Surface charge impact in low-magnesium model of seizure in rat hippocampus

Abstract

Putative mechanisms of induction and maintenance of seizure-like activity (SLA) in the low Mg²⁺ model of seizures are: facilitation of NMDA receptors and decreased surface charge screening near voltage-gated channels. We have estimated the role of such screening in the early stages of SLA development at both physiological and room temperatures. External Ca²⁺ and Mg²⁺ promote a depolarization shift of the sodium channel voltage sensitivity; when examined in hippocampal pyramidal neurons, the effect of Ca²⁺ was 1.4 times stronger than of Mg²⁺. Removing Mg²⁺ from the extracellular solution containing 2 mM Ca²⁺ induced recurrent SLA in hippocampal CA1 pyramidal layer in 67% of slices. Reduction of [Ca²⁺]_o to 1 mM resulted in 100% appearance of recurrent SLA or continuous SLA. Both delay before seizure activity and the inter-SLA time were significantly reduced. Characteristics of seizures evoked in low Mg²⁺/1 mM Ca²⁺/3.5 K⁺ were similar to those obtained in low Mg²⁺/2 Ca²⁺/5mM K⁺, suggesting that reduction of [Ca²⁺]_o to 1 mM is identical to the increase in [K⁺]_o to 5 mM in terms of changes in cellular excitability and seizure threshold. An increase of [Ca²⁺]_o to 3 mM completely abolished SLA generation even in the presence of 5 mM [K⁺]_o. A large variation in the ability of [Ca²⁺]_o to stop epileptic discharges in initial stage of SLA was found. Our results indicate that surface charge of the neuronal membrane plays a crucial role in the initiation of low Mg²⁺-induced seizures. Furthermore, our study suggests that Ca²⁺ and Mg²⁺, through screening of surface charge, have important anti-seizure and antiepileptic properties.

the low Mg²⁺ model of epilepsy was developed several decades ago, and since that time has been widely used as a model to test antiepileptic drugs (Albus et al. 2008; Anderson et al. 1986; Coulter and Lee 1993; Sombati and Delorenzo 1995; Walther et al. 1986). This model of epilepsy has clinical relevance as Mg²⁺ deficits can increase seizure susceptibility to proconvulsant stimuli (Greenberg and Tufts 1934) or even cause seizures in humans (Arnold et al. 1983; Nuytten et al. 1991). There is also evidence that Mg²⁺ concentration in serum and cerebrospinal fluid is lower in patients with generalized tonic-clonic seizures (Afzal et al. 1985; Govil et al. 1981). In addition, intravenously injected Mg²⁺ has an anticonvulsant effect in animal models of epilepsy (Borges and Gucer 1978) and is used to treat seizures, particularly in women with eclampsia (Duley et al. 2010).

The mechanism of induction and maintenance of seizure-like activity (SLA) in the low Mg²⁺ model of seizures has been extensively studied (Albus et al. 2008; Gloveli et al. 1995; Gutierrez et al. 1999; Mody et al. 1987). Generation of SLA in this model depends on facilitation of N-methyl-d-aspartate (NMDA) receptors and on a decrease in the surface charge screening near voltage-activated channels as a result of reduction of positively charged Mg²⁺ ions in the extracellular solution (Heinemann et al. 2006). The seizures in the low Mg²⁺ model of seizures can be stopped during the initial phase (less then 2 h after the first SLA) by application of NMDA receptor blockers or by restoration of extracellular Mg²⁺ concentration. However, during the late phase, neither adding NMDA blockers nor restoration of Mg²⁺ concentration to control levels could suppress epileptic activity (Anderson et al. 1986; Derchansky et al. 2004; Dichter and Pollard 2006). While the role of NMDA receptors in the low Mg²⁺ model of epilepsy is well established, the role of surface charge in this model is not well understood.

Surface charge on the cellular membrane produced by sialic acid, phosphates, charged lipids, charged amino acids, and other hydrophilic residues of channel proteins creates local electrical fields near the channel voltage sensor (Messner et al. 1985; Miller et al. 1983; Roberts and Barchi 1987). The effect of surface charge on the membrane channel can be assessed by changing extracellular concentrations of cations, which produce a screening effect on surface charge. Decreasing extracellular Ca²⁺ or increasing the amount of extracellular negatively charged polysialic acids results in a significant hyperpolarizing shift of activation of voltage-gated Na⁺ channels (VGSC) (Bennett et al. 1997; Frankenhaeuser and Hodgkin 1957; Green and Andersen 1991; Hanck and Sheets 1992; Isaeva et al. 2010). Manipulations with extracellular Mg²⁺ concentrations also affect Na⁺ channel gating properties but to a lesser degree. Ion selectivity of surface charge is not fully understood, although the difference in impact on surface charge between divalent and monovalent cations is well described by the Grahame equation derived from Gouy-Chapman theory (Grahame 1947). The changes in extracellular Mg²⁺ or Ca²⁺ concentrations alter action potential threshold and affect multi-unit activity, spontaneous synaptic activity, spontaneous ectopic discharges, and epileptic activity in in vitro and in vivo models (Feng and Durand 2003; Isaev et al. 2007; Peng et al. 2004).

