Identification of RGS2 and type V adenylyl cyclase interaction sites.

The production of cAMP is controlled on many levels, notably at the level of cAMP synthesis by the enzyme adenylyl cyclase. We have recently identified a new regulator of adenylyl cyclase activity, RGS2, which decreases cAMP accumulation when overexpressed in HEK293 cells and inhibits the in vitro activity of types III, V, and VI adenylyl cyclase. In addition, RGS2 blocking antibodies lead to elevated cAMP levels in olfactory neurons. Here we examine the nature of the interaction between RGS2 and type V adenylyl cyclase. In HEK293 cells expressing type V adenylyl cyclase, RGS2 inhibited Galpha(s)-Q227L- or beta(2)-adrenergic receptor-stimulated cAMP accumulation. Deletion of the N-terminal 19 amino acids of RGS2 abolished its ability to inhibit cAMP accumulation and to bind adenylyl cyclase. Further mutational analysis indicated that neither the C terminus, RGS GAP activity, nor the RGS box domain is required for inhibition of adenylyl cyclase. Alanine scanning of the N-terminal amino acids of RGS2 identified three residues responsible for the inhibitory function of RGS2. Furthermore, we show that RGS2 interacts directly with the C(1) but not the C(2) domain of type V adenylyl cyclase and that the inhibition by RGS2 is independent of inhibition by Galpha(i). These results provide clear evidence for functional effects of RGS2 on adenylyl cyclase activity that adds a new dimension to an intricate signaling network.

Heterotrimeric G protein-mediated receptor signaling pathways are pivotal parts of the intricate and diverse biological processes dictating cellular function. G proteins transduce signals to a variety of effectors including adenylyl cyclase (AC) 1 (1). The hormone-sensitive AC system is a typical archetype of G protein-mediated signal transduction. Appropriate agonistbound, heptahelical receptors activate G s by catalyzing the exchange of GDP for GTP. The GTP-bound ␣ subunit of G s in turn activates AC, increasing the rate of synthesis of cyclic AMP from ATP (1,2). All isoforms of mammalian AC are stimulated by the heterotrimeric G protein G s . Many other regulatory influences including RGS (regulators of G protein signaling) proteins have crucial effects on various AC isoforms (3). These enzymes thus serve critical roles as integrators of diverse inputs.
RGS2 is a 211-amino acid protein that like other members of this family mediates its GAP activity via its core domain (4 -6). Experiments using recombinant RGS2 and membranes prepared from insect cells (Sf9) expressing different AC isoforms indicate that RGS2 suppresses the activity of AC III, the predominant isotype in the olfactory system, and the cardiac isoforms, V and VI (3). RGS2 has also been shown to decrease cAMP accumulation in ␤TC3 insulinoma cells (7). When added to purified recombinant type V AC cytoplasmic domains, recombinant RGS2 decreases cAMP production stimulated by either G␣ s or forskolin. The structural basis for the inhibitory effect of RGS2 remains unknown, although it is likely a direct effect. In vivo, odorant-elicited cAMP stimulation resulted in the opening of cyclic nucleotide gated channels. The microinjection of RGS2 antibody into olfactory neurons dramatically augments the currents observed in whole cell voltage clamp recordings of odorant-stimulated olfactory neurons. These results indicate that the level of RGS2 in an olfactory neuron may regulate the responsiveness of that neuron to olfactory stimulants (3). This study provided evidence for a novel role of RGS proteins as direct regulators of AC activity.
The mechanisms for RGS2 regulation are as yet unclear (8). Phosphorylation of RGS2 by protein kinase C decreases its capacity to negatively regulate phospholipase C␤ activation (9). Signal-induced redistribution may also regulate RGS2 function (10) and have important effects on cellular signaling. RGS2 shares with RGS4 and RGS16 a conserved N-terminal domain that is necessary and sufficient for plasma membrane targeting (10). Until recently, most reports of the intracellular localization of members of the RGS family of proteins have found them to be associated with the membrane fraction or distributed between the membrane and cytosolic fractions (11,12); however, there have been some reports indicating the nuclear localization of RGS proteins (13,14). In human astrocytoma 1321N1 cells, increases in cAMP and phosphoinositide signaling gives rise to a rapid and transient increase in RGS2 expression (15). In these cells both endogenous and transfected RGS2 is found largely, although not entirely, localized in the nucleus (15). Thus, the localization of RGS2 may in fact be controlled in ways not yet appreciated.
