Adenosine A_2A receptor and ecto-5′-nucleotidase/CD73 are upregulated in hippocampal astrocytes of human patients with mesial temporal lobe epilepsy (MTLE)

Abstract

Refractoriness to existing medications of up to 80 % of the patients with mesial temporal lobe epilepsy (MTLE) prompts for finding new antiepileptic drug targets. The adenosine A_2A receptor emerges as an interesting pharmacological target since its excitatory nature partially counteracts the dominant antiepileptic role of endogenous adenosine acting via inhibitory A₁ receptors. Gain of function of the excitatory A_2A receptor has been implicated in a significant number of brain pathologies commonly characterized by neuronal excitotoxicity. Here, we investigated changes in the expression and cellular localization of the A_2A receptor and of the adenosine-generating enzyme, ecto-5′-nucleotidase/CD73, in the hippocampus of control individuals and MTLE human patients. Western blot analysis indicates that the A_2A receptor is more abundant in the hippocampus of MTLE patients compared to control individuals. Immunoreactivity against the A_2A receptor predominates in astrocytes staining positively for the glial fibrillary acidic protein (GFAP). No co-localization was observed between the A_2A receptor and neuronal cell markers, like synaptotagmin 1/2 (nerve terminals) and neurofilament 200 (axon fibers). Hippocampal astrogliosis observed in MTLE patients was accompanied by a proportionate increase in A_2A receptor and ecto-5′-nucleotidase/CD73 immunoreactivities. Given our data, we hypothesize that selective blockade of excessive activation of astrocytic A_2A receptors and/or inhibition of surplus adenosine formation by membrane-bound ecto-5′-nucleotidase/CD73 may reduce neuronal excitability, thus providing a novel therapeutic target for drug-refractory seizures in MTLE patients.

Introduction

Mesial temporal lobe epilepsy (MTLE) is the most common and devastating form of human epilepsy [1]. This disorder is characterized by irreversible biochemical and structural changes in the hippocampus and several neocortical regions [2–4]. In spite of the extensive research of last decades resulting in the discovery of new antiepileptic drugs available for clinical practice, up to 80 % of MTLE patients remain refractory to medication [5]. The leading cause for this concern is the gap in our knowledge regarding the underlying mechanisms responsible for the transformation of brain networks into hyperexcitable connections more prone to intense and/or prolonged neuronal discharges. The last resource treatment for some drug-refractory patients is still the surgical ablation of defective tissue to curb epileptic crisis. Yet, a significant number of MTLE patients do not fulfill the requirements for this surgical procedure and are faced with an unmet medical need. This calls for the urgent finding of new and more accurate pharmacological targets to control seizures and/or the epileptogenic process [1].

Increasing evidences show that the extracellular levels of ATP and adenosine dramatically increase during high-frequency neuronal firing and/or pathologic brain activity, such as prolonged or repeated seizures [6–8]. Once in the extracellular space, adenosine, released directly from cells or produced from the extracellular catabolism of ATP via the ecto-nucleotidase pathway [6–8], controls synaptic transmission through the activation of four membrane-bound receptor subtypes, A₁, A_2A, A_2B, and A₃. The inhibitory A₁ and the excitatory A_2A are the most relevant adenosine receptors in the brain [9]. While A₁ receptors are widely expressed throughout the brain, A_2A receptors are most abundant in the basal ganglia, but these receptors are also present at lower density in other brain regions, namely the hippocampus. The molecular mechanisms by which the A_2A receptor controls neuronal excitability and synaptic plasticity have been extensively studied [10–14]. Activation of the A_2A receptor stimulates glutamate release and prevents its uptake, thus resulting in increased synaptic levels of this excitatory amino acid [11–13]. Excitatory synaptic transmission may be further unbalanced because A_2A receptor activation favors desensitization of presynaptic inhibitory modulation via A₁ receptors [15, 16]. Thus, involvement of the A_2A receptor in diverse pathologies of the central nervous system, including epilepsy [17–25], may be due to counteraction of the neuroprotective role of adenosine via the A₁ receptor [24]. Most studies demonstrating the proconvulsive effect of the A_2A receptor have been performed in rodents, both in mice [17–19, 26] and rats [20–24]. Results from these investigations suggest that selective A_2A receptor antagonists might offer protection against diverse epileptic syndromes, such as temporal lobe epilepsy, highlighting the idea that the A_2A receptor may be an attractive pharmacological target for the treatment of epilepsy. To our knowledge, there are no studies investigating changes in the expression and function of the A_2A receptor in the brain of drug-resistant MTLE human patients as we attempted in the present study.

Although it was believed that the adenosine A_2A receptor was mostly located on hippocampal glutamatergic nerve terminals [27, 28], recent evidences demonstrated that this receptor is also highly expressed in glial cells, both in astrocytes and microglia [11–13, 19]. Astrocytes modulate the activity of neuronal networks upon sustained and intense activity, but it has been recently evidenced that astrocytes can also modulate basal synaptic transmission by the release of purines [29]. Interestingly, γ-aminobutyric acid (GABA) uptake in astrocytes has been attributed to the expression of A₁/A_2A heteromers and to a specific mechanism by which these heteromers signal via inhibitory G_i or facilitatory G_s depending on the extracellular concentration of adenosine [30]. The astrocytic A_2A receptor also seems to be critical for the modulation of glutamate transport, either by decreasing the uptake [12, 13] or increasing the release [11] of this amino acid. In view of this, astrocytic A_2A receptors have been implicated in memory formation [19]. The same research group demonstrated that astrocytic A_2A receptor levels are upregulated in humans with Alzheimer’s disease, as well as in epileptic mice injected with kainate [19].

