Dimers of class A G protein-coupled receptors function via agonist-mediated trans-activation of associated G proteins.

The histamine H1 receptor and the alpha1b-adrenoreceptor are G protein-coupled receptors that elevate intracellular [Ca2+] via activation of Gq/G11. Assessed by co-immunoprecipitation and time-resolved fluorescence resonance energy transfer they both exist as homo-dimers. The addition of the G protein G11alpha to the C terminus of these receptors did not prevent dimerization. Agonists produced a large stimulation of guanosine 5'-3-O-([35S]thio)triphosphate ([35S]GTPgammaS) binding to receptor-G protein fusions containing wild type forms of both polypeptides. For both receptors this was abolished by incorporation of G208AG11alpha into the fusions. Mutation of a highly conserved leucine in intracellular loop 2 of each receptor also eliminated agonist function but not binding. Co-expression of the two non-functional but complementary fusion constructs reconstituted agonist-mediated binding of [35S]GTPgammaS in membranes of HEK293 cells and elevation of [Ca2+]i in mouse embryo fibroblasts lacking both Gq and G11. Co-expression of the histamine H1 receptor- and the alpha1b-adrenoreceptor-G11alpha fusions allowed detection of functional hetero-dimeric complexes, whereas co-expression of histamine H1 receptor-G11alpha with increasing amounts of L151Dalpha1b-adrenoreceptor resulted in decreasing levels of histamine-stimulated [35S]GTPgammaS binding. Co-expression of the alpha1b-adrenoreceptor with a fusion protein incorporating the N-terminal domain and transmembrane helix 1 of the alpha1b-adrenoreceptor and G11alpha did not result in agonist activation of the G protein but did indicate a role for transmembrane helix 1 in dimerization. These data demonstrate that dimers of these class A receptors function via trans-activation of associated G proteins.

The histamine H1 receptor and the ␣ 1b -adrenoreceptor are G protein-coupled receptors that elevate intracellular [Ca 2؉ ] via activation of G q /G 11 . Assessed by coimmunoprecipitation and time-resolved fluorescence resonance energy transfer they both exist as homodimers. The addition of the G protein G 11 ␣ to the C terminus of these receptors did not prevent dimerization. Agonists produced a large stimulation of guanosine 2؉ ] i in mouse embryo fibroblasts lacking both G q and G 11 . Co-expression of the histamine H1 receptor-and the ␣ 1b -adrenoreceptor-G 11 ␣ fusions allowed detection of functional hetero-dimeric complexes, whereas co-expression of histamine H1 receptor-G 11 ␣ with increasing amounts of L151D␣ 1b -adrenoreceptor resulted in decreasing levels of histamine-stimulated [ 35 S]GTP␥S binding. Co-expression of the ␣ 1b -adrenoreceptor with a fusion protein incorporating the N-terminal domain and transmembrane helix 1 of the ␣ 1b -adrenoreceptor and G 11 ␣ did not result in agonist activation of the G protein but did indicate a role for transmembrane helix 1 in dimerization. These data demonstrate that dimers of these class A receptors function via trans-activation of associated G proteins.

5-3-O-([ 35 S]thio)triphosphate ([ 35 S]GTP␥S) binding to receptor-G protein fusions containing wild type forms of both polypeptides. For both receptors this was abolished by incorporation of G208AG 11 ␣ into the fusions. Mutation of a highly conserved leucine in intracellular loop 2 of each receptor also eliminated agonist function but not binding. Co-expression of the two non-functional but complementary fusion constructs reconstituted agonist-mediated binding of [ 35 S]GTP␥S in membranes of HEK293 cells and elevation of [Ca
The concept that G protein-coupled receptors (GPCRs) 1 exist as dimers or higher order oligomers has moved rapidly from hypothesis to being widely accepted (1)(2)(3)(4). A range of approaches has contributed to this understanding. This includes the ability to co-immunoprecipitate differentially epitopetagged forms of a GPCR from cells in which they are co-expressed, and, in intact cells, the application of a number of resonance energy transfer-based techniques. However, the role of dimerization in function and the mechanisms then respon-sible for initiation of signal transduction by the dimer have been more recalcitrant to analysis. Significant progress in this area has recently been achieved for the class C GPCRs. These contain both a long extracellular N-terminal domain, to which agonist ligands bind, and the prototypic seven transmembrane (TM) helix bundle architecture that is the common feature of all GPCR families (for review, see Ref. 5). The functional ␥-aminobutyric acid, type b receptor is a hetero-dimer of two distinct gene products in which trafficking to the plasma membrane requires interaction between the partner polypeptides (6 -9). This indicates that a key role of dimerization is achieving appropriate cellular localization. This is also true for the class A rhodopsin-like GPCRs because non-functional, truncated splice variants can restrict plasma membrane delivery of fulllength GPCRs and, thus, limit their function (10 -12). Chimeric class C GPCRs consisting of the extracellular domain of one GPCR and the TM and intracellular elements of a second, closely related GPCR or in which the intracellular loops of dimer partners are exchanged have provided strong evidence that the mechanism of action of the dimer involves transactivation (13)(14); that is, ligand binding to one element of the dimer results in activation of G protein produced by the other GPCR within the dimer. Again, the ␥-aminobutyric acid, type b receptor has been particularly informative in this regard as only one of the two gene products that forms the dimer is able to bind the agonist ␥-aminobutyric acid. Equivalent systems are not available for the majority of class A GPCRs. However, for the luteinizing hormone receptor, which also has a long N-terminal domain that binds the ligand, partial reconstitution of function has been achieved by co-expression of distinct pairs of mutants (15)(16). Furthermore, in recent studies co-expression of a mutant receptor defective in hormone binding and another mutant defective in signal generation rescued hormone-activated cAMP production (17).
