New Molecular and Structural Determinants Involved in -Adrenergic Receptor Desensitization and Sequestration

As the β-adrenergic receptor (βAR) is resistant to short term agonist-promoted desensitization and sequestration, chimeric β/β receptors were generated to identify the molecular determinants responsible for these regulatory processes in the βAR. By exchanging single or multiple intracellular domains of the βAR for the corresponding regions of the βAR, we show that specific domains can be identified as additive determinants for desensitization, while sequestration is more dependent on global structural conformation. The carboxyl-terminal tail, the third and the second intracellular loops of the βAR provided additive contributions to the desensitization observed upon short term agonist stimulation. The second intracellular loop plays a role which is as important as that of third cytoplasmic loop and carboxyl-terminal tail which had previously been identified as the major determinants of agonist-promoted desensitization. Additive contributions of the cytoplasmic domains of the βAR were also observed for agonist-promoted sequestration. The substitution of the first and second intracellular loops and the carboxyl tail were associated with a β-like sequestration phenotype. However, in contrast to what is observed for desensitization the co-substitution of the third cytoplasmic loop with any of the other domains completely suppressed sequestration. These results suggest that sequestration depends not only on appropriate interactions of multiple molecular determinants within the cytoplasmic region of the βAR but also on conformational determinants that may influence their orientation.

tion are initiated by functional coupling of the receptor with the stimulatory GTP-binding protein G s which in turn activates adenylyl cyclase thus promoting a rise in intracellular cAMP concentration. This signal transduction pathway is tightly controlled by regulatory processes which, on the one hand, prevent hormonal overload (desensitization) and, on the other hand, reset the signaling pathways for further hormonal stimuli (resensitization).
Rapid desensitization of the ␤ 2 AR, which occurs as early as a few minutes following the initiation of the stimuli, results from the uncoupling of the receptor from G s . Phosphorylation of the ␤ 2 AR by cAMP-dependent protein kinase (protein kinase A) and ␤-adrenergic receptor kinase (␤ARK) is known to play a central role in this process (1)(2)(3). Although several phosphorylation sites involved in this uncoupling process have been unambiguously identified (4,5), the existence of other important site(s) has never been ruled out.
Recent studies have suggested that agonist-promoted sequestration is a resensitization mechanism limiting the effects of short term desensitization. Indeed, blocking sequestration was found to significantly delay resensitization which normally occurs upon termination of receptor activation (6,7). According to the proposed model, phosphorylated receptors are sequestered in a subcellular compartment where they are dephosphorylated and become available for recycling to a fully functional conformation to the plasma membrane.
Although significant efforts have been made, an unequivocal identification of the molecular determinants triggering the sequestration process has yet to be achieved. Previous studies using site-directed mutagenesis have suggested the existence of several motifs located in various cytoplasmic domains of the ␤ 2 AR (8 -10), but a coherent functional connection between these molecular determinants is lacking.
Previous studies have also shown that the ␤ 3 AR does not readily undergo rapid agonist-promoted desensitization and sequestration (11)(12)(13). This resistance to rapid regulation and the high level of sequence identity between the ␤ 2 AR and the ␤ 3 AR makes the latter an excellent model to investigate the molecular determinants of desensitization and sequestration. Indeed, an alternative to site-directed mutagenesis in examining the molecular determinants of desensitization is the construction of chimeric receptors. This approach has the advantage of searching for the addition of regulatory phenotypes rather than their loss. Construction of chimeric receptors in which specific domains of the ␤ 2 AR have replaced their counterparts within the ␤ 3 AR has already been successfully used to study their contribution to desensitization. Substitutions of the carboxyl-tail alone (11) or of both the carboxyl-tail and third cytoplasmic loop (12) of the ␤ 3 AR by the corresponding region of the ␤ 2 AR have been shown to confer agonist-promoted desensitization to the ␤ 3 AR. However, neither of the chimeric receptors studied had desensitization profiles comparable with that of the wild-type ␤ 2 AR. This suggests that other regions of the receptor are required for a complete ␤ 2 AR-like desensitization.
