Studying epileptogenesis in a genetic model can facilitate the identification of factors that promote the conversion of a normal brain into one chronically prone to seizures. Synapsin triple-knockout (TKO) mice exhibit adult-onset epilepsy, thus allowing the characterization of events as preceding or following seizure onset. Although it has been proposed that a congenital reduction in inhibitory transmission is the underlying cause for epilepsy in these mice, young TKO mice are asymptomatic. We report that the genetic lesion exerts long-term progressive effects that extend well into adulthood. Although inhibitory transmission is initially reduced, it is subsequently strengthened relative to its magnitude in control mice, so that the excitation to inhibition balance in adult TKOs is inverted in favor of inhibition. In parallel, we observed long-term alterations in synaptic depression kinetics of excitatory transmission and in the extent of tonic inhibition, illustrating adaptations in synaptic properties. Moreover, age-dependent acceleration of the action potential did not occur in TKO cortical pyramidal neurons, suggesting wide-ranging secondary changes in brain excitability. In conclusion, although congenital impairments in inhibitory transmission may initiate epileptogenesis in the synapsin TKO mice, we suggest that secondary adaptations are crucial for the establishment of this epileptic network.
During epileptogenesis, it is assumed that a normal brain is converted into one debilitated by recurrent seizures by a progression of aberrant plasticity events that produce an overly excitable hypersynchronized network (Pitkanen et al. 2009; Reddy and Mohan 2011). Depending upon the type of epilepsy or the epilepsy model, this process may involve a combination of cell dysfunction and/or loss, as well as increased excitability and abnormal rewiring, which culminate in a self-sustaining epileptic state (Giblin and Blumenfeld 2010). Although the underlying etiology is not always clear, in many cases epilepsy can be traced back to an initiating key event such as brain injury or inflammation (Hawkins and Davis 2005; Badawy et al. 2009). Following this initiating event, an extended quiescent latent period may ensue during which epileptogenesis takes place. The onset of epilepsy itself is defined by the subsequent appearance of recurring seizures (Ben-Ari et al. 2008). Consequently, this intervening latent period is considered a critical window of opportunity for therapeutic intervention, during which the onset of epilepsy may be delayed, its severity reduced, or, better yet, epilepsy may be averted altogether (Pitkanen and Lukasiuk 2009; Giblin and Blumenfeld 2010). In genetic epilepsy, the epilepsy-precipitating event is replaced by an underlying genetic predisposition that promotes the occurrence of epileptiform brain activity, which can eventually develop into full-blown epilepsy (Zara and Bianchi 2009). In this case, the affected individual can be viewed as being born in a latent state, which terminates with the onset of epilepsy. Such a predisposition is observed in the case of mutations in the synapsin genes. Mutations in the genes SYN1 and SYN2 are of clinical interest based on their linkage with epilepsy-susceptibility in humans (Garcia et al. 2004; Cavalleri et al. 2007; Lakhan et al. 2010). In mice, deletion of several combinations of the synapsin genes produces an epileptic phenotype (Rosahl et al. 1995; Etholm and Heggelund 2009; Boido et al. 2010; Etholm et al. 2010, 2011; Ketzef et al. 2011). We and others have reported that the synapsin TKO mice consistently exhibit an abrupt transition from a latent seizure-free state to a persistent symptomatic state of sensory-induced tonic-clonic seizures at the approximate age of 2–3 months (Boido et al. 2010; Ketzef et al. 2011). Therefore, this model exhibits the distinct advantage that it allows the purposeful study of the progression of epileptogenesis: It allows studying how a genetic lesion leads to a first seizure and how a chronic state of recurring seizures is consequently established.
The synapsins are abundant phosphoproteins associated with the surface of synaptic vesicles (Cesca et al. 2010). They immobilize a subgroup of resting synaptic vesicles within the synaptic terminal (Pieribone et al. 1995; Orenbuch et al. 2012) and allow their activity-dependent mobilization to release sites (Hilfiker et al. 1999). In their absence, a significant proportion of the vesicle population is lost (Gitler et al. 2004; Siksou et al. 2007; Orenbuch et al. 2012). Importantly, deletion of the synapsins differentially affects excitatory and inhibitory neurotransmission, and specifically tones down basal inhibitory responses (Gitler et al. 2004; Baldelli et al. 2007). This deficit in inhibitory transmission has been credited with the epileptic nature of the synapsin KO mice (Rosahl et al. 1995; Chiappalone et al. 2009; Boido et al. 2010). Indeed, network activity is enhanced in the brains of young TKO mice before the onset of epilepsy (Boido et al. 2010; Ketzef et al. 2011; Farisello et al. 2012). However, the fact that the onset of epilepsy is delayed until the transition to adulthood suggests that the primary genetic lesion is insufficient to directly incur the epileptic phenotype. Rather, this implies that functional changes that occur during the latent period are necessary to initiate epilepsy. Indeed, in a previous study, we concluded that contrary to prevailing models, γ-aminobutyric acid (GABA)ergic transmission is enhanced in the brains of adult epileptic TKO mice (Ketzef et al. 2011). In the current study, we directly probed synaptic and intrinsic properties of neurons in wildtype (WT) and synapsin TKO brains using intracellular recordings. Here, we uncover a gradual derailment of the balance between excitatory and inhibitory synaptic transmission in favor of inhibition. In addition, we identify age-dependent changes in tonic inhibition and in the action potential (AP) duration, both of which are not considered to be directly determined by synapsin. Our results therefore illustrate extensive, long-term network adjustments in the brain of the synapsin TKO mouse during epileptogenesis and the onset of epilepsy.
