Characterization of 1,2-Diacylglycerol Hydrolysis in Human Platelets DEMONSTRATION OF AN ARACHIDONOYL-MONOACYLGLYCEROL INTERMEDIATE*

When platelets are stimulated by thrombin, a phos- phatidylinositol-specific phospholipase C produces a transient rise in 1,2-diacylglycerol. We have now char- acterized the hydrolysis of diacylglycerol by platelet membranes using doubly isotopically labeled sub- strates of defined fatty acid composition, We find that the fatty acid at sn-1 is hydrolyzed faster than that at sn-2 thereby producing a 2-monoacylglycerol intermediate. If hydrolysis had occurred at either position randomly, 1-monoacylglycerol would also be produced. That none was detected indicates that either the sn-1 fatty acid must be cleaved first or that l-monoacyl- glycerol is hydrolyzed by monoacylglycerol lipase much faster than 2-monoacylglycerol. The latter possibility was excluded by the finding that l-monoacyl-glycerol and 2-monoacylglycerol are hydrolyzed at equal rates by platelet membranes.

When platelets are stimulated by thrombin, a phosphatidylinositol-specific phospholipase C produces a transient rise in 1,2-diacylglycerol. We have now characterized the hydrolysis of diacylglycerol by platelet membranes using doubly isotopically labeled substrates of defined fatty acid composition, We find that the fatty acid at sn-1 is hydrolyzed faster than that at sn-2 thereby producing a 2-monoacylglycerol intermediate.
If hydrolysis had occurred at either position randomly, 1-monoacylglycerol would also be produced. That none was detected indicates that either the sn-1 fatty acid must be cleaved first or that l-monoacylglycerol is hydrolyzed by monoacylglycerol lipase much faster than 2-monoacylglycerol. The latter possibility was excluded by the finding that l-monoacylglycerol and 2-monoacylglycerol are hydrolyzed at equal rates by platelet membranes.
The diacylglycerol lipase cleaves diacylglycerols with sn-1 palmitate as rapidly as those with sn-1 stearate. Arachidonate at sn-2 is cleaved twice as fast as sn-2 oleate by monoacylglycerol lipase. The two activities probably represent discrete enzymes since monoacylglycerol lipase activity can be separated from diacylglycerol lipase by fractionation on DEAE-Sepharose, although both are contained in the membrane fraction of platelets.
That the sequential breakdown of 1,2-diacylglycerol also occurs in intact platelets is indicated by our finding of a transient rise in arachidonoyl-monoacylglycerol in thrombin-stimulated platelets. This provides further evidence for a role of the phospholipase C-diacylglycerol lipase pathway in the release of arachidonic acid.
We have described diacylglycerol lipase activity in human platelets (1) and proposed that the combined activities of a phosphatidylinositol-specific phospholipase C, which generates an arachidonate-rich diacylglycerol (2), and the diacylglycerol lipase represent an alternative to phospholipase A2 as a mechanism for arachidonate release from membrane phospholipids. The phospholipase C-diacylglycerol lipase pathway, although less direct than a phospholipase A2, has several attractive features: 1) platelets contain adequate activity, as measured in vitro, to account for the rapid release of arachi-donate observed in stimulated platelets; 2) phosphatidylinositol contains 80% arachidonate at sn-2 (3, 4) which explains why arachidonate is the major unsaturated fatty acid released; 3 ) this pathway links the phosphatidylinositol effect, which has been described in many secretory tissues (5), to the production of arachidonate metabolites. There now is evidence for the phospholipase C-diacylglycerol lipase pathway in human placental membranes (6,7), porcine thyroid (8,9), murine fibrosarcoma cells (lo), rat mast cells (11), 3T3 cells (12), ram seminal vesicles' (13), and rabbit (14), sheep' (13), and human platelets.
The diacylglycerol lipase from human platelets has not been purified or extensively characterized. It is not known whether the fatty acids are hydrolyzed at sn-1 and sn-2 randomly or in an ordered, stepwise reaction. In retrospect, an ordered reaction, with sn-1 hydrolyzed f i s t , was suggested by our experiment in which fatty acid was released simultaneously with glycerol from a substrate labeled in the glycerol and in the sn-2 fatty acid (1). Further, Okazaki et al. ( 7 ) have shown that palmitate is hydrolyzed faster than oleate from a mixture of labeled palmitoyl and oleoyl diacylglycerols using fetal membranes and decidua vera as an enzyme source. Chau and Tai (15) reached the same conclusion using microsomes from human platelets when they showed that 2-monoacylglycerol, but not 1-monoacylglycerol, forms during hydrolysis of a mixture of diacylglycerols, of undefined fatty acid composition, with isotopic label a t sn-1 and sn-2. Neither of these previous studies excluded the possibility that different rates of hydrolysis of possible intermediate 1-and 2-monoacylglycerols could account for the observed effect. Okazaki et al. (7) found that arachidonate is hydrolyzed faster than oleate at sn-2. Chau and Tai (15) found that incubation with platelet microsomes altered the ratio of arachidonate to oleate in substrate diacylglycerols which had been doubly isotopically labeled suggesting preferential hydrolysis of arachidonate. However, in the latter study, it is also possible that the observed results depended on different fatty acids a t sn-1 rather than specificity at sn-2 since their substrates did not have a defined fatty acid composition.
