Exogenous phospholipase D generates lysophosphatidic acid and activates Ras, Rho and Ca2+ signaling pathways.

BACKGROUND: Phospholipase D (PLD) hydrolyzes phospholipids to generate phosphatidic acid (PA) and a free headgroup. PLDs occur as both intracellular and secreted forms; the latter can act as potent virulence factors. Exogenous PLD has growth-factor-like properties, in that it induces proto-oncogene transcription, mitogenesis and cytoskeletal changes in target cells. The underlying mechanism is unknown, although it is generally assumed that PLD action is mediated by PA serving as a putative second messenger. RESULTS: In quiescent fibroblasts, exogenous PLD (from Streptomyces chromofuscus) stimulated accumulation of the GTP-bound form of Ras, activation of mitogen-activated protein (MAP) kinase and DNA synthesis, through the pertussis-toxin-sensitive inhibitory G protein Gi. Furthermore, PLD mimicked bioactive lysophospholipids (but not PA) in inducing Ca2+ mobilization, membrane depolarization and Rho-mediated neurite retraction. PLD action was mediated by Iysophosphatidic acid (LPA) derived from Iysophosphatidylcholine acting on cognate G-protein-coupled LPA receptor(s). There was no evidence for the involvement of PA in mediating the effects of exogenous PLD. CONCLUSIONS: Our results provide a molecular explanation for the multiple cellular responses to exogenous PLDs. These PLDs generate bioactive LPA from pre-existing Iysophosphatidylcholine in the outer membrane leaflet, resulting in activation of G-protein-coupled LPA receptors and consequent activation of Ras, Rho and Ca2+ signaling pathways. Unscheduled activation of LPA receptors may underlie, at least in part, the known pathogenic effects of exogenous PLDs.

Platelet-derived growth factor activation of mitogen-activated protein kinase depends on the sequential activation of phosphatidylcholine-specific phospholipase C, protein kinase C-zeta and Raf-1.

The mechanism of Raf-1 activation by platelet-derived growth factor (PDGF) is poorly defined. We previously reported that, in Rat-1 fibroblasts, PDGF activates a phosphatidylcholine-specific phospholipase C (PC-PLC) and that the product, diacylglycerol, somehow activates protein kinase C-zeta (PKC-zeta). Both PC-PLC and PKC-zeta activities were required for PDGF activation of mitogen-activated protein kinase (MAPK). Now we report that MAPK activation by exogenous PC-PLC depends on Raf-1 activation. PKC-zeta co-immunoprecipitates with, phoshorylates and activates Raf-1, suggesting that in the PDGF- and PC-PLC-activated MAPK pathway, PKC-zeta operates in a signalling complex as a direct activator of Raf-1.

Involvement of phosphatidylcholine-specific phospholipase C in platelet-derived growth factor-induced activation of the mitogen-activated protein kinase pathway in Rat-1 fibroblasts.

The role of phosphatidylcholine (PC) hydrolysis in activation of the mitogen-activated protein kinase (MAPK) pathway by platelet-derived growth factor (PDGF) was studied in Rat-1 fibroblasts. PDGF induced the transient formation of phosphatidic acid, choline, diacylglycerol (DG), and phosphocholine, the respective products of phospholipase D (PLD) and phospholipase C (PC-PLC) activity, with peak levels at 5-10 min. PLD-catalyzed transphosphatidylation (with n-butyl alcohol) diminished DG formation at 5 min but not at later stages of PDGF stimulation. Phorbol ester-induced down-regulation of protein kinase C (PKC) completely blocked PLD activation but not the formation of DG and phosphocholine at 10 min of PDGF stimulation. Collectively, these data indicate that PDGF activates both PLD and PC-PLC. In contrast, epidermal growth factor did not activate PC-PLC in these cells, and it activated PLD only weakly. DG formation by itself, through Bacillus cereus PC-PLC treatment of cells, was sufficient to mimic PDGF in activation of MAPK independent of phorbol ester-sensitive PKC. Since PKC down-regulation blocked PDGF-induced PLD but not MAPK activation, we conclude that PLD is not involved in MAPK signaling. In contrast, MAPK activation by exogenous (bacterial) PLD was not affected by PKC down-regulation, indicating that signals evoked by exogenous PLD differ from endogenous PLD. D609 (2-10 microg/ml), an inhibitor of PC-PLC, blocked PDGF- but not epidermal growth factor-induced MAPK activation. However, D609 should be used with caution since it also affects PLD activity. The results suggest that PC-PLC rather than PLD plays a critical role in the PDGF-activated MAPK pathway.