To identify additional putative determinants involved in rapid desensitization of the ␤ 2 AR and to assess the respective contribution of ␤ 2 AR intracytoplasmic domains to the sequestration process, we have constructed a series of chimeric receptors in which various combinations of the ␤ 3 AR intracellular loops were replaced by the corresponding domains of the ␤ 2 AR. Construction of Chimeric Receptors and Cell Culture-Chimeric ␤ 3 / ␤ 2 AR coding regions illustrated in Fig. 1 were constructed using the following approaches: (i) for the exchange of the third cytoplasmic loop and of the carboxyl terminus, silent restriction sites were generated by site-directed mutagenesis (third cytoplasmic loop: AccI at Val-222 of ␤ 3 AR and at Val-218 of ␤ 2 AR; AvrII at Leu-294 of ␤ 3 AR and at Leu-275 of ␤ 2 AR and carboxyl terminus: PstI at Arg-348 of ␤ 3 AR and at Arg-328 of ␤ 2 AR; ScaI four bases following the stop codon of the ␤ 2 AR. Entire domains of the ␤ 3 AR delimited by these sites were then replaced by the corresponding region of the ␤ 2 AR). (ii) For the exchange of the second cytoplasmic loop, a AgeI restriction site was created at Arg-135 of the ␤ 3 AR. The AgeI-BstEII fragment of this ␤ 3 AR was excised and replaced by a double-stranded synthetic oligonucleotide encoding the corresponding region of the ␤ 2 AR. The remaining two amino acids from the ␤ 3 AR sequence (Arg-152 and Cys-153) were mutated to Asn and Lys by site-directed mutagenesis to complete the sequence of the ␤ 2 AR. (iii) For the first cytoplasmic loop, site-directed mutagenesis of Trp-64, Thr-65, Pro-66, and Met-71 to Lys, Phe, Glu, and Val, respectively, was performed to generate a ␤ 2 AR sequence. All mutations were confirmed by dideoxy sequencing.
Chimeric receptor constructs were subcloned into the eucaryotic expression vector pcDNA3/RSV. This vector was generated by insertion of the BglII-HindIII restriction fragment from the pRc/RSV vector (Invitrogen) into pcDNA3 (Invitrogen) so as to replace the cytomegalovirus promoter. Constructs were then stably transfected in murine L-cells as described previously (12). Geneticin-resistant cells were selected in DMEM supplemented with 10% (v/v) fetal bovine serum, 4.5 g/liter glucose, 100 units/ml penicillin, 100 mg/ml streptomycin, 1 mM glutamine, and Geneticin at a concentration of 400 g/ml. Individual clones were screened for ␤AR expression by radioligand binding assay, using [ 125 I]CYP as ligand.
Chimeras were named starting with their receptor subtype followed by four numbers indicating the origin of the 1st, 2nd, 3rd cytoplasmic loops and of the carboxyl-tail, respectively. For example ␤ 3 -3322 represents a ␤ 3 AR with the first and second cytoplasmic loop of the ␤ 3 AR and the third cytoplasmic loop and carboxyl-terminal tail of the ␤ 2 AR.
Radioligand Binding Assay-Nearly confluent cells, grown as monolayers, were washed with PBS, incubated for 5 min with 2% trypsin/ EDTA at 37°C and resuspended in DMEM supplemented with 10% (v/v) fetal bovine serum. The cells were then centrifuged at 450 ϫ g for 5 min at 4°C and washed twice with ice-cold PBS. Binding assays were carried out using 100 l of cell suspension in a final volume of 250 l containing 50 mg/ml bovine serum albumin and 1 M desipramine.
[ 125 I]CYP at 200 pM (for ␤ 2 AR) or at 600 pM (for ␤ 3 AR) was used as the radioligand. Specific binding was defined as binding displaced by 10 M D/L-propranolol (␤ 2 AR) or 50 M bupranolol (␤ 3 AR). Assays were carried out for 90 min at 25°C and terminated by rapid filtration through Whatman GF/C glass fiber filters previously soaked in PBS containing 0.3% polyethyleneimine (to reduce nonspecific binding). Protein concentrations were determined on broken cell preparations by the method of Bradford (14) using the Bio-Rad protein assay system with bovine serum albumin as standard.
