The Escherichia coli pgpB gene encodes for a diacylglycerol pyrophosphate phosphatase activity.

We provided genetic and biochemical evidence that supported the conclusion that the product of pgpB gene of Escherichia coli exhibited diacylglycerol pyrophosphate (DGPP) phosphatase activity. DGPP phosphatase activity was absent in pgpB mutant cells and was expressed at high levels in cells carrying the wild-type pgpB gene on a runaway replication plasmid. The pgpB mutant has been primarily characterized by a defect in phosphatidate (PA) phosphatase activity and also exhibits defects in lyso-PA phosphatase and phosphatidylglycerophosphate phosphatase activities. The defective PA phosphatase in the pgpB mutant was shown to be a Mg2+-independent PA phosphatase activity of the DGPP phosphatase enzyme. We characterized DGPP phosphatase activity in membranes from cells overproducing the pgpB gene product. DGPP phosphatase catalyzed the dephosphorylation of the β phosphate of DGPP to form PA followed by the dephosphorylation of PA to form diacylglycerol. The specificity constant (Vmax/Km) for DGPP was 9.3-fold greater than that for PA. The pH optimum for the DGPP phosphatase reaction was 6.5. Activity was independent of a divalent cation requirement, was potently inhibited by Mn2+ ions, and was insensitive to inhibition by N-ethylmaleimide. Pure DGPP phosphatase from Saccharomyces cerevisiae was shown to be similar to the E. coli DGPP phosphatase in its ability to utilize lyso-PA and phosphatidylglycerophosphate as substrates in vitro.

DGPP 1 is a novel phospholipid that was first identified from the plant Catharanthus roseus (1). It contains a pyrophosphate group attached to DG (Fig. 1). DGPP has subsequently been found in a variety of plants (2,3) and in the yeast Saccharomyces cerevisiae (4). DGPP is synthesized from PA and ATP via the reaction catalyzed by the membrane-associated enzyme PA kinase (1) and is dephosphorylated to PA via the reaction catalyzed by the membrane-associated enzyme DGPP phospha-tase (4) (Fig. 1). Recent studies have shown that DGPP and PA levels accumulate in plant tissues upon G protein activation (3,5,6). PA accumulation is the result of activated phospholipase C/DG kinase activities and the result of phospholipase D activity (5,6). DGPP accumulation is the result of activated PA kinase activity (3). PA is a phospholipid intermediate used for the synthesis of phospholipids and triacylglycerols (7)(8)(9) and plays a role in cell signaling. PA regulates the activity of several lipid-dependent enzymes (10 -13) and has mitogenic effects in mammalian cells (14 -17). In addition, PA is the source of the signaling lipids DG ( Fig. 1) (18) and lyso-PA (17). It is not yet clear what role DGPP plays in lipid metabolism and cell signaling, but it has been suggested that DGPP may function to attenuate the signaling functions of PA or may serve as a signaling molecule itself (3,4).
DGPP phosphatase has recently been purified to homogeneity from S. cerevisiae (4). The enzyme has a subunit molecular mass of 34 kDa (4). When DGPP is supplied as a substrate in vitro, the enzyme removes the ␤ phosphate of DGPP to generate PA and then removes the ␣ phosphate to generate DG (4). Although DGPP phosphatase can also utilize PA as a substrate in the absence of DGPP, the enzyme has a 10-fold higher specificity constant for DGPP (4). The PA phosphatase activity of the DGPP phosphatase enzyme is distinct from the conventional PA phosphatase enzyme (8,19,20) that is used for the synthesis of phospholipids and triacylglycerols in S. cerevisiae (4). The conventional PA phosphatase enzyme has a Mg 2ϩ ion requirement and is sensitive to inhibition by NEM (19,20). The PA phosphatase activity of the DGPP phosphatase enzyme does not have a Mg 2ϩ ion requirement and is insensitive to NEM (4).
