The design and development of selective agonists acting at the OT (oxytocin)/AVP (vasopressin) receptors has been and continues to be a difficult task because of the great similarity among the different receptor subtypes as well as the high degree of chemical similarity between the active ligands. In recent decades, at least a thousand synthetic peptides have been synthesized and examined for their ability to bind to and activate the different OT/AVP receptors; an effort that has led to the identification of several receptor subtype-selective agonists in the rat. However, owing to species differences between rat and human AVP/OT receptors, these peptides do not exhibit the same selectivities in human receptor assays. Furthermore, the discovery of receptor promiscuity, which is the ability of a single receptor subtype to couple to several different G-proteins, has led to the definition of a completely new class of compounds, referred to here as coupling-selective ligands, which may activate, within a single receptor subtype, only a specific signalling pathway. Finally, the accumulating evidence that GPCRs (G-protein-coupled receptors) do not function as monomers, but as dimers/oligomers, opens up the design of another class of specific ligands, bivalent ligands, in which two agonist and/or antagonist moieties are joined by a spacer of the appropriate length to allow the simultaneous binding at the two subunits within the dimer. The pharmacological properties and selectivity profiles of these bivalent ligands, which remain to be investigated, could lead to highly novel research tools and potential therapeutic agents.
- agonist selectivity
- bivalent ligand
- G-protein-coupled receptor (GPCR)
- therapeutic agent
OT (oxytocin) and AVP (vasopressin) receptors belong to the superfamily of GPCRs (G-protein coupled receptors), which are integral membrane proteins characterized by the presence of seven hydrophobic TM (transmembrane) domains predicted to fold in an α-helical conformation and to be connected by extracellular and intracellular hydrophilic loops [1,2]. The preservation of such a structural architecture among the nearly 2000 GPCRs identified so far clearly indicates that it may serve to accomplish similar functional roles . Data obtained with a variety of different receptors indicate that agonists bind to the extracellular regions and/or to the upper TM core of these receptors, inducing a change in the pattern of the intra-receptor interactions responsible for the organization of the TM domains: this would lead to a re-orientation of the α-helices in the plasma membrane, which in turn would affect the conformation of the intracellular regions, allowing the opening of the docking sites for the G-protein . Several new findings, however, have challenged this classical view, supporting the idea that GPCRs may exist in equilibrium among a number of different molecular conformations, some of which may be associated with inactive and others to active receptor states [4,5]; furthermore, the transition between these different states, regulated by the law of chemical equilibrium described by the ‘allosteric ternary complex’ model , may be differently promoted or stabilized by chemically different ligands. A most recent challenge in molecular pharmacology is the recognition that a single GPCR may activate different signalling pathways by coupling to different G-proteins and/or other signalling intermediates, such as β-arrestins ; this phenomenon is based on the ability of a single GPCR to acquire different ‘active’ conformations, which again may be differently promoted or stabilized by chemically different ligands. Finally, growing experimental evidence indicates that GPCRs do not function as monomers, but as homo/hetero-dimers/oligomers, raising several questions concerning the role of dimerization in GPCR biosynthesis, pharmacology, trafficking and signalling [8–10].
These new insights into the molecular mechanisms of GPCR binding and activation, while posing new challenges to our understanding of how these molecules function at the molecular level, are also opening up new avenues in drug discovery. In particular, the characterization of coupling-selective ligands  and the design of bivalent ligands acting at homo/hetero-dimers/oligomers  represent two very promising strategies that need to be actively explored. To understand fully the new potentiality offered by these analogues, we need to revisit and expand our concept of agonist selectivity, a task that exceeds the purpose of this mini-review, in which we will only try to illustrate a few basic concepts on the molecular basis of agonist selectivity in the OT/AVP receptor family. In particular, we will briefly discuss (i) the state of the art in the development of receptor-subtype-specific analogues, (ii) the development of coupling-selective ligands, and (iii) the strategy to design selective analogues at homo/hetero-dimers.
