Neurotransmitter release can occur independently of action potentials. These miniature release events are caused by release of the smallest functional unit of neurotransmitters from the presynaptic nerve terminals (quantum)—presumably from single synaptic vesicles (Katz,1969; Lisman et al.,2007). The miniature release events are important in controlling a variety of functions in postsynaptic neurons. For instance, they regulate the membrane potential and generation of the action potentials (Carter and Regehr,2002; Otmakhov et al.,1993; Sharma and Vijayaraghavan,2003). Miniature release of glutamate has a strong impact on synaptic transmission and its consequences, as it maintains the density and length of dendritic spines (McKinney et al.,1999) and can also trigger an increase in the intracellular second messenger Ca2+ (Espinosa and Kavalali,2009; Murphy et al.,1994).
Miniature release of neurotransmitters has not been evaluated in certain disorders that are considered to affect synapses. Early-onset generalized DYT1 dystonia (Klein et al.,1998; Ozelius et al.,1997) is one such disorder. It is one of the most common hereditary forms of the neurological disorder dystonia (Ozelius and Bressman,2011), is autosomal-dominant, and is characterized by involuntary movements of the limbs, trunk, and face. Notable features of DYT1 dystonia include: low penetrance (only 30–40% of individuals with the responsible mutation show motor symptoms); variability in motor symptoms, with respect to both severity and the age of onset (ranging between 5 and 25 years of age); and the presence of endophenotypes (subclinical signs) in nonmanifesting (i.e., without motor symptoms) carriers of the mutation. The pathophysiology of DYT1 dystonia is thought to arise from abnormalities in synaptic transmission and the synaptic microstructure (Breakefield et al.,2008; Warner et al.,2010). This notion is based on findings from clinical studies and neuropathologic analyses: the former have revealed enhanced neuronal excitability in patients (e.g., Carbon et al.,2010); the latter have failed to identify neurodegeneration, inflammation or other major macroscopic changes in the central nervous system (CNS) (Hedreen et al.,1988; Rostasy et al.,2003).
DYT1 dystonia is caused by a mutation in the gene that encodes dystonia-associated protein, torsinA. This mutation results in an in-frame deletion of a codon for glutamate (ΔE-torsinA). TorsinA belongs to the “ATPases associated with diverse cellular activities” (AAA+) family of proteins, whose members generally perform chaperone-like functions, assisting in protein unfolding, protein-complex disassembly, membrane trafficking, and fusion (Burdette et al.,2010; Hanson and Whiteheart,2005; White and Lauring,2007). Its homologues are evolutionarily conserved, and found in species ranging from C. elegans to mammals (Breakefield et al.,2001; Jungwirth et al.,2010; Ozelius et al.,1999).
The state of synaptic signaling in ΔE-torsinA mutation carriers is important, but there has not been a study of miniature release of neurotransmitters in animals modeling the mutation carriers. Here, we have analyzed such events in neurons obtained from the ΔE-torsinA knock-in mouse (Goodchild et al.,2005), a model that does not develop overt motor symptoms, yet reproduces the abnormalities in CNS synaptic circuitry and metabolism that have been described in human, nonmanifesting mutation carriers of ΔE-torsinA (Ulug et al.,2011). We have focused on neurons obtained from the hippocampus, as this is among the CNS sites in which torsinA transcript and protein levels are highest (Allen_Mouse_Brain_Atlas,2009; Augood et al.,1999; Walker et al.,2001), and also a site in which glutamate release from nerve terminals has been studied extensively (e.g., Carroll et al.,1998; Malgaroli et al.,1995; McKinney et al.,1999; Noel et al.,1999; Nosyreva and Kavalali,2010).
We used two approaches to analyze miniature release events in the absence of action potentials (i.e., in the presence of TTX, which blocks the voltage-dependent Na+-channel). First, we used patch-clamp electrophysiology to record miniature excitatory postsynaptic currents (mEPSCs), the glutamate receptor-mediated postsynaptic responses to releases of the smallest units of glutamate. Second, we employed a fluorescent, styryl FM dye to label synaptic vesicles as they were recycled, monitoring exocytosis of these synaptic vesicles from individual nerve terminals in the presence of TTX. Both approaches support the hypothesis that miniature release events occur more frequently in heterozygous and homozygous ΔE-torsinA neurons than in their wild-type counterparts, suggesting that this form of neurotransmitter release is enhanced within some CNS regions in ΔE-torsinA knock-in mice.
MATERIALS AND METHODS
Animal care and procedures were approved by the University of Iowa Animal Care and Use Committee, and performed in accordance with the standards set by the National Institutes of Health Guide for the Care and Use of Laboratory Animals (NIH Publications No. 80-23), revised 1996. On postnatal Days 0–1, pups of either sex of the knock-in model of DYT1 dystonia (Goodchild et al.,2005) were genotyped according to the fast-genotyping procedure (within ∼5 h, EZ Fast Tissue/Tail PCR Genotyping Kit, EZ BioResearch LLC, St, Louis, MO), using a published protocol (Goodchild et al.,2005).
Individual newborn pups were genotyped and cultured separately. Primary hippocampal neurons were cultured by a method described in our previous work (Harata et al.,2006), with some modification. Briefly, the CA3-CA1 regions of the hippocampus were dissected on postnatal Days 0–1, trypsinized and dissociated. The cells were plated on 12-mm coverslips pre-seeded with a rat glial feeder layer (Garcia-Junco-Clemente et al.,2010), in 24-well dishes and at a density of 12,000 cells per well. The feeder layer had been seeded in plating medium with the following composition: MEM (Invitrogen, Carlsbad, CA) plus 5 g/l glucose, 0.2 g/l NaHCO3, 100 mg/l bovine transferrin (EMD Chemicals, Gibbstown, NJ), 2 mM GlutaMAX (Invitrogen), 25 mg/l insulin, and 10% fetal bovine serum (FBS, Invitrogen). Feeder layers were maintained in a 1:1 mixture of plating medium and growth medium. The latter had the following composition: MEM plus 4 μM cytosine β-D-arabinofuranoside, 0.5 mM GlutaMAX, NS21 (Chen et al.,2008), and 5% FBS. The hippocampal neurons were used on Days 11–14 of culture. The experimental data were obtained from 4 or 5 different culture batches (animals) for each genotype (wild-type, heterozygous and homozygous littermates).
