Immunogold electron microscopic evidence of in situ formation of homo- and heteromeric purinergic adenosine A1 and P2Y2 receptors in rat brain

Background Purines such as adenosine and ATP are now generally recognized as the regulators of many physiological functions, such as neurotransmission, pain, cardiac function, and immune responses. Purines exert their functions via purinergic receptors, which are divided into adenosine and P2 receptors. Recently, we demonstrated that the Gi/o-coupled adenosine A1 receptor (A1R) and Gq/11-coupled P2Y2 receptor (P2Y2R) form a heteromeric complex with unique pharmacology in co-transfected human embryonic kidney cells (HEK293T). However, the heteromeric interaction of A1R and P2Y2R in situ in brain is still largely unknown. Findings In the present study, we visualized the surface expression and co-localization of A1R and P2Y2R in both transfected HEK293T cells and in rat brain by confocal microscopy and more precisely by immunogold electron microscopy. Immunogold electron microscopy showed the evidence for the existence of homo- and hetero-dimers among A1R and P2Y2R at the neurons in cortex, cerebellum, and particularly cerebellar Purkinje cells, also supported by co-immunoprecipitation study. Conclusion The results suggest that evidence for the existence of homo- and hetero-dimers of A1R and P2Y2R, not only in co-transfected cultured cells, but also in situ on the surface of neurons in various brain regions. While the homo-dimerization ratios displayed similar patterns in all three regions, the rates of hetero-dimerization were prominent in hippocampal pyramidal cells among the three regions.


Background
The adenosine A 1 receptor (A 1 R) is known to regulate Ca 2+ /K + channels, adenylate cyclase, and phospholipase C by coupling to G i/o proteins [1]. In hippocampal astrocytes, P2Y 1 R-and P2Y 2 R-mediated Ca 2+ responses differentially show two forms of activity-dependent negative feedback of synaptic transmission via the phospholipase C beta-IP 3 pathway [2]. Today, the homo-or hetero-dimers of many kinds of GPCRs have been reported [3]. We previously demonstrated that A 1 R associates with P2Y 1 R in co-transfected HEK293T cells and in rat brain homogenates, whereby a P2Y 1 R agonist stimulates A 1 R signaling via G i/o [4,5]. Furthermore, in HEK293T cells co-transfected with A 1 R and P2Y 2 R, the heterodimers display synergistic increases in Ca 2+ signaling, whereby simultaneous activation of the two receptors attenuates A 1 R signaling via G i/o , but synergistically enhances P2Y 2 R signaling via G q/11 [6]. Also, the simultaneous activation of endogenous A 1 R and P2Y 2 R in DDT1MF-2 cells synergistically increases translocation of protein kinase C [7]. Because A 1 R are widely expressed in brain [5], it is likely that these receptors also associate directly in situ; however, direct evidence of their dimerization or precise co-localization in brain has yet to be demonstrated. The aim of the present study is to determine whether A 1 R and P2Y 2 R associate with each other in rat brain by co-immunoprecipitation and looking for receptor complexes via immunogold electron microscopy (IEM).

