Arachidonoyl-diacylglycerol kinase. Specific in vitro inhibition by polyphosphoinositides suggests a mechanism for regulation of phosphatidylinositol biosynthesis.

We previously described the purification of a membrane-bound diacylglycerol kinase highly selective for sn-1-acyl-2-arachidonoyl diacylglycerols (Walsh, J. P., Suen, R., Lemaitre, R. N., and Glomset, J. A.(1994) J. Biol. Chem. 269, 21155-21164). This enzyme appears to be responsible for the rapid clearance of the arachidonate-rich pool of diacylglycerols generated during stimulus-induced phosphoinositide turnover. We have now shown phosphatidylinositol 4,5-bisphosphate to be a potent and specific inhibitor of arachidonoyl-diacylglycerol kinase. Kinetic analyses indicated a Ki for phosphatidylinositol 4,5-bisphosphate of 0.04 mol %. Phosphatidic acid also was an inhibitor with a Ki of 0.7 mol %. Other phospholipids had only small effects at these concentrations. A series of multiply phosphorylated lipid analogs also inhibited the enzyme, indicating that the head group phosphomonoesters are the primary determinants of the polyphosphoinositide effect. However, these compounds were not as potent as phosphatidylinositol 4,5-bisphosphate, indicating some specificity for the polyphosphoinositide additional to its total charge. Five other diacylglycerol kinases were activated to varying degrees by phosphatidylinositol 4,5-bisphosphate and phosphatidic acid, suggesting that inhibition by acidic lipids may be specific for the arachidonoyl-DAG kinase isoform. Given the presumed role of arachidonoyl-diacylglycerol kinase in the phosphoinositide cycle, this inhibition may represent a mechanism for polyphosphoinositides to regulate their own synthesis.

We previously described the purification of a membrane-bound diacylglycerol kinase highly selective for sn-1-acyl-2-arachidonoyl diacylglycerols (Walsh, J. P., Suen, R., Lemaitre, R. N., and Glomset, J. A. (1994) J. Biol. Chem. 269, 21155-21164). This enzyme appears to be responsible for the rapid clearance of the arachidonaterich pool of diacylglycerols generated during stimulusinduced phosphoinositide turnover. We have now shown phosphatidylinositol 4,5-bisphosphate to be a potent and specific inhibitor of arachidonoyl-diacylglycerol kinase. Kinetic analyses indicated a K i for phosphatidylinositol 4,5-bisphosphate of 0.04 mol %. Phosphatidic acid also was an inhibitor with a K i of 0.7 mol %. Other phospholipids had only small effects at these concentrations. A series of multiply phosphorylated lipid analogs also inhibited the enzyme, indicating that the head group phosphomonoesters are the primary determinants of the polyphosphoinositide effect. However, these compounds were not as potent as phosphatidylinositol 4,5-bisphosphate, indicating some specificity for the polyphosphoinositide additional to its total charge. Five other diacylglycerol kinases were activated to varying degrees by phosphatidylinositol 4,5-bisphosphate and phosphatidic acid, suggesting that inhibition by acidic lipids may be specific for the arachidonoyl-DAG kinase isoform. Given the presumed role of arachidonoyl-diacylglycerol kinase in the phosphoinositide cycle, this inhibition may represent a mechanism for polyphosphoinositides to regulate their own synthesis.
Diacylglycerol kinases catalyze the ATP-dependent phosphorylation of sn-1,2-diacylglycerol (DAG) 1 to phosphatidic acid (PA) (1)(2)(3). As such, they are widely regarded as attenuators of the DAG signaling and protein kinase C activation that occur during stimulus-induced PI turnover (4). The recent identification of a specific DAG kinase essential for PI-mediated invertebrate visual transduction is a striking confirmation of the involvement of DAG kinases in the PI cycle (5).
It has recently become evident that DAG kinases are a diverse family of isoenzymes (2). The first of these to be purified and cloned was an 82.6-kDa isoform expressed predominantly in brain and thymus (6,7). Several homologs of this enzyme also have been cloned, each of which has its own highly specific pattern of expression in cells and tissues (8 -12). Additional DAG kinases, which appear distinct from the cloned isoforms described above, also have been reported (13)(14)(15)(16)(17)(18)(19)(20)(21). However, detailed enzymologic data on these are not available at this time. Our laboratory has described and purified a membranebound DAG kinase highly selective for DAG molecular species containing arachidonate as the sn-2 fatty acyl moiety (21)(22)(23)(24). This activity can be distinguished from other DAG kinases by a variety of enzymologic properties in addition to its substrate specificity (21)(22)(23)(24). Arachidonoyl-DAG kinase activity varies widely between different tissues, but it is detectable in all cells and tissues we have examined (21)(22)(23)(24). Given the marked enrichment of animal cell phosphatidylinositols in arachidonate at the glycerol sn-2 position, the substrate selectivity of arachidonoyl-DAG kinase suggests a special role for this isoform in PI biosynthesis (21)(22)(23)(24).