Patch-clamp recording of miniature postsynaptic currents
Electrical recordings were carried out on an inverted microscope (Eclipse-TiE, Nikon, Melville, NY). Membrane currents were monitored by the voltage-clamp mode of conventional whole-cell patch-clamp recording. Patch-pipettes were fabricated from borosilicate glass tubes (1.5-mm outer diameter, PG52151-4, World Precision Instruments, Sarasota, FL) using a pipette puller (P-97 Flaming/Brown Micropipette Puller, Sutter Instrument, Novato, CA), and fire polished (Micro Forge MF-830, Narishige International USA, East Meadow, NY). The resistance of the recording electrode was 2–3 MΩ. The patch-pipettes were positioned with high precision and minimal drifting using a manipulator (PatchStar Micromanipulator, Scientifica, East Sussex, UK). The current and voltage were measured with a patch-clamp amplifier (Axopatch 200B, Molecular Devices, Sunnyvale, CA), which was controlled by the pCLAMP software (Molecular Devices). The membrane currents were acquired at 10 kHz, and filtered at 5 kHz with a built-in 4-pole Bessel filter.
Miniature postsynaptic currents were recorded in the continued presence of 0.5 μM TTX (Tocris Bioscience, Ellisville, MO) in the external solution. Recording of miniature events was started at least 3–4 min after establishing the whole-cell recording, to allow enough time for the internal pipette solution to equilibrate with the intracellular condition. Miniature events were recorded for 2–4 min. In one series of experiments (Supporting Information Fig. S1), glutamate-mediated mEPSCs and GABA-mediated miniature inhibitory postsynaptic currents (mIPSCs) were also isolated in the presence of GABAA receptor antagonist (20 μM (-)-bicuculline methochloride, Tocris Bioscience, Ellisville, MO) and AMPA receptor antagonist (10 μM, 6-cyano-7-nitroquinoxaline-2,3-dione, CNQX, Tocris Bioscience), respectively. In a separate experiment, we confirmed the effectiveness of TTX treatment in our assay, demonstrating that it completely eliminated the inward Na+ currents induced by a 2-ms step depolarization to 0 mV from a holding potential (VH) of −70 mV. Series resistance was measured based on the response to a 10 mV depolarizing voltage step from a holding potential of –70 mV for 100 ms, and was compensated at ∼70%. The recordings were discarded if the series resistance before compensation exceeded 20 MΩ or if the membrane resistance was below 100 MΩ. All experiments were performed at room temperature (23–25°C).
The Tyrode (external) solution had the following composition (in mM): 125 NaCl, 2 KCl, 2 CaCl2, 2 MgCl2, 30 glucose, 25 HEPES, 310 mOsm, pH 7.4. Two internal solutions were used. One, with an intra-pipette Cl− concentration ([Cl−]pipette) of 75 mM, had the following composition (in mM): 60 K-gluconate, 70 KCl, 5 NaCl, 1 EGTA, 4 MgATP, 0.3 GTPNa2, 10 HEPES, 10 phosphocreatine, 5 unit/mg creatine phosphokinase, pH 7.2, 305 mOsm. The second internal solution had the [Cl−]pipette of 9 mM and the following composition (in mM): 130 K-gluconate, 8 NaCl, 1 EGTA, 4 MgATP, 0.3 GTPNa2, 10 HEPES, 10 phosphocreatine, 5 unit/mg creatine phosphokinase, pH 7.2, 305 mOsm. The calculated Cl− equilibrium potentials (ECl) were −15.0 and −69.1 mV for [Cl−]pipette of 75 and 9 mM, respectively. ECl was calculated based on the Nernst equation at 23°C: ECl = −58.8*log(135/[Cl−]pipette) (Kakazu et al.,1999).
It is possible to record mIPSC as an outward current by having ECl more hyperpolarized than the holding potential (e.g. Li et al.,2011), but we did not use this method because the overriding, outward mIPSC would have complicated the analysis of inward mEPSCs.
Analysis of mEPSCs
Miniature events were analyzed using the Mini Analysis Program (version 6.0.7, Synaptosoft, Fort Lee, NJ). Events were selected for analyses using criteria based on threshold amplitude (5x IRMS) and area under the curve (1.5x amplitude). In addition, all the traces were visually examined to protect against software errors. The average baseline noise (IRMS) was 2.94 ± 0.34 pA for wild-type neurons (n = 16), and 3.90 ± 0.49 pA for heterozygous neurons (n = 14) over 51.2 ms, with no statistically significant difference (P = 0.11, mean ± SEM, t-test).
Interevent interval was defined as the time elapsing between measured peaks for two consecutive events. The amplitude of a mEPSC event was calculated by subtracting the pre-event baseline current (averaged over 1 ms) from the peak amplitude. If the rise of an event occurred during the falling phase of a previous event, as might occur during a burst, the baseline was estimated by extrapolating the decay of the first peak using a single exponential function. The amplitude of the event was then calculated by subtracting the extrapolated baseline from the peak amplitude. Rise time was defined as the time for the signal to increase from the onset (0.5% of the peak amplitude) to the peak (100%). Decay time was defined as the time for the signal to decrease from the peak (100%) to 40%. This parameter is susceptible to changes in noise level, especially near the end of the decay period, when the signal amplitude approaches that of the noise. Also multiple overriding events were not excluded in measuring this parameter. Thus the decay phase was also analyzed using another parameter, which is less susceptible to changes in noise level. The decay time constant was measured by fitting a single exponential curve to the 80–20% decay phase of an isolated event. Curve fitting was applied to the first 100 events, in each recorded neuron, that started from a stable baseline and decayed back to baseline (i.e., no other events overrode the decay phase). An area of an event (unitary mEPSC charge) was defined as the area delimited by the pre-event baseline level, onset time (0.5% of peak amplitude) and the time at which the signal decreased to 40% of peak amplitude.
