G-protein-coupled receptor activation induces the membrane translocation and activation of phosphatidylinositol-4-phosphate 5-kinase I alpha by a Rac- and Rho-dependent pathway.

Phosphatidylinositol 4,5-bisphosphate (PI4,5P(2)) mediates cell motility and changes in cell shape in response to extracellular stimuli. In platelets, it is synthesized from PI4P by PIP5K in response to stimulation of a G-protein-coupled receptor by an agonist, such as the thrombin. In the present study, we have addressed the pathway that induces PIP5K I alpha activation following the addition of thrombin. Under resting condition expressed PIP5K I alpha was predominantly localized in a perinuclear distribution. After stimulation of the thrombin receptor, PAR1, or overexpression of a constitutively active variant of G alpha(q), PIP5K I alpha translocated to the plasma membrane. Movement of PIP5K I alpha to the cell membrane was dependent on both GTP-bound Rac and Rho, but not Arf, because: 1) inactive GDP-bound variants of either Rac or Rho blocked the translocation induced by constitutively active G alpha(q), 2) constitutively GTP-bound active variants of Rac or Rho induced PIP5K I alpha translocation in the absence of other stimuli, and 3) constitutively active variants of Arf1 or Arf6 failed to induce membrane translocation of PIP5K I alpha. In addition, a dominant negative variant of Rho blocked the PIP5K I alpha membrane translocation induced by constitutively active Rac, whereas dominant negative variants of either Rac or Arf6 failed to block PIP5K I alpha membrane translocation induced by constitutively active Rho. This implies that the effect on PIP5K I alpha by Rac is indirect, and requires the activation of Rho. In contrast to the findings with PIP5K I alpha, the related lipid kinase PIP4K failed to undergo translocation after stimulation by small GTP-binding proteins Rac or Rho. We also tested whether membrane localization of PIP5K I alpha correlated with an increase in its lipid kinase activity and found that co-expressing of PIP5K I alpha with either constitutively active G alpha(q), Rac, or Rho led to a 5- to 7-fold increase in PIP5K I alpha activity. Thus, these findings suggest that stimulation of a G-protein-coupled receptor (PAR1) leads to the sequential activation of G alpha(q), Rac, Rho, and PIP5K I alpha. Once activated and translocated to the cell membrane, PIP5K I alpha becomes available to phosphorylate PI4P to generate PI4,5P(2) on the plasma membrane.

Although PIP5K I and PIP4K have homologous lipid kinases domains, they differ in substrate specificity and in vivo regulation (23). For example, only PIP5K I requires phosphatidic acid as a cofactor for maximal activity (24,25). In vitro, PIP5K I can associate, in a nucleotide-independent fashion, with the small GTP-binding proteins Rho and Rac (26,27). These GTPases can lead to the activation of PIP5K I, but this effect on lipid kinase activity may be limited to the GTP-bound form of the nucleotide-binding protein (28). There is evidence that the effect of Rac or Rho on PIP5K I is indirect and may be mediated by either Rho-kinase or the GTPase Arf (29 -31). Consistent with these interactions in vitro, is the observation that the addition of a non-hydrolyzable GTP analog, GTP␥S, stimulates PI4,5P 2 production in human placenta membrane and in rat brain (32,33). In HeLa cells, Honda et al. (29) demonstrated that the small G-protein ADP-ribosylation factor, Arf, directly activates PIP5K I␣ in the presence of phosphatidic acid, whereas Rho and Rac had no effect. In contrast, Tolias et al. (34) showed that the addition of recombinant Rac to permeabilized platelets led to PIP5K I␣ activation and PI4,5P 2 production. Therefore, it appears that small GTP-binding proteins affect PIP5K I activity and PI4,5P 2 production. However, it is unclear which GTPase is predominantly responsible.