To deal with the molecular requirements responsible, in both the peptides and receptors, for specific and selective interactions, we need to start by briefly reviewing the data present in the literature on the molecular models of peptide–receptor interactions.
The different AVP and OT receptors cloned so far are classified into three AVP receptors (V1aR, V1bR and V2R) and one OT receptor, OTR, on the basis of their binding and signal transduction properties [13–15]. These receptors are closely related, as their overall similarity varies from 40 to 85%, the most conserved regions being the TM α-helices and the first extracellular loop.
The endogenous ligands for these receptors, represented, in humans and rodents, by OT and AVP, also belong to a family of highly related peptides found in all of the animal kingdom . Structure–function studies using rat bioassays indicate a minimal common requirement for the biological activity of these peptides: a cyclic backbone formed by a disulfide bridge connecting two conserved cysteine residues [17–20].
Owing to this conservation in both peptides and receptors, it is presumable that these peptides interact with the different receptors in a common way. In fact, mutagenesis studies suggest that all these receptors share a common binding pocket situated in the TM region of the receptor [21–23]; furthermore, conserved extracellular residues have also been shown to participate in high-affinity binding and receptor activation [24,25].
A much more difficult task is to deal with the molecular requirements responsible, in both the peptides and receptors, for specific and selective biological and pharmacological activities of different peptides on different receptor subtypes. Substitutions at position 3 within the ring and at position 8 in the C-terminal tripeptide determine the specific receptor interactions of these peptides; generally, a hydrophobic residue is present at position 3, whereas any residue is present at position 8 [17–19]. Structural and molecular modelling studies, while providing very plausible explanations for the observed pharmacological activities for a variety of analogues of OT and AVP (agonists and antagonists) [26–28], have to date failed to provide any really meaningful insights to the design of human or rat receptor-specific AVP and OT agonists or antagonists. Instead, highly selective ligands have been obtained by means of a design strategy which utilized classical structure activity approaches . Some of the most useful receptor-selective ligands obtained with a combined effort between chemist and pharmacologists are listed in Table 1. It should be noted that all of these structurally related analogues lack an absolute receptor-subtype-selectivity, as their receptor-subtype-selectivity is always relative and concentration-dependent.
The OT/AVP receptors are also highly conserved in evolution, as exemplified by the AVT (vasotocin) receptor cloned from a teleost fish, which has 60% identity with the human and rat V1aRs . However, despite this very large receptor similarity, species differences may be responsible for important differences in peptide affinities. Knowing these differences may become crucial when peptidic analogues are used across different species. An example is represented by the ‘selective’ OTR agonist [Thr4,Gly7]OT, which was demonstrated to be highly selective for the rat OTR by in vivo bioassays , a finding confirmed further by binding studies ; as shown in Table 2, [Thr4,Gly7]OT binds with a very good affinity to the rat OTR (Ki=0.8 nM), whereas it has a very low affinity for the rat V1aR (Ki>10000 nM). Furthermore, in the rat, [Thr4,Gly7]OT is more selective than OT, as this peptide binds to the V1aR with a lower Ki (845 nM). However, as also shown in Table 2, this enhanced OTR/V1aR selectivity is lost in the human species, where OT and [Thr4,Gly7]OT bind to the OTR and V1aRs with comparable affinities, indicating that, in humans, the use of [Thr4,Gly7]OT does not present any advantage over OT as far as OTR/V1aR selectivity is concerned.
