Structure and Function of Heterotrimeric G Protein-Regulated Rho Guanine Nucleotide Exchange Factors

Background.

Members of the Gα_12/13 subfamily are potent regulators of the actin cytoskeleton (Buhl et al., 1995) and are important for vascular development and chemotaxis (Offermanns et al., 1997). Given that Rho GTPases were also well established regulators of these responses, the hunt was on for RhoGEFs that could operate directly downstream of Gα_12/13. One candidate was p115RhoGEF, a RhoA-selective GEF containing a region with very weak homology to the catalytic RGS homology (RH) domain characteristic of RGS proteins (Tesmer, 2009). Indeed, this proved to be the binding site for activated Gα₁₂ and Gα₁₃ subunits, although only Gα₁₃ could stimulate GEF activity (Hart et al., 1998; Kozasa et al., 1998; Mao et al., 1998). Soon thereafter, PDZ-RhoGEF (Fukuhara et al., 1999), also known as GTRAP48 (Jackson et al., 2001), and LARG (Fukuhara et al., 2000) were identified as RH-domain containing RhoGEFs. These three proteins form the RH-RhoGEF subfamily, a small subfamily of Gα₁₃-regulated RhoGEFs. LARG is the only one currently known to be activated by Gα₁₂, although this seems to require phosphorylation of the RhoGEF by Tec tyrosine kinase (Suzuki et al., 2003). All three RH-RhoGEFs are widely expressed in mammals (Hart et al., 1996; Fukuhara et al., 1999; Kourlas et al., 2000; Kuner et al., 2002), but p115RhoGEF and PDZ-RhoGEF are most abundantly expressed in hematopoietic cells and the central nervous system, respectively.

In resting cells, RH-RhoGEFs are predominantly localized to the cytosol, with some presence at the cell membrane or bound to actin. Upon activation of Gα_12/13 coupled receptors or coexpression of constitutively active mutants of Gα₁₂ or Gα₁₃, RH-RhoGEFs translocate to the membrane surface (Siehler, 2009; Suzuki et al., 2009a). RH-RhoGEFs also exhibit some receptor selectivity. In HEK293 and PC3 cells, RhoA is activated by PDZ-RhoGEF via LPA receptors and by LARG via thrombin receptors (Wang et al., 2004). The mechanism by which this selectivity occurs is not known, but it is not apparently due to differences in their PDZ domains, because both PDZ-RhoGEF and LARG can bind to the C termini of LPA receptors (Yamada et al., 2005).

The RH RhoGEFs are complex proteins with multiple signaling domains and functional motifs hidden within the low-complexity regions that bracket them. There is considerable evidence that the mechanism of activation is also complex, involving not only recruitment to specific sites at the membrane surface, but also multiple intra- and intermolecular interactions that mediate inhibition of basal activity as well as allosteric activation (Bhattacharyya and Wedegaertner, 2003; Bhattacharyya et al., 2009; Suzuki et al., 2009b; Zheng et al., 2009). The specific details vary among subfamily members (e.g., p115RhoGEF lacks a PDZ domain) and context (e.g., LPA versus thrombin receptors). In the following sections, we summarize what is known about the structure and function of the major domains and other functional motifs of RH-RhoGEFs, highlighting their potential roles in regulating the activity of the DH/PH domains. We end by describing recent solution-based studies of LARG and PDZ-RhoGEF that provide new insights into the molecular mechanism of RH-RhoGEF activation.

The RhoGEF RH Domain.

As the principal binding site for Gα_12/13 subunits, structural analysis of the RhoGEF RH domain was anticipated to explain the molecular basis for selectivity and GAP activity and to shed some light on the mechanism of regulation by heterotrimeric G proteins. The canonical RH domain is a bundle of nine helices that forms a relatively flat domain with two lobes (Tesmer, 2009). In RGS proteins, the α5-α6 loop of the domain interacts with catalytic residues in switch I and II of Gα to accelerate GAP activity (Tesmer et al., 1997a). In contrast, the RH domains of p115RhoGEF and LARG do not have GAP activity. Instead, a short sequence motif positioned immediately N-terminal to the RH domain is required for stimulating GTP hydrolysis in p115RhoGEF and LARG (Chen et al., 2003). The region spanning this N-terminal motif and the RH domain has been termed the “rgRGS” domain (Chen et al., 2001) (Fig. 2a).

