Regulation of P-Rex1 by Phosphatidylinositol (3,4,5)-Trisphosphate and Gβγ Subunits*

P-Rex1 is a guanine-nucleotide exchange factor (GEF) for the small GTPase Rac. We have investigated here the mechanisms of stimulation of P-Rex1 Rac-GEF activity by the lipid second messenger phosphatidylinositol (3,4,5)-trisphosphate (PtdIns(3,4,5)P3) and the Gβγ subunits of heterotrimeric G proteins. We show that a P-Rex1 mutant lacking the PH domain (ΔPH) cannot be stimulated by PtdIns(3,4,5)P3, which implies that the PH domain confers PtdIns(3,4,5)P3 regulation of P-Rex1 Rac-GEF activity. Consistent with this, we found that PtdIns(3,4,5)P3 binds to the PH domain of P-Rex1 and that the DH/PH domain tandem is sufficient for PtdIns(3,4,5)P3-stimulated P-Rex1 activity. The Rac-GEF activities of the ΔPH mutant and the DH/PH domain tandem can both be stimulated by Gβγ subunits, which infers that Gβγ subunits regulate P-Rex1 activity by binding to the catalytic DH domain. Deletion of the DEP, PDZ, or inositol polyphosphate 4-phosphatase homology domains has no major consequences on the abilities of either PtdIns(3,4,5)P3 or Gβγ subunits to stimulate P-Rex1 Rac-GEF activity. However, the presence of any of these domains impacts on the levels of basal and/or stimulated P-Rex1 Rac-GEF activity, suggesting that there are important functional interactions between the DH/PH domain tandem and the DEP, PDZ, and inositol polyphosphate 4-phosphatase homology domains of P-Rex1.

active. The substrate of the P-Rex family is the Rho family GTPase Rac (1)(2)(3), an enzyme that regulates a wide range of cell functions, including cell shape, movement, secretion, phagocytosis, transcription, translation, and the production of reactive oxygen species. Depletion of P-Rex1 from neutrophil-related cell lines results in low reactive oxygen species formation via G protein-coupled receptors (1).
The basic element for Rac-GEF activity is the catalytic Dbl homology (DH) domain. Regulating accessibility of the substrate GTPase to the DH domain is an important mechanism governing GEF activation. In Vav1, the main regulatory phosphorylation site Y174 lies just NH 2 -terminal of the DH domain, and its phosphorylation relieves an autoinhibition between the peptide stretch around Y174 and the DH domain, thus activating the GEF (12). Typically, the DH domain is in tandem with a pleckstrin homology (PH) domain, a phosphoinositide binding domain, the role of which is still not completely understood and seems to vary between GEFs. Crystal structures of Sos1 and Tiam1 DH/PH tandems have shown that the DH and PH domains interact with one another. In Sos1, this interaction precludes access of Rac to its binding site (13), whereas in Tiam1 it does not (14). For Vav, PtdIns(3,4,5)P 3 binding to the PH domain facilitates Vav1 tyrosine phosphorylation, thus enhancing its Rac-GEF activity (15). Similarly, for Sos1, PtdIns(3,4,5)P 3 binding to the PH domain seems to relieve the intramolecular inhibition between DH and PH, enabling Rac binding and, presumably, GEF activation (16). In contrast, the activity of Tiam1 is not affected by binding of another phosphoinositide, PtdIns 3-phosphate, to the PH domain in its DH/PH tandem (17).