This multiplicity of DAG kinases is reminiscent of the diversity seen in the protein kinase C and PI-specific phospholipase C isoenzyme families. As with the phospholipases C and protein kinases C, this diversity most likely reflects multiple mechanisms of enzyme regulation in the cells expressing the various DAG kinase isoforms (26 -28). The detailed mechanisms of DAG kinase regulation are, however, poorly understood. The 82.6-kDa DAG kinase and its homologs are thought to translocate to membranes during cell activation and are activated in vitro by Ca 2ϩ and phosphatidylserine (29). Limited evidence for activation of some DAG kinases by reversible protein phosphorylation also has been reported (30 -32).
Acidic phospholipids have been shown to modulate the in vitro activities of several membrane enzymes involved in stimulus transduction (33)(34)(35)(36)(37)(38)(39)(40)(41)(42)(43). Since levels of these lipids fluctuate during cell stimulation, this modulation is thought to reflect regulation of the enzymes by the in vivo membrane microenvironment. We have now examined the effects of acidic phospholipids on arachidonoyl-DAG kinase in a well characterized, mixed micellar assay system. Phosphatidylinositol 4,5-bisphosphate (PIP 2 ) was shown to be a potent and specific inhibitor of arachidonoyl-DAG kinase. Phosphatidylinositol 4-phosphate (PIP) and PA also were inhibitors, but at 18-fold higher concentrations. Five other DAG kinases were examined and found to be activated to varying degrees by PIP 2 and PA, suggesting that inhibition by acidic lipids may be a specific property of the arachidonoyl-DAG kinase isoform. Given the likely role of arachidonoyl-DAG kinase in PI biosynthesis, these data suggest that inhibition by polyphosphoinositides may represent a mechanism for feedback regulation of this enzyme by PIP 2 , its presumed final product.
Assay of Diacylglycerol Kinase-The DAG kinase assays employed represent minor modifications of our previously described methods (24). For a standard assay, an appropriate volume of DAG stock solution was evaporated under a stream of nitrogen in a glass test tube. To the DAG droplet were added: 50 l of 4 ϫ assay buffer, detergent, phospholipid, histone, dithiothreitol, water, and enzyme to a final volume of 180 l. The reaction was initiated by the addition of 20 l of [␥-32 P]ATP solution. The standard assay contained, in a volume of 200 l, 7.5 mM Triton X-100, 7.5 mM Triton X-114, 50 mM triethanolamine⅐HCl, pH 7.5, 100 mM NaCl, 1 mM MgCl 2 , 0.1 mM [␥-32 P]ATP, 0.3 mM (2.0 mol %) DAG, 1 mM EGTA, 1 mM dithiothreitol, 1 g of histone, 0.75 g of 2,6-di-tertbutyl-4-methylphenol, 10 M diethylenetriaminepentaacetic acid, and enzyme. sn-1-Stearoyl-2-arachidonoylglycerol was the DAG species employed in all assays. For some assays, 50 mM octyl-␤-D-glucopyranoside (octylglucoside) was employed instead of Triton. When phospholipids or other amphiphiles were added to the assays, the concentration of detergent was reduced to keep the total micellar concentration of detergent plus amphiphile equal to 15 mM in the Triton assays and 25 mM in the octylglucoside assays. For assays containing Triton, the detergent monomer concentrations were negligible, and the decrease in total detergent was simply equal to the amphiphiles added. For assays containing octylglucoside, monomeric detergent was assumed to decrease in proportion with the micellar mol fraction (44). The monomer concentration for pure octylglucoside used in these calculations was assumed to be 25 mM (45). Reactions were allowed to proceed for 10 min at 25°C and terminated by the addition of 3.0 ml of CHCl 3 /ethanol (2:1 (v/v)) containing 1.0 mg of dihexadecylphosphate and 1.0 mg of sorbitan trioleate. The organic phase was washed 3 times