Determinants at the N- and C-termini of Gα12 required for activation of Rho-mediated signaling

Background Heterotrimeric guanine nucleotide binding proteins of the G12/13 subfamily, which includes the α-subunits Gα12 and Gα13, stimulate the monomeric G protein RhoA through interaction with a distinct subset of Rho-specific guanine nucleotide exchange factors (RhoGEFs). The structural features that mediate interaction between Gα13 and RhoGEFs have been examined in crystallographic studies of the purified complex, whereas a Gα12:RhoGEF complex has not been reported. Several signaling responses and effector interactions appear unique to Gα12 or Gα13, despite their similarity in amino acid sequence. Methods To comprehensively examine Gα12 for regions involved in RhoGEF interaction, we screened a panel of Gα12 cassette substitution mutants for binding to leukemia-associated RhoGEF (LARG) and for activation of serum response element mediated transcription. Results We identified several cassette substitutions that disrupt Gα12 binding to LARG and the related p115RhoGEF. These Gα12 mutants also were impaired in activating serum response element mediated signaling, a Rho-dependent response. Most of these mutants matched corresponding regions of Gα13 reported to contact p115RhoGEF, but unexpectedly, several RhoGEF-uncoupling mutations were found within the N- and C-terminal regions of Gα12. Trypsin protection assays revealed several mutants in these regions as retaining conformational activation. In addition, charge substitutions near the Gα12 N-terminus selectively disrupted binding to LARG but not p115RhoGEF. Conclusions Several structural aspects of the Gα12:RhoGEF interface differ from the reported Gα13:RhoGEF complex, particularly determinants within the C-terminal α5 helix and structurally uncharacterized N-terminus of Gα12. Furthermore, key residues at the Gα12 N-terminus may confer selectivity for LARG as a downstream effector.


Background
The G12/13 subfamily of heterotrimeric guanine nucleotide binding proteins (G proteins) is comprised of two αsubunits in mammals, Gα 12 and Gα 13 , that have been implicated in a variety of physiological and pathological cellular responses that include proliferation, cytoskeletal rearrangements, migration, and metastatic invasion [1,2]. A diverse set of putative effector proteins have been identified as direct interactors with one or both G12/13 subfamily members; however, the roles of individual Gαeffector interactions in specific cellular responses remain largely undefined [3]. The most extensively characterized G12/13 target proteins are a subset of Rho-specific guanine nucleotide exchange factors (RhoGEFs) that activate the monomeric G protein Rho via tandem Dbl-homology /pleckstrin-homology domains [4]. The Rho monomeric GTPases are known primarily for their role in regulating actin cytoskeletal dynamics, but these proteins also mediate cell polarity, microtubule dynamics, membrane transport pathways, transcription factor activity, cell growth, and tumorigenesis [5]. The G12/13-RhoGEF-Rho axis mediates critical signaling and developmental pathways in model organisms that include Drosophila melanogaster [6], Caenorhabditis elegans [7], and zebrafish [8]. In addition, direct interaction with RhoGEFs is required for mutationally activated Gα 12 to trigger increased invasiveness of breast cancer cells [9]. Activated G12/13 α-subunits trigger Rho activation via binding and stimulation of three distinct RhoGEFs: p115RhoGEF, LARG and PDZ-RhoGEF [10][11][12][13]. This interaction is mediated primarily by a domain, located near the N-terminus of each RhoGEF, that is closely related to the regulator of G protein signaling (RGS) domain that defines the growing family of RGS proteins [14,15]. Although p115RhoGEF, LARG and PDZ-RhoGEF are highly similar in this "RGS homology" (RH) domain [16], these proteins appear to be activated by different mechanisms and play non-redundant roles in G12/13 subfamily-mediated signaling. Purified p115RhoGEF binds Gα 12 and Gα 13 and accelerates GTPase activity for both proteins, but only Gα 13 can stimulate p115RhoGEF to activate RhoA in vitro [10,17]. Interaction of Gα 12 or Gα 13 with purified LARG can trigger its activation of RhoA; however, stimulation by Gα 12 requires prior phosphorylation of LARG by the nonreceptor tyrosine kinase Tec [13]. Furthermore, studies utilizing small interfering RNA to hinder expression of specific RhoGEFs show that LARG is a specific downstream effector of thrombin receptor-mediated signaling, whereas signaling through the lysophosphatidic acid (LPA) receptor is attenuated by blocking PDZ-RhoGEF expression [18]. These results are compelling in light of a separate report that the thrombin receptor shows preferential coupling to Gα 12 , whereas the LPA receptor preferentially utilizes Gα 13 as a conduit to downstream signaling [19]. Although it is possible that Gα 12 stimulates a post-translationally modified form of p115RhoGEF or PDZ-RhoGEF in cells, the evidence to date suggests LARG as the most likely RhoGEF serving as a physiological effector for Gα 12 . Gains in our understanding of the specificity of RhoGEF engagement within the G12/13 subfamily should provide insights into the non-overlapping functions of Gα 12 and Gα 13 in signal transduction.
