Characterization of the Binding Domain of the &Adrenergic Receptor with the Fluorescent Antagonist Carazolol EVIDENCE FOR A BURIED LIGAND BINDING SITE*

The antagonist carazolol has been used as a fluorescent probe for the binding site of the beta-adrenergic receptor (beta AR). The fluorescence properties of carazolol are dominated by the emission of the carbazole group, with the fine structure of the spectrum, but not the quantum yield, sensitive to the environment of the probe. The fluorescence emission spectrum of the bound probe is consistent with an extremely hydrophobic environment in the binding site of the receptor. Binding of carazolol to the purified beta AR increases the polarization of the fluorophore. Exposure to collisional quenchers has demonstrated the bound carazolol to be completely inaccessible to the solvent. Furthermore, the fluorescence of bound carazolol is not quenched by exposure to sodium nitrite, a Förster energy acceptor which has an R0 value of 11.7 A with carazolol. Thus, physical analysis of the binding site of the beta AR by carazolol fluorescence indicates that the antagonist binds to the beta AR in a rigid hydrophobic environment which is buried deep within the core of the protein.

The antagonist carazolol has been used as a fluorescent probe for the binding site of the @-adrenergic receptor @AR). The fluorescence properties of carazolol are dominated by the emission of the carbazole group, with the fine structure of the spectrum, but not the quantum yield, sensitive to the environment of the probe.
The fluorescence emission spectrum of the bound probe is consistent with an extremely hydrophobic environment in the binding site of the receptor. Binding of carazolol to the purified BAR increases the polarization of the fluorophore. Exposure to collisional quenchers has demonstrated the bound carazolol to be completely inaccessible to the solvent. Furthermore, the fluoresence of bound carazolol is not quenched by exposure to sodium nitrite, a F+ster energy acceptor which has an Ro value of 11.7 A with carazolol.
Thus, physical analysis of the binding site of the BAR by carazolol fluorescence indicates that the antagonist binds to the @AR in a rigid hydrophobic environment which is buried deep within the core of the protein.
The P-adrenergic receptor (PAR)' is one of the best characterized members of a family of receptors which mediate their actions through signal transduction pathways involving guanine nucleotide binding regulatory proteins (G-proteins). Binding of agonists to these receptors results in the activation of specific G-proteins, leading to the stimulation or inhibition of effector enzymes and modulation of the levels of intracellular second messengers (1). The cloning of several G-protein coupled receptors has shown them to share common structural features, which presumably reflect their similar mechanisms of action (2). The model which has been proposed for these receptors consist of seven transmembrane helices, connected by hydrophilic loops of varying lengths. The amino terminus, which contains two sites of N-linked glycosylation, is proposed to be exposed extracellularly, thereby dictating the alternating internal and external exposure of the remaining loops. This proposed orientation of the cytoplasmic and extracellular loops of the PAR has been confirmed by immunological analysis using anti-peptide antibodies (3, 4). The majority of primary sequence homology among Gprotein coupled receptors is concentrated within the putative * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Genetic analysis of the @AR has revealed that the ligand binding domain of the receptor involves residues within this conserved hydrophobic core of the protein (5,6). Asp113 in transmembrane helix 3 has been determined to be essential for both agonist and antagonist binding to the receptor, suggesting the formation of an ion pair between the protonated amine group of the adrenergic ligand and the carboxylate side chain of the aspartate residue (7,8). In addition, a combined genetic and biochemical approach suggests that the catechol groups of catecholamine agonists bind and activate the PAR through the formation of hydrogen bonds to serine residues at positions 204 and 207 in transmembrane helix 5 (9). Additional mutagenesis studies have implicated residues in helices 2, 6, and 7 in ligand binding to the receptor (2). These mutagenesis data are in agreement with the results of photoaffinity labeling studies, which suggest the involvement of regions within transmembrane helices 2 and 7 in antagonist binding to the BAR (10,11). Similarly, biochemical and genetic analysis of rhodopsin has revealed that the ligand retinal binds to the opsin protein via formation of a Schiff base with Lys "' in the seventh transmembrane helix (12,13). Glu113 in the third transmembrane helix has recently been shown to contribute the counter-ion for this base (14).
