Rescue of Functional Interactions between the α2A-Adrenoreceptor and Acylation-resistant Forms of Gi1α by Expressing the Proteins from Chimeric Open Reading Frames*

Co-expression of the α2A-adrenoreceptor with a pertussis toxin-resistant (C351G), but not with an also palmitoylation-resistant (C3S/C351G), form of the α subunit of Gi1 resulted in agonist-induced, pertussis toxin-independent, GTP hydrolysis. Construction and expression of a chimeric fusion protein between the receptor and C351G Gi1α generated a membrane protein in which the G protein element was activated by receptor agonist. An equivalent fusion protein containing C3S/C351G Gi1α rescued the ability of receptor agonist to activate this mutant. Fusion proteins of a palmitoylation-resistant (C442A) α2A-adrenoreceptor and either C351G or C3S/C351G Gi1α also responded effectively to agonist. Myristoylation resistant (G2A/C351G) and combined acylation-resistant (G2A/C3S/C351G) mutants of Gi1α are cytosolic proteins. Expression of these as chimeric α2A-adrenoreceptor-G protein fusions restored membrane localization and activation of the G protein by receptor agonist. These studies demonstrate the general utility of generating chimeric fusion proteins to examine receptor regulation of G protein function and that the lack of functional activation of acylation-negative G proteins by a co-expressed receptor is related to deficiencies in cellular targeting and location rather than an inherent incapacity to produce appropriate protein-protein interactions and signal transmission.

A major mechanism for signal transduction across the plasma membrane involves seven transmembrane element G protein-coupled receptors (GPCRs) 1 and their activation of members of the family of ␣␤␥ heterotrimeric guanine nucleotide binding proteins (G proteins) (1)(2). Unlike the GPCRs, none of the individual G protein subunits contain transmembrane-spanning elements although the proteins are membrane-associated. In the case of the ␤␥ complex, post-translational prenylation by the C15 farnesyl or C20 geranylgeranyl groups at a cysteine residue close to the C-terminal tail of the ␥ subunit followed by protein trimming and carboxymethyla-tion acts to anchor the complex to the membrane (3)(4).
A key role in the targeting of specific G protein ␣ subunits to the plasma membrane and their maintenance there has been ascribed to both co-translational myristoylation and posttranslational palmitoylation (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13). These acylations may also contribute to protein-protein interactions between the G protein ␣ subunit and both receptors and the G protein ␤␥ complex (12,14). Addition of myristate occurs only on the ␣ subunits of the G i -family of G proteins because they contain the consensus sequence (MGXXXS) for the enzyme N-myristoyl-CoA transferase. The glycine that is found at codon 2 acts as the acceptor when it is exposed following removal of the initiator methionine. Palmitoylation of either one or two cysteine residues within the first ten amino acids of the ␣ subunits occurs on all of the widely expressed G proteins (4,11).
Although receptors can often be resolved from their cognate G proteins by detergent solubilization of membranes, in a number of examples such solubilization of membranes can result in the maintenance of receptor-G protein contacts (15)(16)(17). The term "precoupled" is often used to describe such avid interactions between receptors and their cognate G protein. As a strategy to create physical precoupling of a receptor and its cognate G protein ␣ subunit, Bertin et al. (18) linked together the ␤ 2 -adrenoreceptor and G s ␣ by generating a fusion protein between these open reading frames. This construct could be expressed, and addition of agonist for the receptor resulted in stimulation of adenylyl cyclase activity (18). We have recently expanded this approach to the ␣ 2A -adrenoreceptor and the inhibitory G protein G i1 ␣ and demonstrated that the addition of agonist to a fusion protein containing the wild-type sequences of both proteins results in activation of the G protein (19).
Although independent co-expression of the ␣ 2A -adrenoreceptor and G i1 ␣ in COS cells allows functional interactions between the receptor and this G protein (20), no functional contacts can be measured following co-expression of the ␣ 2Aadrenoreceptor and acylation-deficient mutants of G i1 ␣ (21) even though at least a C3S mutant can be localized to the membrane. We now demonstrate that agonist-induced signal transduction to the acylation-deficient mutants of G i1 ␣ can be rescued by expressing these polypeptides as fusion constructs with the receptor and that effective agonist-induced signal transduction is unaffected by the palmitoylation potential of the receptor. Oligonucleotides were synthesized on a Millipore Expedite Nucleic Acids Synthesis System.