In the present study, we evaluated the contribution of the effect of lowering Mg²⁺ in the extracellular solution on surface charge in the initial stage of the low Mg²⁺ model of epilepsy. Since variations in [Mg²⁺]_o alter both the activity of NMDA receptors and surface charge, we varied surface charge screening in the low Mg²⁺ model by changing [Ca²⁺]_o. As Ca²⁺ plays an important role in synaptic release, such substitution could not accurately describe the influence of surface charge by Mg²⁺. Also, Mg²⁺ facilitates transmitter release, and, when removed, facilitates presynaptic Ca²⁺ accumulation (Mody et al. 1987). However, considering that synaptic release (Zucker 1993) and release probability (Bollmann et al. 2000; Dodge and Rahamimoff 1967) increases with enhancement of [Ca²⁺]_o in evoked synaptic activity and can facilitate excitatory synaptic transmission of neuronal network (Katz and Miledi 1969; Llinas and Walton 1980), we likely underestimated the effect caused by Mg²⁺ ions. In the present study, we investigated the effect of varying extracellular concentrations of Mg²⁺ and Ca²⁺ on functional properties of VGSC on hippocampal CA1 pyramidal neurons and seizure manifestation in the low Mg²⁺ model of epilepsy.

MATERIALS AND METHODS

Animals.

Wistar rats aged postnatal day 12 were used throughout the study. All experimental procedures were performed in accordance with the guidelines set by the National Institutes of Health for the humane treatment of animals and approved by the Animal Care Committee of Bogomoletz Institute of Physiology.

Extracellular recordings.

Extracellular recordings were performed on temporal lobe slices including neocortical areas (Te2 and Te3), entorhinal cortex, subiculum, and hippocampus (Zhang et al. 1995). Slices were prepared according to the technique described previously (Kopanitsa et al. 2006) with some slight modifications. On the day of the experiment, the rat was deeply anesthetized using isoflurane and decapitated. Cerebellum, frontal lobe region (coronal section), and ventral-lateral areas (sections at the angle 20–30° off the horizontal axis) were removed from the brain. The remaining part of the brain was mounted on the stage of a Vibroslice NVSL (World Precision Instruments, Sarasota, FL) and cut (500 μm) through the hemispheres at an angle of 20–30° of their horizontal planes. All manipulation was performed in freshly prepared ice-cold oxygenated (95% O₂-5% CO₂) artificial cerebrospinal fluid (ACSF) of the following composition (in mM): 126 NaCl, 3.5 KCl, 2.0 CaCl₂, 1.3 MgCl₂, 25 NaHCO₃, 1.2 NaH₂PO₄, and 11 glucose (pH 7.25–7.30). For the experiments, we took three to four slices from the middle portion of the hippocampus. We never observed epileptic activity in the brain slices during normal condition. No behavioral seizures were observed, although EEG/video monitoring was not performed. Brain slices were transferred to an incubation chamber and incubated in an oxygenated ACSF for at least 1 h at room temperature (22–24°C) prior to experiments. For recordings, the slices were transferred to a submersion-type chamber and superfused with oxygenated ACSF at a rate of 2 ml/min (22–24°C). The volume of the recording chamber was 0.8 ml, and liquid covered tested slices by ∼500 μm. Field potentials were recorded from the CA1 pyramidal cell layer with extracellular glass microelectrodes (2–3 MΩ) filled with extracellular solution using a differential amplifier (A-M Systems, Carlsborg, WA). Recordings were digitized at 10 kHz using analog-to-digital converter (NI PCI-6221; National Instruments, Austin, TX) and stored in computer using the WinWCP program (Strathclyde Electrophysiology Software, University of Strathclyde, Glasgow, UK). For each experimental group, no more than two slices per rat were used. In the low magnesium-containing medium, MgCl₂ was omitted from ACSF. Adding 50 μM DAP5 to low magnesium ACSF completely prevented seizure occurrence (n = 7) in accordance with the pivotal role of glutamatergic neurotransmission ictogenesis (data not shown). Application of 50 μM CdCl₂ (Cd²⁺ blocks Ca²⁺ channels thereby not only affecting cellular excitability but also blocking synaptic transmission) completely abolished SLA evoked by low Mg²⁺/1 mM Ca²⁺/3.5 mM K⁺ (n = 5, data not shown), suggesting that this activity depends on activity of calcium channels. Evaluation of the viability of slices, which did not respond with SLA to the application of low Mg/2 mM Ca²⁺ solution, was performed at the end of experiments by application of 100 μM 4-aminopyridine (4AP; potassium channel blocker, data not shown). Only slices responding to application of 4AP with SLA were used in the statistical analysis.

