Identification of amino acid residues that direct differential ligand selectivity of mammalian and nonmammalian V1a type receptors for arginine vasopressin and vasotocin. Insights into molecular coevolution of V1a type receptors and their ligands.

Arginine vasotocin (VT) is the ortholog in all nonmammalian vertebrates of arginine vasopressin (AVP) in mammals. We have previously cloned an amphibian V1atype vasotocin receptor (VT1R) that exhibited higher sensitivity for VT than AVP, while the mammalian V1a type receptor (V1aR) responded better to AVP than VT. In the present study, we identified the amino acid residues that confer differential ligand selectivity for AVP and VT between rat V1aR and bullfrog VT1R (bfVT1R). A chimeric rat V1aR having transmembrane domain (TMD) VI to the carboxyl-terminal tail (C-tail) of bfVT1R showed a reverse ligand preference for AVP and VT, whereas a chimeric VT1R with TMD VI to the C-tail of rat V1aR showed a great increase in sensitivity for AVP. A single mutation (Ile(315(6.53)) to Thr) in TMD VI of V1aR increased the sensitivity for VT, while a single mutation (Phe(313(6.51)) to Tyr or Pro(334(7.33)) to Thr) reduced sensitivity toward AVP. Interestingly the triple mutation (Phe(313(6.51)) to Tyr, Ile(6.53) to Thr, and Pro(7.33) to Thr) of V1aR increased sensitivity to VT but greatly reduced sensitivity to AVP, behaving like bfVT1R. Further, like V1aR, a double mutant (Tyr(306(6.51)) to Phe and Thr(327(7.33)) to Pro) of bfVT1R showed an increased sensitivity to AVP. These results suggest that Phe/Tyr(6.51), Ile/Thr(6.53), and Pro/Thr(7.33) are responsible for the differential ligand selectivity between rat V1aR and bfVT1R. This information regarding the molecular interaction of VT/AVP with their receptors may have important implications for the development of novel AVP analogs.


Arginine vasotocin (VT) is the ortholog in all nonmammalian vertebrates of arginine vasopressin (AVP) in mammals.
We have previously cloned an amphibian V1atype vasotocin receptor (VT1R) that exhibited higher sensitivity for VT than AVP, while the mammalian V1a type receptor (V1aR) responded better to AVP than VT. In the present study, we identified the amino acid residues that confer differential ligand selectivity for AVP and VT between rat V1aR and bullfrog VT1R (bfVT1R). A chimeric rat V1aR having transmembrane domain (TMD) VI to the carboxyl-terminal tail (C-tail) of bfVT1R showed a reverse ligand preference for AVP and VT, whereas a chimeric VT1R with TMD VI to the C-tail of rat V1aR showed a great increase in sensitivity for AVP. A single mutation (Ile 315(6.53) to Thr) in TMD VI of V1aR increased the sensitivity for VT, while a single mutation (Phe 313(6.51) to Tyr or Pro 334(7.33) to Thr) reduced sensitivity toward AVP. Interestingly the triple mutation (Phe 313(6.51) to Tyr, Ile 6.53 to Thr, and Pro 7.33 to Thr) of V1aR increased sensitivity to VT but greatly reduced sensitivity to AVP, behaving like bfVT1R. Further, like V1aR, a double mutant (Tyr 306(6.51) to Phe and Thr 327 (7.33) to Pro) of bfVT1R showed an increased sensitivity to AVP. These results suggest that Phe/Tyr 6.51 , Ile/Thr 6.53 , and Pro/Thr 7.33 are responsible for the differential ligand selectivity between rat V1aR and bfVT1R. This information regarding the molecular interaction of VT/ AVP with their receptors may have important implications for the development of novel AVP analogs.
