluoride. The homogenate was centrifuged a t 1,000 x g for 10 min to remove cell debris and then a t 12,000 x g for 20 min. The supernatant was further centrifuged a t 100,000 x g for 60 min. The resultant pellet (microsomes) was suspended in a minimal volume of the homogenizing buffer and kept a t -80 "C until use.
Thawed microsomes were adjusted to 10 mg of proteidml with 10 mM Hepes-HC1, pH 7.2, containing 1 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride, and 0.02% NaN,. Following the addition of heptylthioglucoside to a final concentration of 2%, the mixture was gently stirred for 30 min and then centrifuged a t 100.000 x g for 60 min. The supernatant was carefully withdrawn, leaving the fluffy layer behind, and then applied to a sulfate-Cellulofine column (2.5 x 5 cm) equilibrated with buffer A (40 mhl potassium phosphate, pH 7.2, 0.06% sucrose monolaurate, and 0.02% NaN,) containing 0.5 M NaCI. After washing with the same buffer, enzyme was eluted with a 300-ml linear NaCl gradient (0.5-3.0 M) in buffer A. Active fractions were collected, brought to 1.6 M with solid ammonium sulfate, and then applied to an ether-Toyopearl column (1 x 13 cm) equilibrated with buffer A containing 1.6 M ammonium sulfate. The column was eluted with a reverse gradient of ammonium sulfate from 1.6 to 0 M in 120 ml of buffer A and subsequently with 30 ml of buffer A. Eluate containing activity was adjusted to pH 8.0 with 1 M &PO, and applied to a zinc-treated chelate-Toyopearl column (1 x 5 cm) equilibrated with buffer A that contained 10 my Tris-HC1, pH 8.0, instead of potassium phosphate buffer, and 0.5 M NaCI. Enzyme was eluted with a 60-ml linear histidine gradient (0-25 mbl) constructed in the same buffer, diluted with 2 volumes of buffer B (20 mhl Tris-HCI, pH 8.0,0.06% sucrose monolaurate, and 0.029 NaN,), and then adsorbed to a Q-Sepharose column (0.7 x 5 cm) equilibrated with buffer B containing 0.15 M NaCI. The column was eluted with a 30-ml linear NaCl gradient (0.15-1 RI) in buffer B. Active fractions were combined, diluted with 1 volume of buffer A, applied to a heparin-Toyopearl column (0.7 x 7 cm) equilibrated with buffer A containing 0.2 M NaCI, and then eluted with a 40-ml linear NaCl gradient (0.2-1.6 M) in buffer A. Eluted enzyme was adsorbed to a 0.8-ml Gigapite column equilibrated with buffer A. The loaded column was sequentially eluted with a 16-ml linear gradient ofpotassium phosphate (0.04-1 M ) in buffer A and then with 5 ml of 1 M potassium phosphate in buffer A to obtain the purified enzyme.
Identification of the Reaction Products-Enzyme was incubated with

1,2-dipalmitoyl-[methyl-:'H1GPC or l-palmitoyl-2-lr4Clpalmitoyl-GPC
under the standard assay conditions. The reaction mixture was mixed with 3 ml ofchloroform/methanol(2:1) and shaken with 0.6 ml of saline. When 1,2-dipalmit0yl-[methyl-~~H]GPC was used as substrate, the upper phase was saved for the analysis of the water-soluble product, concentrated, and separated by thin-layer chromatography (TLC) on a Silica Gel 60 plate (Merck, Darmstadt, Germany) with methanol, 0.6% NaCI, 30% ammonium hydroxide (20:20:1). The entire chromatogram was divided into equal segments, and each segment was counted. When l-palmit0yl-2-['~C1palrnitoyl-GPC was used as the substrate, the lower phase was isolated for the analysis of the lipid product and concentrated in a Vapor Mix evaporator (Toyo Rika, Tokyo, Japan). Radioactive lipids were separated by TLC with chlorofodmethanovacetic acid (13:3:1) and autoradiographed using a Fuji RX film.
Other Analytical Methods-SDS-polyacrylamide gel electrophoresis (SDS-PAGE) was carried out by the method of Laemmli (1970). Protein concentration was determined by the method of Bradford (1976) using the Bio-Rad protein assay kit.

