INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Gap junctions are an important means of intercellular communication and have long been known to play a critical role in the developing nervous system (for review see Rorig and Feller 2000). However, recently it has become clear that gap junctions in the adult brain may also have a significant function in the generation of neuronal population activity. MacVicar and Dudek (1981) were the first to demonstrate (in adult hippocampus) the direct intercellular passage of current when they recorded fast prepotentials (now called spikelets) between pairs of CA3 pyramidal cells. Electrical coupling is not restricted simply to principal cells but has now been demonstrated between pairs of interneurons in the neocortex (Beierlein et al. 2000; Galarreta and Hestrin 1999; Gibson et al. 1999; Tamas et al. 2000), hippocampus (Bartos et al. 2001; Hormuzdi et al. 2001; Venance et al. 2000), and cerebellum (Mann-Metzer and Yarom 1999). It has been suggested that gap junctions play a role in a number of different in vitro models of seizure activity (for reviews see Carlen et al. 2000; Perez-Velazquez and Carlen 2000; Traub et al. 2001). Spikelets have been recorded in CA1 pyramidal cells during spontaneous burst discharges evoked by superfusion with calcium-free artificial cerebrospinal fluid (ACSF) (Valiante et al. 1995) and during ultrafast oscillations (Draguhn et al. 1998). In addition to synaptic excitation and inhibition, gap junctions may also be important for the generation of the ultrafast ripple (approximately 200 Hz) oscillations, which occur with sharp-wave activity in vivo (Buszaki et al. 1992) as these were abolished under halothane anesthesia (Ylinen et al. 1995). Ultrafast oscillations have also been observed in vitro in calcium-free ACSF in the hippocampus (Draguhn et al. 1998). These high-frequency oscillations could be recorded in either the CA1 or the CA3 subfields of the hippocampus and were blocked by several putative modulators of gap junctions, including octanol and halothane (Draguhn et al. 1998). Experimental (Draguhn et al. 1998; Schmitz et al. 2001) and modeling (Traub and Bibbig 2000; Traub et al. 1999) studies have suggested that these high-frequency ripples may be generated by axon-axon gap junctions between pyramidal cells.

Gap junctions are formed by connexins (Cxs), of which 19 members have been identified in the mouse genome, and they assemble in a hexameric stoichiometry to form channels (Willecke et al. 2002). The recently cloned Cx36 (Condorelli et al. 1998; Sohl et al. 1998) is expressed primarily in neurons. Recently we have generated a connexin 36 knockout (Cx36KO) mouse that consequently lacked electrical coupling in pairs of interneurons recorded in both the dentate gyrus and the CA3 region of the hippocampus (Hormuzdi et al. 2001). Previous studies have shown that bath application of either carbachol (20 µM) or kainate (100-300 nM) elicits a persistent gamma-frequency (30-80 Hz) oscillation in the hippocampus (Fisahn et al. 1998) and neocortex (Buhl et al. 1998) that involves both interneurons and pyramidal cells. In slices from the Cx36KO mice, gamma-frequency oscillations, evoked by both carbachol and kainate, were disrupted although not abolished (Hormuzdi et al. 2001).

We now report that, in addition to these disruptions in oscillatory activity and electrical signaling between interneurons in slices from Cx36KO mice, we observed sharp wave-like burst discharges following kainate application. The burst discharges were often accompanied by a fast ripple oscillation and occurred in conjunction with the kainate-evoked -frequency oscillations. These results provide further evidence that long-term loss of electrical signaling in the interneuron network can have a profound effect on the excitability and synchronization properties of the hippocampal neuronal network. A preliminary report of this work has been published in abstract form (Pais et al. 2001).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Preparation of slices, solutions, and drugs

