Metabolic fate of plasma membrane diacylglycerols in NIH 3T3 fibroblasts.

We have examined the metabolism of three radiolabeled 1,2-diacylglycerols (DGs) in NIH 3T3 fibroblasts. Since the lipids used are not appreciably taken up by the cells, we used a phosphatidylserine (PS)-based liposome fusion system to rapidly associate the lipid species with the plasma membrane. When 1,2-[1-14C]dioleoyl-sn-3-glycerol ([14C]DOG) is delivered in this way, it is rapidly converted predominantly to phosphatidylcholine (PC) and triacylglycerol (TG) and to a lesser extent, to monoacylglycerol (MG) and fatty acids (FA), as well as phosphatidic acid (PA) and phosphatidylinositol (PI). We present evidence that [14C] DOG is largely utilized as an intact molecule rather than being broken down to FA and then incorporated to cell lipids. Examination of the metabolism of 1-stearoyl-2-[1-14C]myristoyl-sn-3-glycerol ([14C]SMG) and 1-stearoyl-2-arachidonoyl-sn-3-glycerol ([14C]SAG) reveal important differences. Both produce substantial labeling of PC but [14C]SMG gives rise to the highest proportion of TG and the lowest of PA and PI, whereas [14C]SAG yields the opposite pattern. When phosphatidic acid labeled on its glycerol backbone (1,2-dioleoyl-sn-[U-14C] glycero-3-phosphate) was supplied to the cells via the liposomes, rapid appearance of labeled DG was found which then decreased with concomitant labeling of cellular PC and TG. Only small amounts of the glycerol backbone were recovered in PI. Our experiments identify three types of processes involved in the metabolism of plasma membrane DGs: (i) transferase-catalyzed conversions to PC and TG, (ii) lipolytic breakdown to MG and FA, and (iii) phosphorylation to PA and then conversion to PI. The relative proportions of each DG species converted to these different products are strongly dependent on the fatty acyl composition of the particular DG molecular species, even though formation of PC is the major event in all cases. Since DGs are important second messengers, our study supports the view that conversion to PC and TG can play a key role in DG signal attenuation.

We have examined the metabolism of three radiolabeled 1,2-diacylglycerols (DGs) in NIH 3T3 fibroblasts. Since the lipids used are not appreciably taken up by the cells, we used a phosphatidylserine (PS)based liposome fusion system to rapidly associate the lipid species with the plasma membrane. When 1,2-[ 1-'"C]dio1eoy1-sn-3-g1ycero1 (['"CIDOG) is delivered in this way, it is rapidly converted predominantly to phosphatidylcholine (PC) and triacylglycerol (TG) and to a lesser extent, to monoacylglycerol (MG) and fatty acids (FA), as well as phosphatidic acid (PA) and phosphatidylinositol (PI). We present evidence that ['"C] DOG is largely utilized as an intact molecule rather than being broken down to FA and then incorporated to cell lipids. Examination of the metabolism of 1-stearoy1-2-[1-'"C]myristoy1-sn-3-g1ycero1 (['"CISMG) and l-stearoyl-2-arachidonoyl-sn-3-glycerol (['"C] SAG) reveal important differences. Both produce substantial labeling of PC but ['"CISMG gives rise to the highest proportion of TG and the lowest of PA and PI, whereas ['"CISAG yields the opposite pattern. When phosphatidic acid labeled on its glycerol backbone (1,2dioleoyl-~n-[U-'~C]glycer0-3-phosphate) was supplied to the cells via the liposomes, rapid appearance of labeled DG was found which then decreased with concomitant labeling of cellular PC and TG. Only small amounts of the glycerol backbone were recovered in PI. Our experiments identify three types of processes involved in the metabolism of plasma membrane DGs: (i) transferase-catalyzed conversions to PC and TG, (ii) lipolytic breakdown to MG and FA, and (iii) phosphorylation to P A and then conversion to PI. The relative proportions of each DG species converted to these different products are strongly dependent on the fatty acyl composition of the particular DG molecular species, even though formation of PC is the major event in all cases. Since DGs are important second messengers, our study supports the view that conversion to PC and TG can play a key role in DG signal attenuation.
