A New Approach to Producing Functional Gα Subunits Yields the Activated and Deactivated Structures of Gα_12/13 Proteins

Summary

The oncogenic G_12/13 subfamily of heterotrimeric G proteins transduces extracellular signals that regulate the actin cytoskeleton, cell cycle progression, and gene transcription. Previously, structural analyses of fully-functional Gα_12/13 subunits have been hindered by insufficient amounts of homogeneous, functional protein. Herein we report that substitution of the N-terminal helix of Gα_i1 for the corresponding region of Gα₁₂ or Gα₁₃ generated soluble chimeric subunits (Gα_i/12 and Gα_i/13) that could be purified in sufficient amounts for crystallographic studies. Each chimera bound guanine nucleotides, Gβγ subunits and effector proteins, and exhibited GAP responses to p115RhoGEF and leukemia-associated RhoGEF. Like their wild-type counterparts, Gα_i/13, but not Gα_i/12, stimulated the activity of p115RhoGEF. Crystal structures of the Gα_i/12·GDP·AlF₄⁻ and Gα_i/13·GDP complexes were determined using diffraction data extending to 2.9 and 2.0 Å, respectively. These structures reveal not only the native structural features of Gα₁₂ and Gα₁₃ subunits, which are expected to be important for their interactions with GPCRs and effectors such as Gα-regulated RhoGEFs, but also novel conformational changes that are likely coupled to GTP hydrolysis in the Gα_12/13 class of heterotrimeric G proteins.

Heterotrimeric GTP-binding proteins (G proteins) receive inputs from G protein-coupled receptors (GPCRs) at the cell surface and elicit a wide variety of responses within the cell (1). Activation of GPCRs by extracellular stimuli induces the Gα subunit to release its bound GDP and bind GTP, and facilitates dissociation of Gα·GTP from the β and γ subunits (Gβγ). Both Gα·GTP and Gβγ subsequently bind to and regulate the activity of various effector molecules. A family known as regulator of G protein signaling (RGS) proteins can bind to activated Gα subunits and accelerate their rate of deactivation, thereby serving as GTPase-activating proteins (GAPs) (2). It was shown through biochemical and structural studies of RGS4 that its core helical domain, referred to herein as the RGS homology (RH) domain¹, binds to all three switch regions of Gα and thereby stabilizes the transition state for GTP hydrolysis (3, 4).

Among the four subfamilies of Gα subunits (5), the members of the Gα_12/13 subfamily are distinct in that they are potently oncogenic in cell-based assays (6, 7). Stimulation of Gα_12/13-coupled receptors, such as those for thrombin or lysophosphatidic acid, transforms cells in a manner implicating the activation of Rho family GTPases (8, 9). Downstream effector targets of Gα_12/13 subunits are diverse and include cadherin (10), protein phosphatase 5 (11) and many others (12). However, the best-characterized targets of Gα_12/13 subunits are a family of Rho guanine nucleotide exchange factors (RhoGEFs), including p115RhoGEF, leukemia-associated RhoGEF (LARG), and PDZ-RhoGEF (13–15). Each of these RhoGEFs contains an RH domain (also known as the rgRGS or LH domain)¹ that binds specifically to activated Gα_12/13 subunits (16). Unlike RGS proteins, the RhoGEF RH domain binds to the effector-binding site of the Gα subunit rather than the site traditionally used by RGS proteins (17). A small, acidic domain N-terminal to the RH domain, which is essential for GAP activity but not for the binding of Gα₁₃ (18, 19), occupies the binding site traditionally used by RGS proteins.

Biochemical and crystallographic characterization of Gα_12/13 subunits and their effector complexes requires large amounts of homogeneous, fully-functional Gα₁₂ and Gα₁₃ subunits, which previously have been produced with exceptionally low yields (20, 21). Although Chen et al. were able to express a “Gα_13/i-5” chimera, in which the switch regions and α-helical domain of Gα₁₃ were swapped for those of Gα_i1, the resulting chimera exhibited approximately a 10-fold higher nucleotide exchange rate and 5% of the GAP activity of wild-type Gα₁₃ in response to p115RhoGEF. Even so, it could still be crystallized as a complex with an N-terminal fragment of p115RhoGEF that included its RH domain (17).

Herein we show that substitution of the N-terminal helix of Gα_i1 for the corresponding region of Gα₁₂ and Gα₁₃ generated soluble Gα_i/12 and Gα_i/13 chimera that retained all wild-type biochemical properties while also being expressed at much greater levels than their wild-type counterparts. By determining the crystal structures of the Gα_i/12·GDP·AlF₄⁻ (activated) and Gα_i/13·GDP (deactivated) complexes, we not only show that these proteins are suitable for structural studies, but also provide the first look at the native atomic structures of these oncogenic G protein subunits and a novel conformational change upon deactivation of these subunits that we propose is specific to the Gα_12/13 subfamily. Differences in the structures of Gα₁₂ and Gα₁₃ undoubtedly contribute to their differential ability to couple with receptors and to regulate downstream effectors.