Two forms of PA phosphatase activity exist in mammalian cells. The conventional form of the enzyme associates with the endoplasmic reticulum, has a Mg 2ϩ ion requirement, is inhibited by NEM, and is responsible for the synthesis of phospholipids and triacylglycerols (21). The other form of the enzyme is associated with the plasma membrane, does not have a Mg 2ϩ ion requirement, is insensitive to NEM, and is thought to be involved in lipid signaling pathways (21).
A PA phosphatase activity has been identified from the membrane fraction of Escherichia coli (22), and mutants defective in this activity have been isolated that are defined by the pgpB gene (23)(24)(25). The PA phosphatase activity in pgpB mutants has been measured using the assay conditions described for the conventional Mg 2ϩ -dependent PA phosphatase enzyme (23)(24)(25). However, these assay conditions would not reveal the existence of a Mg 2ϩ -independent PA phosphatase activity such as that displayed by the DGPP phosphatase enzyme. With this in mind, we tested the hypothesis that E. coli possessed a Mg 2ϩ -independent PA phosphatase activity of a DGPP phos-phatase enzyme. In this study we provided both genetic and biochemical evidence that the pgpB gene encoded for a DGPP phosphatase activity that can also utilize PA, lyso-PA, and phosphatidylglycerophosphate as substrates.

Materials
All chemicals were reagent grade. Growth medium supplies were purchased from Difco. Radiochemicals and ENHANCE were from Du-Pont NEN. Scintillation counting supplies were from National Diagnostics. Nucleotides, glycerol 3-phosphate, NEM, Triton X-100, isopropyl ␤-D-thiogalactoside, and bovine serum albumin were purchased from Sigma. Phospholipids were purchased from Avanti Polar Lipids and Sigma. Protein assay reagent was purchased from Bio-Rad. Silica Gel 60 thin-layer chromatography plates were from EM Science. E. coli DG kinase was obtained from Lipidex Inc.

Methods
Strains and Growth Conditions-The E. coli strains used in this work are listed in Table I. Strains CF10 and CF20 are mutants with disrupted alleles in the pgpA and pgpB genes, respectively (25). Strain CF30 is a mutant with disrupted alleles in both the pgpA and pgpB genes (25). Plasmid pTI5-217 is a runaway replication plasmid that contains the pgpB gene under the control of a tac promoter (24). Cultures were grown in LB medium and maintained on LB agar. Growth media were supplemented with ampicillin (100 g/ml), kanamycin (50 g/ml), or chloramphenicol (25 g/ml) as needed. Cultures were grown to the exponential phase of growth at 30°C and then harvested by centrifugation. Strain JM103 bearing plasmid pTI5-217 was shifted to 37°C and incubated for 1 h to increase the plasmid copy number (24). Isopropyl ␤-D-thiogalactoside (2 mM) was then added to the growth medium to induce the expression of the pgpB gene product (24). After incubation for 1 h, the induced cells were harvested by centrifugation.
Preparation of Enzymes-All steps were performed at 5°C. E. coli cells were suspended in 50 mM Tris-maleate (pH 7.0) buffer containing 1 mM Na 2 EDTA, 0.3 M sucrose, and 10 mM 2-mercaptoethanol. Cells were disrupted by sonic oscillation followed by centrifugation at 12,000 ϫ g for 10 min to remove unbroken cells and cell debris. The supernatant was then centrifuged at 100,000 ϫ g for 90 min to obtain the membrane fraction. Membranes were washed, resuspended, and dialyzed with 50 mM Tris-maleate (pH 7.0) buffer containing 20% glycerol and 10 mM 2-mercaptoethanol. These membranes were used as the source of DGPP phosphatase and PA phosphatase activities. S. cerevi-siae DGPP phosphatase (4) and C. roseus PA kinase (2) were purified as described previously.
