Phosphoinositides are localized in various intracellular compartments and can regulate a number of intracellular functions, such as cytoskeletal dynamics and membrane trafficking. Phospholipase Ds (PLDs) are regulated enzymes that hydrolyse phosphatidylcholine (PtdCho) to generate the putative second messenger phosphatidic acid (PtdOH). In vitro, PLDs have an absolute requirement for higher phosphorylated inositides, such as phosphatidylinositol 4,5‐bisphosphate [PtdIns(4,5)P2]. Whether this lipid is able to regulate the activity of PLD in vivo is contentious. To examine this hypothesis we studied the relationship between PLD and an enzyme critical for the intracellular synthesis of PtdIns(4,5)P2: phosphatidylinositol 4‐phosphate 5‐kinase α (Type Iα PIPkinase). We find that both PLD1 and PLD2 interact with the Type Iα PIPkinase and that PLD2 activity in vivo can be regulated solely by the expression of this lipid kinase. Moreover, PLD2 is able to recruit the Type Iα PIPkinase to its intracellular location. We show that the physiological requirement of PLD enzymes for PtdIns(4,5)P2 is critical and that PLD2 activity can be regulated solely by the levels of this key intracellular lipid.
Phospholipase D (PLD) catalyses the hydrolysis of phosphatidylcholine (PtdCho) to generate phosphatidic acid (PtdOH), which remains in the membrane, and choline, a water‐soluble head group (Exton, 1998; Liscovitch et al., 2000). Thus far, two isoforms of PLD have been described, PLD1 and PLD2, both of which exist as two splice variants (Colley et al., 1997; Hammond et al., 1997; Park et al., 1997). The 124 kDa PLD1 appears localized to vesicles derived from the endosomal/lysosomal pathway (Ktistakis et al., 1995; Brown et al., 1998), but has also been detected at the plasma membrane (Brown et al., 1998); it has been suggested that it is involved in the regulation of membrane coating that occurs during vesicle formation (Austin and Shields, 1996; Ktistakis et al., 1996; Chen et al., 1997; Siddhanta et al., 1998) and in the regulation of secretion (Metz and Dunlop, 1990; Stutchfield and Cockcroft, 1993; Cockcroft, 1996; Bi et al., 1997; Morgan et al., 1997; Brown et al., 1998; Caumont et al., 1998). PLD1 activity is regulated by multiple inputs, including phosphatidylinositol 4,5‐bisphosphate [PtdIns(4,5)P2] (Liscovitch et al., 1991, 1994; Pertile et al., 1995; Schmidt et al., 1996b), protein kinase C (PKC) (Conricode et al., 1992; Eldar et al., 1993; Cockcroft, 1996a,b; Kiss, 1996; Ohguchi et al., 1996; Park et al., 1998; Kim et al., 2000), the Rho family proteins Cdc42, Rac (Hess et al., 1997; Plonk et al., 1998) and Rho (Malcolm et al., 1996; Hess et al., 1997; Karnam et al., 1997; Schmidt et al., 1997; Vinggaard et al., 1997), and Arf proteins (Brown et al., 1993; Cockcroft et al., 1994; Ktistakis et al., 1995; Kuribara et al., 1995; Whatmore et al., 1996) that act in an additive manner, leading to greater PLD1 activity than each alone (Hodgkin et al., 1999).
PLD2 is a 106 kDa protein that shares 50–55% homology with PLD1 but lacks the 116 amino acid loop region following the first HKD motif (Plonk et al., 1998). This isoform has been suggested to be localized to the plasma membrane (Plonk et al., 1998) and to an as yet undefined submembraneous vesicle compartment, which translocates to the membrane on stimulation with epidermal growth factor (EGF) (Honda et al., 1999). PLD2 is reported to have a much higher basal activity than PLD1 when expressed in Cos‐7 cells or isolated from insect cells and it has been suggested that cellular proteins, such as fodrin and synucleins (Jenco et al., 1998), act to reduce the basal activity. Consequently, growth factor stimulation, rather than activating the enzyme catalytically, may repress the action of these endogenous inhibitors. Phorbol ester (phorbol 12‐myristate 13‐acetate) and members of the ARF family activate PLD2 2‐fold only; however, deletion of the non‐core N‐terminal amino acids generates a protein with a much lower basal activity, which can be activated by ARF some 13‐fold (Sung et al., 1999). Whether PLD2 can be activated in response to growth factors is still unclear.
