Adenosine is a nucleoside mainly formed by degradation of ATP located intracellularly or extracellularly and acts as a neuromodulator (Fredholm, 1995; Gomes et al., 2011). In the extracellular fluid of the neuronal-glial networks of the Central Nervous System (CNS), it operates as a volume transmission signal through diffusion and flow to modulate the activity of both glial cells and neurons (Fuxe et al., 2010). It also acts as a homeostatic modulator (Gomes et al., 2011). Intracellular adenosine can reach the extracellular space through inter alia equilibrative transporters in the plasma membrane. The effects of adenosine are mediated via four adenosine receptor subtypes: A1R, A2AR, A2BR, A3R (Fredholm, 1995; Fredholm et al., 2011).

The A1R is distributed all over the CNS and found in high densities. It is coupled to G_i/o and inter alia inhibits adenylyl cyclase (AC) and calcium channels, and activates potassium channels. Being located both in synaptic and extrasynaptic positions at the pre and postsynaptic level, it inhibits synaptic transmission and hyperpolarizes nerve cells (Gomes et al., 2011). The A1R is also expressed in astroglia and microglia to fine tune neuronal-glial interactions.

The A2AR has a wide-spread distribution but it is especially enriched in the ventral and dorsal striatum where it is mainly located in the striato-pallidal GABA neurons at a synaptic and extrasynaptic location (Fuxe et al., 2003; Gomes et al., 2011; Brugarolas et al., 2014; Navarro et al., 2016a). It is coupled to G_s/olf proteins, activates AC, enhances glutamate release and enhances the activity of the striato-pallidal GABA neurons involving also inhibition of their inhibitory dopamine D2R located at the postsynaptic level (Fuxe et al., 2007; Gomes et al., 2011).

Unlike the A2AR the A2BR is coupled G_s/q, which leads to PLC activation and increases in intracellular calcium levels (Linden et al., 1999; Fredholm et al., 2011; Goncalves et al., 2015). The A2BR has been scarcely studied in the CNS. A difference from A1R and A2AR is that they are activated by adenosine only at micromolar adenosine concentrations, which may be only reached in pathological states (Müller and Stein, 1996). Their physiological role remains unclear but highly interesting results were recently obtained demonstrating the existence of A2AR-A2BR heteroreceptor complexes, in which A2BR protomer constitutively inhibited the function of the A2AR protomer (Hinz et al., 2017, in press).

With regard to the A3R it was shown to be located in neurons of the hippocampus based on the presence of A3R mRNA levels in hippocampal neurons using single cell PCR analysis (Lopes et al., 2003). The A3R possesses a high affinity for adenosine and is mainly coupled to G_i2.3. It therefore inhibits AC signaling but it can also activate PLC (Fredholm et al., 2011). Activation of A3Rs found in hippocampal nerve terminal membranes using Western blot analysis, produces neuroprotective actions (Boison, 2010; Boison and Shen, 2010; Fishman et al., 2012). It is of high interest that A3R activation selectively brings down persistent pain states through its analgesic properties (Little et al., 2015).

In the current perspective paper we will give an update of the adenosine A2AR heteroreceptor complexes that may be expressed in the CNS and to which extent they may participate and help explain molecular mechanisms underlying the involvement of adenosine receptors in neurodegeneration and neuroinflammation.

It was recently found that the two isoreceptors A2AR and A2BR can form heteroreceptor complexes in living cells and in brain tissue using BRET, FRET, BiFC, and proximity ligation assay (PLA) techniques (Hinz et al., 2017, in press). Remarkably, the demonstration that the A2AR protomer within A2AR-A2BR heteroreceptor complex loses its high affinity binding coupled to marked reduction of the potency of A2AR agonists to activate the Gs-AC pathway with accumulation of cAMP (Hinz et al., 2017, in press). The A2BR is an adenosine receptor with a low affinity for adenosine in contrast to the A2AR with high affinity for adenosine. So far the A2AR-A2BR heterocomplexes have been found in the cerebral cortex and in the hippocampal cortex by using PLA (Hinz et al., 2017, in press).

Based on these exciting findings it seems possible that an upregulation of A2BR expression in the glial-neuronal networks of the brain, which have been reported in neuroinflammation (Feoktistov and Biaggioni, 2011), can block the neurodegenerative effects by A2AR inactivation. Such an upregulation may increase the formation of A2AR-A2BR heteroreceptor complexes which should markedly reduce the A2AR-mediated signaling. These events can take place when the two adenosine receptors are expressed in the same nerve or glial cells. However, it should be considered that through extracellular vesicle (ECV) mediated volume transmission (Guescini et al., 2012; Borroto-Escuela et al., 2015), A2BRs can be transferred to cells expressing A2AR and to form A2AR-A2BR heteroreceptor complexes. Recently, ECV-mediated intercellular communication in the CNS was classified as a component of VT (Borroto-Escuela et al., 2015). The concept of ECV-mediated VT was based on the highly significant work of Simons and Raposo (Simons and Raposo, 2009), which demonstrated the fundamental role of exosomes in intercellular communication. However, it should be underlined that also alterations in endogenous protein expression plays a major role in changing the protein panorama in the cells.

