Neuronal activity modulates the membrane diffusion of postsynaptic γ-aminobutyric acid (GABA)A receptors (GABAARs), thereby regulating the efficacy of GABAergic synapses. The K289M mutation in GABAARs subunit γ2 has been associated with the generalized epilepsy with febrile seizures plus (GEFS+) syndrome. This mutation accelerates receptor deactivation and therefore reduces inhibitory synaptic transmission. Yet, it is not clear why this mutation specifically promotes febrile seizures. We show that upon raising temperature both the number of GABAARs clusters and the frequency of miniature inhibitory postsynaptic currents decreased in neurons expressing the K289M mutant but not wild-type (WT) recombinant γ2. Single-particle tracking experiments revealed that raising temperature increases the membrane diffusion of synaptic GABAARs containing the K289M mutant but not WT recombinant γ2. This effect was mediated by enhanced neuronal activity as it was blocked by glutamate receptor antagonists and was mimicked by the convulsant 4-aminopyridine. Our data suggest the K289M mutation in γ2 confers GABAARs with enhanced sensitivity of their membrane diffusion to neuronal activity. Enhanced activity during hyperthermia may then trigger the escape of receptors from synapses and thereby further reduce the efficacy of GABAergic inhibition. Alteration of the membrane diffusion of neurotransmitter receptors therefore represents a new mechanism in human epilepsy.
γ-aminobutyric acid (GABA) acting on GABAA receptors (GABAARs) mediates fast inhibitory synaptic transmission in the brain. Most GABAARs at cortical synapses are heteropentamers of α1-3, β2/3, and γ2 subunits (Lüscher and Keller 2004). γ2 subunit is required for benzodiazepine binding (Mohler et al. 2002). It also affects the kinetics and conductance of GABAAR channels (Gunther et al. 1995) and is required for postsynaptic clustering (Essrich et al. 1998) and synaptic maintenance (Schweizer et al. 2003). Mutations in the Gabrg2 gene, encoding the γ2 subunit, have been associated with generalized epilepsy syndromes including febrile seizures (FS) and generalized epilepsy with febrile seizures plus (GEFS+) (Macdonald et al. 2010). In particular, the missense mutation K289M shows autosomal dominant inheritance and affects a conserved residue in the short extracellular loop between transmembrane domains II and III. Although this mutation was initially suggested to reduce both membrane expression of γ2 (Kang et al. 2006) and the amplitude of GABA currents in heterologous cells (Baulac et al. 2001; Ramakrishnan and Hess 2004), normal membrane traffic and synaptic aggregation were observed in hippocampal neurons (Eugène et al. 2007). Functionally, the K289M mutation accelerates the deactivation of GABA currents by reducing mean channel open time (Bianchi et al. 2002; Hales et al. 2006). In neurons, this effect accelerates the decay of inhibitory synaptic currents (Eugène et al. 2007), thereby reducing the efficacy of synaptic inhibition.
How may a general reduction of the efficacy of GABAergic synapses specifically promote FS? In heterologous cells, an increase in temperature has been shown to rapidly reduce surface expression of K289M mutant γ2. This reduction may reflect increased receptor endocytosis and might contribute to the emergence of FS in patients (Kang et al. 2006). However, in the absence of synaptic specialization, the mechanisms involved in this temperature-dependent endocytosis as well as its relevance to synaptic GABA signaling are difficult to address in heterologous cells. In neurons, postsynaptic receptor content relies on 1) the insertion/internalization of receptors at the membrane (Collingridge et al. 2004), 2) their lateral diffusion into and out of synapses, and 3) their postsynaptic anchoring through scaffold interactions (Triller and Choquet 2008). GABAARs display free Brownian-type diffusion in extrasynaptic membrane, confined movements within synapses, and rapid translocation between these compartments (Jacob et al. 2005; Thomas et al. 2005; Bogdanov et al. 2006; Lévi et al. 2008; Bannaï et al. 2009; Muir et al. 2010). In addition, the diffusion properties of GABAARs are modulated by neuronal activity. Receptor escape from synapses is facilitated by increased excitatory synaptic activity, resulting in reduced synaptic receptor content and efficacy (Bannaï et al. 2009; Muir et al. 2010). Since neuronal activity is strongly dependent on temperature (e.g., Andersen and Moser 1995; Volgushev et al. 2000; Postlethwaite et al. 2007), these observations predict increased temperature may result in depletion of synaptic GABAARs.
Here, we report that raising temperature alters the postsynaptic clustering of GABAARs and decreases GABA-mediated miniature inhibitory postsynaptic current (mIPSC) frequency in neurons expressing recombined K289M mutant but not wild-type (WT) γ2 subunit. This effect was associated with a rapid increase in the lateral diffusion of synaptic K289M γ2. This temperature-dependent increase in mutant γ2 diffusion was mediated by enhanced excitatory transmission. These results indicate that the K289M mutation in γ2 confers synaptic GABAARs with enhanced sensitivity to increased neuronal activity that likely contributes to the involvement of this mutation in FS.
Materials and Methods
Primary Hippocampal Cultures and Transfection
Hippocampal neurons were prepared from E19 Sprague–Dawley rat pups, as described (Eugène et al. 2007). Cells were plated at a density of 2.5 × 104 cells/cm2 and cultured in a CO2 incubator at 37 °C for 3–4 weeks in Neurobasal medium supplemented with B27 (Invitrogen, Cergy Pontoise, France), 2 mM glutamine, and penicillin/streptomycin. At 14 days in vitro, neurons were transfected with monomeric Red Fluorescent Protein (mRFP) (0.118 μg/cm2) and recombinant Green Fluorescent Protein (GFP)-tagged WT or K289M Gabrg2 constructs (0.394 μg/cm2) using Lipofectamine (Invitrogen) according to manufacturer’s instructions (DNA:lipofectamine ratio 1:3 μg/μL) and used for biological assays within 8–11 days posttransfection.
