Seizure Suppression by Adenosine A1 Receptor Activation in a Mouse Model of Pharmacoresistant Epilepsy


  • Accepted April 6, 2003.

Address correspondence and reprint requests to Dr. D. Boison at Institute of Pharmacology and Toxicology, University of Zurich, Winterthurerstr. 190, CH-8057 Zürich, Switzerland. E-mail:


Summary: Purpose: Because of the high incidence of pharmacoresistance in the treatment of epilepsy (20–30%), alternative treatment strategies are needed. Recently a proof-of-principle for a new therapeutic approach was established by the intraventricular delivery of adenosine released from implants of engineered cells. Adenosine-releasing implants were found to be effective in seizure suppression in a rat model of temporal lobe epilepsy. In the present study, activation of the adenosine system was applied as a possible treatment for pharmacoresistant epilepsy.

Methods: A mouse model for drug-resistant mesial temporal lobe epilepsy was used, in which recurrent spontaneous seizure activity was induced by a single intrahippocampal injection of kainic acid (KA; 200 ng in 50 nl).

Results: After injection of the selective adenosine A1-receptor agonist, 2-chloro-N6-cyclopentyladenosine (CCPA; either 1.5 or 3 mg/kg, i.p.), epileptic discharges determined in EEG recordings were completely suppressed for a period of ≤3.5 h after the injections. Seizure suppression was maintained when 8-sulfophenyltheophylline (8-SPT), a non–brain-permeable adenosine-receptor antagonist, was coinjected systemically with CCPA. In contrast, systemic injection of carbamazepine or vehicle alone did not alter the seizure pattern.

Conclusions: This study demonstrates that activation of central adenosine A1 receptors leads to the suppression of seizure activity in a mouse model of drug-resistant epilepsy. We conclude that the local delivery of adenosine into the brain is likely to be effective in the control of intractable seizures.

Despite optimal treatment with presently available antiepileptic drugs (AEDs) (1), seizures persist in ∼35% of patients with partial epilepsy (2). In addition, pronounced side effects may limit the most favorable use of the available AEDs (3). Although patients with drug-resistant forms of focal epilepsy may undergo resective surgery (4), many patients cannot be treated surgically because of unacceptable risks of loss of brain functions (5). Other strategies for the treatment of refractory epilepsy include vagus nerve stimulation (6), focal cooling (7), and gamma knife therapy (8), but have not led to a major improvement of the therapeutic options for patients with refractory epilepsy. Complex partial seizures, especially those originating from the mesial temporal lobe (mesial temporal lobe epilepsy, MTLE), are the most frequent types of seizures in humans and display the highest percentage of drug resistance (9). A strategy that prevents seizures in drug-resistant epilepsy would be an important therapeutic advance.

Unfortunately, the traditional animal models of MTLE, such as those based on systemic applications of high doses of pilocarpine or kainate, display fundamental differences in the pattern of cell loss or in the nature of seizures compared with those in humans. Recently a novel model was developed that closely replicates the histopathologic changes of human MTLE in the mouse (10,11). In this model, chronic recurrent seizures are resistant to classic AEDs (12).

The model is based on a unilateral injection of a low dose of kainate into the dorsal hippocampus of mice. After a latent period of 2 weeks, spontaneous seizures begin to occur, accompanied by progressive cell loss in the CA1 and CA3 regions of the hippocampus over a period of 4 to 6 weeks (10,13). The immediate death of a subpopulation of γ-aminobutyric acid (GABA)ergic interneurons and of mossy cells underlies the onset of recurrent seizures, whereas the long-term degeneration of pyramidal cells might be owing to recurrent seizures (10). The model represents drug-resistant epilepsy, because hippocampal paroxysmal discharges were not responsive to short-term carbamazepine (CBZ), phenytoin (PHT), or valproate (VPA) treatment (12). This model is therefore suitable to investigate the antiseizure activity of novel treatment approaches in drug-resistant epilepsy.

