Role of iso-receptors in receptor-receptor interactions with a focus on dopamine iso-receptor complexes

The iso-receptor concept

The definition of iso-receptors was based on the criteria defining isozymes (Hebebrand et al., 1988); however, it can be further detailed by using the following five criteria:

1. At the othosteric binding site, they recognize the same endogenous ligand but sometimes with different affinities.

2. They have similar chemical structures; in the case of GPCRs, they are all seven transmembrane domain receptors.

3. They may trigger different intracellular molecular networks.

4. It is possible to differentiate iso-receptors by means of exogenous ligands.

5. They can be differently assembled to form different multimeric receptor complexes.

In general, an iso-receptor consists of one or more protein subunits and/or domains for each of which structurally slightly different forms can exist. By definition, all iso-receptors bind the same endogenous ligand, and in a biological sense, they could have evolved to enable the organism to react differently to such a ligand. In fact, mutations in some subunits can directly affect the binding site, possibly resulting in affinity changes that often are exploited experimentally for identifying receptor subtypes (Agnati et al., 1980; Rudolph and Knoflach, 2011; Changeux, 2013a).

However, the functional features of a receptor can also be modified by mutations in domains not involved in ligand recognition since GPCR can show changes in the response to allosteric effectors, i.e. to ligands binding to sites different from the othosteric binding site (Kenakin and Miller, 2010).

It is noteworthy that almost for each given endogenous ligand acting through a GPCR, a set of iso-receptors for that GPCR can be identified. Thus, the small structural differences occurring between iso-receptors can trigger differences in ligand recognition and/or in G protein/β-arrestin coupling characteristics, leading to quite different signaling properties of the various receptor forms (Fuxe et al., 1998; see also Shenoy and Lefkowitz, 2011; Alexander et al., 2013). DA-R subtypes represent a significant example of this strategy.

GPCRs as components of molecular networks

Differential expression of GIPs and rapid alterations of GPCR/GIP interaction networks are efficient means to regulate GPCR function in a tissue-specific and spatiotemporal manner to trigger appropriate cellular responses (Daulat et al., 2009).

At least five strictly interconnected molecular networks should be considered in analyzing the recognition/decoding processes carried out by GPCR and their turnover (see Figure 1). Let us give some main aspects of these molecular networks and on their main interconnections (refer to cited literature for a more detailed description Agnati et al., 2005a, 2006a, 2008a, 2010a).

The ligand recognition process by the GPCR takes place according to different steric patterns since the location of the agonist-binding site is highly variable for different receptors. Thus, monoamines and acetylcholine bind within the transmembrane (TM) core. Small peptide hormones bind to the amino terminus, the extracellular loops between TM domains and within the TM core. Large amino terminal domains form the binding site for glutamate, GABA, glycoprotein hormones, ions (the Ca²⁺-sensing receptor). Finally, in protease-activated receptors, the agonist is generated by proteolytic cleavage of the amino terminus of the receptor (Ji et al., 1998).

Also, concerning the interaction between GPCR and their cognate G proteins, multiple sites have been identified. These include ICL 2 between TM3 and TM4, ICL 3 between TM 5 and TM 6, and loop 4 formed between TM 7 and the lipid modification on the proximal carboxyl terminus (Bockaert et al., 2004; Shpakov, 2011). Agonist binding causes steric changes in the arrangement of the TM domains that are allosterically transmitted to the associated G-protein heterotrimer (G_αβγ) that dissociate into Gα and Gβγ subunits. Both dissociated Gα and Gβγ subunits are capable of modulating a variety of effectors, including adenylyl cyclase, phospholipase C, and ion channels. As pointed out above, a GPCR can also signal through several G protein-independent pathways. In particular, β-arrestins and kinases, by interacting with GPCR, can activate intracellular signaling pathways that may involve receptor desensitization, coupling to multiple G proteins, Gα or Gβγ signaling, as well as activation of G protein-independent molecular networks (Chilmonczyk et al., 2014). It has also been shown that different agonists with a preferential triggering of a peculiar molecular network can activate the same GPCR. This phenomenon has been called ‘agonist functional selectivity’ or ‘agonist biased signaling’ (Chilmonczyk et al., 2014). As far as trafficking is concerned, the process is strictly linked to signaling since receptor activation triggers not only signaling but also receptor endocytosis. It should also be noted that GPCRs are synthesized in the endoplasmatic reticulum and can dimerize/oligomerize already at this biosynthesis stage (Bulenger et al., 2005); hence, oligomerization could represent an additional mechanism for proper folding, with each protomer acting as a chaperone for the other via interaction of their hydrophobic surfaces. In addition, signaling proteins can be incorporated to the complex already at this early stage (Dupre and Hebert, 2006) and could function as sorting signals for specific complexes.