Monoacylglycerol lipase activity is also present in human platelets (16,17). This could reflect a separate enzyme from diacylglycerol lipase or an additional activity of the same enzyme. Fielding (18) has proposed that monoacylglycerol lipase may have a physiological role in lipolysis of plasma glycerides.
We now report that the hydrolysis of fatty acids from 1,2diacylglycerol is an ordered, two-step reaction with the sn-1 position released fist. 2-Monoacylglycerol accumulates transiently both in enzymatic assays and in thrombin-stimulated platelets. The latter point supports a physiological role for the proposed phospholipase C-diacylglycerol lipase pathway.   Fig. 1. The sn-1 position is hydrolyzed first as shown by a greater accumulation of palmitate at each time. The pattern of release from sn-1 preceding that from sn-2 was observed in five experiments using membrane preparations from different donors (Table I). Although the total release varied by as much as 2.5fold, palmitate release consistently exceeded oleate. In experiments where all of the products of the reaction were separated by TLC, up to 0. 8 Fig.  1. The ratio of palmitate (sn-1) to oleate (sn-2) released was calculated at each time point and an average from the Same time point in different experiments was obtained. The points measured in three or more experiments are shown as are all points from the five experiments. A one-tailed "t"-test was used to evaluate the probability that the observed ratio was significantly different from 1.0. bly detected in the same experiment. The order of release of fatty acids was also determined in two experiments using 1-['4C]stearoyl, 2-[3H]arachidonoyl-sn-glycerol as substrate and the same pattern as that shown in Fig. 1 was obtained, i.e. 1.3-1.8-fold more stearate than arachidonate was found at all points from 5 to 30 min. These results suggest sequential release of the fatty acids with the sn-1 position of 1,2-diacylglycerol hydrolyzed first. An alternative explanation for the results is that the rate of release is identical from both positions of 1,2-diacylglycerol but that the resultant l-monoacylglycerol is hydrolyzed faster than 2-monoacylglycerol. This hypothesis was excluded by our finding of an identical time course of hydrolysis of I-['4C]monoolein and 2-[3H]monoolein as shown in Fig. 2. In other experiments performed with substrate concentrations from 100 PM to 1 mM, rates of hydrolysis ranged from 18 to 66 nmol/min/mg of microsomal protein and were equal with either substrate. The apparent K , obtained from Lineweaver-Burke plots is 530 PM for 1monoolein and 750 p~ for 2-monoolein. These values should be interpreted cautiously since the substrate is hydrophobic and a membrane preparation was the source of enzymatic activity. Nonetheless, the observed order of release from 1,2diacylglycerol cannot be explained by different rates of hydrolysis of I-monoacylglycerol versus 2-monoacylglycerol. A possible, but unlikely, exception to this conclusion is that a 1monoacylglycerol containing a saturated fatty acid is preferred over a 2-monoacylglycerol containing an unsaturated fatty acid.

Diacylglycerol Hydrolysis in Platelets
Transient Accumulation of ArachidonoyE Monoacylglycerol in Thrombin-stimulated Platelets-When washed human platelets, which have been labeled with r3H]arachidonic acid, are stimulated with thrombin, there is a transient rise in 1,2-diacylglycerol (Fig. 3 ) as previously reported by Rittenhouse-Simmons (2). In approximately the same time course (Fig, 3, Table II), an accumulation and then loss of monoacylglycerol also occurs indicating the presence of 2-arachidonoyl-sn-glycerol as an intermediate product of the diacylglycerol lipase. Neither diacylglycerol nor monoacylglycerol accumulated in the absence of thrombin. In other incubations with identical labeling techniques and the same total incorporation of arachidonate, we measured the specific activity of [."H]arachidonate in phosphatidylinositol and released arachidonate. Values ranged between 2500 and 3500 cpm/nmol." Based on these results, we estimate the peak of 1,2-diacylglycerol to be approximately 0.6 nmol/lO" platelets which agrees with the results obtained by Rittenhouse-Simmons (2). In the experiments shown in Table 11 FIG . 4 (right). Fatty acid preference at sn-2 for release from 1,2-diacylglycerol. Diacylglycerol lipase activity in crude membranes was assayed in a reaction mixture which contained equimolar amounts of 1-palmitoyl, 2-[3H]oleoyl-sn-glycerol and I-palmitoyl, 2-['4C]arachidonoyl-sn-glycerol (125 PM each). At the indicated times, the reaction was stopped and the fatty acids were extracted and measured by scintillation spectrometry. 0, arachidonate; A, oleate.