As described above, P-Rex family enzymes are directly activated by PtdIns(3,4,5)P 3 and G␤␥ subunits. Precedent would suggest that the binding site for PtdIns(3,4,5)P 3 lies within its PH domain, but the binding site for the G␤␥ subunits is less predictable. One other Rac-GEF known to interact directly with G␤␥ subunits is p114 Rho-GEF, an enzyme that activates RhoA and Rac1 in vivo (18). p114 Rho-GEF binds G␤␥ subunits both via its DH/PH tandem and its COOH terminus, and overexpression of the p114 Rho-GEF DH/PH tandem together with G␤␥ subunits in NIH3T3 cells results in downstream activation of the serum response element (18). However, it is not known whether G␤␥ subunits regulate the catalytic activity of p114 Rho-GEF. The Rac-GEF activity of one other GEF, Ras-GRF1, is regulated by G␤␥ subunits, but here the effect is indirect, via activation of protein-tyrosine kinases (19). Similarly, the CDC42-GEF activity of PIX␣ is stimulated by G␤␥ subunits indirectly, via their binding to PAK1 (20). Other families of enzymes whose activities are directly regulated by G␤␥ subunits do exist, including phosphoinositide 3-kinase ␥, phospholipase C␤, and adenylyl cyclase II, and the interactions between these enzymes and G␤␥ subunits have been studied in some detail (21)(22)(23). However, no consensus G␤␥ binding motif can be inferred from this work that would predict how G␤␥ subunits act on the P-Rex family. P-Rex family enzymes, like most GEFs, are multidomain proteins. Apart from the catalytic DH domain and tandem PH domain, they also contain two DEP domains (Disheveled, EGL-10, and pleckstrin homology) and two PDZ domains (postsynaptic density disc-large zo-1 homology). These are protein interaction domains that are frequently found in signaling proteins. It is emerging that DEP domains might generally serve to target their proteins to membranes, possibly by interactions with proteins of the SNARE superfamily (24), whereas PDZ domains are known to serve a large range of different adaptor-type functions, influencing the activity and/or targeting of their interacting partners (25). The roles of the DEP and PDZ domains in the P-Rex proteins are unknown. Finally, P-Rex1 and P-Rex2 also have a weak homology over their COOH-terminal half to inositol polyphosphate 4-phosphatase (1, 2), whereas P-Rex2B, a splice variant of P-Rex2, lacks this COOH-terminal half (3). Inositol polyphosphate 4-phosphatase is an enzyme that catalyzes the 4Ј-dephosphorylation of inositol (3,4)-bisphosphate, inositol (1,3,4)-trisphosphate, and PtdIns(3,4)-bisphosphate (26). It regulates the survival of several types of neurons in the cerebellum and hippocampus (27). Although the residues required for phosphatase activity are present in the inositol polyphosphate 4-phosphatase homology (IP4P) domain of P-Rex1 and P-Rex2, we have so far not been able to demonstrate any phosphatase activity. Here, we have studied the mechanisms of regulation of P-Rex1 Rac-GEF activity by PtdIns(3,4,5)P 3 and by G␤␥ subunits through the generation and characterization of a panel of P-Rex1 deletion, truncation, and point mutants.
Generation of Mutant P-Rex1 Proteins-We have constructed a panel of human P-Rex1 proteins consisting of WT P-Rex1 and eight mutant proteins (Fig. 1). The design of the E56A/N238A point mutations ("GEFdead" construct) was based on alignments of the DH domain with those of other Rac-GEFs. Glu 56 in P-Rex1 is equivalent to Glu 1047 in Tiam1 and Asn 238 in P-Rex1 to Asn 1232 in Tiam1 (14). Both residues in Tiam1 are involved in building the interface with Rac, and their mutation to alanine disrupts Tiam Rac-GEF activity (14). The E56A/N238A P-Rex1 construct was cloned using mutagenic primers. The design of the other mutants was based on domain boundary predictions made with SMART and PFAM programs. P-Rex1 lacking the PH domain (⌬PH), P-Rex1 lacking both DEP domains (⌬DEP), and P-Rex1 lacking both PDZ domains (⌬PDZ) were constructed with a short linker (STPSTS) to join the sequence preceding and following the deleted domains. The isolated PH domain (iPH) and the isolated IP4P domain (iIP4P) were cloned with a 5Ј-flanking in-frame EcoRI site and methionine start codon, and with a 3Ј-flanking stop codon and XbaI restriction site. The isolated DH/PH domain tandem (iDHPH) and P-Rex1 lacking the IP4P domain (⌬IP4P) are COOH-terminal deletions that were cloned with a 3Јflanking stop codon and XbaI restriction site.
Mutations were made by standard PCR protocols, and constructs were cloned into pBluescript and sequenced. Constructs were subcloned into pAcoG1 (which gives them an NH 2 -terminal EE-epitope tag) for production of recombinant proteins in Sf9 cells. WT and mutant P-Rex1 proteins were expressed in Sf9 cells by baculovirus infection, purified from Sf9 cells using their EE tag, and stored in phosphate-buffered saline, 1 mM EGTA, 1 mM dithiothreitol, 0.01% Na azide, 50% glycerol, and 2 mg/ml fatty acid-free bovine serum albumin.
