The alpha-subunit of the heterotrimeric G protein G13 activates a phospholipase D isozyme by a pathway requiring Rho family GTPases.

G13 belongs to the G12 family of heterotrimeric G proteins, whose effectors are poorly defined. The present study was designed to test if phospholipase D (PLD) is regulated by G13 and if Rho-type small GTPases are involved. Expression of the constitutively active Q226L mutant of the alpha-subunit of G13 in COS-7 cells stimulated the activity of a rat brain phospholipase D isozyme (rPLD1) co-expressed in the cells. Wild type Galpha13 was ineffective unless the cells were incubated with AlF4-. rPLD1 was previously shown to be activated by constitutively active V14RhoA in COS-7 cells (Park, S. K., Provost, J. J., Bae, C. D., Ho, W. T., and Exton, J. H. (1997) J. Biol. Chem. 272, 29263-29272). When the endogenous Rho proteins of the cells were inactivated by treatment with C3 exoenzyme from Clostridium botulinum, the ability of Galpha13Q226L to activate rPLD1 was greatly attenuated. Co-transfection of dominant negative N19RhoA and N17Rac-1, but not N17Cdc42Hs or N17Ras, also inhibited the activation. Expression of constitutively active Galphaq in COS-7 cells also activated rPLD1, but constitutively active forms of Galphai2 and Galphas were without effect. These findings support an effector role for PLD in G13 signaling and demonstrate a requirement for Rho GTPases in this response.

Phosphatidylcholine (PC) 1 specific phospholipase D (PLD) hydrolyzes its substrate into phosphatidic acid (PA) and choline in response to growth factors and G protein-coupled receptor agonists (1). PA may act as a lipid messenger in the cell, inducing cytoskeletal rearrangements or growth-regulatory responses, and is implicated in the regulation of NADPH-oxidase and intracellular membrane trafficking and fusion events (1,2). PA can also be converted to the protein kinase C (PKC) activator diacylglycerol by phosphatidic acid phosphohydrolase or to the G protein-coupled receptor agonist lysophosphatidic acid by a PA-specific phospholipase A 2 (1).
Analysis of PLD activity from various tissues and subcellular fractions supports the existence of biochemically distinct isozymes differing in subcellular localization and responses to Ca 2ϩ , phosphatidylinositol 4,5-bisphosphate (PIP 2 ), oleate, and detergents in vitro (1,3). Studies of signal transduction pathways in intact cells support the involvement of PKC and Rho GTPases in the regulation of PLD (1,2). In vitro experiments confirm a direct regulatory role for PKC, Rho-family, and ARF GTPases in PLD activation (1,2). PIP 2 also activates PLD in vitro and may be required for PLD regulation in vivo (1).
Mammalian PLD isoforms have been cloned including hPLD1a, hPLD1b, hPLD2, and rPLD2 (4 -7). hPLD1a, a 1072amino acid protein, is specific for PC and is regulated by PKC, ARF, RhoA, and PIP 2 (4,5). hPLD1b encodes a 1034-amino acid splice variant with properties similar to hPLD1a (5). hPLD2 is also PC-specific, has high basal activity both in vitro and in vivo, and is activated by PIP 2 , but not PKC, ARF, or RhoA (6). rPLD1, cloned in our laboratory from a rat brain library using a fragment of hPLD1, is 91% identical to hPLD1b in amino acid sequence and is expressed in a wide variety of tissues (8). It is stimulated when co-expressed with constitutively active V14RhoA in COS-7 cells (8) and also by treatment of the cells with a PKC-activating phorbol ester (8) or lysophosphatidic acid. 2 Like hPLD1, rPLD1 is activated directly by RhoA, Arf, and PKC␣ in vitro. 