Another example of selectivity differences across species concerns dDAVP (1-desamino-[D-Arg8]vasopressin) , widely used clinically for the treatment of diabetes insipidus  on the basis of its potent long-lasting and selective antidiuretic activity in rat bioassays . This selectivity was later confirmed in rat receptor binding assays: rat V2R (Ki=0.3 nM), rat V1aR (Ki=100 nM), (Table 3). However, owing to species differences, dDAVP exhibits a much lower affinity for the human V2R (Ki=23.3 nM) and a slightly higher affinity for the human V1aR (Ki=62.4 nM) (Tables 1 and 3). So in human receptor assays, dDAVP exhibits only a very modest selectivity for the V2R with respect to the V1aR. Within the last decade, dDAVP has also been shown to be a full V1bR agonist in humans and a partial V1bR agonist in rats  (Table 3). With a Ki of 5.8 nM for the human V1bR, dDAVP exhibits a significantly higher affinity for this receptor than for the human V2R1 (Ki=23.3 nM) (Table 3) . In fact, dDAVP has been shown to be a more potent V1bR agonist than a V2R agonist in humans . Thus the search for a V2R agonist which exhibits high selectivity in humans remains a challenging goal.
Coupling-specific agonists (or functional agonists)
In classical pharmacological terms, the classification of agonists is based not on their binding affinity, but on their intrinsic efficacy, i.e. their ability, once bound, to produce a biological response. In the case of GPCRs, such cellular responses are mainly achieved by receptor coupling to different G-proteins that activates specific intracellular signalling pathways.
However, in recent years, it is becoming clear that GPCRs are often promiscuous, as a single receptor subtype may couple to more than one G-protein, thus activating, in the same cells, multiple responses at the same time. Furthermore, certain ligands have been shown to possess different intrinsic efficacy on the different signalling pathways activated by the same receptor, a phenomenon referred to as ‘agonist-directed trafficking of receptor stimulus’, ‘biased agonism’, ‘protean agonism’, ‘differential enhancement’, and, in a recent review, ‘functional selectivity’ . The existence of these analogues is consistent with a multistate model of receptor activation in which single ligands can induce specific receptor conformations capable of differentially promoting the coupling of a single receptor to a restricted subset of G-proteins.
This phenomenon has also recently been described in the OT/AVP receptor family, and in particular on the human V2R and the human OTR.
In the human V2R, inverse agonists acting at the receptor can lead to ligand-selective receptor activation, as shown by Azzi et al. . In HEK-293 (human embryonic kidney) cells stably expressing the human V2R, SR121463B, a non-peptide V2R-selective ligand, was shown to induced an inverse agonist response on the adenylate cyclase pathway, but a stimulation of the MAPK (mitogen-activated protein kinase) pathway, a finding indicating the ability of this analogue to differentially activate distinct signalling partners.
Furthermore, functional selectivity was demonstrated on the human OTR. OTRs are functionally coupled to various G-proteins, including Gαq/11, Gαi, Gαh and possibly Gαs, identifying OTR as a promiscuous receptor , and we have recently shown that atosiban, a peptidic OT derivative, is at the same time an OTR antagonist because of its antagonistic effect on OTR–Gαq coupling, and an OTR–Gαi agonist, thus representing the first pharmacologically coupling-selective agonist of the human OTR (Figure 1) .
The relevance of coupling-selective analogues is in their ability to selectively activate only one signalling pathway. This is particularly relevant as, in some cases, the signalling pathways activated by a single GPCR may act synergistically, but they may also give rise to opposite effects on the same cellular function. For example, in the case of the contraction induced in myometrial cells by OTR coupling to Gαq/11 and to the small G-proteins of the Rho family , both pathways co-operate synergistically. In contrast, in HEK-293 cells stably transfected with human OTRs, receptor coupling to Gαi is responsible for inhibiting cell growth, whereas receptor coupling to a different Gα subunit (possibly Gαq) is linked to cell growth stimulation [39,40]. In this system, atosiban, thanks to its ability to selectively promote only Gαi, inhibits cell growth and at the same time is a competitive antagonist at the OTR–Gαq pathway .
Developing coupling-selective analogues represents a new challenge in molecular pharmacology, because they will activate, for any single receptor subtype, only the ‘appropriate’ downstream signalling pathways, representing a class of compounds highly selective at the single receptor subtype level.