Experimental Procedures

Expression constructs

The Gα_i/13 chimera was created by ligating the N-terminal coding region of rat Gα_i1 (residues 1–28) to Gα₁₃ (residues 47–377) and subcloning the product into the baculovirus transfer vector pFastBacHT_A (Invitrogen, Carlsbad, CA). A construct of Gα_i/12 was created using the same strategy. Each construct therefore includes an N-terminal hexahistidine (His₆) tag followed by a tobacco etch virus (TEV) protease recognition site. The expression construct for glutathione-S-transferase (GST)-p115-RH or GST-LARG-RH was generated by subcloning the coding region of p115RhoGEF (residues 1 to 252) or LARG (residues 319 to 598) into pGEX-KG, respectively.

Production of chimeric Gα subunits

All purification steps were conducted at 4°C using ice-cold buffers supplemented with protease inhibitors. Cells from 4 liters of Sf9 culture were harvested 48 h after infection with 15 ml/L of either Gα_i/13 or Gα_i/12 amplified baculovirus stock, resuspended in 400 ml of Lysis Buffer (20 mM HEPES pH 8.0, 0.1 mM EDTA, 10 mM 2-mercaptoethanol (βME), 1 mM MgCl₂, 100 mM NaCl, and 10 µM GDP), and then lysed by nitrogen cavitation. Lysates were centrifuged at 100,000 × g for 30 min, after which the supernatants were diluted with 1600 ml of Buffer A (20 mM HEPES pH 8.0, 10 mM βME, 1 mM MgCl₂, 100 mM NaCl, 10 µM GDP, and 12.5 mM imidazole, pH 8.0) and loaded onto a Ni-NTA agarose (Qiagen, Valencia, CA) column equilibrated with Buffer A. The column was washed with 20 volumes of Buffer B (Buffer A containing 0.4 M NaCl and 20 mM imidazole, pH 8.0), and the chimera was eluted in 10 fractions of 1 volume of Buffer C (Buffer A containing 150 mM imidazole, pH 8.0). Peak fractions were supplemented with 10% glycerol, and exchanged to Buffer D (20 mM HEPES pH 8.0, 10 mM βME, 1 mM MgCl₂, 100 mM NaCl, 10 µM GDP, 10% glycerol) with Centricon Plus-20 PL-30 (Millipore, Billerica, MA).

Purification of RH proteins

GST-p115-RH was produced in JM109 harboring pGEX-KG/p115-RH and pT-Trx plasmids (22) for 3 hr after induction with 0.1 mM IPTG at 30°C. GST-LARG-RH was similarly expressed in BL21(DE3) CodonPlus RP (Stratagene, La Jolla, CA) harboring the pGEX-KG/LARG-RH plasmid. GST-RH fusions were purified by glutathione Sepharose 4B resin (Amersham Biosciences).

Interaction of chimeric Gα subunits with RH proteins

Gα_i/13 or Gα_i/12 (2 µg) was diluted in Buffer E (20 mM Tris-HCl, pH 8.0, 1 mM EDTA, 10 mM βME, 300 mM NaCl, 10 mM MgCl₂, 30 µM GDP, 0.1% C₁₂E₁₀), either in the absence or presence of AlF₄⁻ (30 µM AlCl₃ and 10 mM NaF), and mixed with 33 µg of GST-p115-RH or 38 µg of GST-LARG-RH in a final volume of 500 µl. After incubation on ice for 20 min, glutathione Sepharose 4B (30 µl resin) was added and samples were incubated with constant mixing for 1 h at 4°C. Beads were pelleted by centrifugation (600 × g, 1 min, 4°C) and washed four times with 400 µl of Buffer E (with or without AlF₄⁻). Beads were mixed with 20 mM reduced glutathione and SDS-PAGE sample buffer and boiled for 10 min. Immunoblotting was performed with anti-Gα₁₃ B860 (21) or anti-Gα₁₂ J168 (20) antibody.

Rho nucleotide exchange assays

His₆-RhoA was expressed in Sf9 cells and purified as described (23). Gα₁₃, Gα_i/13, or Gα_i/12 (5 pmol) was first incubated in the presence of AMF (60 µM AlCl₃, 5 mM MgCl₂, and 20 mM NaF) for 15 min on ice, then incubated with His₆-RhoA (25 pmol) and p115RhoGEF (0.25 pmol) in binding buffer (50 mM Tris-HCl, pH 7.5, 1 mM DTT, 0.5 mM EDTA, 50 mM NaCl, 5 mM MgCl₂, 0.05% C₁₂E₁₀, and 10 µM GTPγS with ~500 cpm/pmol [³⁵S]-GTPγS) in a final reaction volume of 50 µl. Reactions were terminated after incubation for 5 min at 30°C by addition of wash buffer, and GTPγS binding to RhoA was determined by filter binding as described (24).