Enzyme Assays-DGPP phosphatase activity was measured by following the release of water-soluble 32 P i from chloroform-soluble [␤-32 P]DGPP (5,000 -10,000 cpm/nmol) or by following the formation of [ 32 P]PA from [␣-32 P]DGPP (2,000 -5,000 cpm/nmol) as described by Wu et al. (4). The reaction mixture contained 50 mM Tris-maleate buffer (pH 6.5), 0.1 mM DGPP, 2 mM Triton X-100, 10 mM 2-mercaptoethanol, and enzyme protein in a total volume of 0.1 ml. The chloroform-soluble phospholipid product of the reaction, PA, was analyzed with standard PA and DGPP by thin-layer chromatography on potassium oxalatetreated plates using the solvent system chloroform/acetone/methanol/ glacial acetic acid/water (50:15:13:12:4) (4). The positions of the labeled phospholipids on the chromatograms were determined by autoradiography. The amount of labeled phospholipids was determined by scintillation counting.
PA phosphatase activity was measured by following the release of water-soluble 32 P i from chloroform-soluble [ 32 P]PA (10,000 cpm/nmol) (28). The reaction mixture contained 50 mM Tris-maleate buffer (pH 7.0), 0.1 mM PA, 1 mM Triton X-100, 2 mM Na 2 EDTA, 10 mM 2-mercaptoethanol, and enzyme protein in a total volume of 0.1 ml. PA phosphatase activity was also measured in a reaction mixture which included 2 mM MgCl 2 and omitted Na 2 EDTA.
PGP phosphatase activity was measured using a coupled enzyme assay where the PGP substrate of the reaction was generated by the PGP synthase reaction (23). The reaction mixture contained 50 mM Tris-maleate buffer (pH 7.0), 0.5 mM [ 3 H]glycerol 3-phosphate (40,000 cpm/nmol), 0.1 mM CDP-DG, 1.5 mM Triton X-100, 10 mM MgCl 2 , 10 mM 2-mercaptoethanol, 5 mM NEM, 0.1 mg/ml E. coli strain CF30 membrane protein, and 10 pmol/min purified yeast DGPP phosphatase in a total volume of 0.1 ml. The chloroform-soluble phospholipid products of the reaction, PGP and PG, were analyzed by thin-layer chromatography using the solvent system chloroform/methanol/glacial acetic acid/water (50: 30:4:8). The positions of the labeled phospholipids on the chromatograms were determined by fluorography.
All enzyme assays were conducted at 30°C in triplicate. The average standard deviation of the assays was Ϯ5%. The enzyme reactions were linear with time and protein concentration.
A unit of enzymatic activity was defined as the amount of enzyme that catalyzed the formation of 1 nmol of product/min. Specific activity was defined as units/mg of protein. Protein concentration was determined by the method of Bradford (29) using bovine serum albumin as the standard.

PA Phosphatase and DGPP Phosphatase Activities in pgpB Mutant Cells and in Cells
Overexpressing the pgpB Gene Product-The original E. coli pgpB mutant was isolated in a biochemical screen that was designed to isolate cells defective in PGP phosphatase activity (23). Biochemical analysis has  shown that pgpB mutant cells are also defective in PA phosphatase and lyso-PA phosphatase activities (23). These mutant cells exhibit the greatest reduction in PA phosphatase activity (23). We examined the levels of PA phosphatase activity in membranes derived from a well characterized mutant with a pgpB allele that has been disrupted by the insertion of an Amp r gene fragment (25). This mutant has the same properties as the original pgpB mutant (23,25). We measured PA phosphatase activity in the presence of Mg 2ϩ ions as described previously (23)(24)(25). The level of PA phosphatase activity in the pgpB mutant was 75% lower than the activity found in its isogenic wild-type parent strain ( Fig. 2A). We next examined if E. coli possessed a Mg 2ϩ -independent PA phosphatase activity. In these experiments, the reaction mixture also contained Na 2 EDTA, and the membranes used for the enzyme assays were dialyzed to remove divalent cations. Indeed, a Mg 2ϩindependent activity was expressed in E. coli cells. Whether Mg 2ϩ ions were present or not, the levels of PA phosphatase activity were similar in wild type cells and were reduced to the same extent in pgpB mutant cells (Fig. 2A).
