The Nucleotide Exchange Factor Ric-8A Is a Chaperone for the Conformationally Dynamic Nucleotide-Free State of Gαi1

Heterotrimeric G protein α subunits are activated upon exchange of GDP for GTP at the nucleotide binding site of Gα, catalyzed by guanine nucleotide exchange factors (GEFs). In addition to transmembrane G protein-coupled receptors (GPCRs), which act on G protein heterotrimers, members of the family cytosolic proteins typified by mammalian Ric-8A are GEFs for Gi/q/12/13-class Gα subunits. Ric-8A binds to Gα•GDP, resulting in the release of GDP. The Ric-8A complex with nucleotide-free Gαi1 is stable, but dissociates upon binding of GTP to Gαi1. To gain insight into the mechanism of Ric-8A-catalyzed GDP release from Gαi1, experiments were conducted to characterize the physical state of nucleotide-free Gαi1 (hereafter referred to as Gαi1[ ]) in solution, both as a monomeric species, and in the complex with Ric-8A. We found that Ric-8A-bound, nucleotide-free Gαi1 is more accessible to trypsinolysis than Gαi1•GDP, but less so than Gαi1[ ] alone. The TROSY-HSQC spectrum of [15N]Gαi1[ ] bound to Ric-8A shows considerable loss of peak intensity relative to that of [15N]Gαi1•GDP. Hydrogen-deuterium exchange in Gαi1[ ] bound to Ric-8A is 1.5-fold more extensive than in Gαi1•GDP. Differential scanning calorimetry shows that both Ric-8A and Gαi1•GDP undergo cooperative, irreversible unfolding transitions at 47° and 52°, respectively, while nucleotide-free Gαi1 shows a broad, weak transition near 35°. The unfolding transition for Ric-8A:Gαi1[ ] is complex, with a broad transition that peaks at 50°, suggesting that both Ric-8A and Gαi1[ ] are stabilized within the complex, relative to their respective free states. The C-terminus of Gαi1 is shown to be a critical binding element for Ric-8A, as is also the case for GPCRs, suggesting that the two types of GEF might promote nucleotide exchange by similar mechanisms, by acting as chaperones for the unstable and dynamic nucleotide-free state of Gα.


Introduction
As members of the Ras superfamily of regulatory GTP binding proteins, heterotrimeric G protein a subunits (Ga) undergo cycles of activation and deactivation driven by binding and hydrolysis of GTP [1]. Conversion to the basal, inactive state results from the intrinsic GTP hydrolyase activity of the G protein. Reactivation is achieved by replacement of GDP by GTP at the nucleotide binding site, catalyzed by guanine nucleotide exchange factors (GEFs). Although the structural events that accompany GEFcatalyzed nucleotide exchange on small, Ras-like G proteins are relatively well understood [2], the mechanism of heterotrimeric G protein activation remains enigmatic. Agonist-activated, transmembrane G protein-coupled receptors (GPCRs) [3] are the best characterized heterotrimeric G protein GEFs. GPCRs act on plasma membrane-localized G protein heterotrimers that consist of GDP-bound Ga tightly associated with heterodimers of Gb and Gc subunits. Recently, members of a family of predominantly cytosolic proteins, typified by mammalian Ric-8A, were identified as non-receptor GEFs that catalyze nucleotide exchange directly on Ga subunits of the Gi/o/q/12/13 families [4]. Across phylogeny, Ric-8A paralogs act in GPCR-independent pathways to orient mitotic spindles in asymmetric cell division, as demonstrated in (C. elegans [5,6], Drosophila [7], and mammalian cells [8]. Ric-8A is a soluble 59.7 kDa protein predicted to adopt a superhelical structure composed of a-helical armadillo repeats [9]. In contrast to GPCRs, Ric-8A catalyzes the release of GDP directly on Ga subunits, but has markedly weak affinity for Ga bound to GTP or non-hydrolyzable GTP analogs [4]. Upon binding to Gai1NGDP, Ric-8A catalyzes GDP release and forms a stable nucleotide-free Ric-8A:Gai1[ ] complex (empty brackets: ''[ ]'', denote absence of bound nucleotide). In the presence of GTP, the complex dissociates to yield free Ric-8A and Gai1NGTP [4].
