From the Howard Hughes Medical Institute and Department of Molecular Physiology and Biophysics, Vanderbilt University School of Medicine, Nashville, Tennessee 37232
Phospholipase D
(PLD)1 (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.
PLD has been partially purified from many sources (1) and has
recently been cloned from yeast, bacteria, plant, and mammalian sources
(2). The enzymes from Saccharomyces, Ricinus
(castor bean), and Streptomyces have several sequences that
are conserved in the human enzyme (Fig. 1) (2), and
these presumably represent components of the catalytic site. These
sequences are also found in cardiolipin synthase and phosphatidylserine
synthase from Escherichia coli (2, 3) (Fig. 1). These
enzymes also catalyze phosphatidyl transfer, suggesting that PLD is a
member of a larger family of enzymes (2, 3).
The first reported mammalian PLD (hPLD1a) has 1072 amino acids and a
molecular mass of 124 kDa (4). It is specific for PC and was obtained
by using the yeast PLD gene (SPO14) (5) to identify a human
expressed sequence tag for screening a HeLa cDNA library. A shorter
splice variant of hPLD1a with 1034 amino acids (hPLD1b) (Fig. 1), which
has similar regulatory properties, has been identified (6), and another
PLD (PLD2) (Fig. 1), which has 932 amino acids and 51% amino acid
sequence identity to hPLD1a, has been cloned from a mouse embryonic
library (7). In our laboratory, enzymes corresponding to hPLD1a and
hPLD1b have been cloned from rat tissues.2
The regulation of these cloned enzymes will be discussed below. Other
PLDs have been identified in human tissues and C6 glioma cells (8, 9),
and their partial sequences indicate similarity, if not identity, to
hPLD1.
PLD has been purified to a very high degree from pig lung microsomes
(10). It is specific for PC and has a molecular mass of 190 kDa and a
pH optimum of 6.6. Another PLD has been substantially enriched from pig
brain membranes (11). It has a molecular mass of 95 kDa, based on
hydrodynamic measurements, and is markedly stimulated by
PIP2 and the small G proteins ARF and RhoA. Other PLDs have
been purified to a limited extent, including various forms from rat
brain that differ in their pH optima and responses to Ca2+,
PIP2, oleate, or detergents (1, 12), and cytosolic PLDs that are Ca2+-responsive and differ in their substrate
specificity (1, 12). These differences suggest the existence of PLD
isozymes, but since the enzymes are far from homogeneous, this is
unclear.
The subcellular localization of PLD activity shows some interesting
features. The enzyme is enriched in plasma membranes from many tissues
but is also present in high activity in Golgi and nuclei (13-17).
There is significant activity in cytosol but not in mitochondria. In
liver, the plasma membrane enzyme responds more to RhoA than ARF,
whereas the reverse is true for that in other subcellular fractions
(13). Whether these differences reflect differences in the subcellular
distribution of PLD isozymes (7) or other factors remains to be
determined. Studies of PLD isozymes expressed in fibroblasts indicate
that PLD2 localizes predominantly in the plasma membrane, whereas PLD1
is perinuclear, i.e. in endoplasmic reticulum, Golgi, and
late endosomes (7).
Brown et al. (18) discovered that PLD is strongly stimulated
by PIP2, and this has been observed for most but not all
(10, 19, 20) preparations of the enzyme. PI-3,4,5-P3 is
also effective (6),3 but other acidic
phospholipids, including PI-3,4-P2, PI-4-P, and PI, are
nearly or completely ineffective (18, 21). There is evidence that
PIP2 is required for the activation of PLD in intact cells
(22), but it is unclear that physiological changes in PIP2
or PIP3 levels control the enzyme in vivo.
Studies with cloned, purified PLDs indicate that PIP2 and
PIP3 directly activate the enzyme (6).3
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 PIP2 by
PI-phospholipase C with subsequent production of diacylglycerol and
activation of PKC, PC hydrolysis is often secondary to PIP2 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
membrane-associated (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, 23-25).2,3 In particular, studies with cloned PLD purified
from Sf9 cells indicate that PKC 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 PIP2 but does not respond to PKC, Rho, or
ARF. Brain cytosol contains an inhibitor selective for PLD2 (7).
ARF was discovered as a factor that stimulated cholera
toxin-induced ADP-ribosylation of Gs PLD activity is high in Golgi and responds to ARF (12-14), 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 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 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 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).
Ceramides are being increasingly recognized as important
regulators of cell function (53). C6- and
C2-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 The existence of cytosolic proteins that enhance the effects of RhoA
and ARF on PLD has already been described. The actin-binding 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 PIP2
phosphatase activity (68) and presumably inhibits PLD by decreasing
PIP2. Another inhibitor that has a molecular mass of 30 kDa
and does not bind PIP2 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
PIP2. 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
A2 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 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.
Fig. 1.
Alignment of sequences of PLDs and associated
enzymes from various species. Homologous regions are
boxed. aa, amino acids. Modified from Ref.
2.
[View Larger Version of this Image (21K GIF file)]
Fig. 2.
Postulated mechanisms by which growth factors
and agonists whose receptors are linked to heterotrimeric G proteins
activate PLD. Mechanisms involving PKC isozymes and Rho family
proteins are shown, but other mechanisms, e.g. involving ARF
proteins, may operate.
[View Larger Version of this Image (17K GIF file)]
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 domain of PKC
(25). This evidence of PLD regulation by protein-protein 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
PIP2 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).
(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 clathrin-coated
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,
31-35). Two of these factors are PKC
(25) and calmodulin (34), but
the nature of the others remains obscure.
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.
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 (41-43). 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
S and AlF4
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
PIP2 synthesis by PI-4-P 5-kinase (49), and the changes in
PLD could be secondary to alterations in PIP2 levels
(50).
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
Ca2+-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).
2 production and degranulation in
neutrophils in response to chemotactic peptide (75). The generation of
the bactericidal O2 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).