Fusion proteins in which G protein ␣ subunits are linked in-frame to the C terminus of a GPCR have become widely used tools to study the details of information transfer between these proteins (18 -19). Because these can be considered as bi-functional proteins containing the sequence and function of both GPCR and G protein they can be utilized to generate contrasting pairs of non-functional mutants. Herein we generate and analyze fusion proteins incorporating either wild type GPCRs or mutants unable to activate G proteins and wild type or mutant G proteins unable to be activated by GPCRs. We show that co-expression of pairs of these non-functional mutant fusion proteins is complementary and results in the generation of functional dimers. This reconstitution can be both easily quantitated in membrane preparations and monitored in single intact cells. Studies of hetero-dimeric GPCR pairs support a mechanism in which agonists at aminergic class A GPCRs mediate trans-activation of the GPCR-associated G proteins.

EXPERIMENTAL PROCEDURES
Materials-A fibroblast cell line (EF88) (20 -22) derived from a combined G␣ q /G␣ 11 double knockout mouse (23)(24)  . Production and characterization of the anti-G q /G 11 antiserum CQ was described by Mitchell et al. (25). All other chemicals were from Sigma and were of the highest grade available.
Construction of Fusion Proteins-Production and subcloning of wild type and mutated ␣ 1b -adrenoreceptor-G␣ 11 fusion proteins was performed as described in Carrillo et al. (26). Production and subcloning of the human histamine H 1 receptor-G 11 ␣ fusion proteins was performed in two separate stages. In the first step, using the N-terminal primer 5Ј-GATACTGGGCTATCCAAGCTTATGAGCCTCCCCAATTCCTC-3Ј, a HindIII restriction site (underlined) was introduced by PCR upstream of the coding sequence of the human histamine H 1 receptor. Using a C-terminal primer 5Ј-AAGGAAAAAAGCGGCCGCTGGAGCGAATAT-GCAGAATTCTCT-3Ј, a three amino acid spacer (Ser-Gly-Arg) and a NotI restriction site were introduced immediately upstream of the stop codon. Similarly, the mouse G 11 ␣ sequence was amplified by PCR using the N-terminal primer 5Ј-AAGGAAAAAAGCGGCCGCATGACTCTGG-AGTCCATGATGGC-3Ј and the C-terminal primer 5Ј-ATGAAACCGC-TCGAGTCACACCAGGTTGTACTCCTTCAG-3Ј. This introduced NotI and XhoI restriction sites flanking the G 11 ␣-coding sequence. In the second step, the amplified receptor fragment was digested with HindIII/NotI, and the G 11 ␣ fragment was digested with NotI/XhoI. These fragments were purified and ligated into pcDNA 3 vector (Invitrogen) previously digested with HindIII/XhoI. The choice of intracellular loop 2 Leu to Asp mutants was based on the studies of Greasley et al. (27). For co-immunoprecipitation and tr-FRET studies, c-Myc (EQKLISEEDL) or FLAG (DYKDDDDK) epitopes were introduced immediately after the N-terminal methionine. Each construct was fully sequenced before its expression and analysis. Construction of the c-Myc-NtTM1␣ 1b -G 11 ␣ fusion protein was also carried out in two stages. Using the N-terminal primer 5Ј-TTTCCCAAGCTTATGGAACAAAAA-CTTATTTCTGAAGAAGATCTGAATCCCGATCTGGACACC-3Ј a HindIII restriction site (underlined) followed by an ATG codon and the c-Myc tag sequence (italics) was introduced by PCR upstream of the coding sequence of the ␣ 1b -adrenoreceptor. Using a C-terminal primer 5Ј-AATCGGGGTACCGGTGGGCGTCCGCAGGTGCCGAAT-3Ј, a twoamino acid spacer (Lys-Leu) and a KpnI restriction site (underlined) were introduced immediately after the codon for the 80th amino acid of the hamster ␣ 1b -adrenoreceptor. In the second stage the resulting PCR fragment was digested with HindIII/KpnI. Next, the ␣ 1b -adrenoreceptor element of the ␣ 1b -adrenoreceptor-G 11 ␣ fusion protein was removed by digestion with HindIII/KpnI, and the digested PCR fragment was ligated into the resulting vector. Thus, the generated fusion protein contained a c-Myc tag linked in-frame with the first 80 amino acids of the ␣ 1b adrenoreceptor and G 11 ␣.
Transient Transfection of HEK293 Cells-HEK293 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 0.292 g/liter L-glutamine and 10% (v/v) newborn calf serum at 37°C in a 5% CO 2 -humidified atmosphere. Cells were grown to 60 -80% confluency before transient transfection in 60-mm dishes. Transfection was performed using LipofectAMINE reagent (Invitrogen) according to the manufacturer's instructions. Samples were pre-cleared with Pansorbin (Calbiochem) followed by immunoprecipitation with CQ antiserum (25). Finally, the immunocomplexes were washed twice with solubilization buffer, and bound [ 35 S]GTP␥S was measured by liquid scintillation spectrometry.
