Phospholipids Can Switch the GTPase Substrate Preference of a GTPase-activating Protein*

The major cellular inhibitors of the small GTPases of the Ras superfamily are the GTPase-activating proteins (GAPs), which stimulate the intrinsic GTP hydrolyzing activity of GTPases, thereby inactivating them. The catalytic activity of several GAPs is reportedly inhibited or stimulated by various phospholipids and fatty acids in vitro, indicating a likely physiological role for lipids in regulating small GTPases. We find that the p190 RhoGAP, a potent GAP for the Rho and Rac GTPases, is similarly sensitive to phospholipids. Interestingly, however, several of the tested phospholipids were found to effectively inhibit the RhoGAP activity of p190 but stimulate its RacGAP activity. Thus, phospholipids have the ability to “switch” the GTPase substrate preference of a GAP, thereby providing a novel regulatory mechanism for the small GTPases.

GTPase-activating proteins (GAPs) 1 for the small GTPases of the Ras superfamily are potent stimulators of intrinsic GTP hydrolyzing activity and are the major down-modulators of GTPase function. In vitro studies indicate that individual GAPs can regulate multiple members of the GTPase subfamilies, raising a question as to whether context-dependent regulation of the GAPs provides additional specificity in vivo (1). The in vivo regulation of GAPs is poorly understood, but it appears that protein-protein interactions, phosphorylation, and membrane translocation may all play a role (1)(2)(3)(4).
Phospholipids can also regulate GAP catalytic function in vitro. For example, the RasGAP activity of p120 RasGAP and NF1 is inhibited by various acidic phospholipids and fatty acids (5,6). Catalytic function of the RacGAP, n-chimaerin, is inhibited by some phospholipids and stimulated by others (7). Several GAPs for the Arf GTPases depend on phosphoinositides for GAP activity (8 -10).
Here, we report that phospholipids strongly influence the GAP activity of the p190 RhoGAPs (p190A and p190B), which regulate both Rho and Rac GTPases (11,12). Interestingly, some of the phospholipids are potent inhibitors of p190 RhoGAP activity but are stimulators of its RacGAP activity. This finding indicates that phospholipids have the potential to "switch" the GTPase substrate preference for a GAP, thereby providing a novel regulatory mechanism for determining signaling specificity in vivo.
Preparation of Recombinant Proteins-Hexa-histidine-tagged fulllength p190A and p190B proteins and the GAP domain-containing fragment of p190A (residues 1135-1513) were expressed in Sf9 insect cells and affinity purified on a nickel-Sepharose column. Prenylated baculovirus-produced Rac1 and RhoA (RhoA virus was provided by Dr. Matt Hart) were isolated from the membrane fraction of Sf9 cells and purified as described previously (13). Both GTPases were determined to be essentially pure as assessed by SDS-PAGE and Coomassie staining (not shown). Nonprenylated Rac1 and RhoA and the isolated GAP domain (residues 198 -439) and full-length p50RhoGAP were produced as glutathione S-transferase fusions in Escherichia coli.
Measurement of GTP Hydrolysis of Small GTPases-Nucleotide loading of GTPases (1-2 g) was performed with 10 Ci of [␥-32 P]GTP or [␣-32 P]GTP in 100 l of "low magnesium" buffer (50 mM HEPES, 50 mM NaCl, 0.1 mM dithiothreitol, 0.1 mM ATP, 0.1 mM EGTA, 5 mM EDTA, and 1 mg/ml bovine serum albumin, pH 7.4) for 10 min at 30°C for RhoA or at 20°C for Rac1. After incubation, MgCl 2 was added to a final concentration of 10 mM, and proteins were maintained on ice. GTP hydrolysis was determined in most experiments by the nitrocellulose filter binding assay as described previously (11). Incubation was for 15 min at 30°C for RhoA and for 5 min at 20°C for Rac1 (except where otherwise indicated). The reaction was terminated by filtering through 0.45-m nitrocellulose followed by washing with cold wash buffer. In the presentation of data, the intrinsic GTP hydrolysis rate of RhoA and Rac1 is taken into account. Thus, "100%" corresponds to the amount of protein-bound radioactivity retained on the filter in the presence of the respective small GTPase alone.