Both PIP5K I and PIP4K appear to localize to distinct subcellular compartments. PIP5K I␣ has been reported to specifically localize at the plasma membranes or in the nucleus, whereas PIP4K has been reported to be present diffusely within the cell or to be present in the ER, actin cytoskeleton, cytosol, plasma membrane, or nucleus (35)(36)(37)(38). We asked whether localization of PIP5K I␣ played a role in its regulation, because its only known substrates are membrane-bound phospholipids. In the present study, we demonstrate that stimulation of the PAR1 receptor induces a translocation of PIP5K I␣ from the Golgi to the cell membrane. This process involves heterotrimeric G-proteins as well as both Rac and Rho, but not Arf. Coincident with the intracellular translocation of PIP5K I␣, we found that its enzymatic activity increased 5-to 7-fold when co-expressed with either large or small GTP-binding proteins. Therefore, these results delineate a signaling pathway that is initiated by a G-protein-coupled receptor and leads to the production of PI4,5P 2 .

MATERIALS AND METHODS
Reagents and Antibodies-The anti-myc, anti-HA, or the biotinylated monoclonal anti-FLAG (M2) antibodies were purchased from Covance, the anti-GM 130 was purchased from Transduction Laboratories, and the anti-PAR1 monoclonal antibody was a gift from L. F. Brass (University of Pennsylvania, Philadelphia, PA (39)). The fluorescein isothiocyanate-and rhodamine-conjugated goat secondary antibodies were obtained from BIOSOURCE. Alexa 350, the cascade blue-labeled goat anti-rabbit secondary antibody, and Alexa 546, the red dye-conjugated streptavidin were purchased from Molecular Probes. The PI4P was obtained from Roche Molecular Biochemicals, and all remaining chemicals were obtained from Sigma Chemical Co.
Plasmid Construction-The human PIP5K I␣ cDNA was generated by PCR using EST AA054379 (Genome Systems) as template and the following oligonucleotide primers: 5Ј-CG CCA GGA TCC GCC ACC  ATG TCT TCT GCT GCT GAA-3Ј and 5Ј-GG CGC TCT AGA TTA CAA  GTC TTC TTC AGA AAT CAA CTT TTG TTC TAA ATA GAC GTC AAG CAC-3Ј. The primers incorporated BamHI and XbaI sites, as well as a carboxyl-terminal myc epitope (EQKLISEEDL). This product was cloned into BglII-and XbaI-digested pCMV5 (a gift of Mark Stinski, University of Iowa).
Cloning of PIP4K (also known as PIP5K Type II or Type C) was performed by PCR using EST W00481 from Genome Systems. Fulllength sequencing revealed that two nucleotide changes that differed from the published sequence, and altered the predicted translated product from 101 CGK 103 to 101 LRE 103 and V 109 to D. These substitutions likely represent polymorphisms, because multiple independent EST clones (N28765, N33825, and THC154497) contained the identical sequence. The cDNA was amplified using PCR primers CG GCA G GTA CCG GCC ATG GCG ACC CCC GGC and GG CGC TCT AGA TTA CAA GTC TTC TTC AGA AAT CAA CTT TTG TTC CGT CAA GAT GTG GCC AAT, which incorporated an myc-epitope onto the carboxyl terminus and KpnI and XbaI restriction sites to facilitate cloning into pCMV5.
Cell Culture, Transient Transfections-Human Embryonic Kidney HEK 293T and Cos-7 cells grown in Dulbecco's modified Eagle medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin (Life Technologies, Inc.) were transiently transfected by the calcium phosphate technique using a total of 4 g of plasmid per 30-mm tissue culture dish. Care was always taken to include empty vector to some transfections to allow identical quantities of plasmid to be used in every transfection. Twenty-four hours after transfection, the medium was removed and replaced with a fresh medium, and analyzed after an additional 24 h. All cells were grown at 37°C in the presence of 5% of CO 2 .
For the triple staining, cells were first stained with monoclonal anti-HA followed by fluorescein isothiocyanate-labeled goat anti-mouse secondary antibody. The cells were then incubated with polyclonal anti-myc and biotinylated monoclonal anti-FLAG IgG1 (M2) antibodies, followed by staining with the cascade blue-labeled goat anti-rabbit secondary antibody along with Alexa 546-conjugated streptavidin.
