Upregulation of adenosine A2A receptor and downregulation of GLT1 is associated with neuronal cell death in Rasmussen's encephalitis

Abstract

Rasmussen encephalitis (RE) is a severe pediatric inflammatory brain disease characterized by unilateral inflammation and atrophy of the cerebral cortex, drug‐resistant focal epilepsy and progressive neurological and cognitive deterioration. The etiology and pathogenesis of RE remain unclear. Our previous results demonstrated that the adenosine A1 receptor (A1R) and the major adenosine‐removing enzyme adenosine kinase play an important role in the etiology of RE. Because the downstream pathways of inhibitory A1R signaling are modulated by stimulatory A2AR signaling, which by itself controls neuro‐inflammation, glial activation and glial glutamate homeostasis through interaction with glutamate transporter GLT‐1, we hypothesized that maladaptive changes in adenosine A2A receptor (A2AR) expression are associated with RE. We used immunohistochemistry and Western blot analysis to examine the expression of A2ARs, glutamate transporter‐I (GLT‐1) and the apoptotic marker Bcl‐2 in surgically resected cortical specimens from RE patients (n = 18) in comparison with control cortical tissue. In lesions of the RE specimen we found upregulation of A2ARs, downregulation of GLT‐1 and increased apoptosis of both neurons and astroglia. Double staining revealed colocalization of A2ARs and Bcl‐2 in RE lesions. These results suggest that maladaptive changes in A2AR expression are associated with a decrease in GLT‐I expression as a possible precipitator for apoptotic cell loss in RE. Because A2AR antagonists are already under clinical evaluation for Parkinson's disease, the A2AR might likewise be a tractable target for the treatment of RE.

Abbreviations

A2ARs

adenosine A2A receptor

CPS

complex partial seizure

EEG

electroencephalography

EPC

epilepsiapartialis continua

GTCS

generalized tonic clonic seizure

GFAP

glial fibrillary acidic protein

NeuN

neuronal nuclear protein

RE

Rasmussen encephalitis

SPS

simple partial seizure

Introduction

Rasmussen encephalitis (RE) is a severe pediatric inflammatory brain disease characterized by unilateral inflammation and atrophy of the cerebral cortex, drug‐resistant focal epilepsy (epilepsia partialis continua, EPC) and progressive neurological and cognitive deterioration, resulting from progressive loss of function that can be linked to the involved cerebral hemisphere 4, 5, 51, 61. The etiology and pathogenesis of this severely disabling inflammatory disease, in particular, the factors responsible for the characteristic asymmetry are still unclear. Therefore, targeted therapeutic strategies remain elusive. Currently, hemispherectomy is the only effective therapy to control the drug resistant seizures associated with RE but this treatment leads to inevitable postoperative functional deficits 34. Seizures are a prominent clinical feature of RE, whereas inflammatory processes likely play a crucial role in the etiology of the disease. In addition, clinically comorbid cognition deficits are among the most debilitating and persistent concerns of chronic epilepsy associated with RE 25. An intriguing neuropathologic feature of RE is the restriction of the inflammatory process to one brain hemisphere, setting it apart from any other inflammatory disease of the CNS. The pathological characteristics of RE consist of lymphocytic (perivascular lymphocytic cuffing) and microglial infiltrates, neuronal destruction and astrogliosis of the affected hemisphere 33, 34, 54. An antigen‐driven MHC class‐I restricted, CD8+ T cell‐mediated attack against neurons and astrocytes in the affected hemisphere is the leading hypothesis on the pathogenesis in RE 55. Our previous results in patients with RE demonstrated that activation of HMGB1/TLR signaling 26, 34, overexpression of the major adenosine‐removing enzyme adenosine kinase (ADK), and increased expression of the adenosine A1 receptor (A1R) 28, 33, 36, may play crucial roles in the generation of seizures, and possibly epileptogenesis itself. Adenosine is an inhibitory modulator of brain activity, and its anticonvulsant and seizure terminating effects, mediated by a balanced activation of inhibitory A1Rs and facilitatory adenosine A2A receptors (A2ARs), have been illustrated in experimental models of epilepsy 30, 31, 40, 46. Imbalance of adenosine receptor activation (decreased A1R signaling and/or increased A2AR signaling) contributes to the pathophysiology and development of epilepsy. A2ARs in the brain play a crucial role in synaptic plasticity and counteract the A1R‐mediated inhibition of synaptic transmission 14, 30, 31, 41, 47. After brain injury, A2AR expression is increased in both neurons and glial cells, and regarded as a STOP signal of the immune‐inflammatory system 20. Alterations in the cross‐talk between A1R and A2AR signaling are likely involved in seizure generation: prolonged activation of the high‐affinity A1Rs may lead to their internalization and a concomitant increase in A2AR expression 57. Prolonged activation of A2ARs in turn leads to inhibition of glutamate transporter‐I (GLT‐1) function, thereby compromising astroglial glutamate uptake, leading to an increase in the extracellular glutamate concentration as a plausible mechanism for the enhancement of neuronal excitability, epileptic seizures and neurotoxicity 42, 43. Because adenosine signaling plays a crucial role in RE, and because of the tight control of glutamate homeostasis through adenosine, we hypothesized that the etiology of RE is related to changes in A2AR expression and that those alterations might be associated with neuronal death and the development of epilepsy in patients with RE.