Both mammalian PLD1 and PLD2 have an absolute requirement for the higher phosphorylated inositol lipids (Berstein et al., 1992; Colley et al., 1997; Hammond et al., 1997; Sciorra et al., 1999; Hodgkin et al., 2000). In keeping with this, we have recently demonstrated that PLD1 has a functional pleckstrin homology (PH) domain, specific for PtdIns(4,5)P2, that regulates both activity and localization (Hodgkin et al., 2000). Sequence analysis demonstrates considerable homology between the PH domains of PLD1 and PLD2, thus a similar specificity can be presumed.
Cellular PtdIns(4,5)P2 levels are, in part, governed by two families of enzymes, namely the Type I and Type II phosphatidylinositol 4‐phosphate 5‐kinases (PIPkinases) (Hinchliffe et al., 1998; Anderson et al., 1999). The Type II enzymes are composed of three isoforms that synthesize PtdIns(4,5)P2 by the 4‐phosphorylation of PtdIns(5)P (Rameh et al., 1997). The role of these enzymes in the maintenance and synthesis of cellular PtdIns(4,5)P2 pools is not yet understood. Type I PIPkinases comprise at least three distinct isoforms that phosphorylate the 5‐position of PtdIns(4)P (Ishihara et al., 1996; Loijens and Anderson, 1996; Shibasaki et al., 1997). However, in vitro, these enzymes will phosphorylate a number of inositol lipids such as PtdIns, PtdIns(3)P, PtdIns(4)P and PtdIns(3,4)P2 (Zhang et al., 1997; Tolias et al., 1998b). The Type I PIPkinases have been shown to associate with at least two low molecular weight G proteins of the Rho family (Chong et al., 1994; Ren et al., 1996; Tolias et al., 1998a) and recently to be activated by members of the ARF family (Honda et al., 1999; Jones et al., 2000). Previous data demonstrated that the Type I PIPkinases were activated by PtdOH (Jenkins et al., 1994), one product of PLD hydrolysis of PtdCho, and regulation of the Type I PIPkinase by ARF, in vitro, is dependent upon this acidic phospholipid (Honda et al., 1999). The Type I PIPkinases have been shown to be required for secretion in permeabilized cells, for Rho‐mediated ezrin, radixin and moesin (ERM) phosphorylation (Matsui et al., 1999) and Rac‐regulated capping/uncapping of actin in platelets (Tolias et al., 2000). Recent evidence has suggested that PIPkinases may also be involved in ARF‐regulated vesicle coating and budding at the Golgi and lysosomal membranes (Arneson et al., 1999). A role in the internalization of activated receptors has also been suggested. Genetic studies in mice lacking a functional synaptojanin, a PtdIns(4,5)P2 5‐phosphatase, have demonstrated increased PtdIns(4,5)P2 levels with an accumulation of clathrin‐coated vesicles in the cytosol, further supporting a role for PtdIns(4,5)P2 in vesicular trafficking (Cremona et al., 1999).
In most, if not all of these cellular processes, a role for PLD has also been implicated. A feed‐forward cycle has been postulated whereby the generation of PtdIns(4,5)P2 activates PLD, which leads to enhanced PtdOH formation able to activate further the Type I PIPkinase. This would lead to a rapid local increase in both PtdOH and PtdIns(4,5)P2, which has been suggested to be important in both the generation of membrane curvature, vesicle budding and the recruitment/activation of proteins involved in coating of vesicles (Liscovitch and Cantley, 1995). Although enticing, there is a paucity of evidence to corroborate this theory. In this report we demonstrate that both PLD1 and PLD2 interact with the murine Type Iα PIPkinase and that PLD2 is able to lead to the recruitment of Type Iα PIPkinase in porcine aortic endothelial (PAE) cells. Finally, expression of the Type Iα PIPkinase leads to the activation of PLD2 activity in vivo. These data suggest a molecular mechanism by which PtdIns(4,5)P2 can be generated in a localized environment, leading to activation of PLD2.