2 The Rho-related G proteins Rho, Rac, and Cdc42 are members of the Ras superfamily of low molecular weight GTPases. They regulate multiple downstream effects, including the contractility and organization of the actin cytoskeleton (9 -12). Rho is required for cytoskeletal, transcriptional, and PLD responses to some G protein-coupled receptor agonists (2,(12)(13)(14), although the pathways involved and the precise nature of the Rho-requirement remain to be defined. G 13 , a member of the G 12 family of heterotrimeric G proteins, was first identified by the cDNA cloning of its ␣-subunit (15,16). G 13 is expressed in most cell lines and tissues and is especially abundant in human platelets (17). G 13 is coupled to platelet thrombin and thromboxane A 2 receptors (18,19), suggesting a possible role in platelet activation. Immediate downstream effectors have not been identified for the G 12 family, although downstream effects resulting from overexpression of constitutively active mutant forms of G␣ 12 and G␣ 13 have been described, including neoplastic transformation of cultured fibroblasts (20,21), increased amiloride-sensitive sodium-proton exchange (22)(23)(24), induction of immediate early gene expression (13,25), actin stress fiber formation and focal adhesion assembly (26), and activation of the c-Jun N-terminal kinase (JNK) cascade (27)(28)(29). A number of these effects can be blocked by dominant negative mutant forms of Rho-related G proteins, indicating a requirement at some level for these low molecular weight GTPases (13, 16, 23, 26 -30).
The present study examines if PLD is a downstream effector of G 13 and if Rho family GTPases are involved in the signal transduction pathway.

EXPERIMENTAL PROCEDURES
Plasmid Constructs-A partial G␣ 13 cDNA, reverse transcriptionpolymerase chain reaction amplified from mouse brain total RNA, was used to clone a full-length cDNA from a mouse liver cDNA library (Stratagene) using standard methods. A G␣ 13 Q226L mutant was made by overlap-extension polymerase chain reaction and confirmed by DNA sequencing. Wild type and Q226L were subcloned into the HindIII and EcoRI sites of the pcDNA3 expression vector (Invitrogen). rPLD1 in pcDNA3 (8), N19RhoA in pcDNA3, and G␣ q Q209L in pCMV4 were described previously (31). G␣ s QL and G␣ i2 QL in pcDNA3 were provided by Dr. N. Dhanasekaran (Temple University, Philadelphia, PA), N17Cdc42 in pcDNA3 was provided by Dr. Shubha Bagrodia (Cornell University, Ithaca, NY), and N17Rac-1 in pCGT was provided by Dr. Linda Van Aelst (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY).
Cell Culture and Transfection-COS-7 cells (ATCC) were maintained in Dulbecco's modified Eagles's medium (Life Technologies, Inc.) plus 10% fetal bovine serum (Life Technologies, Inc.) under 5% CO 2 . Six-well plates were seeded with 2 ϫ 10 5 cells/well and transfected with 2 g of plasmid DNA and 6 l of LipofectAMINE (Life Technologies, Inc.) per well according to the manufacturer's instructions. Opti-MEM (Life Technologies, Inc.) was substituted for Dulbecco's modified Eagles's medium during reduced serum incubations.
C3 Scrape Loading-Cells were washed 24 h post-transfection, twice with phosphate-buffered saline (PBS) and once with scraping buffer (114 mM KCl, 15 mM NaCl, 5.5 mM MgCl 2 , 10 mM Tris-Cl pH 7.4) and then scraped off the plate in the absence or presence of 5 g/ml Clostridium botulinum C3 exoenzyme (List Biologicals) as described (14). Cells were replated on polylysine and serum-starved 12 h later.