Diacylglycerol kinase in receptor-stimulated cells converts its substrate in a topologically restricted manner.

The regulation of diacylglycerol (DG) kinase activity was studied in fibroblasts and Jurkat T cells. We questioned whether enzyme activity only depends on substrate availability or whether it requires receptor stimulation. To this end, we raised DG levels up to 15-fold by treatment of cells with bacterial phosphatidylinositol-specific phospholipase C (PLC). In detergent cell lysates, DG kinase was readily capable of converting this surplus of DG to phosphatidic acid (PA), but in intact cells the enzyme remained inactive. Stimulation of fibroblasts with bradykinin or endothelin and Jurkat cells with anti-CD3 resulted in DG kinase-mediated formation of PA, but its level was unaffected by PLC pretreatment. Likewise, in streptolysin O-permeabilized fibroblasts, where bradykinin stimulation in the presence of [gamma-32P]ATP and guanosine 5'-O-(thiotriphosphate) generates [32P]PA exclusively via DG kinase, PLC pretreatment did not affect the amount of [32P]PA formed. We conclude that DG kinase acts on DG generated by receptor stimulation, but not on DG generated by exogenous PLC. We propose a model in which DG kinase physically associates with endogenous PLC. Within this complex, receptor-induced DG would then be transmitted ("channeled") from endogenous PLC to the active site of DG kinase, whereas excess DG generated randomly in the plasma membrane by bacterial PLC is inaccessible to this catalytic site.

Rapid attenuation of receptor-induced diacylglycerol and phosphatidic acid by phospholipase D-mediated transphosphatidylation: formation of bisphosphatidic acid.

Generation and attenuation of lipid second messengers are key processes in cellular signalling. Receptor-mediated increase in 1,2-diacylglycerol (DG) levels is attenuated by DG kinase and DG lipase. We here report a novel mechanism of DG attenuation by phospholipase D (PLD), which also precludes the production of another (putative) second messenger, phosphatidic acid (PA). In the presence of an alcohol, PLD converts phosphatidylcholine (PC) into a phosphatidylalcohol (by transphosphatidylation) rather than into PA. We found in bradykinin-stimulated human fibroblasts that PLD mediates transphosphatidylation from PC (donor) to the endogenous 'alcohol' DG (acceptor), yielding bis(1,2-diacylglycero)-3-sn-phosphate (bisphosphatidic acid; bisPA). This uncommon phospholipid is thus a condensation product of the phospholipase C (PLC) and PLD signalling pathways, where PLC produces DG and PLD couples this DG to a phosphatidyl moiety. Long-term phorbol ester treatment blocks bradykinin-induced activation of PLD and consequent bisPA formation, thereby unveiling rapid formation of DG. BisPA formation is rapid (15 s) and transient (peaks at 2-10 min) and is also induced by other stimuli capable of raising DG and activating PLD simultaneously, e.g. endothelin, lysophosphatidic acid, fetal calf serum, phorbol ester, dioctanoylglycerol or bacterial PLC. This novel metabolic route counteracts rapid accumulation of receptor-induced DG and PA, and assigns for the first time a physiological role to the transphosphatidylation activity of PLD, that is signal attenuation.

Phospholipid metabolism in bradykinin-stimulated human fibroblasts. II. Phosphatidylcholine breakdown by phospholipases C and D; involvement of protein kinase C.