Crystallographic studies have revealed important structural aspects of the interaction between Gα 13 and the RH domain of p115RhoGEF, including numerous residues in both proteins that provide contact points [20,21]. Initially, purification of Gα 13 for crystallography required that it be engineered as a chimera in which amino acid sequence within several regions, including the N-and C-termini, was replaced by corresponding sequence from the Gi subfamily protein Gαi 1 [20]. The structure of the Gα 13 : p115RhoGEF-RH complex was later refined in crystallographic studies that utilized a Gα 13 chimera harboring Gαi 1 sequence only at the N-terminus. Because the Gα N-terminus was unstructured in this crystallized complex, any role of this region in RhoGEF interaction remains to be determined. Although the region of Gα 13 downstream of the Switch III region harbors several residues critical for RhoGEF engagement, notably Glu 273 , Thr 274 , Asn 278 , and Arg 279 within the α 3 helix and α 3 -β 5 loop, other regions closer to the Gα 13 C-terminus do not emerge in the crystal structure as providing key RhoGEF contact points [21].
In contrast to Gα 13 , a structure of Gα 12 in complex with a RhoGEF target has not been reported, although a chimeric Gα 12 harboring the N-terminus of Gαi 1 has been crystallized [22]. To examine the full sequence of Gα 12 for structural features mediating its interaction with RhoGEFs, we engineered a series of cassette substitutions within constitutively activated Gα 12 and examined these variants for in vitro binding to the RH domains of LARG and p115RhoGEF, as well as ability to drive the Rho-dependent process of serum response element (SRE) mediated transcription in cells [23]. Our results reveal unexpected regions of Gα 12 as harboring determinants of its functional interaction with RhoGEFs, and also identify key charged amino acids near the Gα 12 N-terminus that may confer selective binding to LARG.

Results
Myc-tagged Gα 12 retains RhoGEF binding, Rho-mediated signaling, and conformational activation To identify mutants of Gα 12 impaired in RhoGEF binding, we first sought to establish an in vitro system in which Gα 12 mutants could be expressed ectopically in cultured cells, rendered soluble in a detergent extract, and detected without interference from endogenous Gα 12 . We engineered the constitutively active Gln 229 Leu variant of Gα 12 (Gα 12 QL ) to harbor a myc epitope tag, flanked by linkers of the sequence SGGGGS and positioned between residues Pro 139 and Val 140 . This insertion site was chosen due to its approximate alignment with the position of green fluorescent protein in Gαq in a prior study [24]. We expressed myc-tagged and untagged Gα 12 QL in HEK293 cells, prepared detergent-soluble extracts, and analyzed these by immunoblotting. As shown in Figure 1A, myctagged Gα 12 QL was detected by both anti-myc and anti-Gα 12 antibodies, with the latter generating a much stronger signal while avoiding an off-target 37 kDa band detected in all samples by the anti-myc antibody. Also, the myctagged protein (~45 kDa) was readily discernible from endogenous Gα 12 and untagged Gα 12 QL (~43 kDa). Next, we subjected myc-Gα 12 QL to pulldown experiments using an immobilized GST fusion of the p115RhoGEF RH domain, as described in Methods. Myc-tagged and untagged Gα 12 QL bound to p115-RH with similar affinity ( Figure 1B), and comparison with mock-transfected cells indicated the~45 kDa band detected by anti-Gα 12 was dependent on transfection with the myc-Gα 12 QL plasmid. Furthermore, LARG-RH and p115-RH showed similar ability to co-precipitate myc-tagged Gα 12 QL ( Figure 1C). To ascertain that myc-Gα 12 is functional as a mediator of cellular signal transduction through Rho, we measured transcriptional activation of a luciferase reporter gene positioned downstream of the serum response element (SRE), a component of the c-fos promoter that provides a readout of Gα 12 -mediated Rho activation [23]. Myc-tagged and untagged Gα 12 QL exhibited similar ability to stimulate this response in HEK293 cells co-transfected with SRE-luciferase ( Figure 1D). Furthermore, trypsin digestion of HEK293 cell lysates harboring myc-Gα 12 QL yielded a protected fragment of~40 kDa, comparable to results observed previously with GTPγS-loaded, purified Gα 12 [25]. An inactive, constitutively GDP-bound (Gly 228 Ala) variant of myc-tagged Gα 12 did not yield this~40 kDa fragment when digested with trypsin ( Figure 1E). Taken together, these results suggest myc-Gα 12 QL undergoes conformational activation and retains normal signaling through the RhoGEF:Rho pathway. Because of the superior sensitivity of anti-Gα 12 antibody in detecting myc-Gα 12 QL , and the easily discernible gel shift of Gα 12 caused by the myc tag and linkers (see Figures 1A and B), we chose to utilize anti-Gα 12 to detect myc-Gα QL by p115RhoGEF. HEK293 cells extracts containing myc-Gα 12 QL were subjected to protein interaction assays (see Methods) using an immobilized GST fusion of the RH domain of p115RhoGEF (RGS) or GST alone (GST). Samples were washed, separated by SDS-PAGE, and analyzed by immunoblotting using antibodies described above. (C) Specificity of myc-Gα 12 QL detection in interaction assays. HEK293 cells transfected with either myc-Gα 12 QL (myc-12 QL ) or the empty pcDNA3.1 plasmid (vect) were lysed and assayed for binding to GST fusions of