We prepared dissociated cultures from hippocampal pyramidal neurons from WT or Clcn3-/- mice (kindly provided by Dr. Thomas Jentsch) at postnatal day 1 as described previously (Guzman et al., 2010). Since Clcn3-/- mice show selective degeneration of the hippocampus starting at an age of about 3 weeks (Stobrawa et al., 2001), we performed our experiments on culture days 14–15 to avoid alterations of neuronal function by potential ultrastructural alterations.

Whole-cell voltage clamp recordings were performed on pyramidal neurons from WT or Clcn3-/- neurons (Guzman et al., 2010) using an Axopatch 200B amplifier (Molecular Devices, Sunnyvale, CA, USA). Patch pipettes with resistances between 3 and 4 MΩ were filled with intracellular solution containing (in millimolars) 137.5 K-gluconate, 11 NaCl, 2 MgATP, 0.2 Na₂GTP, 1.1 EGTA, 11 HEPES, 11 D-glucose, pH 7.3. Only cells with access resistances of 6–10 MΩ were analyzed, and 80–85% of the access resistance was compensated. Currents were filtered at 5 kHz and digitized at 50 kHz. The standard extracellular solution consisted of (in millimolars) 130 NaCl, 10 NaHCO₃, 2.4 KCl, 4 CaCl, 4 MgCl₂, 10 HEPES, 10 D-glucose, pH 7.4 with NaOH. The osmolarity of intra- and extracellular solution was adjusted to 310 mOsm with D-glucose. Action potential-evoked release was studied after addition of 25 μM bicucullin to the extracellular solution. For experiments characterizing miniature EPSCs (mEPSCs) the extracellular solution was supplemented with 1 μM tetrodotoxin (Sigma-Aldrich, Schnelldorf, Germany), 25 μM bicucullin (Tocris Bioscience, Ellisville, MI, USA) and 25 μM APV [(2R)-amino-5-phosphonovaleric acid; (2R)-amino-5-phosphonopentanoate; Tocris Bioscience, Ellisville, MI, USA].

Quantal signals were recorded by clamping neurons to –70 mV for periods of 60 s. mEPSCs were detected as spontaneous events with peak amplitudes >15 pA (~5 times the S.D. of the background noise, e.g., WT: 3.2 ± 0.08 pA, Clcn3-/-; 3.05 ± 0.09 pA, n = 5) and total charges – estimated as integral over the mEPSC >25 fC were analyzed using a commercial software (Mini analysis, Synaptosoft, Version 6.0.3, Decatur, GA, USA). Action potential-evoked EPSCs were recorded using an EPC 10 amplifier controlled by PatchMaster (HEKA, Lambrecht, Pfalz, Germany) software. Cells were clamped to -70 mV, and EPSCs were elicited from hippocampal neurons by local extracellular stimulation as described (Maximov et al., 2007a). Synaptic responses were triggered by 0.5 mA/1 ms current injection at frequency of 0.2 Hz via bipolar electrode (PI2CEA3 concentric bipolar electrode, tip diameter 2–3 μm platinum/iridium, Hofheim, Germany ) placed at a distance of 200–250 μM from the patched cell. For paired-pulse experiments, neurons were stimulated by pairs of extracellular current application with 20 ms interstimulus interval. Peak current amplitudes were measured from baseline current amplitudes determined before the stimulation. EPSCs – evoked by 1 ms stimuli and- usually had a time-to-peak duration of about 2 ms. Artifacts and current peak amplitudes are thus normally easy to separate (Guzman et al., 2010). Moreover, possible contaminations of synaptic currents by stimulation artifacts can be easily detected by the different time courses of EPSC and artifact. We carefully checked all experiments for deviations of measured EPSCs from the typical time course with rapid rise time (20–80% rise time of 0.26–0.50 ms) and decay time constants of 4.6–10.0 ms. EPSCs were analyzed using FitMaster (HEKA) and Origin (OriginLab, USA) software. For comparison of action potential-evoked postsynaptic currents in WT and Clcn3-/- cultures were alternatingly studied at the same setup with the bipolar electrode set to the same distance from the cell body for all evaluated cells. To estimate the size of the readily releasable pool (RRP) we measured synaptic responses evoked by the fast application of hypertonic solution for 5s (500 mM sucrose) using a gravity-fed fast flow system. To test for possible differences in postsynaptic glutamate receptor sensitivity we evoked postsynaptic currents in WT and Clcn3-/- cultures by applying 100 μM L-glutamate glutamate using a FemtoJet perfusion system (Eppendorf, Germany). In these experiments a custom-made perfusion pipette was positioned close to the recording patch-pipette and glutamate was applied for 100 ms. Experiments were performed in the absence as well as in the presence of 200 μM γ-DGG added to the bath solution.

