Journal of Neurophysiology

Mechanisms of Neuronal Hyperexcitability Caused by Partial Inhibition of Na+-K+-ATPases in the Rat CA1 Hippocampal Region

Cyrille Vaillend, Susanne E. Mason, Matthew F. Cuttle, Bradley E. Alger


Extra- and intracellular records were made from rat acute hippocampal slices to examine the effects of partial inhibition of Na+-K+-ATPases (Na+-K+ pumps) on neuronal hyperexcitability. Bath application of the low-affinity cardiac glycoside, dihydroouabain (DHO), reversibly induced interictal-like epileptiform bursting activity in the CA1 region. Burst-firing was correlated with inhibition of the pumps, which was assayed by changes in [K+]o uptake rates measured with K+-ion-sensitive microelectrodes. Large increases in resting [K+]o did not occur. DHO induced a transient depolarization (5–6 mV) followed by a long-lasting hyperpolarization (∼6 mV) in CA1 pyramidal neurons, which was accompanied by a 30% decrease in resting input resistance. Block of an electrogenic pump current could explain the depolarization but not the hyperpolarization of the membrane. Increasing [K+]o from 3 to 5.5 mM minimized these transient shifts in passive membrane properties without preventing DHO-induced hyperexcitability. DHO decreased synaptic transmission, but increased the coupling between excitatory postsynaptic potentials and spike firing (E-S coupling). Monosynaptic inhibitory postsynaptic potential (IPSP) amplitudes declined to ∼25% of control at the peak of bursting activity; however, miniature TTX-resistant inhibitory postsynaptic current amplitudes were unaffected. DHO also reduced the initial slope of the intracellular excitatory postsynaptic potential (EPSP) to ∼40% of control. The conductances of pharmacologically isolated IPSPs and EPSPs in high-Ca/high-Mg-containing saline were also reduced by DHO by ∼50%. The extracellular fiber volley amplitude was reduced by 15–20%, suggesting that the decrease in neurotransmission was partly due to a reduction in presynaptic fiber excitability. DHO enhanced a late depolarizing potential that was superimposed on the EPSP and could obscure it. This potential was not blocked by antagonists of NMDA receptors, and blockade of NMDA, mGlu, or GABAAreceptors did not affect burst firing. The late depolarizing component enabled the pyramidal cells to reach spike threshold without changing the actual voltage threshold for firing. We conclude that reduced GABAergic potentials and enhanced E-S coupling are the primary mechanisms underlying the hyperexcitability associated with impaired Na+-K+ pump activity.


Na+-K+ATPases (pumps) are membrane-bound enzymes that maintain [Na+] and [K+] gradients across the cell membrane. Na+-K+ pumps, which are high-affinity binding sites for cardiac glycosides (CGs), are ubiquitous in neurons (Pietrini et al. 1992) and widely distributed in the mammalian brain (Hauger et al. 1985). At each translocation cycle of operation, the Na+-K+ pumps extrude three Na+ ions and import two K+ions and are therefore electrogenic. Pump activity contributes to the cell resting membrane potential, returns [Na+] and [K+] to their resting transmembrane levels after bursts of activity, and has secondary effects on processes involving monovalent cation gradients such as the Na+-Ca2+ exchanger (Blaustein 1993). Pump activity can be regulated by a variety of factors. Pump activity is stimulated by [Na+]i and [K+]o, and can be specifically inhibited by CGs (MacGregor and Walker 1993). Major disruption of Na+-K+ function causes severe alterations in neuronal function, but very little is known about partial modifications of Na+-K+ pump activity. Interest in physiological regulation of these pumps has been considerably increased since the identification of ouabain-like compounds in the brain (Hamlyn et al. 1991). Therefore inhibition of Na+-K+ pump activity by endogenous CGs may have profound effects on neuronal function and excitability.

In a previous study (McCarren and Alger 1987), we showed that partial inhibition of the pump could contribute to epileptiform discharges by increasing a voltage-dependent, Ca2+ conductance, but effects on synaptic transmission were largely unexplored. The present work was undertaken because of the likelihood that synaptic transmission would also be affected.

A role for Na+-K+ pumps in seizure onset in the brain was first suggested by in vivo studies that showed that the potencies of CGs or a variety of metal ions in inhibiting the Na+-K+ pumps were positively correlated with their potencies in producing convulsions in rats (Donaldson et al. 1971). Decreases in Na+-K+ pump activity occur in animal models of epilepsy and in hippocampal tissue from epileptic patients (Brines et al. 1995; Fernandes et al. 1996) and have been proposed as causal factors in myoclonus epilepsy and ragged red fibers (MERRF) disease, a rare inherited form of human epilepsy (McNamara 1994), and could play a role in other neurological disorders, e.g., bipolar illness (Christo and El-Mallakh 1993), spongiform encephalopathy (Renkawek et al. 1992), and Alzheimer's disease (Rose and Valdes 1994). Disruptions of the genes encoding the Na+-K+ pumps are candidate mechanisms for human temporal lobe epilepsy based on their chromosomal locations and the importance of ion homeostasis in maintaining normal excitability (Ferraro et al. 1999).

Na+-K+ pump dysfunction could play a role in regulation of seizure susceptibility by elevating [K+]o and thereby increasing neuronal excitability (McNamara 1994). Application of ouabain can cause a massive efflux of K+ ions and cell depolarization (Haglund and Schwartzkroin 1990), epileptic activity, and ultimately, cell death (Lees and Leong 1994). However, increased [K+]o and seizure initiation by CG administration are not always correlated (Cordingley and Somjen 1978), suggesting that additional factors may be involved. Dihydroouabain, DHO, a low-affinity analogue of ouabain, reversibly causes interictal-like epileptiform activity in CA1 hippocampal population spike recordings that is not associated with large increases in [K+]o (McCarren and Alger 1987). Synaptic potentials have not been examined extensively, and it remains unclear which effects of DHO underlie the epileptiform activity.

The goal of the present study was to determine how partial Na+-K+ pump inhibition in the rat hippocampal slice preparation leads to the onset of burst potential firing. We confirmed that DHO-induced epileptiform activity cannot be accounted for by large increases in [K+]o or by changes in active or passive membrane properties of CA1 pyramidal neurons. DHO-induced hyperexcitability was characterized by enhanced E-S coupling, which was facilitated by reduced synaptic inhibition and an enhanced late depolarizing potential. Although the mechanisms by which DHO enhanced excitability appeared similar to those underlying the hyperexcitability induced by GABAA-receptor antagonists, they may also involve distinct factors.

Much of the work reported in this report was part of the PhD thesis of S. E. Mason.


Hippocampal slice preparation and solutions

Male Sprague-Dawley rats aged 35–70 days (Charles River Laboratories) were deeply sedated with halothane and decapitated in accordance with the guidelines set forth by the Institutional Animal Care and Use Committee of the University of Maryland School of Medicine. The brain was removed and hippocampi were dissected out and cut into 400-μm-thick transverse slices using a Vibratome Series 1000 (Technical Products International) as previously described (McCarren and Alger 1987). Slices were maintained at room temperature (20–22°C) in a holding chamber at the interface of a physiological saline and a humidified 95% O2-5% CO2 mixture. After ≥1 h of incubation a single slice was transferred to a recording chamber (Nicoll and Alger 1981), where it was held between two nylon nets and continuously perfused with oxygenated saline (29–31°C) at a flow rate of 0.5–1 ml/min.