Material and Methods

Patients and diagnosis

All clinical investigations were conducted according to the declaration of Helsinki, and the local Ethics Committee (Sanbo Brain Hospital, Capital Medical University, China) approved all studies. Informed consent was obtained from parents or legal guardians of all participants. In this study, 18 childhood‐onset RE patients diagnosed with RE according to the typical clinical, MRI and neuropathological findings as proposed in 4 were enrolled in this study. Presurgical evaluation was performed according to previous procedures in Sanbo Brain Hospital, Capital Medical University33, 34, 35, 36, including MRI (spin‐echo T1‐weighted axial and T2‐weighted axial, coronal sequences and fluid‐attenuated inversion recovery images with 5‐mm‐thick axial, sagittal, coronal sections), fluorodeoxyglucose positron emission tomography (FDG‐PET), scalp video‐electroencephalography (EEG), intracranial EEG monitoring, seizure semiology analysis, as well as neuropsychological testing. Interictal/ictal scalp electroencephalography (EEG) was recorded using a video‐EEG monitoring system (Nicolet vEEG; Viasys Healthcare, Madison, USA), with electrodes placed according to the international 10–20 system for all patients. The duration of video‐EEG monitoring ranged from 1 to 3 days, and at least three habitual seizures were recorded during this time span. The main ictal manifestations were categorized according to the semiological seizure classification 1, 38.

Eighteen patients with RE (10 male and 8 female) were identified in the neuropathology archives at Sanbo Brain Hospital, diagnosed between January 2004 and January 2014, with a mean age of seizure onset of 5.27 ± 2.91 years (range: 1.67–11.83), a mean age at surgery of 7.88 ± 3.59 years (range: 2.42–14.5), a mean interval between the seizures and first operation 2.69 ± 2.45 years (range: 0.41–9.83), and a mean duration of follow up 5.28 ± 1.67 years (range: 3–8.5). All patients were Engel Class I 16. Epilepsy was the first clinical manifestation of the condition in all patients. None of the patients had a relevant perinatal history before the onset of the disease. Clinical details of the patient cohort are summarized in Table 1. Brain samples from all 18 RE patients were obtained during neurosurgical operations (functional hemispherectomies or anatomical hemispherectomies). Brain tissue (frontal cortex) obtained from all patients was paraffin embedded. In addition, parts of the brain samples were cryoprotected for Western blot analysis in 5 of these 18 patients. For comparison, six neocortical specimens (five frontal cortex, one occipital cortex) were obtained from well outside the epileptogenic lesion (normal appearing cortex/white matter adjacent to the lesion) in resections for intractable epilepsy in patients with focal cortical dysplasia. Of note, microscopic evaluation was performed on those areas to discard microscopic changes related to dysplasia, in order to avoid perilesion tissue within the border zone between the lesion and normal appearing cortex, as the extent of neuropathological abnormalities in this area may vary among subjects 36. See Table 1 for further demographic data.

Table 1.

Clinical details of RE patients and control cohort. EPC: epilepsia partialis continua; CPS: complex partial seizure; GTCS: generalized tonic clonic seizure; SPS: simple partial seizure.

Patient number	Gender	Age at first seizure (years)	Seizure types	Interval between seizure onset and EPC onset (years)	Interval between seizure onset and hemiplegia (years)	Interval between the seizure and first surgery (years)	Age at surgery (years)	Engel's grading	Follow‐up (years)
1	F	11.83	SPS	1.0	1.0	1.82
(residual stage)	13.75	I	6.33
2	F	1.92	SPS + CPS	0.42	0.42	0.50	2.42	I	6,50
3	F	11.08	SPS	1.0	1.0	1.50	12.58	I	5.83
4	M	4.67	SPS + CPS+GTCS	0.25	0.25	0.41	5.08	I	5.25
5	F	1.67	SPS + GTCS	1.25	1.17	1,33	3.0	I	4.0
6	M	5.33	SPS + GTCS	No EPC	0.67	3.09	8.42	I	5.08
7	F	2.0	SPS + CPS	1.25	1.25	1.5	3.5	I	3.0
8	M	5.08	SPS + CPS+GTCS	1.0	1.0	4.75	9.83	I	5.0
9	M	6.42	SPS + GTCS	0.75	0.75	1.66	8.08	I	6.75
10	M	4.67	SPS + CPS	1.0	1.25	9.83	14.5	I	3.5
11	M	2.75	SPS	0.17	0.25	4.58	7.33	I	4,5
12	F	5.58	SPS + CPS	0.25	0.25	3.0	8.58	I	6.42
13	M	3.67	CPS + GTCS	0	1.0	1.58	5.25	I	8.5
14	M	5	SPS	No EPC	1.0	5.33	10.33	I	8
15	M	5	SPS + CPS+GTCS	1.0	1.0	4.0	9.0	I	6.0
16	F	9.33	SPS + CPS	0.17	0.17	0.58	9.91	I	3.83
17	F	5.58	SPS	0.17	0.17	0.25	5.73	I	3.67
18	M	3.25	SPS + CPS	0	1.0	1.33	4.58	I	3.33
Control groups	Gender	Age at first seizure (years)	Seizure types	Diagnosis	Interval between the seizure and first surgery (years)	Age at surgery (years)
1	M	1.33	SPS + CPS	FCD IIB	2.67	4
2	M	2.08	SPS + GTCS	FCD IIB	1.75	3.83
3	M	0.92	SPS + GT1CS	FCD IIB	2.08	3
4	F	0.5	SPS + GTCS	FCD IIB	1.92	2.42
5	F	3	SPS + GTCS	FCD IIB	8	11
6	F	0.25	SPS + GTCS	FCD IIB	10	17

