Identification of a Conserved Switch Residue Responsible for Selective Constitutive Activation of the b 2 -Adrenergic Receptor*

A cysteine-to-phenylalanine mutation of residue 116 in the third transmembrane domain of the b 2 -adrenergic receptor caused selective constitutive activation of Na 1 /H 1 exchange through a pathway not involving cAMP. This selectivity was identified by comparing binding and signaling characteristics of wild-type (WT) versus C116F mutant receptors transiently transfected into COS-1 cells. Indicating constitutive activity, ligand binding to the C116F mutant showed a 78-fold higher than WT affinity for isoproterenol and a 40-fold lower than WT affinity for ICI 118551. Although agonist-inde-pendent activation of cAMP production was not exhibited by the C116F mutant, a constitutive stimulation of the Na 1 /H 1 exchanger (NHE1) was observed. This was identified by measuring either basal intracellular pH (pH i ) or rate of pH i recovery from cellular acid load. Due to a higher rate of H 1 efflux through NHE1, C116F transfectants exhibited a significantly higher pH i (7.42) than did WT transfectants (7.1). Furthermore, the rate of pH i recovery from acid load facilitated by NHE1 was 2.1-fold faster in mutant transfectants than in WT transfectants. The lower rate seen in the WT case was stimulated by epinephrine, and the higher rate seen in the mutant case was inhibited by ICI 118551. These findings, which show that a C116F mutation of the b 2 -adrenergic recep- tor evokes selective constitutive coupling to NHE1 over cAMP, form the basis of our prediction that multiple and distinct activation states can exist in G protein-coupled binding studies performed with in competition binding experiments. mini-mize variation, and binding experiments were performed simultaneously. Statistical significance in both binding and functional assays identified by the two-tailed t

The ␤ 2 AR 1 is one member of a large family of G proteincoupled receptors that mediate the effects of numerous peptidic and nonpeptidic hormones and neurotransmitters. As with all G protein-coupled receptors, the common structural feature of this family is a single polypeptide chain with seven hydrophobic regions that constitute seven transmembrane-spanning domains. Classification of adrenergic receptors (␣ 1 , ␣ 2 , and ␤), which as a family mediate the physiological effects of the sym-pathetic nervous system, is based on homology of amino acid sequences and pharmacology of various ligands (1). In the case of the ␤ 2 -subtype, dogma suggests that the endogenous catecholamines epinephrine and norepinephrine promote receptor coupling to G ␣s and downstream cAMP production (2,3). Since ␤ 2 ARs are most abundant in smooth muscle, the best characterized effects of cAMP activation include smooth muscle relaxation in bronchial tubes and in vascular tissue (4).
The process of receptor activation and G protein coupling that evokes cellular effects is described by the widely accepted revised ternary complex model (5,6). In this allosteric model, the active conformation of a native receptor is the cornerstone of the agonist-receptor-G protein complex that leads to signaling. Without agonist present, the model predicts spontaneous receptor isomerization between the inactive (R) and active (R*) conformations, with equilibrium under native conditions shifted toward R. At any given time, even though most receptors reside in R, a small population will reside in R*, permitting formation of the R*⅐G protein complex that causes effector activation. The model predicts that the addition of agonist does not directly convert the receptor from R to R*. Rather, the agonist will preferentially bind to receptors already in R*, thereby shifting isomerization equilibrium away from R. Receptor mutations that induce an agonist-independent shift in isomerization equilibrium toward the R* conformation are termed constitutively active and so by definition couple to and evoke second messenger responses in the absence of agonist.