Rac-GEF Activity Assay-Sf9 cell-derived purified human recombinant WT and mutant P-Rex1 proteins were assayed for Rac-GEF activity in vitro essentially as described previously (1). Briefly, 100 nM GDP-loaded purified recombinant Sf9 cell-derived EE-tagged Rac1 was incubated with 50 nM P-Rex1 protein, [ 35 S]GTP␥S, and liposomes (comprised of 200 M each of phosphatidylcholine, -serine, and -inositol) that either did or did not contain varying concentrations of synthetic D/Dstearoyl-arachidonyl-PtdIns(3,4,5)P 3 or purified recombinant Sf9 cellderived G␤ 1 ␥ 2 subunits. Rac was then immunoprecipitated using its EE-tag and its [ 35 S]GTP␥S loading was measured. To control for the fact that cholate (from the G␤␥ subunit storage buffer) inhibits P-Rex1 Rac-GEF activity, mock stimulations were performed for each G␤␥ concentration with the equivalent cholate buffer control.
PtdIns (3,4,5)P 3 Binding Assay-PtdIns(3,4,5)P 3 binding to Sf9 cellderived purified human recombinant WT and mutant P-Rex1 proteins was measured using an assay derived from a protocol we have previously used to identify many PtdIns(3,4,5)P 3 -binding proteins (30). For this, P-Rex1 proteins were pre-cleared with Affi-Gel beads in assay buffer (1ϫ phosphate-buffered saline, 1 mM MgCl 2 , 0.1% Triton X-100, 0.5 mg ml Ϫ1 fatty acid-free bovine serum albumin) and then incubated at 60 nM final concentration with or without 20 M D/D-or L/L-dipalmitoyl-PtdIns(3,4,5)P 3 in assay buffer on ice for 15 min. The samples were then added to D/D-dipalmitoyl-PtdIns(3,4,5)P 3 -coupled Affi-Gel beads or control Affi-Gel beads and incubated for 40 min at 4°C with end-overend rotation. Beads were washed four times in assay buffer before addition of SDS sample buffer. Proteins bound to the beads were detected by SDS-PAGE followed by anti-EE Western blotting.

P-Rex1
Mutagenesis-To study the activation of P-Rex1 by PtdIns(3,4,5)P 3 and G␤␥ subunits, we constructed a panel of deletion, truncation, and point mutants of human P-Rex1. The panel consisted of WT P-Rex1; P-Rex1 with two point mutations, E56A and N238A, in the catalytic DH domain (GEF-dead mutant); P-Rex1 lacking the PH domain (⌬PH mutant); P-Rex1 lacking both DEP domains (⌬DEP mutant); P-Rex1 lacking both PDZ domains (⌬PDZ mutant); P-Rex1 lacking the IP4P domain (⌬IP4P mutant); the isolated IP4P domain (iIP4P mutant); the isolated DH/PH domain tandem (iDHPH mutant); and the isolated PH domain (iPH mutant) (Fig. 1). The WT and mutant P-Rex1 proteins were produced as recombinant NH 2terminal EE epitope-tagged proteins in Sf9 cells and purified using their EE tag. All P-Rex1 proteins were expressed at their expected sizes, WT P-Rex1 and the GEF-dead mutant at 185 kDa, the ⌬PH mutant at 170 kDa, the ⌬DEP mutant at 166 kDa, the ⌬PDZ mutant at 168 kDa, the ⌬IP4P and iIP4P mutants at 93 kDa, the iDHPH mutant at 45 kDa, and the iPH mutant at 15 kDa (Fig. 2).
