Human ADP-ribosylation factor-activated phosphatidylcholine-specific phospholipase D defines a new and highly conserved gene family.

Activation of phosphatidylcholine-specific phospholipase D (PLD) has been implicated as a critical step in numerous cellular pathways, including signal transduction, membrane trafficking, and the regulation of mitosis. We report here the identification of the first human PLD cDNA, which defines a new and highly conserved gene family. Characterization of recombinant human PLD1 reveals that it is membrane-associated, selective for phosphatidylcholine, stimulated by phosphatidylinositol 4,5-bisphosphate, activated by the monomeric G-protein ADP-ribosylation factor-1, and inhibited by oleate. PLD1 likely encodes the gene product responsible for the most widely studied endogenous PLD activity.

Many cellular signaling events are initiated with the generation of biologically active molecules by enzymatic hydrolysis of phospholipids (1). Although the best understood system involves inositol lipids, the production of phosphatidic acid (PA) 1 from PLD-catalyzed hydrolysis of phosphatidylcholine (PC) is also a rapid and widespread response of cells challenged with diverse agonists (Fig. 1A). The range of cell types and signals reported implicate PLD activity in a broad spectrum of physiological processes and diseases, including metabolic regulation, inflammation, secretion, mitogenesis, oncogenesis, neural and cardiac stimulation, diabetes, and senescence (2). In vitro, PA modulates the activity of varied regulatory proteins, including certain protein serine/threonine and tyrosine kinases, Ras GTPase-activating protein, several proteins involved in cytoskeletal organization, and neutrophil respiratory burst oxidase. In vivo, however, PLD's role is untested.
PLD and its modes of regulation have become the subject of intense investigation. PLD activities from many mammalian tissues and cell lines have been reported that differ in their subcellular localization, pH optima, dependence on divalent cations and phosphatidylinositol 4,5-bisphosphate (PIP 2 ), susceptibility to inhibition by fatty acids (e.g. oleate) and detergents (2,3), and extent of activation by small G-proteins such as ARFs and RhoA and -B, Rac, and Cdc42 (4 -10). The essential role of ARF in controlling the binding of the heptameric coatomer complex to the surface of the Golgi cisternae (11) and the essential role of PIP 2 in secretion have focused attention on PIP 2 -dependent ARF-activated PLD activity as a mediator of vesicular trafficking, although the hypothesis has not been tested directly (12). RhoA and -B, Rac, and Cdc42 are implicated in signal transduction pathways that regulate both the actin cytoskeleton and mitogen-activated protein kinase cascades that lead to changes in transcription, suggesting that these small G-proteins act as intermediates in the process by which cell surface receptors activate PLD, but the molecular mechanisms underlying receptor regulation of PLD remain poorly defined. A major impediment toward resolution of these issues has been the absence of molecular definition of PLD genes in animals, particularly since the partially purified PLD preparations described to date may well have contained mixtures in different proportions of PLDs representing distinct gene products. We report here the cloning and characterization of the first animal PC-specific PLD, human PLD1 (hPLD1).

EXPERIMENTAL PROCEDURES
Molecular Isolation of hPLD1-The full-length hPLD1 cDNA was obtained from a Zap II HeLa cDNA library (Stratagene) as described in the text. Of the nine hPLD1 cDNAs sequenced (Sequenase version 2.0, U. S. Biochemical Corp.; a final sequence was determined from both strands), four represented mRNAs that terminated prematurely and five extended past the putative amino terminus by 48 -96 nucleotides. The presumed initiator methionine conforms to the eukaryotic consensus sequence and is the first in-frame methionine in the 5Ј-untranslated region. The coding region is additionally thought to be full length since an in-frame stop codon is located at nucleotides 15-17 of the cDNA. The entire 3Ј-untranslated region was probably not obtained, since a recognizable polyadenylation signal sequence is not present at the 3Ј-end.