The list of non-G protein-binding partners for RGS proteins continues to grow (8,16). It remains to be determined whether interactions of RGS2 with G␣ q/11 , G␣ i , AC, and coat protein (␤-COP) can explain the putative roles for RGS2 in control of behavior, T cell function, and synaptic development in hippocampal CA1 neurons as revealed by RGS2-deficient mice (17). We describe herein a direct interaction between RGS2 and the C 1 domain of type V AC. In addition, we report the nature of the AC-binding site on RGS2.

EXPERIMENTAL PROCEDURES
Materials-The anti-HA (hemagglutinin A) and Ni-NTA-horseradish peroxidase conjugate antibodies were purchased from Roche Molecular Biochemicals and Qiagen, respectively. Anti-G␣ i , anti-G␣ s , and anti-GST antibodies were purchased from Santa Cruz Biotechnology. The anti-RGS2 antiserum was generated in rabbits against a C-terminal peptide (PQITTEPHAT) coupled to keyhole limpet hemocyanin followed by affinity purification using immunizing peptide. Goat anti-rabbit IgG horseradish peroxidase conjugate was obtained from Bio-Rad, and sheep anti-mouse IgG horseradish peroxidase was from Amersham Biosciences. FuGENE 6 transfection reagent was purchased from Roche Molecular Biochemicals, and all other cell culture reagents were obtained from Invitrogen. The cyclic AMP enzyme immunoassay kit was purchased from Assay Designs, Inc. (Ann Arbor, MI).
Tissue Culture and in Vivo cAMP Detection-HEK293 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal calf serum, 1 mM glutamine, 50 g/ml streptomycin, and 50 units/ml penicillin at 37°C with 5% CO 2 . For transient expression of proteins, subconfluent HEK293 cells were plated on 6-well plates and transfected using FuGENE 6 transfection reagent following the manufacturer's protocol with constructs expressing either ␤ 2 -adrenergic receptor (0.5 g/well) or G␣ s -Q227L (0.5 g/well) along with RGS2 (1.0 g/well) and type V AC (0.2 g/well). The vector pcDNA3 (Invitrogen) was used to normalize all of the DNA concentrations to 1.7 g/well. After 36 h of transfection, the cells were starved overnight in medium containing 1% fetal bovine serum, followed by starvation in medium devoid of serum for 2 h.
To measure cAMP accumulation in cells transfected with G␣ s -Q227L or corresponding control DNA constructs, the cells were treated with 1 mM 3-isobutyl-1-methylxanthine for 15 min and subsequently harvested in 250 l of hypotonic lysis buffer (50 mM Tris, pH 7.5, 4 mM EDTA, plus protease inhibitors). A portion of the cell lysate (50 l) was used for Western blotting to monitor protein expression, and the remaining lysate was boiled for 10 min at 100°C and spun at 14000 ϫ g for 10 min to remove cellular debris. The supernatant was used to measure the total cAMP accumulation by cAMP enzyme immunoassay detection. For ␤ 2 -adrenergic receptor activation and corresponding controls, the cells were treated with isoproterenol (0.1 M) for 15 min before subsequent harvesting, lysis, and cAMP detection. The data are represented as the means Ϯ S.E., and statistical significance is determined by Student's t test (p Ͻ 0.05 is considered significant).
Western Blotting-The cell lysates were boiled for 5 min in Laemmli sample buffer and sonicated. The proteins were resolved by SDS-PAGE and transferred to nitrocellulose. The protein expression levels were determined by immunoblotting using anti-HA antibody. Immunoreactive protein bands were detected by a horseradish peroxidase-conjugated secondary antibody and enhanced chemiluminescence.
