Swiss 3T3 cells preferentially incorporate sn-2-arachidonoyl monoacylglycerol into sn-1-stearoyl-2-arachidonoyl phosphatidylinositol.

The sn-1-stearoyl-2-arachidonoyl phospholipids of animal cells appear to be formed by special mechanisms. To determine whether monoacylglycerol (MG) incorporation pathways are involved we incubated quiescent Swiss 3T3 cells with [3H]glycerol-labeled sn-2-arachidonoyl MG, then analyzed the radioactive cell lipids that accumulated. We also examined cell homogenates to identify enzyme activities that might promote the incorporation of sn-2-arachidonoyl MG into other cell lipids. The cell incubation experiments demonstrated rapid labeling of several lipids, including diacylglycerol, lysophosphatidic acid, phosphatidic acid, and phosphatidylinositol. They also demonstrated selective labeling of sn-1-stearoyl-2-arachidonoyl species of phosphatidylinositol, phosphatidylethanolamine, and phosphatidylserine. The cell homogenate experiments identified an sn-2-acyl MG acyltransferase activity, an MG kinase activity that phosphorylates sn-2-arachidonoyl MG in preference to sn-2-oleoyl MG, and a stearoyl-specific acyl transferase activity that converts sn-2-arachidonoyl lysophosphatidic acid into sn-1-stearoyl-2-arachidonoyl phosphatidic acid. The results also showed that this stearoyl transferase could act with other enzymes to convert sn-2-arachidonoyl lysophosphatidic acid into sn-1-stearoyl-2-arachidonoyl phosphatidylinositol. The combined results indicate that Swiss 3T3 cells incorporate sn-2-arachidonoyl MG into phospholipids by at least two different pathways, including one that specifically forms sn-1-stearoyl-2-arachidonoyl phosphatidylinositol.

The en-1-stearoyl-2-arachidonoyl phospholipids of animal cells appear to be formed by special mechanisms. To determine whether monoacylglycerol (MG) incorporation pathways are involved we incubated quiescent Swiss 3T3 cells with [3H]glycerol-labeled sn-2-arachidonoyl MG, then analyzed the radioactive cell lipids that accumulated. We also examined cell homogenates to identify enzyme activities that might promote the incorporation of sn-2-arachidonoyl MG into other cell lipids. The cell incubation experiments demonstrated rapid labeling of several lipids, including diacylglycerol, lysophosphatidic acid, phosphatidic acid, and phosphatidylinositol. They also demonstrated selective labeling of sn-1-stearoyl-2-arachidonoyl species of phosphatidylinositol, phosphatidylethanolamine, and phosphatidylserine. The cell homogenate experiments identified an sn-%-acyl MG acyltransferase activity, an MG kinase activity that phosphorylates sn-2-arachidonoyl MG in preference to sn-2-oleoyl MG, and a stearoyl-specific acyl transferase activity that converts sn-2-arachidonoyl lysophosphatidic acid into sn-1-stearoyl-2-arachidonoyl phosphatidic acid.
The results also showed that this stearoyl transferase could act with other enzymes to convert sn-2-arachidonoyl lysophosphatidic acid into sn-1-stearoyl-2-arachidonoyl phosphatidylinositol. The combined results indicate that Swiss 3T3 cells incorporate sn-2-arachidonoyl MG into phospholipids by at least two different pathways, including one that specifically forms sn-lstearoyl-2-arachidonoyl phosphatidylinositol.