Given the dominant excitatory (proconvulsive) nature of the adenosine A_2A receptor activation in the brain and the need for clarification of its role in drug-resistant human epilepsy, we designed this study to investigate changes in the expression and cellular localization of this receptor in the hippocampus of control and MTLE human patients. Knowing that the A_2A receptor is preferentially activated by adenosine originated from released ATP in the hippocampus [31], we also evaluated the expression and localization of ecto-5′-nucleotidase/CD73 (EC3.1.3.5), the rate-limiting enzyme for extracellular adenosine formation in the brain. It is worth noting that genetic variations in enzymes and/or transporters influencing extracellular adenosine homeostasis, including ecto-5′-nucleotidase/CD73, have been significantly associated with epileptogenesis and posttraumatic epilepsy risk and are, therefore, worth to be explored as therapeutic targets for pharmacological development [32].

Methods

Human brain samples

Epileptic hippocampal samples were obtained from MTLE patients undergoing amygdalohippocampectomy at the Department of Neurosurgery of the Centro Hospitalar do Porto–Hospital Geral de Santo António (CHP-HGSA). This study and all its procedures were approved by the Ethics Committees of CHP-HGSA and Instituto de Ciências Biomédicas Abel Salazar–Universidade do Porto (ICBAS-UP). All MTLE patients signed an informed consent for using the biological material. The amount of tissue removed did not differ from the strict amount necessary for successful epileptic focus ablation surgery. Resection tissue was kept in cold artificial cerebral-spinal fluid (in mM: glucose 10, NaCl 124, KCl 3, MgCl₂ 1, NaH₂PO₄ 1.2, NaHCO₃ 26, CaCl₂ 2, pH 7.40) and used within 2–6 h after collection.

Control brain samples were obtained from four human cadavers submitted to forensic autopsy performed within 4–7 h postmortem; experimental procedure validation and legal constrains for using forensic brain samples were discussed in a previous study from our group [33]. Brain samples were made available by the Instituto Nacional de Medicina Legal e Ciências Forenses–Delegação do Norte (INMLCF-DN), according to Decree-Law 274/99, of July 22, published in Diário da República–1st SERIE A, No. 169, of 22-07-1999, Page 4522, regarding the regulation on the ethical use of human cadaveric tissue for research. After collection, brain tissue was kept in cold artificial cerebral-spinal fluid until use.

Clinical information about the composition of control and MTLE groups is provided in Table 1. The investigation conforms to the principles outlined in The Code of Ethics of the World Medical Association (Declaration of Helsinki).

Table 1.

Comparison of clinical variables of control and MTLE patients

	Control	MTLE
Male/female	3:1	6:4
Age (years) ± SD	60.5 ± 16.3	39.4 ± 8.4
Epilepsy onset (years) ± SD	NA	10.8 ± 6.1
Duration (years) ± SD	NA	28.7 ± 11.6
Cause of death	Acute myocardial infarction (4)	NA
pmi (h) ± SD	5.1 ± 1.3	NA

SD standard deviation, NA not applicable, pmi postmortem interval

Isolation of nerve terminals from human hippocampus

The methodology for isolation nerve terminals (synaptosomes) from the human hippocampus was adapted from previous studies [33–35] and validated here for the first time (see below; Fig. 1). Briefly, the hippocampus was gently homogenized in cold oxygenated (95 % O₂ /5 % CO₂) Krebs solution (in mM: glucose 5.5, NaCl 136, KCl 3, MgCl₂ 1.2, Na₂HPO₄ 1.2, NaHCO₃ 16.2, CaCl₂ 0.5, pH 7.40). Homogenates were filtered through a nylon filter (mesh size 100 μm). The filtrate was left to sit during 30–45 min until pellet formation, which was resuspended into Krebs solution and left at room temperature.

Fig. 1.

Synaptosomes isolated from the hippocampus of MTLE human patients are enriched in synaptic nerve terminals. At the top of the figure are illustrated representative blots of GFAP, synaptophysin, and PSD95 immunoreactivity in total lysates and synaptosomes of the hippocampus of an MTLE patient. GAPDH and β-tubulin were used as reference proteins. Bar graphs show the average composition in GFAP, synaptophysin, and PSD95 of total lysates and synaptosomes from four MTLE patients. Data are expressed as mean ± SD. ***P < 0.001 represents significant differences as compared to total lysates (two-way ANOVA followed by Bonferroni’s multiple comparison test); ns nonsignificant

Western blot analysis

Western blot analysis was performed as previously described [33, 35]. Total lysates and synaptosomes of the human hippocampus were homogenized in radio-immunoprecipitation assay buffer (Tris-HCl 25 mM (pH 7.6), NaCl 150 mM, sodium deoxycholate 1 %, Triton X-100 1 %, sodium dodecyl sulfate (SDS) 0.1 %, EDTA 5 mM, and protease inhibitors). Samples were solubilized in SDS reducing buffer (Tris-HCl 125 mM (pH 6.8), SDS 4 %, bromophenol blue 0.005 %, glycerol 20 %, and 2-mercaptoethanol 5 %), subjected to electrophoresis in SDS-polyacrylamide gels and electrotransferred onto polyvinylidene difluoride (PVDF) membranes. In contrast, the detection of CD73 was performed in nonreducing conditions. Therefore, samples to detect CD73 were dissolved in the above SDS buffer without mercaptoethanol. Membranes were, then, blocked in Tris-buffered saline (in mM: Tris-HCl 10 (pH 7.6), NaCl 150) containing Tween 20 0.05 % and bovine serum albumin (BSA) 5 % and incubated with the primary antibodies: mouse anti-GFAP (1:500, Chemicon, Temecula, CA, USA), mouse anti-synaptophysin (1:750, Chemicon, Temecula, CA, USA), mouse anti-postsynaptic density-95 (PSD95; 1:600, Chemicon, Temecula, CA, USA), rabbit anti-A_2A receptor (1:250; Alpha Diagnostic, San Antonio, TX, USA), and rabbit anti-CD73 (1:500; h5’NT-2_lI₅) available at http://ectonucleotidases-ab.com. Membranes were washed and incubated with horseradish peroxidase-conjugated secondary antibody. For normalization purpose, membranes were incubated with rabbit anti-β-tubulin antibody (1:2500; Abcam, Cambridge, UK) or mouse anti-glyceraldehyde 3-phosphate dehydrogenase (GAPDH) antibody (1:200; Santa Cruz Biotechnology, Dallas, TX, USA) following the procedures described above. The antigen-antibody complexes were visualized using the ChemiDoc MP imaging system (Bio-Rad Laboratories, Hercules, CA, USA). Gel band image densities were quantified with ImageJ (National Institute of Health, USA). To test for specificity of the bands corresponding to A_2A receptor, the anti-A_2A receptor antibody was preadsorbed with a control/blocking peptide (#A2aR21-P, Alpha Diagnostic, San Antonio, TX, USA). The specificity of the band corresponding to CD73 was confirmed by comparing the immunoblot with the one obtained with the preimmune serum (the serum extracted prior to immunization) of the same animal species (rabbit).