[ 3 H]Ligand Binding Studies-[ 3 H]Prazosin binding studies, monitoring expression of the ␣ 1b -adrenoreceptor-containing constructs, were performed as in Carrillo et al. (26). [ 3 H]Mepyramine binding assays, monitoring expression of the histamine H1 receptor-containing constructs were initiated by the addition of 8 g of cell membranes to an assay buffer (50 mM Tris-HCl, 100 mM NaCl, 3 mM MgCl 2 (pH 7.4) containing [ 3 H]mepyramine (0.1-10 nM). Nonspecific binding was determined in the presence of 100 M triprolidine. Reactions were incubated for 30 min at 25°C, and bound ligand was separated from free ligand by vacuum filtration through GF/B filters. The filters were washed twice with assay buffer, and bound ligand was estimated by liquid scintillation spectrometry.
[Ca 2ϩ ] i Imaging-EF88 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) heat-inactivated fetal bovine serum and L-glutamine (1 mM) in a 95% air and 5% CO 2 atmosphere at 37°C. A portion of the cells harvested during trypsinization were plated onto glass coverslips, and after a 24-h growth period they were transfected using LipofectAMINE (Invitrogen) according to the manufacturer's instructions. After 3 h cells were washed twice with Opti-MEM I and then cultured in Dulbecco's modified Eagle's medium growth medium for a further 24 h. A total of 3 g of pCDNA3 containing the relevant cDNA species was used to transfect each coverslip. Transfected EF88 cells were loaded with the Ca 2ϩ -sensitive dye Fura-2 by incubation (15-20 min, 37°C) under reduced light in Dulbecco's modified Eagle's growth medium containing the dye's membrane-permeant acetoxymethyl ester form (1.5 M). Details of the imaging studies and their analysis have been provided previously (28).
GPCR Co-immunoprecipitation Studies-Co-immunoprecipitation studies using FLAG-and c-Myc-tagged forms of the ␣ 1b -adrenoreceptor and histamine H1 receptor constructs were performed as in McVey et al. (29) and Ramsay et al. (30). In the studies with the histamine H1 receptor, 30 units/ml of endoglycosidase F were added. Time-resolved fluorescence resonance energy transfer was performed on intact HEK293 cells using Eu 3ϩ -labeled anti-c-Myc antibodies and allophycocyanin-labeled anti-FLAG antibodies as described in McVey et al. (29).
Miscellaneous-All experiments were performed on at least three independent occasions. Data are presented as the means Ϯ S.E.

RESULTS
We have previously generated a fusion protein between the ␣ 1b -adrenoreceptor and the ␣ subunit of G 11 that binds both agonist and antagonist ligands including [ 3 H]prazosin (26). The addition of the agonist phenylephrine to membranes of HEK293 cells transfected to express this construct resulted in a large stimulation of the binding of [ 35 S]GTP␥S, monitored after the assay by immunoprecipitation using an antiserum against the C-terminal decapeptide of G 11 ␣ (Fig. 1A). Introduction of a G208AG 11 ␣ mutant into the fusion protein essentially eliminated phenylephrine stimulation of [ 35 S]GTP␥S binding when membranes expressing equal amounts of this construct were used (Fig. 1A), because this form of the G protein is unable to release bound GDP. However, this mutation did not alter the binding properties of either [ 3 H]prazosin or phenylephrine ( Fig. 1, B and C, Table I). Previous studies show that mutation of hydrophobic amino acids in intracellular loop 2 of the ␣ 1b -adrenoreceptor can eliminate agonist-mediated signal transduction (27). We, thus, generated a fusion protein between L151D ␣ 1b -adrenoreceptor and G 11 ␣. This also bound both [ 3 H]prazosin and phenylephrine as the wild type fusion protein (Fig. 1, B and C, Table I), but phenylephrine was again unable to stimulate binding of [ 35 S]GTP␥S (Fig. 1A). Co-expression of the two non-functional mutants reconstituted phenylephrine-mediated binding of [ 35 S]GTP␥S (Fig. 1A), and when the membrane amounts employed contained twice as many [ 3 H]prazosin binding sites as used for each individual construct, the level of agonist-mediated [ 35 S]GTP␥S was ϳ70% as great as when employing the wild type fusion construct (Fig.  1A). Reconstitution of function required co-expression of the two mutant fusions. When the two constructs were expressed in separate cell populations and either the cells were mixed before membrane preparation or membranes were prepared individually and then combined before assay, no agonist-stimulated binding of [ 35 S]GTP␥S was observed (Fig. 1D). Such results are consistent with the hypothesis that GPCR dimer-ization is required for agonist function. Furthermore, within the dimer, one GPCR element must activate the G protein physically linked to the partner GPCR. To extend this concept 3, L151D␣ 1b -adrenoreceptor-G 11 ␣; 4 and 5, ␣ 1b -adrenoreceptor-G208AG 11 ␣ ϩ L151D␣ 1b -adrenoreceptor-G 11 ␣. B and C, mutation of L151D in the ␣ 1b -adrenoreceptor or G208A in G 11 ␣ does not modify the binding of agonists or antagonists. Membranes expressing the wild type ␣ 1badrenoreceptor-G 11 ␣ (squares), the L151D␣ 1b -adrenoreceptor-G 11 ␣ (circles), or the wild type ␣ 1b -adrenoreceptor-G208AG 11 ␣ (triangles) fusion proteins were used to measure the binding of varying concentrations of [ 3 H]prazosin (B) or the ability of varying concentrations of phenylephrine to compete with 1 nM [ 3 H]prazosin for binding (C). D, L151D␣ 1b -adrenoreceptor-G 11 ␣ and ␣ 1b -adrenoreceptor-G208AG 11 ␣ reconstitute function only when they are co-expressed. The binding of [ 35 S]GTP␥S in the absence (open bars) or presence (filled bars) of 10 M phenylephrine was measured in HEK293 cell membranes in which L151D␣ 1b -adrenoreceptor-G 11 ␣ and ␣ 1b -adrenoreceptor-G208AG 11 ␣ were co-expressed (1) or in which the two constructs were expressed in separate cell populations that were mixed before membrane preparation (2) or from which membranes were made separately and then mixed before assay (3).   2). This was absent upon separate expression of either a histamine H1 receptor-G208AG 11 ␣ fusion protein or a fusion between L133D histamine H1 receptor and wild type G 11 ␣.
