Excitotoxic mechanisms

Glutamatergic dysregulation is the key step leading to excitotoxicity [56]. Glutamate, however, does not directly damage the neurons. Indeed, the excitotoxic cascade is started by a prolonged activation of glutamate receptors [56] resulting in Na⁺ and Ca²⁺ influx and uptake by mitochondria, which may trigger the production of reactive oxygen species (ROS) and inhibition of ATP production. After glutamate receptor activation both the magnitude and duration of the increase of the intracellular Ca²⁺ concentration are important determinants of whether neurons degenerate. Indeed, Ca²⁺ concentration values much higher (low micromolar) than the typical cytoplasmic concentration in resting conditions (~100 nM) can be tolerated provided they are transient (seconds to minutes). On the other hand, even a low (~500 nM) increase of cytoplasmic Ca²⁺ concentration can kill the neuron if it is sustained for more than 20–30 min [53]. Glutamate receptors are classified into two main classes [66]: ionotropic glutamate receptors (iGluRs) and metabotropic glutamate receptors (mGluRs).

iGluRs are ligand-gated ion channels permeable to various cations that produce glutamate-evoked excitatory currents, while mGluRs control cellular processes.

Three types of iGluRs have been identified [67], namely N-methyl-D-aspartate (NMDA), α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA), and kainate (KA) receptors, all involved in excitotoxic responses, the NMDA receptors being considered to play the major role due to their high permeability to calcium ions [54]. NMDA receptors are tetrameric structures composed of subunits delineating a central pore and requiring simultaneous binding of glutamate and a co-agonist (D-serine or glycine) for activation [68]. Aspartate can also activate NMDA receptors, although with a lower affinity as compared to glutamate [69]. Differences in the type of subunits and in their assembly results in different NMDA receptor subtypes with different functional features (see [54] for a discussion). In this respect, the subtype eNMDA receptor (enriched with GluN2B subunits) has been reported to be particularly involved in activating neurotoxic pathways [70]. AMPA are tetrameric receptors as well, that depending on the subunit composition exhibit different calcium permeability [71]. In particular, AMPA receptors lacking the GluA2 subunit appear to significantly contribute to excitotoxic cell death [72]. As demonstrated by structural biology studies, indeed, when GluA2 is contained within the AMPA receptor the Ca²⁺-permeability is profoundly decreased, due to the presence of arginine at position 607 [73]. A similar property of allowing ion influx following glutamate exposure is also exhibited by KA receptors. They, however, are permeable to sodium and potassium ions and almost impermeable to Ca²⁺ [74]. Furthermore, while AMPA receptors are mostly localized in the postsynaptic membrane, KA receptors are localized both pre- and post-synaptically [75, 76]. Concerning AMPA receptors, an intriguing experimental finding [77] was the observation that rat and mouse cortical neurons secrete exosomes containing two out of the four AMPA subunits which combine to form the receptor tetramer, opening the possibility of an intercellular exchange of elements of the glutamatergic decoding apparatus by microvesicles (the so called “roamer” type of volume transmission [78]) that may represent an interesting mechanism for the modulation of this transmission line.

mGluRs represent a diverse receptor family, including 8 subtypes organized into 3 groups [79]. Group 1 mGluRs (mGluR₁ and mGluR₅) are coupled to G_q protein. When activated, they stimulate inositol triphosphate production and the release of Ca²⁺ from neuronal stores, contributing to the excitotoxic process [80, 81]. On the contrary, the action of group 2 (subtypes 2 and 3) and group 3 (subtypes 4, 6, 7, and 8) mGluRs was reported to decrease NMDA receptor activity and the risk of excitotoxicity [82]. The effect appeared mediated by a regulation of voltage-gated K⁺ channels to induce hyperpolarization of the cell, thus reducing the opening probability of the NMDA receptors [82].

