Anticonvulsant Actions of Gap Junctional Blockers in an In Vitro Seizure Model

Shokrollah S. Jahromi, Kirsten Wentlandt, Sanaz Piran, Peter L. Carlen


Gap junctions (gjs) are increasingly recognized as playing a significant role in seizures. We demonstrate that different types of gap junctional blocking agents reduce the duration of evoked seizure-like primary afterdischarges (PADs) in the rat in vitro CA1 hippocampal pyramidal region, following repetitive tetanization of the Schaffer collaterals. Intracellular acidosis, which is known to block gap junctional communication, decreased the PADs, whereas alkalinization increased the PADs. Cellular excitability was not significantly depressed as determined by input/output relations recorded before and during perfusion of the gj blockers blockers carbenoxolone and sodium propionate. There was a small decrease following 1-octanol perfusion and a large decrease following NH4Cl application. Carbenoxolone diminished PAD duration, but increased neuronal excitability in whole-cell recordings. After robust PADs were established, the expression of several gj proteins including connexins (Cxs) 26, 32, 36, and 43, as measured by Western blotting, was unchanged, although the level of nonphosphorylated Cx43 was decreased. Our data support the concept that blocking gap junctional communication is an anticonvulsant mechanism.


It is now apparent in many neuronal systems that direct physical coupling between neurons via gap junctions is an important mode of intercellular communication. Gap junctional connections permit the passage of small molecules and electrical current (for reviews see Bennett 1996, 2000;Spray and Dermietzel 1996) and have been studied in various parts of the CNS (Church and Baimbridge 1991;Galarreta and Hestrin 1999; Gibson et al. 1999; Hormuzdi et al., 2001; MacVicar and Dudek 1981; Perez Velazquez et al. 1994, 1997;Schweitzer et al., 2000; Valiante et al. 1995). Gap junctions are dynamic structures that can be modulated by increased neuronal activity (Rouach et al., 2000; Yang et al. 1990) and by a large number of intracellular and extracellular factors (Church and Baimbridge 1991; Perez Velazquez et al. 1994; Rorig et al. 1996).

Current evidence strongly suggests a role for gap junctional communication in neuronal synchrony and seizure generation under normal and pathological conditions. The rate of interictal bursts, induced in CA1 neurons by perfusion with a medium containing 0 Mg2+ and 4-aminopyridine, was significantly reduced by octanol and carbenoxolone, both gap junctional blockers (Ross et al., 2000). Synchronized neuronal activity recorded in the absence of chemical synaptic transmission, in the network of cerebellar molecular layer inhibitory neurons, has been shown to depend on gap junctional communication (Mann-Metzer and Yarom 1999). Gap junction blockers, such as sodium propionate and octanol, inhibited spontaneous field burst activities in CA1 pyramidal neurons in a 0 Ca2+ model of epilepsy; i.e., in the absence of chemical synaptic transmission (Perez Velazquez et al. 1994). The involvement of gap junctions in neuronal synchrony and seizures recently has been discussed byDudek et al. (1998), Carlen et al. (2000), Perez Velazquez and Carlen (2000), andTraub et al. (2001).

Here we investigated the role of gap junctional communication in the maintenance of epileptiform discharges, in which seizure-like activities were evoked in CA1 pyramidal neurons by repetitive tetanic stimulation of Schaffer collaterals (Jahromi et al., 2000; Rafiq et al. 1993). Our hypothesis was that gap junctional blockers would depress evoked seizures in an in vitro seizure model, which, if correct, opens up a novel therapeutic target for anticonvulsant therapy.


