Characterization of the Residues in Helix 8 of the Human β1-Adrenergic Receptor That Are Involved in Coupling the Receptor to G Proteins*

Several key amino acids within amphipathic helix 8 of the human β1-adrenergic receptor (β1-AR) were mutagenized to characterize their role in signaling by G protein-coupled receptors. Mutagenesis of phenylalanine at position 383 in the hydrophobic interface to histidine (F383H) prevented the biosynthesis of the receptor, indicating that the orientation of helix 8 is important for receptor biosynthesis. Mutagenesis of aspartic acid at position 382 in the hydrophilic interface to leucine (D382L) reduced the binding and uncoupled the receptor from G protein activation. Mutagenesis of the basic arginine residue at position 384 to glutamine (R384Q) or to glutamic acid (R384E) increased basal and agonist-stimulated adenylyl cyclase activities. R384Q and R384E displayed features associated with constitutively active receptors because inverse agonists markedly reduced their elevated basal adenylyl cyclase activities. Isoproterenol increased the phosphorylation and promoted the desensitization of the Gly389 or Arg389 allelic variants of the wild type β1-AR but failed to produce these effects in R384Q and R384E, because these receptors were maximally phosphorylated and desensitized under basal conditions. In contrast to the membranous distribution of the wild type β1-AR, R384Q and R384E were localized mostly within intracellular punctate structures. Inverse agonists restored the membranous distribution of R384Q and R384E, indicating that they recycled normally when their constitutive internalization was blocked by inverse agonists. These data combined with computer modeling of the putative three-dimensional organization of helix 8 indicated that the amphipathic character of helix 8 and side chain projections of Asp382 and Arg384 within the hydrophilic interface might serve as a tethering site for the G protein.

GPCR for mediating the functions of catecholamines. Activation of the ␤ 1 -AR increases heart rate and myocardial contractility and increases the secretion of renin, all of which contribute to its important role in regulating cardiac function and blood pressure (3). The ␤ 1 -AR is a prominent member of class A GPCR, which are structurally related to the visual receptor rhodopsin that propagates its signal through the activation of the G protein transducin (1). The three-dimensional crystal structure of dark-adapted rhodopsin offers a structural template for other GPCRs, including the assignment of secondary structural elements and the location of highly conserved amino acids (4). This model can accommodate most of the essential elements of functional importance in G protein activation. Therefore, an evaluation of the ␤ 1 -AR can be modeled after rhodopsin.
The cytoplasmic surface of class A GPCR includes three loops and a C-terminal tail. The first, second, and third loops connect adjacent transmembrane helices. A crucial finding in the crystal structure of rhodopsin was the confirmation that residues in helix 7 and the C-terminal domain form a cytoplasmic helical structure termed helix 8 (4). Helix 8 is bound by helix 7 at its N terminus and by two palmitoyl groups inserted in the membrane bilayer at its C terminus (5). In the human ␤ 1 -AR it is predicted that helix 8 is preceded by a short linker sequence of 3-4 amino acids that begin after the tyrosine residue at position 377 of the NPXXY motif. The linker sequence is followed by a stretch of 10 amino acids that are predicted to form an ␣-helical structure that constitutes a novel hallmark of class A of the GPCR superfamily (4). In rhodopsin, helix 8 was involved in coupling rhodopsin to transducin because synthetic peptides corresponding to helix 8 competed with metarhodopsin II in binding transducin (6). In addition, site-directed mutagenesis within helix 8 of rhodopsin indicates that helix 8 is involved in rhodopsin-mediated activation of transducin (7)(8)(9). Analysis of the role of helix 8 in other GPCR revealed a number of diverse functions attributed to this domain. Helix 8 was involved in regulating surface expression of the melanin concentrating hormone receptor-1 (10) and in the internalization of the bradykinin 2 receptor (11,12). Residues within the hydrophilic interface of putative helix 8 of angiotensin II 1A , bradykinin 2 , and oxytocin receptors were involved in coupling these receptors to G q (12)(13)(14).
The studies described earlier have focused on the role of helix 8 in G q -coupled GPCR. Here we characterized site-directed mutants of helix 8 in coupling the human ␤ 1 -AR to the stimulatory regulatory G protein (G s ). Our detailed analyses show that proper orientation of helix 8 regulates the biosynthesis of the ␤ 1 -AR, and the charges of key residues in the hydrophilic interface of helix 8 are involved in proper expression and coupling of the ␤ 1 -AR to G s .

EXPERIMENTAL PROCEDURES
Construction of FLAG-tagged Wild Type and Mutant ␤ 1 -AR-cDNA encoding the Gly 389 variant of the human WT ␤ 1 -AR (15) was a generous gift from R. Lefkowitz (Howard Hughes Medical Institute, Duke University). To allow rapid assessment of cell surface expression of the ␤ 1 -AR, the N-terminal initiator methionine was replaced by the FLAG sequence (DYKDDDDK) resulting in N-FLAG-tagged WT and mutant ␤ 1 -AR. The 8-amino acid FLAG epitope sequence was inserted at the N terminus of the ␤ 1 -AR by the PCR using sense primer 5Ј-AAGCTTA-TGGACTACAAGGACGACGATGACAAGGGCGCGGGGGTGC-TCGTCCTGGGCG-3Ј and antisense primer 5Ј-CATGAATTCTAC-ACCTTGGATTCCGAGGCGAAGCCGGG. The 1.5-kb ␤ 1 -AR cDNA flanked with HindIII (5Ј) and EcoRI (3Ј) sites was cloned into the multiple cloning site in the pUC18 plasmid. Point mutations in helix 8 were generated by site-directed mutagenesis, using the Transformer TM system (Clontech). The sequence of the mutated ␤ 1 -AR was verified by dideoxy sequencing, followed by cloning of the mutated full-length ␤ 1 -AR cDNAs into the mammalian expression vector pcDNA-3.1 (Invitrogen).
Cell Culture and Transient Transfections-HEK-293 cells were cultured in DMEM supplemented with 10% fetal bovine serum until they were ϳ90% confluent. The WT ␤ 1 -AR or its point mutants in pcDNA 3.1 were transiently transfected into HEK-293 cells using the Cytofectene reagent (Bio-Rad) as follows. Plasmid DNA (5 g) was diluted into 200 l of DMEM and then mixed with an equal volume of DMEM containing 12 l of Cytofectene at room temperature for 20 min. Then 4 ml of DMEM was added, and the DNA-lipid complex was layered over the cells for 5 h at 37°C, followed by the addition of an equal volume of DMEM ϩ 10% fetal bovine serum. Stable cell lines for Gly 389 , G389R, D382L, R384E, and R384Q were used where indicated.
Preparation of Membranes and Radioligand Binding Assays-Transiently transfected or stable cell lines on 10-cm culture plates were washed twice with 10 ml of ice-cold PBS, then scraped from the plates, and pelleted by centrifugation at 2,000 ϫ g av for 10 min. The cell pellets were suspended in 10 ml of hypotonic buffer composed of 20 mM HEPES, pH 7.4, 2 mM MgCl 2 , 1 mM EDTA, and 1 mM 2-mercaptoethanol supplemented with 10 g/ml leupeptin and 10 g/ml aprotinin with or without 1 mM phenylmethylsulfonyl fluoride for 10 min on ice. The cells were lysed by 30 up-and-down strokes in a glass-glass homogenizer and then centrifuged at 2,500 ϫ g av for 5 min. The supernatant was recentrifuged at 15,000 ϫ g av for 20 min to pellet the membranes.
