Excitatory effects of gap junction blockers on cerebral cortex seizure-like activity in rats and mice


Address correspondence to Logan J. Voss, Department of Anesthesiology, Waikato Clinical School, Waikato Hospital, Pembroke St, Hamilton, New Zealand. E-mail: vossl@waikatodhb.govt.nz


Purpose: The role of gap junctions in seizures is an area of intense research. Many groups have reported anticonvulsant effects of gap junction blockade, strengthening the case for a role for gap junctions in ictogenesis. The cerebral cortex is underrepresented in this body of research. We have investigated the effect of gap junction blockade on seizure-like activity in rat and mouse cerebral cortex slices.

Methods: Seizure-like activity was induced by perfusing with low-magnesium artificial cerebrospinal fluid. The effect of three gap junction blockers was investigated in rat cortical slices; quinine (200 and 400 μm), quinidine (100 and 200 μm), and carbenoxolone (100 and 200 μm). In addition, the effect of mefloquine was investigated in wild-type mice and connexin36 knockout mice. The data were analyzed for the effect on frequency and amplitude of seizure-like events.

Results: Paradoxical excitatory effects on seizure-like activity were observed for all three agents in rat cortical slices. Quinine (200 μm) and carbenoxolone (100 μm) increased both the frequency and amplitude of seizure-like events. Quinidine (100 μm) increased the frequency of events. Higher doses of quinine (400 μm) and carbenoxolone (200 μm) had biphasic excitatory–inhibitory effects. Similar excitatory effects were observed in adult wild-type mouse cortical slices perfused with mefloquine (5 μm or 10 μm), but were absent in slices from connexin36-deficient mice.

Discussion: In conclusion, we have shown a paradoxical proseizure effect of pharmacologic gap junction blockade in a cortical model of seizure-like activity. We suggest that this effect is probably due to a disruption of inhibitory interneuron coupling secondary to connexin36 blockade.

There has been increased interest in recent years in the role of gap junctions in seizure mechanisms. An important hallmark of seizure activity in the brain is believed to be synchronous neuronal population activity. Direct electrical coupling between neurons in the form of gap junctions is thought to contribute to ictogenesis by facilitating the rapid spread of electrical activity between neurons and by enhancing synchronous activity (Carlen et al., 2000; Perez Velazquez & Carlen, 2000). Traub et al. (2001) have suggested that gap junctions may also be important for seizure initiation.

Many groups have reported anticonvulsant effects of gap junction blockers, strengthening the case for a central role for gap junctions in seizure mechanisms (Carlen et al., 2000; Jahromi et al., 2002; Medina-Ceja et al., 2008). The majority of these studies have focused directly on the hippocampus, or are in vivo studies where the hippocampus may be implicated in cortically recorded phenomena (Gajda et al., 2005); Amabeoku & Chikuni, 1992; Amabeoku & Farmer, 2005). Less is known about the relationship between seizures and pharmacologic manipulation of gap junctions in the isolated cerebral cortex. One recent study has shown that carbenoxolone, a broad-spectrum gap junction blocker, causes paradoxical excitatory effects at the level of individual neurons within the cerebral cortex (Yang & Ling, 2007). An excitatory effect of carbenoxolone has also been observed at the cellular level in the hippocampus (Jahromi et al., 2002). These results seem to contradict previous findings showing inhibitory effects of gap junction blockade on population neural activity (Carlen et al., 2000; Jahromi et al., 2002; Medina-Ceja et al., 2008).

The aim of this study was to investigate the effect of gap junction blockade on seizure-like activity originating from the isolated cerebral cortex. We were particularly interested in whether the excitatory effects of gap junction blockade, previously observed by Jahromi et al. (2002) at the single cell level, could be observed as proseizure effects at the population level. We investigated in adolescent rats the effect of three gap junction blockers—carbenoxolone, quinine, and quinidine—on seizure-like activity induced by perfusion of cerebral cortical slices with low magnesium artificial cerebrospinal fluid (ACSF). A consistent finding was that gap junction blockade had the propensity to enhance low-magnesium seizure-like activity. In a second series of experiments we compared the effect mefloquine on seizure-like activity in wild-type mice and knockout mice lacking the gene for connexin36.


