Functional histamine H₃ and adenosine A_2A receptor heteromers in recombinant cells and rat striatum

Abstract

In the striatum, histamine H₃ receptors (H₃Rs) are co-expressed with adenosine A_2A receptors (A_2ARs) in the cortico-striatal glutamatergic afferents and the GABAergic medium-sized spiny neurons that originate from the indirect pathway of the basal ganglia. This location allows H₃Rs and A_2ARs to regulate the striatal GABAergic and glutamatergic transmission. However, whether these receptors can physically interact has not yet been assessed. To test this hypothesis, a heteromer-selective in vitro assay was used to detect functional complementation between a chimeric A_2AR₃₀₂-Gα_qi4 and wild-type H₃Rs in transfected HEK-293 cells. H₃R activation with the agonist RAMH resulted in Ca²⁺ mobilization (pEC₅₀ 7.31 ± 0.23; maximal stimulation, Emax 449 ± 25% of basal) indicative of receptor heterodimerization. Functional H₃R-A_2AR heteromers were confirmed by co-immunoprecipitation and observations of differential cAMP signaling when both receptors were co-expressed in the same cell. In membranes from rat striatal synaptosomes, H₃R activation decreased A_2AR affinity for the agonist CGS-21680 (pKi values 8.10 ± 0.04 and 7.70 ± 0.04). Moreover, H₃Rs and A_2ARs co-immunoprecipitated in protein extracts from striatal synaptosomes. These results support the existence of a H₃R-A_2AR heteromer with possible physiological implications for the modulation of the intra-striatal transmission.

1. Introduction

The actions of histamine in the periphery on airway constriction, inflammation and gastric acid secretion are well known, whereas its function in the Central Nervous System (CNS) is not yet fully understood. In mammals, histamine actions are mediated by four (H₁R-H₄R) G protein-coupled receptors (GPCRs), of which the H₃ receptor (H₃R) displays the highest expression in the brain. In the CNS, histamine is released from the terminals and dendrites of neurons located in the hypothalamic tuberomammilary nucleus, from where they innervate most of the CNS, including nuclei belonging to the basal ganglia, a subcortical group of structures intimately related to the control of motor behavior (Panula and Nuutinen, 2013; Panula et al., 2015).

The striatum is the main input nucleus of the basal ganglia and integrates motor and sensory information. The primary striatal afferents are glutamatergic axons originating from neurons located in the cerebral cortex and thalamus, and dopaminergic axons of sustantia nigra pars compacta neurons (Bolam et al., 2000). In turn, the axons of the GABAergic medium-sized spiny neurons (MSNs), which account for more than 90% of the striatal neuronal population (Kemp and Powell, 1971), project to the globus pallidus and sustantia nigra pars reticulata (Bolam et al., 2000).

H₃Rs are abundantly expressed in the striatum either as pre-synaptic auto- and hetero-receptors or as post-synaptic receptors, with the latter located on the bodies of the MSNs and GABAergic or cholinergic interneurons (Pillot et al., 2002; González-Sepúlveda et al., 2013; Bolam and Ellender, 2016). The understanding of the physiological role of striatal H₃Rs is complicated by their ubiquitous expression and capability to engage in protein-protein interactions with other GPCRs (Nieto-Alamilla et al., 2016). Of particular interest in the striatum are dopamine and adenosine receptors, which are also highly expressed. Based on the selective expression of dopamine and adenosine receptor subtypes, MSNs can be segregated into two neuronal populations. Neurons that originate the basal ganglia direct pathway (dMSNs) are identified by the expression of D₁-like receptors (D₁Rs) whereas the indirect pathway population (iMSNs) is formed by neurons expressing adenosine A_2A receptors (A_2ARs) and D₂-like receptors (D₂Rs) (Schiffman et al., 1991; Ferre et al., 1997; Tapper et al., 2004). In the striatum, H₃Rs have been shown to form heterodimeric protein-protein complexes with dopamine D₂ and D₁ receptors (Ellenbroek, 2013), and heterotrimers with σ1/D₁ receptors and NMDA/D₁ receptors (Moreno et al., 2014; Rodríguez-Ruíz et al., 2016). In turn, A_2ARs can form dimers with D₂, cannabinoid CB₁ and adenosine A₁ receptors (Ciruela et al., 2006; Carriba et al., 2007; Hillon et al., 2002) and oligomers with D₂ and mGlu5 glutamate receptors (Cabello et al., 2009).

Both H₃ and A_2A receptors modulate intra-striatal synaptic transmission. The activation of Gα_i/o-coupled H₃Rs results in inhibition of GABA, dopamine, acetylcholine and glutamate release (Doreulee et al., 2001; Schlicker et al., 1993; Prast et al., 1999; Arias-Montaño et al., 2001), whereas the activation of Gα_s-coupled A_2ARs also inhibits GABA release but facilitates glutamatergic cortico-striatal transmission (Kirk and Richardson, 1994; Popoli et al., 1995).

Either individually or through their heteromeric interactions with other GPCRs, A_2ARs and H₃Rs have been proposed as targets for the treatment of basal ganglia-related disorders such as Parkinson’s disease and addiction (Ferré et al., 2004; Schwarzschild et al., 2006; Passani and Blandina, 2011; Ellenbroek, 2013). In this work, we hypothesized that striatal H₃Rs and A_2ARs can not only form heteromers with dopamine receptors, but also with each other, and that this interaction forms a unique pharmacological target that could be of potential therapeutic interest. Here we show that H₃Rs co-immunoprecipitate with A_2ARs in HEK293 cells, suggestive of oligomerization. Further, we studied the physical properties of the A_2AR-H₃R interaction by a functional complementation assay. We next showed that the A_2AR-H₃R heteromerization has unique functional pharmacology in terms of cAMP modulation in HEK-293 cells co-expressing the receptors. Using protein extracts from striatal synaptosomes, we confirmed the presence of A_2AR-H₃R heteromers in vivo by co-immunoprecipitation. Finally, we found that in the same preparation H₃R activation resulted in decreased binding affinity of A_2AR for the agonist CGS-21680.