Sequestration Assays-Cells grown in 75-cm 2 flasks were incubated in the presence of 10 M isoproterenol in DMEM containing 10 M ascorbic acid or the vehicle alone for the indicated periods of time. The flasks were then placed on ice, washed twice with ice-cold PBS, and the cells detached mechanically in a buffer containing 5 mM Tris, 2 mM EDTA, pH 7.4, 5 mg/liter soybean trypsin inhibitor, 5 mg/liter leupeptin, and 10 mg/liter benzamidine (buffer A). Cell suspensions were homogenized with a Polytron homogenizer (Janke & Undel Ultra-Turrax T25) for 5 s at maximal setting. The lysate was centrifuged at 450 ϫ g for 5 min at 4°C. The supernatant was layered on top of a 35% sucrose cushion and centrifuged at 150,000 ϫ g for 90 min. As reported previously (15), the light membrane vesicular fraction was found at the 0 -35% interface, whereas the plasma membrane fraction sedimented at the bottom of the sucrose cushion. Each fraction was collected, diluted in buffer A and centrifuged at 200,000 ϫ g for 60 min. The pelleted membranes were resuspended in 50 mM Tris, 5 mM MgCl 2 , pH 7.4, and used immediately for radioligand binding assays. Binding assays were conducted as described above but using membrane preparations instead of cell suspensions. Sequestration levels measured for the wild type ␤ 2 AR using this technique were identical to those measured in whole cell binding assays in which sequestration is defined by the number of [ 125 I]CYP binding sites inaccessible to the hydrophilic ligand CGP12177A (data not shown).
Adenylyl Cyclase Assay-Cells grown in 75-cm 2 flasks were incubated in DMEM containing 10 M ascorbic acid with or without 10 M isoproterenol for the indicated periods of time. Incubations were stopped by washing the cells twice with ice-cold PBS. Cell were then detached mechanically and homogenized in ice-cold buffer A using a polytron homogenizer (Janke & Undel Ultra-Turrax T25). Lysates were centrifuged at 450 ϫ g for 5 min at 4°C. Supernatants were centrifuged at 43,000 ϫ g for 20 min at 4°C and the pellets washed twice in buffer A. The washed membranes were then resuspended in a buffer containing 75 mM Tris, pH 7.4, 5 mM MgCl 2 , 2 mM EDTA, and protease inhibitors.
Adenylyl cyclase activity was measured on these membrane preparations according to the method of Salomon et al. (16). Briefly, the reaction mixture contained: 20 l of membrane preparation ( Determination of Intracellular Cyclic AMP Levels-Cells were washed once in PBS and incubated in the absence or presence of 10 M of isoproterenol or CGP12177A for 15 min at 37°C in PBS 0.5 mM isobutylmethylxanthine, 0.5 mM ascorbic acid. The incubation buffer was discarded and cells lysed in 1 M NaOH for 30 min at 37°C. The lysate was neutralized with 1 M acetic acid and centrifuged in a microcentrifuge at maximum speed for 5 min. The supernatant was used for cAMP determination using [ 3 H]cAMP radioimmunoassay system (Amersham Corp.).