Immunoprecipitation and western blotting of rat brain homogenates
Eight-week-old male Wistar rats were decapitated under anesthesia (Nembutal; 30 mg/kg i.v.), and cortical, hippocampal, and cerebellar tissues were dissected out. The tissues were homogenized with a Polytron homogenizer in 50 mM Tris-acetate, pH 7.4, containing a protease inhibitor cocktail (Roche Applied Science, Manheim, Germany), and the resulting cell suspensions were centrifuged at 30,000 × g for 30 min at 4°C. The pellets were solubilized in ice-cold lysis buffer (50 mM Tris-HCl, pH 7.4, 1% Triton X-100, 300 mM NaCl and a protease inhibitor cocktail) for 60 min at 4°C. The mixture was centrifuged at 18,500 × g for 20 min at 4°C, and the supernatant pre-cleared with Protein G-Sephar-ose™4 Fast Flow (Amersham Bioscience, Piscataway, NJ). The lysate was incubated with rabbit polyclonal anti-A 1 R antibody (anti-A 1 R; 1 μg/ml, Sigma-Aldrich, St. Louis, MO) for 60 min at 4°C. Protein G-Sepharose was added to the mixture, and the incubation continued for an additional 120 min. Protein G-Sepharose was recovered by centrifugation and washed three times with lysis buffer. Immunoprecipitates were eluted with SDS-PAGE sample buffer, resolved by 12% SDS-PAGE, and electrotransferred to nitrocellulose membranes. Receptors on the blot were detected using anti-A 1 R or anti-P2Y 2 R, followed by horseradish peroxidaseconjugated goat anti-rabbit IgG secondary antibody (Sigma-Aldrich). The reactive bands were visualized with enhanced chemiluminescent substrates (Super-Signal West Pico, Pierce, Rockford, IL).
Pre-embedding immunogold electron microscopy (IEM) of transfected HEK293T cells HEK293T cells expressing HA-A 1 R and Myc-P2Y 2 R were fixed with 4% PFA, and permeabilized with 0.25% Triton X-100. Cells were incubated with anti-HA and anti-Myc for 3 h at 4°C. After washing with PBS, cells were incubated with 10-nm gold particle-conjugated goat anti-rat IgG antibody (rat IgG-10, 1:1000, BBI International, Lakewood, CO) and 5-nm gold particleconjugated goat anti-mouse IgG antibody (mouse IgG-5, 1:1000, BBI International) for 4 h at 4°C. After washing, the cells were fixed with 2.5% glutaraldehyde in 0.15 M sodium cacodylate, pH 7.4 for 2 h, washed, and postfixed with 1% osmium tetroxide for 4 h at room temperature. The cells were then dehydrated and embedding resin (Epon 812; NISSIN EM, Tokyo, Japan). Specimens were observed with an H7500 electron microscope (Hitachi, Japan). We quantified the gold staining as follows: The gene-transfected HEK293T cells with the highest numbers of total immuno-reacted gold particles were defined as 100% labeling. Because the cotransfected HEK293T cells that displayed unique pharmacology in our previous study [6] exhibited more than 20% hetero-dimeric gold particles, we used this number as a threshold in the current study. Thus, cells with more than 20% hetero-dimeric particles were defined as being "significantly stained", and those with 20% or less were defined as "not significantly stained".
Post-embedding immunogold electron microscopy of brain tissues Dissected brain tissues were cut into 1.0 mm 3 blocks that were then incubated with lead (II) acetate (Sigma-Aldrich) buffer for 1 h at room temperature, dehydrated through a series of graded ethanol, and embedded in LR-white (NISSIN EM). Ultra thin sections (40 nm) were mounted on 200-mesh nickel grids (NISSIN EM) and incubated in PBS containing 1% BSA for 10 min. After immunostaining with primary antibodies, each specimen was incubated with mouse IgG-5-and IgG-10-nm gold particle-conjugated goat anti-rabbit IgG antibody (rabbit IgG-10) for 6 h at 4°C. For controls, transfected HEK293T cells were embedded with LRwhite under the same conditions as described above. After incubation at 4°C for 12 h with anti-HA (10 μg/ ml) and anti-Myc (10 μg/ml), samples were washed with 1% BSA/PBS. After incubation with gold particleconjugated secondary antibodies for 6 h at 4°C, sections were stained with uranyl acetate for 10 min. "Significant heteromeric staining" was defined as more than 20% of the total number of immuno-reacted gold particles at the cell surface occurring in heteromeric clusters.

Comparison of the numbers of monomers, homo-dimers, and hetero-dimers
The numbers of immunogold particles at the cell surface of each cell type were determined. We defined single particles located independently as monomers (A 1 R and P2Y 2 R in Figure 1), complexes composed of clusters of the same-sized gold particles as "homo-dimers" (A 1 R-A 1 R or P2Y 2 R-P2Y 2 R in Figure 1), and those of different sized gold particles as "hetero-dimers" (A 1 R-P2Y 2 R in Figure 1). Separate calculations were made of particles in cortical neurons ( Figure 1A), hippocampal pyramidal neurons ( Figure 1B), and Purkinje cells ( Figure 1C); gold particles were counted in three cells in each region. We also counted immunogold particles in co-transfected HEK293T cells (please see above, and Figure 1D). The total number of immunoreactive gold particles on each cell surface was defined as 100%. From a total of 12 photos from each brain area (i.e., 36 photos) and from transfected cells that were reacted under the same conditions as the brain sections for each immunostaining, Figure 1 Bar graphs comparing the relative distributions of A 1 R(A1)-and P2Y 2 R(Y2)-immunoreactive elements in each brain region (A-C) and in transfected HEK293T cells (D). The P2Y 2 R-P2Y 2 R, A 1 R-A 1 R and A 1 R-P2Y 2 R dimers are indicated by Y2-Y2, A1-A1 and A1-Y2, respectively. Total number of immunoreactive gold particles on the cell surface was defined as 100%. Each column represents the average frequency (± SD) from three cells. Raw data are shown in the tables under the graphs. Data are means of three independent experiments.