Using limited proteolysis, circular dichroism (CD) spectroscopy, heteronuclear NMR spectroscopy, hydrogen-deuterium exchange mass spectrometry (HD-MS), and differential scanning calorimetry (DSC), we have found that Ric-8A stabilizes Gai1 in a conformationally dynamic and heterogeneous state which, we propose, facilitates GDP release and subsequent GTP binding. We show that the C-terminus of Gai1 is a critical binding element for Ric-8A recognition and activity, as is also the case for GPCRs [10,11], suggesting that the two GEFs may act by convergent mechanisms.

Results
The smallest fragment of Ric-8A with full GEF activity encompasses most of the protein We conducted limited trypsin proteolysis, together with mass spectroscopic and secondary structural analysis, to define a minimal fragment of Ric-8A that retained the activity of the fulllength protein ( Fig. 1A-C). The fragment encompassing residues 1-492 (Ric-8ADC492) exceeded full-length Ric-8A in GEF activity, whereas C-terminal truncations of successive predicted helical regions (DC426, DC453) or truncation of the N-terminus (DN12, DN38) in the background of DC492 retained GDP release activity that was uncoupled from GTPcS binding stimulatory activity (Fig. 1D,E). Truncated proteins (DC402, DC374) bound Gai1NGDP weakly (data not shown) but had no nucleotide release or GEF activity. Because it is both more abundantly expressed in Escherischia coli and appears to biochemically more stable as well as more active than the full-length protein, we chose to conduct subsequent experiments with Ric-8ADC492. To enhance the sensitivity of tryptophan fluorescence assays of GEF activity, we used a non-myristoylated Gai1 mutant in which Trp 258 was substituted with alanine. The W258A mutation did not impair GTP binding, GTPase activity, or susceptibility to the GEF activity of Ric-8A (Fig. 2) [12]. For brevity, we refer to DC492Ric-8A and W258AGai1 as Ric-8A and Gai1, respectively.
Relative to Gai1NGDP, nucleotide-free Gai1 is more accessible to protease digestion, and deficient in secondary structure Trypsinolysis experiments demonstrated that Gai1[ ] was substantially more protease-sensitive than Gai1NGDP (Figures 3A and 3B) and was more rapidly degraded into ,20 kDa fragments. The distribution of cleavage products is different in the free and GDP-bound states. Normalized as mean residue elipticity, the CD spectrum of Gai1[ ] showed an overall reduction of regular secondary structure relative to Gai1NGDP (Figure 4). These results accord with earlier findings that Gai1[ ] is converted into a misfolded species with low affinity for guanine nucleotides [13]. Gai1[ ] bound to Ric-8A was more resistant to trypsinolysis than free Gai1[ ], as indicated by the persistence of fragments labeled 2 though 4 at the 10 minute time point in Figure 3E. Note, for example, that band 1, visible at the 5 minute time point of Gai1[ ], is degraded after 10 minutes of protease exposure ( Figure 3B). The same fragment persisted after 10 minutes in the complex with Ric-8A ( Figure 3E, band 3). Fragments 2-4 encompass the N-terminal residues of the Ras domain beyond the P-loop, together with most or all of the helical domain of Gai1 [14]. Nevertheless, Ric-8Abound Gai1[ ] was still more susceptible to proteolysis than Gai1NGDP ( Figure 3A). In contrast, if bound to Gai1, Ric-8A was more sensitive to protease digestion than free Ric-8A ( Figures 3D   and 3E). After 10 minutes of protease digestion of Ric-8A:Gai1[ ], all Ric-8A fragments with molecular weights greater than ,24 KDa were degraded, yet several fragments of greater length remained intact after free Ric-8A was exposed to trypsin for the same duration. For both free and Gai1-bound Ric-8A, residues 141-348 appears to constitute a relatively protease-resistant protein core (bands 3 and 1 in Figure 3D and 3E, respectively; for reference to the predicted secondary structure of Ric-8A see Figure 1C). Note that no protease inhibitors were present in the trypsin preparation used to generate the data shown in Figure 3, so the extent of Ric-8A degradation is greater than that shown in Figure 1A.