Arginine vasopressin (AVP) 1 is a cyclic nonapeptide that exerts a variety of biological effects in mammals. The primary role of AVP involves the regulation of water and solute excretion from the kidney (1,2). The other physiological functions of AVP include control of blood pressure, platelet aggregation, liver glycogenolysis and neoglucogenesis, adrenocorticotropin release from the adenohypophysis, and aldosterone secretion from the adrenal gland (3)(4)(5)(6). AVP is also implicated in interneuronal communication in the central nervous system and modulates several behavioral functions such as feeding, memory, thermoregulation, and the control of adaptive, social, and sexual processes (7,8).
Vasotocin (VT; [Ile 3 ]AVP) is the AVP counterpart in most nonmammalian vertebrates including lungfishes, amphibians, reptiles, and birds. In amphibians, VT is expressed in both hypothalamic and extrahypothalamic cell bodies (9), indicating that VT acts both as a neurohormone and neurotransmitter as reported for AVP in mammals. At the periphery, VT regulates osmotic and electrolyte balance (10,11) and steroid secretion from frog adrenocortical cells (12,13). In the central nervous system, VT functions as a neuromodulator to control reproductive behavior in amphibians (14,15).
In mammals, the actions of AVP are mediated through three different AVP receptors: V1aR, V1bR, and V2R (16). All these receptor subtypes belong to the G protein-coupled receptor superfamily but differ in their tissue distribution, their relative affinity for synthetic analogs, and their signaling mechanisms (17,18). All three types of receptors are expressed in the central nervous system (19). However, at the periphery, V1aR is mainly expressed in vascular smooth muscle cells and hepatocytes (20), whereas V1bR is exclusively located in pituitary corticotrophs (21,22), and V2R occurs in the kidney (23). Upon ligand stimulation, V1aR and V1bR trigger the phospholipase C/protein kinase C signaling pathway, while V2R activates the adenylyl cyclase/protein kinase A pathway (24,18).
We have recently isolated and characterized a V1aR-type vasotocin receptor (VT1R) in the frogs Rana catesbeiana and Rana esculenta (25). Frog VT1R exhibits the highest sequence identity (58%) with mammalian V1aR among the AVP receptor subtypes and preferentially couples to the phospholipase C/protein kinase C signaling pathway as does V1aR. Frog VT1R shows a high sensitivity to VT but a very poor sensitivity to AVP. It is of interest to note that V1aR exhibits relatively high sensitivity to both AVP and VT, although AVP is slightly more potent than VT. Mutational studies coupled with computational modeling have demonstrated that the VT/AVP binding sites are localized within a narrow cleft delimited by most of the transmembrane regions, 15 Å deep from the extracellular surface of rat V1aR, and that these binding sites may be common to V1bR, V2R, and VTR as these receptors show high conservation of residues responsible for ligand binding (26). Several other studies were also performed to investigate the agonist and antagonist binding sites in V1aR (27)(28)(29)(30)(31). However, little is known about the molecular changes of VT1R to V1aR that lead to the high sensitivity of V1aR to AVP. In the present study, using a well defined chimeric receptor approach and site-directed mutagenesis (32,33), we attempted to determine the amino acid residues in rat V1aR and bullfrog VT1R (bfVT1R) that account for selective ligand sensitivity to AVP and VT. ]VT were purchased from Bachem (Bubendorf, Switzerland). The OT agonist [Thr 4 ,Gly 7 ]OT was purchased from Sigma.

Peptides-Synthetic
Plasmids-The pcDNA3 expression vector was purchased from Invitrogen. The pCMV␤-Gal vector was obtained from Clontech. The c-fos-luc vector, containing the Ϫ711 ϳ ϩ45 sequence of the human c-fos promoter constructed in the pFLASH vector, was a kind gift from Dr. R. Prywes, Columbia University. The full-length bfVT1R cDNA was cloned into the pcDNA3 expression vector at the EcoRI and XhoI enzyme sites as mentioned earlier (25). The rat V1aR cDNA was amplified through PCR from rat liver tissue and inserted at the EcoRI and XhoI sites of the pcDNA3.