The studies contributing to the development of the revised ternary complex model were based primarily on the characterization of ␤ 2 AR signaling, which was assumed to be through a single effector pathway (adenylate cyclase via coupling to G ␣s ). However, recent studies indicate that besides coupling to G ␣s , ␤ 2 ARs also couple to an alternate protein in the G ␣ family (possibly G ␣13 ), which leads to newly discovered cellular effects, including stimulation of NHE1 (7)(8)(9), via Cdc42-and RhoA-dependent pathways (10). Therefore, in the present study, ␤ 2 AR signaling via both cAMP production and Na ϩ /H ϩ exchange was examined in a receptor that was engineered to have a single amino acid mutation of Cys-116 in the third transmembrane domain. Cys-116 is situated approximately one helical turn below Asp-113, which is the putative counterion that binds the protonated amine of adrenergic ligands (11,12). Because of the important role of Asp-113 in ligand binding, any perturbation of the third transmembrane helix could affect both the ligandbinding pocket and second messenger coupling. Interestingly, the C116F mutation induces a receptor conformation that constitutively activates Na ϩ /H ϩ exchange while only maintaining competent coupling to cAMP. This finding is similar to the result observed in an analogous C128F mutation of the ␣ 1b AR, which exhibits selective constitutive coupling to inositol metabolism over the arachidonic acid pathway (13). Work in our laboratory with C116F ␤ 2 AR and C128F ␣ 1b AR has formed the basis of an emerging hypothesis suggesting that receptors are capable of forming multiple activation states that are G protein-specific. An understanding of conformational differences between these distinct states may eventually lead to the design of signaling-specific therapeutic agents. At least in the case of receptors that couple to more than one G protein, we suggest that the current model of receptor activation be revised to include isomerization from the R conformation to distinct active conformations that exhibit G protein selectivity.
Site-directed Mutagenesis-The cysteine-to-phenylalanine mutation was made at residue 116 of the human ␤ 2 AR using cassette replacement as described previously (14). cDNAs were sequenced by the dideoxy method (Sequenase kit, Amersham Corp.) to confirm the mutation. The synthetic human ␤ 2 AR was then subcloned into the eukaryotic expression plasmid pMT2Ј (15) and purified.
Cell Culture and Transient Transfection-COS-1 cells (American Type Culture Collection), known to be ␤AR-negative based on the lack of epinephrine-evoked cAMP responses (16), were grown in Dulbecco's modified Eagle's medium (Sigma) ϩ 10% fetal bovine serum (Life Technologies, Inc.). Cells were plated onto either 150-or 60-mm tissue plates for membrane preparations or second messenger studies, respectively, or onto UV-clear ⌬T dishes (Bioptechs) for fluorescence assay. Transient transfection was achieved by the DEAE-dextran method (17). Efficiency of transfection was typically near 20%. Cells were harvested or assayed 60 h after transfection.
Membrane Preparation and Radioligand Binding-COS-1 cell membranes were prepared as described previously (15). Membrane protein concentration was determined using the Bradford method (18). Competition binding experiments with the antagonist radioligand [ 125 I]CYP were carried out in a final volume of 0.25 ml containing HEM buffer ϩ 0.1% bovine serum albumin, 150 pM [ 125 I]CYP, COS-1 membranes, and varying concentrations of unlabeled ligand. Nonspecific binding was determined in the presence of 10 Ϫ5 M propanolol. Reactions were run at room temperature for 1 h, after which time the mixtures were filtered onto Whatman GF/C glass fiber filters using a Brandel cell harvester. Bound radioactivity was quantitated with a Packard Auto-gamma 500 counter. Binding data were analyzed with the iterative curve-fitting software package GraphPad Prism. Saturation binding studies were performed with increasing concentrations of [ 125 I]CYP (5-600 pM) in the same buffer as used in competition binding experiments. To minimize variation, WT and C116F ␤ 2 AR binding experiments were always performed simultaneously. Statistical significance in both binding and functional assays was identified by the two-tailed Student's t test.
cAMP Determination-Accumulation of cAMP in WT and C116F ␤ 2 AR transfectants was measured using a commercially available cAMP assay system (Amersham Corp.) according to the directions supplied by the manufacturer. Cell extracts were derived from cultures in 60-mm dishes that were preincubated for 30 min with 5 mM theophylline and then for 30 min with both theophylline and increasing concentrations of agonist.