Cx36KO mice were generated as described previously (Hormuzdi et al. 2001). Adult male mice were anesthetized with inhaled isoflurane followed by im injection of ketamine (100 mg kg¹) and xylazine (10 mg kg¹). After the abolition of all pain reflexes, the animals were perfused intracardially with approximately 25 ml of modified ACSF, which was composed of the following (in mM): 252 sucrose, 3.0 KCl, 1.25 NaH₂PO₄, 24 NaHCO₃, 2.0 MgSO₄, 2.0 CaCl₂, and 10 glucose. Following brain removal, 450-µm-thick horizontal slices were cut. Slices were then trimmed and transferred to a holding chamber where they were maintained at room temperature at the interface between normal ACSF (where sucrose was replaced with 126 mM NaCl) and humidified 95% O₂-5% CO₂. For subsequent recording slices were transferred to an interface chamber maintained at 34-35°C. The following drugs were used: bicuculline methochloride (1-6 µM; TOCRIS, UK); carbenoxolone (100-200 µM); carbamylcholine chloride (carbachol, 20 µM); 4-aminopyridine (4-AP, 20-50 µM); trimethylamine (TMA, 1-10 mM; Sigma, UK); and kainic acid (100-300 nM; RBI, UK).

Recording, data acquisition, and analysis

Intracellular recordings were made using 1.5 M KCH₃SO₄ filled glass microelectrodes pulled to resistances of 50-90 M. Extracellular recording electrodes were filled with ACSF (resistance 2-5 M). Data were recorded with an Axoclamp 2A amplifier (Axon Instruments) and recorded on a computer via either an ITC-16 interface (Instrutech, USA) or a CED 1401 (Cambridge Electronic Design). Data were acquired and analyzed on a computer using Axograph software (Axon Instruments) or Spike 2 (Cambridge Electronic Design). Fourier analysis gave the peak frequencies of the oscillations. A measure for power in the -frequency band was determined as the area under the peak in the power spectra between 15 and 60 Hz. The frequency of the fast "ripples" was determined from the auto-correlation of the fast oscillation. As small variations in ripple frequency can occur, the average frequency of six ripples was obtained for each slice/condition. Burst frequency was determined by counting the number of bursts, clearly visible in the recording traces, in a 60-s recording period. In experiments using Cx36KO mice, slices were prepared each day from one Cx36KO mouse, and one wild-type (WT) littermate mouse. To control for small variations in experimental conditions, a slice from each mouse was placed in the recording chamber and simultaneous recordings from Cx36KO and WT mouse slices were then carried out. Subsequent control experiments were performed on either WT littermates from the Cx36KO colony (C57 × SV129) or C57 control mice alone. Results are expressed as mean ± SE standard error and statistical significance was determined with a Student's t-test. A significance level of P < 0.05 was chosen.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Sharp wave-like activity evoked by kainate in the Cx36KO mice