RESULTS
Pharmacological Characterization of the ␤ 3 /␤ 2 AR Chimeras-In order to assess the contribution of the various cytoplasmic domains of the ␤ 2 AR to agonist-promoted uncoupling and sequestration, chimeric receptors were constructed which comprised extracellular and transmembrane regions of ␤ 3 AR origin and various combinations of intracellular domains from ␤ 2 AR and ␤ 3 AR. Conserved amino acid sequences between the ␤ 3 -and the ␤ 2 AR within the putative transmembrane regions and the corresponding junctions with intracellular loops facilitated the exchange of intracellular domains (Fig. 1). For the chimeric receptor identified as ␤ 3 -2222, all the cytoplasmic domains of the ␤ 3 AR were substituted by those of the ␤ 2 AR. Eight additional ␤ 3 /␤ 2 AR chimeric receptors were generated. Clones isolated from Ltk Ϫ cells stably expressing between 150 and 960 fmol of receptor/mg of protein were selected for further study. As shown in Table I, all the chimeras constructed retained affinity constants for [ 125 I]CYP comparable with that of the wild type ␤ 3 AR (500 -3000 pM), which is significantly higher than that of the ␤ 2 AR (51 pM). Further characterization of ␤ 3 -2222 clearly showed that, despite a moderate decrease in the affinity for several ligands, this receptor which contains the largest contribution of ␤ 2 AR sequences, still retained a general order of potency for ␤-adrenergic ligands which is characteristic of the ␤ 3 AR (Table II). Particularly revealing is the relatively low affinity of the ␤ 3 -2222 for isoproterenol, alprenolol, bupranolol, and ICI118551. Furthermore, CGP12177A which is an antagonist at the ␤ 2 AR but has partial agonistic properties toward the ␤ 3 AR, stimulated adenylyl cyclase with an intrinsic activity of 0.6 in cells expressing ␤ 3 -2222, thus con-firming the ␤ 3 AR-like pharmacology of ␤ 3 -2222 (Table II).
Agonist-promoted Desensitization of ␤-Adrenergic-stimulated Adenylyl Cyclase Activity-Previous reports showed that substitution of the third cytoplasmic loop (i3) and/or the carboxyl tail (CT) in the ␤ 3 AR by its ␤2 equivalents partially restored a ␤ 2 -like rapid desensitization profile (11,12). However, neither the CT alone nor the combination of i3 and the CT restored desensitization of a similar magnitude to that observed for the wild type ␤ 2 AR. The chimeras described above were thus used to determine the contribution of each cytoplasmic loop to the desensitization pattern. Desensitization induced by pretreatment of the cells with isoproterenol (10 M) for 2 and 15 min are shown in Fig. 2. As reported previously, isoproterenol caused a rapid desensitization in cells expressing the wild type ␤ 2 AR (Fig. 2, panel I). Desensitization was characterized by a rightward shift of the isoproterenol dose-response curve and by a robust time-dependent reduction of the maximal stimulation promoted by the agonist. Also, consistent with previous observations, sustained stimulation of the ␤ 3 AR-expressing cells (Fig. 2, panel A) induced only a modest shift in the isoproterenol dose response curve and a very small decrease in the maximal stimulation following a desensitization of 15 min.  1. A, topological model of the human ␤ 3 AR. Conserved amino acid residues between the ␤ 3 AR and ␤ 2 AR sequences are represented by filled circles. Arrows indicate the connecting sites between the ␤ 3 AR core sequence and the substituted ␤ 2 AR cytoplasmic domains in the chimeric receptors. B, sequence comparison of the cytoplasmic domains of ␤ 3 AR and ␤ 2 AR. Hyphens indicate identity with the ␤ 3 AR sequence. Triangles indicate protein kinase A phosphorylation sites. Stars indicate potential phosphorylation sites for ␤ARK. Motifs identified as potential determinants of sequestration are overlined.
This confirms that the ␤ 3 AR is largely resistant to rapid agonist-induced desensitization compared with the ␤ 2 AR.