Live-cell and fixed-cell imaging of FM dye
For live-cell imaging, the neurons were subjected to staining by spontaneous activity (basal activity) (Kakazu et al., 2012). They were incubated in 2.5 μM FM4-64 (in MEM, Invitrogen) for 10 min, at 37°C, in the absence of neurotransmitter receptor antagonists and TTX. After staining, neurons were transferred to an imaging chamber (RC-21BRFS, Warner Instruments, Hamden, CT) and washed with dye-free Tyrode solution for 6–7 min, at room temperature. Spontaneous loss of FM dye due to action potentials was suppressed by the application of TTX, CNQX, and the NMDA receptor antagonist D,L-2-amino-5-phosphonopentanoic acid (50 μM, Tocris Bioscience).
After washing, the nerve terminals were imaged for 120 s for the miniature FM release, in the continued presence of TTX, CNQX, and D,L-AP5. They were destained afterwards, by applying the Ca2+ ionophore, ionomycin (5 μM) for 120 s in the continued presence of TTX, CNQX and D,L-AP5 at room temperature. For each coverslip, only a single imaging field was analyzed. This eliminates the possibility that any of the imaged fields was subject to ionomycin-induced FM destaining during an earlier experiment.
For fixed-cell imaging of FM dye, the live neurons were stained with a modified Tyrode solution containing 45 mM KCl and 2.5 μM aldehyde-fixable FM4-64 (FM4-64FX, Invitrogen) for 1 min. Thereafter the neurons were washed for 1 min and chemically fixed in 4% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA) and 4% sucrose in Tyrode solution for 30 min at 4°C, washed for 10 min in Tyrode solution and imaged. The fluorescence intensity of fixable FM4-64 was retained after paraformaldehyde fixation.
All experiments with FM dyes were carried out using extracellular solutions lacking GABAA receptor antagonists.
Immunocytochemistry was used to detect the glutamatergic presynaptic marker, vesicular glutamate transporter 1 (VGLUT1), and the dendritic marker, microtubule-associated protein 2 (MAP2). The cultured neurons were fixed with 4% paraformaldehyde (Electron Microscopy Sciences) and 4% sucrose in Tyrode solution for 30 min at 4°C. After being rinsed with Tyrode solution twice for 5 min each at 4°C, the cells were blocked and permeabilized with 2% normal goat serum (Jackson ImmunoResearch Laboratories, West Grove, PA) and 0.4% saponin in phosphate-buffered saline (PBS, pH 7.4, Invitrogen) (blocking solution), for 60 min at room temperature. Thereafter they were treated with polyclonal, guinea-pig anti-VGLUT1 antibody (AB5905, Chemicon-Millipore, Billerica, MA) (1000x dilution in blocking solution), and polyclonal, rabbit anti-MAP2 antibody (AB5622, Chemicon-Millipore) (400x dilution) overnight (15-21 hours) at 4°C. Following rinsing with PBS, 3 times for 7 min each, the neurons were incubated with goat anti-guinea-pig IgG antibody conjugated with Alexa Fluor 594 (Invitrogen) (1000x dilution in blocking solution), and goat anti-rabbit IgG antibody conjugated with Alexa Fluor 405 (Invitrogen) (1000x dilution) for 60 min at room temperature. They were rinsed with PBS at least five times for 20 min each, and observed directly in PBS.
Fluorescence imaging system
Cells were imaged using an inverted microscope (Eclipse-TiE, Nikon). For imaging FM dyes, we used an EMCCD camera (DU-860, Andor Technology, Belfast, UK). The camera was continuously perfused with chilled water (Oasis 160 liquid recirculating chiller, Solid State Cooling Systems, Wappingers Falls, NY) to maintain a temperature of −80°C, and to thereby reduce noise.
In live neurons, FM dye was excited using a 490-nm light-emitting diode (LED, CoolLED-Custom Interconnect, Hampshire, UK) with 10% intensity, and imaged with an objective lens (Plan Fluor, 40x, NA1.30, Nikon), a filter cube (490/20-nm ex, 510-nm dclp, 650-nm-LP em) and 0.7x coupler. 16-bit images were acquired at 1 frame/s, 20-ms exposure time of camera, EM gain of 50, 10 MHz pixel readout rate, without binning, using the Solis software (Andor). Images were stored in a SIF format (Andor). Neuronal exposure time to excitation light was minimized to 20 ms, by turning the LED on only during image capture (triggered by “Shutter” output of the camera).
In fixed neurons, fixable FM dye was visualized using largely the same set of imaging parameters used for FM4-64 in live neurons, including the 20-ms exposure time of camera. The exceptions were the use of either (1) 10% LED intensity and a 20-ms exposure to LED light (the same as in live-neuron imaging), (2) 100% LED intensity and a 20-ms exposure to LED, or (3) 100% LED intensity and continuous exposure to LED (still with the same 20-ms exposure time of camera).
For immunocytochemical observation, we used an interline CCD camera (Clara, Andor Technology). The camera was cooled at -45°C by an internal fan. Alexa Fluor 594 was excited using a 595-nm LED (CoolLED-Custom Interconnect) with 100% intensity, and imaged with a filter cube (590/55-nm ex, 625-nm dclp, 665/65-nm em), and 2-s exposure. Alexa Fluor 405 was excited using a 400-nm LED (CoolLED-Custom Interconnect) at 100% intensity, and imaged with a filter cube (405/40-nm ex, 440-nm dclp, 470/40-nm em), and 2-s exposure. 16-bit images were acquired with an objective lens (Plan Fluor, 40x), without a coupler (i.e., 1x) and without binning, in a single-image capturing mode of the Solis software (Andor). Images were stored in a SIF format (Andor). For immunocytochemical imaging, differential interference contrast (DIC) optics was first applied to identify neurons of normal morphology; fluorescence images were acquired only afterward, and thus neurons were selected for analysis without regard to fluorescence information.
Image analysis for FM experiments
FM signal was quantified using Image J (v1.43m, W. S. Rasband, NIH) and the associated plug-ins. Acquired images in a time series (“Stack”) were aligned, using the ImageJ-Image Stabilizer plug-in (Kang Li) to correct for small movements in FM signals, and saved in a TIFF format.