Results
Co-localization of A 1 R and P2Y 2 R in transfected HEK293T cells The co-localization of A 1 R and P2Y 2 R in co-transfected HEK293T cells was examined by double immunostaining of HA-A 1 R and Myc-P2Y 2 R as a comparison experiment for the localization of these receptors in brain tissues ( Figure 2). Both receptors were localized mainly on cell surface and cytosolic membranes, but not in the nucleus (Figure 2A, B). Merged images showed their co-localization mainly in cell membranes ( Figure 2C). No signals were observed in non-transfected HEK293T cells, indicating that the immunoreactivity observed in Figure 2  was specific to the expressed receptors (data not shown). These results suggest that both receptors were expressed on cell membranes.

Immunohistochemical studies in rat brain
We examined the expression of A 1 R and P2Y 2 R in brain using immunohistochemical analyses (Figure 2). The specificity of the antibodies against A 1 R and P2Y 2 R was confirmed by the immunocytochemistry of recombinant receptor-expressing cell lines, i.e. antibodies used in this study showed no cross-labeling in A 1 R-and P2Y 2 Rtransfected HEK293T cells (data not shown). Prominent staining of A 1 R and P2Y 2 R were observed especially in Purkinje cells ( Figure 2D-F), interposed cerebellar nuclei ( Figure 2G-H), and hippocampal pyramidal cells ( Figure  2J-L). Comparatively high immunoreactivities were also detected in the piriform cortex, amygdala, hypothalamus, and brainstem (data not shown). Their expressions were mainly restricted to cell bodies and neuronal dendrites. Importantly, co-localization of A 1 R and P2Y 2 R in the cerebellum was observed in cell bodies, except in the nuclear region, in the Purkinje cells and those of the interposed cerebellar lobule nucleus ( Figure 2D-I). In the hippocampal region, pyramidal cell bodies, especially the cell surface membranes, in CA1, CA2, CA3, and the dentate gyrus (CA3; Figure 2J-L, others; data not shown) were intensely stained for both A 1 R and P2Y 2 R. Similar staining patterns were seen in cell bodies of neurons in the cerebral cortex (data not shown).
Co-immunoprecipitation of A 1 R and P2Y 2 R from rat brain Next, we examined whether A 1 R and P2Y 2 R are associated with one another in several brain regions using immunoprecipitation with anti-A 1 R followed by immunoblotting with both A 1 R and P2Y 2 R antibodies ( Figure  3). A 1 R and P2Y 2 R immunoreactivities were present in all three rat brain regions examined ( Figure 3A, B, F). Moreover, in these same regions, anti-A 1 Rs were capable of co-precipitating P2Y 2 R ( Figure 3D), indicating that A 1 R and P2Y 2 R are associated with one another in rat cortex, cerebellum, and hippocampus. The absence of these immunoreactive bands in the presence of anti-P2Y 2 R antigen peptides ( Figure 3C, E) is evidence of their specificity of the antibodies. The specificity of the anti-A 1 R was confirmed by immunocytochemistry of mock-transfected HEK293T cells, and no specific band was detected (data not shown).

Immunogold electron microscopic observations of HA-A 1 R and Myc-P2Y 2 R expressed in HEK293T cells
The immunogold particles were localized singly or in clusters, indicating that both HA-A 1 R and Myc-P2Y 2 R form monomers and homo-dimers. Specificities of the gold-labeled anti-HA and anti-Myc were demonstrated by incubating A 1 R-transfected HEK293T cells with a mixture of both antibodies, and showed that only A 1 Rlabeled particles were present ( Figure 4D). No significant immunoreactivity was detected with both anti-HA and anti-Myc in mock-transfected HEK293T cells or with only secondary antibodies (no primary antibodies) in HA-A 1 R-transfected HEK293T cells (data not shown). Also, when Myc-P2Y 2 R-transfected HEK293T cells were Immunoblotting analyses of extracts of rat brain with anti-A 1 R (A), anti-P2Y 2 R (B), and anti-P2Y 2 R with the control peptide of P2Y 2 R (C). Membrane extracts from each region were immunoprecipitated with anti-A 1 R, and analyzed by immunoblotting with anti-P2Y 2 R (D), anti-P2Y 2 R with the control peptide of P2Y 2 R (E), anti-A 1 R (F). No bands were seen in C, E demonstrating the specificity of the antibodies. It was confirmed that immunoprecpitation without primary antibody resulted in no detectable receptor bands in the immunoblotting (data not shown). Approximate molecular masses are shown in kDa. incubated with both anti-HA and anti-Myc, single particles (monomers) were scattered all over the cells, whereas co-localized, equal-sized particles of Myc-P2Y 2 R (homo-dimers) were only occasionally seen (data not shown). In HEK293T cells co-transfected with both HA-A 1 R and Myc-P2Y 2 R, clusters of different-sized particles were observed mainly at the cell surface ( Figure  4C) might be suggestive that they form heteromeric complexes.
Immunogold electron microscopic observations of A 1 R and P2Y 2 R expressed in rat brain We incubated post-embedded, primary antibody-stained rat brain tissues with secondary antibodies labeled with mouse IgG-5 for A 1 R and rabbit IgG-10 for P2Y 2 R. As negative controls, tissues were stained with only secondary antibodies conjugated with different sized gold particles; no significant immunoreactivities were observed under the experimental conditions used in this study (data not shown). As in the transfected HEK293T cells, we observed clusters of different-sized gold particles at cytoplasmic membranes in cell bodies, indicating the presence of heteromeric complexes of endogenous A 1 R and P2Y 2 R in rat brain ( Figure 4E-G). Significant immunoreactivity was detected in Purkinje cells ( Figure 4F) and hippocampal pyramidal cells ( Figure 4G). Heteroand homo-dimers were detected in significant numbers at the cell surface in both transfected HEK293T cells and native brains.