Immunofluorescence confocal microscopy

Immunofluorescent staining and confocal microscopy analysis was performed as previously described [33, 35]. Brain samples were fixed in 4 % paraformaldehyde, cryopreserved in 30 % sucrose, and stored in a tissue freezing medium at −80 °C. Free floating 30-μm hippocampal sections were incubated for 1 h, with blocking buffer (fetal bovine serum 10 %, BSA 1 %, Triton X-100 0.5 %, NaN₃ 0.05 %) and incubated overnight with the primary antibodies: mouse anti-A_2A receptor (1:300, Chemicon, Temecula, CA, USA), rabbit anti-synaptotagmin 1/2 (1:500, Synaptic Systems GmbH, Goettingen, Germany), rabbit anti-GFAP (1:500, Dako, Glostrup, Denmark), rabbit anti-neurofilament 200 (NF200; 1/900, Abcam, Cambridge, UK), and guinea pig anti-CD73 (h5’NT-1_cI₄ or h5’NT-2_cI₄ at a dilution of 1:200; available at http://ectonucleotidases-ab.com). Sections were rinsed and incubated 2 h with species specific secondary antibodies: donkey anti-rabbit Alexa Fluor 488 (Molecular Probes, Eugene, OR, USA), goat anti-mouse Alexa Fluor 633 (Molecular Probes, Eugene, OR, USA), and donkey anti-guinea pig Dylight 649 (Jackson ImmunoResearch, West Grove, PA, USA). After mounting, sections were observed and analyzed with a laser scanning confocal microscope (Olympus FV1000, Tokyo, Japan). Co-localization was assessed by calculating the Pearson’s linear correlation coefficient (ρ) and the staining overlap for each confocal micrograph using the Olympus Fluoview 4.2 Software (Olympus FV1000, Tokyo, Japan). The ρ value is a measure of pixel-by-pixel covariance in the signal levels of two images (stainings) and varies between +1 and −1, inclusive, where 1 is total positive correlation, 0 is no correlation, and −1 is total negative correlation; because it subtracts the mean intensity from each pixel’s intensity value, the ρ coefficient is independent of signal levels and signal offset (background). Thus, the Pearson’s linear correlation coefficient can be measured in two-color images without any form of preprocessing, making it both simple and relatively safe from user bias [36]. Because (1) ρ may be less sensitive to differences in signal intensity between the components of an image caused by different labeling with fluorocromes, photobleaching or different settings of amplifiers and (2) the negative values of ρ are difficult to interpret when the degree of overlap is the quantity to be measured, the subtraction of the averages of the two colors can be omitted to create the overlap coefficient, which varies between +1 (total overlap) and 0 (no overlap); as with the Pearson’s, this coefficient is not dependent on the magnitude of the signal (gain), but does depend on the background.

Data presentation and statistical analysis

Results are expressed as mean ± standard deviation (SD). n shown in graphs represent the number of individuals in a given situation (unless otherwise indicated). Statistical analysis was carried out using Graph Pad Prism 6.04 software (La Jolla, CA, USA). For multiple comparisons among one or multiple groups, one-way or two-way ANOVA followed by Bonferroni’s multiple comparison test were used, respectively. P < 0.05 values were accepted as significant.

Results

Human hippocampal synaptosomes are enriched in synaptic nerve terminals

Isolation of nerve terminals (synaptosomes) from the human hippocampus followed the methodology described in previous studies using the human neocortex and hippocampi from sheep and rats [33–35]. To validate this methodology, the relative enrichment in synaptophysin-positive nerve terminals compared to GFAP and PSD95 immunoreactivities identifying astrocytic glial cells and postsynaptic densities, respectively, was assessed by Western blot analysis using total lysates and synaptosomal fractions of the human hippocampus. Figure 1 shows that synaptosomes isolated from the hippocampus of MTLE human patients are highly enriched (P < 0.001) in synaptic nerve terminals specifically labeled against synaptophysin as compared to total lysates from the same brain region. Conversely, hippocampal synaptosomes from MTLE patients present smaller (P < 0.001) amounts of the astrocytic cell marker, GFAP, but similar levels of the postsynaptic density marker, PSD95, when compared to total lysates (Fig. 1).

The adenosine A_2A receptor is upregulated in the hippocampus of MTLE patients

Data from Western blot analysis show that the amount of A_2A receptor protein (∼46 kDa) is threefold (P < 0.001) higher in total lysates of the hippocampus of MTLE patients compared to that found in control individuals (Fig. 2a). Augmentation of the A_2A receptor protein in the hippocampus of MTLE patients is much more evident (P < 0.01) in total lysates than in synaptosomal fractions. These results suggest that the A_2A receptor is localized predominantly in nonneuronal cells, most probably glial cells [11–13, 19], which are the most abundant cell contributor to hippocampal total lysates. Two protein species were recognized by the A_2A receptor antibody from Alpha Diagnostic (San Antonio, TX, USA). The higher-molecular-mass species coincided with the reported molecular mass of the A_2A receptor protein (∼46 kDa), whereas the lower-molecular-mass species was more visible (P < 0.001) in total lysates than in synaptosomes of the hippocampus of MTLE patients (Fig. 2a) and may represent proteolytic processing of the larger precursor (see, e.g., [37–40]). Disappearance of the two bands corresponding to the A_2A receptor protein after preadsorption with the corresponding antigen peptide sequence confirms the specificity of the antibody used in this study (Fig. 2a, negative control).

