GPCR-induced dissociation of G-protein subunits in early stage signal transduction

Abstract

G-protein coupled receptors (GPCRs) form a ternary complex of agonist, receptor and G-proteins during primary signal transduction at the cell membrane. Downstream signalling is thought to be preceded by the process of dissociation of Gα and Gβγ subunits, thus exposing new surfaces to interact with downstream effectors. We demonstrate here for the first time, the dissociation of heterotrimeric G-protein subunits (i.e., Gα and Gβγ) following agonist-induced GPCR (α_2A-adrenergic receptor; α_2A-AR) activation in a cell-free assay system. α_2A-AR membranes were reconstituted with the G-proteins (±hexahistidine-tagged) Gα_i1 and Gβ₁γ₂ and functional signalling was determined following activation of the reconstituted receptor:G-protein complex with the potent agonist UK-14304, and [³⁵S]GTPγS. In the presence of Ni²⁺-coated agarose beads, the activated his-tagged Gα_i1his-[³⁵S]GTPγS complex was captured on the Ni²⁺-presenting surface. When his-tagged Gβ₁γ₂ (Gβ₁γ₂his) was used with Gα_i1, the [³⁵S]GTPγS-bound Gα_i1 was not present on the Ni²⁺-coated beads, but rather, it was separated from the β₁γ₂(his)-beads, demonstrating receptor-induced dissociation of Gα and Gβγ subunits. Treatment of the reconstituted α_2A-AR membranes containing Gβ₁γ₂his:Gα_i1 with imidazole confirmed the specificity for the Ni²⁺:G-protein surface dissociation of Gα_i1 from Gβ₁γ₂his. These data demonstrate for the first time, the complete dissociation of the G-protein subunits and extend observations on the role of G-proteins in the assembly and disassembly of the ternary complex in the primary events of GPCR signalling.

• G-proteins
• G-protein coupled receptor
• signalling
• receptor
• dissociation

Introduction

G-protein coupled receptors (GPCRs) represent a superfamily of membrane proteins primarily involved in the transduction of intercellular signals. GPCRs are one of the largest superfamilies in the human genome and are implicated in a number of physiological disorders and diseases. As such, GPCRs represent a significant target for drug discovery programs as illustrated by the fact that approximately 40% of all currently marketed drugs are directed against as many as 30 GPCR types [1]. Of the 747 predicted human GPCRs, 380 are thought to be chemosensory receptors, whereas the remaining 367 GPCRs are predicted to bind endogenous ligands, with 224 of these having been identified, leaving more than 140 orphan GPCRs [2].

GPCRs are structurally characterized by their seven transmembrane (‘serpentine’) spanning domains. GPCR activation occurs from a wide variety of extracellular stimuli such as light, odorants, neurotransmitters, chemokines and hormones. The cycle of formation and reversible interaction and disengagement of the signal transducing protein machinery (i.e., the interaction of GPCR with heterotrimeric G-proteins) governs the cascade of intracellular responses. GPCRs closely associate with the G-proteins [3], Gα and the Gβγ dimer, and activate the G-proteins by promoting binding of GTP in exchange for GDP on the Gα subunit. Both Gα and Gβγ-subunits of activated G-proteins can regulate downstream effectors such as adenylyl cyclases, phospholipases, or ion channels [4], [5]. Based on biochemical experiments and structural studies it is known that on GTP binding, conformational rearrangements in the ‘switch regions’ of the Gα-subunits weaken the interaction with Gβγ-subunits [6–8]. It is generally assumed that the reduced affinity of Gα-GTP for Gβγ causes the G-proteins to dissociate into Gα-GTP and a Gβγ-dimer following GPCR activation. Re-association of the G-protein subunits is then thought to occur upon hydrolysis of the bound GTP which can be accelerated by RGS proteins [9–11]. This classical description (for example [12], [13]) of subunit dissociation has recently been questioned [14]. It is not known definitively whether dissociation of the G-protein subunits or alternatively G-protein conformational changes occurs in cell signalling. However, following agonist activation, it is known that the G-protein complex (heterotrimer) initially engages the GPCR [3] which is followed by a subsequent dissociation process from the GPCR [15], [16].

In this report, we extend previous observations of GPCR signalling and demonstrate that the process of (ligand-induced) receptor-stimulated G-protein activation involves a rearrangement of Gα_i1-GTP leading to the dissociation of the subunits Gβ₁γ₂ and Gα_i1. In these experiments we utilised a baculovirus/Sf9 cell system involving the [³⁵S]GTPγS binding assay, both of which have been used to characterize a wide variety of GPCRs in the first steps of GPCR-induced G-protein signalling [17–19]. Additionally, we demonstrate the use of Ni^2 +-NTA beads as a capture surface for histidine tagged G-proteins.

Materials and methods

Cell culture and protein expression

G-Proteins and the α_2A-adrenergic receptor (α_2A-AR) were prepared from baculovirus infected Sf9 insect cells. Sf9 cells (Invitrogen, Mount Waverley, Vic, Australia) were grown in suspension in Sf-900 II SFM serum-free media (Invitrogen) and maintained at 28°C with agitation at 140 oscillations/min in an orbital shaker. Recombinant baculovirus samples encoded either the α_2A-AR, the G-protein subunits Gα_i1, β₁, γ₂ or the N-terminal hexahistidine-tagged Gα_i1his or γ₂his. For expression of G-proteins or α_2A-AR, Sf9 cells (at 2×10⁶ cells/ml) were infected with the appropriate combinations of Gα_i1 (±hexahistidine-tagged) and Gβ₁γ₂ (±hexahistidine-tagged) or α_2A-AR baculovirus(es) at a multiplicity of infection ratio of 1:2 and harvested at 72 h.