Due to the slow synaptic flux (exit/entry) of receptors, changes in receptor lateral diffusion (5–10 min) precede changes in the density of receptor at synapses (30–60 min) (Lévi et al. 2008). Neurons were therefore preincubated 10 min versus 1 h when studying the behavior of molecules with single particle tracking (SPT) or that of receptor population with cluster imaging and electrophysiology. For SPT experiments, neurons were preincubated 10 min at 27, 31, 37, or 41 °C (depending on the experiment) in imaging medium (see below for composition) following quantum dot (QD) labeling. They were then used within 30 min. For cluster imaging, cells were preincubated in culture medium 1 h in a CO2 incubator set at 41 °C. Cells were then transferred to a recording chamber in imaging medium at the appropriate temperature (37 or 41 °C) and used within 30 min (Figs 1 and 3E,F). For electrophysiology, neurons were preincubated in culture medium 1 h in a CO2 incubator set at 27, 31, 37, or 41 °C depending on the experiment. Cells were then transferred to a recording chamber in recording medium (see below for composition) at 31 °C (Figs 2 and 5) or at 27 °C (Figs 5 and 6). Neurons were then used for experiments within 15 min, a time range too short to significantly affect the number of postsynaptic receptors (Lévi et al. 2008).
Live Cell Imaging and Analysis
Cells were imaged in a temperature-controlled open chamber (BadController V, Luigs & Neumann, Ratingen, Germany) mounted on an inverted microscope (IX71 Olympus, Rungis, France) equipped with a ×60 objective (Numerical Aperture [NA] = 1.42, Olympus). GFP and mRFP were detected using X-Cite 120PC lamp (EXFO, Mississauga, Ontario, Canada) with appropriate filters (excitation: HQ470/40 and D540/25, dichroic: Q495LP and 565DCLP, and emission: HQ525/50 and D605/55, Chroma Technology, Bellows Falls, VT). GFP and mRFP images were acquired with an EMCCD camera (ImagEM; Hamamatsu Photonics, Massy, France) using HC Image software (Hamamatsu). Exposure time was determined on highly fluorescent cells to avoid pixel saturation. All GFP and mRFP images from a given culture were acquired with the same exposure time and acquisition parameters. Quantification was performed using MetaMorph software (Roper Scientific, Evry, France). A user-defined intensity threshold was applied to select clusters and prevent coalescence. Data were obtained from 39 to 55 cells of 3 independent cultures.
Single Particle Imaging
Neurons were incubated for 5 min at 37 °C with a rabbit primary antibody against GFP (20–40 ng/mL, Roche Diagnostics, Meylan, France), washed, and incubated for 5 min at 37 °C with a secondary biotinylated Fab antibody (0.5 μg/mL, Jackson ImmunoResearch, Newmarket, UK). Following washes, coverslips were then incubated for 1 min at 37 °C with streptavidin-coated QDs emitting at 605 nm (0.5 nM, Invitrogen) in borate buffer (50 mM) supplemented with sucrose (200 mM). Cells were then washed and imaged in the presence of appropriate drugs after 5 min preincubation. All washes, incubation steps, and cell imaging were performed in imaging medium prepared with minimum essential medium without phenol red, supplemented with 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) buffer (20 mM), glucose (33 mM), glutamine (2 mM), Na+ pyruvate (1 mM), and B27 supplement (1×) (all from Invitrogen).
Cells were imaged in a temperature-controlled open chamber mounted on an Olympus IX71 inverted microscope equipped with a ×60 objective (NA = 1.42). GFP, mRFP, and QDs were detected using X-Cite 120PC lamp with appropriate filters (excitation: HQ470/40, D540/25, and D455/70; dichroic: Q495LP, 565DCLP, and 500DCXR; and emission: HQ525/50, D605/55, and HQ605/20, GFP and mFRP filters from Chroma Technology; QD filters from Omega Optical, Brattleboro, VT). Real-time fluorescence images were obtained with an integration time of 75 ms with the Hamamatsu ImagEM EMCCD camera with 512 consecutive frames acquired under HC Image. Cells were imaged within 30 min following primary antibody incubation. For each SPT experiment, QDs dynamics were measured on 113 ± 13 QDs per culture from 10 to 20 movies recorded from 2 separate coverslips per culture. Data were obtained from 2 to 11 independent cultures. The proportion of synaptic QDs was 25.9 ± 3.8% (average ± standard error of the mean [SEM]) of the bulk population of QDs.
SPT and Analysis
Single-molecule tracking was performed with custom software (Bonneau et al. 2005) using Matlab (The Mathworks, Meudon, France). Single QDs were identified by their blinking property (Alivisatos et al. 2005). The center of the fluorescence spots was determined with a spatial accuracy of ∼10 nm by cross-correlating the image with a Gaussian fit of the point spread function (for details, see Triller and Choquet 2008). QD trajectories were reconstructed as in Ehrensperger et al. (2007). Subtrajectories of single QDs with ≥20 points without blinks were retained. Synaptic versus extrasynaptic trajectories were determined from overlay of trajectories image and GFP image of GFP-coupled recombinant WT or K289M γ2 clusters. GFP images were first median-filtered (kernel size, 3 × 3 × 1) to enhance cluster outlines. Then, a user-defined intensity threshold was applied to select clusters and avoid their coalescence, and a binary mask was generated. Trajectories were synaptic when overlapping with the mask or extrasynaptic for spots 2 pixels (440 nm) away (Dahan et al. 2003). Diffusion coefficients were calculated from the longest subtrajectories of single QDs in the synaptic and extrasynaptic compartment. For each QD, we calculated the mean square displacement (MSD) and diffusion coefficient (D) within extrasynaptic and synaptic compartments. The size of the confinement domain and dwell time (DT) were calculated for synaptic QDs. Values of the MSD plot versus time were calculated for each trajectory with the following formula: , where xi and yi are the coordinates of an object on frame i, N is the total number of frames in the trajectory, dt is the frame acquisition time, and ndt is the time interval over which displacement is averaged (Saxton 1997). For simple, 2D Brownian mobility, the MSD as a function of time is linear with a slope of 4D, where D is the diffusion constant. If the MSD as a function of time tends to a constant value L, the diffusion is confined in a domain of size L. The diffusion coefficient (D) is determined by a fit on the first 4 points of the MSD as a function of time with MSD(ndt) = 4Dndt + b, where b is a constant reflecting the spot localization accuracy. The area in which diffusion is confined can be estimated by fitting the MSD as a function of time with the following formula:, where L2 is the confined area in which diffusion is restricted and Dmac is the diffusion coefficient on a longtime scale (Kusumi et al. 1993). The size of the confinement domain was defined as the side of a square in which diffusion is confined (Kusumi et al. 1993). For details, see Ehrensperger et al. (2007). Synaptic DT was defined as the duration of detection of QDs at synapses on a recording divided by the number of exits as detailed previously (Charrier et al. 2006; Ehrensperger et al. 2007).