Crystallization of Gα_i/12·GDP·AlF₄⁻ and Gα_i/13·GDP

Gα_i/13 or Gα_i/12 was first incubated with 1.5% (w/w) recombinant TEV protease for 2 hr at 25°C to remove the His₆-tag, and then activated by AlF₄⁻ (20 µM AlCl₃ and 10 mM NaF) for 15 min on ice. This mixture was loaded onto tandem Superdex 200 10/30 gel filtration columns (Amersham Biosciences, Piscataway, NJ) equilibrated in gel filtration buffer (20 mM HEPES pH 8.0, 1 mM EDTA, 2 mM DTT, 5 mM MgCl₂, 50 mM NaCl, 10 µM GDP, 20 µM AlCl₃, and 10 mM NaF). Both Gα_i/13 and Gα_i/12 eluted from the column as apparent monomers. Peak fractions were pooled and concentrated using a Centricon YM-30 (Millipore).

Gα_i/12 was crystallized by mixing 1 µl of 22 mg/ml Gα_i/12 with 1 µl well solution containing 100 mM sodium citrate pH 6.5, 50 mM NaCl, and 14% PEG 8000, and then suspending the drop over 1 ml of well solution at 4°C. Paper-thin, diamond shaped crystals appeared within one week. Gα_i/13 was crystallized similarly, except using 19 mg/ml Gα_i/13 and a well solution containing 100 mM sodium citrate (pH 4.8), 50 mM NaCl, and 10% PEG 2000. Tetragonal rod-shaped crystals appeared in 7 to 20 days. Both crystals were stabilized in a harvesting solution containing all the components of both the gel filtration buffer and their respective well solution, and were serially transferred, in 5% steps at a time, into a final concentration of 20% glycerol in harvesting solution. Crystals were then flash frozen in liquid nitrogen on nylon loops.

X-ray analysis and structure determination

Diffraction maxima were measured from crystals maintained at 100 K at the Advanced Photon Source beam line 17-ID using a Quantum 210 CCD detector. Gα_i/12 crystals exhibited severely anisotropic diffraction with disorder in the maxima that could have been due to warping of the extremely thin crystals. These defects, along with lower resolution diffraction limits, are probably responsible for the relatively high R-factors in the resulting model of Gα_i/12 compared to the Gα_i/13 structure (Table I). Data were reduced using HKL2000 (25). The space group choice and phase problem for Gα_i/13 was solved via molecular replacement with the program PHASER (26) using Gα_i1·GDP·AlF₄⁻ (PDB code: 1AGR) as the search model. The Gα_i/12 structure was determined using the partially refined Gα_i/13 structure as the search model. Each model was refined with several rounds of simulated annealing in CNS_SOLVE (27) and then restrained-refinement in REFMAC5 (28). Two-fold non-crystallographic symmetry restraints were used during refinement of Gα_i/12. Model building was performed with O (29). All reflections were used in the last round of refinement with final combined R-factors of 23.7% and 21.1% for the Gα_i/12·GDP·AlF₄⁻ and Gα_i/13·GDP structures, respectively. Coordinates and structure factors for the Gα_i/12 and Gα_i/13 structures are deposited at the PDB with accession codes 1ZCA and 1ZCB, respectively.

Table I.

Crystallographic data and refinement statistics

X-ray Source: APS 17-ID	Gαi/12·GDP·AlF4−	Gαi/13·GDP
Wavelength (Å)	1.000	1.000
Dmin (Å)	2.9a	2.0
Space group	P21	P43212
Cell constants (Å, °)	a = 57.5, b = 85.2, c = 82.9; β = 106.1, α = γ = 90	a = b = 67.4, c = 175.2; α = β = γ = 90
Unique reflections	15263 (1140)c	27297 (2761)c
Average redundancy	2.5 (1.7)	5.7 (4.0)
Rsym (%)b	10.2 (29.6)	4.9 (38.5)
Completeness (%)	88.7 (67.4)	96.7 (100)
<I>/<σI>	6.9 (1.7)	42.6 (3.4)
Refinement resolution (Å)	29.1 – 2.9	28.5 – 2.0
Total reflections used	14,452	25,848
Protein atoms	5239	2618
Non-protein atoms	74	80
RMSD bond lengths (Å)	0.007	0.015
RMSD bond angles (°)	1.0	1.5
Rworkd	23.9 (35.1)e	20.7 (24.4)e
Rfreef	29.6 (44.3)	25.2 (29.4)

^aDiffraction from this crystal form was anisotropic, with maxima extending beyond 2.9 Å in the b* direction, and from 3.5–4.5Å in orthogonal directions.

^bR_sym = ∑_h∑_i |I(h)_i - I(h)|/ ∑_h∑_i I(h)_i, where I(h) is the mean intensity of i reflections after rejections. A −0.5 I/σ_I cutoff was applied to the Gα_i/12 data set due to its anisotropic data.

^cNumbers in parentheses correspond to the highest resolution shell of data for each set, which was 3.0 – 2.9 Å for Gα_i/12 and 2.07 – 2.00 Å for Gα_i/13.

^dR_work = ∑_hkl‖F_obs(hkl)| -|F_calc(hkl)‖/ ∑_hkl |F_obs(hkl)|; no I/σ cutoff was used during refinement.

^eNumbers in parentheses correspond to the highest resolution shell of data for each set, which was 3.0 – 2.9 Å for Gα_i/12 and 2.05 – 2.00 Å for Gα_i/13.

