Address correspondence and reprint requests to Dr. P. A. Schwartzkroin at University of California, Davis, Department of Neurological Surgery, One Shields Ave., Davis, CA 95616, U.S.A. E-mail: firstname.lastname@example.org
Summary: Purpose: To investigate further the membrane properties and postsynaptic potentials of the CA3 pyramidal cells in mice that display spontaneous seizures because of a targeted deletion of the Kcna1 potassium channel gene (encoding the Kv1.1 protein subunit).
Methods: Intracellular recordings were obtained from CA3 pyramidal cells in hippocampal slices prepared from Kcna1-null and control littermates. CA3 pyramidal cells were activated: orthodromically, by stimulating mossy fibers; antidromically, by activating Schaffer collaterals; and by injecting intracellular pulses of current. Responses evoked under these conditions were compared in both genotypes in normal extracellular medium (containing 3 mM potassium) and in medium containing 6 mM potassium.
Results: Recordings from CA3 pyramidal cells in Kcna1-null and littermate control slices showed similar membrane and action-potential properties. However, in 33% of cells studied in Kcna1-null slices bathed in normal extracellular medium, orthodromic stimulation evoked synaptically driven bursts of action potentials that followed a short-latency excitatory postsynaptic potential (EPSP)-inhibitory PSP (IPSP) sequence. Such bursts were not seen in cells from control slices. The short-latency γ-aminobutyric acid (GABA)A-mediated IPSP event appeared similar in null and control slices. When extracellular potassium was elevated and excitatory synaptic transmission was blocked, antidromic activation or short pulses of intracellular depolarizing current evoked voltage-dependent bursts of action potentials in the majority of cells recorded in Kcna1 null slices, but only single spikes in control slices.
Conclusions: Lack of Kv1.1 potassium channel subunits in CA3 pyramidal cells leads to synaptic hyperexcitability, as reflected in the propensity of these cells to generate multiple action potentials. The action-potential burst did not appear to result from loss of GABAA receptor–mediated inhibition. This property of CA3 neurons, seen particularly when tissue conditions become abnormal (e.g., elevated extracellular potassium), helps to explain the high seizure susceptibility of Kcna1-null mice.
Many different types of potassium ion channels are expressed in the CNS (1). Ion currents activated by these potassium channels possess different physiological and pharmacologic properties and are involved in regulating such parameters as membrane potential and pattern of action-potential discharge (2). Potassium-selective ion channels, therefore, play an extremely important role controlling the processes of excitation. Reduction in outward potassium currents induced by elevated concentrations of extracellular potassium (3–5) or potassium channel blockers (6–9) may lead to neuronal hyperexcitability and even seizure activity in experimental models of epilepsy. These types of epileptiform activities are often associated with bursts of action potentials, enhanced postsynaptic excitation, and reduced postsynaptic inhibition. Mutations in potassium channel genes also can cause epilepsy in humans (10,11), although the underlying cellular mechanisms are still unclear.
The Kcna1 potassium channel gene belongs to the Shaker subfamily of voltage-gated potassium channel genes. When expressed in heterologous systems, cloned Kcna1 gives rise to a rapidly activating, slowly inactivating outward current that demonstrates outward rectification (12–14). This current is sensitive to a number of potassium channel blockers, including 4-aminopyridine, tetraethylammonium, and dendrotoxin. Previous studies indicated that Kv1.1, the Kv channel subunit encoded by Kcna1, forms heteromultimeric channels widely distributed in the brain (15) and, in particular, in axons of CA3 pyramidal neurons in the hippocampus (16,17).
Interestingly, distribution of Kv1 potassium channel subunits of the Kcna gene family correlates with differential seizure susceptibility of different brain areas (18). The hippocampus appears to be one of the more seizure-prone structures of the brain and shows a particularly high level of Kv1.1 immunoreactivity in normal mice (17). The CA3 region—the region of the hippocampus that shows the highest level of Kv1.1 expression—is capable of initiating epileptiform discharges because of its morphologic and physiological organization (19–21). Previously we showed that Kcna1-null mice exhibited spontaneous behavioral and EEG seizures (16) in a pattern typical of temporal lobe epilepsy. In hippocampal slices prepared from Kcna1-null mice, CA3 neurons exhibit axonal properties and postsynaptic activity consistent with their role in epileptiform activities (16). In the current study, we further investigated the properties of CA3 pyramidal neurons to describe the mechanisms underlying enhanced seizure susceptibility in mice lacking the Kv1.1 potassium channel subunit.