The synapsin TKO mice have been described previously (Gitler et al. 2004). SYN 1/2/3 (−/−) homozygous mice are on a C57Bl6/SV129 genetic background. SYN 1/2/3 (+/+) mice, that are the progeny of litter mates of the original mutant mice, served as WT controls. SYN 1/2/3 (−/−) mice backcrossed onto a C57Bl6 background have been shown to exhibit a similar epileptic phenotype (Boido et al. 2010). The mice were raised in a temperature-and-humidity-controlled environment, under a 12:12 h light:dark cycle and were fed ad libitum. Pups were separated at the age of 21–28 days. Mice were monitored for 2 min for commencement of seizures following cage opening during regular handling (Ketzef et al. 2011). Experiments were conducted in accordance with animal protocol IL-32-03-2009 approved by the Ben-Gurion University Institutional Committee for Ethical Care and Use of Animals in Research, in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.
Electrophysiological Patch Clamp Recordings In Vitro
The acute slice preparation was used (Pavlovsky et al. 2003; Seiffert et al. 2004; Ivens et al. 2007). Briefly, mice were deeply anesthetized with isoflurane and decapitated. Brains were quickly removed, and 290–300 μm thick horizontal cortico-hippocampal slices corresponding to plates 143–157 (Paxinos and Franklin 2001) were prepared (Dreier and Heinemann 1991) using a vibroslice (Campden instruments, Loughborough, Leicestershire, United Kingdom) in ice-cold (1–3°C) carbogenated (5% CO2 and 95% O2) high sucrose cutting solution (in mM: 50 NaCl, 25 NaHCO3, 1 NaH2PO4, 7 MgCl2, 0.5 CaCl2, 2.5 KCl, 10 glucose, 150 sucrose). Slices were incubated for 30 min at room temperature (RT) in cutting solution followed by incubation in carbogenated artificial cerebral spinal fluid (ACSF; in mM: 124 NaCl, 26 NaHCO3, 1.25 NaH2PO4, 2 MgSO4, 2 CaCl2, 3 KCl, 10 glucose, pH = 7.4 ± 0.1, osmolarity ∼ 300 mOsm).
After equilibration, slices were transferred to a recording chamber on an upright, fixed-stage microscope equipped with infrared differential interference contrast optics (Zeiss axioscope 2 FS plus, Germany) and a charge-coupled device camera (HQ2 Coolsnap, Roper Scientific Tucson, AZ, United States of America), and were continuously perfused with ACSF at a rate of 1 mL per minute. Recording was performed in layer 5 (L5) of the entorhinal cortex (EC). Patch pipettes were pulled from borosilicate glass (5–7 MΩ, Science Products, Germany) with a P-97 puller (Sutter Instrument, Novato, CA, United States of America). In order to record in current-clamp experiments the membrane potential and the AP waveform, we utilized a potassium-based intracellular solution (in mM): 135 K-gluconate, 6 KCl, 2 MgCl2, and 10 K-4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), pH 7.25. To record postsynaptic responses in the whole-cell voltage-clamp configuration, we used a cesium-based intracellular solution intended to extend the space clamp: 135 Cs-gluconate, 6 CsCl, 2 MgCl2, and 10 Cs-HEPES, pH 7.25 (Fleidervish et al. 2001). The Cs-based intracellular solution was also used to record the potentiation of network activity by the potassium channel blocker 4-AP (Barkai et al. 1995). Although 4-AP and Cs are redundant in the patched neuron, neurons patched in this manner were advantageous for monitoring the enhancement of synaptic release by other neurons. Recordings were obtained with a Multiclamp 700B amplifier, low-pass filtered at 2 kHz, digitized on-line (Digidata 1322A) and recorded using pClamp 9.2 (Molecular Devices, Sunnyvale, CA, United States of America). For whole-cell patch-clamp recordings, cells were first voltage-clamped at −70 mV. Spontaneous excitatory post-synaptic currents (sEPSCs) and miniature EPSCs (mEPSCs) were examined at a holding potential of −70 mV in the absence and presence of 1 μM TTX, respectively. Spontaneous and miniature inhibitory post-synaptic currents (sIPSCs and mIPSCs, respectively) were recorded at a holding potential of 0 mV. Because both EPSCs and IPSCs were recorded for each individual cell, the average excitatory-to-inhibitory (E/I) ratio (amplitude or frequency) of each cell was calculated and then averaged per animal, per genotype and per age group. Stimulation was timed by the pClamp software and was delivered by an isolated stimulation unit (isoflex, AMPI, Jerusalem, Israel) using a concentric electrode placed adjacently to the recorded cell body. The threshold current for evoking responses (Ith) was defined as the lowest current that could still elicit a postsynaptic response in >70% of the trials. A response was defined as a deflection larger than 3 times the standard deviation of the mean holding current appearing <5 ms after the stimulation artifact. According to these criteria, the threshold for stimulation (0.04–0.09 mA) did not vary significantly across, genotype, or modality (n > 8, P = 0.255 Kruskal–Wallis test). High-frequency stimulation was delivered at 2.5-fold the determined threshold intensity. In current-clamp recordings the resting membrane potential was recorded immediately after breaking-in and was filtered at 10 kHz. APs were elicited by short 2–3 ms long 200–600 pA current injections. Repetitive firing was studied by applying 500 ms long current steps with an increment of 20 pA. The input resistance was calculated from the response to a hyperpolarizing −10 pA current step. The series resistance was typically <16 MΩ and was monitored throughout the recordings. Data were used for analysis only if the series resistance remained <20 MΩ and changed by <20% during the recordings. All recordings were performed at RT.