Slice preparation

Neocortical slices were prepared from 5- to 8-week-old Sprague-Dawley rats of either sex, 4-month-old c57/bl6 wild-type mice, and c57/bl6/129sv connexin36 knockout mice (gifted by Prof David Paul, Harvard University). The polymerase chain reaction (PCR) strategy detailed in Deans et al. (2001) was used to confirm the genotype of connexin36 knockout mice. The animals were decapitated while anesthetized with carbon dioxide, in accordance with local animal ethics guidelines. The brain was rapidly removed and cooled in ice-cold ACSF, modified for cerebral protection according to Nowak and Bullier (1996), with composition (in mm): NaCl 92.7; NaHCO3 24; NaH2PO4 1.2; KCl 3; MgCl2 19; CaCl2 0; D-glucose 25; bubbled with carbogen (95% O2; 5% CO2). Coronal slices (400 μm) were cut between bregma −3 mm to −4 mm on a Vibratome (Campden Instruments, Loughborough, United Kingdom) in ice-cold ACSF as noted previously and transferred to a holding chamber containing carbogenated, low-magnesium ACSF of composition: NaCl 124; NaHCO3 26; NaH2PO4 1.25; KCl 5; MgCl2 0; CaCl2 2; D-glucose 10 (pH 7.6). The holding chamber was maintained at room temperature (18–20°C), where the slices were held for at least 1 h before being transferring to the recording chamber. Submerged slices were perfused with carbogenated ACSF (room temperature) at a flow rate of 2 ml/min.

Electrical recording

Spontaneous local field potential activity was recorded using a single 50 μm Teflon-coated tungsten wire positioned in the middle to outer layer of the somatosensory cortex. A silver/silver chloride disc electrode served as a common reference/bath ground. The signal was amplified (1,000×; A-M Systems, Carlsborg, WA, U.S.A.) and bandpass filtered (1 and 3,000 Hz) before analog–digital conversion (Power 1401; CED, United Kingdom) and recording on computer for later analysis (Spike2; CED). The entire recording set-up was enclosed within a grounded Faraday cage to reduce electrical noise.

Rat experimental protocols

Following baseline recording in low-magnesium ACSF for at least 10 min and establishment of a stable pattern of seizure-like activity, one or other of the following experimental sequences were followed.

  • 1Quinine (200 μm, n = 10, from three animals), quinidine [100 μm and 200 μm, n = 10 (two animals) and n = 6 (one animal), respectively], or carbenoxolone (100 μm, n = 14, three animals) was infused in low-magnesium ACSF for 20 min, followed by a further 40 min of perfusion with drug-free low-magnesium ACSF.
  • 2The experiments in 1 demonstrated that all three agents had excitatory effects on seizure-like activity. To examine the dose-dependence of these effects, in a separate set of experiments carbenoxolone and quinine were each perfused individually at an elevated dose [200 μm, n = 8 (two animals) and 400 μm, n = 9 (three animals), respectively], followed by drug-free low-magnesium ACSF for 60 min.

Mouse experimental protocols

To investigate whether the excitatory effects observed with pharmacologic gap junction blockade could be attributed to connexin36 blockade, mice lacking the gene for connexin36 were compared with wild-type mice. Following baseline recording in low-magnesium ACSF for at least 10 min and establishment of a stable pattern of seizure-like activity, the following experimental sequence was followed. In wild-type (n = 8, three animals) and connexin36 knockout (n = 11, three animals) mice, mefloquine (5 or 10 μm) was perfused in low-magnesium ACSF for 20 min followed by a return to drug-free low-magnesium ACSF for 20 min.