Kcna1tm1Tem (Kcna1 targeted mutation 1 from the Tempel laboratory) mutant mice were generated as described in Smart et al. (16) and transferred into the C3HeB/FeJ background. We used C3HeB/FeJ-Kcna1tm1Tem mutant and normal control mice derived from heterozygous parents intercrossed at backcross generations >N5. Offspring (referred to here as Kcna1-null and control) were genotyped before experimentation by using a three-primer polymerase chain reaction (PCR) screen of tail-clip DNA (protocol available online at http://depts.washington.edu/tempelab/index.html). Animals were housed in specific-pathogen–free facilities approved by the Association for Assessment and Accreditation of Laboratory Animal Care. All animal protocols were reviewed and approved by the University of Washington Institutional Animal Care and Use Committee.
One- to two-month-old Kcna1-null and age-matched littermate control mice were decapitated under halothane anesthesia, and their brains were quickly removed into 2–4°C artificial cerebrospinal fluid (aCSF) containing (in mM): 124 NaCl, 3 KCl, 2 CaCl2, 2 MgSO4, 26 NaHCO3, 1.25 NaH2PO4, and 10 dextrose, saturated with 95% O2/5% CO2 gases (pH, 7.4). Transverse hippocampal slices (400 μm thick) were cut with a Vibroslicer (Campden Instruments, Sileby, U.K.) and then transferred to a holding chamber containing gas-saturated aCSF at room temperature (22–24°C) for ≥1 h before recording. During recordings, slices were kept at 32°C at an interface between oxygenated aCSF and humidified gas. Rate of perfusion (0.8–1 ml/min) was kept constant throughout the experiment.
Mossy fibers and Schaffer collaterals were activated by a bipolar stainless steel stimulating electrode. Stimuli (0.1-ms duration) were delivered at 0.1 Hz. Electrodes for intracellular recordings were pulled from borosilicate glass capillary tubing (1 mm outer diameter; Sutter Instruments, Novato, CA, U.S.A.) by using a Sutter Instrument P-87 one-stage electrode puller (San Rafael, CA, U.S.A.). Microelectrodes were filled with 3 M potassium acetate (pH adjusted to 7.4 with KOH) and had resistances of 80–100 MΩ. Only neurons with a resting membrane potential and synaptic responses stable for ≥20 min and input resistance >30 MΩ were included in our analysis. Signals were recorded by using an Axoclamp-2A amplifier (Axon Instruments, Foster City, CA, U.S.A.) in bridge mode. Bridge was monitored throughout the experiment and adjusted if necessary. Cell membrane potential was manipulated by injection of steady positive or negative current through the intracellular microelectrode. In some experiments, action potentials were evoked by brief depolarizing current pulses (duration, 5 ms; amplitude, 0.5–1 nA) injected through the intracellular microelectrode. Membrane current–voltage (I-V) relation, input resistance, and time constant were determined with longer pulses of positive and negative current (duration, 100 ms; amplitude, 0.5 to −0.7 nA). The frequency of action potentials was calculated from duration of current-evoked bursts containing only three action potentials. The duration of bursts was measured from the rising phase of the first to the repolarization phase of the third action potential. Action-potential duration was measured at half-peak. Cell resting membrane potential was verified after withdrawal of the microelectrode from the cell. Data were digitized (Neuro-Corder; Neuro Data Instruments, New York, NY, U.S.A.) and acquired by computer.
Measurements throughout the text are expressed as mean ± SEM, and n indicates the number of neurons studied. The results obtained were compared by using Student's t test and were considered significantly different at p < 0.05.