Membrane preparation

Infected Sf9 cells (up to 1.5 l) were collected and centrifuged at 1000×g for 10 min and re-suspended in 125 ml ice-cold ‘lysis buffer’ (50 mM HEPES pH 8.0, 0.1 mM EDTA, 3 mM MgCl₂, 10 mM β-mercaptoethanol) with protease inhibitors; 0.02 mg/ml phenylmethyl sulfonyl fluoride, 0.03 mg/ml benzamidine, 0.025 mg/ml bacitracin and 0.03 mg/ml Lima bean trypsin inhibitor. Cells were subjected to N₂ cavitation at 500 psi (3400 kPa) for 15 min, followed by sedimentation of nuclei and unbroken cells (750×g, 10 min). Membranes were isolated by centrifugation of the supernatant at 100,000×g for 30 min at 4°C.

Preparation of urea-treated membranes

A modification of the following method was used to remove endogenous G-proteins from Sf9 membranes expressing α_2A-AR [20]. The 100,000×g membrane pellet was resuspended (50 ml per 1 l of original infected Sf9 cell culture) in ‘incubation buffer’ (250 mM Sucrose, 10 mM Tris pH 8.0, 3 mM MgCl₂) containing 7 M urea and protease inhibitors (as above) at 4°C. After 30 min incubation on ice, membranes were diluted to 4 M urea with ‘incubation buffer’ and protease inhibitors and centrifuged at 100,000×g at 4°C for 30 min. The urea-treated membrane pellet was washed twice in ‘incubation buffer’. The final urea-treated membrane pellet was resuspended to approximately 1–3 mg/ml protein and aliquots were rapidly frozen in liquid N₂ and stored at −80°C until use.

Confirmation of α_2A-AR expression

To determine receptor–ligand binding specificity, α_2A-AR membranes were expressed and [³H]MK912 (Perkin Elmer Life Sciences Inc., Boston, MA, USA) binding was carried out on the α_2A-AR membrane preparations. The K_d ranged from 0.5–1.0 nM and B_max was between 5–20 pmol/mg, with >95% specific binding using yohimbine as the antagonist. Additionally, to confirm specificity of the expressed α_2A-AR membranes, the following adrenergic receptor antagonists were used; rauwolscine (α₂-receptor), yohimbine (α₂-receptor), prazosin (α₁-receptor) and propranolol (β-receptor).

Purification of G-proteins

A modification of the following method was used [21]. Frozen membranes (at ≥ 5 mg/ml protein) containing combinations of G-protein α and βγ subunits were thawed and diluted to 5 mg/ml protein with ‘wash buffer’ (50 mM HEPES pH 8.0, 3 mM MgCl₂, 10 mM β-mercaptoethanol, 50 mM NaCl and 10 µM GDP) containing fresh protease inhibitors and 1% (w/v) cholate (final concentration). Membranes were stirred on ice for 1 h to extract. The sample was then centrifuged at 100,000×g for 40 min and the supernatant collected and diluted 5-fold with buffer A (20 mM HEPES pH 8.0, 100 mM NaCl, 1 mM MgCl₂, 10 mM β-mercaptoethanol, 0.5% (w/v) polyoxyethylene-10-lauryl ether and 10 µM GDP). The sample was then loaded onto a 1 ml nickel-nitrilotriacetic acid (Ni(NTA)) column (Qiagen Pty Ltd, Clifton Hill, Vic, Australia) and allowed to pass through (under gravity). The sample flow thru was collected and re-applied to the column. The column containing G-proteins with either Gα_i1his + Gβ₁γ₂, or Gα_i1+Gβ₁γ₂his, was washed with 50 ml of buffer A containing 5 mM imidazole and 300 mM NaCl (pH 8.0) to remove non-specifically bound proteins. The column was warmed to room temperature for 15 min and non His6-tagged G-protein subunits eluted with 1 ml fractions of buffer E containing AlF₄⁻ (20 mM HEPES pH 8.0, 50 mM NaCl, 10 mM β-mercaptoethanol, 10 µM GDP, 1% (w/v) cholate, 50 mM MgCl₂, 5 mM imidazole, 10 mM NaF and 30 µM AlCl₃). The remaining His6-tagged bound G-protein subunits were eluted from the nickel column with buffer E containing an additional 150 mM imidazole. The eluted G-proteins were placed on ice and confirmation of protein expression and purity was carried out by SDS-polyacrylamide gel electrophoresis and Coomassie blue staining. Protein purity was routinely > 95% using this procedure. Eluted fractions containing the appropriate G-protein subunits were pooled (to ≤ 3 ml) and dialysed against 200 ml buffer F (20 mM HEPES pH 8.0, 3 mM MgCl₂, 10 mM NaCl, 10 mM β-mercaptoethanol, 1 µM GDP and 0.1% (w/v) cholate) using a Slide-a-lyzer (Pierce Chemical Company, Rockford, IL, USA) for 1.5 h at 4°C. The dialyzing buffer was then replaced with fresh (cold) buffer F and the process was repeated a total of four times (1.5 h for each dialysis). To aid in the complete removal of residual AlF₄⁻ or imidazole from the G-protein samples, they were further dialysed in 250 ml Buffer F overnight at 4°C. Following complete dialysis, the protein integrity and concentration was confirmed by SDS-polyacrylamide gel electrophoresis of G-protein subunits followed by laser scanning densitometry or by measurement using the Bradford method. G-proteins (5–10 µM) were aliquoted, rapidly frozen in liquid nitrogen and stored at −80°C.

