Distinct Subdomains of Human Endothelin Receptors Determine Their Selectivity to EndothelinA-selective Antagonist and EndothelinB-selective Agonists*

The endothelin (ET) family of peptides acts via two subtypes of G-protein-coupled heptahelical receptors termed ETA and ETB, which have distinct rank orders of affinity to endothelin receptor agonists and antagonists. To delineate which portions of the receptor mol- ecules determine ligand selectivity, we have constructed a series of chimeras between human ETA and ETB receptors and characterized the chimeric recep- tors expressed in heterologous cell lines by competitive radioligand binding analysis and by measuring ago- nist-induced transients of intracellular Ca2+. We dem-onstrate that the binding determinant for the ETB- selective agonists ET-3, BQ3020, and IRL1620 resides within the region spanning the putative transmembrane helices IV-VI and the adjacent loop regions. In contrast, the transmembrane helices I, 11, 111, and VI1 plus the intervening loop regions specify the selectivity for BQ123, an ETA-selective antagonist. BQ123 fitted to a logistic equation (22) by using the nonlinear least squares curve- fitting program UltraFit (Biosoft, Inc.). Ligand affinity and the Hill coefficient were designated as parameters for curve fitting. Both in the binding assays and Ca2+ transient assays, the estimated Hill coefficients were always close to unity (within the range of 0.86-1.20) for all of the wild-type and chimeric receptors examined, being compatible with one-site ligand/receptor interaction without coop- erativity.

(3). They all consist of 21 amino acid residues with two intramolecule disulfide bonds formed between Cys'-Cys'' and Cys3-Cys". The amino acid sequence of each member of the family exhibits a nearly perfect conservation among mammalian species; the only species-related sequence difference known to date is the substitution of Ser4 with Am4 in mouse and rat ET-2, which is also called VIC (4). The sequence of the carboxyl-terminal linear portion is shared by all of the members of the mammalian endothelin family; the amino acid substitutions between the isopeptides are clustered in the amino-terminal loop portions, especially within positions 2-7. Endothelins have a wide variety of biological effects in many different target cell types. Their actions are mediated by specific cell surface receptors that belong to the superfamily of heptahelical G-protein-coupled receptors (5)(6)(7). Two subtypes of endothelin receptor, called ETA and ETB receptors, have been cloned, and both have been shown in many cell types to activate phospholipase C with resultant intracellular Ca2+ transients (8). However, the ETA and ETB receptors have distinct cell type/tissue distributions and thus have different physiological roles (9,10). In many blood vessels, for example, the ETA receptors reside generally in smooth muscle cells and mediate vasoconstrictor responses, whereas the endothelial cells express the ETB receptor, which mediates vasodilator effects via the endothelin-induced release of nitric oxide. They can be pharmacologically distinguished by different rank orders of affinity toward endothelin isopeptides; the ETA receptor is ET-1-selective, showing an affinity rank order of ET-1 2 ET-2 >> ET-3, whereas the ETB receptor exhibits similar affinities to all three isopeptides. In other words, ET-1 can be considered as a nonselective agonist that exhibits similar subnanomolar affinities for both receptor subtypes. ET-3 is a moderately ETB-selective agonist with the affinity to the ETB receptor being 2 orders of magnitude higher than that to the ETA receptor. Recently, a number of synthetic ligands that are highly selective between endothelin receptor subtypes have been developed. For example, the cyclic pentapeptides BQ123, cyclo-(D-Trp-D-Asp-Pro-D-val-Leu), and BQ153, cyclo-(D-Trp-D-Ala(So4)-Pro-D-val-Leu), act as highly ETA-selective antagonists (11)(12)(13). Among highly ETBselective agonists are BQ3020 (N-a~etyl-[Ala"~'~]ET-l(6-21)) (14) and IRL1620 (N-succinyl-[Glug,Ala"~'5]ET-1(8-21)) (15).