Materials-All
Construction of the ␣ 2A -Adrenoreceptor-C351G G i1 ␣ Fusion Constructs-A pertussis toxin-resistant C351G form of rat G i1 ␣ was generated (20) and linked to the porcine ␣ 2A -adrenoreceptor (22). To do so, the ORF of the ␣ 2A -adrenoreceptor DNA was amplified by polymerase chain  reaction using the oligonucleotides: sense, 5Ј-TTGGTACCATGTATC-CTTACGACGTTC-3Ј, and antisense, 5Ј-AAGAATTCCATGGCGATC-CGTTTCCTGTCCCCACGGC-3Ј (restriction sites for KpnI, EcoRI, and NcoI are underlined). The polymerase chain reaction-amplified fragment was digested with KpnI and EcoRI and ligated to pBluescript (Stratagene) through these restriction sites. Introduction of the NcoI site at the 3Ј-end of the ORF resulted in the C-terminal amino acid of the receptor being altered from Val to Ala and removal of the stop codon. The rat C351G G i1 ␣ cDNA contains two NcoI sites, one straddling the ATG start codon and the other 268 bp downstream from this. This 268-bp fragment was removed from C351G G i1 ␣ in pBluescript by digestion with NcoI, and the remaining C351G G i1 ␣ pBluescript cDNA was religated. The shortened cDNA was excised from pBluescript with EcoRI and cloned into the EcoRI site of the ␣ 2A -adrenoreceptor in pBluescript, adjacent to the 3Ј-end of the receptor ORF. The 268-bp fragment was then inserted between the NcoI sites at the 3Ј-end of the ␣ 2A -adrenoreceptor ORF and at the 5Ј-end of the C351G G i1 ␣ ORF. This resulted in production of an in-frame construct whereby the 3Ј-end of the ␣ 2A -adrenoreceptor ORF was exactly adjacent to the 5Ј-end of the C351G G i1 ␣ ORF. The full fusion construct was then excised from pBluescript with KpnI and EcoRI and ligated into the eukaryotic expression vector pCDNA3. The same strategy was used to generate the various acylation-deficient chimeras.
Cell Culture and Transfection-COS-7 cells were maintained in Dulbecco's modified Eagle's medium containing 10% (v/v) fetal calf serum, 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin. Cells were seeded in 60-mm culture dishes and grown to 60 -80% confluency (18 -24 h) prior to transfection with pcDNA3 containing the relevant cDNA species using LipofectAMINE reagent (Life Technologies, Inc.) (20). For transfection, 2.5-2.8 g of DNA were mixed with 10 l of LipofectAMINE in 0.2 ml of Opti-MEM (Life Technologies, Inc.) and incubated at room temperature for 30 min prior to the addition of 1.8 ml of Opti-MEM. COS-7 cells were exposed to the DNA/Lipo-fectAMINE mixture for 5 h. 2 ml of 20% (v/v) fetal calf serum in Dulbecco's modified Eagle's medium were then added to the cells. Cells were harvested 48 h after transfection. In all studies, the cells were treated with pertussis toxin (25 ng/ml) 24 h prior to cell harvest to eliminate possible interactions of the fusion proteins with endogenously expressed G i -family G proteins (19 -20).
Preparation of Membranes-Plasma membrane-containing P2 particulate fractions were prepared from cell pastes that had been stored at Ϫ80°C following harvest. Cell pellets were resuspended in 0.5 ml of 10 mM Tris-HCl, 0.1 mM EDTA, pH 7.5 (buffer A), and rupture of the cells was achieved with 50 strokes of a hand-held Teflon on-glass homogenizer followed by passage (10 times) through a 25-gauge needle. Cell lysates were centrifuged at 1000 ϫ g for 10 min in a Beckman TJ-6 centrifuge to pellet the nuclei, and unbroken cells and P2 particulate fractions were then recovered by centrifugation of the supernatant at 200,000 ϫ g for 30 min in a Beckman TL 100 bench-top ultra-centrifuge using a Beckman TLA 100.2 rotor. P2 particulate fractions were resuspended in buffer A and stored at Ϫ80°C until required.