Tissue preparation

Formalin‐fixed, paraffin‐embedded tissue samples (one representative paraffin block per case containing the complete lesion or the largest part of the lesion resected at surgery) were sectioned at 4 μm and mounted on precoated glass slides (Star Frost, WaldemarKnittel GmbH, Braunschweig, Germany). Sections of all specimens selected in the regions of most severe inflammation were processed for hematoxylin eosin, as well as for immunohistochemistry for CD8, CD3, the neuronal marker NeuN, the astroglial marker glial fibrillary acidic protein (GFAP) and A2AR, GLT‐1 and Bcl‐2, which are described below. For Western blot analysis brain tissue from RE patients (n = 5) and surgical controls (n = 5) was snap frozen in liquid nitrogen and stored at −80°C until further use.

Immunohistochemistry

Fluorescein‐based terminal deoxynucleotidyltransferase‐mediated dUTP nick end‐labeling (TUNEL) (Roche Molecular Biochemicals, Indianapolis, IN) was used for the detection of DNA fragmentation as described previously 29, 30, 31, 37.

Antibodies to A2ARs (polyclonal rabbit ab124780, 1:700, Abcam; England; 1:70), GLT‐1 (polyclonal rabbit ab41621,1:100, Abcam, Cambridge), Bcl‐2 (mouse monoclonal, 1:100; Leica Biosystems Newcastle Ltd, UK), glial fibrillary acidic protein (GFAP; monoclonal mouse, DAKO; Glostrup, Denmark, 1:4000), Iba‐1 (monoclonal mouse ab15690, 1:300, Abcam), neuronal nuclear protein (NeuN; mouse clone MAB377, IgG1; Chemicon, Temecula, CA, USA; 1:2000), CD73 (mouse monoclonal, 1:100; Abcam, Cambridge), CD8 (mouse‐IgG, clone C8/144B, from DakoCytomation, Glostrup, Denmark; 1:50) and CD3 (rabbit‐IgG, clone SP7, from Lab Vision, Fremont, CA, USA, 1:500) were used in routine immunohistochemical analyses of RE human specimens. Sections were counterstained with hematoxylin. To evaluate the specificity of the staining the following control experiments were performed on paraffin‐embedded specimens according to a previous study 34, 65: (i)omission of the primary antibody; (ii) substitution of the primary antibody with a rabbit pre‐immune or non‐immune IgG or a monoclonal mouse IgG of irrelevant specificity; and (iii) pre‐absorption of polyclonal antibodies using an excess of antigen. Those control experiments resulted in the absence of staining. Double‐label immunohistochemistry was performed as described previously 11. After incubation with primary antibodies, the sections were incubated for 2 h at RT with Alexa Fluor® 568 and Alexa Fluor® 488 (anti‐rabbit IgG or anti‐mouse IgG; 1:200; Molecular probes, Eugene, USA). Images were visualized using a Leica microscope under Ex/Em wavelength of 500/550 nm (green), collected using an Optronics DEI‐750 three‐chip camera equipped with a BQ 8000 sVGA frame grabber, and analyzed using Bioquant software (Nashville, TN).

For the semiquantitative immunohistochemical detection of GLT‐1, Bcl‐2 and CD73, we followed our previously published procedures with modifications 29, 30, 31, 32, 34, 35, 36, 37. Two sections from RE patients (n = 18) and controls (n = 6) each were analyzed by scanning GLT‐1 and bcl‐2 immunoreactivity on DAB‐stained slices and the mean area for quantitative analysis was a 2 × 3 mm² field using a Kodak imaging device. Levels of GLT‐1, Bcl‐2 and CD73 were initially quantified as arbitrary density units and subsequently expressed as a percent change from control measurement.

In situ hybridization

RNA in situ hybridization was performed using RNAscope technology according to the manufacturer's instructions for the RNAscope 2.5 HD Reagent Kit‐Brown (Advanced Cell Diagnostics). Paraffin‐embedded tissue sections were deparaffinized by incubating for 60 min at 60°C and endogenous peroxidases were quenched with H₂O₂ for 10 min at room temperature. Slides were then boiled for 15 min in RNAscope Target Retrieval Reagents and incubated for 30 min in RNAscope Protease Plus and hybridized with probes against ADORA2A RNA (Advanced Cell Diagnostics, catalog #500081) and counterstained with hematoxylin and visualized by bright field microscopy.

