Selective inhibition of alpha1A-adrenergic receptor signaling by RGS2 association with the receptor third intracellular loop.

Regulators of G-protein signaling (RGS) proteins act directly on Galpha subunits to increase the rate of GTP hydrolysis and to terminate signaling. However, the mechanisms involved in determining their specificities of action in cells remain unclear. Recent evidence has raised the possibility that RGS proteins may interact directly with G-protein-coupled receptors to modulate their activity. By using biochemical, fluorescent imaging, and functional approaches, we found that RGS2 binds directly and selectively to the third intracellular loop of the alpha1A-adrenergic receptor (AR) in vitro, and is recruited by the unstimulated alpha1A-AR to the plasma membrane in cells to inhibit receptor and Gq/11 signaling. This interaction was specific, because RGS2 did not interact with the highly homologous alpha1B- or alpha1D-ARs, and the closely related RGS16 did not interact with any alpha1-ARs. The N terminus of RGS2 was required for association with alpha1A-ARs and inhibition of signaling, and amino acids Lys219, Ser220, and Arg238 within the alpha1A-AR i3 loop were found to be essential for this interaction. These findings demonstrate that certain RGS proteins can directly interact with preferred G-protein-coupled receptors to modulate their signaling with a high degree of specificity.

Recent studies suggest that RGS proteins may interact with GPCRs to regulate their function (8 -14). In fact, a plant GPCR containing an RGS domain embedded within its C-terminal tail has been identified (15), indicating that RGS proteins and GPCRs are physically and functionally coupled in plants, and suggests that they may have evolved as separate genes in higher organisms allowing for the formation of specific paired complexes. Consistent with this idea, we showed recently (16) that purified RGS2 binds directly to the third intracellular (i3) loops of the G␣ q -coupled M1, M3, and M5 muscarinic acetylcholine receptors but not the G␣ i -coupled M2 and M4. Binding of RGS2 to the M1 receptor was shown to recruit recombinant RGS2 and to regulate the activity of endogenous G␣ q/11 in cell membranes (16). Taken together, these data support the idea that RGS proteins form direct functional complexes with preferred GPCRs in order to modulate the signaling properties of these receptors and their linked G-proteins (17,18). However, previous work has not yet demonstrated direct RGS interactions with full-length receptors in intact systems, the specific amino acids within a receptor that bind the RGS protein, and the general applicability of this phenomenon across GPCRs. ␣ 1 -Adrenergic receptors (ARs) mediate responses to norepinephrine (NE) and epinephrine (19). There are three ␣ 1 -AR subtypes (␣ 1A , ␣ 1B , and ␣ 1D ) which, like M1 receptors, also activate G␣ q and release intracellular Ca 2ϩ . ␣ 1 -ARs play an essential role in the regulation of vascular tone and blood pressure and are an important target for treatment of hypertension (19,20). Most interestingly, a recent report showed that RGS2 knock-out mice have a strongly hypertensive phenotype, with increased mean arterial pressure, renovascular abnormalities, and persistent constriction of the peripheral vasculature (21,22). Thus, we examined the possibility that RGS2 might interact specifically with ␣ 1 -AR subtypes to modulate their signaling. Unexpectedly, we found that RGS2 binds specifically to the i3 loop of the ␣ 1A -AR but not the closely related ␣ 1B -or ␣ 1D -AR. This interaction requires specific amino acids in the ␣ 1A -AR i3 loop and results in inhibition of agoniststimulated responses in intact cells. The findings of this study demonstrate that specific RGS proteins interact directly with preferred GPCRs and that this interaction is essential for controlling RGS specificity and function. ment. In order to perform pull-down assays with purified His-tagged RGS proteins, fragments encoding the i3 loops of ␣ 1A -and ␣ 1B -ARs were subcloned further into the pGEX4T vector to eliminate the His tag. ␣ 1A was amplified as a BamHI/XhoI fragment, and ␣ 1B was amplified as an EcoRI/XhoI fragment and cloned in-frame with an N-terminal GST tag.
The ␣ 1A/B 1-8 constructs encode single, double, or triple amino acid substitutions of the ␣ 1A -i3 sequence with ␣ 1B placed into the GST-␣ 1A/B chimeric template. Substitutions were made as denoted below, where the amino acid number corresponds to position of the residue in the context of the full-length ␣ 1A -AR receptor sequence: . The mutations demonstrating the largest loss in RGS2 binding capacity were then inserted into the full-length ␣ 1A -AR in the pDT vector using the Quikchange site-directed mutagenesis kit (Stratagene) for use in functional assays examining RGS2 inhibition of [ 3 H]InsP formation.