Constructions of Chimeras and Mutants-For domain swapping between bfVT1R and rat V1aR, individual cDNA fragments of interest were amplified through PCR by using Vent polymerase (New England Biolabs, Beverly, MA) and two specific primers, one corresponding to the 5Ј or the 3Ј end of the receptor cDNAs and another corresponding to the region of overlap between the two receptors. The two fragments, one from bfVTR1 and the other from rat V1aR, were subjected to a second round of PCR to generate the chimeric cDNA. All the chimeric constructs were cloned into the pcDNA3 expression vector at the EcoRI and XhoI sites. The single and double mutants were constructed by PCRbased site-directed mutagenesis and then cloned into pcDNA3. The DNA sequences of the chimeras and mutants were analyzed by the dideoxy chain termination method using the DNA Sequencing kit (U.S. Biochemical Corp., Cleveland, OH) to confirm the accuracy of the constructions.
Cell Transfection and Luciferase Assays-CV-1 cells were maintained in Dulbecco's modified Eagle's medium in the presence of 10% fetal bovine serum. For luciferase assays, cells were plated in 24-well plates 1 day before transfection and transfected with SuperFect reagent (Qiagen, Chatsworth, CA) according to the manufacturer's instructions. Approximately 48 h after transfection, cells were treated with the respective ligands for 6 h. For c-fos promoter-driven luciferase assay, cells were maintained in serum-free Dulbecco's modified Eagle's medium 16 -18 h before treatment with the ligand as described previously (32). Cells were harvested 6 h after ligand treatment, and luciferase activity in cell extracts was determined using a luciferase assay system according to the standard method in a Lumat LB9501 luminometer (EB & G Berthold, Germany). The luciferase values were normalized by the ␤-galactosidase values. Transfection experiments were performed in duplicate and repeated at least three times.
Molecular Modeling-The rat V1aR was built by MODELLER 6v2 (34) based on the crystal structure of the bovine rhodopsin (35) as a template. AVP was docked into the putative binding site of the rat V1aR constrained by previously known interactions. The models for rat V1aR/ VT, triple V1aR/AVP mutant, and triple V1aR/VT mutant were built by mutating corresponding residues in the rat V1aR/AVP model, and the models went through energy minimization and molecular dynamics annealing simulations in the MODELLER program. The final geometry was confirmed by PROCHECK (36). The contacts between ligands and receptors were analyzed using Ligplot (37). The figures of the models were drawn using Visual Molecular Dynamics (38).
Data Analysis-Data were expressed as mean Ϯ S.E. of three independent experiments. Data analysis was performed using nonlinear The plasmids containing the wild-type and chimeric receptors were cotransfected with the c-fos-luc reporter vector into CV-1 cells along with ␤-galactosidase as an internal control. Forty-eight hours after transfection, cells were treated with graded concentrations of AVP or VT for 6 h, and luciferase activity was determined.

FIG. 1. Ligand selectivity of rat V1aR and bfVT1R.
A and B, plasmids containing rat V1aR (A) or bfVT1R (B) cDNA were cotransfected with the c-fos-luc reporter vector into CV-1 cells along with ␤-galactosidase as an internal control. Forty-eight hours after transfection, cells were treated with graded concentrations of AVP (E) or VT (Ⅺ) for 6 h, and luciferase activity was determined. regression, and data were expressed using sigmoid dose-response curves. Agonist or antagonist concentrations inducing half-maximal stimulation (EC 50 ) or half-maximal inhibition (IC 50 ) and maximal fold increase (E max ) were calculated using GraphPad PRISM2 software (GraphPad, San Diego, CA). Statistical analysis was performed by one-way analysis of variance followed by Bonferroni test. Data were considered significant if p Ͻ 0.05.