Crystal structures of the p115RhoGEF (Chen et al., 2001) and the PDZ-RhoGEF (Longenecker et al., 2001) RH domains confirmed that they indeed retain most of the canonical RH domain fold, albeit with a ∼65-residue extension that contributes three extra α helices to the C terminus (Fig. 1, b and c). However, these initial structures did not reveal an obvious mode of interaction with Gα_12/13 subunits, in part because the Gα-interacting residues of RGS proteins are not conserved in the RhoGEF RH domain (Tesmer, 2009). Although the entire rgRGS protein of p115RhoGEF was crystallized, the N-terminal GAP motif was not ordered.

The structure of the p115RhoGEF rgRGS domain in complex with a chimeric Gα_13/i1 protein (Chen et al., 2005) revealed a surprising bipartite interaction mediated by the N-terminal GAP motif and the RH domain (Fig. 1b). The Gα_13/i1 chimera used in this study was created to circumvent problems in expressing functional Gα₁₃ in Escherichia coli and contains the α helical domain and switch I and II elements from Gα₁₃, with the remainder from Gα_i1. As a consequence, the effector docking site contains residues that are not native to Gα₁₃, including some that directly contact the RH domain. It is noteworthy that this chimera was never shown to activate p115RhoGEF. The N-terminal motif (residues 22-37) begins with an IIG sequence motif that anchors the peptide on the α-helical domain of Gα_13/i1 and ends with an acidic sequence in which Glu27 and Phe31 contact Gα_13/i1-Arg200 in switch I and Gln226 in switch II, respectively, stabilizing them in a more transition state-like conformation. The structure thereby provides a molecular explanation for the mild GAP activity exhibited by p115RhoGEF. The RH domain (residues 44-233), as if supplanted from switch I by the N-terminal GAP motif, binds to the canonical effector docking site of Gα_13/i1 (Fig. 1b). Instead of using the α5-α6 loop to bind Gα, the interaction is mediated by structural elements from the unique C-terminal extension of the RH domain, in particular the α8-α9 loop, which bears hydrophobic residues that pack into the effector docking site. Consequently, the canonical core of the RhoGEF RH domain can be thought of as a scaffold that positions unique functional motifs/extensions at its N and C termini that are required for high-affinity binding, recognition of the activated conformation of the Gα subunit, and, in p115RhoGEF and LARG, GAP activity.

Recent crystal structures of complexes between a different Gα₁₃ chimera (all but the N-terminal helix was native to Gα₁₃) and the PDZ-RhoGEF rgRGS domain recapitulate the major findings of the Gα_13/i1-p115RhoGEF complex (Chen et al., 2008). As before, the N-terminal motif of the RhoGEF anchors to the α-helical domain of Gα₁₃ using an IIG sequence motif and interacts with the switch regions (Fig. 1c). In this case, however, Glu312 and Tyr315 (analogous to Glu27 and Phe32 in p115RhoGEF) do not make productive interactions with residues in the active site of Gα₁₃, explaining the lack of GAP activity as well as the observation that PDZ-RhoGEF has similar affinity for the guanosine 5′-O-(3-thio)triphosphate- and GDP·AlF₄⁻-bound states of Gα₁₃ (Chen et al., 2008). It is noteworthy that residues in the RH domain of p115RhoGEF that bind to the effector docking site of Gα₁₃ are not conserved in PDZ-RhoGEF (Fig. 1, b and c). Despite this difference, a similar complex is formed, with the effector docking site expanding to accommodate the larger “αE” helix that occupies the position analogous to the α8-α9 loop of p115RhoGEF. However, it is not clear whether the differences in the structure of the effector docking site are a consequence of differences in the chimeric Gα₁₃ subunits used, differences in the RH domains, or both. The structure of the PDZ-RhoGEF rgRGS domain in complex with deactivated Gα₁₃·GDP was also determined, in which switch II retains its active conformation despite the presence of GDP. Thus, the duration of signaling via RH-RhoGEFs could be regulated not only by the rate of GTP hydrolysis but also by the rate at which Gα₁₃·GDP is sequestered by Gβγ (Chen et al., 2008).

The PDZ Domain.

PDZ-RhoGEF and LARG can also be activated by receptors other than GPCRs via interactions mediated by the PDZ domains found at their N termini. PDZ domains are common structural domains that typically bind the C termini of proteins found at the cell membrane (Nourry et al., 2003). Compared with other PDZ domains, the RH-RhoGEF PDZ domain binds to surprisingly diverse biological targets (for review, see Smietana et al., 2008). Perhaps the best characterized is plexin B1, a semaphorin 4D receptor that mediates RhoA activation and axonal growth cone collapse (Aurandt et al., 2002; Hirotani et al., 2002; Perrot et al., 2002; Swiercz et al., 2002). Both RhoGEFs also bind to LPA-1 and -2 receptors, which couple primarily with Gα_12/13 subunits. Interaction between the C termini of these receptors and the PDZ domain was reported to be required for RhoA activation (Yamada et al., 2005).