The extracellular physiological saline (ACSF) comprised (in mM) 120 NaCl, 3 KCl, 2 MgSO4, 1 NaH2PO4, 25 NaHCO3, 2.5 CaCl2, and 10 glucose (pH 7.4 when bubbled with 95% O2-5% CO2). In some experiments, the slices in the recording chamber were perfused with a saline in which the concentration of KCl was raised to 5.5 mM (see results). Slices were allowed to adapt to this new saline for ∼45 min before the start of recordings. Responses were recorded in the same slice in control, experimental, and wash solutions. Drugs were bath applied. Stock solutions DHO (10 mM) were made up in distilled water, and DHO was bath applied at a final concentration of 10–20 μM for 15 min. All other drugs were also prepared as concentrated stock solutions and diluted 1:1,000 in saline. Drugs used includedd,l-2-amino-5-phosphonovaleric acid (d, l-APV (50–100 μM), baclofen (5 μM), CGP 35348 (400 μM), CGP 55845 (5 μM), 8-cyclopentyl-1,3-dipropylxanthine (DPCPX, 1 μM), lidocaine (100–300 μM), LY 341495 (100 μM), 2,3-dioxo-6-nitro-1,2,3,4-tetrahydrobenzo[f]quinoxaline-7-sulfonamide (NBQX, 10 μM), picrotoxin (100 μM), and tetrodotoxin (TTX, 0.5–1 μM). CGP 35348, CGP 55845, and LY 341495 were acquired from Tocris Cookson (Blisville, MO); all other drugs and chemicals were purchased from Sigma (St. Louis, MO).

Electrophysiological recordings

Extracellular recordings were carried out in the CA1 hippocampal area with glass microelectrodes broken to a tip diameter of 4–7 μm and filled with 1 M NaCl or with ACSF (5–15 MΩ). Field population responses were elicited using bipolar concentric stimulating electrodes. Stimulation (100–900 μA; 50–100 μs; 0.05 Hz) was delivered in the stratum radiatum. Presynaptic fiber volleys (FVs) and field excitatory postsynaptic potentials (fEPSPs) were recorded from s. radiatum, and population spikes (PS) were recorded from s. pyramidale. Antidromic CA1 population spikes were evoked by stimulation in the alveus. Input/output (I/O) curves were constructed by taking the slopes of the AMPA- or N-methyl-d-aspartate (NMDA)-receptor-mediated fEPSPs obtained at different intensities of stimulation.

Intracellular current-clamp recordings were made from CA1 pyramidal cells using sharp microelectrodes [80–160 MΩ filled with 1–3 M potassium methyl sulfate (KCH3SO4, ICN, Aurora, OH)]. The pH of the recording solution was adjusted to 7.1 with HEPES. Acceptable cells had stable resting potentials of at least −50 mV, total neuron input resistances of 35–80 MΩ, and overshooting action potentials. Most of the recordings were stable for 1–5 h. Action potentials were elicited either by synaptic stimulation or direct ramp depolarization (−100 to 400 pA). Whole cell voltage-clamp recordings were obtained from CA1 pyramidal neurons using the blind method (Blanton et al. 1989). Whole cell patch electrodes (4–6 MΩ in the bath) were filled with (in mM) 35 KCH3SO4, 120 KCl, 2 [1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid] (BAPTA), 0.2 CaCl2, 1 MgATP, and 1 HEPES. The intracellular solution was adjusted to pH 7.3 with KOH. Series resistance was <12 MΩ at the beginning of an experiment, and cells were discarded if series resistance increased to >20 MΩ. The cells were voltage clamped at a potential near the membrane resting potential (−55 to −70 mV). In these experiments, the “liquid junction potential” was part of the electrode “tip potential.” Tip potentials were measured as the voltage difference between the potential measured by the electrode in the bath before and after breaking the electrode tip back 3–5 mm. Measured in this way the tip potentials were ≤12 mV, and inasmuch as the liquid junction potentials should be smaller than this, they were considered to be acceptable, and no corrections were made.

[K+]o was measured with a K+-sensitive microelectrode (K+-ISM). The microelectrodes had tip diameters of 4–7 μm and were silanized by exposing the tips to dimethyldichlorosilane vapor (Sigma) for 15 min, followed by baking for 1 h at 120°C. After cooling, the tips were filled with a liquid ion-exchanger resin made of 1% weight/volume mixture of potassium tetrakis (4-chlorophenyl) borate (Fluka) in 3-nitro-o-xylene (Aldrich), and backfilled with 150–500 mM KCl. The microelectrodes were calibrated at room temperature (∼22°C) before and after the experiment by a series of test solutions containing 1–50 mM KCl with the NaCl concentration adjusted so that total Cl concentration was 123 mM. Electrode sensitivities ranged from 48 to 58 mV for a 10-fold change in [K+]o, and the output voltages were linearly proportional to the log of the potassium ion activity >3 mM, as described by others (Lewis and Schuette 1975). The K+-ISM measurements are reported as concentration changes (Voipio et al. 1994). Electrodes with atypical sensitivities or with significant drift (>1 mV/30 min) were discarded. To reject common DC field potentials and isolate potential changes across the column of ion exchanger due to shifts in potassium ion activity, the K+-ISM recordings in s. pyramidale of CA1 were differential with respect to a nearby field electrode. No appreciable drift in the extracellular DC potential was recorded by this reference field electrode in control or experimental solutions. The rate of K o+ clearance in response to low-frequency trains of afferent stimuli (5 Hz, 10 s) was monitored by measuring the time required for the transient change in [K+]oto return to 50% of control levels following the train (t 1/2).

Data acquisition and analysis

Signals were amplified with either Axoclamp 2A, Axoclamp 2B, or Axopatch 1-C amplifiers (Axon Instruments). In cases of two-channel simultaneous recordings, one channel was filtered at 2 kHz with an 8-pole Bessel filter (Frequency Devices, Haverhill, MA), and both channels were digitized at 10 kHz using a Digidata 1200 A/D converter (Axon Instruments). Spontaneous miniature inhibitory postsynaptic currents (mIPSCs) were filtered at 1 kHz and digitized at 5 kHz. Off-line analysis was carried out using the Clampfit program in pCLAMP 7.0 (Axon Instruments).


Fiber volleys were quantified by their amplitude, which was defined as the difference between the initial positive and negative peaks. Amplitudes of antidromic population spikes were determined by taking the difference between baseline potential before the stimulus artifact and the negative peak of the response. Orthodromic PS amplitude was determined by subtracting the amplitude of the negative peak from the mean amplitude of the two positive peaks. The initial slope of the fEPSP (V/s) was measured within the first millisecond of the response. Unless otherwise stated, the time course of field potentials was expressed as the mean percentage change from the average baseline responses (% of control). The paired-pulse ratio was quantified as the ratio of the slope of the second fEPSP to that of the first. The changes in basal [K+]o, the amplitude of [K+]oelevation induced by a 5-Hz/10-s train, and the half-time for K o+ clearance (t 1/2) from the end of this train were determined from K+-ISM measurements. The voltage deflections recorded with the K+-ISM were converted to [K+]o in mM according to the semi-logarithmic calibration curves of the electrodes (Lewis and Schuette 1975).


E-S coupling reflects the ability of an EPSP to discharge an action potential in the postsynaptic cell. Extracellular E-S coupling was expressed as the ratio of the PS amplitude to fEPSP initial slope. Extracellular E-S curves were constructed from a range of stimulation intensities in control and experimental salines, and were expressed as the normalized PS amplitude versus the normalized fEPSP slope (maximal responses were defined as 1). The E-S curves were fitted with a sigmoı̈dal equation using the SigmaPlot 4.01 software (SPSS, Chicago, IL). The E50 was defined as the value of the normalized fEPSP slope at which the PS amplitude was 50% of its maximal response and was calculated from the following built-in function in SigmaplotE50=exp{[ln[(cb)(((ad)/(0.5d))1)]]/b} where a = max, b = slope,c = inflection point, and d = min. Intracellular recordings were used to assess the cellular mechanisms underlying the E-S coupling changes induced by our experimental conditions. The probability of firing was represented by the ratio of spike-containing EPSP/inhibitory postsynaptic potential (IPSP) sequences to the total number of sampled EPSP/IPSP sequences in a 5-min recording period. To measure changes in firing probability, we used a subthreshold stimulus intensity that brought the cell to within 2–5 mV of threshold during baseline stimulation. Change in probability of firing was further quantified by measuring the minimal EPSP slope and the minimal stimulus intensity required to evoke an action potential.