The mass-normalized CD spectra of Ric-8A and Ric-8A:Gai1 [ ] show similar degrees of secondary structure formation. Therefore, we infer that Ric-8A-bound Gai1[ ] possesses higher secondary structure content than free Gai1[ ]. Both spectra are indicative of predominantly a-helical structure, whereas Gai1NGDP shows evidence of both a-helical and a-sheet structure, which is characteristic of the Ras-like domain of this and other G proteins [1] (Figure 4). The near absence of b-sheet structure estimated from the CD spectrum of Ric-8A:Gai1[ ] suggests that changes in secondary structure may occur in the a/b Ras-like domain of Gai1 upon binding to Ric-8A and subsequent release of GDP.
Peaks in the 15 N-1 H HSQC spectrum of Gai1 are severely attenuated upon binding to Ric-8A To elucidate the structural properties of Gai1 bound either to nucleotides or to Ric-8A, we acquired 1 H-15 N Transverse Relaxation Optimized (TROSY) Heteronuclear Single Quantum Coherence (HSQC) spectra [15] of [ 15 N]Gai1. The 1 H-15 N TROSY-HSQC spectrum of Gai1NGDP ( Figure 5A) and Gai1NGTPcS (data not shown) showed ,300 moderately well resolved and dispersed peaks, comparable in quality to spectra reported by Abdulaev, et al. [16] for a GDP-bound chimera (Gat/ i) of transducin a (Gat) and Gai1. In contrast, the spectrum of  ; cylinders indicate helical segments predicted using JPRED [51]. Residue codes colored red indicate sites of proteolytic cleavage (see panel A). Residue codes in green indicate N or C-termini of recombinant Ric-8A fragments engineered to coincide approximately with proteolytic sites or predicted secondary structure boundaries: DC492 denotes the Ric-8A fragment comprising residues 1-492. Both N-terminal truncations DN12 and the exchange reaction was quenched with formic acid/acetonitrile at successive time intervals, and the products analyzed by electrospray mass spectrometry (ES-MS). For each time-point, the mass distribution of Gai1 was determined by deconvolution of the raw m/z spectrum ( Figure 6A,B). The mass distribution of Gai1NGDP remained unimodal throughout the 60 minute exchange period ( Figure 6A), whereas that for Gai1[ ] bound to Ric-8A evolved into a multimodal distribution, suggestive of conformational heterogeneity ( Figure 6B). Analysis of these data revealed a nearly four-fold greater initial rate of deuterium exchange in Ric-8A-bound Gai1[ ] than in Gai1NGDP ( Figure 6C). After 60 minutes of exposure to D 2 O, the mass of Gai1[ ] in the complex with Ric-8A increased by ,340 Da, accounting for more than half of all the exchangeable Gai1 protons, versus a ,210 Da mass increase in Gai1NGDP alone. The enhanced rate and extent of deuterium substitution is indicative of greater solvent accessibility at exchangeable sites in Ric-8A-bound Gai1 than in Gai1NGDP, most likely due to amplified breathing motions in the Ric-8A:Gai1[ ] complex [18].
