Levetiracetam (LEV) is an antiepileptic drug with a unique but as yet not fully resolved mechanism of action. Therefore, by use of a simplified rat-isolated nerve-bouton preparation, we have investigated how LEV modulates glutamatergic transmission from mossy fiber terminals to hippocampal CA3 neurons. Action potential–evoked excitatory postsynaptic currents (eEPSCs) were recorded using a conventional whole-cell patch-clamp recording configuration in voltage-clamp mode. The antiepileptic drug phenytoin decreased glutamatergic eEPSCs in a concentration-dependent fashion by inhibiting voltage-dependent Na+ and Ca2+ channel currents. In contrast, LEV had no effect on eEPSCs or voltage-dependent Na+ or Ca2+ channel currents. Activation of presynaptic GABA type A (GABAA) receptors by muscimol induced presynaptic inhibition of eEPSCs, resulting from depolarization block. Low concentrations of Zn2+, which had no effect on eEPSCs, voltage-dependent Na+ or Ca2+ channel currents, or glutamate receptor–mediated whole cell currents, reduced the muscimol-induced presynaptic inhibition. LEV applied in the continuous presence of 1 µM muscimol and 1 µM Zn2+ reversed this Zn2+ modulation on eEPSCs. The antagonizing effect of LEV on Zn2+-induced presynaptic GABAA receptor inhibition was also observed with the Zn2+ chelators Ca-EDTA and RhodZin-3. Our results clearly show that LEV removes the Zn2+-induced suppression of GABAA-mediated presynaptic inhibition, resulting in a presynaptic decrease in glutamate-mediated excitatory transmission. Our results provide a novel mechanism by which LEV may inhibit neuronal activity.
The frequent persistence of seizures in patients with epilepsy treated with different dosing regimens of conventional antiepileptic drugs (Perucca et al., 2000, 2001) necessitates the development of better therapeutics. This may derive from an increased understanding of how current antiepileptic drugs work. Levetiracetam (LEV) is one such antiepileptic drug, used for the treatment of refractory partial epilepsy (Klitgaard, 2001). The precise mechanisms of its anticonvulsant effects are unclear. LEV does not directly inhibit glutamate receptors or voltage-dependent Na+ channels, nor does it potentiate GABA type A (GABAA) receptors; thus, its clinical effects cannot be attributed to any of the three common mechanisms of anticonvulsant drugs (Birnstiel et al., 1997; Klitgaard, 2001). Variable effects on synaptic transmission at hippocampal synapses have been reported, ranging from nil to modest effects on low-frequency responses (Birnstiel et al., 1997; Lee et al., 2009), through to a gradual decrease in the amplitude of excitatory potentials evoked by long stimulus trains (Meehan et al., 2011).
The present study was designed to further clarify the effects of LEV on excitatory transmission at synapses between mossy fibers and CA3 neurons. Aberrant sprouting of mossy fibers is a key feature of temporal lobe epilepsy and results in recurrent excitation of dentate granule cells and additional innervation of CA3 neurons (Represa and Ben-Ari, 1992; Suzuki et al., 1997; Houser et al., 2012). The mossy fiber–CA3 synapses examined here are important model synapses for the enhanced excitability of the hippocampus in temporary lobe epilepsy and are suitable targets for investigating actions of anticonvulsant medications. We focused on how anticonvulsants interacted with physiologic modulators, whose effects can be difficult to quantify with in vitro slice preparations and are frequently not specifically investigated. Zn2+ is an endogenous modulator of neuronal excitability that is at particularly high levels at mossy fiber synapses, with released Zn2+ able to modulate both glutamate and GABAA receptors (Sensi et al., 2009). Both Zn2+-containing mossy fiber terminals and GABAergic transmission show significant changes in the epileptic brain (Coulter, 1999, 2000; Rigo et al., 2002), making them important neuromodulators to consider when investigating anticonvulsant drug actions.
We have established a method for isolation of single nerve cells with functional synaptic-boutons adherent, using an enzyme-free mechanical dissociation procedure (Akaike et al., 2002; Akaike and Moorhouse, 2003). This “synaptic bouton” preparation has the advantages of examining modulation of synaptic transmission in single neurons isolated from surrounding neurons and glia, enabling us to better evaluate the locus of action of various drugs at the single synapse level. Furthermore, we can simultaneously quantify presynaptic and postsynaptic aspects of neurotransmission by measuring the synaptic current amplitude, the synaptic transmission failure rate (Rf), and the paired-pulse ratio (PPR) of evoked excitatory postsynaptic currents (eEPSCs).