Purification of Recombinant Proteins-RGS2 polymerase chain fragments digested with BspHI and BamHI were inserted in-frame with an N-terminal hexahistidine tag in pQE60 using NcoI and BamHI. The proteins were expressed in Escherichia coli and purified under nondenaturing conditions using Ni-NTA-agarose beads (24). The purified protein fractions were dialyzed in buffer containing 10 mM Hepes, pH 7.4, 10 mM ␤-mercaptoethanol, 10% glycerol, 200 mM NaCl, pooled, concentrated, and stored at Ϫ70°C. Type V AC domains were expressed with an N-terminal (VC 2 ) or C-terminal (VC 1 ) hexahistidine tag and purified as previously described (25). GST-VC 1 was purified using glutathione affinity resin. BL21(DE3) cells containing the GST-VC 1 expression plasmid were grown at 30°C in T7 medium until the A 600 reached 1.0. Synthesis of GST-VC 1 was induced with 0.5 mM isopropylthiogalactoside, and incubation continued for 4 h at 30°C. The cells were harvested and resuspended in lysis buffer containing 50 mM Tris, pH 7.7, 1 mM EDTA, 2 mM dithiothreitol, 120 mM NaCl, and protease inhibitors prior to incubation with 0.2 mg/ml lysozyme for 30 min at 4°C and then subsequent incubation with 5 mM MgCl 2 and DNase (0.01 mg/ml) for 30 min. Cellular debris was removed by centrifugation (100,000 ϫ g for 30 min). The supernatant was applied to a glutathione column, which was first washed with lysis buffer containing 400 mM NaCl (25 column volumes) followed by lysis buffer without NaCl (5 column volumes). The protein was eluted with 20 mM Hepes, 1 mM EDTA, 2 mM dithiothreitol, 100 mM NaCl, 5% glycerol, and 12 mM glutathione. The protein was then concentrated and dialyzed overnight in elution buffer lacking glutathione.
All G protein ␣ subunits were synthesized in E. coli and purified as described (26,27). Purified ␣ subunits were activated by incubation with [ 35 S]GTP␥S at 30°C for 30 min (G␣ s ) or 2 h (G␣ i ) (21). Free GTP␥S was removed by gel filtration.
Generation of Adenoviral Recombinants-Adenoviruses for RGS2 and RGS4 were generated as described (28,29). RGS2 and RGS4 were cloned into the adenoviral shuttle vector pACsk2.cmv using the restriction enzymes KpnI and XbaI. Both constructs contained C-terminal HA tags. To obtain recombinant viruses, HEK293 cells were co-transfected with the recombinant shuttle vector and pJM17. Viral plaques were isolated and propagated, and viral DNA was sequenced. HEK293 cells (ϳ70% confluent) were infected with HA-tagged RGS2 and RGS4 adenoviruses (multiplicity of infection ϭ 2). The cells were washed once in ice-cold phosphate-buffered saline and harvested 24 h post-infection in 1 ml of hypotonic lysis buffer (50 mM Tris, pH 7.5, 4 mM EDTA plus protease inhibitors). The cell lysates were subjected to 50 strokes of Dounce homogenization on ice, followed by low speed centrifugation for 5 min to eliminate cellular debris, and subsequently centrifuged at 100,000 ϫ g at 4°C for 20 min. The supernatant was checked for RGS protein expression using anti-HA antibody and used for AC binding assays as described below. Note that the high overexpression resulting from adenoviral expression often leads to a doublet for both RGS2 and RGS4. This is most likely due to proteolytic cleavage of the triple HA tag present at the C terminus. This does not occur with lowered expression of RGS2 and RGS4 in transient transfections.
Adenylyl Cyclase Binding Assays-Pull-down assays were conducted with pure recombinant proteins as well as with HEK293 cell lysates. The lysates were obtained by infecting cells with recombinant adenoviruses encoding RGS2 or RGS4 or by transfecting subconfluent HEK293 cells with constructs expressing wild type or mutant RGS2. Cells at 36 h post-transfection were washed once in ice-cold phosphatebuffered saline and harvested in 300 l of hypotonic lysis buffer (50 mM Tris, pH 7.5, 4 mM EDTA plus protease inhibitors), followed by Dounce homogenization and centrifugation to remove cellular debris. The lysates prepared from HEK293 cells expressing RGS proteins (50 g) were incubated with 15 g of VC 1 (670)-H6 or H6-VC 2 for 30 min and then layered over Ni-NTA-agarose beads. When purified RGS proteins were utilized, GST or GST-VC 1 (15 g) was first bound to glutathione beads. The beads were then washed and mixed with H6-RGS2 or H6-RGS4 (2 M). The reactions were incubated for 2-4 h at 4°C on a rotating mixer, washed twice with wash buffer (20 mM Hepes, pH 8.0, 2 mM MgCl 2 , 200 mM NaCl, 1 mM ␤-mercaptoethanol, and 0.1% C 12 E 10 ), boiled in Laemmli buffer for 5 min, subjected to SDS-PAGE (13% gel), and analyzed by immunoblotting employing the appropriate antibody.