In experiments with temperatures maintained at 33°C, temperature control was performed with the Dual Temperature Controller (TC-144, Warner Instruments).

Intracellular recordings.

Intracellular recordings were performed on pyramidal cells isolated from slices prepared as described above. The hippocampal formation containing CA1 was separated from temporal lobe slices and kept for 30 min in the incubation solution containing (in mM): 150 NaCl, 5 KCl, 1.25 NaH₂PO₄, 26 NaHCO₃, 2 MgCl₂, 2 CaCl₂, and 10 glucose (pH 7.4). Subsequent enzymatic treatment was performed in the same solution with CaCl₂ decreased to 0.9 mM and MgCl₂ to 1 mM containing 0.4 mg/ml of protease (type XXIII) from Aspergillus oryzae for 15 min at 32°C. Then the slices were washed in enzyme-free solution for 15–20 min and transferred into incubation solution. Individual cells were mechanically isolated from the CA1 region of the hippocampus by vibrodissociation method (Vorobjev 1991) in an extracellular solution containing (in mM): 130 NaCl, 5 KCl, 2 CaCl₂, 2 MgCl₂, 20 HEPES, pH 7.4. Patch-clamp recordings in whole cell configuration were performed on pyramidal neurons with small dendritic arborization to minimize space clamp problem. Membrane currents were recorded using 2- to 3-MΩ electrodes pulled from thin-walled 1.5-mm borosilicate glass capillaries (Sutter Instruments) using the Brown-Flaming type puller, P-97 (Sutter Instruments). The intracellular solution contained (in mM): 100 CsF, 20 TEA-C1, 2 MgCl₂, 0.5 CaCl₂, 10 EGTA, 10 HEPES, 2.5 Na₂-ATP, pH 7.2 adjusted with CsOH. The typical external solution consisted of (in mM): 20 NaCl, 95 choline-Cl, 20 CsCl, 2 CaCl₂, 2 MgCl₂, 10 HEPES, 5 TEA-C1, and 25 dextrose, with 400 μM CdCl₂ added to block calcium currents, unless otherwise indicated. Membrane currents were recorded using RK-400 amplifier (Bio-Logic SAS, Claix, France), digitally sampled at 10 kHz and filtered at 3 kHz. Pipette and membrane capacitance were compensated to achieve minimal capacitive transients. Test solutions were applied through gravity-fed tubing; solution exchange in the cell vicinity occurred in less than 20 ms. Cells were always returned to the control solution to determine the extent of recovery. Voltage-gated sodium currents were evoked by a series of depolarizing voltage steps from holding potential of −110 mV or ramp protocols empirically designed to yield the maximum peak sodium current, thereby minimizing channel inactivation (Tombaugh and Somjen 1996). Finally, current-voltage (I-V) curves were corrected by subtracting I-V curves measured in the presence of tetrodotoxin (TTX; 300 nM) used to completely block VGSC Na⁺ conductance g_Na. The voltage dependence of g_Na was computed by dividing the Na current (I_Na) by the driving force for Na⁺ ions, V_m − E_rev. All patch-clamp experiments were performed at room temperature (22–24°C).

Data analysis.

Offline analysis of the recordings was performed using Clampfit (Axon Instruments), Origin 7.0 (OriginLab, Northampton, MA), and GraphPad Prism 5 (GraphPad) software. All the data are represented as means ± SE. One-sample Student's t-tests were used to compare difference scores against zero. Two-sample Student's t-test (2-tailed paired or unpaired, where appropriate) was used to compare population means. Linear regression was used to test correlation between datasets. Proportions of slices with and without SLA in experiments with different extracellular ion concentration were compared using the Fisher exact test. A P value less than 0.05 was considered significant.

Chemicals.

D-(-)-2-amino-5-phosphonopentanoic acid and TTX were obtained from Tocris (Ellisville, MO). All other chemicals were purchased from Sigma (St. Louis, MO).

RESULTS

Changes in [Mg²⁺]_o affect hippocampal VGSC.