The model which emerges from these genetic and biochemical experiments is of a ligand binding site buried within the hydrophobic core of the receptor protein, formed by contributions from residues on several of the transmembrane helices (2,13). Biophysical data to support the model of a buried retinal binding site in rhodopsin has arisen in part from fluorescence quenching studies using retinal as an energy acceptor (15) and from electron spin resonance studies using a spin-labeled retinal analog which forms a stable complex with opsin (16,17). Although the genetic and photoaffinity labeling data suggest that the ligand binding site of the PAR also resides within the hydrophobic domain, there is no direct biophysical or crystallographic evidence to support the assignment of this region to the membrane bilayer. P-Adrenergic ligands are protonated aromatic amines and, as such, are considerably more polar than retinal. The concept of a protonated amine ligand penetrating deep into the transmembrane core of the receptor protein has been challenged with the proposal that the ligand could bind to a more external domain (18). An alternative structure has been proposed for the PAR which provides for a larger extracellular domain with increased secondary structure to accommodate high affinity stereoselective ligand binding (19). Direct physical analysis of the orientation of the ligand in the binding pocket of the BAR would address the validity of using rhodopsin as a model for /3-Adrenergic Receptor Fluorescence the structure of other G-protein coupled receptors. In the present study, we have utilized the high affinity P-adrenergic antagonist carazolol as a fluorescent probe for the ligand binding site of the @AR. Examination of the fine structure and polarization of carazolol fluorescence indicates that the ligand is bound to the receptor in a rigid hydrophobic environment. Furthermore, quenching studies indicate that this ligand binding pocket is buried deep within the core of the receptor protein. EXPERIMENTAL PROCEDURES AND RESULTS'

DISCUSSION
Although the interactions of the PAR with adrenergic ligands have been analyzed by genetic, biochemical, and pharmacological approaches, direct physical information about the structure of the ligand binding site has been lacking. In the present study, we have used the antagonist carazolol as a specific fluorescent probe for the binding site of the PAR. Because of the high affinity of this antagonist for the receptor, it was possible to isolate a PAR-carazolol complex, thus treating carazolol as a pseudo-covalent probe for the active site. The fluorescence properties of this ligand allowed observation of the PAR-carazolol complex without any substantial interference from protein tryptophan fluorescence. Therefore, carazolol could be used as a fluorescent reporter to analyze the environment of the ligand binding site of the @AR. Binding of carazolol to the PAR resulted in a iarge increase in polarization of the carazolol fluorescence. This increase in polarization upon binding to the PAR is indicative of a decrease in the rotation of the carbazole group in the receptor, and suggests the ligand is immobilized in the binding site. Whereas the quantum yield of carazolol was independent of the composition of the solvent, the fine structure of the absorption and fluorescence emission spectra was sensitive to the dielectric constant of the solvent. The structured emission spectrum of carazolol consisted of a high energy peak (varying from 341 to 343 nm) and a lower energy peak (355-358 nm). The ratio of the shotilong wavelength peak increased with increasing polarity of the solvent, providing a convenient measure of the environment of the probe. These comparisons indicated that the binding site of the PAR is very hydrophobic. The environment of carazolol in the receptor binding pocket was observed to be more hydrophobic than 90% ethylene glycol and similar to 90% dioxane.
In addition to altering the relative intensities of the fluorescence peaks, solvents of different polarity were also observed to affect the positions of the peaks. The short wavelength peak displayed a slight blue shift, the magnitude of which correlated with the hydrophobicity of the solvent. However, whereas the peak ratio for carazolol bound to the receptor was similar to that measured in 90% dioxane, the position of the short wavelength peak was more consistent with that of carazolol measured in a dodecyl+D-maltoside solution. It therefore seems likely that the solvent effects on peak position and peak intensity may arise from different mechanisms. One interpretation is that carazolol experiences both specific and general solvent effects. Such divergent solvent effects have been observed for compounds having indole ring systems (25) a polar nitrogen atom (30). Therefore, it is conceivable that, while the overall binding pocket of the @AR is hydrophobic, the nitrogen atom is involved in a specific localized polar environment (for example, a hydrogen bond). Further experiments involving a combination of genetic and biophysical approaches will be necessary to establish the source of this putative ligand-receptor interaction. The receptor-bound carazolol was exposed to a variety of quenchers in order to investigate the accessibility of the binding site of the PAR to the solvent. The compounds which were used had previously been demonstrated to quench the fluorescence of tryptophan.
Because of the similar fluorescence properties of carazolol and tryptophan, it was expected that these compounds would be equally effective in quenching carazolol fluorescence.
The collisional quenchers KI, acrylamide, NaN03, and methyl ethyl ketone were able to quench the fluorescence of free carazolol as shown in Fig. 6 and Table  3. In contrast, these compounds did not appreciably quench the fluorescence of carazolol bound to the PAR. These data suggest that the bound carazolol is not exposed to the solvent to any significant degree.
Since carazolol appears to be shielded from solvent in the binding site of the PAR, it is of interest to compare the parameters of quenching of receptor-bound carazolol with the quenching of internal tryptophans in proteins. Acrylamide quenching has been used to localize buried tryptophan residues and to explore protein flexibility. Examination of a variety of single-tryptophan containing proteins has revealed an inverse correlation between the degree of acrylamide quenching of tryptophan fluorescence and the extent to which the tryptophan residue is buried in the protein (31-33). Of the proteins examined in these studies, ribonuclease T1 had the lowest accurately measurable acrylamide quenching rate (0.2-0.3 M-' ns-'). The tryptophan residue in ribonuclease T1 is almost completely isolated from the solvent, as determined by the position of the fluorescence emission peak and crystallographic measurements (32, 33). The fluorescence of the tryptophan residue of azurin was not measurably quenched by acrylamide (k, < 0.05 Mm1 ns-I); this tryptophan residue is completely sequestered from the solvent (32). In the present study, the bimolecular quenching rate of PAR-bound carazolol by acrylamide was also determined to be below the experimental limits of detection (k, = 0.1 M-' ns-' would be an extreme upper limit). By comparison to the quench constants determined for the occluded tryptophan residues of ribonuclease Tl and azurin, this low quenching rate for PAR-bound carazolol implies a completely buried ligand binding site for the receptor.