Bradykinin (BK) and phorbol 12-myristate 13-acetate (PMA) both stimulate the hydrolysis of phosphatidylcholine (PC) in human fibroblasts, resulting in the formation of phosphatidic acid (PA) and diacylglycerol (DG) (Van Blitterswijk, W.J., Hilkmann, H., de Widt, J., and Van der Bend, R.L. (1990) J. Biol. Chem. 266, 10337-10343). Stimulation with BK resulted in the rapid and synchronous formation of [3H]choline and [3H]myristoyl-PA from the correspondingly prelabeled PC, indicative of phospholipase D (PLD) activity. In the presence of ethanol or n-butanol, transphosphatidylation by PLD resulted in the formation of [3H]phosphatidylethanol or - butanol, respectively, at the cost of PA and DG formation. This suggests that PC-derived DG is generated via a PLD/PA phosphohydrolase pathway. A more pronounced but delayed formation of these products was observed by PMA stimulation. The Ca2+ ionophore ionomycin also activated PLD and accelerated (synergized) the response to PMA. Both [3H] choline and [3H]phosphocholine were released into the extracellular medium in a time- and stimulus-dependent fashion, without apparent changes in the high intracellular levels of [3H]phosphocholine. The protein kinase C (PKC) inhibitors staurosporin and 1-O-hexadecyl-2-O-methylglycerol inhibited BK- and PMA-induced activation of PLD. Down-regulation of PKC by long-term pretreatment of cells with phorbol ester caused a dramatic drop in background [3H]choline levels, while subsequent stimulation with BK, ionomycin, or PMA failed to increase these levels and failed to induce transphosphatidylation. From these results we conclude that PLD activation is entirely mediated by (downstream of) PKC. Unexpectedly, however, BK stimulation of these PKC-depleted cells caused a marked generation of DG from PC within 15 s, which was not seen in BK-stimulated control cells, suggesting PC breakdown by a phospholipase C (PLCc). We conclude that cells stimulated with BK generate DG via both the PLCc and the PLD/PA hydrolase pathway, whereas PMA stimulates mainly the latter pathway. BK stimulation of normal cells leads to activation of PKC and, by consequence, to attenuation of the level of PLCc-generated DG and to stimulation of the PLD pathway, whereas the reverse occurs in PKC-down-regulated cells.

Phospholipid metabolism in bradykinin-stimulated human fibroblasts. I. Biphasic formation of diacylglycerol from phosphatidylinositol and phosphatidylcholine, controlled by protein kinase C.

Stimulation of human fibroblasts with bradykinin (BK) results in the generation of diacylglycerol (DG) and phosphatidic acid (PA). Prelabeling of the cells with [3H]arachidonic acid and [14C]palmitic acid allowed us to quantitate these lipid second messengers and to determine their origin, i.e. DGi and PAi from 3H-enriched inositol phospholipids, and DGc and PAc from 14C-enriched phosphatidylcholine, respectively. BK elicited a biphasic DG response: a first peak at 10-15 s, containing DGi, followed by a second peak at 10-30 min, which is mainly DGc. The latter did not result from de novo lipid biosynthesis. BK also generated free [3H]arachidonate and, to a lesser extent, mono[3H]arachidonoylglycerol. BK stimulation rapidly increased PAi, much more so than PAc, suggesting that DGi, rather than DGc, is the preferred substrate for the enzyme DG kinase. Short pretreatment of the cells with phorbol 12-myristate 13-acetate (PMA) abolished the BK-induced breakdown of phosphoinositides, but did not affect the second-phase DGc level. PMA alone also elicited DGc formation, but more slowly, suggesting a different mechanism. Down-regulation of protein kinase C (PKC) by long term treatment with phorbol ester, prior to BK stimulation, resulted in (i) enhanced DGi and decreased PAi formation, suggesting that DG kinase activity is positively controlled by PKC; (ii) the unexpected manifestation of rapidly formed DGc; (iii) no change in the DGc levels obtained after 30-min BK stimulation, but complete suppression of PMA-induced DGc formation. In contrast, two inhibitors of PKC, staurosporin and 1-O-hexadecyl-2-O-methylglycerol, inhibited both BK- and PMA-induced DGc formation at 30 min, leaving the rapid response towards BK unaffected. The results suggest that the BK-induced rapid and later-phase DG formation and the PMA-induced DG formation are differentially controlled by PKC via mechanisms that differ in the susceptibility to down-regulation or inhibition of PKC.

Effect of dietary lipids on plasma lipoproteins and fluidity of lymphoid cell membranes in normal and leukemic mice.

