New Developments in Phospholipase D*

Phospholipase D (PLD) (1) is present in bacteria, fungi, plants, and animals. It is widely distributed in mammalian cells, where it is regulated by a variety of hormones, growth factors, and other extracellular signals. Its major substrate is phosphatidylcholine (PC), which is hydrolyzed to phosphatidic acid (PA) and choline, but it can also act on phosphatidylethanolamine and phosphatidylinositol in some organisms and tissues. It also catalyzes a phosphatidyl transfer reaction in which a primary alcohol acts as nucleophilic acceptor in place of H2O. The resulting production of phosphatidyl alcohol represents a specific assay for PLD.


Regulation of PLD by Protein Kinase C
There is abundant evidence that PLD is regulated by PKC in most mammalian cells. This comes from studies of the effects of phorbol esters, PKC inhibitors, down-regulation of the enzyme, and overexpression and deletion of specific PKC isozymes (1). Although a role for PKC in the actions of many agonists on PLD in many tissues/cells has been indicated, there are also instances where the enzyme does not seem to be involved (1). Since many of the agonists that activate PLD also stimulate the hydrolysis of PIP 2 by PI-phospholipase C with subsequent production of diacylglycerol and activation of PKC, PC hydrolysis is often secondary to PIP 2 breakdown (1). Activation of PKC is also associated with translocation of the enzyme to cell membranes, and this relocalization is probably required for PKC activation of PLD, which is predominantly membraneassociated (Fig. 2).
The most direct mechanism of control of PLD by PKC would be through phosphorylation of the enzyme. However, in those studies where the effects of PKC on PLD have been studied directly, activation does not involve ATP, i.e. a phosphorylation mechanism (6,(23)(24)(25). 2,3 In particular, studies with cloned PLD purified from Sf9 cells indicate that PKC␣ and PKC␤ can directly activate the enzyme in an ATP-independent manner (6), 3 but other isozymes are ineffective (24). 3 The interaction is not affected by staurosporine and involves the regulatory do-main of PKC (25). This evidence of PLD regulation by proteinprotein interaction does not preclude an additional phosphorylation-dependent mechanism in vivo. Evidence of a role for phosphorylation comes from cell studies with PKC inhibitors that act by interfering with ATP binding (1) and studies of the effects of ATP in cell-free systems (26). However, the latter may be complicated by changes in PIP 2 synthesis (27). It must also be recognized that PKC may act in vivo by phosphorylating a protein(s) involved in PLD regulation or altering the response of PLD to such a protein(s). As noted below, an important aspect of PKC action is its synergistic interaction with ARF and RhoA to activate PLD (Fig. 2).
In contrast to other forms of mammalian PLD, PLD2 has a very high basal activity in vitro and in vivo (7). It is stimulated by PIP 2 but does not respond to PKC, Rho, or ARF. Brain cytosol contains an inhibitor selective for PLD2 (7).

Regulation of PLD by ARF
ARF was discovered as a factor that stimulated cholera toxin-induced ADP-ribosylation of G s ␣ (26). It is now recognized to play a role in vesicle trafficking in Golgi and has been implicated in the fusion of microsomal vesicles and endosomes, the assembly of nuclear membranes, and the formation of clathrincoated vesicles (28). The activation of PLD by ARF was first recognized by the groups of Sternweis and Cockcroft (18,29) and has now been shown using PLD from many sources (1). The enzyme is stimulated by ARF1, -3, -5, and -6, i.e. all classes of ARF, and the myristoylated ARFs are more effective than the non-myristoylated forms (11,30). Studies with cloned PLD purified from Sf9 cells indicate that ARF interacts directly with the enzyme (6). 3 Some reports have indicated that cytosolic factors greatly enhance the effect of ARF on PLD (25,(31)(32)(33)(34)(35). Two of these factors are PKC␣ (25) and calmodulin (34), but the nature of the others remains obscure.
PLD activity is high in Golgi and responds to ARF (12)(13)(14), and the stimulation is inhibited by brefeldin A, an inhibitor of ARF activation (14). As will be discussed below, an important role for ARF-stimulated PLD in the regulation of vesicular transport in Golgi is indicated by the studies of Ktistakis et al. (15). There is also limited evidence of a role for ARF in agonist regulation of PLD in vivo. In HEK cells expressing M3 muscarinic receptors, brefeldin A inhibited carbachol stimulation of PLD (36). In these cells, permeabilization caused a release of cytosolic ARF, which could be inhibited by adding GTP␥S during permeabilization or by prior carbachol treatment (36). Treatment of HL-60 cells with chemotactic peptide also caused an increase in membrane-associated ARF and GTP␥S-stimulated PLD activity (37). These results are consistent with the effect of these agonists on PLD being mediated by membrane translocation of ARF.