Reconstitution and [³⁵S]GTPγS binding

A ‘reconstitution mix’ was prepared on ice consisting of 0.05–0.2 mg/ml α_2A-AR as indicated, 5 µM GDP, 10 µM AMP-PNP, and appropriate Gα and Gβγ proteins (±histidine tagged) at 20–50 nM, 0.2 nM [³⁵S]GTPγS in TMND buffer (50 mM Tris pH 8.0, 100 mM NaCl, 1 mM MgCl₂ and 1 mM DTT). The reactions (100 µl final volume) were initiated by addition of either buffer (control) or 10 µM UK-14,304. 100 µM yohimbine (α₂-AR antagonist), rauwolscine (α₂-AR), prazosin (α₁-AR) or propranolol (β-AR) was added (where indicated) to determine receptor-induced signalling specificity. In some experiments (where indicated), 10 µl of suspended agarose beads in TMND (50% suspension, 45–165 µm diameter beads) pre-charged with Ni^2 +-nitrilotriacetic acid (Ni(NTA) (Qiagen Pty. Ltd., Clifton Hill, Vic, Australia) were included in the assay to allow for G-protein capture via the hexahistidine-tags on the Gα or Gβγ-subunits. Where shown, 100 mM imidazole was used to remove the hexahistidine-tagged G-protein from the Ni(NTA) agarose beads by competition of imidazole with the histidine-tag of G-proteins to Ni^2 +-binding sites. Reactions were incubated at 27°C, with gentle mixing for 90 min. Reactions were either filtered over a Whatman GF/C filter to capture soluble G-proteins or a Whatman No.1 paper filter to capture Ni(NTA)-bound hexahistidine-tagged G-proteins associated with the Ni(NTA) beads or a ‘filter stack’ comprising a top layer of Whatman #1 filter paper (to collect any hexahistidine-tagged G-protein associated with the Ni(NTA) agarose beads) with a GF/C directly beneath it to collect any G-proteins that were not captured by the Ni(NTA) agarose beads. Filters were washed with 3×4 ml washes of ice cold TMN buffer. Filters were dried (separately) and subjected to liquid scintillation counting to determine [³⁵S]GTPγS bound.

Results

There have been numerous studies using the [³⁵S]GTPγS binding assay and baculovirus/Sf9 expression system for GPCRs to quantify activated G-protein signalling following GPCR stimulation (reviewed in [17–19]). The aim of this study was to use a well characterized model GPCR-G-protein system to determine G-protein subunit dissociation. Therefore, we chose to utilize the α_2A-AR and Gα_i1 and Gβ₁γ₂ proteins expressed in the Sf9 insect cell/baculovirus expression system, which has been characterized previously [20]. Additionally, we incorporated a urea extraction step into the existing method of α_2A-AR membrane preparation in order to greatly reduce or eliminate all endogenous G-proteins (Gα and Gβγ) from the insect cell membranes as demonstrated previously [20], [22], [23]. Using [³H]MK912 binding to α_2A-AR/Sf9 membranes, we found that the K_d ranged between 0.5–1.0 nM (this compares to a Kd of 1.25 nM for previously reported data [24] and the B_max was between 5–20 pmol/mg, with >95% specific binding using yohimbine as the antagonist.

In a previous study, Gα_i1 and Gβ₁γ₂ were reconstituted with urea-treated α_2A-AR membranes and signalling was measured by [³⁵S]GTPγS binding [20]. In our study, we reconstituted the purified G-proteins (20 nM) Gα_i1his and Gβ₁γ₂ with urea-treated α_2A-AR membranes (0.1 mg/ml total protein), and measured α_2A-AR stimulation with 10 µM UK14304 in the presence of 0.2 nM [³⁵S]GTPγS for 90 min (GF/C filters were used to capture [³⁵S]GTPγS and although GF/C filters are usually used to capture membranes, we have found that they specifically capture soluble Gα subunits whether obtained from Sf9 cells and myristoylated, or obtained from bacteria and not myristoylated (data not shown)). Under these conditions, significant agonist-induced, receptor-mediated, G-protein [³⁵S]GTPγS binding occurred. In preliminary experiments we found that when using the reconstituted α_2A-AR membranes, inclusion of 5 µM GDP in the reconstitution assay optimized the signal-to-background. Increasing the concentration of GDP decreased the total level of [³⁵S]GTPγS bound without altering the EC₅₀ of UK14304-activated α_2A-AR induced [³⁵S]GTPγS binding (data not shown). However, when lower concentrations of GDP were used, high background binding was observed. Additionally, we obtained the most satisfactory results in preliminary experiments when using G-protein concentrations between 10–50 nM in the presence of α_2A-AR membrane concentrations of 0.05–0.2 mg/ml (with a B_max approximately 10 pmol/mg).

Figure 1(A) shows that the bound [³⁵S]GTPγS was minimal under ‘basal’ conditions (i.e., in the absence of agonist). However, when the agonist UK14304 was included in the reconstitution with α_2A-AR, Gα_i1his and Gβ₁γ₂, the level of [³⁵S]GTPγS binding was significantly increased by approximately 6-fold. Furthermore, when the reconstitution mix also contained the α₂-adrenergic receptor antagonist, yohimbine (100 µM), the agonist response was blocked and equivalent in value to the ‘basal’ or background signal. When the α_2A-AR reconstitution contained Gα_i1 and the hexahistidine-tag on the Gβ₁γ₂-dimer, the increase in UK14304-stimulated [³⁵S]GTPγS binding was significantly higher than the basal or yohimbine-treated samples (Figure 1(B)). Additionally the level of binding when using the histidine tagged Gβ₁γ₂his (+Gα_i1) as shown in Figure 1(B) was similar to that when using histidine tagged Gα_i1his and Gβ₁γ₂ (Figure 1(A)). This confirmed that the histidine tagged Gβ₁γ₂-subunits (Gβ₁γ₂his) and Gα_i1/α_2A-AR reconstitution mix formed a functional complex. In parallel, we also noted that this level of signalling was similar to that when non-histidine tagged proteins were used (data not shown) and this was also comparable to that previously reported when α_2A-AR membranes were reconstituted with non-histidine tagged G-proteins [20].