Slice preparation

Brain slices were prepared from Wistar Rats, 25 to 40 days of age. Animals were anesthetized with halothane and decapitated in accordance with the guidelines of the Animal Care Committee. The brain was quickly removed and placed in ice-cold, continuously oxygenated (95% O2-5% CO2), sucrose-based artificial cerebrospinal fluid (ACSF) for 2–4 min. This solution contained (in mM): 210 sucrose, 26 NaHCO3, 2.5 KCl, 1 CaCl2, 4 MgCl2, 1.25 NaH2PO4, and 10 glucose. The sucrose-based high Mg2+-containing ACSF was used only during the preparation of slices to reduce sodium and neurotransmitter-dependent neurotoxicity and dissection-induced damage to neurons (Jahromi et al., 2000; Rafiq et al. 1993; Rasmussen and Aghajanian 1990). Entorhinal cortex/hippocampus slices were prepared as follows: the brain was hemisected along the midsagittal line and the cerebellum and the forebrain were removed. The dorsal cortex then was cut parallel to the longitudinal axis and the brain was fixed, ventral side up, to an aluminum block with cyanoacrylate glue. The aluminum block was secured at a 12ο angle in a Vibratome (Series 1000, Technical Products International, St. Louis, MO) with the caudal end of the brain facing the blade. Slices (500 μm thick) were incubated for at least 1 h before being transferred to an interface-type chamber for electrophysiological recording. The ACSF used for incubation and for perfusion of slices in the recording chamber was continuously oxygenated and had the following composition (in mM): 125 NaCl, 26 NaHCO3, 5 KCl, 1.8 CaCl2, 0.9 MgCl2, 1.25 NaH2PO4, and 10 glucose. While in the recording chamber, the slices were also aerated with humidified oxygen. The composition of the ACSF used for recording extracellular and whole-cell properties of patch-clamped neurons was the same. At the end of each experiment, the nonhippocampal parts of each slice were dissected away in the recording chamber before freezing it for Western blotting.

Electrophysiological recording

Schaffer collaterals were stimulated by a bipolar electrode (enamel-insulated nichrome wire, 125 μm diameter) placed in the stratum radiatum. Orthodromic extracellular responses were recorded from CA1 pyramidal neurons with borosilicate glass pipettes filled with NaCl (150 mM) and placed in stratum pyramidale. Constant current stimulating square pulses of 100-μs duration were delivered with varied amplitude by a Grass S88 Stimulator (Grass Instruments, Quincy, MA). Responses were filtered (3 kHz), amplified, and recorded with an Axoclamp 2A amplifier (Axon Instruments, Foster city, CA) operating in the bridge mode. Stimulating pulses were applied at five different intensities to evoke subthreshold potential to maximal suprathreshold population spike amplitudes. Input/output (I/O) relations were recorded prior to tetanization, after tetanization, and following drug perfusion. Data acquisition and analyses were performed using pClamp version 6.0.3 software (Axon Instruments).

Sustained epileptiform discharges, previously referred to as primary afterdischarges (PADs) (Jahromi et al. 2000) were evoked in CA1 pyramidal neurons by tetanic stimulation of Schaffer collaterals with repetitive pulses of 100-μs duration at 100 Hz for 2 s (Rafiq et al. 1993). Tetanization of slices was repeated once every 10 min, 6 to 10 times, prior to the application of the drug. The tetanization protocol was also carried out during drug application. PADs were continuously digitized and recorded on videotape (VR-10B, Instrutech Corporation) and later analyzed with software (WCP V1.2) provided by Dr. John Dempster, University of Strathclyde.

Whole-cell recordings were made from CA1 pyramidal neurons following 6–10 episodes of tetanization prior to the application of carbenoxolone and also during the perfusion of slices with this drug. Neurons were patch clamped with borosilicate glass tubing (2 mm OD, 0.5 mm wall thickness, World Precision Instruments, New Haven, CT) filled with a solution containing (in mM): 140 potassium methylsulfate, 2 Mg-ATP, and 10 HEPES (270–280 mOsm, pH 7.2 adjusted with KOH). The resistance of the recording electrodes filled with this intracellular solution ranged from 3 to 5 MΩ. All experiments were performed at 34οC.

