Substrate Specificity of Neutral Phospholipase D from Rat Brain Studied by Selective Labeling of Endogenous Synaptic Membrane Phospholipids in Vitro*

We have designed a novel approach for studying the specificity of neutral phospholipase D from rat brain synaptic plasma membranes for endogenous phospholipid substrates in native membranes. A procedure was established that provides synaptic membranes labeled in selected phospholipids. This labeling procedure ex-ploits the presence of endogenous acyl-coenzyme A synthetase and acyl-coenzyme A:lysophospholipid acyltransferase in synaptosomes for acylating various lysophospholipid acceptors with radioactive fatty acid. With [‘Hlarachidonate for acylation and optimal concentrations of the respective lysophospholipids, membranes were labeled in either of the following phospholipids: phosphatidylcholine (93% of total label in phospholipids), 1-0-alkyl-phosphatidylcholine phosphatidylinositol phosphatidylethanolamine (85%), phosphatidylethanolamine-plasmalogen or phosphatidylserine These membranes were employed to study the substrate specificity of the neutral, oleate-activated rat brain phospholipase D. This phospholipase exhibited almost absolute specificity for the choline-phospholipids

We have designed a novel approach for studying the specificity of neutral phospholipase D from rat brain synaptic plasma membranes for endogenous phospholipid substrates in native membranes. A procedure was established that provides synaptic membranes labeled in selected phospholipids. This labeling procedure exploits the presence of endogenous acyl-coenzyme A synthetase and acyl-coenzyme A:lysophospholipid acyltransferase in synaptosomes for acylating various lysophospholipid acceptors with radioactive fatty acid. With ['Hlarachidonate for acylation and optimal concentrations of the respective lysophospholipids, membranes were labeled in either of the following phospholipids: phosphatidylcholine (93% of total label in phospholipids), 1-0-alkyl-phosphatidylcholine (87%), phosphatidylinositol (90%), phosphatidylethanolamine (85%), phosphatidylethanolamine-plasmalogen (81%) or phosphatidylserine (69%). These membranes were employed to study the substrate specificity of the neutral, oleate-activated rat brain phospholipase D. This phospholipase exhibited almost absolute specificity for the choline-phospholipids phosphatidylcholine and 1-0-alkyl-phosphatidylcholine: 0.34% of the former labeled substrate were transphosphatidylated to phosphatidylpropanol during the assay and 0.28% of the latter. Activity toward other phospholipids was barely detectable and could largely be accounted for by utilization of residual labeled phosphatidylcholine present in those preparations. The phospholipase D exhibited some preference for fatty acids in the C-2 position of phosphatidylcholine in the following order: 2-oleoylphosphatidylcholine (0.67% of this labeled phosphatidylcholine were converted to phosphatidylpropanol), 2-myristoyl-phosphatidylcholine (0.60%), 2-palmitoyl-phosphatidylcholine (0.46%) and 2-arachidonoylphosphatidylcholine (0.34%). The present approach of labeling membrane phospholipids in vitro could be useful in studies of phospholipase specificity as an alternative to the use of sonicated vesicles or mixed detergent-phospholipid micellar systems. *This work was supported in part by grants from the United States-Israel Binational Science Foundation, the Minerva Foundation, and the Irwin Green Research Fund in the Neurosciences. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ Recipient of a Minerva Fellowship. Receptor-mediated activation of mammalian phospholipase D (EC 3.1.4.4., PLD)' by various hormones, neurotransmitters, or growth factors has emerged as a novel putative signal transduction pathway (for review see Refs. 1-3). Whereas evidence for the signal-dependent stimulation of PLD activity in intact cells has accumulated rapidly, there is little information on the molecular and biochemical properties of the enzyme molecule itself. Nevertheless, current evidence suggests that mammalian PLD exists in several forms, which might be distinctly regulated and/or utilize different phospholipid substrates.