Cell Cultures and Transfections-Human embryonic kidney (HEK)293 cells were propagated in Dulbecco's modified Eagle's medium with sodium pyruvate supplemented with 10% heat-inactivated fetal bovine serum, 100 g/ml streptomycin, and 100 units/ml penicillin at 37°C in a humidified atmosphere with 5% CO 2 . Confluent plates were subcultured at a ratio of 1:5 for transfection. HEK293 cells were transfected with 3 g of DNA of each construct for 24 h using Lipofectamine 2000 transfection reagent (Invitrogen), and cells were used for experimentation 48-72 h after transfection. Wild-type Chinese hamster ovary-K1 cells were grown and maintained in Dulbecco's modified Eagle's medium (Cell-Gro) containing 10% fetal bovine serum (Atlanta Biologicals), 1ϫ nonessential amino acids (CellGro), and penicillin/streptomycin.
Preparation of Cell Lysates-To produce cell lysates containing RGS-HA proteins, wild-type Chinese hamster ovary-K1 cells were grown to 80 -90% confluency in 10-cm dishes and transfected by using Lipofectamine 2000 transfection reagent, and cell lysates were harvested as described previously (16).
RGS Pull-down Assays-Pull-down assays were performed using a modified version of the procedure described previously (16). The single modification of this assay involved pull-downs using RGS-His proteins, in which 30 mM imidazole and 80 mM NaCl were included in the buffer to control for differences in the buffers among the purified RGS proteins.
Laser Confocal Microscopy-Cells transiently transfected with HAor GFP-tagged constructs were grown on sterile coverslips, fixed for 30 min with 2% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4, and rinsed three times with phosphate-buffered saline (PBS) containing 0.5% normal horse serum (PBSϩ). For anti-HA immunostaining, fixed coverslips were blocked for 1 h in blocking buffer (PBS containing 1% bovine serum albumin, 5% normal horse serum) containing 0.01% Tri-ton X-100 to permeabilize cells. Anti-HA antibody (1:1000 dilution) was added to coverslips overnight at 4°C in blocking buffer, washed three times with PBSϩ, and incubated with rhodamine red-conjugated antirabbit IgG secondary antibody (1:500 dilution) for 1 h at room temperature in blocking buffer. Coverslips were washed three times with PBSϩ and mounted onto slides using Vectashield mounting medium. Cells were scanned with a Zeiss LSM 510 laser-scanning confocal microscope (Heidelberg, Germany) as described previously (25). For detecting GFP, fluorescein isothiocyanate fluorescence was excited using an argon laser at a wavelength of 488 nm, and the absorbed wavelength was detected for 510 -520 nm. For detecting rhodamine fluorescence, a helium-neon laser at a wavelength of 522 nm was used.

Measurement of [ 3 H]InsP
Formation-Accumulation of [ 3 H]inositol phosphates (InsPs) was determined in confluent 96-well plates. Transiently transfected HEK293 cells were prelabeled with myo-[ 3 H]inositol for 24 h, and production of [ 3 H]InsP was determined by modification of a protocol described previously (45). After prelabeling, medium containing myo-[ 3 H]inositol was removed, and 1 ml of Krebs-Ringer bicarbonate buffer (in mM: NaCl 120, KCl 5.5, CaCl 2 2.5, NaH 2 PO 4 1.2, MgCl 2 1.2, NaHCO 3 20, glucose 11, Na 2 EDTA 0.029) containing 10 mM LiCl was gently added to each well. Cells were incubated with or without 100 M NE for 60 min. The reaction was stopped by addition of 500 l of methanol, and samples were sonicated for 10 s and centrifuged for 5 min at 10,000 ϫ g. Samples were subjected to anion exchange chromatography to isolate [ 3 H]InsPs, which were quantified by scintillation counting. Percent hydrolysis of total myo-[ 3 H]inositol incorporated into lipid converted into [ 3 H]InsPs was counted and expressed as mean Ϯ S.E. Mean values were compared using the one-sample t test, with a p value less than 0.01 considered significant.