Regulation of PLD by Rho Proteins
Rho family proteins regulate many cellular activities including those involving the actin cytoskeleton. The proteins include Rho, which controls the formation of focal adhesions and actin stress fibers, Rac, which regulates lamellipodia formation and membrane ruffling, and Cdc42, which controls the formation of filopodia (38,39). The first evidence that PLD could be regulated by Rho proteins came from a study by Bowman et al. (40). This showed that the stimulatory effect of GTP␥S on PLD in neutrophil plasma membranes was inhibited by RhoGDI, a protein that inhibits GDP dissociation from Rho proteins and thereby blocks their activation. In subsequent studies using plasma membranes from rat liver, HL-60 cells, and neutrophils, it was found that RhoA was the most effective Rho protein to activate PLD, but Rac1 or Cdc42Hs showed some activity (41)(42)(43). RhoA also stimulates PLD in fractions from other cells and partially purified preparations of the enzyme (13,17,31,35,44). Studies with cloned PLD purified from Sf9 cells indicate that RhoA interacts directly with the enzyme and that Rac1 and Cdc42 are also active (6). 3 Interestingly, a combination of RhoA and ARF results in synergistic activation of homogeneous or partially purified PLD (6,25,35,42,44). 3 This suggests the presence of separate but interacting sites for Rho and ARF on PLD. As in the case of ARF, there is evidence that RhoA action on PLD is enhanced by cytosolic proteins (35,43) and by PKC␣ (6,25,43,45). 3 Several bacterial toxins that inactivate Rho proteins have been used to demonstrate the involvement of RhoA in agonist regulation of PLD in vivo. These include the C3 exoenzyme of Clostridium botulinum that ADP-ribosylates RhoA and blocks the activation of PLD in Rat1 fibroblasts by LPA (46). Likewise, Toxin B from Clostridium difficile, which glucosylates and inactivates Rho proteins, blocks the activation of PLD by carbachol in intact cells and by GTP␥S and AlF 4 Ϫ in permeabilized cells (47). The toxin also inhibits IgE receptor-mediated activation of PLD in basophilic leukemia cells (48). While these observations support a role for RhoA in agonist stimulation of PLD, caution should be observed since the small G protein also regulates PIP 2 synthesis by PI-4-P 5-kinase (49), and the changes in PLD could be secondary to alterations in PIP 2 levels (50).
Because of the location of its substrate, PLD must be or must become membrane-associated for activity. Thus its regulators must also be present in membranes or translocate there. As described above, there is evidence for agonist-induced membrane relocalization of ARF (36,37), and this is a well known phenomenon for PKC isozymes. Agonist-induced membrane translocation of Rho family proteins has also been reported recently (46,51,52), and this is probably an important component of the mechanism of PLD activation by certain agonists (Fig. 2).

Other Factors Regulating PLD
Ceramides are being increasingly recognized as important regulators of cell function (53). C 6 -and C 2 -ceramides inhibit agonist or phorbol myristate acetate-stimulated PLD activity in several cell lines (54 -57) and block the stimulation of the enzyme by GTP␥S plus phorbol myristate acetate, ARF, or RhoA in cell extracts (52,56), but dihydro derivatives are ineffective. Since the ceramides also block the membrane translocation of ARF, RhoA, and Ca 2ϩ -dependent PKC isozymes (52,55), part of the inhibition could be due to these effects. In contrast to ceramides, sphingosine and its metabolite sphingosine-1-P activate PLD (58). It is probable that their effects are partly mediated by a sphingosine-1-P receptor coupled to a G protein(s) (59,60).
The existence of cytosolic proteins that enhance the effects of RhoA and ARF on PLD has already been described. The actinbinding protein gelsolin also stimulates PLD but only in the presence of nucleoside triphosphates (61). A role for PLD in control of the actin cytoskeleton has been proposed (62,63), but the mechanisms involved are unclear.
Inhibitors of PLD have been identified (64 -67). One of these is heat-labile with a molecular mass of 150 kDa (65). Its partial sequence (65) indicates that it is synaptojanin, which exhibits PIP 2 phosphatase activity (68) and presumably inhibits PLD by decreasing PIP 2 . Another inhibitor that has a molecular mass of 30 kDa and does not bind PIP 2 has been identified (66). Fodrin, a non-erythroid form of spectrin, has been reported to inhibit PLD (67). Although fodrin has a PH domain, it apparently does not act by binding PIP 2 . Brain contains an 18-kDa inhibitor that is selective for PLD2 (7).