Determination of PLD substrate specificity in vitro is usually carried out by comparing various classes of exogenous phospholipids, added in the form of sonicated vesicles, as substrates. With this method, Saito and Kanfer (4) first identified mammalian PLD, an acid-active enzyme (pH optimum 6.0-6.5) from rat brain membranes which catalyzes the hydrolysis of phosphatidylcholine (PC) and phosphatidylethanolamine (PE), requiring unsaturated fatty acids or bile salts as activators (5). Other mammalian PLD activities have subsequently been reported to preferentially utilize the following substrates: platelet-activating factor and PC (6), PC (7, 8), PE (9, lo), N-acyl-PE (11,12), phosphatidylinositol (PI) (13-E ) , or PI-glycan (16)(17)(18). Recently, we identified and partially characterized in vitro a highly active neutral PLD (pH optimum 7.2) from rat brain synaptic plasma membranes (SPM) (19). This enzyme requires the presence of Na-oleate for activity in vitro and is stimulated by M$+. In that study (19), exogenous sonicated [3H]dipalmitoyl-PC was supplied as a substrate, but the substrate specificity of the enzyme was not further characterized.
The use of exogenous phospholipids in studies of phospholipase substrate specificity can be problematic, as different phospholipids aggregate in different forms according to their physicochemical properties when sonicated in aqueous solutions (20). The poorly defined physical state of the substrate may have considerable influence on the accessibility of the various phospholipids to the enzyme, and hence, on the apparent substrate preference exhibited by the enzyme. Mixed detergent-phospholipid micellar systems have been introduced to solve this problem (21-23). However, in such systems the substrate is not present in a bilayer and, moreover, such The abbreviations used are: PLD, phospholipase D; alkyl-PC, 1-0-alkyl-2-acyl-phosphatidylcholine; BSA, bovine serum albumin; Hepes, N-2-hydroxyethylpiperazine-N"2-ethanesulfonic acid; PA, phosphatidic acid; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PE-plasmalogen, 1-0-alkenyl-2-acyl-phosphatidylethanolamine; PG, phosphatidylglycerol; PI, phosphatidylinositol; PPr, phosphatidylpropanol; PS, phosphatidylserine; SPM, synaptic plasma membranes; TLC, thin layer chromatography; GTPrS, guanosine 5'-3-0-(thio)-triphosphate. systems are applicable only to soluble enzymes or solubilized membrane enzymes.

Materials
Isolation of Synaptic Plasma Membranes from Rat Brain-Male Wistar-derived rats (60 days old) were used throughout the study. Subcellular fractionation was performed according to Jones and Matus (26). Routinely, homogenate5 of six rat brains were pooled and fractionated. The SPM fraction was collected, sedimented at 100,000 X g for 30 min and resuspended in 6 ml of water. Aliquots were rapidly frozen in liquid nitrogen and stored at -135 "C.
Selective Labeling of Endogenous PC-The procedure is based on the activation of 3H-labeled fatty acid by a synaptosomal long chain acyl-CoA synthetase and its incorporation into membrane phospholipids by an endogenous acyl-CoAlysophospholipid acyltransferase in the presence of excess lysophospholipid acceptors. Labeling experiments were carried out at 37 "C for 15 min. The final volume/tube was 100 pl, routinely containing 50 mM Tris-HC1, pH 7.4, 0. After evaporation of the iodine and spraying with EN3HANCE, the plates were subjected to autofluorography. Bands were cut according to the autofluorogram, and the lipids were extracted with 1 ml of methanol/HCl (150:l). Radioactivity was quantitated in a Beckman LS 5000 TD @-counter after addition of 9 ml of scintillation mixture, Lumax/xylene (1:2). For measurement of oleoyl-CoA formation, plates were developed in butanol/acetic acid/water (6015: 25).