Binding of [ 125 I]iodocyanopindolol (ICYP) to 0.5 g of membranes was measured in 50 mM Tris-HCl, pH 7.4, plus 10 mM MgCl 2 binding buffer containing 0.1 mM ascorbic acid for 2 h at 25°C. For saturation binding experiments, ICYP concentrations ranging between 5 and 300 pM were used to calculate the K D and the B max values for ICYP binding by parametric fitting of the data using the Prism 4 software (Graph-Pad Corp.).
For competition binding experiments, ICYP 70 pM was competed with 24 increasing concentrations of isoproterenol ranging from 0.01 nM to 10 M. The IC 50 (high) and IC 50 (low) values for isoproterenol were derived from two-compartment competition to the GTP data. The IC 50 values were converted to the corresponding K I (high) and K I (low) values using Equation 1, Each saturation and competition experiment was in triplicate, and each was replicated between three and five times to determine the means Ϯ S.E.
Cyclic AMP Accumulation and Adenylyl Cyclase Assays-Transiently transfected cells in 6-well plates were switched to DMEM ϩ 25 mM HEPES for 2 h. Appropriate drugs in DMEM/HEPES, supplemented with 300 M of the phosphodiesterase inhibitor isobutylmethylxanthine, were added to the cells for 10 min at 37°C. The reaction was stopped, and 1 ml of 0.1 N cold HCl was added followed by freezing of the entire plate in liquid nitrogen. Frozen plates were quickly thawed at 65°C to break the cells, and the cell extract was lyophilized. The dry pellet was resuspended in assay buffer, and cyclic AMP was quantified by radioimmunoassay (RIANEN Assay System, DuPont).
For the determination of adenylyl cyclase activity, membranes were prepared from cells without phenylmethylsulfonyl fluoride. Fifty g of membrane proteins were incubated in a final volume of 0.1 ml in buffer containing 50 mM Tris-HCl, pH 7.4, 1 mM MgCl 2 , 10 mM phosphocreatine, 1 mM cyclic AMP, 2 mM 2-mercaptoethanol, 1 mg/ml bovine serum albumin, 0.4 mM EGTA, 2 mg/ml creatine kinase, 0.2 mM ATP containing 2 Ci of [␣-32 P]ATP, 1 mM GTP, and increasing concentrations of isoproterenol from 0.1 nM to 1 mM. The assay was initiated by the addition of membranes and terminated after 10 min (16). Cyclic AMP that was formed was separated from ATP by column chromatography, and its specific activity was determined as cyclic AMP formed in pmol/min/mg protein.
Intact Cell Phosphorylation and ␤ 1 -AR Immunoprecipitation-Cell cultures were switched to phosphate-free DMEM supplemented with 12.5 mM HEPES for 60 min. Then they were incubated with 100 Ci of 32 P i /ml for 2 h at 37°C to equilibrate the [ 32 P]ATP pools. The cells were exposed to 1 mM ascorbic acid (control) or 10 M isoproterenol for 10 min at 37°C, followed by rapid aspiration of this medium. The cells were lysed with 1 ml/plate of ice-cold RIPA, and SDS buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 0.2% SDS supplemented with protease and phosphatase inhibitors) was added. The cells were scraped, sonicated briefly, and then centrifuged at 14,000 ϫ g av for 5 min at 4°C. From the supernatant, two 5-l aliquots were removed for protein assay. The remaining volume was processed for immunoprecipitation; each step thereafter was conducted at 0 -4°C (17). Equal amounts of cell lysates were incubated with 40 l of anti-FLAG M2 IgG-agarose beads and incubated at 4°C overnight on a rotating platform. The resin was washed five times with 1 ml of ice-cold RIPA buffer, and then 100 l of Laemmli sample buffer containing 20 mM dithiothreitol was added, and the slurry was incubated at 37°C for 40 min to release the IgG-␤ 1 -AR complex from the resin. The proteins were separated by electrophoresis in 10% acrylamide gels containing 0.1% SDS, dried, and exposed to autoradiography. The relative amounts of 32 P incorporated into the ␤ 1 -AR band were determined by electronic counting of the gel by the InstantImager TM .
Acid Strip Confocal Recycling Microscopy Protocol-HEK-293 cells expressing the FLAG-tagged WT ␤ 1 -AR and its helix 8 point mutants or the pCMV-Tag2 vector (Stratagene) were grown on poly L-lysinecoated glass coverslips and serum-starved at 37°C for 1 h in DMEM supplemented with 25 mM HEPES, pH 7.4. The receptors were labeled with FITC-conjugated to anti-FLAG M2 IgG (5 g/ml) for 1 h at 37°C. Cells were treated with 10 M isoproterenol for 30 min at 37°C to promote agonist-mediated receptor internalizaion. Then the cells were chilled in 4°C Tris-buffered saline to stop endocytosis and exposed to 0.5 M NaCl, 0.2 M acetic acid, pH 3.5, for 4 min on ice to remove antibody bound to the ␤ 1 -AR (18 -20). Cultures were quickly rinsed in warm DMEM supplemented with HEPES and then incubated with 100 M of the ␤-antagonist alprenolol at 37°C for 10, 20, 30, or 45 min. After each time period, the coverslips were rinsed and fixed in 4% paraformaldehyde in 4% sucrose in PBS, pH 7.4, for 10 min at room temperature (19).
To ascertain that the recycled receptor trafficked back to the membrane, coverslips that were incubated with alprenolol for 45 min were exposed to 0.5 M NaCl, 0.2 M acetic acid for 4 min on ice to strip the antibody from receptors that have recycled during that time frame.
Effect of Inverse Agonists on the Localization of Helix 8 Mutants in HEK-293-Cells expressing the WT ␤ 1 -AR, D382L, R384E, or R384Q were incubated with buffer (control condition) or with 100 M of the inverse agonists alprenolol or CGP-20712A for 2 h at 37°C. Cultures were rinsed and fixed with 4% formaldehyde in 4% sucrose in PBS and then permeabilized in Tris-buffered saline containing 1% Triton X-100, 2% normal goat serum, and 1% bovine serum albumin (20). ␤ 1 -AR was labeled with FITC conjugated to anti-FLAG M2 antibody at room temperature for 1 h to label surface-exposed and internalized ␤ 1 -AR. To estimate the effect of the inverse agonists on the distribution of receptors, a circular boundary was drawn around the inner circumference of control and inverse agonist-treated cells to define a 300-nm-wide membrane-delimited area (21). Fluorescence intensity measurements were calculated in the areas outside and inside the boundary to estimate the membranous versus internal pixels. These measurements were repeated in slides with a similar threshold setting in which the radii for each boundary did not differ by Ϯ15%. Data were analyzed using the Mann-Whitney test (Prism 4 program) and expressed as mean Ϯ S.E.