Measurement of Intracellular pH Using BCECF-Transfected COS-1 cells, cultured in ⌬T dishes, were incubated for 15 min at 37°C with 0.3 M BCECF/AM (Calbiochem). Unincorporated dye was washed away by perfusion with fluorescence medium until extracellular fluorescence was undetectable. Intracellularly trapped BCECF fluorescence was measured with a Delta Scan dual wavelength spectrofluorometer (Photon Technology International) functionally linked to an Olympus inverted fluorescence microscope. Cells were alternately excited with 440and 500-nm light while the intensity of the 530 nm emission was measured with a photomultiplier tube. Emission intensities at the two exciting wavelengths were collected every 0.2 s during continuous perfusion of the cells with medium. Using an adjustable shudder, the emission from an area containing only 20 -30 cells was collected for analysis. Cellular autofluorescence and contaminating fluorescence from leaked dye were found to be negligible, but background corrections were made to remove contributions from ambient light. Excitation ratios (500/440 nm) were calibrated for pH i with 10 M nigericin in potassium-based fluorescence medium as described previously (19). Fig.  1 shows the averaged results of seven separate calibrations that indi-cate a linear relationship between the 500/440 nm ratio and pH i in the physiological range (pH 6.3-7.8). For each experimental ratio collected, a pH i value was extrapolated from this calibration plot. For certain experiments, cells were acid-loaded via a 6-min transient exposure to 20 mM NH 4 Cl (20). Rate of recovery of pH i from this induced acidosis was determined by calculating the derivative of the slope of the pH i time course tracing for 10-s intervals through the first 120 s of recovery.

Pharmacological Comparison of WT and Mutant Receptors-
Receptor density and [ 125 I]CYP binding affinity were determined in COS-1 cells transfected with either the WT ␤ 2 AR cDNA or the C116F mutant. [ 125 I]CYP labeled an apparently homogeneous population of binding sites with similar affinity in membranes prepared from both transfected cell populations. Binding of the radioligand was statistically best fit to a one-site model with mean K D values for the WT and C116F receptors of 97.3 and 125.1 pM, respectively (data not shown). When transfecting cells cultured in 150-mm plates with 8 g of plasmid DNA, the mean WT and mutant receptor densities were 0.398 and 0.058 pmol/mg, respectively.
The potency of various ligands to inhibit specific [ 125 I]CYP binding to membranes from WT or C116F receptor-transfected COS-1 cells is shown in Table I. ␤ 2 -Agonists and partial agonists that were tested showed a 6 -78-fold higher binding affinity for C116F mutant receptors than for WT receptors. The largest leftward shift in affinity for the mutant receptor was seen with the full agonists isoproterenol and epinephrine, which showed 78-and 49-fold lower K i values than for the WT receptor, respectively. Comparatively, the antagonists ICI 188551, pindolol, and alprenolol exhibited the binding properties of inverse agonism by showing a 3-40-fold higher affinity for the WT receptor than for the C116F mutant. The most dramatic shift was seen with ICI 118551, which showed a 40-fold higher K i for the mutant receptor. Propanolol, another antagonist tested, exhibited the binding properties of neutral antagonism by not showing a significantly different K i for one receptor compared with the other.
The potential role of G protein precoupling in the observed FIG. 1. pH i versus the 500/440 nm fluorescence excitation ratio. Untransfected and WT and C116F ␤ 2 AR-transfected COS-1 cells were loaded with BCECF and prepared for fluorescence assay. Equilibration of intra-and extracellular pH was achieved by permeabilizing the plasma membrane to H ϩ with potassium-based fluorescence medium containing 10 M nigericin. The pH of the potassium-based fluorescence bathing solution was varied in nine steps as determined by pH electrode: 6.28, 6.46, 6.66, 6.86, 7.07, 7.27, 7.46, 7.67, and 7.87. Each pH step was perfused over the cell layer for 4 min, after which time the 500/440 nm fluorescence excitation ratio was determined. Values shown are the mean ratio Ϯ S.E. calculated at each pH step (n ϭ 7). The standard curves seen in untransfected and WT and C116F receptortransfected cells were similar. Therefore, calibration data from all three cases were averaged to generate the composite curve shown.
shift of agonist affinity seen for the C116F mutant was examined by performing competition studies in the absence and presence of 0.1 mM Gpp(NH)p. K i values and Hill coefficients for isoproterenol binding to mutant receptors were not affected by the presence of Gpp(NH)p (data not shown).