An interesting process contributing to excitotoxicity, but not related to glutamate receptors, is the sustained elevation of glucocorticoids [83]. Glucocorticoid hormones are essential for the adaptive response to stress and an important target organ is the brain [84], where the action of corticosteroids is mediated by the high-affinity mineralocorticoid receptor (MR) and the lower affinity GR. They are members of the nuclear receptor superfamily of ligand-dependent transcription factors [85]. In the absence of hormone, GR resides predominantly in the cytoplasm of cells as part of a large multi-protein complex [85]. Upon binding, GR undergoes a conformational change resulting in the dissociation of the associated proteins. This structural rearrangement allows the translocation of GR into the nucleus, where GR binds directly to target DNA sequences called glucocorticoid-responsive elements (GREs) and regulates the expression of target genes [85]. Measurements of intracellular calcium levels in hippocampal neurons exposed to glutamate and other excitatory amino acids have shown that glucocorticoids disrupt cellular calcium homeostasis and promote calcium overload [86]. The suppression of the production of neurotrophic factors and the suppression of antioxidant defense mechanisms represent other possible endangering mechanisms mediated by glucocorticoids [53]. A rapid effect of glucocorticoids on synaptic glutamate currents has also been recorded in the periventricular nucleus of the hypothalamus mediated by the endocannabinoids [87]. The rapid effects observed, however, were too fast to invoke a genomic mechanism. Thus, the possible involvement of membrane-associated GRs has been suggested [88]. In this respect, several lines of evidence are presently available supporting this glucocorticoid signaling mechanism in the brain [89, 90]. Opioid peptides are another class of stress-related hormones that can affect the excitotoxic process [53]. Dynorphin and nociceptin, for instance, can exacerbate excitotoxicity [91].

Excitoprotective mechanisms

Neurons may implement a variety of mechanisms protecting against neuronal excitoxicity [54]. During action potentials, for instance, potassium channels can significantly limit neuronal excitability. A first example is provided by small-conductance calcium-dependent potassium channels [92]. Being quite sensitive to transient increases of cytosolic calcium [93] they generate a protective, hyperpolarizing signal [92, 94]. The ATP-dependent potassium channels represent a second important class of potassium channels triggering a protective mechanism. Their conductance, indeed, is enhanced by the ATP depletion that follows excitotoxic insults [95] and their activation at presynaptic sites was shown to inhibits glutamate release [96]. Further neuronal responses to cell stress include those activated by neurotrophic factors (such as BDNF) or by transcription factors (e.g., NF-κB and CREB) [53], and the expression of heat-shock proteins [97].

In the context of the present discussion, however, three more mechanisms deserve particular consideration. The first is the production of adenosine following a neural insult [98]. Adenosine, indeed, inhibits excitatory synaptic transmission, mostly through a presynaptic inhibition of glutamate release, an effect mediated by adenosine A₁ receptors. In parallel, the activation of these GPCRs also modulates calcium and potassium channels at the postsynaptic level, leading to a decrease of calcium currents in response to glutamate. The second defensive mechanism involves the inhibitory neurotransmitter GABA that can decrease neural excitability by increasing Cl^– influx through GABA_A receptors [99]. In conditions of sustained excitation, an additional fast-acting neuroprotective mechanism has also been identified, based on the innate capacity of neurons to quench excitation by recruiting GABA_B receptors [100]. Finally, of interest in the present context is emerging research demonstrating the importance of steroid hormones in the regulation of glutamatergic neurotransmission (see [65] for a specific review). The co-localization of AMPA and mGluRs with androgen or estrogen receptors in septum, amygdala, and hypothalamus of rats has been assessed by immunohistochemistry [101], and functional studies indicated a neuroprotective role played by these hormones in ischemic conditions [102]. The effect appears mediated by their receptors, as indicated by studies on estrogen receptors α and β (ER_α and ER_β), showing that specific agonists of these nuclear receptors can protect hippocampal CA1 neurons during ischemia (where glutamate levels are known to be elevated) [103]. In this context, the modulating effects of progesterone, testosterone, and neurosteroids on glutamatergic neurotransmission also deserve consideration [65]. Although more specific studies on this topic would be needed, available data suggest a neuroprotective role for these hormones as well.

Glial cells and excitotoxicity

In addition to the well-known metabolic support given to neurons [53, 54], a key function of astrocytes in the CNS is the regulation of neurotransmitter homeostasis [104], as they can uptake synaptically released neurotransmitters, metabolize them, and release back to neurons precursors of those neurotransmitters as well as regulatory gliotransmitters. Extracellular concentrations of GABA, for instance, are under control of GABA transporters (GAT) expressed by both neurons and astrocytes. Astrocytes mainly express GAT-1 and GAT-3 subtypes and it has been estimated that about 20% of extracellular GABA is taken up by these cells [105].