Labeling experiments with [3H]glycerol and other radioactive precursors have shown that animal cells rapidly form phospholipids that contain palmitic acid and monoenoic, dienoic, or hexaenoic fatty acids, but only slowly form phospholipids that contain stearic acid and arachidonic acid (for reviews, see Refs. [1][2][3][4]. For example, this differential formation has been demonstrated for several different lipids in rat *This research was supported in part by United States Public Health Service Grant RR 00166. 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. $Present address: Publications Dept., Cold Spring Harbor Laboratory, P. 0. Box 100, Cold Spring Harbor, NY 11724. §To whom correspondence should be addressed. liver, including phosphatidylcholine (PC),' phosphatidylethanolamine (PE), phosphatidic acid (PA), diacylglycerol (DG), and phosphatidylinositol (PI) (5-8). To account for these results it has been proposed that sn-1-stearoyl-2-arachidonoyl phospholipids may be formed by a multistep pathway. First, phospholipids that contain palmitic acid in the sn-1 position and a monoenoic, dienoic, or hexaenoic fatty acid in the sn-2 position are formed de nouo by the classical pathways of phospholipid biosynthesis. Then these phospholipids are "remodeled" by reactions that selectively replace the sn-1 palmitoyl group with a stearoyl group and replace the sn-2 fatty acyl group with an arachidonoyl group. In support of this possibility, phospholipase A and acyltransferase activities that might catalyze this type of phospholipid "remodeling" have been identified (see, for example, [9][10][11][12]. Nevertheless, alternative possibilities remain to be excluded. For example, one could imagine that sn-2-arachidonoyl monoacylglycerol (MG), formed from intracellular triacylglycerol (TG) by the action of a TG lipase or formed from sn-1-palmitoyl-2-arachidonoyl PC by the successive action of a phospholipase C and a neutral lipase, might be "recycled" into phospholipids by a stearoyl-specific pathway. We addressed this possibility in the study of quiescent Swiss 3T3 cells that is described below. A preliminary report of some of the results has appeared (13). A stock solution of [3H]MG, stored in toluene at -20 "C, comprised 94% of the sn-2-isomer, as determined by TLC on a Silica Gel-G plate that had been sprayed with 5% boric acid (w/v), dried, and developed in chloroform/acetone/methanol(96:4:2, v/v). For addition to cells the ["HJMG was dried under argon and resuspended (25 pCi/ 50 pl) in a 2% (w/v) solution of bovine serum albumin (BSA) in water by sonication for 3 min. under argon at room temperature. In the BSA suspension, about 20% of the [3H]MG was converted to the sn-1-isomer in 2 h at 4 "C. However, when the [3H]MG was added to cell culture medium at 37 "C, the rate of isomerization was very rapid, with approximately 50% conversion of the sn-2-isomer to the sn-land sn-3-isomers in 12 min. Furthermore, 22-48% of the labeled MG that was taken up by the cells was hydrolyzed to release free ['HI glycerol. The excess of unlabeled glycerol that was included in the medium diluted the specific activity of this free [3H]glycerol.

EXPERIMENTAL PROCEDURES
Cell Harvesting and Lipid Extraction-At appropriate time intervals after labeling, the medium from each dish was decanted, and the cells were frozen at -70 "C in 1.5 ml of 0.15 M HCl in phosphatebuffered saline. Lipids were extracted from the cells with the method of Bligh and Dyer (15). All solvent mixtures contained 0.005% butylated hydroxytoluene.
Lipid Analysis-After extraction the lipids were dried under argon and the individual classes were separated by two-dimensional TLC (16). PI and PS were subsequently resolved on ammonium sulfateimpregnated Silica Gel-H with chloroform/methanol/acetic acid/ water (50:30:8:4, v/v) (17). Neutral lipids were separated on Silica Gel-H with hexane/diethylether/methanol/acetic acid (9020:3:2, v/ v) (18). All chromatography steps were performed under argon. Lipid zones were visualized with 0.1% (w/v) 2,7-dichlorofluorescein in 95% (v/v) methanol under UV light. Lipids were eluted from silica with chloroform/methanol/water (75:25:2, v/v) andchloroform/methanol/ acetic acid/water (50:39:1:10, v/v). Cell lysophosphatidic acid (lysoPA) was analyzed with a special approach. It was extracted from cells by the method of Bligh and Dyer, but with 0.4% trifluoroacetic acid rather than HCI in the second wash. It was then dried down and separated in two successive TLC steps with 1) the first dimension of the two-dimensional method referred to above, and 2) the second dimension of this method applied to aluminum-backed Silica Gel-60 TLC plates. The lysoPA was eluted in chloroform/methanol/acetic acid/water (59:39:1:5, v/v), then dried for counting as with other samples. The overall recovery of lysoPA was approximately 65%.