This brief overview of GPCR-associated molecular networks underlines once more the importance of proteins in the biochemical recognition/decoding processes of intercellular and intracellular signaling. It is noteworthy that many proteins can self-associate into assemblies of two or more monomers working as ‘switches’ that regulate metabolic reactions (as enzymes) or capable of triggering appropriate cellular responses to signals (as receptor-transducer complexes). However, until the 1980s, it was not surmised that GPCRs could assemble into multimeric complexes, and our early investigations suggested for the first time the possible existence of GPCR hetero-oligomers with emergent properties (Fuxe et al., 1981, 1983a; Agnati et al., 1982, 1984; Fuxe and Agnati, 1985, 1987a). In the opening address of the WGC Symposium, Fuxe and Agnati proposed a broad view of the phenomenon by surmising that GPCR, together with different types of proteins, could form molecular assemblages at plasma membrane level capable of operating as integrative microdevices: “Very likely, we will find out that some sophisticated elaborations of information are performed at the membrane level, via interactions within and among different classes of macromolecules” (Fuxe and Agnati, 1987b).

Based on these assumptions and on new experimental data, the concept of HMN was introduced as a molecular microdevice capable of sophisticated integration of several types of chemico-physical signals impinging on the cell plasma membrane (Agnati et al., 2003, 2010a,b).

A basic biochemical mechanism behind this function is the plasticity of the steric conformation of proteins and their ‘stickiness’ (Pauling, 1953). In fact, protein monomers can aggregate and the subunits of a multimeric protein undergo changes in both their conformation (tertiary structure) and arrangements of inter-subunit contacts (quaternary structure).

These steric changes take place in response to the binding of other molecules and/or under the influence of chemical-physical conditions (i.e. the energy landscape) of the microenvironment in which they are embedded (see ‘The receptor mosaic concept’ section; Agnati et al., 2007a, 2010b; Borroto-Escuela et al., 2011a,b). The biochemical and biophysical processes that are involved in the assembly of GPCRs have been investigated, but the field has not been yet completely clarified (Lambert and Javitch, 2014).

Evidence for the existence of receptor complexes

In the 1980s, the first indication of the possible existence of GPCR-GPCR interactions at plasma membrane level was obtained from binding studies on cell membranes and on brain sections (Agnati et al., 1980, 1983a; Fuxe et al., 1981, 1983a) as well as from in vivo functional studies (Agnati et al., 1983b, 1985; Fuxe et al., 1983b, 1984, 1986).

Later on, more sophisticated approaches, such as FRET, BRET, and PLA, provided further support to the assumption (see Trifilieff et al., 2011; Borroto-Escuela et al., 2013a,b; Guidolin et al., 2015), and a new AFM approach (Agnati et al., 2010a), associated with computer-assisted image analysis of immunocytochemical (ICC) preparations and biochemical experiments, was provided to support the assumption of A2A-D2 physical and functional interactions at plasma membrane level. Thus, it has been proposed that to support the possible existence of receptor multimeric complexes at the plasma membrane level and of direct RRI, at least six conditions should be fulfilled. Some of these conditions are schematically illustrated in Figure 2 and can be briefly mentioned as follows (Agnati et al., 2010a; Fuxe et al., 2012; Guidolin et al., 2015):

1. Colocalization of the GPCR in the same microdomain at the plasma membrane level, representing a necessary condition that can be investigated, e.g. by computer-assisted image analysis of ICC preparations (Agnati et al., 2005c);

2. Physical proximity in a nanometer range of the involved proteins, representing a necessary condition that can be investigated, e.g. by biophysical techniques such as BRET, FRET, PLA, and AFM or by biochemical techniques such as coimmunoprecipitation (Skieterska et al., 2013);

3. Allosteric modulation of the affinity of the orthosteric binding site of GPCR-B following the binding of a ligand to the orthosteric site of GPCR-A. In particular, modulation of the intracellular molecular networks following the GPCR cross-talk at the plasma membrane level (i.e. the RRI) provides a strong support for RRI and can be investigated by means of ad hoc biochemical experiments (Agnati et al., 1982, 2003, 2007b, 2008a; Fuxe et al., 2005, 2007a);

4. Corecognition that can be demonstrated by means of bivalent ligands capable of binding at the same time both partners in a pair of interacting receptors (Bhushan et al., 2004; Daniels et al., 2005; Hiller et al., 2013; Yekkirala et al., 2013). This condition can be investigated by means of binding and pharmacological experiments;

5. Cointernalization of the GPCR. Thus, it can be shown that activation of GPCR-A (e.g. D2-R) leads not only to its internalization (i.e. of D2-R) but also to the internalization of GPCR-B (i.e. of A2A-R). This evidence can be obtained, e.g. by ICC associated with computer-assisted image analysis (Genedani et al., 2005; Agnati et al., 2010a);

6. Blockade of the heteromer formation (see Azdad et al., 2009; Borroto-Escuela et al., 2010a; Hasbi et al., 2014) or disruption of the formed hetero-receptor complex (Kenakin and Miller, 2010; Pei et al., 2010) by means of ligands interacting with amino acid sequences involved in the RRI interface. This experimental approach would benefit from a bioinformatics evaluation of the possible amino acid sequences forming the interaction interface (see ‘Interaction interfaces and RM arrangement: a bioinformatics analysis of DA iso-receptors’ section).

Figure 2:

Experimental evidence indicating the existence of Rec-Rec interactions (RRI) at the plasma membrane level.

For comments and their relevance, see text.