Substitution of the first cytoplasmic loop (i1) of the ␤ 3 AR with the corresponding region from ␤ 2 AR did not confer any desensitization phenotype as prestimulation of cells expressing this mutant receptor, for 2 or 15 min, did not affect either the dose-response curves or the maximal stimulation (Fig. 2, panel  B). As expected from previous studies, substitution of either i3 (Fig. 2, panel C) or CT (Fig. 2, panel D) conferred desensitization profiles that are characterized mainly by rightward shifts of the dose-response curves that became clearly evident following 15 min of prestimulation. Moreover, as can be seen in panel F, the contribution of these two domains to the desensitization pattern appear to be additive. Indeed, the desensitization of the ␤ 3 -3322 for 2 and 15 min lead to larger shifts of the doseresponse curves and provoked a sizable reduction in the maximal stimulation observed. However, the extent of desensitization did not reach that observed for the ␤ 2 AR, suggesting that other domains could be required to obtain a full ␤ 2 AR desensitization phenotype. Interestingly, the single substitution of the second intracellular loop (i2) of the ␤ 3 AR by that of the ␤ 2 AR (␤ 3 -3233) was sufficient to promote desensitization. In fact, pretreatment of cells expressing this chimera with isoproterenol induced reductions in agonist-stimulated adenylyl cyclase activity that are at least of the same magnitude as those observed for ␤ 3 -3332 and ␤ 3 -3323 (compare panel E with panels C and D). The contribution of i2 to desensitization is also supported by the observation that agonist-promoted desensiti-zation observed with the double substitution of i2 and CT into the ␤ 3 AR was faster and larger than that conferred by single substitution of CT alone (compare panels G and D). Interestingly, the reduction in agonist-stimulated adenylyl cyclase activity in cells expressing ␤ 3 -3232 was even faster than that observed in cells expressing ␤ 3 -3322. Indeed, a 25% reduction in the maximal stimulation was already evident following a 2-min preincubation period for the ␤ 3 -3232, whereas no change in the maximal stimulation was observed at this time for the ␤ 3 -3322. An additive effect of i2 on desensitization is also evident when comparing ␤ 3 -3222 with ␤ 3 -3322 (compare panels F and H).
The apparently additive contribution of the ␤ 2 AR i2, i3, and CT to the desensitization profiles of the chimeric receptors can be easily appreciated by looking at Fig. 2. Indeed, it can be seen that co-substitution of these three domains leads to a progressive increase in the overall extent of desensitization. However, a quantitative assessment of the additivity is rendered difficult by the fact that desensitization is reflected by changes in two parameters, i.e. a reduction in the maximal stimulation and a rightward shift of the dose-response curves. The assessment of the dose-response shifts is further complicated by the fact that not all chimeras have the same efficacy to stimulate the adenylyl cyclase activity under basal conditions and that in several cases, the desensitized adenylyl cyclase activity did not reach a plateau at the highest isoproterenol concentration used (1 mM), thus making mathematical analysis more difficult. Therefore, the reduction in adenylyl cyclase activity measured for a stim-  ulating concentration of isoproterenol equal to its EC 50 for a given chimera was used as an index of the dose-response rightward shift. Fig. 3 illustrates the amplitude of the changes in stimulation at maximal concentration and at the EC 50 for all the chimeric receptors following a desensitization of 15 min. The effects of the single substitutions of i2, i3, or CT on the desensitization are reflected mainly by a reduction of the response at the EC 50 with only marginal effects on the maximal stimulation. Double or triple substitution of these domains conferred agonist-dependent reduction of both the maximal stimulation and of the stimulation at the EC 50 . Interestingly, the substitution of i1, which has no effect on the desensitization pattern by itself, conferred a slight negative effect on the desensitization of the maximal response when co-substituted along with other cytoplasmic domain of ␤ 2 origin (compare ␤ 3 -3232 and ␤ 3 -3222 with ␤ 3 -2232 and ␤ 3 -2222). This might result from an unfavorable conformational effect, since the affinity for [ 125 I]CYP was also significantly reduced by the co-substitution of i1 (Table I). Since ␤ 3 -2222 and ␤ 3 -2232 also have slightly reduced affinity for isoproterenol, as indicated by their higher K act when compared with ␤ 3 AR (Table I), it might be suggested that the lower extent of the maximal response desensitization of these chimeras is an underestimation resulting from the potentially nonsaturating conditions used. However, this is highly unlikely, since ␤ 3 -3232 and ␤ 3 -3222 have similar elevated K act , and yet they display the greatest desensitization of the maximal response approaching the level observed for the ␤ 2 AR. (17), agonist stimulation leads to a rapid and time-dependent translocation of ␤ 2 AR from the plasma membrane to a light vesicular fraction (Fig. 4). In contrast, no such sequestration is observed in cells expressing the ␤ 3 AR. In fact, isoproterenol promoted an apparent enrichment of the plasma membrane fraction in ␤ 3 -binding sites (expressed in Fig. 4 as negative sequestration). In previous attempts aimed at identifying molecular determinants of sequestration, the exchange of intracellular domains of the ␤ 3 AR with the corresponding regions of the ␤ 2 AR (11,12) led to apparently contradictory observations: the exchange of the CT alone partially restored agonist-promoted sequestration, but no sequestration could be observed in a chimeric receptor harboring both the i3 and the CT of the ␤ 2 AR. Therefore, we tested whether any single or multiple exchanges between intracellular regions of receptors could restore a sequestration profile similar to that observed for the ␤ 2 AR.