Control experiments were carried out by perfusing only low-magnesium ACSF (n = 9, two animals) and low-magnesium ACSF containing the maximum concentration of dimethyl sulphoxide (DMSO) (0.2%, n = 10, three animals). The same experimental sequence as described earlier was followed.

Drug preparation

Carbenoxolone (Sigma, St. Louis, MO, U.S.A.) solution was prepared by adding the appropriate amount of the drug directly to pre-prepared, fresh low-magnesium ACSF. Quinine, quinidine, and mefloquine (Sigma) solutions were prepared similarly, but with the addition of 1–2 ml DMSO per liter (0.1–0.2%) ACSF to help dissolve the drug. All solutions were replaced after no more than 1 week of storage at 1–4°C.

Data analysis

The excitatory effects observed during infusion of the structurally similar drugs quinidine and mefloquine were not dependent upon drug dose; therefore, the data at both doses for these drugs were pooled for analysis (see protocol 1 above). For the statistical comparison of excitatory drug effects, the seizure-like event frequency/amplitude was averaged over the 5 min immediately prior to drug delivery and compared to the average value during the 20 min of drug infusion. The frequency of seizure-like events was calculated as a moving average with a window of 2 min and 50% overlap. Control experiments were analyzed in identical fashion. Comparisons were analyzed using the paired t-test where the data were shown to be normally distributed (Kolmogarov-Smirnov test). Otherwise, paired comparisons were analyzed using the Wilcoxon test. For statistical comparison of biphasic responses, sequential effects were analyzed using repeated measures ANOVA, with Tukey post hoc multiple comparisons tests. Unless otherwise stated, the data are expressed as mean (SD) and p < 0.05 was considered statistically significant.


Seizure-like activity induced by low-magnesium ACSF

Perfusion of cortical slices with low-magnesium ACSF induced a consistent pattern of seizure-like activity (Fig. 1) in all slices, including those from connexin36-deficient animals. There was no statistically significant difference in either the amplitude [mean (SD) 73(99) compared to 72(66) μV] or frequency [2.4(3.0) compared to 2.1(2.3) events/min] of seizure-like events between wild-type and connexin36 knockout mice, respectively. The seizure-like events typically consisted of a large deflection in the local field potential recording, followed immediately by a 4–7 Hz oscillation of varying length, but no longer than 3 s. The amplitude of the initial deflection was usually larger than the oscillation that followed, and ranged from 50–300 μV. Seizure-like events occurred with variable frequency, from 0.5 to 5 events per minute.

Figure 1.

 Profile of seizure-like activity induced in a single cortical slice from a wild-type rat during perfusion with low-magnesium artificial cerebrospinal fluid (ACSF). Each of the high-amplitude lines is a single seizure-like event, one of which is expanded below.

Excitatory effects of gap junction blockers on low-magnesium seizure-like activity in rat slices

In rat cortical slices, all three gap junction blockers induced an increase in the frequency of low-magnesium seizure-like events. In addition, carbenoxolone and quinine significantly increased event amplitude. These results are summarized in Table 1 and examples of the time course of effects are shown in Fig. 2. The drug effect on seizure-like event duration during the period of enhanced activity was highly variable. For carbenoxolone, event duration remained unchanged in six cases, increased in six, and decreased in two. Similar results for event duration were obtained for the other agents tested. Because there was clearly no consistent drug effect, seizure-like event duration was not quantified further.

Table 1.   Seizure-like event frequency and amplitude statistics for the three gap junction blockers
 Baseline (low-magnesium)After drug infusionp
  1. Data are mean [standard deviation (SD)].

Seizure-like event frequency (events/min)
Carbenoxolone 100 μm (n = 14)3.1 (2.2)4.0 (2.1)<0.005
Quinine 200 μm (n = 10)1.8 (0.8)2.1 (0.9)<0.05
Quinidine 100/200 μm (n = 16)2.8 (1.1)3.1 (1.1)<0.02
Seizure-like event amplitude (μV)
Carbenoxolone 100 μm (n = 14)162.0 (98.9)178.1 (111.8)<0.05
Quinine 200 μm (n = 10)145.05 (90.1)157.8 (92.4)<0.05
Quinidine 100/200 μm (n = 16)167.2 (85.6)179.6 (87.8)ns
Figure 2.

