Phosphatidic acid that accumulates in platelet-derived growth factor-stimulated Balb/c 3T3 cells is a potential mitogenic signal.

We developed a monoclonal antibody specific to phosphatidic acid (PA). Using this antibody, a novel method to quantify trace amounts of PA was achieved. With the method, PA can be measured in the range of 20-500 pmol. We applied this method to quantify changes in PA levels in Balb/c 3T3 cells stimulated by platelet-derived growth factor. PA contents were very low in quiescent cells and dramatically increased with time up to 15 min. On the other hand, a biphasic diacylglycerol (DG) increase was found. The early phase showed a transient small peak of DG at 30 s followed by a decrease to 1 min. In the second phase, DG accumulated gradually but very markedly up to 15 min. Treatment with propranolol, a PA phosphohydrolase inhibitor, enhanced the accumulation of PA and inhibited the formation of DG in the second phase. However, R59022, a DG kinase inhibitor, did not influence the accumulation of DG or PA, suggesting that platelet-derived growth factor stimulates mainly phospholipase D-catalyzed hydrolysis of phospholipids rather than phospholipase C-catalyzed hydrolysis in the second phase. PA, even after contaminating lyso-PA was removed, could stimulate DNA synthesis, although lyso-PA was 25 times more potent. Moreover, phospholipase D was found to be a much stronger mitogen than phospholipase C. Phospholipase D treatment caused a biphasic accumulation of PA. PA levels reached a maximum at 1 h, and then decreased between 1 and 2 h; finally, there was a gradual elevation up to 10 h. In this case, there was no significant DG accumulation. On the other hand, phospholipase C treatment induced only DG accumulation without any significant change in PA. These results indicate that PA accumulation, rather than an increase in DG, correlates well with mitogenesis.

We developed a monoclonal antibody specific to phosphatidic acid (PA). Using this antibody, a novel method to quantify trace amounts of PA was achieved. With the method, P A can be measured in the range of 20-600 pmol. We applied this method to quantify changes in PA levels in Balb/c 3T3 cells stimulated by plateletderived growth factor. P A contents were very low in quiescent cells and dramatically increased with time up to 15 min. On the other hand, a biphasic diacylglycerol (DG) increase was found. The early phase showed a transient small peak of DG at 30 s followed by a decrease to 1 min. In the second phase, DG accumulated gradually but very markedly up to 15 min. Treatment with propranolol, a PA phosphohydrolase inhibitor, enhanced the accumulation of P A and inhibited the formation of DG in the second phase. However, R59022, a DG kinase inhibitor, did not influence the accumulation of DG or PA, suggesting that plateletderived growth factor stimulates mainly phospholipase D-catalyzed hydrolysis of phospholipids rather than phospholipase C-catalyzed hydrolysis in the second phase. PA, even after contaminating lyso-PA was removed, could stimulate DNA synthesis, although lyso-P A was 25 times more potent. Moreover, phospholipase D was found to be a much stronger mitogen than phospholipase C. Phospholipase D treatment caused a biphasic accumulation of PA. P A levels reached a maximum at 1 h, and then decreased between 1 and 2 h; finally, there was a gradual elevation up to 10 h. In this case, there was no significant DG accumulation. On the other hand, phospholipase C treatment induced only DG accumulation without any significant change in PA. These results indicate that PA accumulation, rather than an increase in DG, correlates well with mitogenesis.
Many mitogens produce their signals by stimulating the hydrolysis of cellular phospholipids. A stimulation of the breakdown of phosphatidylinositol 4,5-bisphosphate (PIPZ)' to two second messengers, inositol 1,4,5-trisphosphate (IP3) * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ To whom correspondence should be addressed.  (2), and DG activates protein kinase C (3), a kinase shown to be a key enzyme in mitogenic signaling (4,5). Therefore, DG is considered to be an important second messenger regulating cell growth and transformation. More recently, it has been found that many agonists stimulate the hydrolysis of phosphatidylcholine (PC) in addition to PIPp hydrolysis (reviewed in Ref. 6). Moreover, the hydrolysis of PC rather than PIPz has been shown to be the major source of DG formed in stimulated cells (7) (reviewed in Ref. 8). In this case, hydrolysis of PC by either or both phospholipase C and D could lead to DG production, the latter through PA.
PA also accumulates to a level comparable with that of DG in response to growth factors (9)(10)(11)(12). This lipid is not only a crucial intermediate in de nouo lipid synthesis, it also appears to play an important role in cell proliferation (13,14). PA has also been shown to evoke various cellular responses such as the promotion of Caz+ entry into cells and the mobilization of intracellular Ca2+ (15)(16)(17), and also inhibits adenylate cyclase, apparently through interaction with Gi (18). Recently, PA has been found to increase the activity of the ras GTPaseinhibiting protein (19) and to inhibit the activity of ras GTPase-activating protein (GAP) (20). These results suggest that PA may link growth factor receptor signaling to a cellular Ras that is critical in the control of proliferation (14).