^f5% of the truncated data set was excluded from refinement to calculate R_free.

Miscellaneous procedures

GTPγS binding to Gα_i/13 or Gα₁₃ was measured by filter binding assays as described previously (23). Single-turnover GTP hydrolysis by Gα_i/13 or Gα_i/12 was analyzed in the absence (basal) or presence of 100 nM GST-p115-RH or GST-LARG-RH, as described previously for Gα₁₃ or Gα₁₂ (16). EE-tagged, full-length p115RhoGEF and Gα₁₃ were purified as described (24). Atomic representations were created with the program PYMOL (30).

Results and Discussion

Soluble expression of functional Gα_i/12 or Gα_i/13 in insect cells

To increase yields of functional Gα₁₂ or Gα₁₃, we constructed the Gα_i/12 and Gα_i/13 chimera, in which the N-terminal helices of mouse Gα₁₂ or Gα₁₃ were swapped with that of rat Gα_i1 (Fig. 1A and Figure S1). The Gα_i/13 chimera partitioned almost equally to the soluble and crude membrane fractions of Sf9 cells (data not shown). The soluble Gα_i/13 exhibited activation-dependent trypsin protection indicative of a functional Gα subunit (Fig. S2). Similar results were also obtained with Gα_i/12 (data not shown). The yield of Gα_i/13 was about 3.5 mg from 1 liter of Sf9 cell culture after Ni-NTA agarose chromatography (Fig. 1B), representing a vast improvement from previous methods (31). Similarly, the yield of Gα_i/12 was about 4.0 mg from 1 liter of culture (Fig. 1B). The expression of analogous chimeric proteins in E. coli resulted in large yields of soluble protein, but these proteins could not bind guanine nucleotides (data not shown).

Figure 1.

Generation of Gα_i/13 and Gα_i/12 chimeric proteins. (A) Schematic representation of the Gα_i/13 and Gα_i/12 chimera. Residues of Gα_i1 are shown in white, and numbers within the colored regions of the Gα_i/13 or Gα_i/12 chimera correspond to original numbering of residues contributed by Gα₁₃ (black) or Gα₁₂ (gray). (B) Purified His₆-Gα_i/13 or His₆-Gα_i/12 (3 µg) were subjected to SDS-PAGE and stained with Coomassie Brilliant Blue. The position and apparent molecular weight of standard proteins (in kDa) is indicated. (C) GTPγS binding to Gα_i/13 (●) or wild-type Gα₁₃ (○) was measured at 30°C. Data are the mean of duplicate determinations. (D) Gα_i/13 or Gα_i/12 functionally interact with Gβγ. GTPγS binding to Gα_i/13 or Gα_i/12 (3 µg each) was measured either in the absence or presence of Gβγ (20 µg) as indicated. GTPγS binding was quantified after 90 min at 30°C. Data are the mean of duplicate determinations.

Gα_i/12 and Gα_i/13 subunits bind guanine nucleotides and Gβγ

Gα_12/13 subunits possess biochemical properties that are distinct from those of most other Gα subunits. These include rates of nucleotide exchange (~0.01 min⁻¹) and GTP hydrolysis (~0.2 min⁻¹) that are roughly 3–10 fold slower than those of Gα_s and Gα_i subfamilies (20, 21, 32). The activity of purified Gα_i/13 was evaluated by GTPγS binding assays. GTPγS binding to Gα_i/13 was effectively reduced in the presence of AlF₄⁻, which was similar to control samples containing purified Gα_i1 (Fig. S2). Furthermore, purified Gα_i/13 bound GTPγS with similar kinetics as wild-type Gα₁₃ (Fig. 1C). Gα_i/12 also bound GTPγS similarly to wild-type Gα₁₂ (data not shown). Because the N-terminal helix of Gα subunits is known to interact with Gβγ (33–35), we tested whether its substitution affected association of Gα_i/12 or Gα_i/13 with Gβγ, as measured by the reduced binding of GTPγS to Gα in the presence of Gβγ (36). Assays performed in the presence of excess Gβγ showed marked inhibition of GTPγS binding to both chimeric proteins (Fig. 1D), indicating that Gα_i/12 and Gα_i/13 retain the ability to interact with Gβγ.

RhoGEFs exhibit GAP activity for Gα_i/12 or Gα_i/13

Gα₁₂ and Gα₁₃ interact with the RH domain of p115RhoGEF or LARG in an AlF₄⁻-dependent manner (15, 16). As shown in Figure 2, Gα_i/12 and Gα_i/13 also interacted with RH domain-containing fragments of p115RhoGEF or LARG in an AlF₄⁻-dependent manner, as assessed by GST pull-down assays. We also could isolate high affinity complexes of each chimeric G protein with the RH domain of LARG by size exclusion chromatography (data not shown). We next examined whether these same p115RhoGEF or LARG fragments exhibit GAP activity for the chimera. Both fragments accelerated the GTPase activity of Gα_i/12 and Gα_i/13 in single-turnover GTPase assays (Fig. 3A), with a faster response in Gα_i/13 than in Gα_i/12, as previously observed for their wild-type counterparts (15, 16).