Data are reported as means ± SE. Most changes induced by the experimental salines are expressed as percent of control value, which is taken to be 100%. Paired comparisons of parameters measured at various time points (2- to 5-min duration) over the course of the experiment were performed on averaged values. Normality and equal-variance tests were performed to select appropriate statistical tests for comparisons (SigmaStat 2.0, SPSS). Data that successfully passed these tests were analyzed using Student's paired or unpairedt-test for paired versus group comparisons, respectively. Nonparametric Wilcoxon and Mann-Whitney tests based on rank comparisons were performed for paired versus group comparisons, respectively, when data failed the normality and/or equal-variance tests. One-factor ANOVAs followed by post hoc Dunnet tests or ANOVAs with repeated measures followed by post hoc Tukey tests were run for multiple-group comparisons. Friedman repeated-measures ANOVA on ranks followed by post hoc Dunnett tests was performed for multiple comparisons when the normality test failed. Cumulative amplitude distributions of mIPSCs constructed in control and experimental saline were compared using the Kolmogorov-Smirnov (K-S) test with a significance level ofP < 0.005. Correlations were calculated using the Pearson product moment correlation test.


Na+-K+ pump inhibition by dihydroouabain induces epileptiform activity without causing large increases in resting [K+]o

In agreement with previous results (McCarren and Alger 1987), we found that a 15-min bath application of DHO, 20 μM, reversibly caused the appearance of robust extracellular epileptiform burst potentials in hippocampal CA1 (Fig.1A1) in response to single stimuli to s. radiatum. The bursts resembled interictal-like burst potentials recorded in other models of epileptiform bursting. Ictal-like seizure activity—prolonged, spontaneous bursts—was never observed; however, brief spontaneous burst discharges that originated in CA3 and spread into CA1 (Korn et al. 1987) often occurred. Surgical removal of the CA3 subfield blocked these spontaneous events, but it did not prevent evoked burst potentials in CA1 (n = 6, data not shown).

Fig. 1.

Dihydroouabain (DHO) induces epileptiform burst firing and delays K Embedded Image clearance. A: DHO-induced hyperexcitability in CA1 hippocampal area. A1: extracellular population spikes recorded from stratum pyramidale in response to single stimulus pulses delivered to s. radiatum. The same stimulus elicited a single population spike (PS1) in control and wash salines and a multiple-spike response (a burst) in DHO. In this and all figures except Fig. 4, 20 μM DHO was used. Stimulus intensity was adjusted in each slice before the start of the recording to elicit a PS1 of ∼50% of maximum. In all figures, positive is up and the stimulus artifacts have been erased for clarity. A2: PS1 amplitudes plotted as percent of control responses against time (n = 29). Symbols represent means ± SE throughout. A3: amplitudes of the second PS (PS2) in the DHO-induced multiple-spike responses (n = 29) show the development and duration of bursting activity. Peak bursting occurred at 15 min of DHO application. B: changes in [K+]o caused by Na+-K+ pump inhibition. [K+]o was measured with K+-ion-sensitive microelectrodes placed in s. pyramidale.B1: sample traces show the response of a K+-sensitive microelectrode (K+-ISM) to a train of stimuli (5 Hz, 10 s, onset indicated by arrows) delivered to s. radiatum in control, DHO, and 60 min after beginning washout of DHO. DHO increased the amplitude of the [K+]otransient and the half-time of K Embedded Image clearance (t 1/2). B2: half-time (t 1/2 ) of K Embedded Image clearance in control saline (1st bar), during bath application of DHO (4, 12, and 15 min in DHO), and in wash saline (8, 15, 30, 45, and 60 min of wash). The wash period starts at time = 0 and lasts 60 min. The time course of changes in t 1/2 follows the development of DHO-induced bursting activity (cf.A3). Data recorded during 1 h of wash are fitted with a single exponential and show that DHO washed off with a time constant of 12 min.

The time course of bursting activity is represented by the presence of a second population spike (PS2) in the burst potential, as shown in Fig. 1 A3. This “bursting period” generally spanned the period from the 10th min of the DHO application to about the 20th min of DHO wash, although in some slices a wash period of >30 min was required for complete reversal of burst activity (n = 7; not shown). Epileptiform activity induced by lower doses of DHO (i.e., 10 μM) usually took longer to develop and was often characterized by a burst of only two population spikes. Thus a 15-min bath application of 20 μM DHO induces robust yet reversible bursting activity, and in all experiments DHO was washed off after 15 min.

To verify that DHO inhibits the Na+-K+ pumps and to determine whether the burst potentials generated by DHO were correlated with an impairment of Na+-K+ pump activity, we monitored the change in the rate of K o+ uptake using K+-ISMs. Impaired K+ transport by the pumps would be reflected by a decreased rate of recovery of [K+]o to baseline after a short stimulus train (Krnjevic and Morris 1975). Figure1 B1 illustrates a typical K+-ISM response in control, DHO and wash solutions, following a train of afferent stimuli (5 Hz for 10 s). The half-time for [K+]o clearance (t 1/2) slowed an average of 403 ± 59% of control during DHO-induced bursting (i.e., from 4.2 to 17.2 s, n = 10, P < 0.01). Moreover, the amplitude of the train-induced [K+]o transient was larger in DHO (5.0 ± 0.8 mM) than in control (2.8 ± 0.3 mM;n = 10, P < 0.01). Both the changes in [K+]o transient amplitude and clearance rate are signs of reduced Na+-K+ pump activity. Na+-K+ pump inhibition by 20 μM DHO did not produce large increases in resting [K+]o. Indeed, the bursting activity induced by DHO was only accompanied by a modest elevation in resting [K+]o (0.8 ± 0.1 mM, n = 10, P < 0.001), an effect that reversed after 25 min of wash. In the example shown in Fig.1 B1, resting [K+]o was only increased by 0.22 mM, while the t 1/2 was prolonged to 221% of control in agreement with previous results.

Successive stimulus trains were applied over the course of the experiment in five slices to evaluate whether the time course of excitability correlates with inhibition of the Na+-K+ pumps. Figure1 B2 shows that inhibition of the Na+-K+ pumps by DHO reverses with time as DHO is washed off. Data recorded during 1 h of wash were fitted with a single exponential, showing that the effect of DHO washed off with a decay-time constant of 12 min. These data suggest that mild inhibition of the Na+-K+ pumps by DHO is sufficient to increase excitability. In the remainder of this report we examine passive and active properties of CA1 pyramidal cells, synaptically activated potentials, and the coupling between synaptic potentials and burst firing, to understand the mechanisms by which partial Na+-K+ pump inhibition increases neuronal excitability.

DHO induces transient shifts in passive membrane properties of CA1 pyramidal neurons

Time courses of the effects of 20 μM DHO on resting membrane potential (V m), total neuron input resistance (R i), and time constant (τ) are presented in Fig.2 A. R i was measured as the input “chord resistance” (calculated from the magnitude of membrane deflections from the resting potential caused by single 100-ms-long current steps of −100 pA). Within 5 min of application, DHO caused a mean transient depolarization of 5.4 ± 1.0 mV (n = 8,P < 0.001; Fig. 2 A1). Surprisingly, the DHO-induced depolarization ended just before the onset of bursting activity. The transitory depolarization was followed by a long-lasting membrane hyperpolarization that persisted throughout the bursting period (Fig. 2 A1). For the 12 cells that provided the data for Fig. 2, the absolute change in V mduring the hyperpolarization was −3.0 ± 0.9 mV below the resting potential after a 15-min application of DHO (P < 0.01), and it further increased to −6.12 ± 1.4 mV 10 min after starting DHO wash (P < 0.01). After 20 min of DHO wash, the membrane potential depolarized back toward control values. Changes in input resistance and time constant followed a similar time course to changes in V m. As shown in Fig. 2 A2, R i was decreased to 73% of control (i.e., −13.0 ± 2.8 MΩ) at 15 min in DHO (n = 12, P < 0.001), when bursting was fully established. Likewise, the time constant (determined by fitting a single exponential to the charging phase of the voltage transient caused by the dc step) was decreased by −6.7 ± 1.3 ms at the time of peak bursting (n = 12, P < 0.001; Fig. 2 A3).

Fig. 2.