The fact that most patients with refractory epilepsy are resistant to several AEDs acting by different mechanisms argues against epilepsy-induced alterations in specific drug targets. It rather suggests nonspecific or adaptive mechanisms, such as an acquired overexpression of multidrug transporters in the blood–brain barrier (14). Recent evidence demonstrated an overexpression of P-glycoprotein (PGP) and members of the multidrug resistance–associated protein (MRP) family in capillary endothelial cells and astrocytes in epileptogenic brain tissue resected from patients with pharmacoresistant epilepsy (15). Therefore, new therapeutic strategies, which take advantage of endogenous antiepileptic mechanisms of the brain, are less likely to be diminished by the activity of multidrug transporters. For that reason, in the search for new therapeutic approaches for drug-resistant focal epilepsies, the use of adenosine as an endogenous inhibitory neuromodulator (16) appears to be promising.

Adenosine has inhibitory as well as excitatory functions (17,18) mediated through activation of four subtypes (A1, A2A, A2B, and A3) of adenosine receptors (19). Whereas A1 and A3 receptors interact with pertussis toxin–sensitive G proteins of the Gi and Go family, thus mediating mainly inhibitory functions of adenosine, the signaling mechanism of A2A and A2B includes stimulation of adenylyl cyclase via Gs, thus mediating stimulatory effects of adenosine (18). In the brain, the release of various neurotransmitters, in particular of excitatory amino acids, is inhibited through presynaptic A1 receptors (19,20), which are linked via Gi/o proteins to Ca2+ and K+ ion channels (21). Postsynaptically, activation of A1 receptors leads to a stabilization of the membrane potential by modulation of Ca2+ and K+ fluxes (22). In particular, a tonic modulation of the excitatory synaptic input in the hippocampus seems to be exerted by a background activity of G protein–activated inwardly rectifying K+ channels that results from the tonic activation of A1 receptors by ambient adenosine (23). In humans, microdialysis studies demonstrated a release of adenosine during seizure activity, and adenosine was proposed to be a natural mediator of seizure arrest and of postictal refractoriness (24). Pharmacologically, adenosine and its analogues acting on adenosine A1 receptors have potent inhibitory effects on neuronal activity (17) and are effective in seizure suppression and neuroprotection (25–29).

However, it has not been reported previously that adenosine would be efficient in the treatment of pharmacoresistant epilepsy. Here we show that systemic application of the adenosine A1-receptor agonist 2-chloro-N6-cyclopentyladenosine (CCPA) indeed leads to the suppression of epileptic discharges in a mouse model of drug-resistant epilepsy.


Animals and surgery

Experiments were conducted on 12 male NMRI mice (35–40 g; Harlan Netherland, Horst, The Netherlands) housed in individual cages in a 12-h light–dark cycle (light on from 07:00 to 19:00). Food and water were provided ad libitum. All animal procedures were conducted in accordance with the regulations of the local animal-welfare authority. All efforts were made to minimize animal suffering and to reduce the number of animals used.

Under general anesthesia (equithesin, 4 ml/kg i.p.) the experimental mice (n = 12) were stereotactically injected with 50 nl of a 20 mM solution of kainic acid (KA) in 0.9% NaCl (i.e., 1 nmol KA) into the right dorsal hippocampus (coordinates with bregma as reference: anteroposterior (AP) =−1.5, mediolateral (ML) =−1.8, dorsoventral (DV) =−1.9 mm) by using a stainless steel cannula (outer diameter, 0.28 mm) connected to a l-μl microsyringe (Hamilton, Bonaduz, Switzerland) via PE20 tubing filled with sterile water. Each injection was performed over a period of 1 min. At the end of the injection, the cannula was left in place for an additional 1-min period to limit reflux along the cannula track.