Figure 1.  Reconstitution of α_2A-AR with (A), Gα_i1his + Gβ₁γ₂ or (B), Gα_i1+Gβ₁γ₂his in suspension. G-protein subunits (20 nM) were combined with 0.1 mg/ml of α_2A-AR membranes, 5 µM GDP, 10 µM AMP-PNP (‘reconstitution mix’), 0.2 nM [³⁵S]GTPγS and 100 µM yohimbine (where shown) in TMND buffer. The reaction was started by addition of UK14304 (10 µM final concentration). The reaction was incubated for 90 min at 27°C with shaking. The final volume was 100 µl and the entire reaction mix was filtered over a GF/C filter and washed with 3×4 ml with ice-cold TMN buffer (n=3, mean±SEM). Abbreviations; UK, UK14304, Yoh, yohimbine.

In separate experiments, we demonstrate the usefulness of the interactions of histidine tagged G-proteins and Ni^2 +-coated (Ni(NTA) agarose beads. To capture [³⁵S]GTPγS bound to his-tagged Gα_i1 (Gα_i1his) on Ni(NTA) beads we filtered the reaction mix over a paper filter. These paper filters captured the relatively large Ni(NTA) beads, whilst allowing all proteins not associated with the beads to pass through the paper with little or no non-specific binding to the paper. Receptor induced [³⁵S]GTPγS binding to Gα_i1his subunits captured on the surface of Ni(NTA) beads via the hexahistidine tag is shown in Figure 2. The ideal volume of Ni(NTA) beads for such protein capture was between 5–10 µl per assay.

Figure 2.  Ni^2 +-bead capture of α_2A-AR-activated [³⁵S]GTPγS:Gα_i1his. 20 nM of both Gα_i1his and Gβ₁γ₂ were combined with 0.1 mg/ml of α_2A-AR membranes and 5 µM GDP and 10 µM AMP-PNP (‘reconstitution mix’), 0.2 nM [³⁵S]GTPγS, in TMND buffer. Various volumes of Ni(NTA) agarose beads were added to the reconstitution mix as indicated, in the absence (○) or presence (•) of 10 µM UK14304 [final]. The mix was incubated for 90 min at 27°C with shaking. Final volume was 100 µl. The entire reaction was filtered over a Whatman #1 filter paper and washed with 3×4 ml with ice-cold TMN buffer. A representative experiment is shown.

We further characterized the receptor-induced signalling on Ni(NTA) beads using the above approach, by carrying out a dose response experiment using the α_2A-AR agonist UK14304, either in the absence or presence of Ni(NTA) beads. Figure 3 shows the results of a representative experiment. The EC₅₀ values were similar in the absence or presence of Ni(NTA) beads (24 nM and 11 nM, respectively), further indicating that the α_2A-AR-induced signalling response was unaltered by the presence of the Ni^2 +-capture surface and that the pharmacological response was preserved. To further confirm the specificity of the receptors, Figure 4 demonstrates that the α_2A-AR-expressing membranes (Figure 4(A)) in the absence of G-proteins display a pharmacological profile expected of an α_2A-AR i.e., the antagonist specificity was yohimbine (α₂-AR antagonist)≡rauwolscine (α₂-AR)»prazosin (α₁-AR) > propranolol (β-AR) in a competition binding experiment using [³H]MK912. Furthermore, Figure 4(B) demonstrates that the same pharmacological profile resulted when the [³⁵S]GTPγS binding experiment was carried out using Ni(NTA) beads to capture Gα_i1his:[³⁵S]GTPγS. This is not surprising and confirms the receptor-activated G-protein signalling specificity was preserved in the presence of Ni(NTA) beads.

Figure 3.  Agonist dose–response curves with and without Ni^2 +-beads. Various concentrations of UK14304 were incubated with 0.2 mg/ml α_2A-AR containing reconstituted Gα_i1his (50 nM) and β₁γ₂ (50 nM), 5 µM GDP; 10 µM AMP-PNP and 0.2 nM [³⁵S]GTPγS in the absence (□) or presence (▪) of 10 µl Ni(NTA) agarose beads and the basal binding was 40 and 42 fmol/mg, respectively, whilst the maximal (UK14304-stimulated) binding was 300 and 275 fmol/mg, respectively. Data are presented as percent of maximum bound. The mix was incubated for 90 min at 27°C with shaking. Final volume was 100 µl and the entire reaction was filtered over a Whatman #1 filter and washed with 3×4 ml with ice-cold TMN buffer. The EC₅₀ values were 24 nM and 11 nM in the absence and presence of Ni(NTA) agarose beads, respectively. A representative experiment is shown.