We used this synaptic bouton preparation to examine the mechanism of action of LEV, comparing this to another well-studied antiepileptic drug, phenytoin. We report a novel site of LEV action involving inhibition of Zn2+-dependent GABA-induced presynaptic modulation. LEV had a similar effect on GABAA receptor modulation as obtained with the Zn2+ chelators Ca-EDTA and RhodZin-3.
Materials and Methods
Mechanical Dissociation of Hippocampal CA3 Neurons
All experiments were performed in accordance with the Guiding Principles for Care and Use of Animals in the Field of Physiologic Sciences of The Physiologic Society of Japan and were approved by the ethics committee of Kumamoto Health Science University. Wistar rats (11−26 days old, both male and female) were decapitated under pentobarbital anesthesia (50 mg/kg i.p.). The brain was quickly removed and immersed in an ice-cold incubation medium (see below) saturated with 95% O2 and 5% CO2. The hippocampal slices at a thickness of 400 µm were prepared using a vibrating microtome (VR 1200S; Leica, Nussloch, Germany) and then incubated in medium oxygenated with 95% O2 and 5% CO2 at room temperature (21−24°C) for at least 1 hour before mechanical dissociation. For mechanical dissociation, slices were transferred into a 35-mm culture dish (Primaria 3801; Becton Dickinson, Rutherford, NJ) containing the standard external solution (see below), and the CA3 region was identified under a binocular microscope (SMZ345; Nikon, Tokyo, Japan). Details of the mechanical dissociation procedure have been described previously (Akaike et al., 2002; Akaike and Moorhouse, 2003). In brief, mechanical dissociation was accomplished with a fire-polished glass pipette coupled to a vibration device (S1-10 cell isolator; K.T. Laboratories, Tokyo, Japan). The tip of the glass pipette was lightly placed on the surface of the CA3 region and vibrated horizontally (0.2−2.0 mm displacement) at ~50 Hz. Thereafter, the slices were removed from the dish and the mechanically dissociated neurons were left to settle and adhere to the bottom of the dish for at least 15 minutes.
All recordings were obtained from the soma of CA3 pyramidal neurons using conventional whole-cell patch-clamp recordings in voltage-clamp mode. Glutamatergic eEPSCs were recorded at a holding potential (VH) of −65 mV, while voltage-dependent Na+ channel currents (INa) and high-threshold Ca2+ channel currents (IBa) were recorded at VH values of −70 mV and −60 mV, respectively (Multiclamp 700B; Molecular Devices, Sunnyvale, CA) (Wakita et al., 2012). All experiments were performed at room temperature (21−24°C).
Patch pipettes were made from borosilicate capillary glass by a vertical pipette puller (PC-10; Narishige, Tokyo, Japan). The resistance of the recording pipettes filled with the internal (patch pipette) solution (see below) was 3–6 MΩ. Isolated neurons were observed under phase contrast on an inverted microscope (Diapot; Nikon, Tokyo, Japan). Current and voltage were continuously monitored on an oscilloscope (DCS-7040; Kenwood, Melrose, MA). All membrane currents were filtered at 3 kHz (E-3201A Decade Filter; NF Electronic Instruments, Tokyo, Japan), and stored on a computer using pCLAMP 10.2 (Axon Instruments, Foster City, CA). Five-millivolt hyperpolarizing step pulses (30-millisecond duration) were used to monitor the access resistance, and if this changed by more than 20%, the recordings were rejected.
Paired-Pulse Focal Electrical Stimulation of Single Glutamatergic Boutons Using θ Glass Pipettes
Focal electrical stimulation of a single bouton adherent to mechanically dissociated central nervous system neurons has been described previously (Akaike et al., 2002; Akaike and Moorhouse, 2003). The stimulating pipette was made from a glass tube and filled with the standard external solution (see below). The stimulating electrode was placed as close as possible to the soma of a single CA3 neuron from which a whole-cell recording had been successfully obtained. The stimulating pipette was then carefully moved along the surface membrane of the soma and proximal dendrites, while applying stimulus pulses and monitoring for an evoked response. Responses were evoked using a paired-pulse stimulus protocol, with each stimulus pulse of 100-µs duration and of the same intensity, applied at a frequency of 0.2 Hz, using a stimulus isolator (SS-202 J; Nihon Koden, Tokyo, Japan). The stimulus intensity was 0.05−0.09 mA, and the interstimulus interval was 20 or 30 milliseconds. As the stimulation pipette was moved, glutamate-gated synaptic inward currents (eEPSCs) appeared in an all-or-none fashion, indicating that the stimulating pipette was positioned just above a single glutamatergic bouton.