Bath application of carbachol (20 µM) or kainate (100-300 nM) was used as described previously (Buhl et al. 1998; Fisahn et al. 1998) to evoke -frequency oscillations in the hippocampus in vitro in slices from both WT and Cx36KO mice (Hormuzdi et al. 2001). We found that, in addition to -frequency oscillations, kainate also evoked burst discharges in slices from Cx36KO mice but not in slices from WT mice (Fig. 1). Extracellular recordings from stratum radiatum in the CA3 region following bath application of kainate (100-300 nM) revealed a -frequency oscillation in WT slices (Fig. 1Ai). However, in the slices from the Cx36KO mice, kainate-induced -frequency activity was accompanied by sharp wave (SPW) burst discharges (Fig. 1Aii). In total, sharp wave activity was seen in 4/5 Cx36KO mice (8/11 slices) but not in WT mice (11 slices/5 mice). The sharp wave activity occurred simultaneously with the -frequency activity in slices from the Cx36KO mice. However, as recently reported (Hormuzdi et al. 2001), the power of the -frequency activity (see METHODS) evoked by kainate was reduced in the slices from the Cx36KO mice compared with WT (Fig. 1, Bi-ii). In Cx36KO mice the mean power of the -frequency activity evoked with 300 nM kainate was 1.35 ± 0.30 µV² compared with 3.95 ± 1.64 µV² from the WT mice. The emergence of the burst discharges only occurred in the presence of kainate. Bath application of carbachol evoked a -frequency oscillation in Cx36KO mice which had a reduced power compared with WT (Hormuzdi et al. 2001) with a mean of 0.94 ± 0.46 µV² in Cx36KO mice and 3.55 ± 1.37 µV² in WT mice, but no burst discharges were seen in either Cx36KO mice (6 mice; 10 slices) or WT mice (6 mice; 10 slices). In the presence of kainate each burst was followed by a transient suppression of the activity (e.g., Fig. 1Aii) which lasted approximately 2 s. The -frequency activity then increased in amplitude and, just prior to the next burst discharge, large population synaptic events (Fig. 1Bii) were evident. The onset and frequency of the burst discharges was dependent on the concentration of kainate. In 10 slices (Cx36KO mice) the frequency of the burst discharges was measured at increasing concentrations of kainate and only 1/10 slices showed burst discharges at 100 nM kainate, 6/10 at 200 nM, and 7/10 at 300 nM kainate (Fig. 1C). In addition, in 4/8 Cx36KO slices the burst discharges were accompanied by a series of ultrafast (approximately 60-115 Hz) population spikes or ripples (Fig. 2A). These ripples were slower, but phenomenologically similar, to those described in the hippocampus both in vivo (Buszaki et al. 1992; Ylinen et al. 1995) and in vitro (Draguhn et al. 1998). No ripples or SPW activity were seen with kainate application in any of the slices from the WT mice. As our recordings were made in s. radiatum, and the fast ripple oscillations are usually best observed in s. pyramidale (Draguhn et al. 1998; Ylinen et al. 1995), the amplitude of this activity was often small. However, in 4/8 Cx36KO slices fast ripples were evident (Fig. 2A) and the mean frequency of the ripple was determined to be 71.3 ± 14.9 Hz. No ripples were seen in slices from the Cx36KO mice before addition of kainate, although ultrafast oscillations could be seen in calcium-free ACSF in slices from both Cx36KO and WT mice (Hormuzdi et al. 2001) as previously described (Draguhn et al. 1998). Intracellular recordings from CA3 pyramidal cells during sharp wave activity (n = 7) revealed that the extracellular field burst discharge was associated with a large membrane depolarization. In one cell the depolarization was accompanied by a clear train of spikelets occurring prior to the onset of the burst discharge (Fig. 2B).

View larger version (16K): [in this window] [in a new window]   Fig. 1. A: simultaneous extracellular recordings from stratum radiatum in the CA3 region of hippocampal slices from wild-type (WT) (Ai) and Cx36KO (Aii) mice following bath application of kainate (300 nM). -frequency oscillations are present in WT, and in the Cx36KO slice there is some -frequency activity, but burst discharges are also evident. Expanded sections of the traces indicated by the dark bar are shown in (B). In the WT slice (Bi) there is a -frequency oscillation, while in the Cx36KO (Bii) -frequency activity is reduced and large population spikes (*) occur prior to the onset of a burst discharge. C: the frequency of burst discharges in Cx36KO mice was dependent on kainate concentration.

View larger version (10K): [in this window] [in a new window]   Fig. 2. Simultaneous extracellular (Ai) and intracellular (Bi) recordings from a CA3 pyramidal cell from a Cx36KO mouse slice showing burst discharges in the presence of kainate (300 nM). A transient, fast ("ripple") oscillation was evident in the extracellular trace accompanying the burst discharge. Intracellular recording revealed a train of fast spikelets (0.5 mV) and action potentials superimposed on a large depolarization. Expanded sections indicated by the dark bars are shown in Aii and Bii.