Glutamatergic transmission, however, is in a special way regulated by astrocytes (see [63, 106] for specific reviews on the topic). Glutamatergic synaptic activity generates increases, that can exceed micromolar levels, of glutamate concentration in the extra synaptic space [107]. Although many CNS cells contribute to glutamate removal, astrocytes represent the most efficient elements, being responsible for about 90% of glutamate clearance [108]. The main mechanism of glutamate uptake by astrocytes is represented by two types of glutamate transporters (Na⁺-dependent and Na⁺-independent) [63]. High-affinity sodium-dependent glutamate transporters, also known as excitatory amino acid transporters (EAATs), play a major role in the removal of extracellular glutamate by astrocytes and in preventing excitotoxicity [54]. Once taken-up by astrocytes, glutamate can be amidated to glutamine, which is released to the adjacent neurons, where it is converted to glutamate (or GABA) and repackaged into vesicles for release [109]. This astrocytic glutamate-glutamine cycle [63, 110], therefore, is also involved in the maintenance of glutamate homeostasis by sustaining the synthesis of the neurotransmitter [54].

Astrocytes monitor glutamate in their environment by expressing a panel of glutamate receptors [104, 111, 112]. In these cells, the response to glutamatergic signaling is mainly mediated by mGluRs, with mGluR₁, mGluR₃, and mGluR₅ being the most expressed [104, 111]. According to several lines of evidence [104], type 5 mGluR appears as the most relevant, mediating the response of astrocytes to glutamate in many brain areas, such as hippocampus, nucleus accumbens, and thalamus. Ionotropic receptors of the AMPA and NMDA types have also been identified in astrocytes [112, 113]. When compared to neuronal NMDA receptors, astrocytic NMDA receptors appear to exhibit some distinctive features. They, indeed, contain GluN3A receptor subunit, which lowers Ca²⁺ permeability [114], making astrocytes less vulnerable to glutamate-mediated excitotoxicity [113]. Moreover, astrocytic NMDA receptors lack Mg²⁺ block and can be activated without antecedent depolarization [115]. Glutamate receptors also play a role in modulating the activity of glutamate transporters, since group 2 mGluRs can enhance their expression, while group 1 mGluRs and iGluRs downregulate EAATs [63].

The uptake of glutamate into astrocytes can be increased by exogenously administered estradiol, the predominant estrogen in terms of activity [116]. More generally, EAATs may be affected by steroid hormones [65]. The expression of GLT-1 and EAAT3 is upregulated following ischemia, and mRNA and protein levels of these transporters resulted further increased by progesterone and estrogens as compared to ischemia alone [117]. On the contrary, glutamate uptake by astrocytes can be inhibited by adenosine signaling through the adenosine A_2A receptors, which are also expressed by astrocytes, due to a physical association of A_2A with Na⁺/K⁺-ATPases [118].

In addition to uptake, however, many studies indicate that astrocytes can also release glutamate to the adjacent neurons [119], and a notable subpopulation of astrocytes, selectively expressing synaptic-like glutamate-release machinery, has also been identified in defined anatomical locations [120]. Acting as a gliotransmitter [121], glutamate helps to synchronize neuron firing and modulate the excitatory or inhibitory neuronal transmission [63]. In this respect, the possible impact on synaptic transmission of a significantly increased glutamate release by astrocytes is hardly predictable [122]. Extra synaptic mGluRs should be mainly reached. If astroglial glutamate mainly involves the inhibitory mGluR_2-4 (G_i/o coupled) located on the nerve terminals, a reduction of neuronal glutamate release would take place with inhibition of glutamate transmission [123] and reduced toxicity. However, if also extra synaptic and postsynaptic mGluR₁ and mGluR₅ (G_q coupled) are significantly activated by astroglial glutamate, an increase of glutamate synaptic strength may occur, leading to increased intracellular calcium levels [122]. Extracellular signals can modulate astrocytic glutamate release. An example is dopamine, that acting on the D₂ receptors expressed by astrocytes inhibits the release of glutamate by these cells [35]. In the context of the present discussion, of interest are data showing that, in addition to glutamate, astrocytes can release D-serine (essential for NMDA receptors function), but also ATP, GABA [124], lactate [125], and citrate, chelating zinc ions inhibits NMDA receptor [126].

This dual function of astrocytes, with a balance between opposing actions of glutamate removal and release, confirms the complex role played by these cells in conditions potentially leading to excitotoxic processes [63, 113].