The resting membrane potential (RMP) of CA1 neurons in tetanized slices perfused with high-potassium (5 mM) ACSF ranged from −57 to −60 mV. However, prior to recording cellular properties, the RMP in all cells was adjusted to −60 mV by the injection of constant current through the recording pipette. The membrane input resistance (R i) was calculated based on the amplitude of the voltage deflection recorded following the application of hyperpolarizing current pulses (350-ms duration, 0.1 nA amplitude) through the patch-clamp electrode. TheR i was monitored throughout the experiment. Current pulses (350 ms), ranging from −0.1 to 0.6 nA in amplitude, were applied and the resulting (passive) voltage deflections were recorded, generating current/voltage (I-V) plots. Depolarizing current pulses (350 ms), varying in amplitude, were applied and the frequency of action potentials at each current step was measured. Changes in neuronal spike-frequency adaptation were studied by plotting the amplitude of the applied current versus the frequency of action potential firing before and after the application of carbenoxolone. Spike voltage threshold was measured from the first spike evoked following the injection of depolarizing pulses (0.5–1 ms) with varying amplitude. Afterhyperpolarization potentials (AHP) were evoked by the application of suprathreshold depolarizing pulses of 100-ms duration evoking a spike train from a membrane potential of −45 mV.


Carbenoxolone (0.2 mM), ammonium chloride (10 mM), 1-octanol (0.2 mM), sodium propionate (25 mM), and sodium valproate (150 μM) were dissolved directly into the ACSF. However, for the preparation of ACSF containing sodium propionate or ammonium chloride, 25 or 10 mM of sodium chloride was substituted with each of these drugs, respectively.

Western blotting

Control and tetanized slices (15 slices each) were handled/treated similarly in the recording chamber except that control slices were not tetanized. Slices were homogenized in chilled phosphate-buffered solution containing 0.32 M sucrose plus 14 μg/ml aprotinin, 1 μg/ml pepstatin, 2 μg/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, and phosphatase inhibitors (1 mM sodium orthovanadate and 50 mM sodium fluoride). Protein determination was done using Biorad protein assay kit. All fractions were stored at −80°C.

Samples were dissolved in SDS sample buffer to a concentration of 5 μg protein/μl and denatured for 3 min at 95°C prior to loading. The sample proteins were, along with rainbow molecular weight markers (Amersham, Pharmacia, Piscataway, NJ), separated by 10% SDS polyacrylamide gel electrophoresis (BioRad, Missisauga, Ontario, Canada) and transferred from gel to nitrocellulose membrane (Schleicher and Schuell, Keene, NH). The gel was then stained with Gelcode blue stain reagent (Pierce, Rockford, IL) to control for protein transfer. The membrane was blocked with 5% fat-free milk at room temperature for 60 min, rinsed briefly with TBST buffer (10 mM Tris, 150 mM sodium chloride, and 0.05% Tween 20), and then incubated overnight with affinity-purified antibodies outlined below. After washing with TBST buffer four times for 15 min each, the membrane was incubated with horseradish peroxidase-conjugated goat anti-rabbit IgG (H + L) (Promega, Madison, WI) for 1 h and reacted with ECL Western blotting detection reagents (Amersham, Pharmacia) for 2 min. A high-performance autoradiography film (Amersham) was used for 10 s to 1 min and then developed to visualize the antibody binding. Quantity One software (BioRad) was used for the quantitative analyses of protein bands on the film. The integrated intensity of the band (OD value) was determined using both band density and band area.

Affinity-purified rabbit polyclonal IgG antibodies against Cx26 (diluted 1:300) and Cx32 (diluted 1: 400) proteins were obtained from Chemicon (Temecula, CA). An anti-Cx36 (diluted 1:400) as well a rabbit polyclonal anti-Cx43 antibody designated 71–0700 (diluted 1:600) and one monoclonal anti-Cx43 antibody, 13–8300 (diluted 1:600) were obtained from Zymed Laboratories (San Francisco, CA). Previously demonstrated in brain tissue, cardiac tissue, and cultured cells (Li et al. 1998; Nagy et al. 1997), antibody 13–8300 fails to recognize the slower migrating, phosphorylated 43,000 molcular weight forms of Cx43 but reacts with a faster migrating 41,000 molcular weight form, corresponding to dephosphorylated Cx43 (Crow et al. 1990; Laird et al. 1991; Musil et al. 1990; Saez et al. 1997). All primary antibodies were affinity purified.