Analysis of Immunocytochemical Data-All analyses were performed blind to the stimulation history of the culture. Microscope fields had 1-3 cells displaying generally healthy morphology. Six to 10 cells were imaged per culture, and n ϭ 10 cultures were processed per condition. Confocal fluorescence microscopy was performed on all the slides using Zeiss Axiovert LSM 510 (100 ϫ 1.4 DIC oil immersion objective). FITC was excited with the 488-nm argon laser and imaged through the 520-nm long-pass emission filter. Thresholds were set by visual inspection and kept constant for each condition. Z-stacks of images were exported as TIFF files, and individual sections were analyzed with Zeiss LSM 510 and NIH Image 1.6 software as described in Gardner et al. (19). In confocal recycling assays, the time constants () for ␤ 1 -AR recycling were determined by fitting the data to a single exponential decay function of the y ϭ y o ϩ Ae Ϫt/ , where y o and A are constants The effects of inverse agonists on pixel densities inside the boundary were determined by confocal microscopy (19). Internal pixel densities in control cells were arbitrarily set as 100, and the data for each inverse agonist was calculated as % of this value. In acid-striped slides, the boundary method was not use. Instead, total cellular pixel densities were calculated. The ratios of pixel densities in inverse agonist-treated cells to pixel densities in control cells were determined to calculate the effect of the inverse agonist on this parameter. The data represent the mean of the various % values Ϯ S.E. from five slides each involving between 6 and 10 cells.
Molecular Modeling of Wild Type and Mutant ␤ 1 -AR-Homology modeling was performed with the program SegMod (22). The program was used to optimally align the sequences of the human ␤ 1 -AR and bovine rhodopsin, and then the sequence of the ␤ 1 -AR was threaded onto the rhodopsin structure based upon the 1F88 (4) and the 1HZX (23) models of rhodopsin oligomers deposited in the Protein Data Bank. The resulting homology model was then refined according to SegMod protocols. The figures were produced using MOLSCRIPT (24) and rendered with RASTER3D (25) that was also used to estimate the distances in angstroms between the various amino acids.
Data Analysis-Quantitative data were summarized and presented as means Ϯ S.E. from at least four determinations each from triplicate The abbreviations used are as follows: h␣-AR, human ␣-adrenergic receptor; h␤-AR, human ␤-adrenergic receptors; hD-R, human dopamine receptor; rMCH-1R, rat melanin-concentrating hormone 1 receptor; boRHO, bovine rhodopsin; hM-R, human muscarinic receptor; hA-R, human adenosine receptor; hSP-R, human substance-P receptor; hBR 2 -R, human bradykinin 2 receptor; hOX-R, human oxytocin receptor; hANG II -R, human angiotensin II receptor; hBLT1-R, human leukotriene B 4 receptor; h5HT-R, human serotonin receptor. experiments. Least squares linear regression and nonlinear curves were calculated by the Prism 4 program (Graphpad version 4.05) to estimate the K act of each ␤ 1 -AR in stimulating the activity of adenylyl cyclase. Statistical comparisons were analyzed by two-way analysis of variance with post hoc tests using Graphpad Prism program. A value of p Ͻ 0.05 was considered statistically significant.

RESULTS
The crystal structure of bovine rhodopsin indicates that a partially exposed ␣-helix termed helix 8 is located between TM7 and the palmitoylated cysteines in the C terminus. The corresponding region in the human ␤ 1 -AR is the sequence between Pro 381 and Leu 390 ( Table 1). The distribution of side chains in helix 8 of rhodopsin exhibits an amphiphilic pattern, where charged polar groups cluster on one side and hydrophobic ones cluster on the other side of the helix (4,23). Based upon the rhodopsin model, it is expected that the corresponding hydrophobic amino acids in the human ␤ 1 -AR, Phe 383 and Phe 387 , will be buried in the hydrophobic core of the helix facing the inner leaf of the membrane, whereas the charged polar groups Arg 384 , Lys 385 , and Gln 388 would cluster on the other side of the ␣-helix that is exposed to the cytoplasmic environment. As shown in Table 1, the identity and location of the two hydrophobic Phe residues are conserved in class A of GPCR. On the hydrophilic side of the helix, the residue corresponding to Arg 384 is always a basic amino acid such as Arg or Lys, whereas the residues corresponding to Lys 385 and Gln 388 are not conserved.
To test the validity of this model in predicting the structure-activity relationship of these residues, 11 point mutations were introduced into putative helix 8 of the N-terminally FLAG-tagged human ␤ 1 -AR (Table  2). We used the BiotecniX 3D (Gentech Corp.) computer program to verify that none of the point mutations disrupted the predicted ␣-helical secondary structure of helix 8.
Expression and Ligand Binding Properties of Helix 8 Mutants-Immunoblot analyses of detergent extracts prepared from HEK-293 cells that were transiently transfected with the various helix 8 mutants revealed that monomers and dimers of the 73-kDa ␤ 1 -AR were expressed (Fig. 1). The expression patterns of the various helix 8 mutants were comparable with that of the WT ␤ 1 -AR with the exception of the F383H mutant. The highly conserved hydrophobic Phe 383 when mutated to a charged basic His residue (F383H) was not synthesized as a mature protein and did not result in detectable radioligand binding activity ( Table 2).
The other point mutants were tested for the pharmacological character of their radioligand binding profiles ( Table 2). Helix 8 mutants bound the radiolabeled ␤-AR antagonist [ 125 I]ICYP with K D values between 15 and 25 pM, which were comparable with the K D value of the WT ␤ 1 -AR. In addition, the B max values of the various helix 8 mutants were comparable with the B max values of the WT ␤ 1 -AR.
The high and low affinity states for isoproterenol binding to the ␤ 1 -AR were modeled by two-site analysis of the competition isotherms of isoproterenol (Fig. 2). The percentage of receptors in the high affinity state for agonists was comparable among the different helix 8 mutants ( Table 2). The affinities of some point mutants to isoproterenol, however, were not comparable (Fig. 2). Specifically, mutagenesis of the negatively charged amino acid Asp 382 to a nonpolar hydrophobic Leu (D382L) produced a mutant ␤ 1 -AR with ϳ200-fold lower affinity to isoproterenol than the WT ␤ 1 -AR ( Fig. 2 and Table 2). Mutagenesis of the basic Arg 384 amino acid to the acidic and polar glutamic acid or to the neutral and polar glutamine had the opposite effect (Fig. 2). The affinities of the R384E or R384Q to isoproterenol were 10 -50-fold higher than the WT ␤ 1 -AR (Table 2). However, mutagenesis of Arg 384 to the two other basic amino acids Lys (R384K) and histidine (R384H) or

TABLE 2 Ligand binding properties of isoproterenol at the wild type and helix 8 ␤ 1 -AR mutants
[ 125 I]ICYP binding was determined as described under "Experimental Procedures" on membranes derived from HEK-293 cells expressing the wild-type ␤ 1 -AR (Gly 389 or Arg 389 ) or point mutants of helix 8. Saturation isotherms were analyzed by nonlinear regression to determine the K D and B max . Competition binding isotherms for isoproterenol were analyzed using a two-state model to calculate the high and low IC 50 values. The dissociation constants (K I ) of isoproterenol for the high and low affinity state of the receptor were calculated from the corresponding IC 50 values as described under "Experimental Procedures." %R H and %R L indicate the percentage of high and low affinity binding sites, respectively. The results are mean Ϯ S.E. of eight triplicate independent determinations for WT ␤ 1 -AR, R384E, R384Q, and G389R or three determinations for the other mutants. The values for the high affinity binding sites are in nM, while the values for the low affinity binding sites are in M. ND indicates not determined.