Membrane Localization-HEK293 cells (ϳ70% confluent) were transfected with wild type, mutant, or truncated RGS2 constructs. After 36 h, the cells were harvested, briefly sonicated, and subjected to FIG. 2. Interactions between RGS2 and the cytoplasmic domains of type V AC. A, extracts containing HA-tagged RGS4 (R4) and RGS2 (R2) were obtained from adenovirus-infected HEK293 cells (Load). Equal amounts of cell lysates were added in the reactions and incubated with His-tagged C 2 (H6-VC 2 ) and C 1 domains (VC 1 -H6) of type V AC at 4°C for 30 min. Ni-NTA alone was incubated with RGS2 cell extract as a negative control (Ni-NTA Beads). Nontagged GTP␥S-G␣ s and nontagged GTP␥S-G␣ i were incubated with the C 2 and C 1 domain, respectively, as positive controls in the absence of cell extracts. Ni-NTA resin was then added to all of the reactions, mixed for 2 h at 4°C and washed as described under "Experimental Procedures." The proteins were eluted with SDS sample buffer, run on SDS-PAGE, and blotted with anti-HA antibody (Ab). The blots were then stripped and reprobed for G␣ i and G␣ s . The load represents 30% of the total protein present in these reactions, whereas 100% of the elution is run on SDS-PAGE. RGS2 and RGS4 were run as a doublet because of proteolytic clipping of the HA tag in adenovirus-infected cell extracts (see "Experimental Procedures"). B, equal amounts of purified His-tagged RGS4 and RGS2 (Load) were used in additional binding reactions. His-tagged RGS proteins were incubated with glutathione beads alone (Glut. Beads) or with glutathione beads bound with GST (GST-alone) or GST-fused to the C 1 domain of type V AC (GST-VC 1 ). The proteins were incubated for 1 h at 4°C, washed, eluted with SDS sample buffer, run on SDS-PAGE, and blotted with Ni-NTA-horseradish peroxidase antibody to detect both RGS2 and RGS4. The load represents 20% of the total protein used in the binding reactions, whereas 100% of the eluted RGS protein is shown. a low speed spin to remove nuclei and cellular debris. The lysate was then subjected to centrifugation (100,000 ϫ g) at 4°C for 30 min to separate membranes and cytosol. The crude membranes thus obtained were washed twice with cold phosphate-buffered saline, subjected to SDS-PAGE, and probed with anti-HA antibody for detection of RGS2 distribution in the cytosol and membrane.
Adenylyl Cyclase Activity Assays-Adenylyl cyclase activity was measured as described (30). The assays were performed at 30°C in a final volume of 50 l in the presence of 5 mM MgCl 2 . A limiting concentration of the full-length C 1 domain from type V adenylyl cyclase was reconstituted with the C 2 domain (final concentration, 0.5 M) in buffer containing 0.5 mg/ml bovine serum albumin (21). RGS proteins or control buffer were preincubated with adenylyl cyclase domains for 20 min on ice, followed by the addition of GTP␥S-G␣ s and 0.5 mM [␣-32 P]ATP to start the reaction. The reactions were terminated after 10 min, and the products were separated by sequential chromatography on Dowex-50 and Al 2 O 3 .