Selective Labeling of Synaptic Membranes for Subsequent PLD Assay-High specific activity in the phospholipids of choice was achieved by modifying the labeling conditions described for PC as follows. The final volume of the labeling reaction varied with the amount of SPM required for the subsequent PLD assay. No unlabeled fatty acid was included. Labeled fatty acids were added at 1 pCi/3O pg of membrane protein, regardless of their specific activity, yielding the following final concentrations for the different [3H]fatty acids: 1 p M for oleate, 0.05 p M for arachidonate, 0.17 p M for palmitate, and 0.25 p M for myristate. The incubation time of the labeling reaction was extended to 60 min. The labeling was terminated by transferring the tubes to an ice bath. To wash away excess lysophospholipid and labeled fatty acid, 2.5 ml of ice-cold 10 mg/ml BSA were immediately added, mixed carefully to avoid foam, and ice-cold water was added to give a final BSA concentration of 1 mg/ml. The membranes were then sedimented by centrifugation at 100,000 X g for 30 min. For removal of residual BSA, the pellet was resuspended in water by trituration and the centrifugation at 100,000 X g was repeated once without BSA. The membrane pellet was again resuspended in water and this time homogenized manually by 10 strokes in a glass-Teflon homogenizer. The protein concentration of the labeled SPM suspension was determined and the membranes were either used immediately or frozen in liquid nitrogen and stored at -135 "C. PLD lost less than 15% of its activity during the freezing/thawing steps.
PLD Assay with Endogenously Labeled SPM-Unless otherwise indicated, 30 pg of 3H-labeled SPM were incubated for 10 min at 37 "C in the presence of 50 mM Na-Hepes, pH 7.2, 1 mM MgC12, 100 mM 1-propanol, and 0.3 mM oleate, in a final volume of 120 ml. Termination and determination of the products PPr and PA were carried out as described before (19).
Protein Determination-Protein was determined by the method of Bradford (28) Table I

RESULTS
Selective Labeling of PC-The labeling conditions were developed from a procedure which we had utilized previously to investigate the action of neutral SPM-PLD on endogenous phospholipids in uitro (19). The phospholipid labeling obtained with that procedure resulted from acylation of intrinsic membrane lysophospholipids and was therefore rather nonspecific: 15% of the [3H]oleate incorporated into phospholipids, of that -46% labeled PC, 4% labeled PI, and the rest labeled unidentified lipids. Addition of exogenous lyso-PC to the nonselective labeling system significantly increased the incorporation of [3H]fatty acid into PC (Fig. lA). For acylation with [3H]oleate and [3H]arachidonate, the maximal labeling of PC was observed at lyso-PC concentrations of 50 and 100 PM, respectively (Fig. lA). The labeling of other phospholipids decreased concomitantly (not shown). A rate-limiting factor for the acylating enzymes was free fatty acid. In the presence of 50 PM lyso-PC, addition of up to 50 PM oleate (of which labeled oleate contributed 1 p~) further increased the acylation of lyso-PC to PC (Fig. 1B). Unlabeled oleate, at a concentration of 50 PM, was therefore included in the experiments performed to optimize the labeling assay. Lyso-PC and oleate were inhibitory at superoptimal concentrations, probably due to their detergent properties. In the labeling experiments, incubations were generally carried out for 15 min at 37 "C. Time course experiments showed the labeling reaction to progress linearly for at least 30 min (Fig. 2.4) be the rate-limiting step, since hardly any oleoyl-CoA accumulated under these conditions (Fig. 2 A ) . Protein concentrations up to 0.3 mg/ml were within the linear range (Fig. 2B). Optimal MgATP and coenzyme A concentrations were 2.5 mM and 25 p~, respectively (Fig. 3, A and B). Under overall optimal conditions, usually 90% of the [3H]oleate-derived radioactivity present in phospholipids was found in PC and the total radioactivity in PC was 20-fold higher than under the nonselective labeling conditions. Typically, the labeling activity was about 300 pmol/mg/min; the variation between membranes from different preparations was within the range of 150-450 pmol/mg/min. Labeling of Selected Phosphoglycerides-After establishing the conditions for the selective labeling of PC, we modified the assay in order to achieve maximally high specific radioactivity in any one chosen phosphoglyceride with minimal