A preliminary account of this work was presented in the abstract form to the European Histamine Research Society (Márquez-Gómez et al., 2016) and submitted to an electronic pre-print server (Márquez-Gómez et al., 2017).

2. Methods

2.1. Materials

The following drugs and reagents were purchased from Sigma Aldrich (St. Louis, MO, USA): (R)(−)-α-methylhistamine dihydrochloride, histamine dihydrochloride, adenosine deaminase (from bovine spleen), Percoll, quinpirole dihydrochloride. Dimaprit was from Axon MedChem (Reston, VA, USA). CGS-21680 was from Cayman Chemical (Ann Arbor, MI, USA). N-α-[methyl-³H]-histamine (78.3 Ci·mmol⁻¹) and CGS-21680-[carboxyethyl-³H (N)]-(35.2 Ci·mmol⁻¹) were from Perkin Elmer (Boston, MA, USA).

2.2. Molecular cloning

Truncated A_2A or H₃ receptors were generated by PCR using the primers indicated in Supplementary Table 1. After amplification and restriction with HindIII and BamHI enzymes, the truncated A_2A and H₃ receptors were ligated to a pcDNA3.1 plasmid that contained the chimeric Gα_qs4 and Gα_qi4 proteins. Both A_2ARs and H₃Rs (cDNA Resource Center, Bloomsberg, PA, USA) were labeled with a triple hemagglutinin tag (3xHA). To generate the 3xHA tagged receptors, A_2AR and the H₃R were amplified from nucleotide 377 and 458, respectively, towards their amino terminus. The 3xHA tag was amplified from a plasmid that codified for the tagged histamine H₄R (3xHA-H₄R in pcDNA3.1, cDNA Resource Center). The amplified DNA fragments (3xHA-A_2AR377/H₃R458) where then re-introduced into the pcDNA3.1-H₃R or pcDNA3.1-A_2AR backbone to obtain the tagged, full length receptors. The insertion and orientation of the amplified fragment was verified by automated sequencing performed at FESI-UNAM (Los Reyes Iztacala, Estado de México, México).

2.3. Cell culture and transfection

HEK-293T cells (American Type Culture Collection, Manassas, VA, USA) were grown in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), penicillin (50 UI/ml) and streptomycin (0.1 mg/ml) under a humidified atmosphere (5% CO₂ in air) at 37 °C.

For transfections, cells were seeded in 6-well plates (6×10⁵ cells/well; up to passage 15) and incubated for 24 h at 37 °C in a 5% CO₂/air mixture. The next day, 10 μl X-tremeGENE (Roche, Basilea, Switzerland) were mixed with 500 μl Optimem (Life Technologies, San Diego, CA, USA) and the mixture was incubated for 5 min at room temperature before the addition of cDNA in a 1:5 ratio with X-tremeGENE. For cAMP assays a mixture of DNA:Glosensor cAMP plasmid (1:2 ratio) was added to the Optimem/X-tremeGENE solution. When plasmids containing the A_2AR or H₃R were transfected alone, the amount of DNA was preserved by adding empty pcDNA3.1 vector. The transfection mixture was incubated for 20 min at room temperature and then added to the cells, which were incubated for 24 h at 37 °C under a humidified atmosphere (5% CO₂/air).

For co-immunoprecipitation assays, HEK-293T cells, grown in 100-millimeter Petri dishes and at 80% confluence, were transfected using the Lipofectamine 2000 method. Briefly, a mixture of DNA/Lipofectamine 2000 (1:2 ratio) was incubated for 20 min at room temperature before being added to the cells. After incubation for 5 h at 37 °C under a humidified atmosphere (5% CO₂/air), the medium was replaced by high-glucose DMEM supplemented with 10% FBS and 1% antibiotic/antimycotic. Incubation continued for a further 24 h under the same condition.

2.4. cAMP accumulation assay

cAMP assays were performed as described in detail elsewhere (Chiang et al., 2016). Briefly, HEK-293T cells transfected with 3xHA-A_2AR, 3xHA-H₃R or a mixture of both plasmids (1 μg each) together with Glosensor cAMP plasmid (Promega, Madison, WI, USA) were seeded in a white 384-well low-volume plate (25,000 cells/well in 7.5 μl medium) and incubated with Glo-equilibrium medium (7.5 μl, Promega). Drugs under test were added in a 5 μl volume (4x in HBSS solution) and endogenous cAMP luminescence was measured in real-time in a Flexstation 3 apparatus (Molecular Devices, Sunnyvale, CA, USA).

2.5. Calcium mobilization assay

HEK-293T cells transfected with 1 μg of DNA were seeded (2.5×10⁴ cells/well) in a 384-well clear bottom black plate (25 μl-volume) and incubated for 24 h at 37 °C in a humidified 5% CO₂/air atmosphere, covered with Areaseal film (Sigma, St. Louis, MO, USA). Cells were then loaded with 25 μl of the FLIPR Ca²⁺ dye (Molecular Devices, Sunnyvale, CA, USA) and incubated for 1 h at 37 °C. Agonists were added in a 20 μl volume and calcium mobilization was measured in real-time for 2 min in a Flexstation 3 apparatus (Molecular Devices, Sunnyvale, CA, USA) and analyzed as described in detail in van Rijn et al. (2013).