Fig. 2.

The A_2A receptor protein is upregulated in the hippocampus of drug-refractory MTLE human patients. a Representative Western blots of the A_2A receptor immunoreactivity in total lysates and synaptosomes of the human hippocampus of control individuals and MTLE patients; gels were loaded with 100 μg of protein. Two protein species were recognized by the A_2A receptor antibody from Alpha Diagnostic (San Antonio, TX, USA) and both are increased in total lysates, but not in synaptosomes, of hippocampal samples from MTLE patients. Please note that the two bands corresponding to the A_2A receptor protein disappeared after preadsorption of the primary antibody with a control peptide (negative control); GAPDH was used as a reference protein. Data are expressed as mean ± SD, and the number of individuals per group is shown below each bar. **P < 0.01 and ***P < 0.001 (two-way ANOVA followed by Bonferroni’s multiple comparison test). b Representative confocal micrographs from different regions of the human hippocampus demonstrating that immunoreactivity against the A_2A receptor (green) is more evident in the hippocampus of MTLE patients than of control individuals; nuclei are stained with 4′,6-diamidino-2-phenylindole (DAPI; blue); DIC images are shown for comparison; two confocal micrographs were obtained per individual; three individuals from each group (control and MTLE) were analyzed; scale bars = 300 μm. c The immunofluorescence staining against the astrocytic marker, GFAP (red), increases significantly in all hippocampal subregions of MTLE patients compared to control individuals; nuclei are stained with DAPI (blue); two confocal micrographs were obtained per individual; three individuals from each group (control and MTLE) were analyzed; scale bars = 50 μm

Taking into consideration that the adenosine A_2A receptor protein was upregulated predominantly in total lysates of the human hippocampus, we thought it would be interesting to investigate by immunofluorescence confocal microscopy (1) the distribution of the A_2A receptor in hippocampal subregions, and (2) whether it co-localizes with the astrocytic glial cell marker, GFAP, as this also exhibits a higher protein density in total lysates compared to synaptosomal fractions. Confocal micrographs confirmed that patients with MTLE exhibit higher amounts of the adenosine A_2A receptor (green staining), which is evenly upregulated in all subregions of the hippocampus, CA1, CA2, CA3, and dentate gyrus (DG)/CA4, compared to control individuals (Fig. 2b). Figure 2c shows that the immunofluorescence staining of the astrocytic cell marker GFAP also increases in all subregions of the hippocampus, CA1, CA2, CA3, and DG/CA4, of MTLE patients compared to control individuals. Using differential interference contrast (DIC) images, we show in Fig. 2b that boundaries of the characteristic hippocampal regions are less evident in slices from MTLE patients than in control individuals due to hippocampal sclerosis, a neuropathological condition characterized commonly by segmental loss of pyramidal neurons, granule cell dispersion, and reactive gliosis.

Adenosine A_2A receptors are localized predominantly in GFAP-positive astrocytes in the hippocampus of MTLE patients

In order to test whether enrichment of the A_2A receptor protein associates with astrogliosis in the hippocampus of MTLE patients, we performed co-localization studies with antibodies specifically targeting the A_2A receptor and the glial cell marker, GFAP, under the confocal microscope; for comparison purposes, we also tested the degree of co-localization between the A_2A receptor and two specific neuronal cell markers, synaptotagmin 1/2 and NF200.

Figure 3a shows that the adenosine A_2A receptor (stained in green) co-localizes extensively with the astrocytic cell marker, GFAP (stained in red) in all subregions of the hippocampus of MTLE patients, including the Cornu Ammonis (CA1, CA2, CA3, and CA4) and DG. Co-localization between the A_2A receptor and GFAP immunostainings was confirmed by the elevated scores of staining overlap (overlap >0.6) and Pearson’s (ρ > 0.4) coefficients (Fig. 3b) that were obtained by merging the two fluorescence channels (yellow staining; right-hand side of Fig. 3a) in different areas of the hippocampus of MTLE patients (Fig. 3b). It is also worth noting that co-localization of A_2A receptor and GFAP immunofluorescence scores are significantly (P < 0.0001) higher in all subareas of the hippocampus of MTLE patients compared to control individuals (Fig. 3b). The higher magnification image shown in Fig. 3c demonstrates that astrocytic A_2A receptor fluorescence staining is more abundant on the plasma membrane (including processes) compared to the intracellular compartment.

Fig. 3.

The A_2A receptor co-localizes with the astrocytic cell marker, GFAP, in the hippocampus of MTLE patients. a Representative confocal micrographs of different regions of the hippocampus of MTLE human patients stained against the A_2A receptor (green) and GFAP (red). Co-localization is shown by the yellow labeling appearing when merging the two fluorescent channels (right-hand side of the image). Scale bars = 50 μm. b “Box-and-whiskers” graphs plotting Pearson’s coefficient (ρ) and staining overlap scores calculated from an “n” number of confocal micrographs; at least three individuals (control or MTLE) were included in the calculations. These parameters were automatically calculated per image and were used to quantify the co-localization of A_2A receptor and GFAP (yellow staining) in different hippocampal subregions. c A higher magnification confocal micrograph of the A_2A receptor immunostaining in hippocampal glial cells. Image shows a pseudocolor spectral representation of the A_2A receptor immunoreactivity where the signal intensity ranges from dark blue (low reactivity) to red (high reactivity); cell nuclei are stained in red with DAPI. Image scale bar is 20 μm. Glial A_2A receptors exhibit a preferential membrane pattern with sparse intracellular distribution