Thrombin was added and the levels of monoacylglycerol and diacylglycerol were measured as described ( Fig.  3 and "Experimental Procedures"). The time of the peak accumulation of each glyceride and the value at that time (expressed as cpm/ lo9 alatelets) are shown.

Monoacylglycerol
Diacylglycerol (peak Experiment (peak accumulation) accumulation) Substrate Specificity of Diacylglycerol Lipase-Since the predominant species of diacylglycerol derived from phosphatidylinositol is 1-stearoyl, 2-arachidonoyl-sn-glycerol, we investigated whether this is a preferred substrate for hydrolysis. In each of two separate experiments, we measured rates of hydrolysis of arachidonate uersus oleate from diacylglycerols with palmitate at sn-1 and find that there is a 2-fold greater release of arachidonate (Fig. 4) in agreement with observations in placental membranes (7). We have also examined the effect of the fatty acid at sn-1 on diacylglycerol breakdown using substrates with stearate or palmitate in the sn-1 position and arachidonate at sn-2 (Fig. 5) and find that palmitatecontaining substrate is hydrolyzed at the same rate or even slightly faster than that containing stearate.

Time cpm/lOq cells
The Relationship of Monoacylglycerol Lipase and Diacylglycerol Lipase Actiuitie,+"ubcellular fractionation of  platelets was performed and each of the fractions were assayed for both monoacylglycerol and diacylglycerol lipase activities (Fig. 6). Both activities have the same distribution and are found in the membrane-containing fractions. In our respective assays, there consistently was 20 to 50 times as much monoacylglycerol lipase activity as diacylglycerol lipase although the differences decreased when the activities were measured at early time points at high substrate and low enzyme concentrations. Both activities are lost in parallel by heating a t 50 "C or exposure to N-ethylmaleimide, diisopropyl fluorophosphate, or phenylmethylsulfonyl fluoride (data not shown). We have attempted to determine whether the two activities can be separated from each other. As we have described (ZO), solubilization and a 3-to 4-fold purification can be accomplished by sonication in a buffer containing low concentrations of detergent followed by fractionation with ammonium sulfate. Monoacylglycerol and diacylglycerol lipase activities are not separated by these procedures. Attempts to purify the enzyme(s) by ion exchange, hydrophobic dye matrix, and gel fitration chromatography under a variety of conditions have been largely unsuccessful. In most cases, little activity binds to the column suggesting that although the activities are present in a 100,000 X g supernatant they are associated with lipid and/or detergent micelles. Monoacylglycerol lipase activ-  Fig. 7 were assayed for both activities as described in the text. The starting material had 6.5 pmol/min of monoacyIglycero1 lipase and 65 nmol/min of diacylglycerol lipase activities. The recoveries were 97 and 106%, respectively. Open burs, monoacylglycerol (MG) lipase; shaded burs, diacylglycerol (DG) lipase. ity elutes in three peaks from a DEAE-Sepharose column as shown in Fig. 7 . The first peak does not bind to the column, and the others elute with further washing and the addition of NaC1, respectively. When the pooled fractions were assayed for both monoacylglycerol and diacylglycerol activities, the intermediate peak of monoacylglycerol lipase activity (C) was found to be almost devoid of diacylglycerol lipase activity while the other peaks contained both (Fig. 8). Thus, it appears that there is a monoacylglycerol lipase that does not metabolize 1,2-diacylglycerol. This does not resolve the issue of whether a single enzyme that acts on 1,2-diacylglycerol can hydrolyze a monoacylglycerol as well.