Western blot analysis

Preparation and analysis of Western blots were performed as described previously 33, 35 with modifications. Freshly frozen histologically normal surgical control cerebral cortex and RE samples (n = 5 per group) were homogenized in lysis buffer containing 10 mMTris (pH 8.0), 150 mM NaCl, 10% glycerol, 1% NP‐40, Na‐orthevanadate (10.4 mg/ml), 5 mM EDTA (pH 8.0), 5 mM NaF and protease inhibitor cocktail (Boehringer Mannheim, Germany). Protein content was determined using the bicinchoninic acid method. For electrophoresis, equal amounts of proteins (30μg/lane) were separated by sodium dodecylsulfate‐polyacrylamide gel electrophoretic (SDS‐PAGE) analysis. Separated proteins were transferred to nitrocellulose membrane for 1 h and 30 min, using a semi‐dry electroblotting system (BioRad, Transblot SD, Hercules, CA, USA). Blots were incubated overnight in TTBS (20 mMTris, 150 mMNaCl, 0.1% Tween, pH 7.5) / 5% non‐fat dry milk, containing the primary A2AR antibody (polyclonal rabbit ab124780, 1:700, Abcam; England). After several washes in TTBS, the membranes were incubated in TTBS/5% non‐fat dry milk/1% BSA, containing goat anti‐rabbit or goat anti‐mouse antibodies coupled to horse radish peroxidase (1:2500; Dako, Denmark) for 1 h. After washes in TTBS, immunoreactivity was visualized using Lumi–light PLUS Western blotting substrate (Roche Diagnostics, Mannheim, Germany) and digitized using a Luminescent Image Analyzer (LAS‐3000, Fuji Film, Japan). Expression of β‐actin (monoclonal mouse antibody, Sigma, St. Louis, MO, USA 1:50.000) was used as reference. A2AR levels were normalized to internal standards and reported relative to control.

Statistical analysis

Statistical analyses were performed with SPSS for Windows (SPSS 11.5, SPSS Inc., Chicago, IL, USA) using two‐tailed Student's t‐test. P < 0.05 was considered significant.

Results

Clinical features

The clinical features of the cases included in this study are summarized in Table 1. Eighteen patients with RE identified in the neuropathology archives at Sanbo Brain Hospital, diagnosed between January 2004 and January 2014, were recruited in the study. All 18 patients underwent surgical resection for intractable seizures. 17 of 18 patients in the study cohort were in the acute stage of RE and one was in the residual stage, according to Bien's stage‐wise course of RE 5, 27.The seizure types of the patients in the present study included simple partial seizures (SPS), epilepsia partialis continua (EPC), complex partial seizures (CPS) and secondary generalized tonic‐clonic seizures (GTCS). EPC was defined as regular or irregular clonic muscular twitches affecting a limited part of the body, occurring for a minimum of 1 h and recurring at intervals of no more than 10s 29. In the present study, 16 of 18 patients in the study had EPC which appeared 0.61 ± 0.47 (range: 0.00–1.25) years after the first seizure onset. All patients in the study exhibited a focal motor deficit, which appeared 0.76 ± 0.39 (range: 0.25–1.25) years after the first seizure onset (Table 1). MRI and long‐term video EEG results were abnormal in all cases.

The features of inflammatory pathology in RE patients were characterized by the formation of microglial infiltrates (Figure 1A), perivascular cuffing (Figure 1B) and parenchymal infiltrates of lymphocytes (Figure 1C). Double labeling of CD3 and CD8 indicated that the majority of the infiltrating lymphocytes were CD8+/CD3+ positive T‐lymphocytes (Figure 1D–F). In addition, we found neuronal cell loss (Figure 1G), ectopic neurons in the white matter (Figure 1H), enlarged dysmorphic neurons in the white matter (Figure 1I, arrow), occurrence of radial microcolumns in the cortex (Figure 1J), reactive astrogliosis in gray matter (Figure 1K) and white matter (Figure 1L).

Figure 1.

Characteristic neuropathologic findings in brain tissue from RE patients. A. Microglial infiltrates and diffuse microglial activation in the cerebral cortex (arrow, H&E stain). B. Perivascular lymphocytic cuffing in the cerebral cortex (arrow, H&E staining). C. Parenchymal lymphocytic infiltrates in the cerebral cortex (arrows). D–F. Co‐localization of CD3 and CD8 immunoreactivity in lymphocytic infiltrates in the cerebral cortex (D, CD3 immunostaining; E, CD8 immunostaining; F, co‐localization of CD3 and CD8). G. Neuronal loss (arrows, NeuNimmunostaining). H. Ectopic neurons in the white matter (arrow, NeuNimmunostaining). I. Enlarged dysmorphic neurons in the white matter (arrow, inset, NeuN immunostaining). J. Radial microcolumns in the cortex (arrow, NeuN immunostaining). K–L. Reactive astrogliosis (GFAP immunostaining) in gray matter (K) and white matter (L). Scale bar: (A–C, G–L) 25 μm; (D–F) 12.5 μm.