Human ETA and ETB receptors exhibit a high polypeptide sequence identity to each other (~5 5 % overall; -74% within the putative transmembrane helices) (16,17). Since they maintain a clear distinction in ligand binding selectivity de-8547 Human ETAIETB Chimeric Endothelin Receptors spite this structural similarity, we set out to localize the subtype-specific determinants within endothelin receptors.
We constructed a series of recombinant chimeras between human ETA and ETB receptors, expressed them in heterologous cell lines, and characterized those receptors in terms of ligand selectivity by competitive radioligand binding assays and agonist-induced intracellular Ca2+ transient assays. We found that the selectivity toward the ETA-selective antagonist and the ETB-selective agonist is specified by different subdomains of these receptors.

MATERIALS AND METHODS
Reagents-ET-1 and ET-3 were purchased from Peptide Institute (Osaka, Japan). BQl23 was a generous gift from Banyu Pharmaceutical Co., Ltd. (Tsukuba, Japan). '"1-ET-1 and lZ5I-BQ3020 were purchased from Amersham Corp. '251-IRL1620 was a kind gift from Du Pont-New England Nuclear. Fura-P/AM was purchased from Dojin Chemicals (Tokyo, Japan). pME18Sf-vector was a kind gift from Dr. K. Maruyama of the Institute of Medical Science, University of Tokyo.
Chimeric Receptor Constructs-cDNA constructs encoding for chimeric human ETA/ETB receptors were assembled by creating common restriction sites within the wild-type receptor cDNAs (16, 17) by oligonucleotide-directed mutagenesis (18) and splicing the desired restriction fragments from the mutated cDNAs. Regions with well conserved amino acid sequences within the putative intracellular loops (ICL 1-111) and extracellular loops (ECL 1-111) were chosen for creating restriction sites. Care was taken to avoid any amino acid insertions/deletions and to minimize amino acid substitutions at the junction sites as much as possible. The restriction sites introduced and their positions in the deduced amino acid sequences of the ETA/ ETB receptors, respectively, were as follows (see Fig. 1): SnuBI, at Ile82/Ile'03; ApaI, a t P r~"~/ P r o '~~; BstBI, a t A~p~~~/ G l y '~' ; BssHII, at C y~"~/ C y s '~~; NcoI, at ProzZ8/Thrz4'; BglII, at G1u281/G1u299; ClaI, at Arg3'o/Arg357. The conserved EcoRI sites at Asn361/Asn378 within the The number of amino acid residues (circles) is based on the ETA sequence; the sequence gaps introduced to the ETB sequence to align the two polypeptides are depicted by striped circles and the insertions by arrows with the number of inserted amino acid ( a ) residues in boxes. Closed circles denote the amino acid residues that are identical between the aligned human ETA and ETB sequences; open circles designate nonidentical residues. Positions of the restriction sites used to construct chimeric receptors are shown by arrowheads with the names of enzymes. One of the clustered Cys residues within the carboxyl-terminal cytoplasmic tail is assumed to be palmitoylated (wavy line).
seventh transmembrane helices (TM VII) of the wild-type receptors were also utilized. Where an amino acid substitution had to be introduced to create a restriction site, we confirmed that the the ligand binding characteristics of the receptors carrying the point mutation alone were indistinguishable from those of the corresponding wild-type receptors (data not shown). The entire coding sequences of the wild-type and chimeric/mutant cDNAs were then subcloned into the SRa promoter-based mammalian expression vector Competitive Rudioligand Binding Study-COS-7 and Ltk-cells were cultured in monolayers in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. The expression constructs were introduced into COS-7 cells and Ltk-cells by the DEAEdextran method, and the cells were subjected to competitive radioligand binding assays as described (17,20). 48-72 h after transfection, the monolayers of cells were incubated at 37 for 60 min with 2 X lo-" M '251-Tyr13-ET-1, 10"' M 1z51-TyP-BQ3020, or 10"' M lZ5I-Tyr6-IRL1620 (specific activities = 2,000 Ci/mmol) in the presence of various concentrations of cold ET-1, ET-3, or BQ123. After extensive washes, cells were lysed with 0.1 N NaOH, and cell-bound radioactivity was determined. Nonspecific binding was determined in the presence of M unlabeled ET-1 and was approximately 2% of the levels of the specific binding in the absence of the competitor for both cell types.