The existence of receptor multimeric complexes (receptor homomers and receptor heteromers) in cellular models is presently not questioned, and the overall available evidence (obtained through multiple approaches with consistent results) strongly supports the relevance of the phenomenon also in native systems (Borroto-Escuela et al., 2013b; Bouvier and Hébert, 2014; Guidolin et al., 2015). Thus, GPCRs can be assembled into multimeric complexes and, via allosteric interactions (RRIs), integrate signals impinging on cell membrane and trigger cellular responses differing from the ones triggered by the activation of the single GPCR. In this respect, already in the 1980s, the possible functional meaning of RRI at synaptic levels was discussed also in view of the coexistence of monoamines and peptides in the same neuron terminals (Agnati et al., 1983c,d; Fuxe et al., 1983a; Hökfelt et al., 1983, 1986, 1987; Fuxe and Agnati, 1987a). Thus, the presence at postsynaptic level of the cognate receptors assembled in a RM could allow the cross-talk between different transmitters such as monoamines and peptides (Agnati et al., 1980, 2010a; Fuxe et al., 1981, 1983a). Furthermore, it was pointed out that RRIs could occur also in perisynaptic and extrasynaptic RM in view of the possible diffusion of chemical signals in the extracellular space via volume transmission (VT), allowing integration of several intercellular signals at the plasma membrane level of far-located target cells (Agnati and Fuxe, 2000; Agnati et al., 2006a, 2010c).

It is obvious that if GPCRs aggregate the topology of the multimeric assembly, the quaternary organization can play a fundamental role on the integrative recognition/decoding processes carried out by the receptor complex. Hence, we have introduced the term RM as a meaningful pictorial analogy illustrating how different assemblies could be obtained from the same set of tesserae (e.g. GPCR) assembled according to different topologies. Each of these different spatial patterns can lead to a peculiar recognition/decoding process of a certain signal impinging on the cell plasma membrane (Agnati et al., 1982, 2005d, 2009a). In agreement with the mosaic analogy, both the tesserae that assemble into a real mosaic as well as the monomers that assemble into a GPCR-RM have to match their tertiary conformations. However, in the case of GPCR (forming a RM), protein plasticity (e.g. protein flexibility) can favor their matching while protein stickiness (e.g. electrostatic interactions) can favor their assemblage (Azdad et al., 2009; Ferré et al., 2009, 2010; Fuxe et al., 2009a,b, 2012; Navarro et al., 2009, 2010; Borroto-Escuela et al., 2011b; Agnati et al., 2013; Tovo-Rodrigues et al., 2014).

As far as the triggering of the intracellular molecular networks is concerned, it should be noted that the RM concept changed the classical model of signal transduction. In fact, the earlier classical model maintained that the successive steps of the recognition/decoding process were organized in a ‘linear’ manner.

According to this model, the ligand interacts with its cognate receptor and the ligand-receptor complex operates as a functional unit capable of triggering specific signals in the cell (e.g. changes in the second-messenger levels) that cause multiple cellular chemical-physical responses.

Thanks to RRI, a possible divergent activation of different cytoplasmic networks can occur already at the cell plasma level. This functional aspect should be considered also in the frame of the data demonstrating that allosteric modulators can alter the conformation of the receptor, hence not only its recognition/decoding processes but also the recognition/decoding processes of other GPCRs of the RM (Agnati et al., 2006a,b; Kenakin and Miller, 2010). It should be noted that allosteric modulators can reach GPCR from the extracellular, membrane, and intracellular environments (Kenakin, 2010).

Thus, the recognition/decoding processes carried out by a GPCR is conditioned by three environments: the extracellular fluid, the plasma membrane, and the intracellular fluid (see Figure 1 and also Kenakin and Miller, 2010) as well as by the RRI with other GPCRs of the RM.

In particular, the following should be mentioned:

1. The ligand (neurotransmitter or hormone) impinges on the interface between the extracellular environment and the plasma membrane and binds the orthosteric site of the receptor;

2. The interactions of the receptor at the plasma membrane/cytoplasm interface with specific cytosolic molecules (e.g. heterotrimeric G proteins, as well as other key cytosolic proteins) (Oldham and Hamm, 2008);

3. The interactions of the receptor in the plane of the lipid bilayer with other GPCR (Borroto-Escuela et al., 2011a; Guidolin et al., 2015) within the RM or with other membrane-associated proteins and/or lipids (Kenakin and Miller, 2010);

4. The possible role of ‘hub’ proteins (see Glossary table) in plasma membrane microdomains. This important phenomenon has been recently proposed for adenylate cyclase (see ‘A possible criterion for RM classification based on homotropic vs. heterotropic cooperativity’ section and Cooper and Tabbasum, 2014);

5. Actions of physicochemical factors characterizing the three environments with which the GPCR is in contact that can affect the energy landscape of the receptor (Deupi and Kobilka, 2010; Venkatakrishnan et al., 2013; see also below).

Concerning the concept of energy landscape, Motlagh and collaborators have pointed out that proteins, hence also GPCRs, sample a spectrum of conformations that form their ‘energy landscape’ (Frauenfelder et al., 1988, 1991; Hilser et al., 1998, 2006; Motlagh et al., 2014). Thus, proteins are dynamically fluctuating macromolecules, constantly taking conformational excursions away from a canonical native structure (Liu et al., 2006; Motlagh et al., 2014). Membrane potential can affect the energy landscape of GPCR since it has been demonstrated that both the activity of GPCR and their affinity toward agonists are regulated also by membrane potential (Mahaut-Smith et al., 2008; Sahlholm et al., 2011; Chaim et al., 2013; Rinne et al., 2013).