NMR structures of the LARG PDZ domain alone and in complex with a plexin-B1 C-terminal octapeptide have been reported (Liu et al., 2008) (Fig. 1d). A solution structure of the PDZ-RhoGEF PDZ domain is also available (Protein Data Bank entry 2dls). They retain the typical PDZ domain fold, which consists of six antiparallel strands (βA-βF) and two helices (αA and αB). Peptide ligands dock in an extended conformation into a groove located between βB and αB, where they form a β-sheet like interaction with the βB strand. The LARG PDZ domain undergoes an atypical conformational change upon ligand binding relative to other characterized PDZ domains (Fig. 1d), and the apo form exhibits a high degree of conformational flexibility (Liu et al., 2008). These characteristics may facilitate the binding of these PDZ domains to a wide variety of targets.

DH/PH Domain Structures.

The canonical DH domain is composed of 10 or more helices that assemble into an oblong shape reminiscent of a chaise lounge (Worthylake et al., 2000). The last helix (called α6) transverses the longest dimension of the domain and often forms a continuous helix with αN, the first helix of the PH domain (Fig. 1e). The PH domain consists of a flattened seven-stranded antiparallel β-barrel, capped at one end by αC, the hallmark C-terminal helix found in all PH domains. In some PH domains, loops on the opposite end of the barrel from αC are used to bind specific phospholipid head groups. The αN helix is characteristic of PH domains found in DH/PH tandems.

Structures have been determined for the LARG DH/PH domains, both alone and in complex with RhoA (Kristelly et al., 2004) and for the PDZ-RhoGEF DH/PH-RhoA complex (Derewenda et al., 2004). Both RhoA complexes exhibit similar domain organization and intersubunit contacts (Fig. 1f). As in Dbs, another RhoGEF in which the PH domain contributes positively to GEF activity (Rossman et al., 2002, 2003), direct contacts are formed between RhoA and the PH domain. Although this interface accounts for only ∼8% of the total buried surface area in the LARG DH/PH-RhoA complex, mutation of residues in the PH domain-RhoA interface diminished GEF activity to the level of the DH domain alone (Kristelly et al., 2004). Mutation of residues in the PDZ-RhoGEF interface did not lead to significant differences in vitro (Oleksy et al., 2006), and hence the functional importance of the RhoA-PH domain interface seems to vary among RH-RhoGEFs.

Examination of the DH/PH structures reveals several structural elements that could play a role in regulation. The first is a flexible, helical extension at the N terminus of the DH domain (Fig. 1e). Either truncation of the extension or mutation of a tryptophan (LARG-Trp769) buried between the extension and the DH domain reduced GEF activity (Kristelly et al., 2004). However, the LARG-W769D mutation exhibited only a minor defect in serum response element transcription assays in cells (Aittaleb et al., 2009). Sequence analysis predicts that the extension exists in all Lbc subfamily RhoGEFs, suggesting that it is not involved in regulation specific to the RH-RhoGEFs. Instead, recent data suggest that the N-terminal extension could be more important for autoinhibition of basal activity (see Solution Studies).

The second notable feature is a relatively long helical linker between the DH and PH domains that allows for considerable conformational flexibility between the two domains. For example, there is a ∼30° difference in orientation of the PH domain relative to the DH domain between the LARG DH/PH and DH/PH-RhoA structures, the hinge occurring at the junction of the α6 and αN helices. The four unique DH/PH domains in the LARG DH/PH-RhoA structure also exhibit conformational differences in the α6-αN linker (Kristelly et al., 2004). Because the mutation of residues in the PH domain that directly interact with RhoA inhibits GEF activity (Kristelly et al., 2004), these contacts were proposed to alter the conformation of the α6-αN linker region in a manner that facilitates the binding of RhoA. Indeed, residues at the end of α6 make critical interactions with switch II of the bound GTPase (Aghazadeh et al., 1998). Interactions with other proteins or phospholipids that likewise influence the relative orientation of the DH and PH domains could likewise affect LARG activity. Solution studies of the PDZ-RhoGEF DH/PH domains found no evidence for an analogous large conformational change in the linker region (Cierpicki et al., 2009), suggesting that in PDZ-RhoGEF the PH domain may simply serve to stabilize the α6 helix of the DH domain.