the RH domain of p115RhoGEF (p115) or LARG, or GST alone (GST). Immunoblot analysis was performed using anti-Gα 12 antibody as described above. (D) Serum response element (SRE) luciferase activation by myc-Gα 12 QL . HEK293 cells grown in 12-well plates were co-transfected with the plasmids SRE-L (0.2 μg) and pRL-TK (0.02 μg), plus 0.1 μg of the plasmid indicated on the X-axis. Y-axis values show firefly luciferase signal normalized for Renilla luciferase signal within each sample. (E) Trypsin protection assays of myc-tagged Gα 12 . Lysates from HEK293 cells transfected with myc-Gα 12 QL (myc-12 QL ) or the constitutively GDP-bound Gly 228 Ala mutant of wildtype Gα 12 (myc-12 G228A ) were subjected to trypsin digests as described in Methods. Immunoblot analysis was performed using J169 antibody [25] at 1:700 dilution. mutants in which sextets of consecutive amino acids in myc-Gα 12 QL are replaced by the sextet Asn-Ala-Ala-Ile-Arg-Ser ( Figure 2 shows the native amino acid sextet and alphabetical designation for each mutant). This strategy of "NAAIRS" cassette substitutions was chosen due to prediction of this motif being tolerated in the three-dimensional structure of proteins [26], prior use of this approach in mapping functional regions of both retinoblastoma and the telomerase catalytic subunit [27,28], and our previous success employing this strategy to identify Gα 12 determinants of binding to the scaffolding subunit of protein phosphatase-2A and the cytoplasmic tail of polycystin-1 [29,30]. Variants of Gα 12 were expressed in HEK293 cells and tested for interaction with immobilized LARG-RH, as described in Methods. As shown in Figure 3, myc-Gα 12 QL was co-precipitated by a GST fusion of LARG-RH but not by GST alone. Many of these cassette mutants yielded a moderate-to -robust signal in the LARG-precipitated fraction; however, a subset displayed a weak or absent signal ( Figure 3). To assess impairment of LARG binding for each myc-Gα 12 QL variant, we quantified the band intensity for each precipitated sample (pulldown), and divided this by the band intensity in the starting cellular extract (load). These calculations generated a "pulldown:load ratio" for each mutant, and also for the positive control myc-Gα 12 QL that was tested in each experiment. Nearly all cassette mutants were solubilized by our detergent conditions and detected by immunoblotting; exceptions were mutant W, which we did not engineer due to overlap with the insertion site of the myc tag (see Figure 2), and mutant CC due to low expression levels that   produced inconclusive results (data not shown). As shown in Table 1, the majority of cassette mutants exhibited pulldown:load ratios greater than 40% of the ratio determined for myc-Gα 12 QL . However, a number of mutants exhibited lower pulldown:load ratios (<20% of positive control) with a subset generating a ratio less than 10% of the positive control. For all samples, precipitation by immobilized GST yielded no Gα 12 signal (Figure 3), indicating these mutants were not merely binding the GSTglutathione-sepharose complex nor forming insoluble aggregates under these in vitro conditions. Also, we examined the full panel of Gα 12 cassette mutants for interaction with a GST fusion of the N-terminal 252 amino acids of p115RhoGEF (p115-RH). None of the LARG binding- LARG p115 Figure 3 In vitro interaction of Gα 12 mutants with LARG. Immunoblot results for all LARG binding-impaired Gα 12 cassette mutants and selected other mutants are shown. HEK293 cells were transfected with the indicated plasmids (7.0 μg per 10-cm plate) and lysates were prepared for coprecipitation assays as described in Methods. Prior to this step, 5% of each lysate was set aside as starting material (load). Pulldown experiments were performed on 7-9 mutants per experiment, plus myc-Gα 12 QL as a positive control, using equal amounts of GST-LARG-RH (LARG) immobilized on glutathione-sepharose. Immobilized GST was utilized in parallel as a negative control. For all experimental samples, 20% of the volume was analyzed by SDS-PAGE and Coomassie blue staining to verify equal amounts of GST-LARG-RH and GST proteins in the precipitates (data not shown). Immunoblots displayed in this figure are representative of at least three trials per cassette mutant, except for mutants A-D, F-H, V, and KKK that showed minimal impairment in LARG binding after two trials. (Inset) Coomassie blue analysis of GST-fusion constructs expressed in bacteria and immobilized on glutathione-sepharose: GST-LARG-RH (LARG), GST-p115-RH (p115), and GST alone. Molecular weight standards (in kDa) are indicated at right.  