To summarize, the following main structural and operational characteristics of RM affect their integrative actions:

1. Composition of the RM as far as the chemical nature of monomers is concerned (Agnati et al., 2005b), in particular, whether the RM is formed by different GPCRs (upper panel, Figure 3) or by only one type of GPCR or by iso-receptors of the same GPCR (lower panels, Figure 3). These different compositions could lead to hetero-topic cooperativity (upper panel, Figure 3) or homotropic cooperativity (lower panels, Figure 3), respectively;

2. Biochemical interfaces between GPCRs may form more or less stable protein-protein interactions and also may be differently suited to allow the transmission of allosteric signals. As illustrated in Figure 4 (upper panel), it can be surmised that the RRIs outside the plasma membrane are possibly plastic contacts, that is they can be formed and/or broken down more easily than the RRIs inside the membrane or the ones that occur via the scaffold proteins (Agnati et al., 2003, 2005d; Ciruela et al., 2005; Franco et al., 2007; Fuxe et al., 2007b). Furthermore, as suggested in a previous paper (Agnati et al., 2010a), the receptor-receptor biochemical interfaces could have special amino acid sequences operating as ‘check-points’ capable of controlling the allosteric interactions between the two interacting GPCRs (Figure 4; see also ‘Relevance of allostery for the GPCR-GPCR interactions’ section);

3. Topology of the single GPCR in the RM that can be linear, and hence, each GPCR is in contact at most with two GPCRs; or not linear, and hence, each GPCR can be in contact with a variable number of partners (see Figure 5). A peculiar topology is the closed arrangement that has been formally analyzed in previous papers as a molecular reverberating circuit, hence for its possible functional implications for learning and memory (Agnati et al., 2007c; Guidolin et al., 2007);

4. Order of activation of the single monomers of the RM since it can affect in a marked way the integrative action of RM especially if a hub receptor can be surmised in the multimeric complex (Agnati et al., 2008a, 2009b).

Figure 3:

Different types of GPCR oligomers that can be present at plasma membrane level.

It is suggested to distinguish at least three different types of RMs: the hetero-RMs formed by GPCR that recognize different neurotransmitters; the homo-RMs formed by only one type of GPCR (obviously recognizing the same neurotransmitter with the same affinity); and the iso-RMs formed by iso-receptors of the same GPCR that recognize the same neurotransmitter but possibly with different affinities. For further details, see text.

Figure 4:

Schematic representation of the allosteric interaction pathways between GPCRs and within a GPCR.

As pointed out also in a previous article (Agnati et al., 2010a), it is possible to surmise the existence of checkpoints along the allosteric interaction pathways. For further details, see text.

Figure 5:

Relevance of the topology in the assembly of GPCR forming RMs.

Some topologies indicate the possible relevance of the ‘symmetry rule,’ hence that a peculiar cooperative phenomenon can be in operation. For further details, see text.

It is worthwhile to note that the last two mentioned operational characteristics allow different integrative actions by RM that are chemically identical.

Relevance of allostery for the GPCR-GPCR interactions

Monod (1979) defined allostery as ‘the second secret of life’ since allostery could be considered important second only to the genetic code (Fenton, 2008), and as mentioned above and discussed in previous papers, the basic phenomenon mediating RRI is allosterism (see Agnati et al., 2005b, 2010a; Kenakin et al., 2010).

Allostery, or a ‘different shape’, according to the classical view, involves coupling of conformational changes between two widely separated binding sites in a protein; hence, allostery allows an extraordinary functional plasticity to proteins, resulting as a crucial mechanism for living cells. Allosterism is a term nowadays used with broader meanings with respect the one proposed by Monod and coworkers about 50 years ago (Monod, 1979; for a historical and theoretical discussion, see Changeux, 2013a,b).

According to the recent views (see Kenakin and Miller, 2010; Nussinov, 2013; Nussinov and Tsai, 2013), proteins subject to allosteric communication can be monomers or components of large, hetero-oligomeric complexes and allosteric behavior can be even displayed by single domains within a protein (see also Agnati et al., 2007b, 2008b). Since allostery involves the propagation of signals between distant sites in a protein structure (Changeux and Edelstein, 2005), many experimental studies and theoretical models suggest that there is a network of physically interconnected and/or thermodynamically linked residues forming channels along which signals are communicated between binding sites (Goodey and Benkovic, 2008).

Thus, allostery has been proposed as a universal phenomenon whereby a perturbation by an effector at one site of the molecule leads to a functional change at another through alteration of shape and/or dynamics (Nussinov and Tsai, 2013).

As Motlagh et al. (2014) point out, the phenomenon of allosterism is related to the conformational steric plasticity of proteins. Actually, all possible conformations of a protein and, hence, also of GPCR are sampled (or populated) according to their energies. This means that lower energy conformations are sampled more often than those of higher energy are. Because the binding of any ligand increases how often that state is sampled, allosteric ligands and the medium effectively remodel the energy landscape of allosteric proteins such as GPCR. Thus, an important point critical to typifying an allosteric protein ensemble, hence also a RM, is the assessment of the small fraction of the conformations of the ensemble that are stabilized by a certain ligand and/or by some chemico-physical characteristics of the medium. In fact, such a stabilization will cause a redistribution of the probabilities of every state of the ensemble (Motlagh et al., 2014).