[ 3 H]RS-79948 -197 Binding Studies-Binding assays were initiated by the addition of 2-4 g of protein to an assay buffer (10 mM Tris-HCl, 50 mM sucrose, 20 mM MgCl 2 , pH 7.5) containing [ 3 H]RS-79948 -197 (23) (1 nM). K d for this ligand at the fusion protein derived from the wild-type receptor and G protein ϭ 0.35 nM (20). Nonspecific binding was determined in the presence of 100 M idazoxan. Reactions were incubated at 30°C for 45 min, and bound ligand was separated from free by vacuum filtration through GF/C filters. The filters were washed 3 times with 5 ml of assay buffer, and bound ligand was estimated by liquid scintillation spectrometry.
Immunological Studies-Antiserum I1C (24) was produced in a New Zealand White rabbit using a conjugate of a synthetic peptide corresponding to amino acids 160 -169 of the G i1 ␣ subunit and keyhole limpet hemocyanin (Calbiochem) as antigen. The specificity of this antiserum for G i1 ␣ has been demonstrated previously (24). Membrane samples were resolved by SDS-polyacrylamide gel electrophoresis using 10% (w/v) acrylamide gels containing 6 M urea overnight at 100 V. Proteins were subsequently transferred to nitrocellulose (Schleicher & Schuell), probed with relevant antiserum, and visualized as described (25).
High Affinity GTPase Assays-High affinity GTPase assays were performed essentially as described previously (26)

RESULTS
Expression of the porcine ␣ 2A -adrenoreceptor (22) in COS-7 cells allows agonist activation of endogenously expressed G proteins as measured by the capacity of the ␣ 2 -adrenoreceptor agonist UK14304 to stimulate high affinity GTPase activity in membranes of these cells (20). Pertussis toxin treatment prior to cell harvest attenuates this effect, demonstrating it to reflect activation of G i -family G proteins (20). A C351G mutation was introduced into G i1 ␣ to render this protein insensitive to pertussis toxin (20). Co-expression of this modified G protein with the ␣ 2A -adrenoreceptor in COS-7 cells, followed by treatment of the cells with pertussis toxin, now resulted in a robust stimulation of high affinity GTPase activity upon addition of UK14304 (Fig. 1). A C3S mutation of G i1 ␣ prevents this polypeptide from acting as a substrate for post-translational palmitoylation (9 -10). Co-expression of C3S/C351G G i1 ␣ and the ␣ 2A -adrenoreceptor failed to allow UK14304 stimulation of the GTPase activity of this mutated G protein (Fig. 1) although high levels of the G protein can be expressed in the membrane preparation (21). A G2A mutation of G i1 ␣ prevents co-translational myristoylation of this protein (5)(6), and the combination of G2A and C3S mutations totally eliminates acylation of the expressed polypeptide (9) and prevents membrane attachment. Co-expression of these forms of G i1 ␣, which also contained the C351G mutation with the ␣ 2A -adrenoreceptor, did not result in UK14304 stimulation of high affinity GTPase activity (21) as they were virtually entirely cytosolic (data not shown, but see Ref. 21).
The open reading frames of each of C351G G i1 ␣, C3S/C351G G i1 ␣, G2A/C351G G i1 ␣, and G2A/C3S/C351G G i1 ␣ were fused in-frame to the wild-type ␣ 2A -adrenoreceptor cDNA to create chimeric fusion proteins that contained the entire G protein ␣ subunit sequence downstream of the receptor. This process introduced a single substitution of the receptor C-terminal Val by Ala and the maintained presence in the fusion proteins of the initiator Met of G i1 ␣ (Fig. 2).  (23). There were no differences in levels of expression of the individual fusion proteins in concurrently performed transfections (data not shown, but see Fig. 5). Immunoblotting membranes and cytosolic fractions derived from these transfections with the G i1 ␣ specific antiserum I1C (24) resulted in identification of low levels of the endogenously expressed G i1 ␣ as a 41-kDa polypeptide in the membrane fraction of each of mock-transfected cells and all the specific transfects (Fig. 3). An ϳ100-kDa I1C immunoreactive band of the expected size for the fusion proteins was detected only in positively transfected cells (Fig. 3).