change in membrane phospholipid composition. Unlabeled exogenous oleate was omitted from the reaction (this did not alter the selectivity of labeling). idonate labeled all tested lyso-lipids more selectively than oleate (not shown). The following lysophospholipid concentrations yielded maximal specific labeling with arachidonate and were utilized in all later experiments: lyso-PC, l-alkyl-2lyso-PC, and lyso-Ps, 100 pM; lyso-PI, 50 pM; lyso-PE, 250 pM; lyso-PE-plasmalogen, 500 pM; and lyso-PG, 250 pM. The variations in optimal lysophospholipid concentrations and fatty acid efficiencies are likely to be due to the different physicochemical properties of the lyso-lipids, as well as the lyso-lipid/fatty acid preferences of the acyl-CoA transferases involved. The labeling selectivity achieved with [3H]arachidonate was highest for PC (93% of the total label in phospholipids was found in PC), followed by 90% for PI, 87% for alkyl-PC, 85% for PE, 81% for PE-plasmalogen, 71% for PG, and 59% for PS (Table I). -50% of the [3H]arachidonate was incorporated into the membranes and, of that, 98% incorporated into phospholipids. Neutral lipids and residual arachidonate that was not washed out during the termination of the labeling reaction accounted for the rest of the membranebound radioactivity. Utilizing radioactively labeled lyso-PC and lyso-PE we determined that -10% of the lysophospholipid added (7.5 p~ of lyso-PC or 20 pM of lyso-PE) remained in the membranes after the labeling procedure (data not shown). The additional lyso-PC increased the intrinsic lyso-PC level by about 15%. We have further examined the effect of lyso-PE-plasmalogen on PLD activity. No effects of the lysophospholipid were observed at concentrations up to 100 p~; higher concentrations inhibited the PLD activity, with an ICso of 1 mM (not shown). It might therefore be concluded that, although part of added lysophospholipids remains in the membranes after the labeling procedure, the resulting changes in the lyso-lipid content of the membranes are minor ones and do not effect the activity of neutral SPM-PLD toward endogenous substrates. Consequently, the labeling conditions established here provided us with synaptic membranes, labeled in vitro in selected phosphoglycerides, which could be assayed separately as potential substrates in a subsequent PLD assay.
PLD Assay with Selectively Labeled Endogenous PC-PLD activity was determined by measuring the production of PA   [3H]arachidonate (1 pCi/30 pg of membrane protein) and the respective lysophospholipids as detailed under "Experimental Procedures." Lysophospholipid concentrations employed were: lyso-PC, lysoalkyl-PC, and lyso-PS, 100 p~; lyso-PI, 50 p~; lyso-PE and lyso-PG, 250 p~; and Iyso-PE-plasmalogen, 500 PM. Results are expressed as mean k S.D. The numbers in parentheses represent the number of determinations. The total percentages slightly exceed 100% due to the unequal number of determinations. lated by oleate with a similar concentration response relationship (Fig. 4). Maximal PLD activity was observed at 0.3 mM Na-oleate, which was very similar to the values obtained previously with the nonselective labeling method (0.5 mM Naoleate), as opposed to 4 mM needed for activation when exogenous substrate was provided (19). This again suggested that the membrane microenvironment of PLD was not materially altered by the labeling procedure. PA is hydrolyzed by phosphatidic acid phosphohydrolase which is present in these membranes.* Consequently, in the following experiments we determined only the production of PPr, the transphosphatidylation product unique to PLD. Phosphoglyceride Specificity of SPM-PLD-We separately labeled batches of synaptic membranes in either PC, l-alkyl-PC, PS, PI, PE, PE-plasmalogen, or PG with [3H]arachidonate and the respective lyso-phosphoglycerides and performed PLD assays on these membranes. When net phosphatidylpropanol formation was expressed as percent of the selectively labeled substrate present in the sample, 0.34% of the labeled PC were converted to PPr within 10 min, 0.28% of alkyl-PC, 0.06% of PS, 0.02% of PI, 0.01% of PE, and 0% of PE-H. Mohn, unpublished observations. plasmalogen (Fig. 5A). The highest rate of conversion of noncholine phospholipids was observed in PS-labeled membranes which were also least specifically labeled, and where PC accounted for 14% of the label in phospholipids (Table I), This suggested that most of the PPr formed in the PS-labeled