Cassette substitution mutants of myc-Gα 12 QL (see Figure 2 for alphabetical designations) were expressed in HEK293 cells and subjected to protein interaction assays using a GST-fusion of the RH domain of LARG as described in Methods, and for each mutant a pulldown:load ratio was determined and calculated as a percent (left column) of the same ratio for unmodified myc-Gα 12 QL assayed in parallel. Each Gα 12 mutant was analyzed in three independent experiments, except for mutants that appeared in the 70-100% category in two independent experiments. impaired mutants (those with pulldown:load ratio <20% of positive control; see Table 1) yielded a signal intensity in the p115-RH precipitate that exceeded 50% of intensity for the positive control myc-Gα 12 QL (data not shown). The Gα 12 cassette mutant designated OO was among those impaired in LARG binding, consistent with our previous work demonstrating its uncoupling from Rhomediated signaling [31], and several other cassette substitutions within the Switch regions disrupted binding to LARG (mutants HH, LL, MM, NN, QQ, and RR; see Figure 2). However, impaired LARG binding also was caused by substitutions in other regions of Gα 12 (Table 1). Prior crystallographic studies identified several residues in Gα 13 that serve as contact points with p115-RH [20,21]. Table 2 lists Gα 13 residues identified as contact points with p115-RH in these earlier studies, and indicates the corresponding Gα 12 cassette mutant for each Gα 13 residue. From our in vitro binding results (Table 1), it is apparent that most Gα 12 mutants corresponding to RhoGEF-contacting Gα 13 residues displayed partial or severe impairment of LARG binding, mutants V, BB and DDD being exceptions. However, several RhoGEFuncoupling substitutions in Gα 12 (cassette mutants E, I, J, K, M, Z, NN, OO, VV, AAA, EEE, FFF, GGG and HHH) replaced amino acids that do not correspond to Gα 13 contacts with p115-RH. Gα 12 mutants J and K replaced sections of the P-loop, a motif critical in guanine nucleotide binding, and thus would be predicted as impaired in signaling. However, our finding of RhoGEF-uncoupling mutations at the N-and C-termini of Gα 12 was unexpected, because these regions either lacked corresponding contact points in the Gα 13 :p115-RH complex or were disordered in the G12/13 crystal structures (i.e. the N-terminus). To determine whether these N-and C-terminal mutations in Gα 12 are impaired in Rho-mediated signaling, we expressed these variants in HEK293 cells and measured stimulation of SRE-luciferase transcription. All N-and C-terminal mutants impaired in RhoGEF binding were poor activators of this reporter gene ( Figure 4A). Several cassette mutants in the N-and C-terminal regions of Gα 12 that displayed normal binding to LARG (mutants F, V, and KKK) stimulated SRE-luciferase in a manner comparable to the myc-Gα 12 QL positive control ( Figure 4A). With the exception of mutant VV, immunoblot analysis of HEK293 cell lysates revealed expression levels of these mutants similar to myc-Gα 12 QL ( Figure 4B).

Conformational activation of RhoGEF-uncoupled Gα 12 mutants
A concern in our experimental approach was that specific "NAAIRS" cassette substitutions could cause global disruption of Gα 12 shape, so that a mutant might fail to assume an activated conformation. For RhoGEFuncoupled Gα 12 mutants at the N-terminus (i.e. upstream of the P-loop) and C-terminus, we measured protection against trypsin proteolysis. Exchange of GDP for the activating GTP on Gα proteins triggers a conformational change that conceals a trypsin cleavage site within the Switch II region; this property allows the activated state of the Gα protein to be revealed by resistance to trypsin [25,32]. As shown in Figure 5A, mutants E, I, and HHH yielded a protected fragment of approximately 40 kDa that matched the fragment observed following tryptic digestion of myc-Gα 12

QL
. Results for mutant AAA were difficult to interpret; a band of slightly smaller size than undigested AAA was generated by tryptic digestion, but it was unclear whether this matched the~40 kDa trypsin-protected fragment in myc-Gα 12 QL . Other C-terminal mutants we tested− VV, EEE, FFF, and GGG− appeared to match the constitutively inactive myc-Gα 12 G228A which lacked this~40 kDa fragment ( Figure 5A). These results suggest several C-terminal mutants of Gα 12 were sufficiently distorted in shape by the "NAAIRS" substitution to allow trypsin access to proteolytic sites normally not exposed in the GTP-bound state. However, cassette mutants E and I at the N-terminus and HHH at the C-terminus appeared to maintain an activated conformation despite their impairment in RhoGEF binding and SRE stimulation.
We also tested whether RhoGEF-uncoupled cassette mutants at the N-and C-termini of Gα 12 could interact in vitro with other reported binding partners: heat shock protein-90, protein phosphatase-5, the scaffolding Aα  [20,21] are indicated in the left column. Cassette mutants ("NAAIRS" substitution) in which the homologous residue(s) within Gα 12 have been altered are indicated in the right column. See Figure 2 for Gα 12 mutant designations.  subunit of protein phosphatase-2A, and the cytoplasmic tail of E-cadherin [33][34][35][36]. As shown in Figure 5B, each mutant displayed pulldown:load ratios >60% of the positive control, myc-Gα 12 QL , for at least two of these non-RhoGEF targets. Taken as a whole, these findings reveal a subset of mutations at the N-and C-terminus that selectively uncouple Gα 12 from RhoGEFs while preserving conformational activation and ability to bind other downstream proteins.
We next visualized the position of these RhoGEFinteracting regions in the crystal structure of a Gα 12 chimera in which the N-terminal 48 residues were replaced by the N-terminus of Gαi 1 [22]. The native region of Gα 12 replaced in cassette mutant E is not ordered in this structure; however, the regions replaced in the C-terminal mutants EEE-HHH are highlighted ( Figure 6). The sextet replaced in mutant HHH (highlighted in black) resides in the α 5 helix that extends along the Gα 12 surface and approaches the C-terminus at the top of the diagram.

