Altered neurotransmitter release, vesicle recycling and presynaptic structure in the pilocarpine model of temporal lobe epilepsy

Abstract

In searching for persistent seizure-induced alterations in brain function that might be causally related to epilepsy, presynaptic transmitter release has relatively been neglected. To measure directly the long-term effects of pilocarpine-induced status epilepticus on vesicular release and recycling in hippocampal mossy fibre presynaptic boutons, we used (i) two-photon imaging of FM1-43 vesicular release in rat hippocampal slices; and (ii) transgenic mice expressing the genetically encoded pH-sensitive fluorescent reporter synaptopHluorin preferentially at glutamatergic synapses. In this study we found that, 1–2 months after pilocarpine-induced status epilepticus, there were significant increases in mossy fibre bouton size, faster rates of action potential-driven vesicular release and endocytosis. We also analysed the ultrastructure of rat mossy fibre boutons using transmission electron microscopy. Pilocarpine-induced status epilepticus led to a significant increase in the number of release sites, active zone length, postsynaptic density area and number of vesicles in the readily releasable and recycling pools, all correlated with increased release probability. Our data show that presynaptic release machinery is persistently altered in structure and function by status epilepticus, which could contribute to the development of the chronic epileptic state and may represent a potential new target for antiepileptic therapies.

Introduction

Hippocampal mossy fibres exhibit complex structural rearrangements and abnormal synaptic function in mesial temporal lobe epilepsy (McNamara, 1994; Frotscher et al., 2006; Dudek and Sutula, 2007; Danzer et al., 2010). During epileptogenesis, axons of dentate gyrus granule cells sprout and establish abnormal excitatory synapses onto multiple targets, including granule cells, interneurons and CA3 pyramidal neurons (McNamara, 1994; Wenzel et al., 2000; Frotscher et al., 2006; Dudek and Sutula, 2007). While studies have shown alterations in postsynaptic function of excitatory and inhibitory circuits in mesial temporal lobe epilepsy (Coulter, 2000; Gorter et al., 2002), far less attention has been paid to presynaptic dysfunction in the pathogenesis of epilepsy.

Large mossy fibre boutons (MFBs) have multiple release sites with substantial numbers of small and large clear synaptic vesicles containing glutamate (Chicurel and Harris, 1992; Rollenhagen et al., 2007). Large synaptic vesicles are thought to mediate the monoquantal nature of ‘giant’ miniature excitatory postsynaptic currents onto CA3 pyramidal cells, allowing action potential invasion of presynaptic MFBs to trigger neurotransmitter release in amounts that reliably excite CA3 pyramidal neurons. During status epilepticus, excessive glutamate release from MFBs can be highly deleterious to survival of CA3 neurons and is thought to be responsible for the massive and sustained neuronal damage in the CA3 region (Zhang et al., 2009).

Surprisingly little is known about the long-term effects of status epilepticus on vesicular release of transmitter from MFBs. Goussakov et al. (2000) found that, several weeks after kainic acid-induced status epilepticus, mossy fibre, but not associational-commissural, synapses exhibit impaired induction of long-term potentiation, markedly reduced paired-pulse facilitation and augmentation of burst induced mossy fibre-CA3 excitatory field potential (Goussakov et al., 2000). These changes in activity-dependent plasticity were also associated with a significant increase in the size of the readily releasable pool (RRP) of vesicles, suggesting a persistent increase in release probability at mossy fibre–CA3 synapses (Goussakov et al., 2000). Ultrastructural features and targets of abnormally sprouted MFBs in the inner molecular layer have been studied by electron microscopy in models of mesial temporal lobe epilepsy (Cavazos et al., 2003; Frotscher et al., 2006); indeed, redistribution of synaptic vesicles to a ‘strategic position’ in the vicinity of the synaptic cleft has been described after kindled seizures (Hovorka et al., 1989), but these studies did not characterize the arrangement of active zones and morphological features of vesicle pools.

We used the cholinergic agonist pilocarpine to induce prolonged status epilepticus as a model of mesial temporal lobe epilepsy (Turski et al., 1984; Curia et al., 2008), to test the hypothesis that status epilepticus can result in long-term enhancement of vesicular release from MFBs. Assessing presynaptic release properties by patch-clamp electrophysiology in long-term epileptic animals is hampered by the CA3 pyramidal neurons being susceptible to cell death due to excitotoxicity, thus are not optimal to record from as biosensors of transmitter release. Here we utilized two different methods to directly image presynaptic vesicular release by using two-photon laser scanning microscopy of (i) the styryl dye FM1-43 (Pyle et al., 1999), which selectively labels transmitter vesicles for direct visualization of destaining kinetics, in slices from control versus post-status epilepticus rats; and (ii) imaging exocytosis–endocytosis kinetics from individual MFBs in acute hippocampal slices from transgenic mice that selectively express the pH-sensitive vesicle fusion protein synaptopHluorin (SpH) (Li et al., 2005). We also used transmission electron microscopy of rat hippocampus to investigate whether the post-status epilepticus period is associated with ultrastructural modifications at active zones of MFBs in the stratum lucidum correlated with functional changes in vesicular release (Dobrunz and Stevens, 1997; Murthy et al., 1997; Schikorski and Stevens, 1997; Matz et al., 2010).

By each of these measures, we found a persistent enhancement of vesicular release, vesicle endocytosis and ultrastructure reorganization of active zones in MFBs after status epilepticus. These changes are likely to add to observed changes in intrinsic excitability of granule cells and circuit plasticity to increase the excitatory drive from granule cells onto CA3 pyramidal neurons in MTLE.

Materials and methods

Experiments were performed in accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals as approved by New York Medical College and University of Texas Brownsville Institutional Animal Care and Use Committees.

Pilocarpine-induced status epilepticus

Status epilepticus was induced by a single pilocarpine intraperitoneal injection (350 mg/kg in saline) to 30- to 35-day-old Sprague–Dawley rats (Taconic) or SpH21 mice (C57BL/6J background, kind gift from Dr Venkatesh Murthy, Harvard University) following standard procedures (Cavalheiro, 1995; Pacheco et al., 2008). Animals received methylscopolamine nitrate (0.1 mg/kg in saline, subcutaneous) 30 min before pilocarpine to minimize peripheral effects of cholinergic stimulation (Turski et al., 1984). Rats seized for 3 h, mice 1 h (Muller et al., 2009), and motor seizures were terminated with one diazepam injection (10 mg/kg in saline, intraperitoneal). Only animals experiencing continuous status epilepticus during these periods were studied. Controls animals received methylscopolamine and saline instead of pilocarpine.

Hippocampal slice preparation and electrophysiological recordings

Rats and mice 1–2 months post-status epilepticus and age-matched controls were decapitated under deep isoflurane anaesthesia, their brains quickly removed, hemisected in ice-cold sucrose-based cutting solution (Bischofberger et al., 2006) in mM: 87 NaCl, 25 NaHCO₃, 25 glucose, 75 sucrose, 2.5 KCl, 1.25 NaH₂PO₄, 0.5 CaCl₂ and 7 MgCl₂ (equilibrated with 95% O₂/5% CO₂), and 350–400 µm thick horizontal hippocampal slices cut with a vibratome (DSK model DTK-1000). Slices were submerged in cutting solution at 32°C for 30 min, then transferred to a holding chamber in room temperature artificial CSF (in mM: 126 NaCl, 3 KCl, 1.25 NaH₂PO₄, 1.3 MgCl, 2.5 CaCl₂, 26 NaHCO₃, 10 glucose) saturated with 95%O₂/5%CO₂. Evoked field excitatory postsynaptic potentials were recorded in the proximal dendritic field in stratum lucidum for at least 15 min before starting an experiment.