Molecular species of PI, PA, PS, PE, PC, and DG were analyzed by reverse-phase high performance liquid chromatography (HPLC) according to the method of Nakagawa and Horrocks (19). A model "45 chromatograph (Waters, Milford, MA) and an Ultrasphere-ODS column (Beckman Instruments) were used. Molecular species analysis of PA formed in incubation experiments with cell membranes was performed by the method of Nakagawa and Waku (20). To verify the positional distribution of sn-1-acyl groups, PC, PE, and PI that had been labeled during a 1-h cell incubation experiment with radioactive sn-2-arachidonoyl MG were isolated as described above, then hydrolyzed with phospholipase D for 15 min (PC or PE) or 4 h (PI) at room temperature (21). The resulting PAS were purified by TLC and methylated with diazomethane (20). The dimethylated PAS were hydrolyzed with TG lipase for 1 min at 37 "C to generate sn-2-acylglycerodimethylphosphate. The reaction was quenched with chloroform/methanol (1:2, v/v), and the products were extracted by the method of Bligh and Dyer (15). The resulting sn-2-acyl-glycerodimethylphosphate was purified on TLC plates with chloroform/methanol (85:15, v/v), then acetylated with acetic anhydride/pyridine (5:1, v/v) for 3 h at room temperature. The molecular species of sn-lacetyl-2-acyl dimethyl PA were separated by reverse-phase HPLC with acetonitrile/isopropyl alcohol/methanol/water (45:25:15:15, v/ v) at a flow rate of 1.0 ml/min. Fractions were dried and counted as previously described. Several synthetic phospholipids were treated by the same method and used as standards for the HPLC analysis. Each yielded only one peak, which corresponded to that predicted on the basis of the positional specificity of the lipase reaction (22).
Statistical tests were performed with the Prophet computer system (Bold, Beranek, and Newman, Cambridge, MA). Means and standard deviations reported in the tables and shown on the graphs were calculated for replicate assays from original Prophet data tables using the "compute stats" command available in the Prophet applications package (23).
Assay of MG Acyltransferase Activity-MG acyltransferase was determined by measuring the formation of DG from sn-2-arachidonoyl MG and [l-'4C]palmitoyl-CoA according to the method of Coleman and Haynes (24). Assays were performed in the presence or absence of exogenous sn-2-arachidonoyl MG. In order to calculate the specific activity of MG acyltransferase, radioactivity that accumulated in DG and TG in the absence of added substrate was subtracted from that which accumulated in DG and TG in the presence of added substrate. The corrected TG counts were divided by 2 and added to the corrected DG counts. Enzyme activity was expressed as picomoles of [14C]palmitate incorporated into DG/milligram of protein/minute. (With this assay protocol we were able to obtain specific activities similar to those of Coleman and Haynes for MG acyltransferase in a liver microsomal preparation from 6-day-old rats.) Assay of MG Kinase Activity-MG kinase activity was determined by measuring the formation of 32P-labeled lysoPA from [y-"PIATP and unlabeled sn-2-arachidonoyl MG. The latter was prepared as follows: sn-1-stearoyl-2-arachidonoyl DG was suspended in 2% (w/v) BSA in 50 mM Tris HCl, pH 7.9, by sonication for 3 min under argon at room temperature, then hydrolyzed for 2.5 min at 37 "C with TG lipase from R. arrhizus (2.4 X lo4 unit/lO mg of substrate). The reaction was quenched with chloroform/methanol (1:2, v/v). sn-2-Arachidonoyl MG was separated from undigested DG and free stearic acid with a silica SEP-PAK cartridge (Waters). Lipid extracts were dried under argon and redissolved in toluene/hexane (l:l, v/v) (solvent A) for loading onto the cartridge. Stearic acid and DG were eluted with a mixture of 75% solvent B in solvent A (where solvent B was toluene/ethylacetate (3:1, v/v) containing 1.2% (v/v) formic acid). sn-2-Arachidonoyl MG was eluted with chloroform/methanol (1:2, v/v), dried under argon, and redissolved in chloroform for storage. The overall yield of sn-2-arachidonoyl MG from this procedure was -30%. The final product contained <5% of the sn-1-isomer.