with 2.0 ml of 1.0% HClO 4 , 0.1% H 3 PO 4 in H 2 O/ethanol (4:1 (v/v)). The volume of the final organic phase was 2.25 ml. Incorporation of 32 P into PA was determined by Cerenkov counting 1.2 ml of this CHCl 3 phase. The Triton X-100/X-114 mixture was used instead of Triton X-100 because of the more favorable surface dilution behavior of the mixture (24). The concentrations of ATP and DAG employed were near the apparent K m for these substrates, where activity should be maximally sensitive to effects of the phospholipids being studied (24). The use of ethanol instead of methanol in the extractions and the addition of sorbitan trioleate prevented the highly polar lipids used in these studies from forming interfacial emulsions that would have impaired product recovery. The Triton X-100/X-114 stock solution was prepared as described previously, but using 29 g of Triton X-100 and 25 g of Triton X-114 in place of the 50 g of Triton X-100 (24). The cloud point of the Triton X-100/X-114 mixture was 49°C. Phospholipid stock solutions were generally 20 mol % in detergent, either Triton or octylglucoside, containing, additionally, 0.2 mol % 2,6-di-tert-butyl-4-methylphenol and 0.1 mM diethylenetriaminepentaacetic acid. Concentrations of phospholipid solutions were confirmed by determination of organic phosphorus (46). Stock solutions of DAG, ATP, 4 ϫ assay buffer, dithiothreitol, and histone were prepared as described previously (24). All assays were linear with the time and protein concentrations employed, and in all cases the 32 P reaction products comigrated as a single spot with authentic phosphatidic acid on thin-layer chromatography, which was performed as described previously (47). All DAG kinase activities presented are averages of at least duplicate determinations, which in all cases agreed within 10%. All kinetic curves were fitted directly to rate equations by nonlinear least squares regression using algorithms supplied with the SigmaPlot graphics package.
Diacylglycerol Kinases-Arachidonoyl-DAG kinase was purified to apparent homogeniety from bovine testis as described previously (24). Protein concentrations of the arachidonoyl-DAG kinase used in this work were too low to be reliably assayed, and activities are therefore expressed as pmol/min/l of the original pooled fractions (24). Testis cytosol DAG kinase, and cytosolic DAG kinases from 3T3 cells were also prepared as described previously (24). The 3T3 cell DAG kinases, referred to in this work as 3T3 DAG kinase A and 3T3 DAG kinase B, are identical to the type I and II enzymes described by Stathopoulos et al. (15). Porcine thymus cytosol DAG kinases were prepared according to the procedure of Sakane et al. through the DEAE cellulose step (14).
The activity eluting at 100 mM NaCl, referred to in this work as thymus DAG kinase A, is the 82.6-kDa brain enzyme subsequently cloned by that group (7). The activity eluting at 200 mM NaCl is referred to as thymus DAG kinase B.
Organic Syntheses-Procedures for preparation of the acidic phospholipid analogs used in this work are described below. Electrospray ionization mass spectrometry (ESI-MS), including tandem mass spectrometry (MS/MS), of the products was performed as described by Kerwin et al. (48) except that acidic compounds were injected in 5 mM methanolic HCl. Neutral compounds were injected in 5 mM methanolic ammonium acetate. Characteristic ions are expressed as m/z (relative intensity). Some structures were additionally confirmed by 1 H NMR at 200 MHz in Me 2 SO-d 6 , 20% deuterium chloride (20:1 (v/v)). Chemical shifts are reported as ␦ (parts/million) relative to tetramethylsilane. Oxalate-impregnated plates for TLC of multiply phosphorylated compounds were prepared as described by Cunningham et al. (49).