Measurement of Intracellular Ca2+ Transients-We used mouse Ltk-cells for Ca2+ transient assay in this study because this cell line was found to give highly reproducible, large Caz+ responses to endothelin agonists after transfection with the expression vectors. Ltkcells expressing the wild-type and chimeric receptors 48-72 h after transfection were loaded with the Ca'+-sensitive fluoroprobe Fura-2/ AM as described (17). Intracellular Ca2+ transients evoked by various concentrations of ET-1 and ET-3 were monitored by a JASCO CAF-110 fluorescence spectrophotometer with dual excitation at 340 nm/ 380 nm and emission at 500 nm. In some experiments, lo-' M BQ123 was added 10 min prior to the addition of the endothelins. The cytosolic concentration of Ca2+ ([Ca*+]i) was estimated as described (21). Endothelins induced an acute [Ca"], increase in the transfected cells which was followed by lower plateau [Ca2+Ii values. The peak [Ca2+], values from the initial transients were used to draw the doseresponse curves.
Data Analysis-All data were the means of at least two independent experiments done in duplicate or triplicate. Raw data obtained from the radioligand binding and Caz+ transient assays were fitted to a logistic equation (22) by using the nonlinear least squares curvefitting program UltraFit (Biosoft, Inc.). Ligand affinity and the Hill coefficient were designated as parameters for curve fitting. Both in the binding assays and Ca2+ transient assays, the estimated Hill coefficients were always close to unity (within the range of 0.86-1.20) for all of the wild-type and chimeric receptors examined, being compatible with one-site ligand/receptor interaction without cooperativity.

Characterization of Wild-type Endothelin Receptors-COS-
7 cells transfected with wild-type human ETA and ETB receptor cDNA constructs expressed specific binding sites for lZ5I-ET-1 (Fig. 2). The cells transfected with empty vector DNA had no detectable levels of specific ET-1 binding (data not shown). At the lZ5I-ET-l concentration of 2 X lo-'' M, approximately 0.5-1 X lo4 molecules/cell of the radioligand were bound in the absence of competitors. The radioligand was displaced in a competitive manner by ET-1, ET-3, or BQ123. ET-1 had a similar binding affinity to both subtypes of receptor: apparent Ki values for the ETA a n d ETB receptors were 3.5 X M and 9.5 X 10"' M, respectively ( Table 1). ET-3 was virtually equipotent with ET-1 in displacing lZ5I-ET-1 from the ET, receptor (Ki = 2.0 X lo-' M), while being nearly 300 times less potent than ET-1 for the ETA receptor (Ki = 1.0 X 1O"j M). BQ123 was highly selective for ETA with Ki values for the ETA and ETB receptors of 2.5 X lo-* M and 3.1 X 10-~ M, respectively. Ltkcells transiently transfected with the same constructs also showedvirtually identical ligand binding characteristics. We detected no specific binding of Human ETA/ETB Chimeric Endothelin Receptors 8549 , and BQ123 (A). Levels of '=I-ET-l binding are expressed as percentages of the specific binding in the absence of competitor.   and BQ123 to wild-type and chimeric human endothelin receptors determined by competitive binding assay with lZ5I-ET-l as radioligand -   Furthermore, ET-3 produced a considerably smaller maximum response for ETA-expressing cells as compared with ET-1; the maximum responses to ET-3 (at 10"j M ) were 28 and 100% of those to ET-1 in the cells expressing the ETA and ETB receptors, respectively. In the ETA-expressing cells, BQ123 M) inhibited an ET-1 M)-induced [Ca'+Ii increase by 95% (Fig. 4). In contrast, the same concentration of the antagonist inhibited an ET-1 (1O"j M)-induced Ca2+ response by only 9% in the cells expressing the ETB receptor. BQ123 at up to M had no detectable agonist activity for either receptor subtype in this assay (data not shown).