2.6. Synaptosome preparation

All procedures were approved and controlled by the Cinvestav Animal Care Committee and were in accord with the rules issued by the National Institutes of Health (NIH Publications No. 8023) and the Mexican Council for Animal Care. Wistar rats (males, 250–300 g, provided by Unidad de Producción y Experimentación para Animales de Laboratorio; Cinvestav, Mexico City) were decapitated, the brain was quickly removed from the skull and the forebrain was separated and deposited on a metal plate placed on ice. The striata from 3–5 animals were dissected using forceps and the whole tissue was placed in 5 ml 0.32 M sucrose solution containing 10 mM Hepes, 1 mg/ml bovine serum albumin and 1 mM EDTA (pH 7.4 with NaOH). The tissue was homogenized using 10 strokes of a hand-held homogenizer (400 rpm), the homogenate was centrifuged (1000xg, 10 min, 4 °C) and the supernatant was pelleted at 14,000xg (12 min, 4 °C). The pellet was re-suspended in 5 ml of a Percoll solution (45 %, v:v) in Krebs-Henseleit-Ringer buffer (in mM: NaCl 140, Hepes 10, D-glucose 5, KCl 4.7, EDTA 1, pH 7.3 with NaOH). After centrifugation (2 min, 14,000xg, 4 °C), the upper phase was collected and brought up to 20 ml with Krebs-Ringer-Hepes (KRH) solution (in mM: NaCl 113, NaHCO₃ 25, Hepes 20, D-glucose 15, KCl 4.7, CaCl₂ 1.8, MgCl₂ 1.2, KH₂PO₄ 1.2, pH 7.4 with NaOH). The suspension was centrifuged (20,000xg, 20 min, 4 °C) and the pellet (synaptosomes) was re-suspended in KRH solution unless otherwise indicated.

2.7. Electron microscopy

Striatal synaptosomes were isolated by the Percoll method as above. Sample preparation and electron microscopy were performed as described in detail elsewhere (Morales-Figueroa et al., 2015).

2.8. Radioligand binding assays with striatal membranes

Striatal synaptosomes were re-suspended in lysis solution (Tris-HCl 10 mM, EGTA 1 mM, pH 7.4) and incubated for 20 min at 4 °C before centrifugation (20 min, 20,000xg, 4 °C). The pellet (synaptosomal membranes) was re-suspended in 1 ml KRH solution containing adenosine deaminase (2 U/ml). After incubation for 30 min at 37 °C, the suspension was brought to 20 ml with KRH solution and centrifuged (20 min, 20,000xg, 4 °C). The membranes were re-suspended in incubation solution (Tris-HCl 50 mM, MgCl₂ 5 mM, pH 7.4) and 130 μl aliquots (40 μg protein) were incubated with 10 μl of increasing concentrations of CGS-21680 or RAMH (20x) and 50 μl of [³H]-NMHA (8 nM) or [³H]-CGS-21680 (48 nM). After 1 h at 30 °C ([³H]-NMHA) or 2 h at 25°C ([³H]-CGS-21680), incubations were stopped by rapid filtration through Whatman GF/B filters pre-soaked (2 h) in 0.3 % polyethylenimine (PEI). Filters were washed 3 times with 1 ml ice-cold buffer solution (50 mM Tris-HCl, pH 7.4), soaked in 4 ml scintillation solution and the tritium content was determined by scintillation counting.

2.9. Co-immunoprecipitation assays

Striatal synaptosomes were obtained as described above and then re-suspended in 1 ml of lysis solution (Tris-HCl 50 mM, NaCl 150 mM, Triton X-100 1%, SDS 0.05%, protease inhibitor 1 μl/ml). HEK-293T cells were dislodged in RIPA solution. For both synaptosomal and cell samples, protein was extracted by sonication (3 cycles, 30 sec, 8 kHz). The sample was centrifuged (20 min, 6,000xg) and the protein extract (supernatant) was collected. Nonspecific binding was removed by incubation (1 h, 4 °C) with 10 μl of AG beads (Santa Cruz Biotechnology; Dallas, TX, USA) under rotatory rocking. The beads were pelleted (2 min, 6000xg) and the supernatant was used for the assay. Protein quantification was performed by the BCA method.

The protein extracts (500 μg protein from striatal synaptosomes and 200 μg from HEK-293T) were incubated with 2 μg of the primary antibodies (anti-A_2AR Abcam, cat. ab3461, lot. GR238882–9; anti-H₃R, Abcam, cat. ab84468, lot. GR27494–1, in a 1:100 dilution) together with AG beads (Santa Cruz Biotechnology sc-2003, 1:5 antibody to beads ratio) for 16 h at 4°C with constant rotational rocking. Antibodies against the CD-81 protein (Santa Cruz Biotechnology, cat. sc-70803, lot. B0609) or D₂R (Santa Cruz Biotechnology, cat. sc-5303, lot. A03013) were used as negative controls for the striatal synaptosomes or HEK-293T cells, respectively. The bead-antibody complex was pelleted (2 min, 6000xg) and a 30 μl aliquot of the supernatant was used as a load control. In both protein extracts, 30 μl of the total protein was used as input. The complex was dissociated for 60 min at 48°C in loading buffer (10% β-mercaptoethanol and 50% Laemmli buffer in H₂O) and separated by electrophoresis in a 10% SDS-polyacrylamide gel (20 min at 80 V and then 65 min at 120 V). Semi-dry transfer was performed at 15 V for 95 min. Membranes were blocked with 5% nonfat dry milk diluted in TBS-Twin 0.05% solution (overnight, 4°C). After extensive washing, 5 ml of TBS-Twin 0.05% solution containing 5% BSA and the blotting antibody (anti-A_2AR, anti-HA, Cell Signaling, cat. C29F4, lot. 3724S or anti-H₃R, in a 1:1000 dilution) were added to the membrane and incubated overnight at 4 °C. Membranes were incubated with the secondary antibody (Invitrogen, HRP-conjugated anti-rabbit IgG, cat. 65–6120, 1:5000, non-fat dry milk 5% in TBS-Twin 0.05%) at room temperature for 2 h. For the loading controls membranes were incubated for 1 h with antibodies directed to β-tubulin (Invitrogen, cat. 322600, lot. 1235662) or α-actin (Sigma, cat. A5228, lot. 128K4843) in a 1:5000 dilution in TBST 0.05%. After washing, membranes were incubated for 1 h with the secondary antibody (α-mouse, Santa Cruz, cat. sc-2005) diluted in TBST 0.05%. Blot images were obtained by chemiluminescence in an X-ray film (Kodak).

2.10. Statistical analysis

Data show the mean ± standard error of the mean (SEM). Statistical analysis was performed with student t-test with Prisma GraphPad software, version 5.0 (San Diego, CA, USA).