The reactions involved in the metabolism of diacylglycerol (DG)' are likely to be determinants in the function of these second messenger molecules. To date, most of the data used to analyze this aspect of DG messenger function has been derived from studies carried out in platelets. In this system, it is believed that lipolysis and phosphorylation to phosphatidic acid (PA), leading in turn to formation of phosphatidylinositol (PI), are the major events removing the DGs (1, 2). Thus, when dioctanoylglycerol (diCs) metabolism was studied in platelets, a significant conversion of diCs to PI was observed (3). Because diC8 is considerably more polar than natural signaling DGs, the results obtained in intact cells using this synthetic DG may not be representative of the metabolic fate of naturally occurring DGs. To enable us to study the metabolism of longer chain DGs, we here employ a liposome fusion system to rapidly label the plasma membrane of cultured cells with selected molecular species of radioactive DGs. Using this approach, we examined the metabolism of three different labeled DGs, 1,2-[14C]dioleoyl-sn-glycerol ([I4C]   [14C]SMPC was separated from the catalyst by passage through Rexyn 1-300 ion exchange resin and purified by TLC, using commercial nonlabeled 1-stearoyl-2-myristoyl-PC as standard. After thin layer chromatography of a mixture of the commercial and the radiolabeled synthetic compound, no chemical or radioactive impurities were detected by prolonged autoradiography (10 days at -70 "C) and by chemical staining with molybdenum blue and charring (6). This mixture also yielded the expected reaction products upon treatment with phospholipases C, D, and Az.  0.036-0.07 pmol) of the pure radiolabeled phospholipids (without addition of unlabeled compound) were resuspended in 0.5 ml of 50 mM Tris-HCl,30 mM sodium borate, 0.04% NaN3, pH 7.4, containing 25 p1 of phospholipase C from Bacillus cereus, commercial preparation. After 30 min of incubation at 37 "C, the DG formed was extracted with 4 X 2 ml of water-saturated diethyl ether. The solvent was evaporated under nitrogen and the DGs were used directly in the experiments as described below. The phospholipids were completely hydrolyzed to DG under these conditions as shown by TLC.
Labeling Procedures labeled lipids to the PM of cultured NIH 3T3 fibroblasts. The protocol We tested a number of liposome systems for delivering the radiofinally selected involved phosphatidylserine/phosphatidylcholine (PS/PC) liposomes which were induced to associate with the PM by brief pretreatment of the cells with polylysine (for the principles involved, see Stegmann et al. (7)). The procedure comprised three steps: 1) treatment of the cells for 1 min with 25 pg/ml of polylysine in PBS at room temperature; (ii) exposure to negatively charged PS/ PC liposomes carrying the radioactive lipids (0.1 ml of liposome suspension + 0.9 ml of PBS) for 5 min at room temperature; and (iii) removal of the liposome suspension, washing twice with DMEM, and incubation for the indicated time periods at 37 "C, after which the lipids were extracted as described below. The liposome suspension was prepared as follows: aliquots of CHCl, solutions containing 0.5 pmol of bovine heart PS and 0.5 pmol of egg PC were mixed with 4 pCi of the radiolabeled lipids, which were also dissolved in CHC13. The solvent was evaporated under nitrogen and further in uacuo for 30 min and 1 ml of 20 mM Tris-HC1,150 mM NaC1,O.l mM EDTA, pH 7.4, was added. The lipids were suspended by vigorous vortexing with the aid of a glass bead and finally sonicated (2 X 1 rnin), under N,, with the microtip of an Ultrasonic W-385 sonicator at setting 4. The liposomes were then centrifuged for 10 min at 10,000 rpm in the SS34 rotor of a high speed Sorvall centrifuge to remove any metal particles originating from the sonicator probe. To monitor the interaction of the lipid vesicles with the cells, we used two procedures: (a) N-Rh-PE, a nonexchangeable fluorescent phospholipid analog, incorporated into the liposomes (1 mol %) and ( b ) 5(6)-carboxyfluorescein (CF). In this case, we prepared the liposomes containing 0.15 M CF dissolved in 10 mM Tris-HC1, pH 7.4. These liposomes were then passed through a Sephadex G-50 column (1 X 20 cm), equilibrated, and eluted with 0.15 M NaCl, 10 mM Tris-HC1, pH 7.4, to separate untrapped CF. At the concentration used, CF is selfquenched and, upon dilution, fluorescence increases markedly (8). Fluorescence microscopy revealed that the N-Rh-PE-labeled liposomes rapidly attach to the surface of the cells pretreated with polylysine. Diffuse fluorescence over the whole cells indicated fusion between plasma membranes and vesicles. There was, however, substantial punctate fluorescence, which remained cell-associated despite extensive washing. This suggests that part of the lipids associated with the cells are not incorporated to the membranes and are not available for metabolic processing. This possibility is supported by the observation that part of the lipids delivered by the liposome system remain unmetabolized after 2 h of incubation. When we used the CF-labeled liposomes, the cells showed green fluorescence, which indicates dilution of the CF in the cytoplasm and therefore, fusion of liposomes and PM.