Two-photon laser scanning microscopy

Fluorescence was visualized with a customized two-photon laser-scanning Olympus BX61WI microscope with a ×60/0.90 NA water immersion infrared objective lens and an Olympus multispectral confocal laser scan unit. The light source was a Mai-Tai™ laser (Solid-State Laser Co.), tuned to 825 nm for exciting Alexa Fluor 594, and 890 nm for exciting SpH and FM1-43. Epifluorescence was detected with photomultiplier tubes of the confocal laser scan head with pinhole maximally open and emission spectral window optimized for signal over background. A 565-nm dichroic mirror (Chroma Technology) separated green and red fluorescence to eliminate transmitted or reflected excitation light. Depending on the fluorophore, HQ525/50 (SpH) or HQ605/50 (FM1-43 or Alexa Fluor 594) filters were placed in the ‘green’ pathways. Images were acquired with Fluoview FV300 software (Olympus America). Although there were no signs of photodamage, we used the lowest intensity needed for adequate signal-to-noise ratios. Pixel images (512 × 512) were acquired, 0.15 µm/pixel in x−y axes and, in some cases, 8–10 µm steps of 0.5–1 µm in the z-axis. For imaging MFBs, glass microelectrodes with broken tips (R < 1 MΩ) were dipped in 1% Alexa Fluor 594 10-kDa dextran (Invitrogen) in distilled water until the tip filled by capillarity action, desiccated for 2–3 days until dye formed a crystal particle at the tip. During the experiment, dye coated tips (with sealed backends) were inserted ∼50–100 µm into the dentate gyrus granule cell layer for 15–20 min, then slices were transferred to a holding chamber in artificial CSF at 32°C for 30 min, followed by room temperature for 1–1.5 h, before imaging. Using NIH ImageJ, greyscale images of Alexa Fluor 594 filled MFBs flanked by at least one axonal process were processed by automated Otsu thresholding to eliminate investigator bias (Lee et al., 2011), and cross-sectional area of binary images measured. A total of 8–10 images were acquired in a z-axis stack in 0.2–0.5 µm steps for 3D reconstruction (Supplementary Video 3). For SpH fluorescence experiments, slices were perfused with 25 µM 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX; Tocris) and 50 µM D-(-)-2-Amino-5-phosphonopentanoic acid (D-AP5; Tocris). A 600 stimulus 20-Hz train in the stratum lucidum evoked SpH fluorescence increases (imaged within 100 µm from stimulating electrode) in the proximal region (the first 100 µm) of CA3 apical dendrites and images were acquired every 30 s. While extracellular brain pH changes during seizures, first to alkaline and later to acidic (Xiong and Stringer, 2000), we use SpH as a pH-sensitive indicator of vesicular release in brain slices bathed in artificial CSF (containing 25 µM CNQX and 50 µM D-AP5 to prevent synaptically driven action potentials and epileptiform activity) in our recording chamber with pH maintained at or ∼7.4. To calculate half-time of post-stimulus decay of SpH peak fluorescence intensity, (t_1/2) was calculated for each punctum by single exponential fits to destaining curves using the equation: y₀ + A₁e^(−x/t₁⁾, with t₁ the decay constant. Data were analysed with OriginPro and presented with KaleidaGraph (Synergy software).

FM1-43 loading and destaining of the readily releasable vesicle pool

FM1-43 was loaded into the RRP as previously described (Stanton et al., 2003, 2005). After recording a 15 min stable baseline field excitatory postsynaptic potential (∼50% of maximum amplitude), 25 µM CNQX and 50 µM D-AP5 were bath applied to prevent synaptically driven action potentials and epileptiform activity from accelerating dye release or inducing synaptic plasticity. Presynaptic boutons were loaded by 5-min bath applied 5 µM FM1-43 in normal artificial CSF, followed by hypertonic artificial CSF (800 mOsm) for 30 s to load the RRP, returned to normal artificial CSF with 5 µm FM1-43 for 1 min, then washed out with dye-free artificial CSF (Supplementary Fig. 2). Stimulus-induced destaining was evoked by a 50 stimulus 20-Hz train each 30 s, and measured after 30 min in dye-free artificial CSF plus 100 µM ADVASEP-7 to scavenge and remove extracellular FM1-43. To measure spontaneous release, after FM1-43 loading, slices were continuously bathed in 1 µM tetrodotoxin, 25 µm CNQX and 50 µM D-AP5 and extracellular [K⁺] raised to 15 mM [K⁺]o (High[K+]o) to facilitate action potential-independent release (Axmacher et al., 2004). For some slices, 8–10 images were acquired in 0.5–1 µm in the z-axis at each time point, maximal intensity z-axis projections were made for this subset of the final stack. Circular regions of interest were selected around centres of bright, punctate fluorescence spots at least 2 µm in diameter within 100 µm of the CA3 pyramidal cell body layer. Activity-dependent destaining time courses were generated by correcting for bleaching, subtracting background fluorescence and normalizing each region of interest to pre-stimulus intensity as previously reported (Stanton et al., 2003, 2005; Tyler et al., 2006). Vertical bars show SEMs for all normalized and corrected boutons.

Tissue preparation for immunohistochemistry

Age-matched control and epileptic SpH mice (<6 months post-status epilepticus) were anaesthetized with 100 mg/kg ketamine, the brain removed and immersed for 48 h in 4% formaldehyde in 0.1 M sodium phosphate buffer (pH 7.4) plus 0.9% NaCl (phosphate-buffered saline). After a phosphate-buffered saline wash, brains were cryoprotected in 30% sucrose in phosphate-buffered saline for ∼48 h at 4°C, frozen in Optimum Cutting Temperature Compound (EMS Co.) at −20°C, and 40 μm thick coronal sections sliced with a cryostat (HM500 Microm, Zeiss) and stored in 150 mM Tris plus 150 mM NaCl (Tris-buffered saline; pH 7.4) and 0.1% NaN₃. Several sections were Nissl-stained to assess seizure-mediated neuronal loss and cytoarchitectonic boundaries. Free-floating sections from controls and post-status epilepticus were processed simultaneously for VGluT1 and NeuN immunofluorescence as described (Pacheco et al., 2008). The extent of cell loss in the CA3 subfield was assessed at ×400 magnification by using grid morphometric techniques (Gorter et al., 2003). A 240 × 240 µm box was placed over the region of interest (CA3 area) and NeuN-labelled cells in the pyramidal layer, presumed neurons, were counted by an observer masked to experimental group/condition. Counts of three sections per animal were averaged. Final neuron counts were calculated as the density of neurons per 0.1 mm². It should be noted that this technique provides a relative estimate and not absolute calculation of the number of neurons in a region. Statistical differences in the number of neurons per square millimetre between control and pilocarpine-treated epileptic groups were determined by Student's t-test with significance set at P < 0.05. The cross-sectional diameter of SpH-positive VGluT1-labelled puncta in the granule cell layer and supragranular region was measured in three confocal images of dentate gyrus per animal per group using the measuring tool of Fluoview (Olympus).