Expression
vector neutralization of its charge with an amino acid of comparable mass such as Phe (R384F) reduced the binding affinity of isoproterenol to these mutants by ϳ5-20-fold, and these differences in the K I were statistically significant (p Ͻ 0.05). Therefore, it is possible to alter the binding affinity of the ␤ 1 -AR to agonists through selective mutations in Arg 384 .
In the next series of experiments, we determined the effect of mutating the surface-exposed Lys 385 on the binding affinities of isoproterenol. Mutagenesis of Lys 385 to another basic amino acid such as Arg (K385R) or to an acidic amino acid such as Glu (K385E) did not alter the binding affinities of these mutants to isoproterenol (Table 2). However, mutagenesis of Lys 385 to the hydrophobic Met residue (K385M) markedly reduced the binding affinity of isoproterenol (Fig. 2). We also investigated the effect of Gly 389 or Arg 389 on the binding affinity of isoproterenol because these residues are expressed in this position at an allelic frequency of 0.26 and 0.76, respectively (26). The affinity of isoproterenol for the Arg 389 allele was ϳ3-fold higher than its affinity for the Gly 389 allele, in agreement with other more detailed studies on this allelic variation (26).
Coupling of Point Mutants in Helix 8 to G s /Adenylyl Cyclase-Activation of the ␤ 1 -AR promotes the conversion of GDP-G s into GTP-G s leading to the activation of adenylyl cyclase that catalyzes the conversion of ATP into cyclic AMP. To investigate the role of helix 8 in coupling the ␤ 1 -AR to G s , basal and isoproterenol-stimulated cyclic AMP accumulation were measured in cells expressing comparable amounts (between 0.8 and 1.3 pmol of ␤ 1 -AR/mg protein) of WT ␤ 1 -AR or its helix 8 mutants (Fig. 3A). Isoproterenol produced a graded increase in cyclic AMP accumulation in cells expressing the WT ␤ 1 -AR that was maximal at 50 nM. Expression of either F383H or D382L did not increase cyclic AMP accumulation in response to isoproterenol because the F383H mutation prevented the biosynthesis of the ␤ 1 -AR, and the D382L mutation apparently uncoupled the receptor from G s . Based on their binding parameters in Table 2, the Arg 384 point mutants were divided into mutations that either increased or decreased the binding affinity of isoproterenol for the ␤ 1 -AR. These differences were manifested in their respective potencies in promoting cyclic AMP accumulation. Arg 384 mutations that reduced the affinity of isoproterenol in binding to the ␤ 1 -AR, such as R384F and R384K, were associated with markedly diminished receptor-G s coupling efficacy and with 70% reduction in maximal levels of cyclic AMP accumulation. Arg 384 muta-FIGURE 2. Competition binding isotherms of isoproterenol to the wild type ␤ 1 -AR or its helix 8 mutants.
[ 125 ]ICYP binding was determined as described under "Experimental Procedures" on membranes derived from HEK-293 cells expressing the WT ␤ 1 -AR or mutants in helix 8 of the ␤ 1 -AR. The results were representative of three to eight independent experiments whose mean values are reported in Table 2. tions that increased the binding affinity of isoproternol, specifically R384E and R384Q, increased the receptor-G s coupling efficacy. These point mutants were exquisitely sensitive to ␤-agonists, because 1 nM isoproterenol produced close to maximal stimulation of cyclic AMP accumulation (Fig. 3B). The point mutants of Lys 385 did not affect the potency of isoproterenol in stimulating cyclic AMP accumulation, indicating that this amino acid had no appreciable effect on the efficacy of ␤ 1 -AR coupling to G s (Fig. 3C). Mutagenesis of G389R increased the potency of isoproterenol and reduced its 50% efficacy (EC 50 ) from 5 Ϯ 2 to 2 Ϯ 1 nM, in agreement with the data in Table 2.
The effect of helix 8 point mutants on receptor coupling efficacy to G s was assessed by measuring basal and isoproterenol-stimulated adenylyl cyclase activities in membranes prepared from cells expressing comparable densities of each ␤ 1 -AR construct (Fig. 4). The basal level of membrane-associated adenylyl cyclase activity in cells transfected with the empty vector was 8 Ϯ 3 pmol/mg/min, and this level increased to 310 Ϯ 67 pmol/mg/min upon exposing the membranes to 10 M forskolin. As described in Fig. 4A, isoproterenol-mediated activation of adenylyl cyclase by D382L or F383H was comparable with that of the empty pcDNA vector. In cells expressing Gly 389 or Arg 389 variants of the WT ␤ 1 -AR, basal activities of adenylyl cyclase were 15 Ϯ 3 and 18 Ϯ 4 pmol/mg/min, respectively (Fig. 4B). The EC 50 value of the Gly 389 var-iant in stimulating adenylyl cyclase was 11 Ϯ 3 nM and that of the Arg 389 variant was 3 Ϯ 2 nM (p Ͻ 0.05), but their maximal stimulations of adenylyl cyclase activities were comparable (155 Ϯ 23 pmol/mg/min for the Gly 389 versus 167 Ϯ 30 pmol/mg/min for Arg 389 , n ϭ 6, p Ͼ 0.1).
The basal activities of adenylyl cyclase for R384E or R384Q were 62 Ϯ 11 pmol/mg/min. Maximal activation of adenylyl cyclase by R384Q was 218 Ϯ 38 pmol/mg/min (n ϭ 5), which was attained at 10 nM or less of isoproterenol. Similarly, maximal activation of adenylyl cyclase by R384E occurred at 12 nM isoproterenol (EC 50 ϭ 0.4 Ϯ 0.06 nM), but the absolute values for its maximal activation were comparable with those achieved by the WT ␤ 1 -AR (Fig. 4B). The profile of the other Arg 384 or Lys 385 point mutants in activating adenylyl cyclase was comparable (EC 50 ϭ 79 Ϯ 17 nM), but their maximal activation of the cyclase was 45 Ϯ 11% of that attained by the WT ␤ 1 -AR.
The allosteric ternary complex model for GPCR activation predicts that constitutively active GPCR display increased affinity for agonists and decreased affinity for inverse agonists as compared with wild type GPCR (27). Metoprolol is a selective ␤ 1 -AR blocker that is widely used for the management of essential hypertension, which acted as a weak inverse agonist in cells expressing constitutively active ␤ 1 -AR mutants (28). Basal adenylyl cyclase activity in the presence of 100 M metoprolol was reduced by 25 Ϯ 3% in WT ␤ 1 -AR membranes, but the magnitude of metoprolol-mediated inhibition of adenylyl cyclase in R384E or R384Q membranes was significantly higher at 52 Ϯ 4 and 55 Ϯ 6%, respectively (Table 3). Therefore, mutagenesis of R384E or R384Q produced a gain of function profile that is more commonly associated with constitutively active receptors.