Neurons were superfused with a recording medium containing (in mM) 125 NaCl, 20 D-glucose, 10 HEPES, 4 MgCl2, 2 KCl, and 1 CaCl2 (pH = 7.4) in a recording chamber maintained at 31 °C. mIPSCs were recorded in whole-cell mode in the presence of TTX (1 μM), NBQX (20 μM), and D,L-APV (100 μM), with an internal solution containing (in mM) 135 CsCl, 10 HEPES, 10 ethyleneglycol-bis(2-aminoethylether)-N,N,N′,N′-tetra acetic acid (EGTA), 4 MgATP, 1.8 MgCl2, and 0.4 Na3GTP (pH = 7.4). Currents were recorded at −70 mV, with an Axopatch 200B amplifier (Molecular Devices, Wokingham, UK), filtered at 2 kHz and digitized at 20 kHz. Access and input resistance were monitored with −5 mV voltage steps. mIPSCs were detected and analyzed offline using Detectivent software (Ankri et al. 1994). In some experiments, spontaneous synaptic (Fig. 5) or intrinsic (Fig. 6) activity were recorded in voltage-clamp or current-clamp mode, respectively. Spontaneous synaptic activity was recorded in the absence of any postsynaptic receptor antagonist, with a solution containing (in mM) 115 CsMeSO4, 11.5 CsCl, 10 HEPES, 10 EGTA, 4 MgATP, 0.4 Na3GTP, and 1.8 MgCl. Spontaneous intrinsic activity from a resting membrane potential of −60 mV with a solution containing (in mM) 120 KMeSO4, 8 KCl, 10 HEPES, 10 EGTA, and 3 MgCl2.
Peptide Treatment and Pharmacology
The following peptides and drugs were used: myr-P4 (50 μM; Tocris Bioscience, Bristol, UK), TTX (1 μM; Latoxan, Valence, France), NBQX (10 μM), DL-AP5 (100 μM), R,S-MCPG (500 μM; Ascent Scientific, Bristol, UK), and 4-AP (100 μM; Sigma-Aldrich, Lyon, France).
Data are presented as mean ± SEM. Means were compared using the nonparametric Mann–Whitney rank-sum test unless otherwise stated. Tests were performed using SigmaStat software (SPSS, Bois Colombes, France). Cumulative distributions were compared using the Kolmogorov–Smirnov test under StatView (SAS, Grégy-sur-Yerres, France). Differences were considered significant for P values above 5%.
Loss of Synaptic Aggregates of K289M Mutant γ2 Subunit upon Temperature Elevation
Hippocampal neurons were transfected with either WT or K289M (K289M) mutant GABAAR γ2 subunit constructs with GFP fused at their N-terminus. The postsynaptic aggregation of recombinant γ2 subunits was examined using live-cell imaging of GFP in hippocampal neurons maintained at 37 °C. As previously reported (Eugène et al. 2007), the K289M mutation did not affect the membrane expression or aggregation of recombinant γ2. At 37 °C, numerous punctae of recombinant γ2 subunits were detected both on the soma and dendrites of neurons expressing either WT or K289M γ2 (Fig. 1A). Most GFP-labeled recombinant γ2 clusters were instantaneously quenched by live exposure to bromophenol blue (5 mM), indicating that recombinant γ2 was inserted in the membrane (Supplementary Fig. 1). Large recombinant γ2 clusters were mostly localized at inhibitory synapses, as revealed by their close apposition to presynaptic varicosities immunoreactive for the GABA synthesis enzyme, glutamic acid decarboxylase (GAD; 73.7 ± 3.8%, n = 287, GAD positive synapses on 17 dendrites from 6 cells; data not shown).
In heterologous cells, several epilepsy-related mutations in Gabrg2 have been suggested to result in reduced membrane expression of γ2 upon hyperthermia (Kang et al. 2006). However, membrane trafficking and clustering of GABAAR are likely differently regulated in heterologous cells and neurons. We therefore compared the aggregation of recombinant γ2 in primary hippocampal neurons maintained for 1 h at 37 versus 41 °C. In these experiments, the survival of neurons was not compromised as evidenced using the vital die trypan blue (not shown). One-hour exposure to 41 °C had no detectable effect on recombinant WT γ2 clustering but induced a significant decrease in the number of K289M mutant γ2 clusters per 10 μm dendritic length (−27% of control, WT 37 °C, n = 2.2 ± 0.3 from 55 cells; WT 41 °C, n = 2.3 ± 0.3 from 50 cells; KM 37 °C, n = 2.2 ± 0.3 from 50 cells; KM 41 °C, n = 1.6 ± 0.3 from 39 cells; 3 cultures; P = 0.7 for WT and P < 0.05 for K289M γ2; Fig. 1A,B). These observations suggest an increase in temperature of a few degrees significantly reduce postsynaptic aggregation of mutant but not WT γ2 and predict a functional reduction in the efficacy of synaptic inhibition in neurons expressing K289M mutant γ2.
Functional Impact of Temperature Rise on GABAergic Synaptic Transmission in Hippocampal Neurons
In order to examine the functional impact of the temperature-induced reduction of clustering of mutant γ2, we compared the properties of mIPSCs in hippocampal neurons expressing either WT or K289M mutant γ2 (Fig. 2). mIPSCs were pharmacologically isolated by tetrodotoxin (TTX, 1 μM) and the glutamate receptor antagonists D,L-AP5 (100 μM) and NBQX (20 μM). As previously reported (Eugène et al. 2007), mIPSCs recorded from neurons expressing either WT or K289M mutant γ2 had similar frequency (11.4 ± 2.0 vs. 12.1 ± 1.0 Hz, P = 0.9), mean amplitude (−39.4 ± 5 vs. −34.8 ± 3.6 pA, P = 0.6), and onset kinetics (10–90% time to peak, 0.83 ± 0.05 vs. 0.77 ± 0.04 ms, P = 0.33; n = 9–11 cells for both WT and K289M) (Fig. 2A–C). However, their decay time constant was accelerated by 28.8% (14.5 ± 1.0 vs. 10.3 ± 0.4 ms, P < 0.01) in neurons expressing the K289M mutant as compared with WT recombinant γ2 (Fig. 2B,C). Therefore, in steady-state conditions, the major synaptic effect of the K289M mutation is to reduce current charge through GABAARs with no apparent change in unitary conductance or mean number of receptor per synapse (Eugène et al. 2007).