In addition to the PS/PC liposomes, we used two alternative systems for delivering radioactive DGs: cardiolipin (CL) and DDAB-PE liposomes. They were prepared as described above for the PS/PC liposomes using 2 pmol of CL or 1 mg of PE + 0.4 mg of DDAB, respectively, for every 4 pCi of the radiolabeled DGs employed. In the experiments with CL liposomes, we took advantage of the induction of fusion of this kind of liposomes by acidic conditions (9). In this case, a two-step procedure was followed. First, the cells were washed twice with cold PBS and incubated in 0.9 ml of PBS + 0.1 ml of liposome suspension at 4 "C for 1 h. The cells were then washed twice with DMEM and incubated at 37 "C for 30 and 60 min, after which the lipids were extracted as described below.
A mixture of PE and a cationic detergent has been shown to form liposomes that fuse with the PM and are an efficient means for cell transfection with exogenous DNA (10). In this work we have used PE and DDAB as cationic lipid. This mixture is also an efficient DNA carrier for transfection of cells? We employed these liposomes as described for the PS/PC ones, except that the step of exposure to polylysine was omitted.

Cell Treatments
In the experiments involving the DG kinase inhibitor R59022 (11) and the DG lipase inhibitor RG 80267 (121, the cells were preincubated with these agents for 30 min prior to the addition of liposomes. R59022 and RG 80267 were prepared in 200 X concentrated stock solutions in dimethyl sulfoxide. Controls received the same amount of dimethyl sulfoxide (0.5% final concentration).

Lipid Extraction
After incubation for the indicated time periods, lipids were extracted by a modification of the procedure of Bligh and Dyer (13). The medium from the dishes was aspirated and 1.9 ml of ice-cold CH30H/Hz0 (kO.9, v/v) was added. The cells were then scraped into glass tubes containing 1 ml of CHC1,. The tubes were vortexed and centrifuged to separate the phases. The lower phase was collected, evaporated under nitrogen, and the lipids dissolved in CHC13.

Thin Layer Chromatographic Separation of Lipids
Two procedures were used for lipid analysis: (a) samples of labeled cellular lipids were separated on Silica Gel 60 or LK6OD plates, J. K. Rose, personal communication.
developed in two steps in the same direction: first, with CHC13/ CH30H/H20 (65:35:2.5, v/v) to a height of 12 cm (for separation of polar lipids) and then, with petroleum ether (30-60 "C)/diethyl ether/ acetic acid (9020:2, v/v) to the top (to separate neutral lipids). Radioactive lipids were located by autoradiography, exposing Kodak XAR films for 48 h a t -70 "C. (b) We also analyzed samples of lipids on separate plates for neutral or polar lipids using the same solvent mixtures as described above. For quantification, unlabeled TLC standards were added to the samples and lipid spots were detected by exposure to iodine vapor, marked, and scraped into scintillation vials after allowing evaporation of the iodine. The samples were counted in a Packard Tri-Carb Liquid Scintillation Analyzer 1900 CA, using Ecoscint scintillation fluid.

