Inhibition of phospholipase D by a protein factor from bovine brain cytosol. Partial purification and characterization of the inhibition mechanism.

A specific protein inhibitor of partially purified bovine brain phospholipase D (PLD) was identified from bovine brain cytosol. The PLD inhibitor has been enriched through several chromatographic steps and characterized with respect to size and mechanism of inhibition. The inhibitor showed an apparent molecular mass of 30 kDa by Superose 12 gel exclusion chromatography and inhibited PLD activity with an IC50 of 7 nM. The inhibitor had neither proteolytic activity nor phospholipid-hydrolyzing activity. Because phosphatidylinositol 4,5-bisphosphate (PIP2), which is included in substrate vesicles, is an essential cofactor for PLD, we examined whether the inhibition might be mediated by sequestration of PIP2. PIP2 hydrolysis by phospholipase C (PLC)-β1 was not affected by the inhibitor and the inhibitor did not bind to substrate vesicles containing PIP2. In contrast, a PH domain derived from PLC-δ1, which could bind to PIP2, showed a nearly identical inhibition of both PLC-β1 and PLD activities. Thus, the PLD inhibition by the inhibitor is due to the specific interaction with not PIP2 but PLD. The suppression of PLD activity by the inhibitor was largely eliminated by the addition of ADP-ribosylation factor (ARF) and GTPγS. We propose that the inhibitor plays a negative role in regulation of PLD activity by PIP2 and ARF.

Phospholipase D (PLD) 1 plays an important role in signal transduction of a variety of cells (1,2). The enzyme hydrolyzes phosphatidylcholine (PC) to produce phosphatidic acid (PA) and choline. PA has been implicated as a biologically active molecule and can also be metabolized via PA phosphohydrolase to form diacylglycerol (3)(4)(5). The diacylglycerol from PC has been known as a sustained activator of protein kinase C (PKC) (5).
While PLD-activating mechanisms have been described, the negative regulation of PLD is poorly understood. The existence of PLD-inhibitory factors has been suggested by several research groups (22)(23)(24)(25)(26). It was recently suggested that an inhibitor of ARF-dependent PLD activity exists in bovine brainderived membranes (25). In addition, an inhibitor of ARFdependent PLD activity in permeabilized HL-60 cells was detected in bovine brain cytosol (26).
The present work was performed to characterize the inhibitor and its mechanism of inhibition of PLD. Herein, we demonstrate that a 30-kDa inhibitor from bovine brain cytosol specifically inhibits PLD activity.

Measurement of PLD Activity
PLD activity was measured essentially as described previously by Brown et al. (9) with slight modifications. Partially purified bovine brain PLD was reconstituted in mixed phospholipid vesicles with the column fractions containing PLD-inhibitory activity. The phospholipid vesicles (65.56 M phospholipids) comprised PE, PIP 2 , and dipalmitoyl-[methyl-3 H]choline in a molar ratio of 16:1.4;1 (total volume of 150 l), and the reconstitution buffer was 50 mM HEPES (pH 7.5), 3 mM EGTA, 2 mM CaCl 2 , 3 mM MgCl 2 , 1 mM dithiothreitol, and 80 mM KCl. The reaction mixture was incubated at 37°C and terminated by the addition of 1 ml of chloroform/methanol/HCl (50:50:0.3) and 0.3 ml of 1 N HCl. The mixture was shaken and centrifuged at 2000 ϫ g for 5 min. An aliquot of the supernatant (0.5 ml containing released [ 3 H]choline) was removed, and the released [ 3 H]choline was counted by liquid scintillation spectroscopy. To determine whether the PLD inhibitor decreases the transphosphatidylation activity of PLD, PLD activity was measured under the same condition except that the radiolabeled substrate was replaced with dipalmitoyl[2-palmitoyl-9 -10-3 H]PC (0.5 Ci) and 1% ethanol (v/v). The produced [ 3 H]phosphatidylethanol (PEtOH) was separated by thin layer chromatography as described (27). The PEtOHproducing activity as well as the choline-releasing activity of PLD was decreased by addition of the inhibitory fraction (data not shown).