Functional nerve terminals (boutons) in live neurons were identified based on differences in fluorescence intensity between the 5-frame averages taken immediately before and after ionomycin-induced destaining. Regions-of-interest (ROIs, 3x3 pixels, 2.4x2.4 μm) were assigned on isolated, fluorescent puncta in the difference image, if five or more of the 9 pixels showed an intensity above 30 arbitrary units (a.u.), which was approximately the standard deviation of the background intensity obtained from the bare coverslip area. Changes in ROI intensity were measured using ImageJ-Time Series Analyzer V2.0 (Balaji Jayaprakash), and were exported to Microsoft Excel. ROIs were excluded if they exhibited any of the following changes in intensity: an increase during recording, a sudden decrease before application of ionomycin or a long latency after application of ionomycin.
Miniature FM release was quantified as the cumulative change in the absolute intensity of FM4-64 between times 0 and 120 s of imaging (ΔFMMinis). It should be noted that individual release events cannot be visualized under the conditions used here, since the signal intensity from a single vesicle is low compared to the high noise level created by other stained vesicles in the same terminal (Aravanis et al.,2003).
The FM dye photobleaching was assessed in the fixed neurons stained with fixable FM4-64. ROIs of the same size were chosen from isolated fluorescent puncta that corresponded to nerve terminals. Changes in FM intensity in the ROIs during imaging were normalized by setting the absolute intensity at the start of imaging experiments as “1” and setting the absolute intensity when the camera shutter was closed as “0.” The normalized curves were fit with single exponential functions: FM(t) = exp(-t/τPB), where FM(t) is the relative intensity after time t of imaging; t is time in seconds after start of imaging; τPB is a time constant of photobleaching. Our measurement shows that our imaging system caused only minimal photobleaching (Supporting Information Fig. S5), and thus we did not need to correct for this.
For figure production, the contrast and brightness of the FM images were changed linearly using ImageJ, such that the parameters remained the same for all images in a single figure.
Image analysis for counting the numbers of somata and nerve terminals
Images were analyzed using Image J. Neuronal density was measured by counting the number of somata of live neurons, in a field imaged with 4x phase-contrast objective lens. Each field encompassed 3.74 mm2.
The number of glutamatergic nerve terminals in our preparation was counted, based on positive immunocytochemical staining for VGLUT1. Positive signals were identified after the intensity of image background was subtracted, and the pixels were selected for the intensity above a threshold. A cluster of contiguous positive pixels was counted as a potential glutamatergic terminal. To focus on the population of VGLUT1 in nerve terminals that make synapses on postsynaptic dendrites, we limited our analysis to the isolated VGLUT1 signals making contact with the MAP2-positive region.
For figure production, the images of VGLUT1 and MAP2 were binarized, and overlaid on gray-scale DIC images using ImageJ.
All chemical reagents were purchased from Sigma-Aldrich unless otherwise specified. Reagents were applied to live neurons in the imaging chamber by bath perfusion (600 μl/min) and the “Y-tube” method, a fast application system that allows exchange of the external solution within 30 ms, with an average travel rate of ∼100 μm/ms near the cell (Harata et al.,1996,1999; Kira et al.,1998). This configuration ensures that the entire surface of the neuron is constantly exposed to fresh solution.
The distributions of the parameters of miniature events were compared for statistical significance using the nonparametric Kolmogorov–Smirnov test (Figs. 1, 2, and 4) (Sulzer and Pothos,2000; Van der Kloot,1996) because they were not Gaussian (normal) (Supporting Information Fig. S1, panel E). Specifically, a cumulative histogram was created by (Sulzer and Pothos,2000; Van der Kloot,1996): sorting the data from a single experiment in ascending order; assigning each data point a rank number and a fractional value based on that rank (i.e., fractional value = rank number/total number of events); and plotting the fractional values of all events cumulatively, with the y-axis indicating the fractional value for each data point and the x-axis indicating the specific value of the parameter for the data point (e.g., mEPSC amplitude). This normalizes the distributions between y = 0 and y = 1. Each curve of the final cumulative histogram was then obtained by averaging the cumulative histograms obtained from individual neurons of the same genotype, in order to eliminate giving different weights to neurons based on the occurrence of different numbers of events.
Means were compared using the unpaired Student's t test (Fig. 3). All significance values provided are two-tailed P values.
Synaptic events in the absence of stimuli are defined as “miniature” when they are observed in the presence of TTX, and as “spontaneous” when they are observed in the absence of TTX.
Basic properties of cultured neurons obtained from ΔE-torsinA knock-in mice
We evaluated the basic properties of cultured neurons obtained from ΔE-torsinA knock-in mice because they have not been reported before. Using the whole-cell recording mode of the patch-clamp method (Fig. 1A), we measured the passive membrane properties: resting membrane potential, membrane capacitance and membrane resistance (Table I). No significant differences were observed among the neurons of the three genotypes (P > 0.05 between wild-type and either heterozygous or homozygous neurons, for any parameter, N = 11–24 neurons, t-test). The recording conditions were equivalent, as there was no significant difference in the series resistance before compensation (P > 0.05 between wild-type and either heterozygous or homozygous neurons, t-test) (Table I). This lack of change in the passive membrane properties in mutant neurons is consistent with previous reports based on the analysis of striatal neurons of ΔE-torsinA transgenic mice (Pisani et al.,2006; Sciamanna et al.,2009,2011). Additionally, neuronal density, as measured on the day of electrophysiological recording, was the same for all genotypes (P > 0.05 between wild-type and either of heterozygous or homozygous neurons, N = 10–12 image fields, t-test) (Table I).
Table I. Basic properties of the cultured neurons obtained from ΔE-torsinA knock-in mice
Numbers represent mean ± SEM of N = 11–24 neurons in each genotype (top four properties), and N = 10–12 image fields in each genotype for the neuronal density measurement. There was no statistically significant difference between wild-type and either heterozygous or homozygous neurons (P > 0.05, t-test).