Fixation and embedding for transmission electron microscopy

Tissue was prepared for transmission electron microscopy using standard methods (Miranda et al., 2011). Seven post-status epilepticus and six age-matched Sprague–Dawley rats were anaesthetized with ketamine (50 mg/kg, intraperitoneal injection), transcardially perfused with saline followed by fixative (2% paraformaldehyde, 2% glutaraldehyde, 0.002% CaCl₂ in 0.1 M cacodylate and Hank's buffered salt solution; pH 7.3–7.4), and their brains were removed, sliced in 5-mm thick slabs, and post-fixed in fresh fixative overnight at room temperature. Serial 250 -μm thick sections were cut with a vibratome in Hank's buffered salt solution (1000 Plus, Vibratome Co.), and the hippocampus and surrounding cortex dissected away. Samples were osmicated in 2% buffered osmium tetroxide, a 50 : 50 mixture of 4% osmium tetroxide and 4% potassium ferrocyanide in 0.2 M cacodylate buffer (EMS Co.), then contrasted in 1% aqueous uranyl acetate. Tissue was flat-embedded in 50 : 50 Spurr/Epon (EMS Co.) and polymerized at 60°C for 24 h before resin blocks were pseudo-randomly selected and trimmed to ∼1 mm³ trapezoids containing the CA3 pyramidal cell layer and stratum lucidum. To identify field CA3, 1–2 μm sections were stained with 0.1% cresyl violet acetate solution and viewed on a transmitted light scope. Ultrathin sections (55 ± 5 nm) were cut through stratum lucidum of CA3 with a Leica Ultracut ultramicrotome and collected on Formvar-coated copper slotted grids.

Transmission electron microscopy inspection and identification of mossy fibre boutons

Images 26 000× (FEI Tecnai) were used to identify regions of interest in field CA3 containing cross-sectioned or parallel bundles of axons and large MFBs (1–5 µm diameter) with numerous synaptic vesicles. For blind analysis of morphometry, coded images were imported into NIH ImageJ. Active zones were distinguished by the following criteria: (i) synaptic vesicles in close proximity to a presynaptic density; (ii) asymmetry between pre- and postsynaptic densities; and (iii) widening of the synaptic cleft (Rollenhagen et al., 2007). Asymmetric (excitatory) non-perforated synapses with well-defined presynaptic compartments, clear synaptic clefts and postsynaptic density were analysed. Perforated synapses refer to a presumably more efficacious synapse subtype found in the CNS characterized by a discontinuous postsynaptic density (Calverley and Jones, 1987; Itarat and Jones, 1992; Adkins et al., 2008). The number of perforated and non-perforated synapses was compared between groups

Active zone length was measured by a continuous line following the contour of presynaptic membrane on the synaptic cleft. Synaptic cleft width was averaged from three measures of distance between pre- and postsynaptic membranes (centre and ∼20 nm from cleft ends). Postsynaptic density area was measured as the contour around electron-dense aggregates and postsynaptic membrane. The active zone length and quantification of synaptic vesicles was performed in single ultrathin sections using a modified 2D cross-sectional method as described previously (Miranda et al., 2011) using NIH ImageJ.

The RRP was defined as the number of synaptic vesicles <60 nm from the presynaptic membrane, and the releasable pool of vesicles 60–200 nm from the presynaptic membrane (Rollenhagen et al., 2007). Docked synaptic vesicles consisted of vesicles immediately adjacent to or fused to the active zone presynaptic membrane. Perimeter and mean area of synaptic vesicles were also measured, and number of active zones exhibiting ultrastructural signs of bulk endocytosis indicative of intense synaptic activity (Meunier et al., 2010).

Differences in vesicle pool size were assessed by non-parametric Kolmogorov–Smirnov two-sample test (MiniAnalysis, Synaptosoft), and differences in means of normally distributed measurements were evaluated using two-tailed Student's t-test. Significance level was preset to P < 0.05. Correlation analysis between parameters was performed with non-parametric Spearman analysis, since most parameters were not normally distributed. Correlations were classified as weak [Spearman rho (r) value < 0.40], moderate (0.4 < r < 0.7) and strong (r > 0.7).

Results

Two-photon laser scanning microscopy imaging of large mossy fibre boutons of synaptopHluorin-expressing transgenic mice in acute hippocampal slices

SpH is a fusion protein that consists of a pH-sensitive pHluorin fused to the C-terminus luminal domain of the vesicle SNARE protein synaptobrevin (Miesenbock et al., 1998). Under resting conditions, the luminal pH of the synaptic vesicle is acidic (pH ∼5.5), resulting in proton dependent quenching of SpH fluorescence. When vesicle exocytosis is triggered and glutamate released, the lumen of the vesicle is exposed to the more alkaline pH of the extracellular space (pH ∼7.2), resulting in a 20-fold increase in SpH fluorescence. When the vesicle membrane is retrieved by endocytosis and the vesicle reformed, it undergoes rapid reacidification by the vesicular ATPase, which returns SpH to its quenched state. The SpH21 transgenic mice line we used in this study expresses SpH preferentially at glutamatergic synapses (Li et al., 2005; Burrone et al., 2006; Bayazitov et al., 2007). A representative two-photon laser scanning image of the CA3 region of an acute hippocampal slice from this mouse (Fig. 1A) contains bright GFP-positive boutons >2 µm in diameter, proximal to CA3 pyramidal cell bodies in the region innervated by mossy fibre axons of dentate granule cells (Blackstad et al., 1970; Claiborne et al., 1986). Associational-commissural CA3 synapses were significantly smaller (<1 µm in diameter) and more distal to our region of interest (rectangular box, Fig. 1A).

Figure 1.

Visualizing MFBs in acute hippocampal slices from SpH21 transgenic mice. (A) Live cell two-photon laser scanning image of field CA3 of SpH-expressing glutamatergic terminals in a control hippocampal slice (post natal 60 days of age). The arrows depict representative fluorescence puncta (∼4–5 µm in diameter) in the proximal (stratum lucidum, Str. Luc.) region of CA3 pyramidal cells (seen here as a ghost layer) that are likely to be giant MFBs. Distal puncta shown within the rectangular box are notably smaller in size and likely represent associational-commissural synapses in stratum radiatum. (B) Schematic representing hippocampal circuitry with the circle in the mossy fibre pathway (MFP) depicting the area of imaging Lightning symbol depicts the area of local stimulation and the pipette in the dentate granule cell layer (DG) showing the area of Alexa Fluor 594 dextra containing pipette insertion. Squares and pentagons depict CA3 and CA1 pyramidal cell layers, respectively. (C) Alexa Fluor 594 dextran filled mossy fibre axons and giant MFBs (arrows) showing characteristic en-passant arrangement along the axonal projections, visualized using 825 nm excitation. (D) Same region as (C) but visualized with 890-nm excitation to see SpH native fluorescence. CA3 pyramidal cell bodies can be seen as ghost cells in stratum pyramidal, and bright giant MFBs in stratum lucidum. (E–F) Merged images of (C) and (D) showing co-localization (F: arrows) between SpH and Alexa Fluor 594 dextran-labelled puncta. (F) Digital zoom of image from the inset box in (E). Note not all SpH-positive puncta co-localize with Alexa Fluor 594 puncta (broken arrows). PP = perforant path; SC = Schaffer collaterals; Str. Ori. = stratum oriens; Str. Pyr. = stratum pyramidale. Scale bars: A = 20 µm; C–E = 20 µm; and F = 4 µm.