DHO-induced changes in passive membrane properties of CA1 pyramidal neurons. A: time courses of the changes in resting membrane potential (V m, A1), membrane input resistance (R i,A2), and time constant (τ, A3) in 12 pyramidal neurons exposed to DHO in a 3-mM [K+]o-containing saline. B: DHO-induced changes in V m are reduced in 5.5 mM [K+]o. B1: relative changes in V m induced by 20 μM DHO in 3 mM [K+]o (○,n = 20) and 5.5 mM [K+]o(●, n = 24). B2: sample traces of simultaneously recorded intracellular (top) and extracellular responses (bottom) in a slice exposed to 20 μM DHO in 5.5 mM [K+]o. The intracellular responses were elicited by subthreshold stimuli delivered to s. radiatum. In the top panel, the intracellular excitatory/inhibitory postsynaptic potential (EPSP/IPSP) sequence recorded in control saline (· · ·) is superimposed on the records in DHO and the wash for comparison. The cellV m is indicated on the leftof each record. Intracellular action potentials in DHO are truncated for clarity. B3: time courses of changes in the extracellular PS2 amplitude (top) and pyramidal cellV m (bottom).

When applying long hyperpolarizing current steps (200–400 ms), we noticed that the delayed membrane potential “sag” (Purpura et al. 1968) caused by repolarizingI h current (originallyI q) (Halliwell and Adams 1982) was much reduced in DHO (see for example, Fig.4 B1). The reduction in membrane potential sag could mean that the I h was inhibited or, on the contrary, persistently activated at rest by DHO. To distinguish between these possibilities, we tested the effects of DHO onR i in the presence of 2 mM Cs, which blocks I h (Halliwell and Adams 1982; Magee 1998), and found that in these conditions, DHO had a significantly reduced effect on R i (P < 0.05,n = 7). These results suggest, but do not prove, that DHO-induced reduction in R i is attributable in part to activation ofI h at resting membrane potential levels where it is normally inactivated. More work must be done to test this hypothesis, however.

In a previous investigation, it was found that bath application of 10 μM DHO in a saline containing 5.4 mM [K+]o was not associated with long-lasting membrane hyperpolarization or a significant decrease in R i (McCarren and Alger 1987). The discrepancy between our present observations and those of McCarren and Alger could depend on the dose of DHO and/or [K+]o. To investigate this issue, we studied the effects of 10–20 μM DHO applications in a saline containing 5.5 mM [K+]o. As shown in Table1, the effects of DHO on the Na+-K+ pumps, as represented by the increases in thet 1/2 for [K+]o recovery, were independent of the concentration of DHO or of [K+]o. In line with the previous report (McCarren and Alger 1987), DHO applied in a 5.5 mM [K+]o-containing saline increased resting [K+]o by <1 mM. However, in 5.5 mM [K+]o, DHO did not hyperpolarize the membrane, but only depolarized it during the bursting period (Fig. 2 B1 and Table 1). Nevertheless, epileptiform bursting was observed in all conditions (see e.g., Fig.2 B, 2 and 3). Hence, although the reason for the dramatic difference in results obtained in 3- and 5.5-mM [K+]o-containing saline is not known, these data show that onset and recovery of extracellular and intracellular burst activity are independent of changes inV m andR i.

View this table:
Table 1.

DHO-induced changes in [K+]o, passive membrane properties, and monosynaptic IPSP properties during the period of hyperexcitability peak

Partial Na+-K+ pump inhibition enhances coupling between fEPSPs and population spikes

We considered that DHO-induced hyperexcitability could be mediated through enhanced coupling of excitatory inputs and firing efficiency (E-S coupling). To assess this, we plotted the E-S coupling ratio (i.e., the amplitude of the first population spike divided by the slope of the fEPSP) in control and DHO-containing saline. Figure3 A shows that the E-S coupling ratio was greater in DHO (1.98 ± 0.3) than in control (0.92 ± 0.03; P < 0.001; n = 29). The onset of enhanced E-S coupling was temporally correlated with bursting activity, and DHO induced a leftward shift of the E-S coupling curves (Fig. 3 B). The normalized fEPSP slope that generated a PS amplitude at 50% of maximum (i.e., theE 50) was significantly smaller in DHO (0.35 ± 0.03) than in control (0.53 ± 0.03) at the time of maximum bursting activity (n = 9; P < 0.01). In principle, enhanced E-S coupling could reflect a decrease in IPSPs, a decrease in the voltage threshold for intracellular spike firing, or an increase in EPSPs, and we investigated each of these possibilities.

Fig. 3.

DHO enhances extracellular field EPSP (fEPSP)-to-spike (E-S) coupling.A: time course of the effects of DHO on extracellular E-S coupling. E-S coupling is represented by the ratio of the amplitude of PS1 (mV) to the fEPSP slope (V/s); n = 29, same slices as in Figs. 1 A and 7B. An increase in the ratio indicates a given population spike is triggered by a smaller fEPSP in the presence of DHO than in control. B: the leftward shift of the E-S coupling curves, indicating that a larger PS1 was produced by a given fEPSP slope (n = 9) in DHO. The amplitudes of PS1 normalized to maximal amplitude are plotted against the fEPSP slope normalized to maximal slope in control (○) and DHO (15 min, ●). A leftward shift indicates enhanced E-S coupling. C: picrotoxin mimics and occludes the effect of DHO on the E-S coupling curve. When 100 μM picrotoxin was applied first, the E-S curve was shifted to the left (▴) of the control curve (○). Subsequent application of DHO produced only a slight rightward shift of the E-S curve (●). D: group data showing that picrotoxin reduced the stimulus intensity required to trigger an intracellular initial spike (P < 0.05, n = 4).

DHO-induced hyperexcitability is associated with a decrease in GABAergic inhibition

E-S coupling is known to be affected by several pharmacological agents. Potentiation of E-S coupling can be blocked by A1 adenosine-receptor antagonists (Fujii et al. 1999) or by the mGluR-selective antagonist (5)-α-methyl-4-carboxyphenylglycine (MCPG) (Breakwell et al. 1996). E-S coupling is enhanced by suppression of GABAergic IPSPs (Chavez-Noriega et al. 1989). We found that DHO-induced enhancement of field potential E-S coupling was not prevented by bath application of the NMDAR antagonist APV (100 μM, n = 3), the mGluR antagonist LY 341495 (100 μM, n = 3), the A1 receptor antagonist DPCPX (1 μM, n = 4), or LY 341495 plus APV (n = 3). Picrotoxin (PTX) induced a large leftward shift in the coupling, which was shown by decrease in theE 50 from 0.49 ± 0.02 to 0.24 ± 0.06 (Fig. 3 C, n = 3,P < 0.05). DHO produced no further shift in E-S coupling (E 50 = 0.24 ± 0.01;n = 3) in the presence of picrotoxin. Figure3 D illustrates that picrotoxin lowered the stimulus intensity threshold for population spike initiation and that DHO did not have a significant effect on spike threshold in the presence of picrotoxin. This suggested that DHO might suppress IPSPs; however, this could not be confirmed at the level of field potential recordings.

We therefore analyzed the effects of DHO on monosynaptic IPSPs recorded in CA1 pyramidal neurons (APV and NBQX present to block ionotropic glutamatergic transmission). As shown in Fig.4 A, DHO greatly suppressed IPSP amplitude [to 24.6 ± 3.2% of control at the end of the 15-min DHO application (n = 6; P < 0.001) following a time course that was temporally correlated with the bursting period (cf. Fig. 1)]. The IPSP amplitude remained depressed in DHO even when the cells were transiently repolarized with DC injection to the control resting membrane potential (not shown), showing that IPSP suppression by DHO was independent of the membrane hyperpolarization. DHO reduced both the early (<100 ms) and late (>100 ms) components of the synaptic response (see traces in Fig.4 A1), implying that Na+-K+ pump inhibition altered both GABAA and GABAB components of IPSP, respectively, although we did not investigate the GABAB component extensively.

Fig. 4.