Western blotting of Cx43 was conducted to determine the relative levels of phosphorylated and nonphosphorylated forms of Cx43. Phosphorylation status is typically deduced from the progressively slower mobility of Cx43 in gels. However, resolution of bands was poor due to the necessary inclusion of phosphatase inhibitors, which explicitly deteriorates band separation (Li and Nagy 2000;Mikalsen et al. 1997, Nagy et al. 1997).


In our assessment of the drug effect, we measured PAD duration in tetanized slices before and after drug application. Slices producing PADs shorter than 8 s were not included in our analysis. For statistical analyses and graphic presentations, drug effects were expressed as percentages of control and presented as means ± SE. For Western blotting, results are shown as means. Paired or unpairedt-tests were applied as required. Drug effects were considered significant at P < 0.05.


Extracellular recordings

Primary afterdischarges were reliably produced in all slices examined. Slices showing widespread depressions following tetanization (25%) were excluded from our analysis. The duration of PADs increased with each subsequent tetanization ranging from 8 to 70 s following 6–10 tetanic stimuli (Fig. 1,A and B).

Fig. 1.

Primary afterdischarges recorded from the CA1 pyramidal neuronal layer in a hippocampal-parahippocampal slice, following tetanic stimulation of Schaffer's collaterals before and after perfusion with 1-octanol.A: Primary afterdischarge (PAD) after the first tetanic stimulation for 2 s at 100 Hz (solid bar). B: PAD recorded after the sixth tetanization, showing that the PAD duration progressively increases with subsequent tetanizations. PADs recorded following 30 min of 1-octanol perfusion (C), and after 68 min washout (D) with drug-free ACSF. Note that 1-octanol clearly blocks PADs and that this drug effect is fully recovered with washout. I/O field responses before (E) and after the sixth tetanization (F), following 30 min perfusion with octanol (G) and 60 min washout of drug effect (H). Note that the field responses are moderately depressed after 30 min perfusion with octanol (G), but they recovered 60 min after drug washout (H).

Perfusion of four slices from different animals with 1-octanol (200 μM) blocked the duration of PADs to 0 within 15–30 min (Fig.1 C). This effect was reversible and the PADs recovered by 1 h after return to the drug-free (control) ACSF (Fig.1 D). The amplitude of the I/O responses recorded following 30 min of perfusion with 1-octanol (Fig. 1 G) were slightly reduced (15%), but fully recovered 60 min after drug wash out (Fig.1 H).

Sodium propionate (25 mM), which is known to cause cytoplasmic acidification (Rorig et al. 1996), also reduced the duration of PADs to 59 ± 9% (n = 7, mean ± SE) of control (Fig. 2,A--D). The duration of PADs recovered within10 min after wash out of drug (Fig. 2 D). Sodium propionate did not reduce the amplitude of the I/O responses (Fig. 2,E–H).

Fig. 2.

Propionic acid reduced the duration of primary afterdischarges (PADs) in hippocampal-parahippocampal slices but did not diminish field responses to single stimuli. A: PAD recorded after the first tetanization of 2 s duration at 100 Hz (solid bar). B: PAD recorded after the 7th tetanization. Note the duration of PADs increased with successive tetanizations. C: propionic acid significantly reduced the duration of the PAD. This effect reversed following return to control ACSF (D).E-H illustrate I/O relations before (E) and after the seventh (F) tetanization, 30 min after drug perfusion (G) and 30 min after drug washout (H). Propionic acid suppressed PADs without diminishing I-O field responses to single stimuli. Arrows point to the epileptiform discharges seen after the seventh tetanization and following drug washout with ACSF.