We have been developing antibodies to phospholipids related to signal transduction, such as PIP and PIP2, in order to evaluate the role of these lipids (21,22).
We report here a very sensitive method based on specific anti-PA antibodies to quantify levels of PA and its variants upon mitogenic stimulation. Our results point out that PA accumulation, rather than DG increase, correlates well with mitogenesis.
Cell Culture-Balblc 3T3 cells were maintained a t 37 "C in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10%  calf serum. The cells were plated at 6 X IO5 cells/9-cm dish. Two days after plating, subconfluent cells were serum-starved by washing 2 times with DMEM containing 5 pg/ml transferrin and 100 pg/ml bovine serum albumin and then incubated for 24 h at 37 "C in the same medium.
Production of Anti-PA Monoclonal Antibody-Anti-PA antibody was made by a method similar to that described before (21,22). Briefly, BALB/c mice were immunized with liposomes containing dimyristoyl phosphatidylcholine, cholesterol, PA (egg, 0.5 mg), and lipid A (molar ratio, 1:1.5:1:0.08) every 2 weeks for 9 months. The immune spleens were fused with P3-X63-Ag8 cells using polyethyleneglycol (M, 3,350). Two weeks after cell hybridization, hybridoma supernatants were evaluated by ELISA. The hybridoma cells secreting anti-PA antibody were doubly cloned by limiting dilution. The cells were inoculated into pristane-primed mice, and the resulting ascites fluid was used to measure PA content. The antibody was found to be IgG2b by subclass analysis.The specificity of the antibody was evaluated by ELISA and TLC immunostaining as described before (21,22).
Mass Analysis of PA-Cells treated with growth factors, phospholipase C or D, were incubated at 37 "C for the indicated time. The incubations were terminated by aspirating the medium, washing with phosphate-buffered saline (PBS), and immediately adding 1.5 ml of ice-cold methanol. The cells were scraped and transferred to tube and the dishes were further washed with 1.5 ml of chloroform/methanol (1/2, v/v). Then chloroform (1.5 ml) was added to the tube, and the lipids were extracted with 1 N HCl as described before (22). The lipids were spotted on thin layer chromatography (TLC) plates (Polygram, Macherey-Nagel) and developed in chloroform/methanol/water (60/ 25/4, v/v/v). The plates were soaked overnight in PBS containing 3% bovine serum albumin and 1% polyvinylpyrrolidone 25, and then treated for 2 h a t room temperature with anti-PA antibody ascites diluted 500-fold with PBS. The plates were washed 4-5 times with PBS containing 0.05% Tween 20, reacted with peroxidase-conjugated anti-mouse immunoglobulins, and stained with a Konica staining kit (Konica, Tokyo). The content of PA was measured by an area thymidine (1 pCi/ml) was added to each well, and incubation was continued for 3 h. The cells were then washed with PBS and treated with 0.25% trypsin followed by the addition of 10% trichloroacetic acid. The cells were harvested by a cell harvester, and thymidine incorporation was measured. For the detection of saturated and unsaturated PAS, a TLC plate was first immunostained with KA-1 and then sprayed with phosphorous-detection reagent (24).
Other Methods-IPa and DG contents were measured using IP, and DG assay kits (Amersham Corp.).