Desensitization of Helix 8 Point Mutants-Resistance to agonist-mediated desensitization could account for the functional differences between the various helix 8 mutants because desensitization controls the duration of the cellular signal generated by the GPCR (29). Desensitization was studied by comparing adenylyl cyclase activities in membranes prepared from cells that were exposed to buffer or 10 M isoproterenol for 20 min (Fig. 5). In the case of the Gly 389 variant of the WT ␤ 1 -AR, desensitization was manifested by a 48% reduction of maximal adenylyl cyclase activity and by a right-ward shift of the dose-response curve that increased the EC 50 from 15 Ϯ 3 to 32 Ϯ 6 nM (Fig. 5A). The EC 50 value of G389R in activating adenylyl cyclase also increased from 5 Ϯ 2 to 23 Ϯ 6 nM in membranes prepared from cells exposed previously to isoproterenol (Fig. 5B). The dose-response curves of R384E or R384Q were neither shifted to the right nor appreciably desensitized, because the EC 50 values in both cases were between 2 and 5 nM (Fig. 5,  C and D). The dose-response curves for the other Arg 384 point mutants such as R384F or Lys 385 point mutants such as K385E in activating  adenylyl cyclase were shifted to the right causing a 3-4-fold reduction in their EC 50 values and about 50% reduction in their maximal adenylyl cyclase activities (Fig. 5, E and F). Therefore, neutralization or reversal of the charge of Arg 384 resulted in constitutively active ␤ 1 -ARs that were resistant to agonist-induced desensitization. Phosphorylation of Helix 8 Point Mutants-Another characteristic of constitutively active and desensitized GPCR is that they are constitutively phosphorylated (30). To determine whether the different extent of desensitization among the various helix 8 mutants could be explained by differences in their phosphorylation, basal and isoproterenol-mediated phosphorylation of these receptors was determined in cells expressing equivalent amounts of receptor densities (Fig. 6). As shown in Fig. 6, exposure of cells expressing Gly 389 to isoproterenol resulted in receptor phosphorylation that was rapid (maximal response occurred in ϳ3 min) and resulted in a 530 Ϯ 80% increase in 32 P incorporation above basal levels (n ϭ 6). Basal phosphorylation of G389R was higher than Gly 389 , which is expected because basal cyclic AMP levels of G389R were higher (Fig. 3). The phosphorylation of G389R in response to isoproterenol was increased by 200 Ϯ 43% above basal (n ϭ 3), whereas in the F383H mutant we did not observe appreciable 32 P incorporation either in the presence or absence of isoproterenol. The magnitudes of the increase in the Lys 385 mutants were comparable with that observed in the WT ␤ 1 -AR (data not shown). Basal phosphorylation levels of the constitutively active R384E and R384Q were significantly higher than the WT ␤ 1 -AR (410 Ϯ 50%, n ϭ 3). The incorporation of 32 P into R384E and R384Q did not appreciably increase in response to isoproterenol, indicating that these receptors were already maximally phosphorylated. Therefore, mutagenesis of R384E and R384Q resulted in constitutively phosphorylated and desensitized ␤ 1 -AR mutants.
Distribution and Recycling of Helix 8 ␤ 1 -AR Mutants-The distribution of the WT ␤ 1 -AR and the other helix 8 mutants in HEK-293 cells was determined by confocal microscopy (Fig. 7). The cells were transfected for 2 days with the various FLAG-tagged ␤ 1 -AR constructs or with empty pCMV-Tag 2 that expresses a 37-amino acid FLAG-tagged peptide. The pCMV-Tag2 vector was used to compare the distribution of the FLAG tag alone with that of the FLAG-tagged ␤ 1 -AR. To visualize the various FLAG-tagged constructs, the cells were labeled for 1 h at 37°C with FITC-conjugated anti-FLAG IgG and then fixed and imaged with the Zeiss LSM-510 (Fig. 7A). The images revealed that the various FLAG epitopes were expressed in the cell membrane (Fig. 7A, panels a,  j, s, bЈ, and kЈ). To confirm these data, the cells were exposed to mild acidic conditions to remove the FITC-labeled anti-FLAG antibody from the extracellular lining of the cell membrane (Fig. 7A, panels b, k, t, cЈ,  and lЈ). FITC anti-FLAG IgG labeling was stripped from cells expressing the various FLAG-tagged ␤ 1 -AR (Fig. 7A, panels k, t, cЈ, and lЈ). However, the localization of the FLAG peptide in cells expressing pCMV-Tag-2 was consistent through out the entire experiment (Fig. 7A, panels  a-i).
Exposing the various cell lines to isoproterenol for 30 min promoted the internalization of Gly 389 , G389R, and the other helix 8 mutants into distinct intracellular punctate vesicular structures that are associated with the internalized GPCR phenotype (Fig. 7A, panels l, u, dЈ, and mЈ). As expected, the internalized ␤ 1 -AR constructs were resistant to acid treatment (Fig. 7A, panels m, v, eЈ, and nЈ).
To determine the recycling kinetics of the various ␤ 1 -AR constructs, the cells were exposed to the ␤-antagonist alprenolol to segregate receptor recycling from internalization. Gly 389 and G389R recycled rapidly (Fig. 7A, panels n-q and w-z) and completely as indexed by acid stripping (Fig. 7A, panels r and aЈ). The recycling of R384F and K385E appeared to be similar to that of the WT ␤ 1 -AR (Fig. 7A, panels fЈ-iЈ and oЈ-rЈ). The recycling kinetics of Gly389, G389R, R384F, and K385E were comparable at a t 0.5 of 18 Ϯ 5 min (Fig. 7B).
In separate sets of experiments, we determined the cellular localization of the R384E, R384Q, and D382L mutants in permeabilized cells to observe cell surface and internalized ␤ 1 -AR populations (Fig.  8A). Unlike the WT ␤ 1 -AR, which was localized to the cell surface (Fig. 8A, panel a), these mutant receptors were mainly located in intracellular cytoplasmic vesicles (Fig. 8A, panels e, i, and m). Because the inverse agonist metoprolol markedly reduced basal adenylyl cyclase activities of R384E and R384Q (Table 3), we investigated whether metoprolol would also affect cellular localization of these receptors. Cells expressing the ␤ 1 -AR, as described in Fig. 8A, were treated with 100 M of metoprolol for 2 h, fixed, permeabilized, and then labeled with FITC to detect intracellular and membranous populations of the ␤ 1 -AR. The images reveal that continuous exposure to 100 M metoprolol for 2 h did not affect the distribution of The slides were incubated with FITC-anti FLAG M2 antibody for 1 h at 37°C and then exposed to 10 M isoproterenol (ISO)for 30 min. The cells were acid-washed (0.5 M NaCl, 0.2 M acetic acid for 4 min at 4°C) to strip off the antibody bound to the extracellular lining of the cell membrane. ␤ 1 -AR recycling was initiated by exposing the cells to 100 M of the ␤-adrenergic antagonist alprenolol at 37°C. After each period, the cells were fixed and then visualized by confocal microscopy. At the end of the recycling period, the cells in panels i, r, aЈ, jЈ, and tЈ were subjected to a second acid wash to strip the antibody from the recycled receptors. Each image is a representative image of at least 10 separate experiments involving five or more cells. B, quantification of ␤ 1 -AR recycling kinetics by confocal recycling assays. The LSM-510 software was used to determine the pixel densities after each recycling period. The time constants () for ␤ 1 -AR recycling were determined by fitting the data to a single exponential decay function of the y ϭ y o ϩ Ae Ϫt/ , where y o and A are constants. Each scale bar represents 5 m.

FIGURE 6. Isoproterenol-mediated phosphorylation of the wild type and its helix 8 mutants in HEK-293 cells.