Resting membrane potential (mV)
−64.75 ± 1.31
−62.55 ± 2.60
−63.47 ± 2.37
Membrane capacitance (pF)
99.39 ± 13.46
87.96 ± 9.50
87.29 ± 9.40
Membrane resistance (mΩ)
369.21 ± 58.20
249.76 ± 53.90
246.60 ± 42.95
Series resistance (mΩ)
13.67 ± 0.83
14.32 ± 0.87
14.09 ± 1.02
Neuronal density (/mm2)
2.27 ± 0.36
2.94 ± 0.27
2.31 ± 0.22
mEPSCs can be recorded in isolation from mIPSCs without relying on pharmacological inhibition
In one approach to evaluate the miniature release of glutamate, we recorded mEPSCs using a patch-clamp technique (Fig. 1A). One of the standard methods for recording mEPSCs in isolation from miniature inhibitory postsynaptic currents (mIPSCs) involves applying a competitive or noncompetitive antagonist of GABAA receptor to the external solution (Supporting Information Fig. S1). This is a necessary step when the [Cl−]pipette is high (e.g., 75 mM), because both mEPSCs and mIPSCs are inward currents at a negative holding potential (VH) (e.g., −70 mV) (Supporting Information Fig. S1). However, there is concern that such antagonists might change an existing balance between excitatory and inhibitory synaptic systems. We thus eliminated mIPSCs without using GABAA receptor antagonists, by changing the reversal potential of IPSCs to near the holding potential and thereby reducing the ionic driving force of Cl− (Torborg et al.,2010). This was achieved by setting [Cl−]pipette at 9 mM and VH at −70 mV (Supporting Information Fig. S2). Under these conditions, the mIPSC amplitude (in the presence of TTX) remained below the detection limit, even when the patch-clamped neuron received GABAergic inputs, as evidenced by the presence of spontaneous IPSCs (sIPSCs, in the absence of TTX) (Supporting Information Fig. S3A). Using this method, we found that the interevent intervals of not only sEPSCs but also mEPSCs in individual neurons were affected by GABAA receptor antagonists (Supporting Information Fig. S3B and Fig. S4; also see Supplementary Results). Thus, experiments using GABAA receptor antagonists to assess physiological mEPSC properties must be interpreted with caution.
Interevent intervals of mEPSCs are shortened in ΔE-torsinA mutant neurons
We addressed whether mEPSCs are affected in the context of DYT1 dystonia, using the electrophysiologic conditions described above (TTX, VH = −70 mV and [Cl−]pipette = 9 mM, in the absence of GABAA receptor antagonists). For this purpose, we recorded mEPSCs in cultured neurons obtained from wild-type, heterozygous, and homozygous ΔE-torsinA knock-in mice (Goodchild et al.,2005). Representative traces are shown in Figure 1B.
We measured the interevent intervals of the recorded mEPSCs (Fig. 1B, arrows). Although they varied widely within each genotype, they were distinctly shorter in neurons obtained from mutant mice (Fig. 1C, red curve for heterozygotes and green curve for homozygotes) than in neurons obtained from wild-type mice (black curve). The difference from the wild-type neurons was statistically significant, as assessed by the Kolmogorov-Smirnov test (heterozygous vs. wild-type neurons, P = 10−156; homozygous vs. wild-type neurons, P = 10−193; based on n = 10,440, 17,463, and 13,049 mEPSC events for wild-type, heterozygous and homozygous neurons, respectively). These results demonstrate that miniature releases of glutamate occur more frequently in the mutant mice. The fact that this change is present in mutant neurons of both genotypes underscores an importance of the change.
Imaging-based assessment supports enhanced miniature vesicular release by ΔE-torsinA mutation
The shortening of the interevent intervals could be a consequence of either an increase in the frequency of release from individual nerve terminals, or an increase in the number of glutamate-releasing nerve terminals. To test the former possibility directly, we used a second approach, based on live-cell imaging with fluorescent styryl FM dyes (Gaffield and Betz,2006). FM dye was loaded into recycling synaptic vesicles (staining); the subsequent release of FM dye from synaptic vesicles was measured, with the decrease in intensity of FM fluorescence (destaining) serving as a measure of vesicle exocytosis. In conventional imaging experiments, FM destaining is measured during stimulus-triggered vesicle exocytosis (Gaffield and Betz,2006; Harata et al.,2006). In our analysis, however, we assessed destaining in the absence of an external stimulus (Wilhelm et al.,2010) and in the presence of TTX. This allowed us to quantitate cumulative release via miniature events over a set period of time.
For live-cell imaging of ongoing miniature synaptic activity, the recycling synaptic vesicles were loaded with an FM dye (FM4-64 staining). The neurons were washed extensively with dye-free solution, after which imaging was started in the continued presence of TTX and in the absence of stimulation (Fig. 2Aa,b). After 2 min of observation (Fig. 2Ac and 0–120 s in Fig. 2Ae), we applied the Ca2+ ionophore ionomycin, which forces synaptic vesicle exocytosis (Piedras-Renteria et al.,2004) even in the presence of TTX. Positive responses to ionomycin (Fig. 2Ad,e) were used to limit our analysis to FM signals from the nerve terminals. The expanded graph (Fig. 2Af) illustrates the miniature FM dye release from wild-type neurons (black), heterozygous neurons (red), and homozygous neurons (green). The cumulative amount of FM dye released during a 2-min observation period (ΔFMMinis) was significantly higher in mutant neurons vs. their wild-type counterparts (Fig. 2Ba) (heterozygous vs. wild-type neurons, P = 9.4 x 10−6; homozygous vs. wild-type neurons, P = 5.3 x 10−9; based on “n” = 328, 387, and 380 nerve terminals for wild-type, heterozygous and homozygous neurons, respectively; Kolmogorov-Smirnov test). These results suggest that miniature vesicle release occurred more frequently at individual nerve terminals of heterozygous and homozygous neurons than at those of their wild-type counterparts.
One important assumption of our imaging method is that the decay of FM intensity reflects miniature vesicle release rather than photobleaching of the FM dye. In order to test this assumption, we measured the amount of photobleaching in our preparation. We stained the cultured neurons with an aldehyde-fixable form of the FM dye (fixable FM4-64). It has an aliphatic amine for crosslinking by paraformaldehyde. Chemical fixation of the stained neurons allows long-term imaging of FM dye in a nerve terminal by blocking any vesicle release and eliminating other potential mechanisms of FM dye loss from nerve terminals, for example diffusion of an FM molecule within a contiguous lipid membrane. Photobleaching of this FM dye occurred at a rate of only 2–3% over 2 min (time constant of 5736 s) under the imaging conditions used for live-cell imaging. The expected amount and time course of photobleaching are illustrated in Figure 2Af (thin continuous curve), based on the measured time constant and the measured absolute intensity of FM4-64 at time 120 s. This indicates that the amount of photobleaching is small in comparison to the signal, and thus we did not correct for it.