Localization experiments were performed a minimum of three times, and ϳ30 -60 cells per experiment were examined by two or more viewers. Confocal micrographs of at least six to eight representative cells of each experimental condition were performed by the University of Pennsylvania Cancer Center Confocal Microscopy Core Facility. Confocal images were acquired from a TCS 4D Upright microscope and processed on an IBM OS9 workstation, using Scanware software.
Co-immunoprecipitation Experiments-After transient transfection, 10-cm tissue culture plastic dishes of confluent HEK 293T cells were washed three times with cold PBS and lysed with 1 ml buffer A (1% Triton X-100; 0.14 M NaCl; 1 mM MgCl 2 ; 1 mM EGTA; 20 mM HEPES, pH 7.5, containing 1 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 2 g/ml aprotinin, 0.2 mM Na 3 VO 4 , and 50 mM NaF). The lysates were clarified by centrifugation at 13,000 ϫ g for 30 min at 4°C. Immunoprecipitation was performed using 9E10, a monoclonal antibody against the myc epitope. The immune complexes are collected on protein G-Sepharose, washed twice with buffer A, three times with the kinase buffer B (40 mM HEPES, 20 mM EGTA, 0.2 mM EDTA, 0.1 mM NaCl, 10 mM MgCl 2 , pH 7.5, containing 1 mM dithiothreitol) and subjected to immunoblotting and kinase lipid assay.
Assay of PIP5K I Activity-A standard assay for phosphorylation of PI4P was carried out in an incubation medium of 80 l of buffer B containing 125 M ATP, 125 M PI4P, and 2.5 Ci of [␥-32 P]ATP. The enzyme reactions were incubated at 37°C for 20 min. After stopping the reaction with 1 M HCl (2 volumes) and extracting the lipids with an equal volume of chloroform-methanol (1:1, v/v), the 32 P-labeled PI4,5P 2 products were resolved by thin-layer chromatography using waterchlorophorm-methanol-NH 4 OH (25:70:100:15, v/v) as a solvent system. Immediately before chromatography, the thin-layer plates were precoated with 1% potassium oxalate in water-methanol (1:1) and baked at 110°C for 35 min. Unlabeled PI4P and PI4,5P 2 standards were run in parallel to samples to monitor lipid migration and were visualized by exposure to iodine vapor. The regions of the TLC plates that contained PI4,5P 2 spots were carefully excised, placed in a screw-capped scintillation vial, and then subjected to scintillation counting. Counts were normalized for immunoprecipitated myc-PIP5K I␣ as determined by quantitative 125 I-immunoblotting.

Stimulation of a G-protein-coupled Receptor Leads to the
Membrane Recruitment of PIP5K I␣-Previous reports demonstrated that stimulation of the predominant thrombin receptor on human platelet, PAR1, leads to an increase in PI4,5P 2 synthesis (41,42). Using Cos-7 cells transfected with tagged PIP5K, we investigated whether stimulation of this G-proteincoupled receptor was also associated with an intracellular redistribution of PIP5K I␣. Under resting conditions, expressed PIP5K I␣ was located adjacent to the cell nucleus (Fig. 1). Because of the eccentric perinuclear distribution, we simultaneously co-stained cells with antibodies against the Golgi marker, adaptin, as well as the epitope tag on PIP5K I␣. We found that resting PIP5K I␣ localized in proximity with ␥-adaptin (Fig. 1). Confocal microscopy verified this localization and showed that PIP5K I␣ colocalized most closely with a marker of the cis-Golgi cisternae, GM-130 ( Fig. 2A). Thus, under resting conditions, PIP5K I␣ is co-localized within the early trans-Golgi network. It should be noted that the Golgi has previously been described to contain PIP5K I activity (43,44) and that this intracellular localization is also similar to the related lipid kinase PI4K␤ (45,43).
We next tested whether stimulation of the PAR1 thrombin receptor would alter the location of PIP5K I␣. As shown in Fig.  3, stimulation of PAR1 with the receptor-specific activating peptide SFLLRN resulted in the translocation of PIP5K I␣ to the cellular membrane. After 1 h of stimulation with the peptide agonist, 30 -40% of cells redistributed PIP5K I␣ toward their plasma membrane. Additional studies using transfected HEK-293 cells yielded the same result (not shown). Therefore, these data indicate that stimulation of the thrombin receptor initiates a signaling cascade that can induce the movement of PIP5K I␣ toward the cell membrane.