Figure 2.

Gα_i/13 or Gα_i/12 chimera interact with RH domains in an activation-dependent manner. GST-p115-RH or GST-LARG-RH were incubated with purified His₆-Gα_i/13 (upper panels) or His₆-Gα_i/12 (lower panels), either in the absence or presence of AlF₄⁻, and then GST-RH was pulled down using glutathione Sepharose 4B resin. Protein eluted from washed beads was resolved by SDS-PAGE and either immunoblotted (IB) with anti-Gα₁₃ (B860) or anti-Gα₁₂ (J168) antibody, or stained by Coomassie Brilliant Blue (CBB).

Figure 3.

Activity of purified Gα_i/13 and Gα_i/12 in reconstitution assays. (A) Single-turnover hydrolysis of GTP bound to Gα_i/13 (upper panel) or Gα_i/12 (lower panel) was measured at 15°C in the absence (□) or presence of 100 nM GST-p115-RH (●) or 100 nM GST-LARG-RH (▲). Data are from one experiment, which is representative of three experiments. (B) Stimulation of p115RhoGEF activity by Gα_i/13 but not Gα_i/12. GTPγS binding to His₆-RhoA was measured after incubation for 5 min at 30°C in the presence of p115RhoGEF (5 nM) and indicated AlF₄⁻-activated Gα subunit (100 nM). Data are the mean of duplicate determinations from one experiment, which is representative of three experiments. (C) Dose-dependent stimulation of p115RhoGEF activity by Gα_i/13. GTPγS binding to His₆-RhoA by p115RhoGEF (5 nM) was evaluated after 5 min at 30°C, and included the indicated concentration of either AlF₄⁻-activated wild-type Gα₁₃ (○) or Gα_i/13 (●). Samples lacking RhoA but containing p115RhoGEF and either AlF₄⁻-activated wild-type Gα₁₃ (△) or Gα_i/13 (▼) were assayed in parallel.

Regulation of RhoGEF activity by Gα_i/13 or Gα_i/12

Gα₁₃, but not Gα₁₂, directly stimulates p115RhoGEF-mediated nucleotide exchange on RhoA (13). To assess whether our chimera could similarly regulate RhoGEF activity, we measured GTPγS binding to RhoA in the presence of p115RhoGEF and AlF₄⁻-activated Gα_i/12 or Gα_i/13. As shown in Figure 3B, Gα_i/13, but not Gα_i/12, could stimulate p115RhoGEF. The dose-response curve of Gα_i/13 for p115RhoGEF activation is similar to that of wild-type Gα₁₃ (Fig. 3C). Thus, Gα_i/12 and Gα_i/13 regulate effector molecules with specificity similar to their wild-type counterparts.

In summary, Gα_i/12 and Gα_i/13 chimeras retain the specific biochemical properties of their wild-type counterpart subunits. Although the N-terminal helices of Gα subunits are involved in binding Gβγ subunits, regulation of effector molecules, and specific interaction with receptors (33, 37, 38), substitution of the N-terminal helix of either Gα₁₂ or Gα₁₃ with that of Gα_i1 did not affect its interaction with Gβγ or RhoGEFs insofar as our assays could detect. Thus, the native N-terminal helix of Gα_12/13, while important for receptor selectivity (38), is not essential for the regulation of RhoGEFs.

Structure determination of Gα_i/12 and Gα_i/13

Both Gα_i/12 and Gα_i/13 were activated with AlF₄⁻ prior to crystallization, and their structures were determined by molecular replacement (Table I). As is typical for structures of monomeric Gα subunits, the N-terminal helices and extreme C-termini are disordered. The structure of Gα_i/12·GDP·AlF₄⁻ spans residues 54 to 371, and therefore contains only wild-type residues (Fig. 4A). Gα_i/13 was resolved in a GDP-bound state that most closely resembles the deactivated structures of Gα_i and Gα_t (Fig. 4B, Fig. 5B). In its structure, only one residue derived from Gα_i at the chimeric N terminus is observed, and a portion of switch II (residues 226 to 232) and two residues in the α4-β6 loop (residues 339 to 340) are additionally disordered (Fig. 4B). It is somewhat surprising that the Gα_i/13 protein crystallized in a deactivated state because both Gα_i/12 and Gα_i/13 were activated using the same protocol, formed stable complexes with RH domains (Fig. 2) and were resistant to trypsin digestion (Fig. S2). However, AlF₄⁻ binding is reversible, and different crystallization conditions likely lead to different proportions of proteins in activated and deactivated states. The relatively low pH used for the crystallization of Gα_i/13 (pH 4.8), for example, could destabilize the activated, AlF₄⁻-bound state and allow the deactivated protein to crystallize (39). Previously reported Gα·AlF₄⁻ complexes were crystallized at pH values 5.5 or higher.

Figure 4.