 Time course of seizure-like events from selected individual slices from wild-type rats for (A) carbenoxolone (cbx) and (B) quinine. Each vertical line in the graphs represents a single seizure-like event. The height of each line is the maximum amplitude of each event.

When carbenoxolone and quinine were infused individually at higher doses (200 and 400 μm, respectively), similar excitatory effects were observed, followed by a dramatic reduction in seizure-like activity (Fig. 3). The effect was somewhat more convincing for carbenoxolone, as illustrated in Fig. 3A. These inhibitory effects were not observed with the lower doses of carbenoxolone and quinine (100 and 200 uM, respectively) (see Fig. 2).

Figure 3.

Effect of (A) carbenoxolone (200 μm) and (B) quinine (400 μm) on low-magnesium seizure-like event amplitude and frequency in wild-type rats. *p<0.05; **p<0.001 compared to baseline values; repeated measures ANOVA, Tukey post hoc test.

Excitatory effects of the gap junction blocker mefloquine on low-magnesium seizure-like activity in mouse slices

To further document the role of connexin36 gap junction blockade in the proseizure effects observed in the rat cortical slices, the effect of pharmacologic intervention was investigated in wild-type and connexin36 knockout mice. Consistent with the rat slice data, mefloquine (5 or 10 μm) induced an increase in seizure-like event frequency [2.4(1.0) to 3.0(0.9) events per min, p < 0.0005 paired t-test] and amplitude [73(60) to 99(103)μV, p < 0.05, Wilcoxon test] in wild-type mice (see Fig. 4). In contrast, no significant mefloquine effect on event frequency was observed in connexin36 knockout animals, whereas the amplitude of events decreased [72(38) to 66(39) μV, p < 0.05, paired t-test] slightly during mefloquine infusion (see Fig. 4).

Figure 4.

 Effect of mefloquine (5 and 10 μm) on low-magnesium seizure-like event amplitude and frequency in (A) wild-type and (B) connexin36 knockout mice. The graphs plot the relative change in each parameter. *p<0.05; **p<0.0005 compared to pre drug test.

Overall, neither low-magnesium on its own nor DMSO (0.2%) in low-magnesium had a statistically significant effect on either seizure-like event frequency or amplitude. However, it was noted that in 4 of 10 slices (from a single animal), 0.2% DMSO appeared to have a marked and reversible antiictogenic effect, seen as a mean (SD) reduction in event frequency of 43(28)% in these four slices. A small reduction in event amplitude was also noted in each of these slices [mean (SD) reduction of 10(8)%].


In previous reports, pharmacologic blockade of gap junctions has been consistently associated with anticonvulsant effects, in both the cortex and the hippocampus (Carlen et al., 2000; Jahromi et al., 2002; Gajda et al., 2005; Nilsen et al., 2006; Bostanci & Bagirici, 2007; Medina-Ceja et al., 2008). Our study has shown that in the cerebral cortex, the anticonvulsant effects of quinine, quinidine, and carbenoxolone are preceded by excitatory phenomena, reflected in an increase in the frequency and amplitude of seizure-like events. To our knowledge, this is the first report directly linking gap junction blockade with excitatory effects on population activity within the cerebral cortex.

Gajda and colleagues (2005) anecdotally reported “pro-convulsive” effects of “high” dose (>80 μm) quinine in penicillin-induced neocortical seizures in vivo; however, the data relating to this effect were not presented. They attributed these excitatory effects to unspecified nonspecific (i.e., extra-gap junctional) effects of the drug. Mefloquine and quinine have also been reported to have proconvulsant effects in vivo when administered systemically at high dose (approximately 250 and 600 μm, respectively) (Amabeoku & Chikuni, 1992; Amabeoku & Farmer, 2005). Interestingly these effects seem to be related to a disturbance in GABAergic mechanisms (Amabeoku & Chikuni, 1992; Amabeoku & Farmer, 2005).