The porcine ␣ 2A -adrenoreceptor has a cysteine residue 9 amino acids from the C terminus that has been shown to act as an acceptor for post-translational palmitoylation (27). Although a C442A mutation of the receptor has been reported not to interfere with G protein activation (27), we considered whether potential palmitoylation of this cysteine might alter agonist regulation of the fusion proteins as palmitoylation is believed to provide an extra site of membrane anchorage and generate a "fourth intracellular loop" in the receptor (11). Furthermore, the C-terminal tail of the ␣ 2A -adrenoreceptor is only some 20 amino acids long, which is relatively short within the family of G protein-coupled receptors, and we wished to consider if the extra flexibility following removal of the site of palmitoylation might influence regulation of fusion protein activation by agonist. Fusion proteins containing the pertussis toxin-insensitive C351G G i1 ␣ in concert with G2A, C3S, and  1 and 2) and those expressing wild-type (lanes 3-10) or C442A (lanes 11-18) ␣ 2A -adrenoreceptor fusion proteins containing C351G G i1 ␣ (lanes 3, 4, 11, and 12), C3S/C351G G i1 ␣ (lanes 5, 6, 13, and 14), G2A/C351G G i1 ␣ (lanes 7, 8, 15, and 16), or G2A/C3S/C351G G i1 ␣ (lanes 9, 10, 17, and 18) were resolved into P2 particulate (even numbered lanes) and cytosolic (odd numbered lanes) preparations, resolved by SDS-polyacrylamide gel electrophoresis, and immunoblotted to detect the presence of G i1 ␣. In cells expressing each of the fusion proteins, a polypeptide of some 100 kDa was detected.  FIG. 2. Construction of ␣ 2A -adrenoreceptor-G i1 ␣ fusion proteins. Fusion of the N terminus of C351G G i1 ␣ to the C terminus of the ␣ 2A -adrenoreceptor resulted in the receptor C-terminal amino acid (valine) being converted to alanine and maintenance of the methionine, which normally functions as initiator in G i1 ␣. Arrows represent the amino acids altered from the wild-type sequence in the various fusion proteins generated and examined. The marked cysteine residues are the known target sites for post-translational palmitoylation in both the receptor and the G protein.
G2A/C3S mutations were constructed with the C442A ␣ 2Aadrenoreceptor (Fig. 2) and expressed in COS-7 cells. Each of these constructs was expressed to a similar level as those containing the wild-type receptor (Fig. 3), and each was able to stimulate pertussis toxin-insensitive high affinity GTPase activity in response to UK14304 to similar extents and with similar EC 50 values (2.2-2.8 ϫ 10 Ϫ7 M in individual transfections) as the fusion proteins containing the wild-type receptor sequence (Fig. 4B).
Co-expression of the ␤ 1 ␥ 2 complex with the wild-type ␣ 2Aadrenoreceptor-C351G G i1 ␣ fusion protein resulted in substantially greater maximal stimulation of GTPase activity by UK14304 than achieved without excess ␤ 1 ␥ 2 (Fig. 5A). This did not reflect increased levels of expression of the fusion protein ( Fig. 5B) or alterations in the EC 50 for UK14304 (data not shown but see (19)). Additional expression of ␤ 1 ␥ 2 also increased the maximal GTPase activity in response to UK14304 of each of the G2A/C351G, C3S/C351G, and G2A/C3S/C351G forms of G i1 ␣ constrained in fusion proteins with the wild-type ␣ 2A -adrenoreceptor (Fig. 5A), consistent with interaction of the ␤ 1 ␥ 2 complex with all of the individual fusion proteins. This is despite the known role of the N-terminal region of the G protein ␣ subunit, which is constrained in the fusion constructs, in ␤␥ binding (28 -29). DISCUSSION Acylation of G protein ␣ subunits plays a key role in the membrane targeting and association of these proteins (3)(4)11). Furthermore, prevention of myristoylation of the ␣ subunit of G i -like G proteins reduces their affinity of interaction with the ␤␥ complex and can render them cytoplasmic (5)(6). Prevention of palmitoylation also limits membrane association and has been reported to limit or prevent interactions with receptors (7,9,12,14). However, it has been unclear whether poor activation of acylation-defective G protein ␣ subunits by receptors is simply a problem associated with the lack of appropriate targeting and thus proximity to a receptor or is inherently due to the acylation status of the G protein.