All experiments were performed with at least three different preparations with WT and Clcn3-/- mice, and all comparisons were between cultures from WT and Clcn3-/- littermates. All experiments were performed at room temperature (22–23°C).

After transferal to a 200 μm deep aluminum platelet (Microscopy Services, Flintbek, Germany) fresh hippocampal tissues were frozen using a Leica HPM 100 (Wetzlar, Germany), followed by freeze substitution in 0.1% tannic acid at -90°C for 48 h in a Leica AFS2 (Wetzlar, Germany). Solutions were then changed to 2% OsO4, and the temperature was raised initially to -20°C and after 7 h to 4°C. Samples were washed in acetone and embedded in EPON. Fifty nanometers ultrathin sections were post-stained with 4% uranyl acetate and lead citrate (Reynolds, 1963), mounted onto form var coated copper grids, stained with 4% uranyl acetate and lead citrate (Reynolds, 1963) and then examined in a Morgagni TEM (FEI, Hillsboro, OR, USA), operated at 80 kV. Images from different tissues blocks of the same animal were taken with a 2K side mounted Veleta CCD camera, binned to 1024 × 1024 pixels. Experiments were performed on hippocampi of at least three different WT and Clcn3-/- animals.

Neuronal cultures were blocked and permeabilized with 5% goat serum (Sigma-Aldrich, Schnelldorf, Germany ) and 1% Triton X100 for 45 min after fixation in PBS containing 4% paraformaldehyde at room temperature. Primary antibodies against MAP2 (rabbit polyclonal, Synaptic Systems, Göttingen, Germany) and anti-VGLUT1 (mouse monoclonal, Synaptic Systems, Göttingen, Germany) were diluted in 1% goat serum and 0.1% Triton X100 in PBS and then added to cells for 60 min. Subsequently, a secondary antibody linked to Alexa-Fluor 633 and 488 (Invitrogen, Darmstadt, Germany) was applied for 60 min. After washing in PBS, coverslips containing cells were either imaged immediately or mounted on a glass slide with 2 μl Fluoromount-G (SouthernBiotech).

Images were acquired using a Leica DM IRD (Wetzlar, Germany) inverted microscope equipped with a 63× oil objective and analyzed with NIH imageJ 1.45. Immuno-positive spots were determined using a threshold-based detection routine with a threshold adjusted to dendritic background signals. Immunosignals were quantified as mean fluorescent intensity, and synaptic density was determined by counting the number of VGLUT1-stained puncta per 50 μm dendrite length identified by MAP2 staining.

Cell surface expression of GluR1 receptors was assayed with a modification of cell surface biotinylation methods (Du et al., 2004). Hippocampal cultures obtained from Clcn3-/- mice and WT litter mate were washed with ice-cold PBS (with calcium and magnesium pH 7.4; Invitrogen, Darmstadt, Germany) to prevent receptor internalization. After three washes with PBS cells were incubated with sulfo-NHS-LC biotin (0.25 mg/ml in ice-cold PBS) (ThermoScientific, Rockford, IL, USA) for 30 min. The reaction was stopped by removal of the above solution and incubation for 20 min in ice-cold PBS containing 10 mM glycine. After threefold washing and lysis with RIPA buffer [in millimolars): 20 HEPES, 100 NaCl, 1 EGTA, 1 Na-orthovanadate, 50 NaF, protease inhibitor (Roche), 1% NP40, 1% deoxicholate, 0,1% SDS pH 7.4], biotinylated proteins were precipitated with immune pure immobilized streptavidin (Pierce, ThermoScientific, Rockford, IL, USA). Biotinylated proteins were separated in 10% SDS-PAGE gel and transferred to nitrocellulose membrane. Membranes were probed with an anti-GluR1 antibody (rabbit polyclonal, 1:700, abcam, Cambridge, UK) or anti-actin (rabbit, poloclonal, 1:1000, sigma, St. Louis, MO, USA) followed by peroxidase-conjugated goat anti-rabbit IgG (1:25000, Invitrogen, Darmstadt, Germany). Immunoreactive bands were visualized by enhanced chemiluminiscence (ECL, AppliChem, Darmstadt, Germany) using a GeneGnome chemiluminiscence imagine system (SYNGENE, Cambridge, UK) and ImageJ software.