Finally, in all three DH/PH structures, a solvent-exposed patch of conserved hydrophobic residues is observed in the PH domain (Fig. 1e,f). Depending on the crystal structure, the patch forms either a 2-fold crystallographic or a quasi-2-fold molecular interface that buries ∼800 Ų of surface area. Located on the side of the PH domain β barrel, the patch would be in position to bind adjacent peripheral membrane proteins or domains at the cell membrane. Mutation of residues in the hydrophobic patch of LARG had no effect on nucleotide exchange activity in vitro but abrogated the ability of LARG to induce RhoA activation and stress fiber formation in cells (Aittaleb et al., 2009). Because the activity of these mutants could be rescued by fusion with nonspecific membrane targeting motifs, the hydrophobic patch seems to contribute to the proper localization of LARG by interacting with unknown target(s) at the cell membrane. However, membrane targeting by constitutively activated Gα₁₃ did not rescue activity, suggesting that the hydrophobic patch also plays an essential role in GPCR-mediated regulation, although it does not yet seem to involve direct interactions with activated Gα₁₃ (Aittaleb et al., 2009). The hydrophobic patch is conserved among all Lbc subfamily members, once again indicating a common functional role. Indeed, mutation of the hydrophobic patch of Lbc likewise abolished its ability to promote RhoA-mediated transcriptional activation in cells (Aittaleb et al., 2009).

Other Functional Motifs.

p115RhoGEF, LARG, and PDZ-RhoGEF undergo homo- and heterodimerization mediated by their C-terminal tails, and specific coiled-coil motifs have been identified in p115RhoGEF and LARG (Eisenhaure et al., 2003; Chikumi et al., 2004; Grabocka and Wedegaertner, 2007) (Fig. 2a). Deletion of the C terminus augments the ability of RH-RhoGEFs to induce RhoA activation in cells, suggesting a negative regulatory role for dimerization, presumably via the recruitment of unknown inhibitory factors (Eisenhaure et al., 2003; Chikumi et al., 2004). A proline-rich motif positioned immediately C-terminal to the PDZ-RhoGEF PH domain (residues 1081-1119) has been shown to be essential for plasma membrane localization of PDZ-RhoGEF and is required for PDZ-RhoGEF induced cortical actin reorganization and cell rounding (Togashi et al., 2000). The RH-DH/PH linker region also contains functional elements. In addition to an acidic autoinhibitory motif (Zheng et al., 2009), the RH-DH/PH linker of PDZ-RhoGEF contains a unique actin-binding motif (Banerjee and Wedegaertner, 2004; Banerjee et al., 2009).

Solution Studies.

Two recent solution-based studies have provided new insights into RH-RhoGEF regulation (Fig. 3a). Using surface plasmon resonance, the first study showed that Gα₁₃ could interact not only with the RH but also with the DH/PH and C-terminal regions of LARG (Suzuki et al., 2009b), consistent with earlier observations using coimmunoprecipitation (Wells et al., 2002; Bhattacharyya and Wedegaertner, 2003). The presence of all three domains (RH, DH, and PH) was required for formation of the highest affinity complex, and the K204A mutation of Gα₁₃, which abrogates binding to the RH domain (Nakamura et al., 2004; Grabocka and Wedegaertner, 2005), did not eliminate binding to the DH/PH domain. Thus, Gα₁₃ seems to exhibit at least two LARG binding surfaces: one for the rgRGS domain that is dependent on the activation status of Gα₁₃ and another for the DH/PH domains that uses activation-independent structural elements, potentially the C-terminal region of the Ras-like domain (Kreutz et al., 2007).

Truncation studies of PDZ-RhoGEF identified a conserved acidic motif (residues 706-712) a few amino acids upstream of the N-terminal helical extension of the DH domain that serves an autoinhibitory role (Zheng et al., 2009). NMR spectroscopy comparing the chemical shifts in DH/PH fragments with or without this acidic motif suggests that it interacts with basic residues on the DH domain, in particular Arg867 and Arg868. Because these residues also interact with the bound GTPase, their interaction with the acidic motif would form an autoinhibited state reminiscent of that of Vav (Aghazadeh et al., 2000). Indeed, charge-reversing mutations of four of the acidic residues in the motif (the “activation box”; Fig. 2a) enhanced GEF activity ∼20-fold in a DH/PH fragment of PDZ-RhoGEF. The enhancement was curiously only ∼2-fold for the same mutants in the context of a larger fragment that also contained the PDZ and RH domain. This may indicate that the acidic motif and the PDZ and RH domains act synergistically to autoinhibit activity. Acidic motifs also occur in the RH-DH linker regions of p115RhoGEF and LARG, although it remains to be seen if they function similarly. If the activation box is used in all Lbc subfamily members to regulate basal activity, it may provide a functional explanation for the N-terminal extension of the DH domain, which folds in such a manner that it would juxtapose the acidic motif of PDZ-RhoGEF with Arg867 and Arg868 on the surface of the DH domain.