In the current study, we have taken a highly novel approach to address this question. This has involved the construction of chimeric fusion proteins between the ␣ 2A -adrenoreceptor and the ␣ subunit of the G protein G i1 . To do so involved the apparently simplistic expedient of linking the N terminus of the G protein directly to the C terminus of the receptor. The strategy used resulted in a minimal alteration to the sequence of the protein in the region of fusion. Indeed, only the Cterminal amino acid of the receptor was altered (Val to Ala), and the initiator Met of the G protein, which would normally be removed, remained in the sequence of the new protein (see Fig.  2). Only a single previous example has examined the potential for signal transduction following expression of a GPCR fused to its cognate G protein ␣ subunit. In the case of the ␤ 2 -adrenoreceptor-G s ␣ chimera, addition of agonist was able to cause activation of adenylyl cyclase following expression of the fusion protein in a cell line that genetically lacks endogenous G s ␣ (18).
As well as using the wild-type receptor and G protein, fusion chimeras were generated between the wild-type ␣ 2A -adrenoreceptor and mutant forms of G i1 ␣ in which the sites that are normally palmitoylated and myristoylated in the G protein were modified. Expression of each of these constructs could be detected in the membrane fraction following transient transfection of each of these in COS-7 cells. This was achieved both in binding assays using the highly selective and high affinity ␣ 2 -adrenoreceptor antagonist [ 3 H]RS-79948 -197 (23) to detect the ligand binding site of the receptor (Fig. 5) and by immunoblotting cytosolic and membrane fractions of these cells with the specific G i1 ␣ antiserum I1C (24) (Fig. 3). G i1 ␣ is expressed endogenously in COS-7 cells at low levels and could be detected as a 41-kDa polypeptide by this antiserum in membranes from both mock and positively transfected cells. In contrast, membranes of positively transfected but not mock-transfected cells displayed the presence of an I1C reactive polypeptide of some 100 kDa, which corresponds to the chimeric fusion proteins (Fig. 3). Similar levels of the fusion proteins were expressed whether the G protein element of the fusion protein was wild type at the N terminus or contained G2A, C3S, or both mutations (Fig. 3), and none of these immunoreactive proteins were detected in the cytosolic fractions.
As with many GPCRs (11), the ␣ 2A -adrenoreceptor is also a target for post-translational palmitoylation, at Cys442 within the C-terminal tail. Because this acylation is proposed to create a "fourth intracellular loop" in the receptor structure and because this receptor has only a short C-terminal tail, we also created chimeric fusion proteins between a C442A mutant of the ␣ 2A -adrenoreceptor and the various forms of G i1 ␣ detailed above. Although previous studies have indicated this mutation in the receptor not to interfere with agonist-mediated G protein activation, we wished to assess whether potential greater flexibility provided by a longer "linker" between the seventh transmembrane element of the receptor and the N terminus of the G protein ␣ subunit would affect agonist activation of the G protein. All of the C442A-␣ 2A -adrenoreceptor-G i1 ␣ fusions were also expressed and to similar levels as the versions that included the wild-type receptor (Fig. 3).
As an aid to subsequent analysis, we generated all of the fusion proteins using a form of G i1 ␣ in which Cys-351, which is the normal target for pertussis toxin-catalyzed ADP-ribosylation, was substituted by Gly. We have previously demonstrated that C351G G i1 ␣ is not a substrate for pertussis toxin-catalyzed ADP-ribosylation (20). It can, however, be activated by the ␣ 2A -adrenoreceptor, with the only clear difference, compared with the wild-type G protein, being that some 10 -15-fold higher concentrations of agonist are required to produce the same degree of stimulation (20). This is presumably a reflection of the alteration in conformation of the C terminus of the G protein, an element known to be a key receptor contact site (30). Such mutations, however, allow receptor regulation of the mutated G protein to be studied in isolation following pertussis toxin-treatment of cells to eliminate potential functional contacts between the receptor and the endogenously expressed G i -family G proteins (20,31,32). As such, all of the functional experiments herein were performed following pertussis toxin treatment of the cells for times and with amounts of toxin that cause modification of all of the endogenous G i -like G proteins (20).