RESULTS
Purification of PLD from Pig Lung-After testing a number of detergents, we found heptylthioglucoside (Shimamoto et al., 1985) to be most suitable for the solubilization of PLD from pig lung microsomes. When examined a t a fixed microsomal protein concentration of 10 mg/ml, this nonionic detergent was most effective a t 2% with a recovery of 70-100% PLD activity in the 100,000 x g supernatant. Octylglucoside, octylthioglucoside, and Mega-9 (n-nonaoyl-N-methylglucamide) gave less satisfactory results. Triton X-100, CHAPS, CHAPSO, Mega-8 (n-octanoyl-N-methylglucamide), and Mega-10 (n-decanoy1-Nmethylglucamide) yielded poor results. After solubilization with heptylthioglucoside, PLD was subjected to sequential column chromatographic procedures. To keep enzyme soluble and minimize nonspecific adsorption, we included a low concentration of sucrose monolaurate (0.06%) in all solutions used for column chromatography. Solubilized enzyme was first separated on a sulfate-Cellulofine column (Fig. LA 1. Over 99% of protein passed through the column while about 40% of the PLD activity was adsorbed to the column. PLD was eluted with a NaCl gradient, resulting in 64-fold purification with 32% yield. The eluted enzyme was applied to a hydrophobic column, ether-Toyopearl. The use of hydrophobic adsorbents with larger aliphatic or aromatic hydrocarbon chains resulted in low yields. The ammonium sulfate eluate was then loaded onto a chelate column that had been pretreated with ZnC1, (panel B ) . PLD could be eluted by a gradient increase in histidine concentration. Probably because of inhibition by contaminated zinc (see below), total activity was decreased to 15% but was restored considerably after Q-Sepharose chromatography. As the final step, we used hydroxyapatite (panel C). Overall, PLD was purified 2,200-fold from pig lung microsomes with a yield of 0.4%. Table I  and the radioactive lipids were separated by TLC. As shown in Fig. 2 A , the radioactive lipid produced in the absence of ethanol comigrated with authentic PA. In the presence of ethanol, however, the same fractions catalyzed the formation of phosphatidylethanol at the expense of PA. I t should be noted that phosphatidylethanol formation occurred in the absence of Ca2+ (see "Experimental Procedures"). These results provide conclusive evidence for the widely accepted view that mammalian PLD catalyzes not only hydrolysis but transphosphatidylation as well (Kobayashi and Kanfer, 1987;Gustavsson and Alling, 1987;Bocckino et al., 198713). Unlike transphosphatidylation catalyzed by base-exchange enzyme (Kanfer, 1972), ethanolamine and serine were not utilized as the phosphatidyl acceptor (data not shown).
To identify the water-soluble product, purified PLD was incubated with 1,2-dipalmitoyl-[rnethyl-3HlGPC, the upper layer of chlorofodmethanol extraction was separated by TLC, and the chromatogram was analyzed for radioactive product as described under "Experimental Procedures." Radioactivity comigrated with authentic choline, and no radioactivity was associated with phosphocholine and GPC (data not shown). Thus, choline was the sole water-soluble product. In the routine PLD assay, released choline was extracted from the upper layer of chlorofodmethanol extraction with butyronitrile in the presence of tetraphenyl-boron  and counted (see "Experimental Procedures"). Taken together, these results demonstrate that the purified enzyme catalyzed the hydrolysis of PC to PA and choline and, in the presence of ethanol, the formation of phosphatidylethanol. Properties of Purified PLD-The molecular mass of purified PLD was estimated to be 190 kDa by SDS-PAGE under the reducing conditions (Fig. v)). SDS-PAGE performed under the nonreducing conditions gave the same results (data not shown). PLD showed an anomalous behavior in gel filtration matrices, such as Sephacryl S-200, S-300, and S-400 (Pharmacia), Toyopearl HW65 (Toso), and Protein Pak G-300 (Waters, Milford, MA). Its elution was markedly retarded in these gels probably because of interaction with the gels. Thus it was difficult to determine the native molecular mass by gel filtration. As shown in Fig. 3, PLD showed a rather sharp pH profile with maximum activity at pH 6.6 in sodium dimethylglutarate buffer. A similar pH profile was obtained with sodium phosphate buffer. Activities measured in these two buffers were not significantly different. Similar pH optima were reported for partially purified rat brain PLD (Taki and Kanfer, 1979) and detergent-solubilized bovine lung PLD . Choline formation was proportional to the incubation time up t o 80 min under the standard assay conditions. Fig. 4 shows the effect of varying concentrations of dipalmitoyl-GPC on enzyme activity. PLD did not follow a normal saturation kinetics. The substrate dependence curve was sigmoidal as previously noted with rat brain synaptosomes (Kobayashi and Kanfer, 1987). Half-maximal activity was attained at 0.8 mM dipalmitoyl-GPC. This value was close to the K , value measured with partially purified rat brain enzyme, 0.75 mM (Taki and Kanfer, 1979).