B(N-I)A(II-III)B(IV-VI)A(VII)B(C) B(N)A(I-III)B(IV-VI)A(VII)B(C)
Characterization of Chimeric Receptors-We constructed two systematic series of chimeric endothelin receptors by progressively substituting the structure of the ETA receptor with that of the ETB receptor (see "Materials and Methods"). The progressive substitutions from the carboxyl terminus resulted in the series of A/B chimeras, whereas the aminoterminal substitutions gave rise to B/A chimeras. Thus, for example, chimera A(N-II)B(III-C) hereafter designates a chimeric receptor consisting of the ETA sequence from the amino-terminal putative extracellular tail through the TM 11, followed by the ETB sequence from the TM I11 through the carboxyl-terminal cytoplasmic tail. Each chimeric molecule was expressed in COS-7 cells, and the cells were subjected to a competitive radioligand binding assay. All chimeric receptor constructs we tested conferred similar densities of lZ5I-ET-l binding sites in these cells; between 0.5 and 1 x lo4 molecules/ cell of lZ5I-ET-l (at 2 X lo-" M ) were bound in the absence of competitors. Table I  Provided with this uniformity of the affinity to ET-1 in all chimeric receptors, we divided the K; values for ET-3 and BQ123 by those for ET-1 in each receptor construct and used these affinity ratios (designated hereafter as RET-^ and R B Q~~~, respectively) as an indicator of the ligand selectivity exhibited by each recombinant receptor (Table I).
Determinant for Selectivity to BQ123"Replacement of the amino-terminal extracellular tail of ETA with the corresponding region of ETB caused little change in the affinity to BQ123 (chimera B(N)A(I-C)). However, further progressive replace-

chimeras B(N-I)A(II-C), B(N-II)A(III-C), and B(N-III)A(IV-C)
). Replacement of the carboxyl-terminal half of TM VI1 together with the carboxyl-terminal cytoplasmic tail of ETA with the corresponding region of ETB caused no significant change in BQ123 selectivity (chimera A(N-VII)B(C)). Further substitution of the amino-terminal half of TM VI1 plus the carboxyl-terminal half of ECL I11 resulted in an 11-fold increase in the R~4123 value (chimera A(N-VI)B(VII-C)). The selectivity to BQ123 did not change very much when further substitution of the ETA sequence from the carboxyl-terminal side was carried out through TM IV and ICL I1 (chimeras A(N-V)B(VI-C), A(N-IV)B(V-C), and A(N-III)B(IV-C)). However, when the TM 111 together with the carboxyl-terminal half of ECL I was further substituted (chimera A(N-II)B(III-C)), an additional =lOO-fold increase of RBQ123 was observed.
These results suggest that the ETA sequences spanning from TM I through TM I11 as well as the amino-terminal half of TM VI1 are likely to be important for its high affinity binding to BQ123. To test whether these regions were also sufficient to define the binding determinant for the ETAselective antagonist, we constructed the chimeras B(N)A(I-III)B(IV-VI)A(VII)B(C). This chimeric receptor indeed exhibited the ability to bind BQ123 with high affinity, with an RBQ123 value similar to that for the wild-type ETA receptor (Table I and Fig. 5). Furthermore, BQ123 M ) inhibited the ET-1-induced Ca2+ transient response by 96% in Ltkcells transfected with this chimeric construct (Fig. 4). BQ123 showed no detectable agonist activity at up to M. Al- Determinant for Selectivity to ET-3-Progressive replacement of the sequence of the ETB receptor with that of ETA from the amino-terminal extracellular tail through TM I11 caused little changes in RET. 3  Peptide (log M) -1 (0) and BQ123 (A). Levels of lZ5I-ET-l binding are plotted as percentages of the specific lZ5I-ET-from the wild-type ETA receptor in terms of RET.^ values.