HEK-293 cells expressing the various FLAG-tagged ␤ 1 -AR were metabolically labeled with 32 P i O 4 in phosphate-free DMEM and exposed to either to buffer (Ϫ) or 10 M isoproterenol (ISO) (ϩ) for 10 min. Cells were lysed in radioimmune precipitation buffer, and equal amounts of protein were incubated overnight with anti-FLAG M2 IgG conjugated to agarose. Anti-FLAG immunoprecipitates were subjected to SDS-PAGE, and the gels were dried and subjected to autoradiography. The amount of 32 P incorporated into␤ 1 -AR monomers (73 kDa) was determined by electronic autoradiography. The average counts/ min Ϯ S.E. incorporated into each ␤ 1 -AR monomer band from n ϭ 3 separate experiments were plotted. Statistical analyses were based on analysis of variance with post hoc tests.
the WT ␤ 1 -AR (Fig. 8A, panel b) but caused the redistribution of intracellular R384E and R384Q to the cell membrane (Fig. 8A, panels f and j). A circular boundary was drawn around the inner circumference of cell expressing R384E and R384Q to determine the distribution of these receptors between membranous and intracellular compartments. The density of the pixels residing inside the boundary versus those residing outside the boundary was used as an index for internalized and membranous ␤ 1 -AR, respectively. In control R384Q and R384E, 84 Ϯ 7% of these receptors was intracellularly distributed. To estimate the effect of the inverse agonist, internal pixel densities in control cells were arbitrarily set as 100, and the data for each inverse agonist were calculated as percent of this value (Fig. 8B). Metoprolol reduced the percent of intracellular R384Q and R384E to 11 Ϯ 4 and 8 Ϯ 3%, respectively (Fig. 8B).
The phenomenon of inverse agonist-mediated redistribution of the internalized constitutively active GPCR is referred to as "externalization" (31). If externalized R384E and R384Q are inserted properly into the cell membrane, then FITC-conjugated anti-FLAG IgG bound to the N-terminal FLAG epitope will be oriented extracellularly. In this case, a second acid wash would strip FITC-IgG from the externalized receptor population. In agreement with these assumptions, we observed reduced cell fluorescence in metoprolol-treated R384E and R384Q cells after a brief acid treatment (Fig. 8A, panels h  and l). Metoprolol-mediated translocation of R384Q to the plasma membrane resulted in a 24% increase in the density of cell surfacebinding sites on R384Q expressing cells as measured by [ 3 H]CGP-12177 binding (data not shown). We screened another inverse ago-nist CGP-20712A (28) and a pure antagonist (propranolol) for their ability to "externalize" R384E and R384Q. CGP-20712A was equally efficacious to metoprolol in "externalizing" R384E and R384Q (Fig.  8A, panels g and k), whereas propranolol did not cause significant externalization (data not shown). The effect of the inverse agonists metoprolol or CGP-201712A on the distribution of D382L was also investigated (Fig. 8A, panels n and o). The intracellular distribution of D382L was not altered by the inverse agonist metoprolol as determined by the pixel distribution method (Fig. 8A, panel n) or by the acid-strip method (Fig. 8A, panel p, and Fig. 8B). CGP-20712A reduced by 30% the density of intracellular pixels of D382L (Fig. 8B).
The predicted three-dimensional structure of the 8th ␣-helix in Fig. 9 reveals that the highly conserved hydrophobic residues Phe 383 and Phe 387 were oriented toward the hydrophobic milieu of the inner cell membrane or lipid raft. In addition, helix 8 is bordered at its N terminus by the linker sequences of Cys 379 , Ser 380 , and Pro 381 that project from TM7 and at its C terminus by a pair of palmitoylated cysteines that are buried into the cytoplasmic face of the membrane. These structures at both ends of helix 8 tether the predicted helix to the membrane, and along with the hydrophobic phenyl alanines orient helix 8 into a partially buried ␣-helix on one side and an exposed ␣-helix on the opposite side. Our predictions indicate that the residues in the exposed top portion of helix 8 are Lys 385 and Arg/Gly 389 . The other residues such as Ala 386 and Leu 390 are aligned on one side of the 8th ␣-helix (Fig. 9A), whereas Asp 382 , Arg 384 , and Gln 388 lie on the other side of putative helix 8 (Fig. 9B).

FIGURE 8. Effect of inverse agonists on the distribution of R384Q, R384E, and D382L in HEK-293 cells.
A, cells stably expressing the indicated FLAG-tagged ␤ 1 -AR were grown on poly-L-lysine-covered coverslips. The slides were incubated either with buffer or with 100 M of metoprolol or CGP-20712A for 2 h at 37°C. The slides were fixed in 4% paraformaldehyde in 4% sucrose in PBS, pH 7.4, for 10 min at room temperature, and then the cells were permeabilized and incubated with FITC-conjugated to anti-FLAG M2 antibody. In panels d, h, l, and p, the slides were acid-washed (0.5 M NaCl, 0.2 M acetic acid) to strip FITC-IgG bound to the extracellularly oriented FLAG epitope. B, the distribution of receptors between the membranous and intracellular compartments was determined by drawing a circular boundary around the inner circumference of the cell. The density of pixels residing inside the boundary versus those residing outside the boundary was used as an index for internalized and membranous ␤ 1 -AR, respectively. To estimate the effect of the inverse agonist on intracellular ␤ 1 -AR distribution, internal pixel densities in control cells were arbitrarily set as 100, and the data for each inverse agonist was calculated as percent of this value. In slides (panels d, h, l, and p), total cellular pixel densities were divided by the internal pixel densities in control cells to calculate their percentile values. The data represent the mean Ϯ S.E. from five slides each involving between 6 and 10 cells. Each scale bar represents 5 m.

DISCUSSION
Several reports indicate that the putative 8th ␣-helix formed between the NPXXY motif in TM7 and by membrane insertion of conserved palmitoylated cysteines may be involved in the interaction of GPCR with their cognate G proteins. Synthetic peptides from the C terminus of the ␣-subunit of transducin interacted with rhodopsin and kept it in the active state (9). This interaction, however, was abolished in mutants of rhodopsin in which the Asn 310 , Lys 311 , and Gln 312 sequences in helix 8 were replaced with the corresponding sequences (Ser, Pro, and Asp) in the ␤ 1 -AR or the ␤ 2 -AR. In other experiments, the substitution of the Asn-Lys-Gln sequence at the N terminus of helix 8 in rhodopsin with the Ser-Pro-Asp sequence inhibited the activation of transducin by rhodopsin (8). These data are intended to supplement other reports showing that G protein subunits interact directly or indirectly with helix 8 (6). Discrete point mutations within the Asn 310 -Lys 311 -Gln 312 sequence in helix 8 of rhodopsin engineered in Lys 311 revealed that Lys 311 mutants were not defective in transducin activation (8). In another study, it was shown that both Asn 310 and Lys 311 were required for correct folding of rhodopsin (32). Therefore, the Asn-Lys-Gln sequence is involved in proper folding of rhodopsin, but the specific residue(s) in this sequence that are involved in promoting rhodopsin-transducin interactions are still obscure.