Stimulation of cAMP Formation-Functional coupling of WT and C116F ␤ 2 ARs to the cAMP pathway was examined by measuring accumulation of cAMP. For this experiment, the transfection protocol was modified to assure similar receptor expression levels in WT and C116F transfectants. This was accomplished by titering down the amount of WT cDNA introduced to the cells until B max was approximately equal to values seen in the mutant case. Mean receptor expression in titered WT transfectants was 0.063 pmol/mg compared with 0.056 pmol/mg in mutant transfectants. Fig. 2 compares the isoproterenol concentration dependence of cAMP production in cells expressing similar levels of either WT or C116F receptors. Arguing against constitutive coupling to G ␣s and cAMP production, the C116F mutant receptor did not evoke a response in the absence of isoproterenol, and it did not affect the potency of isoproterenol to evoke a response compared with the WT receptor. The agonist concentration curves seen in the two transfected cell populations were nearly identical, with mean EC 50 values of 90.8 and 79.3 nM, respectively. Like isoproterenol, epinephrine concentration-response curves and EC 50 values seen in WT and C116F transfectants were not different (data not shown).
Stimulation of Na ϩ /H ϩ Exchange-A possible alternative constitutive coupling of C116F ␤ 2 ARs to G ␣13 and downstream NHE1 was initially examined by studying the effect of this hypothetical constitutive activity on basal pH i . Continuous stimulation of H ϩ extrusion through the Na ϩ /H ϩ exchanger by a constitutively active C116F mutant receptor would likely cause cytosolic alkalosis. Therefore, using basal pH i as a marker of constitutive activity, resting pH i levels in COS-1 cells were determined. Fig. 3 illustrates that C116F receptortransfected cells had a significantly more alkaline pH i (7.42) than either WT transfectants (7.1) or untransfected cells (7.04). This was despite 6.8-fold more receptors on WT than on mutant cell membranes. As expected, intracellular alkalosis in the C116F transfectants was inhibited by inactivating the mutant receptors with the inverse agonist ICI 188551 (10 Ϫ5 M).
The apparent constitutive coupling of the C116F ␤ 2 AR to Na ϩ /H ϩ exchange was further examined by measuring the rate of pH i recovery from cellular acid load. COS-1 cells transfected with either the WT or C116F ␤ 2 AR cDNA were loaded with H ϩ by a 6-min exposure to 20 mM NH 4 Cl. Rate of recovery from this acid load (dpH i /dt) was compared in WT and C116F transfectants under various conditions. In an attempt to factor out effects on H ϩ extrusion rate caused by the ion's concentration gradient, comparisons were only made between groups of cells that loaded to a similar minimum pH i in the range between 6.35 and 6.4. A time course of the pH i changes that occur during acid load in C116F receptor-transfected cells is shown in Fig. 4. The initial alkalosis upon addition of NH 4 Cl was due to the rapid internalization of NH 3 and its association with H ϩ to form NH 4 ϩ (20, 21). Upon abrupt removal of the NH 4 Cl, intracellular NH 4 ϩ dissociates to form NH 3 , which rapidly leave the cells, and H ϩ ions, which are retained intracellularly. Recovery of the cells from this acid load was detected as a net alkalinization of the cytosol. Also shown in Fig. 4, facilitation of this pH i recovery was confirmed to be NHE1-specific based on sen-  Selective Constitutive Signaling in ␤ 2 -Adrenergic Receptors sitivity to 50 mM dimethylamiloride (22). The time course of pH i recovery following acid load was compared in COS-1 cells expressing either WT or C116F ␤ 2 ARs. A significantly more rapid recovery was seen in C116F transfectants than in either WT transfectants or untransfected control cells (Fig. 5A). As summarized in Fig. 5B, cells expressing the mutant receptor showed a 2.1-fold higher dpH i /dt than WT transfectants. WT receptor-transfected cells and untransfected cells recovered from the acid load with a similar rate. Furthermore, activation of WT receptors with 10 Ϫ6 M epinephrine stimulated dpH i /dt to the rate seen in C116F transfectants, whereas treatment with 10 Ϫ5 M ICI 118551 had no effect. Comparatively, treatment of the C116F mutant receptor with 10 Ϫ5 M ICI 118551 significantly reduced dpH i /dt to a value near that seen in WT transfectants, whereas stimulation with 10 Ϫ6 M epinephrine had no effect. DISCUSSION In the current revised ternary complex model