Expression of a Type I PIPkinase in Cos‐7 cells leads to the activation of an endogenous PLD
Cos‐7 cells transfected with LacZ (cont) or murine Type Iα PIPkinase were treated as controls or stimulated with 12‐O‐tetradecanoylphorbol 13‐acetate (TPA) or lysophosphatidic acid (LPA) and PLD activity was assessed. PLD activity is measured easily by the inclusion of a primary alcohol, such as butanol, which is used as a nucleophile, in place of water, leading to the production of phosphatidylbutanol (PtdBut). Unlike PtdOH, which can be synthesized in a cell by at least three separate mechanisms, the more metabolically stable PtdBut can only be formed by the action of PLD. Transfection of the Type Iα PIPkinase leads to an increase in endogenous PLD activity [from 856 ± 75 to 1432 ± 86 U (where U are arbitrary phosphoimager units); Figure 1]. As only 50% of the cells become transfected under these conditions, this suggests that the true stimulation would be at least 4‐fold. This increase is equivalent to that seen in vivo using transfection with constitutively activated RhoA or with ARF‐1, known activators of PLD1 (data not shown). TPA induced an increase of 1391 ± 81 U in non‐transfected cells; however, in Type I PIPkinase‐transfected cells TPA gave an increase of 2626 ± 207 U. Expression of Type I alone led to an increase of 575 ± 52 U, and added to the increase from the TPA treatment this would account for 1966 U of PLD activity. This suggests that Type Iα PIPkinase‐induced PLD activity is unlikely to occur through increased PtdIns(4,5)P2 formation and subsequent hydrolysis leading to enhanced PKC activation. The fact that the activity in Type Iα PIPkinase‐transfected cells stimulated with TPA is greater than the sum of the two activities alone may suggest that TPA may be able to activate the Type Iα PIPkinase.
Analysis of the phosphatidylinositol lipids after transfection of the Type I PIPkinase demonstrated a 2‐fold increase in the level of PtdIns(4,5)P2 (Figure 1). These data are consistent with the hypothesis that transfection of the Type I PIPkinase enhances PtdIns(4,5)P2 synthesis, leading to the stimulation of an endogenous PLD activity.
Co‐expression of Type Iα PIPkinase leads to the activation of PLD2
To investigate which isoform of PLD is activated by the Type Iα PIPkinase we studied the effect of its expression together with either PLD1 or PLD2 on PtdBut formation. As the extent of transfection differs between experiments, the counts obtained in the LacZ alone transfection have been subtracted. Thus, the data represent the PLD activity due to the transfection of the various constructs. Transfection of PLD1 into Cos‐7 cells led to an increase in the basal activity that was stimulated potently by TPA, but we have been unable to demonstrate enhanced activation by co‐transfection with the Type Iα PIPkinase (data not shown). This was also shown to be the case when the cells were stimulated with LPA. Although not conclusive, these data suggest that the endogenous PLD that is activated by expression of the Type Iα enzyme is not PLD1.
Transfection of PLD2 led to a small increase in both PtdOH (102 U) and PtdBut (327 U) in serum‐starved cells. Co‐transfection of PLD2 with the Type Iα PIPkinase led to a much larger increase in both PtdOH and PtdBut formation (1756 and 2216 U, respectively. Type I PIPkinase alone yielded an increase of 306 and 42 U in PtdOH and PtdBut formation, respectively (Figure 2A). No enhancements of these changes were seen after stimulation with EGF, although MAP‐kinase was activated by this agonist (Figure 2B). The gel also shows that there was approximately equal green fluorescent protein (GFP)–PLD2 and Type I PIPkinase expression in the transfected cells (each lane represents a single transfection and is the protein recovered from the interface of the lipid extractions used to generate the PtdBut data) (Figure 2A). These data suggest that co‐expression of the Type I PIPkinase is able to lead to activation of PLD2 in vivo.
Murine Type Iα PIPkinase interacts with PLD
To assess whether PLD2 and Type Iα PIPkinase interact, we co‐transfected Cos‐7 cells with haemagglutinin (HA)‐tagged PLD2 and EE‐tagged Type Iα PIPkinase. Immunoprecipitation was carried out using antibodies directed against the Glu‐Glu (EE) tag (Type I PIPkinase), followed by western blot analysis using the anti‐HA antibody (PLD2). Analysis of the whole cell lysates showed that HA‐PLD2 was expressed equally (Figure 3A, lanes 5 and 7), but was only immunoprecipitated by the EE antibody when co‐expressed with the EE‐Type Iα PIPkinase (Figure 3A, lane 3). These data demonstrate that PLD2 is able to interact with the Type Iα PIPkinase.