Structures of Gα_12/13 subunits in activated and deactivated states. (A) The activated Gα_i/12·GDP·AlF₄⁻ complex. The region of the αB-αC loop distinct from that of Gα_i/13 is colored purple. The three conformationally flexible “switch regions” of the Gα subunit are red and labeled with Roman numerals (I–III). A fourth, apparently Gα_12/13-specific element (the “spring”) is likewise colored red. The Mg²⁺·GDP·AlF₄⁻ ligand complex is shown as a ball-and-stick model, with carbons colored gray, oxygens red, nitrogens blue, phosphates yellow, fluorines purple, aluminum cyan and magnesium black. Waters are shown as red spheres. (B) The deactivated Gα_i/13·GDP complex adopts an unusually open conformation in which the α-helical domain has rotated ~8.5° away from the Ras-like domain. Switch II is almost completely disordered, and switch III has rotated away (up in the figure) from the nucleotide binding site. (C) Stereo view of the C^α traces of Gα_i/12·GDP·AlF₄⁻ (green with red switch regions) and Gα_i/13·GDP (black with yellow switch regions), superimposed using their Ras-like domains. The glutamate side chain in the “spring” of each α-helical domain (E175 in Gα₁₂, E172 in Gα₁₃) is shown as a stick model.

Figure 5.

The active sites and α-helical domains of activated and deactivated Gα_12/13 subunits. (A) The activation-dependent “spring” of the α-helical domain. In the deactivated (GDP-bound) state, Gα_i/13-E172 (yellow) is extended to form a salt bridge with Lys292 and packs against the nucleotide. In the activated (AlF₄⁻-bound) state, the spring (red) adopts a curled conformation and Gα_i/12-Glu175 does not contact the bound nucleotide. Only the Ras-like domain of Gα_i/13 is shown for clarity. Electron density from a 2.5 σ |F_o|-|F_c| omit map from the Gα_i/12·GDP·AlF₄⁻ crystal structure is shown as green wire cage. (B) Superposition of the activated and deactivated conformations of the Gα₁₃ α-helical domain. The C^α-trace of the α-helical domain (residues 74 to 201) and the β6-α5 loop (residues 345 to 357) of the Gα_i/13·GDP structure are colored black, and the α-helical domain from the Gα_13/i-5·p115RhoGEF complex is colored green. The αD-αE1 loop in the α-helical domain and the linker between the Ras-like and α-helical domains (N-terminal to αA) exhibit the most profound conformational changes upon nucleotide hydrolysis. Electron density from a 1.0 σ 2|F_o|-|F_c| map derived from the Gα_i/13·GDP crystal structure is shown as green wire cage.

Comparison of the Gα_i/12 and Gα_i/13 structures

Like other Gα subunits, Gα_i/12 and Gα_i/13 are composed of a nucleotide-binding domain homologous to that of Ras, into which an α-helical domain is inserted (Fig. 4). The most dramatic difference between Gα_i/12·GDP·AlF₄⁻ and Gα_i/13·GDP is an ~8.5° rotation of the α-helical domain away from the Ras-like domain in Gα_i/13·GDP (Fig. 4C). Among previous Gα_i and Gα_t crystal structures, the Ras-like and α-helical domains differ in relative orientation only by ~2.5° between their activated and deactivated states. Thus, the Gα_i/13 structure is unusually “open” in conformation. This is most likely due to its activation state and not to differences in primary sequence between Gα₁₂ and Gα₁₃ (see below discussion).

In addition to the three switch regions, which are well known to exhibit activation-dependent conformational changes, the Ras-like domains of Gα_i/12 and Gα_i/13 exhibit differences in the α4-β6 loop, which is disordered and one residue longer in Gα_i/13. In other Gα subunits, this loop, along with the C-terminus, is thought to contribute to receptor specificity (reviewed in (40)). More subtle differences are also evident in the backbone of the α3-β5 loop, a region known to interact with effector proteins in Gα_s and Gα_t (4, 41), although this could be affected by an effector-like crystal contact in the Gα_i/12 crystal lattice (Fig. S3). Excluding these regions, the Ras-like domains of Gα_i/12 and Gα_i/13 are quite similar, as might be expected by their 70% sequence identity. They superimpose with a root mean squared deviation (RMSD) of 0.75 Å for 175 equivalent C^α atoms (Fig. 4C).

The α-helical domains of Gα_i/12 and Gα_i/13 exhibit more profound differences. In Gα_i/13, the N-terminus of αA is kinked due to the presence of Pro86 (Asp93 in Gα₁₂), and the αB-αC loop is four residues longer and harbors an additional helix (αB1; Fig. 4B). Omitting these regions, the domains superimpose with an RMSD of 0.7 Å for 101 equivalent C^α atoms. Intriguingly, the αD-αE1 loop of Gα_i/13 (residues 171 to 173, immediately adjacent to the nucleotide binding site) adopts a strikingly different conformation from the analogous loops in the activated Gα_i/12 and Gα_13/i-5 structures, even though these loops have identical primary structure (Fig. 5A, 5B). Therefore, the αD-αE1 loop also appears to exhibit a novel activation-dependent conformational change.