Contrary to that observed with GFAP, Fig. 4 shows that there is reduced overlap (overlap <0.25) and linear correlation (Pearson’s coefficient, ρ < 0.03) between the A_2A receptor immunoreactivity and two neuronal cell markers, synaptotagmin 1/2 (Fig. 4a–c) and NF200 (Fig. 4d–f), which are specific for nerve terminals and axon fibers, respectively. Under these conditions, we could hardly see the yellow staining when merging the two fluorescent channels in any subareas of the hippocampus of MTLE patients (Fig. 4a, d). Using the Pearson’s coefficient (ρ), which reflects the linear correlation of covariance signals of the two fluorescent probes without background interference, we found no significant differences (P > 0.05) when comparing immunofluorescence signals against the A_2A receptor and NF200 in the hippocampus of control individuals and MTLE patients (Fig. 4e), while some degree of overlap becomes apparent in axon fibers of CA1, CA2, and DG subregions of the hippocampus of MTLE patients (Fig. 4f). Co-localization scores depicted in Fig. 4b, c anticipate a partial loss of the A_2A receptor staining in synaptotagmin-positive nerve terminals of CA1, CA2, and DG subregions of the hippocampus of MTLE patients compared to control individuals. It, thus, appears that co-localization scores of the A_2A receptor and the two neuronal markers change in opposite directions. Whether this means a redistribution of the A_2A receptor from synaptic nerve terminals (stained by synaptotagmin 1/2) to extrasynaptic nerve fibers (stained by NF200) in the hippocampus of MTLE patients compared to control individuals requires further investigations. Notwithstanding this, co-localization scores with the two neuronal markers are minor considering the significant enrichment of the A_2A receptor observed in GFAP-positive astrocytes of the epileptic human hippocampus (Fig. 3).

Fig. 4.

In the hippocampus of MTLE human patients, the A_2A receptor does not co-localize with neuronal cell markers, synaptotagmin 1/2 and NF200, which are specific for synaptic nerve terminals and axon fibers, respectively. The upper two panels show representative confocal micrographs of different regions of the hippocampus of MTLE human patients labeled with antibodies against the A_2A receptor (green) and (a) synaptotagmin 1/2 (red) or (d) NF200 (red). The absence of yellow staining denotes lack of co-localization when merging the two fluorescent channels (right-hand side of the panels). Scale bars = 50 μm. Bottom panels show “box-and-whiskers” graphs plotting Pearson’s coefficient (ρ) and staining overlap scores calculated from an “n” number of confocal micrographs; at least three individuals (control or MTLE) were included in the calculations. These parameters were automatically calculated per image and were used to quantify the co-localization of A_2A receptor and synaptotagmin 1/2 (b, c) or NF200 (e, f) (yellow staining) in different hippocampal subregions

Increased expression of ecto-5′-nucleotidase/CD73 in the hippocampus of MTLE patients: the enzyme is localized in GFAP-positive astrocytes in close proximity of adenosine A_2A receptors

Given our previous findings suggesting that excitatory A_2A receptors are preferentially activated by adenosine formed from released adenine nucleotides in rat hippocampal, neuromuscular and myenteric synapses [31, 41], we evaluated the amount and distribution of ecto-5′-nucleotidase/CD73 in the human hippocampus compared to those of the A_2A receptor, as this is the rate limiting enzyme for extracellular adenosine formation in the brain. Figure 5a shows that the hippocampus of MTLE patients exhibits higher (P < 0.001) amounts of ecto-5′-nucleotidase/CD73 protein (∼55 kDa) than control individuals. Upregulation of ecto-5′-nucleotidase/CD73 is mostly due to proteins found in total lysates of the epileptic hippocampus, because immunoblot bands using synaptosomal fractions were almost undetectable (Fig. 5a). The band corresponding to the ecto-5′-nucleotidase/CD73 protein disappeared in immunoblots obtained with the preimmune serum obtained from the same animal (rabbit) prior to immunization (Fig. 5a), which functions as a negative control under these circumstances.

Fig. 5.

Upregulation of ecto-5′-nucleotidase/CD73 in astrocytes of the hippocampus of MTLE human patients. a Representative immunoblot analysis of the ecto-5′-nucleotidase/CD73 protein in total lysates and synaptosomes of the hippocampus of control individuals and MTLE patients. Please note that the band corresponding to the molecular weight of ecto-5′-nucleotidase/CD73 protein (∼55 kDa) disappeared in the immunoblot obtained with the preimmune serum from the same animal (negative control). Gels were loaded with 40 μg of protein; GAPDH was used as a reference protein. Data are expressed as mean ± SD, and the number of individuals per group is shown below each bar. **P < 0.01 (two-way ANOVA followed by Bonferroni’s multiple comparison test). b Representative confocal micrographs of the CA4/DG subregion of the human hippocampus demonstrating that the immunoreactivity against ecto-5′-nucleotidase/CD73 (red) is more evident in MTLE patients than in control individuals. DIC images are shown for comparison; scale bars = 300 μm. c Representative higher magnification confocal micrographs of the hippocampus of control and MTLE human patients demonstrating that ecto-5′-nucleotidase/CD73 (stained in red) co-localizes with the astrocytic cell marker, GFAP (stained in green); co-localization is shown by the yellow labeling appearing when merging the two fluorescent channels (right-hand side of the image). Scale bars = 50 μm. Images are representative from at least three individuals per group (control and MTLE)

Like that occurring for the adenosine A_2A receptor, confocal micrographs of CA4/DG hippocampal regions show that immunostaining of ecto-5′-nucleotidase/CD73 in MTLE patients is more evident than in control individuals (Fig. 5b). This is highlighted in Fig. 6a where one can see that the A_2A receptor and the ecto-5′-nucleotidase/CD73 immunofluorescence staining is more intense in the CA4 region of the hippocampus of MTLE patients than in controls. Parallelism between adenosine A_2A receptor and ecto-5′-nucleotidase/CD73 staining intensity is further strengthened because we found that the enzyme (stained in red) also co-localizes with GFAP (stained in green) in the hippocampus of MTLE patients (Fig. 5c; compare to Fig. 3a).

Fig. 6.

