G‐Protein‐Coupled Receptor Heteromers as Key Players in the Molecular Architecture of the Central Nervous System

Summary

The overall architecture of the nervous system, especially the CNS, is remarkable. The anatomy of the nervous system is constituted not only by macroscopic and microscopy identifiable regions and neuronal cell types, but also by protein complexes whose identification and localization require sophisticated techniques. G‐protein‐coupled receptors (GPCRs) constitute an example of proteins that are the key factors in the framework needed to sustain brain and nerve structure and function. The versatility underlying nervous system anatomy takes advantage of a recently discovered feature of GPCRs, the possibility to form heteromers that, placed at specific neuronal subsets and at specific locations (pre‐, post‐, or peri‐synaptic), contribute to attain unique neural functions.

Introduction

G‐protein‐coupled receptors (GPCR) constitute a very populated protein family and the most versatile group of cell surface proteins involved in signal transduction. In humans, more than 1% of the genome codes for about 500 of these type of proteins, 90% of which are expressed in the central nervous system (CNS) 1, 2. GPCRs have seven transmembrane domains where the N‐terminus is extracellular, whereas the C‐terminus is facing the cytosolic side. GPCRs can be activated by a variety of endogenously active substances (first messenger and/or neurotransmitter) including, but not limited to, amino acids, peptides, proteins, lipids, and purine nucleosides and nucleotides. The perception of environmental stimuli (light, taste, and smell) is also mediated by GPCRs. Activation of the receptors leads to signal transduction mediated by four main heterotrimeric G‐protein subclasses (G_s, G_i, G_q, and G_12/13) and results in the regulation of all kinds of cellular events. Receptor engagement leads to changes in the levels of intracellular second messengers (mainly cAMP and Ca²⁺) and, often, involves Ser/Thr protein kinases and phosphatases 3.

Although it was for long suspected that GPCRs could form homodimers, heterodimers, or higher‐order oligomers, technical problems prevented demonstration of such oligomerization until the late 90s. In fact, an ever increasing number of reports using a variety of techniques have shown that GPCRs do not usually behave as independent functional units but forming oligomers between themselves (homomers) or with other receptors (heteromers) 1, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14. Whereas homomers do not introduce new physiological concepts or challenges, heteromerization results in novel functional units and not a mere addition of the signals of the receptors composing the heteromer. Roles for GPCR heteromers have been proposed starting with the quality control of receptor folding and cell surface targeting of newly synthesized receptors 15, 16, 17. Heteromers also underlie pharmacological diversity as the ligand binding to one receptor may influence the binding of a hormone/neurotransmitter to the second receptor; the affinity of the second binding event may be higher or lower 18, 19. From a pharmacological point of view, the signaling properties of a given ligand may also affect the selectivity of interaction between the GPCR and its G‐protein resulting in various possibilities for G‐protein coupling 20. Internalization is another event that may be modified by heterodimerization 21, 22.

Different GPCR Heteromers in Different Neural and Neuronal Locations

One of the most expressed GPCRs in the CNS is the cannabinoid receptor 1 (CB₁ receptor). CB₁ receptor has been extensively studied, and from the very beginning, it was considered a presynaptic receptor 23, 24. The physiological substrate is unique as the endocannabinoid‐producing enzymes are located in the postsynaptic neuron, whereas endocannabinoids act on presynaptic CB₁ receptors 25. Recent data show, however, that exceptions may occur. On the one hand, a postsynaptic location for the second cannabinoid receptor subtype (CB₂) has been reported in hippocampus 26. Concerning this second cannabinoid receptor, it is interesting to note that until pretty recent work in a few laboratories 27, 28, it was thought that the CB₂ receptor was not expressed in neurons. On the other hand, Dr. Lanciego's laboratory has just identified postsynaptic CB₁ receptors forming heteromers with CB₂ receptors in neurons of the primate globus pallidus (data submitted). Following identification of postsynaptic CB₁–CB₂ receptor heteromers, one may wonder why these receptors are in a small subset of neurons in such location. It is anticipated that CB₁–CB₂ receptor heteromers are performing there a specific task that neither CB₁ nor CB₂ receptors could do. More importantly, the neuronal mechanisms involved in placing CB₁ and CB₂ receptors and CB₁–CB₂ receptor heteromers pre‐ or postsynaptically may be instrumental to know the intricacies that underlie the anatomy of the nervous system at the molecular level.