In some experiments, the noncompetitive glutamate-receptor antagonist kynurenic acid (10 mM, Sigma) and/or the γ-aminobutyric acid (GABA)A-receptor antagonist bicuculline (10 μM, Sigma) was added to aCSF to block excitatory and inhibitory synaptic transmission.
Intracellular recordings were obtained in normal aCSF from CA3 pyramidal cells in hippocampal slices prepared from Kcna1-null (23 neurons from four animals) and control (21 neurons from four animals) mice. Membrane I-V relations were investigated by injecting pulses of negative and positive current through the intracellular microelectrode (Fig. 1A, C). I-V plots were similar in both genotypes, although slight rectification at membrane potentials more negative than −100 mV was observed in six of 10 neurons in control slices (Fig. 1B). Neurons from Kcna1-null slices usually demonstrated a linear I-V relation in the hyperpolarizing direction (Fig. 1D); only three of eight recorded cells displayed rectification similar to that seen in control slices. Action potentials evoked by depolarizing current pulses had similar waveforms (Fig. 1A and C) and frequency (Fig. 1E and F) in both genotypes. Passive membrane properties and action-potential parameters were comparable in control and Kcna1-null slices (Table 1).
Table 1. Passive membrane properties and action-potential parameters of CA3 pyramidal cells
Resting membrane potential (mV)
Input membrane resistance (MΩ)
Time constant (ms)
Action potential amplitude (mV)
Action potential duration (ms)
65.7 ± 1.3 (n = 21)
56.5 ± 3.3 (n = 21)
48.5 ± 3.2 (n = 10)
92.4 ± 1.7 (n = 21)
0.74 ± 0.05 (n = 21)
64.5 ± 1.3 (n = 23)
58.6 ± 3.1 (n = 23)
50.3 ± 3.7 (n = 8)
89.8 ± 1.4 (n = 23)
0.76 ± 0.02 (n = 23)
In normal aCSF, mossy fiber stimulation in slices from control mice induced a sequence of postsynaptic potentials that included a fast initial excitatory postsynaptic potential (EPSP, arrow in Fig. 2A, −60 mV), and a fast inhibitory postsynaptic potential (IPSP, circle in Fig. 2A, −60 mV); the latter was followed by a slow IPSP (square inFig. 2A, −60 mV). This pattern of responses was recorded in all 23 neurons studied. The “normal” type of postsynaptic response also was recorded in 14 of 21 cells in Kcna1-null slices (not shown). In these responses, peak amplitude of the fast IPSP in control slices (17.2 ± 0.7 ms, n = 9) was similar to that of Kcna1-null slices (17.4 ± 0.6 ms, n = 8). Mossy fiber–evoked postsynaptic responses were induced at different membrane potential levels when steady positive or negative current was injected through the intracellular microelectrode (Fig. 2A, −80 and −100 mV). The fast IPSP, measured at 15 ms after mossy fiber activation, had a linear voltage dependency and reversed in polarity close to −70 mV in both control (Fig. 2B) and Kcna1-null slices (not shown).
However, in seven of 21 cells recorded in Kcna1-null slices, mossy fiber–evoked responses were different. At the resting membrane potentials, these responses consisted of an initial EPSP (arrow in Fig. 2C, −60 mV) and a fast hyperpolarization (circle in Fig. 2C, −60 mV), followed by a slow depolarizing envelope that was capped by a burst of action potentials. The slow depolarization was followed by a long-lasting hyperpolarization (Fig. 2C, −60 mV). The fast hyperpolarization reversed in polarity when steady hyperpolarizing current was injected through the intracellular microelectrode; voltage dependency of this event, measured at 15 ms after mossy fiber stimulation, was linear, and the reversal potential was close to −70 mV (Fig. 2D). The voltage dependency of this event is similar to that of the fast IPSP recorded in control and “normal”Kcna1-null slices. The action-potential burst was not completely blocked by membrane hyperpolarization, and the amplitude of underlying slow depolarization was enhanced as membrane potential was hyperpolarized (Fig. 2C), as would be expected for a synaptically driven event.