All mice were then implanted with a bipolar electrode inserted into the injected hippocampus and a monopolar surface electrode placed over the cerebellum (reference electrode). The bipolar electrode was formed of two twisted-enamel insulated stainless steel wires (diameter, 170 μm; distance between the tips, 0.4 mm) connected to a male connector. It was aimed at the right hippocampus with the same coordinates as for the injection site. The monopolar electrode was made of the same enamel-insulated stainless steel wire (diameter, 250 μm) soldered on a male connector (Wire pro, Farnell, France). It was inserted into the skull so that only the tip (0.5 mm) protruded onto cerebellar tissue. The electrodes were fixed to the skull with cyanoacrylate and dental acrylic cement. The mice were then allowed to recover from anesthesia before being placed in the EEG recording chamber. To provide a control for the antiepileptic efficacy of CBZ, a group of genetically epilepsy prone rats (GEPR-9s; n = 8) was used. The animals were tested as described previously (29).

Electroencephalographic recordings

EEG activities of freely moving animals placed in a Faraday cage were recorded by using a digital acquisition computer-based system (MP100WSW System; Biopac Systems, Inc., Santa Barbara, CA, U.S.A.; six channels; sampling rate, 200 Hz). Before starting the EEG recordings, a period of 1 h was allowed for the habituation of the animals to the test cage. Status epilepticus was recorded for ≤3 h after the end of surgery in all KA-injected mice (n = 12). Each animal was then recorded 1 week after surgery during the latent phase of epileptogenesis and again 2 weeks after KA injection when the chronic phase of epilepsy became evident. These recordings were taken for a period of 3 h during the resting phase of the animals (14:00–17:00).

Drug administration

The effects of the AED CBZ and of the adenosine A1-receptor agonist CCPA on the frequency of recurrent seizures were assessed in mice exhibiting typical hippocampal paroxysmal discharges beginning ∼2 weeks after KA injection. CBZ (RBI, Natick, MA, U.S.A.) was first dissolved in Molecusol (2-hydroxypropyl-cyclodextrin; RBI) and then in distilled water at a final concentration of 15 mg/ml. A dose of 30 mg/kg, i.p., was injected into each test animal (n = 12 injections). CCPA was dissolved in 20% dimethylsulfoxide (DMSO) in 0.9% saline (vol/vol). To test the antiseizure efficacy of CCPA, two different doses were injected: 1.5 and 3 mg/kg, i.p.; n = 18 injections, each. Control injections of vehicle [20% DMSO in 0.9% saline (vol/vol)] alone were performed a total of 6 times. To exclude potential effects of the activation of peripheral A1 receptors by CCPA, 8-sulfophenyltheophylline (8-SPT), a nonselective non–brain permeable adenosine-receptor antagonist, dissolved in 10% polyethylene glycol (PEG) in saline (vol/vol), was used. In control experiments, 8-SPT (20 mg/kg, i.p.) was injected 15 min before CCPA or vehicle administration (n = 6 for each dose of CCPA, or vehicle). All pharmacologic compounds were tested in a randomized order during the third and fourth weeks after KA treatment, when all animals were in the chronic phase of seizure activity. After a reference EEG-recording period of 2 hours, the animals were treated accordingly and were recorded for another 6 hours. Drugs were administered 3 times per week in single doses in a randomized order; thus each animal received a total of six injections and was used as its own control, with a delay of ≥2 days between injections.

Rat model of audiogenic seizures

The seizure susceptibility of GEPR-9s (30) was tested by auditory stimulation. Individual animals were placed in a plastic cylinder (40-cm diameter), which was mounted in a sound-proof box (50 × 50 × 50 cm). After 60 s for habituation, the animals were exposed to a convulsive sound stimulus (120 dB, assembled from frequencies of 12–16 kHz). The stimulus was applied for 30 s or until wild running occurred. Seizure response was assessed as previously described (30) on a seizure scale ranging from 0 (no response) to 9 (one running phase followed by a complete tonic convulsion of the body, including hindlimb extension).

Histologic controls

After completion of the experiments (5 weeks after KA injection), mice were transcardially perfused with 4% paraformaldehyde and 15% saturated picric acid solution in phosphate buffer (0.15 M, pH 7.4). The brains were removed and cryoprotected in 10% DMSO in phosphate-buffered saline (PBS; vol/vol). The brains were then cut into 40-μm coronal sections by using a sliding microtome. Histologic analysis was performed after cresyl violet staining to verify (a) the location of the KA injection, (b) the location of the hippocampal electrode, and (c) the pattern of neuronal loss and the dispersion of dentate gyrus granule cells, as previously reported (11,31).