RESULTS
Specificity of Anti-PA Antibody-Anti-PA antibody KA-1 binds to PA very specifically. On ELISA, KA-1 showed no cross-reactivity with lyso-PA, phosphatidylglycerol, PI, PIP, PIP, or 1,2-diolein, and very little with cardiolipin (7%) and tetraoleoyl bis-PA (5%) (Fig. lA). TLC immunostaining showed that antibody KA-1 reacted only with PA when rat brain total phospholipids (30 pg of phosphorus/spot) were used as antigens (Fig. lB, left). Furthermore, KA-1 reacted with dioleoyl PA, dilinoleoyl PA, 1-palmitoyl-2-oleoyl PA, or 1-stearoyl-2-arachidonoyl PA, but not with dimyristoyl PA, dipalmitoyl PA, distearoyl PA, or didecanoyl PA on both ELISA (Table I) and TLC (Fig. 1C). These results suggest that the antibody reacts with PA molecules that contain an unsaturated fatty acid at the AP position. Next, we tried to establish a method for measuring PA content by TLC immunostaining. By this method, PA content could be measured in the range of 20-500 pmol (Fig. 2). To ascertain the reliability of this assay, a fixed amount of PA (10, 20, 50, or 100 pmol) was added to cells (1 x lo6 cells) and the PA was extracted and quantified. PA levels shifted in proportion to the amount of PA added to the original cells with a recovery of greater than 90%. Using this assay system, we were able to measure trace amounts of PA without the use of a radioactive precursor. DG and PA Formation in Response to PDGF-To determine how DG and PA were formed in response to PDGF-stimulation, the time course for the changes in DG and PA levels was examined. As shown in Fig. 3, increase of DG was biphasic, which is consistent with previous results (22). The early phase showed a transient small peak at 30 s followed by a decrease to 1 min. In the second phase, which occurred after 1 min, DG accumulated gradually but very markedly to about 600 pmol/106 cells up to 15 min. In the previous report, we showed that DG in the early peak was derived from PIP2, while most of the DG in the second phase came from other lipids (22). On the other hand, PA content increased steadily with time to about 250 pmol/106 cells from 1 to 15 min. In addition, PA content was very low (less than 20 pmol/106 cells) in growth-arrested cells, while that of DG was fairly high (about 320 pmol/106 cells).
Effect of Propranolol and R59022 on PDGF-induced DG and PA Formation-Propranolol is known to be an inhibitor of PA phosphohydrolase and has been used to examine the pathway of DG (25). R59022 has been used as an inhibitor of DG kinase (26). Here, we used these drugs to investigate the route of PDGF-induced formation of DG and PA. As shown in Fig. 4, propranolol treatment enhanced accumulation of PA and inhibited the formation of DG in the second phase (15 min after stimulation). However, R59022 treatment did not influence the accumulation of DG and PA even at 15 min after stimulation. These results suggest that DG in the second phase is derived from PA rather than formed directly by phospholipase C.
Mitogenic Effect of PA, Lyso-PA, DG, Phospholipase C, Phospholipase D, and PDGF-Exogenously added PA and lyso-PA have been shown to stimulate DNA synthesis. However, recently Jalink et al. (27) reported that commercial sources of PA were contaminated with lyso-PA. Therefore, we used purified PA to examine whether PA by itself has any mitogenic effect. As shown in Fig. 5A, purified PA induced DNA synthesis, but its potency was about 25 times less than lyso-PA. There is still the possibility that PA is degraded to lyso-PA during incubation. However, the mitogenic effect of PA was found to be much stronger than that of DG (I,% diolein or ~-cr-l-oleoyl-2-acetoyl-sn-3-glycerol). If PA formed endogenously in response to PDGF-stimulation is as mitogenic as exogenously added PA, phospholipase D treatment should cause an enhancement of DNA synthesis. Indeed, phospholipase D was found to be a strong mitogen compared with phospholipase C (Fig. 5B), suggesting that PA or lyso-PA rather than DG plays an important role in PDGF-induced cell growth. Time Course of DG and PA Formation in Response to Phospholipase C, Phospholipase D, and PDGF-Next, we examined the time course of changes in DG and PA levels when cells were treated with PDGF, phospholipase C, or phospholipase D in order to see the correlation between PA levels and cell proliferation (Fig. 6). In this case, a long term time course experiment was planned. In PDGF-stimulated cells, both DG and PA contents reach a maximum at 15 min and then gradually decrease up to 4 h. When cells were treated with phospholipase D, PA showed a biphasic change. In the first phase, PA levels reached a maximum at 1 h followed by a decrease to 2 h. In the second phase, PA increased gradually up to 10 h. In this case, there was no significant DG accumulation. On the other hand, when cells were treated with phospholipase C, DG accumulation occurred without any marked increase in PA. DG accumulated rapidly up to 1 h and then remained stable for a long time. Lyso-PA Formation by PDGF Treatment-In case of PDGF-stimulation of cells prelabeled with ["P]PO$-,, 32P incorporation into lyso-PA increased by 1.8-fold, while incorporation into PA increased 3.8-fold. Radioactivity incorporated into lyso-PA was 2,137 cpm, which was 3.4% of the radioactivity incorporated into PA (Table 11). In this isotope labeling method, PDGF-induced enhancement of PA synthesis was only 3.8-fold, while lo-15-fold enhancement was detected with the mass level-determining method using anti-PA antibody (Fig. 3). This discrepancy may be accounted for by the difference of PA species formed in response to PDGF. Indeed Fig. 7A showed that two distinct PAS (arrows 3 and 4 ) and lyso-PAs (arrows 1 and 2) were detected on autoradiography and the spots which could be visualized by iodine vapor (arrows 2 and 4 ) and be stained with KA-1 (arrow 4 ) were increased by PDGF stimulation. Arrows 3 and 4 were confirmed to be saturated and unsaturated fatty acid-containing PAS by using authentic PAS (Fig. 7B). These results suggest that most PA synthesized in response to PDGF are PA species having unsaturated fatty acid at the A, position. Therefore it is very important to measure unsaturated fatty acid PA mass levels to know the change in PA.

