Receptor- and Nucleotide Exchange-independent Mechanisms for Promoting G Protein Subunit Dissociation*

Mechanisms for heterotrimeric G protein activation that do not rely on G protein coupled receptor activation are becoming increasingly apparent. We recently identified (cid:1)(cid:2) subunit-binding peptides that we proposed bound to a “hot spot” on (cid:1)(cid:2) subunits, stimulating G protein dissociation without stimulating nucleotide exchange and activating G protein signaling in intact cells. AGS3, a member of the activators of G protein signaling family of proteins, also activates G protein signaling in a nucleotide exchange-independent manner, and AGS3 homologues are involved in asymmetric cell division during development. Here we demonstrate that a consensus G protein regulatory (GPR) peptide from AGS3 and related proteins is sufficient to induce G protein subunit dissociation and that both the GPR and hot spot-binding peptides promote dissociation to extents comparable with a known G protein activator, AMF. Peptides derived from adenylyl cyclase 2 and GRK2 prevented formation of the heterotrimeric complex but did not alter the rate of (cid:3) subunit dissociation from (cid:1)(cid:2) subunits. These data indicate that these nucleotide ex-change-independent G protein activator peptides do not simply compete for (cid:3) interactions with (cid:1)(cid:2) subunits, but actively promote subunit dissociation. Thus, we (cid:2) ARK-ct peptide, WKKELR- DAYREAQQLVQRVPKMKNKPRS. The GPR consensus motif peptide, TMGEEDFFDLLAKSQSKRMDDQRVDLAG, was synthesized and pu- rified by Biosynthesis, Inc. (Lewisville, TX). All the peptides were dissolved in water. Preparation of Biotinylated (cid:2) 1 (cid:3) 2 Subunits— The cDNA for rat (cid:2) 1 subunit was subcloned into a baculovirus transfer vector for expression of amino-terminal fusions of a biotin acceptor peptide and the biotinylated (cid:2) 1 subunit was expressed as described previously (7). Biotinylated

Heterotrimeric G proteins activated by G protein-coupled receptors mediate a wide variety of cellular processes (1). The mechanisms by which G protein-coupled receptors activate G proteins have not been fully defined, but involve interactions between the activated receptor, G protein ␣ subunits, and perhaps G protein ␤␥ subunits. This interaction leads to the exchange of GDP for GTP on the G protein ␣ subunit, leading to a conformational change resulting in dissociation of the ␤␥ subunits from the ␣ subunits (1,2). The free ␣GTP and ␤␥ subunits interact with downstream targets and regulate their activities.
Multiple mechanisms for G protein activation that do not rely on G protein-coupled receptors or even nucleotide exchange are becoming increasingly apparent. We recently identified a receptor independent mechanism for activation of G protein ␤␥ subunit signaling by peptides derived from a random peptide phage display screen that we have proposed bind to a "hot spot" on ␤␥ subunits (3,4). Protein interaction hot spots are regions on protein surfaces thought to have unique characteristics suited to driving protein-protein interactions that are often selected for in random peptide screens (5,6).
Activators of G protein signaling (AGS1-3 proteins) 1 (7,8) were isolated from a genetic screen in yeast to look for activation of the ␤␥-mediated mating pathway. AGS3 binds to ␣ subunits and activates the signaling pathway without stimulating nucleotide exchange on the G␣ subunit (7). Sequences similar to a 25-30 amino acid repeat region in AGS3 were found in multiple other proteins and suggested to be a signature G protein regulatory (GPR) motif (7). This motif was also independently postulated to be a G protein-binding motif and termed the GoLoco motif (9). Synthetic peptides representing this motif inhibit GDP release from ␣ subunits (10 -12) and have been co-crystallized with G protein ␣ subunits (13).