Increased Expression of A2ARs Within the Lesions of RE

The expression of A2ARs was studied by immunohistochemistry in brain specimens of RE patients and surgical control. Consistent with low expression levels of the A2AR in the cerebral cortex 53 we found only negative or weak A2AR immunoreactivity in control gray (Figure 2A,E) and white (Figure 2B,F) matter. In contrast, we discovered overexpression of A2AR in specimen from RE patients, which were broadly characterized by (I) globally (neurons and astrocytes) increased expression of A2ARs in lesional areas and (II) ectopic expression in the cytoplasm (Figure 2G–J). Specifically, within the gray matter of RE lesions marked perinuclear (cytoplasmic) staining of A2ARs became evident in remaining neurons (Figure 2G, arrow and inset). In the white matter from RE specimens, strong A2AR immunoreactivity was observed in hypertrophied astrocytes (Figure 2D,H, arrow, inset), especially in perivascular hypertrophied astrocytes (Figure 2J). In addition, endothelial cells within the lesions of the cortex displayed high levels of A2AR immunoreactivity (Figure 2J, arrow). Marked cytoplasmic staining of A2ARs in ectopic neurons (enlarged dysmorphic neurons) was detected in the white matter (Figure 2I, arrows). In addition, in situ hybridization demonstrated that the mRNA expression of A2ARs increased in RE compared to control specimens (Figure 3).

Figure 2.

Expression of A2ARs within the lesions of RE. A,B,E,F. Weak immunostaining for A2ARs in control cortical gray matter (A,E) and white matter (B,F). Negative or weak expression in neurons in cortical gray matter (E, arrow, inset) and sparse expression on glial cells in control white matter (F, arrow, inset). C,G. Marked ectopic cytoplasmic expressionof A2ARs in the remaining neuronal cells (G, arrow, inset)in gray matter within the lesion of RE. D,H. Marked ectopic cytoplasmic expression of A2ARs in reactive astrocytes (H, arrow, inset) in white matter within the lesion of RE. C and D shows absence of neuronal or astroglial immunoreactivity after pre‐absorption respectively. I,J. Marked expression of A2ARs in perivascular reactive astrocytes in white matter (I,J), and endothelial cells within the lesions of the cortex (J, arrow). In addition, marked cytoplasmic staining of A2ARs in ectopic neurons (enlarged dysmorphic neurons) was detected in the white matter (I, arrow). Scale bars: (A–D) 50 μm; (E–H) 25 μm, (I,J) 6.25 μm.

Figure 3.

In situ hybridization to assess A2AR mRNA within the lesions of RE. A: Marked expression of A2AR mRNA within the lesions of RE; B: Very weak or negative expression of A2AR mRNA in control specimen. Scale bars: 12.5 μm.

Double labeling demonstrated ectopic cytoplasmic expression of A2ARs in NeuN‐positive neurons (Figure 4A–D, arrows) and in GFAP‐positive reactive astrocytes (Figure 4E–H, arrows) within the lesion area of RE. We also detected A2AR‐positive activated microglial cells (Iba‐1) within the lesional cortex (Figure 4I–L, arrows). Western blot analysis was performed to quantify the total amount of A2ARs in homogenates of surgically retrieved cortex from RE patients and surgical control cortex (Figure 5A). Compared to control specimens, a significant increase (P < 0.05) in A2AR expression was demonstrated in RE (Figure 5B).

Figure 4.

A2AR expression in neurons, astrocytes and microglia in RE (case 8). A‐D. Co‐locolization of NeuN (A, green) and A2ARs (B, red) demonstrated ectopic cytoplasmic expression of A2ARs in the remaining neurons within the lesion area (D, arrow). E–H. Co‐localization of GFAP (E, green) and A2ARs (F, red) indicated ectopic cytoplasmic expression in reactive astrocytes within the lesion area (H, arrows). I–L. Co‐localization of Iba‐1 (I, green) and A2ARs (J, red) within the lesion area (L, arrows). Merged images (D,H,L, yellow) confirm localization of the A2AR in neurons, astrocytes and microglia (C,J,K). Representative micrographs of nuclear staining with DAPI. Scale bars: 15 μm.

Figure 5.

Western blot analysis of A2ARs in RE and control specimens. A. Representative immunoblots of A2ARs in total homogenates of RE and control specimens. β‐actinimmunoreactivity was used to normalize for equal protein loading. B. Quantitative analysis of A2AR expression levels. Data are mean ± SEM (n = 5 per group). *P < 0.05 vs control.

Reduced Expression of GLT‐1 Within Lesions of RE

Similar to previous results 3, 60, we found a patchy distribution of GLT‐1 immunoreactivity in control gray matter, characterized by a regular expression profile with strong homogenous abundance in glial cell processes (Figure 6A,E). In control white matter, only few GLT‐1 expressing astrocytes were found (Figure 6B,F). However, in white and gray matter of RE no discernible or only sparse GLT‐1 expression was observed, (Figure 6C,D,G,H, arrows, inset). Quantification of immunodensities revealed a significant decrease in GLT‐1 immunoreactivity within the gray matter of RE specimens (n = 18) compared to control (n = 6) (Figure 6I).