Differential uncoupling of Gα 12 from LARG and p115RhoGEF
We next sought to identify specific residues within these N-and C-terminal sextets of Gα 12 that mediate RhoGEF interaction. To examine putative surface residues, we performed charge substitutions in the native regions corresponding to cassette mutants E, I, and HHH, and examined these variants for SRE-luciferase activation. None of the single-residue charge-reversals in the regions encompassed in mutants I or HHH caused significant decrease in SRE signaling (data not shown). However, a double charge-reversal in the mutant E region, converting Glu 31 and Glu 33 to Arg residues, caused a near-complete loss of SRE activation in HEK293 cells despite normal levels of protein expression ( Figure 7A). We next examined this Gα 12 mutant, designated Glu 31/33 Arg, for binding to the RH domains of LARG and p115RhoGEF. As shown in Figure 7B, a selective loss of RhoGEF binding was observed: the Glu 31/33 Arg charge-reversals severely disrupted LARG-RH binding relative to non-mutated myc-Gα 12 QL (pulldown:load ratio~18% of control) but had minimal effect on p115-RH binding (ratio~86% of control). In trypsin protection assays, the Glu 31/33 Arg mutant yielded a protected fragment at the same molecular weight (~40 kDa) as observed for the myc-Gα 12 QL positive control, suggesting its ability to attain an activated conformation ( Figure 7C). The intermediate intensity of this band (approximately a midpoint between activated Gα 12 and the constitutively inactive Gly 228 Ala variant) may be due in part to the mutational introduction of Arg residues providing additional sites for trypsin proteolysis. Taken as a whole, these findings not only provide evidence that the structurally uncharacterized N-terminus of Gα 12 plays a role in its functional interaction with RhoGEFs, but also reveal individual charged residues in this region as candidates for conferring specificity of Gα 12 for LARG among the RH-containing RhoGEFs.

Discussion
The G12 subfamily members Gα 12 and Gα 13 are welldocumented as utilizing RhoGEFs as downstream signaling effectors. Crystallographic studies by Chen et al. [20] and Hajicek et al. [21] have provided intricate structural details of the interaction between Gα 13 and the RH domain of p115RhoGEF, identifying a set of Gα 13 residues that directly contact this target protein. The structure of Gα 12 also has been elucidated, using a chimera comprised of amino acids 49-379 of Gα 12 preceded by amino acids 1-28 of Gαi 1 [22]. However, a Gα 12 :RhoGEF complex has not been reported. In the current study, we utilized in vitro and cell-based approaches to examine the interaction between Gα 12 and two putative target RhoGEFs, LARG and p115RhoGEF. Using immobilized RGS-homology (RH) domains of these RhoGEFs, we identified several substitutions of native amino acids in Gα 12 that disrupted its binding to these proteins and blocked its ability to stimulate the Rho-dependent process of SRE-mediated transcription. Although our results indicated that a number of common determinants Figure 6 Structural position of Gα 12 C-terminal determinants of RhoGEF binding. The structure of N-terminally Gαi 1 -substituted Gα 12 (PDB accession code 1ZCA, [22]) as a GDP•AlF 4 activated complex was analyzed using PyMOL software. The native Gα 12 region substituted for the sequence "NAAIRS" in the C-terminal mutants EEE, FFF, and GGG is highlighted in orange, and the sextet substituted in mutant HHH is highlighted in black. The bound GDP molecule is highlighted in blue. in Gα 12 and Gα 13 mediate RhoGEF binding, several RhoGEF-uncoupling mutations in Gα 12 did not correspond to regions of RhoGEF contact within Gα 13 ; these include amino acid sextet substitutions in the C-terminal α 5 helix as well as the structurally uncharacterized N-terminus . Several of these Gα 12 mutants exhibited protection from tryptic digestion as well as unimpeded binding to other, non-RhoGEF targets, indicating their impaired interaction with RhoGEFs is not caused by failure to attain an activated conformation and suggesting the shapes of other effectorbinding surfaces in these Gα 12 mutants remain intact as RhoGEF interaction is disrupted.
Although Gα 12 and Gα 13 share 67% amino acid identity and bind several common downstream targets, several functional differences between these Gα proteins suggest their signaling mechanisms are not redundant [1,3]. Both Gα 12 and Gα 13 bind LARG and p115RhoGEF [10,12], and both of these RhoGEFs accelerate GTPase activity of purified Gα 12 and Gα 13 in single-turnover assays [13,17]. Whereas Gα 13 stimulates both p115RhoGEF and LARG to trigger guanine nucleotide exchange on RhoA in vitro, Gα 12 can only stimulate LARG under these experimental conditions, and in a manner dependent on prior phosphorylation of LARG by the tyrosine kinase Tec [10,13]. Also, activated Gα 12 is more potent than Gα 13 in recruiting the RH domain of p115RhoGEF to the plasma membrane, and specific mutations in p115RhoGEF disrupt Gα 12 but not Gα 13 in triggering this localization [37]. At the cellular and organismal levels, it is increasingly clear that Gα 12 and Gα 13 utilize non-overlapping signaling pathways. Mice lacking Gα 13 die early in embryogenesis due to defects in vascular development and thrombin-induced cell migration, but mice lacking Gα 12 do not display these developmental defects. However, knockout of Gα 12 combined with absence of Gα 13 causes earlier lethality than Gα 13 knockout alone, and in mice lacking one Gα 13 allele, at least one Gα 12 allele must be present for normal embryonic development [38,39]. Furthermore, LPA-induced activation of mTOR complex 2 leading to activation of PKC-δ requires Gα 12 but not Gα 13 [40]. Because of these differences, plus the increasing list of Gα 12 -specific effector proteins (including another RhoGEF, AKAP-Lbc, that is activated exclusively by Gα 12 within the G12/13 subfamily), we believe the Gα 12 :RhoGEF interface cannot be defined summarily by structural features of the Gα 13 :RhoGEF complex.