Molecular Determinants of Isoproterenol-dependent Sequestration-As reported previously
Sequestration of the ␤ 2 AR in Ltk Ϫ cells reached a maximum of 34% after 15 min of isoproterenol stimulation. Thus, we first screened L cells expressing the chimeric receptors described above by measuring sequestration following incubation with 10 M isoproterenol for 5 and 15 min (Fig. 4). The single substitution of either CT, i2 or i1 of the ␤ 3 AR with the corresponding regions of the ␤ 2 AR partially restored agonist-promoted sequestration (␤ 3 -3332: 12%; ␤ 3 -3233: 8%; ␤ 3 -2333: 7%, following a stimulation of 15 min). In contrast, substitution of i3 had no apparent effect on the sequestration pattern. Positive effects on sequestration of the CT, i2, and i1 appeared partially additive as the level of sequestration observed for ␤ 3 -3232 and ␤ 3 -2232 tended to be greater than those observed when each of these domains were substituted alone. However, it should be noted that substitution of i3 had a dominant negative effect over all other substitutions. Indeed, no agonist-promoted sequestration was observed in any of the ␤ 3 AR chimeric receptors containing an i3 of ␤ 2 AR origin. Based on these data, one could postulate the existence of a specific sequence located in i3 of the ␤ 2 AR that inhibits sequestration. If such a negative sequence exists, substitution of this ␤ 2 AR domain by the corresponding region of the ␤ 3 AR, would be expected to lead to a "super-sequestration" profile. We studied the sequestration profile of such a chimeric ␤ 2 -2232 receptor: the sequestration pattern was indistinguishable from that of the ␤ 2 AR wild type (data not shown) arguing against the existence of specific sequences within the i3 loop that inhibit sequestration. Alternatively, one could argue that the lack of sequestration of ␤ 3 -2222 results from its reduced affinity for isoproterenol, which is reflected by a 14-fold increase in the K act when compared with the ␤ 3 AR (Table I). This is highly unlikely since ␤ 3 -3232 has an even higher K act , but undergoes sequestration which reached levels comparable with that attained for the ␤ 2 AR. Also, increasing the concentration of isoproterenol used to promote sequestration to up to 1 mM did not induce any sequestration of ␤ 3 -2222 (data not shown).
To further characterize chimeric receptors showing positive sequestration, this process was studied for longer periods of time. For the three single substitutions (␤ 3 -2333, ␤ 3 -3233, ␤ 3 -3332), agonist-promoted sequestration reached its maximum between 15 and 30 min of stimulation (10%-15%) and remained at that level for up to 60 min (Fig. 5). Although sequestration was observed for these three chimeras, the level of sequestration never reached that observed for the ␤ 2 AR (30%). In contrast, sequestration of ␤ 3 -3232 and ␤ 3 -2232 attained levels observed for the ␤ 2 AR albeit with somewhat slower kinetics. Indeed, sequestration levels were equivalent to those of the ␤ 2 AR after 60 min of stimulation. These results suggest that three ␤ 2 AR cytoplasmic domains, i1, i2, and CT, provide positive sequestration signals that may be somewhat additive but are not sufficient to restore an entirely normal ␤ 2 AR sequestration profile.