Regulation of P-Rex1 Basal Rac-GEF Activity-First, we assayed the basal in vitro Rac-GEF activity of WT P-Rex1 and of those mutant P-Rex1 proteins that contain the DH domain, by measuring GTP␥S loading of Rac (Fig. 3). In the absence of any stimulus, WT P-Rex1 catalyzed the GTP␥S loading of only a limited amount of Rac (2% of the total Rac in the assay in 10 min), as expected. The P-Rex1 protein with the E56A/N238A double mutation in the DH domain was catalytically inactive, confirming formally that the DH domain of P-Rex1 is indeed the catalytic site and that it behaves like a DH domain typical for GEFs of Rho family GTPases. The ⌬PH mutant had 10 times elevated basal activity compared with WT P-Rex1, which shows that, in the absence of any stimulus, the PH domain helps to keep the WT enzyme in an inactive state. However, removal of all domains COOH-terminal to the PH domain, leaving just the DH/PH domain tandem (iDHPH), created a highly constitutively active protein. Titration of the amount of iDHPH protein in the assay suggested that its basal activity was at least 60 times that of WT P-Rex1. Hence, the PH domain is not sufficient to negatively regulate the basal activity of P-Rex1, i.e. additional regulation clearly comes from other parts of the WT enzyme. Indeed, ⌬DEP and ⌬PDZ mutants also had 5 and 15 times elevated basal activities, respectively, showing that the DEP and PDZ domains participate in keeping the WT enzyme in an inactive basal state. In contrast, removal of the COOH-terminal half of P-Rex1, which is largely occupied by the IP4P domain (⌬IP4P mutant) did not significantly change the basal Rac-GEF activity of P-Rex1 (but see below).
PtdIns (3,4,5)P 3 -dependent Regulation of P-Rex1 Rac-GEF Activity-Next, we measured the direct PtdIns(3,4,5)P 3dependent stimulation of the in vitro Rac-GEF activities of WT P-Rex1 and of those mutant P-Rex1 proteins that contain the DH domain, using as stimulus D/D-stearoyl/arachidonyl-PtdIns(3,4,5)P 3 , the naturally occurring form of PtdIns(3,4,5)P 3 (Fig. 4). PtdIns(3,4,5)P 3 had no effect on the GTP␥S loading of Rac in the absence of any GEF protein. In contrast, the Rac-GEF activity of WT P-Rex1 was directly stimulated more than 30-fold by PtdIns(3,4,5)P 3 in a dose-dependent manner, with maximal activation at concentrations above 10 M and an EC 50 of about 1.5 M. This is similar to the effects we described in our original report on P-Rex1 (1). As expected, the E56A/N238A double mutant could not be activated by PtdIns(3,4,5)P 3 , confirming again that it is GEF activity dead. Importantly, the elevated basal activity of the ⌬PH protein could not be further stimulated by PtdIns(3,4,5)P 3 , whereas that of the iDHPH mutant could (when the latter was titrated down 12-fold to be within the range of our assay). These results show that the PH domain is necessary and the DH/PH tandem sufficient for PtdIns(3,4,5)P 3 -dependent regulation of P-Rex1 Rac-GEF activity.   2. Purification of recombinant WT P-Rex1 and mutant P-Rex1 proteins. Human WT P-Rex1 and P-Rex1 mutants were cloned, expressed with NH 2 -terminal EE-epitope tags in Sf9 cells, and purified using their EE tag as detailed under "Experimental Procedures." To assess purity and size of the P-Rex1 proteins, 10 l of each preparation were analyzed by SDS-PAGE followed by Coomassie staining, except for the ⌬IP4P mutant, of which 20 l were loaded. The band at 66 kDa is bovine serum albumin from the P-Rex1 protein storage buffer. The iPH mutant (see arrow) was loaded onto this gel prior to the addition of bovine serum albumin.

P-Rex1 Regulation
The Rac-GEF activities of all other P-Rex1 mutants, including those with elevated basal activities, could also be stimulated by PtdIns(3,4,5)P 3 . The EC 50 values of their stimulation by PtdIns(3,4,5)P 3 were similar to that of WT P-Rex1, between 0.3 and 1.2 M (with these slight differences unlikely to be significant), and their maximal activation by PtdIns(3,4,5)P 3 was also similar to WT P-Rex1, all catalyzing GTP␥S loading of around 60% of the total Rac in the assay within 10 min. Only the ⌬IP4P mutant was significantly less active (GTP␥S loading 30% of Rac within 10 min), suggesting that the presence of the IP4P domain in WT P-Rex1 might somehow promote full-scale PtdIns(3,4,5)P 3 -dependent stimulation. However, another more mundane explanation for its reduced activity could be that the IP4P mutant has simply come out of its purification procedure with slightly lower specific activity. In any event, the IP4P domain plays at best a modulatory role in PtdIns(3,4,5)P 3dependent stimulation of P-Rex1. In summary, apart from the PH domain, none of the other P-Rex1 domains contribute fundamentally to the stimulation of P-Rex1 Rac-GEF activity by PtdIns(3,4,5)P 3 .