Co-expression of these two mutants reconstituted agonist activation of the G protein (Fig. 2). Again, after co-expression of the two mutants, membranes expressing a 2-fold higher number of 3 H-labeled antagonist binding sites produced nearly as high a level of [ 35 S]GTP␥S binding upon the addition of histamine as the wild type histamine H1 receptor-G 11 ␣ fusion protein expressed in isolation.
As an extension to these studies we attempted to monitor dimerization and functional reconstitution in a single cell. To do so we employed Ca 2ϩ imaging using EF88 cells. These are a line of mouse embryo fibroblasts that derive from a G q ␣/G 11 ␣ double knock-out mouse (20 -21). They, thus, require expression of both a functional GPCR and a functional Ca 2ϩ -mobilizing G protein to produce elevation of intracellular [Ca 2ϩ ] (28,31). Upon introduction of fusions between wild type forms of either the histamine H1 receptor or the ␣ 1b -adrenoreceptor, and G 11 ␣ appropriate agonists produced elevation of intracellular [Ca 2ϩ ] (Fig. 3). This occurred only in positively transfected cells. Because EF88 cells are recalcitrant to transfection, we co-transfected with enhanced green fluorescent protein (GFP) to allow visualization of the positively transfected cells. Only those cells that were positive for GFP responded to agonist ligands (Fig. 3A). For both the histamine H1 receptor and the ␣ 1b -adrenoreceptor the fusions containing either the nonagonist-responsive GPCR or the G protein mutant failed to elevate intracellular [Ca 2ϩ ] (Fig. 3, B and C). However, coexpression of the pairs of complementary fusions again resulted in effective signal generation (Fig. 3, B and C).
We also wished to demonstrate directly the ability of both the isolated GPCRs and the GPCR-G protein fusions to form dimers. Constructs were N-terminally epitope-tagged with either the c-Myc or FLAG tags. After co-expression in HEK293 cells of both tagged forms of the ␣ 1b -adrenoreceptor, but not their separate expression, followed by cell mixing, immunoprecipitation with anti-FLAG antibodies resulted in the presence of anti-c-Myc immunoreactivity in the precipitate (Fig. 4A). SDS-PAGE demonstrated the presence of bands identified by the c-Myc antibody of apparent size 53 and 110 kDa, consistent with monomeric and dimeric forms of the ␣ 1b -adrenoreceptor. Anti-c-Myc immunoreactivity was also observed near the top of the gel, and this may represent either a higher order oligomer or aggregated protein (Fig. 4A). When equivalent experiments were performed with the ␣ 1b -adrenoreceptor-G 11 ␣ fusion protein similar results were obtained except that the anti-c-Mycreactive bands were now of apparent mass 95 and 190kDa, consistent with the anticipated size of monomeric and dimeric forms of this fusion protein (Fig. 4A). Similar results were obtained for FLAG-and c-Myc-tagged forms of the histamine H1 receptor (Fig. 4B). The monomeric form of the isolated  EF88 cells were transfected to express GPCR-G 11 ␣ fusion proteins, and GFP and the ability of agonist ligands to elevate intracellular Ca 2ϩ were monitored. A, only positively transfected cells respond to agonist. Cells were co-transfected with the wild type histamine H1 receptor-G 11 ␣ fusion and GFP. In the field shown only a single cell expressed GFP (left). Basal (center) and 1 mM histamine (right)-stimulated Ca 2ϩ was then monitored in these cells. The warmer color represents elevated [Ca 2ϩ ]. B, EF88 cells were transfected with GFP and histamine H1 receptor-G 11 ␣ (black, n ϭ 6), histamine H1 receptor-G208AG 11 ␣ (blue, n ϭ 10), L133D histamine H1 receptor-G 11 ␣ (green, n ϭ 12), and both histamine H1 receptor-G208AG 11 ␣ and L133D histamine H1 receptor-G 11 ␣ (red, n ϭ 8). The response of GFP-positive cells to 1 mM histamine was then measured over time. n ϭ the number of individual cells quantitated. C, EF88 cells were transfected with GFP and ␣ 1b -adrenoreceptor-G 11 ␣ (black, n ϭ 8), ␣ 1b -adrenoreceptor-G208AG 11 ␣ (blue, n ϭ 10), L151D ␣ 1b -adrenoreceptor-G 11 ␣ (green, n ϭ 8), and both ␣ 1b -adrenoreceptor-G208AG 11 ␣ and L151D ␣ 1b -adrenoreceptor-G 11 ␣ (red, n ϭ 12). The response of GFP-positive cells to 10 M phenylephrine (Phe) was then measured over time. n ϭ the number of individual cells quantitated. receptor migrated as a ϳ50-kDa polypeptide with the dimeric form migrating as anticipated for a polypeptide of some 100 kDa (Fig. 4B). Again, as with the ␣ 1b -adrenoreceptor, a series of higher molecular mass species were also detected. When using the histamine H1 receptor-G 11 ␣ fusion protein both the monomeric and dimeric species were also easily detected (Fig. 4B).