Radioligand Binding-Confluent 150-mm plates were washed with PBS and harvested by scraping. Cells were collected by centrifugation, homogenized with a Polytron, centrifuged at 30,000 ϫ g for 20 min, and resuspended in PBS. Radioligand-binding sites were measured by saturation analysis of specific binding of the ␣ 1 -AR antagonist radioligand 125 I-BE2254 (20 -800 pM). Nonspecific binding was defined as binding in the presence of 10 M phentolamine. The pharmacological specificity of radioligand-binding sites was determined by displacement of 125 I-BE2254 (50 -70 pM) by NE. Data were analyzed using nonlinear regression (26).

RGS2 Selectively
Associates with the i3 Loop of the ␣ 1A -AR-To determine whether selected B/R4 RGS proteins directly associate with ␣ 1 -ARs, we examined the capacity of closely related RGS2 and RGS16 to associate with the i3 loops of all three subtypes (␣ 1A , ␣ 1B , and ␣ 1D ) by using pull-down assays. As shown in Fig. 1, RGS2 was capable of binding to the ␣ 1A -i3, but not to the ␣ 1B -i3 or ␣ 1D -i3, whereas RGS16 did not associate with any of the three ␣ 1 -AR subtype i3 loops. To identify a specific RGS2 binding domain within the ␣ 1A -i3, we created ␣ 1A/B -i3 chimeras encoding amino acids 214 -239 of ␣ 1A fused to amino acids 259 -280 of ␣ 1B and ␣ 1B/A chimeras encoding amino acids 234 -258 of ␣ 1B fused to amino acids 239 -260 of ␣ 1A (Fig. 1B). Chimeras were then fused to GST and tested for their capacity to associate with RGS2-His by using pulldown assays. Most interestingly, we found that RGS2 robustly associated with the GST-␣ 1A/B chimera (Fig. 1C), indicating that the N-terminal half of the ␣ 1A -i3 contains an RGS2-binding motif. In contrast, we found RGS2 interacted very weakly or not at all with the GST-␣ 1B/A chimera, suggesting the first 26 amino acids of the i3 loop is predominantly responsible for RGS2 interaction. Therefore, these data indicate that an RGS2binding motif may be contained within the proximal domain of the ␣ 1A -i3.
RGS2 Co-localizes at the Plasma Membrane with ␣ 1A -ARs in HEK293 Cells-To determine whether RGS2 associates with ␣ 1A -ARs in a cellular context, we co-transfected GFP-tagged RGS proteins with HA-tagged ␣ 1 -ARs into HEK293 cells, and we examined their cellular localization by using confocal microscopy. However, to ensure that these effects are not a result of receptor and/or RGS overexpression, we performed radioligand binding experiments using 125 I-BE2254 on cells tran-siently expressing HA-␣ 1A -AR and HA-␣ 1B -ARs, and we found that both constructs express between 100 and 300 fmol/mg protein (data not shown), which is well within the physiological range for expression of these receptors in native systems (27). In addition, we found that when titrating the concentration of GFP-tagged RGS constructs to be used in transfection from 0.5 to 6 g of cDNA per 30-mm plate, 3 g of construct resulted in ϳ60% transfection efficiency with the majority of the cells exhibiting low to moderate fluorescence (data not shown). In agreement with previous studies, RGS2 (10,28) and RGS16 (16) were found primarily in the nucleus, whereas ␣ 1A -and ␣ 1B -ARs were found at the plasma membrane ( Fig. 2A) (29,30). However, upon co-transfection with HA-␣ 1A -ARs, RGS2-GFP was primarily localized at the plasma membrane (Fig. 2B, upper panels), whereas RGS2-GFP remained sequestered within the nucleus when co-transfected with HA-␣ 1B -ARs (Fig.  2B, middle panels). In addition, RGS16-GFP cellular localization was unchanged upon co-expression with HA-␣ 1A - (Fig. 2B, lower panels) or HA-␣ 1B -ARs (data not shown), indicating this association occurs in an RGS-specific manner. Therefore, these experiments support our findings from the pull-down assays demonstrating that RGS2 can selectively associate with ␣ 1A -ARs.