membranes was derived from PC. Assuming the PC conversion rate to be 0.34%, it was calculated that about 80% of PPr formed in PS-labeled membranes is indeed PC-derived. To further evaluate the contribution of PC to PPr formation, the latter was expressed as a percentage of the labeled PC present in each preparation. In PS-labeled membranes, 0.24% of the labeled PC were converted to PPr within 10 min; 0.33% of the PC were converted in PI-labeled membranes and 0.34% in PE-labeled membranes (Fig. 5B). Hence, the low PLD activity observed in SPM labeled in phospholipids other than PC or alkyl-PC most likely originates from residual labeled PC present in such preparations. (The low PLD activity ( 4 0 0 of Rat Brain Phospholipase D 11135 cpm) observed in such preparations explains the substantial variance between some experiments.) The conversion of PG was not measurable, since a labeled degradation product of PG comigrated with PPr on the TLC plate; this degradation product was not formed by PLD action, since it was also present in zero time controls. Altogether, these results show that SPM-PLD is highly specific for PC and alkyl-PC. Preference for Fatty Acids in C-2 Position of PC-Batches of synaptic membranes were labeled in PC with 1 pCi/30 pg of membrane protein of either [3H]arachidonate, [3H]palmitate, [3H]myristate, or [3H]oleate, in the presence of optimal lyso-PC concentrations, without adding unlabeled fatty acid. Optimal lyso-PC concentrations for selective PC labeling were 50 PM for labeling with oleate and 100 p~ for labeling with other fatty acids. PLD utilized the different C-2-fatty acylated PCs in the following order: 2-oleoyl-PC (0.67% of labeled 2oleoyl-PC were converted to PPr in 10 min) > P-myristoyl-PC (0.60%) > 2-palmitoyl-PC (0.46%) > 2-arachidonoyl-PC (0.34%) (Fig. 6). Although the differences were not statistically significant, the results seem to suggest that PLD prefers certain fatty acids in the C-2 position of the choline-containing substrate molecules. It should also be noted that the eggyolk lyso-PC utilized in these experiments as an acceptor is itself somewhat heterogenous, a fact that possibly masked differences that are due to the variation in the C-2 fatty acid.

DISCUSSION
Problems Associated with Determination of Phospholipase Substrate Specificity-Determination of the substrate specificity of lipolytic enzymes is not a trivial biochemical problem. Lipids aggregate in an aqueous environment in various forms, each according to its physicochemical properties and other prevailing conditions (20). Obviously, the lipid aggregate form would influence its susceptibility to enzymic hydrolysis by affecting parameters such as the effective surface concentration. This difficulty is compounded in the case of lipases which are integral membrane proteins, as fusion of substratecontaining and enzyme-containing vesicles may be rate limiting for catalysis. Such fusion is likely to be greatly affected by the nature of the lipid substrate, again influencing apparent rates of hydrolysis. One approach to this problem involves the use of a detergent-phospholipid mixed micellar system, which had been introduced by Dennis et al. tages of this approach are (i) in introducing the phospholipid substrate in a defined physical environment (the detergent micelle) and, (ii) in enabling quantitative kinetic analysis of enzyme-substrate interaction based on the phospholipid surface concentration. This approach had been utilized to analyze phospholipases such as phospholipase Az (21) and phosphatidic acid phosphohydrolase (23), as well as enzymes that require lipid cofactors, e.g. protein kinase C (22). However, one disadvantage of mixed micellar systems is that the phospholipid is not present in its physiological environment, i.e. the membrane phospholipid bilayer. In addition, the presence of a detergent may conceivably affect substrate-enzyme interaction and, hence, apparent substrate specificity. Finally, phospholipases which are integral membrane proteins must be solubilized prior to their analysis in a mixed micellar system; their removal from the bilayer, which involves exchange of the "annular" lipids with detergent molecules, is liable to affect their substrate specificity. We described here a novel alternative approach to that problem which is complementary to the use of mixed micelles. An in uitro procedure for selective labeling of endogenous phosphoglycerides in synaptic membranes was developed, thus