Drugs were applied by perfusion. Recordings were performed at least 20 min after the commencement of drug application. N-methyl-d-aspartate (NMDA), non-NMDA, or GABA type A (GABAA) receptors were blocked with 30 μM 2-amino-5-phosphovaleric acid (APV), 10 μM 6-cyano-7-nitroquinoxyline-2, 3-dione (CNQX), and 50 μM bicuculline methiodide (BMI), respectively. Transient potassium currents were blocked with 50 μM 4-aminopyridine (4-AP). Tonic GABAA receptors were blocked with 50 μM Gabazine (SR-95531). Voltage-gated sodium channels were blocked with 1 μM Tetrodotoxin (TTX). Materials were purchased from Sigma-Aldrich (Rehovot, Israel).
Data Analysis and Statistics
Data analysis was performed using the pClamp 9.2 (Clampfit) and IBM SPSS Statistics 19 (SPSS) programs. The frequency and amplitude of synaptic events were assessed for 2 min for each experimental condition. Synaptic events were automatically detected by the pClamp 9.2 software using an event detection template-based search tool. Event frequencies and amplitudes were averaged across neurons, animals, and experimental groups. When data were distributed mostly not-normally (Shapiro–Wilk test, P > 0.05) nonparametric tests were used: Either 2-tailed Mann–Whitney U-test for 2-group comparisons or the Kruskal–Wallis test, for multiple comparisons, and specifically for testing age-dependency. Data are expressed as mean ± standard error of the mean throughout. Statistical significance was set at P ≤ 0.05.
Real time-polymerase chain reaction (RT-PCR) analysis was performed essentially as described (de Moura et al. 2010). Briefly, brains of young (3–4 weeks old) and adult (9–12 months old) WT and TKO mice (n = 5 for each group) were quickly removed in ice-cold ACSF that was prepared with diethylpyrocarbonate (DEPC)-treated double-distilled water. The cerebral cortex, dorsal to the hippocampus was collected from each hemisphere, as described (Ketzef et al. 2011). The tissue was frozen immediately in liquid nitrogen and kept at −80°C until use. RNA was extracted using the RNeasy Lipid Tissue mini kit (Qiagen, MD, United States of America) according to manufacturer's protocol. RNA samples were quantified by spectrophotometry. Complementary DNA (cDNA) was synthesized by reverse transcription with the Verso cDNA synthesis kit (ABgene, Epsom, Surrey, United Kingdom). The cDNA was amplified using the Applied Biosystems 7300 Real Time PCR System, using SYBR green (Applied Biosystems, Foster City, CA, United States of America). The Primer sequences (forward/reverse) are as follows: GABAAα5: TTATTCTTACTGGGAATGGACAATGG/TTAAACCGCAGCCTTTCATCTTTC, GABAAδ: GACTACGTGGGCTCCAACCTGGA/ACTGTGGAGGTGATGCGGATGCT, GAPDH: CAATGTGTCCGTCGTGGATCT/GTCCTCAGTGTAGCCCAAGATG (Sigma-Aldrich, Rehovot, Israel). GAPDH was used as an endogenous reference gene. The following PCR conditions were used: Preheating to 50°C for 2 min, denaturation at 95°C for 15 min, and 40 cycles of amplification and quantification (15 s at 95°C and 60 s at 60°C). Reactions were carried out in triplicate. The 4 experimental groups (adult WT, young WT, adult TKO, and young TKO) were normalized by the value of the adult WT sample. Statistical analysis was carried out using the Kruskal–Wallis 1-way analysis of variance applied simultaneously to the 4 experimental groups, followed by the Mann–Whitney test used pairwise, with significance set at P < 0.05.
Slices From Adult Synapsin TKO Mice Exhibit Aberrant Multiphasic Hypersynchronous Activity
In a previous study, we showed that slices from adult synapsin TKO mice, but not those from young TKO mice or WT mice of any age, exhibit epileptiform activity upon stimulation (Ketzef et al. 2011). Because we had used extracellular recordings, we could not determine unambiguously the specific contribution of glutamatergic and GABAergic transmission to these responses. To do so now, we performed measurements in the whole-cell voltage-clamp configuration in the absence of receptor blockers. The chloride reversal potential was set to −70 mV, allowing us to measure glutamatergic and GABAergic ionotropic currents independently by setting the membrane holding potential to −70 or 0 mV, respectively (Supplementary Fig. 1). This configuration allowed us, at the same time, to preserve the local network connectivity, which is critical for the study of epilepsy. Recordings were performed from layer 5 pyramidal neurons of the EC; this area was chosen because the temporal cortex, which includes both the hippocampus and the EC, is often involved in the generation of epilepsy in human patients (Schwartzkroin and McIntyre 1997). More importantly, we previously observed epileptiform activity in slices from symptomatic adult synapsin TKO mice in the EC, rather than the hippocampus, in the absence of convulsing agents (Ketzef et al. 2011).
Measurements in slices from adult synapsin TKO mice (>7 months of age) revealed broad all-or-none, multiphasic and prolonged evoked responses in 58% of the sampled cells (n = 7 out of 12) obtained from 60% (n = 3 out of 5) of the examined mice (Fig. 1A). Both excitatory and inhibitory components exhibited similar characteristics, suggesting that they reflect extensive network activity. In addition, spontaneous hypersynchronous seconds-long ictal activity composed of multiple sequential high-amplitude synaptic events was evident in a subset of adult synapsin TKO cells (3 cells, Fig. 1B). Such extended responses, whether evoked or spontaneous, were never observed in age-matched WT controls or in young TKO slices (adult WT: n = 8 cells/4 animals, young WT: n = 12/5, young TKO: n = 14/5), suggesting that this phenomena emerges due to the properties of the slices rather than the recording conditions. Analyzing the input/output response curves revealed a significant enlargement in the evoked inhibitory component exclusively in adult TKO slices (P = 0.029, repeated-measures ANOVA, Fig. 1C,D), which was not observed in the excitatory component (P = 0.36). This confirms our previous indirect evidence based on extracellaular recordings that showed an age-dependent increase in inhibitory transmission in the synapsin TKO brain (Ketzef et al. 2011).