Kainate induced sharp wave activity in Cx36KO mice is blocked by carbenoxolone

Several types of seizure-like activity evoked in vitro have been shown to be sensitive to gap junction blockers (for review see Carlen et al. 2000; Perez-Velazquez et al. 1994) and physiological SPWs recorded in vivo are blocked by halothane anesthesia (Ylinen et al. 1995). Similarly, -frequency activity evoked by either carbachol (Traub et al. 2000) or kainate (E. H. Buhl and A. Fisahn, unpublished observations) is blocked by carbenoxolone. We therefore investigated whether the gap junction blocker carbenoxolone had any effect on the kainate evoked burst discharges in slices from the Cx36KO mice (n = 7). Figure 3A shows an example of one experiment in which, with bath application of 300 nM kainate, there is a gamma-frequency oscillation interspersed with burst discharges. Following 30-45 min bath application of 200 µM carbenoxolone, the burst discharges were abolished (Fig. 3Aii). At this time point the -frequency oscillation persisted, although it now occurred in transient bursts as opposed to a persistent oscillation. With longer carbenoxolone applications (60 min), gamma-frequency activity was completely blocked (Fig. 3Aiii). The group data (Fig. 3, B and C) show that after 30 and 60 min of carbenoxolone application burst frequency was reduced by approximately 45 and 100%, respectively (n = 3). The gamma-frequency activity was also blocked by carbenoxolone with reductions of 90 and 96% after 30 and 60 min, respectively.

View larger version (16K): [in this window] [in a new window]   Fig. 3. A: extracellular recording from CA3 s. radiatum of a Cx36KO mouse slice during sharp wave activity evoked by Ai bath application of kainate (300 nM). Aii: after 30 min bath application of carbenoxolone (200 µM), all sharp wave activity was abolished and the remaining gamma-frequency activity occurred in transient epochs of rhythmic activity. Aiii: after 60 min application of carbenoxolone, the -frequency activity was fully abolished. B: decline in burst frequency with time into carbenoxolone application (n = 3). C: the power (15-60 Hz) of the -frequency activity with increasing time into carbenoxolone application (n = 7).

Sharp wave activity is not replicated by disruption of GABAergic inhibition alone

We next tested whether a disruption and/or reduction of GABAergic inhibition, as a consequence of the loss of Cx36 in the KO mice, may account for the burst discharges observed in the presence of kainate. Thus we bath applied kainate to control slices to establish a -frequency oscillation and subsequently added increasing concentrations of the -aminobutyric acid-A (GABA_A) receptor antagonist bicuculline (Fig. 4A). Low concentrations of bicuculline (1-4 µM) (6 mice; 6 slices) reduced the power of the -frequency activity by 63 ± 5.4% (Fig. 4, Aii-iii) but failed to evoke any burst discharges. At higher concentrations of bicuculline (6 µM; n = 2) -frequency activity was greatly reduced or abolished before bursts discharges became evident (Fig. 4Aiv).

View larger version (16K): [in this window] [in a new window]   Fig. 4. Extracellular recording from the CA3 region in a control mouse slice demonstrating -frequency activity (Ai) in the presence of kainate (300 nM). Addition of bicuculline (2 µM) caused a reduction in -frequency activity (Aii) that was even more marked with (4 µM) bicuculline (Aiii). Gamma-activity was abolished before the onset of burst discharges with 6 µM bicuculline (Aiv).

Sharp wave activity was not evoked by increasing gap junction conductance

The sharp wave activity observed in the Cx36KO mice might be a consequence of a compensatory change in gap junction conductance via other connexins. We therefore attempted to replicate this situation using bath application of TMA, which is reported to open gap junctions as a result of intracellular alkalinization (Lee et al. 1996; Spray et al. 1981). Following the establishment of kainate-evoked -frequency oscillations, bath application of TMA (2-6 mM) to three control slices failed to elicit any burst discharges (data not shown) and the power of the gamma-frequency activity was also unchanged. Increasing the concentration of TMA to 10 mM caused the -frequency activity to collapse but burst discharges and/or ultrafast ripples were not observed at any concentration of TMA used here.