To determine whether the GPR peptides and ␤␥ hot spotbinding peptides uniquely target critical sites on G protein subunits to promote subunit dissociation or if they are simply steric competitors of ␣-␤␥ binding, we analyzed and compared the effects of multiple peptides believed to interact at the ␤␥-␣ subunit interface for their ability to induce the ␣ subunit dissociation from ␤␥ subunits. We conclude that both the hot spot and GPR motif consensus peptides have the unique ability to dissociate heterotrimers by a mechanism that most likely involves conformational changes in the ␤␥ and ␣ subunits, respectively.

EXPERIMENTAL PROCEDURES
Peptides-SIGK, QEHA, and ␤ARK-ct peptide (643-670) were synthesized by Alpha Diagnostics International, purified by high performance liquid chromatography to greater than 90% purity, and their identity was confirmed by mass spectrometry analysis. The SIGK peptide was derived from the previously described SIRK (SIRKALNILGYP-DYD) peptide using a doping mutagenesis and rescreening strategy (14). Since SIGK had an apparently higher affinity for ␤␥ than SIRK but whose properties were otherwise similar to SIRK, this peptide was used throughout the studies described here. The sequences of these peptides were as follows: SIGK, SIGKAFKILGYPDYD; QEHA, QE-HAQEPERQYMHIGTMVEFAYALVGK; ␤ARK-ct peptide, WKKELR-DAYREAQQLVQRVPKMKNKPRS. The GPR consensus motif peptide, TMGEEDFFDLLAKSQSKRMDDQRVDLAG, was synthesized and purified by Biosynthesis, Inc. (Lewisville, TX). All the peptides were dissolved in water.
Preparation of Biotinylated ␤ 1 ␥ 2 Subunits-The cDNA for rat ␤ 1 subunit was subcloned into a baculovirus transfer vector for expression of amino-terminal fusions of a biotin acceptor peptide and the biotinylated ␤ 1 subunit was expressed as described previously (7). Biotinylated * This work was supported by Grant GM60286 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Measurement of ␣-␤␥ Interactions by Flow Cytometry-Binding of fluorescein isothiocyanate-labeled myristoylated ␣ i1 (F-␣ i1 ) to biotinylated ␤ 1 ␥ 2 (b-␤␥) subunits was measured using a flow cytometry assay (4,16). F-␣ i1 was kindly provided by Dr. Richard Neubig and was prepared by reacting purified myristoylated ␣ i1 with fluorescein isothiocyanate, followed by dialysis and repurification by ␤␥-agarose chromatography (16). The resulting F-␣ i1 has a specific activity of 11 pmol of [ 35 S]GTP␥S bound per g and 0.9 mol of dye/mol of protein incorporated. Biotinylated ␤␥ (50 pM final concentration) was mixed with streptavidin beads in HEDNMLG buffer (20 mM Hepes, pH 8.0, 1 mM EDTA, 1 mM DTT, 150 mM NaCl, 1.2 mM Mg 2ϩ , 0.1% C 12 E 10 , 10 M GDP) at room temperature. After 30 min, the beads were washed twice by centrifugation in a microcentrifuge with HEDNMLG buffer and resuspended in the same buffer at 10 5 beads/ml (50 pM ␤␥). For ␣ subunit dissociation experiments, the beads with bound 50 pM ␤␥ subunits were premixed with 300 pM F-␣ i for 10 min prior to the addition of the different peptides or ␣ i1 . For equilibrium binding measurements 300 pM F-␣ i1 and peptides or ␣ i1 were added simultaneously. The amount of F-␣ i1 bound to beads with b-␤␥ was assayed at the times indicated in the figure legends using a BD Biosciences FACs Calibur flow cytometer. Nonspecific binding, determined by the simultaneous addition of 300 pM F-␣ i1 and 50 nM myristoylated ␣ i1 subunits to the b-␤␥ bound beads, was 10 -20% of the total signal and was subtracted from the mean channel numbers from each experiment unless otherwise indicated.