Effects of Divalent Cations and Detergents-Ca2+
has been implicated in the activation of PLD (Augert et al. 1989;Huang et al., 1991;Kanaho et al., 1992). In HL-60 cells and neutrophils, Ca2+ was required for G protein-mediated activation of PLD Olson et al., 1991;Geny et al., 1993;Brown et al., 1993;Cockcroft et al., 1994). Ca2+ requirement was also demonstrated in the protein kinase C-dependent activation, for example, of CCL39 cell membrane PLD (Conricode et al., 1994). We were interested in examining whether Ca2+ had a direct effect on PLD enzyme. As shown in Fig. 5, purified enzyme exhibited no absolute requirement for Ca2+ and M e .
EDTA had no effect. Low Ca2+ and Mg2' had no significant effect, but higher Ca2+ and Mg2' increased PLD activity to a maximum of 2.5-and 1.7-fold at 1 and 2 mM, respectively. Zn2+ was inhibitory. Other divalent cations, such as Cu2+, Co2+, and Ni2+, were also inhibitory (data not shown). These results are fairly well consistent with the observations obtained with partially purified rat brain PLD and synaptosomal membrane PLD (Taki and Kanfer, 1979;Chalifa et al., 1990).
Several reports showed the stimulation of PLD activity by detergents (Chalifour and Kanfer, 1982;Kobayashi and Kanfer, 1987;Kanoh et al., 1991;Huang et al., 1992). We examined the effects of some detergents on purified enzyme. Although purified enzyme already contained 0.06% sucrose monolaurate, its final concentration in the assay mixture was calculated to be 0.012%, a concentration considerably lower than its critical micellar concentration, 0.021%. Thus the effect of the endoge- nous detergent was thought to be negligible. As shown in Fig. 6, sucrose monolaurate exerted an aberrant effect on PLD activity. PLD activity was inhibited by low concentrations of this detergent but restored considerably by higher detergent concentrations. Unlike sucrose monolaurate, Triton X-100 showed only inhibitory effect with little restoration of activity even at high concentrations. The exact mechanism for the paradoxical effects of sucrose monolaurate is not clear. Probably, this detergent had dual effects: l) "surface dilution effect," a decrease in the surface concentration of substrate at the water-lipid interphase by the addition of detergent (Deems et al., 1975;Hendrickson and Dennis, 1984) and 2) activation of PLD. The former would account for the enzyme inhibition at low detergent concentrations, and the latter would explain the partial restoration of enzyme activity at high detergent concentrations. N-Ethylmaleimide, a thioreactive reagent, inhibited PLD activity by 75% at 10 mM.
Effect of Fatty Acid and Complex Lipids-The stimulation of PLD by fatty acid has been studied with membrane preparations (Chalifour and Kanfer, 1982;Kobayashi and Kanfer, 1987;Chalifa et al., 1990;Siddiqui and Exton, 1992). But ambiguity remained as to whether or not fatty acid affected the enzyme itself. Fig. 7 shows the effect of oleic acid on purified PLD. Maximal activation occurred at a concentration of 2 mM.  The optimal oleate concentration varied somewhat with the concentration of PC and amount of enzyme protein present in the assay mixture. In Table I1 the effects of various fatty acids were compared at fixed concentrations of 2 mM. The data show that all the unsaturated fatty acids examined were effective, whereas all the saturated fatty acids were ineffective. Arachidonic acid was the most effective fatty acid. Interestingly, Mg2' was needed for maximal activation of PLD by fatty acid (Fig. 7).
The stimulatory effects of M$+ and Ca" (see above) required the presence of fatty acid (data not shown).
We next examined the effects of various complex lipids on PLD activity by adding the lipids at a concentration of 0.1 mM to the standard reaction mixture instead of oleic acid. Whereas lyso-PC, PE, PI, PA, and DAG were inhibitory to the enzyme, lyso-PA and PIP, were slightly stimulatory (about 20% increase in activity). This weak stimulatory effect of PIP, was unexpected because Brown et al. (1993) reported that HL-60 PLD was strongly stimulated by this phospholipid. This discrepancy may be caused by different isoforms of PLD present in HL-60 cells and lung or simply by different assay conditions. Phosphatidylserine had no effect.
Substrate Specificity-We examined the activity of purified PLD toward different phospholipids. Enzyme was incubated with 1 mM l-palmitoyl-2-[l-'4C]palmitoyl-GPC, l-a~yl-2-[1-'~C]arachidonoyl-sn-glycero-3-phosphoethanolamine, l-[l-l4C1palmitoyl-GPC, and 1,2-diacyl-sn-glycero-3-phospho[2-3H]inositol under the standard assay conditions, and we examined whether radioactive PA or lyso-PA could be formed from the former three phospholipids by TLC and autoradiography. When 1,2-diacyl-sn-glycero-3-phospho[2-JHlinositol was used as substrate, the incubation mixture was shaken with 3 ml of chlorofodmethanol(2:l) and 0.6 ml of saline, and the aqueous phase was used to examine whether or not radioactive inositol was formed. But the release of radioactive inositol could not be demonstrated (data not shown). As shown in Fig. 8, PA was liberated from PC but not from PE. Lyso-PA formation from lyso-PC could not be demonstrated. Thus PE, lyso-PC, and PI were not utilized by the enzyme. These results clearly show that the enzyme was specific to PC.