We made one additional interesting observation in this series of experiments: PtdIns(3,4,5)P 3 was inhibitory for the Rac-GEF activities of both mutants lacking the IP4P domain, the iDH/PH and ⌬IP4P proteins, at concentrations above 3 M. The suspected lower specific activity of the ⌬IP4P protein (see above) cannot account for the striking difference in the PtdIns(3,4,5)P 3 dose-response curve of this mutant compared with WT. This result suggests that the presence of the IP4P domain might serve to prevent inactivation of the WT P-Rex1 enzyme at high PtdIns(3,4,5)P 3 concentrations.
G␤␥ Subunit-dependent Regulation of P-Rex1 Rac-GEF Activity-Next, we assayed the G␤␥ subunit-dependent stimulation of the in vitro Rac-GEF activities of WT P-Rex1 and those mutant P-Rex1 proteins that contain the DH domain, using purified recombinant Sf9 cell-derived G␤ 1 ␥ 2 subunits as stimulus (Fig. 5). In this series of experiments, GEF activities were inhibited by the cholate in the G␤␥ storage buffer, so we performed cholate buffer controls throughout, and we normalized results to the basal activity of WT P-Rex1 to allow direct comparison of all mutants. G␤␥ subunits activated Rac to a limited extent in the absence of any GEF (about 7% of the total Rac in the assay in 10 min). In contrast, the Rac-GEF activity of WT P-Rex1 was strongly and directly stimulated by G␤␥ subunits. The activation was 30-fold at 0.3 M G␤␥ subunits, similar to that described in our original report on P-Rex1 (1). We did not attempt to determine EC 50 values for G␤␥ subunitdependent Rac-GEF activity, as even the highest G␤␥ subunit concentrations we could achieve in our assay were non-saturating. As expected, the E56A/N238A GEF-dead mutant could not be activated by G␤␥ subunits. Surprisingly, the Rac-GEF activities of all other mutants could. Notably, both the iDHPH mutant and the ⌬PH mutant, having only the DH domain in common, were activated by G␤␥ subunits to a similar degree as WT P-Rex1. This surprising result infers that the DH domain mediates G␤␥ subunit-dependent regulation of P-Rex1 Rac-GEF activity.
The lower degree of activation observed with ⌬DEP and ⌬PDZ mutants (about half and one-third of WT P-Rex1 activity, respectively, at 0.3 M G␤␥ subunits) might indicate that these domains have a moderate modulatory effect on G␤␥ subunitdependent activation of WT P-Rex1 Rac-GEF activity. In contrast, the seemingly much lower activation of Rac by the ⌬IP4P mutant is certainly an underestimate, created in part by its relatively low basal activity in this series of experiments compared with those presented in Figs. 3 and 4 (an effect of the cholate) and possibly by its suspected lower specific activity (see above). We therefore do not consider it likely that the IP4P domain has a major influence on G␤␥ subunit-dependent activation of P-Rex1, although we cannot exclude a modulatory role. In summary, we have learned that the DH domain is sufficient for G␤␥ subunit-dependent activation of WT P-Rex1 Rac-GEF activity, and that no other P-Rex1 domains play a fundamental role in this mode of regulation.