Alternatively, where noted, the amount of free [ 32 P]phosphate following GTP hydrolysis was determined after separation from [␥-32 P]GTP by active charcoal (14). In Fig. 2, C and D, the amount of hydrolyzed [ 32 P]phosphate is represented after subtraction of the values obtained in the absence of GAP proteins (i.e. in the presence of RhoA or Rac1 alone).
Preparation of Lipid Vesicles-Lipids were dissolved in chloroform or a chloroform/methanol (95:5) mixture and dried in a nitrogen atmosphere. To prepare liposomes, lipids were rehydrated at room temperature in 100 l of buffer (25 mM HEPES, 50 mM NaCl, pH 7.5) followed by vortexing until the mixture was homogeneously opalescent. Alternatively, the vortexed lipid suspension was sonicated for 20 min at 4°C. Phospholipids were used at 0.1 mg/ml (ϳ200 M).

RESULTS
Phosphatidylserine Inhibits p190 RhoGAP Activity and Promotes RacGAP Activity-We tested the possibility that phospholipids can regulate the catalytic activity of the p190 Rho-GAPs. We examined GAP activity using RhoA and Rac1 as substrates and initially tested the effects of PS. As seen by others, intrinsic GTP hydrolysis rates differ significantly for the RhoA and Rac1 GTPases. In a typical experiment, 50% of RhoA-bound [␥-32 P]GTP is hydrolyzed in 15 min at 30°C, whereas with Rac1, 30% of GTP hydrolysis occurs in 5 min at 20°C (Fig. 1A). Notably, the intrinsic enzymatic activity of neither of the small GTPases is influenced by PS.
The purified p190A and p190B proteins exhibit a comparable catalytic GTPase stimulating activity on prenylated RhoA and Rac1 GTPases. When p190 GAP activity is assayed in the presence of PS, the RhoGAP activity of both p190 proteins is substantially inhibited. In a typical experiment, the amount of GTP-bound Rho remaining after incubation with p190A or p190B increases from 50% to 80 and 90%, respectively, in the presence of PS (Fig. 1, B and C). In striking contrast, when Rac1 is used as substrate, the presence of PS in the GAP assay results in a considerable decrease in GTP-bound Rac, when compared with reactions without PS (Fig. 1, B and C). This effect of PS is seen both with p190A and p190B and indicates that PS stimulates the RacGAP activity of the p190 proteins.
The RhoGAP-inhibiting and RacGAP-promoting effect of PS is also seen with an isolated C-terminal fragment of p190 that contains the catalytic domain (Fig. 2, A and B). The extent of the PS effect depends on the amount of GAP protein in the reaction. At a relatively high GAP concentration (roughly equimolar ratio of the GTPase and the GAP), the influence of PS is negligible. However, at lower GAP concentrations (the GTPase is in 10 -100-fold excess over the GAP) there is an almost complete switch in the substrate specificity. Thus, in the presence of PS, at "catalytic concentrations," p190 becomes an active RacGAP but poor RhoGAP.
In more than 60 independent experiments, this effect of PS on p190 GAP activity was highly significant, and the findings were reproducible using several independent preparations of the GAP and the GTPases. On average, in the presence of PS, the level of GTP-bound RhoA following p190 incubation was increased from 29.2 Ϯ 6.0% to 65.8 Ϯ 6.9%, whereas the level of GTP-bound Rac1 decreased from 62.6 Ϯ 3.75% to 37.1 Ϯ 3.9% (Fig. 2, E and F). Enhancement of RacGAP activity by PS is also seen at 30°C (decreasing the reaction time to 90 s), indicating that the opposing effect of PS on substrate specificity is not due to the temperature difference in the RhoGAP and RacGAP assays.