A series of issues have been raised about the meaning and validity of GPCR dimerization data that rely exclusively on co-immunoprecipitation (3,32). We, thus, monitored dimerization of both the isolated ␣ 1b -adrenoreceptor and the ␣ 1b -adrenoreceptor-G 11 ␣ fusion protein in intact HEK293 cells using tr-FRET. When co-expressing c-Myc-and FLAG-tagged forms of either the isolated GPCR or the fusion protein, a clear energy transfer signal was obtained in intact cells upon the addition of a combination of Eu 3ϩ -labeled anti-c-Myc antibodies as energy donor and allophycocyanin-labeled anti-FLAG antibodies as energy acceptor (Fig. 5A). An energy transfer signal was not obtained when the tagged forms of the GPCR constructs were expressed in separate populations of cells that were mixed before the addition of the antibodies. Equivalent results were obtained in HEK293 cells expressing N-terminal c-Myc-and FLAG-tagged forms of both the histamine H1 receptor and the histamine H1 receptor-G 11 ␣ fusion protein (Fig. 5B).
To examine the possibility of hetero-dimerization between the histamine H1 receptor and the ␣ 1b -adrenoreceptor and the mechanism of G protein activation by GPCR dimers we coexpressed a FLAG-tagged form of the histamine H1 receptor and the c-Myc-tagged form of the ␣ 1b -adrenoreceptor. After immunoprecipitation with anti-FLAG antibodies and SDS-PAGE, c-Myc immunoreactivity was detected in polypeptides of apparent molecular mass 50 and 100 kDa, consistent with the immunoprecipitation of histamine H1 receptor-␣ 1b -adrenoreceptor hetero-dimers that are only partially separated by the electrophoresis conditions employed (Fig. 6A). tr-FRET studies after co-expression of the FLAG-tagged form of the histamine H1 receptor and the c-Myc-tagged form of the ␣ 1b -adrenoreceptor confirmed the presence of histamine H1 receptor/␣ 1b -adrenoreceptor hetero-dimers at the cell surface (Fig. 6B), although the absolute level of the signal indicated that these heterodimers formed less efficiently than the corresponding homodimer pairs (see the y axis of Figs. 5, A and B, compared with Fig. 6B). As in the homo-dimer studies, no tr-FRET signal was observed when separate cell populations expressing each of these receptors were mixed before analysis (Fig. 6B).
When L133D histamine H1 receptor-G 11 ␣ was co-expressed in EF88 cells with ␣ 1b -adrenoreceptor-G208AG 11 ␣, phenylephrine was able to elevate intracellular [Ca 2ϩ ], but histamine was not (Fig. 7A). This can only occur if the ␣ 1b -adrenoreceptor activates the G protein physically linked to the L133D histamine H1 receptor. When the protocol was reversed by co-expression of L151D␣ 1b -adrenoreceptor-G 11 ␣ and histamine H1 receptor-G208AG 11 ␣, histamine now caused elevation of intracellular [Ca 2ϩ ], but phenylephrine did not (Fig. 7B). To extend this type of analysis the histamine H1 receptor-G 11 ␣ fusion was co-expressed with the isolated L151D ␣ 1b -adrenoreceptor that is unable to activate G protein and, thus, stimulate binding of [ 35 S]GTP␥S. Histamine stimulation of [ 35 S]GTP␥S binding was significantly reduced in comparison to membranes expressing the same level of only the histamine H1 receptor-G 11 ␣ fusion (Fig. 8A). Such data are consistent with the L151D ␣ 1b -adrenoreceptor generating inactive hetero-dimers with histamine H1 receptor-G 11 ␣ and indicate that the histamine H1 receptor in the hetero-dimer does not activate the G protein physically associated with it. The remaining signal produced by histamine in the co-transfection reflects that some functional histamine H1 receptor-G 11 ␣ homo-dimer is still formed in the presence of L151D ␣ 1b -adrenoreceptor. Indeed, when we co-expressed histamine H1 receptor-G 11 ␣ with increasing amounts of L151D ␣ 1b -adrenoreceptor cDNA, the ability of histamine to cause [ 35 S]GTP␥S binding in membranes expressing the same number of histamine H1 receptor binding sites decreased as levels of L151D ␣ 1b -adrenoreceptor cDNA were increased (Fig. 8A). Similar results were obtained after co-transfection of L151D ␣ 1b -adrenoreceptor with the histamine H1 receptor-G 11 ␣ in EF88 cells. Histamine stimulation of intracellular [Ca 2ϩ ] was reduced markedly (Fig. 8B).