Figure 4.  Antagonist competition curves. To determine IC₅₀ values at the α_2A-AR, the following adrenergic receptor subtype-specific antagonists were used: rauwolscine (•), yohimbine (○), prazosin (▪) or propranolol (□) as indicated. (A) Direct receptor binding of [³H]MK912 (1 nM) with various concentrations of different adrenergic receptor subtype-specific antagonists as indicated. The reaction mix was filtered over GF/C filters and washed with 3×4 ml with ice-cold TMN buffer. The calculated IC₅₀ values for rauwolscine, yohimbine, prazosin or propranolol for the α_2A-AR membrane preparation were 0.091 µM, 0.141 µM, ≈73 µM and ≈275 µM, respectively. (B) Receptor-stimulated binding of [³⁵S]GTPγS in the presence of UK14304 and the indicated amounts of selected antagonists. The experiment was carried out using 20 nM of both Gα_i1his and Gβ₁γ₂ combined with 0.1 mg/ml of α_2A-AR membranes, 5 µM GDP and 10 µM AMP-PNP (‘reconstitution mix’), 0.2 nM [³⁵S]GTPγS in the presence of various concentrations of the different adrenergic receptor subtype-specific antagonists as indicated and 10 µl Ni(NTA) agarose beads (n=3, mean±SEM). UK14304 (1 µM final concentration) was added to start the reactions and the mix was incubated for 90 min at 27°C with shaking. Final volume was 100 µl and the entire reaction was filtered over a Whatman #1 filter and washed with 3×4 ml with ice-cold TMN buffer. The IC₅₀ values for each of the antagonists was determined as 0.051 µM (rauwolscine: selective α₂-AR antagonist); 0.080 µM (yohimbine: selective α₂-AR antagonist); 8.3 µM (prazosin: selective α₁-AR antagonist) and 86.9 µM (propranolol: β-AR antagonist).

In Figure 1(A) and 1(B), the reconstitution mix containing G-proteins was filtered over a GF/C filter only, to enable binding and capture of the [³⁵S]GTPγS-labelled G-proteins. In Table I we demonstrate the usefulness of a ‘filter stack’. The filter stack was comprised of a Whatman #1 paper (which do not trap proteins, but do capture the relatively large Ni(NTA) beads used) overlaid onto a Whatman GF/C (which bind proteins). The initial aim was to demonstrate the successful capture of histidine-tagged G-proteins on the surface of Ni(NTA) beads, by measurement of [³⁵S]GTPγS bound to the purified G-protein subunits. Data in the first row of Table I, indicated by ‘No Ni(NTA) beads’, shows that the Gα proteins (whether monomeric or as the heterotrimer) are captured by the GF/C filter placed beneath the paper filter, as shown by radioactivity associated with the filter. The β₁γ₂his did not bind significant radioactivity, as expected. In the second row, indicated by ‘Ni-NTA beads’, with non-histidine tagged Gα_i1 there was negligible binding to the beads, as indicated by no radioactivity on the paper filter. When purified Gα_i1his was treated in the same way, the complex bound to the Ni(NTA) beads, and was captured on the paper filter. When Gα_i1his was combined with β₁γ₂, the complex bound to the Ni(NTA) beads. The complex of Gα_i1 and β₁γ₂his was also shown to bind to Ni(NTA) beads, although some of the Gα_i1 apparently dissociated and bound to the GF/C filter (or it is possible that not all the Gα_i1 present was bound to the β₁γ₂his). In the final row, indicated by ‘Ni-NTA beads + imidazole’, the presence of imidazole changed the binding of radioactivity in a manner consistent with the displacement of his-tagged G proteins from Ni(NTA) beads in every case.

Table I.  Demonstration of his-tag specific capture of G-proteins on Ni^2 +-beads. Purified samples of (50 nM) Gα_i1, Gα_i1his, β₁γ₂his, Gα_i1hisβ₁γ₂ or Gα_i1β₁γ₂his were incubated with 5 nM [³⁵S]GTPγS in the absence or presence of Ni(NTA) beads (±200 mM imidazole) for 90 min at 27°C. The samples were then filtered over a filter paper stack as described in Materials and Methods, and depicted in Figure 7. Each filter was counted by liquid scintillation counting separately and the results are shown (in the table, ‘–’ indicates negligible binding associated with the filter, whereas ‘ + ’ indicates significant binding was associated with that filter, by measurement of radioactivity). The Whatman #1 ‘Paper’ filters retain the relatively large (approximately 45–165 µm diameter) Ni(NTA) beads and associated G-proteins if they contain a histidine tag, whilst GF/C filters bind all proteins that are not immobilized on the surface of Ni-NTA beads.

	Gαi1		Gαi1his		β1γ2his		Gαi1hisβ1γ2		Gαi1β1γ2his
	Paper	GF/C	Paper	GF/C	Paper	GF/C	Paper	GF/C	Paper	GF/C
No Ni-NTA beads	−	+	−	+	−	−	−	+	−	+
Ni-NTA beads	−	+	+	−	−	−	+	−	+	+
Ni-NTA beads + imidazole	−	+	−	+	−	−	−	+	−	+

With the above observations using purified G-proteins, our next aim was to determine whether it was possible to utilize the filter stack to follow ligand-induced, receptor activated G-proteins. Therefore, in the following experiments, we used the α_2A-AR/G-protein reconstitution mix in the absence or presence of Ni(NTA) beads and then filtered this over a filter stack and determined the level of [³⁵S]GTPγS binding associated with the paper filter and the GF/C filter, separately. In order to demonstrate receptor specificity we used yohimbine and to demonstrate specificity of capture of G-proteins on the surface of Ni(NTA) beads we used imidazole (where shown). In Figure 5 the reconstitution contained α_2A-AR, Gα_i1his and β₁γ₂. Following activation of the GPCR-G-protein complex with UK14304, we observed that there was minimal [³⁵S]GTPγS binding associated with the paper filters (Figure 5(A)) in the filter stack. However, in contrast, Figure 5(B) shows the [³⁵S]GTPγS was captured by the GF/C filter placed beneath the paper filter in the filter stack, suggesting that the [³⁵S]GTPγS bound G-protein passes through the porous paper filter without hindrance and was subsequently captured by the GF/C filter. This demonstrated the usefulness of the filter stack protocol.