Optimal MG kinase activity was attained in a reaction mixture (35 pl) containing 20 pg of sn-2-arachidonoyl MG and 12 pg of PS (dried under argon, then resuspended in 2% BSA (w/v) in MOPS, pH 7.25, by sonication for 3 min under argon at room temperature), 150 mM MOPS, pH 7.25, 3 mM DTT, 54 mM MgC12, 0.5 mM [y-"PIATP (700-1200 cpm/pmol and 5-15 pg of cell protein. Assays were performed at room temperature for 4 or 5 min. Under these conditions, about 80% of the MG substrate was maintained as the sn-2-isomer. The assays were quenched with 0.6 ml of chloroform/methanol/HCl (66:33:1, v/v), carrier PA and lysoPA were added, and the phases were separated by addition of 0.5 ml of chloroform/methanol/water (3:48:47, v/v). The lower phase was dried under argon or in a Speed Vac (Savant Instruments, Farmingdale, NY), and the reaction products, 32P-labeled lysoPA and 3'P-labeled PA, were resolved by TLC on plastic-backed Silica Gel-60 plates with chloroform/acetone/methanol/acetic acid/water (30:40:10:10:5, v/v). The lipids were visualized by exposure to I? vapor and by autoradiography. Enzyme-specific activity was expressed as picomoles of '*P incorporated into product/ milligram of protein/minute. Protein concentrations were measured with the methods of Bradford (25) and Lowry et al. (26).
Assay of DG Kinase Activity-DG kinase was determined at room temperature, in a final volume of 35 pl, by a modification of a previously reported method (27). The reaction mixtures contained 50 mM MOPS, pH 7.25, 1 mM DTT, 18 mM MgCl,, 73 mM octyl glucoside, 3.3 mM PS, 2 mM substrate DG, and 0.5 mM [y-"P]ATP (700-1200 cpm/pmol). 5-20 pg of protein were used for each assay. After 2 min, the reaction was quenched and the reaction mixture was treated as described in the assay protocol for MG kinase.
Assay of LysoPA Acyltransferase Activity-LysoPA acyltransferase activity was determined by measuring the formation of "'P-labeled P A from "'P-labeled sn-2-arachidonoyl lysoPA. The latter was prepared as follows: sn-2-arachidonoyl MG was suspended with PS in 2% (w/v) BSA in 375 mM MOPS, pH 7.2, by sonication for 3 min under argon at room temperature. The MG was phosphorylated in a reaction mixture (140 pl 4 min and quenched with 2.4 ml of chloroform/methanol/HCl (66:33:1, v/v). Then 2 ml of chloroform/methanol/water (3:48:47, v/ v) were added, the phases separated, and the lower phase dried under argon. LysoPA in the extracted material was purified on a silicabased cation exchange column (Bond Elut-PRS cartridge column, Analytichem International, Harbor City, CA). The extract was loaded onto the column in chloroform; neutral lipid and a small amount of P A were removed by washing the column with 8 ml each of chloroform/methanol (982, v/v) and chloroform/methanol (97. 5:2.5, v/v). LysoPA was then eluted with 7 ml of chloroform/methanol (95:5, v/ v). The purity of lysoPA prepared by this method was determined to be >95%.
Membrane-associated enzyme activity was measured as follows. The '"P-labeled sn-2-arachidonoyl lysoPA in a polypropylene tube was dried down under argon, then resuspended at an apparent concentration of 10 p~ in 100 mM Tris-HC1 buffer, pH 7.5, containing 2% BSA. The reaction mixture (50 pl) containing 50 pmol 32P-labeled sn-2-arachidonoyl lysoPA, 1% BSA, 50 mM NaF, 2 mM Na3V04, 2.5 mM DTT, 0.5 mM EDTA, and a 10-pl suspension of Swiss 3T3 cell membranes was incubated for up to 30 min at 37 "C. The reaction was stopped by addition of 0.9 ml of chloroform/methanol/HCl (66:33:1, v/v). After addition of carrier PA and lysoPA the lipids were extracted by addition of 0.75 ml of chloroform/methanol/water (3/ 48/47, v/v). The lower phase was dried under argon, and the PA was separated from lysoPA by TLC as described for the MG kinase assay.