Figure 6.

Downregulation of the GLT‐1within the Lesions of RE. A,E. In control gray matter, marked staining and patchy distribution of GLT‐1 immunoreactivity detected in glial cell processes (E, inset, arrows). B,F. Sparse GLT‐1 expression on astrocytes in control white matter (F, inset, arrows). In specimens of RE, no detectable or only sparse GLT‐1 expression was detected on astrocytes in gray matter of RE (C,G, inset arrows) and white matter (D,H, inset arrows). Scale bars: A–D, 50 µm; E‐H, 25 µm. (I) Scanned immunodensities of GLT‐1 expression show asignificant decrease (P < 0.01) in GLT‐1 expression within the RE (n = 18) grey matter compared with controls (n = 6).

Reduced Expression of Bcl‐2 Within Lesions of RE

In control tissue, Bcl‐2 immunoreactivity was mainly localized within the neuronal cytoplasm (Figure 7A,E, inset, arrows), or astrocytic cytoplasm (Figure 7B,F, inset, arrows) of gray and white matter, respectively. In RE, however, no detectable or only sparse Bcl‐2 immunoreactivity was detected in gray and white matter (Figure 7C, 7, G, H). Quantitative analysis demonstrated a significant decrease in the immunodensity of Bcl‐2 within the RE (n = 18) grey matter compared to controls (n = 6) (Figure 7I).

Figure 7.

Downregulation of Bcl‐2 within RE lesions. A,B,E,F. Bcl‐2 immunoreactivity in neurons of control grey matter (A,E, inset, arrows) and in the astrocytes within control white matter (B,F, inset, arrows). C and G. No detectable (G, inset, arrows) or sparse weak neuronal expression (G, inset, arrowheads) of Bcl‐2 in gray matter of RE. D and H. no detectable (H, inset, arrows) or sparse weak astrocytic expression (H, inset, arrowheads) in white matter of RE. Scale bars: A–D, 50 µm; E–H, 25 µm. (I) Immunohistochemistry analysis demonstrated a significant decrease (P < 0.05) in the density of Bcl‐2 immunoreactivity within the RE (n = 18) gray matter compared with controls (n = 6).

In gray matter of RE (Figure S1A–D), no detectable (Figure S1D, arrow) or minimal colocalization of A2AR and bcl‐2 was detected in a few cells with neuronal morphology (Figure S1D, arrowhead). In white matter of RE (Figure S1E–H), no colocalization of A2ARs and bcl‐2 in cells with astrocytic morphology was demonstrated (Figure S1H, arrows).

Colocalization of A2AR with TUNEL in Lesions of RE

Next, we assessed apoptosis by evaluating the expression of TUNEL. Both in gray matter (Figure 8A–D) as well as in white matter (Figure 8E–H) we found an abundance of TUNEL‐positive cells. Every TUNEL positive cell showed also A2AR immunoreactivity suggesting a strong association between high cytoplasmic levels of the A2AR and apoptosis.

Figure 8.

Co‐localization of A2ARs and TUNEL labeling in RE lesions (case 8). A–D. Co‐localization of A2AR (red) (A) with TUNEL (green) (B) (D, arrows) in the gray matter of RE. E–H. Co‐localization of A2ARs (red) (E) with TUNEL (green) (F) (H, arrows) in the white matter of RE. All slices were conterstained with the nuclear marker DAPI. Scale bars: 15 μm.

Increased Expression of CD73 within the Lesions of RE

We also studied the expression of CD73 by immunohistochemistry in brain specimens of RE patients and surgical control. Overexpression of CD73 was found in both gray matter (Figure 9A,E)) and whiter matter (Figure 9B,F) from specimens of RE patients. In contrast, only weak or negative CD73 immunoreactivity was observed in control gray (Figure 9C,G) and white matter (Figure 9D,H). Quantification of immunodensities revealed a significant increase in CD73 immunoreactivity within the lesion of RE specimens (n = 18) compared to control (n = 6) (Figure 9I).

Figure 9.

Expression of CD73 within the lesions of RE. A,E. Marked ectopic cytoplasmic expression of CD73 in the remaining neuronal cells (E, arrow, inset) in gray matter within the lesion of RE. B and F. Marked cytoplasmic expression of CD73 in reactive astrocytes (F, arrow, inset) in white matter within the lesion of RE. C,D,G,H. Very weak or negative immunostaining of CD73 in control cortical gray matter (C,G) and white matter (D,H). I. Immunodensities of CD73 expression show a significant decrease (P < 0.05) in CD73 expression within the RE (n = 18) compared with controls (n = 6). Scale bars: (A–D) 50 μm; (E‐H) 25 μm.