Miniature vesicular release is enhanced at a majority of nerve terminals
We also addressed the question of whether a majority of nerve terminals, or only a small subpopulation, contributed to the increase in ΔFMMinis. For this purpose, we compared the distributions of ΔFMMinis among different genotypes. We reasoned that if only a small population of boutons contributed to the increase in ΔFMMinis, with the remaining population retaining the features seen in wild-type neurons, then the shape of the distribution of ΔFMMinis would be distorted in comparison to that of wild-type neurons. On the other hand, if most of the boutons in mutant neurons contributed to the increase in ΔFMMinis, the distribution of ΔFMMinis would be shifted proportionally from that in wild-type neurons. Interestingly, the cumulative histograms of ΔFMMinis intensity in mutants can be fit well when the values in wild-type histogram are multiplied by a constant factor, 1.20 in the case of heterozygotes and 1.34 in the case of homozygotes (Fig. 2Bb and c). Thus, simple multiplication of wild-type distribution by constant factors reproduced the intensity distribution in mutant neurons, indicating that a majority of the nerve terminals in mutant neurons contributed to increased miniature FM release (ΔFMMinis).
Taken together, our imaging results demonstrate that miniature vesicle release is more frequent at individual nerve terminals of heterozygous and homozygous neurons. It is likely that this change plays a significant role in the shortening of mEPSC interevent intervals demonstrated in Figure 1, because a majority of hippocampal neurons and nerve terminals in culture are glutamatergic (Benson et al.,1994; Hartman et al.,2006), as is also the case in situ (Jinno and Kosaka,2010; Megias et al.,2001).
There is no change in the number of nerve terminals of heterozygous ΔE-torsinA neurons
The electrophysiological and imaging data suggest that miniature glutamate release events from individual nerve terminals occur more frequently in ΔE-torsinA mutant neurons than in their wild-type counterparts. We next examined whether changes in the number of glutamate-releasing nerve terminals underlie the shortened interevent intervals (Fig. 1C). This was achieved by counting the number of glutamatergic nerve terminals. Glutamatergic terminals were identified by immunocytochemical staining for the glutamatergic marker, vesicular glutamate transporter 1 (VGLUT1). The measurement was confined to synaptic VGLUT1 puncta that we defined as the puncta that colocalized with or were adjacent to any part of the dendritic tree identified by co-staining with the dendritic marker MAP2 (Figs. 3A–3C). Such puncta correspond to the nerve terminals that release glutamate and affect the mEPSCs detected by electrophysiology. The numbers of the terminals, normalized to the averaged values in the wild-type neurons, were 1.00 ± 0.09 in wild-type neurons, 0.88 ± 0.07 in heterozygous neurons, and 1.34 ± 0.10 in homozygous neurons (mean ± sem, N = 37, 28 and 38 image fields, respectively). The value in heterozygous neurons did not differ significantly from that in wild-type neurons (p > 0.05, t-test) (Fig. 3D), suggesting that the numbers of glutamatergic nerve terminals did not contribute to the shortening of mEPSC interevent intervals in heterozygous neurons. This is partly supported by the fact that the densities of neuronal somata were indistinguishable among different genotypes (Table I), although the number of somata is related to nerve terminal numbers only indirectly.
Notably, the number of glutamatergic nerve terminals was higher in homozygous neurons than in their wild-type counterparts (p < 0.02, t-test) (Fig. 3D), and this could partly account for the shorter interevent intervals of mEPSCs in homozygous neurons. It is possible that testing of a much larger sample size would reveal more glutamatergic terminals in heterozygous neurons than in wild-type neurons. Nevertheless, the observed difference in phenotype between heterozygous and homozygous neurons (Fig. 3D) indicates that the homozygous neurons have abnormalities beyond those exhibited by their heterozygous counterparts.
Individual mEPSC events are modified in the context of the ΔE-torsinA mutation, without a change in synaptic strength
We analyzed individual mEPSCs to assess the amount of excitability imparted to the postsynaptic neurons by a single, miniature glutamate release event (quantal glutamatergic transmission). Known as synaptic strength, this factor is defined as the unitary synaptic charge, and measured as the area of the mEPSCs (pA * ms) (Thiagarajan et al.,2002). We evaluated the synaptic strength and the associated parameters of individual mEPSC events (Fig. 4).
Representative mEPSC traces are shown for a wild-type neuron (Fig. 4Aa, black), a heterozygous neuron (Fig. 4Ab, red), and a homozygous neuron (Fig. 4Ac, green). An overlay of the raw traces illustrates the subtle differences in absolute values (Fig. 4Ad). An overlay of traces normalized to the peak amplitudes illustrates the differences in the decay time course (Fig. 4Ae). When the statistical significance was assessed at P = 10−30 by the Kolmogorov-Smirnov test, the heterozygous neurons (red curves) showed slightly larger mEPSC amplitude (Fig. 4B, P = 10−31) and a smaller decay time (Fig. 4C, P = 10−241) than wild-type neurons (black curves). The result on the decay time is supported by measurement of the decay time constant, which is less sensitive to noise (Fig. 4D, P = 10−36; assessed at P = 10−10 due to lower values of “n”). These changes, in conjunction with a normal rise time (Fig. 4E, P = 10−27), seem to have compensated for each other, because the synaptic strength in heterozygous neurons was indistinguishable from that of wild-type neurons (Fig. 4F). Some of the changes in homozygous neurons (green curves) were opposite to those observed in heterozygous neurons: in the former the amplitude was smaller (Fig. 4B, P = 10−123) and the decay time constant was longer, relative to those in the wild-type neurons (Fig. 4D, P = 10−11), while the rise time was normal (Fig. 4E). However, these changes, similarly as in heterozygous neurons, resulted in a synaptic strength indistinguishable from that of wild-type neurons (Fig. 4F). In light of the complexities inherent in the mechanisms underlying the changes in these individual parameters of mEPSCs (amplitude and decay time course), we have not attempted to dissect them in the current study (Supporting Information text).