We compared expression of VGLUT1 mRNA levels in cultured neurons from four different WT and Clcn3-/- preparations. Total RNA was prepared using the PureLink RNA Mini Kit (Ambion, Darmstadt, Germany ). Reverse transcription was performed from 0.1 μg RNA using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Darmstadt, Germany), and VGLUT1 mRNA was quantified with the 2^-ΔΔCT method using a specific TaqMan Assay for VGLUT1 (Mm00812886_m1, StepOnePlus System, Applied Biosystems, Darmstadt, Germany).

All summary data are given as mean ± SEM. Paired Student’s t analysis was used to test statistical differences with ^*p <0.05, ^**p <0.01, ^***p < 0.001 levels of significance. Authors warrant that any human and/or animal studies have been approved by an appropriate institutional review committee.

Figure 1 shows representative whole-cell patch clamp recordings from miniature excitatory postsynaptic currents from WT and Clcn3-/- neurons. Cells were held at -70 mV, and quantal signals were acquired for a period of 60 s in the presence of 1 μM TTX, 25 μM APV and 25 μM bicuculline. Miniature excitatory postsynaptic currents (Figure 1A) display comparable time courses in WT as well as in Clcn3-/- neurons (Figure 1B, insert), but distinct amplitudes (Figure 1C; WT 28.9 ± 0.6 pA, n = 40 and Clcn3-/- 35.2 ± 1.4 pA, n = 47, p < 0.001) and frequencies (Figure 1D; WT 2.2 ± 0.3 Hz, n = 40 and Clcn3-/- 4.0 ± 0.5 Hz, n = 47, p < 0.01) as illustrated by the amplitude frequency distribution in Figure 1E.

The observed effects on quantal signals might be caused by changes in the glutamate concentration in the synaptic cleft after exocytosis of individual glutamatergic vesicles. To probe ClC-3-dependent alterations in the synaptic glutamate concentration we took advantage of γ-DGG, a rapidly dissociating competitive antagonist of AMPA receptors which block AMPA receptors with lower efficacy at higher glutamate concentrations (Liu et al., 1999). We found that application of 200 μM γ-DGG reduces the mean peak mEPSC amplitude at WT and mutant neurons, however, the effect is less pronounced in the Clcn3-/- than in WT (Figure 2A). The cumulative frequency distribution of mEPSC amplitudes revealed that γ-DGG shifted the distribution of events towards lower values in WT than in Clcn3-/- neurons (Figure 2B). The efficiency of the blocker in reducing the mean peak mEPSC amplitude was ~1.5 fold stronger for WT than for Clcn3-/- neurons, as expected for higher glutamate concentration in Clcn3-/- than in WT synaptic clefts. We found that application of γ-DGG reduces mean peak mEPSC amplitudes by 18 ± 2 % for WT (n = 12, three different cultures), but only by 12 ± 2% (n = 12, three different cultures) for Clcn3-/- neurons (p < 0.05; Figure 2C).