It is proposed that such processes with increased formation of A2AR-A2BR heteroreceptor complexes may represent an important mechanism to counteract the neurodegenerative and toxic effects in models of brain disease of the expression and in vitro activation of A2ARs (as discussed above). Targeting and increasing expression of A2AR-A2BR heteroreceptor complexes via gene therapy involving the A2BR gene or a pharmacological approach may be a novel strategy for treatment of neurodegenerative disease in which A2AR over activation and/or overexpression plays a detrimental role. The pharmacological analysis can involve the identification of the transcription factors activating the expression of the A2BR as well as the use of a brain-penetrant heterobivalent compound built up of an A2A receptor antagonist pharmacophor and an A2BR agonist pharmacophor. Such a heterobivalent compound may assist in the heteromerization process. The A2AR-A2BR heteromer may be in balance with A2AR and A2BR homomers and other types of A2AR and A2BR heteroreceptor complexes. This homo and heteroreceptor panorama may vary among brain regions and within glial-neuronal networks of the same region and also among individual neurons and different types of glial cells in the same region.

It should be mentioned that A2BRs can also counteract the A1R produced inhibition of the glutamatergic transmission of the hippocampus (Goncalves et al., 2015). It seems likely that these findings reflect the existence of A1R-A2BR heteroreceptor complexes in glutamate nerve terminals, where the two adenosine receptor subtypes seem to be coexpressed.

In line with the above proposal is the early demonstration that A1R-A2AR heteromers exist in striatal and hippocampal glutamate nerve terminals where A1R and A2AR receptors were found to be colocalized (Rebola et al., 2005) and form an isoreceptor complex in the rat brain (Figure 1). The major receptor-receptor interaction found appears to be an A2AR agonist produced reduction of A1R affinity. Therefore, at high concentrations of adenosine, which can activate A2ARs, an increase of glutamate release is found. In astrocytes a similar mechanism maybe found involving A1R-A2AR heteromers which via Gi/o and Gs proteins modulate GABA transport (Cristovao-Ferreira et al., 2011, 2013). It was proposed that the structure might be based on heteromers formed from homomers (Navarro et al., 2016b).

Taken together, it seems possible that with increased concentrations of adenosine the allosteric receptor-receptor interactions in the A1R-A2AR and putative A1R-A2BR isoreceptor complexes will favor glutamate release which may contribute to enhanced excitation and possible excitotoxicity.

As clearly pointed out by Gomes et al. (2011), it is still not clear if the trophic factors and their tyrosine kinase receptors (RTK) play a significant role in the regulatory actions of adenosine via A2AR activation, since these signals are supposed to produce neuroprotection. A2AR receptor antagonists in fact counteract neurodegeneration in a number of models. Nevertheless, there is one interesting and significant publication that demonstrates that activation of A2AR receptors can enhance Trk neurotrophin receptor signaling which in this preparation developed independently of neuroptrophins (Lee and Chao, 2001). Furthermore, in a beautiful paper FGF2 and A2AR agonists were found to act via FGFR1 and A2AR to enhance synaptic plasticity in the striatum (Flajolet et al., 2008). In this case an FGFR1-A2AR heteroreceptor complex was demonstrated (Flajolet et al., 2008; Borroto-Escuela et al., 2013a; Figure 2). It seems likely that the results obtained in the Lee and Chao paper (Lee and Chao, 2001) also reflects the existence of heteroreceptor complexes between TrkA and A2AR and TrkB and A2AR in which A2ARs enhance Trk signaling.

In line with these findings it was also found that chronic in vivo treatment with an A2AR antagonist blocked the BDNF induced facilitation of LTP. Thus, a reduction of BDNF action was achieved (Jeronimo-Santos et al., 2014).

Based on the work presented in this article it appears that treatment with A2AR antagonists and increased formation of A2AR-A2BR heteroreceptor complexes should represent novel strategies for alpha-synyclein induced neurodegeneration. However, in the case of deficits in neurotransmission plasticity A2AR agonists may be the way to go for treatment in view of their ability to enhance TRK signaling.

It was recently demonstrated that A2ARs can induce a dysfunction of the hypothalamic-pituitary-adrenal (HPA) axis by targeting the function of glucocorticoid receptors (GR; Batalha et al., 2016). Overactivation of A2ARs leads to loss of plasma glucocorticoid levels and reduction of hippocampal levels of GR. The A2AR activation affected GR function by modulating the transcriptional activity of GR and their nuclear location. These are exciting findings and show that the A2AR is a major modulator of the function of GR. It can represent one important mechanism for the cognitive enhancement with improvement of memory produced by A2AR antagonists (Batalha et al., 2016).