Mice of the GR/A strain were fed four different isocaloric semipurified diets, enriched in either (1) saturated fatty acids (palm oil), or (2) polyunsaturated fatty acids (corn oil), or (3) palm oil plus cholesterol, or (4) a fat-poor diet containing only a minimal amount of essential fatty acids. We have studied the effects of these dietary lipids on the density profile and composition of the plasma lipoproteins and on the lipid composition and fluidity of (purified) lymphoid cell membranes in healthy mice and in mice bearing a transplanted lymphoid leukemia (GRSL). Tumor development in these mice occurred in the spleen and in ascites. While the fatty acid composition of the VLDL-triacylglycerols still strongly resembled the dietary lipids, the effects of the diets decreased in the order VLDL-triacylglycerols greater than HDL-phospholipids greater than plasma membrane phospholipids. Diet-induced differences in the latter fraction were virtually confined to the content of oleic acid and linoleic acid, and they were too small to affect the membrane fluidity, as measured by fluorescence polarization using the probe 1,6-diphenyl-1,3,5-hexatriene. Healthy mice were almost irresponsive to dietary cholesterol, but in the tumor bearers, where lipoprotein metabolism has been shown to be disturbed, the cholesterol diet caused a substantial increase in the low- and very-low density regions of both blood and ascites plasma lipoproteins. The cholesterol-rich diet also increased the cholesterol/phospholipid molar ratio and lipid structural order (decreased fluidity) in GRSL ascites cell membranes, but not in the splenic GRSL cell membranes. We conclude that the composition of plasma lipoproteins and cell membrane lipids in GR/A mice is subject to exquisite homeostatic control. However, in these low-responders to dietary lipids the development of an ascites tumor may lead to increased responsiveness to dietary cholesterol. The elevated level of membrane cholesterol thus obtained in GRSL ascites cells did not affect the expression of various cell surface antigens or tumor cell growth.

Accumulation of an alkyl lysophospholipid in tumor cell membranes affects membrane fluidity and tumor cell invasion.

Tumor cells grown in the presence of 1-O-alkyl-2-O-methylglycero-3-phosphocholine (AMG-PC) accumulated this ether lipid in their membranes. Depending on the cell type and the dose of the compound, up to 17% of the total phospholipids of the purified plasma membranes consisted of authentic AMG-PC. Extensive incorporation of the agent resulted in a decrease in plasma membrane fluidity and inhibition of tumor cell invasiveness in embryonic chick heart fragments. The extent of AMG-PC incorporation and fluidity change was not strictly correlated with the degree to which tumor cell invasion was inhibited.

A metabolite of an antineoplastic ether phospholipid may inhibit transmembrane signalling via protein kinase C.

In our search for the mechanisms by which the drug 1-O-alkyl-2-O-methylglycero-3-phosphocholine (AMG-PC) inhibits tumor growth and metastasis, we have detected a metabolite, 1-O-alkyl-2-O-methylglycerol (AMG), in membranes of MO4 mouse fibrosarcoma cells grown in the presence of the drug. Synthetic AMG inhibited the activation of highly purified human protein kinase C by diacylglycerol in the presence of phosphatidylserine. Furthermore, AMG also inhibited the receptor-specific binding of 3H-phorbol-12,13-dibutyrate to human HL-60 promyeloid leukemia cells in a dose-dependent fashion. AMG-PC was not effective or much less so in these assays. We suggest that interaction of the metabolite AMG with protein kinase C may inhibit stimulus-response coupling in tumor cells and may thus potentially contribute to the mechanism by which AMG-PC exerts its anticancer activities.

Quantitative contributions of cholesterol and the individual classes of phospholipids and their degree of fatty acyl (un)saturation to membrane fluidity measured by fluorescence polarization.