According to such a view, every state of a RM can have potentially different affinities for the orthosteric ligands and bind different allosteric modulators; hence, it may trigger different cytoplasmic molecular networks.

Let us briefly discuss some aspects of the allosteric interactions in a RM and the possible role of cooperativity for RRI (Agnati et al., 2005b, 2010a).

Allostery occurs also within a single GPCR (as shown in the lower right panel of Figure 4; Agnati et al., 2010a); hence, the complexity of the integrative actions carried out by a RM depends on allosteric interactions within and between the GPCRs of the RM. Furthermore, GIPs, by a fine-tuning of receptor steric conformations, likely affect their predisposition to form heteromers as well as their aptitude of showing peculiar allosteric interactions. It should be also pointed out that GPCR assemblage can occur by different biochemical mechanisms that likely have also different functional impacts on the integrative action of the RM (Tovo-Rodrigues et al., 2014).

An overview is given in the upper panel of Figure 4 that should be analyzed in the frame of the articles cited by recent reviews (see Kenakin and Miller, 2010; Borroto-Escuela et al., 2014; Ferré et al., 2014; Guidolin et al., 2015) and of the experimental work of Woods and coworkers on the relevance of electrostatic interactions (Ciruela et al., 2004; Woods et al., 2005) to generate receptor complexes and on the role of certain kinases (as for instance adenylate cyclase) to impart specificity to the interactions (Woods and Jackson, 2013). It is also noteworthy that the noncovalent complexes so formed appear more stable than covalent bonds (Jackson et al., 2005, 2006). Once the receptor assemblage is formed (and this is not a completely solved problem), the relevant aspect is how the allosteric interactions among GPCRs in the multimeric complex occurs.

A peculiar aspect of allosteric interactions is the cooperativity phenomenon that can be observed in macromolecule assemblages when the number of binding sites or their affinity for a ligand is a nonlinear function of the ligand concentration. Thus, the phenomenon of cooperativity is displayed by molecular systems involving elements, which act nonindependently of each other since the binding of one ligand molecule to one site either increases (positive cooperativity) or decreases (negative cooperativity) the affinity of the other site(s) for a ligand. Such effects can be either homotropic, i.e. one ligand species affects the binding of the same ligand species, or heterotropic, i.e. one ligand species affects the binding of a different ligand species.

In the present article, emphasis will be given to the former since homotropic cooperative phenomenon is likely in operation in RM formed by both homoreceptors and iso-receptors.

Thus, as proposed in a previous paper, cooperativity can be a criterion to classify different types of RM that can display different integrative capabilities (Agnati et al., 2005b).

A possible criterion for RM classification based on homotropic vs. heterotropic cooperativity

Since allostery is a basic phenomenon in RRI, cooperativity plays an important role in the RM integrative actions. On this basis, it has been proposed that the two types of cooperativity, namely homotropic and heterotropic cooperativity, could allow distinguishing two types of functionally different RM (see Agnati et al., 2005b):

1. RM-type 1 (RM1): The receptors of the multimeric complex recognize/decode the same chemical signal (Nt). Thus, it is suggested to distinguish the homo-RMs (i.e. formed by one type of iso-receptor) from the iso-RMs (i.e. formed by different iso-receptors of the same GPCR) (see lower panels of Figure 3). A well-studied case is that of RM formed by the different subtypes of DA-R (Nimchinsky et al., 1997; Strange, 2005; Rashid et al., 2007; Fiorentini et al., 2008; Guitart et al., 2014; Fuxe et al., 2015). Since the monomers of these RM recognize the same Nt homotropic cooperativity is possible.

2. RM-type 2 (RM2): These are hetero-RMs formed by different types of receptors. Homotropic cooperativity is not possible, but allosteric interactions can occur, causing a change in the recognition/decoding processes of the monomers (see upper panel of Figure 3). Thus, it has been shown that depending on which partner is costimulated, the affinity of the D2 receptor for dopamine is either reduced (as in the adenosine A2A-D2 hetero-oligomer; Fuxe et al., 2003, 2014) or enhanced (as in the somatostatin SST5-D2 hetero-oligomer; Rocheville et al., 2000; Agnati et al., 2005b).

It can be surmised that if DA binding to a RM1 is considered, the iso-receptor with the highest affinity for DA (hub receptor of the RM1) regulates the affinity of the other iso-receptors of the RM1 positively or negatively in the case of positive or negative homotropic cooperativity, respectively.

Thus, in the presence of a low concentration of the ligand (e.g. DA), the positive cooperativity favors the activation also of the other iso-receptors of the RM1, while the negative cooperativity leads to the preferential activation of the hub receptor. In the presence of high concentrations of the ligand, negative cooperativity avoids the saturation of the no-hub receptors.