RESULTS
Cell Incubation Experiments with (3H]MG-To investigate the ability of quiescent Swiss 3T3 cells to incorporate sn-2arachidonoyl MG into PI and other phospholipids, we incubated the cells with exogenous [3H]MG in the presence of a large excess of unlabeled glycerol, then measured the radioactivity that accumulated in several lipid classes. Under the conditions of our experiments ("Experimental Procedures"), the cells incorporated about 5% of the added t3H]MG into cellular lipids (other than MG) during a 1-h incubation. By contrast, cells that were incubated with free [3H]glycerol of the same specific activity (in the absence of unlabeled free glycerol) incorporated only 0.1-0.2% of the label into lipids (not shown). The cells' ability to incorporate [3H]MG into lipids was noteworthy because the MG isomerized rapidly in the incubation medium and was partially hydrolyzed in the presence of cells (see "Experimental Procedures").
A comparison of the cell lipids that were labeled with [3H] MG with those that were labeled with free [3H]glycerol revealed a major difference in the relative distribution of radioactivity (Fig. 1). The PI, PE, and PC of the [3H]MG-labeled cells respectively accounted for 35, 22, and 25% of the combined radioactivity in eight different lipid classes, whereas the values for the corresponding lipids of the [3H]glycerol-labeled cells were 3, 12, and 63%. These results provided evidence that the cells preferentially incorporated the [3H]MG into PI and PE.
In further studies of the cells' ability to incorporate this MG into lipids, we incubated the cells with [3H]MG for 0,0.5, and 1 h, then changed the medium on some of the cells that had been labeled for 1 h to medium that contained no label, and continued the incubation for an additional 24 h. Upon measuring the radioactivity in the cells that had been exposed to ['HIMG for no more than a few seconds, i.e. after 0 h of incubation, we observed significant labeling of DG, TG, lysoPA, PA, and PI (  lysoPA were labeled at this early time point raised the possibility that the cells might have contained two different MG incorporation pathways, one initiated by an MG acyltransferase and one initiated by an MG kinase.
Very little PS was labeled during the 1 h of incubation with ['HIMG, but the labeling of PS increased 8-fold during the subsequent 24 h chase (Fig. 2). The net increase in radioactive PS was comparable with the net decrease in radioactive PE, which suggested that the PS might have been formed by a head group exchange reaction involving PE (8,9). Analysis of the distribution of radioactivity among molecular species of the lipids that were labeled during the time course experiments (see "Experimental Procedures") demonstrated that most of the radioactivity in each lipid class comigrated with arachidonic acid-containing species (Table  11). Furthermore, when radioactive PI, PE, and PC from cells that had been incubated for 1 h with [3H]MG ("Experimental Procedures"), were separately analyzed by a procedure that involved mild treatment with TG lipase, 82-98% of the activity recovered in the phospholipid products was found to comigrate with arachidonic acid-containing species. This provided evidence that only sn-2-arachidonoyl MG had been incorporated into these phospholipids, although other isomers of the labeled MG were clearly present in the cell incubation medium ("Experimental Procedures").
The most noteworthy result of the molecular species analysis was that the distribution of radioactivity among individual arachidonic acid-containing species differed in the different lipid classes. In PI the sn-1-stearoyl-2-arachidonoyl species accounted for about 90% of the recovered radioactivity at each time point. This was the highest proportion of radioactivity observed in this species in any lipid class (Table 11). Somewhat lower proportions of radioactive sn-l-stearoyl-2arachidonoyl species were found in PS and PE, while the lowest proportions were found in PC and DG, where sn-lstearoyl-2-arachidonoyl species accounted for only about 40% of the total radioactivity. The remainder of the radioactivity in PC and DG was mainly present in sn-1-palmitoyl-2-arachidonoyl and sn-1-myristoyl-2-arachidonoyl species. A similar distribution of radioactivity among the three arachidonic acidcontaining species was found in both lipid classes. These results provided evidence for the presence of MG incorporation pathways of differing acyl chain specificity.