Analysis of drug effects

The efficacy of seizure suppression after drug injection was assessed by counting and comparing the number of seizures in EEG recordings from the same animal in a 1-h period before and after drug injection. In addition, the average length of the seizures was determined.


Development of progressive chronic epilepsy after intrahippocampal kainic acid injection

In the mouse kainate model of epilepsy, as a consequence of the KA injections, three distinct stages of epileptogenesis become evident in EEG recordings (12): (a) status epilepticus on awakening from surgery, (b) latent phase, ≤2 weeks after KA injection, and (c) chronic phase from 2 weeks to several months after KA injection.

To set the baseline for the following drug-treatment studies, EEGs were monitored during the three distinctive stages of epileptogenesis.

  • 1As a direct consequence of the KA injection, status epilepticus lasting for ≥3 h became evident in EEG recordings taken on awakening from anesthesia (Fig. 1a).
  • 2After cessation of the status epilepticus, a latent period of ≤2 weeks followed. During this time, only occasional single spikes were noticed in the EEG (Fig. 1b).
  • 3After this latent period, spontaneous recurrent seizure activity developed in all of the mice during the second week after KA injection. The seizure activity was characterized by repeated discharges lasting ∼41 ± 10 s, separated by interictal periods in which only isolated interictal spikes were observed (Fig. 1c). On average ∼22 ± 5 of these seizures were observed per hour. This recurrent seizure activity is considered to be of focal origin (12). This seizure pattern remained constant during the course of the experiments. In contrast, secondarily generalized seizures, which became apparent with clonic movements of the forelimbs, were observed only rarely and therefore not suited for a systematic analysis. A total of only three secondarily generalized seizures was observed during the course of the experiments.
Figure 1.

Representative EEG recordings displaying the development of progressive chronic epilepsy after intrahippocampal kainic acid (KA) injection. After the injection of KA, EEG recordings from the ipsilateral hippocampus were performed during three distinct stages of epileptogenesis. a: The status epilepticus was characterized by typical, continuous spikes and waves lasting for ≤10 h after KA injection. b: The latent phase during the first 2 weeks after KA injection was characterized by only occasional spikes. c: The chronic phase commenced 2 weeks after the injection of KA and was characterized by repeated discharges separated by interictal periods. The chronic seizure pattern remained stable until the termination of the experiments (5 weeks after KA injection).

Pharmacoresistance in kainic acid–treated mice

The mouse KA model of epilepsy is considered to be a model of pharmacoresistant epilepsy, because intraperitoneal injections of CBZ, VPA, and PHT failed to suppress seizure activity during the chronic phase of epilepsy (12). To provide a control for the pharmacoresistance in our group of KA-treated mice, CBZ (30 mg/kg, i.p.) was injected at different time points 3 and 4 weeks after KA injection (n = 12). Control animals received vehicle injections (n = 6). In EEG recordings taken during a period of 1 h preceding vehicle or CBZ injections, recurrent seizure activity was evident, with an average of 21.7 ± 5.4 seizures per hour, lasting 47.3 ± 13.0 s in the vehicle control group (Fig. 2a, Table 1), and with an average of 20.4 ± 5.1 seizures per hour, which lasted 39.4 ± 8.5 s, in the CBZ group. This seizure activity was not diminished after the injection of vehicle (Fig. 2b, Table 1) or CBZ (Fig. 2c, Table 1), with an average of 18.2 ± 9.0 seizures per hour, each lasting 52.2 ± 12.6 s after saline injection and with an average of 22.7 ± 4.2 seizures, each lasting 44.3 ± 9.5 s after CBZ injection (Fig. 3, Table 1). This experiment demonstrated the failure of CBZ to suppress seizure activity in KA-treated mice. This pharmacoresistant epileptic activity can thus be used as a baseline for the comparison of antiepileptic effects achieved by experimental AEDs. To rule out a general lack of efficacy of our dose and preparation of CBZ, the drug was tested in a different model of epilepsy. For this control we used GEPR-9s, an animal model of audiogenic seizures (30) (n = 8), which reproducibly displayed grade 9 seizure activity after each audiogenic test stimulus, as described previously (28). After injection of CBZ (30 mg/kg, i.p.), seizures were transiently suppressed, with mean seizure grades of 0.7 ± 0.3, 1.2 ± 0.2, and 9.0 ± 0 determined by the application of an audiogenic test stimulus 1, 3, and 24 h after drug injection, respectively. This control experiment indicates the general antiepileptic efficacy of the drug and dosage used.