Comparison of the ␤␥ Hot Spot-dependent Dissociation Mechanism with AMF-induced Subunit
Dissociation-We wanted to determine whether the ␤␥ hot spot-dependent mechanism for subunit dissociation was similar in magnitude to classically described mechanisms for subunit dissociation. AMF (AlF 4 Ϫ plus Mg 2ϩ ) is a well characterized mediator of G protein subunit dissociation. We chose AMF for the comparison rather than GTP or GTP␥S, because the rate of dissociation of the G protein subunits by AMF is not limited by the GDP release rate.
To measure ␣ i binding and dissociation from ␤␥, we used a flow cytometry assay developed by Sarvazyan et al. (16) that measures protein ␣-␤␥ binding at concentrations of ␣ and ␤␥ near the K d for their interaction. In this assay, biotinylated ␤␥ (b-␤␥) was immobilized on the surface of beads and fluorescein labeled ␣ i1 (F-␣ i1 ) was added. F-␣ i1 that bound to b-␤␥ on the beads was detected by the flow cytometer. The binding was concentration-dependent, and on and off rates could be measured.
To demonstrate that the ␤␥ hot spot-binding peptide (SIGK) and AMF could inhibit ␣ subunit interactions with ␤␥, they were compared for their effects on the initial binding of F-␣ i1 to b-␤␥. Excess unlabeled myristoylated ␣ i1 was used to measure nonspecific binding. SIGK, AMF, and ␣ i all inhibited formation of the heterotrimeric F-␣ i1␤␥ complex to comparable extents, indicating they were equally effective at preventing heterotrimer formation (Fig. 1A).
Next we measured the effects of SIGK on subunit dissociation. First we determined the concentration of SIGK required for dissociation of an F-␣ i1 ␤␥ complex (Fig. 1B). Increasing concentrations of SIGK caused a progressive increase in disruption of the preformed complex with maximal dissociation observed between 20 and 30 M peptide. To compare the SIGKmediated dissociation rates with AMF, we compared a maximally effective concentration of SIGK with a standard concentration of AMF (30 M AlCl 3 , 10 mM NaF, 10 mM MgCl 2 ) that should be sufficient to activate all of the ␣ subunits (17). Both SIGK and AMF enhanced the release of ␣ subunits from ␤␥ subunits relative to the intrinsic ␣ i1 off rate (measured by addition of a 50-fold excess of unlabeled myristoylated ␣ i1 ) (Fig.  1C). The k off with AMF was 0.62 min Ϫ1 and with SIGK was 0.5 min Ϫ1 compared with the intrinsic off rate of 0.1 min Ϫ1 . The intrinsic off rates are difficult to calculate accurately, because the extent of intrinsic dissociation in this short time course was only 20%. Significantly more intrinsic dissociation occurs with longer time courses, but it is during this initial dissociation phase that the greatest effects of the peptides are observed. Since the off rates induced by SIGK and AMF are similar, it indicates that the ␤␥ subunit-binding peptide-dependent dissociation mechanism has the potential to increase the ␣-␤␥ dissociation rate to an extent comparable with the classical G protein ␣ subunit activation dependent dissociation.