Steady-state fluorescence polarization (P) measurements, using the probe 1,6-diphenyl-1,3,5-hexatriene, in a large variety of well-defined liposomes at 25 degrees C allowed a quantitative description of the contributions of cholesterol, sphingomyelin, and (un)saturation of fatty acyl groups in the various phospholipids to the structural order (or the mutual affinity) of membrane lipids. The P values for liposomes prepared from lipid extracts of natural (purified) membranes of various origins could be more or less predicted (calculated) from the relative contributions of the individual lipid components. In all cases, the polarization varied with the cholesterol/phospholipid molar ratio (C/PL) according to the equation P = Pplat -(Pplat -Pzero) exp(-alpha C/PL), in which Pzero refers to the polarization without cholesterol and Pplat is a maximal plateau value, reached at a high C/PL (greater than 1). The "cholesterol-ordering coefficient" alpha of the phospholipids was found to increase with the fraction of sphingomyelin or dipalmitoylphosphatidylcholine molecules, indicating that the susceptibility of phospholipids to be ordered by cholesterol is increased by these compounds. Pzero increases curvilinearly with the fraction of either of these molecules. Pplat increases linearly with the fraction of saturated acyl chains for most phospholipids. Highly unsaturated fatty acyl chains (e.g., 20:4 and 22:6) strongly depress Pplat but not Pzero. The results suggest that such phospholipids are unlikely to associate with cholesterol and may thus create extremely fluid membrane domains. The disproportionation of cholesterol in the cell can be understood by the differing composition of the phospholipids in plasma membranes and endomembranes and their ordering susceptibility (affinity) toward cholesterol.

Catalytic hydrogenation of fatty acyl chains in plasma membranes; effect on membrane lipid fluidity and expression of cell surface antigens.

Optimal reaction conditions were established for hydrogenation of plasma membranes of living murine GRSL leukemia cells, using the water-soluble catalyst Pd(QS)2 (QS, sulphonated alizarine; C14H6O7NaS). Under these conditions more than 80% of the cells remained viable. Analysis by gas chromatography revealed that hydrogenation occurred predominantly in the 18:2, 20:4 and 22:6 fatty acyl chains of the membrane phospholipids. Under the same conditions hydrogenation was also performed in purified plasma membranes from GRSL cells and from rat liver, and in liposomes prepared from the total lipid extracts of these membranes. Hydrogenation increased the lipid structural order parameter in the membranes, as measured by fluorescence polarization. This increase was more pronounced in the liposomes (46%) than in the plasma membranes (17-25%). Hydrogenation increased the expression of a 15 kDa antigen on the surface of viable GRSL cells, as measured in a Fluorescence Activated Cell Sorter, using monoclonal antibodies. The expression of four other antigens, among which H-2k, was not or much less affected by this treatment.

Alterations in biosynthesis and homeostasis of cholesterol and in lipoprotein patterns in mice bearing a transplanted lymphoid tumor.

The membrane fluidity of murine lymphoid GRSL tumor cells has been shown to depend on their site of growth in the host. Tumor cells located in ascites, in contrast to those in the enlarged spleen, show a very high plasma membrane fluidity, mainly due to a decreased level of cellular (membrane) cholesterol. Yet, the rate of cholesterol biosynthesis in the tumor cells as estimated by the activity of HMG-CoA reductase and the incorporation of [14C]acetate into cholesterol was extremely high when compared to various lymphoid cells in normal control mice. Also the rate of biosynthesis and the cholesterol content in liver and spleen of tumor-bearing mice were substantially higher than in the organs of control mice. Blood plasma cholesterol, however, was decreased in tumor-bearing mice as a result of altered lipoprotein patterns. Outgrowth of the tumor was accompanied by a strong reduction in plasma high-density lipoproteins. Low-density lipoproteins became transiently increased, but eventually all lipoproteins, and consequently the plasma cholesterol content decreased to very low levels, especially so in the ascites plasma. The low transfer of [14C]cholesteryl ester-labeled lipoproteins between blood and ascites plasma after either intravenous or intraperitoneal injection suggested a hampered flow between the two compartments. Also apparent differences in cholesteryl ester fatty acid composition between lipoproteins of the blood and ascites plasma indicated the lack of a rapid equilibration between the two compartments. The results suggest that the limited availability of lipoproteins as an additional source of cholesterol to the rapidly growing ascites cells promotes their high membrane fluidity.

Differences in membrane lipid composition and fluidity of transplanted GRSL lymphoma cells, depending on their site of growth in the mouse.