A previous report (10) indicates that RGS proteins are recruited to the plasma membrane when co-expressed with GPCR in an agonist-independent manner. To examine the effects of receptor ligand binding on ␣ 1A -AR/RGS2 co-localization, HEK293 cells transiently co-expressing HA-␣ 1A -ARs and RGS2-GFP were stimulated with norepinephrine, niguldipine, and prazosin, respectively. Addition of 100 nM of the ␣ 1A -AR subtype-selective antagonist niguldipine for 30 min did not alter the co-localization of ␣ 1A -AR/RGS2 at the plasma membrane (Fig. 2C, upper panel). Incubating cells with the putative ␣ 1 -AR inverse agonist prazosin (100 nM, 30 min) also did not alter the plasma membrane localization of ␣ 1A -AR/RGS, although the proteins appeared to be clustered and nonoverlapping in their distribution (Fig. 2C, middle panel). However, stimulation of cells with 10 M NE for 30 min resulted in marked ␣ 1A -AR internalization with an apparent disruption of the ␣ 1A -AR-RGS2 complex (Fig. 2C, lower panel). This was also observed following 5 and 15 min of stimulation with 10 M NE and with 5, 15, and 30 min of stimulation with 1 M NE (data not shown). Therefore, these data suggest that agonist stimulation results in ␣ 1A -AR internalization, which is followed by subsequent RGS2 dissociation from the receptor and redistribution to the cytosol.
RGS2 Selectively Inhibits ␣ 1A -AR Functional Responses in HEK293 Cells-We next examined if RGS2 can selectively regulate ␣ 1A -AR functional responses, by transiently transfecting ␣ 1 -ARs into HEK293 alone and in combination with RGS2 proteins and assaying for NE-stimulated [ 3 H]InsP formation. Both HA-and GFP-tagged RGS2 significantly inhibited [ 3 H]InsP formation by ␣ 1A -ARs in response to 100 M NE by 60.6 Ϯ 3.3 and 61.4 Ϯ 5.3%, respectively (Fig. 3, left). Transfection with pEGFP or pcDNA3.1 vector alone had no effect on ␣ 1A -AR signaling. However, HA-and GFP-tagged RGS2 had no significant effect on ␣ 1B -AR-stimulated [ 3 H]InsP formation (Fig. 3, right). Combined with our previous findings, these data indicate that RGS2 selectively associates with ␣ 1A -ARs at the plasma membrane to facilitate uncoupling of the receptor with G-protein-mediated functional responses.
Effect of RGS2/␣ 1A -AR Association on Agonist Binding Affinity-It is generally accepted that agonists display multiple affinity states for GPCRs, as a result of G-protein binding to GPCR i3 loops. Typically, binding of the G-protein induces low affinity agonist binding interactions with the GPCR, and uncoupling of the G-protein by GTP decreases agonist affinity by Ͼ10-fold. In our studies, we find that RGS2 can directly associate with the ␣ 1A -i3. We therefore tested whether RGS2 affected agonist affinity for binding to the ␣ 1A -AR. To examine this, we harvested membranes from HEK293 cells stably expressing wild-type ␣ 1A -ARs, which were then preincubated for 30 min at 4°C with concentrations of purified RGS2-HA reported previously to be sufficient for maximal association with M1 receptors in isolated Chinese hamster ovary cell membranes (16). After 30 min, the affinity of the nonselective adrenergic receptor agonist NE was determined using 125 I-BE2254 competition radioligand binding. As reported in Table  I, NE bound with low affinity to ␣ 1A -ARs when expressed alone. However, preincubation with 10, 30, or 100 nM RGS2 caused no significant change in affinity, suggesting that RGS2 binding to the ␣ 1A -AR does not affect ligand-receptor interactions.
The N Terminus of RGS2 Determines Selectivity of Receptor Association-Previously, we and others have shown the Nterminal domains of RGS proteins are required for promoting association with the G␣ subunits and with GPCRs (16, 31-33). However, the functional significance of this interaction was not examined in intact cells. Therefore, we examined the capacity of a chimera containing the N-terminal portion of RGS2 fused to the RGS domain and C-terminal portion of RGS16 (N2/ RGS16-HA) to associate with and inhibit the signaling of ␣ 1A -ARs (Fig. 4A). In pull-down assays, we found RGS2-HA and N2/RGS16-HA selectively associated with ␣ 1A -i3 GST-fusion proteins, whereas RGS16-HA did not (Fig. 4A). In HEK293 cells transiently co-transfected with HA-␣ 1A -ARs alone or with RGS constructs, co-expression with RGS2-HA resulted in significant inhibition of ␣ 1A -AR stimulation of [ 3 H]InsP formation, whereas co-expression with RGS16-HA caused no significant decrease in the ␣ 1A -AR maximal response (Fig. 4B). However, co-expression with N2/RGS16-HA caused a decrease in NEstimulated [ 3 H]InsP formation to the level observed with RGS2 (Fig. 4B). Taken together, these data suggest that the N-terminal region of RGS2 is responsible for association with and inhibition of ␣ 1A -AR function, through a direct association with the i3 loop.