Thermodynamic stability of both nucleotide-free Gai1 and Ric-8A increase upon complex formation We used differential scanning calorimetry (DSC), by which change in heat capacity (Cp) is measured as a function of DN38 were also C-terminally truncated at residue 492 and comprised residues 12-492 and 38-492, respectively. (D) Kinetics of intrinsic (open symbols) or Ric-8A-stimulated (filled symbols) GDP release (squares) from, or GTPc binding to (circles) myristoylated Gai1 were determined by a filter binding assay using radiolabeled nucleotides as described [4].  Intrinsic and Ric-8A-catalyzed kinetics of binding of GTPcS to wild-type Gai1, W258A-Gai1, ND25Gai1, GaiCD9 and Gai1-GasC12 were measured using a fluorescence binding assay [12,47]. 400 ml of protein (1 mM) in the GDP bound form was equilibrated for 10-15 min at 25uC in a cuvette. A 10-fold excess of GTPcS was added and fluorescence at 340 nm upon excitation at 290 nm was monitored in the absence (open bars) or presence (filled bars) of Ric-8A (1 mM). Error bars represent +/2 one standard deviation apparent first-order rate constants determined in three replicates. doi:10.1371/journal.pone.0023197.g002  [19]. At temperatures below and above the thermal unfolding transition of a protein, the Cp exhibits a linear, typically positive, dependence on the temperature of the native and unfolded states, respectively. In the region of the thermal transition, Cp exceeds that of both the denatured and native states as hydrophobic groups are increasingly exposed to the aqueous solvent, and reaches a maximum value at T m [20,21,22].
Gai1NGDP underwent an irreversible cooperative unfolding transition with T m = 52uC (Figure 7, blue trace). The irreversible nature of the transition of this and the other proteins and protein complexes reported here precludes accurate determination of the enthalpy of unfolding, but allows comparison of the significant thermal features of the four species when measured at equivalent scan rates. Only a weak transition near 33uC was observed for Gai1[ ] (Figure 7, dashed blue trace), which exhibited changes in Cp that were close to the detection limits of the instrument. The nucleotide-free protein thus appears to be conformationally heterogeneous or disordered [23], consistent with its high protease sensitivity ( Figure 3B The C-terminus of Gai1 is a specific and critical recognition element for Ric-8A binding and GEF activity The mechanism by which Ric-8A catalyzes nucleotide exchange is similar in some respects to the analogous reaction catalyzed by GPCR at Gbc-bound Ga subunits. Experimental evidence . Nucleotide-free Gai1 is relatively unstructured in comparison to Gai1NGDP, but regains helical secondary structure in the complex with Ric-8A. Circular dichroic spectra were normalized as mean residue elipticity, and predicted secondary structure assignments are: Ric-8A, red: 87% a-helix; Ric-8A:  indicates that specific recognition and binding of the C-terminus of Ga is crucial to the action of Ric-8A, just as it is for GPCRs [10,11]. First, a yeast two-hybrid screen of a rat brain library using a bait construct comprising Ric-8A residues 1-297 yielded a prey clone expressing the C-terminal 81 residues of Gai1, that also interacted with a full-length Ric-8A bait construct ( Figure 8A). Second, the peptide Gai1C18, which is composed of a sequence of amino acid residues identical to that of the 18 C-terminal residues of Gai1, inhibited Ric-8A-catalyzed exchange of GTPcS for GDP with an IC 50 of 23 mM ( Figure 8B). Isothermal calorimetric measurements indicated that Gai1C18 binds directly to Ric-8A with a K d of 12 mM ( Figure 8C). Third, Gai1CD9, a Gai1 truncation mutant lacking the nine C-terminal residues of the native protein, fails to serve as a substrate for Ric-8A although it retains GTP binding and hydrolytic activity [24] (Figure 2). Finally, substitution of the C-terminal twelve residues of Gai1 with the corresponding residues of Gas, a Ga protein that does not bind to Ric-8A [4], abrogated susceptibility to the GEF activity of Ric-8A ( Figure 2) but did not impair GTP binding activity. These results are in accord with recent findings that pertussis toxin-  catalyzed ADP ribosylation at the C-terminus of Gai1 [8] and that truncation of the twelve Gai1 C-terminal residues [25] blocks Ric-8A binding and GEF activity. On the other hand, truncation of 25 residues from the N-terminus of Gai1 did not affect its susceptibility to the GEF activity of Ric-8A (Figure 2).