The pgpB gene has been cloned, sequenced, 2 and shown to be the structural gene for the PA phosphatase activity defined by the pgpB mutation (24). The protein product encoded by the pgpB gene is expressed at elevated levels in cells bearing the wild-type pgpB gene under the control of a tac promoter on a runaway replication plasmid (24). These cells also overexpress PA phosphatase, lyso-PA phosphatase, and PGP phosphatase activities (24). The overexpression of the pgpB gene product also resulted in elevated levels (370-fold) of the PA phosphatase activity that was measured in the absence of Mg 2ϩ ions ( Fig. 2A).
We next examined the levels of DGPP phosphatase activity. In wild-type cells, the specific activity of DGPP phosphatase (Fig. 2B) was approximately 6-fold greater than that of the PA phosphatase activities measured in the presence or absence of Mg 2ϩ ions ( Fig. 2A). DGPP phosphatase activity was not detected in pgpB mutant cells (Fig. 2B). Furthermore, the specific activity of DGPP phosphatase from cells that overexpress the pgpB gene product was 310-fold greater than that found in wild-type cells (Fig. 2B).
A pgpA mutant was isolated in the same biochemical screen that identified the pgpB mutant (23). This mutant exhibits a defect in PGP phosphatase activity, while PA phosphatase and lyso-PA phosphatase activities are present at wild-type levels (23). We examined PA phosphatase (measured in the presence or absence of Mg 2ϩ ions) and DGPP phosphatase activities in a mutant in which the pgpA gene was disrupted (25). This mutant has the same properties as the original pgpA mutant (23,25). The levels of PA phosphatase ( Fig. 2A) and DGPP phosphatase (Fig. 2B) activities from pgpA mutant cells were the same as that from its isogenic wild-type parent (Fig. 2B).
Properties DGPP Phosphatase and Mg 2ϩ -independent PA Phosphatase Activities-The membrane fraction of strain JM103/pTI5-217 was used for the characterization of DGPP phosphatase and Mg 2ϩ -independent PA phosphatase activities. The elevated expression of the pgpB gene product from these cells represents a considerable enrichment of these activities that is equivalent to a nearly 600-fold purification over that expressed in the cell extract of wild-type E. coli.
We examined the thermolability of DGPP phosphatase and PA phosphatase activities. Membranes were heated at 50°C, and samples were removed at various time intervals. DGPP phosphatase and PA phosphatase activities were then measured at 30°C under standard assay conditions. In this experiment, DGPP phosphatase and PA phosphatase activities were inactivated with identical kinetics (Fig. 3). A t1 ⁄2 of 55 s for the inactivation of both phosphatase activities was calculated from a replot of the log activity versus time of incubation (Fig. 3).
The time dependence of the DGPP phosphatase reaction was examined using [␣-32 P]DGPP as the substrate (Fig. 4). By using the ␣-labeled substrate we could distinguish the removal of the ␤ phosphate of DGPP from the removal of the pyrophosphate moiety of DGPP. The water-soluble fraction of the reaction was also analyzed. DGPP phosphatase catalyzed the dephosphorylation of the ␤ phosphate of DGPP, and after an initial lag period, catalyzed the dephosphorylation of the ␣ 2 GenBank TM accession no. M23628. phosphate of DGPP (Fig. 4B). Thus, DGPP phosphatase dephosphorylated DGPP to PA and subsequently dephosphorylated PA to DG.
The kinetics of activity toward DGPP and PA were examined using Triton X-100/phospholipid-mixed micelles. This mixed micellar system allows the analysis of lipid-dependent enzymes using surface dilution kinetics, a model system which mimics the physiological surface of the membrane (30). Positive cooperative kinetics were exhibited with respect to the surface concentration of DGPP (Fig. 5). The data were analyzed according to the Hill equation using the EZ-FIT Enzyme Kinetic Model Fitting Program (31). This analysis yielded a Hill number of 1.45, a K m value for DGPP of 2.3 mol %, and a V max value of 2,167 nmol/min/mg. Positive cooperative kinetics were also exhibited with respect to the surface concentration of PA (Fig.  5). Analysis of this data according to the Hill equation gave a Hill number of 3.9, a K m value for PA of 3.1 mol %, and a V max value of 313 nmol/min/mg. The specificity constants (V max /K m ) for DGPP and PA were 942 and 101, respectively.