Figure 2.

Representative EEG recordings taken from two animals after the intraperitoneal injection of antiepileptic drugs. During a time frame of 3–4 weeks after kainic acid treatment the drugs under study were administered 3 times per week in single doses in a randomized order. The animals were recorded for a total of 2 h before and 6 h after drug injection. Representative sections of 5 min duration (top traces) are displayed. Close-up views of 30 s each are displayed in the bottom traces. The recordings displayed were performed (a) 1 h before vehicle injection; (b) 1 h after vehicle injection; (c) 1 h after carbamazepine injection (30 mg/kg, i.p.); (d) 1 h after 8-SPT injection (20 mg/kg, i.p.) co-injected with vehicle; (e) 1 h before CCPA (1.5 mg/kg, i.p.) injection preceding recording g; (f) 1 h before CCPA (3.0 mg/kg, i.p.) injection preceding recording h; (g) 1 h after CCPA injection (1.5 mg/kg, i.p.); (h) 1 h after CCPA injection (3.0 mg/kg, i.p.); (i) 1 h after coinjection of 8-SPT (20 mg/kg, i.p.) with CCPA (1.5 mg/kg); (j) 1 h after coinjection of 8-SPT (20 mg/kg, i.p.) with CCPA (3 mg/kg, i.p.). The recordings a, b, f, h, j, and the recordings c, d, e, g, i were taken from two individual animals. CBZ, carbamazepine; 8-SPT, 8-sulfophenyltheophylline; CCPA, 2-chloro-N6-cyclopentyladenosine.

Table 1. Paired comparison of seizure activity in KA-treated mice before and after drug injection
Drug treatmentSeizures/h before injectionSeizures/h after injectionStatistical comparisonaAverage seizure length before injection[s]Average seizure lengthafter injection[s]Statistical comparisona
  1. Seizure activity in KA-treated mice was monitored in EEG recordings performed during the phase of chronic seizure activity (3 to 4 weeks after KA injection). The total number of seizures and the average length of each seizure were determined within a 1-h period before drug injection and compared with the total number of seizures and the average length of each seizure within a 1-h period beginning 15 mins after drug injection. Statistical analysis of paired treatment groups (before versus after drug injection) was performed by Wilcoxon signed-ranks tests. The respective p values are given. Errors are given as ±SDs.

  2. Vehicle, 20% DMSO in 0.9% saline; CBZ, carbamazepine; CCPA, 2-chloro-N6-cyclopentyladenosine; 8-SPT, 8-sulphophenyltheophylline.

  3. a Because of background noise in EEG recordings during parts of the 1-h intervals before or after drug injection, 13 paired recordings were not analyzed in detail.