One-hour exposure to a temperature of 41 °C had no significant effect on the mean amplitude, frequency, or decay kinetics of mIPSCs in hippocampal neurons expressing recombinant WT γ2 (Fig. 2A–C). The mean amplitude of mIPSCs was also unaffected by temperature rise in neurons expressing mutant K289M γ2 (36.2 ± 6.6 vs. 34.8 ± 3.6 pA) and was comparable to that of neurons expressing WT γ2 (39.3 ± 4.5 vs. 39.4 ± 5.7 pA, P = 0.3; Fig. 2A–C), suggesting the mean number of receptors per synapse was unchanged. However, mIPSC frequency was reduced by 67% after temperature increase in neurons expressing mutant K289M as compared with WT γ2 (6.2 ± 1.8 vs. 18.6 ± 5.1 pA, P < 0.05; Fig. 2A,C), suggesting that the number of inhibitory synapses containing functional receptors was reduced. This is in agreement with live-imaging data showing a reduced density of inhibitory synapses with K289M γ2 clusters upon temperature increase (Fig. 1).
If some synapses containing recombinant K289M γ2 had lost their postsynaptic receptors upon temperature elevation, then the relative abundance of synapses containing only endogenous WT receptors should increase. Since both types of receptors can be distinguished based on their deactivation kinetics, we would predict mIPSC decay would be slowed in neurons expressing K289M mutant but not WT γ2. Consistent with this prediction, the mean decay time constant of mIPSCs recorded in neurons expressing K289 mutant γ2 increased after 1 h at 41 °C as compared with control (89 ± 5 vs. 71 ± 3% of WT, P < 0.01; Fig. 2B,C). These results suggest that GABAARs containing K289M mutant γ2 subunit may escape from inhibitory synapses upon temperature rise and/or may be partially replaced by receptors containing endogenous WT γ2.
Raising Temperature Promotes Synaptic Escape of Recombinant GABAARs Containing K289M γ2
GABAARs diffuse laterally in the neuronal plasma membrane and rapidly shift between extrasynaptic and synaptic sites (Jacob et al. 2005; Thomas et al. 2005; Bogdanov et al. 2006; Lévi et al. 2008; Bannaï et al. 2009; Muir et al. 2010). Since membrane dynamics properties control receptor content at synapses (Choquet and Triller 2003; Triller and Choquet 2005, 2008), we asked whether enhancing temperature might specifically affect the lateral diffusion of GABAARs containing the K289M mutant γ2 subunit.
The mobility of recombinant γ2 was analyzed using QD-based SPT (Dahan et al. 2003; Bannaï et al. 2006). The surface recombinant γ2 subunits were labeled with an antibody raised against GFP and subsequently labeled with an intermediate biotinylated Fab fragment and streptavidin-coated QD. We first examined the impact of a rise in temperature on the lateral diffusion of recombinant γ2 subunits for bulk population of QDs (i.e., independent of their synaptic vs. extrasynaptic localization). Within 5–10 min after temperature reached 41 °C, the surface explored by individual recombinant WT and K289M γ2 was reduced. Consequently, cumulative distributions of both WT and K289M γ2 diffusion coefficient (D) were shifted toward lower values (WT 37 °C, D = 8.2 ± 0.5 × 10−2 μm2s−1, n = 268; WT 41 °C, D = 6.7 ± 1.0 × 10−2 μm2s−1, n = 270; KM 37 °C, D = 8.8 ± 1.8 × 10−2 μm2s−1, n = 252; KM 41 °C, D = 4.1 ± 0.3 × 10−2 μm2s−1, n = 347; 2 cultures; P < 0.001 for both WT and K289M γ2; Fig. 3A). Protein traffic and endocytosis are temperature-dependent processes and are accelerated at higher temperature. Receptors undergoing endocytosis are immobilized in confined areas, such as coated pits (Tardin et al. 2003; Petrini et al. 2009). Therefore, the reduced mobility and increased confinement of both WT and K289M γ2 upon raising temperature suggest that QD-stained recombinant γ2 may be trapped in endocytic membrane domains. QD staining of receptors in SPT experiments provides access to only a small fraction of the entire membrane population of receptors. This precludes visualization of newly membrane-inserted receptors not yet engaged in internalization and the lateral mobility of which may be sensitive to an increase in temperature.