Fragment of TMDs VI to VII Confers Differential Ligand
Selectivity-Differential ligand selectivity of rat V1aR and bfVT1R was examined using a c-fos promoter-driven luciferase assay (25,39). Rat V1aR had higher sensitivity toward AVP than VT as revealed by EC 50 value ( Fig. 1A and Table I). The bfVT1R showed a high sensitivity to VT but very low sensitivity toward AVP ( Fig. 1B and Table I). VT differs from AVP by a single amino acid at position 3 (Phe for AVP and Ile for VT). Thus, it is likely that some critical amino acids responsible for interacting with Phe 3 of AVP or Ile 3 of VT are differentially arranged between V1aR and bfVT1R. As an initial step toward identifying selectivity-conferring regions, large segments of TMDs, intracellular loops (ICLs), and extracellular loops (ECLs) were swapped between rat V1aR and bfVT1R, and a series of reciprocal chimeras were constructed (Table I). The first set of chimeric receptors is composed of the amino-terminal fragment of bfVT1R in rat V1aR and vice versa (V1aR/ VT1R (N) and VT1R/V1aR (N) ). The second set includes the region from the amino terminus to half of ICL2 of one receptor followed by the sequence of the other receptor (V1aR/ VT1R (N-ICL2) and VT1R/V1aR (N-ICL2) ). The third set contains a substitution of the TMD VI to the carboxyl-terminal tail (C-tail) of one receptor with the other (V1aR/VT1R (TMD6-C) and VT1R/ V1aR (TMD6-C) ). The EC 50 values of AVP and VT for the different chimeric receptors are indicated in Table I. Replacement of the amino-terminal domain of rat V1aR by that of bfVT1R (V1aR/ VT1R (N) ) led to a decrease in sensitivity to both AVP and VT. However, no significant changes in ligand sensitivity were observed in the reciprocal chimera, VT1R/V1aR (N) . The V1aR/ VT1R (N-ICL2) chimera showed a reduced sensitivity to both AVP and VT, while the VT1R/V1aR (N-ICL2) chimera showed an increased sensitivity to VT. Interestingly V1aR/VT1R (TMD6-C) showed a great decrease in sensitivity to AVP but a marked increase in sensitivity to VT, resulting in a reverse ligand selectivity (Table I and Fig. 2A). The opposite VT1R/ V1aR (TMD6-C) chimera showed a great increase in sensitivity to AVP but a modest increase in sensitivity to VT (Table I and Fig.  2B). This result indicates that the residues determining ligand selectivity lie mainly in the sequences between TMDs VI and C-tail.
To further dissect the subdomains responsible for the ligand selectivity, three additional bfVT1R chimeras were constructed: VT1R with a fragment from ECL-3 to C-tail of rat V1aR (VT1R/V1aR (ECL3-C) ), VT1R with TMD VII to C-tail of rat V1aR (VT1R/V1aR (TMD7-C) ), and VT1R with the C-tail of rat V1aR (VT1R/V1aR (C) ) ( Table II). Compared with VT1R/V1aR (TMD6-C) , VT1R/V1aR (ECL3-C) exhibited a lowered sensitivity to AVP. VT1R/V1aR (TMD7-C) showed a significantly decreased sensitivity to AVP. Moreover VT1R/V1aR (C) revealed a marked decrease in sensitivity to AVP such that it had an EC 50 value for AVP similar to that of wild type bfVT1R. These results strongly suggest that the fragment between TMD VI to TMD VII is crucial for the ligand selectivity of mammalian and nonmammalian V1a-type receptors to AVP and VT.
Identification of Individual Amino Acids Important for Determining Ligand Selectivity-To further identify the molecular determinants underlying such a differential ligand selectivity, the amino acid sequences of TMD VI to TMD VII within V1a-type receptors were analyzed (Fig. 3). As it has been demonstrated previously that nonapeptide hormones bind into a cleft 15 Å deep defined by the upper halves of TMDs (26), these The plasmids containing the VT1R chimeric receptors were cotransfected with the c-fos-luc reporter vector into CV-1 cells along with ␤-galactosidase as an internal control. Forty-eight hours after transfection, cells were treated with graded concentrations of AVP or VT for 6 h, and luciferase activity was determined.

FIG. 2. Comparison of ligand selectivities of chimeric V1aR and VTR.