Changes in Spontaneous Inhibitory Transmission in the TKO Network
To determine the progression of changes that occur before and after the onset of epilepsy, we examined spontaneous synaptic activity at 4 different stages of the progression of epilepsy: In presymptomatic 2–3-week-old mice, 1.5-month-old mice (±5 days) prior to experiencing their initial seizure, just symptomatic 3.5-month-old mice (±5 days), and epileptic adult mice (7–14-month-old mice).
We first studied the properties of miniature quantal synaptic events, recorded in the presence of the voltage-gated sodium-channel blocker TTX (Fig. 2A,C). These inform us about the postsynaptic response to single presynaptic quanta (Del Castillo and Katz 1954). No differences were observed in the amplitude of both excitatory and inhibitory miniature synaptic events in slices obtained from WT mice or from age-matched TKO presymptomatic mice, whether 2–3 weeks old (nWT = 15/7, nTKO = 8/7) or ∼1.5 months of age (n = 13/4 for both genotypes, P > 0.25, Mann–Whitney test, Fig. 2E,F). In contrast, in slices taken from ∼3.5 months old TKO mice that recently started seizing, the mEPSC amplitude was increased by 30.8% compared with WT slices (nWT = 12/4, nTKO = 10/3, P = 0.002, Fig. 2A,B, left panel in E), consistent with heightened excitability. However, in older mice (>7 months old) the mEPSC amplitude reverted to WT values (P = 0.275, nWT = 11/7, nTKO = 22/8), in parallel to an increase in the mIPSC amplitude (P = 0.001, Fig. 2C,D, middle panel in E), perhaps reflecting compensation (see below) (Echegoyen et al. 2007). Because the waveform of mEPSCs and mIPSCs of the youngest and oldest mice did not differ between genotypes (Supplementary Fig. 2), we tentatively ruled out a contribution of channel kinetics to the observed differences in the mPSC amplitudes. No differences in the frequency of the miniature events of both modalities were observed between genotypes for all age groups (P > 0.34, Fig. 2F), consistent with the observed similarity in WT and TKO neuronal populations (Ketzef et al. 2011). No age-related trend was revealed in mEPSC or mIPSC amplitudes (P > 0.157, Kruskal–Wallis test), and only in WT slices did the frequency of mIPSCs decrease with age (P = 0.005, Fig. 2E,F).
Unlike miniature events, spontaneous transmission, which is recorded in the absence of TTX, is additionally dependent on the excitability of the network and on the synaptic transfer function that translates network activity into actual synaptic transmission. Effects of the deletion of the synapsins on spontaneous transmission were apparent already at the earliest age we examined. Specifically, both the amplitude and frequency of the spontaneous inhibitory events (sIPSCs) were lower in slices from 2- to 3-week-old mice (nTKO = 6/6, nWT = 7/7, P = 0.002 and 0.035, respectively, Mann–Whitney test, Fig. 3A,B, middle panels in E,F), in agreement with previous reports of weaker inhibitory transmission in cultures (Gitler et al. 2004; Baldelli et al. 2007) and in slices (Farisello et al. 2012). Since no difference had been observed in the mIPSC amplitude in this age group (Fig. 2), the decrease in the average sIPSC amplitude suggests a reduction in either the excitability of inhibitory neurons or their release probability. However, this difference between genotypes gradually reversed. In slices from 1.5-month-old presymptomatic mice (n = 6/4) and also in 3.5-month-old mice that had just started to exhibit seizures (n = 6/4), the characteristics of the sIPSCs reverted to WT values (P > 0.329, Fig. 3 middle panels in E,F). Moreover, in epileptic mice >7 months of age, both the amplitude and frequency of the sIPSC were significantly larger than in WT tissue (nTKO = 7/7, nWT = 5/5, P = 0.021 and 0.05, respectively, Fig. 3C,D, middle panels in E,F), in agreement with our previous conclusion that inhibition is enhanced in adult TKO mice (Ketzef et al. 2011). Examination of the time dependence of the alteration in the sIPSC properties revealed that in WT mice the amplitude decreased with age (P = 0.041, Kruskal–Wallis test, Fig. 3E, middle panel), a tendency which was inverted in the TKO slices (P = 0.012). Consequently, inhibition in the TKOs is relatively enhanced. In contrast to the substantial effect of deletion of the synapsins on the sIPSCs, the amplitude and the frequency of sEPSCs were not affected at any age we examined (Fig. 3, E,F, left panels).
A Changing Balance Between Excitation and Inhibition
When studying epilepsy, determining the balance between excitation and inhibition is of the utmost importance (El-Hassar et al. 2007; Turrigiano 2011). Because the E/I ratio provides a normalized measure of this balance, we examined the E/I ratio of the amplitude and of the frequency of miniature and spontaneous synaptic activity. In agreement with the results presented above, the E/I ratio for the amplitude of mPSCs peaked in 3.5-month-old mice (P = 0.05) and returned to WT values in epileptic mice >7 months of age (Fig. 2E, right panel). Furthermore, no difference was observed in the E/I ratio for mPSC frequencies between WT and TKO (Fig. 2F, right panel). When examining the E/I amplitude ratio for spontaneous synaptic activity, we found that it was significantly higher in the youngest TKO slices compared with age-matched WTs (P = 0.014, Mann–Whitney test, Fig. 3E, right panel), consistent with the smaller amplitude of GABAergic transmission in the TKOs at this age. Although this observation could be interpreted as contributing to epilepsy, synapsin TKO mice at this age do not yet suffer from seizures (Ketzef et al. 2011). In contrast, in older mice the E/I amplitude ratio no longer differed between TKO and WT (P > 0.522). Moreover, a gradual increase in the E/I frequency ratio of spontaneous events that was observed in WT slices (P = 0.026 Kruskal–Wallis test, Fig. 3F, right panel) was absent in the TKO ones (P = 0.931). To conclude, a relative shift towards inhibition was observed in the TKO adults. One possibility to explain this observation is that an increase in inhibition reflects a compensatory mechanism that raises the threshold to induce seizures, in agreement with the progressing decline in seizure susceptibility in TKO mice after the age of 6 months (Supplementary Fig. 3). A similar trend has been reported (Cambiaghi et al. 2012), albeit initiating at a younger age. On the other hand, enhanced inhibition has been shown to promote epilepsy in other models (Ouardouz and Carmant 2012). To more fully document the effects on inhibition, we reexamined this critical observation using pharmacological tools.