Binding of PtdIns (3,4,5)P 3 to P-Rex1-To investigate in more detail the interaction between PtdIns(3,4,5)P 3 and P-Rex1, we measured next direct binding of PtdIns(3,4,5)P 3 to WT P-Rex1 and all mutant P-Rex1 proteins using a PtdIns(3,4,5)P 3 bead binding assay (Fig. 6). This assay is based on a protocol we have previously employed to identify many PtdIns(3,4,5)P 3 -binding proteins (30). Protein binding to PtdIns(3,4,5)P 3 beads was compared with binding to control beads and considered specific when it could be competed-off by the naturally occurring D/D form of dipalmitoyl-PtdIns(3,4,5)P 3 but not its L/L-dipalmitoyl-PtdIns(3,4,5)P 3 diastereoisomer. It should be noted that this assay does not define absolute binding affinities but shows the relative abilities of proteins to interact with immobilized PtdIns(3,4,5)P 3 and to be recovered after a series of rapid washes. We found that WT P-Rex1 and all P-Rex1 mutants except the iIP4P protein could bind PtdIns(3,4,5)P 3 in a stereospecific manner. The minimal domain required for PtdIns(3,4,5)P 3 binding was the isolated PH domain. Together with the results from the Rac-GEF activity assays, this result strongly suggests that PtdIns(3,4,5)P 3 binding to the PH domain is sufficient for PtdIns(3,4,5)P 3 -dependent regulation of P-Rex1 Rac-GEF activity.

FIG. 3. Rac-GEF activity of WT P-Rex1 and P-Rex1 mutants. The basal
Rac-GEF activity of each P-Rex1 protein was assessed in vitro using 100 nM purified GDP-loaded Sf9 cell-derived EE-Rac1 as substrate, by measuring its GTP␥S loading as detailed under "Experimental Procedures." The concentration of WT P-Rex1 and all mutant P-Rex1 proteins in the assay was 50 nM, except for the iDHPH domain, which was used at both 50 (1) and 4 nM (2) to estimate the extent of its basal activity. Data are mean Ϯ S.E. from three to five experiments performed in duplicate, except for the GEF dead mutant, which is mean Ϯ range from two experiments and high-dose iDHPH (1), which is from one experiment.
possibility that an additional site for PtdIns(3,4,5)P 3 binding might exist. We speculated that this putative additional site could lie within the IP4P domain, because PtdIns(3,4,5)P 3 binding to the ⌬IP4P mutant was also markedly reduced. However, seen that we suspect the ⌬IP4P protein to have reduced specific activity (see above), a proportion of it may be misfolded, and furthermore, in view of the result that the iIP4P mutant failed to bind PtdIns(3,4,5)P 3 , we consider it unlikely that there is an additional PtdIns(3,4,5)P 3 binding site in the IP4P domain. Hence, on the basis of the panel of mutants we have made, we cannot at present say where, or if at all, this additional binding site exists.  4. PtdIns(3,4,5)P 3 -dependent stimulation of the Rac-GEF activities of WT P-Rex1 and P-Rex1 mutants. The Rac-GEF activities of WT P-Rex1 and the P-Rex1 mutants were assayed as described in the legend to Fig. 3, except with the indicated concentrations of synthetic D/D-stearoyl/arachidonyl-PtdIns(3,4,5)P 3 , as detailed under "Experimental Procedures." The concentration of WT P-Rex1 and all mutant P-Rex1 proteins in the assay was 50 nM, except for the iDHPH tandem, which was 4 nM. Data are mean Ϯ S.E. from three to five experiments performed in duplicate, except the GEF dead mutant, which are mean Ϯ range from two experiments, and "No-GEF" data, which are from one experiment representative of three.

DISCUSSION
In this paper, we have investigated the regulation of P-Rex1 Rac-GEF activity by PtdIns(3,4,5)P 3 and G␤␥ subunits. In essence, we show that PtdIns(3,4,5)P 3 activates P-Rex1 via the PH domain and that G␤␥ subunits activate P-Rex1 via the DH domain.