Discussion
The experiments described in this report provide insight into the mechanism of Ric-8A-catalyzed exchange of GDP for GTP on Gai1. In this reaction, Ric-8A:Gai1[ ] is a stable intermediate that does not readily dissociate in the absence of GTP or nonhydrolysable GTP analogs. We have shown that, within this complex, Gai1[ ] adopts a considerably more dynamic conformation than nucleotide-bound Gai1, but is more structured and less susceptible to proteolysis than free Gai1[ ]. This is in sharp contrast to most nucleotide-free complexes of small G proteins with cognate GEFs, in which both the GEF and G protein components are typically well ordered structures [2].
Ric-8A-catalyzed nucleotide exchange proceeds through a stable (in the absence of GTP) but loosely structured intermediate. Examples of enzymes that stabilize proteins in unfolded or disordered states include chaperones such as GroEL [26], and AAA+ ATPase unfoldases of ClpXP proteases and related proteases that degrade mis-folded proteins [27]. However, Ric-8A functions differently from these, in that it is not coupled to an exergonic reaction (e.g. ATPase activity), but does exhibit high substrate specificity and catalyzes a discrete chemical transformation. We propose that the catalytic power of Ric-8A derives, in part, from its ability to act (in rough analogy with GroEL and other unfoldases) as a chaperone for a partially unfolded or disordered conformation Gai1[ ], thereby reducing the activation energy barrier to GDP release and GTP binding, while disfavoring unproductive side reactions that would lead to Gai1 deactivation and aggregation. In the partially unstructured state induced and stablilized by Ric-8A, the nucleotide binding site of Gai1 may be more solvent-accessible than in the nucleotide-bound state.
The mechanism by which Ric-8A catalyzes nucleotide exchange may be similar in some respects to the analogous reaction catalyzed by GPCR at Gbc-bound Ga subunits. Recognition and binding of the Ga C-terminus is crucial to the action of both exchange factors. In a manner analogous to that proposed for GPCRs, Ric-8A could promote nucleotide release by gripping, and perhaps tensioning, the C-terminus of Ga and thereby weaken local tertiary structure (a5 helix, b5 and b6 strands) that is allosterically coupled to the purine binding site and switch regions [11,28,29]. Indeed, the substantial reduction in fluorescence emission of Trp 211 in switch II of Gai1 upon binding to Ric-8A provides evidence for such perturbations [12]. Whether Ric-8A directly engages the switch regions of Ga is uncertain. Such interactions might be precluded since Ric-8A is able to form a transient ternary complex with Gai1NGDP:AGS3 [12]. AGS3, a guanine nucleotide dissociation inhibitor, comprises GPR/Go-Loco motifs [30,31] that partially block the switch I/switch II interface [32]. Similarly, direct interactions between GPCRs and the switch regions of Ga are problematic on stereochemical grounds [33]. It is also noteworthy in this context that inportin-b, a protein involved in the transport of protein cargo into the nucleus, and also a presumptive structural analog of Ric-8A, has been shown to act as a GEF for Ran1NGDP [34]. Crystallographic analysis demonstrates that importin-b induces conformational changes in the switch regions of Ran1NGDP [35].
Both  [37]. Recent evidence obtained from double electron electron resonance (DEER) spectroscopy shows that, in the activated rhodopsin-heterotrimer complex, nucleotide-free Gai1 is conformationally heterogeneous, and that the Ras-like and helical domains of Gai1, between which nucleotide is bound, swing away from each other [38]. The NMR and HD-MS data presented here suggest that nucleotide-free Gai1 bound to Ric-8A undergoes conformational exchange, with interconversion times that are possibly in the ms-ms range. The DSC melting profile of Ric-8A:Gai1[ ] is suggestive of a non-cooperative unfolding transition at lower temperature followed by a discrete transition near 50uC. It seems reasonable to attribute the former to a conformationally heterogeneous and dynamic Gai1[ ] and the latter to Ric-8A, which in the complex with Gai1[ ] is more thermostable yet also more protease accessible than unbound Ric-8A, suggesting that Ric-8A itself may undergo some structural change upon binding to Gai1. However, structural assignment of transitions in the DSC spectra is speculative. It remains to be determined which segments of Gai1 become mobile within the nucleotide free complex with Ric-8A, and importantly, to confirm that induction or stabilization of a partially disordered or conformationally flexible state in Gai1 in fact reduces the kinetic energy barrier to GDP release and GTP binding.