PAR1 is typically coupled to phospholipase C by G␣ q . To determine whether PAR1 mediated its affect on the intracellular distribution of PIP5K I␣ via G␣ q , we tested whether the relocalization of PIP5K I␣ could be mimicked by co-expression of PIP5K I␣ with an active variant of the ␣ subunit. The Q209L mutation in G␣ q (HA-G␣ q Q209L) is constitutively in the GTPbound state. In 100% of cells expressing G␣ q Q209L and PIP5K I␣, PIP5K I␣ was easily identified as being associated with the cell membrane (Fig. 4). Confocal microscopy of cells expressing active G␣ q and PIP5K I␣, verified PIP5K I␣ was present on the cell membrane (Fig. 2B). Although these confocal images revealed that the majority of PIP5K I␣ was now on the cell membrane, in some cells a fraction of PIP5K I␣ was still localized with the Golgi. These results indicate that the subcellular distribution of the PIP5K I␣ can be regulated by the thrombin receptor and at least one of the G-proteins which normally associate with it.
We also found that stimulation of the EGF receptor by 1-h stimulation of 25 ng/ml EGF initiated cell membrane association of PIP5K I␣ (data not shown). This demonstrates that the relocalization of PIP5K I␣ is not limited to stimulation of a G-protein-coupled receptor. In addition, we have found that expressed PIP5K I␣ in Jurkat T-cells has a perinuclear distribution when the cells are plated on poly-L-lysine. However, PIP5K I␣ translocates to the cellular membrane when the cells are plated on either fibronectin or C305 (an IgM antibody that binds and activates the T-cell antigen receptor). This implies that the signaling pathway leading to membrane association of PIP5K I␣ ultimately involves effectors common to multiple receptor families, including G-protein-coupled, growth factor, immunoglobulin supergene, and integrin receptors.
The Role of Low Molecular Weight GTP-binding Proteins-Although there is agreement that low molecular weight GTPbinding proteins contribute to PIP5K I␣ activation, the identity of the relevant proteins is controversial. Because the low molecular weight GTP-binding proteins Rac and Rho become activated after stimulation of G-protein-coupled, growth factor, immunoglobulin supergene, and integrin receptors (42, 46 -49), we tested whether the membrane localization of PIP5K I␣ could be influenced by these proteins. As shown in Fig. 5, when PIP5K I␣ was co-expressed along with constitutively GTPbound variants of Rac (Rac L61) or Rho (Rho L63), PIP5K I␣ was recruited from the Golgi to the cellular membrane in 100% of cells. This effect by Rac or Rho was regulated by the nucleotide-bound state, because overexpression of constitutively GDP-bound variants of either Rac (Rac V12N17) or Rho (Rho FIG. 1. Intracellular localization of unstimulated PIP5K I␣. Cos-7 cells were transiently transfected with myc epitope-tagged PIP5K I␣. The cells were fixed and stained with both polyclonal anti-myc and monoclonal anti-␥ adaptin as a marker of Golgi. Shown is anti-myc epitope staining, anti-␥ adaptin staining, and phase microscopy. This demonstrates a predominantly perinuclear distribution of PIP5K I␣ under resting conditions that coincide with the Golgi .   FIG. 2. Confocal microscopy PIP5K I␣ and G␣ q . A, Cos-7 cells were transiently transfected with myc epitope-tagged PIP5K I␣ and stained with polyclonal anti-myc (green) and monoclonal anti-GM130 (red). Shown is an overlay confocal micrograph demonstrating that unstimulated PIP5K I␣ localizes with GM130. B, Cos-7 cells were transiently transfected with myc epitope-tagged PIP5K I␣ along with constitutively active HA-tagged G␣ q . The cells were fixed and stained with both polyclonal anti-myc and monoclonal anti-HA. Shown is an overlay confocal image of anti-myc (red) and anti-HA (green) staining. This demonstrates that active G␣ q stimulates PIP5K I␣ to localize on the cell membrane.  N19) failed to induce the PIP5K I␣ translocation (Fig. 5). Arf is another low molecular weight GTP-binding protein that has been shown to interact in vitro with PIP5K I␣. In contrast to the effect of Rac and Rho on PIP5K I␣ intracellular localization, we found that neither constitutively GTP-bound variant of Arf1 (Fig. 5) or Arf6 (not shown) induced PIP5K I␣ translocation. Therefore, this demonstrates that Rac and Rho, but not Arf, are capable of inducing membrane translocation of PIP5K I␣. This also shows that PIP5K translocation is regulated by which guanine nucleotide is bound to the low molecular weight GTPbinding protein.
