Regulation of phosphatidic acid phosphohydrolase by epidermal growth factor. Reduced association with the EGF receptor followed by increased association with protein kinase Cepsilon.

An important component of receptor-mediated intracellular signal transduction is the generation of lipid second messengers. Lipid second messenger production is a complex process involving a variety of regulatory enzymes that control the intracellular response to the extracellular signal. Phosphatidic acid (PA) is generated in response to phospholipase D and can be converted to other lipid second messengers including diacylglycerol (DG) and lysophosphatidic acid. PA is converted to DG by PA phosphohydrolase (PAP). We report here that PAP activity can be detected in epidermal growth factor (EGF) receptor immunoprecipitates. Following treatment with EGF, there is a substantial reduction in the PAP activity that co-precipitates with the EGF receptor. The loss of EGF receptor-associated PAP activity occurs with a concomitant increase in PAP activity associated with the ε isoform of protein kinase C (PKC). The PAP activity associated with PKCε was dependent upon the PKC co-factors phosphatidylserine and DG but was independent of the kinase activity of PKCε. These data suggest a novel signaling mechanism for the regulation of lipid second messenger production and implicate PAP as an important regulatory component for lipid second messenger production in receptor-mediated intracellular signaling.

One of the earliest responses to extracellular signals is the metabolic conversion of membrane phospholipids into intracellular second messengers (1,2). The best studied second messenger-generating system is the activation of phospholipase C␥ (PLC␥), 1 which catalyzes the hydrolysis of phosphatidylinositol 4,5-bisphosphate (PIP 2 ) to diglyceride (DG) and inositol 1,4,5-trisphosphate (3). Another mechanism for generating DG is through the activation of type D phospholipases (PLD) that catalyze the hydrolysis of phosphatidylcholine (PC) to phosphatidic acid (PA) and choline (4). PA can be converted to DG by a PA phosphohydrolase (PAP). Regulation of PAP is likely to be important because PA can be converted to biologically active molecules other than DG. Lyso-PA can be generated from PA and has been shown to mediate several biological activities (5)(6)(7). PA has been reported to enhance ARF-GTPase-activating protein, which inactivates the Ras superfamily GTPase ARF (8), which has been shown to be a co-factor for the activation of PLD (9,10). PA has been reported to inhibit Ras-GTPase-activating protein (11,12), and Ras has also been implicated in the activation of PLD (13,14). PA stimulates phosphatidylinositol 4-kinase activity (15), which leads to the production of PIP 2 , the substrate for PLC␥ and a co-factor for PLD (9,16). Another suspected role for PA is in facilitating both membrane fusion and vesicular transport (8). Thus, control of cellular PA levels is likely to be important for the control of responses to extracellular stimuli mediated by phospholipid metabolism.
We previously demonstrated that EGF induces an increase in DG that was generated from PC-derived PA (17), suggesting a role for PAP. In this report, we describe a novel mechanism for regulation of lipid second messenger production in response to EGF.

EXPERIMENTAL PROCEDURES
Cells and Cell Culture Conditions-A431 cells, obtained from the American Type Culture Collection were maintained in Dulbecco's modified Eagle medium supplemented with 10% bovine calf serum (Life Technologies, Inc.). Cells were routinely placed in serum-free medium prior to treatment with EGF to reduce background PAP activity. Materials-[ 3 H]PA was obtained from New England Nuclear. PIP 2 , PC, phosphatidylserine (PS), DG were obtained from Sigma. The anti-EGF receptor antibody was obtained from Oncogene Science; antibodies specific for the ␣, ⑀, and PKC isoforms were obtained from Transduction Laboratories.