To independently confirm that large SpH expressing boutons were mossy fibre terminals, we used anterograde, bulk labelling of mossy fibres with Alexa Fluor 594 dextran introduced into the granule cell layer of the dentate gyrus (Fig. 1B). During a 1.5- to 2-h incubation, Alexa Fluor 594 dextran was taken up by granule cells and transported anterogradely to label mossy fibre axons and presynaptic boutons (Fig. 1C, Supplementary Fig. 1A and Supplementary Video 3). Figure 1E and F illustrate that some (solid arrows) but not all (broken arrows) SpH and Alexa Fluor 594-positive boutons were co-labelled confirming that the large (>2 µm in diameter) SpH fluorescent puncta within the proximal 60 µm of the CA3 cell body are excitatory MFBs (Fig. 1D). Incomplete co-localization is likely because Alexa Fluor 594 labelled a subset of mossy fibre axons in stratum lucidum, along with sparse SpH expression in the excitatory terminals (Li et al., 2005; Burrone et al., 2006).

Pilocarpine-induced status epilepticus causes disorganization of dentate gyrus and CA3 cytoarchitecture in synaptopHluorin-expressing transgenic mice

In the hippocampal CA3 region of both control and post-status epilepticus mice (Fig. 2A), immunostaining for the presynaptic glutamate transporter VGluT1 was intense in the mossy fibre projection area in stratum lucidum, where it co-localized with native SpH fluorescence, consistent with the glutamatergic nature of presynaptic terminals expressing SpH in SpH21 mice (Li et al., 2005; Burrone et al., 2006). There was also ∼51% reduction in CA3 stratum pyramidale of post-status epilepticus mice in the density of neurons (117.9 ± 35.4 cells/0.1 mm², n = 6) that stained positively for the neuronal marker NeuN, compared with age-matched controls (234.4 ± 55.2 cells/0.1 mm², n = 6, Student's t-test, P < 0.01, Fig. 2A), consistent with seizure-induced loss of CA3 pyramidal neurons following pilocarpine-induced status epilepticus reported previously (Mello et al., 1993; Schauwecker, 2012).

Figure 2.

Confocal images of SpH, VGluT1 and NeuN expression in CA3 (A) and dentate granule cell (B) layers of control and post-status epilepticus (SE) SpH transgenic mice. (A) Native SpH fluorescence image (green), VGluT1 immunofluorescence image (red) and NeuN immunofluorescence image (purple) in control and post-status epilepticus mice showing co-localization of SpH with VgluT1 labelling in MFBs. A dramatic loss of cells was noticed in CA3 stratum pyramidale by NeuN staining (SpH + VGluT1 + NeuN, column 3) and SpH and VGluT1-positive terminals appear disorganized in stratum lucidum of post-status epilepticus compared with controls (SpH + VGluT1 + NeuN, column 4). Punctate SpH signals were also scattered through the pyramidal cell layer (s.p) surrounding cell somata that were stained with NeuN antibody. The column on the right (column 4) shows fluorescence intensity profile and co-localization along line scans (1: control and 2: post-status epilepticus) through single MFBs indicated in the respective ×2 magnification images (scale bar = 25 µm) of the rectangular boxes seen in column 3 (scale bar = 50 µm). Scale bar for line scans = 2 µm. (B) In the dentate granule cell layer of control and post-status epilepticus, (NeuN labelling), SpH fluorescence (green) and VGluT1 immunofluorescence (red) were intense in the hilus (asterisk) and in the inner molecular layer (arrow head). In the control (B, top row) SpH-positive puncta around putative granule cell somata were devoid of VGluT1 expression (line plots: lines 1 and 2), while SpH and VGluT1 signals co-localized in the inner molecular layer (line plots: lines 3 and 4), as shown by respective line scans of fluorescence intensity through puncta in the granule cell layer (1 and 2) versus the inner molecular layer (3 and 4). In post-status epilepticus (B, bottom row) SpH and VGluT1 co-localize in both the dentate granule cell layer (SpH + VGluT1) and inner molecular layer. NeuN immunofluorescence illustrates a lack of marked granule cell loss in post-status epilepticus dentate gyrus (SpH + VGluT1 + NeuN). Stratum oriens scale bar = 2 µm.

In the dentate gyrus, we detected robust SpH fluorescence in the inner molecular layer and hilus of both control and post-status epilepticus mice (Fig. 2B), closely co-localized with VGluT1 immunofluorescence (Fig. 2B). VGluT1 puncta positive for SpH fluorescence in the inner molecular layer exhibited 39.5% larger cross-sectional diameters (Fig. 2B, 1.84 ± 0.54 µm, Student's t-test, P < 0.001) and 47.5% higher normalized mean fluorescence intensity (Fig. 2B, post-status epilepticus: VGluT1) relative to background-subtracted baseline intensity [204.8 ± 35.5 arbitrary units (a.u), n = 29, Student's t-test, P < 0.0001] in pilocarpine-treated post-status epilepticus mice compared with controls (Fig. 2B, control: VGluT1 138.8 ± 32.2 a.u and mean diameter of puncta: 1.31 ± 0.37 µm, n = 29). Interestingly, SpH puncta in granule cell somata in the control dentate gyri exhibited little immunostaining for VGluT1, whereas the granule cell layers of post-status epilepticus hippocampi showed strong co-immunolabelling for VGluT1 (Fig. 2B), suggesting either a possible upregulation of VGluT1 expression in existing terminals or neo-synaptogenesis manifested by the appearance of ectopic glutamatergic (VGluT1 positive) terminals in both inner molecular and granule cell layers during epileptogenesis, as previously reported in animal models of MTLE (Mello et al., 1993; Esclapez et al., 1999; Epsztein et al., 2005; Pacheco et al., 2006; Boulland et al., 2007).

Pilocarpine-induced status epilepticus persistently increases size and vesicular release rate of mossy fibre boutons in synaptopHluorin-expressing mice

Live cell imaging of MFBs in acute hippocampal slices was done by bulk loading a group of granule cells and their axons with Alexa Fluor 594-dextran. Dye-filled excrescences (Fig. 1B) were classified as the main body of giant MFBs if they had at least a 4 µm² cross-sectional area (Claiborne et al., 1986; Acsady et al., 1998; Danzer et al., 2010). The mean area of MFBs 1–2 months after pilocarpine induced-status epilepticus was significantly enhanced (∼21.09%, Fig. 3C, P = 0.008, Mann–Whitney U-test, n = 7) compared with aged matched controls (n = 7). A frequency distribution histogram of individual MFBs from epileptic animals revealed a significant rightward shift in the curves (Fig. 3B; P < 0.05, Kolmogorov–Smirnov test), consistent with increased MFB volumes.

Figure 3.