To define regions of PIPkinase that interact with PLD2, deletion mutants of the Type Iα PIPkinase (Figure 3B) were constructed and expressed as recombinant glutathione S‐transferase (GST)‐tagged proteins in Escherichia coli. These were purified using glutathione beads and then used to affinity purify PLD2 that had been expressed in Cos‐7 cells (Figure 3B, blot). As controls, either GST alone or GST–Type IIα PIPkinase was used. The full‐length Type Iα PIPkinase was able to affinity purify PLD2. Deletion of the first 120 amino acids (δ1–420) did not prevent affinity purification of PLD2. However, deletion of the first 306 amino acids (δ1–912) completely abolished PLD2 binding. Deletions in the other direction demonstrated that even removal of the C‐terminal 425 amino acids (δ325–1626), leaving only the N‐terminal 108 amino acids, still allowed binding of PLD2. It should be noted that there is no overlap between δ325–1626 and δ1–420, suggesting that either these two regions form distinct PLD2 binding sites or that they are both part of a single binding site. No affinity purification of PLD2 was achieved with either GST alone or GST–Type II PIPkinase.
Although we were unable to show that Type Iα PIPkinase could regulate PLD1 activity in vivo, we were able to demonstrate that these two proteins interact (Figure 4). After co‐transfection of the two proteins, wild‐type PLD1 was immunoprecipitated and the Type Iα PIPkinase activity associated with this enzyme was assessed. Co‐transfection of the two cDNAs led to an enhancement of the PIPkinase activity immunoprecipitated with PLD1 (there was an increase in the PIPkinase activity in Type I‐transfected cells alone, which may represent immunoprecipitation of endogenous PLD1) (Figure 4). Immunoprecipitation of the Type I activity, using an anti‐myc antibody, demonstrated that its expression, either in the presence or the absence of PLD1, was equivalent (Figure 4, inset). No PIPkinase activity was associated with PLD1 immunoprecipitates when it was co‐expressed with the Type IIα PIPkinase. These data demonstrate that both isoforms of PLD are able to interact with the Type Iα PIPkinase, but not with the Type II enzyme. However, PtdIns(4,5)P2 generated by this enzyme can only activate PLD2 in vivo.
PLD2 recruits the Type Iα PIPkinase when co‐expressed in PAE cells
To assess whether Type Iα PIPkinase and PLD2 interact in vivo, we co‐expressed these proteins in confluent PAE cells by microinjection of the cDNAs and stained for the various proteins 3 h later. This experiment was carried out using both GFP‐tagged PLD2 and myc‐tagged Type I PIPkinase or GFP‐tagged Type Iα PIPkinase and HA‐tagged PLD2. In both cases the same result was observed. PLD2 was localized in a submembraneous vesicular compartment (Figure 5A, 2), which did not co‐localize with markers for either the endoplasmic reticulum or the Golgi (data not shown). Expression of the Type I PIPkinase alone showed a complex staining pattern associated with the plasma membrane, cytosol and in intracellular structures that resemble microtubuli (Figure 5A, 1). Co‐expression of PLD2 and Type Iα PIPkinase resulted in a dramatic relocalization of the Type Iα PIPkinase to the submembraneous PLD2‐positive patches (Figure 5B). This co‐localization was also seen when viewing the cells through a Z section (Figure 5C). As the Type Iα PIPkinase is able to associate with a number of different proteins (Rac1, PLD1 and PLD2, PKD and ARFs) and to test the requirement for PIPkinase activity for this recruitment, we also tested the localization of the GFP–δ325–1626 deletion mutant, which contains only the first 108 amino acids but is still able to bind to PLD2 (Figure 3C). When injected alone, the GFP–δ325–1626 was completely cytosolic (Figure 5A, 3); however, co‐injection with PLD2 also led to a complete co‐localization at the sub‐membraneous patches (Figure 5D). Thus, PLD2 is able to recruit the Type Iα PIPkinase to its own intracellular location.