In order to circumvent this problem, we thus examined the impact of hyperthermia on the lateral mobility of recombinant γ2 while pharmacologically blocking endocytosis using bath application of myristoylated QVPSRPNRAP (myr-P4) peptide. This peptide interferes with dynamin and amphiphysin interaction (Marks and McMahon 1998), thereby blocking a crucial step for endocytosis (Wigge et al. 1997; Marsh and McMahon 1999). In conditions of dynamin-dependent endocytosis blockade, lateral diffusion of WT and K289M γ2 did not differ at 37 °C (WT 37 °C, D = 2.2 ± 0.2 × 10−2 μm2s−1, n = 260, KM 37 °C, D = 2.2 ± 0.4 × 10−2 μm2s−1, n = 360; P = 0.03; 2 cultures; Supplementary Fig. 2). This is coherent with the notion that, at steady states, the mutation does not change the number of receptors per synapse (Eugène et al. 2007, Fig. 1). Furthermore, myr-P4 prevented immobilization of QD-bound γ2 upon raising temperature to 41 °C (Fig. 3B), suggesting that reduced diffusion observed at 41 °C in the absence of the peptide reflected increased confinement of QDs within clathrin-coated pits (Fig. 3A). However, in these conditions, hyperthermia specifically accelerated diffusion of QD-bound mutant (KM 41 °C, D = 3.8 ± 0.3 × 10−2 μm2s−1, n = 403, P < 0.001) but not WT γ2 (WT 41 °C, D = 2.2 ± 0.3 × 10−2 μm2s−1, n = 339, P = 0.1; 2 cultures; Fig. 3B). Extrasynaptic versus synaptic trajectories were segregated by comparison with GFP images of γ2 clusters (see Materials and Methods). Trajectories were at inhibitory synapses when overlapping with γ2 clusters (e.g., red in Fig. 4A) or extrasynaptic (e.g., blue in Fig. 4A) for trajectories 2 pixels (440 nm) away (Dahan et al. 2003). We found that the increased mobility of K289M mutant γ2 at 41 °C was observed for both synaptic and extrasynaptic receptors (synaptic: KM 37 °C, D = 2.1 ± 0.7 × 10−2 μm2s−1, n = 133; KM 41 °C, D = 3.8 ± 0.7 × 10−2 μm2s−1, n = 114, P < 0.001; extrasynaptic: KM 37 °C, D = 2.2 ± 0.2 × 10−2 μm2s−1, n = 192; KM 41 °C, D = 3.8 ± 0.6 × 10−2 μm2s−1, n = 289, P = 0.006; 2 cultures; Fig. 3C). Thus, lateral diffusion of GABAA receptors containing mutant but not WT γ2 subunit is enhanced upon raising temperature from 37 to 41 °C.
We then asked whether elevated temperature might relieve constraints on K289M γ2 diffusion at inhibitory synapses. The MSD versus time relation for K289M γ2 trajectories showed a steeper slope, suggesting that trajectories were less confined at inhibitory synapses at 41 versus 37 °C (data not shown). Accordingly, the mean size of the confinement domain (L) was increased for K289M but not for WT γ2 trajectories upon raising temperature (WT 37 °C, L = 208.7 ± 18.1 nm, n = 84; KM 37 °C, L = 194.9 ± 17.3 nm, n = 125; WT 41 °C, L = 221.7 ± 37.3 nm, n = 38; KM 41 °C, L = 319.9 ± 41.0 nm, n = 91; 2 cultures; P = 0.48 for WT and P = 0.002 for KM; Fig. 3D). Enhanced lateral mobility and lower diffusion constraints on synaptic receptors may alter the time receptors spend at synapses and thereby influence synaptic receptor content (Triller and Choquet 2008). We asked whether temperature impact the time GABAARs containing K289M mutant γ2 spend at synapses. A temperature jump from 37 to 41 °C did not alter DT of WT γ2 (WT 37 °C, DT = 18.6 ± 1.3 s, n = 154; WT 41 °C, DT = 18.8 ± 1.2 s, n = 189; 2 cultures; P = 0.89; Fig. 3E). In contrast, K289M γ2 DT were significantly decreased at 41 °C (KM 37 °C, DT = 21.3 ± 1.3 s, n = 171; KM 41 °C, DT = 13.5 ± 1.5 s, n = 189; 2 cultures; P < 0.0001; Fig. 3E), indicating a faster escape of mutant receptors from the synaptic domain at 41 versus 37 °C.
Reduced synaptic DT usually correlates with depletion of postsynaptic receptor clusters (Bannaï et al. 2009; Charrier et al. 2010). Since hyperthermia reduced postsynaptic aggregation of mutant γ2 (Fig. 1), we asked whether this reduction was dependent on endocytosis of surface receptors. In the presence of myr-P4 to block dynamin-dependent endocytosis, 1-h exposure to 41 °C significantly reduced (−57% of control) the number of γ2 clusters per 10 μm dendritic length in neurons expressing the K289M mutant but not the WT subunit (WT 37 °C, n = 2.5 ± 0.3 from 27 cells; WT 41 °C, n = 1.9 ± 0.2 from 30 cells; KM 37 °C, n = 2.4 ± 0.3 from 31 cells; KM 41 °C, n = 1.1 ± 0.2 from 27 cells; 2 cultures; P = 0.07 for WT and P < 0.001 for K289M γ2; Fig. 3F,G). We conclude that reduced mutant γ2 clustering observed at 41 °C did not result from increased endocytosis but rather reflected a rapid escape of receptors from synapses and receptor depletion from the postsynaptic membrane.
Endocytosis is known to be a highly temperature-dependent process. An alternative to pharmacological blockade of clathrin-dependent endocytosis may then be to study γ2 diffusion in a lower temperature range. We thus examined the effects of a rise in temperature from 27 to 31 °C on the diffusion of recombinant γ2 (Fig. 4). As observed at 41 versus 37 °C in the presence of myr-P4, raising temperature from 27 to 31 °C did not alter the exploratory behavior (Fig. 4A), diffusion coefficients (synaptic: WT 27 °C, D = 2.7 ± 0.3 × 10−2 μm2s−1, n = 83; WT 31 °C, D = 3.0 ± 0.4 × 10−2 μm2s−1, n = 81, P > 0.9; extrasynaptic: WT 27° C, D = 4.6 ± 0.3 × 10−2 μm2s−1, n = 362; WT 31 °C, D = 4.7 ± 0.4 × 10−2 μm2s−1, n = 246, P = 0.1; 11 cultures; Fig. 4B,C) or confinement (WT 27 °C, L = 339 ± 32 nm; n = 39; WT 31 °C, L = 420 ± 45 nm; n = 38; from at least 3 cultures; P = 0.7; Fig. 4D) of WT QD-γ2. In contrast, the exploratory behavior (Fig. 4A) and the lateral mobility of the mutant γ2 subunit increased at 31 versus 27 °C for both synaptic and extrasynaptic trajectories (synaptic: KM 27 °C, D = 2.7 ± 0.2 × 10−2 μm2s−1, n = 177; KM 31 °C, D = 4.0 ± 0.3 × 10−2 μm2s−1, n = 108; P < 0.001; extrasynaptic: KM 27 °C, D = 4.1 ± 0.2 × 10−2 μm2s−1, n = 638; KM 31 °C, D = 4.7 ± 0.2 × 10−2 μm2s−1, n = 479; 11 cultures, P = 0.03; Fig. 4E,F). The steeper slope of the MSD plots for K289M γ2 trajectories (Fig. 4G) and the increase in the mean size of the confinement domain L (KM 27 °C, L = 287 ± 18 nm, n = 84; KM 31 °C, L = 426 ± 32 nm, n = 50; ≥3 cultures; P = 0.01) illustrated reduced confinement at inhibitory synapses at 31 versus 27° C. The time spent by K289M γ2 at synapses was also reduced after raising temperature from 27 to 31 °C (KM 27 °C, DT = 15.0 ± 1.0 s, n = 289; KM 31 °C, DT = 11.9 ± 1.0 s, n = 217; ≥3 cultures; P = 0.09; data not shown). Therefore, the membrane dynamics of K289M mutant but not WT γ2 subunit is sensitive to a rise in temperature both in the 27–31 and 37–41 °C range. We therefore used this lower temperature range in experiments to further elucidate the mechanisms involved in this phenomenon.