The sequence that corresponds to the Asn-Lys-Gln sequence of rhodopsin in the ␤ 1 -AR and in the ␤ 2 -AR is Ser 380 -Pro 381 -Asp 382 (Table 1). We analyzed the role of Asp 382 in ␤ 1 -AR-mediated activation of G s because this residue is outside the Ser-Pro sequence that was involved in proper folding of rhodopsin. D382L was deficient in its activation of G s , indicating that Asp 382 was a potential player in coupling helix 8 to G s . Analysis of the role of the N terminus of predicted helix 8 in the human leukotriene B 4 (BLT-1) receptor indicates that this domain is involved in sensing the status of the BLT-1 receptor coupling to the G protein because deletion of the N-terminal sequence of helix 8 prevented GTP␥S-mediated inhibition of leukotriene B 4 binding (33,34). These mutant BLT-1 receptors also exhibited high B max values accompanied with more prolonged intracellular signaling than in the wild type BLT-1 receptor (33).
The role of the highly conserved Phe residue that projects into the intracellular leaflet of the plasma membrane has generated modest attention because mutagenesis of Phe 313 in rhodopsin and Phe 318 in the melanin-concentrating hormone 1 receptor or Phe 313 in the angiotensin II 1A receptor to Ala produced properly folded and biologically active receptor molecules (10,13,35). We opted instead to test for the effect of a charge substitution at this position, and we found out that it prevented the biosynthesis of the ␤ 1 -AR perhaps by interfering with the proper folding of the receptor. Therefore, a hydrophobic uncharged residue at this position is required for proper expression of the GPCR, indicating that the formation of an amphiphilic 8th ␣-helix is important for folding of the full-length receptor.
To determine the role of the highly conserved Arg 384 residue, we engineered a variety of mutations to test the effect of different chemical structures on the coupling of the ␤ 1 -AR to G s . Mutagenesis of the basic Arg 384 to the two other basic amino acids, Lys and His, or to uncharged Phe indicated that side chain projections play a major role in coupling the ␤ 1 -AR to G s . Our results strongly indicate that Arg at this position provides normal basal levels and displays the highest affinity for isoproterenol in binding to the receptor among those Arg 384 mutants with similar basal activities (Table 2). Furthermore, Arg 384 promoted the highest efficacy in activating adenylyl cyclase among this group of Arg 384 mutants (Fig. 3). The mutation involving a charge reversal to glutamic acid produced a receptor with an exceedingly high affinity in binding to isoproterenol and high basal levels of adenylyl cyclase activity. R384E and R384Q displayed characteristics associated with the constitutively active ␤ 1 -AR phenotype. The pharmacological character of constitutively active ␤ 1 -and ␤ 2 -AR was gleaned from mutagenesis of Leu 322 in helix 6 of the ␤ 1 -AR into Lys and from compound mutations in the cytoplasmic neck of helix 6 of the ␤ 2 -AR (36). The involvement of residues in the cytoplasmic neck of helix 6 in coupling the ␤-AR to G s and in constitutive activation is well documented (37), but our finding that a single point mutation in helix 8 can constitutively activate the ␤ 1 -AR has not been described before. R384E and R384Q displayed the pharmacological profile predicted by the allosteric ternary complex model for constitutively active receptors (36). This model predicts that constitutively active receptors manifest increased affinity for agonists and increased intrinsic activity for inverse agonists as compared with wild type GPCR (27). In agreement with the predictions of the allosteric ternary complex model, the affinity of isoproterenol in binding to R384E and R384Q was markedly increased by 50 -75-fold, whereas the distribution of the low versus high affinity sites was maintained. Another prediction of the ternary complex model stipulates that ␤ 1 -adrenergic inverse agonists would cause stronger inhibition of basal adenylyl cyclase activity in the R384E and R384Q mutants than in the WT ␤ 1 -AR (27,28). The inverse agonist metoprolol preferentially reduced the basal activity of adenylyl cyclase in the R384E and R384Q mutants, indicating enhanced activity of inverse agonists toward the constitutively active ␤ 1 -AR phenotype (Table 3).
Another characteristic of the R384E and R384Q mutants was their resistance to desensitization, which we attribute to constitutive desensitization of these ␤ 1 -AR constructs (30). This finding is surprising con- sidering that the Gly 49 variant of the ␤ 1 -AR displayed the pharmacological character of a constitutively active receptor but was desensitized to a greater extent than the Ser 49 variant of the WT ␤ 1 -AR (38); but this finding is in agreement with the ternary complex model that stipulates that constitutively active receptors are also constitutively desensitized (30).
There were major qualitative differences, however, in the desensitization of adenylyl cyclase in WT ␤ 1 -AR versus R384E and R384Q. The fold increase in adenylyl cyclase activity in control WT ␤ 1 -AR in response to isoproterenol was ϳ9 Ϯ 2-fold over basal, whereas in the desensitized state the fold increase was reduced to 5 Ϯ 2-fold (Fig. 5). The fold increase in isoproterenol-mediated activation of adenylyl cyclase in control or desensitized R384E and R384Q mutants was 3-fold. We attribute the resistance of the R384E and R384Q mutants to desensitization to two complementary processes. The first is that these receptors were constitutively phosphorylated. The counts/min of 32 P incorporated into R384E and R384Q in the absence of isoproterenol were equivalent to the counts/min incorporated into the WT ␤ 1 -AR after isoproterenol. Furthermore, the phosphorylation of R384E and R384Q did not increase after isoproterenol. Because desensitization is a consequence of ␤-adrenergic receptor kinase-mediated phosphorylation (39), basal phosphorylation of R384E and R384Q was already maximal, which blunted further desensitization. The second process is that desensitization involves agonist-mediated sequestration and internalization of the GPCR (40). This phenomenon was exhibited by the WT ␤ 1 -AR, which was rapidly internalized with a t 0.5 of 5 Ϯ 2 min and recycled in a cAMP-dependent protein kinase-dependent manner with a t 0.5 of 18 min (19) (Fig. 7A). The R384E and R384Q mutants, however, were already internalized, and their internal receptor pixels did not appreciably increase after isoproterenol. What accounts for constitutive internalization of R384E and R384Q? To address this question we exposed cells expressing R384E and R384Q to the inverse agonist metoprolol. If R384E and R384Q mimic the behavior of the agonist-activated WT ␤ 1 -AR, then they would be constitutively activated, internalized, and recycled. Thus treating R384E and R384Q with an inverse agonist would be expected to restore their membranous distribution because inverse agonists generate a strong inhibitory effect on constitutively active GPCR (41). What we observed was the redistribution of R384E and R384Q to the cell surface, indicating that when inverse agonists blocked constitutive internalization, R384E and R384Q recycled back and were re-inserted into the cell membrane. Therefore, the R384E and R384Q mutations produced a ␤ 1 -AR phenotype that was constitutively internalized and recycled. These receptors were not degraded because the densities of R384E and R384Q in broken cells were not increased, even though metoprolol increased the densities of R384E and R384Q in membrane fractions by ϳ25%. The behavior of the constitutively active ␤ 1 -AR in this regard was different from the behavior of the constitutively active ␤ 2 -AR. The constitutively active ␤ 2 -AR was associated with constitutive down-regulation and desensitization that resulted in low expression levels (30). This phenotype of the ␤ 2 -AR was reversed by overnight treatment with an inverse agonist that resulted in strong up-regulation of receptor levels and in the stimulation of agonist-promoted activation of adenylyl cyclase (30). Therefore, in this regard, the ␤ 1 -AR resembled the behavior of the constitutively active angiotensin II 1A receptor, which was constitutively internalized and recycled, and its density was not appreciably increased by inverse agonists (31). What is the role of the amino acid corresponding to Arg 384 in coupling other GPCR to their cognate G protein? In the angiotensin II 1A receptor, Lys 310 corresponds to Arg 384 in the human ␤ 1 -AR (Table 1). K310Q and K311Q mutants displayed normal binding parameters and displayed normal activity in activating phospholipase C in response to angiotensin II (13). The corresponding mutations in the melanin-concentrating hormone receptor 1 , R319Q, or R319E reduced the B max by 60%, increased the EC 50 for calcium efflux by 7-fold, and altered the distribution of these receptors from the plasma membrane into the perinuclear zone and cytoplasm (10). Thus, Arg 384 functions in a receptor or G proteinspecific manner.