Effect of MgCl 2 on the Rate of ␣ Subunit Dissociation from the Heterotrimeric Complex-Magnesium is known to profoundly affect the interaction of ␣ and ␤␥ subunits, and dissociation by activation of the ␣ subunit with either GTP␥S or AlF 4 Ϫ is strongly dependent upon Mg 2ϩ concentration (1,18). Magnesium stabilizes interactions of AlF 4 Ϫ and GTP␥S at the nucleotide-binding site and directly influences subunit dissociation through an undefined lower affinity site. For the experiment in Fig. 1C, SIGK-mediated dissociation was measured with 1.2 mM MgCl 2 and 1 mM EDTA (0.2 mM free Mg 2ϩ ). To determine whether Mg 2ϩ was required for subunit dissociation by SIGK, we measured SIGK-mediated dissociation in the absence of MgCl 2 , in the presence of 1 mM EDTA (Fig. 1D). SIGK was able FIG. 1. The hot spot-binding peptides are nearly as effective as AMF at promoting subunit dissociation. A, SIGK inhibits the heterotrimeric G␣␤␥ complex formation. 300 pM F-␣ i was preincubated with 50 nM unlabeled myristoylated ␣ subunit, 10 M SIGK, or AMF (60 M AlCl 3 , 10 mM NaF, 10 mM MgCl 2 ) in HEDNMLG buffer followed by incubation with biotinylated ␤ 1 ␥ 2 (50 pM) bound to streptavidin beads. After 30 min, F-␣ i1 binding to the b-␤␥ beads was measured using flow cytometer. For each sample, 3000 beads were assessed. B, SIGK concentration dependence for dissociation of heterotrimeric complex. 300 pM F-␣ i1 was mixed with 50 pM b-␤␥ in the HEDMLNG buffer and incubated for 15 min followed by the addition of the indicated concentrations of SIGK. The amount of F-␣ i1 bound was measured 2 min after addition of the peptide. C, comparison of the dissociation rates of F-␣b-␤␥ complex. 300 pM F-␣ i1 was preincubated with b-␤␥ as in B, and dissociation was initiated by the addition of 50 nM myristoylated ␣ i1 subunits (OE), 25 M SIGK (f), or AMF (30 M AlCl 3 , 10 mM NaF, 10 mM MgCl 2 ) (). Intrinsic dissociation rates were not significantly different if 25, 50, or 100 nM myristoylated ␣ i1 were used. Solid lines are fits of the data to single exponential dissociation functions. D, magnesium is not required for SIGK-mediated dissociation but enhances the rate of dissociation. The rate of F-␣ i1 dissociation from a preformed heterotrimeric F-␣ i1 ␤␥ complex was measured as in Fig. 1, B and C. The additions were: 50 nM myristoylated ␣ i1 subunits with 1 mM EDTA and no added MgCl 2 (OE) or 11 mM MgCl 2 (‚), 25 M SIGK with 1 mM EDTA and no added MgCl 2 (f), or 11 mM MgCl 2 (ᮀ). The curves are fit to single exponential decay functions. All data are mean Ϯ S.E. from pooled data from three separate experiments.
to significantly enhance subunit dissociation in the absence of Mg 2ϩ , indicating Mg 2ϩ is not required for this dissociation mechanism. Addition of Mg 2ϩ at a free concentration of 10 mM increased the intrinsic off rate of F-␣ i as expected from 0.1 min Ϫ1 in the absence of Mg 2ϩ to 0.37 min Ϫ1 with 10 mM Mg 2ϩ . The rate of ␣ subunit dissociation by SIGK increased from 0.43 min Ϫ1 without Mg 2ϩ to 1.14 min Ϫ1 with 10 mM Mg 2ϩ .
Effect of ␤␥-Binding Peptides on the Interaction of F-␣ i1 with ␤␥-Since SIGK and other hot spot-binding peptides enhance the off rate of ␣ from ␤␥ subunits, it suggests they do not act simply by competition for reformation of spontaneously dissociating ␣-␤␥ complexes. Such a mechanism would not increase the dissociation rate relative to the intrinsic off rate of ␣ dissociation from ␤␥. Based on this we have suggested that the peptides stabilize a conformation of ␤␥ subunits that has a lower affinity for ␣ leading to an enhanced rate of ␣ subunit dissociation (4). An alternative mechanism might be that a small peptide could compete for one of the two major contacts of ␣ subunits with the sides and top of the ␤ subunit torus during transient separation of one of these contacts, thereby leading to an enhancement of the dissociation rate. If there were nothing unique about the hot spot-binding peptides, and it were simply a competitor at the ␣-␤␥ interface, this model would predict that any peptide that bound at the ␣ subunit interface with ␤␥ should enhance the off rate.