Increased Lateral Diffusion of K289M γ2 upon Warming Involves Excitatory Synaptic Activity
Most biological processes are temperature dependent. In particular, increased temperature may affect the fluidity of the plasma membrane and thereby influence lateral diffusion of transmembrane proteins, such as GABAARs. Alternatively, temperature may act to increase synaptic transmission (Schiff and Somjen 1985; Moser et al. 1993; Volgushev et al. 2000) and indirectly promote activity-dependent modulation of postsynaptic receptor diffusion (Lévi et al. 2008; Bannaï et al. 2009; Muir et al. 2010). We asked whether the effect of raising temperature on the lateral diffusion of γ2 may be intrinsic or rather mediated by postsynaptic activity.
We first examined the effect of raising temperature in the range 27–31 °C on spontaneous synaptic activity in hippocampal neurons. Mixed excitatory and inhibitory spontaneous postsynaptic currents (spPSCs) were recorded in hippocampal neurons first at 27 °C and 15–20 min after raising temperature to 31 °C (Fig. 5A,B). SpPSC frequency increased from 20% to 146% in 5 of 6 recorded neurons (mean frequency = 161.6 ± 22.4% of control, n = 6, Wilcoxon signed-rank test P < 0.05). Therefore, even a modest change in temperature was sufficient to induce a rapid increase in spontaneous synaptic activity. This is in agreement with previous data showing enhanced synaptic excitatory transmission after raising the brain temperature from 29 to 33 °C (Schiff and Somjen 1985).
Since GABAAR membrane dynamics are regulated by synaptic activity (Lévi et al. 2008; Bannaï et al. 2009; Muir et al. 2010), we asked whether blocking intrinsic or synaptic activity may prevent the increased diffusion of K289M mutant γ2 upon raising temperature. We compared the effect of raising temperature from 27 to 31 °C in the presence or absence of the sodium channel blocker TTX alone (1 μM), D,L AP5 (100 μM) alone or in combination with TTX (1 μM) and other glutamate receptor antagonists (NBQX, 10 μM; D,L-AP5, 100 μM; and R,S-MCPG, 500 μM; Fig. 5C–E). TTX alone did not prevent the temperature-induced acceleration of K289M mutant γ2 (Control, KM 27 °C, D = 6.9 ± 0.8 × 10−2 μm2s−1, n = 273; Control, KM 31 °C, D = 7.4 ± 0.4 × 10−2 μm2s−1, n = 345; TTX, KM 31 °C, D = 8.0 ± 0.4 × 10−2 μm2s−1, n = 247; 3 cultures; P < 0.001 for KM 27 vs. 31 °C and P = 0.01 for KM 31 °C vs. KM31 + TTX; Fig. 5C–E). In contrast, application of D,L AP5 alone completely abolished the effect of temperature on the diffusive properties of K289M mutant γ2 (D,L-AP5, KM 31 °C, D = 4.2 ± 0.3 × 10−2 μm2s−1, n = 345; P < 0.001; Fig. 5C,E). Application of all antagonists reduced the diffusion of the recombinant subunit below that observed in control conditions (TTX + AP5 + NBQX + MCPG, KM 31 °C, D = 3.2 ± 0.3 × 10−2 μm2s−1, n = 212; 3 cultures; P < 0.001; Fig. 5C,E). These results suggest that the K289M mutation potentiates the sensitivity of lateral diffusion of the receptor to excitatory synaptic activity.
This conclusion predicts the K289M mutation in γ2 may increase the lateral diffusion of GABAARs when excitatory synaptic activity is enhanced, independent of a rise in temperature. We examined this issue by comparing the membrane dynamics of recombinant WT and K289M γ2 subunits before and after application of the potassium channel blocker 4-aminopyridine (4AP, 100 μM). In untransfected neurons recorded at 27 °C, application of 4AP rapidly led to a robust increase in both spPSP frequency and firing that persisted throughout the application of the drug (Fig. 6A). Although 4AP has previously been shown to increase lateral diffusion of endogenous GABAARs in hippocampal neurons (Bannaï et al. 2009 and Supplementary Fig. 3), it had little or no effect on the confinement (Fig. 6B) or diffusion coefficients of recombinant WT γ2 (Control, WT 27 °C, D = 4.4 ± 0.3 × 10−2 μm2s−1, n = 398; 4AP, WT 27 °C, D = 4.6 ± 0.5 × 10−2 μm2s−1, n = 201; 2 cultures; P = 0.75; Fig. 6B,C). In contrast, 4AP significantly increased the exploratory behavior (Fig. 6B) and diffusion coefficient of K289M mutant γ2 (Control, KM 27 °C, D = 3.0 ± 0.2 × 10−2 μm2s−1, n = 509; 4AP, KM 27 °C, D = 4.1 ± 0.3 × 10−2 μm2s−1, n = 330; 2 cultures; P < 0.001). Therefore, GABAARs containing the recombinant K289M mutant but not WT γ2 subunit show enhanced lateral diffusion during sustained neuronal activity. Altogether, these results demonstrate that the K289M mutation in the γ2 subunit confers synaptic GABAARs with enhanced sensitivity to increased neuronal activity. We conclude GABAA receptors containing K289M mutant γ2 show increased diffusion and faster escape from synapses in conditions of enhanced neuronal activity, such as upon temperature elevation.