Co-expression of two distinct GPCRs must result in the presence of the respective homo-dimers as well as providing the potential for hetero-dimer formation. We wished to ensure that the reconstitution of Ca 2ϩ signaling observed upon co-expression of L133D histamine H1 receptor-G 11 ␣ with ␣ 1b -adrenoreceptor-G208AG 11 ␣ did not reflect that only ␣ 1b -adrenoreceptor and histamine H1 receptor homo-dimers were present and that the ␣ 1b -adrenoreceptor-G208AG 11 ␣ homo-dimers were simply able to contact and activate G 11 ␣ linked to L133D histamine H1 receptor-G 11 ␣ homo-dimers. To enhance the levels of appro-priately membrane-targeted G protein we generated a construct in which G 11 ␣ was linked to the C terminus of a c-Myctagged form of the N-terminal and first TM region of the ␣ 1b -adrenoreceptor (c-Myc-Nt-TM1␣ 1b -G 11 ␣). This was transfected into HEK293 cells. Immunoblots of membrane fractions clearly demonstrated its expression as a doublet of 53 and 47 kDa whether detection was via anti-c-Myc (Fig. 9A) or anti-G protein antisera (data not shown). Based on immunodetection by the anti-c-Myc antibody, levels of c-Myc-Nt-TM1␣ 1b -G 11 ␣ were significantly greater than those of the c-Myc-␣ 1b -adrenoreceptor-G 11 ␣ fusion protein (Fig. 9A). [ 35 S]GTP␥S binding assays, at the end of which the c-Myc-Nt-TM1␣ 1b -G 11 ␣ construct was immunoprecipitated with anti-c-Myc antibodies, confirmed this construct did not bind [ 35 S]GTP␥S in response to phenylephrine (Fig. 9B). Parallel experiments showed that the anti-c-Myc antibodies did capture phenylephrine-stimulated binding of [ 35 S]GTP␥S to the full-length c-Myc-tagged ␣ 1b -adrenoreceptor-G 11 ␣ fusion protein (Fig. 9B). However, co-expression of c-Myc-Nt-TM1␣ 1b -G 11 ␣ with the isolated ␣ 1b -adrenoreceptor equally did not result in significant stimulation of [ 35 S]GTP␥S binding in anti-c-Myc immunoprecipitates (Fig.  9B), and this was also true when c-Myc-Nt-TM1␣ 1b -G 11 ␣ was co-expressed with the ␣ 1b -adrenoreceptor-G208AG 11 ␣ fusion protein (Fig. 9B). Thus, simply increasing the concentration of membrane-associated G protein did not allow the ␣ 1b -adrenoreceptor or ␣ 1b -adrenoreceptor fusion protein homo-dimers to activate this G protein. It was possible that co-expression of c-Myc-Nt-TM1␣ 1b -G 11 ␣ with forms of the ␣ 1b -adrenoreceptor did not result in their physical proximity and that this might account for the lack of activation of the membrane-tethered G 11 ␣ by phenylephrine. To assess this we co-expressed c-Myc-Nt-TM1␣ 1b -G 11 ␣ with the FLAG-tagged ␣ 1b -adrenoreceptor-G 11 ␣ fusion protein. This pairing produced a strong tr-FRET signal after the addition of a combination of Eu 3ϩ -labeled antic-Myc and allophycocyanin-labeled anti-FLAG antibodies (Fig.  9C). This, again, was not observed when the two constructs were expressed in different cell populations that were mixed before analysis (Fig. 9C). These data demonstrate the proximity of co-expressed c-Myc-Nt-TM1␣ 1b -G 11 ␣ and ␣ 1b -adrenoreceptor-G 11 ␣. Certain other equivalent ␣ 1b -adrenoreceptor TM-G 11 ␣ constructs did not behave in this fashion, 2 and these data are, thus, consistent with the concept that TM1 of the ␣ 1badrenoreceptor may be a contact interface for receptor homodimerization. Overall the data from co-expression of the pairs of inactive histamine H1 receptor-and ␣ 1b -adrenoreceptor-G protein fusions argue strongly that function must result from trans-activation of the G proteins within the dimers.

DISCUSSION
That GPCRs can exist as dimers is now widely accepted (1)(2)(3)(4). However, the basis and importance of this for function has been less easy to establish. Early studies on the ␤ 2 -adrenoreceptor indicated that a peptide corresponding to TMVI of the receptor was able to prevent dimerization and limit agonist stimulation of adenylyl cyclase activity (33). The general applicability of these observations, however, remains to be exemplified. Indeed, although recent studies on the leukotriene B 4 BLT1 receptor also support a key role for TMVI in homo-dimer formation (34), studies on the dopamine D2 receptor suggest a key role for TMIV (35). Furthermore, application of atomic force microscopy to the study of rhodopsin in native disc membranes (36 -37) suggests that inter-dimeric contacts may involve TM segments IV and V. Distinct approaches provide evidence of roles for TM1 and TM2 in the stabilization of dimers of the yeast ste2 receptor (38) and of TM segments I, II, 2 J. J. Carrillo and G. Milligan, manuscript in preparation.