The ionic composition of the incubation medium was 124 mM NaCl, 5 mM KCl, 1.2 mM KH2PO4, 24 mM NaHCO3, 2.4 mM CaCl2, 1.3 mM MgSO4, and 10 mM glucose saturated with 95% O2 and 5% CO2. The pH was adjusted to 7.4. The standard external solution used for recordings contained 150 mM NaCl, 5 mM KCl, 2 mM CaCl2, 1 mM MgCl2, 10 mM glucose, and 10 mM HEPES. The modified external solutions for recording INa contained 60 mM NaCl, 100 mM choline-Cl, 10 mM CsCl, 10 mM glucose, 0.01 mM LaCl3, 5 mM tetraethylammonium (TEA)-Cl, and 10 mM HEPES, and for IBa contained 145 mM choline-Cl, 5 mM CsCl, 5 mM BaCl2, 1 mM MgCl2, 10 mM glucose, and 10 mM HEPES. All external solutions were adjusted to a pH of 7.4 using Tris-base. The composition of the internal pipette solution for glutamatergic EPSCs was 5 mM CsCl, 135 mM CsF, 5 mM TEA-Cl, 2 mM EGTA, 10 mM HEPES, and 5 mM 2-([2,6-dimethylphenyl]amino)-N,N,N-triethyl-2-oxoethanaminium (QX-314) bromide. The internal pipette solution for INa measurements was 105 mM CsF, 30 mM NaF, 5 mM CsCl, 5 mM TEA-Cl, 2 mM EGTA, 10 mM HEPES, and 2 mM ATP-magnesium salt (ATP-Mg), and for IBa, the solution was 80 mM Cs-methanesulfonate, 60 mM CsCl, 5 mM TEA-Cl, 2 mM EGTA, 10 mM HEPES, and 2 mM ATP-Mg. All pipette (internal) solutions were adjusted to pH 7.2 with Tris-base. ATP-Mg was dissolved in the internal solution just before use. Glutamatergic eEPSCs were isolated from GABAergic eIPSCs by recording at a VH of −65 mV, which is close to the Cl− equilibrium potential (ECl) (Wakita et al., 2012).
6-Cyano-7-nitroquinoxaline-2,3-dione (CNQX) and (S)-AMPA were purchased from Tocris Cookson (Ellisville, MO). Muscimol, phenytoin, N-methyl-d-aspartate (NMDA), Ca-EDTA, TEA-Cl, EGTA, ATP-Mg, and 4-aminopyridine (4AP) were purchased from Sigma-Aldrich (St. Louis, MO). RhodZin-3 was purchased from Invitrogen (Eugene, OR). LEV was purchased from LKT Laboratories (St. Paul, MN). All test solutions containing drugs were applied by a “Y-tube system” allowing rapid solution exchange, within ~20 milliseconds.
Glutamatergic eEPSCs evoked by paired-pulse focal electrical stimuli were counted and analyzed in preset epochs before, during, and after each test condition using the pCLAMP 10.2 (Axon Instruments). The amplitude, Rf, and PPR of eEPSCs were analyzed with pCLAMP 10.2 (Katsurabayashi et al., 2004). When the P1 or P2 current was a failure, the PPR was not included in the mean PPR data. For the time-course plots of PPR (Figs. 1–3, 6–8, and 10), when the P1 and/or P2 were failures, the PPR value was plotted as 0. Effects of LEV, phenytoin, muscimol, and Zn2+ on the current amplitude, Rf, and PPR for eEPSCs were normalized as relative changes from their respective controls. The relative INa and IBa responses with various concentrations of phenytoin and Zn2+ were similarly normalized to the control peak currents in standard solution without phenytoin or Zn2+. To examine exogenous glutamate (IGlu), NMDA (INMDA), and α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA) receptor responses (IAMPA), we applied 10 μM glutamate, 30 μM NMDA, and 3μM (S)-AMPA, respectively, concentrations that elicit little desensitization. The concentration-dependent effects of Zn2+ and phenytoin on these responses were quantified by normalizing the peak current amplitudes in the presence of drugs to those in the absence of drugs. Concentration-inhibition curves of INa, IBa, IGlu, and INMDA for Zn2+ or phenytoin, and concentration-response curve of IAMPA were fitted by a sigmoidal dose-response equation with Origin Pro 7.5 software (OriginLab Corporation, Northampton, MA).