Local generation of PtdIns(4,5)P2 is required for TPA‐stimulated PtdBut formation
To determine whether the Type I PIPkinase activity was required for the activation of PLD2, we constructed a kinase‐dead (KD) version of the Type I PIPkinase by substituting a single amino acid in the ATP‐binding loop as described previously (Ishihara et al., 1998). This mutant was expressed in Cos‐7 cells, immunoprecipitated and assayed for PIPkinase activity. The wild‐type enzyme yielded 109 146 U, while the KD mutant produced 1435. This mutant was transfected into Cos‐7 cells, in the presence or the absence of PLD2, and the activity of the expressed PLD2 was assessed. Overexpression of the KD mutant significantly reduced the basal activity of PLD2 (Figure 6A). The reason for the incomplete inhibition probably resides in the fact that both PLD2 and Type Iα KD were co‐transfected. Thus, the expression of the KD may not be high enough to completely abrogate the heterologously expressed PLD2 activity (anti‐GFP antibodies showed that GFP–PLD2 with GFP–wild‐type Type Iα PIPkinase were expressed at equimolar levels).
Previous data showed that PtdIns(4,5)P2 was also required for TPA‐stimulated PLD activity in vivo. We therefore investigated whether the kinase activity was also required. PLD2 was co‐expressed together with the δ325–1626 construct, which we showed was able to bind PLD‐2, but is inactive with respect to kinase activity. Expression of this mutant results in a 50% decrease in the TPA‐induced PLD2 activity (Figure 6B). Western analysis showed that there was no significant change in the expression of PLD2 in these transfections. Identical data were obtained using the KD mutant Type Iα instead of the δ325–1626 mutant (Figure 6A, striped bars). These data suggest that interaction with Type Iα PIPkinase is required for the generation of PtdIns(4,5)P2, which is required for the activation of PLD2 by TPA.
Type Iα PIPkinase and PLD interact in vivo
Although the previous data are consistent with the hypothesis that PLD family members and the Type Iα PIPkinase interact, these experiments were carried out with overexpressed proteins. To assess the in vivo significance, we employed anti‐peptide antibodies specific for the Type I PIPkinase and assessed whether they were able to immunoprecipitate PLD activity assayed in vitro. PAE cells were grown to confluency, lysed using a mild detergent buffer and the proteins were immunoprecipitated using specific antibodies to the Type I or Type II PIPkinase as a control. After extensive washing, the beads were assayed for PIPkinase activity by phosphorylation of PtdInsP to PtdIns(4,5)P2 in the presence of (N‐[6‐[7‐nitrobenz‐2‐′2‐oxa‐1,3‐diazol‐4‐yl]amino]caproyl) (NBD)– PtdCho as a substrate for PLD. The reactions were carried out for 20 min and the products were separated by thin layer chromatography (TLC). PIPkinase activity was assessed by phosphoimager analysis of the 32P incorporated into PtdIns(4,5)P2, while PLD activity was assessed by the generation of NBD–PtdOH. As a control, HA‐tagged PLD2 was expressed in Cos‐7 cells and immunoprecipitated using the anti‐HA antibody (Figure 7, lane 3). No PLD activity was found when either protein G– Sepharose or protein G–Sepharose coupled to anti‐Type II PIPkinase antibodies was used (Figure 7). In contrast, anti‐Type I PIPkinase antibodies immunoprecipitated PLD activity. As the assays were carried out in the presence of 10 μM PtdIns(4,5)P2, the increased PLD activity is not a reflection of the amount of conversion of PtdIns(4)P to PtdIns(4,5)P2. These data suggest that, in vivo, PLD can associate with Type I PIPkinase isoforms in PAE cells.
A number of studies have demonstrated that PLD activity requires the presence of PtdIns(4,5)P2. It has been suggested that the inositide interacts with a site within domain IV (Sciorra et al., 1999), but contradicting evidence has now pointed to the effect being mediated through a PtdIns(4,5)P2 selective PH domain (Hodgkin et al., 2000). The PH domain binding site appears to have two functions, as point mutations abolish PLD activity, whilst deletion of the domain prevents membrane association. The demonstration that Rho family proteins are able to regulate PIPkinase activity (Chong et al., 1994; Ren et al., 1996; Tolias et al., 1998a) together with the use of a number of bacterial toxins, such as the Clostridium difficile toxin B (which inactivates Rho family proteins) and Clostridium sordelli toxin (which inhibits both Ras and Rho family proteins), has been used to implicate the importance in vivo of phosphoinositides in the regulation of PLD activity (Schmidt et al., 1996a,b, 1997). These data are, however, limited as the small molecular weight G proteins, which are targets for these toxins, are also potential activators of PLD1. In toxin B‐pretreated HEK 293 cells, the reduced GTPγs‐mediated PLD activity could be fully restored by the re‐addition of PtdIns(4,5)P2 (Schmidt et al., 1996b). These data, although implying a specific role for PtdIns(4,5)P2 in the regulation of PLD activity, do not demonstrate that PLD activity can be regulated in vivo by the levels of this lipid. In this study we demonstrate that the endogenous PLD in Cos‐7 cells can be stimulated solely by the overexpression of a Type I PIPkinase. Furthermore, we suggest that this occurs through the interaction between the two proteins. Recruitment of the Type Iα PIPkinase to the intracellular compartment where PLD2 resides would lead to enhanced synthesis of polyphosphoinositides. Although both PLD1 and PLD2 have an absolute requirement for PtdIns(4,5)P2 and are able to interact with the Type I PIPkinase, we were only able to show activation of PLD2. As PLD1 activity is regulated by multiple, independent inputs: two separate GTPases, a Rho family member and an ARF family member and PKC, in a phosphorylation‐independent manner, it may be that PtdIns(4,5)P2 levels alone are not enough to regulate this enzyme. PLD2, in contrast, is not regulated by the same multiple inputs.