Sharp wave activity was replicated in control mice by bath application of 4-AP

Bath application of the potassium channel blocker 4-AP (applied once -frequency activity had been established with 300 nM kainate) produced a pattern of activity in control slices that was qualitatively very similar to the activity observed in the Cx36KO mice. In total 13/16 slices showed burst discharges with application of 20-80 µM 4-AP. Figure 5B illustrates one example in which 4-AP (25 µM) resulted in the emergence of burst discharges occurring in conjunction with a -frequency oscillation. The effects of 4-AP were concentration dependent with 0/6 slices showing any burst discharges at 10 µM but 13/16 slices exhibited burst responses at 20-60 µM 4-AP. The mean burst frequency with 4-AP at 30-80 µM was significantly faster than that seen in the Cx36KO mice (P < 0.05; mean of 0.43 ± 0.22 Hz). Intracellular recordings from CA3 pyramidal cells (n = 5) during the 4-AP-evoked burst discharges (Fig. 5Dii) revealed a depolarizing response. No spikelets were seen in the intracellular recordings with 4-AP although, as observed in the Cx36KO mice, the 4-AP bursts in control mice (10/13) could also be accompanied by a fast ripple-like event (Fig. 5D). In 4/10 of these slices the ripples were measurable and the mean frequency of this activity in the presence of 4-AP was 55.3 ± 3.0 Hz. In six slices the power of the -frequency activity occurring with kainate alone was compared with that occurring after addition of 20 µM 4-AP. Overall the power of the gamma activity increased in 5/6 slices after application of 4-AP (mean increase 166 ± 62%) but, because of the high variability, was not significant. However, with longer duration applications, or higher concentrations of 4-AP, the burst discharges occurred so frequently that the -frequency oscillation was often reduced (data not shown). We also tested the ability of carbenoxolone to block the 4-AP-evoked burst discharges. Similar to the results obtained in slices from Cx36KO mice, application of carbenoxolone abolished the burst discharges evoked by application of 4-AP in slices from control mice (Fig. 6). Carbenoxolone application (n = 4) for 30 and 60 min significantly (P < 0.05) decreased the burst discharge frequency by 17 ± 11.5 and 83 ± 14%, respectively (Fig. 6B).

View larger version (18K): [in this window] [in a new window]   Fig. 5. A: extracellular recording from a control mouse slice in the CA3 region showing -frequency activity evoked by bath application of 300 nM kainate. With the addition of 4-AP (B) burst discharges occur in conjunction with the -frequency activity which is increased in this example. C: an expanded section from A. Extracellular recordings reveal ripple events accompanying the burst discharges (Di), while the simultaneous intracellular (Dii) recording reveals a large depolarization with partial spikes. Insets show autocorrelations.

View larger version (20K): [in this window] [in a new window]   Fig. 6. A: extracellular recording from a control mouse slice in the CA3 region showing -frequency activity evoked by bath application of 300 nM kainate (Ai). With the addition of 4-AP (Aii), burst discharges occurred in conjunction with the -frequency activity which were blocked by addition of carbenoxolone. B: group data show the reduction in burst discharge frequency with increasing time into carbenoxolone application.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In this study we report that in slices from Cx36KO mice, in addition to altered gamma-frequency oscillations, application of kainate (but not carbachol) results in the onset of sharp wave-like activity. The burst discharges in Cx36KO slices occurred intermittently during the -frequency activity that was reduced in power compared with WT mice (Hormuzdi et al. 2001). The generation of burst discharges required the absence of Cx36-containing gap junctions since the concentration of kainate used never induced burst discharges in slices from WT mice. Extracellular recordings revealed that the burst discharges, in several slices from Cx36KO mice, were often accompanied by transient, fast, ripple-like oscillations. Similar fast population spikes superimposed on an "epileptiform" field potential in the CA3 region have been described in several in vitro models of seizure-like activity (Schwartzkroin and Prince 1977; Wong and Traub 1983). These fast oscillations are qualitatively similar to the fast ripples that have been described in conjunction with sharp waves in the hippocampus in vivo (Buzsaki et al. 1992; Penttonen et al. 1999; Ylinen et al. 1995) and have also been recorded prior to epileptic seizures in humans (Traub et al. 2001). Recently Maier et al. (2002) have also reported burst discharges and high-frequency ripples in Cx36KO mice generated separately (Guldenagel et al. 2001). However, their findings differed somewhat from our results as they reported burst discharges and ripples in both control mice and Cx36KO mice, with the Cx36KO mice exhibiting less frequent burst discharges than the WT mice. We did not see any sharp wave activity in our WT mice, either in normal ACSF or after application of carbachol or kainate, to evoke the -frequency activity, and the occurrence of the bursts in our study was specific for the Cx36KO mice in the presence of kainate.