Another glial cell type involved in excitotoxicity is microglia, the resident immune system of the CNS. They are plastic cells, exhibiting a variety of cell morphologies and cell surface markers. They are activated by molecules from damaged neurons and, like peripheral macrophages, they are usually described as characterized by two activation states [127]. In the classical M1 phenotype microglia are cells producing proinflammatory factors. By contrast, in the M2 phenotype, they contribute to neuroprotection and injury repair. A recently characterized process of endocytosis of full-length tau protein by microglial cells [128, 129], triggered by the purinergic G protein-coupled receptor 12 (P2Y₁₂), provides an example of protective action in Alzheimer’s disease. Increasing evidence, however, suggests that this M1/M2 dichotomy is over-simplified (see [130, 131] for detailed discussions of the topic). Different sub-types of M2-like microglia, indeed, have been identified, and simultaneous expression of classical M1 and M2 markers within individual microglial cells has been reported [132]. These findings indicate that individual microglial cells may be able to adopt complex phenotypes that exhibit both inflammatory and restorative functions. To detect molecular patterns associated with tissue damage, microglial cells exploit a large set of receptors, including all the glutamate receptor types except mGlu₇ [131, 133]. Ionotropic [133, 134] and group I/II metabotropic [135] glutamate receptor activation is in general associated with the secretion of cytotoxic or inflammatory factors. Conversely, group III [131, 136] metabotropic receptors were reported to mainly trigger protective actions. Upon specific activation, microglia can also release glutamate, contributing to excitotoxicity.

A schematic view of the main mechanisms is provided in Figure 1, and examples of excitotoxic mechanisms contributing to acute and degenerative CNS diseases are reported in Table 2.

Display full size

Figure 1. 

Schematic view of the main processes involved in excitotoxicity. Red lines represent processes favoring excitotoxicity; blue lines represent protective processes. In neurons [53, 54], the binding of kainate (KA) receptors results in Na⁺ influx and membrane depolarization, leading to the opening of voltage-dependent Ca²⁺ channels (VDCC). Further calcium influx is associated to the activation of some forms of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors and most importantly of N-methyl-D-aspartate (NMDA) receptors. The activation of group 1 metabotropic glutamate receptors (mGluRs) also contributes to increase intracellular Ca²⁺ by mobilizing intracellular calcium stores. The resulting increase in cytoplasmic calcium levels may induce Ca²⁺ uptake into the mitochondria that, when excessive, leads to the production of reactive oxygen species (ROS) and inhibits ATP production followed by cell damage or death. The action of activated group 2/3 mGluRs, however, can decrease NMDA receptor activation and protect to some extent the cell. Concerning astrocytes [54, 63], the release of protective factors and, most importantly, their activity of glutamate uptake through excitatory amino acid transporters (EAATs) channels, represent important processes to prevent excitotoxicity. Astrocytes, however, can also release glutamate. Microglial cells monitor the extracellular environment by a large panel of receptors. Depending on the phenotype they assume when activated, microglial cells can trigger pro-inflammatory actions and release of glutamate, or neuroprotective processes [130]

Neuronal receptor complexes involving glutamate receptors

As Table 3 illustrates, glutamate receptor subtypes were found to participate in the formation of a quite large number of receptor complexes (see [20, 180] for specific reviews).

mGluRs are class C GPCRs and dimerization is mandatory for these receptors to exert their functions. However, in addition to homodimers, this family of glutamate receptors was shown to form a number of heterodimers as well. A first set of heterodimers involves mGluRs of different types (e.g., mGluR₁-mGluR₂, mGluR₁-mGluR₅, mGluR₂-mGluR₃, mGluR₂-mGluR₄). Molecular details that govern dimerization and activation of these heterodimers have been recently explored by Wang and collaborators [159] by determining cryo-electron microscopy structures of dimers in distinct conformational states. The results showed that in the inactive states the heterodimers may assume distinct conformations at both the extracellular domains and TM domains. Upon activation, in contrast with mGluR homodimers, the heterodimers assume configurations suggesting an asymmetric mode [159, 181] of signal transduction with only one of the subunits coupling to the G protein. The choice of the subunit for G protein activation likely depends on the stability of the initial inactive conformation, highlighting the complexity of the mGluRs signaling mechanism. A second set of identified receptor complexes containing mGluRs involves receptors from a different family, such as adenosine (A₁ and A_2A), dopamine (D₂), serotonin (5HT_2A), and opioid (MOR) receptors [20], suggesting receptor complexes as a relevant molecular organization to tune glutamatergic transmission. In these heterodimers, synergistic interactions may occur. An example is the mGluR₂-5HT_2A receptor complex where 5HT_2A-dependent phosphorylation of mGluR₂ at Ser843 promotes mGluR₂ signaling [182]. On the other side, antagonistic interactions were found to characterize some heterodimers containing mGluRs. An example is the mGluR₁-A₁ receptor complex in cerebellar Purkinje cells. In this heterodimer, the activation of adenosine A₁ receptor attenuates mGluR₁-mediated neuronal responses to glutamate [161]. In this context, of great interest is also obtained evidence for the existence extrasynaptically around striatal glutamate synapses of mGluR₅-D₂ heterodimers [165] and of D₂-A_2A-mGluR₅ heterotrimers [23]. In these heteromers, the inhibitory action of dopamine D₂ receptors on glutamate release is effectively counterbalanced by mGluR₅ and adenosine A_2A receptors, contributing to enhancement of the neuron excitability [20].