Cell-free Assay of PLD Using Human Neutrophils
PLD activities of plasma membrane from human neutrophil were measured as described (14). Cells were labeled with [ 3 H]alkyllysophosphatidylcholine (1.5 Ci/2 ϫ 10 7 cells/ml) for 90 min at 37°C. Plasma membranes and cytosol were isolated as described (28). Incubations containing plasma membrane plus cytosol were carried out in the presence of 10 M GTP␥S, 1 M CaCl 2 , and 1.6% ethanol, and the reaction was terminated by transfer to chloroform:methanol (1:2). The produced [ 3 H]PEtOH was separated by thin layer chromatography as described (27).

Preparation of Bovine Brain Membranes and Cytosol
Bovine brains were obtained from a local slaughter house. All steps were performed at 4°C. Six brains (about 2 kg) were homogenized with a polytron homogenizer in 4 volumes (8000 ml) of buffer A (20 mM Tris-HCl, pH 7.4, 1 mM EGTA, 1 mM EDTA, 0.2 mM phenylmethylsulfonyl fluoride, and 0.2 mM ␤-mercaptoethanol). The homogenate was filtered through two layers of cheesecloth, and the resultant pellet (membranes) and supernatant (cytosol) were obtained by centrifugation at 100,000 ϫ g for 60 min. Membranes were washed once by suspension in buffer A and the repeated centrifugation.