describing the activation of G protein-coupled receptors (5, 6), agonists do not directly drive the inactive receptor to assume the active conformation. Rather, the model predicts that agonists will preferentially bind to a receptor's R* conformation, thus stabilizing R* and shifting the isomerization equilibrium away from R. Conversely, inverse agonists, which preferentially bind to R, promote receptor inactivation by shifting the equilibrium away from R* (23,24). Constitutively active mutant receptors that exhibit an increase in agonist affinity and/or a decrease in inverse agonist affinity do so because a shift in equilibrium toward the active conformation has already occurred. A mutation-induced shift of a receptor toward R* will, by definition, facilitate agonist-independent coupling to G proteins, thus FIG. 5. Rate of pH i recovery from acid load. COS-1 cells that either were untransfected (UT) or were transfected with the WT ␤ 2 AR cDNA (WT) or the C116F ␤ 2 AR cDNA (C116F) were loaded with BCECF and prepared for fluorescence assay. Acid loading of the cells was accomplished by exposure to 20 mM NH 4 Cl as described for Fig. 4. A, after minimum pH i was achieved, recovery of pH i was monitored by measuring excitation ratios every 0.2 s for 120 s. The tracings represent the recovery of pH i seen in representative groups of 20 -30 cells. B, the rate of pH i recovery from acid load, termed dpH i /dt, was determined in untransfected cells and in WT and C116F transfectants under various conditions. Besides the untreated cases that summarize the pH i tracings shown in A, 10 Ϫ5 M ICI 118551 was added to both WT and C116F transfectants for 16 h prior to and during assay (WT ϩ ICI and C116F ϩ ICI, respectively). Epinephrine (10 Ϫ6 M) was also given to both WT and mutant transfectants for 15 min prior to and during assay (WT ϩ Epi and C116F ϩ Epi, respectively). The bars represent the mean dpH i /dt Ϯ S.E. determined from separate measurements made on groups of 20 -30 cells. The numbers of separate measurements are reported in the bars and are representative of four separate transfections. Asterisks indicate data that are significantly different from the WT case based on a two-tailed Student's t test (p Ͻ 0.001). Mean receptor expression in WT and C116F receptor-transfected cells was 0.398 and 0.058 pmol/mg of protein, respectively. causing constitutive activation of effector pathways. Based on these tenets of the model, the higher than WT binding affinity of common ␤ 2 -agonists seen for the C116F ␤ 2 AR (Table I) is strongly suggestive of constitutive receptor activity. Consistent with receptor theory (6), the degree of high affinity shift was proportional to the intrinsic efficacy of the agonist tested. Also supporting this hypothesis, the ␤ 2 -inverse agonist ICI 118551 exhibited a higher affinity for the WT receptor than for the C116F mutant (Table I). Again, it appears that the degree of shift to lower affinity was proportional to the inverse efficacy of the antagonist tested. ICI 118551, which is known to be a highly efficacious inverse agonist (24), showed the largest rightward shift in affinity for the mutant receptor (Table I). As expected, alprenolol and propanolol, which are classified either as weak inverse agonists (23) or as neutral antagonists (24), exhibited smaller rightward affinity shifts. Overall, these changes in ligand binding affinity are intrinsic to the mutant receptor and are not due to altered G protein precoupling as evidenced by the inability of Gpp(NH)p to affect K i values or Hill coefficients (data not shown).
The C116F mutation could evoke these shifts in ligand binding affinity in one of two ways. First, since the 116th residue is located in the third transmembrane domain only one helical turn below Asp-113, the counterion for the protonated amine of catecholamines (11,12), it seems possible that it could indirectly affect interaction of the receptor with ligand per se. However, since the function of the receptor has also been altered and correlates to an activational paradigm, this scenario seems unlikely. Rather, we suggest that the mutation has affected receptor conformation to mimic the activated state. Since we have postulated that the equivalent mutation in the ␣ 1b AR (C128F) is involved in modulation of an important salt bridge constraint that stabilizes the inactive conformation (25), it is possible to envision that C116F is influencing a similar constraining factor.