Assay of PAP Activity-A431 cells were grown to confluence in 150-mm culture dishes and then placed in serum-free medium overnight as described previously (17). The cells are then washed in cold isotonic buffer, scraped from the plates, and suspended in 2 ml of hypotonic buffer A (20 mM Tris-HCl, pH 7.3; 5 mM NaCl; 5 mM Na 2 HPO 4 ; 1 M ZnCl; 1 mM EDTA; 1 mM EGTA; 0.2 mM phenylmethylsulfonyl fluoride; 1 g/ml leupeptin; 1 g/ml aprotinin; 400 M Na 4 VaO 4 ). The suspended cells were then broken by Dounce homogination (15 strokes with type B pestle). Disrupted cells were centrifuged at 500 ϫ g for 10 min to clear nuclei and unbroken cells. The post-nuclear homogenate was used as the whole cell lysate. The post-nuclear homogenate can be separated into membrane and cytosolic fractions by centrifugation at 100,000 ϫ g for 60 min. The supernantent is saved as the cytosolic fraction. The membrane fraction is recovered from the pellet by resuspending in buffer A. Liposomes are made by mixing lipids in chloroform, drying under a stream of nitrogen, and resuspending in assay buffer (20 mM Tris-HCl, pH 7.3, 5 mM Na 2 HPO 4 , 140 mM NaCl, 0.1 mM EGTA, 2 mM MgCl 2 , 0.5 mM CaCl 2 ) with vortexing and then sonicating for 3 min. Unless otherwise indicated, the liposome suspension was 5 g/ml dipalmitoyl-PA (ϳ0.1 mol fraction), 10 g/ml PC (from egg yolk), 10 g/ml PIP 2 , 30 g/ml PS, 3 g/ml DG (Sigma), and 20,000 cpm [ 14 C]dipalmitoyl-PA (100 Ci/mmol). The sonicated solution is kept at room temperature for 2 h to equilibrate. The PAP assay is initiated by the addition of 20 l of homogenate (40 g of protein) to prewarmed (37°C) liposome suspension. Where appropriate, fatty acid-free bovine serum albumin was added to maintain constant protein concentration (1 mg/ml) in the reaction mixture. The final volume of the reaction mixture was 140 l. The mixture is incubated at 37°C for 15 min unless otherwise indicated. The reaction consumed less than 20% of the [ 14 C]PA. The reaction is terminated by the addition of 800 l of chloroform/methanol (6:2 v/v). DG is resolved by thin layer chromatography as described previously (18).
Immunoprecipitations-membrane fractions were treated with buffer A containing 1% Triton X-100 and 140 mM NaCl. Insoluble material was cleared by centrifugation (12,000 ϫ g, 30 min). The 1 ml of supernatant (1 mg of protein) was then incubated with the indicated antibody (2 g) overnight at 4°C with shaking. The antigen-antibody complexes were then recovered with 20 l of either protein A-Sepharose or protein G-agarose suspensions (Santa Cruz Biotechnology) (2 h, 4°C with shaking) by microcentrifugation and washing four times in buffer A containing 1% Triton X-100 and 140 mM NaCl. Liposome suspensions were then added to the immunoprecipitates, and PAP activity was determined as described above.

RESULTS
PAP Activity in A431 Cells-There are currently no direct assays for PAP activity in intact cells. However, PAP activity can be measured in vitro by adding cell extracts or lysates containing PAP to liposomes containing radiolabeled PA and examining the conversion of PA to DG. An in vitro PAP assay based on strategies developed by Brindley and co-workers (19,20) was used to investigate PAP activity in A431 cells. In A431 cells PAP activity could be detected in both cytosolic and membrane fractions (Table I). It has been reported that in several different cell types, two distinct PAP activities exist that can be distinguished on the basis of a differential sensitivity to Nethylmaleimide (NEM) and Mg 2ϩ (19 -21). In A431 cells, we found that virtually all of the PAP activity, both membrane and cytosolic, was sensitive to NEM (Table I). This difference was not likely due to differences in experimental protocol, because we also find that in 3Y1 rat fibroblasts 80% of the PAP activity was resistant to NEM treatment (data not shown). NEM-resistant PAP was reported to be insensitive to Mg 2ϩ , whereas the NEM-sensitive PAP activity was stimulated by Mg 2ϩ (19). As shown in Table I, the NEM-sensitive PAP activity from both membranes and cytosol was dependent upon Mg 2ϩ . Thus, we were unable to detect significant biochemical differences between the membrane and cytosolic PAP activities in A431 cells. These properties are characteristic of the PAP previously des-ignated PAP1 (19). As expected, the PAP activity was sensitive to the amphiphilic cations propranolol, chlorpromazine, and sphingosine, which have been shown previously to inhibit PAP activity (19).