DISCUSSION
Many growth factors have been shown to enhance the hydrolysis of PIP, to IP, and DG (1). However, Hill et al. (31) reported that PDGF-induced activation of PIPZ-phospholipase C is not required for the induction of DNA synthesis in C3H10T1/2 cells. On the other hand, it is becoming clear that such mitogens also stimulate the hydrolysis of phosphatidylcholine, most probably by phospholipase C and phospholipase D (6). Phospholipase C hydrolyzes phosphatidylcholine to generate DG that activates protein kinase C directly. It has also been shown that phospholipase D participates in phosphatidylcholine breakdown in stimulated cells. In this case, PA is liberated and some of the PA are converted to DG through the action of PA phosphohydrolase. Although PA might be an important lipid involving in cell proliferation, it has so far been difficult to measure PA mass levels because its level is very low. We developed an antibody specific to PA and applied it to quantify PA levels in an investigation of the biological function of PA. Together with radioisotope technique, we found that PAS formed by PGDF stimulation were PA species which contained unsaturated fatty acid at A2 position. Furthermore, by an application of this method to examine which pathway (phospholipase Cor phospholipase D-catalyzed phosphatidylcholine degradation) is more prominent in PDGF-stimulated cells, we conclude that the phospholipase D-mediated pathway predominates in PDGF-stimulated Balb/c 3T3 cells, because propranolol inhibits PDGFinduced DG formation (Fig. 4). Therefore, it seems likely that PDGF stimulates phospholipase D rather than phospholipase C activity in Balb/c 3T3 cells in the second phase where prolonged accumulation of PA is observed. This idea is supported by the fact that R59022, a DG kinase inhibitor, does not cause an increase in the accumulation of DG in stimulated cells. However, Larrodela et al. (28) reported that PDGF activates only DG formation without any detectable change in PA in Swiss 3T3 cells and suggested that phospholipase C is the main route activated by PDGF. In Balb/c 3T3 cells, a very marked increase in PA in response to PDGF was observed. Therefore, activation pathway, via either phospholipase C or phospholipase D, may differ among cells or tissues. Our data also show that quantity of PA enhanced by the stimulation is almost same as that of DG, although, the PA level is very low in arrested cells (lower than 20 pmol/106 cells), in contrast to abundantly existed DG level (around 300 pmol/106 cells) (Fig. 3). These results suggest that the dramatic elevation in PA levels upon growth factor stimulation might play a critical role in triggering cell proliferation. We showed that PA accumulation in stimulated cells also continues as long as DG accumulation. Considering that PA and DG are potent mitogens, it is essential to determine whether DG or PA plays the more important role in cell growth. To clarify whether endogenously formed PA stimulates DNA synthesis without concurrent DG formation, we examined the effect of phospholipase D on cell growth. As shown in Fig. 6, phospholipase D treatment caused a very marked increase in PA formation without DG formation. Even at 10 h after phospholipase D treatment, significant DG accumulation was not observed. Similarly, phospholipase C treatment caused only DG accumulation. These data suggest that conversion of P A to DG is very low in phospholipase D-treated cells in contrast with PDGF-stimulated cells. As shown in Fig. 5, phospholipase D treatment causes a stronger mitogenic response than phospholipase C treatment, suggesting that PA by itself can induce DNA synthesis without DG formation. In this case, lyso-PA, rather than PA, may play the major role in cell proliferation since lyso-PA seems to be formed in PDGF-stimulated cells (Table 11, Fig. 7). However, we could not detect a mitogenic effect of PA when PA was injected into cells (29). This discrepancy may be explained by the fact that injected PA forms micelles which are not effectively incorporated into membranes. Recently, PA has been demonstrated to have an inhibitory effect on Ras GAP and to increase the activity of the GTPase-inhibiting protein (20, 19). The interaction of PA with these two regulatory proteins may increase a Ras activity by keeping Ras as GTP complex and stimulate cell proliferation. Indeed, Ras activation has been demonstrated in PDGF-stimulated cells (30). Therefore, a dramatic increase in PA in PDGF-stimulated cells may cause Ras activation leading to cell proliferation. However, it is not clear whether PA induces cell growth through direct interaction with Ras or not, although PA-induced mitogenesis is inhibited by the injection of anti-Ras antibody into cells (14). This problem remains to be resolved in future.