The effect of ammonium chloride (NH4Cl) on the activity of tetanized slices was time dependent. Ammonium chloride is known to cause cytoplasmic alkalinization followed by acidification on wash out (Xiong et al., 2000). Shortly (2–3 min) after the application of NH4Cl (n = 6), spontaneous field burst activities (secondary afterdischarge) were observed (Fig. 3 B). Eight to 10 min later, PADs disappeared (Fig. 3 C), and the slices either did not respond to single short-duration (0.1 ms) stimulating pulses (n = 4/6) or the amplitude of the population spike was reduced (n = 2/6). These drug effects reversed 30–40 min after return to drug-free ACSF; i.e., the PADs (Fig. 3 D) and I/O responses to single stimuli appeared (Fig.3 G) following drug wash out.

Fig. 3.

Application of ammonium chloride (10 mM) to tetanized slices initially evokes spontaneous secondary afterdischarges. A: control PAD recorded after the fifth tetanization. Solid bar indicates tetanization for 2 s at 100 Hz. B: spontaneous secondary afterdischarge recorded shortly (3 min) after perfusion of the slice with NH4Cl. C: complete blockade of the PAD following return to control ACSF. D: recovery of the PAD after 37 min of drug washout with control ACSF. I/O field responses to single stimuli before (E) and after the fifth tetanization (F) and during the late washout phase (G). Arrows point to the epileptiform discharges which are seen after the fifth tetanization but are absent during the acidification phase.

Carbenoxolone (200 μM), a known gap junctional blocker, effectively reduced the duration of PADs in tetanized slices. Ten minutes after the application of this drug, the PAD duration dropped to 51.8 ± 14.8% of control in five slices, did not change in one, and increased in another slice. However, following 20 to 40 min of drug perfusion, the PAD duration was reduced in all slices (n = 7) to 11.5 ± 8.7% of control (Fig. 4,A–C). This effect was prolonged and could not be reversed after return to the control ACSF for ≤60 min. I/O field responses to single stimuli were not reduced by this drug (Fig. 4, D–F).

Fig. 4.

Carbenoxolone reduced the duration of primary afterdischarges recorded from CA1 neurons in a tetanized hippocampal-parahippocampal slice.A: PAD after the third tetanization. Solid bar indicates 2 s of tetanization at 100 Hz. B: PAD recorded following the fifth tetanization. C: PAD recorded 32 min after the perfusion of the slice with carbenoxolone. Note that the duration of the PAD is markedly reduced. This blocking effect of carbenoxolone could not be reversed with drug washout for ≤60 min. I/O field responses to single stimuli (D–F) were not suppressed by carbenoxolone. Arrows point to the multiple population spikes indicating epileptogenic activities after the 4thtetanization.

Sodium valproate (150 μM), a commonly used anticonvulsant, also reduced the duration of PADs to 57.7 ± 6.6 (n = 5) percent of control 1 h after perfusion. However, as we have shown (Jahromi et al., 2000), phenytoin, another known anticonvulsant did not significantly diminish PAD durations in this model. For comparative purposes, the effects of gap junction blockers and sodium valproate have been graphically presented in Fig.5.

Fig. 5.

PAD durations following application of 1-octanol, sodium propionate, ammonium chloride (NH4 Cl), sodium valproate and carbenoxolone (carb) are plotted as % of control. The thicker lines, representing actions of 1-octanol and ammonium chloride, denote 0 PAD duration following the application of these two drugs.