Vehicle (n = 6)   21.7 ± 5.418.2 ± 9.0p = 0.2747.3 ± 13.052.2 ± 12.6p = 0.07
   (n = 4)    (n = 4)(n = 4)   
CBZ (30mg/kg; n = 12)   20.4 ± 5.122.7 ± 4.2p = 0.2439.4 ± 8.544.3 ± 9.5p = 0.04
   (n = 8)     (n = 8)   
CCPA (1.5 mg/kg; n = 18)   22.2 ± 5.40p = 0   41.7 ± 10.90p = 0   
   (n = 17)    (n = 17)  
CCPA (3.0 mg/kg; n = 18)   24.4 ± 3.40p = 0   45.0 ± 11.80p = 0   
   (n = 15)    (n = 15)  
8-SPT (20 mg/kg; n = 6)   21.0 ± 6.722.5 ± 6.4p = 0.2639.7 ± 7.643.1 ± 4.2p = 0.29
   (n = 4)     (n = 4)   
8-SPT + CCPA (1.5 mg/kg; n = 6)19.5 ± 9.10p = 0   35.8 ± 8.00p = 0   
   (n = 6)     (n = 6)   
8-SPT + CCPA (3.0 mg/kg; n = 6)   20.2 ± 5.40p = 0   39.3 ± 8.40p = 0   
   (n = 5)     (n = 5)   
Figure 3.

Suppression of carbamazepine-resistant seizures by central adenosine A1–receptor activation. The total number of seizures in EEG recordings were counted within a 1-h period before drug injection and compared with the total number of seizures within a 1-h period beginning 15 min after drug injection. Vehicle, 20% dimethylsulfoxide in 0.9% saline; CBZ, carbamazepine; CCPA, 2-chloro-N6-cyclopentyladenosine; 8-SPT, 8-sulfophenyltheophylline. Error bars: ±SD.

Suppression of pharmacoresistant seizure activity by adenosine A1-receptor activation

To evaluate the potential of adenosine A1-receptor activation for the suppression of pharmacoresistant seizure activity, the selective A1-receptor agonist CCPA (32) was used. KA-treated mice were injected with two doses of CCPA (1.5 and 3 mg/kg, i.p.) during the phase of chronic pharmacoresistant seizure activity (3–4 weeks after KA-injection; Fig. 2e and g). In contrast to recordings taken during 1 h before drug injection (Fig. 2e and f, Table 1), seizures were completely suppressed for 2.8 ± 0.9 h after the injection of the low dose of CCPA (1.5 mg/kg, i.p.) and for 4.0 ± 0.9 hours after the high dose of CCPA (3 mg/kg, i.p.; Fig. 2g and h; Table 1). Thus seizure suppression appeared to be dose dependent, because 3 mg/kg CCPA led to a longer-lasting suppression of epileptic discharges and interictal spikes than did injections of 1.5 mg/kg. Seizure suppression was transient and was subsequently restored to preinjection levels; therefore each animal acted as its own control. The complete lack of seizures 1 h after the injection of CCPA (Fig. 3, Table 1) is in striking contrast to the preinjection seizure activity, which was characterized by 22.2 ± 5.4 seizures per hour lasting 41.7 ± 10.9 s, for the low-dose CCPA group, and by 24.4 ± 3.4 seizures lasting 45.0 ± 11.8 s for the high-dose CCPA group (Table 1). Most important, after two different doses of CCPA, seizure activity was completely suppressed, whereas injections of vehicle or CBZ failed to be efficient in seizure control. We conclude that activation of the adenosine system is sufficient to suppress pharmacoresistant seizure activity.

Seizure suppression by CCPA is mediated by central A1 receptors

Apart from the CNS, adenosine A1 receptors are prominent in the cardiovascular system (19). Activation of these peripheral receptors may lead to a reduction of heart rate, blood pressure, and body temperature (33) and thus may contribute to seizure suppression. Although we did not observe any obvious peripheral side effects of CCPA at the concentrations used, to rule out a peripheral contribution of the seizure-suppressive potential of CCPA, a control experiment was performed by coinjecting CCPA with the nonselective adenosine-receptor antagonist 8-SPT, which cannot cross the blood–brain barrier (34). Injections with 8-SPT (20 mg/kg, i.p.; n = 18) were followed after 15 min by an injection of CCPA (either 1.5 or 3 mg/kg; n = 6 each; Fig. 2i and j; Table 1) or the corresponding vehicle (n = 6; Fig. 2d; Table 1). It was demonstrated that in combination with 8-SPT, the CCPA-induced seizure suppression was maintained and comparable with the seizure suppression achieved when the respective doses of CCPA were given alone (Figs. 2g–j and 3). In contrast, injection of 8-SPT with vehicle did not alter the seizure response significantly, with preinjection values of 21.0 ± 6.7 seizures per hour lasting 39.7 ± 7.6 s compared with postinjection values of 22.5 ± 6.4 seizures per hour each lasting 43.1 ± 4.2 s (Table 1). We conclude that seizure suppression by CCPA in drug-resistant epilepsy is mediated by activation of central adenosine A1 receptors.