Preparation of Substrates for Functional Assays-Phospholipids were purchased from Avanti Polar Lipids. PIP 2 was isolated as described (13 ated, selected, and propagated using standard methods (14). Monolayers of Sf9 cells (30 ϫ 10 7 cells in a 225-cm 2 flask) were infected with recombinant baculoviruses at a multiplicity of 10 and cultured at 27°C for 48 h. The cells were washed in ice-cold phosphate-buffered saline, scraped into ice-cold lysis buffer (25 mM Tris, pH 7.5, 2 mM EDTA, 1 mM dithiothreitol, 0.1 mM benzamidine, 0.1 mM phenylmethylsulfonyl fluoride), and disrupted by sonication, and the resultant suspension was centrifuged at 2000 ϫ g for 10 min at 4°C. The supernatant was centrifuged at 100,000 ϫ g for 1 h at 4°C. The supernatant from this second centrifugation was removed to give the cytosolic fraction and the pellet resuspended in lysis buffer. The resuspended pellet was centrifuged at 2000 ϫ g for 5 min. The supernatant obtained constituted the membrane fraction.
Transient Transfection Assays for hPLD1 Activity-A fragment of the hPLD1 cDNA encoding the entire open reading frame (nucleotides 93-3603 of the cDNA) was subcloned in-frame downstream of the cytomegalovirus promoter and the Flu epitope tag in the mammalian expression vector pCGN. COS-7 cells were transfected with 3 g of the resulting plasmid, pFlu-hPLD1, using lipofectamine (Life Technologies, Inc.). Forty-eight hours later, the cells were assayed for PLD activity.
PLD Assay-The PLD assay procedure measures release of the choline headgroup from radiolabeled PC and was based on a protocol previously described (4). The time course of choline release varied linearly over time and in proportion to the amount of protein added.
PLD Product Analysis-A detailed analysis of the water-soluble PLD reaction products was performed by cation-exchange HPLC using a 1-ml column of Source 15S resin (Pharmacia Biotech Inc.). For these analyses, incubations containing [ 3 H]PC (see above) were terminated by addition of the organic solvent mixtures described below, substituting H 2 O for HCl. Samples were applied to the column, which was washed with 10 ml of H 2 O and eluted with a 30-ml linear gradient of 0 -1 M CH 3 COONH 4 in water at a flow rate of 1 ml/min. 1-ml fractions were collected, and their radioactivity was determined by liquid scintillation counting. Compounds were identified by reference to authentic standards. For analysis of phospholipid products, the vesicles were of standard composition except that approximately 10 -20 ϫ 10 3 dpm of [ 32 P]PC, [ 32 P]PE, or [ 32 P]PI (specific radioactivity, 10,000 dpm/nmol) was substituted for the [ 3 H]PC. Incubations were exactly as described above except that transphosphatidylation assays contained 2% v/v EtOH. The assays were terminated by addition of 222 l of 1:1 CHCl 3 : MeOH. After mixing and centrifugation, the lower phases were removed, dried under vacuum, and analyzed by TLC on oxalate-impregnated Whatman 60A silica gel plates in a solvent system of CHCl 3 : MeOH:acetic acid (13:3:1, v/v). Products were visualized by autoradiography and identified by their mobilities relative to authentic standards.