The most surprising finding of our study is that the DH domain of P-Rex1 mediates G␤␥ subunit-dependent stimulation of P-Rex1 Rac-GEF activity. We can conclude this from our results showing that the Rac-GEF activities of the iDHPH mutant and of the ⌬PH mutant, having only the DH domain in common, can both be stimulated by G␤␥ subunits. This is, to FIG. 5. G␤␥ subunit-dependent stimulation of the Rac-GEF activities of WT P-Rex1 and P-Rex1 mutants. The Rac-GEF activities of WT P-Rex1 and the P-Rex1 mutants were assayed as described in the legend to Fig. 3, except with the indicated concentrations of purified Sf9 cell-derived recombinant G␤ 1 ␥ 2 subunits (filled symbols) or mock stimulated with the appropriate cholate buffer controls (open symbols), as detailed under "Experimental Procedures." The concentration of WT P-Rex1 and all mutant P-Rex1 proteins in the assay was 50 nM. Data are mean Ϯ S.E. from three to five experiments performed in duplicate, except the GEF dead mutant, which are mean Ϯ range from two experiments and "No-GEF" data, which are from one experiment representative of three. our knowledge, the first time that a Dbl family member has been shown to undergo positive regulation of GEF activity through its DH domain. We had expected the G␤␥ subunits to act either via the DEP, PDZ, or PH domains. The DEP and PDZ domains were likely candidates as they are well known protein interaction domains (see references in Introduction), and the PH domain was another likely candidate as the PH domains of some other signaling proteins, including GRK2 and phospholipase C␤2, have previously been shown to bind G␤␥ subunits (31,32). We have attempted to measure direct binding of G␤␥ subunits to our P-Rex1 mutants, by immobilizing P-Rex1 proteins and measuring their ability to interact with free G␤␥ subunits. However, these assays were difficult because of unspecific binding of G␤␥ subunits to the beads, and the results were inconclusive.
The region within the G␤␥ subunits that interacts with their various effectors, e.g. GRK2, phospholipase C␤, or G␣ subunits, is conserved (22,23,31). Hence, it is likely that P-Rex1 will compete with these other signaling enzymes for G␤␥ subunit binding. The structural components of the P-Rex1 DH domain required for interaction with G␤␥ subunits are unknown. We expect, in analogy with what is known for other G␤␥ effectors (31), that many residues within the DH domain are going to be involved in building the interface, by folding into a concave groove that can accommodate the G␤␥ subunits. Which residues these are cannot be easily inferred from the primary sequence. Presumably, binding of G␤␥ subunits to one side of the DH domain causes a conformational change within the domain that enables access of GDP-Rac to its binding pocket. Structural work will be necessary to elucidate the mode of action of G␤␥ subunit binding to and activation of the P-Rex1 DH domain.
One intriguing possibility emerging from our work is that more Dbl family GEFs might be regulated by G␤␥ subunits than presently appreciated, as all of these enzymes possess a DH domain. A screen of several recombinant GEFs or isolated DH domains in a GEF activity assay like ours could address this possibility. p114 Rho-GEF would be a likely candidate to investigate, as it is already known to bind G␤␥ subunits somewhere within its DH/PH tandem (18).
Our work has also given us some insight into the functioning of the PH domain of P-Rex1. First, the PH domain mediates PtdIns(3,4,5)P 3 regulation of P-Rex1 Rac-GEF activity. This we concluded from the facts that the Rac-GEF activity of the ⌬PH mutant cannot be stimulated by PtdIns(3,4,5)P 3 , whereas that of the iDHPH mutant can, and that PtdIns(3,4,5)P 3 binds to the iPH mutant. Second, we found that the PH domain participates in maintaining a low basal activity of P-Rex1 in the absence of any stimulus (our ⌬PH mutant has elevated basal activity). Hence, the role of the P-Rex1 PH domain seems to be similar to that proposed for Vav1 and Sos1, where, in the absence of stimulus, the PH domain interacts with the DH domain, restricting access of Rac to the catalytic site, whereas PtdIns(3,4,5)P 3 binding to the PH domain promotes a conformational change that allows access of Rac and catalysis (16). However, PtdIns(3,4,5)P 3 binding to Vav1 or Sos1 promotes, at best, a 2-fold stimulation of their Rac-GEF activities (15,33), so there clearly is some difference in mechanism between Vav1/ Sos1 and P-Rex1 regulation by PtdIns(3,4,5)P 3 . The only other Rac-GEF significantly activated by PtdIns(3,4,5)P 3 is SWAP-70, where PtdIns(3,4,5)P 3 also binds to the PH domain, but the mechanism of action must be different still, as, uniquely, in SWAP-70 the PH domain is located NH 2 -terminal to the DH domain (34).