Comparison of the frequencies of monomers, homodimers, and hetero-dimers
We counted gold particles on the surfaces of cells in the cortex, cerebellum, and hippocampus and classified them as monomers (A 1 R or P2Y 2 R), homo-dimers (A 1 R-A 1 R or P2Y 2 R-P2Y 2 R), or hetero-dimers (A 1 R-P2Y 2 R). While the homo-dimerization ratios (A 1 R-A 1 R/P2Y 2 R-P2Y 2 R) displayed similar patterns in all three regions ( Figure 1A-C), the rates of hetero-dimerization were prominent in hippocampal pyramidal cells among the three regions.

Discussion
The present study provides the first detailed evidence of an interaction between endogenous A 1 R and P2Y 2 R in brains using co-immunoprecipitation and IEM. The homo-dimerization of A 1 R was previously analyzed in our laboratory by computational prediction, coimmunoprecipitation, and BRET analysis [9]. In the present study, we might suggest the existence of homodimers (A 1 R-A 1 R and P2Y 2 R-P2Y 2 R) using IEM. Very interestingly, the percentage of A 1 R homo-dimers was higher than that of P2Y 2 R in both rat brain and transfected HEK293T cells (Figure 1). By contrast, the ratios of heteromeric gold-particle clusters were different in the cortex, hippocampus, and cerebellum. Importantly, both homo-dimeric and hetero-dimeric gold-particles were much fewer at inner cytoplasmic membranes than at the cell surface (data not shown). In general, most GPCRs dimers have been observed on the cell surface [10,11]. Total numbers of hetero-dimers observed on the cell surface and in the cytoplasm were obviously different (data not shown) and may reflect the process of receptor maturation and association of the A 1 R-P2Y 2 R complex.
In the hippocampal region, the strong presence of hetero-dimers coincided with the relative signal intensity of the co-immunoprecipitation band ( Figure 3D lane 3). In the previously reported electron microscopic analysis of A 1 R and P2Y 1 R co-localization in hippocampus, the A 1 R density was relatively higher than that of P2Y 1 R at the presynaptic membrane [12]. They suggested that the hetero-dimerization or cross-talk of A 1 R and P2Y 1 R is involved in regulation of glutamate release. The relative distributions of immunoreactivities for GABA B R2 and GABA B R1 were also different in the basal ganglia and globus pallidus/substantia nigra, which suggests the possible co-existence and hetero-dimerization of two types of receptors at various pre-/postsynaptic sites [13]. From the present study, it can be speculated that the A 1 R/ P2Y 2 R hetero-oligomer might be responsible for down regulation, via hippocampal Ca 2+ secretion, of synaptic functions [14]. The abundant formation of A 1 R/P2Y 2 R hetero-oligomers in hippocampus revealed in this study supports the idea that the unique signal transduction generated by hetero-dimerization, including the enhancement of Ca 2+ signaling via G q/11 , observed in transfected cells also occurs in hippocampus.
List of abbreviations GPCR: G protein-coupled Receptor; A 1 R: A 1 adenosine receptor; P2Y 1 R: P2Y 1 purinergic receptor; P2Y 2 R: P2Y 2 purinergic receptor; IEM: immunogold electron microscopy revising. NH was responsible for experimental design and revised and polished the manuscript. All authors have read and approved final manuscript.