4-aminopyridine (4-AP), a potassium channel blocker and a known convulsant, enhances synaptic release at both excitatory and inhibitory terminals (Barkai et al. 1995; Avoli et al. 2002). As a consequence, it potentiates network activity, as evidenced by paroxysmal depolarizing shift (PDS) events observed in all adult slices (>7 months old, nWT = 8/5, nTKO = 8/6, Fig. 4A). Similar PDS events were observed when the reporter cell was held at either 0 mV (P = 1.00) or −70 mV (p = 0.094, Mann–Whitney test, Fig. 4B), consistent with the induction of extensive network activity involving both excitatory and inhibitory components. 4-AP also caused a substantial increase in the amplitude and frequency of spontaneous synaptic release (Fig. 4C). In adult (>7 months old) WT slices 4-AP augmented the amplitude of both components similarly (Supplementary Fig. 4), resulting in no net change in the E/I amplitude ratio (n = 8/5, P = 0.237, Fig. 4D). In contrast, in age-matched TKO slices, 4-AP unmasked additional inhibitory capacity (Supplementary Fig. 4), producing a significant decrease in the E/I ratio (n = 8/6, P = 0.017). 4-AP increased the frequency of excitatory and inhibitory sPSCs differently in WT and TKO slices. In TKO slices both components were enhanced similarly (Supplementary Fig. 4), resulting in no change in the E/I frequency ratio (P = 0.776, Fig. 4D, right). In contrast, in WT slices 4-AP increased the frequency of sIPSCs to a larger extent than sEPSCs (Supplementary Fig. 4), causing a significant decrease in the E/I frequency ratio (P = 0.018, Fig. 4D), which reached the same level observed in TKO slices. A possible explanation for this difference is that in TKO slices sIPSCs were more frequent prior to the addition of 4-AP (Fig. 3F, right panel). This result consolidates our previous conclusions regarding an increase in inhibition in adult TKO mice and is in agreement with the maintenance of balance between excitation and inhibition even after 4AP application in the synapsin TKO mice (Ketzef et al. 2011).
An increase in inhibitory tone can mask an increase in excitation, by dampening the total excitability of the network. To determine the maximal excitatory capability of a network unhampered by inhibition, we measured PDS events produced when the GABAA receptor blocker BMI (Rohrbacher et al. 1998; Psarropoulou and Descombes 1999; Schiller 2002) was applied to slices from adult >7-month-old mice (Fig. 4E). Using extracellular recordings, we previously observed hypersensitivity of adult TKO slices to BMI (Ketzef et al. 2011). Here, we observed that the charge transfer during BMI-induced PDS events in adult TKO slices (n = 10/6) was larger compared with those recorded in age-matched WT slices (n = 9/6, P = 0.035, Fig. 4F). This result indicates that the total excitatory release capability was higher in the TKO slices when unrestricted by inhibition. Conversely, it also implies that in TKO slices inhibition is enhanced, since it is efficient enough to counteract stronger excitation.
Age-Dependent Alteration in the Time Course of Responses to High-Frequency Stimulation
We further reexamined the kinetics of synaptic depression, an additional known effect of deletion of the synapsins. When synapses are challenged with high-frequency stimulation, an eventual gradual decrease in synaptic strength is observed in many synapse types (Pan and Zucker 2009). This type of synaptic depression occurs, at least partially, because of depletion of the readily releasable vesicles. Consequently, its time course is regulated to a large degree by the recruitment of reserve vesicles from the recycling pool (Pan and Zucker 2009; Denker and Rizzoli 2010). Because the synapsins are key regulators of reserve vesicles, manipulations that inactivate the synapsins have repeatedly been shown to substantially accelerate synaptic depression (Pieribone et al. 1995; Rosahl et al. 1995; Hilfiker et al. 1999; Gitler et al. 2004; Baldelli et al. 2007; Bogen et al. 2009). We measured the time course of synaptic depression in response to 20 Hz stimulation (Luthi et al. 2001; Jensen et al. 2007) in WT and synapsin TKO slices from young presymptomatic 2–3-week-old mice and also from adult >7 months old epileptic mice, an age group in which synaptic depression has not been previously examined. In agreement with previous recordings performed in various synapsin KOs (Gitler et al. 2004; Hvalby et al. 2006; Bogen et al. 2009; Farisello et al. 2012), we observed that depression of glutamatergic transmission in young TKO slices was significantly faster than that observed in age-matched WT slices (nWT = 12/5, nTKO = 14/5, P = 0.03 repeated-measures ANOVA calculated for the first 100 stimuli, Fig. 5A), whereas depression of inhibitory responses was similar for both genotypes (P = 0.167). However, when the same experiments were performed on slices from adult epileptic mice (>7 months of age, nTKO = 10/5, nWT = 8/4), no differences were observed in the kinetics of depression for both glutamatergic and GABAergic transmission (P = 0.701 and 0.431, respectively, Fig. 5B). It should be noted that as our recordings were performed in the absence of receptor blockers, local network connectivity within the EC may sculpt the kinetics of synaptic depression, in parallel to purely presynaptic considerations. To conclude, differences that were originally observed in the young synapsin TKO tissue were lost by the time when seizures were firmly established. This argues against the acute participation of glutamatergic synaptic depression kinetics in seizure initiation and propagation, although a role in epileptogenesis cannot be ruled out.