Partial Purification of PLD from Bovine Brain Membrane
PLD was prepared from bovine brain membranes essentially as described previously (25). Membranes were solubilized by buffer B (20 mM Hepes, pH 7.4, 1 mM EDTA, 1 mM dithiothreitol, 0.4 M NaCl, and 0.1 mM phenylmethylsulfonyl fluoride) containing 1% (w/v) sodium cholate, and the mixture was incubated for 1 h at 4°C with stirring. Insoluble material was removed by centrifugation at 100,000 ϫ g for 60 min. PLD was partially purified through two sequential chromatographic steps including Sephadex G-50 and SP-Sepharose columns. Fractions containing PLD activity through SP-Sepharose were pooled and used for measurement of PLD activity. PIP 2 was the essential component for detection of PC hydrolysis by the PLD preparation. This PIP 2 -dependent PLD activity did not show GTP␥S-dependent activation, suggesting that the preparation is free from PLD-activating G proteins. We also confirmed that the PLD preparation contained neither ARF nor RhoA, as judged by immunoblot analysis using anti-ARF and anti-RhoA antibodies.
Purification of an Inhibitory Factor of PLD Activity from Bovine Brain Cytosol Step 1: Chromatography with Heparin-Agarose-Cytosol (1 liter, 5 g of protein) from a bovine brain was loaded onto a heparin agarose column (5 ϫ 20 cm) equilibrated with buffer C (20 mM Hepes, pH 7.0, 1 mM EDTA, 1 mM EGTA). The column was eluted with a 1.2-liter gradient of 0.05-0.5 M NaCl in buffer C and followed by a 1.2-liter gradient of 0.5-1.5 M NaCl in buffer C. The major peak of PLD-inhibitory activity was eluted at 0.25-0.35 M NaCl. The inhibitory fraction from a heparinagarose column was combined with the inhibitory fractions from another five identical preparations.
Step 2: Chromatography with Butyl-Toyopearl-The PLD-inhibitory fractions from Step 1 were adjusted to 3 M with solid potassium chloride and then applied to a butyl-Toyopearl column (5 ϫ 20 cm) equilibrated with buffer B containing 3 M KCl. Proteins were eluted with 2.4 liters of linear gradient 3-0 M KCl in buffer B. The PLD-inhibitory activity was eluted at 1.0 -0.5 M NaCl.
Step 3: Chromatography with TSK Gel DEAE-5PW HPLC-Fractions containing PLD-inhibitory activity from Step 2 were pooled (300 ml), concentrated to 40 ml on an Amicon YM10 filter, and dialyzed against buffer C (20 mM Tris-HCl, pH 7.4, 1 mM EGTA, 1 mM EDTA). The dialyzed fractions were loaded onto a TSK gel DEAE-5PW (21.5 ϫ 150 mm) column equilibrated with buffer C. Proteins were eluted (flow rate, 5 ml/min) with equilibration buffer for 5 min, followed by a linear gradient from 0 to 0.3 M KCl in 45 min and a second linear gradient from 0.3 M to 1 M KCl in 10 min. The PLD-inhibitory activity was eluted at 0.25-0.3 M KCl.
Step 4: Chromatography with TSK Gel Blue-5PW-Fractions containing PLD-inhibitory activity from Step 3 were pooled, and the pooled fractions were dialyzed against buffer B and loaded onto TSK gel blue-5PW column (7.5 ϫ 75 mm) equilibrated with buffer B. Bound proteins were eluted (flow rate, 1 ml/min) with equilibration buffer for 5 min, followed by a linear gradient from 0 to 0.4 M KCl in 40 min. The  PLD-inhibitory activity was eluted at around 0.3 M KCl.
Step 5: Chromatography with TSK Gel Hydroxyapatite-5PW-The PLD-inhibitory fractions from Step 4 were pooled and dialyzed against buffer D (20 mM potassium phosphate, pH 7.5) and loaded onto a TSK gel hydroxyapatite-5PW column (7.5 ϫ 75 mm) equilibrated with buffer D. Bound proteins were eluted with 40 ml of a linear gradient from 0.02 to 0.3 M potassium phosphate in 40 min. The PLD-inhibitory activity was eluted around at 0.2 M potassium phosphate.
Step 6: Chromatography with TSK Gel SP-5PW-The PLD-inhibitory fractions from Step 5 were pooled and dialyzed in buffer E (20 mM Mes, pH 6.5, 1 mM EGTA, 1 mM EDTA). The dialyzed fraction was loaded onto TSK gel SP-5PW column (7.5 ϫ 75 mm) equilibrated with buffer E. Proteins were eluted with 40 ml of a linear gradient from 0.02 to 0.3 M NaCl in 40 min. PLD-inhibitory activity was eluted around at 0.2 M NaCl.
Step 7: Gel Filtration with Superose 12-Fractions containing PLDinhibitory activity from Step 6 were pooled and concentrated to 0.4 ml with an Centricon-10. The preparation was applied to a Superose 12 column (HR 10/30) equilibrated with buffer C containing 150 mM NaCl at a flow rate of 0.4 ml/min. Aliquots of 2 l of each fraction were tested for their ability to inhibit PLD activity.

Preparation of Pleckstrin Homology Domain of PLC-␦1
cDNA corresponding to pleckstrin homology (PH) domain (residues 1-133) of PLC-␦1 was amplified by the polymerase chain reaction from PLC-␦1 cDNA (29). The amplified cDNA was then subcloned into pGEX-3T plasmid. The recombinant protein was introduced into the Escherichia coli strain DH5␣ and expressed as glutathione S-transferase (GST) fusion protein. The fusion protein (GST-PHPLC␦1) was purified by using glutathione-Sepharose column chromatography. GST-PHPLC␦1 fusion protein prepared was about 95% pure as judged by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and Coomassie staining.

Measurement of Binding of the Inhibitor to Phospholipid Vesicles
The binding ability of the PLD-inhibitory factor to substrate vesicles was estimated by centrifugation assay as described previously (30). The final assay volume was 0.15 ml containing 50 mM HEPES (pH 7.5), 3 mM EGTA, 2 mM CaCl 2 , 3 mM MgCl 2 , 1 mM dithiothreitol, 80 mM KCl, and 15 g of bovine serum albumin plus the indicated concentration of phospholipid vesicles and PLD-inhibitory fraction. After incubation at room temperature for 10 min, the mixture was centrifuged at 100,000 ϫ g for 60 min. An aliquot of the supernatant was removed for measurement of the PLD-inhibitory activity.