Alterations in synaptic inhibition and excitation have long been proposed to underlie several in vitro models of seizure-like activity (for reviews, see Ben-Ari and Cossart 2000; Meldrum and Chapman 1999; Olsen and Avoli 1997; Olsen et al. 1999; Treiman 2001). In addition, neuronal electrical signaling via gap junctions has also been shown to be important in several in vitro models of seizure-like activity (Carlen et al. 2000; Perez-Velaquez et al. 1994; Ross et al. 2000; Traub et al. 2001). Recent work in mice has suggested that, in the adult, Cx36 is preferentially expressed in hippocampal interneurons (Hormuzdi et al. 2001). Therefore our studies identifying sharp wave activity in the slices from adult Cx36KO mice demonstrates that perturbed electrical signaling in the interneuronal network can (in combination with kainate but not carbachol) result in burst discharges in vitro. It is also possible that, rather than being due to the loss of Cx36, this effect is the result of a compensatory upregulation of different connexin(s) in pyramidal cells. Since the fast ripple oscillations, presumed to be generated by gap junctions in the pyramidal cell network (Draguhn et al. 1998; Traub et al. 1999), persist in Cx36KO mice, we believe that gap junctions containing other connexins are present in the Cx36KO mice. However, we think it unlikely that upregulation of other connexin(s) has contributed to the burst discharges reported here because these fast oscillations were not appreciably different in Cx36KO when compared with WT (Hormuzdi et al. 2001). Nonetheless, to investigate this possibility further, we increased gap junction conductance in control slices by bath application of TMA, which is thought to open gap junctions as a result of intracellular alkalinization (Lee et al. 1996; Spray et al. 1981). Previous work has shown that transient -frequency oscillations evoked by tetanic stimulation in the CA1 region of the hippocampus can be followed by an ultrafast oscillation and burst discharges in the presence of TMA (Traub et al. 2000). However, in control mice TMA, when added to kainate, did not elicit any sharp wave activity or ultrafast oscillations during the persistent -frequency oscillation.

Interestingly, although in Cx36KO mice the gamma-frequency activity evoked by both kainate and carbachol application was reduced (Hormuzdi et al. 2001), the combination of burst discharges and a -frequency oscillation was only seen following application of kainate. Thus the differential effects of carbachol and kainate on the occurrence of sharp wave activity in Cx36KO slices may provide a clue to the underlying mechanism. Both kainate and carbachol have multiple, complex effects on neuronal excitability (for reviews see Ben-Ari and Cossart 2000; Frerking and Nicoll 2000). Kainate can either decrease GABA release on inhibitory terminals (Fisher and Alger 1984; Rodriguez-Moreno et al. 1997) or increase GABA release due to an increase in the firing rate of GABAergic interneurons (Cossart et al. 2001). Likewise, carbachol can also modulate inhibition either by exciting interneurons (Behrends and ten Bruggencate 1993; Pitler and Alger 1992) or by decreasing GABA release from their terminals (Behrends and ten Bruggencate 1993). However, the sharp wave activity evoked by kainate in Cx36KO mice was not simply a consequence of a global reduction of synaptic inhibition per se, since partial blockade of GABAergic inhibition using low concentrations of the GABA_A receptor antagonist bicuculline did not reproduce the characteristic pattern of burst discharges, occurring concomitantly with the -frequency activity. Bath application of low concentrations of bicuculline in control mice caused a decrease in the power of the -frequency activity, but burst discharges occurred only at higher concentrations of bicuculline, when all -frequency activity was abolished. Therefore a nonspecific reduction of synaptic inhibition is unlikely to account for the differences between WT and Cx36KO mice. In a separate study of Cx36KO mice Maier et al. (2002) found that, following application of 100 µM 4-AP, there was no significant difference in the occurrence of complex bursts or brief interictal events between WT and Cx36KO mice. This, combined with the fact that in our study burst discharges in Cx36KO mice occurred only with kainate and not carbachol, suggests that a nonspecific change in the threshold for seizure-like activity cannot account for the neuronal activity pattern we observed in Cx36KO mice.