Regulation of Tonic Inhibition in the Synapsin TKO Brain
Another example for late changes in neurotransmission properties is the magnitude of tonic inhibition. Tonic inhibition was quantified by measuring the effect of 50 μM Gabazine on the holding current of neurons that were voltage-clamped to 0 mV. The offset in the holding current is proportional to the blockage of high-affinity GABAA receptors, which mediate tonic inhibition (Farrant and Nusser 2005; Isaacson and Scanziani 2011). Gabazine produced a larger offset in slices from young TKO mice when compared with WT controls (n = 7/3 for each condition, P = 0.035 Kruskal–Wallis test, P = 0.009 young TKO vs. young WT, Mann–Whiney test, Fig. 6A,B). In contrast, no difference was found in the magnitude of tonic inhibition in adult TKO and WT neurons (P = 0.916), due to downregulation of tonic inhibition in the adult TKO slices (Fig. 6B). In the hippocampus, a decrease in tonic inhibition, rather than an increase, has been observed in synapsin TKO mice (Farisello et al. 2012). Therefore, we measured the mRNA levels of the α5 and δ subunits of the GABAA receptor, which are reported to mediate extrasynaptic tonic inhibitory currents (Brunig et al. 2002; Fritschy and Brunig 2003; Wei et al. 2003; Farrant and Nusser 2005; Goodkin and Kapur 2009). Quantification was performed by RT-PCR based on cDNA isolated specifically from the temporal and entorhinal cortices, excluding the hippocampus. While no difference was observed in the quantity of the GABAAδ subunit mRNA (P = 0.226 Kruskal–Wallis, n = 5 animals for each group), as was also reported in the hippocampus (Farisello et al. 2012), the level of the GABAAα5 subunit mRNA was significantly higher in the young TKO group (P = 0.014 Kruskal–Wallis, Fig. 6C), in full correlation with our electrophysiological recordings. mRNA levels of the GABAA α1, α2, β2, and γ2 subunits, which are primarily involved in phasic rather than tonic inhibition (Nusser et al. 1998; Rudolph et al. 1999), have been reported to be unaffected in the EC of synapsin TKO mice (Ketzef et al. 2011). To conclude, higher initial levels of tonic inhibition (Fig. 6B) reverted in the adult TKO to WT levels. It is of interest that tonic inhibition exhibited an opposite pattern compared with the phasic inhibition (Fig. 3), perhaps hinting of co-regulation.
Deletion of Synapsins Alters the Intrinsic Properties of Layer 5 Pyramidal Neurons in Adult TKO Mice
The results we presented so far relate to synaptic or perisynaptic properties, in line with the well-established function of the synapsins in the presynaptic terminal (Gitler et al. 2004). However, since our results illustrate the occurrence of long-term and delayed functional changes, we conjectured that nonsynaptic compensation may also occur. To test this hypothesis, we examined the intrinsic properties of EC layer 5 pyramidal neurons in presymptomatic 2–3-week-old mice and in epileptic mice older than 7 months of age (Fig. 7). The resting membrane potential and input resistance were unaffected irrespective of age or genotype (P = 0.583 and 0.722 Kruskal–Wallis test, respectively). The AP threshold, measured in response to a brief depolarization (see Methods section) was likewise unaffected (P = 0.167). In contrast, the AP amplitude decreased with age (P = 0.04 Mann–Whitney test for both genotypes), as has been previously reported in rats (McCormick and Prince 1987; Oswald and Reyes 2008) and in monkeys (Chang et al. 2005; Luebke and Chang 2007). This alteration is probably unrelated to deletion of the synapsins, since the effect was genotype-insensitive (P = 0.131 for young WT and TKO mice, P = 0.079 for adult mice, Mann–Whitney test). In contrast, parameters related to the AP waveform, that is, the AP half-width and the maximal slope of the AP upstroke were affected in a manner that was genotype-dependent. Although both parameters were similar in young mice of both genotypes (P = 0.934 and 0.280, respectively), in adult WT mice the AP duration was shorter by 30% compared with age-matched synapsin TKO mice, whereas the slope was larger by 34% (P = 0.002 and P = 0.003, Fig. 7A,C). Closer examination revealed that an age-dependent reduction in the AP duration that took place in WT mice (P < 0.001 and P = 0.014, and see McCormick and Prince 1987; Luebke and Chang 2007) did not take place in the synapsin TKO mice (P = 0.179 and 0.089). Because a difference in the AP waveform can affect the firing rate during AP bursts (Chang et al. 2005), we also examined the number and frequency of APs induced by a different stimulation protocol consisting of long-lasting depolarizing current injections. Although the number of APs was insensitive to age or genotype (P = 0.184, repeated-measures ANOVA), the 2nd to 3rd AP instantaneous frequency was significantly higher in adult WT slices compared with the other groups (P = 0.019 repeated-measures ANOVA, for stimulation intensities of 70 pA or more), in keeping with their shorter AP. To conclude, our results illustrate the occurrence of delayed alterations in a nonsynaptic property of neurons, which is directly relevant to neuronal excitability, and which could be a contributing factor for epilepsy in the adult synapsin TKO mice.