Inverse agonists or antagonists did not alter the distribution of D382L (Fig. 8A, panels m-p). D382L displayed low binding affinities to agonists and did not increase cyclic AMP or activate cAMP-dependent protein kinase in response to isoproterenol. Because cAMP-dependent protein kinase is required for efficient recycling of ␤ 1 -AR through the endosomal pathway (19), we speculate that the D382L was not efficiently delivered to membranes by the biosynthetic and/or by the recycling routes.
Whatever is the structural basis for the role of helix 8 in coupling the GPCR to G proteins is currently obscure. A fundamental observation gleaned from docking the three-dimensional structures of rhodopsin and transducin is that transducin binds to rhodopsin dimers in such a way that one molecule of rhodopsin binds transducin and the other serves as a docking platform (42). In this complex, the activation of one signaling rhodopsin in the oligomer is needed for productive coupling with transducin. The contact sites between transducin and rhodopsin involve the tips of helices 3, 6, and 8 (42). There were additional interactions between Glu 59 and Lys 68 of the ␥-subunit of transducin with Gln 312 and Thr 319 , respectively, in helix 8 of rhodopsin (42). In addition, mutational and cross-linking studies indicate that the Asn 310 -Gln 312 region in helix 8 of rhodopsin interacts with residues 340 -350 in the C-terminal sequence of the ␣-subunit of transducin (8,9). These and other pieces of data indicate that helix 8 participates in binding to transducin and in rhodopsin-mediated activation of transducin. The analogue situation in other GPCR is not known because rhodopsin is the only GPCR for which the crystal structure has been determined. A similar situation might occur in the human ␤ 1 -AR because our data in Fig. 1 reveal that dimers and monomers of the human ␤ 1 -AR were expressed in HEK-293 cells. Furthermore, modeling of helix 8 of the human ␤ 1 -AR on the three-dimensional structure of rhodopsin revealed that many amino acids were conserved. Our modeling data indicate that the most functionally active amino acids in predicted helix 8 of the ␤ 1 -AR were Asp 382 , Phe 383 , and Arg 384 . Asp 382 and Arg 384 were assigned to the same side of the 8th ␣-helix and may form a grove that is involved in coupling the ␤ 1 -AR to G s .
Currently there is no definite rule that defines the coupling between the GPCR and a given G protein. In rhodopsin, a stretch of three consecutive amino acids (Asn 310 -Lys 311 -Gln 312 ) in the N terminus of helix 8 was shown to interact directly with the C terminus of the G protein transducin (8,9). The amino acids in putative helix 8 of selected GPCR that have been implicated in coupling the GPCR to G proteins are shaded in Table 1. In the angiotensin II 1A receptor, a string of three amino acids (Tyr 312 -Phe-313 -Leu 314 ) in the C terminus of helix 8 was essential for the coupling and activation of G q (13). In the melaninconcentrating hormone receptor 1 , a succession of three amino acids (Arg 319 -Lys 320 -Arg 312 ) participated equally in cell expression, whereas Arg 319 and Lys 320 were partially involved in coupling the receptor to G i and G q (10). In the oxytocin receptor, the charged amino acids His 335 , Glu 339 , and Arg 343 that are separated from one another by one turn of the putative ␣-helix were involved in oxytocin-mediated activation of phospholipase C by G q (14). In the bradykinin 2 receptor, Arg 311 in helix 8 (11) and Tyr 320 in the C-terminal tail (43) were involved in coupling the receptor to phosphatidylinositol hydrolysis via G q .
The GPCR with the most intriguing correlations between the structure of its hypothetical helix 8 and mutant receptor function is the BLT-1 receptor (34). The putative helix 8 of the BLT-1 receptor is linked to the NPXXY motif of helix 7 by a long linker and lacks the cysteine residue, which is palmitoylated in many GPCR (44). Substitution of Leu 304 and Leu 305 in helix 8 by alanine caused constitutive BLT-1 recep-tor activation, increased the B max values, and prevented leukotriene B 4 -mediated internalization of these receptors in COS-7 cells (45). BLT-1 receptors with truncated or substituted helix 8 maintained high affinities for binding leukotriene B 4 in HEK-293 and Chinese hamster ovary cells, even in the presence of excess amounts of GTP␥S, and these changes were associated with prolonged Ca 2ϩ mobilization and cellular metabolic activation (33,34). These studies suggest that predicted helix 8 in the BLT-1 receptor might play an important role in the inactivation of BLT-1 by sensing the status and composition of the G proteins that are coupled to it (33,34,46).
In the human ␤ 1 -AR, three consecutive amino acids (Asp 382 -Phe 383 -Arg 384 ) were implicated in regulating the expression, localization, and coupling of this receptor to G s . Thus several structural determinants within helix 8 appear to influence the proper recognition of the G protein by the ␤ 1 -AR. The involvement of different motifs within predicted helix 8 in binding to G proteins is dictated by differences in amino acid composition and sequences of helix 8 among the various GPCR and by the diversity of the G proteins that interact with these receptors. One would expect that the G q -coupling motif in helix 8 would be different from the G s -coupling motif because these G proteins have different ␣-, ␤-, and ␥-subunits with varying contact points among them to the GPCR.
The role of putative helix 8 in coupling the GPCR to G proteins illustrates the importance of amino acids outside the classical GPCR binding pocket in regulating receptor-mediated signaling effects. These observations along with the realization that GPCR and G proteins do not exist as 1:1 complexes but rather as heterodimers, involving more than one molecule of each, provides additional mechanisms for GPCR-mediated activation of the G protein. In this regard, helix 8 may serve as a novel site for allosteric modulation of the GPCR (47). Allosteric modulation relies on the organization of multiprotein subunits into discrete conformations capable of undergoing reversible transitions in the absence of ligand (48). The activity of these complexes can be regulated by ligand binding to domains that are topographically distinct from the binding pocket, "the orthosteric site," within the hydrophobic core of the class A GPCR. The amino acids of helix 8 lie outside the catecholamine binding pocket (49) and are involved in heterooligomerization of the GPCR with the G protein. Furthermore, distinct motifs in helix 8 are used by each GPCR in coupling to their specific G proteins. These observations suggest that allosteric ligands designed to interact with helix 8 might be specific for each GPCR because of the diverse nature of helix 8 sequences among the various GPCR subclasses (50) and the distinctiveness of their G protein binding and interacting pockets. These novel allosteric agents, could either "tune up" or "tune down" the signaling of the agonist-occupied GPCR with more selectivity than traditional "orthosteric" agents (50)