Figure 5.  Reconstitution of α_2A-AR with Gα_i1his + Gβ₁γ₂ in suspension. 20 nM of both Gα_i1his and Gβ₁γ₂ were combined with 0.1 mg/ml of α_2A-AR membranes, 5 µM GDP, 10 µM AMP-PNP (‘reconstitution mix’), 0.2 nM [³⁵S]GTPγS and 100 µM yohimbine (where shown) in TMND buffer. The reaction was started with 10 µM UK14304 (final) and the reaction was incubated for 90 min at 27°C with shaking. Final volume was 100 µl. The entire reaction was filtered over a filter ‘stack’ that was comprised of a Whatman #1 paper filter on top of a GF/C filter. The sample was washed with 3×4 ml with ice-cold TMN buffer. (A) The Whatman #1 filter data set and (B) GF/C filter data set were counted (by scintillation) separately (n=3, mean±SEM). Abbreviations; UK, UK14304, Yoh, yohimbine.

We then pursued the hypothesis that the receptor activated (and dissociated) [³⁵S]GTPγS-(his tagged)-G-protein may be captured onto Ni(NTA) beads via histidine-Ni^2 + binding (Figure 6(A)). Therefore, the ‘filter stack’ experimental protocol as in Figure 5 was repeated with the addition of 10 µl of Ni(NTA) beads to the reconstituted α_2A-AR membranes containing Gα_i1his and Gβ₁γ₂. Interestingly, UK14304-stimulated [³⁵S]GTPγS binding was almost entirely associated with the paper filters on the filter stack, whereas in Figure 5(B) the receptor-stimulated [³⁵S]GTPγS binding was associated with the GF/C filters. Figure 6(A) demonstrates that in this assay the UK14304-stimulated level of [³⁵S]GTPγS bound was increased approximately 3-fold over the basal or yohimbine-treated level. Since basal and yohimbine-treated levels of bound [³⁵S]GTPγS were minimal, this result suggested that the UK14304-stimulated [³⁵S]GTPγS binding was due to the [³⁵S]GTPγS:Gα_i1his bound specifically to the Ni(NTA) beads via the Ni^2 +-histidine interactions (and retained on the paper filter) using the ‘filter stack’. To determine whether the interaction of [³⁵S]GTPγS:Gα_i1his with Ni(NTA) was specific, imidazole was included in some experiments in the presence of the Ni(NTA) beads (Figure 6(A), 4th bar from left) and confirmed the specificity, i.e., there was minimal retention of [³⁵S]GTPγS bound to the paper filter in the presence of imidazole. However, the [³⁵S]GTPγS binding in this experiment was found to be associated with the GF/C filter which was beneath the paper filter in the filter stack as shown in Figure 6(A) (8th bar). This suggested that the α_2A-AR-activated [³⁵S]GTPγS-G-protein was unable to bind to the Ni^2 +-coated surface of beads in the presence of imidazole, and that the [³⁵S]GTPγS-G-protein passed through the paper filter and was caught on the GF/C. Therefore, the histidine tagged Gα_i1his was specifically prevented from binding the surface of Ni(NTA) beads with imidazole and importantly, this did not effect the level of signalling activity ([³⁵S]GTPγS binding), since a similar level of binding was observed on the surface of the paper filter (ie [³⁵S]GTPγS-G-protein) in the absence of imidazole (compare Figure 6(A) 2nd bar with Figure 6(A) 8th bar). Essentially, with this experimental construct, any [³⁵S]GTPγS-labelled histidine tagged G-protein that was bound to the Ni(NTA) beads (and hence retained on the paper filter) was captured and measured.

Figure 6.  α_2A-AR-activated [³⁵S]GTPγS:G-protein subunit dissociation. (A), Ni^2 +-bead capture of α_2A-AR-activated [³⁵S]GTPγS:Gα_i1his. 20 nM of both Gα_i1his and Gβ₁γ₂ were used and (B), α_2A-AR-activated [³⁵S]GTPγS:Gα_i1/β₁γ₂his dissociation. 20 nM of both Gα_i1 and β₁γ₂his were used. In both (A) and (B) the appropriate combination of G-proteins were combined with 0.1 mg/ml of α_2A-AR membranes and 5 µM GDP and 10 µM AMP-PNP (‘reconstitution mix’), 0.2 nM [³⁵S]GTPγS, 100 µM Yohimbine or 100 mM imidazole (where shown) in TMND buffer. 10 µl of Ni(NTA) agarose beads were added to the reconstitution mix and the reaction was started with 10 µM UK14304 [final]. The mix was incubated for 90 min at 27°C with shaking. Final volume was 100 µl. The entire reaction was filtered over a ‘filter stack’ that was comprised of a Whatman #1 filter paper on top with a GF/C filter beneath it. The bead sample was washed with 3×4 ml with ice-cold TMN buffer. Whatman #1 paper filters and GF/C filters were counted (scintillation) separately as indicated (n=3, mean±SEM). Abbreviations; UK, UK14304.

In separate control studies (data not shown) we demonstrated that if functional G-protein subunits lacked the hexahistidine-tag (i.e., Gα_i1 and Gβ₁γ₂), or if the Ni(NTA) beads were pre-treated with EDTA to remove Ni^2 + from the NTA chelate surface, UK14304-induced [³⁵S]GTPγS binding was absent from the Ni(NTA) beads. These data further support the necessity for the Ni^2 + and histidine residue interactions for bead capture of the G-proteins.