To prepare mono-O-hexadecylpentaerythritol trisphosphate, pentaerythritol (70 g, 0.51 mol) was partially acetylated with acetic acid (89 ml, 1.55 mol) by refluxing under a Dean-Stark trap in 150 ml of benzene containing 2 g of toluenesulfonic acid. After 24 h, 28 ml (1.55 mol) of water had collected in the trap, and the reaction was terminated. The mixture was deionized with 50 ml of Dowex AG 501-X8. Evaporation of the benzene yielded 134 g of clear, syrupy product (theoretical 136 g). To prepare the hexadecyl ether, acetylated pentaerythritol (20 g, 76 mmol) was stirred with 80 ml of dimethylformamide and 80% NaH (1.1 g, 37 mmol) added. After 1 h, H 2 evolution had ceased, and 1-bromohexadecane (10 g, 33 mmol) was added. The reaction was stirred for an additional 2 h and then quenched by addition of 150 ml of ice water. This mixture was extracted 3 times with 50 ml of diethyl ether, and the pooled extracts passed over a 50-ml bed of silicic acid. To remove the acetate blocking groups, the ether was evaporated, and the residue dissolved in 100 ml of methanol. Concentrated HCl (5 ml) was added, and the mixture refluxed. Methyl acetate (azeotrope with methanol, b.p. 54°C) was allowed to distill over. After 30 min, evolution of methyl acetate had ceased. The mixture was refluxed for an additional hour, after which, the methanol was evaporated. The product was precipitated at Ϫ20°C from H 2 O/ethanol (1:1 (v/v)) and washed at Ϫ20°C with petroleum ether/diethyl ether (2:1 (v/v)). The residue was dissolved in 80 ml of CHCl 3 and extracted with 80 ml of H 2 O/ethanol (4:1 (v/v)) followed by 60 ml of H 2 O. Ethanol (9 ml) was added to the CHCl 3 , and the solution passed over a 50-ml bed of silicic acid. The solvents were evaporated, and the product was crystallized from petroleum ether/ diethyl ether (1:1 (v/v)). The final yield was 4.7 g. The product was homogeneous by TLC on silicic acid (R F 0.08, benzene/ethyl ether/acetic acid, 25: Some additional product could be obtained by concentrating the mother liquor, but it was contaminated with a small amount of faster migrating material presumed to be the dihexadecyl ether. To prepare the trisphosphorylated product, monohexadecylpentaerythritol (2.0 g, 5.6 mmol), dimethylaminopyridine (0.5 g, 4 mmol), and triethylamine (4.0 ml, 28.7 mmol) were dissolved in 20 ml of CHCl 3 and added slowly to diphenylchlorophosphate (4.0 ml, 25 mmol) in 20 ml of CHCl 3 . The reaction was stirred for 1 h and then quenched by pouring slowly, with stirring, into 100 ml of 0.5 M NaHCO 3 together with an additional 30 ml of CHCl 3 . The CHCl 3 phase was washed with 125 ml of 0.5 M NaHCO 3 and passed over a 10-ml bed of silicic acid. The product migrated as a single spot on TLC (R F 0.75, CHCl 3 /toluene/ethanol, 20:20:5 (v/v)). To remove the phenyl protecting groups, the CHCl 3 was evaporated, and the syrupy product was dissolved in 180 ml of ethanol/ H 2 O/acetic acid (4:1:1 (v/v)). 0.5 g of PtO 2 was added, and the mixture was reduced with H 2 at 25°C and 60 psi in a Parr hydrogenation apparatus. Hydrogen uptake was largely complete after 30 min, and the reaction was continued overnight. A total of 137 mmol of H 2 was consumed (theoretical 142 mmol). The catalyst was removed by filtration, and the solution was concentrated to about 15 ml by evaporation.
This material was dissolved in 100 ml of H 2 O. The pH was adjusted to 11.0 by addition of 10 N NaOH, and the product was precipitated at 0°C by the addition of 100 ml of methanol. The precipitate was collected by filtration and washed successively with 50% ice-cold aqueous methanol, absolute ethanol 3 times, ethyl acetate 2 times, and diethyl ether 3 times. Yield was 3.86 g (95%). The final product migrated as a single spot on oxalate-impregnated TLC plates (R To prepare 2-hexadecylglycerol-1,3-bisphosphate, 2-hexadecylglycerol was first prepared from 1-bromohexadecane and 1,3-benzylidene glycerol according to the procedure of Serdarevich and Carroll (50) except that the alkylation reaction was carried out in NaH/dimethylformamide as described above. The product migrated as a single spot on TLC (R F 0.14, benzene/diethyl ether/acetic acid, 25 Hexadecyl phosphate was prepared by phosphorylation of hexadecyl alcohol with excess phosphoryltriimidazole as described previously (51) and crystallized as the monosodium salt from ethanol/water. The final product migrated as a single spot on TLC with the expected R F (51). Structures of these polyphosphorylated lipid analogs are shown in Fig. 1.