Ecto-5′-nucleotidase/CD73 partially co-localizes with the adenosine A_2A receptor in the hippocampus of MTLE patients. a Representative confocal micrographs from different regions of the human hippocampus demonstrating that immunoreactivity against the ecto-5′-nucleotidase/CD73 (red) partially overlaps the A_2A receptor immunofluorescence staining (green) in the hippocampus of control and MTLE patients. Co-localization is shown by the yellow labeling appearing when merging the two fluorescent channels (right-hand side of the image). Scale bars = 50 μm. b Pearson’s coefficient (ρ) and the staining overlap parameters (mean ± SD) calculated from two to four confocal micrographs per MTLE individual; at least three patients were included in the calculations. These parameters were automatically calculated per image and were used to quantify the co-localization of A_2A receptor and ecto-5′-nucleotidase/CD73 (yellow staining) in different hippocampal subregions

Figure 6 shows that there is a partial co-localization of the adenosine A_2A receptor (green) and ecto-5′-nucleotidase/CD73 (red) in all subregions of the hippocampus of MTLE patients, namely CA1, CA2, CA3, and CA4. Yellow staining denoting co-localization of the adenosine A_2A receptor and ecto-5′-nucleotidase/CD73 appeared when merging the two fluorescence channels (Fig. 6a, right-hand side of the image). The scores obtained for the staining overlap (overlap >0.4) and the Pearson’s coefficient (ρ > 0.1) in different areas of the epileptic hippocampus were moderate, although statistically different (P < 0.05) from the null hypothesis (Fig. 6b).

Discussion and conclusions

The present study shows, for the first time, that the adenosine A_2A receptor is upregulated (by about threefold) in the hippocampus of drug-resistant MTLE human patients compared to control individuals. This feature is in agreement with previous studies using animal models showing that the A_2A receptor exists at relative low levels in the hippocampus [42, 43], but its amount dramatically increases in epileptic animals [19, 20, 44]. As a matter of fact, increasing evidences show that activation of the adenosine A_2A receptor favors seizure activity in different epileptic syndromes [17, 18, 20, 22–24]. Neuronal excitation in epilepsy likely leads to enhanced synaptic A_2A activation, which may aggravate synaptotoxicity and thereby the degeneration of normal circuitry contributing to the progressive course of epilepsy. Activation of the A_2A receptor has a proconvulsive effect on piriform cortex kindled seizures in the rat [23], and it decreases the seizure threshold of hyperthermia-induced convulsions in young rats [24]. Conversely, administration of the A_2A receptor antagonist, SCH58261, prior to the induction of a status epilepticus with pilocarpine increased its latency and decreased both the incidence and the mortality rate of this condition associated with TLE in rats [21, 22]. Likewise, intracerebroventricular injection of the A_2A receptor antagonist, ZM241385, decreased the duration of after-discharges, motor seizures, and stage 5 convulsions in amygdala-kindled rats [21, 22]. Adenosine A_2A receptor-deficient mice are partially resistant to limbic seizures induced by pentylenetetrazol, but deletion of the A_2A receptor did not protect against maximal electroshock-induced generalized tonic-clonic seizures originated from brainstem structures [17, 18].

It is also worth noting that increases in extracellular ATP amounts and, consequently, augmentation of the levels of its metabolite, adenosine, have been reported during high-frequency neuronal firing and/or prolonged or repeated epileptic seizures [6–8]. This may be critical for A_2A receptor activation because adenosine resulting from the extracellular catabolism of released ATP, via the ecto-5′-nucleotidase/CD73 pathway, preferentially activates this receptor subtype in the hippocampus and other excitatory synapses (e.g., neuromuscular junction, myenteric plexus) [31, 41]. The tight association between ecto-5′-nucleotidase/CD73 and the A_2A receptor has been elegantly and definitively documented by co-immunoprecipitation and proximity ligation assays in the striatum [45]. Previous studies demonstrated that ecto-5′-nucleotidase/CD73 is up-regulated in rat models of temporal lobe epilepsy induced by kainate or pilocarpine injections [46]. Increased ecto-5′-nucleotidase/CD73 expression was also found in the DG molecular layer of TLE human patients, implicating this enzyme in reactive synaptogenesis in the hippocampus [47]. A strong subgranular ecto-5′-nucleotidase/CD73 labeling was also seen by these authors in GFAP-stained cells of the DG of control and TLE patients. In keeping with these findings, we show here that ecto-5′-nucleotidase/CD73 is overexpressed in GFAP-positive astrocytes of the hippocampus of human MTLE patients and that this enzyme co-localizes, at least partially, with the A_2A receptor. Therefore, ecto-5′-nucleotidase/CD73 is positioned ideally to promote A_2A receptor activation after conversion of released adenine nucleotides into adenosine and, hence, its blockade may prevent adenosine-mediated neuronal excitation. Nonenzymatic properties of CD73 (e.g., adhesive properties to extracellular matrix components influencing cellular contacts) have also been demonstrated, but whether these features influence epileptic synaptogenesis and/or astrogliosis requires further investigations.

Adenosine has been considered the main endogenous antiepileptic molecule, via the activation of presynaptic inhibitory A₁ receptors, and it has been proposed that augmentation of adenosine levels could be a solution to resolve drug-refractory epilepsies [48–50]. Our study suggests that this must be taken carefully since increased extracellular levels of adenosine may not only activate inhibitory A₁ receptors [9], but can trigger proconvulsive effects via excitatory A_2A receptors in epileptic brain regions. Considering that the A_2A receptor is upregulated in the epileptic human hippocampus adenosine-mediated excitatory actions are more likely to occur in epileptic patients. Unbalance between inhibitory and excitatory effects of adenosine may be further exaggerated because binding to A₁ receptors may be reduced [51] and the expression of A₁ receptors may be downregulated in the hippocampus of intractable MTLE patients (manuscript in preparation). Experimentally, dynamic changes in A₁ receptor signaling or expression have been described as a direct consequence of acute seizures. In the cerebral cortex of neonatal rats, neuroprotective receptor density changes consisting of an increase in the A₁ receptor density and downregulation of the A_2A receptor, accompanied by a loss of 5′-nucleotidase activity, were only temporarily verified from 48 h to 5 days after hyperthermia-induced seizures [52]. It is, therefore, advisable when proposing adenosine augmenting strategies for treating epilepsy to consider simultaneous blockage of excitatory A_2A receptors with selective antagonists to avoid losing the anticonvulsive adenosine effect mediated by A₁ receptors.