Data are reported as the mean ± S.E.M. of these normalized values. Possible differences in the current amplitude, Rf, and PPR distribution of eEPSCs were tested by analysis of variance (ANOVA) and post hoc Dunnet’s test. Values of P < 0.05 were considered significant.
Effects of Phenytoin on eEPSCs, INa, IBa, and IGlu
Focal electrical stimulation of single boutons synapsing onto an isolated CA3 neuron induced fast inward currents that were completely and reversibly abolished by adding 10−20 µM CNQX, AMPA/KA receptor antagonist, indicating clearly that they are AMPA/KA receptor-mediated eEPSCs (Henze et al., 2002, Wakita et al., 2013). Paired-pulse stimulation at a frequency of 0.2 Hz produced paired EPSCs in which the second eEPSC amplitude (P2) was approximately 138.0 ± 5.0% (n = 66 neurons [boutons]) of the first EPSC amplitude (P1). As shown in Fig. 1A, P1 and P2 eEPSCs appeared in an all-or-none fashion, as reported previously (Shin et al., 2012a,b; Wakita et al., 2012), with approximately 13.2 ± 1.5% (n = 43 neurons [boutons]) of stimuli producing no responses or response failures.
Given that the therapeutic range of plasma concentrations of phenytoin is 40−80 µM (Hayes and Kootsikas, 1993), we investigated the effects of various concentrations of phenytoin between 1 and 100 µM on the paired glutamatergic eEPSCs responses that responded to both the first (P1) and second (P2) paired focal stimuli. Phenytoin at concentrations of 1−10 µM had no effect on the eEPSCs, but higher concentrations (≥30 µM) significantly suppressed the P1 amplitude and increased the Rf and PPR (P2/P1) of eEPSCs in a concentration-dependent fashion. Figure 1, A and B, shows representative current traces (inset) and experiments illustrating the time course of effects of 30 µM (A) and 100 µM (B) phenytoin on the P1 eEPSC current amplitude (Aa, Ba) and the PPR (Ab, Bb) for eEPSCs evoked by paired-pulse stimulation with 20-millisecond interpulse intervals. Phenytoin at 30 µM significantly decreased the P1 amplitude (to 60.2 ± 13.0% of control, n = 5; P < 0.05) and increased the Rf (to 194.3 ± 23.2% of control, n = 5; P < 0.05) and the PPR (to 160.1 ± 19.3% of control, n = 5; P < 0.05). A higher concentration (100 µM) further decreased P1 amplitude (to 18.2 ± 7.5% of control, n = 5; P < 0.05) and further increased the Rf (to 550.2 ± 95.4% of control, n = 5; P < 0.01) and the PPR (to 261.6 ± 30.4% of control, n = 5; P < 0.05) (Fig. 1C). It is evident that phenytoin acts presynaptically to decrease the excitatory glutamatergic synaptic transmission.
The inhibition curves for INa and IBa by phenytoin are shown in Fig. 1D. The currents were elicited by a depolarizing step pulse from a VH of −70 mV to −20 mV (INa) and from −60 to +10 mV (IBa). Phenytoin decreased both IBa and INa in a concentration-dependent manner, with a greater potency for inhibition of INa, where significant inhibition was observed at concentrations ≥ 10 μM. The potency of inhibition of IBa and eEPSCs by phenytoin was similar, with significant inhibition observed at concentrations ≥ 30 μM. Phenytoin also inhibited the glutamate-induced whole-cell response in a concentration-dependent manner, with significant inhibition observed at concentrations ≥ 10 μM (Fig. 1E).