Among the Gα 13 residues that provide contact points with p115RhoGEF in crystallographic studies [20,21], many have corresponding residues within Gα 12 , and therefore we paid particular attention to Gα 12 cassette mutants corresponding to these key Gα 13 residues (see Table 2). For example, the Gα 12 mutant HH replaced residues corresponding to Gα 13 residues Arg 200 , and  Figure 2) and the double charge substitution mutant Glu 31/33 Arg were compared to myc-Gα 12 QL (12 QL ) in SRE-luciferase assays under the cell transfection conditions described in Figure 4. A constitutively GDP-bound variant of wildtype myc-Gα 12 (12 G228A ) was assayed in parallel as a negative control. Results shown are the mean of three independent experiments, and error bars indicate range. (B) Protein-protein interaction assays. Detergent-soluble extracts from transfected HEK293 cells transfected with myc-Gα 12 QL , the Glu 31/33 Arg mutant, or empty pcDNA3.1 plasmid (vector) were subjected to co-precipitation assays as described in Methods, using GST-fusions of either LARG-RH (LARG), p115RhoGEF-RH (p115), the N-terminal domain of the Gα 12 target radixin [46], or no adduct (GST). Prior to the precipitation step, 5% of each lysate was set aside as starting material (load). Table values show the pulldown:load ratio for Glu 31/33 Arg as a percent of the positive control value (12 QL ), with mean +/-range presented for three independent experiments. (C) Trypsin protection of the Glu 31/33 Arg mutant, in comparison to constitutively GTP-and GDP-bound Gα 12 . Assays were performed as described in Methods. Results shown are representative of two independent experiments. Lys 204 , both of which provide contact points with p115-RH. In another Gα 12 cassette mutant, termed RR, a substituted residue corresponds to Arg 260 within Gα 13 ; this residue provides a key contact with amino acids within the βN-αN region of p115RhoGEF. Also, Gα 12 cassette mutants Q, R, and S contain altered residues in the Gα 12 helical domain that correspond to p115-RH interacting residues in Gα 13 . Among the Gα 12 mutants corresponding to p115-RH contact points in Gα 13 , most showed impaired RhoGEF interaction and poor stimulation of SRE-mediated signaling. However, several differences between Gα 12 and Gα 13 were noted, particularly in the helical domain. Gα 12 cassette mutant V alters residues that correspond to two contact points within the Gα 13 :p115-RH complex; however, this mutant showed minimal impairment in RhoGEF binding in vitro and stimulated SRE-mediated transcription robustly in cells. Gα 12 mutant BB, which removes a Phe corresponding to a Gα 13 contact point with p115-RH, displayed a slight impairment in SRE-mediated transcriptional activation and no impairment of RhoGEF binding. In addition, Gα 13 utilizes a C-terminal residue (Arg 335 ) as a contact point with p115-RH, but the corresponding Gα 12 cassette mutant (DDD) exhibited normal binding to RhoGEFs and only modest impairment in SRE signaling. However, because this cassette mutant preserves the corresponding Arg residue in Gα 12 (DRKRRN substituted for NAAIRS), it is possible this Arg in Gα 12 participates in RhoGEF binding despite the alteration in adjacent amino acids.
Aside from the N-and C-terminal mutants of Gα 12 that show impaired RhoGEF binding, we have identified other RhoGEF-uncoupling mutations in Gα 12 that lack corresponding Gα 13 contact points for p115-RH (see Tables 1 and 2). None of the native Gα 12 residues replaced in cassette mutants M and Z match p115-RH contact points in Gα 13 , and thus may indicate Gα 12 -specific determinants of RhoGEF interaction. Impaired RhoGEF binding also was observed in Gα 12 mutants J and K; however, this most likely was due to these substitutions disrupting the canonical GXGXXGKS guanine nucleotide binding motif [41]. Although our results suggest a core similarity in the mechanisms utilized by Gα 12 and Gα 13 to engage RhoGEF targets, it is apparent that several determinants of RhoGEF binding are unique to Gα 13 . We have identified determinants that may be unique to Gα 12 or potentially important for both G12/13 subfamily members in RhoGEF engagement. Studies of Gα 13 variants harboring corresponding mutations will be important in distinguishing these possibilities.