Liposome composition can also affect in vitro PAP activity (19). We therefore examined the effect of the phospholipid composition of the liposomes used in the PAP assay. As shown in Table II, PAP activity was stimulated by the inclusion of PIP 2 , although PAP activity was not dependent upon PIP 2 as demonstrated for PLD (9,16). PIP 2 is a substrate for PLC␥, and it has been reported that PA can stimulate PLC␥ activity (22). The PA stimulation of PLC␥ activity was dependent upon Ca 2ϩ (22); however, as shown in Table I, Ca 2ϩ had no effect upon the PAP activity in the presence of PIP 2 . As shown in Table II, the addition of DG (the product of PLC␥) to the liposomes actually inhibited the PAP activity in the presence of PIP 2 . Thus, the effect of PIP 2 is not likely due to an effect of PLC␥ on PAP activity. The addition of PS and DG was also stimulatory for PAP activity (Table II). The effect of PS and DG, which are co-factors for PKC activity, suggested the possible PKC involvement in PAP activity. The effect of EGF on the PAP activity from A431 cells was examined, and although EGF reproducibly elevated PAP activity, the effect varied with the composition of the liposomes (Table II).
PAP Activity Co-immunoprecipitates with the EGF Receptor and Is Reduced upon Treatment with EGF-The EGF receptor has previously been demonstrated to be in complexes with molecules that transduce intracellular signals (1,23,24). We therefore investigated whether PAP activity could be detected in EGF receptor immunoprecipitates. Membranes were isolated from EGF-treated A431 cells and untreated controls. Detergent lysates of the membrane fractions were treated with antibodies raised against the EGF receptor. Immunoprecipitates were recovered, and PAP activity was examined. As shown in Fig. 1A, PAP activity could be easily detected in the EGF receptor immunoprecipitates. However, in contrast to expectations, activity was lost upon EGF treatment (Fig. 1A). The loss of PAP activity associated with the EGF receptor in response to EGF was both time- (Fig. 1A) and dose-dependent ( Fig. 1B) with the loss in activity being detectable within 1 min and at 1 nM EGF. This reduction in PAP activity was not due to a reduced ability of the antibody to precipitate EGF-treated receptor, because there were no differences in the amount of EGF receptor precipitated from the EGF-treated and untreated cells (Fig. 1C). This suggested that either the PAP activity was TABLE I PAP activity in A431 cells PAP activity in membrane and cytosolic fractions from A431 cells was determined as described under "Experimental Procedures." PAP assays were performed using liposomes containing PA, PC, and PIP 2 as shown in Table II. The effect of removing either Mg 2ϩ or Ca 2ϩ , adding NEM (2 mM), and heat denaturation (55°C, 20 min) on PAP activity is shown. The effect of the amphiphilic cations propranolol (2 mM), sphingosine (0.1 mM), and chlorpromazine (0.3 mM) on membrane PAP activity was determined by inclusion of these compounds in the liposome preparations. The relative effects are shown in parentheses after normalizing to the PAP activity found in the complete assay mix. The data represent the average of duplicates (Ϯ range) from a representative experiment repeated three times. ND, not determined.  reduced or that PAP protein was released from the receptor in response to EGF. Because we could detect overall increases in total PAP activity in response to EGF, we considered the first possibility unlikely. Thus, the data in Fig. 1 suggest that the PAP associated with the EGF receptor is released in response to EGF treatment. Table II it was shown that PAP activity was enhanced by the presence of PS and DG in the liposomes. This suggested an involvement of PKC. Jaken and co-workers have characterized a number of proteins that interact directly with PKC isoforms (25). We therefore investigated the possibility that PAP activity could be associated with PKC isoforms present in A431 cells. The predominant PKC isoforms present in A431 cells are the ␣, ⑀, and isoforms. 2 PAP activity in PKC isoform immunoprecipitates from lysates of A431 cells that had been either treated or untreated with EGF was determined. As shown in Fig. 2A, a very strong EGF-dependent PAP activity was detected in PKC⑀ but not PKC␣ or immunoprecipitates. The lack of PAP in the PKC␣ and immunoprecipitates was not due to an inability to precipitate PKC␣ and because these isoforms could be detected at levels comparable with the ⑀ isoform in the immunoprecipitates (data not shown). A PKC⑀ peptide against which the antibody was raised prevented precipitation of PAP with the PKC⑀ antibody ( Fig. 2A). The kinetics of association of PAP with PKC⑀ was just slightly behind the EGF-induced dissociation of PAP from the EGF receptor (Fig. 2B).   1. PAP activity in EGF receptor immunoprecipitates. A, the membrane fraction from A431 cells that were either untreated or treated with EGF (100 nM) for the indicated times were harvested as in Table I and then lysed with 1% Triton. The lysate was then incubated with an anti-EGF receptor antibody (12 h, 4°C). Antigen-antibody complexes were recovered with protein G-agarose and added directly to the complete liposome mixture described in Table II, and PAP activity in the immunoprecipitates was determined. The data presented are from a representative experiment that was repeated three times. The dose response (B) for the effect seen in A was determined as shown. C, the amount of EGF receptor immunoprecipitated in EGF-treated (100 nM, 5 min) and untreated A431 cells was determined by Western blot analysis as described previously (29). The PAP activity in these anti-EGF receptor immunoprecipitates is also presented.