Postsynaptic responses also were recorded in five cells from Kcna1-null slices (one animal) during application of the GABAA-receptor antagonist bicuculline. Under these conditions, mossy fiber stimulation evoked bursts of action potentials with an underlying slow membrane depolarization (Fig. 2E). At membrane potentials close to the resting levels, this depolarization was not preceded by a fast hyperpolarization (Fig. 2E, −55 mV), as it was observed in “bursting”Kcna1-null cells recorded in normal aCSF. Only the initial fast EPSP preceded burst discharge (arrow in Fig. 2E, −55 mV), and it increased in amplitude at more negative membrane potentials (arrows in Fig. 2E, −75 mV and −95 mV). The voltage dependency of the postsynaptic event measured at 15 ms after mossy fiber activation was very different from that recorded in control and Kcna1-null neurons in normal aCSF. This event always remained positive, even at depolarized membrane potentials (Fig. 2F), suggesting that the GABAA-mediated IPSP indeed had been blocked.
Previously we showed that moderate elevation of extracellular potassium (6 mM) in aCSF induced spontaneous field-potential discharges in Kcna1-null, but not in control slices (16). Therefore excitability of CA3 pyramidal cells was evaluated in aCSF containing an elevated concentration of potassium (6 mM). The noncompetitive glutamate-receptor blocker, kynurenic acid (10 mM), was added to block spontaneous burst discharges driven by recurrent glutamatergic synaptic excitation in the CA3 region. Antidromic activation of CA3 pyramidal cells was achieved by Schaffer collateral stimulation. Under these conditions, subthreshold stimulation did not induce any responses in CA3 pyramidal cells (Fig. 3A, Control). Stimulation at threshold intensity evoked a single action potential in cells from control slices, whereas neurons recorded in Kcna1-null slices responded with a burst of action potentials (Fig. 3A, Kcna1-null, −63 mV). The averaged number of action potentials was 1.3 ± 0.2 (seven cells from two animals) in control slices and 3.3 ± 0.3 (seven cells from one animal) in Kcna1-null slices (Fig. 3B). A slow membrane depolarization underlay burst discharges in cells from Kcna1-null slices. Burst discharge, and the associated depolarization, could be blocked by injection of steady negative current through the intracellular microelectrode (Fig. 3A, Kcna1-null, −67 mV), indicating the voltage-dependent nature of this slow depolarization.
Short pulses of depolarizing current (duration, 5 ms; amplitude, 0.5–1 nA) also were injected through the intracellular microelectrode into CA3 pyramidal cells when the bathing medium contained the noncompetitive glutamate receptor blocker kynurenic acid (10 mM) and elevated extracellular potassium (6 mM). Under these conditions, threshold current injected in neurons in control slices usually evoked a single action potential (Fig. 3C, Control), whereas cells in Kcna1-null slices generated bursts of action potentials (Fig. 3C, Kcna1-null, −62 mV). The averaged number of action potentials was 1.1 ± 0.1 (n = 7) in control slices and 3.5 ± 0.4 (n = 7) in Kcna1-null slices (Fig. 3D). A slow membrane depolarization, similar to that evoked by antidromic activation, was associated with these burst discharges and could be blocked by membrane hyperpolarization (Fig. 3C, Kcna1 null, −65 mV). The intensity of depolarizing intracellular current necessary to evoke an action-potential discharge in control slices (0.53 ± 0.05 nA, n = 7) was similar to that in Kcna1-null slices (0.61 ± 0.06 nA, n = 7).
Our data demonstrate that CA3 pyramidal neurons have similar I-V relations, passive membrane properties, and action-potential parameters in both control and Kcna1-null slices. These results are consistent with previous findings of unaltered membrane properties and action-potential parameters of CA3 and neocortical cells in Kcna1-null slices (16,22). These data indicate that lack of the Kv1.1 channel subunit does not affect membrane properties of the cell soma in CA3 pyramidal cells—probably because these channels are localized primarily to axons and terminals (16,17)—too distant for their absence to affect responses to somatically injected current.