Our results demonstrate that heterozygous and homozygous mutant neurons retain the normal synaptic strength for each mEPSC event (Fig. 4F). They also indicate that, in these mutant neurons, the total excitation of the postsynaptic neurons will be increased during a given amount of time, because the overall miniature release events were more frequent (Figs. 1 and 2).
We have shown that the mEPSC interevent intervals in heterozygous and homozygous ΔE-torsinA neurons are shorter than in their wild-type counterparts, and that more FM dye is released from preloaded synaptic vesicles in the mutant neurons, during miniature events recorded over a fixed amount of time. These data were obtained in the presence of TTX, but in the absence of GABAA receptor antagonists, i.e., without interfering with inhibitory activity of the neuronal network. Given that a majority of hippocampal neurons and nerve terminals are glutamatergic, our results suggest that glutamate release via miniature events occurred more frequently in the ΔE-torsinA mutant neurons.
Presynaptic effects of ΔE-torsinA mutation
This study is the first to provide evidence, at the single synapse level, in support of a presynaptic effect of ΔE-torsinA on glutamatergic synaptic transmission. To date, the analysis of DYT1 dystonia models has resulted in only one report of spontaneous glutamate release (sEPSCs) (i.e., recorded in the absence of TTX), and no report on mEPSCs (i.e., recorded in the presence of TTX). The analysis of sEPSCs was carried out in the striatum of transgenic mice overexpressing human ΔE-torsinA, and the frequency of sEPSCs was indistinguishable from that in nontransgenic mice or transgenic controls overexpressing wild-type torsinA (Sciamanna et al.,2009). Unfortunately, it is difficult to compare these data directly with ours, given the difference between sEPSCs and mEPSCs. Also, the sEPSCs recorded in the earlier study were measured in the presence of the GABAA receptor antagonist bicuculline, whereas the mEPSCs in our study were measured in its absence. It will be of interest to establish whether mEPSCs occur more frequently in the striatum, because a positive answer would suggest that the effect of ΔE-torsinA is more widespread throughout the brain, and the opposite answer would be consistent with the observed differences in mEPSCs reflecting differences of the systems studied—e.g., with respect to the brain region (striatum vs. hippocampus) and the animal model (transgenic vs. knock-in mice, with different levels of torsinA expression).
Our results indicate that the overall impact of miniature events on the postsynaptic neurons will be greater in the mutants than in wild-type neurons. This notion is supported by several findings: first, the observed mEPSC events were more frequent; second, there was no change in the synaptic strength (unitary synaptic charge) despite modest changes in the individual mEPSC parameters; and third, cumulative FM release as a consequence of miniature events (ΔFMMinis) was higher in mutants. These miniature events can have a significant influence on neuronal functions, although the extent to which glutamatergic synaptic transmission contributes to setting of the ambient glutamate concentration in the extracellular space remains a matter of debate (Featherstone and Shippy,2008; Hascup et al.,2010; Herman et al.,2011). The mEPSCs consistently influence the membrane potential (Carter and Regehr,2002; Otmakhov et al.,1993; Sharma and Vijayaraghavan,2003), and the density and length of dendritic spines (McKinney et al.,1999). They can also have context-dependent effects. For instance, the small quantities of glutamate released through miniature events are able to suppress protein synthesis within dendrites maintained under normal conditions (Sutton et al.,2006), yet enhance protein synthesis in the dendrites of neurons subjected to endoplasmic reticulum stress (Nosyreva and Kavalali,2010). The importance of the miniature release of glutamate is further highlighted by the fact that this phenomenon is altered in neurological and psychiatric disorders (e.g., Li et al.,2011; Pratt et al.,2011).
We have recently shown that the evoked release of FM dyes from nerve terminals is facilitated in both heterozygous and homozygous ΔE-torsinA neurons during strong stimulation (Kakazu et al.,2012). This, together with the current finding that miniature releases of FM dyes and glutamate were more frequent in these mutant neurons, indicates a general enhancing influence of ΔE-torsinA on the presynaptic mechanisms of glutamatergic synaptic transmission. Such an influence is broadly consistent with an enhancement of evoked glutamatergic transmission demonstrated at cortico-striatal synapses in animal models of DYT1 dystonia. The long-term depression was lost in transgenic mice overexpressing human ΔE-torsinA (Martella et al.,2009) and heterozygous ΔE-torsinA knock-in mice (Dang et al.,2012). In addition, long-term potentiation (LTP) was enhanced in amplitude, and did not recover to pre-LTP levels (lack of synaptic depotentiation) (Martella et al.,2009). All these changes in long-term synaptic plasticity are expected to enhance network excitation.
Unknown mechanisms underlying presynaptic changes
How can torsinA influence presynaptic functions? The details of the underlying mechanism remain unknown. However, torsinA has been found to interact with at least two binding partners, snapin and the subunit 4 of COP9 signalosome (CSN), both of which are involved in regulating vesicle exo-endocytosis (Granata et al.,2011; Granata et al.,2008). According to one hypothesis, the presence of the ΔE-torsinA mutation leads to mislocalization of snapin and other synaptic proteins to sites outside presynaptic nerve terminals, thus causing synaptic abnormalities. However, it is likely that the process is not straightforward (Granata et al.,2009). For example, cultured cerebral cortical neurons obtained from snapin knock-out mice showed a reduced frequency of mEPSCs and a reduced number of synapses (Pan et al.,2009), in contrast to our results (Figs. 1 and 3). It is possible that loss of presynaptic proteins, such as snapin, is a primary cause of the observed abnormalities, but that regional differences underlie the specific effects in cerebral cortical neurons of the snapin knock-out mice (Pan et al.,2009) vs. the hippocampal neurons of the ΔE-torsinA knock-in mice (current results). Alternatively, ΔE-torsinA could impact mEPSCs by indirect mechanisms, for example involving snapin and other binding partners, and thereby leading to general changes in synapses or in the endomembrane system (Warner et al.,2010).