Because PIP4K has also been reported to become stimulated after thrombin stimulation of platelets (38), we next tested whether small GTP-binding proteins also affected the intracellular distribution of this lipid kinase. In contrast to our findings with PIP5K I␣, neither the GTP-nor GDP-bound variants of Rac or Rho affected the localization of PIP4K (Fig. 6). Therefore, these observations suggest that, although Rac and Rho can contribute to the membrane localization of PIP5K I␣, they have no effect on the intracellular distribution of PIP4K.
Are Rac or Rho Critical for G␣ q -mediated PIP5K I␣ Recruitment to the Membrane?-Because activation of both high, and low, molecular weight GTP-binding proteins lead to the translocation of PIP5K I␣, we examined whether they are components of the same signaling pathway. To begin to address this question, we tested whether dominant negative variants of Rac or Rho affected the ability of G␣ q to induce membrane association of PIP5K I␣. As shown in Fig. 7, GDP-bound variants of either Rac or Rho functioned as competitive inhibitors (dominant negatives) and blocked the ability of G␣ q to induce cell membrane association of PIP5K I␣. Triple staining verified the simultaneous expression of PIP5K, G␣ q , and the dominant negative GTPase in all analyzed cells. Pharmacologic inhibition of Rho by C3-exotoxin also inhibited G␣ q -initiated translocation of PIP5K I␣, confirming the necessity of Rho (not shown). These data imply that stimulation of the PAR1 receptor leads to the cellular membrane binding of PIP5K I␣ through a pathway that involves both Rac and Rho.
Because the dominant negative variants of either Rac or Rho block G␣ q -mediated PIP5K I␣ translocation, both Rac and Rho participate in a single signaling pathway. Nobles and Hall (50) have previously demonstrated that activation of Rac leads to the activation of Rho. Thus, it is possible that the effect of Rac on PIP5K I␣ localization is indirect and requires activation of Rho. To test this hypothesis, we expressed constitutively active Rac L61 alone, or along with dominant negative Rho N19, and analyzed the effect on the intracellular distribution of PIP5K I␣. Coexpression of Rac L61, Rho N19, and PIP5K I␣ in the Cos-7 cells was confirmed by triple antibody staining (Fig. 8).
As we had seen previously, constitutively active Rac induced membrane localization of PIP5K I␣; however, this effect was blocked by co-expression of a dominant negative Rho. This suggests that the effect on PIP5K I␣ by Rac is mediated by activation of Rho.
We performed several different experiments to verify the specificity of our observations. First, we found that pharmacologic inhibition of Rho by C3-exotoxin also inhibited Rac-stimulated membrane translocation of PIP5K I␣ (not shown). Second, we tested whether dominant negative Rac would block PIP5K I␣ membrane translocation initiated by constitutively active Rho. As shown in Fig. 8, expression of dominant negative Rac did not influence Rho-stimulated PIP5K I␣ membrane FIG. 5. GTP-bound RAC or Rho can induce the membrane localization of PIP5K I␣. Cos-7 cells were transfected with myc-PIP5K I␣ alone or along with constitutively inactive, or active, variants of Rac or Rho. Shown is PIP5K I␣ transfected alone, along with constitutively inactive HA-Rac V12N17, constitutively active HA-Rac L61, constitutively inactive HA-Rho N19, constitutively active HA-Rho L63, or constitutively active HA-Arf1 QL. Cells were stained with anti-myc (panels on left) to detect PIP5K I␣, and with anti-HA (panels on right) to detect various small GTPases. Together, this figure demonstrates that either GTP-bound Rac, or Rho, can induce the membrane relocalization of PIP5K I␣. In contrast, GDP-bound Rac or Rho, and GTP-bound Arf, have no effect. The white bar corresponds to 30 m. association. As a third test of specificity, we have found that dominant negative Arf6 N27 also did not influence PIP5K I␣ translocation induced by either Rac or Rho. Together these three lines of evidence demonstrate that Rac and Rho are in a single signaling pathway that leads to the membrane localization of PIP5K I␣ and that in these cells PIP5K I␣ translocation is mediated most directly by Rho.