The filter binding assay of GTPase activity can potentially be FIG. 1. PS inhibits p190 RhoGAP activity and stimulates its RacGAP activity. GTP hydrolysis of prenylated RhoA and Rac1 GTPases was measured in a filter binding assay. A, intrinsic GTPase activity of RhoA and Rac1 is not affected by PS. 100% corresponds to the filter-bound radioactivity measured at 0 min in the ice-cold sample. B and C, PS inhibits the RhoGAP activity but stimulates the RacGAP activity of p190A (B) and p190B (C). In B and C, 100% corresponds to the filter-bound radioactivity measured at the end of the respective incubation period, in the presence of the GTPase alone. Where indicated, full-length p190A or p190B was present at 10 nM and PS was added at 0.1 mg/ml. The results of three independent experiments performed in duplicate are shown. misleading for two reasons: 1) GTPase activity cannot be distinguished from nucleotide release because both result in decreased radioactivity on the filter, and 2) binding of protein to the filter is based on electrostatic interactions and might be modified by lipids. To exclude these possibilities, we performed two control experiments. First, we performed assays using [␣-32 P]GTP instead of [␥-32 P]GTP. Neither p190 alone nor in combination with PS has any effect on the amount of radioactivity detected on the filters, indicating that nucleotide remains associated with the GTPase throughout the reaction (data not shown). Next, we repeated the experiment shown in Fig. 2, A and B, using the charcoal precipitation method for measuring GTP hydrolysis. In this assay, [␥-32 P]GTP is bound to and sedimented by activated charcoal, allowing the detection of hydrolyzed radioactive phosphate in the supernatant. Fig. 2, A-D, illustrates the results of a typical experiment where the two techniques were applied in parallel. The findings confirmed the observed opposing effects of PS on p190 RhoGAP and RacGAP activity.
In previous studies, we noted differences in the interaction of p190A with prenylated or nonprenylated small GTPases (15). All of the experiments described above were conducted with prenylated RhoA and Rac1. No effect of PS on p190 GAP activity is detected when nonprenylated RhoA or Rac1 proteins are used as substrates, indicating that the lipid-sensitive GAP activity of p190 depends on proper C-terminal modification of its substrate GTPases (data not shown).
Several Phospholipids Affect the RhoGAP and RacGAP Activity of p190 -It was previously reported that the p120 Ras-GAP can be sequestered nonspecifically by various lipids (5,6). Therefore, we investigated the specificity of the effect of PS on p190 GAP activity. First, we examined the effect of a low concentration of the negatively charged, ionic detergent, SDS, on p190 activity (Fig. 3A). We empirically determined an SDS concentration (0.01%) that does not detectably interfere with nucleotide binding or intrinsic GTP hydrolysis by RhoA and Rac1. At that concentration, SDS substantially inhibits both the RhoGAP and the RacGAP activity of p190. Thus, the negatively charged detergent similarly affects GAP activity toward the two small GTPases, possibly by sequestering the GAP in lipid micelles. This is distinct from the observed opposing effect of PS on the interaction of p190 with RhoA and Rac1 and reflects a more specific effect of the phospholipid. Moreover, the fact that PS stimulates the RacGAP activity of p190 is not consistent with a simple sequestration mechanism.
Next, we examined the effects of a variety of physiological phospholipids on p190 GAP activity (Fig. 3, B and C). PC and PE exhibit no detectable effect on either the RhoGAP or Rac-GAP activity of p190, indicating that the hydrophobic environment per se is not responsible for the observed effect of PS. PI and PIP 2 affect p190 like PS; they both inhibit RhoGAP activity and enhance RacGAP activity of p190. Interestingly, another negatively charged phospholipid, phosphatidic acid, has a distinct effect; it inhibits RhoGAP activity to a similar degree as PS but does not detectably influence RacGAP activity. These findings indicate that it is not simply the negative charge of the phospholipid that influences the substrate specificity of p190 and that several cellular phospholipids have the potential to alter the substrate preference of p190.