Pilocarpine-induced status epilepticus (SE) leads to persistent increases in dentate gyrus giant MFB area. (A) Live cell two-photon images of Alexa Fluor 594 dextran-loaded giant mossy fibre terminals from field CA3 of acute hippocampal slices from control and post-status epilepticus mice. The arrow shows a filopodia-like projection arising from the giant MFB. (B) Frequency distribution histogram of individual MFB areas from control (black columns, total of 336 boutons, n = 7) vs. 1–2 months post-SE mice (red columns, 394 boutons, n = 7). (C) Mean MFB area for control (black column, 9.05 ± 0.52 µm²) and epileptic (red column, 11.47 ± 0.48 µm²) mice. Data plotted as mean ± SEM, *P < 0.05, Mann–Whitney U-test. Scale bar A = 5 µm.

To determine whether post-status epilepticus leads to functional differences in transmitter release from excitatory MFBs, we utilized two-photon live cell imaging in slices from SpH21 mice. To trigger vesicular release from MFBs in CA3 stratum lucidum (Fig. 1B–F), granule cell axons in slices from control and post-status epilepticus animals were stimulated with a single, continuous train of 600 stimuli at 20 Hz (Fig. 1B and Supplementary Video 4A and B, respectively), a paradigm that recruits both the RRP and rapidly recycling vesicle pools (Li et al., 2005). Representative time-lapse images (Fig. 4A) of control and post-status epilepticus SpH-expressing MFBs showed robust, cumulative increases in SpH fluorescence intensity during the stimulus train, followed by return of fluorescence to baseline levels ∼40 s after stimulus termination. Figure 4C plots the normalized mean SpH fluorescence responses in control versus post-status epilepticus animals, which showed significantly larger stimulus-evoked rises in SpH fluorescence (2.02 ± 0.15, red circles, n = 8, P < 0.05, Student's t-test) compared with controls (1.47 ± 0.03, black circles, n = 10).

Figure 4.

Pilocarpine-induced status epilepticus (SE) enhances vesicular release and endocytosis at mossy fibre terminals in CA3 stratum lucidum. (A) Time-lapse two-photon images from control and post-status epilepticus SpH-expressing MFBs in the proximal apical dendritic region of field CA3. Solid arrows indicate puncta that showed activity-dependent fluorescence changes during a 600 pulse/20 Hz stimulus train. The broken arrows in (A, bottom row) show there were some slowly fluorescing boutons in the epileptic slices. (B) Representative time course of normalized fluorescence intensity of individual boutons from a control (black traces) and a post-status epilepticus (red traces) slice in response to the 600 pulse/20 Hz mossy fibre stimulation (black bar shows duration of the train). (C) Normalized, evoked SpH fluorescence increases in response to a 600 pulse/20 Hz mossy fibre stimulus train, in MFBs from control (filled black circles, F_peak = 1.47 ± 0.03, n = 10) versus post-status epilepticus (open red circles, F_peak = 2.02 ± 0.15, n = 8) slices. F_peak was significantly increased in post-status epilepticus slices (P < 0.05, Student's t-test; all values mean ± SEM). (D) Frequency distribution histogram of normalized peak SpH fluorescence for all MFBs in post-status epilepticus (F_peak = 1.76 ± 0.04, 100/115 puncta and 3.89 ± 0.10, 15/115 puncta) versus control slices (F_peak = 1.41 ± 0.025, 93 puncta). Inset is a cumulative histogram of normalized peak SpH fluorescence, P < 0.001, Kolmogorov–Smirnov test. (E) Mean fluorescence values of scaled SpH fluorescence decay (data normalized to respective peak fluorescence values from (C) after cessation of stimulation, control (filled black circles) and post-status epilepticus (open red circles). Inset represents mean ± SEM of individual decay constants derived by a single exponential fit to the SpH fluorescence decay curve. Control (black bar), τ = 10.4 ± 0.77 s, n = 67 and post-status epilepticus (red bar), τ = 7.23 ± 0.32 s, n = 133 (*P < 0.05, Student's t-test). Scale bar: A = 5 µm.

This result could be due to either a general increase in release probability of existing MFBs or appearance of a distinct sub-population of terminals with higher release probabilities. To resolve this distinction, we plotted a distribution histogram (Fig. 4D) of the normalized peak SpH fluorescence values, which is a function of the total number of vesicles released during the stimulus train. Control and post-status epilepticus distributions of peak SpH fluorescence for all individual MFBs indicate the appearance of a new sub-population of ∼3-fold higher release rate MFBs in the post-status epilepticus animals. The cumulative probability distribution (Fig. 4D inset) showed a significant increase in peak SpH fluorescence attained for the post-status epilepticus MFBs compared with controls (P < 0.001, Kolmogorov–Smirnov test).

Taken together, these results suggest that, in the first 1–2 months post-status epilepticus, morphological changes develop in the MFB cytoarchitecture, correlated with both an increase in vesiclular release rate of MFBs and appearance of a high release rate sub-population of terminals.

Pilocarpine-induced status epilepticus persistently enhances vesicular endocytosis in mossy fibre boutons of synaptopHluorin-expressing mice

To determine whether there are changes in rate of vesicle endocytosis in MFBs, we used the decay kinetics of SpH fluorescence after the end of the stimulus train. Since vesicle endocytosis is the rate-limiting step for decay in vesicle fluorescence (Sankaranarayanan and Ryan, 2000), determining the decay constant (τ) of this fluorescence gives an estimate of the rate of vesicle endocytosis. We first normalized the mean peak stimulus-evoked SpH fluorescence increases to 1.0 for control and post-status epilepticus slices, to compare directly the time courses of decay. As shown in Fig. 4E, the mean rate of decay of SpH fluorescence from peak to baseline for epileptic slices was significantly faster (red circles) compared with control slices (black circles, P < 0.05, Student's t-test), indicating an enhanced speed of endocytosis. We also fit single exponential decay functions to curves for individual MFBs to determine the decay time constants in post-status epilepticus vs. control MFBs. As shown in the Fig. 4E inset, chronic post-status epilepticus led to a significant decrease in decay time constants [red bar, τ = 7.23 ± 0.32 s versus control slices (black bar), τ = 10.4 ± 0.77 s, P < 0.05; Student's t-test] also consistent with acceleration in the rate of vesicle retrieval.

FM1-43 measurement of changes in vesicular release from the rat mossy fibre bouton readily releasable pool at early and late time points post-status epilepticus

As a second, independent measure of vesicular release, we estimated the time course of neurotransmitter release in slices from post-status epilepticus Sprague–Dawley rats using two-photon imaging of stimulus-evoked release of the styryl dye FM1-43 as described previously (Stanton et al., 2003, 2005; Zakharenko et al., 2003; Winterer et al., 2006; Bailey et al., 2008). We loaded FM1-43 selectively into vesicles of the RRP with a 30-s hypertonic shock (Supplementary Fig. 2) and similar to SpH fluorescence, we visualized bright puncta in the proximal region of field CA3 > 2 µm in diameter (Fig. 5A), presumed MFBs (Galimberti et al., 2006). More distal puncta that took up FM1-43 were much smaller in size, consistent with associational-commissural synaptic terminals. Dye release was triggered by repetitive bursts of 50 mossy fibre stimuli at a frequency of 20 Hz, spaced 30 s apart (Fig. 5B and Supplementary Fig. 2), to allow binding and clearance by the scavenger ADVASEP-7 (100 µm) (Zakharenko et al., 2003) and minimize potential movement artefacts.