DISCUSSION
Despite extensive investigation during the past years, molecular mechanisms involved in short term ␤AR regulation have not been completely elucidated. We took advantage of the high degree of homology existing between the ␤ 2 AR and the ␤ 3 AR (69% within putative membrane spanning domains and corresponding junctions with intracellular loops) and of their distinct profile of regulation to identify novel molecular determinants of receptor desensitization and sequestration. Current hypothetical models suggest that molecular determinants of ␤ 2 AR regulation are located in intracellular domains (18). Sequence homology between ␤ 2 AR and ␤ 3 AR facilitated the exchange of unmodified intracellular domains and the construction of functional chimeric receptors. We previously showed that the chimeric receptor strategy is particularly adapted to study molecular basis of receptor function (19). This approach, complementary to site-directed mutagenesis studies, allows assessment of the contribution of entire structural domains without preconceived notions of the precise residues involved.
The chimeric receptors constructed in the present study conserved pharmacological properties characteristic of the ␤ 3 AR. In particular, ␤ 3 -2222 which contains the largest proportion of ␤ 2 AR derived sequence maintained all the pharmacological trademarks of the ␤ 3 AR including the agonistic properties of the ␤ 2 AR antagonist CGP12177A. Previous studies, based on molecular modelling and pharmacological characterization of the ␤ 3 -and ␤ 2 AR suggested that ␤ 2 -antagonists with ␤ 3 -agonist properties, such as CGP12177A, may adopt a stacked conformation in the ␤ 2 AR binding pocket, leading to antagonistic effects while they would adopt an extended conformation in the less encumbered ␤ 3 -binding site. This last conformation may allow interactions with specific residues implicated in signal transduction (20). The ␤ 3 -like pharmacological properties maintained in ␤ 3 -2222 suggest that the intracellular domains do not affect the overall organization of the binding pocket determined by the positioning and the orientation of the transmembrane domains.
As reported previously (12), substitution of the cytoplasmic domains of the ␤ 3 AR with those of ␤ 2 AR containing all known specific consensus sequences for receptor phosphorylation by ␤ARK and protein kinase A (i3, CT) (4,5,15,21,22) failed to confer a ␤ 2 AR-like desensitization profile. The present report clearly shows that additional molecular determinants involved in receptor desensitization are also present in the ␤ 2 AR i2. The presence of this domain alone is sufficient to confer a desensitization level at least equivalent to that provided by CT and i3. Furthermore, when substituted in combination with the other domains, additive effects on the level of agonist-promoted desensitization were found. Consistent with the contribution of i2 to receptor desensitization is the observation that Phe-139 located in i2 of the ␤ 2 AR is apparently involved in G-protein coupling (8). The recent finding that phosphorylation of Tyr-141 within the ␤ 2 AR i2 favors its coupling with G s (23) also suggest that this domain plays an important role in the regulation of receptor-G s interaction. The contribution of i2 to the desensitization process could result from its interaction with previously identified proteins, which regulate receptor function, such as ␤ARK or ␤-arrestin thus stabilizing their interactions with domains already characterized. Alternatively, i2 may contain new unidentified sites which promote receptor uncoupling. The observation that substitution of i2 alone is sufficient to confer receptor desensitization would support the latter.
All molecular determinants of ␤ 2 AR uncoupling identified so far correspond to phosphorylation targets for protein kinases.
In a previous report mutation of all putative ␤ARK, protein kinase A and protein kinase C phosphorylation sites significantly reduced agonist-promoted phosphorylation and desensitization but did not completely abolish them (5). This is consistent with the idea that additional phosphorylation sites may exist and be involved in receptor desensitization. Two serine residues Ser-137 and Ser-143 present in i2 of ␤ 2 AR are absent from the ␤ 3 AR. These residues might be the target of another kinase. One serine (Ser-137) is contained in the potential phosphorylation consensus site S/T P X K/R, which has been shown to be a preferred substrate for cdc2 kinase (24). Additional experiments are required to assess whether this region contains phosphorylation sites involved in receptor uncoupling and to identify the putative kinase participating in such regulation.