Since it was not definitive at this point whether the Gα-subunit was labelled with [³⁵S]GTPγS as the entire G-protein heterotrimer or Gα as a complex dissociated from the Gβγ subunits, we carried out further experiments to test this. Firstly, it should be noted that we demonstrated that Gβ₁γ₂his binds to the Ni(NTA) beads in the presence of (non-histidine tagged) Gα_i1 (Table I). Nevertheless, this was theoretically possible since when the Gβ₁γ₂his and Gα_i1 subunits were purified from Sf9 cell extracts, they were bound and eluted from the same Ni(NTA) beads used in the above experiments. With the observation that the α_2A-AR membrane/Gα_i1:Gβ₁γ₂his forms a functional signalling complex following UK14304 stimulation (Figure 1(B)) and that β₁γ₂his:Gα_i1:[³⁵S]GTPγS binding does occur at the Ni(NTA) bead surface, our next experiment aimed to determine whether agonist-induced α_2A-AR membranes would lead to the dissociation of the heterotrimeric G-protein subunits using this methodology. Therefore, the experiment was conducted in a similar manner to that as for Figure 6(A), except the histidine-tag was on the Gβ₁γ₂ subunit, i.e., β₁γ₂his (Figure 6(B)). In this experiment the α_2A-AR membranes were reconstituted with Gα_i1 and Gβ₁γ₂his followed by the addition of Ni(NTA) beads. Using the histidine tagged Gβ₁γ₂his in such a manner, we anticipated that the Gβ₁γ₂his would attach to the surface of Ni(NTA) beads (as shown in Table I). Therefore, if [³⁵S]GTPγS was associated with Ni(NTA) beads (i.e., retained on the paper filter), this would indicate the presence of a heterotrimeric complex of [³⁵S]GTPγS:Gα_i1:β₁γ₂his:Ni(NTA) beads suggesting no dissociation of the G-proteins. Alternatively, if the [³⁵S]GTPγS was found to pass through the paper filter and was retained on the GF/C, essentially as [³⁵S]GTPγS:Gα_i1, this would suggest Gα_i1 and Gβ₁γ₂his dissociation had occurred. Figure 6(B) shows that there was minimal [³⁵S]GTPγS bound to the paper filter in the filter stack (i.e., on beads), under all conditions tested (basal, +UK14304, +UK14304 + Yohimbine or +UK14304 + imidazole) suggesting that there was little or no Ni(NTA) bead associated [³⁵S]GTPγS bound to G-proteins. Figure 6(B) shows that the UK14304-stimulated [³⁵S]GTPγS labelled Gα_i1 was associated with the GF/C filter that was placed beneath the paper filter in the filter stack, i.e., the activated and dissociated [³⁵S]GTPγS:Gα_i1 passed through the paper filter and was retained by the GF/C filter. Furthermore, the presence of imidazole did not inhibit the [³⁵S]GTPγS:Gα_i1 from binding to the GF/C filter (see Figure 6(B), 6th and 8th bars) and further confirmed the specificity of the dissociation of the G-protein subunits. Since the β₁γ₂his was shown to bind the Ni(NTA) surface of beads in Table I, as well as during the purification process using the crude membrane fraction, it is likely that the [³⁵S]GTPγS:Gα_i1 bound to the GF/C filter was in the form of [³⁵S]GTPγS:Gα_i1 only and not [³⁵S]GTPγS:Gα_i1:β₁γ₂his (depicted in Figure 7). Therefore, this was direct evidence that in the reconstituted α_2A-AR membranes with Gα_i1 and Gβ₁γ₂, there was dissociation of the G-protein subunits following agonist-induced receptor activation of the G-protein heterotrimer, and this was demonstrable without the use of detergents.

Figure 7.  Schematic demonstrating the specificity of the ‘filter stack’. Following α_2A-AR-stimulated G-protein activation, as measured by [³⁵S]GTPγS binding to Gα subunits, the reaction mixture is filtered over a ‘filter stack’ comprising a top layer of Whatman #1 filter paper (to collect Ni^2 +-coated (Ni(NTA)) agarose beads containing hexahistidine-tagged G-proteins), with a GF/C directly beneath it (to collect G-proteins that are not captured by the Ni(NTA) agarose). (A), The [³⁵S]GTPγS bound to Gα_i1his is attached to the Ni(NTA) bead via the heaxahistidine-Ni^2 + interaction, and is thus retained on the upper Whatman #1 filter in the stack. (B), Following α_2A-AR-stimulated G-protein activation using Gα_i1 and β₁γ₂his, the activated Gα_i1:[³⁵S]GTPγS complex is not retained on the Ni(NTA) beads but is captured by the GF/C filter.

Discussion

The functional consequences of the proposed cellular ternary complex model of agonist, GPCR, and G-proteins, includes the rapid binding of GTP to the Gα subunit, release of the receptor and the Gβγ dimer. As a result of these changes, there is exposure of new Gα and Gβγ surfaces which enable interaction with downstream effectors such as adenylate cyclase [4], [25]. It is known that following activation of the GPCR by its specific agonist (or partial agonist), disassembly of ligand, GPCR and G-proteins occurs [15], [16]. Although the dissociation of the G-protein heterotrimer (Gαβγ) has also been thought to occur as a result of GPCR activation [6–8], there has been no definitive evidence to demonstrate GPCR-induced dissociation of G-protein subunits using GPCR expressing membranes. Nevertheless, it was thought that this process is an absolute prerequisite for the exposure of new Gα and Gβγ subunit surfaces to interact with downstream cytoplasmic effector proteins. In this study using a reconstituted cell-free system we provide for the first time, direct evidence to substantiate the hypothesis that GαGTP and Gβγ subunits dissociate following activation of the ligand-activated GPCR/G-protein ternary complex. This was confirmed using a modification of the widely accepted [³⁵S]GTPγS binding assay as a measure of reconstituted and functional GPCR signalling [17], [18].