All of the expressed chimeric fusion proteins were able to stimulate high affinity pertussis toxin-insensitive GTPase activity in membranes of the transfected COS-7 cells upon addition of the ␣ 2 -adrenoreceptor agonist UK14304. Agonist stimulation of high affinity GTPase activity is a classical assay to measure both GDP-GTP exchange and then subsequent GTP hydrolysis by a G protein ␣ subunit (25)(26). As such, these results demonstrate both that the G protein ␣ subunit is folded appropriately to exchange guanine nucleotide and to act as an enzyme and that information, presumably mediated via conformational change, can be transmitted from the binding of agonist to the receptor on to the G protein. These data thus further demonstrate the general utility of generating chimeric fusion proteins to examine receptor regulation of G protein function (18 -19).
G2A and G2A/C3S mutants of G i1 ␣ are essentially completely cytosolic (9), and thus it is not surprising that they are not activated by receptor agonists (21). Physical proximity, as guaranteed from the fusion protein approach, showed that the lack of functional activation of such acylation negative mutants of G proteins by a co-expressed receptor is related to deficiencies in cellular targeting and location rather than an inherent incapacity to produce appropriate protein-protein interactions for signal transmission. As the fusion chimeras containing the G2A and C3S mutants required similar concentrations of UK14304 to cause half-maximal effects, this also argues that acylation is not inherently required to produce effective protein-protein contacts between the receptor and G protein ␣ subunit but rather is to position the G protein appropriately. It was noted, however, that the fusion constructs containing both the G2A and C3S mutations in G i1 ␣ displayed somewhat lower GTPase activity in response to maximally effective concentrations of UK14304 rather than the other constructs (Fig. 4). The reason for this requires further investigation. The C3S mutation of G i1 ␣ results in expression of a protein that is partially cytoplasmic but partially membrane-associated (9). As such, by increasing levels of expression of C3S G i1 ␣, it is possible to obtain levels of this mutant at the membrane as high or higher than that following expression of wild-type G i1 ␣ (21). Despite this, the independently expressed ␣ 2A -adrenoreceptor is unable to cause significant activation of C3S G i1 ␣ (Fig. 1). The reason for this remains to be resolved as the fusion proteins containing C3S G i1 ␣ were as effectively stimulated by UK14304 as any of the others (Fig. 4).
Given the nature of the physical linkage between the receptor and G protein in the fusion proteins, the knowledge that the N terminus of the G protein ␣ subunit plays a central role in interaction with the ␤␥ complex (28 -29), the concept that the ␤␥ complex may play a key role in receptor interactions with the ␣ subunit (33)(34)(35), and the understanding that G protein ␣ subunit acylation is important in interactions with the ␤␥ complex, it was of considerable interest to observe that coexpression of excess ␤ 1 ␥ 2 along with any of the ␣ 2A -adrenoreceptor-G i1 ␣ fusion proteins resulted in greater maximal UK14304 stimulation of GTPase activity (Fig. 5). A trivial explanation for this observation based on higher steady-state levels of expression of the fusion protein in the presence of excess ␤␥ was eliminated by performing 3 H-antagonist binding studies (Fig. 5). These results imply interaction of the fusion protein with the ␤␥ complex, but understanding the details of this will require further study. It is of interest in this regard to note, however, that Taylor et al. (36) have previously indicated a role for ␤␥ in receptor stimulation of the GTPase activity of the ␣ 2A -adrenoreceptor.
Overall these studies demonstrate the receptor-G protein fusion approach to be a novel and useful means to study receptor-G protein interactions and indicate a key role for G protein acylation in cellular targeting but not intrinsically in transmission of information between receptors and G proteins.