Metabolic Fate of [l-'4C]Dioleoyl Glycerol (r4C]DOG) and
[l-14C]Oleic Acid (r4C]18:1)-We employed the PS/PC liposome system to label the plasma membrane of NIH 3T3 fibroblasts. The cells were briefly exposed to polylysine immediately before a short incubation with the liposomes containing either [14C]DOG or ['4C]181. These were then removed and the cells washed as described under "Experimental Procedures." The labeled cells were incubated at 37 "C for 30 min, at the end of which their lipids were extracted and analyzed by TLC. In Fig. 1   Results are expressed as percentages of the total radioactivity incorporated into cellular lipids other than the precursor used, marked with an asterisk (*). In the case of ["CIDOG, the total radioactivity in all lipids except DG averaged 24,708 cpm which corresponds to 22.5% of the total radioactivity associated with the cells. The remainder of the radioactivity is probably present in nonfused liposomes that account for the punctate fluorescence in N-Rh-PE-labeled liposomes (see "Experimental Procedures"). In the case of  DOG is carried out at 18 "C instead of 37 "C, the same basic pattern arises, although relatively higher amounts of MG and fatty acids (FA) accumulate in the cells (not shown). We also investigated two alternative liposome carriers: cardiolipin liposomes, which fuse with the plasma membrane at acidic pH (9), and DDAB-PE liposomes, that fuse at neutral pH values (10). We found that the PS/PC system was the most efficient in labeling the cells, followed by the cardiolipin liposomes. Parallel experiments using the three systems showed the same basic patterns of cellular lipid labeling irrespective of the system used (data not shown). Effects of R59022 and RG 80267 on the Metabolism of r4Cl DOG-Cells were preincubated with 10 PM R59022, a DG kinase inhibitor (ll), 30 PM RG 80267, inhibitor of DG lipase (12), or the same amount of solvent (0.5% dimethyl sulfoxide), for 30 min and then they were labeled with PS/PC liposomes carrying [14C]DOG. After incubation of the labeled cells for 60 min, the cellular lipids were extracted and separated as described under "Experimental Procedures." The results in Table I show that R59022 does not affect the conversion of [14C]DOG to PC and TG but strongly reduces the incorporation to PA and PI. In the presence of RG 80267, the amount of [14C]DOG converted into FA and MG is markedly reduced, but the amounts converted to PC and TG are not altered.
Metabolism of r4C]Myristateand r4ClArachidonate-la-  Calculation of ratios of radioactivity recovered in TG or PI with respect to that in PC are shown and correspond to the data displayed in Fig. 4.  Fig. 1). The predominant species labeled as a result of SMG metabolism is TG, with a smaller but significant amount of PC labeled, but not PE or PI, and only a barely detectable amount of PA. In contrast, in the case of SAG, the major species labeled is PC, with much smaller labeling of TG but significant labeling of PA and PI. In Fig. 3b Table 11, we compare the ratios of TG and PI to PC for the two DGs and the corresponding fatty acids. The results shown in Table I1 emphasize the differences discussed between the fates of the two DG molecular species and also between labeling the cells with either DGs or free fatty acids.

TG/PC
In Fig. 4, the time course of conversion of [14C]SMG and [14C]SAG to other lipids is shown separately for polar and neutral lipids. Examination of the conversion to polar lipids indicates that PC is a major end product for both DGs. PA is formed notably only in the case of [14C]SAG and seems to have an early peak at or before 15 min. Labeling of PI increases steadily with time, but the amounts formed do not account for the decrease observed in PA after 15 min. Thus, P A is being converted to other lipids in addition to PI. The neutral lipid panel shows the striking conversion to TG in the case of [14C]SMG. This lipid also becomes labeled from [14C]SAG but to a lesser extent and in a transient fashion. TG labeled from [14C]SAG peaks at 30 min and thereafter decreases, whereas that labeled from SMG as the precursor appears to be a more stable cellular lipid. Both DGs are subject to lipolytic attack, indicated by the appearance of FA and MG.
Metabolism of f4C]Glycerol Backbone-labeled Dioleoyl P h sphatidic Acid-We have prepared 1,2-dioleoyl-sn-[U-14C]glycero-3-phosphate ([14C]PA) from [U-'4C]-sn-3-glycerol phosphate by acylation with oleic anhydride. The metabolism of this compound delivered via the PS/PC liposomes was examined in NIH 3T3 fibroblasts. Initially, all the radioactivity is present exclusively as PA. After 15 min of incubation, considerable conversion to DG and PC is evident. The only other radioactive compounds detected are TG and minor amounts of MG and PI. Fig. 5 quantitative information on the time course of ["C] PA metabolism illustrates that conversion to DG is transient and that PC is the major destination of the glycerol backbone initially associated with the plasma membrane (Fig. 5). The second lipid pool in which this backbone is recovered is TG.