Preparation of ARF
Recombinant ARF was overexpressed in E. coli as described (31). For the N-myristoylation of ARF, pBB131 vector containing the N-myristoyltransferase gene-1 (NMT) was cotransformed into the cells. The overexpressed ARF was purified to near homogeneity through sequential chromatography on DEAE-cellulose and Superose 12 gel filtration column.

Miscellaneous Methods
Bovine plasma gelsolin was purified from bovine serum as described (32). The final preparations of plasma gelsolin were about 70% pure, as judged by Coomassie staining of sample after SDS-PAGE. PLC-␤1 was purified from bovine brain as described previously (33). Protein concentrations were estimated by the assay of Bradford and Rubin (34) with bovine serum albumin as the standard. SDS-PAGE was according to Laemmli (35).

Bovine Brain Cytosol Contains an Inhibitor of PLD Activity-
Previous studies have demonstrated several PLD-activating factors, such as ARF, RhoA, and Cdc42 in cytosolic fractions (9 -19). However, we found that bovine brain cytosol inhibited the PIP 2 -dependent PLD activity both in the absence and in the presence of GTP␥S (Fig. 1). This result suggests that strong inhibitory constraints of PLD activity may exist.
Partial Purification of an Inhibitor of PLD Activity-A PLDinhibitory factor was purified from bovine brain cytosol about 1000-fold with a yield of 0.08% (Table I)  RhoGDI (GDP dissociation inhibitor) by heparin-agarose chromatography (data not shown). The inhibitor was eluted at a volume corresponding to a molecular mass of approximately 30 kDa using Superose 12 column chromatography ( Fig. 2A). When the column fractions from the Superose 12 were loaded onto SDS-PAGE, a protein band with a molecular mass of about 50 kDa correlated with the inhibitory activity (Fig. 2B). However, the protein band did not coincide with the inhibitory activity in the TSK gel SP-5PW column chromatographic step (data not shown). Although we tried five times to purify the inhibitor with many different combinations of column chromatographic steps, any candidate protein band was not observed in SDS-PAGE with either Coomassie or silver staining (data not shown). The inhibitory fraction from the Superose 12 column decreased PLD activity in a dose-dependent manner (Fig. 3). When the inhibitory fractions were boiled at 100°C for 5 min, the inhibitory activity disappeared (Fig. 4A). Tryptic digestion also resulted in the loss of inhibitory activity (Fig.  4B). From these results, the inhibitory factor appears to be a protein.
Specific Inhibition of PLD Activity by the Inhibitor-Since PLD requires PIP 2 in the substrate vesicles as a cofactor, a PIP 2 -binding protein is a candidate for the PLD-inhibitory factor. In addition to its role as a cofactor, PIP 2 is used as substrate for PLC. Therefore, we investigated the effect of the inhibitory factor and PIP 2 -binding protein on PLC and PLD activities. As shown in Fig. 5A, PLD activity decreased by increasing concentrations of the inhibitor, while PIP 2 hydrolysis by PLC-␤1 was not changed. To understand the effect of a PIP 2 -binding protein on both enzyme activities, we used the PH domain of PLC-␦1. The PH domain of PLC-␦1 mediates binding to PIP 2 (30,36,37). The PH domain was expressed in E. coli as a GST-fused protein (GST-PHPLC␦1). PLD and PLC-␤1 activities were inhibited by the addition of GST-PHPLC␦1 in a dose-dependent manner as shown in Fig. 5B. About 0.7 M GST-PHPLC␦1 inhibited the PLD and PLC activities to 60% in the presence of 5 M PIP 2 -containing vesicles.
The PLD Inhibitor Does Not Bind to Substrate Vesicles-We investigated the ability of the PLD inhibitor to bind to sub-