The mechanism behind the regulation of PLD family members by phosphoinositides is not clear, as it has been demonstrated that PLD2 activity may be negatively regulated in cells by proteins such as fodrin and synucleins (Jenco et al., 1998) and by actinin and amphiphysins (Lee et al., 2000; Park et al., 2000). It is possible that the role of polyphosphoinositides is to alleviate the inhibition by these proteins. The recent demonstration that ARF family members are able to regulate Type Iα PIPkinase activity (Honda et al., 1999; Jones et al., 2000), together with the data from this study, suggest that ARF‐mediated regulation of PLD activity may occur through enhanced local PtdIns(4,5)P2 synthesis. Alternatively, phosphoinositides may regulate ARF activation. The regulation of ARF by phosphoinositides is complex. PtdIns(4,5)P2 has been shown to mediate the activation of an ARF GAP and its interaction with ARF (Randazzo, 1997); this would lead to an increase in the GDP loading. However, PtdIns(4,5)P2 has also been suggested to positively regulate the activity of an ARF–guanine nucleotide exchange factor (ARNO), leading to an increase in the levels of GTP bound ARF (Paris et al., 1997). Further complexity arises through the ability of PtdIns(4,5)P2 to stimulate guanine nucleotide exchange on ARF (Terui et al., 1994). At present, there are no methods to look specifically for activated ARF in vivo and the importance in vivo of ARF family members in regulating PLD2 remains controversial.
A number of studies have previously suggested a link between PtdIns(4,5)P2, PtdOH and vesicle formation (Liscovitch and Cantley, 1995; Pertile et al., 1995; Arneson et al., 1999). The generation of PtdIns(4,5)P2 by PIPkinase, and the subsequent activation of PLD2 to generate PtdOH, positively regulates PIPkinase, leading to high local concentrations of these lipids in the membrane, which may be important in the induction of membrane curvature. It may also be important in the recruitment of adapter proteins involved in the generation of a coat required to make the vesicle. Whether this feedback occurs is not known. Previous data from Moritz et al. (1992) showed that treatment of membranes with bacterial PLD, which leads to increased PtdOH formation, also led to the activation of an endogenous PIPkinase.
The generation of PtdOH at membranes may be a general mechanism for regulating PIPkinase at these intracellular domains. Recently, it has been demonstrated that endophillin‐1, which is able to stimulate synaptic vesicle budding, acts as a lysophosphatidic acid acyl‐transferase generating PtdOH. The authors suggest that this may be required to activate a PIPkinase able to generate PtdIns(4,5)P2 and thus leading to the recruitment/activation of dynamin (Schmidt et al., 1999). A role for PtdIns(4,5)P2 in clathrin‐coated vesicle formation has been genetically established in mice that are homozygous for a deletion of synaptojanin, a PtdIns(4,5)P2 5‐phosphatase (Cremona et al., 1999).
The data presented in this paper are consistent with the hypothesis that PLD2 is able to lead to the recruitment of the Type Iα PIPkinase that leads to enhanced PtdIns(4,5)P2 production, which is able to regulate PLD2 activity. This study provides the first evidence in vivo for activation of PLD by PtdIns(4,5)P2 and places PLD2 as a downstream effector of the Type Iα PIPkinase. This may have important implications as a number of studies have suggested that both PIPkinase and PLD activities are up‐regulated in the development of cancer.