In addition to serving as a collisional quencher, NaNO* has a significant spectral overlap with tryptophan (33, 34) and carazolol (present study). Therefore, this compound could also quench the fluorescence of either tryptophan or carazolol by acting as a resonance energy acceptor. As expected, enhanced quenching of free carazolol by this compound was observed when compared to NaN03 (Fig. 6 and Table 3), which does not exhibit any significant spectral overlap. This enhanced quenching was similar to that which has been observed for NaNOz quenching of N-acetyltryptophanamide (33,34). Because it could quench by energy transfer, NaNO* was able to effectively quench the fluorescence of the buried tryptophan in ribonuclease Tl, while NaN03 was not (34). However, in the present study we observed that NaNO* was not able to quench the fluorescence of carazolol in the binding site of the PAR, again suggesting that carazolol bound to the receptor is buried more deeply than the tryptophan in ribonuclease T1. In a similar study, the quenching of Tb3+ fluorescence by energy transfer from retinal in the binding site of rhodopsin was used as a measure of the depth to which retinal was buried in the opsin protein.
The Ro for retinal and Tb3' was determined to be 46. 7 A. Assuming that Tb3' was in the rapid diffusion limit, the distance of closest approach between bound retinal and Tb3' was calculated to be 22 A (15). In the present study, the R. ralue for NaNO* and carazolol was determined to be 11.7 A. If the distance of closest approach between bound carazolol and NaN02 was also 22 A, and assuming that NaN02 was in the rapid diffusion limit (see "Appendix"), then the quenching rate of the fluorescence of PAR-bound carazolol by NaNOz would have been 0.07 M-i ns-'. This value is clearly below the limits of detection for NaN02 quenching in the present study. For the model compound 11-9 (carbazole) undeconic acid in dodecyl-P-D-maltoside micelles, we detected a rate constant of 0.644 M-i ns-' for NaN02 quenching.
This rate constant corresponds to a distance of closest approach of 10.3 A, demonstrating that the fluorescence of the carbazole group could be quenched if it were within 10 A of the surface of the receptor. A detection limit for NaN02 quenching of carazolol fluorescence of 0.6 M-' ns-' can be assigned on the basis of the error measurement (Table 3) and verified by comparison to 11-9 (carbazole) undeconic acid. This limit corresponds to a distance of closest approach of 10.9 A. Thus, the failure to detect any quenching of bound carazolol fluorescence by NaN02 indicates that the carazolol molecule is buried in the @AR at a depth of >ll A.
The demonstration by collisional and energy transfer quenching that the carazolol is not bound near the extracellular surface of the PAR but, rather, resides in a deeply buried hydrophobic domain, agrees with the results of deletion mu-"."M " "05 0010 0 015 [Quencher] (M) tagenesis studies (5,6), which indicate that the ligand binding domain of the PAR involves residues within the hydrophobic core of the protein. According to the model which has been proposed (2), the amino acid residues which have been demonstrated to be important for ligand binding to the @AR, including Aspu3, SerZo4, SerZo7, and Phe*"' (hamster &-adrenergic receptor nomenclature), would be located on various transmembrane helices, all approximately 30-40% of the way into the membrane bilayer. The determination that the carbazole fluorophore is buried at least 10.9 A into the protein provides direct biophysical evidence to support such a model, and is consistent with the hypothesis that this region of the protein forms a membrane-spanning bundle. The parameters of the carazolol binding site which have been determined in the present study are similar to those previously determined for the retinal binding site in mammalian opsin. Using similar fluorescence techniques, retinal was estimated to be 22 A from the membrane surface (15). In ESR studies, a spin-labeled retinal analog was found to be highly immobilized in the binding site of rhodopsin and inaccessible to water-soluble reagents (16), as well as being sequestered from the phospholipid bilayer (17). The observation that the ligand binding domains of rhodopsin and the PAR share these physical characteristics is consistent with the previous observation of the similar hydropathicity profiles and primary structures of these two proteins (20). The results of the present study suggest that, despite the structural differences in the ligands which bind to the @AR and rhodopsin, the structures of the ligand binding sites of these proteins may be similar. The fluorescence properties of bound carazolol provide physical evidence that, like rhodopsin, the ligand binding domain of the PAR is in a constrained, hydrophobic /I-Adrenergic Receptor Fluorescence environment sequestered away from the surface of the protein. _ 50000, quenched than hl""J ca,P2"l"l. this would lead 1" an apparent lime dependen, quenching of the BAR.
Slern-Yolmer plolr far ,he quenching of ca.raolol !lua,ercence by i-M402 are shown in Fip. 6. and the qucnchiny, conslants given in Table 3 Over 97% of the signal was tit lo a lifetime of 13.9 ns.