To try to distinguish between these two mechanisms, we tested two ␤␥-binding peptides thought to bind to ␤␥ subunits at the ␣ subunit interface. QEHA is a 27-residue peptide derived from the second catalytic domain of adenylyl cyclase 2 (amino acids 956 -982) and inhibits ␤␥ regulation of several effectors including K ϩ channels, phospholipase C-␤, and adenylyl cyclase (19). The IC 50 for QEHA effects on most processes was 50 -200 M. Cross-linking of the QEHA peptide to the ␤ subunit is prevented by the ␣ subunit, suggesting it binds to ␤␥ within the ␣ subunit-binding site (19,20). Another peptide derived from the COOH-terminal region of GRK2 (␤ARK-ct, 643-670) (21) also binds to ␤␥ subunits, has an IC 50 of 100 M for its effects, and has properties consistent with binding at the ␣ subunit-binding site on ␤␥ subunits. This notion is supported by the recently determined co-crystal structure of ␣ subunits with GRK2 (22), demonstrating binding of the region of GRK2 corresponding to this peptide to a region on ␤␥ that overlaps with the ␣ subunit-binding site.
To confirm that these peptides block ␣-␤␥ interactions, the ability of these peptides to block binding of F-␣ i1 to ␤␥ was tested by flow cytometry. Both peptides inhibited the heterotrimer formation with QEHA inhibiting by 90% and ␤ARK-ct peptide (300 M) by more than 65% compared with 85% by SIGK ( Fig. 2A). This supports the idea that QEHA and ␤ARKct peptide can inhibit ␣-␤␥ interactions probably by directly competing for ␣ binding to ␤␥.
To determine whether QEHA and ␤ARK-ct peptides can stimulate release of ␣ i1 from the heterotrimer, we measured the off rates in the presence of these peptides at concentrations that significantly inhibited ␣ binding to ␤␥. Neither of these peptides caused any significant enhancement of the rate of dissociation of the ␣ subunits from the preformed complex (Fig.  2B). There are some minor differences in the rate of dissociation for QEHA, ␤ARK, and the intrinsic dissociation rate, but these differences were not consistently reproducible and were not significant. Thus, while being able to compete for ␣-␤␥ interactions they are unable to promote dissociation of ␣ from ␤␥. These data strongly suggest that the ␤␥ hot spot-binding peptides act through a unique mechanism that does not involve simple steric occlusion of ␣ binding sites on ␤␥ subunits.
Effect of a GPR Consensus Motif Peptide on Association and Dissociation of G␣␤␥-A class of proteins that stimulate G protein ␤␥ subunit signaling by a nucleotide exchange-independent mechanism is the AGS proteins. AGS3 binds to ␣ subunits, inhibits GDP release, and promotes ␤␥ subunit signaling in yeast and is involved in establishing ␤␥ and ␣ subunit-dependent asymmetric cell division in Caenorhabditis elegans and Drosophila. A 28-amino acid (GPR) motif consensus pep- FIG. 2. Binding of peptides to ␤␥ at ␣ subunit-binding surfaces does not enhance the rate of subunit dissociation. A, competition for binding of F-␣i1 to ␤␥ in the presence of ␤-subunit-interacting peptides QEHA and ␤ARK. Biotinylated-␤␥ was bound to streptavidin beads in HEDNMLG buffer. QEHA, ␤ARK-ct-(643-670) (300 M each), or 25 M SIGK were added to b-␤␥ bound beads followed by addition of 300 pM F-␣ i1 , incubation for 30 min at room temperature, and measurement of the bound ␣ i1 by flow cytometry. Maximal binding of F-␣ i1 was determined in the absence of any competitors (control). B, comparison of kinetics of F-␣i1 dissociation from ␤␥ by SIGK, QEHA, and ␤ARK peptides. Peptides or excess myristoylated ␣ i1 subunits were added to a complex formed with 300 pM F-␣ i1 and 50 pM ␤␥ subunits as in Fig tide derived from AGS3 and other related proteins is able to bind to ␣ subunits and inhibit nucleotide exchange in a manner comparable with larger protein fragments of AGS3 (10,23). Since a GPR-like peptide from RGS14 binds to ␣ subunits near the ␣-␤␥ interface, it was proposed that these peptides enhance ␤␥ subunit-dependent processes by competing for rebinding of ␣ to ␤␥ by steric occlusion of the ␤␥ binding site on ␣ (13).