A role for the C-terminus of G12/13 subfamily proteins in RhoGEF engagement has been suggested by prior studies. Kreutz et al. [42] engineered chimeras of Gα 12 and Gα 13 that were interchanged downstream of the Switch III region, and demonstrated the C-terminal 114 amino acids of Gα 13 as sufficient for its unique ability to stimulate purified p115RhoGEF to activate RhoA. Also, a chimeric Gα 13 in which the region downstream of Switch III was replaced by the corresponding region of Gαi 2 displayed loss of ability to stimulate SRE-mediated transcriptional activation [43]. Initial crystallographic studies of Gα 13 :RhoGEF interaction utilized a chimeric Gα 13 harboring Gαi 1 sequence at the C-terminus, and determinants of RhoGEF binding were not found downstream of the Switch regions in this protein [20]. Subsequent crystallographic work utilizing Gα 13 with native C-terminal sequence did identify residues slightly downstream of the Switch III region as critical for RhoGEF engagement [21], and also revealed a more distal residue in the C-terminal region (Arg 335 ) as providing a contact point with the RH domain of p115RhoGEF. However, no residues at the extreme C-terminus of Gα 13 , including the α 5 helix, were found to mediate RhoGEF binding. Our results suggest differences between Gα 12 and Gα 13 in the role of the C-terminus, as several substitutions near the extreme C-terminus of Gα 12 disrupted RhoGEF interaction, most notably the cassette mutant HHH within the α 5 helix.
The N-terminus provides the greatest amino acid sequence divergence between Gα 12 and Gα 13 . Gα subunits utilize this region for interaction with Gβγ [44], and in Gα 12 and Gα 13 this region confers specificity of coupling to thrombin and LPA receptors, respectively [19]. Importantly, Gα 13 is a more potent stimulator of RhoGEF activation in vitro than a chimeric Gα 13 harboring the N-terminus of Gαi 1 , indicating a possible role of the Gα 13 N-terminus in RhoGEF activation [21]. However, specific determinants within the N-terminus of G12/13 subfamily proteins that mediate binding to effectors, including RhoGEFs, have not been reported. The 48-residue region at the N-terminus of Gα 12 has not been characterized in crystallographic studies, because its replacement by the Gαi 1 N-terminus was necessary for obtaining sufficient quantities of purified protein [16,22]. Furthermore, the N-terminus was disordered in crystallographic analysis of both the aforementioned Gαi 1 /Gα 13 hybrid and a more recent structure of fulllength Gα 13 [21], suggesting the Gα 12 N-terminus may be refractory to crystallographic analysis even if native sequence is utilized. Our approach of employing cassette substitution mutants throughout the length of Gα 12 has provided an indirect means of circumventing this obstacle, and has revealed specific N-terminal regions as possible determinants of RhoGEF interaction. Importantly, our discovery that mutations in this N-terminal region (cassette mutants E and I) cause loss of RhoGEF binding allowed us to focus on putative surface residues in these substituted regions, ultimately revealing Glu 31 and Glu 33 as critical for Gα 12 interaction with LARG and stimulation of SRE-mediated transcription. Our finding that charge substitutions of these N-terminal Gα 12 residues disrupted binding to the LARG-RH domain but had minimal effect on interaction with the corresponding domain of p115RhoGEF was intriguing, and suggested these residues play a role in targeting Gα 12 preferentially to LARG. It is possible that Gα 12 harbors sufficient RhoGEF-interacting surfaces for in vitro binding to p115RhoGEF, but that a functional, physiological interaction (i.e. with LARG) requires this N-terminal region. Our RhoGEF binding results for Gα 12 cassette mutant E, as well as the more specific Glu 31/33 Arg mutant, were surprising in light of earlier findings that RhoGEF binding was preserved in a Gα 12 chimera containing the Gαi 1 N-terminus [22]. It is possible that "NAAIRS" substitution and particularly the Glu 31/33 Arg chargereversals cause a more dramatic change to this RhoGEF binding surface than occurs when Gαi 1 sequence is introduced. Cassette mutant E and the Glu 31/33 Arg mutant are impaired in activating the Rho-dependent readout of SRE-mediated transcriptional activation in cells, and it remains to be determined whether the Gαi 1 /Gα 12 chimera is similarly impaired in stimulating this pathway.
Because previous phosphorylation of LARG by Tec is a requirement for Gα 12 , but not Gα 13 , for in vitro activation of Rho, it will be important to determine whether this phosphorylation event regulates interaction of LARG with Gα 12 , particularly its N-terminus and C-terminal α 5 helix. Furthermore, as suggested by Hajicek et al. [21], it is conceivable that post-translational modification of p115RhoGEF in cells modulates its responsiveness to Gα 13 or could potentially render it a target of Gα 12 . A challenge for future studies of Gα 12 -and Gα 13 -mediated signaling will be to determine the combinations of G12/ 13 subfamily α-subunits and RhoGEFs that activate Rho in response to different signaling inputs, and in different cell and tissue types.

Conclusions
Gα 12 and Gα 13 define the G12/13 class of heterotrimeric G protein α-subunits, which participate in numerous signaling pathways through stimulation of RhoGEFs that subsequently activate Rho. Although these proteins are non-redundant in their stimulation of effectors and their cellular and organismal roles, only Gα 13 has been characterized in the structural basis of its interaction with RhoGEF targets. However, the involvement of Gα 12 in stimulating SRE-mediated transcription, cell rounding, c-Jun N-terminal kinase activation, cell growth, and metastatic invasion supports a physiological role for a Gα 12 -RhoGEF-Rho axis in developmental pathways and disease progression [45]. Therefore, an improved understanding of the structural aspects of Gα 12 :RhoGEF interaction likely will be of broad importance. Our results provide several key additions to this structural model: 1) characterization of the Gα 12 :RhoGEF interacting surface by identifying regions in Gα 12 that mediate binding; 2) unexpected roles of the Gα 12 N-terminal region and C-terminal α 5 helix in engagement of RhoGEFs; 3) identification of specific residues near the Gα 12 N-terminus that may mediate its selectivity for LARG as an effector protein. To date, no structural studies have examined the interaction of Gα 12 with RhoGEFs. Our hope is that mutant-based strategies will augment such crystallographic approaches and provide key details toward understanding the structural aspects and biological role of this Gα:effector interaction.