Critical Micelle Concentrations-Critical micelle concentrations (CMCs) of the multiply phosphorylated lipid analogs were determined in 100 mM NaCl by monitoring the enolic tautomer of benzoylacetanilide at 320 nm as described by Shoji et al. (52). The measured CMCs were not dependent on the benzoylacetanilide concentration, and under no condition did the micelles contain greater than 1 mol % of the probe. The benzoylacetanilide (Aldrich) was crystallized 2 times from hexane/ diethyl ether/ethanol, 1:1:1, prior to use.
Triton Phase Partitioning-Partitioning of phosphorylated lipid analogs between the aqueous and micellar pseudophases of the DAG kinase assay mixture was determined by preparing 1 mol % solutions of the analogs in assay buffer (minus ATP) and inducing formation of two phases by raising the temperature to 55°C, which is above the cloud point of the Triton X-100/Triton X-114 mixture (53). The upper aqueous phase was removed and replaced with an equal volume of detergentfree buffer, and the mixture was cooled to 25°C to allow reformation of a single phase (53). Partitioning of the phosphorylated amphiphiles was then assessed by assaying the two solutions for Triton (⑀ 275 ϭ 1.49 mM Ϫ1 cm Ϫ1 , isopropanol/H 2 O, 1:1 (v/v)) and organic phosphorus (46).  Fig. 2. The two PA species gave identical inhibition curves, indicating that the presence of an sn-2 arachidonoyl group is not required for this effect. The inhibition curve with PIP was identical to that of PA (data not shown). Phosphatidylinositol 4,5-bisphosphate was an 18-fold more potent inhibitor than PA (Fig. 2). We have previously shown that other phospholipids have negligible effects on arachidonoyl-DAG kinase at these concentrations (24). Phosphatidylinositol, in particular, caused minimal inhibition at less than 5 mol % (24).

Inhibition of Arachidonoyl-DAG Kinase by Acidic
Addition of 20 mol % phosphatidylcholine had no effect on the PIP 2 inhibition. The addition of up to 50 M Ca 2ϩ or 5 mM Mg 2ϩ also was without effect. In the presence of 20 mol % octadecyltrimethylammonium chloride, 50% inhibition occurred at 1.4 mol % PIP 2 and 75% inhibition at 2.2 mol %. 2 Presumably, the decreased inhibitory potency of PIP 2 in OTAC reflects complexation by the quaternary amine. However, it does demonstrate that PIP 2 is capable of inhibiting the enzyme even in micelles bearing a net positive charge. Examination of surface dilution behavior by varying the micelle concentration while holding the surface concentrations of PIP 2 and 18:0 -20:4 DAG constant demonstrated that activity in the presence of 0.04 or 0.10 mol % PIP 2 was dependent on the PIP 2 surface concentration and independent of micelle concentration (data not shown). This is consistent with our previous observation that several other phospholipids do not alter the surface 2 Octadecyltrimethylammonium chloride is also an activator of DAG kinases (24). The stimulation of arachidonoyl-DAG kinase by OTAC in this assay system is 2.2-fold both at 10 and 20 mol %. Assuming each PIP 2 binds five octadecyltrimethylammonium ions, binding of the quaternary amine cannot account for the effect of PIP 2 . dilution behavior of arachidonoyl-DAG kinase (24). Inositol polyphosphates, including D-myo-inositol 1,4,5-trisphosphate and inositol hexaphosphate had no effect on arachidonoyl-DAG kinase activity at 1.0 mM, nor did they reverse the inhibition by PIP 2 (data not shown).
Inhibition of Arachidonoyl-DAG Kinase by Other Multiply Phosphorylated Amphiphiles-The foregoing observations suggest that arachidonoyl-DAG kinase inhibition by PIP 2 is highly specific for this phospholipid. Moreover, the inability of PI to inhibit at these concentrations and the much weaker inhibition by PIP suggest that the phosphomonoesters are major structural determinants of this effect. None of the naturally occurring lipids examined, however, is as highly charged as PIP 2 , and it was possible that addition of any highly charged amphiphile to the assay would cause similar inhibition. To test this, amphiphilic compounds bearing one, two, or three phosphate monoesters in their headgroups ( Fig. 1) were prepared and examined with arachidonoyl-DAG kinase. As shown in Fig. 3, increasing the number of phosphates in the headgroup did increase DAG kinase inhibition. However, these analogs were not as potent as PA and PIP 2 . Concentrations of the mono-and bisphosphorylated analogs required for 50% inhibition were 3and 7-fold higher, respectively, than the concentrations of PA and PIP 2 required for this level of inhibition. Even the trisphosphorylated analog was less than half as potent a DAG kinase inhibitor as PIP 2 . These results strongly indicate that the inhibition is specific in some way for the structure of PIP 2 and not just a function of its total charge.