Whole-cell recording: carbenoxolone

To further investigate the relationship between cellular excitability and blocking gap junctional communication, we have recorded intracellularly measured electrophysiological properties before and after the addition of a commonly used gap junctional blocker, carbenoxolone. Following 6–10 episodes of tetanic stimuli, at which time robust PAD was established, whole-cell recordings from CA1 neurons were performed and changes in cellular intrinsic properties were recorded, before and during the application of carbenoxolone. The RMP of tetanized CA1 pyramidal neurons ranged from −57 to −60 mV. However, as previously mentioned, the RMP was maintained at −60 mV by the injection of a hyperpolarizing constant current through the recording electrode. The membrane input resistance (R i) significantly increased 30–40 min after the perfusion with this drug, from 45.9 ± 5.0 (n = 10) to 77.4 ± 5.9 MΩ, (n = 8), which is equivalent to 164 ± 11.9% (n = 8) of the initial value (Fig.6, A andB). The spike voltage threshold was measured from the first spike evoked after the injection of stimulating current pulses of varying amplitudes and duration of 20–40 ms. The voltage threshold decreased to 70.5 ± 4.2% (n = 12) of control, i.e., from 29.1 ± 3.4 mV depolarization to 21.0 ± 3.0 mV, 20–30 min after drug application (Fig. 6, C andD). Stimulating current pulses (350 ms), ranging from 0.1 to 0.55 nA, were injected through the recording electrode and the frequency (Hz) of evoked action potentials was measured. Spike frequency adaptation was reduced (n = 10) and the average spike frequency was increased 20–40 min following the application of carbenoxolone (Fig. 6, E and F). The effect of carbenoxolone on the long-lasting AHP was variable. It increased the AHP in five neurons (224.7 ± 56.3% of control) but reduced it in four neurons (50.8 ± 14.8% of control). We conclude that carbenoxolone essentially increased neuronal excitability while at the same time uncoupling and desynchronizing the neuronal network responsible for the seizure-like activity.

Fig. 6.

A: voltage responses of a CA1 neuron to a family of hyperpolarizing and depolarizing current pulses, recorded before (control) and 30 min after the application of carbenoxolone. The membrane potential was kept at −60 mV. B: current/voltage relationship is plotted for CA1 neurons (n = 8) before (control) and 30–40 min following drug perfusion (carbenoxolone). As seen, carbenoxolone significantly increased the membrane input resistance in these neurons.C: responses of a CA1 neuron to current pulses of 20–30 ms duration and −0.1 to +0.4 nA before (C) and 30–40 min after (D) the application of carbenoxolone. Note that the voltage threshold required for evoking action potentials is significantly reduced. E: trains of action potentials recorded from a CA1 neuron stimulted with a 350 ms duration, 0.5 nA current pulse before (control) and 30 min after carbenoxolone. Note that the frequency of action potential firing has increased after perfusion with carbenoxolone. Action potentials are truncated (because of the sampling rate). F: average spike frequency (Hz) plotted (n = 10) for a family of stimulating currents (0.1–0.55 nA). Note that spike-frequency adaptation was significantly decreased by carbenoxolone.

Connexin protein expression

Measurement of protein expression by Western blotting for connexin (Cx) Cx43, Cx36, Cx32, and Cx26 showed no differences in the level of expression of these connexins between control and tetanized slices. However, the level of nonphosphorylated Cx43, as detected by antibody 13–8300, was significantly reduced as a result of tetanization (Fig.7). Since there is no change in the overall expression of Cx43, this suggests a redistribution from the dephosphorylated to the phosphorylated form of Cx43.

Fig. 7.

Western blots showing connexin protein expression in the rat hippocampal slices before and after repetitive tetanization.A: there were no significant differences in Cx26, Cx32 and Cx36 protein expression between control (lane 1) and tetanized slices (lane 2). The blot probed for Cx 43t (antibody 71–0700) shows detection of both phosphorylated and nonphosphorylated Cx 43 in control and tetanized slices, whereas the blot probed for Cx 43 days (antibody 13–8300) shows detection of only the nonphosphorylated form of Cx43. Repetitive tetanization induces a decrease in the nonphosphorylated Cx43. B: histogram plot of the relative changes in Cx protein expression according to optical density (OD). White bars and solid bars represent control and tetanized slices, respectively. *Indicates statistical significance (P < 0.05) between control and the tetanized slices.