Even though the binding data allude to a constitutive activity induced by the C116F mutation, agonist-independent generation of a second messenger signal must be identified to firmly establish this hypothesis. Classically, it is held that the ␤ 2 AR evokes cellular effects through a coupling to cAMP production via G ␣s (2,3). Based on this, agonist-dependent and -independent effects on cAMP production were compared in WT and C116F receptor-transfected COS-1 cells. Not only did the C116F mutation fail to potentiate the isoproterenol dose response (EC 50 ) relative to the WT receptor, it also failed to evoke a response in the absence of the agonist (Fig. 2). Similar results were seen with epinephrine (data not shown). These findings indicate that the C116F mutation does not facilitate constitutive coupling to G ␣s . Rather, the mutant and WT receptors show equal competency in the generation of agonist-evoked cAMP responses. If the ␤ 2 AR indeed only signals through a coupling to G ␣s , these data would suggest that the C116F mutation affects ligand binding without inducing the receptor to assume an activated conformation that is competent to couple to a second messenger signal.
As was mentioned, a coupling of ␤ 2 ARs to G ␣s has been convincingly established in the literature. Because of this, when studying ␤ 2 -receptor pharmacology, the inclination has been to measure cAMP as a marker of G protein coupling. Over the past decade, however, a strong case for an alternate pathway has emerged. Early work by Strader et al. (7) showed that ␤-receptor mutants, uncoupled from G ␣s due to deletion of G protein-interacting domains (residues 222-229 and 258 -270), exhibited marked increases in agonist affinity. Similarly, mutagenesis of Asp-130 (in the third transmembrane domain) also led to increases in agonist binding affinity despite receptor uncoupling from G ␣s (26). The G ␣s -uncoupled 222-229 deletion mutant was later shown by Barber and Ganz (8) to activate Na ϩ /H ϩ exchange via an unknown G protein. This study established the novel hypothesis that G ␣s -independent cellular responses were evoked by ␤ 2 ARs. Since then, Barber and coworkers (9) have identified G ␣13 as responsible for activating a pathway that stimulates Na ϩ /H ϩ exchange through the ubiquitous NHE1. At least one other member of the G ␣ subfamily, G ␣12 , has since been implicated in the regulation of various isoforms of NHE1 (27), although the G ␣12 effects are inhibitory. These studies collectively support a hypothesis implicating G ␣13 in coupling of the ␤ 2 AR to Na ϩ /H ϩ exchange.
With G ␣s -independent coupling of ␤ 2 ARs to NHE1 having been established, questions are raised about possible effects of the C116F ␤ 2 AR mutant on Na ϩ /H ϩ exchange. NHE1, which is present on the plasma membrane of all mammalian cells, fulfills several distinct physiological functions including control of pH i and maintenance of cellular volume (28). NHE1 function, which is commonly measured by monitoring pH i with the fluorescent dye BCECF, was examined via two approaches. Basal pH i was determined in WT and C116F receptor-transfected COS-1 cells. Due to continuous H ϩ extrusion during the posttransfection period, possible constitutive activation of NHE1 induced by the C116F mutation would be detected via identification of a cytosolic alkalosis relative to pH i values seen in the WT case. However, since pH i is not a direct measure of NHE1 activity, the actual rate of NHE1 function was also determined in WT and C116F transfectants by measuring the rate of pH i recovery following NH 4 Cl-induced cellular acid load.
In the first experiment, intracellular alkalosis was evident in C116F receptor-transfected COS-1 cells relative to pH i levels seen in WT transfectants (Fig. 3). This was despite a 6.8-fold lower number of receptors in the C116F case. Locking the mutant receptor in the inactive state with the inverse agonist ICI 118551 blocked the effect, with treated mutants exhibiting a basal pH i similar to the WT case (Fig. 3). These findings not only provide strong evidence of a constitutive activation of Na ϩ /H ϩ exchange in C116F receptor-transfected cells, but the effect is shown to be ␤ 2 -receptor-specific based on inhibition by the ␤ 2 -selective inverse agonist.