Materials and methods
Unless otherwise stated all reagents were of analytical grade. PtdCho, phosphatidylserine (PtdSer), PtdIns, PtdInsP and PtdInsP2 were purified from bovine brain. NBD–PtdCho was purchased from Molecular Probes. All radiochemicals were purchased from Amersham International. Anti‐HA tag antibody was generated from clone 12CA5, while the myc epitope was from clone 9e10. All secondary antibodies were purchased from Dako Products.
A cDNA encoding the murine Type Iα PIPkinase was generated from a murine brain cDNA library using PCR and cloned into either Pjex 4T (Pharmacia) for expression in bacteria, or pCDNA3 for expression in cos cells. Deletions were generated using PCR and cloned into the above vectors and into pEGFP‐C2 for GFP fusion protein expression. PLD1 and HA‐PLD2 were subcloned into pCDNA3. GFP–PLD2 was cloned into PEGFP‐C2
Cos‐7 cells were routinely maintained at ∼30% confluency and were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 8% fetal calf serum (FCS) (v/v). The cells were split 1 day before transfection to 30%, and transfected the following day using DEAE–dextran. Briefly, cells were washed twice with phosphate‐buffered saline (PBS) and incubated with plasmid DNA [2 μg per 6 cm dish in 560 μl of PBS + 30 μl of DEAE–dextran (10 mg/ml)] for 30 min. Four millilitres of DMEM–8% FCS containing 80 μM chloroquine were added and left for a further 2.5 h. This medium was replaced with 2 ml of DMEM–8% FCS–10% DMSO, left for 2 min and replaced with fresh DMEM–8% FCS. Transfection was carried out for a further 24 h, after which the cells were labelled and treated as below. PAE cells were maintained in DMEM–8% FCS and routinely passaged to 30% confluency. Microinjection of plasmids (0.1 μg/ml) into the nucleus of PAE cells was carried out using an automated Eppendorf injector. The cells were allowed to recover for 3 h and then were fixed and processed as described below.
In vivo PLD assay
Cells were transfected as above, left for 24 h and labelled overnight with [32P]orthophosphate (10 μCi) in phosphate‐free DMEM. The cells were washed twice with RPMI‐salts buffer and incubated in this buffer containing butan‐1‐ol (0.3% v/v) for 30 min. This medium was aspirated, and 0.45 ml of 2.4 M HCl added. The cells were maintained and scraped on ice, removed to a clean Eppendorf tube and the dishes washed with 0.5 ml of methanol, which was pooled with the HCl. Half a millilitre of chloroform containing 5 μg of Folsch lipid extract was added, together with 0.25 ml of water. The two phases were mixed vigorously, centrifuged (2 min at 21 000 g in an Eppendorf bench centrifuge) and the lower phase was removed carefully, so as not to disturb the protein interface (see below), and was washed once with theoretical upper phase (chloroform:methanol:1 M HCl 15:245:235, 0.7 ml), with the lower phase being removed to a clean Eppendorf tube. The first upper phase was then back extracted with 0.2 ml of chloroform, which after mixing and centrifugation was removed to the tube containing the theoretical upper phase wash. This was mixed and centrifuged and the lower phase removed to the tube containing the first lower phase. The samples were dried and kept at −20°C. Samples were analysed for PtdBut formation and for PtdIns/PtdInsP/PtdInsP2 labelling (see the following section). To assess the expression levels of various constructs, the protein layer at the initial interface was recovered after lipid removal by the addition of 0.5 ml of methanol to the upper phase. The Eppendorf tube was centrifuged (21 000 g for 10 min), and the protein pellets were washed once with 70% acetone. These were allowed to dry and were resuspended overnight in 50 mM Tris pH 8.0, 8 M urea. SDS–PAGE sample buffer was added, the samples were boiled, resolved by SDS–PAGE and transferred to nitrocellulose. The blots were probed with various antibodies (at dilutions indicated in the figure legends) and visualization was carried out using ECL according to the manufacturer's instructions (Amersham International).