Our results clearly indicate that different gap junctional blockers significantly reduce the duration of epileptiform activities recorded as primary afterdischarges following repetitive tetanic stimulation. Despite the importance of gap junctional intercellular communication, relatively few agents are known to block electrotonic coupling in a specific manner. The gap junction channel is well insulated from the extracellular space that it spans, and there is restricted access to allow direct channel modulation. Higher alcohols such as heptanol and octanol, as well as the anesthetic, halothane, block junctional conductance in a wide variety of cell types and tissues. These compounds, which act rapidly in a few minutes, are suggested to limit gap junctional communication by dissolving in the membrane lipids, inducing changes in membrane fluidity (Johnston et al. 1980; Rose and Ransom 1997;Cotrina et al. 1998), thus interfering with the proper functioning of gap junction channels. Glycyrrhetinic acid derivatives have been shown to reduce intercellular coupling via gap junctions (Takens-Kwak et al. 1992; Guan et al. 1996; Tordjmann et al. 1997; Taylor et al. 1998; Chaytor et al. 1999;Mambetisaeva et al. 1999; Boitano and Evans, 2000). 18α-glycyrrhetinic acid, 18β-glycyrrhetinic acid and carbenoxolone mediate concentration dependant inhibition of junctional conductance on a slower time scale of tens of minutes. They are thought to act indirectly, via activation of protein kinases, G proteins, transport ATPases and gap junction phosphorylation states. Why octanol and NH4 Cl completely blocked the PAD and the other treatments did not, is currently unclear.

In a recent review of anticonvulsant mechanisms, gap junctional blockade is not yet mentioned as a target for seizure therapy (Bazil and Pedley 1998). We show here that the anticonvulsant, sodium valproate, reduced tetanus-induced PAD at a clinically relevant concentration. We have previously shown (Jahromi et al., 2000) that topiramate blocked epileptiform discharges in this model at 100 μM, probably too high a concentration to be clinically relevant, and that phenytoin, another known anticonvulsant, had no effect. These data suggest that this model could be relevant to the problem of pharmacologically “intractable” epilepsy. In this model, all gap junctional blockers significantly suppressed epileptiform discharges. Whether gap junctional blockade is part of the mechanism of action of valproate is not known. However our data suggest that gap junctional blockade may be an attractive target for novel anticonvulsant development, particularly in the context of treating intractable epilepsy.

Gap junctional blockers or acidosis depressed spontaneous epileptiform discharges in hippocampal slices exposed to a zero calcium perfusate (Perez Velazquez et al. 1994; Xiong et al., 2000; Schweitzer et al., 2000). Xiong et al., (2000) also showed that the initiation of the field burst resulted in intracellular acidification, which terminated the bursting activity after reaching a certain level. Also synchronized activities, recorded along the dorsal dentate gyrus, were blocked by focal application of an acidic medium. Neuronal resting membrane potential and action potentials in zero calcium in CA1 neurons (Perez Velazquez et al. 1994) and in dentate granule neurons in normal ACSF (Schweitzer et al., 2000) did not significantly change following changes in the pHosufficient to block spontaneous field burst activities. Furthermore, gap junction blockers octanol, oleamide and low pH, suppressed the field burst activities without affecting epileptiform discharges recorded from single units (Schweitzer et al., 2000). Likewise, in our experiments, carbenoxolone increased neuronal excitability whereas it significantly reduced the duration of field burst activities in CA1 pyramidal neurons. These data strongly suggest that field bursts depend on the synchronization of individual responses from a large number of neurons via gap junctional connections.

Ross et al. (2000) demonstrated that the high-frequency of spontaneous bursts of population spikes, generated by the hippocampal CA1 pyramidal neurons in a 0 Mg2+ and 50 μM 4-aminopyridine medium, were reduced by carbenoxolone (100 μM). This action of carbenoxolone, which is also known to act as a mineralocorticoid agonist, was not blocked by the mineralocorticoid antagonist, spironolactone (1 μM). Activation of mineralocorticoid receptors, which are found in CA1 pyramidal neurons, results in increased excitability and reduced spike frequency adaptation in these neurons (Joels and de Kloet 1990), as seen by us as well with carbenoxolone. Therefore the blocking effect of carbenoxolone on synchronized epileptiform discharges appears to be unrelated to its action on mineralocorticoid receptors at the single cell level. In our experiments, as in those of Ross et al. (2000), the blocking action of carbenoxolone on epileptiform discharges was slowly or not at all reversed with washout (40–70 min).