Identification of Amino Acids within the ␣ 1A -i3 Responsible for Binding of RGS2-Thus far, we have determined that RGS2 associates with the ␣ 1A -AR through a direct association with the proximal half of the i3 loop. Next, we initiated studies to identify specific amino acids within the ␣ 1A -i3 that are responsible for promoting this interaction. In Fig. 1, we showed RGS2 binds the ␣ 1A/B but not the ␣ 1B/A -i3 chimera. By comparing the sequence homology between the ␣ 1A -i3 and ␣ 1B -i3 (Fig.  5A), nonhomologous amino acids in the proximal half were identified between the receptors to target for substitution mutation, and used to create a series of ␣ 1A -i3 mutants in which specific amino acids in the ␣ 1A -i3 were replaced with the corresponding amino acids in the ␣ 1B -i3, using the ␣ 1A/B as a template. A total of eight mutant ␣ 1A -i3 constructs were created using PCR, each involving between 1 and 3 amino acid substitutions (Fig. 5A). Constructs were then expressed as GST fusion proteins and were used for pull-down assays with purified RGS2-His. Of the eight ␣ 1A/B -i3 mutants examined, the constructs containing a Lys 219 -Ser 220 to Glu-Ala  mutation demonstrated the most severe loss in RGS2 binding (Fig. 5B).
To determine whether the loss of binding correlated with a decrease in RGS2 inhibition of ␣ 1A -AR functional responses, the Lys-Ser/Glu-Ala, Leu-Lys/Val-Met, and Arg/Ser mutations were introduced into full-length ␣ 1A -ARs via site-directed mutagenesis, using the full-length ␣ 1A -AR in pDT vector as a template. Full-length ␣ 1A -ARs carrying the Lys-Ser/Glu-Ala, Leu-Lys/Val-Met, and Arg/Ser mutations were then transiently transfected into HEK293 alone or in combination with RGS2-HA and assayed for NE-stimulated [ 3 H]InsP formation. In comparison to full-length ␣ 1A -ARs (B max ϭ 321 Ϯ 32 fmol/mg protein), each of the three ␣ 1A -AR mutants demonstrated relatively equal (B max values, Lys-Ser/Glu-Ala ϭ 401 Ϯ 15, Arg/ Ser ϭ 300 Ϯ 24 fmol/mg protein) or higher (B max Leu-Lys/Val-Met ϭ 710 Ϯ 45 fmol/mg protein) binding site densities. As shown in Fig. 6, RGS2 caused a significant decrease in the efficacy of NE for stimulating ␣ 1A -AR-mediated phosphatidylinositol hydrolysis (Fig. 6A). However, ␣ 1A -ARs containing either the Lys-Ser/Glu-Ala (Fig. 6B) or Arg/Ser mutations (Fig.  6C) were rendered insensitive to RGS2, indicating that these mutations abrogated the direct association between RGS2 and ␣ 1A -ARs. However, ␣ 1A -ARs containing the Leu-Lys/Val-Met mutation remained susceptible to RGS2 inhibition of functional responses (Fig. 6D) indicating that in the context of the fulllength receptor in a cellular environment, these amino acids alone are not required for RGS2 association with receptor. Taken together, these data suggest that RGS2 associates with the ␣ 1A -AR through a direct interaction within the proximal half of the ␣ 1A -i3 and that three amino acids within this domain (Lys 219 , Ser 220 , and Arg 238 ) appear to be critical for this interaction.

DISCUSSION
Many RGS proteins, particularly those of the B/R4 family, show little selectivity in inhibiting G-protein signaling when assayed as purified proteins in reconstituted systems, but they appear to have a high degree of specificity in intact cells (5). Recent reports (8 -11,16) suggest this may be due to the capacity of GPCRs to recruit RGS proteins to the plasma membrane and thus regulate the specificity of RGS function, although these studies provide only indirect evidence in support of this hypothesis. Our recent work (16) demonstrates that RGS2 binds directly and selectively to the intracellular third loop of the M1 muscarinic receptor to modulate receptor and G q/11 signaling in cell membranes. However, this and other previous work has not demonstrated direct RGS interactions with fulllength receptors in intact systems, the specific amino acids within a receptor that binds RGS proteins, or the general applicability of this phenomenon across GPCRs. In this study, we demonstrate that RGS2 directly associates with ␣ 1A -ARs through an interaction between the N-terminal domain of RGS2 and three specific amino acids in the proximal half of the ␣ 1A -AR i3 loop. Through this association, ␣ 1A -ARs recruit RGS2 to the plasma membrane, resulting in inhibition of agonist-stimulated responses. In contrast, RGS2 was incapable of directly associating with or regulating the function of the closely related ␣ 1B -AR. Thus, we demonstrate here that this direct RGS interaction is selective for specific G q/11 -linked receptors and that this interaction is necessary to confer RGS specificity of function.