In the first set of experiments, we investigated the influence of [Mg²⁺]_o and [Ca²⁺]_o on the amplitude and gating properties of voltage-activated Na⁺ channels specifically for the case of hippocampal pyramidal neurons involved in SLA. In the presence of internal Cs⁺ and external TEA⁺, Cs⁺ and Cd²⁺ step depolarization from −110 mV at 2 mM Ca²⁺ evoked an inward current with a threshold near −50 mV (Fig. 1A). This current was TTX sensitive, peaked between −30 mV and −10 mV, and reversed near +36 mV, the calculated reversal potential (E_rev) for Na⁺. On the basis of its properties, we concluded that this current is mediated by sodium ion influx through VGSC. While step depolarizations are more conventional to record the activity of voltage-gated channels, ramps are very useful for the frequent repetition of I-V plots to monitor changes in both current amplitude and voltage dependence over time (Tombaugh and Somjen 1996). Ramp depolarization was performed from −110 mV to +60 mV over 11 ms (15.5 mV·ms⁻¹). The ramp protocol used in our study was empirically designed to minimize effect of channel inactivation on the peak of Na⁺ current (Fig. 1B). The step and ramp protocols resulted in I-V relationship (Fig. 1C) with similar negative resistance branches. At higher voltages, slight differences were observed mainly due to the channel inactivation. Because of individual I-V experiment duration and more precise determination of I-V curve peak in further experiments, we used ramp protocols for determination of the changes in gating of sodium channels induced by divalent cations.

Fig. 1.

Two voltage clamp protocols for voltage-gated Na⁺ channels result in virtually the same current-voltage (I-V) curves. A: conventional step-protocol. Top: neuron was clamped at −70 mV and depolarized for 20 ms to voltages from −70 to −110 mV. Bottom: corresponding I_Na elicited by this protocol after subtraction of the same traces elicited in the presence of 0.3 μM TTX. B: empirically designed ramp (velocity 15.46 mV·ms⁻¹) from −110 to 60 mV. C: after normalization, both voltage protocols resulted in the quasi-identical I-V curves indicating equivalence of both protocols.

We next conducted a series of experiments to quantify the influence of Mg²⁺ on hippocampal Na⁺ channel activity to compare with well-described effects of Ca²⁺. Since the Mg²⁺ chelating agents have a very similar affinity for Ca²⁺, the comparison of the roles of Mg²⁺ and Ca²⁺ in high concentrations could be used for prediction of their relative effectiveness at low concentrations. We have tested the effect of [Mg²⁺]_o starting from 2 mM on I_Na in the presence of 1 mM [Ca²⁺]_o. Similar experiments were made for [Ca²⁺]_o in 1 mM [Mg²⁺]_o. No consistent shift in the E_rev of I_Na was observed during exposure of different [Ca²⁺]_o and [Mg²⁺]_o. In consistence with previously published data, Mg²⁺ affected voltage dependence of I_Na less than Ca²⁺ (1:1.4) (Hanck and Sheets 1992). An increase in divalent ion concentration to 5 mM resulted in a decrease of I_Na peak by 16.3 ± 1.6% (P < 0.01, n = 9) and 15.5 ± 2.2% (P < 0.01; n = 11) and shift of V_1/2 by 8.6 ± 0.2 mV (P < 0.01, n = 9) and 6.9 ± 0.6 mV (P < 0.01; n = 11) for [Ca²⁺]_o and [Mg²⁺]_o, respectively. Application of 10 mM Ca²⁺ decreased the I_Na peak by 28.00 ± 4.3% (P < 0.01, n = 9) and shifted the V_1/2 by 16.5 ± 1.02 mV (P < 0.01, n = 9); the same concentration of Mg²⁺ decreased I_Na peak by 29.4 ± 3.1% (P < 0.01, n = 11) and shifted V_1/2 by 12.1 ± 0.7 mV (P < 0.01, n = 11). Decreases in [Ca²⁺]_o or [Mg²⁺]_o to 1 mM significantly facilitated the I_Na peak by 13.9 ± 1.9% (P < 0.01, n = 9) or 9.4 ± 1.8% (P < 0.01, n = 11) and shifted half-activation V_1/2 by −6.3 ± 0.5 mV (P < 0.01, n = 9) or −4.2 ± 0.1 mV (P < 0.01, n = 11), correspondingly. Changes in both current amplitude and voltage dependence of I_Na arose simultaneously and reversed upon washout. Also, divalent cations enhanced the previously described block of g_Na (Fig. 2B), which was statistically significant (R² = 0.895, P = 0.054 for Ca²⁺, and R² = 0.875, P = 0.065 for Mg²⁺), with negligible differences between the effects of 10 mM [Ca²⁺]_o and 2 mM [Ca²⁺]_o (less than 2.4%), at potentials corresponding to the I_Na peak (Armstrong and Cota 1991). We did not find any significant difference between the action of Ca²⁺ and Mg²⁺ on I_Na amplitude in the range of tested concentrations and voltages. Removal of Mg²⁺ ions from extracellular solution shifts leftward the half-activation V_1/2 by 2.04 ± 0.6 mV (P < 0.01, n = 11) and facilitates the amplitude of I_Na by 4.5 ± 2.9% (P < 0.01, n = 11) compared with 1 mM Mg²⁺ solution (Fig. 2A). Due to an ill-predictable residual [Mg²⁺]_o after exclusion of Mg²⁺ from extracellular solution, these data were not included into the cumulative data plots in Fig. 2.