Immobilized receptor and G-protein interactions have been characterized previously with cell-free systems using surface plasmon resonance [15] or on beads using flow cytometry approaches [16], [26–30]. The kinetics of assembly and disassembly of the ternary complex was measured using a bead-based approach with derivatized surfaces. In such studies the assembly and disassembly of the receptor (e.g., recombinant β₂-adrenergic receptor fused with Green Fluorescent Protein [16]) from the G-protein heterotrimer complex was measured, but the Gα and Gβγ subunit dissociation was not directly measured. Therefore, we provide data that extends the concept of ternary complex disassembly to include dissociation of the individual G-protein subunits.

We acknowledge that a limitation of this study is the use of heterologous cells, however, based on the results shown in Figures 3 and 4, the pharmacological profile of the characterized receptor preparation is as expected for α_2A-AR, i.e., potencies were yohimbine≡rauwolscine»prazosin > propranolol. Additionally, this type of expression system has been well characterized and is considered by others as a ‘gold standard’ for demonstrating experimentally and in vitro, the first steps of GPCR signalling [17–19]. Therefore, at least in terms of this limited cell-free system, we were able to demonstrate G-protein subunit dissociation. Ideally, it would be most appropriate to demonstrate such G-protein dissociation (or otherwise) in unmodified mammalian cells to support or refute the G-protein dissociation hypothesis. This aspect of G-protein signalling is being rigorously studied by many groups and definitive answers still await as to whether G-proteins dissociate or rearrange themselves upon receptor activation. Some groups have attempted to answer this by incorporating large recombinant fusion proteins into the receptors and/or G-protein subunits which could be problematic in interpretation [3], [14], [31]. In our study we used hexahistidine-tagged Gα or Gγ subunits which exhibited normal functional interactions with the α_2A-AR and subsequent [³⁵S]GTPγS binding. Also, it was possible to avoid using unacceptably high (non-physiologic) concentrations of proteins, GDP and GTP. Only minor modifications of the G-proteins (i.e., histidine tagging) were required and this did not appear to affect the signalling and functioning of the ligand:receptor:G-protein ternary complex. Furthermore, the level of signalling in solution (i.e., in the absence of Ni(NTA) beads was comparable to that obtained as a captured ternary complex assembly on the Ni(NTA) beads, suggesting this type of assay construct may prove useful for high throughput drug discovery applications.

Interestingly, we noted that the order of addition of the G-proteins and α_2A-AR to the beads was important (data not shown). In some experiments we successfully coated Ni(NTA) beads with Gα_i1his or with Gβ₁γ₂his and captured Gα_i1 as determined by [³⁵S]GTPγS binding to the beads. However, on addition of the α_2A-AR (after coating the beads with G-proteins) we were unable to detect functional signalling as determined by agonist-induced [³⁵S]GTPγS binding. This suggested that although the G-protein(s) likely remained functional, the liganded α_2A-AR was unable to interact with the G-proteins to elicit a signalling response. This may have been in part due to the orientation of the G-proteins captured on the Ni(NTA) beads being such that steric hindrance of the G-protein:α_2A-AR occurred resulting in the inability to form a ternary or GPCR:G-protein complex assembly. Additionally, this also suggested that in our experiments where the α_2A-AR was reconstituted with the G-proteins prior to addition of Ni(NTA) beads, the histidine tagged G-proteins could only bind to the Ni^2 + presenting surface of the beads after the event of activation and dissociation of G-proteins from the receptor complex. This information will also help contribute to the understanding of the order of activation of G-protein signalling for future mathematical models. Furthermore, the efficiency of the Ni(NTA) beads to capture all available free (non-bound) Gα_i1his-GTP was demonstrated by using a high concentration of Ni(NTA) beads in the assay system (e.g., between 5–10 µl in a final assay volume of 100 µl). Since the experiments were done in 100 µl, with approximately 0.05 nM Gα_i1-[³⁵S]GTPγS bound to the Ni(NTA) surface, this equates to approximately 5×10⁻¹⁵ mol of Gα_i1his bound to the Ni(NTA) beads. Therefore, if these experiments were to be repeated using non radioactive methodologies such as fluorescence-based technologies, it would be necessary to develop a sensitive method capable of measuring such low quantities of activated Gα_i1-GTP.

In summary, using a cell-free approach, we have demonstrated that G-protein activation involves a rearrangement of the heterotrimeric G-protein complex such that dissociation of the Gβγ-dimer and the Gα subunits occur using a G_i-coupled GPCR. The net effect of this is likely to involve the exposure of new (previously hidden) G-protein surfaces available to interact with downstream effector proteins, such as phospholipases, adenylate cyclases and ion channels. Although we provide data here for only the Gα_i class of G-proteins using the α_2A-AR as the model receptor, it is likely that this observation may extend across all classes of G-proteins, such as G_i/o, G_s, G_q/11, however, this needs to be tested. It should now be possible to incorporate the G-protein subunit association and dissociation kinetics into the ligand:receptor:G-protein ternary complex model and thus determine the kinetics and timing of the entire GPCR signalling cycle. The use of a technically simple system as reported here, eliminates complex biochemical mechanisms that exist to discriminate, integrate and modulate signal transduction in vivo allowing us to demonstrate the dissociation of the subunits in a cell-free system.

We wish to thank Mrs Sharon Burnard and Ms Kelly Bailey for providing expert technical assistance and to Dr George Lovrecz and staff (CSIRO Molecular & Health Technologies, VIC, Australia) for up-scaling the insect cell cultures. This work was supported by the CSIRO's Emerging Science Area for Nanotechnology funding scheme. WRL wishes to thank the Australian Academy of Science for travel support. We also thank Prof Richard R. Neubig (University of Michigan) for generously providing the recombinant baculoviruses encoding the α_2A-adrenergic receptor, Gα_i1, Gα_i1his, β_1, γ₂ and γ₂his.