GRSL lymphoma cells were isolated from various growth sites in the host. The relative membrane lipid fluidities of these cells and of normal lymphoid cells were estimated by fluorescence polarization, using the probe diphenylhexatriene and by measuring the (free) cholesterol/phospholipid molar ratio in whole cells. The results indicate that the membrane fluidity (reciprocal of the lipid structural order) of the lymphoma cells increases in the order of their location: peripheral blood less than spleen less than mesenterial lymph node less than ascites fluid. The membrane fluidities of normal lymphocytes from thymus, mesenterial lymph node and spleen were about the same, but higher than of peripheral blood lymphocytes, and between those of the lymphoma cells from lymph node and spleen. These results are confirmed by more extensive analysis on purified plasma membranes from the splenic and ascitic GRSL lymphoma cells and from normal splenocytes and thymocytes. The significantly higher lipid order parameter found in the GRSL plasma membrane isolated from the spleen as compared to those from the ascites cells could be fully explained by the differences measured in the major chemical determinants of the fluidity, i.e., the cholesterol/phospholipid ratio, the sphingomyelin content and the degree of saturation of the fatty acyl groups of the phospholipids. It was also found that the cholesterol/phospholipid ratio in erythrocyte membranes isolated from the peripheral blood of the tumor bearers was higher than in those from normal control mice. The observed differences in membrane fluidity between distinct subsets of tumor cells may be relevant to the sensitivity of these cells to immune attack or to drugs.

Rigid plasma-membrane-derived vesicles, enriched in tumour-associated surface antigens (MLr), occurring in the ascites fluid of a murine leukaemia (GRSL).

Extracellular membraneous vesicles of GRSL leukaemia cells were isolated from the ascites fluid bathing the cells in vivo, and from cell washes. Mammary tumour virus-induced antigens (MLr) expressed on the surface of the cells are enriched on these vesicles as compared to plasma membranes isolated from the cell homogenate. The lipid fluidity of the vesicles is much smaller than that of the plasma membranes, and the content of the pertinent lipid parameters, cholesterol and sphingomyelin, are accordingly greatly increased. The extracellular vesicles which are also enriched in sialic acid and 5'-nucleotidase are apparently derived from the plasma membrane, probably at least partly by exfoliation of selected parts or domains of the surface of living cells. An analogy between this shedding of vesicles, the formation of endocytotic vesicles and the budding of viruses is noted; all these processes select or assemble rigid lipid domains of the cell membrane. The possible participation of surface microvilli and sub-lethal autolysis in the process of shedding is discussed.

Effect on glutaraldehyde fixation on lectin-mediated agglutination of mouse leukaemia cells.

The effect of glutaraldehyde fixation on lectin-mediated agglutination of murine leukaemia (GRSL) cells was investigated using 2 assay methods which differed in the shear forces to which the agglutinated cells were subjected. First, lectin and cells were allowed to interact under conditions in which shear forces were minimized and the degree of agglutination was evaluated microscopically by the appearance and size of the cell aggregates. This assay demonstrated that concanavalin A (con A)-, wheat germ agglutinin (WGA)- or Ricinus communis agglutinin I (RCAI)-mediated cytoagglutination was unaffected (WGA and RCAI) or somewhat enhanced (con A) by prior fixation of the cells with glutaraldehyde. Secondly, an electronic particle counter was used to measure the disappearance of single cells and concomitant appearance of cell aggregates as a function of the lectin concentration. In this assay, in which the aggregated cells are subjected to significant shear forces during dilution and cell counting, agglutination of GRSL cells by each of the 3 lectins was drastically inhibited by prior fixation of the cells with glutaraldehyde. This assay also demonstrated enhanced nonlectin-induced cell aggregation after fixation. In both cytoagglutination assays about the same lectin concentration was required for threshold agglutination of unfixed cells. Comparatively, the results of the 2 cytoagglutination assays indicate that a fraction of the lectin-mediated bonds between unfixed cells is shear resistant and that fixation of the cells either weakens these bonds or inhibits their formation. Morphologically, cells prefixed with glutaraldehyde were sperical at all lectin concentrations, with a continuous dense distribution of cell surface-bound con A, labelled directly with haemocyanin or indirectly using the peroxidase-diaminobenzidine reaction. Unfixed cells showed angular and toadstool-shaped deformations, especially at the highest lectin concentrations, the agglutinating surfaces being flattened against each other over extended areas. The distribution of con A label was continuous and dense between the apposed surfaces and discontinuous on free surfaces. In the presence of con A the free surfaces of prefixed cells exhibited more microvilli than the surfaces of non-prefixed cells. These results favour the view that fixation prevents the formation of shear-resistant, lectin-mediated bonds between cells, not by restricting the lateral mobility of lectin receptors, but by impairing the apposition of rigid cell surfaces.