FIG. 5. Displacement of 1261-ET-1 binding to COS-7 cells expressing chimeric receptor B(N)A(I-III)B(IV-VI)A(VII)B(C) by unlabeled ET
Similarly, progressive replacement of the ETB sequence with ETA from the carboxyl-terminal cytoplasmic tail through the carboxyl-terminal half of ECL I11 (chimeras B(N-VII)A(C) and B(N-VI)A(VII-C)) resulted in little change in RET-^. However, further substitution of the amino-terminal half of ECL 111, TM VI, and most of ICL I11 with the ETA sequence caused a significant increase of the selectivity ratio to 55 (chimera These results suggest that the structure of the ETB receptor spanning ICL I1 through the amino-terminal half of ECL I11 was necessary to render the receptor able to bind ET-3 with high affinity. To confirm further that the above mentioned region from the ETB receptor is sufficient to provide high affinity binding of ET-3, we constructed the chimera A(N-III)B(IV-VI)A(VII-C). Indeed, this chimeric receptor had virtually equal affinities for ET-1 and ET-3 both in the radioligand binding assay and in the [Ca"]i transient assay (Table  I and Figs. 2 and 3). In contrast, the chimeric receptor B(N-IV)A(V)B(VI-C) displayed an intermediate affinity to  its RET.^ value was 7 times lower than the wild-type ETA yet 20 times higher than the ETB. This indicates that although the sequence of the ETB receptor spanning ECL 11 and TM V significantly contributes to the binding determinant for ET-3, this structure alone is not sufficient to form the complete determinant.
There are two charge-modifying amino acid substitutions between the ETA and ETB receptors in their TM IV and TM VI: namely, AspZ4l and Leu347 in ETB versus Valzz5 and Lys330 in ETA. To examine if these substitutions of charged residues have any effect on the binding affinity of ET-3, we introduced the following point missense mutations at these positions: ET~(v225D), ET~(K3301), ET~(v225D/K3301). However, we found that the ET-3 selectivities of these mutant receptors were all indistinguishable from the respective wild-type receptors (data not shown).
A Chimeric Receptor with High Affinity to Both ETa and ETB-selective Ligands-Since TM I, 11, 111 and VI1 from the ETA receptor (including the intervening loop regions) were sufficient to form the high affinity binding determinant for BQ123 in the chimeric receptors, we expected that the "ETBlike" chimera A(N-III)B(IV-VI)A(VII-C) could still maintain high affinity to BQ123. This was actually the case; the chimeric receptor had an affinity to BQ123 which was very similar to native ETA receptor (Table I and Fig. 2). Therefore, it seemed that this particular chimeric receptor could accept both ETA-and ETB-selective ligands with high affinity. We performed competitive binding studies on this chimeric construct by using the highly ETB-selective agonists 1251-BQ3020 and '251-IRL1620 as radioligands and the highly ETA-selective antagonist BQ123 as competitor. We found that the chimeric receptor specifically bound these ETB-selective radioligands in a manner similar to the native ETB receptor (Fig. 6). Moreover, the specific binding of the ETB-selective radioligands was completely abolished by M BQ123 in the chimeric receptor (Fig. 6). In contrast, in the case of the wildtype ETB receptor, the radioligand binding was affected by M BQ123 only slightly. The apparent K; values for BQ123 in displacing the specific binding of lZ5I-BQ3020 and lZ5I-IRL1620 in the chimeric receptor were 1.9 X M and 1.0 X M, respectively (Fig. 7). These values were similar to the Ki values for BQ123 observed in the wild-type ETA receptor by using '"I-ET-l as radioligand (Table I and Fig. 2).