Figure 5.

Pilocarpine-induced status epilepticus (SE) enhances evoked, but not action potential independent, presynaptic vesicular release of FM1-43 from the RRP in MFBs. (A) Two-photon image of a control field CA3 showing FM1-43 loaded RRP of MFBs. The solid arrows show staining of giant mossy fibre terminals in the proximal region of CA3 stratum lucidum. The rectangular box outlines FM1-43 staining of puncta distal to the CA3 pyramidal cell bodies and much smaller in size, likely to be associational-commissural synapses. (B) Representative time-lapse images of FM1-43 destaining from the RRP using repetitive mossy fibre stimulus bursts (50 stimuli/20 Hz each 30 s) to evoke release from MFBs in control and 2-month post-SE slices, respectively. [B (1 and 4)] Pre-stimulus baseline fluorescence, [(B (2 and 5)] fluorescence intensity after 15 trains, and [B (3 and 6)] fluorescence intensity after 30 trains of mossy fibre stimulation. Solid arrows indicate regions of interest >2 µm in diameter that showed robust stimulus dependent FM1-43 destaining. (C) Time course of FM1-43 destaining of MFBs in 1–2 months (2 M) post-SE slices (filled red circles, 0.36 ± 0.03, n = 6), aged-matched controls (filled black circles, 0.49 ± 0.04, n = 5), >11 month (11 M) post-SE slices (open red circles, 0.79 ± 0.04, n = 5) and aged-matched controls (open black circles, 0.62 ± 0.04, n = 6). All points are mean ± SEM of normalized fluorescence decay at 30th stimulus burst. P < 0.05, Student's t-test. (D) Time course of first 4.5 min of action potential independent FM1-43 destaining in presence of tetrodotoxin (TTX; 1 µm) and 15 mM [K+]o from the RRP in MFBs of 1–2 months post-SE slices (filled circles, n = 4) and aged matched control (open circles, n = 5) slices (all points are mean ± SEM). Decay after 4.5 min in High[K+]o: Control = 0.86 ± 0.028 (∼14% of pre-stimulus baseline) and post-SE = 0.80 ± 0.034 (∼20% of pre-stimulus baseline; P > 0.20, Student's t-test). (E) Late phase of spontaneous release. Time course of renormalized spontaneous FM1-43 destaining after 10–25 min in High[K+]o. Control normalized destaining = 0.85 ± 0.02 and post-SE destaining = 0.81 ± 0.02 (P > 0.20, Student's t-test, all points are mean ± SEM). Scale bars: A = 20 µm; B = 10 µm.

Figure 5C shows that mean time courses of RRP loaded stimulus-evoked FM1-43 destaining from MFBs in slices up to 2 months post-status epilepticus was associated with a significantly enhanced release of FM1-43 (∼64% from pre-stimulus baseline), when compared with age matched controls (∼51% from pre-stimulus baseline, P < 0.05, Student's t-test). Interestingly, 11 months post-status epilepticus rats showed a substantial slowing in MFBs release rate in the epileptic group (∼21% from pre-stimulus baseline) versus aged matched controls (∼37% of baseline, P < 0.05, Student's t-test).

Action potential-independent release of FM1-43 from the mossy fibre bouton readily releasable pool is not persistently altered by pilocarpine-induced status epilepticus in rats

The above data suggest changes in stimulus evoked, but not necessarily spontaneous release rate post-status epilepticus. To measure presynaptic spontaneous release from the RRP, we loaded FM1-43 by hypertonic shock (800 mOsm/20 s), followed by bath application of tetrodotoxin (1 µM), and extracellular [K⁺] raised to 15 mM [K⁺]o (High[K+]o) to facilitate action potential-independent release (Axmacher et al., 2004). To mark the early phase of action potential-independent release, we monitored the magnitude of FM1-43 destaining within the first 4.5 min of High[K+]o application (Fig. 5D), which showed no significant difference between control (∼15%) and post-status epilepticus (∼20%) MFBs (P > 0.20, Student's t-test). To determine whether steady-state spontaneous release was altered, we examined the late phase of spontaneous release 10–25 min after application of High[K+]o (Fig. 5E). As in the early phase, there was no significant difference between the two groups after 25 min of High[K+]o, indicating action potential dependent and independent vesicular release from the RRP are differentially regulated in the post-status epilepticus state.

Status epilepticus elicits long-term ultrastructural reorganization of active zones in rat mossy fibre boutons

Previous literature has shown clear correlations between release probability and synapse morphological parameters such as sizes of the RRP (Dobrunz and Stevens, 1997) and rapidly-recycling pools (Murthy et al., 1997), active zone size (Schikorski and Stevens, 1997; Matz et al., 2010) also known as ‘release sites’, postsynaptic density (Schikorski and Stevens, 1997) and presynaptic bouton size (Matz et al., 2010). Since we found increased release probability post-status epilepticus as measured by both SpH (Fig. 4C) and FM1-43 destaining (Fig. 5C), we examined possible ultrastructural rearrangements in MFBs in CA3 stratum lucidum in post-status epilepticus and age-matched control Sprague–Dawley rats (Fig. 6A).

Figure 6.

Transmission electron microscopy of active zones (AZ) in MFBs of control and post-status epilepticus (SE) rats. (A) Representative transmission electron microscopy images of control (left) and post-SE (right) MFBs illustrating active zones on synaptic excrescences (se) showing an apparent increase in synaptic vesicle density in post-SE MFBs. Boxed areas are depicted at higher magnification (×4) in images (B and C), showing arrangement of vesicles in the active zones of control and post-SE MFBs. Notice the larger density of vesicles and length of active zones in post-SE MFBs. (D) Frequency histograms and bar graph representation of data for active zone length, number of synaptic vesicles (SV) in RRP and releasable vesicle pool (RP), and number of docked synaptic vesicles. Notice appearance of increased number of active zones exhibiting larger lengths post-SE (>800 nm). (E) Cumulative histograms for these variables in control versus post-SE groups revealed significant rightward shifts toward larger sizes (Kolmogorov–Smirnov (K–S) two-sample test). Scale bars: A = 2 µm; B–E = 500 nm. D = dendrite.

Transmission electron microscopy shows that large MFBs (2–5 μm in diameter) had multiple active zones facing postsynaptic densities, contained mitochondria of various sizes, and were filled with numerous small and large clear synaptic vesicles distributed throughout the terminal (Fig. 6A), as described previously (Amaral and Dent, 1981; Chicurel and Harris, 1992; Rollenhagen et al., 2007). Asymmetric active zones were distinguished at MFBs synapses by the dense accumulation of synaptic vesicles in close proximity to the presynaptic density and characteristic widening of the synaptic cleft.