We have shown that the K289M mutation in the γ2 subunit affects the membrane dynamics and postsynaptic aggregation of GABAARs in conditions of increased temperature or neuronal activity. Raising temperature reduced the clustering of mutant γ2 and decreased the efficacy of synaptic inhibition in neurons expressing mutant but not WT recombinant γ2. A rise in temperature, in conditions of reduced endocytosis, increased the diffusion coefficients and decreased the confinement and synaptic DT of K289M γ2, thereby favoring its escape from synapses. This effect was likely due to the enhanced neuronal activity induced by the temperature rise since the increase in K289M γ2 dynamics was reversed by pharmacological blockade of excitatory synaptic transmission and mimicked by the convulsant drug 4-aminopyridine. We conclude the K289M mutation in γ2 confers GABAARs with enhanced sensitivity to neuronal activity that may then trigger the escape of receptors from synapses and thereby further reduce the efficacy of GABAergic inhibition. We suggest that this reflects a conformational change of γ2 that may impair receptor–scaffold interactions at synapses.
Temperature-Induced Loss of Mutant K289M γ2 Clusters at Inhibitory Synapses
We have shown that postsynaptic receptor clusters containing the K289M mutant γ2 rapidly disappear upon temperature increase. This effect was detected as a reduced number of GFP punctae per 10 μm dendritic length in neurons expressing recombinant GFP-Gabrg2 bearing the K289M mutation. It was correlated with a reduced frequency of mIPSCs suggesting the number of functional GABAergic synapses was reduced. This effect is unlikely to reflect changes in presynaptic function since it was only observed in neurons expressing recombinant mutant but not WT γ2. Currents carried by GABAARs containing K289M mutant γ2 show faster decay (Eugène et al. 2007), reflecting faster deactivation kinetics (Bianchi et al. 2002; Hales et al. 2006). We took advantage of this physiological signature to compare the relative proportion of synapses containing the mutant subunit before and after heating. Whereas mIPSC decay was significantly faster in neurons expressing K289M mutant versus WT γ2 at 37 °C, this difference became nonsignificant after 1-h exposure to 41 °C, suggesting the contribution of synapses devoid of mutant γ2 to mIPSCs had increased.
This observation reveals that all GABAergic synapses do not express receptors containing recombinant γ2 in transfected neurons. Consistent with this conclusion, a fraction of GAD immunopositive terminals were not facing GFP clusters in neurons expressing recombinant γ2 and yet colocalized with endogenous γ2 clusters (data not shown). The specificity of recombinant γ2 synaptic incorporation may reflect distinct behaviors of the 2 splice variants γ2S and γ2L that are expressed in hippocampal neurons (Gutiérrez et al. 1994; Khan et al. 1994; Miralles et al. 1994). The present study was conducted using recombinant γ2L. This variant might be preferentially associated with synaptic GABAARs (Baer et al. 2000) and differs from γ2S by an extra 8 amino acids within the cytoplasmic loop with putative phosphorylation site by protein kinase C (Moss et al. 1992). Postsynaptic aggregation of γ2 requires the anchoring protein gephyrin (Essrich et al. 1998; Kneussel et al. 1999, 2001; Lévi et al. 2004) and γ2L- but not γ2S-gephyrin interaction, and postsynaptic clustering is regulated by Protein Kinase C (Meier and Grantyn 2004). Therefore, γ2S and γ2L clustering to distinct postsynaptic sites may reflect different interactions with gephyrin. Alternatively, γ2 isoforms may be part of heteropentamers with different content in α subunits with various binding properties to scaffolding molecules (Tretter et al. 2008).
Although mIPSC frequency was reduced upon raising temperature, likely reflecting a loss of postsynaptic receptors at synapses containing mutant γ2, the mean amplitude of mIPSCs was unaffected. This may be explained if 1) the affinity of scaffolding molecules for GABAARs somehow acts to maintain receptor content constant by replacing mutant γ2 containing receptors by receptors containing endogenous γ2 or 2) most synapses containing mutant γ2 were totally depleted from postsynaptic receptors after 1 h at 41 °C. Both hypotheses predict that mIPSC decay time would increase after heating in neurons expressing mutant γ2 (Fig. 2). However, it seems unlikely that inhibitory synapses maintained their postsynaptic receptor content since a partial reduction in synaptic GABAAR content can be achieved in conditions of increased neuronal activity (Bannaï et al. 2009; Muir et al. 2010). We conclude that exposure of neurons to 41 °C very rapidly leads to a complete loss of postsynaptic GABAARs containing K289M mutant γ2 at some, but not all, inhibitory synapses.
Facilitation of Mutant γ2 Diffusion at Inhibitory Synapses by Neuronal Activity
We report that raising temperature by a few degrees is sufficient to increase lateral diffusion of mutant but not WT recombinant γ2 in hippocampal neurons. Accelerated membrane turnover including endocytosis of individual GABAARs precluded single-molecule tracking experiments to be conducted at 41 °C. Indeed, both WT and KM recombinant γ2 were slowed down and confined upon hyperthermia. This reduced diffusion likely reflected increased trapping of receptors in endocytotic pits since pharmacological blockade of dynamin-dependent endocytosis revealed increased diffusion and decreased synaptic DT of K289M recombinant γ2. Reduction in synaptic DT often correlates with depletion of postsynaptic receptor clusters (Bannaï et al. 2009; Charrier et al. 2010). The loss in postsynaptic receptor content is likely due to reduced receptor trapping by the subsynaptic scaffold. This may reflect a reduction in 1) the number of gephyrin molecules at synapses (Bannaï et al. 2009; Charrier et al. 2010), 2) the receptor ability to interact with gephyrin (Zita et al. 2007; Lévi et al. 2008), and/or 3) gephyrin oligomerization (Bedet et al. 2006; Calamai et al. 2009; Charrier et al. 2010). Temperature-induced loss of K289M γ2 clusters was also detected after blockade of dynamin-dependent endocytosis, suggesting that reduced clustering of mutant γ2 at 41 °C did not result from increased internalization but rather reflected a rapid escape of receptors from synapses and receptor depletion from the postsynaptic membrane.