We further examined whether BQ123 still acted as an antagonist for the chimeric receptor A(N-III)B(IV-VI)A(VII-

Wild-type B
FIG. 6. Specific binding of the ETB-selective ligand '"I-BQ3020 (filled bars) and '261-IRL1620 (striped bars) to COS-7 cells expressing the designated receptor constructs. Cells were separately incubated with each labeled peptide (10"' M) in the absence (-) and presence (+) of M BQ123. Levels of specific binding are expressed as percentages of those seen in the wild-type ETB receptor in the absence of BQ123.  (Fig. 4). BQl23 did not have a detectable agonist activity at up to M.

DISCUSSION
All of the chimeric receptor constructs we examined in this study conferred similar densities of specific 'T-ET-l binding sites when expressed in COS-7 cells. Furthermore, all chimeric receptors exhibited high affinities to ET-1 which were comparable with those observed in the wild-type ETA and ETB receptors. Taken together with the fact that ET-1 is a nonselective ligand for both receptor subtypes, these findings suggest that the details of tertiary structure of the ETA and ETB receptors are similar to each other and well maintained in all chimeric receptors tested. This may at least partly be because of the high level of polypeptide sequence similarity seen between the two receptor subtypes. By systematically constructing and analyzing chimeric endothelin receptors exhibiting these prerequisite properties, we demonstrated the clearly distinct binding determinants for subtype-selective agonists and antagonists. A separation of agonist and antagonist binding sites has recently been reported for human progesterone receptor (23). It is of interest to examine whether the complete separation of agonist and antagonist binding determinants can be seen also in other family of heptahelical G-protein-coupled receptors, especially in receptors for other

Human ETAIETB Chimeric Endothelin Receptors
peptide ligand families such as tachykinins (24). The amino acid sequences of the ETA and ETB receptors are most dissimilar to each other in their putative aminoterminal extracellular tails, which have predicted N-glycosylation sites. In fact, there is no detectable sequence similarity between these portions of the two receptor subtypes at all. It is therefore somewhat surprising that mutual swapping of the extracellular tails between the ETA and ETB receptors had no appreciable effect on their ligand binding characteristics (chimeras B(N)A(I-C) and A(N)B(I-c)). The function of these parts of the receptor molecules is not clear at present; one possibility is that they may assist the receptor polypeptides to fold and be expressed on the cell surface in the proper transmembrane orientations.
BQ123 is an established competitive antagonist selective for the ETA receptor (11). The three-dimensional structures of BQ123 and ET-1 in solution have been demonstrated with proton NMR by several independent laboratories (25-28). However, so far it has been unclear as to which part of the ET-1 structure may be mimicked by the antagonist. We demonstrated in this study that the chimeric receptor A(N-III)B(IV-VI)A(VII-C) could achieve a high affinity binding of both BQ123 and the two ETB-selective agonists, BQ3020 and IRL-1620. Furthermore, we found that BQ123 could compete with iodinated BQ3020 or IRL1620 for the specific binding to this chimeric receptor. The latter ligands are linear, amino-terminally truncated derivatives of ET-1. These findings strongly argue for the idea that BQ123 mimicks a structure within the carboxyl-terminal, linear portion of ET-1. In this regard, the amino acid sequence of BQ123 appears to have certain similarity to the carboxyl-terminal sequence of ET-1. The motif found in BQ123, -D-Val-Leu-D-Trp-D-Asp, may mimic the carboxyl terminus of endothelins, -1le-Ile-Trp-COOH. We also demonstrated that the determinant for the high affinity binding of BQ123 resides within the TM 1-111 and VI1 (including adjacent loops) of the ETA receptor. This is in agreement with the findings recently reported by Adachi et al. (29) that the ECL I from the ETA receptor is an important determinant for BQ123 binding. Taken together with the above argument, it seems plausible to consider that these portions of the receptor molecules contain at least a part of their binding domain for the carboxyl termini of endothelins, which includes the TrpZ1 residue essential for the binding of the peptide to either receptor subtype (30).