The bis-and trisphosphorylated amphiphiles were freely soluble in water, behaving as detergents. It was important to show under the assay conditions used that these compounds incorporate completely into Triton micelles, as failure to do so would render the calculated mol fractions invalid. The CMC in 100 mM NaCl of monohexadecyl-pentaerythritol trisphosphate was 34 M, and for 2-hexadecylglycerol bisphosphate it was 16 M. The CMC of PIP 2 under these conditions was less than 2 M. Assuming ideal partitioning between the aqueous and micellar pseudophases (44), greater than 99% of these amphiphiles must be in the micelles. 3 Partitioning of these com-pounds into Triton micelles was also estimated by cloud point separation as described under "Experimental Procedures." These results indicated that greater than 98% of the bis-and trisphosphorylated analogs and approximately 80% of the dodecyl phosphate were in the micelles 4 (data not shown). The inability of these compounds to inhibit arachidonoyl-DAG kinase as effectively as PIP 2 cannot, therefore, be due to a failure to partition into the micelles.
Mechanism of PIP 2 Inhibition-Double-reciprocal plots of arachidonoyl-DAG kinase dependence on 18:0 -20:4 DAG and mate the partitioning of the multiply phosphorylated amphiphiles into the micellar pseudophase of the DAG kinase reaction mixture. 4 The Triton phase partitioning was performed at 55°C. At this temperature, the partitioning of the ionic amphiphiles into mixed micelles is expected to be less than at 25°C (55). Therefore, these measurements probably underestimate the extent of incorporation. MgATP are shown in Fig. 4. The intersection of all the plots on the 1/DAG axis suggests a random order equilibrium mechanism, with K m ϭ K s ϭ 153 M for MgATP and K m ϭ K s ϭ 3.3 mol % for 18:0 -20.4 DAG. Random order of addition is the only type of equilibrium mechanism that can cause this type of kinetic behavior (56). A nonequilibrium, steady state mechanism is excluded by our previous observation that octadecyltrimethylammonium bromide is a partially noncompetitive activator of arachidonoyl-DAG kinase (24). Given these kinetics, the activator must be accelerating some rate-limiting step after binding of both substrates (56). In its absence, binding of substrates must thus be in a near equilibrium state, excluding any type of nonequilibrium mechanism for the kinetics in Fig. 4. Double-reciprocal plots of the effects of PIP 2 on the 18:0 -20:4 DAG dependence are shown in Fig. 5A. Inhibition was strictly noncompetitive with a K i of 0.04 mol %. The noncompetitive kinetics with respect to DAG are consistent with the absence of selectivity for an sn-2 arachidonoyl group in PA (Fig. 2). Strictly competitive kinetics were observed when the PIP 2 inhibition was examined with respect to MgATP, again with a K i of 0.04 mol % (Fig. 5B). Inhibition by PA also was strictly competitive with MgATP but with a K i of 0.7 mol %, indicating that the mechanism is the same as that of PIP 2 (Fig. 6). This result also implies that octadecyltrimethylammonium bromide cannot be activating the enzyme by complexation of PA (generated by the reaction) and consequent reversal of PA mediated inhibition. Since inhibition by PA is competitive with MgATP, reversal of any such inhibition should decrease the apparent K m for this substrate. However, we have shown previously that activation of arachidonoyl-DAG kinase by octadecyltrimethylammonium cation is due to an increase in V m and that the apparent K m for MgATP is unchanged (24). Such a mechanism is also excluded by the observation that chlorpromazine, another cationic amphiphile, readily reversed PA inhibition but did not activate DAG kinase in the absence of added PA (data not shown).