We have found that CA3 pyramidal cells in Kcna1-null slices are capable of generating late bursts of action potentials in normal aCSF in response to mossy fiber activation. This type of activity was never seen in control slices, where neurons responded only with initial short-lasting EPSPs, followed by short- and long-lasting IPSPs. Membrane depolarizations underlying late bursts in Kcna1-null slices demonstrated properties of a PSP: they could not be blocked by steady membrane hyperpolarization and increased in amplitude as membrane was hyperpolarized. Similar membrane depolarization-associated bursts of action potentials were described in CA3 cells when GABAA receptor–mediated inhibition was blocked (19–21) or reduced because of elevated extracellular potassium (3,4,23). These bursts are driven by glutamatergic synaptic excitation, synchronized via recurrent collaterals among CA3 pyramidal cells. We conclude that similar interactions underlie bursts of action potentials in Kcna1-null slices and are likely due to repetitive discharge of depolarized axon collaterals, and/or the slow repolarization of terminals, lacking the Kv1.1 subunit contribution to low voltage–activating potassium currents (24).
In our experiments, synaptically driven bursts were recorded only in a portion of Kcna1-null neurons studied. It is important to take into consideration that not only the properties of a single cell, but also the circuitry through which these cells interact, are key contributors to the abnormal (i.e., burst-firing) activity. Variability in the network connectivity of an individual neuron could explain the variable synaptic phenomena observed in the Kcna1-null neurons.
In the experiments conducted in normal aCSF, burst discharges recorded in Kcna1-null slices were preceded by a fast membrane hyperpolarization. Our data suggest that these hyperpolarizations are mediated by GABAA receptors because they have time course and voltage dependency similar to those of GABAA receptor–mediated IPSPs in “normal” responses evoked by mossy fiber activation. This postsynaptic event also was blocked in Kcna1-null neurons recorded in the presence of the GABAA-receptor antagonist bicuculline. Therefore in this model of epilepsy, intensive excitatory synaptic interactions among neurons may take place under conditions in which GABAA receptor–mediated inhibition is still present. GABA-mediated events preceding similar discharges also have been detected in different short-term models of epileptiform activity (4,8). It is still unclear from our experiments whether a quantitative change in GABAA-mediated inhibition is present, although the inhibitory drive onto neocortical pyramidal neurons (22) and cerebellar Purkinje cells (25) is augmented in Kcna1-null slices under “normal” conditions. Given that alterations of inhibition in a population of CA3 pyramidal neurons can modulate synaptically driven burst discharges, it seems likely that variation in inhibitory strength contributes to the variability of postsynaptic responses in Kcna1-null slices.
Antidromic activation of CA3 pyramidal cells evoked bursts of action potentials in Kcna1-null slices and only single action potentials in control slices when excitatory synaptic transmission was blocked and extracellular potassium was moderately enhanced. These spikes are not driven by synaptic excitation, but rather are generated by axonal activity invading the cell soma. These results suggest that deletion of the Kv1.1 potassium channel, which has primarily axonal localization in hippocampus, affects CA3 pyramidal cell excitability. This excitability could result from increased action-potential initiation at trigger zones or from increased probability of action-potential propagation through branch points into axon collaterals (26). Enhanced excitability of axons in the hippocampus, brainstem, and peripheral nervous system in Kcna1-null mice also has been suggested previously (16,24,27). In the current study, the triggering zone for bursts in Kcna1-null slices could not be located precisely along the axon. It is likely that some of these zones are located close to the cell body (e.g., initial segment or first node of Ranvier), because brief pulses of current injected into the soma also initiated bursts of action potentials (similar to those induced by antidromic activation) in Kcna1-null slices.
Bursts of action potentials generated in axons and/or terminals have been shown to contribute to neuronal excitability in other models of epileptiform activity (28,29). Critically, spikes generated in axons propagate in both directions and may greatly increase neurotransmitter release. We conclude that axonal hyperexcitability in Kcna1-null mice may facilitate propagation of activity via excitatory recurrent collaterals in the CA3 region, and thus may enhance excitatory synaptic neurotransmission and postsynaptic action-potential discharge.
Acknowledgment: This work was supported in part by NIH grant NS18895 to P.A.S. and RO1 DC03805 to B.L.T.