A and B, plasmids containing the V1aR/VT1R (TMD6-C) (A) or VT1R/V1aR (TMD6-C) (B) cDNA were cotransfected with the c-fos-luc reporter vector into CV-1 cells along with ␤-galactosidase as an internal control. Forty-eight hours after transfection, cells were treated with graded concentrations of AVP (E) or VT (Ⅺ) to the V1aR/VT1R (TMD6-C) (A) or VT1R/V1aR (TMD6-C) (B) chimeric receptor for 6 h, and luciferase activity was determined. The dose-response activity was compared with that of the wild type rat V1aR (dashed lines, • for AVP and f for VT). chi, chimera; wt, wild type. regions were examined more carefully. In addition, the sequence of chicken VT1R was also considered as it shows a relatively high sensitivity to AVP among nonmammalian VT1Rs (40).
Three individual amino acid residues Phe 313(6.51) , Ile 315(6.53) , and Pro 334 (7.33) in the rat V1aR were chosen and mutated to those in bfVT1R. Phe 307(6.51) in the hV1aR is known to be important for the interaction with antagonists but not agonists (41). However, mutation of Phe 6.51 3 Tyr in rat V1aR led to a drastic decrease in the AVP sensitivity, while the sensitivity for VT remained unchanged (Fig. 4, A and B, and Table III). Mutation of Ile 6.53 to Thr in rat V1aR did not significantly alter sensitivity for AVP but increased sensitivity for VT (Fig. 4, A and B, and Table III). This mutation increased receptor efficacy for both VT and AVP (Fig. 4, A and B, and Table III). Interestingly the Pro 7.33 3 Thr mutation led to a significant decrease in sensitivity to AVP, but this mutation did not alter sensitivity to VT (Fig. 4, A and B, and Table III). Based on these results we constructed a triple rat V1aR mutant (F6.51Y/I6.53T/P7.33T) to further explore the combined effect of these amino acids. This F6.51Y/I6.53T/P7.33T mutant showed approximately a 1000-fold reduced sensitivity for AVP, while it had a similar sensitivity for VT compared with the wild type receptor (Fig.  4C). This result is consistent with that obtained with the V1aR/ VT1R (TMD6-C) chimera that showed a drastic decrease (about 1000-fold) in AVP sensitivity.
In bfVT1R, individual reciprocal amino acids, Tyr 306(6.51) , Thr 308(6.53) , and Thr 327 (7.33) , were replaced by those of rat V1aR. The Tyr 6.51 3 Phe mutation resulted in a slight increase in AVP sensitivity. The Thr 6.53 3 Ile mutation led to a significant decrease in both VT and AVP sensitivity. Interestingly the Thr 7.33 3 Pro mutation resulted in a significant increase in sensitivity to AVP (Fig. 5, A and B, and Table III). Three other mutants were constructed to determine the relative importance of the Thr 304(6.49) , Asn 320 (7.26) , and Ile 322(7.28) residues in bfVTR. The Thr 6.49 3 Ala mutation led to an impairment of receptor activity (data not shown). The Asn 320(7.26) 3 Ile and the Ile 322(7.28) 3 Thr mutations in ECL-3 had no effect in AVP sensitivity (Table III). As we observed increased ligand sensitivity for AVP in the Y6.51F and T7.33P mutants, we generated the double mutant Y6.51F/T7.33P. This double mutant showed a 100-fold increase in AVP sensitivity and a 10-fold increase in VT sensitivity (Fig. 5C). This is also comparable with that of the chimeric VT1R/V1a (TMD6-C) receptor. To examine the effect of Thr 308(6.53) together with the Y6.51F and T7.33P mutations we constructed a triple bfVT1R mutant (Y6.51F/T6.53I/ T7.33P). This mutant did not display significant change in AVP and VT sensitivity compared with the double bfVT1R mutant Y6.51F/T7.33P. Taken together, these results suggest that Ile/ Thr 6.53 in TMD VI is involved in VT selectivity, whereas the Phe/Tyr 6.51 in TMD VI and Pro/Thr 7.33 in TMD VII are important for the interaction with AVP.