RESULTS
Cloning of a Human PC-specific PLD cDNA-To identify cDNAs encoding PC-specific PLD activity, we took advantage of a screen that had uncovered a yeast PC-specific PLD gene (SPO14 (15)), which in turn identified a GenBank humanexpressed sequence tag (EST) encoding a significantly similar peptide sequence. A hybridization probe was generated from random-primed HeLa cDNA using polymerase chain reaction and primers specific to the EST and used to screen a HeLa cDNA library. Sequence analysis of hybridizing clones revealed a large open reading frame (ORF; Fig. 1B) encoding a 1072amino acid protein. hPLD1 appears to be remarkably devoid of recognized domain structures, in contrast to the various isoforms of phospholipase C (PLC), which contain SH2, SH3, or PH domains. Furthermore, no similarity exists between hPLD1 and PLC or phosphatidylinositol-glycan-specific PLD (16). As described below, hPLD1 is activated by PIP 2 , yet bears no similarity to proteins known to bind PIP 2 or IP 3 , such as PLC-␦, PI 3-kinase, or the IP 3 receptor (17). On the other hand, a data base search using hPLD1 identified homologs in numerous widely disparate organisms, demonstrating that PLD1 is a member of a novel but highly conserved gene family (Fig. 1C). Only one of these cDNAs was recognized to encode a PC-PLD M37842; Pseudomonas putida ORF, X55704; Streptococcus PLD, D16444; Bacillus subtilis ORF, Z49782.  (18)); the remainder constitutes ESTs or hypothetical open reading frames. A comparison of the related sequences reveals the location of several blocks of highly conserved amino acids, one or more of which might constitute critical portions of the catalytic or cofactor binding sites. Two regions in particular (amino acids 455-490 and 892-926) are highly conserved in all of the PLD1-related genes and contain an invariant charged motif, HXKXXXXD. PLD is thought to be intracellular and membrane-associated (not an integral membrane protein). Consistent with this prediction, PLD encodes neither a signal peptide nor a hydrophobic transmembrane sequence. hPLD1 Activity-To investigate its catalytic properties and activation requirements, hPLD1 was expressed in Sf9 insect cells using baculovirus and in COS-7 cells. A recombinant protein of approximately 120 kDa was observed ( Fig. 2A and data not shown), matching closely the theoretical size of 124 kDa and suggesting that little if any post-translational processing occurs. By definition, PLD catalyzes the hydrolysis of PC to PA and choline. To determine the activity encoded by recombinant hPLD1, control and hPLD1-encoding baculovirus-infected Sf9 cells were assessed using a standard headgroup release assay that measures the amount of tritiated headgroup (e.g. [ 3 H]choline) liberated by hydrolysis of the labeled substrate [ 3 H]PC (Fig. 2B). In lanes 1 and 3, modest levels of endogenous PLD activity are observed in control Sf9 cells (either uninfected or (data not shown) infected with an irrelevant construct), the majority of which (94%) is membraneassociated. In contrast, Sf9 cells infected with hPLD1-encoding baculovirus exhibit substantial cytosolic (32-fold above control) and membrane-associated (15-fold above control) PLD activity (lanes 2 and 4). To confirm that hPLD1 encodes a PLD activity, cation-exchange HPLC was used to analyze the water-soluble product(s) which co-eluted with a labeled choline standard (Fig.  2C), and thin layer chromatography was used to demonstrate concurrent production of PA (Fig. 2D, lane 1). Both membraneassociated and cytosolic PLD activities have been described in mammalian tissues and cell lines (2); it has been suggested that the membrane-associated and cytosolic PLD activities have distinct biochemical properties and thus might derive from different gene products. In contrast, we demonstrate here that a recombinant PLD activity derived from a single gene product is located both in the cytoplasm and in association with the membrane. This observation suggests instead that hPLD1 can exist as a stable soluble protein and that controlled interaction with substrate-containing phospholipid surfaces may be a physiologically important mode of regulation for this enzyme, as has been shown for phospholipase A 2 and PLC (19).
hPLD1 Selectivity-In addition to PC, PLD activities capable of hydrolyzing both PI and PE have been reported, and the issue of whether a single gene product mediates one or multiple activities has remained unresolved. To assess hPLD1's substrate selectivity, the standard assay was carried out using [ 32 P]PE and [ 32 P]PI (Fig. 2D, lanes 3 and 5). The results revealed that hPLD1 is unable to hydrolyze PE or PI. All PLD activities described to date also function as transphosphatidylases in the presence of primary alcohols, catalyzing the transfer of the phosphatidyl group from an appropriate substrate to the alcohol and thus generating phosphatidyl alcohol (20). To determine whether hPLD1 was capable of transphosphatidylase activity, EtOH was added to the reaction mixture and the products analyzed by TLC (Fig. 2D, lane 2). The results demonstrate that hPLD1 catalyzes the formation of [ 32 P]phosphatidylethanol when presented with [ 32 P]PC. Since PC-specific PLD is the only enzyme capable of catalyzing this particular reaction, we conclude that hPLD1 must be a PC-specific PLD.
Requirements for hPLD1 Activation-Two distinct rat brain PLD activities have been previously reported: one that is activated by PIP 2 and ARF but inhibited by oleate and a second that is insensitive to PIP 2 and ARF but requires oleate (6). Analysis of hPLD1 demonstrates that it is activated by PIP 2 and strongly inhibited by oleate (Fig. 3, A and B). The stimulation by PIP 2 (11-fold) is comparable in both magnitude and concentration dependence with effects reported previously for endogenous PLD activities (6,9,21). Similarly, the inhibition by oleate is comparable with that previously reported for a rat brain ARF-stimulated PLD activity (6).