Discussion

RE is a rare, chronic progressive inflammatory neurological disorder of uncertain etiology affecting mostly children. It is associated with hemispheric atrophy, pharmacoresistant focal epilepsy, cognitive deterioration and progressive neurological deficits. Even though etiology and pathogenesis of RE are still unclear, increasing experimental and clinical evidence supports a link between inflammation, epilepsy (both in terms of epileptogenesis and the long term consequences of seizures), and cognitive dysfunction associated with epilepsy. Those pathological findings are in line with a prominent role of inflammatory processes in various epileptic disorders 50. In line with inflammatory processes, disruption of adenosine metabolism and signaling has been associated with epilepsy and its cognitive comorbidities 7, 43, 46. Adenosine, acts as an endogenous neuromodulator with potent anti‐ictogenic 8, 21, anti‐epileptogenic 32, 34, 36, anti‐inflammatory 45, 64 and pro‐cognitive 56 functions. Our previous data indicated that overexpression of adenosine kinase and resulting adenosine defciency in sclerotic lesions of the RE brain can be an important contributor for the development of pharmacoresistant focal seizures, inflammation and cognitive deterioration 33.

Increased Immunoreactivity of A2ARs in RE

Neuronal excitability in the brain is modulated by activation of G protein coupled adenosine receptors (A1, A2A, A2B, A3) 33. Therefore, both the receptor expression levels as well as the availability of endogenous adenosine to activate the receptors, play a crucial role in the regulation of neuronal excitability 6. The anticonvulsant efficacy of adenosine has been demonstrated in various rodent models of epilepsy, including a model of pharmacoresistant epilepsy 21, 39. It has been demonstrated that the anticonvulsant activity of adenosine in vivo is mediated by activation of the G protein coupled A1R 29, 30, 31, 32, 35. Through activation of pre‐synaptic A1Rs, adenosine modulates neuronal activity by decreasing the presynaptic release of various neurotransmitters, in particular glutamate 52. Through activation of postsynaptic A1Rs adenosine activates K+ channels, leading to a hyperpolarization of postsynaptic neurons and NMDA receptor inhibition 63. In addition, as depolarizing GABA(A)R signaling has been implicated in seizure initiation and progression, the adenosine‐induced attenuation of depolarizing GABA(A)R signaling may represent an important mechanism for seizure inhibition 24. Our previous results indicated that increased neuronal A1R expression in RE may be an endogenous adoptive response of the brain to prevent seizure spread and seizure‐induced brain damage beyond the affected cerebral hemisphere 18, 36. In contrast to the established role of A1Rs in epilepsy, there are only few and controversial literature reports on the role of the A2AR in epilepsy 11. The predominant opinion is that the activation of A2ARs exerts proconvulsant activity 22. In the present study we found striking overexpression of A2ARs in reactive astrocytes, microglial cells and remaining neurons in the focal lesions in RE. In addition, endothelial cells in walls of blood vessels within the lesion cortex displayed A2AR positive immunostaining, suggesting a possible role of endothelial A2ARs in the inflammatory response. Notably, we found a major ectopic expression of the A2AR in the cytoplasm of neurons and astrocytes. These unexpected findings are in line with the literature describing ectopic overexpression of A2ARs in the cytoplasm of neurons and astrocytes in pathological situations such as models for Alzheimer's and Parkinson's disease 23, 42, 43. The role and pathological implications of cytoplasmic A2AR expression are currently unknown. The high levels of cytoplasmic A2AR expression precluded us from quantifying A2AR expression in the plasma membrane, separately. We propose that once membrane insertion of the receptor is saturated, any excess of the receptor accumulates in the cytoplasm. An increase in transmembrane A2ARs is expected to have a pro‐epileptic net effect, based on several studies reported in the literature. In an in vitro study, the application of the adenosine A2AR antagonist ZM241385 shortened the duration of epileptiform activity, suggesting that the activation of A2ARs may compromise the A1R‐mediated anticonvulsant properties 17. The intensities of pentylenetetrazol‐ or pilocarpine‐induced seizures, as well as the percentages of convulsing mice, were significantly reduced in A2AR knockout mice, indicating that activation of A2ARs modulates excitatory neurotransmission and exacerbates limbic seizures 15. In line with those findings, the increased expression of A2ARs in neurons and reactive astrocytes in lesioned areas of the brain of RE patients, suggests a disruption of synaptic plasticity, proconvulsant activity, and progression of brain inflammation 13, 20.

Adenosine and adenosine receptors are increasingly recognized as crucial therapeutic targets for controlling cognition under normal and disease conditions because of their dual roles as neuromodulators and metabolic regulators of the brain 9. In addition to mediating seizure inhibition, adenosine has been identified as an important regulator of behavior and disruption of adenosine metabolism has been linked with cognitive and psychiatric phenotypes 6. Being a key upstream modulator of major neurotransmitter systems including glutamatergic and GABAergic neurotransmission, adenosine acts as a crucial regulator of cognitive processes 28. Our previous data indicated that pathological overexpression of the major adenosine‐metabolizing enzyme adenosine kinase might play a crucial role in the development of cognitive comorbidities in RE 28, 33. In line with those findings, the increased expression of astroglial A2ARs in RE might additionally affect cognitive function through a novel mechanism involving astrocyte‐driven neuronal adaptation processes 28. The effects of A2ARs on memory are based on the ability of A2ARs to affect working memory and especially reference memory performance 23. Overexpression of A2ARs and the pharmacological activation of A2ARs are sufficient to trigger memory impairment 13. On the other hand, the pharmacological or genetic blockade of A2ARs impedes memory deterioration 13. Adenosine A2AR antagonists prevent a delayed memory deficit resulting from a single convulsive episode in early life 12. Dysregulation of A2AR expression in the brain of RE patients thus might contribute to the cognitive deterioration in RE patients.