Molecular Cloning and Protein Expression
The open reading frame of rat Ric-8A and truncation variants (encompassing residues 1-492, 12-492, 1-453, 1-426, 1-402, 1-374, 12-492 and 38-492) were amplified by PCR and subcloned into the pET-28a vector for expression as N-terminally hexahistidine tagged proteins. Proteins were expressed in Escherichia coli BL21 (DE3)-RIPL cells in LB media containing ampicillin (120 mg/L) and induced with 300 mM isopropyl b-d-thiogalactopyranoside (IPTG) at 20uC. After overnight growth at 20uC, cells were lysed by sonication at 20uC in lysis buffer (50 mM Tris, pH 8.0, 250 mM NaCl, 2 mM DTT, and 2 mM PMSF. The cell lysate was clarified by centrifugation and loaded onto a column containing 5 ml of nickel NTA-agarose (GE Healthcare). After extensive washing with lysis buffer, proteins were eluted from the resin with buffer (50 mM Tris, pH 8.0, 150 mM NaCl and 2 mM PMSF) containing 250 mM imidazole and dialyzed in a low ionic strength buffer (50 mM Tris, pH 8.0, 2 mM DTT, and 2 mM PMSF). The dialysate was loaded onto a UNO-Q matrix (Bio-Rad) and eluted with a 0-500 mM NaCl gradient on an AKTA FPLC system (GE Healthcare). Pure Ric-8ADC492 eluted from the matrix at 165-175 mM NaCl.
Rat Gai1 was expressed as a tobacco etch virus protease (TEV)cleavable, N-terminal glutathione-S-transferase (GST) fusion protein as described [12]. W258A-Gai1, in which the tryptophan residue at position 258 is substituted by alanine, was generated by use of the QuikChange (Stratagene) kit according to the manufacturer's protocol, using the pDEST-15 vector harboring wild type GST-Gai1 as a template. To generate ND25-Gai1, from which the N-terminal 25 residues of the native protein are deleted, attB-modified primers corresponding to amino acids 25 to 35 and 343 to 353 of Gai1 were used for PCR amplification and cloning of the fragment into the pDEST15 vector. W258-Gai1 and ND25-Gai1 were expressed and purified as described [12].
The plasmid pBN905, which expresses rat Gai1DC9, lacking the C-terminal nine residues of the native protein, fused in-frame to intein-CBD cDNA in the pTXB3 expression vector (New England Biolabs), was a kind gift from Dr. T.J. Baranski, Washington University, St. Louis, MO. Gai1DC9 was expressed and purified as described [39]. With the exception of experiments summarized in Figure 1, 2 and 8A, all other experiments were performed with Ric-8ADC492 and W258A-Gai1, which we henceforth refer to as Ric-8A and Gai1, respectively.

Preparation of 15 N-labeled proteins
[ 15 N]Gai1 was prepared as described with minor modifications [42]. Briefly, transformed E. coli cells were grown in minimal media supplemented with [ 15 N]NH 4 Cl (Cambridge Isotopes, 99.8% purity) and [ 15 N]Bioexpress Cell Growth media (10 ml of 106 concentrate/liter of media) (Cambridge Isotope Labs), induced with 500 mM IPTG at 20uC and allowed to express Gai1 overnight at the same temperature. The purification protocol was identical to that used for native proteins and the yields were approximately one third lower.