Numerous reports have implicated monomeric G-proteins as PLD1, a PC-specific Phospholipase D Gene Activated by ARF 29642 regulators of PLD activities in a variety of mammalian cells and tissues (2, 4 -6). We found that recombinant human ARF1 strongly activates hPLD1 expressed in either baculovirus or mammalian cells (Fig. 3, C and D) and does so with a magnitude comparable with that previously reported. DISCUSSION PLD activities have been detected in essentially all organisms (2). Mammalian PLD activities in many tissues and cell lines have been studied in detail, although no clear biochemical classification of their properties has emerged. Moreover, attempts to devise such a classification have been complicated by reports of activities that differ in their subcellular localization, metal ion dependence, phospholipid and detergent requirements in exogenous substrate assays, and activation by various G-proteins and unidentified accessory proteins (2). Failure to isolate any of the enzymes in sufficient purity has left this matter unresolved. The PLD enzyme we have identified has properties that clearly implicate it as being responsible for at least part of the PLD activity previously observed in cell extracts and partially purified preparations. It is possible that some of the other PLD activities that have been described are mediated by additional members of the hPLD1 gene family. We have identified a second mammalian PLD gene (ϳ60% identical; data not shown), presenting an immediate possibility for another part of the reported human activity. Moreover, the existence of at least two human genes suggests that at least one reason that related but different properties have been reported for PLD has been that the activity obtained from different cell lines or tissues or using different purification techniques is actually composed of different mixtures of at least two distinct gene products that may have different biochemical properties and/or requirements for activation.
ARF-activated PLD is present in Golgi vesicles (22). The demonstration that hPLD1 is activated by ARF1 lends support to the hypothesis that PLD and specifically hPLD1 are involved in intravesicular membrane trafficking. Previous reports using partially purified PLD had raised the possibility that PLD and ARF interact directly (6,9). Our results extend previous efforts by using a single recombinant, purified protein as a PLD source, as well as recombinant, purified ARF as the activator. The data strengthen the hypothesis that interaction between ARF and hPLD1 is direct, although rigorous proof will require further experiments. Preliminary results suggest that at least some members of the Rho/Rac family of small G-proteins activate hPLD1 as well (data not shown). It will be important to determine which of the numerous small G-proteins reported to activate endogenous PLD actually activate hPLD1 to a significant extent.
Preliminary in vivo expression studies of PLD1 (and the second mammalian gene) in mouse embryos indicate that expression is detected at high levels in different selected regions of the brain and spinal cord (data not shown). These results raise the possibility that the PLD genes may also play a role in selected signal transduction events. In addition, there may be other members of the mammalian PLD gene family that are as of yet unidentified. The work presented in this paper paves the way for a molecular definition of the PLD enzymes. Ultimately, this advance should provide essential information for future studies designed to reveal the cellular and physiological function of these enzymes and their products.
FIG. 3. Effects of PIP 2 , oleate, and ARF1 on hPLD1 activity. PLD activity associated with membranes from Sf9 cells infected with the hPLD1 baculovirus was determined under standard assay conditions except that (A) the PIP 2 content of the vesicles was varied or (B) varying amounts of sodium oleate or (C) varying amounts of ARF1 were added to the incubations. 50 M GTP␥S was included in all assays containing exogenously added ARF1. The data shown are means Ϯ S.E. of triplicate determinations. Standard errors fell within the symbols for some data points. ϳ0.1 g of membrane protein was present in each sample. hARF-1, human ARF1. D, ARF1 activation of hPLD1 expressed in mammalian cells. COS-7 cells were transfected using Lipofectamine (Life Technologies, Inc.) with 3 g of LacZ or hPLD1 driven by a cytomegalovirus promoter and assayed after 48 h for PLD activity in the presence or absence or exogenously supplied ARF1 and GTP␥S. PLD1, a PC-specific Phospholipase D Gene Activated by ARF 29643