Adenosine is a regulatory molecule present in every cell and tissue. In the course of evolution, it was probably one of the first hormones and of the first neuromodulators. Its role in the nervous system is very important and is mediated by the three adenosine receptors that are present in the CNS (A₁, A_2A, and A_2B); the fourth receptor does not seem to be highly expressed in neurons. The adenosine A_2A receptor is found in a variety of locations (central and peripheral), but it is heavily expressed in the striatum 29. Whereas the A₁ receptor when expressed in neurons was considered to be presynaptic, A_2A has been mainly considered a postsynaptic receptor. It has been demonstrated that A₁ may regulate neurotransmitter release, whereas it has been also shown that the A_2A receptor may regulate the action of a given neurotransmitter in the postsynaptic neuron. However, it turns out that both A₁ and A_2A may be pre‐ or postsynaptic depending on the neural region and neuronal subtype. In the case of the A_2A receptor, there is a careful immunohistochemical study by Rosin et al. 29, which shows that—in the rat—“at least some of these central adenosine A_2A receptors are found presynaptically.” This finding was further substantiated at the ultrastructural level 30. What determines the specific location of an adenosine receptor in a specific neuron and for what reason? The answer, which is to achieve a differential effect by a given neuromodulator, requires further players that seem to be required for a correct architecture within the final—and complex—nervous system anatomical scenario. It then seems that neurons place the A₁ receptor presynaptically and the A_2A receptor postsynaptically by default, but that there are neuronal mechanisms that may interfere with the default program to attain certain goals. As detailed below, the two receptors are brought together to the nerve terminal to build up a very unique module of neurotransmitter release control.

Differential GPCR Distribution in the Cortico‐Striatal Pathways

GPCR heterocomplexes play a crucial role in the integration of information in the striatum, where medium spiny neurons (MSNs) constitute the 95% of the population. These neurons are inhibitory as they use GABA as neurotransmitter. The two main inputs are as follows: glutamatergic afferents from cortical neurons and thalamus and dopaminergic afferents from substantia nigra pars compacta (SNc) 31.

MSNs may be divided into two subtypes depending on the region to which they project and according to the expression of peptide and neurotransmitter receptors. Direct pathway MSNs express dynorphin, substance P, dopamine D₁ receptors (coupled to G_s), and adenosine A₁ receptor (coupled to G_i). G_s leads to increases in cAMP levels, whereas G_i leads to the opposite. These striatonigral neurons project directly to the globus pallidus (pars interna) and to the substantia nigra. Additionally, cortical glutamatergic boutons arriving to striatonigral MSNs express adenosine A_2A receptors presynaptically forming, at least in part, heteromers with adenosine A₁ receptors (see section Heteromers located presynaptically). Stimulation of the direct pathway results in motor activation. Indirect pathway MSNs express enkephalin, dopamine G_i‐coupled D₂ receptors, and G_s‐coupled adenosine A_2A receptors; they are also known as striatopallidal MSNs and project to globus pallidus (pars externa). Stimulation of indirect pathway MSNs results in motor inhibition 32, 33, 34, 35, 36. A balance between the direct and the indirect pathway is required for proper motor control; GPCRs and GPCR heteromers directly participate in achieving this balance.

Neurodegenerative Diseases Impacting on Basal Ganglia Function

A main input of basal ganglia in terms of motor control is the globus pallidus where direct and indirect pathways converge. These GABAergic inhibitory neurons project to the thalamus which projects back to the cortex via glutamatergic efferents. An adequate equilibrium between both pathways produces normal/controlled movement. Dopamine, which is mainly produced by neurons from SNc, is a key regulator for correct functioning of basal ganglia; it induces motor activation via activation of dopamine D₁ receptor in the direct pathway and inhibition of neurotransmission via activation of dopamine D₂ receptors in the indirect pathway. Parkinson's disease patients, whose nigral neurons degenerate, experience movement depression or hypokinesia due to depletion of nigral dopamine. On the other hand, in Huntington's disease patients, hyperkinetic chorea movements are due to a gradual depression of the indirect inhibitory pathway 37.

Nowadays, the basal ganglia circuitry is being revisited to consider it not as a simple anatomical substrate where the connectivity and functional interactions occur unidirectionally, but as a complex structure with microcircuits sustained and reinforced by reciprocal innervations 38. In those circuits GPCR and GPCR heteromers play a crucial role; it should be noted that two adjacent but morphologically indistinguishable neurons might respond differently to dopamine as each of them may contain a different dopamine‐containing heteromer (A₁–D₁ in the direct or A_2A–D₂ in the indirect pathway, shown in Figure 1). It has been recently identified the occurrence of heteromers in models of Parkinson's disease 39, meaning that current therapies for the disease (L‐DOPA mainly) are targeting dopamine receptor‐containing heteromers. This fact opens new concepts in the way pharmaceutical companies select targets to combat neurological/neurodegenerative/neuropsychiatric diseases. Interestingly, L‐DOPA treatment leads to disruption of A_2A–D₂ receptor heteromers and this correlates with dyskinesia 39, 40. Whether disruption of heteromers may be in full or in part cause of dyskinesias merits further experimental effort.