One puzzling observation was that in spite of fact that heterozygous and homozygous neurons shared some properties, certain others differed. The shared features were the more frequent miniature release of glutamate (shortened interevent intervals of mEPSCs, Fig. 1), the more frequent miniature release of FM dyes (increased amount of ΔFMMinis, Fig. 2), unchanged rise time and synaptic strength of mEPSCs (Fig. 4), and the lack of changes in membrane properties and somata density (Table I). Discrepancies arose with respect to: the number of VGLUT1-positive nerve terminals (increased in homozygous neurons; Fig. 3); and the amplitudes and decay time constants of mEPSCs (changes in opposite directions; Fig. 4). We also noticed both similarities and differences among mutant neurons when we evaluated the evoked release of FM dyes from nerve terminals (Kakazu et al.,2012). Although most features were shared, the size of the total recycling pool of synaptic vesicles was unchanged in heterozygous neurons, but reduced in homozygous neurons. Currently, we have no clear explanation for why heterozygous and homozygous neurons share some features but not others. It is possible that homozygous neurons are subject to additional changes that are qualitatively different from those arising from gene dosage. This notion is supported by a number of reported differences between heterozygotes and homozygotes. One is the early postnatal lethality of the latter: homozygous mice die within a few days (Goodchild et al.,2005) or up to 21 days after birth, depending on the mouse strain (Tanabe et al.,2012), whereas heterozygous mice do not (Goodchild et al.,2005; Tanabe et al.,2012). Another difference is the presence of the nuclear membrane abnormality in the latter: in homozygous neurons, vesicular structure is present within the lumen of nuclear envelope, but this is not the case in heterozygous neurons (Goodchild et al.,2005; Kim et al.,2010).
Of related interest is the fact that the miniature release of FM dyes was increased in both heterozygous and homozygous neurons (Fig. 2), whereas the size of the total recycling pool, as evaluated by the evoked release of FM dyes, was unchanged or even reduced (see above). A general correlation has been noted among the miniature, spontaneous and evoked responses (e.g., Prange and Murphy,1999), but according to a recent proposal, distinct pools of synaptic vesicles can be independently regulated to achieve different types of release (Ramirez and Kavalali,2011; Ramirez et al.,2012). It is thus possible that the increased miniature release of FM dyes is in part due to an increase in the size of the pool of vesicles involved in the miniature release. It is also possible that there are changes in the amount of FM dye released per vesicle during exocytosis, and the sensitivity of the exocytosis machinery to cytoplasmic Ca2+. Further details of the synaptic vesicle pools and presynaptic functions in the context of DYT1 dystonia await examination.
Implications for DYT1 dystonia
In the knock-in model of DYT1 dystonia studied here, the presence of one wild-type and one mutated allele produce a condition paralleling that in patients with the autosomal-dominant DYT1 dystonia (heterozygous for ΔE-torsinA). Importantly, the levels of torsinA protein are not higher than wild-type in either the knock-in mice (Goodchild et al.,2005) or the patients (Goodchild and Dauer,2004; Goodchild et al.,2005). This is in contrast to other experimental models in which the protein is overexpressed.
We acknowledge that the primary neuron cultures used in the current study represent a very simplified experimental system. The number of synapses and the environment in culture will be different from those in other preparations, such as acute brain slices and the whole brain. The relevance of our results will need to be tested subsequently in other preparations of higher complexity. Notwithstanding this drawback, our results have at least two important implications for the pathophysiology of DYT1 dystonia as described below.
First, our study demonstrates that the ΔE-torsinA neurons have a presynaptic phenotype. This point is consistent with the accumulating lines of evidence indicating that mouse models of DYT1 dystonia exhibit abnormalities in synaptic transmission involving the neurotransmitters, e.g. glutamate, GABA, acetylcholine, dopamine, adenosine and serotonin (e.g. Grundmann et al.,2007; Martella et al.,2009; Page et al.,2010; Sciamanna et al.,2009; Sciamanna et al.,2011; Yokoi et al.,2011; Zhao et al.,2008). Furthermore, it has been suggested that ΔE-torsinA can influence a subset of presynaptic functions (Balcioglu et al.,2007; Bao et al.,2010; Granata et al.,2011; Granata et al.,2008; Hewett et al.,2010; Kakazu et al.,2012; Page et al.,2010; Sciamanna et al.,2009). Our results demonstrate that presynaptic abnormalities can exist in a glutamatergic system. They are consistent with the notion that multiple neurotransmitter systems can be affected (Grundmann et al.,2007).
Second, the enhancement of mEPSC properties at the cellular level might be associated with the suggestion that CNS excitability is enhanced in patients with DYT1 dystonia (Carbon et al.,2010), although defective GABAergic inhibition was proposed as a potential mechanism based on clinical analysis (Edwards et al.,2003). The motor symptoms of DYT1 dystonia have motivated many researchers to direct their analyses to brain regions associated with motor control, for example basal ganglia. Recent brain imaging and clinical electrophysiology analyses have revealed that other brain regions are also affected (e.g., cerebellum, thalamus and cerebral cortex) (Neychev et al.,2011), and that human ΔE-torsinA carriers can exhibit subtle but recognizable abnormalities, even in the absence of overt motor symptoms. The identified endophenotypes include a wide variety of defects in sensory processing and sensory-motor coordination (Carbon et al.,2011; Fiorio et al.,2007), electrophysiological responses (Edwards et al.,2003), and sensorimotor cortical activity (Carbon et al.,2010). Additionally, carriers of the ΔE-torsinA mutation have been found to be susceptible to recurrent major depression (Heiman et al.,2004). Thus, it is conceivable that the ΔE-torsinA mutation impacts a larger spectrum of CNS regions than was previously appreciated. Although the neurons of the hippocampus may not be the major cause of the motor symptoms observed in dystonia, our results have the potential to shed light on disease mechanisms that are broadly relevant across multiple components of the CNS. An important next step will be to evaluate the presynaptic functions of other neurotransmitters in addition to glutamate, and those functions in other brain regions, in order to assess the universality of our findings.
The authors thank members of the Harata lab for helpful discussions throughout the execution of this project.