DISCUSSION
The present study has shown that PLD can be efficiently solubilized from lung microsomes with a nonionic detergent, heptylthioglucoside. The solubilized enzyme was purified to homogeneity by sequential chromatographic procedures in the presence of sucrose monolaurate. The molecular mass of the purified enzyme was considerably large, 190 kDa. A similar value, 200 kDa, was also reported for the molecular mass of rat brain PLD determined by gel filtration (Taki and Kanfer, 1979). This size is even larger than those of PI phospholipases C-p and "y known to be activated by G protein and tyrosine kinase, respectively (for reviews, see Rhee et al. (1989) and Rhee and Choi (1992)). Several factors have been implicated in the regulation of PLD activity, such as G protein, protein kinase C, and Ca2+. Although very little is known about the regulation of lung PLD, we expect that this 190-kDa sequence might contain a domain responsible for the control of enzyme activity by these factors. The specific activity of the present enzyme is 3.7 pmol min" mg", but this activity will be considerably increased when the regulatory mechanism for lung PLD activity is elucidated and reconstitution of the regulation is accomplished. Purified PLD obtained here will provide a suitable test system in such studies.
The major difference between Taki and Kanfer's enzyme preparation and the present one lies in the substrate specificity. The former utilizes both PC and PE, and PE gave even higher V, , , than PC. Their results are inconsistent with the recent ones obtained by Horwitz and Davis (1993), who showed that PC was the best substrate for PLD in rat brain microsomes by assaying phosphatidylbutanol synthesis using t3HIbutanol as labeled substrate. PE and phosphatidylserine were reported to give relatively small activities. In contrast, the present results are consistent with the finding of Honvitz and Davis (1993) and Wang et al. (1991) as well. The latter authors showed that octylglucoside-solubilized membrane PLD from bovine lung was specific to PC.
The hydrolysis of phospholipid other than phosphatidylinosi-to1 in response to agonist has been intensively studied. Several investigators showed that PE was also hydrolyzed together with PC in certain cell-agonist systems, such as NaF-stimulated mesangial cell microsomes (Harris and Bursten, 19921, hydrogen peroxide-or linoleic acid hydroperoxide-stimulated endothelial cells (Natarajan et al., 1993), and phorbol ester-or thrombin-stimulated amnion cells (Mizunuma et al., 1993). But the results of most studies point to PC as the principal phospholipid susceptible to agonist-induced hydrolytic reaction. PC has been identified as the major hydrolyzable phospholipid based on the analysis of the fatty acid composition of DAG (Grove and Schimmel, 1982) or PA (Bocckino et al., 1987a), identification of the selectively decreasing phospholipid (Daniel et al., 1986;Bocckino et al., 1987a;Osada et al., 19921, and analysis of the molecular species of DAG (Augert et al., 1989;Pessin et al., 1990;Leach et al., 1991) or phosphatidylethanol (Holbrook et al., 1992). These observations strongly support the view that the PLD purified here is the signal-transducing PLD. Apparently, the present PLD is not involved in the agonistinduced hydrolysis of PE seen in some tissues (see above), where a different isoform of PLD (for example, cytosolic PLD with broader substrate specificity)  or some other enzymatic mechanism (Guy and Murray, 1983) may play a role.
PLD activity on PI has been detected in neutrophils (Balsinde et al., 1989) and Madin-Darby canine kidney cells (Huang et al., 1992). The enzyme is Ca2+-dependent and cytosolic, thus different from the present PLD. These findings, together with the present results, suggest that there are several PLDs with different substrate specificities. The present enzyme should be adequately called phosphatidylcholine phospholipase D. An additional finding of some interest is the lack of activity toward lyso-PC in the purified enzyme. Recently, lyso-PA has attracted much attention as an intercellular signaling molecule (van Corven et al., 1989;van de Bend et al., 1992;van Corven et al., 1993). This phospholipid is thought to be synthesized from PA by the action of a specific phospholipase 4 in platelets (Billah et al., 1981;Gerrard and Robinson, 1989) and secreted in response to stimuli (Eichholtz et al., 1993).
Theoretically, sequential action of, first, phospholipase 4 and, second, PLD on PC could be considered as an alternative route for the synthesis of lyso-PA, but the present results seem to exclude this possibility.