DHO reduces GABAA -ergic neurotransmission. A: DHO reduces monosynaptic IPSP amplitude in CA1 pyramidal neurons. A1: sample traces of IPSPs recorded in the presence of 10 μM 2,3-dioxo-6-nitro-1,2,3,4-tetrahydrobenzo[f]quinoxaline-7-sulfonamide (NBQX) and 50 μM d,l-2-amino-5-phosphonovaleric acid (APV) in control, 10 μM DHO and wash salines. A2: IPSP amplitude expressed as percent of control amplitude plotted against time (n = 6). Breaks in the time course are due to interruptions made to perform other measurements. B1: monosynaptic IPSPs evoked (stimulus given at ↓) during positive and negative current pulses. IPSP amplitude was reduced in DHO independently of the membrane potential. B2: example of IPSP amplitudes plotted against the membrane potentials measured before the stimulus artifact in control (○), DHO (●), and wash (▵) salines. Linear regression fits shown in the figure were used to calculate the reversal potential of IPSP. Correlation coefficients for the regression lines were >0.9 for each of the 7 cells tested.

A reduction in IPSP size by DHO may be caused by partial blockade of the synaptic conductance (g IPSP) and/or a shift in the IPSP reversal potential (E IPSP). To assess these factors, we constructed I-V relationships for the IPSPs evoked at a range of membrane potentials in seven cells (Fig. 4 B). As illustrated in Figs. 4 B, 1 and 2, DHO reversibly decreased the slope of the IPSP-V relationship and produced a small depolarizing shift in E IPSPafter 15 min of application. We found thatg IPSP was significantly smaller in DHO (11.3 ± 3.07 nS) than in control (20.4 ± 3.6 nS), i.e., it was reduced by 49 ± 5.2% (n = 7;P < 0.001). E IPSPshifted by +4.9 ± 1.06 mV during the bursting period (n = 7; P < 0.01). To examine DHO effects on IPSPs in the absence of marked changes in passive membrane properties, we also studied monosynaptic IPSPs in a saline containing 5.5 mM [K+]o as DHO has minimal effects on membrane properties in this saline (Fig.2 B and Table 1). The DHO-induced reduction of IPSP amplitude was less dramatic in 5.5 mM [K+]o; i.e., 20 μM DHO only reduced IPSP amplitude to 51.7 ± 16.6% of control (vs. 24.6 ± 3.2% of control in 3 mM [K+]o); however,g IPSP, was reduced to ∼50–60% of control in all conditions (see Table 1). Conversely, we found that, in the absence of DHO, exogenously increasing baseline [K+]o from 3 to 5.5 mM shifted E IPSP by +6.5 mV (n = 7) but did not produce bursting activity. Because this represents a larger shift inE IPSP than that induced by DHO in 3 mM [K+]o, a shift inE IPSP cannot explain DHO-induced bursting activity. It appears that the decrease ing IPSP is the most significant factor in DHO-induced burst firing.

DHO could reduce g IPSP either by reducing the amount of inhibitory transmitter released or by modifying the properties of postsynaptic GABAA receptors. Using conventional “blind” whole cell voltage-clamp techniques (Blanton et al. 1989), we found that DHO did not change mIPSC amplitude (42.5 ± 2.8 pA) compared with control (44.4 ± 1.7 pA, n = 5 cells; P > 0.01, K-S test, see Fig. 5 C), suggesting that the 50% reduction of IPSP is not caused by postsynaptic factors. DHO did increase spontaneous mIPSC frequency from 2.35 ± 0.19 to 3.17 ± 0.23 Hz during the bursting period (100–500 events per cell, n = 5, P < 0.05, Fig. 5,A and B), showing that it does affect GABAergic nerve terminals directly.

Fig. 5.

DHO increases miniature IPSC (mIPSC) frequency without affecting mIPSC amplitudes. A: sample traces showing spontaneous mIPSCs recorded under whole cell voltage clamp in control (left) and DHO (right); 10 μM NBQX, 50 μM APV, and 0.5 μM TTX were present in the saline.B: DHO (10 μM) increases mIPSC frequency. Each vertical bar represents the mean number of mIPSCs per min. The horizontal bars indicate the 5-min periods from which data were analyzed. C: Cumulative amplitude distributions for mIPSCs calculated from 100–500 events over 3- to 5-min periods in control (open symbols) and after 15 min of DHO application (filled symbols). Cumulative amplitudes are normalized to the maximal amplitude (100%). Amplitude of mIPSCs was not affected by DHO (n = 5), K-S test, P < 0.005.

DHO does not alter action-potential threshold

DHO might enhance spike firing by decreasing the threshold for action-potential initiation. We measured the action-potential threshold by injecting either rectangular current pulses (“step depolarization,” SD), or current ramps (“ramp depolarization”) into cells in the presence or absence of DHO (Fig.6). We also measured spike amplitudes and half-widths in 15 cells and observed that with either method of measurement spike amplitudes were reduced slightly (∼10%) but significantly (Fig. 6 B, left). Spike properties were measured in 5.5 mM [K+]o to minimize changes in Vm, and when necessary Vm was held constant by small DC injection. Similarly, half-widths were broadened significantly (∼20%, Fig.6 B, middle). These alterations in action potential properties probably reflect a slight increase in [Na+]i caused by the partial Na+-K+ pump inhibition. However, the threshold for spike firing measured with either technique was unchanged in DHO (−48.21 ± 1.95 mV) compared with control (−48.09 ± 1.67 mV; n = 10,P = 0.8, NS, Fig. 6 B, right). Therefore we conclude that DHO does not enhance E-S coupling by decreasing the voltage threshold for action-potential generation.

Fig. 6.

DHO reduces action-potential amplitude without affecting threshold.A, top: examples of intracellular action potentials elicited either with brief rectangular current pulses (inset) or 400-ms-long current ramps.Insets: the action-potential amplitude was decreased and half-width was increased slightly. Action-potential voltage threshold as determined by the ramp currents was not significantly altered.B, histograms summarize the data from 15 cells. Action-potential amplitudes were significantly reduced (P < 0.01), and the half-widths prolonged (P < 0.001), by DHO whether they were induced by rectangular pulses (“step depolarization,” SD) or current ramps (“ramp depolarization,” RD). Action-potential thresholds measured from action potentials induced with either method were not significantly different between control and DHO conditions (P > 0.1).

Partial Na+-K+ pump inhibition also reduces glutamatergic transmission

Suppression of inhibition could explain the DHO-induced hyperexcitability; however, it is also possible that DHO directly increases excitability. Surprisingly, we found that DHO reduced both the intracellular (Fig. 7 A) and extracellular (Fig. 7 B) EPSPs. The initial slope of the intracellular EPSPs began to decline after ∼5 min in DHO and was reduced to 43 ± 4.7% of control (n = 9,P < 0.01) at the end of the DHO application when maximum bursting activity was observed (Fig. 7 A2). The fEPSP slope declined to an average of 69 ± 4.9% of control at the time of bursting (n = 29, Fig. 7 B). Interestingly, the EPSPs did not recover after >20 min of DHO wash, whereas the population spike in the same slices, which represents the ability of the cells to fire action potentials, did recover (see Fig.1 A2). This indicates that the slice preparation and principal cells were not damaged by a neurotoxic effect of DHO application. Thus DHO-induced bursting activity was not temporally correlated with enhanced EPSPs. The actual magnitude of intracellular EPSP suppression is unclear with the mixed EPSP-IPSP responses shown in Fig. 7 A.

Fig. 7.

DHO reduces glutamatergic transmission. A: DHO reduces the slope of the depolarizing component of intracellular EPSP/IPSP sequences. A1: synaptic activation of CA1 pyramidal neurons by a single stimulus delivered to the s. radiatum in control, DHO, and 20 min after switching to the wash. Top andbottom traces show responses evoked by suprathreshold and subthreshold stimuli, respectively. Note again the suppression of IPSPs. A2: time course of changes in EPSP slope (mean % of control). DHO reduced the slope of EPSPs evoked by subthreshold stimulus intensities by 40–50%, suggesting a decrease in glutamatergic transmission (n = 9).B: DHO reduced the initial slope of fEPSPs recorded from s. radiatum simultaneously with the orthodromic PS in the 29 slices shown in Fig. 1 A. Top traces show that the fEPSP slope was reduced in DHO and 20 min after starting the wash from DHO. Control record is superimposed on DHO and wash traces (· · ·). Note that repetitive population spikes are detected after the peak of the fEPSPs in DHO. The plotbelow shows the time course of DHO effects on the fEPSP slope (mean % of control), and illustrates that fEPSP slope was reduced at peak bursting (15 min of DHO application) and did not recover after 20 min in wash saline.