Trypsin Protection Assays
To samples containing Ric-8A, Gai1NGDP, Gai1[ ] or Ric-8A:Gai1 (50 mM in 50 mM Tris-HCl, pH 8.0, 150 mM NaCl and 2 mM DTT), L-1-p-tosylamido-2-phenylethyl chloromethyl ketone (TPCK)-treated trypsin (Sigma) was added at a molar ratio of 1000:1. Samples were incubated at 4uC for 5, 10, 15 and 25 minutes. For each time point, a 10 ml aliquot was withdrawn, diluted in SDS-PAGE loading buffer, boiled, separated by SDS-PAGE and visualized by Coomassie staining. Proteolytic products were eluted from the gel slices and subjected to MALDI-TOF mass spectrometry on a Voyager DE3 (Applied Biosystems) or by electrospray mass spectometry using on a Agilent 6520 QTOF. Limited trypsinolysis for mass spectrometric identification of large Ric-8A fragments (Figure 1) was conducted in the presence of a 1:2 molar ratio of bovine pancreatic aprotinin to TPCK-trypsin. Proteolytic products were eluted from the gel slices and subjected to electrospray mass spectrometry and N-terminal sequencing at the Protein Chemistry Core Facility of the University of Texas Southwestern Medical Center.
Pull-down assays for Gai1 binding to Ric-8A fragments Equimolar amounts of Ric-8A and Gai1 (10 mM protein in 50 mM TRIS. HCl, pH 8.0, 150 mM NaCl, 2 mM DTT and 0.05% C12E10) were incubated overnight at 4uC. 10 ml of a 50% slurry of Ni +2 IMAC (BioRad) resin was then added to the mixture, incubated for one hour, washed thrice with 500 ml of wash buffer (50 mM Tris, pH 8.0, 250 mM NaCl, 1 mM DTT and 2 mM PMSF) and the proteins retained on the beads were visualized by Coomassie stained SDS-PAGE.

Peptide synthesis
An amidated peptide, Gai1C18, corresponding to the Cterminal 18 residues of rat Gai1 (DAVTDVIIKNNLKDCGLF) was synthesized using standard FMOC chemistry by the Protein Chemistry Core laboratory at UT Southwestern Medical Center at Dallas and purified to near homogeneity by HPLC (Agilent Technologies) on a pre-packed C18 matrix (Waters). Mass of the peptide was confirmed by MALDI-TOF mass spectrometry (Voyager DE, Applied Biosystems).

Peptide competition assays
Exchange of GTPcS for GDP bound to Gai1 or Gai1CD9 was followed by monitoring the change in the tryptophan fluorescence of Gai1, as described [12]. Gai1NGDP (1 mM) in 20 mM HEPES, pH 8.0, 100 mM NaCl, 10 mM MgCl 2 , 1 mM DTT, and 0.05% C12E10 in a reaction volume of 400 ml was allowed to equilibrate for 10-15 min at 20uC in a quartz fluorescence cuvette. GTPcS (final concentration, 10 mM) was added to the reaction mixture in the absence or presence of 1 mM Ric-8A, and the increase in fluorescence at 340 nm was monitored upon excitation at 290 nm [47]. Exchange kinetics were also measured in the presence of 1 mM Ric-8A and (5-50 mM) Gai1C18. Protein and peptide mixtures were preincubated for one hour before addition of GTPcS. Fluorescence measurements were conducted using an LS55 spectrofluorometer (PerkinElmer Life Sciences) attached to a circulating water bath to maintain a steady sample temperature of 20uC. Excitation and emission slit widths were set at 2.5 nm. All exciting light was eliminated by use of a 290 nm cut-off filter positioned in front of the emission photomultiplier.

Circular Dichroism spectroscopy
Gai1NGDP, Ric-8A, Gai1[ ] or the nucleotide free binary complex of the two proteins at 4 mM each in 25 mM HEPES, pH 7.2, 150 mM NaCl and 2 mM DTT and, in the case of Gai1 [ ], 20% v/v glycerol, were dispensed into a 300-ml quartz cuvette with a 1 mm path length. CD spectra in the range of 195-245 nm were measured at a scan rate of 1 nm/min using a PiStar-180 CD spectrometer (Applied Photophysics). The scans were repeated thrice; the data were averaged and the CD spectra of the buffer was subtracted. The optical path and the cuvette chamber were continually flushed with a nitrogen flow throughout the course of the experiment. Secondary structure analysis was performed using K2D2 [48].