In our opinion a direct interaction of the membrane located A2ARs with cytosolic-perimembrane GR should also be considered. Glucocorticoid receptors also exist at the neuronal membrane and can regulate nongenomic corticosteroid signaling (Groeneweg et al., 2012). Thus, a direct receptor-receptor interaction between the intracellular domains of the A2AR and GR becomes possible. Through the formation of this complex the A2AR can modulate the function of GR in two ways. The first possibility is that the A2AR-GR complex became recruit to the lysosome. As a results only the intracellular domains of the A2AR may remain linked to the GR and be translocated to the nucleus as a new component of GR. The link of these A2AR fragments to the GR element can substantially alter the GR transcriptional activity and represent a novel mechanism for modulation of gene transcription. Another alternative is the A2AR is available to recruit GR to the plasma membrane to an increased extent. In this way less GR will be available for translocation into the nucleus. In this way the transcriptional activity of GR can become reduced, instead the nongenomic GR signaling become dominant.

Interestingly, recent work demonstrated that the GR binds to GREs in β-arrestin 1 and 2 and modulates their gene expression (Oakley et al., 2012) which will became reduced by the A2AR-GR heteroreceptor complex. Such mechanisms may in fact have a major impact on the GR function leading to a marked disturbance of synaptic plasticity and memory. Thus, it is proposed that such a A2AR-GR heteroreceptor complex may lead to a mechanism that can change the hippocampal transcription panorama. The hippocampal connection to the hypothalamus may also be affected leading to hypothalamic-pituitary-adrenal dysfunction. Future work will establish if this hypothesis is of value.

We have previously introduced the theory that long-term memory becomes possible by the translocation of parts of the heteroreceptor complexes into the nucleus over the perinuclear membrane and promote the activation capability of a particular set of transcription factors (Fuxe et al., 2014a,b; Borroto-Escuela et al., 2015). These may then induce the transcription of unique adaptor proteins that upon expression can bind to the heteroreceptor complexes in the cytoplasmic component of the plasma membrane and link them to the cytoskeleton and major scaffolding proteins (Fuxe et al., 2014b). In this way they become stabilized with conserved allosteric receptor-receptor interactions.

It should be noted that also adenosine A3Rs take part in heteroreceptor complexes. So far A1R-A3R complexes have been demonstrated in the cerebral cortex and dorsal hippocampus (Hill et al., 2014) and A2AR-A3R isoreceptor complexes in the dorsal hippocampus mainly in the pyramidal cell layer using proximity ligation assay (Borroto-Escuela et al., 2016a; Figure 1). The function of complexes including A3R has not yet been clarified.

The exciting work performed on A2AR mediated neurodegeneration performed by several groups (Gomes et al., 2011; Santiago et al., 2014; Laurent et al., 2016; Ferreira et al., 2017a) opens up a major role for A2AR antagonists in neurodegenerative disease involving alpha-synucleins. In this perspective article we have proposed that adenosine A2AR iso- and heteroreceptor complexes and their allosteric receptor-receptor interactions can be key players in balance with corresponding homorecepor complexes. In future work it will therefore be crucial to characterize the heteroreceptor complexes with FRET/BRET and proximity ligation assay and their allosteric receptor-receptor interactions with radioligand binding assays and their signaling pathways with luciferase reporter gene assays. The receptor interface is usually different between a receptor heteromer and the corresponding homomers. Therefore, it becomes possible to selectively interfere with formation of a receptor heteromer and thus its function vs. corresponding homomers through use of interface interacting peptides (Borroto-Escuela et al., 2010b, 2018).

Also a scheme was introduced how alpha-synuclein monomers/oligomers/ synuclein fibrils can modulate the adenosine A2AR homo-heteroreceptor panorama in the plasma membrane through intramembrane and intracellular interactions enhancing neurodegeneration (Figure 2). They can exist both in neurons, astroglia and microglia. The role of putative A2AR-D2R-NMDAR and A2AR-mGluR5-NMDAR heteroreceptor complexes were inter alia discussed in relation to glutamate synapses and excitotoxicity. Future work is necessary to test the impact of these adenosine A2AR-containing complexes on A2AR mediated neurodegeneration, especially when mediated by alpha-synucleins.

We confirm and declare that all authors meet the criteria for authorship according to the ICMJE, including approval of the final manuscript, and they take public responsibility for the work and have full confidence in the accuracy and integrity of the work of other group authors. They have substantially contributed to the conception or design of the work. Also they have also helped revising it critically for important intellectual content; and final approval of the version to be published. In addition, they have contributed in this last version of the manuscript in writing assistance, technical editing, language editing, and proofreading.