In the second experiment, cells transfected with the C116F ␤ 2 AR showed a significantly faster rate of pH i recovery from acid load than WT transfectants. Confirming facilitation by NHE1, H ϩ extrusion during pH i recovery was blocked by dimethylamiloride, a specific NHE1 antagonist (22) (Fig. 4). Recovery rate, quantitated by calculating dpH i /dt, was Ͼ2-fold greater in mutant transfectants than in WT transfectants (Fig.  5). As before, this significant difference was seen despite a 6.8-fold lower receptor number in the C116F receptor-transfected cells. Further implicating the ␤ 2 AR as a regulator of NHE1, activation of the WT receptor with epinephrine caused an increase in dpH i /dt to values seen in the C116F case, whereas inactivation of the C116F receptor with ICI 118551 reduced dpH i /dt to values seen in the WT case. These results (a) corroborate earlier findings demonstrating coupling of ␤ 2 ARs to NHE1, (b) indicate selective constitutive coupling of the C116F ␤ 2 AR to NHE1, and (c) establish a role for Cys-116 in regulating selective coupling of the ␤ 2 AR to either the cAMP pathway or NHE1. It should be noted that coupling of the ␤ 2 AR to NHE1 has been demonstrated in cell lines that endogenously express the receptor (29). Therefore, it is unlikely that the results presented in this report are due to a promiscuous coupling of the receptor to G ␣13 in COS-1 cells.
Interestingly, epinephrine did not affect the rate of H ϩ extrusion seen in C116F receptor-transfected cells (Fig. 5). It is possible that the C116F mutation had itself induced a maximal shift of receptor equilibrium toward R*. Treatment with agonist under these conditions would not evoke additional stimulation. Alternatively, it is also possible that NHE1 was functioning at its maximal rate either because of strong constitutive activation induced by the mutant or due to the action of other mechanisms compensatory to the large H ϩ efflux. These factors would render additional stimulation of the signal by agonist undetectable. Overall, the present findings cannot distinguish between these two possibilities. One possible compensating factor, extracellular Na ϩ concentration, is not likely rate-limiting in this case because Na ϩ /H ϩ exchangers normally operate at saturation with respect to Na ϩ at physiological Na ϩ levels (30).
Based on the data presented in this report, a model is established that shows selective constitutive activation of NHE1 over cAMP in ␤ 2 ARs possessing a C116F mutation (Fig. 6). The model predicts that under basal conditions, the C116F mutant will preferentially isomerize to a conformation that couples putatively to G ␣13 (R*⅐G ␣13 ). Fewer receptors will isomerize to the uncoupled state (R), while fewer still will reside in the G ␣s -coupled conformation (R*⅐G ␣s ). In the presence of agonist, competent coupling to G ␣s is finally achieved due to a shift in equilibrium that leads to a significant number of receptors in the G ␣s -coupled conformation. Therefore, according to our model, the mutation selectively induces a putative R*⅐G ␣13 conformation that does not restrict agonist induction of the WT receptor activation scenario (i.e. coupling to both second messenger pathways).
The importance of Cys-116 in controlling the transition of the receptor to a selectively coupled conformation suggests a role for this amino acid position as a "switch" that determines the trafficking of the receptor between multiple active states. The hypothesis that Cys-116 can function as a switch is supported by earlier studies of ␣ 1b -adrenergic and AT 1 receptors. Cys-116 in the ␤ 2 AR is 14 residues N-terminal to the highly conserved DRY sequence found on the cytoplasmic border of the third transmembrane domain. DRY, an important functional domain for the binding of G proteins and the catalysis of GDP release (31), provides a convenient reference for comparing the C116F ␤ 2 mutation with similar point mutations in ␣ 1b ARs, AT 1 receptors (Fig. 7), and possibly other G protein-coupled receptor systems.
In the ␣ 1b AR system, the 14th position N-terminal to the DRY domain is Cys-128 (Fig. 7). Analogous to the C116F ␤ 2 AR, a C128F mutation of the ␣ 1b AR exhibited a selective constitutive coupling to one second messenger over another. In the case of the ␣ 1b AR mutant, constitutive coupling to the inositol pathway via a pertussis toxin-insensitive G protein was achieved, whereas only competent coupling to the pertussis toxin-sensitive arachidonic acid pathway occurred (13). Comparatively, an A293E mutation of this receptor exhibited constitutive activity (32) that was distinct from the C128F mutant because of constitutive coupling to both pathways (13). The ability of the C128F mutant to achieve a presumed G q -coupled/G i/o -uncou-pled state in the absence of agonist could simply be a manifestation of more efficient coupling to G q , the supposed primary second messenger pathway in the ␣ 1b AR system. It was argued, however, that the R state of the C128F mutant receptor did not isomerize to a single active conformation. Rather, it was postulated that R can isomerize to at least two distinct active states: one coupled to G q and one coupled to G i/o . Given the results of this earlier study, either the G ␣13 /NHE1 pathway is a more efficient coupler to ␤ 2 ARs than the G s /cAMP pathway, or the conclusions of the C128F ␣ 1b AR study are valid.