Maximum DGPP phosphatase activity was observed at pH 6.5 (Fig. 6). DGPP phosphatase activity was independent of any divalent cation requirement and was stimulated (35%) by 2 mM Na 2 EDTA. DGPP phosphatase activity was not significantly affected by the addition of 5 mM NEM to the assay system. The addition of Mn 2ϩ ions to the assay resulted in a dose-dependent inactivation of DGPP phosphatase activity (Fig. 7). This inactivation followed positive cooperative kinetics (Hill number of 3) and the IC 50 value for MnCl 2 was calculated to be 34 M. PA phosphatase activity was also inhibited by Mn 2ϩ ions (data not shown). Furthermore, the Mg 2ϩ -independent PA phosphatase activity was stimulated (25%) by 2 mM Na 2 EDTA and was not significantly affected by 5 mM NEM. The addition of 1 mM MnCl 2 to the assay systems for DGPP phosphatase and PA phosphatase (measured in the presence or absence of Mg 2ϩ ions) resulted in the total inhibition of all three activities in wild-type cells, in cells that overproduced the pgpB gene product, and in pgpA mutant cells (data not shown).
The DGPP Phosphatase from S. cerevisiae Can Utilize Lyso-PA and PGP as Substrates-Based on the ability of the pgpB gene product to utilize DGPP, PA, lyso-PA, and PGP as substrates, we examined the ability of pure DGPP phosphatase from S. cerevisiae to utilize lyso-PA and PGP as substrates. Lyso-PA phosphatase activity was measured by following the dephosphorylation of [ 32 P]lyso-PA under the optimum assay conditions for DGPP phosphatase activity (4). The yeast DGPP phosphatase enzyme catalyzed the dephosphorylation of lyso-PA in a dose-dependent manner (Fig. 8) was measured by a coupled enzyme assay using PGP synthase activity to generate the PGP substrate for the reaction. Membranes from E. coli strain CF30 were used as the source of PGP synthase. This strain contains disrupted alleles in both the pgpA and pgpB genes and is partially defective in PGP phosphatase activity (25). Thus, the PGP synthesized via the PGP synthase reaction accumulates and can be used for the reaction catalyzed by the yeast DGPP phosphatase. The yeast DGPP phosphatase was incubated with CDP-DG, [ 3 H]glycerol-3-phosphate, and the E. coli membranes. Following incubation, the chloroform-soluble products of the reaction were analyzed by thin-layer chromatography. In this experiment, yeast DGPP phosphatase catalyzed a time-dependent reaction utilizing the PGP generated in the PGP synthase reaction for the synthesis of PG (Fig. 9).

DISCUSSION
In this study we tested the hypothesis that the E. coli pgpB gene encoded a DGPP phosphatase activity. The product of the pgpB gene is a 28-kDa membrane-associated enzyme which has the ability to dephosphorylate PA, lyso-PA, and PGP in vitro (23)(24)(25). We questioned whether the PA phosphatase activity defined by the pgpB gene (23) was in fact a Mg 2ϩ -independent activity of the DGPP phosphatase enzyme similar to the activity characterized in S. cerevisiae (4). We showed that E. coli possessed a PA phosphatase activity that was independent of any Mg 2ϩ ion requirement and was insensitive to inhibition by NEM. The expression of the Mg 2ϩ -independent PA phosphatase activity was reduced in pgpB mutant cells and elevated in cells overexpressing the wild-type pgpB gene. A residual amount of PA phosphatase activity was present in the pgpB mutant indicating that the pgpB gene product was not the only source of PA phosphatase activity in E. coli. DGPP phosphatase activity was present in E. coli and its expression was not detected in the pgpB mutant. Thus, the lack of DGPP phosphatase activity was the most pronounced phenotype of the pgpB gene.