Effects of LEV on eEPSCs, INa, and IBa
Human plasma concentration of LEV 1−14 hours after administration of a typical clinical dose (1500 mg) is 44−185 µM (Koubeissi et al., 2008). Thus, we similarly investigated clinically relevant concentrations of LEV (10, 100, and 300 µM) on eEPSCs. LEV at all these concentrations had no effect on the P1 amplitude, Rf, or PPR of glutamatergic eEPSCs (Fig. 2, A and B). The application of 30 µM 4AP, a nonselective K+ channel blocker, following application of LEV, caused a marked increase in eEPSC P1 amplitude and a marked decrease in Rf and PPR, confirming that the synapses examined in this set of data could be modulated as reported previously (Shin et al., 2012a). In addition, the highest relevant concentration of LEV (300 µM) had no effect on either INa or IBa (Fig. 2C). The results demonstrate that LEV alone has no direct pre- or postsynaptic effect on glutamatergic transmission at these mossy fiber–CA3 excitatory synapses.
Effects of Zn2+ on eEPSCs, INa, IBa, IGlu, IAMPA, and INMDA
Zn2+ and other divalent cations (e.g., Co2+, Cd2+, Mn2+), and trivalent cations (e.g., La3+), are known to decrease voltage-dependent Ca2+ currents in neurons (Akaike et al., 1981) and thus may be expected to decrease glutamatergic eEPSCs. Low concentrations of Zn2+ (0.1−3 µM) had no effects on glutamatergic eEPSCs, but concentrations over 10 µM significantly and concentration-dependently decreased the eEPSC P1 current amplitude and increased the Rf and PPR of eEPSCs (Fig. 3, A–C). Figure 3, A and B shows representative current traces and sample experiments indicating the effects of 10 µM (A) and 100 µM (B) Zn2+ on eEPSC parameters evoked by paired-pulse stimulation (30 millisecond intervals). Figure 3C summarizes the effects of Zn2+ at 0.1−300 µM on the current amplitude, Rf, and PPR of P1 eEPSCs. The concentration of Zn2+ that produced a half-maximal effect to decrease P1 and increase PPR and Rf was between 30 and 100 µM, with a good correlation among the effects on these parameters.
The concentration-dependent inhibition curves for Zn2+ at various concentrations (1−1000 µM) on INa and IBa are shown in Fig. 3D. Zn2+ decreased INa, but only at the high concentrations, beyond 300 µM. In contrast Zn2+ decreased IBa in a concentration-dependent manner from much lower concentrations, with an IC50 of 33.2 µM. Zn2+ could also inhibit IGlu, but concentrations greater than 300 μM were required (Fig. 3E).
Zn2+ is known to have effects on both AMPA (Sun et al., 2010) and NMDA receptors (Westbrook and Mayer, 1987; Molnár and Nadler, 2001). Application of (S)-AMPA, a selective AMPA receptor agonist, induced the AMPA-receptor-mediated whole-cell currents in a concentration-dependent manner (Fig. 4A). The presence of 10−300 μM Zn2+ potentiated the AMPA-receptor-mediated currents, whereas 1000 μM Zn2+ inhibited these currents (Fig. 4B). Figure 5A shows that NMDA (30 μM)-induced whole-cell inward current responses (INMDA) (VH = −40 mV) were only recorded in our conditions when we added glycine (1 μM) to the external solution and removed Mg2+. Hence, NMDA-receptor activation is unlikely to contribute to EPSCs recorded under our standard conditions with Mg2+ present and glycine free. Zn2+ concentration-dependently inhibited these NMDA-induced currents, with inhibition observed above 3μM and an approximate IC50 of 30 μM (Fig. 5B).
Together, the results clearly indicate that Zn2+ decreases eEPSCs at concentrations between 10 and 100 µM, and eEPSC inhibition likely results from suppression of presynaptic voltage-dependent Ca2+ channels and decreased glutamate release.
Interactions among Muscimol, Zn2+, and LEV on Presynaptic GABAA Receptors
We have previously demonstrated that mossy fiber glutamatergic terminals possess functional GABAA receptors that modulate eEPSCs (Yamamoto et al., 2011). In confirmation, Fig. 6 shows the effects of muscimol, a selective GABAA receptor agonist, on the glutamatergic eEPSCs. Muscimol at very low concentrations (0.03 µM) had a tendency to increase the P1 eEPSC current amplitude, and it decreased the PPR as we reported previously (Shin et al., 2011). Muscimol at higher concentrations (≥1 µM) significantly inhibited the P1 current amplitude and increased the Rf and the PPR. The muscimol-induced inhibition of eEPSCs can be prevented by adding 10 µM bicuculline, a selective GABAA receptor antagonist (Yamamoto et al., 2011).