DISCUSSION
Okazaki et al. (7) concluded that diacylglycerol lipase from placental tissue cleaves the sn-1 position of 1,2-diacylglycerol prior to cleaving the sn-2 position. Chau and Tai (15), using platelet microsomes and diacylglycerols of undefined fatty acid composition, which were labeled in either the sn-1 or sn-2 position, observed the accumulation of 2-monoacylglycerol but not 1-monoacylglycerol and also concluded that the sn-1 position of diacylglycerol is cleaved first. We have confirmed and extended these results using substrate diacylglycerols of defined fatty acid composition. We find that: 1) the sn-1 position is cleaved before the sn-2 position; 2) 2-monoacylglycerol, but not 1-monoacylglycerol, accumulates during the reaction; 3) 1-and 2-monoacylglycerols are hydrolyzed at equal rates by platelet microsomes, thereby excluding the possibility that 1-monoacylglycerol does not appear as an intermediate because of preferential hydrolysis; 4) arachidonate at sn-2 is cleaved slightly faster (-2-fold) than oleate at the same position; and 5) either palmitate or stearate is readily hydrolyzed from sn-1 indicating a lack of specificity for the fatty acid in this position.
The fact that the sn-1 position of diacylglycerol is cleaved first and that platelet microsomes cleave palmitoyl-and stearoyl-diacylglycerol readily suggests that the specificity of fatty acid release is not dictated by the substrate specificity of diacylglycerol or monoacylglycerol lipases. Although the 2fold preference for arachidonate over oleate at sn-2 may have some physiological relevance, the major factor controlling which fatty acids are released from stimulated platelets appears to be the fatty acid composition of phosphatidylinositol. Since arachidonate comprises 80% of the sn-2 position in platelet phosphatidylinositol (3, 4), it is the major fatty acid released. In fact, oleate and linoleate are also released upon thrombin stimulation of platelets in proportion to their occurrence in phosphatidylinositol (3, 26).
An alternative route of metabolism for the 1,2-diacylglycerol which accumulates transiently following the stimulation of platelets with thrombin is conversion to phosphatidic acid. Platelets contain diglyceride kinase (27) and there is a rise in phosphatidic acid following thrombin stimulation (4). Rittenhouse-Simmons measured the level of diacylglycerol in thrombin-stimulated platelets that had been preincubated with indomethacin which results in inhibition of the diacylglycerol lipase but not diacylglycerol kinase ( 2 8 ) . She found a 5-fold increase in diacylglycerol accumulation, but no augmentation of phosphatidic acid accumulation and she concluded that much of the 1,2-diacylglycerol is hydrolyzed in vivo to release arachidonate. This conclusion is supported by the observation of Habenicht et al. (12) who found a transient accumulation of monoacylglycerol, as well as diacylglycerol, in stimulated 3T3 cells. Phosphatidic acid has been proposed as both an activator of, and substrate for, phospholipase A2 (29). It has also been suggested that phosphatidic acid is subsequently converted back to phosphatidylinositol, thereby completing the futile "phosphatidylinositol cycle." We and others (3, 30) previously have shown that cycling back to phosphatidylinositol is not the fate of most of the 1,2-diacylglycerol formed by thrombin-stimulated human platelets. The phosphatidylinositol which is resynthesized after thrombin stimulation is deficient in stearate and arachidonate as compared to unstimulated cells and therefore is not a result of cycling but of de nouo synthesis . Broekman et al. (4) reached a conflicting conclusion probably because in their experiments less than one-half as much phosphatidylinositol was resynthesized as in ours (3). We now have provided further evidence for the functional significance of the lipase pathway with our finding of a thrombin-induced increase in arachidonoyl monoacylglycerol (Fig. 3). This compound is a predicted intermediate in the serial release of fatty acids from 1,2-diacylglycerol ( Fig.  1) and indicates that diacylglycerol is metabolized by ordered fatty acid hydrolysis in intact platelets as well as in assays of membranes using exogenous diacylglycerol.
The monoacylglycerol and diacylglycerol lipase activities of human platelets may represent different enzymes. Chau and Tai (15) reached this conclusion based on differences in pH optimum and Okazaki, et al. (7) also favored two distinct enzymes in placental tissues based on different rates of inhibition and different percentages of the two activities in the supernatant following homogenization and centrifugation. In our experiments, both activities were found in the same subcellular fractions (Fig. 6), were inactivated equally well by several inhibitors and heat, were both solubilized with detergent, and were found in the same ammonium sulfate fractions. However, a portion of the monoacylglycerol lipase activity was separated from fractions containing both activities suggesting that monoacylglycerol lipase is a separate enzyme (Fig. 8). Whether there is another enzyme which has only diacylglycerol lipase activity or both activities remains undetermined in spite of our attempts at purification.
We conclude that most of the 1-stearoyl, 2-arachidonyl-s~~glycerol which appears in thrombin-stimulated platelets is acted upon by a diacylglycerol lipase to yield 2-arachidonoylsn-glycerol. This monoacylglycerol undergoes hydrolysis catalyzed by a monoacylglycerol lipase, which is probably a separate enzyme, to release arachidonate for further metabolism.