Fig. 2.

Changes in [Mg²⁺]_o (on top of constant 1 mM of Ca²⁺) as well as in [Ca²⁺]_o (on top of constant 1 mM of Mg²⁺) affect voltage sensitivity and conductance of I_Na. A: cumulative data for the shift of half-activation (V_1/2) obtained from 9 and 11 individual I-V curves for different [Ca²⁺]_o or [Mg²⁺]_o, correspondingly. Representative traces are demonstrated. Using linear fit of this data, we found that the increase in [Mg²⁺]_o and [Ca²⁺]_o leads to depolarization shift in voltage sensitivity (V_1/2) of I_Na by 16.28 ± 0.82 mV and 22.8 ± 1.5 mV per decade of concentration correspondingly. B: the changes in sodium conductance (g_Na) were calculated for peaks of I_Na as described in the text. Increase in the concentration of divalent cations leads to statistically significant (R² = 0.895, P = 0.054, n = 9 for Ca²⁺ and R² = 0.875, P = 0.065, n = 11 for Mg²⁺), but negligible, blockade of I_Na (less than 5% for 10-fold increase in [Ca²⁺]_o or [Mg²⁺]_o).

Induction and manifestation of low-Mg²⁺-induced epileptiform activity depending on extracellular K⁺ and Ca²⁺ concentrations.

In the next set of studies using the low [Mg²⁺]_o model of seizures on temporal lobe slices, we compared the effect of changing [Ca²⁺]_o concentrations and depolarization caused by increasing [K⁺]_o on induction and patterns of SLA. Removing Mg²⁺ from extracellular solution induced recurrent SLA (rSLA) in hippocampal CA1 pyramidal layer in 67% of slices (10/15). In the remainder of the slices, only increased multiple unit activity was observed (data not shown). Figure 3A represents the example of rSLA which first occurred within 48.3 ± 4.2 min following removal of Mg²⁺ from the extracellular solution. rSLA lasted 3.6 ± 0.9 min and recurred with 15.0 ± 2.1-min intervals. Increasing [K⁺]_o from 3.5 mM to 5 mM in low [Mg²⁺]_o allowed us to record epileptic discharges in all tested slices (n = 14) with significantly decreased delay time between magnesium removal and SLA onset (34.6 ± 4.4 min, P = 0.04, Fig. 4A). Two kinds of SLA were observed in response to application of low Mg²⁺/2 mM Ca²⁺/5 mM K⁺ modified ASCF: rSLA (6/14 slices) and continuous SLA (cSLA; 8/14 slices) (Fig. 3, B1 and B2). rSLA recorded in 5 mM [K⁺]_o was insignificantly longer (P = 0.39) than rSLA recorded in 3.5 mM [K⁺]_o (5.4 ± 2.1 min, Fig. 4B) but had significantly decreased (P < 0.01) intervals between SLA (6.6 ± 0.4 min, Fig. 4C). cSLA lasted at least 90 min, and the frequency of epileptic discharges during this activity did not change with a mean value of 0.4 ± 0.2 Hz.

Fig. 3.

Probability and manifestation of low-magnesium-induced epileptiform activity depend on extracellular potassium and calcium concentrations. Extracellular field potential recordings in hippocampal CA1 pyramidal layer in rat temporal lobe slices. Depletion of Mg²⁺ in the extracellular solution induced recurring seizure-like activity (SLA) in the presence of 3.5 mM [K⁺]_o/2 mM [Ca²⁺]_o (A) and presence of 5 mM [K⁺]_o/2 mM [Ca²⁺]_o (B1) and continuous SLA in the presence of 5 mM [K⁺]_o/2 mM [Ca²⁺]_o (B2). Examples of recurrent (rec) SLA (rSLA) and continuous (cont) SLA (cSLA) are outlined with boxes shown on an expanded scale below. Enlargements of different phases of rSLA: initial-bursting activity (a), tonic (b), and clonic (c) activity shown in A and B1 (3rd row). C: cumulative histogram shows probability of low magnesium to induce rSLA and cSLA in CA1 pyramidal layer in the presence of 3.5 mM [K⁺]_o (white) and of 5 mM [K⁺]_o (gray). Cumulative data from the records in the presence of 2 mM [Ca²⁺]_o (open columns) and 1 mM [Ca²⁺]_o (hatched columns). Number of slices used for analysis shown in parenthesis.