Phospholipids Do Not Regulate the RhoGAP and RacGAP Activity of p50RhoGAP-p50RhoGAP is another member of the RhoGAP family, which, like p190, can utilize both Rho and Rac GTPases as substrates, prefers to interact with prenylated GTPases (15), and binds weakly to phosphoinositides (16). We determined that PS, PA, PC, and PE do not affect either the RhoGAP or the RacGAP activity of full-length p50RhoGAP or the isolated GAP catalytic domain (Table I). These experiments were conducted with prenylated GTPases. However, no effect of any tested phospholipid was observed when nonprenylated Rho or Rac was used as substrate (data not shown). Together, these findings suggest that p190 GAP activity exhibits a specific sensitivity to several physiological phospholipids that have the potential to switch its preference for particular small GTPase substrates. DISCUSSION We find that several physiological phospholipids can regulate the catalytic activity of the p190 GAPs. The observation that several of the tested lipids inhibit the p190 RhoGAP activity but stimulate its RacGAP activity is provocative, suggesting that phospholipids can alter the GTPase substrate specificity of a GAP. Although phospholipids have previously been shown to inhibit or stimulate catalytic activity for some GAPs, this is the first demonstration that a phospholipid can exert an opposing regulatory effect on a GAP depending on the provided GTPase substrate. Interestingly, phosphorylation of a Rac/ Cdc42 GAP, called MgcRacGAP, allows it to function as a RhoGAP as well (4). Thus, phosphorylation may be another mechanism by which the substrate preference of GAPs is regulated. A notable difference between these findings is that in our studies, the effect of phospholipids is to switch the GTPase preference of the GAP, whereas MgcRacGAP phosphorylation Phospholipids Regulate p190 GAP Specificity 5057 results in the acquisition of an additional substrate interaction. The fact that the phospholipid effect on p190 depends on GTPase prenylation raises a general consideration for the analysis of GAP specificity. Notably, in every previously reported case of a phospholipid-mediated regulation of GAP activity, the GTPases used for analysis were prepared in bacteria and, therefore, lacked prenylation. Thus, analogous assays with prenylated GTPases could reveal phospholipid effects on substrate preference by other GAPs similar to those described here. Our findings with p50 RhoGAP, however, indicate that not all GAPs are regulated in this manner, even with prenylated GTPases. The absence of detectable phospholipid regulation of p50 RhoGAP also indicates that the observed effects of phospholipids are not simply due to effects on the GTPases that alter their sensitivity to all GAPs but, rather, are specific effects on the interaction between p190 and the Rho and Rac GTPases.
The fact that the C-terminal fragment of p190 RhoGAP is sensitive to phospholipids suggests that a direct interaction of the phospholipids within or near the catalytic region is probably required for regulation. We imagine that a lipid-mediated subtle conformational change is a likely mechanism for altering substrate preference, and future structural studies would be required to formally test this hypothesis. For some GAPs, plextrin homology domains mediate lipid binding (8,17); however, p190 proteins do not have plextrin homology domains. Notably, the difference in the effect of PA versus PI, PS, and PIP 2 on p190 GAP activity is reminiscent of findings with the ArfGAPs, where two separate phospholipid regulatory sites were identified (16). The requirement for a prenylated C terminus of the GTPases in phospholipid sensitivity is consistent with our previous finding that prenylation plays a role in the GAP-GTPase interaction (15), suggesting that prenylation can be an important determinant of GTPase-specific recognition by GAPs.
The physiological significance of phospholipid-mediated GAP regulation has been difficult to establish and remains a challenge to the GTPase field. Significant issues include the fact that dozens of cellular GAPs can potentially regulate a particular GTPase, and so it is difficult to study the regulation of a GTPase by any one GAP in "isolation" in vivo. There is also the fundamental problem that the extraction of GAPs from cellular membranes, their frequent site of action, requires the use of detergents that may disrupt the interaction of the GAPs with regulatory lipids.
Although the physiological role of these phospholipid effects is difficult to assess, the ability of PI, PS, and PIP 2 to switch the substrate preference for p190 RhoGAP from Rho to Rac sug-gests some potential scenarios for in vivo regulation. The nonrandom distribution of PI and PS within various cellular membranes raises the possibility that the p190 substrate preference could be influenced by p190 localization to a particular membrane "subdomain." In addition, the signaling role of PIP 2 , as a product of phospholipase C activation, makes it possible for p190 substrate preference to be regulated by various extracellular signals that impinge on phospholipase C.
In summary, we find that the GTPase substrate preference for the p190 GAPs can be regulated by several physiological phospholipids. The fact that these lipids can effectively convert p190 to either a RhoGAP or a RacGAP provides a novel regulatory mechanism by which GTPase-mediated signaling can be regulated in a context-dependent manner.