3. Results

3.1. Co-immunoprecipitation of A_2ARs and H₃Rs expressed in HEK-293 cells

We first studied by co-immunoprecipitation whether H₃Rs and A_2ARs interact physically upon co-transfection in HEK-293T cells. 3xHA-H₃Rs were detected after immunoprecipitation of wild type (WT) A_2ARs (Figure 1A). HEK-293T cells lack expression of the D₂ receptor, and an antibody targeting this receptor was therefore used as a negative control, yielding no signal. This result supports dimerization between H₃Rs and A_2ARs.

Figure 1.

Adenosine A_2A and histamine H₃ receptors co-immunoprecipitate in transfected HEK-293 cells. A. Co-immunoprecipitation of the hemagglutinin-tagged H₃R (3xHA-H₃R) with the A_2AR in protein extracts from HEK-293 cells. The ~45 kDa band corresponds to the 3xHA-H₃R, and was not observed with the negative control. Input represents 30% of the total protein. An antibody against the D₂ receptor (α-D₂R) was used as a negative control. The blot is representative of 3 independent experiments. The whole blot is shown in Supplementary Figure 5.

3.2. Functional complementation of A_2ARs and H₃Rs expressed in HEK-293 cells

To further study the potential interaction between A_2ARs and H₃Rs in a recombinant cell system, we employed a functional complementation assay, previously used to study physical interactions between GPCRs (Han et al., 2009; van Rijn et al., 2013). This assay relies on the fusion of chimeric Gα_qs4 or Gα_qi4 proteins to the truncated C-terminal tail of a GPCR to generate a nonfunctional receptor, rescuable through homo- or hetero-dimerization with a WT, untruncated receptor. It was previously reported that fusion of a chimeric G protein to a GPCR truncated near helix 8 produces a receptor that is unable to signal by itself but can be rescued when transfected with a full-length receptor (van Rijn et al., 2013). Therefore, we first generated an A_2AR truncated to 302 residues (A_2AR₃₀₂) and H₃Rs of 427, 421 and 411 residues (H₃R₄₂₇, H₃R₄₂₁ and H₃R₄₁₁), and the truncated receptors were then fused to the chimeric G proteins (Figure 2A).

Figure 2.

Study of the possible A_2AR-H₃R dimerization by functional complementation assay. A. Basis of the assay. The G protein-coupled receptor (GPCR; grey cylinders) is truncated at its carboxyl terminus to generate a nonfunctional GPCR, which is then fused to the chimeric G protein (green hexagons). The chimeric G protein is formed by a Gα_q protein in which 9 or 10 residues of the C-terminus were substituted by the corresponding sequence of a Gα_s4 or Gα_i4 protein (represented by X). Under these conditions, Ca²⁺ mobilization can only be elicited when the complex GPCR-Gα_qx is in close proximity to a wild type (WT) GPCR, either the same (homodimerization) or a different receptor (heterodimerization). B. Activation of the H₃R with its agonist RAMH resulted in Ca²⁺ mobilization only when it was co-expressed with the chimeric Gα_qi4 but not when transfected alone. C. When activated with its agonist CGS-21680, the A_2AR was uncapable to induce Ca²⁺ mobilization either alone or when co-transfected with the chimeric Gα_qs4 protein. D. Functional complementation by homodimerization of the truncated A_2AR₃₀₂, either bound to Gα_qs4 (A_2AR₃₀₂-Gα_qs4) or Gα_qi4 (A_2AR₃₀₂-Gα_qi4), with the native A_2AR was not observed. E. Activation of the truncated H₃R₄₂₇ bound to Gα_qi4 (H₃R₄₂₇-Gα_qi4) induced Ca²⁺ mobilization on its own and homodimerization with the WT-H₃R did not modify the response. No signal was observed with activation of the H₃R₄₂₇-Gα_qs4 construct when it was co-expressed with the WT-H₃R. In all graphs data are means ± SEM from 3 replicates from a representative experiment. Where SEM bars are not visible, they are smaller than the symbol size. The quantitative analysis is shown in Tables 1 and Supplementary Table 2.

Activation of the H₃R in transfected CHO-K1 cells and rat striatal neurons in primary culture induces Ca²⁺ mobilization (Cogé et al., 2001; Rivera-Ramirez et al., 2016), and in the striatum A_2AR activation favors Ca²⁺ entry by modulating voltage-activated Ca²⁺ channels (Kirk and Richardson, 1995; Gubitz et al., 1996), which are endogenously expressed by HEK-293 cells (Berjukow et al., 1996; Thomas and Smart, 2005). In HEK-293 cells, the H₃R selective agonist RAMH did not induce any discernible Ca²⁺ response but did so when the H₃R was co-transfected with Gα_qi4 proteins (Figure 2B). The A_2AR did not induce Ca²⁺ signaling when activated by the selective agonist CGS-21680, but also no response was observed when co-transfected with Gα_qs4 proteins, suggesting that A_2ARs do not activate this chimeric protein (Figure 2C). It has been reported that not all GPCRs are amenable to signal through chimeric G-proteins (Conklin et al., 1996). A different Gα_s-coupled receptor, the histamine H₂ receptor, was capable to induce Ca²⁺ release when co-transfected with Gα_qs4 proteins and stimulated with the selective agonist dimaprit (Supplementary Figure 1A), discarding Gα_qs4 malfunction. Consistent with the finding that A_2ARs did not signal efficiently via chimeric G-proteins, functional complementation by homodimerization of A_2AR₃₀₂-Gα_qs4 or A_2AR₃₀₂-Gα_qi4 with the full-length A_2AR was not observed (Figure 2D).

The C-tail of the H₃R was truncated such that the H₃R-Gα_qi4 fusion protein would display limited Ca²⁺ mobilization when expressed alone, but a pronounced Ca²⁺ signaling when co-expressed with WT-H₃Rs. The H₃R₄₂₁-Gα_qi4 and the H₃R₄₁₁-Gα_qi4 constructs did not induce Ca²⁺ mobilization, and could not be rescued via homo-dimerization with WT-H₃Rs (Supplementary Figure 1B). However, Ca²⁺ mobilization induced by H₃R₄₂₇-Gα_qi4 upon activation and co-transfection with WT-H₃Rs produced a similar increase in calcium release (Figure 2E). While not optimal, this behavior allowed for the study of the interaction between the H₃R₄₂₇-Gα_qi4 and the A_2AR as described below.