Ligand Sensitivity of Mutant Receptors for Agonists and Antagonists-To further elucidate the role of the Phe/Tyr 6.51 , Ile/ Thr 6.53 , and Pro/Thr 7.33 residues in differential ligand sensitivity, the sensitivity of wild type and mutant receptors for agonists and antagonists was also examined. First we studied The sequences of TMD VI, ECL-3, and TMD VII of the mammalian V1aR and nonmammalian VT1R have been aligned. The amino acid residues that were mutated in the TMD VI and TMD VII of rat V1aR and bfVT1R are indicated as boxes.
Molecular Modeling-We constructed three-dimensional models of wild type rat V1aR (Fig. 7, A and C) and triple rat V1aR mutant (F6.51Y/I6.53T/P7.33T) (Fig. 7, B and D) based on the crystal structure of the bovine rhodopsin and docked with AVP and VT. TMDs I to V are shown by gray cylinders, whereas TMD VI and VII are represented by yellow ribbons. The side chains of residues Phe 6.51 , Ile 6.53 , and Pro 7.33 of the receptor and the third amino acid residue of AVP or VT are shown as ball and stick models in the following colors: carbon atoms in cyan, oxygen atoms in red, nitrogen atoms in blue, and sulfur atoms in yellow. Our model showed essential interactions that have been suggested previously (26 -28). Tyr 2 of AVP was found to have hydrophobic interactions with the Trp 310(6.48) and Phe 313(6.51) residues located in TMD VI of wild type rat V1aR as described previously (26). It has been suggested that the Arg 8 residue of VT/AVP may interact with the Tyr 115(2.68) residue in ECL1 of human V1aR (27). In rat V1aR, Tyr 2.68 has been replaced by Ser. Our model depicts that Arg 8 forms a hydrophobic interaction with Ser 2.68 (data not shown). In addition, the carboxyl-terminal tripeptide Pro 7 -Arg 8 -Gly 9 interacts with the patch of Leu 42(1.23) to Arg 46(1.27) , which is a key determinant in AVP binding (28). As AVP and VT differ by a single amino acid at position 3, Phe 3 of AVP or Ile 3 of VT are likely responsible for differential ligand selectivity. In this model, Phe 3 of AVP was found to have a close contact with Phe 6.51 , Phe 6.52 , and Gln 6.55 located in TMD VI (Fig. 7A). In the triple V1aR mutant (F6.51Y/I6.53T/P7.33T), mutation of ,des-Gly-NH 2 9 ]VT (ࡗ) in the presence of 10 nM VT for 6 h, and luciferase activity was determined and compared with the respective wild type receptors (V1a antagonist, ƒ; V2 antagonist, ϩ; and OT antagonist, ‚). mt, mutant; wt, wild type; anta, antagonist. Phe 6.51 to Tyr led to a loss of interaction with Phe 3 of AVP (Fig.  7C). The amino acids in rat V1aR interacting with VT were essentially the same as those interacting with AVP except for Ile 3 of VT that was found to form hydrophobic interactions with Val 219 (5.39) in TMD V and Ile 6.53 and Gln 6.55 in TMD VI (Fig.  7B). However, replacement of Ile 6.53 with Thr in the triple rat V1aR mutant led to loss of interaction with Ile 6.53 . Ile 3 moved downward and had a hydrophobic interaction with Phe 6.52 that is placed between Phe 6.51 and Thr 6.53 (Fig. 7D). DISCUSSION The present study has demonstrated that replacement of three amino acids (Phe 6.51 , Ile 6.53 , and Pro 7.33 ) in rat V1aR with those found in bfVT1R leads to an altered AVP/VT selectivity. Reciprocally substitution of the equivalent residues in bfVT1R by those of rat V1aR (Y6.51F/T7.33P) provoked a remarkable increase in AVP sensitivity. As AVP and VT differ from each other by a single amino acid at position 3 (Phe for AVP and Ile for VT), the Phe 3 residue of AVP is proposed to directly or The plasmids containing the triple rat V1aR (F 6.51 Y/I 6.53 T/P 7.33 T) mutant and the double bfVT1R (Y 6.51 F/T 7.33 P) mutant cDNA were cotransfected with the c-fos-luc reporter vector into CV-1 cells along with ␤-galactosidase as an internal control. After forty-eight hours, cells transfected with the triple rat V1aR (F 6.51 Y/I 6.53 T/P 7.33 T) mutant or the double bfVT1R (Y 6.51 F/T 7.33 P) mutant cDNA were treated with different concentrations of natural ligand, agonists, and antagonists for 6 h, and luciferase activity was determined and compared with the respective wild type receptors. Antagonists were treated in the presence of 10 nM VT. ND, not determined; wt, wild type.