DISCUSSION
In this investigation, we have used a new methodology to associate [14C]DOG with the plasma membrane of NIH 3T3 fibroblasts. Upon incubation of the labeled cells, the radioactivity is recovered mainly in PC and TG and, to a lesser extent, in other lipids, like MG, FA, PA, and PI. The pattern observed is markedly different from the one resulting from labeling with ['4C]18:1 under the same conditions. These results indicate that incorporation of [14C]DOG into cellular lipids does not proceed to any significant extent through prior lipolytic breakdown to free fatty acids, and imply that the DGs supplied directly enter the biosynthetic pathways leading to PC and TG. Since the enzymes catalyzing these conversions are known to reside in the endoplasmic reticulum (14), our finding provides the first experimental evidence for a transport process conveying naturally occurring DGs from the plasma membrane (PM) to internal membranes. Previously, Pagano and co-workers (15)(16)(17)   remained to be determined. Indeed, the relatively polar nature of the lipids utilized makes them significantly different in their physicochemical properties compared to longer chain natural lipids. The importance of chain length was emphasized by studies of Nichols and Pagano (18). Thus, when these authors compared the spontaneous diffusion rates of Clz-NBD-PC and C6-NBD-PC, they found that the latter transfers spontaneously more than 2 orders of magnitude faster than the first. For this reason, our experiments utilizing radioactively labeled DGs that occur naturally, like DOG, SMG, and SAG, are critical in providing convincing evidence of DG transporting mechanisms in intact cells. They also help to clarify how PM DGs are metabolized. We found similar results using three alternative and unrelated liposome systems for delivery of radiolabeled DOG. The three systems have different efficiency in labeling the cells, but the patterns obtained are closely similar. This strongly suggests that any alterations of the PM lipid composition induced by the liposomes do not affect the processes studied. Furthermore, in our previous investigation on the metabolism of cell permeant diCs,3 no liposome system was used, ruling out membrane alterations, and the results were essentially the same as those found here. Incubation of [14C]DOG-labeled cells at 18 "C, a temperature that blocks fusion of pinocytic vesicles with lysosomes (19), also results in labeling of PC and TG predominantly. This supports the notion that lysosomal degradation of the radiolabeled lipids plays no significant role in the processes observed.
The results of our experiments using the inhibitor of DG lipase RG 80267 are consistent with the notion that release of FA from DG does not precede the incorporation of label from [14C]DOG into cellular lipids, as strong inhibition of the appearance of MG and FA does not result in decreased amounts of PC and TG. This further supports the view that the DGs are incorporated as intact molecules. The inhibitor of DG kinase had strong inhibitory effects on PI formation, which suggests that this compound effectively inhibits a DG kinase involved in the PI cycle. More importantly, the observation that PC and TG labeling is unaffected by R59022 suggests that the transfer of DG molecules to the sites of biosynthesis of these two cellular lipids does not require conversion to PA, i.e. DGs are likely to be translocated as such within the cells. The mechanism by which such transfer can take place deserves further study. The fact that lysosomal degradation appears not to be involved suggests that transport on endosomal membranes is not a prevailing process for DG transfer. Our analysis of the lipids associated with the cells shows that substantial amounts of DGs remain unmetabolized even after prolonged incubation and our fluorescence microscopy observations indicate that, in addition to fusion, liposomes can become rapidly attached to the cells in a punctate pattern. These results are consistent with the notion that part of the liposomes associate with the cells without fusion and that DGs contained in these liposomes remain excluded from cellular metabolic processes. They emphasize that close apposition of lipid bilayers is not sufficient for the exchange of DGs. Translocation of DGs within the cells, thus, calls for DG transport proteins, similar to those known to occur for different phospholipids and fatty acids (20,21). Nichols and Pagano (18) have investigated the effect of several purified phospholipid exchange proteins on the transfer of C12-NBD-DG. They found enhancement of DG transfer with bovine liver nonspecific exchange protein. However, a search for specific DG transfer proteins has not yet been reported.