Fig. 4.

Different characteristics of low magnesium induced SLA depending on lowering calcium and increasing potassium in extracellular solution. Delay time of SLA (A), duration (B), and recurring time (C) of rSLA, and relative duration of tonic phase during rSLA (D) in the presence of 3.5 mM [K⁺]_o (white) and of 5 mM [K⁺]_o (gray). Cumulative data from the records in the presence of 2 mM [Ca²⁺]_o (columns without hatches) and 1 mM [Ca²⁺]_o (hatched columns).

In the next set of experiments, we decreased [Ca²⁺]_o to 1 mM. In all tested slices (n = 14), perfusion with low Mg²⁺/1 mM Ca²⁺ ASCF in the presence of 3.5 mM [K⁺]_o evoked SLA with the similar value of rSLA and cSLA as was shown for experiments using low Mg²⁺/2 mM Ca²⁺/5 mM K⁺ modified ACSF (Fig. 3C). Comparative analysis between SLA induced by low Mg²⁺/2 mM Ca²⁺/3.5 mM K⁺ and low Mg²⁺/1 mM Ca²⁺/3.5 mM K⁺ saline demonstrated that onset of SLA recorded in 1 mM Ca²⁺ solution was shorter (26.9 ± 2.7 min vs. 48.3 ± 4.2, P < 0.01), rSLA was longer (6.2 ± 0.9 min; P = 0.086), and recurring intervals of rSLA were shorter (5.5 ± 0.6 vs. 15.0 ± 2.1 min; P = 0.002) (Fig. 4). The frequency of discharges during cSLA recorded in 1 mM Ca²⁺ modified solution was 0.3 ± 0.1 Hz. Increasing [K⁺]_o concentration to 5 mM in low Mg²⁺/1 mM Ca²⁺ did not lead to further alteration of the onset, duration, and interburst intervals of SLA (Fig. 4). The only characteristic of rSLA induced in low Mg²⁺/1 mM Ca²⁺/5 mM K⁺ ACSF, which was significantly different from other experimental groups presented in Fig. 4, was the relative duration of the tonic-like phase during rSLA. The duration of the tonic-like phase during rSLA was significantly longer in low Mg²⁺/1 mM Ca²⁺/5 mM K⁺ than in low Mg²⁺/1 mM Ca²⁺/3.5 mM K⁺ modified ACSF (23.4 ± 3% vs. 10.7 ± 2.6%; P = 0.01) (Fig. 4D).

Next we examined the probability of inducing SLA in low Mg²⁺/3 mM Ca²⁺/3.5 mM K⁺ modified ACSF. Three-hour exposure to this solution did not lead to SLA appearance in any of the tested slices (n = 8). Application of low Mg²⁺/3 mM Ca²⁺/5 mM K⁺ also did not lead to epileptic activity (n = 6).

These experiments demonstrate that the incidence, duration, and frequency of SLA in the low [Mg²⁺]_o model of epilepsy is highly dependent on the concentrations of [K⁺]_o and [Ca²⁺]_o.

Suppression of low-Mg²⁺-induced seizure by increasing extracellular Ca²⁺ concentration.

As demonstrated above, incubation of slices with low Mg²⁺/1 mM Ca²⁺ ACSF always led to SLA. In the next set of experiments, we attempted to estimate the effective [Ca²⁺]_o required to stop SLA. Slices were perfused with low Mg²⁺/1 mM Ca²⁺/3.5 mM K⁺ ACSF, and 1 h after SLA onset we increased [Ca²⁺]_o to completely block seizure activity (Fig. 5A). In 7 of 19 slices (36.8%), we were able to stop SLA after increasing [Ca²⁺]_o to 2 mM; in 6 of 19 (31.6%) slices, SLA was abolished after increasing [Ca²⁺]_o to 3 mM; and in 4 of 19 slices (21.1%), we were able to stop SLA only after increasing [Ca²⁺]_o to 4 mM (Fig. 5C). SLA in the remaining slices (n = 2) were resistant even to application of 5 mM [Ca²⁺]_o. SLA completely recovered after decreasing the extracellular concentration of Ca²⁺ to 1 mM (n = 6, Fig. 5B).