FIG. 2. PAP activity can be co-immunoprecipitated with PKC⑀ after EGF treatment.
A, the membrane fraction from A431 cells that were either untreated or treated with EGF (100 nM, 5 min) were harvested and then lysed with 1% Triton as in Fig. 1. The lysate was then incubated with antibodies specific for the ␣, ⑀, and PKC isoforms (12 h, 4°C). The antigen-antibody complexes were recovered with protein A-Sepharose for PKC␣ and ⑀ and protein G-agarose for PKC, and PAP activity was determined as in Fig. 1. To establish that the effect was specific for PKC⑀, a PKC⑀-specific peptide against which the antibody was raised was included in the immunoprecipitation and the ability to immunoprecipitate PAP activity was determined. B, the time course for association of PAP activity with PKC⑀ after EGF treatment was determined as shown.

PAP Activity Associated with PKC⑀ Is Dependent upon the Presence of DG and PS in Liposomes, but Not PKC Kinase
Activity-The in vitro PAP activity from A431 cells was significantly enhanced when PS and DG were included in the liposomes used in the PAP assay (Table II). We therefore examined the effect of DG and PS in the liposomes when the PKC⑀associated PAP activity was determined. In the absence of DG and PS, we were unable to detect any PAP activity in the PKC⑀ immunprecipitates in response to EGF (Fig. 3). The DG and PS requirement suggested that PKC⑀ activity is important in the in vitro liposome assay. However, as shown in Fig. 3, neither ATP nor ATP␥S had any effect on the PAP activity associated with PKC⑀, suggesting that the kinase activity is not required. Consistent with this result, staurosporine, which inhibits PKC by competing for ATP binding, also had no effect upon the in vitro PAP activity in the PKC⑀ immunoprecipitates (Fig. 3B). These surprising results suggest that although the PKC cofactors DG and PS are required to observe the PAP activity associated with PKC⑀, the kinase activity of PKC⑀ is apparently not required for the in vitro PAP activity. DISCUSSION In this report, data have been presented suggesting a novel signaling mechanism in which PAP associated with the EGF receptor becomes associated with PKC⑀. Upon EGF treatment, PLC␥ is activated (26). The generation of DG by PLC␥ could lead to the observed increase in membrane localization of PKC⑀ in response to EGF. PAP could then become associated with membrane-bound PKC⑀. Although the data presented here do not demonstrate that PAP activity is actually elevated in response to EGF, the kinetics of the EGF-induced association of PAP with PKC⑀ correlate well with the EGF-induced increase in PC-derived DG reported previously (17). It is possible that there are no significant changes in the specific activity of PAP in response to EGF and that regulation is accomplished by the induced change in the association of PAP from the EGF receptor to an association PKC⑀. We were unable to detect PAP activity in anti-phosphotyrosine immunoprecipitates, 2 suggesting that the PAP is not a direct substrate of the EGF receptor. Thus, a molecular mechanism for the putative dissociation of PAP from the EGF receptor and association with PKC⑀ remains to be determined.
The requirement for PS and DG in order to see the PKC⑀associated PAP activity suggested a requirement for PKC activity. However, because ATP, ATP␥S, nor the ATP analog staurosporine had any effect on the PKC⑀-associated PAP activity, it is possible that the PS and DG are functioning to localize PKC⑀ and the associated PAP to the liposomes where the PA substrate is present. In this regard, it is of interest that PKC has been shown to stimulate PLD activity via an ATPindependent mechanism (27,28). These observations, along with data presented here, suggest that PKC(s) may have kinase-independent roles in the regulation of intracellular signals. It is possible that PKC⑀ may serve to either allosterically modify PAP or to localize PAP to appropriate membrane locations where PA substrate is localized. PKC could be important for positioning PAP in a molecular complex where the enzyme functions as a component of a signaling complex that generates DG, which activates the associated PKC, which would then phosphorylate other substrates.