There was a significant increase in number of active zones per MFB in post-status epilepticus rats (130 active zones in six control rats, 5.1 ± 1.36 active zones per MFB; 286 active zones in seven post-status epilepticus rats, 7.7 ± 3.05 active zones per MFB, Student's t-test, P < 0.05). The number of perforated synapses was also significantly increased in MFBs from post-status epilepticus animals (46 of 286, 16.1%) compared to controls (12 of 130, 9.2%, Student's t-test, P < 0.05). The majority of electron microscopy variables failed to follow normal distributions, necessitating use of a non-parametric Kolmogorov–Smirnov two-sample test to assess between group differences in distributions. As reported previously (Chicurel and Harris, 1992; Rollenhagen et al., 2007), individual active zones varied substantially in shape and size; both large and small active zones were found in control (104–887 nm) and post-status epilepticus (105–1837 nm) hippocampus. Frequency distributions revealed the presence of a distinct group of synapses of larger length in epileptic animals that was absent in controls (Fig. 6D). A cumulative histogram indicated a significant leftward shift towards larger individual active zone lengths in MFBs post-status epilepticus (Fig. 6E), compared with controls (Table 1, Kolmogorov–Smirnov test, P < 0.005). There was also a significant increase in mean postsynaptic density area in the post-status epilepticus group (Table 1) compared to controls (Table 1, Kolmogorov–Smirnov test P < 0.005, ∼37% increase). In contrast, no significant changes were detected in average synaptic cleft width between control and post-status epilepticus active zones (Table 1).

Table 1.

Summary of quantitative analysis of structural variables in active zones of MFBs

	Control, Mean ± SD	Post-status epilepticus, Mean ± SD	Percentage of control	K–S, P
Active zone ultrastructural variables
Length of active zone (nm)	364.91 ± 44.81	485.27 ± 59.63	133.5	<0.005
Length of synaptic cleft (nm)	26.93 ± 3.91	29.95 ± 2.46	107.4	0.31
PSD area (mm2)	12.31 ± 3.03	16.90 ± 3.31	136.9	<0.005
Number of SVs in RRP	7.43 ± 3.12	8.94 ± 1.78	120.3	<0.001
Number of SVs in the releasable pool	31.07 ± 6.05	33.44 ± 7.13	107.6	<0.05
Number of docked SVs per active zone	1.96 ± 0.68	2.63 ± 0.56	134.2	<0.0005
Docked SVs per active zone length (SV/mm)	5.48 ± 1.75	5.88 ± 0.90	108.9	<0.05
Percentage of docked SVs of RRP	27.92 ± 6.45	29.56 ± 3.24	105.85	0.59

Measurements were obtained from analysis of active zone variables in MFBs. Statistical comparisons were made using the Kolmogorov–Smirnov (K–S) two-sample test, with statistical significance set at P < 0.05. Values are presented as means ± SEM.

PSD = postsynaptic density; SV = synaptic vesicle.

It has been previously suggested that a rapid refilling of the RRP from a larger releasable vesicle pool is a key mechanism in ensuring fidelity of mossy fibre-CA3 pyramidal cell neurotransmission (Suyama et al., 2007). To determine whether ultrastructural organization of synaptic vesicle pools is altered post-status epilepticus, we measured the number of vesicles docked, within 60 nm of the active zone, 60–200 nm from an active zone, and >200 nm from an active zone, in MFBs of control versus epileptic animals. The RRP was defined as the sum of docked vesicles and those within 60 nm of the active zone, while the releasable pool was defined as vesicles 60–200 nm away from an active zone (Suyama et al., 2007). Compared with controls, post-status epilepticus increased number of docked (+34.2%), RRP (+20.3%), and releasable pool (+7.6%) synaptic vesicles (Fig. 6D and Table 1). A significant difference was detected in the analysis of the cumulative distributions of these variables by Kolmogorov–Smirnov test (Fig. 6E and Table 1). Although the percentage of docked vesicles relative to RRP size was not significantly different (Table 1), the average number of docked vesicles per length of active zone was significantly higher for post-status epilepticus (+8.9%, Table 1). Additionally, the number of synaptic vesicles in each of these pools was significantly correlated with the length of individual active zones in both control (RRP: r = 0.33, P < 0.001 and releasable pool: r = 0.41, P < 0.001) and post-status epilepticus MFBs (RRP: r = 0.57, P < 0.001 and releasable pool: r = 0.55, P < 0.001, Supplementary Fig. 3).

In order to assess endocytosis, we measured the number and percent of clathrin-coated vesicles 0–200 nm from an active zone (Fig. 7A) and the area of membranous regions apparently internalized at the active zone, indicative of ‘bulk endocytosis’ (Fig. 7B), in MFBs of control and post-status epilepticus. There were no statistical differences in percent of synapses exhibiting one or two putative clathrin-coated vesicles at active zones between control (22.7 ± 10.1%) and post-status epilepticus (22.0 ± 7.1%, Fig. 7C) animals, or percent synapses exhibiting bulk endocytosis (control: 48.7 ± 13.1%, epileptic: 40.7 ± 15.8%). In contrast, the area of large elliptical or irregular membranous structures at the active zone was significantly higher at MFB synapses post-status epilepticus (5662 ± 385 nm²) compared with controls (2917 ± 287 nm², Kolmogorov–Smirnov test, P < 0.001, Fig. 7D), suggesting increased rate of endocytosis.

Figure 7.

Representative transmission electron microscopy images of active zones in MFBs exhibiting structural signs of clathrin-mediated endocytosis and ‘bulk endocytosis’ in control versus epileptic rats. (A) Putative clathrin-coated (dark) vesicles (white arrowheads) located proximal to presynaptic membrane active zones synapsing on spines (S) of control and post-status epilepticus (SE) MFBs. (B) Irregular membranous structures (black arrowheads) near active zones on spines were observed in both control and post-SE MFBs. Note these structures were larger in post-SE rats. (C) Mean ± SEM% active zones positive for clathrin-coated vesicles, showing no difference in control versus post-SE rats (P > 0.20, Student's t-test). (D) Cumulative histogram plot of bulk endocytosis area showing a significant rightward shift on the size distribution towards larger values in post-SE MFBs (red) compared with controls (blue, P < 0.001, Kolmogorov–Smirnov test). Inset: mean area of bulk endocytosis in control (black) versus post-status epilepticus (red) rats (*P < 0.05, Student's t-test. Scale bars: A, B = 500 nm.

Taken together, our electron microscopy data show pilocarpine-induced status epilepticus leads to a profound and long-lasting rearrangement of MFB transmitter release sites, resulting in significant increases in number of release sites per MFB, length of individual release sites, and RRP and releasable pool vesicle pools sizes that may underlie persistent increases in functional transmitter vesicular release rates, magnitude and recycling properties.

Discussion

We describe here a set of long-term alterations in presynaptic morphology and synaptic vesicle recycling at mossy fibre-CA3 terminals of rats and mice subjected to pilocarpine-induced status epilepticus, a model of temporal lobe epilepsy that results in spontaneous recurrent seizures within 2 months in ∼97% of rats (Leite et al., 1990; Cavalheiro et al., 1991; Mello et al., 1993; Priel et al., 1996) and mice (Cavalheiro et al., 1996; Muller et al., 2009). In rats, the latent period ranges between 1 and 6 weeks (Curia et al., 2008), with the mean latent period between 15–18 days (Leite et al., 1990; Priel et al., 1996), and depends on the duration of status epilepticus, where a 2-h status epilepticus corresponds to about a 7-day long latent period (Goffin et al., 2007). In mice, the latent period is ∼14 days (Cavalheiro et al., 1996), therefore, an overwhelming majority of our animals likely developed spontaneous seizures prior to experiments. Observed presynaptic changes include increases in (i) MFB size; (ii) number of release sites per MFB; (iii) number of vesicles in the RRP and releasable pool; (iv) active zone length; (v) action potential-driven vesicular release rate measured with either FM1-43 or in SpH-expressing transgenic mice; and (vi) enhanced vesicle endocytosis. These alterations persisted for at least 1 month following a single sustained status epilepticus.