Heterodimers resulting from direct RRI of iGluRs (NMDA, AMPA) with other receptor types have also been identified. Both D₁ and D₂ dopamine receptors, for instance, may form receptor complexes with the NMDA receptor, and their activation reduces the operation of NMDA channels [152, 153]. In particular, the interaction between dopamine D₁ receptor and the GluN1A subunit of NMDA appears associated with a D₁-induced protection from excitotoxicity [152]. Opposingly, the interaction in the NMDA-MOR heterodimer is synergistic [154]. In this receptor complex heteromerization occurs by electrostatic interactions between the two C-termini, which dissociate when morphine activates PKC, leading to enhanced calcium currents through the NMDA channels [20]. Cross-antagonism is observed in the NMDA-A_2A heterodimer [155] and in the NMDA-D₁-H₃ heterotrimer [157]. In both receptor complexes, indeed, the exacerbation of NMDA receptor function was reduced by antagonists of the A_2A and H₃ receptors respectively. Furthermore, in the heterotrimer H₃ antagonists were also found to counteract the effect of an overstimulation of the D₁ receptor. A more complex reciprocal interaction has been observed in the NMDA-mGluR₅ heteromer. Antagonistic interactions, indeed, have been reported [156] with inhibition of the NMDA currents, but synergistic interactions have also been observed [183] and in the hippocampus, both types of actions were noted [20]. An explanation could be the dynamic association of scaffolding proteins to the heteroreceptor complex [184].

A further example of the important role played by proteins interacting with protomers is provided by the AMPA-D₂ heterodimer, where D₂ agonist regulation of the AMPA receptor-mediated neurotoxicity is made possible by an increased coupling of the intracellular loop 3 of the D₂ receptor to NSF (ATPase N-ethylmaleimide-sensitive factor) protein [151]. An enhancement of glutamate excitotoxicity, on the contrary, characterizes the AMPA-IFNγ heterodimer [150], a neuron specific calcium-permeable complex whose formation is triggered by IFNγ.

Experimentally assessed molecular models of receptor complexes involving glutamate receptors are shown in Figure 2A.

Display full size

Figure 2. 

Examples of receptor complexes potentially involved in the modulation of excitotoxic processes. (A) Three inactive configurations were exhibited by the metabotropic glutamate receptor 2-3 (mGluR₂-mGluR₃) heterodimer according to Wang and collaborators [159]. Available data suggest that mGluR₂-mGluR₃ heterodimers are likely present at mossy fibers terminals in the CA3 region of the hippocampus, a region critically involved in memory processing [185]. Molecular models (Protein Data Bank codes: BJCU, BJCV, BJCZ) are shown. They differ at both the extracellular and transmembrane (TM) domains. The arrangement of TM in the three structures is schematically illustrated below each model. (B) Receptor complex formed by adenosine A_2A and dopamine D₂ receptors as described by Borroto-Escuela and collaborators [186]. The TM domains forming the interface are indicated. In astrocytes the A_2A-D₂ heterodimer controls the release of glutamate from striatal astrocytic processes and a potential role of this receptor complex in neuropsychiatric disorders has been suggested [45]. (C) Possible structure of the heterodimer between glucocorticoid and mineralocorticoid human receptors, obtained by alignment methods [187] starting from the experimentally assessed structures of the two receptor homodimers (PDB codes: 6BSF and 4TNT). It has been reported that in the hippocampus mineralocorticoid receptor (MR) and glucocorticoid receptor (GR) bind as heterodimers to the same glucocorticoid-responsive element (GRE) sites after an acute stress challenge [170]