Perhaps importantly for our observation of kainate-induced sharp wave activity in Cx36KO mice, kainate, but not carbachol, has been shown to depolarize axons and increase antidromic action potentials in both the mossy fibers of dentate granule cells (Kamiya and Ozawa 2000; Schmitz et al. 2001) and the hippocampal interneurons (Semyanov and Kullman 2001). Our data, therefore, suggest that loss of electrical signaling in the interneuronal network in Cx36KO mice may expose a kainate-induced increase in pyramidal cell axonal excitability that can trigger burst discharges. In support of the idea of a possible kainate-induced increase in pyramidal cell nonsynaptic excitability in slices from the Cx36KO mice, we found that the combination of burst discharges and -frequency activity could be qualitatively replicated in control mice slices by bath application of the potassium channel blocker 4-AP (Storm 1988). In vitro 4-AP application elicits a number of distinct behaviors including interictal-like burst discharges (Perreault and Avoli 1991, 1992; Rutecki et al. 1987) and a long-lasting depolarization mediated by GABA (Perreault and Avoli 1991, 1992). 4-AP is also known to increase ectopic action potentials, i.e., action potentials arising at regions distant from the axon hillock and soma (Avoli et al. 1998; Perreault and Avoli 1991, 1992; Traub et al. 1995). We found that 4-AP, when added along with kainate, evoked burst discharges in control slices that were very similar to those seen in the slices from the Cx36KO mice. As with the burst discharges seen in Cx36KO mice, an episode of fast ripple-like activity often accompanied the onset of the burst discharge.

Modeling studies have shown that spikelets and/or ripples prior to the onset of a burst discharge can be simulated by connecting cells in the network solely by axonal gap junctions in the pyramidal cells (Draguhn et al. 1998; Traub et al. 1999, 2001). Electrical coupling between hippocampal pyramidal cells was first demonstrated by MacVicar and Dudek (1981), although the exact location of the electrical coupling was unknown. These authors also suggested that the fast prepotentials, initially proposed to be dendritic spikes (Spencer and Kandel 1961), were in fact action potentials from the electrically coupled cell (MacVicar and Dudek 1981). To date there has been no direct evidence demonstrating the passage of current across individual pairs of pyramidal cell axons; however, recent studies have provided compelling experimental evidence for such connections (Draguhn et al. 1998; Schmitz et al. 2001). Although we have not identified the site of origin of the spikelets seen in the Cx36KO mice, their rapid upstroke and decay would be consistent with an axonal rather than dendritic origin (Draguhn et al. 1998). The block of burst discharges and the -frequency activity in the Cx36KO mice with the gap junction blocker carbenoxolone suggested a role for gap junctions formed of connexins other than Cx36. As with the bursts occurring in the slices from the Cx36KO mice, carbenoxolone also blocked the burst discharges evoked with 4-AP in control slices. A detailed study of the effects of carbenoxolone on hippocampal pyramidal cell excitability has recently been reported (Schmitz et al. 2001). This report showed that the intrinsic cell properties remained unaltered in the presence of carbenoxolone but that the amplitude of the recorded spikelets was decreased by 35-86%. However, as carbenoxolone could affect gap junctions in both neurons and glial cells, we cannot draw any firm conclusions about the locations, or cellular distributions, of remaining gap junctions in the Cx36KO mice.

A reduction of synaptic inhibition may be a salient mechanism in the induction of seizure-like activity in vitro. The present data suggest that a disruption of electrical signaling within inhibitory networks may also be sufficient to tip the balance of network activity toward the generation of burst discharges, similar to the sharp waves and interictal events observed in vivo. These and previous (Hormuzdi et al. 2001) observations of changes in network activity suggest an additional dimension to communication between individual elements in an interneuron network where nonsynaptic, direct, electrical interactions appear to be important for the appropriate expression of inhibition-based physiologically relevant brain rhythms.