FIG. 4. Effects of boiling and trypsin treatment of the inhibitor on PLD-inhibitory activity.
Boiling of PLD-inhibitory fraction exhibited the loss of inhibitory activity. Aliquots (100 g) of the inhibitory fraction (TSK gel blue-5PW) was boiled at 100°C for 5 min, and aggregated proteins were removed by centrifugation. The PLD-inhibitory activities of the supernatant were measured for boiled fractions (closed circle) and control fractions (open circle) as described in Fig. 3. A, tryptic digestion was performed in the reaction mixture containing 50 mM Hepes-NaOH (pH 7.5), 3 mM CaCl 2 , 1 mM EGTA, and trypsin. Indicated amounts of the inhibitory fraction (TSK gel blue-5PW) and trypsin were incubated in the reaction mixture at 37°C for 6 h. For measuring the effect of trypsin, buffer from TSK gel blue-5PW was added in the mixture instead of the inhibitory fraction. PLD-inhibitory activities of the incubated mixtures were analyzed. B, the data shown are the mean Ϯ S.E. of three independent experiments. strate vesicles containing PIP 2 . The PLD inhibitor or GST-PHPLC␦1 was incubated with the indicated concentrations of lipid vesicles containing PE, PIP 2 , and PC (16:1.4:1, molar ratio), and lipid vesicles were pelleted by centrifugation. GST-PHPLC␦1 was precipitated by addition of phospholipid vesicles containing PIP 2 in a dose-dependent manner, whereas the PLD-inhibitory factor did not bind to substrate vesicles under PLD assay conditions (Fig. 6).
The Inhibitor Fraction Contains Neither Protease nor Lipase-In addition to PIP 2 binding, an alternative mechanism is that the inhibitor may be a PIP 2 -hydrolyzing PLC. However, this is unlikely because we could not detect PIP 2 -hydrolyzing activity by the inhibitor, and PIP 2 -hydrolyzing activity did not coincide with the PLD-inhibitory activity during chromatographic steps (data not shown). Hydrolysis of the phospholipids in the substrate vesicle by lipases might inhibit the PLD activity by changing the physical environments of the substrate vesicles. To test whether the PLD inhibitor might have lipase activity, we incubated the inhibitor with substrate vesicles. The phospholipids were then analyzed by thin layer chromatography. The PLD inhibitor did not hydrolyze PE, PC, and PIP 2 (data not shown). A protease might also inhibit the PLD by hydrolyzing the enzyme. To exclude the possibility of a contaminating protease in the PLD-inhibitory fraction, the PLD was preincubated with the inhibitor for 30 min at 37°C before adding substrate vesicles. Preincubation failed to affect the potency of the inhibitor (data not shown), indicating that a time-dependent hydrolysis of the PLD enzyme was not taking place. Also, the PLD-inhibitory activity was not affected by the addition of several protease inhibitors, including phenylmeth-ylsulfonyl fluoride, leupeptin, and aprotinin (data not shown).
The Inhibitor Has Little Effect on PLD Activity in the Presence of ARF-PIP 2 -dependent PLD activity was activated up to 10-fold by the addition of 1 M ARF and 10 M GTP␥S (Fig. 7A). To address the possible role of ARF in the regulation of PLD by the inhibitor, we examined whether maximal activation of PLD activity by ARF could overcome the inhibition of PLD activity caused by the inhibitor. When ARF was absent, 1.6 g of the inhibitory fraction decreased PLD activity to about 50% of PLD activity in the absence of the inhibitor. However, in the presence of ARF, the PLD activity was minimally affected by the inhibitor (Fig. 7B).
The PLD Inhibitor Does Not Decrease PLD Activity of Neutrophils-It has been reported that cytosol from HL-60 cells (as opposed to cytosol from bovine brain) was effective in stimulating PLD activity (25,26). PLD activity of neutrophil plasma membranes was stimulated by the addition of neutrophil cytosol and GTP␥S (28). We examined the effect of the PLDinhibitory fraction on PLD activity in the assay system using neutrophil plasma membrane. As shown in Fig. 8, neutrophil PLD activity was not affected by adding different amounts of the PLD inhibitor. DISCUSSION The present study has shown that an inhibitor from bovine brain cytosol potently inhibits PIP 2 -dependent PLD activity. The inhibitor specifically regulated PLD activity and did not bind to substrate vesicles containing PIP 2 . In addition, the suppression of PLD activity by the inhibitory protein was re- versed by the ARF.