providing a labeled substrate for neutral synaptosomal PLD within its native membranes. The feasibility of this approach stems from the presence of two enzymes, acyl-CoA synthetase and acyl-CoA:lyso-phospholipid acyltransferase, in synaptosomal membranes (24). These enzymes reacylate endogenous lysophospholipids normally produced by phospholipase A2 (29). Selectivity of phospholipid labeling was accomplished in the present study by incorporating excess of a chosen lysophospholipid into membranes and its dominance over intrinsic lysophospholipids (mostly lyso-PC) in the competition for fatty acid acylation by [3H]arachidonate. (It is noteworthy that our approach would have yielded still higher selectivity of labeling had we used radiolabeled lysophospholipids and acylated them with unlabeled fatty acids; however, labeled lysophospholipids other than lyso-alkyl-PC and lyso-PC are not commercially available to date.) The major advantage of this approach over previously described ones is in providing a nearly physiological milieu for phospholipasephospholipid interaction. The lysophospholipid reacylation enzymes are also present in microsomes from most tissues (29), which could therefore be labeled by a procedure similar to that which was used here for synaptic membranes. In addition to avoiding the problems associated with use of sonicated vesicles and mixed micellar substrates, the present approach may potentially be used to study enzyme activation by physiological mechanisms, e.g. receptors, G proteins, and protein kinases.
Substrate Specificity of Neutral Rat Brain SPM-PLD-The present approach was employed for studying the substrate specificity of neutral rat brain SPM-PLD, thus demonstrating its utility. We showed that neutral synaptosomal PLD is specific for the choline-phosphoglycerides diacyl-PC and alkyl-acyl-PC. The small PLD activity observed in PS-labeled membranes is most likely derived from residual labeled PC in these membranes. This could be deduced by expressing the PLD activity as percent of the labeled PC in those preparations. Thus, a nearly absolute specificity of this enzyme for choline-phosphoglycerides was demonstrated. In addition, suggestive evidence is provided for some specificity of SPM-PLD for the C-2 fatty acyl moiety of its PC substrate. These properties clearly distinguish the neutral rat brain PLD from certain other mammalian PLDs which utilize other phospholipids as substrate (see below). A similar preference for PC over PE was reported for the transphosphatidylation activity Substrate Specificity of Rat Brain Phospholipase D of the acid-active rat brain PLD by Chalifour et al. (30). Qian and Drewes (31) have studied PC hydrolysis by a neutral GTPyS-activated PLD from canine brain; the substrate specificity of this enzyme has not been examined, and its relationship to the neutral rat brain PLD discussed here is unknown. Numerous studies carried out in (3H]choline-labeled intact cells have indicated the existence of a PC-hydrolyzing PLD (for review, see Refs. 32 and 33); the neutral rat brain SPM-PLD activity assayed in uitro (Ref. 19 and the present study) is likely to represent one or more of these cellular enzymes.
Labeling intact cells with other phospholipid head groups has suggested the existence of PLDs which hydrolyze PE (9) and PI (34). This conclusion is borne out by in vitro studies, employing exogenous sonicated phospholipid substrates, which demonstrate the existence of PLDs that prefer noncholine-containing phospholipids. A PLD that hydrolyzes PE rather than PC was described in membranes isolated from NIH-3T3 cells (9). A PI-utilizing PLD was identified in neutrophil extracts (13,14) and human plasma (15). An Nacyl-PE-specific PLD was found in dog brain and rat heart (11,12). Recently, a cytosolic PLD utilizing PE, PC, and PI in that order of preference was demonstrated in various bovine tissues (10). Another soluble PLD, which specifically releases PI-glycan-anchored cell-surface proteins from the membrane, was purified from human plasma and bovine serum, cloned, and sequenced (16)(17)(18). As sonicated phospholipid vesicles were utilized in the above-mentioned studies, the substrate specificity of these PLDs, when it was examined, cannot be regarded as conclusive. The present approach of selective labeling of membrane phospholipids in uitro can be employed to elucidate the substrate specificity of these phospholipases and help establish their physiological significance.