1,2,3,4-Tetrahydro-6-nitro-2,3-dioxo-benzo[f] quinxaline-7-sulfonamide (NBQX) is known to block non-NMDA receptors to the same extent regardless of the glutamate concentration (Liu et al., 1999). Application of NBQX to WT or Clcn3-/- cultures attenuated quantal signals (Figure 2E), and shifted the mEPSC amplitude cumulative frequency distribution to the left (Figure 2F). However, in contrast to γ-DGG, NBQX inhibition was not different between WT and Clcn3-/- (Figure 2G; mean peak mEPSC amplitudes reduction for WT 17.3 ± 1.4%, n = 11 and Clcn3-/- 19.0 ± 1.7, n = 13, p = 0.22, three different cultures). Figures 2D,H depict control mean mEPSC amplitudes from the cultures used in these experiments, illustrating consistently increased quantal signals in Clcn3-/- neurons in the absence of blockers. Taken together, our results indicate that synaptic cleft glutamate concentrations are higher in Clcn3-/- than in WT neurons.

Enhanced glutamate release might be associated with enlarged SV in Clcn3-/- neurons. We performed an ultrastructural analysis of the size of SV at synapses in hippocampal slices from WT and Clcn3-/- neurons (Figures 4A,B). Synapses can be readily identified by a heavily stained postsynaptic density in close proximity to presynaptic terminals filled with clusters of SV. Synaptic vesicle sizes were estimated by measuring the outer vesicle diameter in different blocks of the same animal from at least three independent preparation, with mean values of WT vesicles of 38.2 ± 0.2 nm (n = 350 SV, 22 synaptic terminals) and of 43.8 ± 0.2 nm in Clcn3-/- neurons (n = 550 SV, 15 synaptic terminals, p < 0.001). The distribution of synaptic vesicle outer diameters exhibits a significant shift towards larger values in Clcn3-/- neurons (Figure 4B). Assuming a spheroidal geometry of SV these values correspond to 1.5 fold increased vesicle volumes in knock-out as compared to WT neurons (Figure 4C). The increased magnitude of synaptic events in Clcn3-/- neurons thus corresponds morphologically to increased vesicle sizes.

Vesicular glutamate transporters (VGLUTs) are necessary for glutamate accumulation in SV, and the quantal size of excitatory postsynaptic currents is therefore critically dependent on expression levels of these transporters (Wilson et al., 2005). We stained WT and Clcn3-/- neurons with anti-VGLUT1 antibodies and estimated protein expression of VGLUT by determining intensity levels at distinct locations. No differences in event numbers or in cumulative frequency distributions of VGLUT1 intensities at synapses were found. (WT n = 724 synapses and Clcn3-/- n = 943 synapses, p = 0.6; Figures 5A,B). Moreover, mRNA levels of VGLUT1 were not significantly different between WT and Clcn3-/- neurons (four different cultures, p = 0.52; Figure 5C).

GluR1 is highly expressed in hippocampal pyramidal neurons and changes in its number are known to alter synaptic strength (Harms et al., 2005; Han and Stevens, 2009). We determined total expression levels and surface density of GluR1 AMPA receptor subunit using western blotting of full lysates and surface biotinylation without any discernible differences between WT and Clcn3-/- neurons (Figures 5D,E). We furthermore compared WT and Clcn3-/- postsynaptic currents evoked by direct application of 100 μM L-glutamate. No difference was observed between WT (2.9 ± 0.2 nA, n = 7) and Clcn3-/- (3.1 ± 0.6 nA, n = 6, p = 0.38) neurons. To exclude the possibility that saturation might have masked possible differences in receptor activation experiments were repeated in the presence of 200 μM of γ-DGG. As expected, evoked current in responses to 100 μM L-glutamate were smaller in the presence of 200 μM γ-DGG than the corresponding responses to 100 μM L-glutamate alone, but no difference was observed for WT (2.2 ± 0.2 nA, n = 6) and Clcn3-/- (1.9 ± 0.1 nA, n = 5, p = 0.2).

The frequency of mEPSCs critically depends on the number of synapses in our experimental system. We determined the numbers of synapses for WT and mutant neurons by identifying dendrites by MAP2 staining, and determining the density of synapses as number of VGLUT1-positive dots per 50 μm dendrites (Figures 5F,G). This approach provided similar synapse numbers for WT and Clcn3-/- (p = 0.36). Taken together, these results indicate that neither the number of synapses nor expression levels/sensitivity of AMPA receptors and VGLUTs differ in WT and Clcn3-/- neurons.