In the continued presence of 1 µM muscimol, the addition of 1 µM Zn2+ increased the eEPSC P1 current amplitude and decreased Rf and PPR (Fig. 7, A and C). Recall that this concentration of Zn2+ had no effect by itself on the eEPSC, or on INa, IBa, IGlu, IAMPA, or INMDA. The modulatory effects of Zn2+ persisted throughout the experiment and restored the eEPSC parameters back toward the control (premuscimol) condition. However, when 300 µM LEV was applied in the continuous presence of 1 µM muscimol and 1 µM Zn2+, there was a decrease in the eEPSC P1 current amplitude and an increase in the Rf and PPR. The eEPSC parameters in response to LEV returned to the values observed with 1 µM muscimol in the absence of Zn2+ (Fig. 7, B and C). Recall also that LEV had no effect by itself on eEPSC parameters, and in the presence of 1 µM muscimol without Zn2+, LEV (300 µM) also had no effect on the eEPSCs (Fig. 8). These results indicate that LEV reverses the Zn2+-induced suppression of GABAA receptor–mediated presynaptic inhibition.
We next examined the actions of Zn2+ chelators on eEPSCs. We used two Zn2+ chelators, Ca-EDTA (Vogt et al., 2000) and RhodZin-3 (Sensi et al., 2003). Ca-EDTA is not a high-affinity chelator, but has much higher affinity for Zn2+ than other extracellular cations (Vogt et al., 2000). RhodZin-3 is a high-affinity chelator with nanomolar affinity for Zn2+ and no affinity for Ca2+ (≤40 µM), Mg2+ (≤500 µM), or Na+ (≤200 mM) (Sensi et al., 2003). Both of these agents (1 mM Ca-EDTA, 5 µM RhodZin-3) had no effect by themselves on glutamatergic eEPSCs (Fig. 9). However, the application of 1 mM Ca-EDTA in the continuous presence of 1 µM muscimol and 1 µM Zn2+ decreased the eEPSC P1 current amplitude and increased the Rf and PPR (Fig. 10A). Similarly, in the same experimental conditions, the application of 5 µM RhodZin-3 decreased the eEPSC P1 current amplitude and increased the Rf and PPR (Fig. 10B). The similarity between the actions of these Zn2+ chelators and those of LEV is consistent with the assumption that LEV reverses the Zn2+-induced disinhibition of GABAA-receptor mediated presynaptic inhibition.
Our present study compared the mechanisms of action of antiepileptic drugs on excitatory synaptic transmission at mossy fiber to CA3 neuron synapses in isolated hippocampal neurons. Phenytoin inhibited these glutamatergic synaptic responses in a concentration-dependent manner that correlated with its inhibition of voltage-dependent Na+ and Ca2+ channel currents in CA3 neurons (Fig. 1). Phenytoin also inhibited the postsynaptic CA3 soma membrane glutamate response to bath-applied glutamate in a concentration-dependent manner. Therefore, the inhibitory effects of phenytoin on eEPSC involve both direct postsynaptic glutamate receptor inhibition and the suppression of glutamate releases due to inhibition of the presynaptic voltage-dependent Na+ and Ca2+ channels. However, the presynaptic effect is likely to be more relevant. Application of 100 µM phenytoin markedly suppressed eEPSCs (Fig. 1C), including complete failures in some neurons (as seen in Fig. 1B), whereas the direct glutamate receptor response was only decreased to 64% of control at this concentration (Fig. 1E).
In contrast, LEV had no effect itself on glutamatergic eEPCSs (Fig. 2). Previous reported effects of LEV on glutamate neurotransmission are controversial. LEV has been reported to inhibit glutamatergic eEPSCs recorded from granule cells of the dentate gyrus via inhibition of presynaptic P/Q-type calcium channels (Lee et al., 2009). However, LEV was also reported to increase elevated K+-induced glutamate release in rat neocortex (Kammerer et al., 2011).