Comparison of Gα_12/13 subunits with other Gα subfamilies

In line with their sequence identities (43% and 39%, respectively), the Ras-like domains of Gα_12/13 are more similar to Gα_i than Gα_s, with RMSDs of superposition of 1.1 Å and 2.2 Å for 184 and 170 equivalent C^α atoms, respectively. Regions exhibiting obvious structural differences among the Ras-like domains of Gα_12/13, Gα_i, and Gα_s are the α4-β6 loop, which projects further away from the Ras-like domain in Gα_12/13 subunits, and the β5-α4 loop (Fig. 4C). The cleft formed between the α2 (switch II) and α3 helices and their C-terminal loops is increasingly recognized as the effector binding site of Gα (17, 41, 42). In Gα_12/13 subunits, the α3-β5 loop is most similar to that of Gα_t. Differences between the α-helical domains of Gα_12/13 and Gα subunits from other subfamilies have been described previously (17).

Activation-Dependent Conformational changes in Gα_12/13 subunits

Comparison of the structures of Gα_i/12·GDP·AlF₄⁻, Gα_i/13·GDP and Gα_13/i-5·GDP·AlF₄⁻ reveals conformational changes that appear to be linked to GTP hydrolysis in Gα_12/13 subunits (Fig. 4 and Movie S1). This important deactivation mechanism has previously only been described for the Gα_i subfamily of G proteins. In the Gα_i/12·GDP·AlF₄⁻ structure, the three switch regions and the α-helical domain adopt a conformation similar to those of the activated structures of Gα_i, Gα_t and Gα_s. In Gα_i/13·GDP, switch I retains this activated conformation except for residues 206 to 210 at the C terminus of the switch, which shift by as much as 2 Å towards the α-helical domain (Fig. 4C), a conformation previously observed in several GDP-bound structures of Gα_i (34, 43). Less of switch II is disordered in the Gα_i/13·GDP structure than in Gα_i1·GDP, and switch III rotates 13° away from the active site to a position nearly identical to that observed in the Gα_t·GDP complex (44).

As described above, the α-helical domain of Gα_i/13 also appears to undergo activation-dependent conformational changes (Fig. 4C, 5B) by rotating away from the Ras-like domain and by alteration of the αD-αE1 loop (residues 171 to 173 in Gα_i/13). In other subfamilies, Gα₁₃-Glu172 is substituted by aspartate, which interacts with an invariant lysine in the Ras-like domain (Gα₁₃-Lys292) and packs against the purine ring and ribose sugar of GDP. The unusually open conformation of Gα_i/13·GDP allows the longer Gα₁₃-Glu172 side chain to form an analogous salt bridge (Fig. 5A), although in electron density maps this side chain appears less ordered than its neighboring residues. Contrarily, in activated structures of Gα₁₂ and Gα_13/i-5, the αD-αE1 loop is puckered, apparently to avoid steric collisions, and Gα₁₃-Glu172 makes relatively few contacts with the bound nucleotide (Fig. 5). Thus, the αD-αE1 loop thus acts like an activation-dependent “spring” that appears to push the α-helical domain away from the Ras-like domain in the Gα_i/13·GDP structure. This transition may be facilitated by the loss of contacts between the α-helical domain and switch III upon GTP hydrolysis.

To better understand whether these conformational differences are coupled to GTP hydrolysis and not to differences in primary structure, we need atomic structures of the activated and deactivated forms of both Gα_i/12 and Gα_i/13. Currently, Gα_i/12·GDP has not been crystallized, and we only have the Gα_13/i-5 chimera to represent the activated structure of Gα₁₃. Despite this, the 70% sequence identity of Gα₁₂ and Gα₁₃ and the 100% identity of their αD-αE1 loops suggest that they undergo similar conformational changes upon deactivation. Indeed, the activated structures of Gα_i/12 and the Gα_13/i-5 chimera are remarkably similar despite their even higher sequence disparity. Furthermore, in the Gα_i/13·GDP structure, the αD-αE1 loop adopts the same conformation as those of activated and deactivated structures of Gα_i, Gα_t and Gα_s, suggesting that this is the energetically preferred state, which can only be attained in Gα_12/13 subunits after GTP hydrolysis and outward rotation of the helical domain. However, as in any crystal structure, we cannot totally exclude the possibility that the unusually open conformation of the Gα_i/13·GDP structure was also influenced by lattice contacts or crystallization conditions.

If the open conformation of the Gα_i/13·GDP structure and the structural transition of the αD-αE1 loop indeed represent activation-dependent conformational changes in Gα_12/13 subunits, what might their physiological significance be? Because the Ras-like and α-helical domains of Gα₁₃ are both known to interact with the N-terminal fragment p115RhoGEF (17), the open conformation of deactivated Gα_i/13·GDP could help discourage interactions with p115RhoGEF or LARG upon GTP hydrolysis. Another consequence of the open conformation of Gα_i/13·GDP is that the nucleotide appears less solvent accessible in deactivated Gα_i/13 than in activated Gα_12/13 subunits (~14.5 Ų of accessible surface area in the Gα_i/12·GDP·AlF₄⁻ and Gα_13/i-5·GDP·AlF₄⁻ complexes, and 6.9 Ų in the Gα_i/13·GDP complex). The opposite is true in Gα_i subunits (e.g. 18.8 Ų in the structure of Gα_i·GDP, and 3.5 Ų in Gα_i·GDP·AlF₄⁻). These differences in the substrate binding site could influence rates of nucleotide exchange and/or hydrolysis, and thus the duration of signals in response to the activation of Gα_12/13-coupled receptors.