PLD Assay-Cells were serum-starved (0.5% fetal bovine serum in Opti-MEM) at 24 h post-transfection for 16 h in the presence of 1 Ci/ml [ 3 H]myristate (NEN Life Science Products), washed with PBS, and incubated in serum-free medium (Opti-MEM) for 50 min. PLD activity was then assayed as described previously (14). Cells were incubated in 0.3% 1-butanol for the indicated times. Cells were then washed with ice-cold PBS and stopped with methanol. Lipids were extracted, and the phosphatidylbutanol product was resolved by thin layer chromatography as described (14). Bands co-migrating with a phosphatidylbutanol standard were quantitated by scintillation counting. 13 -To study the effect of G 13 signaling on PLD, we expressed wild type G␣ 13 and GTPase-deficient, constitutively active G␣ 13 Q226L together with a Rho-responsive PLD isoform, rPLD1, by liposome-mediated transient transfection of COS-7 cells. High level expression of the proteins and similar levels of rPLD1 expression between different co-transfections were confirmed by Western analysis (not shown). Expression of wild type G␣ 13 had little or no effect on either endogenous PLD activity or that of rPLD1. G␣ 13 Q226L also had no effect on the activity of the endogenous PLD, but markedly stimulated rPLD1 activity (Fig. 1A). This activation of rPLD1 by G␣ 13 Q226L was consistent with an effector role for PLD in G 13 signaling. However, since G␣ 13 signaling was chronically activated by the prolonged expression of the Q226L mutant in this experiment, it was not clear how direct or indirect the activation of rPLD1 might be.

PLD Activation by G␣
To rule out mechanisms arising from long term activation of the pathway or consequently altered expression of PLD regulators, we examined G␣ 13 signaling in an acutely regulated system. Aluminum fluoride activates heterotrimeric G proteins due to its ability to mimic the ␥-phosphoryl group of GTP when complexed with the GDP-bound ␣-subunit (32). We expressed wild type G␣ 13 with and without rPLD1 and then treated the cells with aluminum fluoride for 30 min (Fig. 1B). Fluoride treatment of the cells yielded a small stimulation of the endogenous PLD activity, which was not affected by overexpression of wild type G␣ 13 . rPLD1 activity was potently stimulated by fluoride treatment when co-expressed with G␣ 13 , but not when expressed alone. This G␣ 13 -dependent acute stimulation of rPLD1 by aluminum fluoride effectively rules out mechanisms requiring long term activation of G␣ 13 and is consistent with an effector role for PLD in G 13 signaling.
Requirement for Rho Family GTPases-To investigate the potential role of Rho in the regulation of rPLD1 by G␣ 13 , we inhibited its function by treating the cells with C3 exoenzyme from C. botulinum, an ADP-ribosyltransferase that inactivates Rho (33). Efficient ADP-ribosylation was confirmed by in vitro assays with C3 (33), which showed that cellular Rho from the C3-scraped cells was efficiently modified as indicated by a subsequent loss of ADP-ribosylation in vitro ( Fig. 2A). We then looked at the effect of C3 toxin on the activation of rPLD1 by G␣ 13 Q226L. C3 had little or no effect on the endogenous PLD activity or that of the expressed rPLD1. It did, however, significantly diminish the effect of G␣ 13 Q226L on rPLD1 activity, consistent with a requirement for Rho in the activation of rPLD1 by active G␣ 13 (Fig. 2B).
To further address the requirement for Rho and related GTPases, we expressed dominant negative mutant forms of these G proteins together with G␣ 13 Q226L and rPLD1 and FIG. 1. Effect of mutational and aluminum fluoride-stimulated activation of G␣ 13 on rPLD1 activity. A, PLD activity was measured in COS-7 cells transfected with 0.6 g/well of pcDNA3 vector containing no insert, G␣ 13 , or G␣ 13 Q226L, with and without 0.6 g of co-transfected rPLD1 cDNA. Vector was added when needed to adjust the total amount of DNA to 2 g/well. Cells were serum-starved and labeled with [ 3 H]myristate as described under "Experimental Procedures" and incubated with 0.3% 1-butanol for 1 h. Radioactivity incorporated into phosphatidylbutanol was quantitated as described under "Experimental Procedures." The results are plotted as the means of at least four independent experiments Ϯ S.E. B, COS-7 cells were transfected as above with the indicated combinations of G␣ 13 and rPLD1. PLD activity was measured in cells preincubated with 0.3% 1-butanol for 10 min followed by 10 M AlCl 3 ϩ 10 mM NaF for 30 min. The results are the means of three independent experiments Ϯ S.E. G␣ 13 Activates Phospholipase D 4824 measured PLD activity. These dominant negative mutants are thought to inhibit signaling through endogenous Rho-related small G proteins by forming stable, inactive complexes with guanine nucleotide exchange factors required for their activation (34). N17Ras, N19RhoA, N17Rac, and N17Cdc42 alone had no effect on rPLD1 activity (not shown). When co-expressed with both G␣ 13 Q226L and rPLD1, N19RhoA and N17Rac partially blocked the activation of rPLD1 while N17Ras and N17Cdc42 did not (Fig. 3). Combination of N19RhoA and N17Rac completely blocked the activation. These results are consistent with a requirement for exchange factor-mediated regulation of Rho family GTPases such as Rho and Rac in the activation of rPLD1 by constitutively active G␣ 13 .