Structural Studies of p63RhoGEF.

After delineation of the minimal fragment of p63RhoGEF required for high-affinity binding to Gα_q, the Gα_q-DH/PH-RhoA complex was isolated and crystallized. Its structure determination revealed that the C-terminal helix of the p63RhoGEF PH domain was unusually long (Fig. 1g), the hydrophobic residues on the extension of this helix binding directly into the effector docking site of Gα_q (Lutz et al., 2007). Mutation of these hydrophobic residues abrogates binding (Lutz et al., 2007; Rojas et al., 2007). Other residues, principally from the β1-β4 sheet of the PH domain, form specific contacts that help to dictate specificity for the Gα_q/11 subfamily. The interface is highly reminiscent of the interface between the PH domain of GRK2 and Gβγ (Lodowski et al., 2003), establishing a theme for how PH domains serve as protein-protein interaction modules. It is noteworthy that the manner in which p63RhoGEF engages Gα_q is distinct from how the rgRGS domain engages Gα₁₃ in that the GAP binding site on switch I is left freely accessible to RGS proteins. p63RhoGEF lacks significant GAP activity on its own but can allosterically regulate the binding of RGS proteins (Shankaranarayanan et al., 2008).

Although the principal interaction is via the PH domain, the C-terminal region of the Gα protein, including its C-terminal α5 helix, forms direct interactions with the DH domain and residues in the DH/PH domain interface. Mutation of some of these residues in p63RhoGEF leads to a loss of Gα_q activation in vitro but had only minor effects on binding, indicating that these contacts are most important for regulation of activity (Lutz et al., 2007). More recently, mutation of Tyr356 in the C terminus of Gα_q has also been shown to abrogate activation of p63RhoGEF (A. Shankaranarayanan, C. A. Boguth, S. Lutz, C. Vettel, F. Uhleman, M. Aittaleb, T. Wieland, and Tesmer JJ, submitted). Because the equivalent residue in Gα₁₆ is isoleucine, this specific contact may help explain why Gα₁₆ binds yet fails to activate p63RhoGEF.

Compared with structures of other characterized DH/PH tandem domains, the PH domain of p63RhoGEF in complex with Gα_q is held in an unusual orientation relative to the DH domain, being rotated ∼50° around the axis of the α6 helix relative to those of Dbs and TrioN, the closest homologs of known structure (Snyder et al., 2002; Chhatriwala et al., 2007). The PH domain makes no direct contacts with the substrate RhoA. Although the structure of p63RhoGEF in the absence of Gα_q is not yet available, it is anticipated that the domain-bridging interactions of Gα_q help to constrain the DH and PH domains in this conformation and that the DH/PH domains adopt a more Dbs-like conformation in the basal state.

Mechanism of Activation.

Unlike the RH-RhoGEFs, p63RhoGEF is predominantly found localized to actin-rich structures (Souchet et al., 2002), and overexpressed p63RhoGEF is localized to the cell surface (A. Shankaranarayanan, C. A. Boguth, S. Lutz, C. Vettel, F. Uhleman, M. Aittaleb, T. Wieland, and Tesmer JJ, submitted). Therefore, membrane translocation is not expected to play a major role in activation. In addition, unlike RH-RhoGEFs and Dbs, the PH domain of p63RhoGEF is autoinhibitory, the isolated DH/PH fragment having far less activity than the isolated DH domain (Lutz et al., 2004, 2007; Rojas et al., 2007). Thus, the PH domain must somehow interfere with the binding of RhoA. Indeed, the binding of Gα_q greatly enhances the affinity of RhoA for the DH/PH domains (A. Shankaranarayanan, C. A. Boguth, S. Lutz, C. Vettel, F. Uhleman, M. Aittaleb, T. Wieland, and Tesmer JJ, submitted). Because the PH domain does not inhibit GEF activity when added in trans (A. Shankaranarayanan, C. A. Boguth, S. Lutz, C. Vettel, F. Uhleman, M. Aittaleb, T. Wieland, and Tesmer JJ, submitted), the inhibition mediated by the PH domain seems to require a covalent linkage between the DH and PH domains, once again implicating the α6-αN linker as a region of regulatory importance.