Effect of Large and Small GTP-binding Proteins on the Lipid Kinase Activity of PIP5K I␣-To determine whether membrane localization of PIP5K I␣ is associated with an increase in lipid kinase activity, we performed in vitro kinase assays using immunoprecipitated PIP5K I␣. As shown in Fig. 9, the ability of PIP5K I␣ to generate PI4,5P 2 from PI4P increased 5-to 7-fold when it was co-expressed with activated large, or small, GTP-binding proteins. In contrast, constitutively inactive variants of Rac and Rho had an inhibitory effect on PIP5K I␣ lipid kinase activity. GDP-bound Rac inhibited PIP5K I␣ activity by 40% (p Ͻ 0.05) and GDP-bound Rho inhibited PIP5K I␣ activity by 60% (p Ͻ 0.04). These experiments imply that the same signaling pathway that induces the relocalization of PIP5K I␣ also induces an increase in its lipid kinase activity.

DISCUSSION
The goal of this work was to investigate the signaling pathway initiated by stimulation of a G-protein-coupled receptor leading to the activation of PIP5K I␣ and production of PI4,5P 2 . Our in vivo findings demonstrate a pathway that sequentially involves the activation of G␣ q , Rac, and Rho. This signaling pathway simultaneously results in the membrane recruitment and enzymatic activation of PIP5K I␣. We hypothesize that both relocalization and biochemical activation of PIP5K I are critical for its ability to regulate cell motility and shape change in response to extracellular stimuli.
Our observations extend previous data on the subcellular localization of PIP5K I␣. Using a variety of transfection or viral infection systems, other investigators have reported PIP5K I␣ in the cell membrane, cytoplasm, and nucleus. To our knowledge, our data are the first to visualize an agonist-mediated translocation of this lipid kinase from one defined cellular location to another. Our data is consistent with previous cell fractionation studies of platelets that revealed that PIP5K enzymatic activity migrates after thrombin stimulation from a Triton X-100-soluble to -insoluble fraction (38).
Schwartz and colleagues (28) were the first to demonstrate that PIP5K I␣ is regulated by small GTP-binding proteins. Their in vitro observations were that PIP5K I␣ catalytic activity in cell lysates increased in the presence of GTP-bound Rho but not in the presence of either GDP-bound Rho or GTP-bound Rac. Consistent with this observation, Oude Weernink et al. (30) found that Rho-kinase may be the direct mediator of this effect. Although there is agreement in the literature that small GTPase regulates PI4,5P 2 synthesis, the specific GTPase required and mechanism of this regulation is controversial. For example, Carpenter and co-workers (27) have found that Rac activated PIP5K I␣, whereas Rho has little effect. Alternatively, Honda et al. (29) reported that GTP-bound Arf6, but not Rac or Rho, is required for activation of PIP5K I␣ after stimulation of a growth factor receptor. Although these authors demonstrate that purified Arf1 and Arf6 activated PIP5K I␣ in vitro, we did not find any influence of constitutively active Arf1, or dominant negative Arf6, on the intracellular localization of PIP5K I␣ under the conditions of our experiments. Our results studying the G-protein-coupled PAR1 thrombin receptor are most consistent with a combined Rac and Rho effect. Furthermore, our observations, that dominant negative, or pharmacologic, inhibition of Rho blocks the membrane recruitment of PIP5K I␣ by Rac, indicate that the Rac effect is indirect and ultimately requires the activation of Rho in these cells. Discrepancies between our observations and those of other investigators might be attributed to differences in experimental design including: 1) in vitro versus in vivo studies (the latter allows the influence of endogenous proteins such as Rho, which could become activated by Rac), 2) different cell types utilized for expression experiments, and 3) potential disparate effects between different isozymes of small GTPases.