Co-transfection of the ‘inert’ A_2AR₃₀₂-Gα_qi4 and H₃Rs induced Ca²⁺ mobilization after activation of the latter receptors with RAMH, suggestive of close physical proximity between the H₃R and both the A_2AR and the fused Gα_qi4-protein. H₃R activation failed to induce functional complementation in cells co-transfected with A_2AR₃₀₂-Gα_qs4 (Figure 3A). Given that H₃Rs are Gα_i/o-coupled, and the inability of the A_2AR to signal through chimeric Gα_qi4- or Gα_qs4-proteins, it was not surprising that A_2AR activation with the selective agonist CGS-21680 did not produce any response through H₃R₄₂₇-Gα_qs4 or H₃R₄₂₇-Gα_qi4 (Figure 3B). Interestingly, the co-expression of WT-A_2ARs prevented RAMH-induced Ca²⁺ mobilization mediated by the H₃R₄₂₇-Gα_qi4 (compare Figure 2E and Figure 3C). We briefly explored this pharmacological response further by analyzing Ca²⁺ mobilization in HEK-293T cells co-expressing the H₃R₄₂₇-Gα_qi4 and the A_2AR or the A_2AR₃₀₂-Gα_qi4 and the H₃R and activating both receptors, but did not observe any additive or antagonistic effect (Supplementary Figure 1C and 1D).

Figure 3.

Ligand modulation of the A_2AR-H₃R interaction in HEK-293 cells. Activation of the WT-H₃R with its agonist RAMH led to functional complementation when co-expressed with the A_2AR₃₀₂-Gα_qi4 but not with A_2AR₃₀₂-Gα_qs4. B. Ca²⁺ mobilization was not observed when the WT-A_2AR was co-transfected with the truncated H₃R₄₂₇-Gα_qs4 or H₃R₄₂₇-Gα_qi4, in accord with the incapability of the receptor to activate the chimeric proteins. C. As shown in Figure 1D, activation of the H₃R₄₂₇-Gα_qi4 resulted in Ca²⁺ mobilization, and this response was prevented when it was co-expressed with the WT-A_2AR, suggesting a preference of the H₃R₄₂₇-Gα_qi4 to form heterodimers. D. The well-studied A_2AR-D₂R heterodimer was used as a positive control for these experiments. Activation of the D₂R with increasing concentrations of the agonist quinpirole caused marked Ca²⁺ mobilization only when co-expressed with the chimeric A_2AR₃₀₂-Gα_qi4. E. The histamine H₄ receptor (H₄R) as a negative control. No functional complementation was observed when the receptor was co-expressed with the A_2AR₃₀₂-Gα_qi4, showing the specificity of the A_2AR-H₃R interaction. F. Functional complementation between the A_2AR₃₀₂-Gα_qi4 and the WT-H₃R was not observed when the latter receptor was activated with the endogenous agonist histamine. For all graphs data are means ± SEM from 3 replicates from representative experiments. Where SEM bars are not visible, they are smaller than the symbol size. The quantitative analysis is shown in Table 1 and Supplementary Table 2.

A_2ARs have been shown to form heteromers with D₂Rs (Ferré et al., 2003) and robust functional complementation was accordingly observed when A_2AR-Gα_qi4 was co-expressed with WT-D₂Rs (Figure 3D), supporting that our results are due to heterodimerization and not to stochastic interactions.

3.3. The putative A_2AR-H₃R heteromer displays ligand bias

To test for the selective nature of the A_2AR-H₃R interaction in the functional complementation assay, we investigated if A_2AR-Gα_qi4 would also allow the structurally and physiologically similar histamine H₄ receptor (H₄R) to induce Ca²⁺ signaling. Co-activation with histamine of H₃Rs and H₄Rs elicited Ca²⁺ mobilization only when co-transfected with Gα_qi4 (Supplementary Figure 2A and 2B, respectively). Neither histamine nor RAMH induced Ca²⁺ mobilization when the H₄R was co-expressed with A_2AR₃₀₂-Gα_qi4, supporting the specificity of the A_2AR-H₃R interaction (Figure 3E). Unexpectedly, whereas RAMH induced Ca²⁺ mobilization when the H₃R was co-expressed with A_2AR₃₀₂-Gα_qi4, histamine did not (compare Figure 3A with Figure 3F). This result suggests that the A_2AR-H₃R heteromer displays agonist bias.

3.4. A_2AR-mediated signaling is increased by H₃R co-activation

To study the pharmacology of the putative A_2AR-H₃R heteromer we next tested cAMP signaling using WT receptors in HEK-293T cells. In cells transfected with the A_2AR, incubation with CGS-21680 resulted in a concentration-dependent increase in cAMP levels in accordance with the receptor’s coupling to the Gα_s signaling pathway (Figure 3A). CGS-21680 induced a similar cAMP response in HEK293T-A_2AR cells co-expressing H₃Rs (Figure 4A). Co-activation of the receptors resulted in a decrease in baseline, in agreement with the Gα_i/o coupling of the H₃R, but led to an augmentation of A_2AR-mediated cAMP formation as noticed by a 2.5-fold change of baseline (versus 1.5-fold in control). This result suggests that the A_2AR signaling becomes more efficacious when heterodimerization with the H₃R occurs (Figure 4A, Supplementary Figure 2C, and Table 2). The increase in the A_2AR signaling was not observed in the absence of RAMH (1.5-fold of baseline) suggesting an agonist-dependency of the response (Figure 4A and Supplementary Figure 2C).

Figure 4.