rPLD1 Is Activated by Mutationally Active ␣-Subunits of the G 12 and G q Families-To determine the selectivity of G 12 family proteins in the activation of rPLD1, we examined its regulation by other heterotrimeric G protein ␣-subunits. Accordingly, we expressed mutationally active ␣-subunits of the G 12 , G q , G i , and G s families with and without rPLD1 and then measured PLD activity (Fig. 4). We found that the activity of rPLD1 was potently stimulated by constitutively active G␣ 13 Q226L and G␣ q Q209L, but not by G␣ i2 Q205L or G␣ s Q227L. DISCUSSION This report supports the hypothesis that rPLD1, a widely expressed rat hPLD1b homolog, serves as a downstream effector in G 13 signaling. Both mutational activation and fluoride stimulation of expressed G␣ 13 led to potent stimulation of co-expressed rPLD1 activity (Fig. 1). 3 The acute regulation by fluoride-stimulated G␣ 13 rules out mechanisms requiring long term activation of the pathway and is consistent with an effector role for rPLD1. 4 Mutational and fluoride-stimulated activation of G␣ 13 had no effect on endogenous PLD (Fig. 1). This is probably due to differences between rPLD1 and the endogenous PLD in COS FIG. 2. Effect of loading COS-7 cells with C3 exoenzyme of C. botulinum on rPLD1 activation by G␣ 13 Q226L. A, transfected COS-7 cells were scrape-loaded with bovine serum albumin or C3 toxin and replated as described under "Experimental Procedures." Twentyeight h later, cellular extracts were prepared and treated in vitro with C3 toxin in the presence of [ 32 P]NAD as described (33) to measure the extent of ADP-ribosylation of Rho proteins. The extracts were then resolved by SDS-polyacrylamide gel electrophoresis and subjected to autoradiography. B, COS-7 cells transfected with the indicated plasmid constructs were scrape-loaded with bovine serum albumin or C3 toxin and replated as described under "Experimental Procedures." Cells were starved and labeled as described and then incubated with 0.3% 1-butanol for 1 h. Radioactivity incorporated into phosphatidylbutanol was quantitated as described under "Experimental Procedures." The results are plotted as the means of three independent experiments Ϯ S.E .   FIG. 3. Effect of dominant negative RhoA, Rac, Cdc42, and Ras mutants on rPLD1 activation by G␣ 13 Q226L. PLD activity was measured in COS-7 cells transfected with 0.5 g/well each of plasmid constructs encoding G␣ 13 Q226L and rPLD1 and 1 g/well of the indicated dominant negative mutant GTPases (0.5 g ϩ 0.5 g when N19RhoA and N17Rac were combined). Vector was added to adjust the total amount of DNA per well to 2 g. Cells were starved and labeled as described under "Experimental Procedures" and incubated with 0.3% butanol for 1 h. Radioactivity incorporated into phosphatidylbutanol was quantitated as described, and results were expressed as a percentage of the maximal response. The results plotted are the means of at least three independent experiments Ϯ S.E.