The inhibitory factor was eluted with a apparent molecular mass of 30 kDa from a Superose 12 column. If we consider the molecular mass of the inhibitor to be 30 kDa, 7 nM (0.03 g) of the inhibitory fraction is sufficient to decrease PLD activity to about 50% (Fig. 3). Since it was not purified to homogeneity, the inhibitor seems to decrease PLD activity at a concentration considerably less than 7 nM, indicating that the inhibitor is very potent. In contrast, the inhibitor had no effect on PLCmediated PIP 2 hydrolysis.
To examine the specificity of inhibition, the effects of PIP 2binding protein versus the inhibitor on PLD activity were tested. Because PIP 2 is essential for PLD activity as a cofactor, masking of PIP 2 by PIP 2 -binding proteins should cause the inhibition of PLD activity. The PH domain of PLC-␦1 decreased PLD activity at a concentration 100-fold higher than inhibitor (Fig. 5B). We also found that gelsolin, which has PIP 2 -binding properties, inhibited PLD activity at a concentration above 1 M (data not shown). Compared to the PIP 2 -binding proteins, the inhibitor exhibits a far greater potency. Moreover, the inhibitor did not bind to the substrate vesicles containing PIP 2 . These results suggest that the inhibitor does not mask the PIP 2 , but suppresses PLD activity by specific binding to PLD. In addition to PIP 2 binding, an alternative mechanism for the PLD-inhibitory effects is that the inhibitor may be a PIP 2hydrolyzing PLC or other lipase. This is unlikely because we could not detect the hydrolysis of PE, PC, and PIP 2 by the inhibitor.
It has been reported that several GTP-binding proteins and PIP 2 are involved in the activation of PLD (9 -21). In this report, we showed that the PIP 2 -dependent PLD activity was potently decreased by the inhibitor and that the suppression of PLD activity was overcome by ARF. Recent reports demonstrate that ARF-mediated PLD activation is synergistically modulated by RhoA, PKC, calmodulin, or unidentified proteins (15-19, 31, 38). In concert with ARF, these activating factors may play a role in reversing the suppression of PLD activity by the inhibitor. However, nothing is known about the cellular role of the PLD inhibitor and how the PLD-activating factors relate to the negative regulation by the inhibitor. Because the substrate of the PLD enzyme, PC, is abundant in cell membranes and PIP 2 stimulates the PLD activity, we speculate that this inhibitor suppresses the PIP 2 -dependent PLD activity in unstimulated cells. If PLD enzyme were activated by the PLDstimulating factors, the inhibitory constraint might be eliminated.
Recently, it was reported that an inhibitor from bovine brain cytosol eluted with a high molecular mass of 300 kDa from a Superose 12 gel filtration column (26). These data were obtained using streptolysin-O-permeabilized HL-60 cells to measure PLD-inhibitory activity. However, we could not detect an inhibitory fraction with such a high molecular mass when exogenous substrate vesicles were used for measurement of PLD activity. Thus, different forms of PLD inhibitor proteins, which can be detected under different assay conditions, may exist in bovine brain cytosol. In this report, we showed that the low molecular weight PLD inhibitor did not decrease PLD activity of neutrophil membranes. The clear indication for the different PLD isozymes between neutrophil and brain has not been reported yet. However, previous studies have shown that neutrophil PLD might be activated by distinct regulation mode, such as the existence of 50-kDa protein as a major activating factor, and little activation of PLD by ARF (14,31). Thus, it is tempting to speculate that PLD inhibitors may have a specificity toward different PLD isoforms.
In summary, our observations have elucidated specific inhibition of PLD activity by an inhibitory protein with a low molecular mass, and have demonstrated the relationship between the inhibitor and ARF-dependent activation on the regulation of PLD activity. To elucidate the cellular role and the biochemical mechanism of the inhibitor, it will be important to know the molecular identity.