H₃R activation enhances A_2AR-mediated cAMP signaling in HEK-293 cells. A. Activation of the Gα_s-coupled A_2AR with its agonist CGS-21680 induced cAMP formation and H₃R co-activation enhanced A_2AR efficacy. Co-expression of the H₃R did not modified the A_2AR functional response. Values for pEC₅₀ and maximal effect (Emax) are given in Table 2. B. H₃R activation with RAMH decreased forskolin-induced cAMP formation in accord with the inhibitory nature of the Gα_i/o-coupled receptor. A_2AR expression and receptors co-activation lead to a change in the H₃R signaling. Data are means ± SEM from 5 replicates from a representative experiments. Where SEM bars are not visible, they are smaller than the symbol size. Values for pIC₅₀ and maximal effect (Imax) are given in Table 2.

Table 2.

Pharmacological characteristics of the cAMP formation assay in transfected HEK-293 cells.

Transfection	Agonist	pEC50	Emax (%)	pIC50	Imax (%)
A2AR	CGS	8.0 ± 0.4	152 ± 7	-	-
A2AR + H3R	CGS	8.7 ± 0.5	153 ± 7	-	-
A2AR + H3R	+RAMH	8.0 ± 0.2	248 ± 9b	-	-
H3R	RAMH	-	-	8.1 ± 0.3	−26 ± 2
H3R ± A2AR	RAMH	10.6 ± 1	114 ± 3	-	-
H3R ± A2AR	+CGS	9.9 ± 0.6	120 ± 2	-	-

Data are means ± SEM from 3 experiments. Statistical comparisons were performed between two agonists for the same transfection, or between the transfection with the chimeric receptor and the transfection of the chimeric receptor plus the wild type receptor. ns, no significant difference

^aP < 0.05

^bP < 0.01, Student’s t test.

Both RAMH and CGS-21680 were assayed at 100 nM. RAMH, R-α-methylhistamine; CGS, CGS-21680.

In HEK293T-H₃R cells, RAMH inhibited forskolin-stimulated cAMP formation congruent with the Gα_i/o coupling of the H₃R receptor. However, when the A_2AR was co-transfected, RAMH-mediated H₃R signaling shifted to facilitate cAMP formation instead. Activation of the A_2AR caused an expected increase in the baseline but did not significantly modify the effect on cAMP formation. This result suggests that in the heterodimer, A_2AR signaling prevails over the H₃R signaling (Figure 4B and Supplementary Figure 2D; Table 2).

3.5. H₃R activation decreases A_2AR binding affinity in synaptosomal membranes

The identification of A_2AR-H₃R heteromers was performed in recombinant cell systems over-expressing the receptors. As previously mentioned A_2ARs and H₃Rs are co-expressed in iMSNs and cortico-striatal projections. We therefore used rat striatal synaptosomes to seek for pharmacological traces that supported the existence of native A_2AR-H₃R dimers.

Electron microscopy confirmed in the purified synaptosomal preparation the vast presence and conserved structure of isolated nerve terminals, characterized by a delimited membrane and the presence of mitochondria and synaptic vesicles (Supplementary Figure 3A and B). In protein extracts from striatal synaptosomes, immunoprecipitation of the A_2AR resulted in a band of ~45 kDa, corresponding to the expected migration of the H₃R. No signal was detected when an irrelevant antibody (α-CD81) was tested. As a positive control the H₃R was immunoprecipitated and detected as a band of ~45 kDa (Figure 5A). When the reverse approach was employed and the H₃R was immunoprecipitated from the striatal protein extracts, a band of ~45 kDa corresponding to the expected migration of the A_2AR was observed. This band was also observed when the A_2AR was immunoprecipitated as a positive control but not detected in the negative control (Figure 5B). This result supports that the interaction A_2AR-H₃R is constitutively present in the striatal nerve terminals.

Figure 5.

Co-immunoprecipitation of the A_2A and H₃ receptors in striatal synaptosomes. A. The H₃R co-immunoprecipitated with the A_2AR in a protein extract of Percoll-purified striatal synaptosomes. The band of ~45 kDa corresponds to the expected migration of the H₃R. No band was observed in the negative control (α-CD81). B. Co-immunoprecipitation of the A_2AR with the H₃R in a protein extract of striatal synaptosomes. The band of ~45 kDa corresponds to the expected migration of the A_2AR. The figure depicts representative blots, repeated a further 4 times with similar results. Full blots are shown in Supplementary Figure 6.

In binding studies with synaptosomal membranes the A_2AR agonist CGS-21680 (10 and 100 nM) did not affect the affinity of the H₃R for its agonist RAMH (Figure 6A), whereas RAMH (100 nM) decreased by two-fold the affinity of the A_2AR for CGS-21680 (Figure 6B, 6C and Table 3), with no effect on maximal binding (Bmax). This pharmacology appears to be specific for pre-synaptic H₃Rs and A_2ARs, because we did not observe a similar decrease in A_2AR affinity for CGS-21680 in membranes from the whole striatum, in which post-synaptic membranes constitute the major component (Figure 6D and Table 3).

Figure 6.

H₃R activation decreases A_2AR affinity for the agonist CGS-21680 in synaptosomal membranes but not in membranes from the whole striatum A. CGS-21680 (10 and 100 nM) did not modify the H₃R affinity for its ligand RAMH in membranes isolated from striatal synaptosomes. B. In the same preparation, the H₃R agonist RAMH (100 nM) decreased the A_2AR affinity for [³H]-CGS-21680. Values are means ± SEM from 3 replicates from a representative experiment. C. Analysis of 3 independent experiments. The statistical analysis was performed with paired Student’s t test. D. H₃R activation with RAMH (100 nM) failed to decreased the A_2AR affinity for [³H]-CGS-21680 in membranes from the whole striatum. Values are means ± SEM from 3 replicates from a representative experiment. Where SEM bars are not visible, they are smaller than the symbol size.

Table 3.