Experiments with Cell Homogenates-To identify enzymes that might have contributed to these pathways, we first measured the activity of MG acyltransferase, an enzyme that has been studied extensively in the intestinal mucosa and liver (see, for example, Ref. 30). We found that the cells contained a low level of MG acyltransferase activity, which was mainly associated with membranes (Table 111). A comparison of the enzyme's activities toward sn-2-arachidonoyl MG and sn-2oleoyl MG provided little evidence of enzyme specificity. In three experiments the mean ratio of the enzyme's activity toward the two substrates (arachidonoyl MG/oleoyl MG) was 0.8. Furthermore, both substrates were converted into TG (not shown), presumably by the coupled action of the MG acyltransferase and a DG acyltransferase. This suggested that the MG acyltransferase might have initiated the incorporation of [3H]MG into DG and TG in the incubation experiments with intact cells.
Because we had also observed rapid labeling of lysoPA in these experiments we next measured the MG kinase activity of the cell homogenates. The cells did indeed contain this type of activity, and more than 90% of it was associated with the cell membranes, as had been found for the MG acyltransferase. A comparison of the membrane-associated enzyme's activities toward sn-2-arachidonoyl MG and sn-2-oleoyl MG in several experiments demonstrated a 2-3-fold preference for the arachidonic acid-containing substrate (data not shown).
The membranes of Swiss 3T3 cells also contain a DG kinase that preferentially phosphorylates arachidonic acid-containing substrates (27). T o examine the relation between this enzyme and the MG kinase, we extracted cell membranes with 0.3 M KCl, then measured the activities of the two enzymes in the extract and residue. The KC1 extract and the residue each contained about one-half of the recovered MG kinase activity, and in each case the MG kinase showed an approximately 2-fold preference for sn-2-arachidonoyl MG as compared with sn-2-oleoyl MG (Table IV). In contrast, the residue contained essentially all of the arachidonoyl-specific DG kinase activity, and this activity showed a 10-11-fold preference for sn-1-stearoyl-2-arachidonoyl DG as compared with sn-1-stearoyl-2-oleoyl DG (Table IV). These results showed that the MG kinase activity differed from the arachidonoyl-specific DG kinase activity.
To search for an enzyme activity that could catalyze the conversion of sn-2-arachidonoyl lysoPA into sn-l-stearoyl-2arachidonoyl PA, we incubated cell homogenate fractions with 32P-labeled sn-2-arachidonoyl lysoPA in the presence or absence of potential cofactors. The cell membranes converted the radioactive sn-2-arachidonoyl IysoPA into PA in the absence of added cofactors, whereas the high speed supernatant fraction showed little or no activity (Fig. 3A). Moreover, the membranes formed even more PA when they were incubated with 32P-labeled lysoPA in the presence of added CoA and/or ATP (Fig. 3B). However, the distribution of radioactivity among molecular species of the PA differed under these different incubation conditions. Selective formation of sn-lstearoyl-2-arachidonoyl PA occurred in the absence of added cofactors or in the presence of added GOA (Fig. 4, A and B ) . We confirmed the identity of this species of PA by hydrolyzing the PA with lipase and analyzing an sn-1-acetyl-%acyl derivative of the resulting lysoPA by HPLC (data not shown).