RESULTS AND DISCUSSION
RGS2 but Not RGS4 Directly Binds to the C 1 Domain of Type V Adenylyl Cyclase-Mammalian AC consists of a short N terminus, a set of six transmembrane spans, a cytoplasmic domain (designated C 1 ), followed by a second set of six transmembrane spans, and a C-terminal cytoplasmic domain (C 2 ) (2). Inhibition of the activities of types III, V, and VI AC reported earlier (3) could be attributed to the direct binding of RGS2 with either cytoplasmic domain of the enzyme. To demonstrate that RGS2 directly binds to AC, we performed binding assays using hexahistidine-tagged C 1 and C 2 domains of type V AC. These cytoplasmic domains are fully active when reconstituted with each other and retain most of the properties of the native enzyme (21,(31)(32)(33). An excess of His-tagged C 1 and C 2 domains from type V AC were incubated with HEK293 extracts containing C-terminal HA-tagged RGS2 or RGS4 ( Fig. 2A). Nontagged GTP␥S-G␣ i or GTP␥S-G␣ s were used as positive controls for the C 1 and C 2 domains, respectively. Ni-NTA resin was used to pull down the interacting proteins that were eluted with SDS sample buffer, followed by immunoblotting for the respective proteins. Approximately 30% of the HA-tagged RGS2 present in the HEK293 extracts was bound to the C 1 domain of type V AC. No binding of RGS4 to the C 1 domain was observed, although the C 1 domain tightly bound G␣ i . Neither RGS2 nor RGS4 bound to Ni-NTA beads alone or to the C 2 domain, although this protein was capable of binding G␣ s (22).
Because the above experiment utilized HEK293 extracts, it was still possible that a second protein was required to mediate this binding event. Therefore, we tested the direct interaction of E. coli purified His-tagged RGS proteins with the C 1 domain fused to GST versus GST alone. Glutathione resin was used to pull down the interacting proteins that were detected by immunoblotting with the Ni-NTA-horseradish peroxidase antibody or an antibody directed against GST. Because of the limited expression of GST-VC 1 , an excess of purified RGS proteins was used in these reactions. RGS2 associated with GST-VC 1 but not with GST alone (Fig. 2B). No interaction was observed with RGS4. These results further support our hypothesis of a direct interaction between RGS2 with type V AC.

RGS2 Inhibits Type V AC in Vivo Independent of RGS2 GAP Activity-Previous studies have shown that RGS2 inhibits G␣ sinduced type III AC activation in vivo (3). To verify that RGS2
inhibits G␣ s -induced type V activation in vivo, we transfected HEK293 cells with the expression vectors for type V AC and a GTPase-deficient form of G␣ s , G␣ s -Q227L, in the presence or absence of RGS2. Co-expression of G␣ s -Q227L and type V AC increased cAMP accumulation 3-fold (Fig. 3A). Additional expression of RGS2 reduced G␣ s -Q227L-stimulated cAMP accumulation to near basal levels. A similar result was obtained with nontagged RGS2, ruling out any possible unwarranted effects of the HA tag (data not shown). The inhibition by RGS2 is unusual because many RGS proteins including RGS1, RGS4, and GAIP have no effect or show increased cAMP accumulation (34 -36). However, RGS-PX1, RGS3, and RGS13 have also been reported to have inhibitory effects on cAMP accumulation by either G␣ s GAP activity or other unknown mechanisms (36 -38).
To test the role of RGS GAP activity, we have utilized a mutation within the highly conserved GAP domain. Replacement of Asn 128 with Ala in RGS4 reduces GAP activity and binding to G proteins by over 3 orders of magnitude (39,40). Structural data also indicates that Asn 128 in RGS4 plays a crucial role in the interaction of RGS and G proteins (41). We made the corresponding mutation in RGS2, Asn 149 to Ala, and examined its effect on cAMP production. This mutant form of RGS2 also suppressed cAMP accumulation to the same level as wild type RGS2 (Fig. 3A). Hence, a mutation created at the G␣-RGS binding interface in RGS2 that disrupts GAP activity does not limit its ability to inhibit cAMP. However, deletion of the N-terminal amino acids of RGS2 (⌬NT1, ⌬NT2, and ⌬NT3) abolished its ability to inhibit cAMP generated by type V AC. Even the shortest deletion of 19 amino acids (⌬NT1) was sufficient to eliminate RGS inhibitory activity (Fig. 3A). Wild type, mutant, and truncated forms of RGS2 were all expressed to similar levels as detected by immunoblotting using anti-HA antibodies.
Similar results were observed in HEK293 cells, transfected with type V AC and the ␤ 2 -adrenergic receptor (␤ 2 AR). Wild type and mutant (N149A) RGS2 inhibited isoproterenol-stimulated cAMP accumulation to a similar extent (Fig. 3B), whereas N-terminal truncations of RGS2 displayed no significant inhibition.