Neuronal receptor complexes involving receptors modulating glutamate homeostasis

Glutamate homeostasis and glutamatergic transmission can be regulated by a variety of extracellular signals. A first example is the typical neuromodulator adenosine [98], whose extracellular levels increase under noxious conditions. The high affinity adenosine A₁ receptor plays a major role in neuroprotection, mediating a significant inhibitory action on synaptic transmission and neuronal excitability [188]. However, under noxious conditions the low affinity, facilitatory, adenosine A_2A receptors are up regulated [98] and may form heteroreceptor complexes with the A₁ receptors [167]. The major RRI in the heterodimer appears to be a reduction of A₁ affinity by A_2A agonists. Thus, at high concentrations of adenosine, able to activate A_2A receptors, an increase of glutamate release was found, contributing to enhanced excitation and possible excitotoxicity [122]. This regulation, therefore, inverts the usual inhibitory action of adenosine acting on A₁ receptors, with A_2A actions surpassing A₁-mediated effects.

As demonstrated by studies on cortico-striatal glutamate terminals, adenosine A_2A receptors can also form heteromers with dopamine D₂ receptors [8, 189]. Being that the D₂ receptor is involved in the inhibition of glutamate release, this receptor complex significantly modulates glutamatergic transmission [190]. The A_2A-D₂ heterodimer (Figure 2B) is probably one of the most studied (see [191] for a recent review) and is characterized by reciprocal antagonistic interactions between the protomers. A_2A agonists, indeed, lead to a reduction of the affinity of the D₂ agonist-binding site [192], while D₂ receptor activation inhibits the A_2A-induced increase in cAMP accumulation [193] and was shown to slow and partially inhibit the binding of A_2A agonists to the receptor [194]. Furthermore, new allosteric binding sites were found to emerge following the formation of the receptor complex. Homocysteine, for instance, can bind to the heterodimer without interfering with the RRI between A_2A and D₂ receptors, acting as an allosteric antagonist of the dopamine D₂ receptor and amplifying the effect of A_2A agonists [49].

Quite recent interesting findings indicate that adenosine A_2A receptor is a major modulator of glucocorticoid signaling [169] as well. As mentioned before, glucocorticoids by acting on their GR receptor can disrupt cellular calcium homeostasis, promote calcium overload, and contribute to the excitotoxic process [86]. The possibility of a direct interaction between the membrane A_2A receptor and the cytosolic-perimembrane GR has been suggested. It would involve direct RRI between the intracellular domains of A_2A and GR, leading to a recruitment of GR and a reduction of its transcriptional activity (see [122] for a discussion). In this context, of substantial interest is a study on the hippocampus [170] showing that acute stress challenges result in an increased interaction of nuclear corticoid receptors GR and MR with their genomic recognition sites. Beyond expectancies, under the analyzed condition they interact with those sites not just as homodimers, but also as heterodimers (Figure 2C). These findings indicate an additional level of complexity in brain glucocorticoid action and may offer suggestions for future research in the almost unexplored field of RRI between nuclear receptors. In this respect, in view of the excitoprotective effects exhibited by steroid hormones (estrogens in particular) [67], of interest is evidence reporting a direct interaction between ER_α and mGluR₁ in rats undergoing hormonal treatment [171]. Following treatment with estradiol an increased internalization of both mGluR₁ and ER_α was also observed [195]. These findings support the possibility of estrogen/glutamate receptor complexes mediating an interaction between hormonal and glutamatergic signaling.

Concerning GABA transmission, a signaling pathway that can decrease neural excitability [99], heteromerization between GABA_A and dopamine D₅ receptors has been demonstrated by Liu and collaborators [168]. The results indicated that the formation of the complex was dependent on the co-activation of the monomers and allowed a reciprocal crosstalk, leading to a reduction in GABA_A signaling and a reduced coupling of D₅ to the G_s protein.

Receptor complexes in astrocytes and microglia

Although less investigated than in neurons, many available data indicate that RRI may play a significant role in astrocytes [111, 196].