Functional enhancement in vesicular transmitter release from MFBs correlates well with previously reported structural changes (number of docked synaptic vesicles or the RRP, area of postsynaptic density, active zone length) that track vesicle release probability (Schikorski and Stevens, 1997; Matz et al., 2010), suggesting that the morphological changes are probably components of the response to hyperactivation that underlie enhanced transmitter release. This agrees with a previous report that kainic acid-induced seizures lead to an increase in glutamate release from the RRP and loss of paired pulse facilitation (Goussakov et al., 2000), suggesting that seizures of a variety of causes may lead to persistent increases in basal release probability.

The significant increase we observed in FM1-43 destaining post-status epilepticus and the correlated increase of vesicle pool sizes measured by transmission electron microscopy suggest an increase in both initial release probability and size of the readily releasable and rapidly recycling vesicle pools. Previous reports suggest that the size of the recycling pool is an important determinant of sustainability of release with prolonged activation (Suyama et al., 2007), consistent with the increases we observed post-status epilepticus. While our transmission electron microscopy SpH and Alexa Fluor 594-dextran loading data all indicate that MFBs show increased size and larger numbers of synaptic vesicles for months after status epilepticus; the new population of terminals, with significantly larger size and faster release rate, could be from modification of existing terminals, ectopic MFBs, or both, and remains to be determined. Also, since individual MFBs contain multiple active zones (Hallermann et al., 2003), we cannot tell whether increased release probability at MFBs (Figs 4D and 5C) occurs at the level of individual active zones switching to a high release state, or whether there is a uniform increase in release probability across all active zones. Additional changes such as increases in the number of active zones per bouton and size of the RRP and releasable pool in individual active zones also suggest persistent increase in glutamate release that could be a key contributor to death of CA3 pyramidal neurons characteristic of hippocampal sclerosis. Increases in the density of perforated synapses have been associated with synaptogenesis and post-lesional compensatory plasticity (Itarat and Jones, 1992; Luke et al., 2004; Adkins et al., 2008). Increased incidence of perforated mossy fibre synapses after status epilepticus suggest that synaptic connectivity undergoes structural remodelling favouring synapses, with potentially higher transmission efficiency during the process of epileptogenesis.

Using SpH fluorescence, we derived a decay constant of 10.4 ± 0.77 s for control boutons (Fig. 4E), similar to previous reports of an endocytic time constant of 9.0 ± 5.5 s estimated using capacitance recordings at MFBs (Hallermann et al., 2003). Our observations of accelerated recovery of stimulus-evoked SpH fluorescence in slices from post-status epilepticus mice, but lack of differences between the number of clathrin-coated vesicles in electromicrographs between the post-status epilepticus and control groups, could be due to an enhanced rate of endocytosis that may or may not depend on clathrin, increased local recycling of vesicles to the RRP (Stanton et al., 2003) or to the recycling pool after exocytosis (Wu and Wu, 2009), increase in bulk endocytosis (retrieval of larger membrane area, Table 1) or combinations of the above. Also, tissues used for electromicrographs were not electrically stimulated prior to fixation, and thus do not necessarily reflect activity-dependent endocytic events but, more likely, the steady-state of the presynaptic bouton ultrastructure.

The 1–2 month time point post-status epilepticus used in our experiments corresponds to the period of occurrence of spontaneous seizures in both rats (Leite et al., 1990; Cavalheiro et al., 1991; Priel et al., 1996; Dudek and Sutula, 2007) and mice (Cavalheiro et al., 1996; Muller et al., 2009; Schauwecker, 2012). Interestingly, in contrast to the early increase in MFB release probability, we found that, at very long survival periods post-status epilepticus (11–12 months), rates of MFB release were markedly reduced. The reduced transmitter release from the RRP in these animals could either reflect long-term compensatory changes or dysfunction of presynaptic release machinery. Our observation that action potential independent RRP release rate in tetrodotoxin (1 µm) was not altered by pilocarpine seizures indicates that action potential dependent and independent regulation of the RRP is very different, perhaps due to potentially distinct pools of vesicles and mechanisms between evoked and spontaneous release (Fredj and Burrone, 2009; Chung et al., 2010). We cannot rule out the possibility that new vesicle pools may contribute to action potential independent release of glutamate post-status epilepticus.

Our functional data demonstrating that post-status epilepticus state leads to increased vesicular release and endocytotic rates, and electron microscope data showing increased presynaptic active zone length, membrane invaginations, clathrin-coated vesicles and vesicle pool sizes, are all consistent with the conclusion that post-status epilepticus state is associated with long-term increases in transmitter release from MFBs, and perhaps other glutamatergic terminals. Since brain-derived neurotrophic factor is highly expressed in presynaptic MFBs (Danzer and McNamara, 2004) and can elicit long-term enhancements in both transmitter release and size of the RRP (Zakharenko et al., 2003; Tyler et al., 2006), neurotrophins seem a likely candidate mediator of these presynaptic alterations.

Whether increase in stimulus evoked release following-status epilepticus results from alterations in Ca²⁺ influx, release of Ca²⁺ from internal stores per se, or whether there are downstream changes in SNARE protein sensitivity to Ca²⁺ is not known. A recent study (Pacheco et al., 2008) found that pilocarpine seizures also cause downregulation of large conductance calcium-activated potassium (BK) channels in MFBs, reductions that could slow hyperpolarization and increase terminal Ca²⁺ influx. Levels of the Ca²⁺ buffer calbindin D-28K are also lowered in MFBs after pilocarpine seizures (Carter et al., 2008), a change that could further increase presynaptic Ca²⁺ influx, dysregulate Ca²⁺ homeostasis and promote multivesicular release (Hallermann et al., 2003).

In summary, our study is the first, to our knowledge, using the pilocarpine model of temporal lobe epilepsy in SpH-expressing transgenic mice, a new tool to investigate presynaptic alterations in epilepsy with potential application for other CNS diseases that may involve presynaptic dysfunction. Our studies indicate that the early phase of the pilocarpine model of temporal lobe epilepsy is associated with persistent structural changes in mossy fibre presynaptic terminal size, vesicle pools and active zones that are correlated with functional increases in rates of action potential-driven vesicular release and associated endocytotic recycling. The time course and mechanisms underlying these changes suggests the presynaptic terminal as a novel target for new therapeutics to treat epilepsy, especially temporal lope epilepsy.

Funding

National Institutes of Health (grants GM081109, P20 MD001091-06 and NS063950); Department of Defense (grant PR100534 to E.R.G.S.); National Institute of Neurological Diseases and Stroke (grants NS056093 to J.V., NS072966 to L.V. and NS044421 to P.K.S.); Department of Defense (grant PR100534P1 to P.K.S.).

Supplementary material

Supplementary material is available at Brain online.

Glossary

Abbreviations

MFB

mossy fibre bouton

RRP

readily releasable pool

SpH

SynaptopHluorin