Analysis of binding assays in rat striatal membranes

	pKi	Bmax (%)
[3H]-NAMH
Synaptosomes
RAMH	9.09 ± 0.22	100.0 ± 0.3
RAMH + CGS 10 nM	8.99 ± 0.02ns	110.0 ± 4.0ns
RAMH + CGS 100 nM	9.08 ± 0.06ns	103.0 ± 4.1ns
[3H]CGS-21680
Whole striatum
CGS	8.06 ± 0.15	100.0 ± 14
CGS + RAMH	8.11 ± 0.04ns	88.0 ± 6.0ns
Synaptosomes
CGS	8.10 ± 0.04	100.0 ± 9.0
CGS + RAMH	7.70 ± 0.04a	88.1 ± 10.0ns

Data are means ± SEM from 3–5 experiments. For [³H]-NAMH binding assays the statistical analysis was performed with one-way Anova and Dunnett´s post hoc test. For [³H]-CGS-21680 binding, values in the presence of RAMH were compared with the corresponding control with Student’s t test.

^aP < 0.001, ns, not significant. CGS, CGS-21680. RAMH (R-α-methylhistamine) was tested at 100 nM in all the experiments.

4. Discussion

The H₃R has been proposed as a potential novel drug target for the treatment of drug use disorders, depression, schizophrenia, and Parkinson’s disease. However, its wide brain expression and effects on other neurotransmitter systems may result in adverse effects when H₃R selective drugs are administered systemically. The A_2AR has also been proposed as an option for the treatment of Parkinson’s disease and possesses a strategic distribution in the striatum for targeting and modulating the cortico-dMSNs projections and the activity of iMSNs. The results presented herein may lead to an alternative to specifically target H₃Rs located in either iMSNs or cortical afferents synapsing onto dMSNs projections.

4.1. A_2AR-H₃R interaction

The primary finding of this study was the identification, for the first time, of A_2AR-H₃R heteromers, not only in recombinant cell systems but also in rat striatal nerve terminals.

The basis of the functional complementation assays requires a receptor to be nonfunctional and this was obtained by truncation of the C-terminus. Herein we showed that truncation of the H₃R from 445 to 411 or 421 residues was sufficient to prevent H₃R-mediated G protein activation of the H₃R-Gα_qi4 fusion protein, yet we were unable to functionally rescue its signaling, indicating that the remaining C-tail may be too short to connect with an interacting GPCR. However, truncation to 427 amino acids allowed the H₃R to remain functional.

Exposure of HEK-293T cells transfected with H₃R₄₂₇-Gα_qi4 to the H₃R agonist RAMH resulted in Ca²⁺ mobilization (Figure 1E). As mentioned in the Results section, this response was not expected. In an optimal scenario, the receptor truncation would have prevented or diminished the receptor capability to activate G proteins and the H₃R₄₂₇-Gα_qi4 function was expected to be either complemented or increased by the co-expression and co-activation with the WT-H₃R. However, these conditions allowed for the detection of the loss of H₃R function with the solely presence of the A_2AR, suggesting a preference of the H₃R to form hetero-dimers over homo-dimers. It may be possible that truncating the H₃R at the amino acids 421–427 will result in a receptor short enough to abolish the homo/monomeric signaling, but long enough to functionally complement with a full length GPCR.

In order for a GPCR to activate G proteins, the integrity of helix 8 appears to be required. This structural requirement has been demonstrated for D₁Rs and opioid κ and μ receptors (van Rijn et al., 2013). The H₃R third intracellular loop also appears to play a key role in the receptor-G protein coupling on the basis of the decreased signaling of the H₃R₃₆₅ isoform, which lacks 80 residues in the third intracellular loop (Riddy et al., 2016), and reduced signaling was induced by the A280V mutation in the same loop (Flores-Clemente et al., 2013). Further consideration to the H₃R carboxyl tail should therefore be made when assessing the H₃R-G protein coupling.

An important issue when describing a novel heterodimer is to show changes in the signaling profiles of the receptors involved in the dimer. In this regard, we found enhanced signaling efficacy of the agonist CGS-21680 when the H₃R was co-transfected and co-activated in the HEK293T-A_2AR cells. This effect can be explained by a H₃R-mediated facilitation of the A_2AR-G protein coupling, leading to an increase in the receptor efficacy to activate Gα_s proteins and thus to produce cAMP. Given that in the absence of RAMH, CGS-21680 behaved the same in HEK293T-A_2AR and HEK293T-A_2AR/H₃R cells, receptor expression does not appear to account for the observed effects.

Canonically H₃Rs couple to Gα_i/o proteins, which inhibit adenylyl cyclase activity and accordingly activation of the receptor with RAMH lead to a decrease in cAMP formation. Similar to the observed in the Ca²⁺ mobilization assays, expression of the A_2AR shifted the H₃R-mediated cAMP response from inhibition to an increase in the cAMP formation. In the Ca²⁺ mobilization assays we did not observe signaling of the H₃R through Gα_qs4 proteins bound to the truncated A_2AR, and we are thus not considering a H₃R change in signaling pathways as an explanation for this effect. Therefore, we hypothesized that in the A_2AR-H₃R heteromer the A_2AR signaling prevails over that of the H₃R, hampered possibly by a steric impediment. This is in line with the loss of Ca²⁺ signaling when the H₃R-Gα_qi4 was co-expressed with the A_2AR.

An unexpected but interesting finding was the potential biased-signaling at the A_2AR-H₃R heteromer, as evidenced by the incapability of the endogenous ligand histamine to signal at the A_2AR-H₃R heteromer compared to the exogenous agonist RAMH. This discrepancy suggests that RAMH induces conformational changes in the H₃R that allow the interaction to occur, whereas histamine-induced changes appear not sufficient for heteromerization, or at least for the heteromer to signal. Ligand bias could also be explained by agonist residence time, this is, the time that a particular drug remains in its binding pocket and that will directly determine the time that a given receptor maintains an active conformation. The active state induced by histamine may therefore have a shorter duration compared with that induced by RAMH, preventing the former from activating the chimeric G proteins bound to the A_2AR₃₀₂. Binding studies with striatal membranes and histamine show that the first hypothesis holds better because histamine induced an increase in the affinity of the A_2AR for its agonist CGS-21680 (Supplementary Figure 4), opposite to the decrease in affinity change induced by RAMH in the same preparation. This result suggests that histamine and RAMH lock the H₃R in different conformational states that affect its interaction with the A_2AR.