We also found that Zn2+ concentration-dependently reduced eEPSCs, with the pattern of responses (increased PPR and Rf) consistent with a presynaptic site of action. The inhibition of eEPSCs was concurrent with inhibition of somatic voltage-dependent Ca2+ channels (≥3 µM), and inhibition of these channels in presynaptic terminals is likely to mediate the inhibition of glutamatergic excitatory transmission (Fig. 3). These responses were consistent with studies using in vivo microdialysis in rat hippocampus (Takeda et al., 2004) and rat mossy fiber synaptosomes (Bancila et al., 2004) that showed Zn2+-induced inhibitory presynaptic glutamate release. In contrast, Huang et al. (2008), using mice hippocampal slices, showed that Zn2+ potentiates the efficacy of the hippocampal mossy fiber–CA3 pyramid synapse by a mechanism that requires tropomyosin-related kinase receptor subtype B. We are unsure why these studies differ, but it may relate to different sensitivities of the presynaptic terminals to Zn2+, resulting from different experimental methods and preparations. Here we would like to make sure that the “synaptic bouton” preparation allows a more direct examination of drug effects at the single synapse level, devoid of potential influences from extrasynaptic sites and surrounding neurons and glia cells.
Muscimol, a selective GABAA receptor agonist, significantly decreased the amplitude of eEPSCs, concurrent with an increase in PPR (Fig. 6). These results illustrate the importance of presynaptic GABAA receptors in controlling glutamate release. GABA interneurons in the hippocampus project to neurons and to the glutamatergic presynaptic mossy fiber terminals. Activation of these presynaptic GABAA receptor channels can cause a biphasic modulation of glutamatergic transmission, with enhanced release at low levels of activation and with inhibition of release at higher concentrations (Jang et al., 2006; Yamamoto et al., 2011). Under physiologic conditions, Na+-K+-2Cl− cotransporter type 1 (NKCC-1) transports Cl− into nerve terminals and thereby facilitates a depolarizing Cl− efflux upon activation of GABAA receptors (Kakazu et al., 1999; Jang et al., 2002). This GABAA-induced presynaptic depolarization induces facilitation of sEPSC frequency, even in CA3 pyramidal neurons dissociated from 4-week-old rats (Han et al., 2009). The same presynaptic GABAA-induced “depolarization” causes a decrease in evoked release of glutamate, via a depolarization block of Na+ channels (Fig. 6). The presynaptic inhibition of eEPSCs and the increase in spontaneous release by muscimol are blocked by the pretreatment with bumetanide, a selective blocker of NKCC-1, confirming that Cl− efflux and depolarization mediate this effect (Yamamoto et al., 2011).
A clinically relevant concentration of LEV (300 µM) applied with 1 µM muscimol did not further change eEPSCs (Fig. 8). In contrast, Zn2+ significantly inhibited the muscimol-induced presynaptic GABAA receptor activation, resulting in an increase in eEPSC amplitude (Fig. 7). Zn2+ is a well known physiologic GABAA receptor negative modulator, which binds at three discrete binding sites to mediate subunit-dependent effects. One binding site is located within the ion channel pore itself, with the other two residing in the externally facing amino (N)-terminus, at the interfaces between α and β subunits (Hosie et al., 2003). The precise subunit combination of the presynaptic GABAA receptors at mossy fiber–CA3 synapses is unclear, although the possible presence of various subunit combinations (δ, α1, α4, α5, γ2) has been implicated by the high neurosteroid and muscimol sensitivity of the synapses (Ruiz et al., 2010; Kim et al., 2011). Interestingly, the subunit expression pattern of GABAA receptors in the dentate granule cell layer changes in the epileptic hippocampus (Brooks-Kayal et al., 1998). Specifically, expression of α1 subunits decreases while expression of α4 and δ subunits increases (Brooks-Kayal et al., 1998). Although α1 subunits can exhibit a low sensitivity to Zn2+, association with γ2-containing GABAA receptors results in a high Zn2+ sensitivity (White and Gurley, 1995; Burgard et al., 1996; Fisher and Macdonald, 1998). Replacement of γ2 with a δ subunit further increases Zn2+ sensitivity of GABAA receptors (Saxena and Macdonald, 1996). Subunit changes observed in the epileptic brain may make them more sensitive to the action of Zn2+ and hence antagonism of Zn2+ actions on GABAA receptors by levetiracetam may be particularly important.