Hydrogen-Deuterium Exchange Mass Spectrometry
Hydrogen-deuterium exchange of the Gai1NGDP or Ric-8A:Gai1[ ] was analyzed by automated reverse-phase HPLC coupled to electrospray ionization TOF mass spectrometry. The HPLC consisted of an Agilent 1100 HPLC with a G1377a autosampler, and the ESI-TOF was a Bruker microTOF. Following initiation of the reaction by ten-fold dilution of protein stock (1 mg/ml Gai1NGDP or Ric-8A:Gai1[ ], in 20 mM sodium phosphate, pH 6.8, 100 mM NaCl and 1 mM DTT) into D 2 O, the reaction mixture was pipetted into a sealed autosampler vial and the autosampler was used to draw aliquots at regular time intervals. Quenching of the exchange reaction was achieved by rapid binding of the protein onto a C4 reverse phase cartridge from Michrom Bioresources (861 mm) and subsequent washing and elution. The column and autosampler were pre-equilibrated with 20% (v/v) acetonitrile, 80% H 2 O and 0.1% formic acid (w/ v), pH 2.2, prior to sample loading. Immediately following sample (0.5 ml) injection, the solvent composition was changed to 100% acetonitrile, 0.1% formic acid. By using a rapid step gradient and very high flow rates of 600 ml/min, the sample was minimally delayed in the flow path to the mass spectrometer, eluting at approximately 0.4 minutes. The column system was equilibrated at 4uC to minimize back-exchange. Data processing was performed with the Bruker Data Analysis software package, version 4.0. The Maximum Entropy devolution routine was used to perform charge-deconvolution for the spectral range of 700 m/ z to 1400 m/z, which encompassed the majority of the observed distribution of protein signal. The deconvoluted spectra were exported to ORGIN software and the centroid masses for Gai1 were calculated and plotted as a function of time.

Differential Scanning Calorimetry (DSC)
For DSC analysis, Gai1NGDP, Ric-8A and Ric-8A:Gai1[ ] were dialyzed against degassed DSC sample buffer: 25 mM PIPES pH 7.2, 150 mM NaCl and 1 mM TCEP, and additionally for the Gai1NGDP sample, 20 mM GDP. DSC buffer for Gai1[ ] contained 20% glycerol (v/v). Immediately before DSC analysis, protein samples were clarified by centrifugation at 14,000 RPM for 10 min in a bench-top Eppendorf microfuge. Protein concentrations after dilution, if required, were determined by least squares fitting of predicted protein extinction coefficients to spectra in the 220-420 nm range measured on a HP diode array instrument. The measured values were 3.6 mM for Gai1NGDP, 5.9 mM for Ric-8A, 5.1 mM for Ric-8:Gai1[ ] and 7.9 mM for Gai1[ ]. DSC measurements were conducted using a Microcal capDSC with autosampler (MicroCal, GE Healthcare). After establishing a thermal history by running water vs. water scans, three buffer against water scans were conducted for each sample using the corresponding dialysate solution to obtain the buffer Cp over the experimental temperature range. Following this, two buffer vs protein scans were performed. Protein samples were rescanned once to check for thermal reversibility.
A typical thermal cycle involved cooling the instrument to 20uC after which a 10 min. thermal equilibration was initiated. Following thermal equilibration, scanning of the sample was performed at a scan rate of 1uC/min over a 20uC-70uC range using passive feedback gain mode and a filtering period of 5 seconds. Once the experimental high temperature limit was reached (70uC), the instrument was cooled back to the starting temperature.
Data analysis was performed using Origin 7.0 by first subtracting the last buffer scan from the protein thermal scan. After normalizing the data to the protein concentration, a progressive baseline estimation was performed by calculating the fractional contribution of the native and denatured state to the sample Cp at each point beneath the excess heat capacity function, thus producing a smoothly varying function of temperature [50]. Data presented in Figure 7 were corrected by subtraction of the temperature-dependent change in Cp of the buffer. A weighted average thermal profile for Gai1[ ] and Ric-8A was computed using the expression Cp Av (T) = w Ric-8A Cp Ric-8A (T)+w Gai1