We have demonstrated that the TM IV-VI and the adjacent loop regions of the ETB receptor constitute the high affinity binding determinant for the ETB-selective agonists, including the amino-terminally truncated linear ET-1 derivatives. This is in sharp contrast to the case of the heptahelical receptors for tachykinins, in which the region spanning from TM I1 to ECL I1 (together with a minor contribution of the aminoterminal tail) has been recently reported to specify isopeptide selectivity (24). It is interesting to note that, like endothelin family, tachykinins have a common carboxyl-terminal motif and divergent amino-terminal portions. These observations indicate a significant difference in the mode of receptor/ligand interactions between the endothelin and tachykinin systems, despite these apparent similarities in the general configurations of isopeptides and receptors.
The findings presented in this study provide further insight into the possible mechanism for the ligand selectivity of endothelin receptors. It is plausible to consider that the structures of endothelin isopeptides and their derivatives are comprised of two distinct subdomains: the amino-terminal disulfide loop portion that is variable among isopeptides, and the carboxyl-terminal linear hydrophobic region that is highly conserved. The currently available information on the structure-activity relations of endothelin derivatives suggests the following general rules (30, 31): (i) both the amino-terminal loop structure from ET-1 (or other nonselective ligands such as sarafotoxin S6b) and the common carboxyl-terminal linear structure are required for high affinity binding to the ETA receptor; (ii) in contrast, the ETB receptor only requires the carboxyl-terminal half of the ligand. In this context, the present results suggest that the TM IV-VI of endothelin receptors may interact with the amino-terminal loop portion of the ligands. Incorporating these considerations, we present a hypothetical model for the binding determinants of endothelin receptors and their ligands (Fig. 8). Thus, it is conceivable that the amino-terminal loop domain of ET-1 functions in a manner similar to a classical "address" domain suggested for a number of peptide ligand family (32). The TM IV-VI of the ETA receptor interacts selectively with this address domain of ET-1 to promote a full activation of receptor, which is brought about by the carboxyl-terminal "message" domain of ET-1. However, the corresponding address domains of the ETB-selective ligands are either invalid (ET-3) or missing (BQ3020, IRL1620), resulting in the inability to interact with the ETA receptor. In contrast, the ETB receptor, with its subtly different structures within the TM IV-VI region, may be highly promiscuous for the ligands' address domains. Alternatively, it is tempting to speculate that the TM IV-VI domain of the ETB receptor may function as a self-contained address recognition domain that does not require a valid address information presented by the ligand. Further, it is also conceivable that the mode of interaction between the cyclic pentapeptide antagonist BQ123 and the ETA receptor Endothelin agonists are presumably comprised of the carboxyl-terminal message sequence common to all of the natural and synthetic isopeptides (filled squares) coupled with the amino-terminal address domain, which involves the disulfide loop portion of the isopeptides. The region spanning the TM IV-VI (and the intervening loop portions) of the ETA receptor interacts with the address domain specific to ET-1. ET-3 has an address domain that can interact only weakly with this address recognition subdomain of the ETA receptor. The highly ET~-selective agonists BQ3020 and IRL1620 have no valid address domain because of the amino-terminal truncations. However, in the case of the ETB receptor, the corresponding address recognition subdomain can interact with a much wider spectrum of ligands' address portions. Alternatively, the ETs receptor presumably has an internal, self-content address domain (depicted with a striped hemisphere) so that it requires no external address sequence. The cyclic pentapeptide antagonist BQ123 interacts with the subdomain of the ETA receptor covering the TMs 1-111 and VI1 plus the adjacent loop portions (note that despite the apparent separation of these transmembrane helices in the primary structure, they are presumably contiguous to each other in the tertiary structure common to the rhodopsin superfamily (5)). However, because of a subtle difference in the mode of binding, the antagonist is unable to evoke the conformational alteration necessary to activate the receptor and also becomes exempt from the requirement of interactions at the address domain.

Human ETAIETB Chimeric Endothelin
Receptors 8553 is significantly different from agonist/receptor interactions so that it does not require the interaction at the addressing domain.