indirectly interact with Phe 6.51 and Pro 7.33 . Thus, it can be postulated that Phe 6.51 of rat V1aR interacts with Phe 3 of AVP probably through an aromatic-aromatic interaction (42). It is of interest to note that the Phe 6.51 residue is also highly conserved in the receptors for biogenic amines and that it plays a key role in interaction with the catechol ring of the ligand (43). Indeed our modeling study indicates that the Phe 3 residue of AVP is involved in hydrophobic interaction with Phe 6.51 of wild type V1aR, while the Ile 3 residue of VT forms a hydrophobic interaction with Ile 6.53 . It has been reported previously that mutation of Phe 6.51 to a non-aromatic amino acid (Leu or Val) in human V1aR does not significantly affect the agonist binding affinity (41). However, the mutation of Phe 6.51 to Tyr in rat V1aR causes a substantial decrease in AVP sensitivity. Thus, the polar nature of the Tyr residue may interfere with the interaction with AVP. Since AVP has a constrained conformation due to the presence of a disulfide bond between Cys 1 and Cys 6 , it is possible that the presence of Tyr at position 6.51 might perturb the optimal positioning of AVP in the binding pocket. In the triple V1aR mutant, substitution of Phe 6.51 with Tyr would lead to a loss in the interaction with Phe 3 of AVP, which may weaken ligand receptor interaction. On the other hand, mutating Ile 6.53 to Thr places Phe 6.52 ahead in the ligand binding crevice, resulting in a potent hydrophobic interaction between Phe 6.52 and Ile 3 of VT. Pro 7.33 seems to facilitate a favorable conformation for AVP docking as the mutation of Tyr 6.51 to Phe alone in bfVT1R was not sufficient to increase AVP sensitivity. A double bfVT1R mutant where Tyr 6.51 and Thr 7.33 were mutated to Phe and Pro, respectively, displayed a marked increase in AVP sensitivity. This suggests that the Pro kink in TMD VII alters the apparent position of TMD VI in the binding pocket, thereby exposing Phe 6.51 in the ligand binding pocket, which becomes more accessible for AVP. It has been demonstrated that activation of rhodopsin results in a rigid body movement of TMDs III and VI (44). The motion of TMD VI is also a feature of the ␤ 2 -adrenergic receptor (45). Thus, it can be postulated that TMD VI of V1aR undergoes displacement following AVP binding, which is affected in the absence of the Pro kink. Such a conformation is also favored by the other natural nonapeptides like VT, OT, and MT as shown by the increased sensitivities of VT, OT, and MT for the bfVT1R double mutant Y6.51F/T7.33P.
In the other nonmammalian V1a-type receptors that also show reduced sensitivity for AVP, Phe 6.51 is highly conserved, although these receptors lack Pro in TMD VII that is otherwise well conserved in the mammalian V1aR. The mammalian V1b, V2, and OT receptors also contain a conserved Pro residue either in ECL3 or TMD VII. Phe 6.51 is highly conserved in all members of the AVP/OT receptor family. Chicken VT1R shows higher sensitivity for AVP than the frog VT1R and fish VT1R that contain a Pro residue in ECL3. These observations also provide evidence that the AVP binding pocket is common in all other members of the AVP/OT receptor family. While Phe 6.51 and Pro 7.33 are important for the AVP selectivity of rat V1aR, Ile 6.53 appears to be involved in the selectivity of VT as the mutation of rat V1aR (Ile 6.53 to Thr) increased the sensitivity to VT, while a single mutation of bfVT1R (Thr 6.53 to Ile) decreased VT sensitivity. It is noteworthy that, compared with the double mutant (Y6.51F/T7.33P), such a decrease in VT sensitivity was not observed in the triple bfVT1R mutant Y6.51F/T6.53I/ T7.33P, indicating that the increase in VT sensitivity by the Y6.51F and T7.33P mutations may compromise the effect of Thr 6.53 to Ile mutation.