Effects of Acidic Phospholipids on Other DAG Kinase Isoforms-Diacylglycerol kinases are now known to be a large family of isoenzymes (2). It was thus of interest to examine whether inhibition by PIP 2 or PA is seen with other DAG kinases. The octylglucoside assay was used for these experiments because other DAG kinases tend to have low activity in the Triton (24). As shown in Fig. 7A, five cytosolic DAG kinase isoforms were activated, in some cases markedly, by phosphatidic acid. As shown in Fig. 7B, PIP 2 also activated these five DAG kinases. Only the arachidonoyl-DAG kinase was inhibited by PIP 2 or PA. The same general effects were seen in Triton (Table I), although, as expected, the activities were lower than in octylglucoside. This apparent absolute specificity for the arachidonoyl-DAG kinase isoform is strong additional evidence that the inhibition reflects a specific interaction of PIP 2 with the enzyme and is not an artifact of the assay method employed. Expressed as a mole fraction, PIP 2 was a less potent arachidonoyl-DAG kinase inhibitor in octylglucoside than in Triton. The behavior of PIP 2 in octylglucoside as a substrate for phospholipase C-␥1 is also very different from its behavior in Triton (34). With the phospholipase C, kinetic effects of enzyme tyrosine phosphorylation could be observed only in a Tritonbased assay and not in octylglucoside (34). Given the many differences between these two detergents, the significance of these observations is unknown. With all five of the other DAG kinase isoforms, the maximal activities observed with octylglucoside/PA were similar to those obtained in the deoxycholate assay normally used for these enzymes (6,24), raising the possibility that PA and PIP 2 may regulate the activities of these enzymes in vivo. This may indicate a small degree of cooperativity or simply be a nonspecific effect due to formation of mixed micelles containing multiple PA molecules, which would be expected at the higher PA concentrations used. The apparent K i for PA, obtained by extrapolation, was 0.7 mol % in two separate experiments. The PA concentrations used were: , 0.9 mol %; , 0.6 mol %; q, 0.3 mol %; E, no PA. DISCUSSION We have shown arachidonoyl-DAG kinase to be potently and specifically inhibited by PIP 2 . The K i for PIP 2 inhibition of the enzyme was 0.04 mol %. Assuming a Triton aggregation number of 140, this corresponds to one PIP 2 molecule/20 micelles in the assay mixture. 5 As arachidonoyl-DAG kinase inhibition was dependent on the PIP 2 surface concentration and independent of micelle concentration, the mechanism must involve intramicellar binding of PIP 2 to the enzyme. Rapid equilibration of the solubilized lipids and enzyme between micelles presumably permits this low level of PIP 2 to effectively inhibit the enzyme (24,59). The inability of D-myo-inositol 1,4,5-trisphosphate and inositol hexaphosphate to inhibit arachidonoyl-DAG kinase is further evidence that PIP 2 must incorporate into the micelle before it can inhibit the enzyme. The observation that PI does not inhibit arachidonoyl-DAG kinase below 5 mol % points to the phosphomonoesters as major determinants of the PIP 2 effect. This conclusion is also supported by the potent inhibition seen with the multiply phosphorylated analogs. These analogs were, however, clearly less potent than PIP 2 itself, indicating that some structural feature of PIP 2 not related to its total charge is also important. The competitive inhibition kinetics with respect to MgATP raise the possibility that the PIP 2 phosphomonoester binding sites overlap the binding site for trisphosphate moiety of the MgATP, although an allosteric mechanism cannot be excluded. The observations that five other DAG kinase isoforms are not inhibited by PIP 2 or PA are strong additional evidence that this is a specific property of arachidonoyl-DAG kinase and not a nonspecific effect of the mixed micellar assay.
Acidic lipids, including polyphosphoinositides and phosphatidic acid, have been shown to activate several membrane associated enzymes of stimulus transduction, including PI-specific phospholipase C-␥1 (33), several protein kinase C isoforms (35)(36)(37), phospholipase D (38), ADP-ribosylation factor 1 (39), and an ADP-ribosylation factor GTPase-activating protein (40). Phosphatidylinositol-4-phosphate 5-kinase appears to be inhibited by PIP 2 and activated by PA (41,42). Casein kinase I also appears to be inhibited by PIP 2 (43). In some of these cases, as with the DAG kinases, these effects were shown to occur only with certain isoforms of these enzymes (37,42). The surface concentration of PIP 2 in Triton required for half-maximal DAG kinase inhibition, 0.04 mol %, compares well with the apparent K m of phospholipase C-␥1 for PIP 2 in Triton, which is 0.03-0.3 mol %, depending on the activation state of the enzyme and presence or absence of a PA cofactor (33,34). Also, activation of protein kinase C by PIP 2 in Triton occurs at 0.08 -0.2 mol % (35,36). In other studies, the use of very different assay methods renders comparisons with the present work difficult. Taken together, however, these results do suggest that signaling cascades may be regulated by the lipid microenvironment of subcellular membranes in which they are located. Inasmuch as PIP 2 and PA undergo rapid flux during cell stimulation (60 -65), this hypothesis seems not unreasonable.