Previous studies have suggested the existence of several motifs located in various cytoplasmic domains of the ␤ 2 AR involved in sequestration. However, no clear connection could be established between these motifs that leads to an unequivocal identification of the molecular determinants triggering the sequestration process. In their studies, Hausdorff et al. (25) showed that site-directed mutagenesis of a subset of serine residues, believed to be ␤ARK phosphorylation sites, blocked agonist-promoted sequestration. In particular, substitution of Ser-356 and Ser-364 by glycine residues completely blocked sequestration. However, mutations of additional serines and threonines in this region restored a normal sequestration phenotype (4). The authors concluded that Ser-356 and Ser-364 are not required for sequestration but that their mutation leads to conformational changes interfering with the sequestration process. Also suggesting that receptor conformation may influence sequestration is the recent report by Green and Liggett (9), indicating that a proline-rich sequence located in the third cytoplasmic loop of the ␤ 1 AR prevents the efficient sequestration of this receptor subtype. Recently, a tyrosine residue (Tyr-326) located at the interface between the seventh transmembrane domain and the carboxyl tail has been proposed as a specific determinant for ␤ 2 AR sequestration (10). Although this residue may be required, it is certainly not sufficient to confer an agonist-promoted sequestration phenotype. Indeed, a tyro- sine residue within a NPXXY motif identical to that of the ␤ 2 AR is also present in a similar position in the ␤ 3 AR. However, the ␤ 3 AR subtype is not sequestered upon agonist stimulation (11,12). In addition, mutation of the tyrosine residue located in the NPXXY motif of the gastrin-releasing peptide receptor or of the Type 1 angiotensin II receptor did not affect their agonist promoted sequestration arguing against a general role for this sequence (26,27). More recently, Ferguson et al. (28) proposed that the reduction of sequestration caused by the mutation of Tyr-326 in the ␤ 2 AR resulted from the inability of this mutant receptor to act as a substrate for ␤ARK. They proposed that ␤ARK-mediated phosphorylation facilitates ␤ 2 AR sequestration. Although that may be the case, it is clear from previous studies that phosphorylation by ␤ARK is not an absolute requirement nor is it the signal initiating the sequestration process. Indeed, it has been shown that ␤ 2 AR lacking all putative ␤ARK phosphorylation sites can readily be sequestered upon agonist stimulation (3,7,10,28). The presence of an hydrophobic residue in the DRYXXI(V)XXPZ sequence (where Z is the hydrophobic residue) within the second cytoplasmic loop of the ␤ 2 AR has also been proposed as being important for receptor sequestration (8). Such a hydrophobic residue is conserved in identical position in the ␤ 3 AR (DRYLAVTNPL), suggesting that the presence of this residue is not sufficient to facilitate agonist-promoted sequestration.
Our data support the notion that interaction between multiple intracellular domains of the ␤ 2 AR contribute to sequestration phenotypes. Clearly, none of the cytoplasmic domains (which contain the various sequestration signals previously proposed), when substituted alone, could confer a ␤ 2 AR-like sequestration pattern. In fact, i1, i2 and CT allowed very modest agonist-promoted sequestration, while the association of the second intracytoplasmic loop with the carboxyl terminus of the ␤ 2 AR in the chimeric ␤ 3 -3232 and ␤ 3 -2232 receptor restored sequestration levels similar to that of the ␤ 2 AR albeit with slower kinetics. These results suggest that CT and i2 of the ␤ 2 AR play major roles in the sequestration process. The contribution of CT is consistent with the recently proposed facilitator role of CT ␤ARK phosphorylation sites in the sequestration (28). However, the mere presence of these motifs is not sufficient to assure a sequestration phenotype. Indeed, no sequestration was detected in any of the chimeric receptor harboring the third cytoplasmic loop from ␤ 2 AR origin. This negative effect is clearly not attributable to the presence of a specific signal preventing sequestration, since it is compatible with normal sequestration of the wild type ␤ 2 AR. These data therefore suggest that, together with the concerted participation of multiple cytoplasmic domains, the adoption of an appropriate conformation resulting from specific interactions among intra-cytoplasmic domains is required for proper sequestration.
In conclusion, we have shown that in addition to the carboxyl tail and the third cytoplasmic loop, the second cytoplasmic loop of the ␤ 2 AR is involved in the process of agonist-promoted desensitization. Also, sequestration does not merely depend on the presence of specific domains (i.e. i2 and CT) but largely relies on the proper arrangement of all the cytoplasmic domains.