Fig. 5.

Propensity to stop rSLA by increasing extracellular calcium concentration. A: extracellular field potential recorded from the CA1 pyramidal cell layer simultaneously obtained from two temporal lobe slices (top and bottom). rSLA was induced by low Mg²⁺/3 mM Ca²⁺/3.5 mM K⁺ ACSF. B: examples of rSLA outlined with boxes in A are shown on expanded scales. C: cumulative histogram shows probability of different extracellular calcium concentrations to abolish rSLA induced by low Mg²⁺/3.5 mM K⁺ ACSF.

To assist the ability of Ca²⁺ to suppress seizures in physiological temperatures, we performed experiments at 33°C (n = 10). SLA occurred after 19.1 ± 1.7 min of low Mg²⁺/1 mM Ca²⁺ ACSF application. [Ca²⁺]_o, which is needed to suppress SLA in 33°C, was distributed as follows: 1) in two slices (20%) SLA were abolished by increasing to 2 mM; 2) in four slices (40%) by 3 mM; 3), in three slices (30%) by 4 mM; and 4) in one slice even 5 mM [Ca²⁺]_o was not enough to stop seizures.

DISCUSSION

The role of NMDA receptors in the low Mg²⁺ model of epilepsy is well described. Increased NMDA currents appear to be a critical trigger of seizure susceptibility in this model (DeLorenzo et al. 1998; Traub et al. 1994). However, Mg²⁺ ions have a potent ability to screen membrane surface charge, and a decrease in this screening could facilitate activation of the voltage-gated channels.

Divalent cations have been reported to cause a voltage-dependent block of Na⁺ channels (Armstrong and Cota 1991). Indeed, we have found that a tenfold increase in [Ca²⁺]_o or [Mg²⁺]_o results in a statistically significant, but virtually negligible, block of I_Na peak by ∼5%. Thus, in the model of pyramidal neurons, we fully support the data by Armstrong and Cota (1991) that a direct block of Na⁺ channels by divalent cations at physiological voltages has no substantial role in I_Na alteration, at least when tested at the concentrations used here.

Application of low Mg²⁺/2 mM Ca²⁺ solution evoked rSLA in 2/3 slices tested. Increasing [K⁺]_o from 3.5 to 5 mM, a concentration that is conventionally used with low [Mg²⁺]_o model of seizures on the brain slices (Dreier and Heinemann 1990; Jones and Heinemann 1987; Khosravani et al. 2005; Pohl et al. 1992), significantly increased seizure occurrence (P = 0.025, exact Fisher test). Such an alteration of [K⁺]_o decreases neuronal resting membrane potential by ∼8 mV (Isaev et al. 2007) and facilitates seizure occurrence (Isaev et al. 2005). Two kinds of seizure activity manifestation were observed at increased [K⁺]_o: cSLA and rSLA, which is in agreement with the data obtained previously (Dreier and Heinemann 1991; Yaari et al. 1986). The most prominent difference in rSLA recorded in 3.5 mM and 5 mM [K⁺]_o was in the interval between rSLA but not rSLA duration. It is interesting that a similar observation on the low [Mg²⁺]_o model of seizures was made with lowering of pH (Velísek et al. 1994). rSLA duration in this model may depend on intrinsic properties of the network rather than on other factors.

We found only a slight difference between results obtained in the experiments with increasing [K⁺]_o to 5 mM and lowering [Ca²⁺]_o to 1 mM. A common method to induce a low [Ca²⁺]_o model of seizure activity in a hippocampal slice is to perfuse it with ACSF containing a concentration of Ca²⁺ below 0.5 mM. It was shown that non-synaptic mechanisms play an important role in the generation of SLA in this model (Feng and Durand 2003; Haas and Jefferys 1984; Taylor and Dudek 1984). In our study, blockade of voltage-gated Ca²⁺ channels by CdCl₂ stopped SLA induced by perfusion with 1 mM [Ca²⁺]_o (see materials and methods). However, saturating concentration of CdCl₂ had no effect on bursting in the low Ca²⁺ model of seizures (Bikson et al. 1999). These results show that SLA induced in low Mg²⁺/1 mM Ca²⁺ ACSF is induced by lowering Mg²⁺ rather than lowering Ca²⁺.