Mutations in the synapsin genes have been associated with familial epilepsy in humans (Garcia et al. 2004; Cavalleri et al. 2007; Lakhan et al. 2010), generating significant interest in mice in which various combinations of the 3 synapsin genes have been knocked out (Cesca et al. 2010). The epileptic mouse models supply an opportunity to experimentally probe the properties of neurons and of networks in vivo and in vitro, and have provided substantial information concerning their synaptic deficits. Deletion of the synapsins, even exclusively of synapsin I, causes a significant decrease in basal inhibitory transmission and in the amount of vesicles in inhibitory terminals (Gitler et al. 2004; Baldelli et al. 2007). Because reduced inhibition is expected to enhance brain excitability, this observation was proposed as the basis for the epileptic nature of the synapsin KO mice (Li et al. 1995; Terada et al. 1999; Gitler et al. 2004; Baldelli et al. 2007; Cesca et al. 2010). However, the onset of epilepsy in these mice is delayed for months, in apparent contradiction to the fact that the deficit in inhibitory transmission is evident already in the youngest mice examined. This is consistent with the delay in the onset of epilepsy which has been observed in other genetic models as well as in patients with known mutations (King and LaMotte 1989; Loiseau et al. 1998), suggesting that developmental changes in network-activity regimes may be crucial for defining the timing of the onset of seizures. Considering current progress in therapeutic sequencing and the advent of tailored personalized medicine, prior knowledge of the expected delay in the onset of genetic epilepsy may provide a critical window of opportunity to avert its onset in affected individuals. This realization stresses the utility of genetic animal models as test beds for probing the complex process of epileptogenesis.
Our results suggest that deletion of the synapsins initiates a complex cascade of events that involves direct but also secondary and compensatory changes in brain activity. To clarify which changes were related to the deletion of the synapsins, we characterized in parallel age-related alterations in synaptic and extrasynaptic properties in WT and synapsin TKO mice. Recordings were performed from layer 5 pyramidal neurons in the EC because this area is often involved in the initiation of seizures (Schwartzkroin and McIntyre 1997), but more importantly because there, rather than in the hippocampus, is where we have observed epileptiform activity in synapsin TKO slices in the absence of blockers (Ketzef et al. 2011).
Overall, the spontaneous inhibitory input onto L5 neurons decreased with age in WT mice, producing an increase in the E/I ratio (Fig. 3). Similar age-related alterations in WT animals have been reported previously in other higher brain areas, such as the auditory cortex (Caspary et al. 2008), the hippocampus (Stanley and Shetty 2004; Kanak et al. 2011), and the visual cortex (Leventhal et al. 2003), indicating that this process represents a normal trend, and is not in itself a contributing factor for epilepsy. Although we did not explore the mechanisms underlying this trend in WT brains, we note that it could be caused by a decrement in the population of inhibitory neurons (Stanley and Shetty 2004), by changes in the excitability or release probability of some inhibitory neurons, or by a weaker excitatory input onto them. The E/I balance has been shown to be involved in gain control and in the basic mechanisms underlying oscillatory waves (Isaacson and Scanziani 2011), and thus its disruption is expected to lead to severe dysfunction. Indeed, in synapsin TKO slices the E/I trend change was inverted. Spontaneous inhibitory transmission was initially weaker (Fig. 3), in agreement with previous studies illustrating reduced basal release capacity (Gitler et al. 2004; Baldelli et al. 2007). The absence of effect on mIPSC amplitudes at this age (Fig. 2) suggest either weaker excitatory input onto inhibitory neurons or a lower release probability of inhibitory cells (Baldelli et al. 2007). The E/I amplitude and frequency ratios equalized during the latent period and even reversed after the emergence of seizures, resulting in a relatively stronger inhibitory tone in adult seizing mice, reminiscent of that observed in the case of the epileptic Kcna-1 KO mice (van Brederode et al. 2001). In parallel, we observed a transient increase in the excitatory mEPSC amplitude when mice start experiencing seizures, which declined later (Fig. 2). An intriguing possibility is that homeostatic mechanisms (Turrigiano 2011), which have been observed also in relation to inhibitory transmission (Lau and Murthy 2012), differentially modify activity in WT and TKO brains due to their divergent initial conditions; we cannot determine whether this is the case, but note that in slices from adult mice the amplitude of mIPSC in the TKO slices was eventually larger than in WT ones (Fig. 2), in agreement with postsynaptic secondary changes. To summarize, our key finding indicates that phasic inhibitory transmission is significantly enhanced rather than reduced in the EC by the time that epilepsy is established (Fig. 3). It is therefore unlikely that a deficit in phasic inhibitory transmission is directly responsible for epilepsy in the synapsin TKO mice, as was previously postulated, although it could participate in its genesis. Overall our results illustrate significant and complex alterations in the course of normal age-dependent reorganization of inhibitory transmission. The observation that inhibition is enhanced, together with the lack of difference in the number of inhibitory neurons in adult TKO and WT brain slices (Ketzef et al. 2011) hints that the intrinsic excitability of inhibitory neurons or the excitatory input onto them are enhanced. This hypothesis should be confirmed in future studies by direct recordings from well-defined inhibitory neurons.
In contrast to the situation observed for phasic inhibition, tonic inhibition was initially stronger in slices from young TKO mice and weakened to WT values in the adults. This observation is congruent with changes in the mRNA levels of the α5 GABAA receptor subunit that participates in tonic inhibition (Fig. 6). It is possible that tonic extrasynaptic inhibition in young TKO mice compensates for weak phasic inhibition, thus postponing the initiation of seizures. Following the same logic, tonic inhibition is down-regulated in the adult, as phasic inhibition increases (Fig. 3), but in a manner that still allows the induction of seizures. Nevertheless, we note that the increase in phasic inhibition may be compensatory in nature, in agreement with the decline in seizure propensity after its peak at the age of 6 months (Supplementary Fig. 3). An alternative possibility is that enhanced inhibition (van Brederode et al. 2001) may promote epilepsy by synchronizing network activity (Avoli and de Curtis 2011) making the network epilepsy-prone. At this time, we cannot distinguish between these alternatives.