How might temperature influence lateral diffusion and clustering of synaptic GABAARs? Most biological processes may be accelerated at higher temperatures yet neuronal activity seemed most likely to mediate this effect. While action potentials may be reduced in amplitude and/or frequency at higher temperatures (Andersen and Moser 1995; Volgushev et al. 2000), synaptic transmission on the contrary is enhanced (Katz and Miledi 1965; Schiff and Somjen 1985; Moser et al. 1993). This latter effect involves a combination of presynaptic (increased vesicle pool recycling, Pyott and Rosenmund 2002; Kushmerick et al. 2006) as well as postsynaptic factors (acceleration of postsynaptic receptor activation, Postlethwaite et al. 2007). Consistent with these observations, we observed an increased frequency of spPSCs in hippocampal neurons upon raising temperature. Increased lateral diffusion of mutant γ2 at higher temperatures was reversed by the NMDA receptor antagonist D,L-AP5 and even more so by a combination of glutamate receptor antagonists and TTX. On the other hand, TTX alone had little or no effect, while 4-aminopyridine mimicked the effect of temperature. These results point to a major role of postsynaptic glutamate receptor activation on the temperature-dependent modulation of mutant γ2 diffusion. Accordingly, GABAAR lateral diffusion has previously been shown to be strongly dependent on postsynaptic NMDAR activation and Ca2+-calcineurin signaling (Bannaï et al. 2009; Muir et al. 2010). In particular, diffusion coefficients of synaptic GABAARs were directly correlated with intracellular Ca2+ concentrations. This modulation might involve a Ca2+-dependent modulation of either the stability of gephyrin scaffold (Hanus et al. 2006) or the receptor–gephyrin interactions through the α2 subunit (Tretter et al. 2008), or both. Noticeably, this modulation seems to differentially affect receptors containing endogenous versus recombinant WT γ2 (Fig. 6 and Supplementary Fig. 3). The greater sensitivity to excitatory synaptic activity of endogenous as compared with recombinant WT γ2 diffusion will need to be further explored and might reflect differences between splice variants.
We show that the sensitivity of GABAARs to Ca2+-dependent modulation of their diffusion is enhanced by the K289M mutation in γ2. This effect might reflect an indirect increase in neuronal excitation due to the reduced efficacy of synaptic inhibition in neurons expressing the mutant γ2 (Eugène et al. 2007; Fig. 2). However, in this case, one would expect faster diffusion and reduced clustering of K289M versus WT γ2 even at steady state. Instead, the K289M mutation alters neither receptor clustering nor inhibitory synaptic transmission at 37 °C (Eugène et al. 2007; Figs 1 and 2). This argues against an indirect impact of the mutation on γ2 diffusion through enhanced excitatory drive. Instead, we propose the mutation per se may act to affect the allosteric conformation of the receptor. The K289 residue is lining the external mouth of the pore of the channel, between transmembrane segments II and III (O'Mara et al. 2005). Therefore, it seems unlikely the K to M mutation may directly compromise gephyrin–receptor or γ2-α2 interactions. However, the mutation is known to affect receptor allosteric conformation and favors its closed state (Bianchi et al. 2002; Hales et al. 2006). We thus propose this allosteric conformation may unmask a region or residue of γ2 that controls its diffusion (e.g., Muir et al. 2010). Therefore, favoring the closed state of GABAARs would enhance the sensitivity of their lateral diffusion to Ca2+ and promote their escape from inhibitory synapses.
Gabrg2 Mutations and the Mechanisms of Febrile Seizures
Our results demonstrate the K289M mutation in the GABAAR γ2 subunit affects GABAergic signaling in 2 distinct ways depending on the level of neuronal activity. We had previously shown expressing K289M recombinant γ2 in hippocampal neurons results in accelerated IPSC kinetics, which likely leads to both smaller and faster IPSPs in these cells (Poncer et al. 1996; Eugène et al. 2007). This effect is dominant since it was observed in neurons expressing endogenous WT γ2. In addition, we now show the K289M mutation also increases the sensitivity of lateral diffusion of synaptic GABAARs to neuronal activity. This effect does not result in a significant difference in membrane expression or clustering of mutant γ2 as compared with WT (Eugène et al. 2007 and this study). This suggests that the level of spontaneous activity in primary hippocampal cultures may not be sufficient to lead to a steady-state decrease in synaptic clustering of K289M mutant γ2. Instead, this effect may become prominent only in conditions of increased neuronal activity, sufficient to induce activation of postsynaptic NMDARs. This may occur during febrile episodes, which lead to respiratory alkalosis and subsequent neuronal hyperexcitability (Schuchmann et al. 2006). It may then further attenuate synaptic inhibition by reducing the number rather than the efficacy of functional GABAergic synapses and thereby precipitate seizures.
The mechanisms by which fever leads to seizures remain poorly understood. Enhanced temperature-dependent endocytosis has been proposed as a common mechanism for all Gabrg2 mutations associated with FS in humans (Kang et al. 2006). Although attractive, this hypothesis seems unlikely to account for the involvement of other Gabrg2 mutations in FS. In neurons, the R43Q mutation largely compromises membrane targeting of γ2 and prevents synaptic aggregation with other subunits (Eugène et al. 2007; Frugier et al. 2007; Tan et al. 2007). Instead, this mutation mostly reduces GABA signaling through a reduction of tonic inhibition. Similarly, although some surface expression of the Q351X mutant γ2 was reported in heterologous cells, none is detected in neurons expressing this mutant (Eugène et al. 2007). Therefore, the possible contribution of an activity-dependent escape from synapses of these mutant γ2, similar to that described here for the K289M mutant, seems unlikely to account for the functional phenotype associated with these mutations. Other γ2 mutations associated with FS (R139G, Audenaert et al. 2006; W390X, Sun et al. 2008) will need to be further explored in neurons to confirm whether the mechanism described in the present study can be generalized to other mutants.
Avenir program of Institut National de la Santé et de la Recherche Médicale (to J.C.P.) and grants from the city of Paris; the Fondation Electricité de France; and the Fondation pour la Recherche Médicale (to J.C.P).
We thank Steve J. Moss and Christel Depienne for kindly providing the original WT and K289M mutant Gabrg2-GFP constructs and Norbert Ankri for sharing and assistance with Detectivent software. Conflict of Interest: None declared.