Figure 1.

Differential receptor distribution in striatal synapses. Reported localization of different receptor homodimers and heteromers and their inter‐receptor interactions, negative modulation in red lines, and positive modulation in blue arrows. Receptor oligomers localized presynaptically control glutamate release inhibiting (in red) or enhancing it (in blue). Postsynaptic oligomers modulate adenylate cyclase inhibiting cAMP production (in red) or enhancing it (in blue). Further studies are required to elucidate the localization of other receptor homodimers or heteromers and their specific signaling pathways.

Heteromers Located Presynaptically

To our knowledge, O'Kane and Stone 41 were among the first to report a functional interaction—in hippocampus in vitro—between adenosine A₁ and A_2A receptors. The interaction was deciphered using synthetic compounds due to the fact that the natural agonist for the two receptors is adenosine. In other words, it is strange that two receptors for the same agonist are establishing a kind of competition. Even more strange is to find cells expressing two receptors, one coupled to G_s and another to G_i, for the same hormone/neurotransmitter. Expression of two different receptors for the same neurotransmitter is not exclusive to adenosine but occur for a variety of neurotransmitters (a detailed description is out of the scope of the present article).

As it was not expected that two receptors for adenosine were physically interacting, the A₁–A_2A receptor heteromer was first assayed as a negative control. The results were, however, contrary to the expectations, and therefore, the A₁ and A_2A receptor heteromer was detected—in transfected cells—by Ciruela et al. 42. The existence of an intramembrane (inter‐receptor) interaction in the A₁–A_2A receptor heteromer results in various actions that require the heteromer and cannot be obtained by individual receptors. For instance, stimulation of adenosine A_2A receptor decreases adenosine A₁ receptor affinity for agonist binding. But this would not explain why the two receptors (and the A₁–A_2A receptor heteromer) are expressed in glutamatergic terminals coming from the cortex to the striatum. The explanation for this conundrum lies in the fact that these specific heteromers behave as sensors of the adenosine concentration and make the neuron respond accordingly (Figure 2). On the one hand, weak extracellular adenosine concentrations preferentially stimulate adenosine A₁ receptors, which would inhibit glutamatergic transmission. On the other hand, under a strong adenosine release to the extracellular space, adenosine A_2A receptor is activated and this activation blocks adenosine A₁ receptor‐mediated effects and results in potentiation of glutamate release 42, 43. Overall this GPCR heteromer system works as a switch that produces opposite outcomes depending on the adenosine concentration.

Figure 2.

Adenosine concentration switching device in glutamatergic nerve terminals expressing A₁–A_2A receptor heteromers. Concerning regulation of glutamate release, completely opposite outputs result at low or high adenosine concentrations. At low concentrations, the A₁ receptors become activated and Gi‐mediated signalling occurs. At high concentrations, the A_2A receptors become activated and both A₁ receptors loss affinity for adenosine and Gs‐mediated signaling occur. The mechanism underlying this phenomenon, that is, Gi‐blockade when the A_2A receptor is activated is not known, but it is originated within the receptor heteromer and likely impacts on the coupling to Gs and Gi proteins.

In the striatal MSNs, cannabinoid CB₁ receptors may participate in different heteromeric configurations and their activation leads to inhibition of glutamate release by blocking the calcium channels that participate in neurotransmitter release. At this presynaptic level, cannabinoid CB₁ receptors in the direct pathway may heteromerize with adenosine A_2A receptors to give rise to A_2A‐CB₁ receptor heteromers (Figure 1) 44, 45. It is not known whether adenosine A_2A receptor‐mediated inhibition of cannabinoids modulation of synaptic events 46 is dependent on a specific action of the heteromer or on a cross talk occurring at the level of intracellular signaling cascades. It should be noted that in neuroblastoma cells endogenously expressing CB₁ receptor and A_2A receptors (and therefore A_2A–CB₁ receptor heteromers), CB₁ receptor signaling via G_i is dependent on A_2A receptor activation 47.