FIG. 6. The ␣ 1b -adrenoreceptor and the histamine H1 receptor can form hetero-dimeric complexes. A, a FLAG-tagged form of the histamine H1 receptor (flag H1) and a c-Myc-tagged form of the ␣ 1badrenoreceptor (myc ␣ 1b ) were expressed either individually or together (flag ϩ myc) in HEK293 cells. Cells expressing the two constructs individually were also mixed before analysis. Samples were immunoprecipitated with anti-FLAG and, after SDS-PAGE and transfer, immunoblotted with anti-c-Myc-antibodies. B, cells either co-expressing FLAG H1 and c-Myc ␣ 1b (filled bars) or separate populations of cells expressing either of the two constructs that were then mixed were treated with a combination of Eu 3ϩ -labeled anti-c-Myc and allophycocyanin-labeled anti-FLAG antibodies, and tr-FRET was measured. and/or IV in the complement C5a receptor (39). This range of results is fascinating, and combinations of direct experimentation and bio-informatic approaches (40 -41) are likely to be required to provide understanding. Given the variation in potential orientation and organization between GPCRs, it is thus interesting that in experiments described herein, we have also provided evidence consistent with a role for TM1 in the dimerization of this receptor (Fig. 9C). As well as homo-dimers, many pairs of GPCRs are able to form hetero-dimers (1)(2)(3)(4), and there is evidence that this results in production of distinct pharmacology and alterations in function (42)(43)(44)(45)(46)(47). Although yet to be tested directly, there are suggestions that in such heterodimers, the individual GPCRs may contribute different elements to the dimer interface (48). We have demonstrated the capability of the ␣ 1b -adrenoreceptor and the histamine H1 receptor to form hetero-dimers. It is, thus, of importance to note that both of these GPCRs are expressed in arterial smooth muscle cells (49 -50) and may, thus, form hetero-dimers in vivo.
Although class C is the smallest family of GPCRs, studies on the importance of dimerization and the mechanisms of signal transduction within the dimer have advanced most rapidly in this group (5). This reflects both that the ␥-aminobutyric acid, type b receptor is an obligate hetero-dimer between two distinct but related GPCR gene products and that all members of this family have a large N-terminal extracellular domain that is responsible for binding of agonist ligands. It has, thus, been relatively easy to generate chimeric class C receptors that mix and match the extracellular and transmembrane and intracellular domains of individual GPCRs. The rhodopsin-like class A GPCRs comprise greater than 80% of the entire family present in the human genome (51)(52). However, few class A GPCRs have structural features to facilitate such an approach, and GPCRs with aminergic agonists bind the ligands deep within the cleft produced by the architecture of the seven TM helices. By employing fusion proteins between both the ␣ 1b -adrenoreceptor and the histamine H1 receptor with the G protein G 11 ␣, we now show that these GPCRs dimerize and that this is not compromised by the addition of the G protein to the C-terminal tail of the GPCR. Equally, by employing tr-FRET to detect GPCR dimers in intact cells we were able to demonstrate the presence of these complexes at the cell surface. This also was not compromised by the addition of the G protein sequence to the C-terminal tail of the GPCRs. Furthermore, by introducing mutations that prevent agonist activation of the G protein into either the GPCR or the G protein, we produced pairs of distinct, non-functional fusion proteins that were able to restore ago-  Agonist function in heterodimeric complexes proceeds via trans-activation of receptor-associated G proteins. EF88 cells were transfected to co-express L151D ␣ 1b -adrenoreceptor-G 11 ␣ ϩ histamine H1 receptor-G208AG 11 ␣ (A) or L133D histamine H1 receptor-G 11 ␣ and ␣ 1b -adrenoreceptor-G208AG 11 ␣ (B). The effect of 10 M phenylephrine (Phe) or 1 mM histamine (Hist) on cellular [Ca 2ϩ ] i was then assessed. In both panels, co-transfection with GFP demonstrated that only the cell in the field that responded to agonist was transfected.
nist-mediated function when co-expressed. Many of the studies utilized combinations of epitope tagging and/or mutagenesis of the fusion proteins. As such, it was important to confirm that these modifications did not cause substantial alterations in the binding of ligands compared with the parental fusion proteins. We demonstrated that the binding of both agonist and antagonists ligands was not markedly different for any of the modified constructs (Table I), and these values were similar to those of the corresponding wild type GPCRs. Functional reconstitution was monitored in two ways. First, agonists were able to produce elevation of intracellular [Ca 2ϩ ] in EF88 cells only after co-expression of two mutants that were each non-functional in isolation. EF88 cells lack expression of phospholipase C-coupled G proteins, and thus, it is necessary to introduce both a suitable GPCR and G protein into these cells to generate a Ca 2ϩ signal (31). This assay had the obvious benefit that Ca 2ϩ imaging allowed us to monitor functional dimerization in single cells and in the absence of excess G protein. However, although [Ca 2ϩ ] i can be calculated from such studies, this is an amplified signal and, thus, not ideal for quantitation of the effectiveness of dimerization. One of the earliest steps that can be measured in the signal transduction cascade is agonistinduced guanine nucleotide exchange on the G protein. This can be monitored conveniently by the binding of [ 35 S]GTP␥S. Historically, the characteristics of guanine nucleotide exchange by different G protein families had meant that this assay was only suitable for the pertussis toxin-sensitive members of the G i -subfamily (53). However, recent addition of end of assay immunocapture steps has allowed it to be adapted to also