When a clinically relevant concentration of LEV was applied in the continued presence of 1 µM muscimol and a low concentration of Zn2+ (1 µM), the eEPSC amplitude returned to levels seen prior to Zn2+ application (Fig. 7). The most important finding in this study was that LEV reversed the Zn2+-induced disinhibition of the presynaptic GABAA receptor-induced depolarization blockade. In this way LEV decreased eEPSC amplitude. This interpretation is consistent with a study by Rigo et al. (2002) on cultured cerebellar granule cells from epileptic mice, who also reported that LEV could antagonize the inhibitory effects of Zn2+ on whole-cell postsynaptic GABA responses. Our study extends this to presynaptic GABAA receptor actions and is the first study to report antagonizing the inhibitory effects of Zn2+ in acutely isolated neurons. This interpretation was further supported by the similar actions of LEV and Zn2+ chelation. The well-known Zn2+ chelators, Ca-EDTA (Vogt et al., 2000) and RhodZin-3 (Sensi et al., 2003), did not affect glutamatergic transmission by themselves (Fig. 9), but they reversed the antagonizing effects of Zn2+ on muscimol-induced presynaptic GABAA receptor–mediated inhibition (Fig. 10). The effects of both Zn2+ chelators were approximately the same as that obtained with LEV. The molecular basis by which LEV is functionally antagonizing the Zn2+ effects on GABAA receptor activation is unclear at present. However, both studies by Rigo et al. (2002) and ours propose that antagonism of allosteric Zn2+ modulation by LEV is a novel mechanism of action.
The interactions between LEV and Zn2+ may be clinically important, when considering the hippocampal hyperexcitability and altered synaptic circuits in temporal lobe epilepsy (Coulter, 1999, 2000). In both human temporal lobe epilepsy and in animal models, there is a marked sprouting of mossy fiber terminals to innervate new targets in the inner molecular layer of the dentate gyrus. Zn2+ is colocalized and coreleased with glutamate from nerve terminals (Assaf and Chung, 1984; Howell et al., 1984), with mossy fiber terminals being particularly rich in Zn2+ (Lin et al., 2001). Large amounts of Zn2+ and glutamate are coreleased during seizures as a result of this mossy fiber sprouting, and may diffuse into extracellular spaces to inhibit GABAA receptor-mediated presynaptic inhibition onto dentate granule cells. Coupled with the ability of Zn2+ to facilitate glutamate-evoked synaptic currents in hippocampal slices (Lin et al., 2001), this could induce a vicious circle of excitation, which may contribute to the hyperexcitability and seizures in the epileptic hippocampus. Our results suggest that LEV may, at least partly, mediate some of its antiepileptic effects via preventing this Zn2+ modulation of presynaptic GABAA receptors.
Palma et al. (2007) reported an additional aspect to LEV’s interactions with GABAA receptors, directly reducing the extent of GABAA receptor whole-cell response run-down. This effect was only seen in oocytes injected with membranes isolated from the mesial temporal lobe epilepsy neocortex and in human mesial temporal lobe epilepsy slices, but not in control brains, and after prolonged LEV incubations. LEV-induced alleviation of GABAA receptor response run-down would work in concert with the LEV actions reported here to facilitate inhibition.
LEV has shown a clinically favorable outcome in temporary lobe epilepsy (Jehi et al., 2012), and the results presented here provide an additional facet to the mechanisms by which LEV may exert its antiepileptic effects.
The authors thank Dr. A. Moorhouse (University of New South Wales, Sydney, Australia) for valuable comments and critical reading of the manuscript.
Participated in research design: Akaike, Wakita, Kogure.
Conducted experiments: Wakita.
Performed data analysis: Wakita.
Wrote or contributed to the writing of the manuscript: Kotani, Wakita, Akaike.
- Received August 13, 2013.
- Accepted November 19, 2013.
M.W. and N.K, contributed equally to this work.
This work was supported by Otsuka Pharmaceutical Co., Ltd., and by Grants-in-Aid from Kumamoto Health Science University (to M.W. and N.A.) and by a Grant-in-Aid from Kitamoto Hospital, Koshigaya (to N.K. and N.A.).
- α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid
- ATP–magnesium salt
- evoked excitatory postsynaptic current(s)
- GABA type A
- paired-pulse ratio
- failure rate
- Copyright © 2014 by The American Society for Pharmacology and Experimental Therapeutics