Another important finding from this study concerns the demonstration that the A_2A receptor is localized predominantly on the plasma membrane of GFAP-positive hippocampal astrocytes of drug-resistant MTLE human patients in close proximity to ecto-5′-nucleotidase/CD73. A recent study reached a similar conclusion using epileptic rats injected with kainate and hippocampal samples from patients with Alzheimer’s disease [19]. This suggests a more general involvement of increased astrocytic A_2A receptors in diseases associated with neuroexcitotoxicity. On the other hand, we failed to demonstrate significant co-localization of the A_2A receptor with two neuronal cell markers, synaptotagmin 1/2 (nerve terminals) and NF200 (axon fibers), in the hippocampus of MTLE human patients. As a matter of fact, our data showed a negative correlation between the A_2A receptor immunofluorescence signal and the nerve terminal marker, synaptotagmin, while the staining overlap with the axonal marker, NF200, increased discreetly in CA1, CA2 and DG regions of the epileptic hippocampus compared to control tissues. Thus, our findings suggest that upregulation of the A_2A receptor parallels astrogliosis, which is considered a hallmark of hippocampal sclerosis in MTLE patients. The epileptogenic relevance of this phenomenon compared to the putative (even if modest) redistribution of A_2A receptors from injured hippocampal nerve terminals to aberrant sprouting axons deserves elucidation in future studies.

Epilepsy research has traditionally been focused on neurons, as is reflected by the number of antiepileptic drugs acting primarily at neuronal target proteins, but growing evidence is available on the roles played by astroglia in epilepsy phenomena leading to new alternatives for shaping reciprocal neuro-astroglia coupling in drug-refractory epilepsy [53]. Astrocytes influence the pathogenesis and pathophysiology of epilepsy by creating an excitatory feedback loop via the release of gliotransmitters, such as glutamate, D-serine, and ATP, and/or by acting upstream on the homeostatic control of uptake, degradation, and recycling of neurotransmitters and neuromodulators, including adenosine (reviewed in [48]). In addition to their local modulatory role to exacerbate or synchronize neuronal firing on shorter timescales of milliseconds to minutes, long-lasting volume and morphological changes (e.g., perisynaptic branches swelling, extension, or retraction of processes) of reactive astrocytes may transform these cells into aberrant synaptic network integrators due to inhibition of neurotransmitter clearance, defective potassium buffering capacity, altered Ca²⁺ signal propagation, and uncoupling of gap junction-mediated cell communication (reviewed in [54]). The astrocytic localization of the A_2A receptor strengthens a role for these cells in epilepsy, which may be associated with the proposed modulation of glutamate transport and release [11–13]. Activation of astrocytic A_2A receptor may result in increases in the extracellular levels of glutamate leading to neuronal excitation and/or excitotoxicity. The mechanism underlying the A_2A receptor-mediated control of glutamate outflow by astrocytes seems to involve activation of protein kinase A and intracellular Ca²⁺ mobilization [11], while the mechanism responsible for the downmodulation of glutamate uptake by the A_2A receptor depends on a decrease in Na⁺/K⁺-ATPase activity, and, thereby, the disruption of the transmembrane Na⁺ gradient [13] that is required for glutamate uptake by GLT-1 [55]. Although the intracellular transduction pathway implicated in the A_2A receptor-mediated inhibition of glutamate uptake in astrocytes may differ, the final result (i.e., collapse of the Na⁺-driving force to take up glutamate) is quite similar to that we have found to be involved in the P2X7 receptor-mediated downmodulation of glutamate uptake in cortical nerve terminals [33, 35]. This evidences a hazardous concerted action of high ATP levels activating nerve-terminal P2X7 receptors followed by stimulation of astrocytic A_2A receptors after ATP catabolism into adenosine by co-localized ecto-5′-nucleotidase/CD73. On the other hand, adenosine A_2A receptor-mediated GABA uptake by neighboring astrocytes may contribute to neuronal excitation [14], since astroglia dominate the buildup of extrasynaptic GABA that controls the glutamatergic neuronal excitatory drive. In a previous study, Cristovão-Ferreira et al. [30] elegantly demonstrated that higher extracellular adenosine levels, such as those occurring during epileptic crisis, favor astrocytic GABA uptake by switching the coupling of A₁/A_2A heteromers from inhibitory G_i to excitatory G_s protein proteins signaling. Mixed signal transduction and lateral stabilization of A₁/A_2A heteromers in the 2:2 conformation has been confirmed recently in heterologous systems [56], but the pathophysiological consequences of altered receptors stoichiometry motivated by epileptic changes is worth to be investigated in the near future. Notwithstanding this, it is safe to conclude that more emphasis should be put on astroglial targets, like GABA and glutamate transporter modulators, which may indirectly couple, via A_2A receptors, to and modulate neuronal function responsible for the severity of epileptic condition.

Temporal lobe epilepsy is thought to be triggered by an insult to the brain (e.g., brain injury, acute seizure, stroke, high fever, infection), which is followed by chronic processes required for the progression and maintenance of epileptogenesis. Both processes are tightly linked to disruption of adenosine homeostasis and astroglial dysfunction ultimately leading to hippocampal sclerosis that is a characteristic pathological hallmark of MTLE (reviewed in [48]). The acute or initiating phase (lasting 2–3 h) is characterized by an acute surge in adenosine associated with a transient downregulation of astrocytic adenosine kinase (ADK), which commands the cellular uptake of the nucleoside from the extracellular milieu. Besides controlling neuronal excitability, high levels of adenosine can transiently reduce epigenetic DNA methylation, which precipitates epileptogenesis by favoring the transcription and expression of initiating genes [57]. In humans, the following weeks or months are critical for epileptogenesis as inflammatory processes are activated leading to microglial and astroglial activation (reviewed in [58, 59]). The onset of chronic seizures coincides with the emergence of astrogliosis and overexpression of ADK, resulting in extracellular adenosine deficiency. As we show here in the human epileptic hippocampus, astrogliosis is not only associated with upregulation of ADK but also with increased expression of A_2A receptors and ecto-5′-nucleotidase/CD73 in GFAP-positive astrocytes (see also [12, 19]). Astrocytic upregulation of A_2A receptors together with ecto-5′-nucleotidase/CD73, the major adenosine producing ecto-enzyme in the brain, may be relevant because A_2A receptors play a significant role in the control of astrocyte physiology. Previous studies indicate that activation of A_2A receptors stimulates astrogliosis either, directly, via an Akt/NF-κB-dependent pathway [60], or triggered by brain-derived neurotrophic factor (BDNF) [61]. The mechanisms underlying maladaptive, yet possibly interdependent, processes leading to upregulation of ADK, A_2A receptors, and ecto-5′-nucleotidase/CD73 in astrocytes of the epileptic hippocampus remain elusive.