DNA constructs
Plasmids encoding 1) a fusion of glutathione-S-transferase (GST) to amino acids 320-606 of LARG (GST-LARG-RH), and 2) amino acids 1-252 of p115RhoGEF with an N-terminal myc epitope tag were kindly provided by Tohru Kozasa (Univ. of Ill., Chicago). We used PCR to subclone the p115RhoGEF sequence into pGEX-2T (GE Healthcare) to produce GST-p115-RH. All "NAAIRS" amino acid substitution mutants within myc-tagged Gα 12 Gln 229 Leu (myc-Gα 12 QL ) were engineered as described previously [29]. Single amino acid substitutions were engineered in myc-Gα 12 QL using the QuikChange II W sitedirected mutagenesis system (Agilent Technologies), and this system was used to engineer a constitutively inactive Gly 228 Ala variant (myc-Gα 12 G228A ) within a plasmid encoding myc-tagged, wildtype Gα 12 (provided by Pat Casey, Duke University). The luciferase reporter plasmid SRE-L was a gift from Channing Der (University of North Carolina Chapel Hill).

Expression and immobilization of GST fusion proteins
GST fusion constructs were transformed into BL21 (Gold)-DE3 cells (Stratagene). Cells were grown under 75 μg/ml ampicillin selection to OD 600 of 0.5−0.7, and recombinant protein expression was induced using 0.5 mM isopropyl-β-D-thiogalactopyranoside (Fisher Scientific). After 3 h, cells were lysed on ice using 0.32 mg/ml lysozyme (MP Biomedicals), and GST fusion proteins were bound to glutathione-sepharose 4B (GE Healthcare) as described previously [31,34]. Following three washes in 50 mM Tris pH 7.7 supplemented with 1 mM EDTA, 1 mM dithiothreitol, and 150 mM NaCl, samples were snap-frozen in aliquots and stored at −80°C.

Protein interaction assays
HEK293 cell extracts were diluted in NAAIRS Lysis buffer lacking polyoxyethylene-10-lauryl ether, using sufficient volume to dilute this detergent in the samples to 0.05% (w/v). Next, sepharose-bound GST fusion proteins were added and allowed to incubate for approximately 2 h at 4°C with continuous inversion. A percentage of the diluted extract was set aside as starting material prior to sepharose addition. Next, samples were centrifuged at 1,300 g, and pellets were washed three times and then subjected to SDS-PAGE and immunoblot analysis using an antibody specific to the Gα 12 N-terminus (Santa Cruz Biotechnology) or the myc 9E10 epitope tag (Zymed), followed by alkaline phosphatase conjugated secondary antibodies (Promega). For each variant of myc-Gα 12 QL , the Gaussian intensity of the~45 kDa band from the precipitated material and the corresponding band from the starting material were quantified using a Kodak Gel Logic 100 system equipped with Molecular Imaging 5.X software (Carestream Health, New Haven CT).

Reporter gene assays
HEK293 cells grown in 12-well plates were transfected with 0.2 μg SRE-luciferase plasmid (encoding firefly luciferase) and 0.02 μg pRL-TK plasmid encoding Renilla luciferase, plus plasmids encoding variants of myc-Gα 12 QL . Reporter assays for SRE-mediated transcriptional activation were performed as described previously [31]. Briefly, cells were washed with phosphate-buffered saline and lysed in 1X passive lysis buffer (Promega), and lysates were analyzed using a Dual-luciferase assay system and GloMax 20/20 luminometer (Promega). Light output due to firefly luciferase activity was divided by output from Renilla luciferase activity to normalize samples for transfection efficiency.

Trypsin protection experiments
HEK293 cells grown in 10-cm dishes were transfected with various Gα 12 constructs using Lipofectamine 2000 (Invitrogen), and tryptic digestions were performed as a modification of the procedure of Kozasa and Gilman [25]. Briefly, cells were lysed in 50 mM Hepes pH 8.0, 1 mM EDTA, 3 mM dithiothreitol, 1% polyoxyethylene-10-lauryl ether containing the same protease inhibitors as NAAIRS Lysis buffer (see above) but at two-fold lower concentration. Samples were cleared by centrifugation at 70,000 g for 1 h, and supernatants were diluted 20-fold in volume using 50 mM Hepes pH 8.0, 1 mM EDTA, 3 mM dithiothreitol, 10 mM MgSO 4 . Samples were digested with 10 μg/ml TPCK-treated trypsin (New England Biolabs) for 20 min at 30°C, and proteolysis was terminated by addition of 100 μg/ml lima bean trypsin inhibitor (Worthington, Lakewood NJ). Samples were analyzed by SDS-PAGE and immunoblotting using J169 antisera specific to the Gα 12 C-terminus, provided by Tohru Kozasa (Univ. of Ill., Chicago).