We suggest that the proconvulsant activity of quinine and its analogs is due to blockade of connexin36 gap junctions, based on the following evidence. First and most important, the excitatory effects of mefloquine observed in wild-type mice were not observed in mice lacking the gene for connexin36. Mefloquine potently and specifically (between 1 and 5 μm) blocks connexin36 gap junctions (Cruikshank et al., 2004), strongly suggesting that the excitatory effects observed in the wild-type animals were a specific connexin36 effect. Second, if nonspecific actions were responsible for the excitatory effects observed in this study, it is unlikely that completely different classes of drug (carbenoxolone and quinine/quinidine) would show similar effects. Third, connexin36 gap junctions are maximally blocked by quinine at a dose of approximately 100 μm (Srinivas et al., 2001) and twice this dose has no direct effects on neuronal excitability (Bikson et al., 2002). The dose of quinine used in the present study was 200 μm. Therefore, the proconvulsive effects of quinine were observed at a dose associated with strong gap junction blockade, but that has no direct effects on neuronal excitability.

The results of this study are consistent with recent findings showing that carbenoxolone (a broad-spectrum gap junction blocker (Gajda et al., 2005; Nilsen et al., 2006) causes paradoxical excitatory effects at the level of individual neurons within the cerebral cortex (Yang & Ling, 2007). In this study by Yang and Ling, an increase in excitatory postsynaptic potential amplitude was shown to be indirectly caused by uncoupling of (GABAergic) inhibitory interneurons. Connexin36 gap junctions are restricted primarily to a subclass of interneurons, parvalbumin-containing basket cells (Deans et al., 2001; Liu & Jones, 2003; Markram et al., 2004; Baude et al., 2007), which synapse onto pyramidal cells in the region of the soma or proximal dendrites (Deans et al., 2001; Liu & Jones, 2003; Markram et al., 2004). Blocking direct electrical communication within this network of cells by closing gap junctions is likely to have a disinhibitory effect on pyramidal cell activity, hence the excitatory effects observed at both the cellular level (Yang & Ling, 2007) and at the population level in the present study.

In this study we did not observe any difference in baseline seizure activity between wild-type and connexin36 knockout animals. Given the excitatory effects observed with pharmacologic blockade of connexin36 gap junctions, we might have expected the connexin36 knockout animals to show enhanced baseline seizure activity. The most likely explanation is that connexin36 knockouts develop compensatory responses to chronic gap junction blockade. Wide-reaching compensatory changes in neuronal gene regulation have been documented for connexin knockouts (Iacobas et al., 2005; Spray & Iacobas, 2007), although the functional correlates of these changes remains a matter for further investigation. One report has shown morphologic and functional neuronal effects at the subcortical level in connexin36 knockouts (De Zeeuw et al., 2003). Preliminary results from our laboratory have failed to identify any compensatory changes in the mRNA levels for a range of glial and neuronal connexins and pannexins (results not shown). An alternative, and less likely explanation in our view, is that the excitatory effects of mefloquine in wild-type animals are due to off-target effects (see subsequent text for further discussion). If this was the case, however, one would expect similar excitatory effects in both wild-type and knockout animals.