In this study, we examined the fate of different DG molecular species. We found that two naturally occurring DGs, SAG and SMG, are metabolized at different rates. We have identified three kinds of metabolic reactions affecting these DGs. In order of importance they are: (i) conversion to either PC or TG, which involves the action of two enzymes: CDP-cho1ine:DG cholinephosphotransferase and acyl-CoA:DG acyltransferase, respectively; (ii) lipolytic breakdown to yield MG and FA, and (iii) phosphorylation to PA by DG kinase. The resulting PA can be further metabolized to PI. It can also be hydrolyzed back to DG (see below). The rates at which these different processes take place are markedly dependent on the DG molecular species examined. Both [14C]SMG and [ 14C]SAG are good precursors for cellular PC, but they differ in the rates at which they are converted to TG and PA/PI.
[14C]SMG is a good precursor for TG and a poor precursor for PA/PI. The opposite is true for [14C]SAG. These observations most probably reflect the substrate specificities of the different enzymes involved. Our findings agree well with a recent report showing that DG kinase can show marked substrate specificity towards arachidonate-containing DGs in Swiss 3T3 cells (22).
Recently, Simpson et al. (23), studied the metabolism of exogenous monoacylglycerols radiolabeled on their glycerol moieties in Swiss 3T3 fibroblasts. They reported that a number of enzymes prefer arachidonoyl-to oleoyl-MGs. Their results led them to propose that an initial phosphorylation of the monoarachidonoylglycerol channels the glycerol backbone to P A and PI, whereas acylation to DG directs the labeled glycerol backbone to PC and TG. Our present findings are in excellent agreement with this proposal.
We studied the labeling patterns obtained using two very different fatty acids, ['4C]14:0 and [l4C]20:4. These experiments provide useful information regarding two aspects. First, we found that markedly different patterns arise again when the labeled fatty acid is provided as a free compound or esterifying position sn-2 of a DG. Second, they show that labeling of cellular PI with the free fatty acids is much stronger than that observed with the corresponding DGs, as reflected in the ratios of labeling of PI to PC shown in Table   11. These findings indicate that PI biosynthesis is more closely connected to de novo synthesis of phospholipids than to recycling of DGs from the PM, as compared to PC.
We examined the fate of the radiolabeled glycerol backbone when supplied to the PM as [14C]glycerol-dioleoyl PA via PS/ PC liposomes. The results have two noteworthy aspects. First, the PA is rapidly converted to DG, strongly indicating the presence of PA phosphohydrolase activity at the PM. This is in agreement with results reported previously using fluorescent phospholipid analogs (24). The DG formed is thereafter rapidly cleared from the cells. Second, the glycerol backbone labels cellular lipids, precisely as expected if conversions to PC and TG were the major processes withdrawing DGs, since these are almost the only lipids in which the radioactivity is recovered in these experiments. Formation of PI is, indeed, a very minor event. These results considerably strengthen the conclusions reached using fatty acyl-labeled DGs discussed above.
It is now generally accepted that, in addition to phosphoinositides, PC is an important source of second messenger DGs (25,26). It should be noted, therefore, that if the PI cycle was an important cellular means of DG removal, its operation would lead to an expansion of the PI pool at the expense of PC. This is against available experimental evidence (27). It has also been shown that phorbol esters stimulate both PC breakdown and biosynthesis (28)(29)(30). Thus, it seems reasonable to propose that a PC cycle could be activated in connection with signaling events (31). Our finding, that naturally occurring DG molecular species associated with the PM are precursors of cellular PC, substantiates the notion that such a cycle can indeed take place.
From this work, a new picture for the metabolic fate of plasma membrane DGs arises. In it, new processes emerge as key components of the cellular mechanism for DG signal termination. We recognized that the conversions catalyzed by transferases that act on DG molecules to yield PC or TG are the most important metabolic processes affecting cellular DGs initially associated with the PM. On the other hand, breakdown by DG lipase and phosphorylation to PA by DG kinase, two pathways demonstrated in platelets, seem to play a less important role in 3T3 fibroblasts. The relative rates of these possible conversions appear to be dictated by the fatty acyl composition of the particular molecular species. Indeed, differential channeling of the diverse molecular species may explain how, upon cell stimulation by DG-enhancing agonists, some of the DG species represent only brief signals, while others, lasting longer, can support sustained cell activation (32).