We directly tested whether the GPR peptide could inhibit ␣-␤␥ binding and/or promote subunit dissociation. 1 M GPR peptide was able to inhibit ␣ binding to ␤␥ to an extent greater than that observed with 10 M SIGK (Fig. 3A). The IC 50 for inhibition of ␣-␤␥ interactions was ϳ250 nM (Fig. 3B), comparable with that observed for the ability of the peptide to inhibit GDP dissociation from ␣ subunits. Next we tested whether 1 M GPR peptide could cause dissociation of a preformed ␣-␤␥ complex. The GPR peptide caused a rapid dissociation of the G protein subunits, about 2-fold faster (0.95 min Ϫ1 ) than that observed with 25 M SIGK (0.5 min Ϫ1 ) and about 13-fold higher than the intrinsic off rate of the F-␣ i1 subunit (Fig. 3C). The GPR peptide very potently promoted dissociation (Fig. 3D) and was about 10-fold more potent than SIGK (compare Fig. 3D with Fig. 1B). Thus, the GPR consensus peptide is a very potent and effective promoter of G protein subunit dissociation. Since they dramatically increase the dissociation rate of ␣ from ␤␥, it strongly suggests that the GPR peptides act by a non-competitive mechanism. DISCUSSION We compared the effects of the GPR peptides and ␤␥ hot spot-binding peptides with peptides that are thought to bind at the G protein ␤␥ subunit-␣ subunit interface. We show that AC2-and ␤ARK-derived peptides are capable of blocking ␤␥-␣ subunit interactions consistent with previous data, suggesting that they bind at the ␣-␤␥ interface. These competitor peptides were unable to enhance the rate of G protein subunit dissociation, while SIGK and GPR peptides significantly enhanced subunit dissociation. The enhanced rates of G protein subunit dissociation by SIGK and GPR peptides were comparable with a known activator of G proteins, AMF. This strongly suggests that neither the SIGK peptides nor the GPR peptides are simply acting by preventing reassembly of dissociated subunits to lead to G protein activation. The GPR-and ␤␥-binding peptides are unique in that they induce subunit dissociation, probably through conformational alterations of ␣ or ␤␥ subunits respectively.
In the x-ray crystallographic structural model of a GPR-like peptide from RGS 14 in a complex with ␣ i1 (13), conformations of switch I and II are altered relative to heterotrimeric ␣ i GDP ␤␥. These conformational differences could result in subunit dissociation via a mechanism analogous to the GTP or AlF 4 Ϫ dependent conformational changes in the switch regions of the ␣ subunit that contribute to subunit dissociation (24).
GPR proteins play important roles in regulation of asymmetric cell division in Drosophila (25) and in C. elegans (26). In particular, they appear to regulate the polarized distribution of free ␤␥ and ␣ subunits derived from heterotrimers required for correct orientation of mitotic spindles. In ␣ subunit immunoprecipitates from Drosophila sensory organ precursor cells, Pins (an AGS3 homologue in Drosophila) and a peptide repre-senting the GPR motif from Pins disrupted ␣-␤␥ interactions when added during the immunoprecipitation (25). Our results that directly measure subunit dissociation are consistent with these results.
The data presented also support a model where a conformational change in ␤␥ subunits induced by hot spot-binding peptides results in destabilization of interactions with ␣ subunits and an increase in the k off for subunit dissociation. ␤␥ subunits have indeed been shown to undergo conformational changes upon binding of phosducin (27,28), although the functional significance of these changes are not entirely understood. A mechanism for ␤␥-binding peptide-mediated enhancement of subunit dissociation, whatever the details of the mechanism, is clearly very distinct from other mechanisms that exist for promoting subunit dissociation by either GPR peptides or nucleotide binding that involve conformational alterations of the switch regions of the ␣ subunits.