The stimulation of some other DAG kinases by PIP 2 and PA, while raising the possibility that these enzymes may be activated by PA or polyphosphoinositides, does not support a role for these isoforms in regulating the level of PIP 2 . Stimulation of these isoforms by phosphatides may reflect a mechanism for their recruitment to specific intracellular membranes during cell stimulation (66,67). However, detailed studies of this phenomenon have yet to be performed.  7. Effects of acidic phospholipids on DAG kinase isoenzymes. Panel A shows the effect of PA on the activities of arachidonoyl-DAG kinase and five other DAG kinase isoforms. Assays were performed in the octylglucoside assay described under "Experimental Procedures" in the presence of the indicated surface concentrations of PA. Phosphatidic acid concentrations greater than 20 mol % could not be tested because of insolubility in the assay mixture above this concentration. Panel B shows the effect of PIP 2 on these same DAG kinases. In these assays, the octylglucoside contained 10 mol % phosphatidylserine in addition to the indicated concentrations of PIP 2 . The basal activities are slightly lower with the arachidonoyl-DAG kinase and slightly higher with the other isoforms because of the phosphatidylserine. Assays at 1.0 and 3.0 mol % PIP 2 in octylglucoside without phosphatidylserine showed the same pattern of inhibition and activation. To facilitate comparison, activities in both panels are expressed as percentages of the maximal activity observed with that isoform in the octylglucoside/PA assay. The DAG kinase isoforms used, maximal activities, and mol fractions of PA at which these activities were observed are as follows: q, arachidonoyl-DAG kinase, 33 pmol/min/l, 0 mol %; E, 3T3 DAG kinase A, 1.9 nmol/min/mg, 7.5 mol %; ⌬, 3T3 DAG kinase B, 1.0 nmol/min/mg, 20 mol %; Ⅺ, thymus DAG kinase A, 25 nmol/min/mg, 20 mol %; , thymus DAG kinase B, 12.3 nmol/min/mg, 20 mol %; f, testis cytosol DAG kinase, 8.2 nmol/min/mg, 20 mol %.
The marked enrichment of animal cell phosphatidylinositols in arachidonate strongly suggests a role for arachidonoyl-DAG kinase in PI synthesis (21,25). Given this, the observed inhibition may represent a mechanism for PIP 2 to regulate its own synthesis. Work from several groups has shown that the arachidonate-enriched DAG pool that arises during stimulusinduced PI turnover is rapidly phosphorylated by a diacylglycerol kinase and that much of this PA is ultimately converted back to PI (21, 60 -65). Other DAG species are phosphorylated much more slowly (60 -65). As other DAG kinases do not exhibit fatty acyl selectivity, these results strongly suggest arachidonoyl-DAG kinase is the isoform responsible for this DAG phosphorylation. Phosphorylation of arachidonoyl-DAG under these conditions coincides with a transient drop in the cellular level of PIP 2 (63, 68 -70). Regulation of arachidonoyl-DAG kinase by PIP 2 feedback is thus entirely consistent with available in vivo data.
The present work demonstrates PIP 2 to be a potent and specific in vitro inhibitor of arachidonoyl-DAG kinase. To what extent the mixed micellar system reflects the in vivo regulation of this enzyme remains to be determined. Arachidonoyl-DAG kinase is an integral membrane protein and is likely to be highly localized in the cells in which it is found. The intramicellar mechanism of PIP 2 inhibition in vitro suggests an intramembranous in vivo mechanism. Any regulation of this enzyme by PIP 2 would then require the phosphoinositide to be present in the same membrane compartment as the DAG kinase. The subcellular localization of the PI cycle, while likely to be of critical importance to its regulation, is understood only to a limited extent (71,72). Future studies of in vivo DAG kinase regulation will have to address these issues of compartmentation and multiple enzyme isoforms. Cloning of arachidonoyl-DAG kinase, which is in progress in our laboratory, will facilitate new approaches to these questions. The interaction with PIP 2 will be amenable to site-directed mutagenesis, and the subcellular location of the enzyme will be open to immunocytochemical approaches. Such studies promise a better understanding of PI-mediated stimulus transduction in animal cells.