In particular, it has been proposed that cooperativity could play a role in the differentially tuning receptor sensitivity of intrasynaptic vs. extrasynaptic RM (Agnati et al., 2005b). Hence, we have hypothesized that negative cooperativity between receptors in RM1 is a mechanism important at the synaptic level where the transmitter concentrations are high (in the order of millimolar concentrations), whereas positive cooperativity between receptors in RM1 is a mechanism important at the extrasynaptic level since the concentration of the chemical signal is usually in the nanomolar range (Vizi, 2000).

In the case of RM1 formed by the same iso-receptor (i.e. a homomer RM1), it can be surmised that the topological arrangement (see Figure 5) and the possible allosteric effects of signals originated in one of the three environments (i.e. extracellular, intramembrane, or intracellular) can govern which iso-receptor will operate as a hub in RM1. Furthermore, it can be surmised that something like a ‘symmetry rule’ (Ackers et al., 1992) can be in operation in the allosteric interactions of RM formed by iso-receptors arranged according to a peculiar topology (see Agnati et al., 2005e, and lower panel of Figure 5).

In the case of hemoglobin, the symmetry rule states that a quaternary switching from the tense (T) form (i.e. the ‘deoxy’, low-affinity state) to the relaxed (R) form (i.e. ‘oxy’, high-affinity state) takes place whenever haem-site binding creates a tetramer with at least one ligated subunit on each dimeric half-molecule.

Although it is obvious that the symmetry rule has been proposed for hemoglobin and might not have a general validity for every tetrahomomeric protein, it clearly indicates that RM1s might have some, until now not deeply investigated, biochemical characteristics that can affect their integrative functions. It is therefore possible to surmise that especially the topological arrangements of iso-receptors indicated in the lower panel of Figure 5 can show allostery according to some sort of a symmetry rule. Such can be the case of DA-R tetramers that may mirror the hemoglobin structural organization. At least for D2-receptors, the assumption that dimers can be assembled into tetramers is supported by experimental observations (Zawarynski et al., 1998).

Actually, it has been shown that D2-receptor oligomerization is required for D2-receptor trafficking to the cell-surface; hence, dimers might represent the ‘building blocks’ of high-order oligomers (Lee et al., 2003) and D2-receptor tetramers may be possible. Previous data have indicated that also D3-receptor tetramers seem to be the dominant form in the cerebral cortex (Nimchinsky et al., 1997). Thus, if the restrictions imposed by the symmetry rule are in operation, RM formed by D2 or D3 tetramers could detect not only dopamine binding but also whether it has occurred on the same dimer or on the two dimers that form the tetramer via the absence or presence of the cooperative behavior as dictated by the symmetry rule (Agnati et al., 2005b). As discussed below (‘Biochemical and functional aspects of RM formed by DA iso-receptors’ section), it has been demonstrated that also D1-D3 is a tetrameric heteromer, constituted by two interacting D1R and D3R homodimers coupled to Gs and Gi proteins, respectively (Guitart et al., 2014).

Based on these data, it would be interesting to analyze the peculiar integrative capability of RM1 tetramers maybe having as starting assumption the symmetry rule.

Interaction interfaces and RM arrangement: a bioinformatics analysis of DA iso-receptors

As mentioned above, the possibility that iso-receptors could assemble according to different topologies, leading to different receptor complexes, dramatically increases the spectrum of recognition/decoding processes available to the cells even when the chemical signal impinging at plasma membrane level is the same. Bioinformatics methods represent a key tool to investigate this topic, since they allow a wide theoretical exploration useful to orient the experimental work. Two aspects, in particular, can be well addressed by bioinformatics analysis:

1. The first aspect concerns the identification of the possible interfaces that a receptor protein can exploit to interact with a receptor partner. The nature of the interface is important, since it influences the allosteric processes and, consequently, the functional features of the resulting complex. In this respect, the evidence that intrinsically disordered domains play a significant role in protein-protein interaction (Hilser and Thompson, 2007) was particularly intriguing, and the analysis of intrinsic disorder is a presently followed strategy to identify GPCR regions potentially linked to RRIs (Agnati et al., 2008b; Tovo-Rodrigues et al., 2014). The analysis of disordered domains, however, does not help in the prediction of putative interaction sites located in the TM helices and other methods are needed to estimate favorable contact surfaces on the TM helices of GPCR (see Nemoto and Toh, 2005).

2. The second aspect concerns the estimate of the possible geometrical arrangements the receptor complex can assume. This feature is a consequence of a number of conditions, including not only the physical properties of the interacting proteins (i.e. the available interaction interfaces) but also the microenvironment surrounding the interacting partners (i.e. the energy landscape; Frauenfelder et al., 1991), and can deeply influence the functional behavior of the RM (Agnati et al., 2007b).

An illustrative analysis focused on the dopamine iso-receptors is described in the Appendix. The interaction interfaces identified on these class A GPCR are reported in Table 1. As shown, each receptor can theoretically exploit multiple interaction interfaces located in both the TM helices and the intracellular domains. As far as the TM domains are concerned, TM5 and TM6 were predicted as possible interaction sites in all the DA-Rs. This result is consistent with available experimental data indicating these domains as the most frequently observed interfaces between class A GPCR heteromers (Guo et al., 2008; Borroto-Escuela et al., 2010a). Alternative interaction interfaces at the level of the TM4 domain were predicted for the D1-like family of DA-Rs. This result is also consistent with experimental data reporting the involvement of TM4 helix in a variety of interactions between GPCR (Fotiadis et al., 2003; Wu et al., 2012). Concerning the intracellular domains, ICL 3 was a predicted site of interaction in all the DA-Rs, while the C-terminal tail resulted as an additional possible interface in the D1-like family of DA-Rs. These results show consistency with the experimental results by Woods et al. (2005) demonstrating that strong electrostatic interactions can occur between intracellular domains (namely ICL3 and C-terminus) that are linked to the presence of a negatively charged serine-phosphate-containing intracellular motif of one receptor and a positively charged arginine-rich motif of a second receptor.