Biological Significance of PLD
PLD could exert its biological effects by several mechanisms. The first is by changing the properties of cellular membranes by altering their lipid composition. Thus, by causing local changes in PC and PA and releasing polar choline, the physical properties of the membranes could be substantially changed. A second mechanism is by generating PA. This lipid would probably remain in the membrane but could interact with proteins located in the membrane or cytosol. Many proteins have been shown to have their activities changed by PA in vitro (1), but evidence that they are targets of the lipid in vivo is largely lacking. Since transphosphatidylation may be a major function of PLD in vivo, the possibility should be considered that the products of this reaction may also have signaling functions.
A third mechanism of biological action of PLD arises from the fact that PA is rapidly converted to DAG in most cells through the action of phosphatidate phosphohydrolase. Thus the late phase of activation of PKC produced by agonists in many cells is mainly attributable to DAG derived from PLD action (1,69). A fourth function of PLD is the generation of LPA through the action of a specific phospholipase A 2 on PA. LPA is becoming recognized as a major extracellular signal produced by activated platelets and probably other cells (70). A final possibility relates to the formation of choline. Because of its high resting cellular levels, this compound probably does not have a signaling function, but it could serve as a substrate for acetylcholine synthesis in neurons (71).
Roles for PLD have been proposed for many cellular functions, but space permits consideration of only a few. The potential function of PLD in the regulation of cell proliferation remains controversial. PA and bacterial PLD are mitogenic in several cell lines (1,72), although some of the effects may be due to LPA contamination or formation (1,70), and there is a lack of correlation between PLD activity and mitogenesis in some cell lines (73). Other signaling pathways are undoubtedly involved in growth control. In Saccharomyces cerevisiae, PLD was first recognized as the product of the SPO14 gene, which is essential for meiosis (5). The mechanisms by which PLD could control the cell cycle are unknown. However, Raf-1 kinase, which is involved in signal transduction from several receptors, has a binding site for PA and is translocated to membranes under conditions where PLD is activated (74).
As alluded to above, PLD has been implicated in the regulation of vesicle trafficking by ARF in Golgi. In Chinese hamster ovary cell lines displaying high basal PLD activity in Golgi, ARF is not necessary for coatamer assembly (15). The formation of coated vesicles is also inhibited by ethanol, which reduces PA formation due to the production of phosphatidylethanol, and treatment of Golgi membranes with bacterial PLD promotes coatamer binding (15). Studies with ethanol and exogenous PLD in other cell types have also provided evidence that PLD is involved in vesicle transport (27). It is not known whether its action is due to the generation of PA, a membrane fusigen, or to other membrane lipid changes. Colley et al. (7) have observed that PLD2 provokes cortical reorganization mediated by actin polymerization and undergoes redistribution from the plasma membrane in serum-stimulated cells. They have postulated that PLD1 and PLD2 are involved in vesicle translocation from the perinuclear region and in receptor-mediated endocytosis, respectively. Other studies have implicated PA in the control of actin polymerization (62,63).
PLD has been implicated in O 2 . production and degranulation in neutrophils in response to chemotactic peptide (75). The generation of the bactericidal O 2 species, which is due to activation of NADPH oxidase, is inhibited by ethanol and mimicked by the addition of PA (75). These data, which implicate PA in the control of NADPH oxidase, are supported by in vitro findings (76,77). However, these studies also provide evidence for the involvement of DAG (77) and indicate that the two lipids synergize to activate the enzyme (77).

Concluding Comments
Although substantial advances have recently been made in the enzymology and regulation of PLD, major questions remain. In particular, the various isozymes need to be identified and their structure/function relationships and regulatory properties defined. Molecular biological approaches should now make this possible. The mechanisms by which growth factors, hormones, and other agonists regulate the enzyme need to be clarified, with emphasis on the roles of PKC, Rho proteins, ARF, and certain lipids. Finally, but most importantly, the physiological functions of PLD need to be defined, and molecular biological approaches should also provide major help in this area.