The anti‐inflammatory properties of adenosine in the brain are largely mediated by A2ARs and to a lesser degree by less abundant A3Rs 66, but only indirect evidence has been provided that this pathway is activated in the brain of RE patients 28. A2ARs have been demonstrated to increase in expression both in neurons and glial cells upon injury, and to act as a STOP signal of the immune‐inflammatory system 28. Therefore, A2ARs can affect glial reactivity and control neuroinflammatory processes 20. In addition, A2ARs are also located in endothelial cells of brain capillaries, where they play an important role in controlling brain vascular function 48. In the present study, the overexpression of A2ARs in reactive astrocytes as well as in endothelial cells, suggests an association with the neuroinflammatory processes of RE.

A recent study indicated that the extracellular CD73‐mediated formation of adenosine provides an important source for the activation of A2AR, as demonstrated by the coordinated overexpression of CD73 and A2AR in Parkinson's disease models 44. In the present study, we observed overexpression of CD73 in the specimens of RE, implicating that increased CD73‐mediated adenosine formation might contribute to the activation of A2ARs in RE.

Association of Increased A2ARs and Decreased GLT‐1 with Cell Death in RE

The interaction of A2ARs with GLT‐1 modulates glutamate signaling 41, 42, 43, and thereby may influence cell death and cognitive processes associated with RE. In primary cultured astrocytes as well as in an ex vivo preparation enriched in glial plasmalemmal vesicles, it has been demonstrated that the prolonged activation of A2ARs leads to the reduction of the activity and expression of GLT‐1, and thereby to a sustained decrease in glutamate uptake 42. It has been suggested that the dysfunction of astrocytic A2ARs triggers an astrocyte‐to‐neuron wave of communication resulting in disrupted glutamate homeostasis via modulation of GLT‐1 activity 31. This mechanism contributes to an excessive extracellular accumulation of glutamate, and is considered to play a major role in the evolution of most neurodegenerative disorders 2. Because excessive concentrations of extracellular glutamate play a crucial role in excitotoxicity and neuronal cell death, the GLT‐1 mediated glutamate uptake by astrocytes is an endogenous mechanism to prevent excitotoxicity and cell death 58. In the present study, astrocytes of the RE brain showed a concomitant reduction of GLT‐1 with an increase in A2AR expression. In line with a downregulation of GLT‐1 as precipitator for increased glutamate‐mediated neurotoxicity, we also found deficits in Bcl‐2 expression and increased TUNEL immunoreactivity in RE brains. Bcl‐2 is a well characterized protective signaling molecule preventing cell death either through anantiapoptotic activity 19, 49 or through a pro‐survival activity 50. The coexpression of A2ARs in RE brain in TUNEL positive cells, suggests the interesting hypothesis that dysregulation of A2AR expression in RE may contribute through increased neuronal cell death through a mechanism including deficits in GLT‐1 expression, concomitant increases in glutamate‐mediated excitotoxicity, and deficiency in Bcl‐2 mediated anti‐apoptotic mechanisms. The phenomenological findings described here in samples from human RE patients, justify future mechanistic studies in suitable model systems.

In conclusion, the increased expression of A2ARs in RE described here may contribute to the epilepsy phenotype in RE, as well as to inflammation, neuronal atrophy, and cognitive deficiency. If increased expression of A2ARs plays a functional role in the development of RE, then, A2ARs maybe a therapeutic target in the treatment of RE. Because A2AR antagonists are already in clinical development 10 and considered as promising neuroprotective agents 47, 59, 62, they may provide a spectrum of benefits for RE patients, including anti‐ictogenic, neuroprotective, anti‐inflammatory and pro‐cognitive activities.

Authors' contributions

Immunohistochemistry, western blot, as well as the analysis of the data were performed by He X, Chen F, Yifan Zhang, Gao Q, Wang J, Chen Y and Luan G. He X, Chen F helped Li T in drafting and preparing the manuscript for submission. The overall experimental design was conceived and supervised by Li T. Guan Y, Zhou J and Zhai F helped in the selection and collection of brain tissues. Boison D provided advice and co‐wrote the manuscript. All authors read and approved the final manuscript.

Conflict of interests

The authors declare that they have no conflict of interests. DB is co‐Founder of PrevEp LLC, and served as scientific consultant to Hoffman LaRoche AG.

Supporting information

 

Figure S1. Double labeling A2ARs and bcl‐2 in RE lesions. (A‐D) No detectable (D, arrow) or minimal co‐localization of A2ARs and bcl‐2 in a few cells with neuronal morphology in the grey matter of RE (D, arrowhead). (E‐H) Lack of co‐localization of A2ARs and bcl‐2 in the white matter of RE in cells with an astrocytic morphology (H, arrows). Scale bars: 15 μm.

Data Availability Statement

No additional unpublished data are available.