TABLE I1 Distribution of radioactivity in molecular species of lipids from 3T3 cells labeled with ['HIMG
Quiescent 3T3 cells were incubated with ['HIMG (25 pCi/dish in the presence of 7000-fold excess of unlabeled glycerol) for 0, 0.5, and 1 h. The medium on some of the cells that had been labeled for 1 h was subsequently changed to one containing no radiolabeled precursor, and the cells were incubated for a further 24 h. After the appropriate pulse or chase times, cells were harvested and the lipids extracted and analyzed as described under "Experimental Procedures." Molecular species of the individual lipid classes were resolved by reverse-phase HPLC. The percentage distribution of radioactivity in the species was calculated from the total radioactivity recovered in arachidonoyl and non-arachidonoyl species. The data represent the means f standard deviation of four replicates within a single experiment. Similar results were obtained in three separate experiments, though in one of the experiments somewhat lower percentages were obtained for sn-1-stearoyl-2-arachidonoyl PA. 140-204, sn-lmyristol-2-arachidonoyl phospholipid species; 16:O-20:4, sn-1-palmitoyl-2-arachidonoylphospholipid species; 180-20:4, sn-1-stearoyl-2-arachidonoyl phospholipid species.  DG k i m e activities in KC1-extracted Swiss 3T3 cell membranes 3T3 cell total membrane fraction was isolated, extracted with 0.3 M KC1 and subfractionated into supernatant and pellet fractions. All fractions were adjusted to the same volume. The MG kinase activity toward sn-2arachidonoyl MG or sn-2-oleoyl MG was measured in BSA, and the DG kinase activity toward sn-l-stearoyl- 2arachidonoyl (18:O-20:4) DG or sn-l-stearoyl-2-oleoyl(180-181) DG was measured in octyl glucoside. The assays were performed as described under "Experimental Procedures." Similar results were obtained in two other exDeriments. The activity of the CoA-dependent stearoyl-specific enzyme could be coupled with that of other membrane-associated enzymes to promote incorporation of 32P-labeled sn-2-arachidonoyl lysoPA into PI (Table V). Furthermore, analysis of the PI that was formed ("Experimental Procedures") in the presence or absence of added CoA demonstrated that all of the radioactivity was present in the sn-1-stearoyl-2-arachidonoyl species (data not shown).

DISCUSSION
The results of this study show that quiescent Swiss 3T3 cells incorporate radioactive sn-2-arachidonoyl MG into several cell lipids including sn-1-stearoyl-2-arachidonoyl PI. Furthermore, they suggest that two or more MG incorporation pathways are involved and that the pathway that forms PI is initiated by the successive action of an MG kinase and a CoAdependent stearoyl transferase. The evidence for the participation of these enzymes in P I formation can be summarized as follows. 1) When quiescent Swiss 3T3 were incubated for a few seconds with radioactive sn-2-arachidonoyl MG, significant amounts of radioactive lysoPA, sn-1-stearoyl-2-arachidonoyl PA, and sn-1-stearoyl-2-arachidonoyl P I accumulated (Table I).
2) The sn-1-stearoyl-2-arachidonoyl species of P I accounted for about 90% of the radioactivity that was present in this phospholipid class at all incubation times. 3) Examination of cell homogenates revealed the presence of an MG kinase activity that phosphorylated sn-2-arachidonoyl MG in preference to sn-1-oleoyl MG (Table IV). 4) When cell membrane preparations were incubated with 32P-labeled sn-2arachidonoyl lysoPA in the presence or absence of added CoA, radioactive sn-1-stearoyl-2-arachidonoyl PA was formed selectively (Fig. 4). 5) The membrane preparations apparently converted this PA into sn-1-stearoyl-2-arachidonoyl PI in the presence of CTP and inositol (Table V), although we cannot exclude the possibility that an sn-2-arachidonoyl CDP-MG intermediate may have been involved. It is interesting to compare those results with those of Nakagawa and colleagues (31), who incubated alveolar macrophage microsomes in the presence of ['4C]glycerol 3-phosphate, CTP, and inositol and found that only 2-3% of the label that was converted into P I was present in the sn-1-stearoyl-2-arachidonoyl species. It appears that the results of Nakagawa and colleagues reflect the de nouo pathway of P I synthesis.
The MG kinase-initiated pathway appears to branch after the first step because we found that the membranes of Swiss 3T3 cells contained both a CoA-dependent stearoyl transferase and an ATP-dependent acyltransferase that could catalyze the conversion of 32P-labeled sn-2-arachidonoyl lysoPA into PA. The ATP-dependent acyltransferase reaction formed a mixture of sn-1-palmitoyl-2-arachidonoyl and sn-l-stearoyl-2-arachidonoyl PA (Fig. 4), which suggests that the acyl-CoAdependent acyltransferase that we found in detergent extracts of the membranes was probably involved.