measure activation of the G s and G q family G proteins (53). In all the [ 35 S]GTP␥S binding assays we initially measured the level of expression of each of the GPCR-G protein fusions by using saturation 3 H-labeled antagonist binding studies. This allowed us to add membrane amounts containing defined quantities of the constructs to the assays. We have previously demonstrated that there is a linear increase in agonist-stimulated [ 35 S]GTP␥S binding with the addition of increasing amounts of a GPCR-G 11 ␣ fusion protein (31). When co-expressing the histamine H1 receptor-G208AG 11 ␣ and L133D histamine H1 receptor-G 11 ␣ fusion proteins, it required the presence of twice the number of 3 H-labeled antagonist binding sites to generate approximately the same amount of agonist-stimulated [ 35 S]GTP␥S binding as when only the wild type histamine H1 receptor-G 11 ␣ fusion protein was expressed. This provides supporting evidence that the functional element is a dimer. If the functional histamine H1 receptor is a dimer, then stochastically, when co-expressing the two non-functional mutant fusions, half of the dimers produced should be non-functional because they will represent homo-interactions between either of the non-functional forms, i.e. histamine H1 receptor-G208AG 11 ␣ or L133D histamine H1 receptor-G 11 ␣. Only 50% of the dimers would be expected to be hetero-dimers containing one copy of histamine H1 receptor-G208AG 11 ␣ and one of L133D histamine H1 receptor-G 11 ␣ and, thus, be functional. These studies are also consistent with the idea that aminergic class A GPCRs function via trans-activation of their associated G proteins. The copy of the G protein in the dimer that can be activated is linked to the non-functional form of the GPCR, whereas the functional form of the GPCR is associated with non-functional G protein. Further studies, however, were required to provide greater support for this mechanism. To do so we took advantage of the known capacity of structurally related GPCRs to form hetero-dimers. Initial studies demonstrated that when co-expressed the histamine H1 receptor and the ␣ 1b -adrenoreceptor could be co-immunoprecipitated. Furthermore, co-expression in EF88 cells of L133D histamine H1 receptor-G 11 ␣ and ␣ 1b -adrenoreceptor-G208AG 11 ␣ resulted in phenylephrine but not histamine-mediated elevation of [Ca 2ϩ ] i . This can only occur if the ␣ 1b -adrenoreceptor activates the G protein physically linked to the inactive histamine H1 receptor (Fig. 10). When the experiment was reversed such that the inactive ␣ 1b -adrenoreceptor was linked to the wild type G protein and the wild type histamine H1 receptor was linked to the mutant G protein, histamine was functional, but phenylephrine was not. We extended this idea by examining the effectiveness of histamine to stimulate binding of [ 35 S]GTP␥S when the histamine H1 receptor fusion protein was co-expressed with increasing amounts of the isolated but inactive L151D ␣ 1b -adrenoreceptor. The effect of histamine was reduced. Such information is consistent with the concept than increasing levels of a histamine H1 receptor-G 11 ␣-L151D ␣ 1badrenoreceptor hetero-dimer reduces the amounts of the histamine H1 receptor-G 11 ␣ homo-dimer and that histamine binding to the hetero-dimer is unable to activate the G protein that is physically associated with the histamine H1 receptor. In this situation phenylephrine was inactive because L151D ␣ 1b -adre- noreceptor is unable to stimulate any G protein. A number of reports have indicated that GPCR-G protein fusions can interact with and activate endogenously expressed G proteins as well as the G protein element of the fusion (54 -55). However, in these studies the GPCR-G protein fusions were expressed at very high levels that are within the range in which nonspecific "bystander" (56) effects have been reported due to physical proximity and crowding in the membrane. Use of EF88 cells eliminated the possibility of interaction with endogenous G proteins as they do not express G q ␣ or G 11 ␣, and thus, effects have to reflect activation of the fused G proteins. Moreover, after introduction of the G208A mutation into the G protein element of the fusions, agonist stimulation of [ 35 S]GTP␥S binding in membranes of transfected HEK293 cells was virtually abolished. This indicates that at the level of expression achieved, there was virtually no activation of endogenous G q ␣ or G 11 ␣ in HEK293 cells even though both are expressed (25). In the hetero-dimerization experiments in HEK293 cells, excess G protein is introduced in a 1:1 molar ratio with the second GPCR due to the 1:1 stoichiometry of GPCR and G protein defined by the fusion. To assess if the results could be ascribed simply to the presence of the extra G protein we provided extra G protein via an alternate strategy. To do so we generated a form of G 11 ␣ linked to the N terminus and first TM region of the ␣ 1b -adrenoreceptor. Equivalent constructs for other G proteins have been employed previously (55,(57)(58), and it has been suggested that the link to a TM ␣ helix provides the G protein in a particularly effective orientation for activation (55). Although this construct could be expressed to markedly higher levels than the ␣ 1b -adrenoreceptor-G 11 ␣ fusions, the G protein was not activated by phenylephrine whether expressed alone or in combination with either the isolated ␣ 1b -adrenoreceptor or an ␣ 1b -adrenoreceptor-G 11 ␣ fusion. As discussed above, tr-FRET analysis confirmed the physical proximity of the ␣ 1b -adrenoreceptor-G 11 ␣ fusion and G protein linked to the TM1-containing construct after their co-expression. These studies provided further evidence that the reconstitution of signal with co-expression of non-functional pairs of GPCR-G protein fusions must reflect an internal trans-activation within the reconstituted dimer and are also consistent with a key role of TM1 as a dimer interface in the ␣ 1b -adrenoreceptor.