FIG. 4. Effect of expression of mutationally active G protein
␣-subunits of the G 12 , G q , G i , and G s families on rPLD1 activity. PLD activity was measured in COS-7 cells transfected with 0.6 g/well of each of the indicated cDNAs encoding mutationally active G protein ␣-subunits both with and without rPLD1 co-transfection (0.6 g/well). Vector was added to adjust the total amount of DNA to 2 g/well. Cells were serum-starved and labeled as described under "Experimental Procedures" and incubated with 0.3% 1-butanol for 1 h. Radioactivity incorporated into phosphatidylbutanol was quantitated. The results are plotted as the means of four independent experiments Ϯ S.E. cells. Previous work in this laboratory has indicated the absence of Rho-reponsive PLD activity in COS cells (8). This has been observed both in cells transfected with V14RhoA and also with in vitro assays performed by incubation of COS cell extracts with RhoA and GTPS (8). In both cases, the Rho response could be reconstituted by expression of rPLD1 (8).
We found that C3 toxin, which is selective for Rho (33), significantly blocked rPLD1 activation by G␣ 13 (Fig. 2). Dominant negative N19RhoA and N17Rac-1 alone partially but significantly blocked rPLD1 activation and, in combination, completely blocked the effect (Fig. 3). These data support a requirement for guanine nucleotide exchange factor-dependent regulation of Rho family G proteins such as Rho and Rac in this response (34). Cytochalasin D treatment of the cells did not block the response (not shown), suggesting that the inhibitory effects of C3 toxin, N19RhoA, and N17Rac are not simply due to a general disruption of signaling responses as a consequence of altered cytoskeletal organization or function induced by these treatments. All together, these data are consistent with the involvement of Rho family G proteins in the signaling pathway between G␣ 13 and rPLD1.
We showed that rPLD1 was also activated by mutationally active G␣ q , but not by G␣ i2 or G␣ s . This observation is consistent with reports of PLD activation by G q -coupled receptors and PKC-activating phorbol esters (1). rPLD1 is stimulated by the ␣and ␤-isozymes of PKC in vitro 2 and by phorbol ester when expressed in COS-7 cells (8). Our results indicate that rPLD1 is selectively activated by ␣-subunits of the G 12 and G q families, although ␤␥-subunit mediated regulation of rPLD1 or activation of other PLD isoforms by heterotrimeric G proteins other than G q and G 12 family members remain a possibility. G␣ 13 Q226L expression triggers actin stress fiber formation and focal adhesion assembly (26), JNK activation (27)(28)(29), sodium-proton exchange (23,24), serum response element-dependent gene transcription (13), and apoptosis (30) by pathways also inhibited by dominant negative Rho family G proteins. For several reasons, rPLD1 activation is likely to occur either upstream of these effects or by an entirely separate pathway. First, Rho family G proteins are direct regulators of rPLD1 3 (5), while the G␣ 13 -mediated effects listed above are likely to be downstream of a cascade of Rho-initiated signals. Second, cytochalasin D disrupts cytoskeletal responses, but did not inhibit rPLD1 activation (not shown). Apoptotic responses and gene induction typically require hours to develop, while rPLD1 is activated within minutes by fluoride-stimulated G␣ 13 (Fig. 1). Finally, activation of the JNK cascade by G␣ 13 in COS-7 cells is blocked by N17Ras, but not N19RhoA (27)(28)(29), while rPLD1 activation is blocked by N19RhoA, but not N17Ras (Fig. 3).
Phospholipases are key components of transmembrane signal transduction pathways. PLD-catalyzed production of phos-phatidylcholine-derived messengers could mediate critical downstream responses in G 13 signaling. Further study is needed to more precisely define the role of PLD in G 13 -mediated responses. We suggest that G 13 may play a role in the regulation of Rho-responsive PLD activity by G protein-coupled receptors.