RGS2 and RGS2-N149A but Not ⌬NT1 Bind to the C 1 Domain of AC-Using binding assays as described in Fig. 2A, we assayed whether mutant and truncated RGS2 proteins were deficient in binding the C 1 domain from type V AC. Cell extracts from HEK293 cells expressing HA-tagged proteins were incubated with His-tagged C 1 domain. Strong interactions were observed with the C 1 domain for both wild type and mutant RGS2-N149A but not the ⌬NT1 truncated protein (Fig.  4). Incubation with the C 1 protein is sufficient to completely clear the cytosol of wild type and N149A-RGS2 protein, whereas almost all of the ⌬NT1 truncated protein remains. Hence, deletion of the N-terminal 19 amino acids is sufficient to eliminate detectable interactions between RGS2 and AC.
The N-terminal 19 Amino Acids of RGS2 Are Sufficient for Inhibition of AC-The RGS2 box domain is located from amino acids 72-199, as defined by structural homology to the RGS4 FIG. 4. Binding of wild type, mutant (N149A), and N-terminal  truncated (⌬1-19) RGS2 with the C 1 domain of type V AC. HAtagged wild type (Wt), mutant (N149A), and truncated RGS2 (NT1, ⌬1-19) were obtained from transfected HEK293 cells. Equal amounts of protein were loaded in the reactions (LOAD). His-tagged C 1 domain of type V AC was incubated with extracts containing HA-tagged wild type, mutant and truncated RGS2 as described in the legend to Fig. 2A. Unbound (SUPERNATANT) and the eluted bound proteins (VC 1 -H6 Elution) were run on SDS-PAGE and blotted with HA-antibody. The samples represent 100% of the load, supernatant, and the eluted bound proteins present in the reactions.
box (41). Zeng et al. (42) have shown that the N-terminal residues (1-33) of RGS4 confer high affinity and receptor selective signaling via G q . Deletion of the N-terminal domain decreased the potency of RGS4 inhibition of G q signaling by 10 4 -fold. RGS2 and RGS4 share a limited homology outside the RGS box from amino acid 42-72, but the extreme N terminus of RGS2 is 20 amino acids longer and displays no homology with the N terminus of RGS4. Although the N-terminal domain of RGS2 appeared to be a likely candidate to interact with AC, we could not rule out additional sites of interaction.
RGS2 contains 12 amino acids at the C terminus that are unique to this RGS protein. In addition, residues within the RGS core domain could also play a role. This domain shares many hydrophobic amino acids among all RGS family mem-  6. Alanine scanning mutations of the N-terminal 19 amino acids of RGS2: cAMP accumulation, localization, and binding assays. A, a series of mutations were constructed within the N-terminal 19 amino acids of RGS2. Groups of three amino acids were mutated to alanine as shown. B, cAMP production was measured in HEK293 cells transfected as indicated with type V AC, G␣ s -Q227L, and wild type (Wt) and mutant RGS2 constructs (RGS2-MFL, RGS2-VQH, RGS2-DCR, and RGS2-PMD). The protein expression levels were determined by immunoblotting using anti-HA antibody. The data are expressed as the means Ϯ S.E. from two different experiments each performed in duplicate. C, HA-tagged wild type and two mutants of RGS2 (N149A and VQH) were obtained from transfected HEK293 cells. Equal amounts of each protein was loaded in the reactions (LOAD). His-tagged C 1 domain of type V AC was incubated with all three protein extracts at 4°C for 30 min. Unbound proteins (SUPERNATANT) and the bound eluted proteins (VC 1 -H6 Elution) were run on SDS-PAGE and blotted with HA-antibody. The load and supernatant represent 60% of the protein present in the reaction, whereas 100% of the VC 1 -H6 eluted protein is shown. D, the subcellular localization of wild type, mutant (RGS2-VQH), and truncated (⌬NT1) RGS2 was determined as described under "Experimental Procedures." Equal amounts of protein from whole cell lysate (LOAD), cytosol, and membrane fractions were loaded on SDS-PAGE and detected by immunoblotting. bers; however, significant differences between RGS2 and RGS4 lie within ␣ helices 3 and 6 of the RGS fold (41). To determine the precise regions of RGS2 critical for inhibition of type V AC, we created two additional truncations of RGS2. RGS2-⌬box is devoid of the RGS core domain and contains only the N-terminal 71 amino acids. RGS2-⌬C lacks the 12amino acid overhang at the C terminus. Type V AC and G␣ s -Q227L were co-expressed in the presence or absence of constructs directing the expression of full-length RGS2 and the ⌬box and ⌬C truncations in HEK293 cells (Fig. 5A). Both truncations inhibited cAMP production to the same extent as wild type RGS2, indicating that the N terminus is sufficient for RGS2-mediated inhibition with no significant contributions from the RGS box or C terminus. Similar results were obtained by co-transfection with the ␤ 2 AR and stimulation with isoproterenol (Fig. 5B).