The fate of the PA formed by the ATP-dependent acyltransferase in intact cells remains to be determined, but one possibility is that the PA may be converted (via DG) into PE. Formation of P E by this branch of the MG kinase-initiated pathway might account for the preferential labeling of P E that occurred in the intact cell incubation experiments ( mM ATP + 4 mM magnesium acetate; D, with CoA + ATP + magnesium acetate. The PA that was formed was methylated and analyzed by reverse-phase HPLC as described under "Experimental Procedures." In the case of the incubation performed in the absence of cofactors (panel A ) the PA from two separate incubation replicates was pooled in order to obtain a sufficient amount of radioactivity for HPLC analysis. As a blank, a parallel incubation was done without a source of enzyme, and the radioactivity in the effluent from the HPLC column was determined. The values obtained were then subtracted from those obtained for the test samples. The arrows in the figure indicate the retention times of standards; dashed lines indicate the absorbance of carrier lipids at 206 nm; solid lines indicate radioactivity. Similar results were obtained in a second experiment.

TABLE V
Production of PI from sn-2-arachidonoyl lysoPA by 3T3 cell membranes in the presence of CTP and myoinositol "'P-Labeled sn-2-arachidonoyl lysoPA (50 pmol) was incubated with the membrane fraction (80 pg) at 37 "C with 2.5 mM CTP, 20 mM magnesium acetate and/or 5 mM myoinositol in the presence of the PA phosphohydrolase inhibitors, NaF (50 mM) and Na3V0, (2 mM). The incubation periods were 5 min for those with 0.5 mM CoA and 20 min for those without CoA. Products were separated by twodimensional TLC as described under "Experimental Procedures." A similar result was obtained in a separate experiment. 1). However, the relatively high content of labeled sn-lstearoyl-2-arachidonoyl PE observed (Table 11) suggests that a stearoyl-specific pathway, such as the stearoyl-specific branch of the MG kinase-initiated pathway, may also have been involved. Furthermore, the stearic acid-containing PE that was formed may have been converted into PS. An MG acyltransferase reaction may have formed most of the radioactive DG that accumulated in the cell incubation experiments. The cells clearly contained such an activity (Table 111), and studies by previous investigators have shown that sn-%acyl MG acyltransferase activities in other cells can initiate MG incorporation pathways that form DG and TG (30,32). However, the specificity of the Swiss 3T3 cell enzyme remains to be fully characterized. We demonstrated that it shows no preference for sn-2-arachidonoyl MG over sn-2oleoyl MG, but have yet to examine its preference for different acyl-CoAs. Thus, we do not know whether the distribution of labeled DG species that we observed in the cell incubation experiments would be consistent with its activity.
The MG acyltransferase reaction that formed most of the cell DG may also have formed most of the cell PC. The major reason for believing this is that the distribution of radioactive species of DG was very similar to the corresponding distribution found for PC ( Table 11). The simplest way to account for this similarity would be to postulate that the MG acyltransferase formed sn-1,Z-DG and that a DG:phosphorylcholine transferase reaction subsequently converted this DG into PC (Fig. 5). However, direct evidence for this remains to be obtained, and we cannot exclude the possibility that some sn-2,3-DG also may have been formed.
Much more work will be needed to clarify the precise nature of these MG incorporation pathways. None of the enzymes involved has been isolated or fully characterized, so it is not clear how many enzymes or enzyme isoforms may be present or how they may relate to other enzymes that have been described. For example, we presume that the MG kinase in Swiss 3T3 cell membranes is related in some way to the soluble MG kinase that was recently found in bovine brain (33) because both enzymes phosphorylate sn-2-arachidonoyl MG in preference to sn-2-oleoyl MG. But the relation between the two enzymes remains to be determined.
Both the CoA-dependent stearoyl transferase and the ATPdependent acyltransferase that convert sn-2-arachidonoyl lysoPA into PA also remain to be characterized. We are currently examining the CoA-dependent enzyme in an effort to identify the stearoyl donor, establish the basis for the CoA requirement, determine the enzyme's specificity for different sn-%acyl lysophospholipids, and relate its activity to that of a recently discovered but still uncharacterized CoA-dependent stearoyl transferase activity in brain membranes (34).