Growing experimental evidence indicates that GPCRs do not function as monomers, but as dimers/oligomers, raising several questions regarding the role of dimerization in GPCR biosynthesis, pharmacology, trafficking and signalling [8–10].
In the OT/AVP receptor family, the existence of homo- and hetero-dimers has been reported in in vitro cellular systems [41–45], and their existence in vivo is an important and challenging issue that needs be investigated. In particular, the existence of heterodimers opens up a completely new field in the drug design of analogues acting selectively at the different heterodimers.
This new challenge has started to be explored by using bivalent ligands, in which two agonists and/or antagonists are joined by a spacer of appropriate length to allow the simultaneous binding at the two subunits within the dimer . By joining analogues provided with different receptor selectivity and activity, a completely new class of compounds could be produced. As part of a study on chimaeric peptides, the group of Wheatley synthesized the first bivalent ligands of a linear antagonist for the rat V1aR and V2R . We recently reported the syntheses and some preliminary pharmacological properties of bivalent ligands for the human OTR, V1aR and V1bR . Suberic acid, utilized as reported in , served as the spacer joining ornithine or lysine residues in an OT–V1aR antagonist d(CH2)5[Tyr(Me)2]OVT, and two linear V1aR–OT antagonists, HO-Phaa-D-Tyr(Me)-Phe-Gln-Asn-Arg-Pro-Lys-NH2 (where Phaa is phenylacetyl) and HO-Phaa-D-Tyr(Me)-Phe-Gln-Asn-Arg-Pro-Arg-Lys-NH2. The resulting bivalent ligands exhibit high affinities, in the nanomolar range, for the human OTR and V1aRs expressed in heterologous cell systems . As OTR–V1aR, OTR–V1bR and V1aR–V1bR heterodimers may be postulated to exist in the central nervous system as well as in the peripheral organs and tissues, the pharmacological potentiality of bivalent ligands now needs to be explored. However, the lack of experimental evidence of heterodimers formation in vivo and the difficulty to set up pharmacological assays for the screening and identification of the properties of bivalent ligands, represent a real challenge in the field. In fact, as in the case of bivalent ligand targeting opioid receptors [49,50], bivalent ligands may be characterized by novel features that may lead to the generation of a new class of therapeutics.
Far from being a static field, the molecular pharmacology of GPCRs is a very active and moving area in which even well-established concepts such as the definition of selective agonists are subjected to critical re-evaluation and expansion. From classical receptor-subtype-selective agonists, we are moving towards the development of coupling-selective analogues, molecules capable of exerting selective effects within a single receptor subtype. Furthermore, the existence of receptor homodimers and heterodimers is leading to the design of bivalent analogues which may exhibit completely new selective profiles, as well as unexpected therapeutic actions, that we are eager to explore in the years to come.
This work was supported by a grant from the Italian Association for Cancer Research (AIRC 2006) and Fondazione Cariplo grant number 2004/1419: “New targets for the diagnosis and prevention of human diseases: genomics and proteomics of GPCRs” and by National Institutes of Health (NIH) grant GM-25280. We thank Ms Ann Chlebowski for her assistance in the preparation of this manuscript.
Family Resemblances? Ligand Binding and Activation of Family A and B G-Protein-Coupled Receptors: A Biochemical Society Focused Meeting held at GlaxoSmithKline, Stevenage, U.K., 24–25 April 2007. Organized and Edited by D. Poyner (Aston, U.K.) and M. Wheatley (Birmingham, U.K.).
Abbreviations: AVP, vasopressin; dDAVP, 1-desamino-[D-Arg8]vasopressin; GPCR, G-protein-coupled receptor; HEK-293, human embryonic kidney; OT, oxytocin; OTR, OT receptor; Phaa, phenylacetyl; TM, transmembrane; V1aR (etc.), AVP receptor type 1a (etc.)
- © The Authors Journal compilation © 2007 Biochemical Society