Heteromers Located Postsynaptically

Postsynaptic A_2A receptor is found in enkephalin MSNs of the indirect pathway. It is mainly found perisynaptically and adjacent to dopaminergic synapses 48. Three different populations of heteromers may be found: the A_2A–A_2A receptor homodimer, the A_2A–D₂ receptor heteromer 49, 50, the A_2A–D₂–mGlu₅ receptor heterotrimer 51 and the A_2A–CB₁–D₂ receptor heterotrimer 52, 53 (Figure 1).

The antagonistic interactions between A_2A receptors and D₂ receptors have been demonstrated at biochemical, functional, and behavioral levels. The first indication about an antagonist relationship was obtained from behavioral analysis of Parkinson's disease animal models 54. More information to understand the consequences of the interaction of A_2A and D₂ receptors was obtained in membrane preparations from rat striatum, where stimulation of A_2A receptors produced a decrease in the affinity of D₂ receptors for agonists due to conformational modification in the D₂ receptor‐binding site 55. After this and other more recent studies, the use of A_2A receptor antagonists in Parkinson's disease treatment was proposed. Nowadays, the A_2A receptor antagonist KW‐6002, which preferentially blocks A_2A receptor‐forming heteromers with D₂ receptors 56, has been approved in Japan for Parkinson's disease treatment 57, 58. As indicated above, L‐DOPA is targeting D₂‐containing receptor heteromers, whereas KW‐6002 is targeting A_2A‐containing receptor heteromers.

Heterotrimers formed by A_2A, D₂, and mGlu5 receptors are also present in striatum. A_2A receptors co‐immunoprecipitate with mGlu5 receptors in co‐transfected cells and they colocalize in striatal tissue 59. Radioligand binding assays in rat striatum membranes showed that stimulation of mGlu5 receptors also produced a decrease in the affinity that D₂ receptors display for agonists. Moreover, when A_2A receptors and mGlu5 receptors are costimulated, the inhibitory effect on D₂ receptors is stronger than the reductions induced by stimulation of either receptor alone 60; due to trimer formation (or otherwise to cross talk at the intracellular level), these data suggest tight inter‐receptor interactions. Upon the D₂ receptor‐mediated motor control, mGlu5 receptor agonists and antagonists produced similar effects as, respectively, A_2A receptor agonists and antagonists. A selective mGlu5 receptor agonist may preferentially inhibit motor activation induced by D₂ receptor agonists 60, whereas mGlu5 receptor antagonists may reduce the effects of D₂ receptor antagonists 61. Furthermore, A_2A and mGlu5 receptor agonists and antagonists can result in synergistic effects at the behavioral level 59, 60, 62. A_2A–D₂–mGlu5 receptor interactions provide the rationale for the possible application of mGlu5 receptor antagonists or combined A_2A and mGlu5 receptor antagonists in Parkinson's disease treatment 62, 63, 64.

Tebano et al. 65 described that A_2A receptor activation potentiates the synaptic effects of the CB₁ receptor. Furthermore, CB₁ receptor‐induced depression of synaptic transmission may be prevented by inactivation of A_2A receptors 66, 67. A postsynaptic mechanism would depend on the interaction between A_2A receptors and CB₁ receptors in the enkephalinergic MSNs and probably also on the interaction with dopamine receptors as some of the effects of A_2A–CB₁ interactions depend on D₂ receptor function 68. A_2A–CB₁–D2 receptor heterotrimers have been detected in HEK cells 53, rat striatum 40, and primate caudate nucleus 39. Thus, it seems that distinct CB₁ receptor‐containing heteromers differentially located at pre‐ or postsynaptic membranes can account for the described diverse relationship between adenosine and cannabinoid receptors.

Relevance of Heteromers in CNS Disorders

Evidence in recent years points to heteromers as therapeutic targets in neurological or neuropsychiatric disorders. A heteromer deeply characterized in the CNS is the G_q/11‐coupled dopamine D₁–D₂ receptor heteromer, which has been identified in rat striatal neurons. Whereas individual receptors, D₁ and D₂, couple to, respectively, G_s and G_i and modify cAMP levels, the heteromer switches coupling to G_q/11, calcium mobilization and a cascade of events that may include accelerated neuronal growth and increased expression of BDNF. It has been proposed that this specific dopaminergic signaling may have an important role in drug addiction, Parkinson's disease, and schizophrenia 69, 70.