Analogous to both Cys-128 in the ␣ 1b AR and Cys-116 in the ␤ 2 AR, Asn-111 resides at the 14th position N-terminal to the DRY domain of the AT 1 receptor (Fig. 7). In the AT 1 system, the constitutively active mutant N111G exhibited higher than WT activation of inositol metabolism under basal conditions (33). However, this constitutive activation was only ϳ50% of the agonist-evoked response seen in the WT receptor. Noda et al. (33) suggested that the N111G mutation induced a transition of the receptor from R to a partially activated intermediate (RЈ). Supporting their model, agonistic ligands induced full activation of inositol metabolism by the N111G mutant, suggesting that transition of the mutant receptor from RЈ to the fully activated R* conformation could be achieved. In agreement with conclusions drawn in the present study, the authors postulated that Asn-111 is a switch that controls isomerization between multiple and distinct activation states of the receptor. Given the diverse nature and lack of amino acid conservation between G protein-coupled receptors, the behavior of Asn-111 in the AT 1 receptor and of Cys-128 and Cys-116 in the adrenergic receptors suggests that, in general, this particular amino acid position may function as a switch that regulates trafficking between distinct receptor conformations.
The findings presented in this report corroborate previously published work (7-9) describing a coupling of ␤ 2 ARs to NHE1, putatively through G ␣13 . Additionally, we identify novel selective constitutive activation of NHE1 over the cAMP pathway induced by a Cys-to-Phe mutation at residue 116 of the ␤ 2 AR. Based on this, we suggest an important role for Cys-116 as a switch residue that controls isomerization between at least two G protein-specific activation states of the ␤ 2 AR. When considering the present results along with the data of earlier studies, we hypothesize that, in general, the amino acid located at the 14th position N-terminal to the conserved DRY domain of G protein-coupled receptors is a critical entity in the control of receptor isomerization between R, R*, and other distinct activation states that may exist.
This work supports Kenakin's "cubic" ternary complex model (34) by predicting the formation of two different ternary complexes when a single receptor interacts with two different G proteins (Fig. 8A). In contrast to this model, however, we argue that such a receptor can adopt two distinct activated conforma-FIG. 6. Selective constitutive activation of NHE1 by the C116F ␤ 2 AR. Under basal conditions (without agonist present), the C116F ␤ 2 AR isomerizes preferentially to the G ␣13 -coupled active state (R*⅐G ␣13 ). R*⅐G ␣s is the G ␣s -coupled active state; R is the inactive state. FIG. 7. Alignment of residues and the switch position in the third transmembrane helix of the AT 1 and ␣ 1b -and ␤ 2 -adrenergic receptors. Using the highly conserved DRY domain as a reference, Cys-116 in the ␤ 2 AR, Cys-128 in the ␣ 1b AR, and Asn-111 in the AT 1 receptor are in the same relative position. All three amino acids cause constitutive activation when mutated and are implicated as switch residues that control receptor isomerization to selective activation states.
tions (R* 1 and R* 2 ), each selectively interacting with a different G protein. Based on this, we offer the bicubic model (Fig. 8B) as a revision of Kenakin's current model, which predicts only a single active conformation (R*) that partitions among the G proteins. In either scheme, agonists may be developed that selectively traffic the receptor to a particular activated conformation. Based on this, such signaling-specific agonists would offer vast advantages over promiscuous drugs. These compounds would not only extend our understanding of receptorpathway-response coupling, but they might also belong to an important new class of therapeutic agents. FIG. 8. Cubic ternary complex models. A, the cubic ternary complex model of Kenakin (34) in which a single activated state of the receptor partitions into different ternary complexes; B, the proposed revision of the cubic ternary complex model in which a single receptor can form two distinct activated conformations that selectively interact with different G proteins. A, agonist; R, receptor; G, G protein.