Many different connexins have been localized in the CNS. More specifically, Cx43 and Cx30 have been found to exist in astrocytic membranes, and Cx32 is present in oligodendrocytes (Rash et al. 1998, 2001). Currently, Cx36 is the first connexin to be localized only in neurons (Rash et al., 2001). In our experiments, Western blotting did not show any significant differences in the level of protein expression for Cx26, Cx32, Cx36 and Cx43 in the control and tetanized slices. Only the expression level of dephosphorylated Cx43, not the total Cx43 protein expression, appeared reduced. How this relates to the development of seizure-like activity is unclear. Currently, although the majority of connexins, excluding Cx26, have been shown to be phosphoproteins, there are no commercially available antibodies to assess the phosphorylation state of Cx32 and Cx36. The phosphorylation of Cx43 is a controversial topic and appears to influence gap junctional communication in both a positive and negative manner (Crow et al. 1990; Musil et al. 1990; Brissette et al. 1991; Kadle et al. 1991; Laird et al. 1991; Berthoud et al. 1992). For instance, activation of PKC in different cell types has been shown to have variable effects on channel conductance. Even in the same cell type, neonatal rat cardiomyocytes, all three responses to TPA (12-O-tetradecanoylphorbol-13-acetate) have been reported: a decrease in junctional conductance (Munster et al. 1993), an increase in junctional conductance (Kwak et al. 1995), and little change in junctional conductance (Spray et al. 1990). This suggests that in addition to regulating the channels' conductance state (and consequently permeability), phosphorylation can lead to changes in other channel parameters.

In spinal motor neurons, axotomy enhances gap junction coupling without significant changes in the expression of Cx36, Cx37, Cx40, Cx43 and Cx45, connexins which are normally seen in adult motor neurons (Chang et al., 2000). These authors have suggested that the existing and constitutively expressed connexins in motor neurons may be responsible for re-establishing gap junctional connections in these injured neurons. It is also becoming increasingly evident that Cx36 plays a major role in synchronizing interneurons (Deans et al., 2001), which can play a major role in seizure generation in the CA1 region following tetanization (Perez Velazquez and Carlen, 2000)

In hippocampal tissues from patients showing complex partial seizures in the medial temporal area and particularly in the hippocampus, analyses of Cx43 mRNA and Cx43 protein, with Northern and Western blotting respectively, showed no up-regulation or increases in the expression level of this connexin (Elisevich et al. 1997). Li and Nagy (2000) found an increase in Cx43 dephosphorylation in the spinal cord dorsal horn after electrical stimulation of sciatic nerve A-fibers or after cutaneous stimulation of C-fibers with capsaicin. Recently it has been proposed that up-regulation of astrocytic Cx43 may exacerbate seizures in human mesial temporal lobe epilepsy (Fonseca et al., 2002). We noted a decrease in the nonphosphorylation state of Cx43 following tetanus-induced seizure activity. Whether this change is directly relevant to the seizure activity measured in our model is unclear.

We showed that evoked epileptiform activities were blocked by gap junctional blockers, without depressing the excitability of single neurons in the case of carbenoxolone. In the light of our experimental results and published evidence, we conclude that gap junctional connections are fundamental to the synchronization of neuronal activities as seen in in vitro epileptogenesis.


We thank F. Vidic for technical support.

This work was supported by Canadian Institutes of Health Research.


  • Address for reprint requests: S. S. Jahromi, Toronto Western Research Institute, Division of Cellular and Molecular Biology, MacL. Pavilion, 12th Floor, Rm. 413, Toronto Western Hospital, University of Toronto, 399 Bathurst Street, Toronto, Ontario M5T 2S8 Canada (E-mail address: sjahromi{at}


View Abstract