To determine if the DHO effect is exerted on the glutamatergic synapses themselves, we studied isolated EPSPs with intracellular recordings in saline containing 100 μM picrotoxin and a high concentration of divalent cations (4 mM CaCl2 and 4 mM MgSO4; “high-Ca/high-Mg saline”) to block epileptiform firing. The CA3 region was removed from these slices to prevent spontaneous epileptiform activity, which originates in CA3, from invading CA1. Synaptic stimulation in the presence of picrotoxin usually elicits complex depolarizing waveforms, even with high-Ca/high-Mg saline. To attempt to measure the properties of the monosynaptic EPSP without contamination by other conductances, we determined the time of occurrence of the EPSP peak from the differentiated depolarizing waveform and measured the peak at that time in the original traces (light vertical line in Fig.8 A). Somewhat surprisingly, we did not find a consistent effect of DHO on EPSPs in the high-Ca/high-Mg saline: in three cells, g EPSP was reduced by approximately the same extent as wasg IPSP, i.e., to 54.9 ± 10.6% of control (P < 0.05, n = 3). In three other cells, g EPSP was apparently unaffected by DHO, decreasing ≤10% in each case. DHO had no effect on membrane properties of either of the groups. We also checked DHO's effects on IPSPs in high-Ca/high-Mg saline in 5 cells (data not shown) and found that DHO caused the same shift in IPSP reversal potential (of +5.4 ± 1.1 mV; P < 0.01) and decrease in conductance (to 59.7 ± 5.4% of control, open circle;P < 0.05, n = 5), as it did in normal saline (see Table 1). We conclude that DHO can reduce both glutamatergic and GABAergic transmission but that the reduction in gEPSP is not always readily apparent.

Fig. 8.

A late depolarizing potential is present in DHO. A: depolarizations in high-Ca/high-Mg-containing saline plus 100 μM picrotoxin (CON) and then the same saline plus picrotoxin and 20 μM DHO (DHO) at −75 mV; the averages of 10 traces in each condition are shown. The EPSP was measured at the point of the thin vertical line. Note the complex waveform that follows the initial EPSP.B: depolarizing waveforms in control, DHO, and at 6 min of DHO wash. The control trace (dashed line) is overlaid on the DHO and 6-min wash traces on the left. On theright, the DHO and 6-min wash traces are scaled to control. C: stimuli given in the same conditions as inA at a range of membrane potentials. Note relative insensitivity of the peak of the late depolarization to changes in membrane potential both in control and DHO conditions.

Conceivably, DHO-induced bursting could arise simply from a failure of polysynaptic inhibition. Because the Schaffer collaterals make monosynaptic contacts onto pyramidal cells, whereas polysynaptic pathways produce both feedforward and feedback IPSPs, a uniform suppression of polysynaptic conduction could cause a greater relative suppression of IPSPs than of EPSPs. However, neither lidocaine (100–300 μM, 10–15 min, n = 17), NBQX (0.5–2 μM, 15 min, n = 3), the GABAB-selective agonist baclofen (5 μM, 2 min,n = 2), nor perfusion with a low-calcium/high-magnesium saline (1.25 mM Ca2+/4 mM Mg2+, n = 3) produced epileptiform activity. Hence, uniform suppression of synaptic transmission in the slice does not cause hyperexcitability.

DHO increases a late depolarizing potential

A late depolarizing potential (Fig. 8) that is detectable in picrotoxin represents a possible key factor in the development of hyperexcitability produced by partial Na+-K+ pump inhibition. Picrotoxin greatly reduced the stimulus intensity necessary to trigger an initial spike (104.75 ± 8.78 μA) compared with control (195.25 ± 17.78 μA) (Fig. 3 D, n = 4,P < 0.05) and DHO did not decrease action-potential threshold in the presence of picrotoxin. Thus the effects of DHO could be occluded by blocking GABAA-ergic IPSPs with picrotoxin, suggesting again that suppression of IPSPs is the primary mechanism for DHO-induced bursting.

The amplitude of the late depolarizing potential was only decreased to 94.7 ± 11.7% of control by DHO (Fig. 8 B,P > 0.9, NS). To determine if this late potential is a type of conventional EPSP, we attempted to measure its conductance by evoking it at a subthreshold range of membrane potentials but found that its amplitude did not change significantly over this range (Fig.8 C). This could mean either that the late depolarization is not a conductance-increase phenomenon or that it is produced in an electrotonically distant part of the cell, such that our somatically located microelectrode is incapable of affecting it. In either case, we infer that the early onset of this potential obscured the true EPSP in those cases noted in the preceding text in which DHO appeared not to affect g EPSP. The traces in Fig.8 B show an isolated depolarizing waveform in control, DHO, and at 6 min of DHO wash. As is clear in the scaled traces, the duration was significantly increased by DHO from 77.6 ± 2.6 to 90.1 ± 2.8 ms at the time of bursting (P < 0.05,n = 6). Thus although the nature of this potential is unclear, it could be responsible for the apparent imbalance between excitation and inhibition that characterizes the bursting period. On the other hand, DHO had less-pronounced effects on passive cell properties in high-Ca/high-Mg saline. In this saline, plus picrotoxin, DHO failed to generate a transient depolarization of the membrane potential, and the peak long-lasting hyperpolarization was only ∼50% of that in normal saline. This raises the question of whether the late depolarizing potential that is measured in high-Ca/high-Mg saline is actually relevant to the bursting that is seen in normal saline.

To investigate this question, we again studied synaptically evoked responses in normal saline. Application of 20 μM DHO gave rise to burst potentials in response to smaller EPSPs in both extracellular (Fig. 9 A, left) and intracellular (Fig. 9 A, right, B, andC) recordings. The minimal intracellular EPSP slope required to evoke an action potential was decreased from 4.2 ± 0.5 V/s in control to 1.1 ± 0.2 V/s in DHO (n = 15,P < 0.05). The latency for action-potential firing was increased in DHO, suggesting that, despite the reduction in EPSP slope, the late depolarization enabled the cell to reach the voltage threshold for firing (Fig. 9 C). Indeed, the duration of the depolarizing component of the evoked response was dramatically increased in DHO in normal saline (from 15.23 ± 1.70 to 68.86 ± 8.56 ms; n = 3, P < 0.05). The voltage threshold for action potentials induced by synaptic stimulation was unchanged during DHO-induced bursting activity (n = 10, P > 0.05, NS, Fig.9 C, top traces). We first assumed that the late potential was NMDAR dependent because NMDAR-dependent responses are controlled by GABAA-ergic IPSPs. However, the late depolarization was not blocked by 100 μM bath-applied APV (n = 2), suggesting that it is largely non-NMDAR dependent. In any case, we infer that DHO-induced suppression of IPSPs facilitates E-S coupling of CA1 pyramidal neurons by enhancing a late depolarizing potential, which eventually drives the cell past spike threshold despite the smaller size of the EPSP itself. Because of its prolonged duration, the late potential is capable of initiating the multi-action-potential burst.

Fig. 9.