Inspection of residues in the α6-αN linker region indicate that Arg341 of p63RhoGEF might sterically and electrostatically repel Arg68 in switch II of RhoA if the PH domain were in a more Dbs-like orientation relative to the DH domain. An R341A mutation indeed enhances the basal activity of p63RhoGEF but, surprisingly, does not eliminate activation by Gα_q (A. Shankaranarayanan, C. A. Boguth, S. Lutz, C. Vettel, F. Uhleman, M. Aittaleb, T. Wieland, and Tesmer JJ, submitted). In fact, the R341A mutation allows Gα_q to enhance activity to levels greater than those of the DH domain alone, indicating that Gα_q can activate p63RhoGEF not only by removing an autoinhibitory restraint but also by activating the DH domain allosterically. This is presumably carried out via the interactions of the C-terminal region of Gα_q with the DH domain, as observed in the crystal structure, which in turn alters the conformation of the α6-αN linker (A. Shankaranarayanan, C. A. Boguth, S. Lutz, C. Vettel, F. Uhleman, M. Aittaleb, T. Wieland, and Tesmer JJ, submitted). A schematic model for Gα_q-mediated activation of p63RhoGEF and related DH/PH domains is shown in Fig. 3b.

Mechanism of Activation.

In cells, P-Rex1 is activated when both Gβγ and PI3K are overexpressed; in vitro, PIP₃ and Gβγ act synergistically to activate the enzyme (Welch et al., 2002). Not all Gβγ subunit isoforms can activate P-Rex1 equally, although the ones with the most activity seem to be the same ones that activate PI3K the most (Mayeenuddin et al., 2006). Gγ₁ and Gγ₁₁, two of the Gγ subunits that were less effective at activating P-Rex1, are farnesylated instead of geranylgeranylated, suggesting that their ability to stably associate with the cell membrane is important for regulation by Gβγ. Phosphorylation of P-Rex1 by PKA inhibits activation by Gβγ by approximately 50-fold (Mayeenuddin and Garrison, 2006), and thus P-Rex1 can be regulated by GPCRs via both direct (Gβγ) and indirect (PIP₃ and PKA) pathways.

To delineate the regions that are important for Gβγ and PIP₃ binding, truncation studies of P-Rex1 were conducted. The isolated DH/PH tandem domain fragment exhibited higher activity than the full-length protein, indicating that other domains of the protein are involved in autoinhibition of basal activity. Of all the domains, the PH domain was found to be most important for PIP₃ activation of P-Rex1, although it lacks obvious sequence signatures for phospholipid binding found in other PIP₃-binding PH domains, and there was evidence for at least one other PIP₃ binding site in the protein (Hill et al., 2005).

Identification of the Gβγ binding-site has not been as clear-cut. The isolated DH/PH tandem domain of P-Rex1 was reportedly activated by Gβγ, as was a variant with the PH domain deleted, suggesting that the DH domain, being the only domain in common between the constructs, was sufficient for binding and regulation by Gβγ (Hill et al., 2005). However, direct activation of the isolated DH domain was not shown. Coimmunoprecipitation assays using various fragments of P-Rex2b suggested instead that the PH domain contained the binding site for Gβγ (Li et al., 2005). A more recent study of P-Rex1, again using coimmunoprecipitation, indicated that the DH domain does not directly bind Gβγ. Rather, an intramolecular complex formed by the second DEP, first PDZ, and IP4P domains creates a binding site for Gβγ. PKA phosphorylation, presumably at consensus sites found in these domains, weakened these interdomain contacts and thus inhibited Gβγ binding (Urano et al., 2008). This latter result, however, is difficult to reconcile with the fact that P-Rex2b seems to be regulated by Gβγ and altogether lacks an IP4P domain (Li et al., 2005) and that the DH/PH tandem domains of P-Rex1 are sufficient to observe PIP₃- and Gβγ-mediated membrane translocation (Barber et al., 2007). Thus, although PIP₃-mediated membrane recruitment represents one component of the activation mechanism for P-Rex subfamily RhoGEFs, the molecular basis for allosteric control of enzymatic activity by Gβγ remains obscure, except that it probably results in the release of the inhibition mediated by domains C-terminal to the DH domain (Fig. 3c).