Another example of late-onset changes in synaptic properties relates to synaptic depression (Fig. 5). Depression of excitatory synapses was initially faster in TKO slices compared with WT slices. This difference was lost in the adult, because synaptic depression in WT animals accelerated with age, reaching the same rate originally observed in the TKO slices. One of the mechanisms that determine the kinetics of synaptic depression is the availability of reserve vesicles (Pan and Zucker 2009) and has been suggested to be the basis for the faster depression in young TKO neurons (Gitler et al. 2004). Although changes in vesicle populations have been described during synapse maturation in culture (Mozhayeva et al. 2002), the time frame of these changes was limited to very young neurons, and it remains to be determined whether in vivo the division of vesicles into pools changes with age in a relevant manner. Alternatively, synaptic depression may be sculpted by the probability of release, with higher release probability leading to faster depression. Our observation that excitatory evoked responses were not enlarged with age in WT slices (Fig. 1) does not support this explanation. More importantly, because the kinetics of synaptic depression in the adult WT equalized to those measured in adult TKO epileptic mice, this parameter is unlikely to contribute directly to the generation of seizures, but could participate in the alteration of network properties that promotes the epileptic condition.
Clear evidence for secondary alterations in activity patterns in the synapsin TKO mice came from our observation that AP properties, which were initially similar in neurons of young mice, diverged in the adults. Apparently, the APs of the adult TKO neurons are longer (Fig. 7) because an age-dependent reduction in duration that occurs in WT neurons failed to take place in the TKOs. This observation is significant because longer APs enhance synaptic transmission (Wheeler et al. 1996; Shu et al. 2006; Ali et al. 2007; Kole et al. 2007), a factor that can contribute to more prominent synaptic activity, and potentially to epilepsy. We propose that the early synaptic defect activates varied plasticity events that indirectly alter other key nonsynaptic neuronal properties by an as-yet-undetermined mechanism.
It is probable that some of the adjustments we describe are compensatory in nature, in view of the lower ictal activity observed in frankly old TKO mice (Boido et al. 2010) and the reduced seizure susceptibility in adult TKO mice (Supplementary Fig. 3, and Cambiaghi et al. 2012). Stronger phasic inhibition (Fig. 3), a shift in the E/I ratio of the responses upon 4-AP application (Fig. 4), and a lower instantaneous AP frequency (Fig. 7) are arguably consistent with compensation. However, at this point we cannot rule which events are compensatory and oppose epilepsy, and which are secondary alterations that promote it. On the basis of the varied roles played by different subpopulations of interneurons (Isaacson and Scanziani 2011), it is conceivable that specific alterations in the function of subgroups of interneurons could affect network activity in considerably dissimilar ways. Future studies that attempt to separately manipulate the various components in isolation may answer this important question.
Although epileptiform activity has been recorded in the hippocampus of synapsin TKO mice (Boido et al. 2010), this was performed in the presence of the convulsant 4-AP. A later study examined the synaptic and extrasynaptic alterations in the TKO hippocampus (Farisello et al. 2012). These studies differed from our own in several aspects. First, the impairment in inhibition in the hippocampus persisted into adulthood (∼6 months of age). In contrast to our observation of transiently stronger tonic inhibition in the EC (Fig. 6), in the hippocampus tonic inhibition was weaker regardless of age. Finally, in the hippocampus the distribution of the GABAA α5 subunit was unaffected by deletion of synapsins, in contrast to our observations in the cortex (Fig. 6), and unlike studies in other animal epilepsy models (Houser and Esclapez 2003; Lauren et al. 2003; Zhang et al. 2007; Goodkin and Kapur 2009) and in human patients (Brooks-Kayal et al. 1999; Porter et al. 2005; Loup et al. 2006; de Moura et al. 2010) that reported changes in other GABAA receptor subunits. The differences between the results obtained from the hippocampus and the EC could partially be due to the sparing of local recurrent connections in the horizontal cortico-hippocampal slice preparation that allow propagation of synchronized activity in the EC. Therefore, seizure activity may entail interplay between reciprocally interconnected cortical and subcortical areas, as has been previously proposed (Jones and Woodhall 2005; Etholm et al. 2011; Farisello et al. 2012).
To conclude, our study concerning epilepsy in the synapsin TKO mice reveals a complex and long-term pathological progression involving multiple parallel pathways. Some of these are synaptic and can be discussed in the context of the original presynaptic lesion, but others are clearly unrelated. These reflect secondary plasticity events, which could be classified as aberrant, on the one hand, or compensatory, on the other. A more complete understanding of the underlying mechanisms may assist in directed targeting of therapeutic agents aimed at alleviating or even avoiding epilepsy. In this context, it is crucial to determine how treatments that manipulate GABAergic transmission (Treiman 2001) affect the epileptic condition during its different stages. Does strengthening GABAergic transmission in the latent phase delay the onset of seizures? Does GABAergic transmission exacerbate the manifestation of epilepsy in the adult? If so, does reducing inhibition paradoxically lessen the symptoms? The second theme that emerges from our study is the time scale of the effect of the deletion of the synapsins. We observed that a congenital synaptic defect progressively affects network properties well into adulthood, indicating that the cascade of actions and reactions persists and develops, before and also after a clinically relevant condition is evident. Delineation of the functional and temporal links between the various involved systems is therefore necessary in order to more fully understand epileptogenesis in general, and specifically in this model.
Supplementary material can be found at: http://www.cercor.oxfordjournals.org/.
This work was supported by grant 76-08-09 from the National Institute for Psychobiology in Israel, grant 3-0000-6217 from the Chief Scientist of the Ministry of Health of Israel, and the Rich Foundation for Education, Culture and Welfare, Luzern Switzerland. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
We thank Professors Ilya Fleidervish and Alon Friedman for the critical reading of the manuscript and for their professional advice throughout the project. Conflict of Interest: None declared.