Mutational Analysis of RGS2 Identifies a Series of Residues Important for Inhibition of Type V AC-The N-terminal 19 amino acids of RGS2 clearly play a pivotal role in inhibiting type V AC-generated cAMP production. We conducted a limited alanine scanning analysis of the first 19 amino acids in the N-terminal region of RGS2 to identify residues important for RGS2 function. Sets of three residues along the 19-amino acid stretch were mutated to alanine as shown in Fig. 6A. HEK293 cells were transfected with expression vectors for type V AC and G␣ s -Q227L, in the presence of full-length wild type RGS2 or mutant RGS2 (MFL, VQH, DCR, and PMD). With the exception of VQH, the RGS2 mutants displayed similar inhibition FIG. 7. Mutations in the G␣ i -binding site of type V AC prevent inhibition by carbachol but have no effect on inhibition by RGS2. A, production of cAMP was measured in HEK293 cells transfected as indicated with G␣ s -Q227L and wild type and mutant type V AC (E411A and L472A) constructs in the presence or absence of carbachol (1 M) for 15 min. B, cAMP production was measured in HEK293 cells transfected with constructs directing the expression of wild type and mutant type V AC (E411A and L472A) and G␣ s -Q227L in the presence and absence of wild type RGS2 (1 g/well). The protein expression levels were determined by immunoblotting using anti-HA antibody (data not shown). The data are expressed as the means Ϯ S.E. from three different experiments each performed in duplicate. C, cAMP production was measured in HEK293 cells transfected with constructs as described for B, except that a variable amount of RGS2 DNA was used (25 ng, 50 ng, 1 g, and 2 g). The data are shown as percentages of the fold activation in the absence of RGS2 for each type V AC construct, where the fold activation is simply the level of cAMP of type V AC plus G␣ s -Q227L as compared with G␣ s -Q227L alone. The data are expressed as the means Ϯ S.E. from a single experiment (n ϭ 2) and are representative of two different experiments each performed in duplicate. D, wild type (q, 60 nM), L472A (, 80 nM), and E411A (f, 225 nM) mutant VC 1 (670) was reconstituted with 0.5 M VC 2 and assayed with 400 nM activated G␣ s in the presence of the indicated concentrations of purified RGS2. The activities are expressed as percentages of control values (1630, 1950, and 240 nmol/min mg for wild type and L472A and E411A mutant VC 1 (670), respectively) and are representative of two different experiments performed in duplicate. wt, wild type. of cAMP accumulation as wild type RGS2 (Fig. 6B). The mutation VQH displayed no inhibition of cAMP production by type V AC, although it expressed as well as if not better than any of the other mutant proteins. In addition to blocking inhibition of cAMP production by RGS2, mutation of the residues VQH to alanine also prevented binding to the C 1 domain of type V AC (Fig. 6C). The N-terminal 19-amino acid stretch of RGS2 displays no homology with RGS3 or RGS13, which have been reported to impair G s signaling (36,37), nor with any other mammalian protein sequence in the GenBank TM data base.