DHO enhances intracellular (E-S) coupling without changing the synaptically evoked action-potential threshold. DHO causes a subthreshold stimulus intensity to become suprathreshold.A: simultaneously recorded extracellular and intracellular responses, in control and DHO. Each group of traces shows a series of responses to single synaptic stimuli that increase in intensity (80–190 μA) from bottom totop. Arrowheads indicate threshold for action-potential initiation. DHO decreased the stimulus intensity required to bring the cell to the threshold for action-potential firing. B: plot of EPSP slope against stimulus intensity in control (○) and DHO (15 min, ●) for the cell shown inA. DHO decreased the EPSP slope at which the pyramidal cell fired action potentials (↑, ↓). C: single events triggered by synaptic activation in control (· · ·) and DHO (—). Responses induced by suprathreshold (top) and subthreshold stimuli (bottom) are shown. The top traces correspond to the traces marked (↑, ↓) in A. The synaptically activated late depolarization triggered action potentials in control and DHO even though the EPSP had a flatter slope in DHO than in control. The voltage threshold for action-potential firing was not changed in DHO (- - -), although the onset of the spike was delayed because of the slow rise of the late depolarization. The bottom traces illustrate the subthreshold late depolarization.

DHO-induced decrease in neurotransmission is associated with reduced presynaptic fiber volley

If, as our results suggest, the primary cause of DHO-induced burst firing is a suppression of synaptic transmission, then it is important to understand how partial inhibition of the Na+-K+ pump suppresses synaptic transmission. It could arise from activation of presynaptic inhibitory receptors (e.g., mGluRs, GABAB) by glutamate or GABA (Davies and Collingridge 1996; Gereau and Conn 1995). However, antagonists of GABAB receptors (CGP 35348 at 400 μM,n = 4, or CGP 55845 at 5 μM, n = 4), metabotropic glutamate receptors (LY 341495 at 100 μM,n = 5), or adenosine A1 receptors (DPCPX at 1 μM, n = 4) all failed to prevent DHO-induced bursting activity and reduction of excitatory transmission. Even concomitant application of LY 341495 and APV (100 μM) did not prevent DHO-induced bursting (n = 3). The paired-pulse facilitation ratio (interpulse intervals of 50 ms) was not affected in DHO (n = 6), which suggests that Na+-K+ pump inhibition by DHO does not significantly alter the presynaptic mechanisms involved in the Ca2+-dependent release of glutamate.

Alternatively, because DHO does affect action potential properties (Fig. 6), it could suppress synaptic potentials by impairing axonal conduction in afferent fibers. We found that DHO induced a consistent decrease in the amplitude of the presynaptic fiber volley (FV), the small biphasic response immediately preceding the fEPSP, whether this was measured in normal saline or in the absence of EPSPs (Fig.10 A, top traces,50 μM APV and 10 μM NBQX present). We determined that DHO also affected conduction in CA1 pyramidal-cell axons by observing that antidromic population spike amplitudes were reduced in DHO (Fig.10 A1, bottom traces). The relative reduction of fiber spike amplitude was independent of stimulus intensity over the range 150–600 μA (n = 9, see insets in Figs. 2, 3, and 10 A).

Fig. 10.

DHO reduces fiber excitability and FV-fEPSP coupling. A: DHO reduces fiber excitability. A1: fiber volleys (FV,top) and antidromic population spikes (PS,bottom) recorded simultaneously, in control (a) and after 10 min in wash saline (b). The control trace is superimposed on the other records for comparison (· · ·). Fiber volleys evoked by single stimuli (∼400–600 μA) were isolated with NBQX and APV. The plots below show the time course of DHO effects on the amplitude of FV (A2) and antidromic PS (A3). Amplitudes are expressed as mean percentages of control (n = 9). During the breaks in the time courses, input-output curves were constructed in control (○) and wash salines (●). The curves, insets, show the changes in normalized FV amplitudes plotted against a range of stimulus intensities (150–600 μA). B: DHO decreases the FV-fEPSP coupling. fEPSPs were evoked by a range of stimulus intensities (100–600 μA) in control (○) and after 10 min in wash saline (●). The slopes of the normalized fEPSPs are plotted against the corresponding FV amplitudes. Each symbol represents the mean FV amplitude ± SE and corresponding mean fEPSP slope ± SE for a given stimulus intensity. Sample traces are shown at the top of B, 1 and 2. Traces on the right show initial fEPSP slope in control and DHO.B1: DHO reduces the FV-fEPSP coupling for fEPSPs mediated by non-NMDA glutamate receptors. APV was present when sample traces above were obtained but not for the data represented in the graph. B2: DHO reduces the FV-fEPSP coupling for EPSPs mediated by NMDA receptors (NBQX present) to a greater extent.

Hence, DHO-induced suppression of excitatory synaptic potentials is associated with impaired axonal conduction. If this was the sole factor in DHO-induced reduction of excitatory transmission, then DHO should not alter the FV-fEPSP coupling because both potentials would be reduced to the same extent. However, when we constructed normalized FV-fEPSP curves in control saline and DHO for a range of stimulus intensities (100–600 μA), we found that DHO induced a downward shift of the FV-fEPSP coupling curve, i.e., fEPSPs associated with a given FV were smaller in DHO than in control saline (Fig. 10 B). For example, the fEPSP slope induced by a 1-mV amplitude FV was reduced by 26% in DHO compared with control (P < 0.001), suggesting a decoupling between presynaptic FV and synaptic transmission. In contrast, lidocaine (100 μM), a local anesthetic that decreased both FV and antidromic population spike amplitudes (76 and 67% of control, respectively, n = 6), produced a very similar reduction (74%) of fEPSP slopes. Thus lidocaine did not decouple FV and fEPSP suppression, and this suggests that DHO has an additional suppressive effect downstream of action potential conduction to the terminals.

If DHO only affected synaptic transmission through presynaptic factors, then both AMPAR-mediated and NMDAR-mediated fEPSPs should be similarly reduced. Surprisingly, DHO induced a greater reduction of the NMDAR component (Fig. 10 B2) than of the initial, AMPAR-mediated, component of the fEPSP (Fig. 10 B1). The NMDAR component (measured in 0.1 mM [Mg2+]o and NBQX) of the fEPSP slope induced by a 1-mV amplitude FV was reduced by 65% in DHO compared with control (n = 6, P < 0.05), compared with a reduction of the initial, AMPAR-dependent component of the EPSP of only 26%. Even high stimulus intensities (to 900 μA), which elicited FV amplitude of ∼2 mV, were associated with NMDAR fEPSP slopes of only 35–40% of maximal control amplitude (Fig.10 B2). Thus DHO suppressed the NMDAR more than the AMPAR component of the fEPSP, suggesting that DHO may suppress glutamatergic transmission by both pre- and postsynaptic mechanisms. We propose that GABAA-ergic IPSPs are also suppressed largely because of interference with conduction in the axonal arborization of interneurons, although because these axons do not form a simple coherent bundle the way that glutamatergic Schaffer collaterals do, we were unable to assess this question directly.


We report that partial inhibition of the Na+-K+ pumps by DHO induces numerous changes in neuronal membrane properties and synaptic function. Increases in [K+]o and alterations of passive membrane properties of pyramidal neurons are not required for DHO-induced onset of interictal-like burst firing in the CA1 hippocampal area. Rather, DHO application enhances EPSP-to-spike coupling by reducing inhibitory neurotransmission and induction or unmasking a prolonged depolarizing potential. Synaptic potentials are suppressed mainly by presynaptic mechanisms, probably by decreasing action-potential invasion of nerve terminals. Our study highlights the importance of the Na+-K+pumps in the regulation of neuronal excitability and suggests that disruption of the pump activity is a candidate mechanism in the development of epileptiform activity.