Structural insights into the interactions of Gα₁₂ and Gα₁₃ with effectors

The best-characterized Gα_12/13 effector is the N-terminal fragment of p115RhoGEF, which includes its RH domain (17). Because the Gα_13/i-5-p115RhoGEF structure determination employed a Gα subunit with a chimeric effector-binding site, we modeled the p115RhoGEF RH domain complex with our essentially wild-type Gα_i/13 and Gα_i/12 structures. The resulting model reveals that the chimeric residues in the Gα_13/i-5-p115RhoGEF interface either do not make significant contributions or are conservatively substituted in Gα_i/13 and Gα_i/12. In line with our observation that the RH domains of p115RhoGEF and LARG bind Gα_i/12 and Gα_i/13 equally well (Fig. 2), the structures of Gα_i/12 and Gα_i/13 have no significant amino acid differences among the residues that contribute to their respective RH domain binding sites. However, Gα₁₂ and Gα₁₃ subunits do exhibit differences in their GAP response to p115RhoGEF and LARG. The region immediately N-terminal to the RH domain of p115RhoGEF is responsible for its GAP activity and binds in the cleft formed between the Ras-like and α-helical domains (17, 19), where it contacts αA and the αB-αC loop. While the interacting residues within αA are conserved in Gα₁₂ and Gα₁₃, their αB-αC loops have dramatically different structure (Fig. 4). Diminished or detrimental interactions between the shorter αB-αC loop of Gα_i/12 and p115RhoGEF could therefore account for the higher GAP response of Gα_i/13 with respect to Gα_i/12.

Another relatively-well characterized effector target of activated Gα_12/13 subunits is cadherin (10). Recently, it was shown that mutation of residues 244 to 249 within β4 of Gα₁₂ lead to uncoupling of Gα₁₂ from p115RhoGEF and LARG, but not E-cadherin (45). Based on the crystal structure of Gα_i/12, this substitution is expected to disrupt the effector-binding site of the G protein, in accordance with the loss of p115RhoGEF and LARG binding. Cadherin either binds elsewhere on Gα_i/12, or is less reliant on a properly structured effector-binding pocket.

Conclusions

Elucidation of the molecular mechanisms by which Gα subunits regulate their own activity as well as the activity of their effector target is required not only to understand the regulation of many cellular processes, but also to develop drugs designed to modulate such processes. Gα₁₂ and Gα₁₃ have similar biochemical properties (20, 21), but distinct functional roles in cell signaling, such that Gα₁₃ knockout mice die around embryonic day E10 due to a vascular system formation defect (46), whereas Gα₁₂ knockout mice are viable and without any obvious phenotype (47). In this study, we demonstrate a simple and efficient method to produce fully functional, soluble forms of Gα₁₂ and Gα₁₃ that enabled us to determine their crystal structures. These structures reveal molecular differences not only between these two subunits, but also with those of other Gα subunits. By determining the activated structure of Gα₁₂ and the deactivated structure of Gα₁₃, we fortuitously observed novel conformational changes that are likely coupled to GTP hydrolysis. If true, then it can no longer be assumed that the transition upon deactivation will necessarily be the same in all Gα subfamilies. In the future, these chimeric subunits can be used in biochemical and crystallographic analyses to decipher their differential ability to regulate the growing number of effectors that interact with Gα_12/13 subunits (12). Remarkably, the chimeric approach we used to generate soluble Gα_12/13 subunits in this paper can be applied successfully to other, formerly intractable Gα subunits, even from other subfamilies, as evidenced by our successful structural studies of the analogous Gα_i/q chimera in complex with GRK2 and Gβγ (48).

Supplementary Material

SUPPORTING INFORMATION AVAILABLE

Figure S1. Alignment of the amino-terminal regions of Gα_i1, Gα₁₂, and Gα₁₃.

Figure S2. Gα_i/13 is activated by aluminum tetrafluoride.

Figure S3. Pseudo-effector crystal contact between the effector-binding region of Gα_i/12 and a symmetry-related α-helical domain.

Movie S1. Conformational changes proposed to occur upon deactivation of Gα_12/13 subunits.

This material is available free of charge via the Internet at http://pubs.acs.org

Abbreviations

βME

β-mercaptoethanol

DTT

dithiothreitol

EDTA

ethylenediaminetetraacetic acid

G protein

guanine nucleotide binding protein

Gα

heterotrimeric G protein α subunit

Gβγ

heterotrimeric G protein β and γ subunits

GPCR

G protein-coupled receptor

GDP

guanosine-5’-diphosphate

GST

glutathione-S-transferase

GTP

guanosine-5’-triphosphate

HEPES

4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid

LARG

leukemia-associated RhoGEF

p115RhoGEF

p115 Rho guanine nucleotide exchange factor

RGS

regulator of G protein signaling

RH

RGS homology

SDS

sodium dodecyl sulfate