Histologic controls

To verify that in all experimental mice used in the current study, chronic seizure activity was accompanied by corresponding KA-induced hippocampal lesions that characterize this model, brain sections of all treated mice (n = 12) were analyzed histopathologically 5 weeks after KA injection, as described (11,12). In all cases, a typical pattern of hippocampal damage was evident in the injected brain hemisphere, characterized by extensive neuronal loss in the CA1 and CA3c regions and in the hilus of the dentate gyrus (Fig. 4b). In addition, an enlargement of the dentate gyrus accompanied by a prominent dispersion of the granule cells was observed in all animals. These changes were restricted to the injected hippocampus (Fig. 4b), and no histopathologic changes were observed in the contralateral brain hemisphere (Fig. 4a). In all animals, the tip of the hippocampal electrode was located in the enlarged upper blade of the dentate gyrus (Fig. 4b). The histologic analysis described here shows that recurrent chronic seizure activity in KA-treated mice is accompanied by cellular rearrangements in the hippocampus reminiscent of those occurring in human MTLE.

Figure 4.

Morphologic alterations after unilateral injection of kainic acid (KA) into the dorsal hippocampus, as seen in transverse sections stained with cresyl violet at, 4 weeks after injection. a: Contralateral, noninjected hippocampus, with the normal histology of the hippocampal formation. b: Ipsilateral, KA-injected side. A dispersion and enlargement of the dentate gyrus (DG) and CA3 region is evident. Pycnotic cells are discernible in the hilus of the DG. The location of the bipolar intrahippocampal electrode is indicated with an asterisk (*). Scale bar, 0.5 mm.


Several investigations support the notion that adenosine acts as an endogenous anticonvulsant (17). After the onset of seizures, adenosine levels in the brain increase rapidly, and its levels have been shown to remain elevated postictally (24). This may be an endogenous mechanism to prevent further seizure development and/or propagation (35). Thus, in a wide range of experimental paradigms, pharmacologic activation of adenosine A1 or A2A receptors reduced convulsions induced by a variety of chemical and electrical stimuli (29,35–38). Despite their profound anticonvulsive effects, adenosine-receptor agonists have not proved clinically useful because of peripheral, mainly cardiovascular side effects (38). Effective strategies that make use of the anticonvulsant effects of adenosine-receptor activation may require a novel approach. It was recently demonstrated that focal seizures elicited in the hippocampus of fully kindled rats can be suppressed by engineered adenosine-releasing cells grafted near the seizure focus (25,28). In the context of a future patient-oriented adenosine-based therapy, it is important to assure that adenosine is able to suppress pharmacoresistant seizures. Although antiseizure effects of the A1-receptor agonist 2-chloroadenosine have been described in rats kindled in the amygdala (39,40), a model of TLE in which PHT responders and nonresponders have been described (41), to our knowledge, adenosine-receptor activation has never been compared directly with a standard AED in a model of pharmacoresistant epilepsy. We now demonstrate for the first time by direct comparison with a standard AED that activation of the adenosine system leads to the suppression of pharmacoresistant seizure activity. This is of importance because the high proportion of pharmacoresistant epilepsy (20–30%) continues to remain a significant health issue.