Our data indicated that the DGPP phosphatase and PA phosphatase activities were due to a single enzyme. DGPP phosphatase activity was overproduced in cells overexpressing the pgpB gene product and the extent of its overproduction paralleled that of the Mg 2ϩ -independent PA phosphatase activity. Both activities were inhibited by Mn 2ϩ ions and were insensitive to inhibition by NEM. In addition, DGPP phosphatase and PA phosphatase activities exhibited the same kinetics of temperature inactivation. The substrate dependence experiments demonstrated that DGPP was a better substrate for the enzyme when compared with PA. The specificity constant for DGPP was 9.3-fold higher than that for PA. In addition, the enzyme exhibited greater cooperative kinetic behavior toward PA when compared with DGPP. This substrate preference was consistent with the elevated DGPP phosphatase activity relative to the PA phosphatase activity found in the various E. coli strains.
The enzymological properties (e.g. dephosphorylation of the ␤ phosphate of DGPP followed by the dephosphorylation of the ␣ phosphate of DGPP, pH optimum, stimulation by Na 2 EDTA, inhibition by Mn 2ϩ ions, and insensitivity to NEM) of the E. coli DGPP phosphatase were similar to those described for the enzyme from S. cerevisiae (4). Moreover, the yeast DGPP phosphatase was similar to the E. coli enzyme with respect to the utilization of lyso-PA and PGP as substrates in vitro.
PA phosphatase enzymes have been purified from rat liver (32,33) and porcine thymus (34) membranes. Interestingly, these mammalian enzymes (32)(33)(34) share properties that are strikingly similar to those of the PA phosphatase activity of the DGPP phosphatase enzymes from S. cerevisiae (4) and E. coli. For example, the mammalian enzymes do not have a Mg 2ϩ ion requirement, they are insensitive to inhibition by NEM, and they are inhibited by Mn 2ϩ ions (32)(33)(34). In addition, the rat liver Mg 2ϩ -independent PA phosphatase, like the DGPP phosphatase enzymes from S. cerevisiae and E. coli, can utilize lyso-PA as a substrate in vitro (35). It will be interesting to examine whether this rat liver Mg 2ϩ -independent PA phosphatase can utilize DGPP as a substrate.
The rat liver Mg 2ϩ -independent PA phosphatase also utilizes ceramide 1-phosphate and sphingosine 1-phosphate as substrates in vitro (35). These lipid phosphate compounds as well as PA and lyso-PA have been shown to be mediators of cell activation and signal transduction in mammalian cells (17, 36 -39). It has been suggested that the rat liver Mg 2ϩ -independent PA phosphatase may play a role in regulating the balance of these lipid mediators (35). However, it is unclear whether the rat liver Mg 2ϩ -independent PA phosphatase uti- lizes all of these substrates in vivo. Similarly, it is unclear whether the DGPP phosphatase enzymes from S. cerevisiae and E. coli utilize all of their respective substrates in vivo. It is known that the PGP phosphatase activity encoded by the E. coli pgpB gene product is not essential for the synthesis of PG in vivo (25).
An understanding of the function of DGPP in phospholipid metabolism and cell signaling will require a combination of genetic, molecular, and biochemical approaches. This will require the cloning of those genes that encode enzymes involved in DGPP metabolism and the purification and characterization of the products of these genes. Our laboratories have initiated these studies by purifying a PA kinase from plants (2) and purifying a DGPP phosphatase from S. cerevisiae (4). In preliminary studies, we have identified PA kinase activities in S. cerevisiae (4) and in E. coli 3 and are developing methods for their purification. The finding of DGPP phosphatase and PA kinase activities in S. cerevisiae (4) and E. coli suggests a potential role of DGPP metabolism in these organisms that may be similar to that described in plants (3). The identification of the pgpB gene as the gene encoding for a DGPP phosphatase in E. coli should facilitate genetic and molecular studies aimed at gaining an understanding of the role(s) of DGPP metabolism in E. coli and may also facilitate the isolation of genes encoding for DGPP phosphatase from eukaryotic organisms.