In our patch-clamp experiments, the shift in half-activation of I_Na between 2 mM and 1 mM [Ca²⁺]_o was approximately −6.3 mV. This hyperpolarizing shift leads to a decrease in the threshold of spike generation as well as depolarization of neuronal membrane by increasing [K⁺]_o, produces, in terms of decreased threshold, similar effect. The observed slight difference between 1 mM [Ca²⁺]_o and 5 mM [K⁺]_o in the SLA onset (P = 0.14) and interburst intervals of rSLA (P = 0.13) tentatively can be explained by an increased inhibitory influence of GABA(A) currents with cellular depolarization (Isaev et al. 2005, 2007). Alternatively, high [K⁺]_o increases synaptic transmitter release, and lowering of [Ca²⁺]_o has an effect on the balance between excitation and inhibition (Jones and Heinemann 1987). At the same time, an increase of [Ca²⁺]_o to 3 mM resulted in positive shift of the Na⁺ channel activation and required a stronger cellular depolarization for seizure generation. In this condition, we were not able to evoke SLA by low Mg²⁺ ACSF with 3.5 mM or 5mM [K⁺]_o. Additionally, differences between increased [K⁺]_o and decreased [Ca²⁺]_o could be explained by direct influences of K⁺ and Ca²⁺ on synaptic transmission. Stimulus-induced synaptic potentials decrease with removal of Ca²⁺ from the extracellular membrane (Rausche et al. 1990). The [Ca²⁺]_o at which synaptic transmission is blocked increased with increasing [Mg²⁺]_o and decreased when [K⁺]_o was changed from 3 to 5 mM both in CA1 and the dentate gyrus. However, with an elevation of [K⁺]_o from 5 to 8 mM, synaptic transmission increased in CA1, but was decreased in the dentate gyrus (Rausche et al. 1990).

Cellular depolarization by [K⁺]_o in solution containing 1 mM [Ca²⁺]_o did not lead to changes in either the delay of the SLA onset (P = 0.84) or intervals between rSLA (P = 0.4). This result is in agreement with previously obtained data by Pan and Stringer (1997) where the authors found a 100% probability of inducing cellular bursting in 1 mM [Ca²⁺]_o and ≥4 mM [K⁺]_o solution. The only significant difference in rSLA manifestation in this condition was a ratio of tonic activity to rSLA. In previous studies, we reported that oversialylation of cellular membrane induced by blockade of neuraminidase, a specific enzyme that cleaves extracellular polysialic acids, with the specific blocker NADNA (N-acetyl-2,3-dehydro-2-deoxyneuraminic acid), significantly increases the duration of the tonic phase of SLA induced by lowering Mg²⁺ in hippocampal slice culture (Isaeva et al. 2010). The tonic phase of SLA is associated with depolarization of both the dendritic and somatic areas of the cell, whereas the clonic phase is associated with synchronized rhythmic burst generation mediated by the activation of slow spikes in the dendrites (reviewed by McCormick and Contreras 2001). We propose that the combined effect of cellular depolarization and shift of Na⁺ channel activation to more negative voltages results in synchronization of dendritic and somatic discharges.

We found that after a 1-h application of low Mg²⁺ ACSF in ∼30% of slices in room temperature and 40% in 33°C, SLA became persistent even to 3 mM [Ca²⁺]_o. We suggest that mechanisms involved in transition from initial seizures to the late stage (Dichter and Pollard 2006) take part from the onset of the seizure discharges.

Our data suggest that both removal of Mg²⁺ block of NMDA channels and a decrease in the surface charge screening are required conditions to induce seizures in the low Mg²⁺ model of epilepsy. Both cellular depolarization and hyperpolarizing shift of the voltage-gated channel activation cause similar effects on neuronal network activity due to reduced voltage difference between the resting membrane potential and potentials where Na⁺ channels are activated. Sole change in either the resting membrane potential or in the surface charge affect delay time between application of low Mg²⁺ solution and SLA, frequency of rSLA and manifestation of seizure activity, but not duration of rSLA.

These observations provide evidence that Mg²⁺ can reduce cerebral excitability by screening surface charge and support the usefulness of Mg²⁺ as an anti-seizure and anti-epileptic agent. This study further suggests that surface charge may represent a novel therapeutic target in the treatment of epilepsy.

GRANTS

This work was supported by the Science and Technology Center in Ukraine Grant #5214, National Institute of Neurological Disorders and Stroke Grants NS-041595 and NS-073083, and Christopher Donalty and Kyle Coggins Memorial Grant from Citizens United for Research in Epilepsy (CURE).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: D.I., V.K., O.K., G.L.H., and O.M. conception and design of research; D.I., G.I., E.I., and A.S. performed experiments; D.I., E.I., and O.M. analyzed data; D.I., O.K., G.L.H., and O.M. interpreted results of experiments; D.I. and O.M. drafted manuscript; D.I., O.K., and G.L.H. edited and revised manuscript; D.I., O.K., and G.L.H. approved final version of manuscript; E.I. and O.M. prepared figures.