We cannot completely rule out pharmacologic effects on gap junctions other than connexin36 in this study. Connexin45, for example, is expressed in cortical neurons of adult mice (Maxeiner et al., 2003) and rats (Condorelli et al., 2003) and quinine partially blocks this gap junction at a dose of 300 μm (Srinivas et al., 2001). Effects on astrocytic gap junctions may also be relevant. Astrocytes are important regulators of synaptic excitability (Fellin & Haydon, 2005; Tian et al., 2005; Fellin et al., 2006) and changes in astrocytic signaling can alter neuronal population activity (Tian et al., 2005). Mefloquine at a dose of 10 μm causes a 40% blockade of connexin43 (Cruikshank et al., 2004), an astrocytic gap junction protein in the central nervous system. Furthermore, the effect of quinine and its derivatives on connexin30, another astrocytic gap junction protein (Rash et al., 2001), has not been documented. “Off-site” effects, however, do not provide a complete explanation for the observed results, for two reasons. First, the effect of mefloquine was abolished in the connexin36 knockout animals, strongly implying a specific connexin36 effect. Second, the doses of quinine (200 μm) and mefloquine (5–10 μm) used are such that effects on connexin45 and connexin43, respectively, are likely to have been minimal.

A major drawback of blocking gap junctions pharmacologically is the possibility of non–gap junctional drug effects. Mefloquine has been shown to disrupt calcium homeostasis in neurons, although at a concentration four times that used in the present study (Caridha et al., 2008). Carbenoxolone is known to have extra–gap junctional effects. For example, 100 μm carbenoxolone has direct inhibitory effects on neural network activity independent of its effects on gap junctions (Rouach et al., 2003; Chepkova et al., 2008). In addition, carbenoxolone has mineralocorticoid actions, which may reduce seizure propensity (Nilsen et al., 2006). Quinidine and its analogs are also known to have a multitude of actions, many unrelated to gap junction effects (Gattass & De Meis, 1978; Garlid et al., 1986; Pederson et al., 1986). The potential importance of these off-target effects in studies that have documented a reduction in seizure activity with these agents cannot be underestimated, and call into question the significance of gap junctions in ictogenesis. In the present study, a reduction in seizure-like activity was observed at the higher drug doses only, where nonspecific pharmacologic effects may predominate. At the very least, the present results strongly suggest that connexin36 gap junction coupling is not critical for ictogenesis in the low-magnesium cortical model.

In hippocampal (Bikson et al., 2002) and cortical seizure models (Gajda et al., 2005), quinine has been shown to increase seizure frequency concurrently with a reduction in seizure duration (and amplitude). An increase in event frequency may result from a reduction in refractory period due to the shorter duration of each seizure event (Bikson et al., 2002). This explanation does not account for the proseizure effects observed in our study. The baseline duration of each seizure-like event in the present study was already much shorter than in these other models (from 1 to 3 s) and was not significantly reduced during the period of heightened activity.

A potential confounder in this study is the effect of DMSO, a versatile solvent that is widely used to aid delivery of drug compounds that are poorly soluble in aqueous solution. A DMSO concentration of 1% is reported to be a “safe” experimental concentration (Kahler, 2000); however, a close look at the literature reveals that much lower concentrations (as low as 0.02%) might have significant effects on central nervous system function (Tsvyetlynska et al., 2005; Nasrallah et al., 2008). At the maximum dose used (0.2%), no consistent or significant effect on seizure-like event characteristics was observed, supporting earlier reports that this dose does not modulate neuronal excitability in the mammalian cortical slice preparation (Rosen & Andrew, 1991; Nasrallah et al., 2008). However, a marked reduction in seizure-like activity was noted in a subset of slices from a single animal, which was reversible with a return to low-magnesium ACSF. This may explain the small, but significant reduction in seizure-like event amplitude during mefloquine infusion in connexin36 knockout animals. The possibility of an antiseizurogenic effect of DMSO, although an important issue that deserves further attention, does not compromise the present results; if anything, this would have minimized the excitatory effects that we observed.

In conclusion, we have shown a paradoxical proseizure effect of pharmacologic gap junction blockade in a cortical model of seizure-like activity. We suggest that this effect is due to a disruption of inhibitory interneuron coupling resulting from connexin36 blockade. This finding also suggests that in the low-magnesium model, connexin36 gap junctions are not critical for ictogenesis.


This work was supported by the Marsden Fund of New Zealand (07-UOW-037). We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Disclosure: None of the authors has any conflict of interest to disclose.