Under basal conditions, we have found that the majority of PIP5K I␣ is localized to the Golgi. Our findings are in contrast to those published by Shibasaki et al. (35) who found that adenovirus-expressed PIP5K I␣ was bound to the cell membrane, and its expression was associated with extensive actin changes even in the absence of stimulation. This discrepancy in results might at least partially be due to the adenovirus. Adenovirus infection alone has been previously shown to induce actin organization and cell spreading (51)(52)(53)(54). Our studies indicate that PIP5K I␣ only supports cell spreading and actin changes after the kinase has been stimulated and translocated to the cell membrane. 2 Our findings are consistent with two previous publications demonstrating that PIP5K I is associated with the Golgi (43,44). This indicates that it co-localizes with PI4K, the lipid kinase that generates the predominant PIP5K substrate, PI4P (45,55). After stimulation of the PAR1 receptor, we have found that PIP5K I␣ relocates to the plasma membrane. Our results are consistent with previously established observation that PI4,5P 2 is most abundant in the plasma membrane (56,57). It is conceivable that several small GTP-binding proteins contribute to the localization of these lipid kinases and that some are required for trafficking to the Golgi and others for transport to the cell membrane to coordinate polyphosphoinositide synthesis.
How GTP-bound Rho recruits PIP5K I␣ to the cell membrane is unclear. Several investigators have demonstrated that Rho family members shuttle on and off the membrane; this association is regulated by their GTP/GDP-bound state (58 -60). It is conceivable that Rho is constitutively bound to PIP5K I␣ and that GTP binding merely serves to help translocate the complex to the cell membrane where it can phosphorylate PI4P to generate PI4,5P 2 on the plasma membrane. We have found that the in vitro activity of immunoprecipitated PIP5K I␣ is increased when it has been co-expressed with GTP-bound Rho. This demonstrates that Rho-mediated membrane recruitment PIP5K I␣ alone is not sufficient to also explain the resultant increased production of PI4,5P 2 . Therefore, this implies that the membrane recruitment and the increase in lipid kinase activity, although both are mediated by GTPases, have a separate mechanism of action. Consistent with this hypothesis, we have found that catalytically inactive PIP5K I␣ is still capable of being recruited to the cell membrane. 2 In conclusion, our results support the hypothesis that stimulation of a G-protein-coupled receptor leads to the sequential activation of heterotrimeric G-proteins, Rac and then Rho. Once Rho becomes activated, it induces the activation and membrane recruitment of PIP5K I␣ to generate PI4,5P 2 . In contrast, heterotrimeric G-proteins, Rac and Rho, appear to have no effect on PIP4K activation or localization. The mechanism of its regulation, as well as the identification of other signaling components of the PIP5K I␣ activation pathway, are areas of active investigation. FIG. 9. Effect of GTP-binding proteins on PIP5K I␣ activity. HEK-293T cells were transiently transfected with myc-PIP5K I␣ alone, or along with either a constitutively active GTPase (HA-G␣ q QL, HA-Rac L61, or HA-Rho L63) or a dominant negative GTPase (HA-Rac V12N17 or HA-Rho N19). The cells were lysed, and immunoprecipitated myc-PIP5K I␣ was split into two. One part was used in an in vitro lipid kinase reaction using PI4P as the exogenous substrate, fractionated by thin-layer chromatography, excised, and quantitated on a scintillation counter. The second part of the immunoprecipitate was fractionated by SDS-polyacrylamide gel electrophoresis, 125 I-immunoblotted with antimyc to quantitate the myc-PIP5K I␣ in each immunoprecipitate. The graph shows the mean Ϯ S.E. of relative PI4,5P 2 production from three to four experiments after normalization for immunoprecipitated PIP5K I␣.