The molecular interaction of ligand and receptor was further examined using AVP/OT agonists and antagonists. The triple rat V1aR mutant F6.51Y/I6.53T/P7.33T and the double bfVT1R mutant Y6.51F/T7.33P displayed an altered agonist and antagonist sensitivity compared with the wild type receptors. The triple rat V1aR mutant F6.51Y/6.53T/P7.33T was completely non-responsive to [deamino-Cys 1 ,Val 4 ,D-Arg 8 ]-AVP, although wild type V1aR was activated by [deamino-Cys 1 ,Val 4 , D-Arg 8 ]-AVP (Table IV and Fig. 6C). [d(CH 2 ) 5 1 ,Tyr(Me) 2 ]AVP was less potent in suppressing the triple rat V1aR mutant F6.51Y/6.53T/P7.33T activity than wild type V1aR, while [d(CH 2 ) 5 1 ,Tyr(Me) 2 ,Thr 4 ,Orn 8 ,des-Gly-NH 2 9 ]VT was more potent in suppressing the activity of the mutant receptor (Table  IV and Fig. 6, E and F 5 1 ,Tyr(Me) 2 ]AVP is structurally very similar to AVP except for the first and second residues. Thus, it appears that the Phe 3 residue of [d(CH 2 ) 5 1 ,Tyr(Me) 2 ]AVP is likely to interact with Phe 6.51 of rat V1aR and that the absence of Phe 6.51 in the mutant led to a decrease in the potency of the antagonist in suppressing the mutant receptor activity. Conversely the presence of a Phe residue in place of Tyr 6.51 in the double bfVT1R mutant favors [d(CH 2 ) 5 1 ,Tyr(Me) 2 ]AVP binding, which therefore suppresses the mutant receptor activity more potently than the wild type. These results suggest that different agonists select different spectra of conformations and that different agonists produce different receptor active states (46 -48).
The present study also provides some clues toward receptor-ligand coevolution of the AVP/OT family. Thus, our results indicate that the evolution of the V1a-like receptor might have preceded that of the ligands as the AVP sensitivity is higher in birds (35) than in amphibians (this study). During evolution, V1aR acquired an ability to respond to AVP with a high sensitivity. Change of VT to AVP is also important not only for achieving a high potency to V1aR but also for lesser potency to the OT receptor. It is of interest to note that mammalian OT receptors exhibit a relatively high sensitivity toward VT but low sensitivity to AVP (25). Thus, the change in the peptide sequence from VT to AVP allows circumventing the interference of AVP with OT activity such as milk ejection or uterine contraction in mammals. It should also be considered that the double bfVT1R mutant Y6.51F/T7.33P exhibited a better sensitivity to OT than the wild type, while the triple rat V1aR mutant F6.51Y/I6.53T/P7.33T showed reduced sensitivity to OT compared with the wild type. This finding indicates that evolutionary changes in amino acids at positions 6.51, 6.53, and 7.33 did not contribute to lesser OT selectivity of V1aR.
In conclusion, we have identified three amino acid residues (Phe/Tyr 6.51 , Ile/Thr 6.53 , and Pro/Thr 7.33 ) in the rat V1aR and bfVT1R that may be critical for ligand selectivity for AVP and VT. This study provides important information on the molecular interaction of VT/AVP with their receptors, which should prove useful for the development of novel AVP analogs. Further this study provides a clue to understanding the molecular coevolution of AVP/OT-related peptides and their receptor family.