A major consequence of the possible existence of multiple interaction interfaces is that the assemblage of receptor molecules to form a RM could occur in a number of different geometrical arrangements. Bioinformatics predictions support this view. As illustrated in the Appendix, two types of arrangements can be predicted for the D1-D2 dimer: one involving an interaction between TM domains of the partners, and the second based on an interaction at the level of the intracellular domains. As discussed in the section that follows, the latter is consistent with recent experimental data (Hasbi et al., 2014).

Biochemical and functional aspects of RM formed by DA iso-receptors

In principle, all the DA iso-receptors may be involved in RM of the type RM1 that have different localizations and differential mechanisms to activate specific intracellular molecular pathways. Thus, as discussed in the ‘Possible speculations on the functional relevance of iso-receptors’ section, a complex interplay among the DA iso-receptors in integrating and mediating critical brain functions can take place.

Although these investigations are still at their beginning, important data are already available. In fact, in vitro and in vivo studies exist (see Fuxe et al., 2015, for a recent review) demonstrating that the following DA iso-receptor complexes are possible (Borroto-Escuela et al., 2014) and capable of ‘emergent’ recognition/decoding properties:

1. D1-D2 (Lee et al., 2004);

2. D1-D3 (Marcellino et al., 2008);

3. D2-D4 (Borroto-Escuela et al., 2011b; Gonzàles et al., 2012);

4. D2-D3 (Scarselli et al., 2001); and

5. D2-D5 (So et al., 2009).

Thus, DA iso-receptors can form novel pharmacological entities through the formation of receptor heteromeric complexes by oligomerization not only with other GPCR but also with other DA-iso-receptors (Chun et al., 2013; Hasbi et al., 2014).

Let us consider the first two DA homoisomers (according to the nomenclature of Figure 3) that have been investigated with detailed studies and have important pharmacological implications:

D1-D2 iso-RM: Dopamine D1R and D2R receptors are the most abundant dopaminergic receptors in the striatum, and the presence of D1-D2 oligomers with unique functional properties was first shown in transfected cells using different methods (Rashid et al., 2007; Hasbi et al., 2014).

Thus, a clear segregation between the pathways expressing these two receptors has been reported in certain subregions, and in addition, the presence of D1-D2 oligomers within a unique subset of neurons, forming a novel signaling transducing functional entity, has been shown (Hasbi et al., 2011). Furthermore, for the dopamine control of motor behavior or the dopamine-mediated reward, a concomitant stimulation of D1R and D2R receptor by means of specific drugs potentiates the effect exerted by each other even if the same drugs are ineffective when administered separately (Dziedzicka-Wasylewska, 2004).

This D1/D2 synergism demonstrates an obligatory participation of both receptors, and it can be based on the activation of dimers D1-D2. Recently, the amino acid sequence from the D1R C tail involved in forming an interaction interface with D2R has been identified (Hasbi et al., 2014). Thus, it was determined that D1R receptor carboxyl tail residues 404Glu and 405Glu were critical in mediating the interaction with the D2R receptor. Furthermore, a peptide capable of disrupting the D1-D2 oligomer, switching its G-protein coupling, abolishing its signaling, and functioning as a selective antagonist in vitro and in vivo has been synthesized.

The use of this peptide corresponding to a sequence containing these two aa and the six flanking residues on either side of this site not only blocked the interaction of the D1R and D2R shown by BRET analysis and coimmunoprecipitation studies but also blocked the calcium signal activated by the oligomer, usually seen as a fingerprint of D1-D2 oligomer activation. Furthermore, the treatment of cells expressing the D1-D2 oligomer with the D1 peptide switched the preference of D1R interaction from Gq in a D1-D2 oligomer context to a preference for interaction with Gs, usually seen in a D1R homomer context. As pointed out by Hasbi (Hasbi et al., 2011; O’Dowd et al., 2012), this D1-D2 receptor oligomer-calcium signal may represent a first common biochemical bridge between the dopaminergic system-CaMKII-BDNF, synaptic plasticity, and the occurrence of drug addiction and schizophrenia.

It should be noted that such a peptide may represent a novel pharmacological tool to elucidate the functional and behavioral consequences of D1-D2 oligomer activity, and, in a broader view, it gives further support to GPCR oligomers as new targets for drugs (Agnati et al., 2011; Guidolin et al., 2015).