Membrane Localization of Wild Type, ⌬NT1 (1-19) RGS2, and RGS2-VQH-Localization can be a key determinant of RGS2 function (10). One trivial explanation for the loss of inhibition by RGS2-⌬NT1 and the mutation of residues VQH would be a possible mislocalization of the truncated or mutated protein to regions away from the plasma membrane. Earlier reports identified amino acid residues 33-67 as an amphipathic ␣-helical domain that was sufficient for plasma membrane targeting (10). To rule out possible additional effects from the N-terminal 19 amino acids, we fractionated HEK293 cells expressing wild type, ⌬NT1 truncated RGS2, and the mutant RGS2-VQH. All three forms of RGS2 were present in the plasma membrane and in the cytosol of HEK293 cells (Fig.  6D). Therefore, it is unlikely that gross mislocalization of the truncated or mutant RGS2 proteins results in a loss of activity.

RGS2 Does Not Interact with the G␣ i -binding Site on Type V AC and Is Independent of Inhibition by G␣ i -Because RGS2
bound to the C 1 domain of type V AC, we tested whether RGS2 may utilize the G␣ i inhibitory binding site. We made use of two mutations in type V AC, ACV-E411A and ACV-L472A, that showed a 60-fold reduction in the IC 50 for G␣ i -mediated inhibition of type V AC (21). The G␣ i -binding site was defined as a cleft formed by the ␣2 and ␣3 helices of the C 1 domain. The amino acids Glu 411 and Leu 472 reside on the ␣2 and ␣3 helices, respectively. Mutation of these residues had no effect on the activation of type V AC by G␣ s , the synergy with forskolin, or the K m for ATP. Although these mutant proteins were extensively characterized in vitro, they had not been tested in vivo. Therefore, we transfected constructs expressing wild type or the mutant type V AC with G␣ s -Q227L and tested for their ability to be inhibited by the muscarinic agonist, carbachol (Fig. 7A). As expected, the wild type AC was inhibited upon the addition of carbachol, but neither mutant of type V AC was affected. Hence, these proteins are deficient in G␣ i inhibition both in vitro and in vivo.
The E411A and L472A mutations were next tested for their potential roles in RGS2-mediated inhibition. HEK293 cells were transfected with G␣ s -Q227L and the constructs directing the expression of wild type or mutant type VAC, in the presence or absence of RGS2. Both mutant proteins were inhibited by RGS2 expression to a similar extent as wild type AC (Fig. 7B). However, these experiments were performed with maximal levels of RGS2. To determine whether the mutations at the G␣ i -binding site had any effects on the IC 50 for RGS2-mediated inhibition, increasing concentrations of DNA-encoding RGS2 were used, resulting in a ϳ100-fold range of RGS2 expression as analyzed by Western blotting (data not shown). In this case, the fold activation of type V AC by G␣ s -Q227L is shown relative to G␣ s -Q227L alone. Once again, the degree of inhibition by RGS2 is very similar for wild type and L472A and E411A mutants of type V AC across a large range of cellular RGS2 concentrations (Fig. 7C).
Finally, the IC 50 for inhibition of the cytoplasmic domains (C 1 and C 2 ) of type V AC by purified RGS2 was determined for wild type C 1 and L472A and E411A mutant C 1 domains upon reconstitution with the C 2 domain from type V AC (Fig. 7D).
RGS2 inhibits G␣ s -stimulated AC activity ϳ70% with very similar IC 50 values for the wild type and mutant C 1 domains (0.4 -1.0 M). Therefore, the Glu 411 and Leu 472 residues, critical for G␣ i inhibition, are not required for inhibition by RGS2. Thus, RGS2 most likely utilizes a binding site on the C 1 domain of type V AC that is distinct from the G␣ i site of interaction. Although RGS2 is generally a poor GAP for G␣ i , it is possible that AC may promote such interactions leading to either reduced inhibition by G␣ i or possibly some type of synergy between the two regulators. Under conditions where type V AC was not fully inhibited by RGS2, no synergistic effects were observed with both carbachol and RGS2 (data not shown). Hence, inhibition by RGS2 appears to be independent of G␣ i regulation.
Conclusion-AC is regulated by many different signals. The present study reinforces the important link between RGS proteins and cellular signaling and identifies a novel binding site for the interaction of RGS2 and a downstream effector protein, type V AC. Further studies are required to completely understand the mechanism of this inhibition and the exact site of interaction on the C 1 domain of type V AC. Our appreciation of RGS proteins as more than simply inhibitors of G protein signaling is ever expanding. The direct link between RGS2 and AC is yet another example of the diverse roles that RGS proteins play in cellular signaling.