We found, furthermore, that the PH domain alone is not sufficient for keeping WT P-Rex1 catalytically inactive in the absence of stimulus (our iDHPH mutant has very high basal activity). The same applies to the DH/PH domain tandems of Trio (35) and several GEFs for other Rho family GTPases (36). Hence, negative regulation by large parts of GEF protein in addition to the PH domain is common, even though the domains that confer this negative regulation must differ between GEF families, as their domain structures vary a great deal (except the DH/PH tandem). In P-Rex1, the domains that participate with the PH domain in its negative regulation are the DEP and PDZ domains (their removal results in increased basal activity). One question we cannot address with our present panel of P-Rex1 mutants is how much the PH domain contributes to basal guanine-nucleotide exchange catalyzed by the DH domain. In several other GEFs, although the DH domain is always sufficient for activity, the DH/PH domain tandem shows higher basal activity than the isolated DH domain, which has led to the suggestion that the DH/PH domains together act as the true functional catalytic domain (35,36).
As mentioned above, we have further learned that PtdIns(3,4,5)P 3 binds to the isolated PH domain of P-Rex1, which was not unexpected, as PH domains are well known phosphoinositide binding domains. Whereas many PH domains contain a clear consensus site for PtdIns(3,4,5)P 3 binding (36), those in the DH/PH tandem of GEFs usually do not. For exam-FIG. 6. PtdIns(3,4,5)P 3 binding to WT P-Rex1 and P-Rex1 mutants. WT P-Rex1 protein or mutant P-Rex1 proteins were incubated with PtdIns(3,4,5)P 3 beads (PtdIns(3,4,5)P 3 coupled to Affi-Gel beads) or control Affi-Gel beads with or without competing free D/D-or L/Ldipalmitoyl-PtdIns(3,4,5)P 3 . The beads were then washed and protein bound to the beads quantified by anti-EE Western blotting, as detailed under "Experimental Procedures." The total control represents 2.5% of the P-Rex1 mutant protein in each assay. Data are representative Western blots from at least two experiments for each mutant. ple, the critical lysine and arginine residues considered to mediate PtdIns(3,4,5)P 3 binding to the PH domains of Grp1, BTK, or PDK1 (37) are not conserved in P-Rex1. Hence, we did not attempt to pinpoint the exact residues for PtdIns(3,4,5)P 3 binding in the P-Rex1 PH domain, as it was difficult to identify residues whose mutation would eliminate binding. One surprising result was that PtdIns(3,4,5)P 3 also binds to our ⌬PH mutant, suggesting that an additional binding site for PtdIns(3,4,5)P 3 might exist. Two types of domains in P-Rex1, apart from the PH domain, have been shown to interact with phosphoinositides in other proteins, the PDZ and IP4P domains (26,38). The PDZ domain of syntenin binds PtdIns(4,5)P 2 , mediating its membrane binding, and inositol polyphosphate 4-phosphatase uses PtdIns(3,4)P 2 as one of its substrates (26). However, our ⌬PDZ mutant bound PtdIns(3,4,5)P 3 seemingly as well as WT P-Rex1, and our iIP4P mutant did not bind PtdIns(3,4,5)P 3 at all, so we cannot at present speculate where this putative second binding site might be.
Our work has also yielded some information on the role of the DEP, PDZ, and IP4P domains of P-Rex1. The DEP and PDZ domains help to maintain the low basal activity of the wild type enzyme, as discussed above, whereas the IP4P domain seems to protect the enzyme from inactivation at high concentrations of PtdIns(3,4,5)P 3 . Neither of these domains are fundamentally required for PtdIns(3,4,5)P 3 -or G␤␥ subunit-dependent stimulation of P-Rex1 Rac-GEF activity. However, deletion of the DEP and PDZ domains (and possibly, although not likely, the IP4P domain) reduces the maximal activation of the enzyme by G␤␥ subunits, whereas activation by PtdIns(3,4,5)P 3 is unaffected, suggesting that the DEP and PDZ domains may somehow participate in full-scale activation of P-Rex1 by G␤␥ subunits.
Finally, activation of P-Rex family enzymes by PtdIns(3,4,5)P 3 or G␤␥ subunits can occur independently by one or the other stimulus, the mechanisms of which we have investigated here, but also in synergism (1,2). Hence, conformational changes and/or membrane translocation induced by one of the stimuli must influence the level of P-Rex activation by the other. Future work should enable us to address the mechanism of this synergistic activation.