Heteromers containing the serotonin receptor 5‐HT_2A seem to play a key role in the psychotic actions of the hallucinogenic actions of the diethylamide of the lysergic acid (LSD) and of 2,5‐dimethoxy‐4‐iodoamphetamine (DOI), which behave as 5‐HT_2A agonists. In fact, oligomerization of serotonin 5‐HT_2A and dopamine D₂ receptors leads to cross talk by which 5‐HT_2A receptor‐mediated phospholipase C (PLC) activation is synergistically enhanced by the concomitant activation of the D₂ receptor. Moreover, the D₂ receptor‐mediated adenylate cyclase inhibition is decreased by the costimulation with 5‐HT_2A receptor agonists. This allosteric cross‐modulation provides a new framework to understand the psychotic actions of the hallucinogens acting via 5‐HT_2A receptors 71, 72. Interestingly, another player has entered into scene to explain the actions of these drugs. Some of the neuropsychological responses induced by hallucinogens seem to involve another heteromer formed by 5‐HT_2A and mGlu2 receptors. By competition binding studies in mouse sensory cortex membranes, it has been shown that the binding affinity of hallucinogenic 5‐HT_2A receptor agonists is increased by an mGlu2/3 receptor agonist, while DOI decreases the binding affinity of mGlu2/3 agonists. Again, there is an allosteric cross‐modulation that supports the view of multiple signaling possibilities due to the occurrence of multiple 5‐HT_2A receptor‐containing heteromers. It should be noted that complementary data obtained in different model systems and in mGlu2 and 5‐HT_2A knockout mice prove that antipsychotic effects of mGlu2 activation or 5‐HT_2A inhibition require the presence of the two receptors and thus suggest a key role of this heteromer in higher cerebral functions 73, 74, 75, 76.

Conclusions

GPCRs are the key factors in the molecular architecture of the CNS. The stoichiometry of GPCR complexes, which is still under study in in vitro cell models, and the exact composition of the pre‐ or postsynaptic complexes, that is, whether they contain 1, 2, 3… different receptors, are important details in the anatomical ultrastructure of the CNS. Examples of heteromers relevant for CNS function are indicated in Table 1.

Table 1.

Examples of receptor heteromers, their potential effects, and diseases for which they may become therapeutic targets

Heteromer	Effect	Disease
CB1‐CB2	Stress, pain, memory, learning, reproductive function, development	Mental illness, cancer, metabolic disorders
A1‐A2A 41, 42, 43	Sleep, arousal, cognition, learning, motor activity	Neurodegenerative diseases: Alzheimer's, Parkinson's
A2A‐D2 50, 54, 55, 56, 57, 58	Motor activation, memory	Huntington's disease, dyskinesia, Parkinson's
A2A‐CB1 44, 45, 46, 47, 65, 66, 67	Psychomotor activation	Neuropsychiatric disorders, drug abuse
D1‐D2 69, 70	Motor activation, neuronal growth	Parkinson's, schizophrenia, drug addiction
5HT2A‐D2 71, 72	Cognitive control, weight gain	Drug abuse, psychotic drugs, schizophrenia
5HT2A‐mGlu273, 74, 75, 76	Impulsive behavior, stress	Psychiatric disorders: schizophrenia, depression
A2A‐CB1‐D2 39, 40, 53, 68	Motor activation	Parkinson's, drug abuse, schizophrenia
A2A‐ D2‐mGlu559, 60, 61, 62, 63, 64	Motor activation	Parkinson's

The exact pre‐ or postlocalization of receptors in the neuron is fundamental to obtain a very precise action. At first, a given GPCR in the CNS had just a given effect, proximally modifying the level of second messengers, and overall regulating either neurotransmission (if postsynaptic) or neurotransmitter release (if presynaptic). The picture now is that a given GPCR may be both pre‐ and postsynaptic and play very different roles depending on its localization and on the exact structure of the GPCR complex in which it participates.

GPCR heteromers are formed by two or more receptors and their action is unique (and not a mere addition of the action of individual receptors within the heteromer). This explains, for instance, why two opposite‐coupled GPCRs, one leading to activation and the other leading to inhibition of adenylate cyclase, may be found in the same cell. The heteromer can achieve something that two individual receptors cannot achieve (the above‐described A₁–A_2A receptor heteromer, which is postsynaptically expressed in glutamatergic neurons arriving to the striatum, is a paradigmatic example).

The CNS uses heteromers to achieve different outcomes for a given neurotransmitter input. In fact, individual receptors being activated by a given neurotransmitter can only provide one response. In contrast, the occurrence of heteromers implies that a given neurotransmitter may lead to differential signaling depending on the heteromer context of the targeted receptor. Indeed, the heteromer context may vary from cell to cell and from place to place within a given neuron.

For all the above reasons, a final conclusion would be that dynamics of GPCR and of GPCR heteromers expressed in different regions and/or trafficking to different neuronal localizations are fundamental for neural transmission and plasticity.

Conflict of Interest

The authors declare no conflict of interest.