4.2. Functional relevance of the A_2AR-H₃R heterodimer

Our results suggest a pre-synaptic location of the A_2AR-H₃R heterodimer in the striatum. A previous report indicates equal distribution of A_2ARs in total and synaptosomal membranes from rat striatum (Rebola et al., 2005), with a preferential location on the post-synaptic density fraction over the pre-synaptic active zone fraction (49.2 ± 3.3 % and 26.9 ± 3.3 % of total immunoreactivity, respectively). H₃Rs are expressed pre- and post-synaptically (Ellenbroek and Ghiabi, 2014), but they seem to be highly enriched in the terminals of striato-pallidal neurons (iMSNs) yielding a value of 1,327 ± 79 fmol/mg protein (Morales-Figueroa et al., 2014). This distribution supports our proposal for the heterodimer location and suggests a role for the A_2AR-H₃R heteromer in the pre-synaptic modulation of glutamatergic and GABAergic transmission.

In the striatum, A_2ARs have a specific location in iMSNs (Schiffman et al., 1991) and cortico-dMSNs terminals, but not in the cortico-iMSNs projections (Quiróz et al., 2009) whereas the H₃R is ubiquitously expressed throughout the striatum (Nieto-Alamilla et al., 2016). Both receptors can modulate the cortico-striatal glutamatergic and intra-striatal GABAergic transmission, but whereas the H₃R consistently inhibits neurotransmitter release in either location, the A_2AR facilitates cortico-striatal glutamate release but inhibits GABA release from the iMSNs collaterals.

Our data depicts thus a scenario where the A_2AR and the H₃R preserve their canonical signaling pathways (Gα_s and Gα_i/o, respectively), and upon dimerization the H₃R facilitates the A_2AR signaling. This receptor functionality is preserved in the cortico-dMSNs projections, with A_2AR activation enhancing glutamate release (Popoli et al., 1995) and H₃Rs exerting the opposite effect (Arias-Montaño et al., 2001). Activation of the A_2AR-H₃R heterodimer in these terminals may further facilitate glutamate release, resulting in hyper-stimulation of the dMSNs population.

Contrary to the A_2AR and H₃R antagonic modulation of glutamatergic transmission, both receptors inhibit GABA release from iMSNs collaterals. This may be due to the A_2AR functional duality showed in this neuronal population, as demonstrated by its capability to activate Gα_s and Gα_q proteins (Gubitz et al., 1995; Kirk and Richardson, 1996). Therefore, we hypothesized that the activation of the A_2AR-H₃R heterodimer in this location will further inhibit GABA release from iMSN nerve terminals, facilitating the activation of the dMSNs. According to this hypothesis, the overall activation of the striatal A_2AR-H₃R heterodimer would increase dMSNs activity. It is therefore tempting to speculate that co-activation of the heterodimer may have relevance for the treatment of the autism and obsessive and compulsive disorder, in which alterations in cortico-striatal glutamatergic transmission and MSN excitability have been shown (Welch et al., 2008; Naaijen et al., 2017).

Furthermore, it has been reported that the A_2AR antagonists KW-6002 and SCH-442416 can distinguish receptors located in the cortical afferents targeting dMSNs or iMSNs, respectively (Orrú et al., 2011). Although there are no agonists capable to differentiate between A_2AR populations, this information may be a valuable tool for targeting and modulating the A_2AR and H₃R pharmacology in a location-specific manner, using bivalent ligands directed to the heterodimer.

5. Conclusion

This study presents for the first time evidence for an A_2AR-H₃R heterodimer based on functional complementation and co-immunoprecipitation assays in HEK-293 cells, where co-activation of the receptors leads to enhanced A_2AR signaling and attenuation of H₃R functionality. In rat striatal tissue, the interaction occurs in synapses where H₃R activation modifies the binding affinity of the A_2AR.

Table 1.

Pharmacological characteristics of the A_2AR-H₃R functional complementation assays in transfected HEK-293 cells

Transfection	Agonist	Emax (%)	pEC50
H3R + Gαqi4	RAMH	419 ± 25	7.73 ± 0.31
	Histamine	504 ± 22b	6.19 ± 0.17b
H3R427-Gαqi4	RAMH	753 ± 41	7.14 ± 0.21
H3R427-Gαqi4 + H3R	RAMH	523 ± 41b	7.38 ± 0.34
A2AR302-Gαqi4 + H3R	RAMH	449 ± 25	7.31 ± 0.23
	RAMH + CGS	429 ± 14ns	7.63 ± 0.15
H4R + Gαqi4	Histamine	725 ± 51	6.35 ± 0.25
	RAMH	551 ± 27a	6.15 ± 0.16
H2R + Gαqs4	Dimaprit	860 ± 37	6.15 ± 0.12
A2AR302-Gαqi4 + D2R	Quinpirole	612 ± 34	6.47 ± 0.22

Data are means ± SEM from 3 experiments. Statistical comparisons were performed between two agonists for the same transfection, or between the transfection with the chimeric receptor and the transfection of the chimeric receptor plus the wild type receptor. ns, no significant difference

^aP < 0.05

^bP < 0.01, Student’s t test.

Both RAMH and CGS-21680 were assayed at 100 nM. RAMH, R-α-methylhistamine; CGS, CGS-21680.

Abbreviations

A_2AR

Adenosine A_2A receptor

cAMP

Cyclic adenosine monophosphate

D₁R

Dopamine D₁ receptor

D₂R

Dopamine D₂ receptor

dMSNs

Direct pathway striatal medium-sized spiny neurons

GABA

γ-Aminobutyric acid

GPCRs

G-protein coupled receptors

H₂R

Histamine H₂ receptor

H₃R

Histamine H₃ receptor

H₄R

Histamine H₄ receptor

iMSNs

Indirect pathway striatal medium-sized spiny neurons

MSNs

Striatal medium-sized spiny neurons

RAMH

(R)-α-methylhistamine

WT

Wild type