Adenosine A_2A and dopamine D₂ receptors provide the first example. Both receptor types were found to form receptor heteromers in astrocyte processes [172]. From the functional standpoint, the activation of D₂ receptors inhibited glutamate release, while the activation of A_2A receptors, per se ineffective, abolished the D₂-induced release inhibition [197]. Heterodimers between oxytocin receptor (OTR) and A_2A or D₂ receptors have also been recently identified [174, 175]. Functionally, both receptor complexes appear involved in the regulation of glutamatergic transmission. The interaction between OTR and D₂ receptors appears facilitatory, leading to an increase of D₂ receptor affinity by oxytocin. Actually, subthreshold concentrations of dopaminergic agonists (too low to activate the astrocytic D₂ receptor) become effective in the presence of oxytocin [175]. On the contrary, in the A_2A-OTR heterodimer the interaction was antagonistic, since the activation of the adenosine receptor abolished the oxytocin-induced inhibition of the release of glutamate by astrocytes [174].

In astrocytes a demonstrated heteromeric association involving the adenosine A₁ receptor is with purinergic P2Y₁ receptors [173]: As indicated by coimmunoprecipitation methods, these receptors colocalize on astroglial membranes where they organize into receptor complexes. The pharmacological results indicated that within the complex P2Y₁ receptor activation induces A₁ receptor desensitization. Thus, it has been suggested that this heteromer could play a significant role in the astrocytic modulation of glutamatergic neurotransmission during excitotoxic processes, when large amounts of adenosine and purines are released [173].

Adenosine receptors are also involved in the regulation of GABA uptake, which occurs via the modulation of GATs by the A₁ and A_2A receptors [196]. By coimmunoprecipitation and BRET assays it has been demonstrated [105] that in astrocytes these receptors can be organized as A₁-A_2A receptor complexes. Coupled to two different G proteins, G_s and G_i/0, both regulate GABA transport in an opposite way, with the A₁ protomer mediating inhibition of GABA transport and the A_2A protomer mediating facilitation of GABA transport into astrocytes. Due to difference in affinity of the two receptor types, at low levels adenosine preferentially binds the A₁ protomer, while at high concentrations adenosine activates the A_2A protomer. The receptor complex, therefore, was suggested to operate as a dual amplifier to control ambient GABA levels [105].

As briefly illustrated before, some receptor complexes (e.g., A_2A-D₂ and A₁-A_2A) are expressed in both neurons and astrocytes. In this respect, however, it is reasonable to assume [47] that some difference in terms of conformation and interaction of the protomers within the heteromers could occur because of differences (as, for instance, membrane potential [198] and lipid composition [199]) between the two cell types in the membrane microenvironment.

For the present discussion is also of interest the reported RRI between purinergic ionotropic receptors (P2X) (ligand-gated cationic channels) purinergic receptors expressed in microglia and controlling microglia activation [200]. P2X₄ and P2X₇ are the dominant forms of microglial P2X receptors [133]. Although still a matter of debate [201], the possible occurrence of P2X₄-P2X₇ heteromers has been reported in these cells [176], probably allowing a more sophisticated regulation of cytokine production and early inflammatory gene expression [133, 176].

As discussed before, microglia can also exert neuroprotective actions. In this respect, there is evidence that endogenous cannabinoids may favor a switch of microglial cells to an anti-inflammatory phenotype [202], and activation of cannabinoid type 2 (CB₂) receptor has been proposed to be the mechanism triggering these effects [203]. A number of heteroreceptor complexes involving cannabinoid receptors in microglia have been identified. The first example is the receptor complex CB₁-CB₂ between the two types of cannabinoid receptors [177], where the activation of one receptor blunts the response of the partner, leading to a wide spectrum of effects when reached by endocannabinoids or by synthetic molecules acting on cannabinoid receptors [204]. CB₂ receptor was also found to form heterodimers with the adenosine A_2A receptor [178], and the orphan receptor G protein-coupled receptor 18 (GPR₁₈) [179]. In the A_2A-CB₂ receptor complex, the blockade of the A_2A receptor leads to increased CB₂ signaling, while bidirectional cross-antagonism was observed in the GPR₁₈-CB₂ heteromer. Thus, both the receptor complexes are of interest from a pharmacological standpoint, since the use of antagonists targeting A_2A or GPR₁₈ receptors could be useful in the microglia-mediated protection of neuronal death in neurodegenerative diseases [178, 179].