In the present study, we chose a mouse model of pharmacoresistant epilepsy in which standard AEDs such as CBZ, VPA, and PHT consistently fail to be effective in the suppression of seizure activity (12). In this mouse model, the injection of a tiny amount of KA into the dorsal hippocampus leads to the development of recurrent chronic seizure activity, which is associated with characteristic cellular rearrangements within the injected hippocampus (Fig. 4). The progress of hippocampal cell degeneration after a KA injection is well characterized (10,11,42,43) and was shown to be associated with the development of pharmacoresistant seizure activity (12).

We were now able to demonstrate that KA-induced pharmacoresistant seizures can be suppressed by CCPA, an adenosine A1 receptor–selective agonist, whereas CBZ fails to reduce seizure activity in the same animals (Figs. 2 and 3; Table 1). CCPA (38) has previously been used to suppress seizure activity in various models of epilepsy (29,36–38). The doses used here (1.5 and 3 mg/kg) are in the range of those used to achieve seizure suppression in a rat model of audiogenic brainstem epilepsy (29). CCPA-mediated seizure suppression was dose dependent and could not be diminished by co-injecting the non–brain-permeable adenosine-receptor antagonist 8-SPT (Figs. 2 and 3). These experiments demonstrate for the first time directly that activation of central adenosine A1 receptors by CCPA is sufficient for the suppression of pharmacoresistant seizure activity. The results of this study thus go beyond those conducted with 2-chloroadenosine in amygdale-kindled rats, a model of TLE in which CBZ induced increases in focal seizure threshold in all kindled rats (41).

In pharmacoresistant epilepsy, the efficacy of conventional AEDs appears to be diminished by multidrug transporters, which have evolved for the excretion of naturally occurring toxins and xenobiotics (14). Whereas most AEDs are potential substrates for efflux carriers of the blood–brain barrier, such as PGP and MRPs (14), the antiepileptic efficacy of adenosine or adenosine-related compounds is not likely to be diminished by efflux through the blood–brain barrier for the following three reasons:

  • 1Adenosine and related purine nucleosides are transported from blood to brain via carrier-mediated transport across the blood–brain barrier. Recently the adenosine transporter of the blood–brain barrier was cloned and identified as CNT2 (44). In contrast to other sodium-dependent cotransporters described at the blood–brain barrier thus far, which are primarily efflux systems (45), CNT2 is one of the first sodium-dependent cotransporters that mediate transport of substrate in the blood-to-brain direction (44). Because of the existence of this concentrative mechanism of blood–brain barrier adenosine transport, adenosine and related compounds are unlikely subjects for efflux mechanisms thought to be operative in pharmacoresistant epilepsy.
  • 2Because adenosine concentrations within the brain appear to be primarily regulated by the activity of adenosine kinase, an “enzymatic” blood–brain barrier for circulating adenosine has been postulated (46). Therefore, an efficient efflux system for adenosine did not need to be evolved.
  • 3Finally, the highly hydrophilic ribose moieties of adenosine and its analogues could prevent these compounds from being eliminated by the drug-transport systems thought to be operative in pharmacoresistant epilepsy.

Despite their profound antiseizure activity in a variety of models, A1-receptor agonists, when given systemically, are not potential AEDs because of profound peripheral, mainly cardiovascular, effects (38). However, as demonstrated in this and in a previous study (29), the systemic coapplication of a brain-permeable adenosine-receptor agonist with a non–brain-permeable adenosine-receptor antagonist, such as 8-SPT, may provide a means to circumvent cardiovascular side effects of systemic adenosine-related drugs.

In conclusion, our data demonstrate that activation of central adenosine A1 receptors leads to the suppression of seizures in a mouse model of pharmacoresistant epilepsy. The results presented here, together with our previous findings on the antiseizure effect of adenosine-receptor activation in rat models of epilepsy (25,26,28,29), suggest that the local delivery of adenosine into the brain is likely to be effective in the control of intractable seizures. This study thus supports a potential anticonvulsant therapy for pharmacoresistant epilepsy mediated by adenosine-releasing cell grafts.


Acknowledgment:  This work was supported by grant 31-59109.99 of the Swiss National Science Foundation and by the NCCR on Neural Plasticity and Repair.