Previously, RGS2 has been shown to selectively block G␣ q/11 function in reconstituted systems (34 -36) and interact with a variety of G␣ q/11 -coupled receptors (8,11,16). Therefore, we expected RGS2 would associate with all three ␣ 1 -ARs, because each subtype signals through G␣ q/11 (19). Most unexpectedly, RGS2 interacted specifically with the ␣ 1A -but not the closely related ␣ 1B -or ␣ 1D -AR subtypes. This specificity appeared to be due to RGS2 associating with specific amino acids in the ␣ 1A -AR i3 loops, as replacement with the corresponding amino acids in the ␣ 1B -AR completely abrogated the capacity of RGS2 to bind and inhibit ␣ 1A -AR agonist-stimulated responses. These data demonstrate that binding motifs located within GPCR i3 loops are responsible for RGS binding and functional specificity. In addition, these data demonstrate that an RGS protein will not necessarily interact with all GPCRs linked to a common signaling pathway. Our findings show that RGS2 is selectively recruited to the plasma membrane by the unstimulated ␣ 1A -AR to regulate receptor and G-protein signaling. These findings are consistent with our previous findings with M1 muscarinic receptors (16) and those of others (10) showing specific membrane recruitment of certain RGS proteins by unstimulated receptors. Taken together, these findings suggest that receptors and RGS proteins can form preferred functional pairs and predict a model where GPCR and RGS are prebound prior to signal initiation. By forming a complex with a specific GPCR at the plasma membrane, the RGS is positioned to modulate the linked Gprotein upon agonist activation. In this way, the receptor determines RGS protein/G-protein coupling and functional specificity. Consistent with this model, we found that RGS16 is capable of blocking receptor-mediated G q/11 signaling only when it contained the N terminus of RGS2 and is capable of binding ␣ 1A -AR. Further studies will be needed to confirm these models.
Based on a number of recent studies, it is now apparent that RGS proteins play a prominent role in regulating GPCR function in the cardiovascular system (37). Both protein and mRNA for RGS members are detectable in a variety of cardiac (38 -42) and vascular tissues (43,44), where they appear to modulate GPCR functional responses (42,43,45,46). In a recent report, RGS2 knock-out mice demonstrate a profound hypertensive phenotype and significant alterations in cardiovascular function and remodeling (21,22). Preliminary evidence also implicates other RGS proteins in the development of cardiovascular disease, including stroke (47), cardiac hypertrophy (48), and congestive heart failure (42,49). The ␣ 1A -AR subtype is known to contribute strongly to the regulation of blood pressure (50) and vascular remodeling (51,52). Therefore, the hypertensive phenotype observed in RGS2 knock-out mice may be due to an increased activity of the ␣ 1A -AR in the peripheral vasculature. Thus, based on the results of this and previous studies, RGS proteins appear to be attractive new therapeutic targets for development of novel pharmaceutical agents (37,53).
This study presents evidence that RGS2 binds directly to ␣ 1A -ARs to modulate their function. This interaction occurs with a high degree of specificity through identified domains within the receptor and the RGS protein. Thus, we propose that RGS proteins form stable, functional complexes with preferred GPCRs to selectively modulate the signaling functions of those receptors and linked G-proteins. Such direct interactions are likely to play significant roles in the still poorly understood specificity of RGS actions in vivo.  5. Identification of ␣ 1A -i3 amino acids critical for association with RGS2. A, sequence comparison of ␣ 1A -and ␣ 1B -i3 loops. ␣ 1A/B -i3 mutants were created by replacing the indicated amino acids in ␣ 1A/B -i3 with the corresponding amino acids in the ␣ 1B -i3 and numbered A/B 1-8. B, ␣ 1 -i3 constructs were fused to GST, isolated, and purified. Purified RGS2-His (0.5 g) was incubated with equal amounts of each i3 loop construct or with GST alone. Western blot analysis was performed using anti-His antibody. Results are representative of three individual experiments.