Specificity and physiological relevance of DHO-induced epileptiform activity

We attribute the effects produced by DHO to its pump- blocking effects: i.e., it decreased the rate of K o+ clearance and increased the amplitude of the [K+]o rise triggered by a short stimulus train (Krnjevic and Morris 1975). DHO also decreased the amplitude of Na+-dependent action potentials and extracellular fiber volleys, which may be expected from the accumulation of [Na]i that follows modest Na+-K+ pump inhibition (Blaustein 1993). Finally, the onset and reversal of DHO-induced bursting activity was temporally correlated with the DHO effects on K o+ uptake rate. Nevertheless, the cardiac glycosides may have other cellular actions besides Na+-K+ pump inhibition (Brosemer 1985; Santana et al. 1998; Thieren and Blostein 2000), and we cannot rule out the contribution of other factors to the effects we observed. The high-specificity binding site for CGs is on the catalytic α subunit of the Na+-K+pump, of which three different isoforms (i.e., α1, α2, and α3) exist in nervous tissue (Sweadner 1989). The α2 and α3 isoforms predominate in neuronal rather than glial cells and have high affinity for CGs in the dose range used in our experiments. Therefore our results can probably be attributed to blockade of the neuronal Na+-K+ pumps, with little involvement of the glial pump. In emphasizing that partial inhibition of Na+-K+ATPases has particularly important effects on GABAergic transmission, our data are in agreement with recent reports. Both fluid percussion injury (Ross and Soltesz 2000) and a brief period of high-frequency stimulation (Ross and Soltesz 2001) produced long-lasting effects on interneurons in the dentate gyrus by causing an inhibition of the interneuronal Na+-K+ pumps. Their results show that interneurons became significantly depolarized and made hyperexcitable by pump inhibition. Probably the differences in experimental protocols is largely responsible for the differences between their results and ours. The observations of Ross and Soltesz were mimicked and occluded by 100 μM strophanthidin, which will inhibit the pumps much more profoundly than the lower concentrations of dihydroouabain that we used.

Impaired Na+-K+ pump activity alters passive membrane properties of CA1 pyramidal neurons

Na+-K+ pump inhibition by DHO application had minor effects on passive membrane properties of pyramidal neurons. DHO induced a transient depolarization of ∼5–6 mV before the bursting activity began, but the onset of bursting was associated with increased membrane conductance and, unexpectedly, a long-lasting hyperpolarization of ∼3–6 mV. The Na+-K+ pump exports three Na+ for each K+ imported into the cell and therefore produced an electrogenic outward, hyperpolarizing, current that contributes to setting the membrane potential. Blocking this current will depolarize the membrane. We calculate that the DHO-induced, transient depolarization (5.4 mV) could be caused by blocking an electrogenic pump current of 87 pA [which would be similar to the pump current measured in other cells (Mitsutoshi et al. 1998; Senatorov et al. 1997; Shimura et al. 1998)], given the input resistance of our cells (62 MΩ). The actual electrogenic effect of partial Na+-K+ pump inhibition in our case is unclear, however. The concentrations of DHO that we used only affect the most ouabain-sensitive pump isoforms, α2 and α3, and not the α1 isoform, which plays the major role in “housekeeping” processes of the cell. Nevertheless, block of an electrogenic current could explain a small depolarization but not the DHO-induced hyperpolarization that predominates when the cells become hyperexcitable. The hyperpolarization was dependent on [K+]o, i.e., it was much reduced in 5.5 mM [K+]o. The larger changes in passive membrane properties in 3 mM [K+]o saline may relate to regulation of Na+-K+pump activity. Increases in [K+]o may antagonize the binding of CGs to the pumps (Gleitz and Peters 1997) or enhance the activity of unblocked pumps (Skou 1991), for example. A detailed study will be required to investigate this issue. DHO did induce burst firing in all experimental conditions, indicating that changes in passive membrane properties of pyramidal neurons were not required for DHO-induced hyperexcitability.

Partial Na+-K+ pump activity alters synaptic responses

DHO-induced evoked burst discharges and population-spike burst responses resemble those induced by GABAA-receptor antagonists. We found that DHO reduced g IPSP and shiftedE IPSP slightly in the depolarizing direction. E IPSP varies as a function of both Cl and HCO3 electrochemical gradients, which, in turn, are controlled by various pumps that couple Cl and HCO 3 transport to that of Na+ or K+(Kaila 1994). Hence, DHO-induced changes inE IPSP probably reflect changes in transmembrane [Na+] and [K+ ] gradients. Decreases in IPSP amplitude did not come about by a reduction in postsynaptic GABAA receptor sensitivity because TTX-insensitive mIPSC amplitudes were not altered. It therefore appears that the DHO-induced decrease in g IPSPis the result of a decrease in action-potential-induced GABA release. The increase in action-potential-independent mIPSC frequency may result from a slight increase in nerve terminal [Ca2+]i caused by pump inhibition (Blaustein 1993) that can induce spontaneous release of neurotransmitters. An increase in spontaneous transmitter release may contribute to a decrease in action-potential-initiated transmitter release. Note that even the nerve terminal, a very confined region that should have a much higher input resistance than the cell soma, should also only depolarize by a few millivolts when the electrogenic pumps are blocked, unless the density of the Na+-K+ pumps is higher in the terminal than elsewhere. If the density is the same, the magnitude of the pump current will diminish in proportion to the membrane surface area. At present there is no reason to think the terminal pump density is greater, but if it is, then the block of the electrogenic current would play a greater role in influencing synaptic transmission.

DHO also reduced evoked EPSPs. We believe that the generalized reduction in synaptic transmission is caused by DHO-induced reduction in the amplitude of action potentials in afferent fibers of s. radiatum. A reduction in the fiber volley could be due to the DHO-induced decrease in individual action potential amplitudes and/or to changes in fiber excitability, but both factors would lead to a decrease in total neurotransmitter output. In addition DHO could have a more-direct effect on transmitter release because it also affected FV-fEPSP coupling, i.e., a given fiber volley was associated with a smaller fEPSP in DHO. Finally, the greater reduction of the NMDAR component of the fEPSP suggested that postsynaptic mechanisms might also be involved in the DHO reduction of glutamatergic neurotransmission. The major point is that enhancement of the EPSP was not a factor in DHO-induced burst firing. Thus despite the clear epileptogenic effects of IPSP suppression, the approximately equivalent suppression of both EPSPs and IPSPs meant that another mechanism was necessary to explain burst initiation by DHO.

Cellular mechanisms of initiation of epileptiform activity in DHO

The occlusion of DHO's enhancement of E-S coupling by picrotoxin suggested that disinhibition was largely responsible for the enhanced coupling. The induction, or unmasking, of a late depolarizing synaptic component by DHO enabled the cell to reach spike threshold even though the EPSP itself was smaller. Although the ionic mechanisms of the late depolarizing potential in DHO are unknown, the potential itself appears similar to depolarizing potentials induced by synaptic stimulation in the presence of picrotoxin (Dingledine and Gjerstad 1980; Dingledine et al. 1986; Wong and Prince 1978). The depolarizing potential in disinhibited neurons (∼150–250 ms) is far longer than that of the normal, fast EPSP. It may consist of two synaptic potentials (i.e., AMPA and NMDA) plus a depolarization triggered intrinsically. Activation of mGluRs can also enhance E-S coupling (Breakwell et al. 1996). However, although APV slightly reduced the duration of the depolarizing potential in DHO, it failed to prevent epileptiform burst discharges, and bath application of the mGluR antagonist, LY341495, plus APV, also failed to prevent DHO-induced E-S coupling and bursting activity. Therefore disinhibition of intrinsic conductances may play a role in multiple action potential generation as well. A previous study suggested that a voltage-dependent Ca2+-dependent conductance participates in DHO-induced burst firing in disinhibited cells (McCarren and Alger 1987). If this conductance is activated by evoked synaptic potentials, it could participate in depolarizing the cell and trigger action potentials.

The Na+-K+ pumps regulate many cellular processes directly and indirectly. Which of these constitute the ultimate cause of the changes in synaptic potentials that we have observed must be investigated in future work.


The work was supported by National Institute of Neurological Disorders and Stroke Grants RCI NS-36612 and RO1 NS-36612 to B. E. Alger (the 2nd grant also provided support for C. Vaillend and M. F. Cuttle). S. E. Mason was supported by the Membrane Training Program at the University of Maryland School of Medicine.

Present addresses: C. Vaillend, Laboratorie de Neurobiologie de l'Apprentissage de la Memoire et de la Communication, CNRS, UMR, 8620, Université de Paris-Sud 91405, Orsay, France; and M. F. Cuttle, School of Biological Sciences, University of Southampton, SO16 7PX, UK.


  • * These authors contributed equally to this work.

  • Address for reprint requests: B. E. Alger, Dept. of Physiology, University of Maryland School of Medicine, 655 W. Baltimore St., Baltimore, MD 21201 (E-mail: balger{at}


View Abstract