Under distinct conditions, the coordinated induction or repression of A_2A receptors and ecto-5′-nucleotidase/CD73 expression strongly supports the view that these two molecules are tightly interconnected [62, 63]. Upregulation of astrocytic A_2A receptors may be a compensatory reaction to reduced adenosine basal tone. Increased sensitivity to extracellular adenosine may also result from recruitment of latent intracellular receptors to the plasma membrane [64]. While previous studies associate internalization of A_2A receptors to abnormal increases in extracellular adenosine (see, e.g., [65]), more recent data ague against this by showing that A_2A receptors may be mobilized to the plasma membrane under anoxic conditions coinciding with high extracellular adenosine concentrations [66]. It is unclear which of these processes dominate during epileptogenesis, but our results clearly indicate that the A_2A receptor immunolabeling prevails on the plasma membrane of GFAP-positive hippocampal astrocytes of MTLE patients in detriment of a sparse intracellular receptor pool. Notwithstanding the mechanism leading to increased sensitivity to extracellular adenosine by A_2A receptors overexpression in astrocytic plasma membrane, it may be detrimental during epileptic seizures. This is because high-frequency neuronal firing increases dramatically ATP release and the nucleotide can be rapidly converted into adenosine by overexpressed ecto-5′-nucleotidase/CD73 near the receptor sites, thus enhancing the signal to noise ratio at excitatory synapses. On the contrary, increased A_2A receptors expression may trigger compensatory upregulation of ADK expression to limit reactive astrogliosis and neuronal excitation. Yet, this mechanism might be insufficient to compensate excessive adenosine formation from released ATP during epileptic seizures and, thus, self-reinforcement of the maladaptive circuit that promotes reactive astrogliosis, aberrant nerve sprouting and neuronal damage.

In conclusion, upregulation of A_2A receptors localized in hippocampal astrocytes of drug-resistant MTLE human patients suggest that blockade of these receptors with selective antagonists may prevent neuronal excitation and confer neuroprotection, as previously documented in epileptic animal models [10, 21, 22, 26]. Therefore, the hippocampal astrocytic A_2A receptor may be an attractive pharmacological target to increase seizure threshold and manage therapeutically drug-refractory MTLE in alternative to the pure neurocentric view that dominated antiepileptic drugs research for decades [67]. The exact mechanism by which the A_2A receptor promotes excitability in epileptic human patients remains unknown and difficult to investigate given the scarcity of human brain samples available for functional studies. Nevertheless, our findings makes it highly likely that modulation of glutamate transport operated by activation of overexpressed astrocytic A_2A receptors [11–13] might be involved, as it results in augmentation of extracellular glutamate levels and, thus, promote neuronal excitability/excitotoxicity in the human epileptic hippocampus. Upregulation of GABA uptake by astrocytes promoted by A_2A receptors activation may also concur to increase the severity of epileptic seizures and to drug-refractoriness. The tight association between ecto-5′-nucleotidase/CD73 and A_2A receptors expression in astrocytes of the epileptic human hippocampus allow us to consider the possibility that the manipulation of ecto-5′-nucleotidase/CD73 activity might also afford an antiepileptic benefit similar to selective A_2A receptor blockade. Moreover, given the hazardous concerted action of extracellular ATP accumulation and consequent adenosine formation favoring nerve-terminal P2X7 and astrocytic A_2A receptor activation in the epileptic human brain, one may speculate about the potential benefit of combining selective A_2A antagonists together with blockade of the P2X7 channel (see, e.g., [33, 35, 68–70]) to control intractable MTLE.

Abbreviations

ρ

Pearson’s coefficient

ADK

Adenosine kinase

BSA

Bovine serum albumin

BDNF

Brain-derived neurotrophic factor

DG

Dentate gyrus

GAPDH

Glyceraldehyde 3-phosphate dehydrogenase

GABA

γ-Aminobutyric acid

GFAP

Glial fibrillary acidic protein

MTLE

Mesial temporal lobe epilepsy

NF200

Neurofilament 200

PSD95

Postsynaptic density-95

SD

Standard deviation

SDS

Sodium dodecyl sulfate

Compliance with ethical standards

Funding

This study was supported by the University of Porto/Santander Totta, Liga Portuguesa Contra a Epilepsia (LPCE), Tecnifar and Fundação para a Ciência e Tecnologia (FCT, Fundo Europeu de Desenvolvimento Regional - FEDER funding and COMPETE, projects PIC/IC/83297/2007 and Pest-OE/SAU/UI215/2014). J. Sévigny received support from the Canadian Institutes of Health Research (CIHR, MOP – 93683, MOP – 102472). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. JMC was hired under the scope of FCT Portugal CIÊNCIA 2008 Programme (FSE-POPH-QREN, CONT_DOUT/117/ICBAS-UP/215/10824/2/2008); ABB was in receipt of a PhD studentship by FCT (SFRH/BD/79259/2011); JS was a recipient of a “Chercheur National” research award from the Fonds de Recherche du Québec–Santé (FRQS). The authors acknowledge the collaboration of Dr. Bárbara Leal in the collection of clinical information from patients with epilepsy. Authors also thank Mrs. M. Helena Costa e Silva and Belmira Silva for their technical assistance.

Conflict of interest

The authors declare that they have no conflict of interest.

Ethical approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed consent

Informed consent was obtained from all individual participants included in the study.