D1-D3 iso-RM: Ferré and collaborators have demonstrated that also D1-D3 is a tetrameric oligomer, constituted by two dimers, each formed by interacting D1R and D3R protomers coupled to Gs and Gi proteins, respectively (Guitart et al., 2014). Coactivation of both receptors causes the canonical negative interaction at the level of adenylyl cyclase signaling, a strong recruitment of β-arrestin-1, and a positive cross-talk of D1R and D3R agonists at the level of mitogen-activated protein kinase (MAPK) signaling. Furthermore, D1R or D3R antagonists counteracted β-arrestin-1 recruitment and MAPK activation induced by D3R and D1R agonists, respectively (cross-antagonism). These investigations demonstrate the functional selectivity of allosteric modulations within the D1-D3 oligomer, which can be involved with the reported behavioral synergism of D1R and D3R agonists.

Thus, the proper modulatory actions via suitable selective drugs of this oligomer could be of importance for neuropsychiatric diseases and in drug addiction.

Analogies from the information theory

As mentioned above, communication among cells takes place by means of signals capable of being recognized/decoded by the target cells. These signals activate intracellular molecular cascades to produce the appropriate cell responses. In the information theory, ‘entropy’ is a measure of the information content of an event based on the unpredictability of its outcome. It will be high when there is no a priori way to predict the result of the event.

By applying this concept to living systems, we could say that the expression of multiple iso-receptors for a given chemical signal by cells and tissues is a way to increase the entropy of that signal, i.e. its information content, since it can provide a pattern of possible results after being recognized and decoded by the target cell.

Therefore, the presence of iso-receptors and their organization within RM allow the use of a reduced number of signals, since the target cells can recognize and decode them in different ways. This means that the target cells can produce appropriate different responses notwithstanding that they have been reached by the same triggering signals.

An analogy with some electronic apparatuses could be exploited to describe what has been stated above. In particular, the use of the ‘multiplexer’ and ‘demultiplexer’ concepts (see Glossary table for a definition) to model the iso-receptors role in information handling by the cell may be suggested.

Thus, two ‘biochemical devices’ can be briefly discussed as it follows:

• Multiplexer. Actually, a RM can operate as a multiplexer when an allosteric modulator can act like a ‘selector’ by modifying the allosteric interactions between the partner protomers. A known example is homocysteine, which can disrupt the A2A-D2 allosteric interactions by interacting with a check-point controlling the allosteric pathway between A2A and D2 in the RM (see Agnati et al., 2010a).

• Demultiplexer. A demultiplexer is a logic circuit allowing the selection of an output (among a set of available possibilities) based not simply on the features of the input signal but also on which iso-receptor is activated and the RM to which it belongs as well as all factors mentioned in Figure 1.

Possible speculations on the functional relevance of iso-receptors

The triggering of different intracellular molecular networks by the same neurotransmitter, thanks to its binding to different iso-receptors, has a great functional relevance since it allows the following:

1. Redundancy, i.e. alterations in a biochemical network in some cells do not prevent the proper integrative actions of the cellular network since it may be compensated by the activation by the same transmitter of other molecular networks that may carry out similar tasks.

2. A more complex pattern of cellular responses since several intracellular molecular networks are activated by the same neurotransmitter and can either work in parallel (redundancy) or interact with each other (integration) or independently but in a complementary fashion (complex patterns).

Against this background, we would like to introduce the concept that the extraordinary capability of information handling by CCN in the CNS is achieved not so much by an increase in the neurotransmitter numbers but thanks to iso-receptors, RM, and plasma membrane microdomains. These factors lead to an astonishing increase in the modes for the recognition/decoding processes that control the cellular biochemical machineries.

The special role for such a phenomenon of iso-receptors can be summarized as follows:

1. Iso-receptors can be coupled to different intracellular molecular networks.

2. Iso-receptors can selectively interact with other receptors in the membrane, and these RRIs can modify either the recognition (e.g. affinity) and/or the decoding processes.

3. Iso-receptors can selectively interact with other membrane molecules (e.g. lipids of the LRs, integrins, or tetraspanins; Chini and Parenti, 2004; Little et al., 2004; Agnati et al., 2009c) and/or extracellular molecules (e.g. tenascin; Dityatev and Schachner, 2006) and/or intracellular molecules (e.g. β-arrestins; Reiter and Lefkowitz, 2006).

Just to give an idea of the potential possibility of integrative actions carried out by RM thanks to the iso-receptors, let us examine the possible number of assembling of iso-receptors for 5HT considering only RM1 formed by two or four iso-receptors for such neurotransmitter. This means that not all of the possible oligomers formed by 5HT with other GPCRs are taken into account (González-Maeso et al., 2008; Borroto-Escuela et al., 2010b; Łukasiewicz et al., 2010). It has been suggested that 5HT can be recognized and decoded by at least 20 different iso-receptors (Pytliak et al., 2011). This means that, at least in principle, 400 iso-receptor dimers and 160 000 iso-receptor tetramers could be obtained. If the possible different topological arrangements and the order of activation of the partner protomers are also taken into account, the number of theoretically possible RM1s hugely increases.

This, of course, is just a theoretical extrapolation, since each cell expresses only a limited number of iso-receptor types. However, it underlines the potential capabilities of RM as recognition/decoding devices.

As far as the possible increase in the multiplicity of triggering the intracytoplasmic molecular networks, a very important finding is the biochemical and functional characterizations of the D1-D2 and D1-D3 heteromers mentioned above since they could be a potential therapeutic targets for neuropsychiatric disorders (Guitart et al., 2014; Hasbi et al., 2014).