Cells were transfected as described and left for 48 h, after which they were lysed [1 ml lysis buffer (50 mM Tris pH 8.0, 50 mM KCl, 10 mM EDTA, 1% NP‐40)], scraped, and nuclei and cellular debris removed by centrifugation (14 000 r.p.m., 4°C Eppendorf centrifuge). Immunoprecipitations were carried out using monoclonal antibodies against epitope tags 12CA5 (anti‐HA tag), 1E10 (anti‐myc tag) and an antibody derived against the EE tag. Immunoprecipitation of the endogenous Type Iα PIPkinase from PAE cells was carried out using an anti‐peptide antibody. The immunoprecipitates were collected using protein G–Sepharose, and washed five times with IP wash buffer (50 mM Tris pH 7.5, 150 mM NaCl, 5 mM EDTA 0.1% Tween‐20), then once with PIPkinase buffer. Immunoprecipitations were either used for western blotting or were used to assess the PIPkinase and/or PLD activities as described below.
Lipid vesicles were prepared using 1 nmol of PtdInsP isolated from pig brain together with 3 nmol of PtdOH. Reactions were carried out at 30°C for 5 min in PIPkinase buffer (50 mM Tris pH 7.4, 10 mM MgCl2, 1 mM EGTA, 70 mM KCl) containing cold ATP (20 μM) and 1 μCi of [32P]ATP in a final volume of 100 μl. Reactions were quenched with 0.5 ml of chloroform:methanol [1:1 (v:v)] and the phases were split by the addition of 125 μl of 2.4 M HCl. The lower phases were removed to a new tube, dried and separated by TLC either using an acidic (chloroform:methanol:acetone:glacial acetic acid:water 240:78:72:70:42) or an alkaline solvent [chloroform:methanol:ammonia (28%):water 45:35:2:8]. Incorporation into PtdIns(4,5)P2 was quantitated using a phosphoimager.
To assess PLD and PIPkinase activity in the same reaction the following lipid vesicles were made: 3 nmol PtdOH, 1 nmol PtdInsP, 1 nmol PtdIns(4,5)P2, 10 nmol PtdSer and 2 nmol NBD‐labelled PtdCho. Reactions were carried out for 20 min, after which they were extracted as above and reaction products separated using the acidic solvent system. Incorporation into PtdIns(4,5)P2 was carried out as above, while the PLD activity was assessed by the production of NBD–PtdOH.
Type Iα PIPkinase deletions
Deletions in the 3′ direction were carried out using a constant 5′ primer tagged with the myc epitope and a unique restriction site, while deletions in the other direction were with a constant 3′ primer tagged with the myc epitope. PCR was carried out using Pfu and the reaction products were restricted and cloned into either pCDNA3 for eukaryotic expression, or in a GST vector for bacterial expression and purification. In some cases, the products were cloned into a GFP vector. Site‐directed mutagenesis was carried out using the Quick Change kit from Stratagene. All constructs were verified by sequencing.
GST–PIPkinase affinity purification of PLD2
Lysates were prepared from Cos‐7 cells expressing HA‐PLD2 and were incubated with GST fusion proteins purified from bacterial lysates by incubation with glutathione–Sepharose. Incubations were carried out for 2 h, after which the affinity beads were washed three times with immunoprecipitation buffer (see above). The beads were then placed in SDS loading buffer, separated by SDS–PAGE, transferred to nitrocellulose and probed using antibodies against the HA tag. GST alone, GST–Type IIα PIPkinase or glutathione beads were used as controls for this experiment. No binding to any of these was found.
Immunolocalizations in PAE cells
Constructs were microinjected into the nucleus of PAE cells, allowed to recover for 3 h and then fixed with formaldehyde (3.6% in PBS). The cells were permeabilized with Triton X‐100 (0.1% in PBS) and blocked with PBS–bovine serum albumin (1%). Coverslips were incubated with primary antibody [PIPkinase (1:50), anti‐HA (1:100)] for 1 h, washed in PBS three times and incubated with the corresponding secondary antibody conjugated to either fluorescein isothiocyanate or Texas red. The coverslips were washed with PBS, after which they were mounted using vectashield. Fluorescence was viewed using a Leica confocal microscope.
We would like to acknowledge the radioisotope department at the NKI for their constant help and advice, N.Ktistakis (The Babraham Insitute) for the PLD1 plasmid and antisera and members of H3 for advice and encouragement. N.D. is an AVL fellow. C.D‘S. is supported by the Dutch Cancer Society grant number NKB 99‐2055. Work in M.J.O.'s laboratory is supported by the Wellcome Trust.
- Copyright © 2000 European Molecular Biology Organization