Activation of calcineurin underlies altered trafficking of α2 subunit containing GABAA receptors during prolonged epileptiform activity

Fast inhibitory signalling in the mammalian brain is mediated by gamma-aminobutyric acid type A receptors (GABAARs), which are targets for anti-epileptic therapy such as benzodiazepines. GABAARs undergo tightly regulated trafficking processes that are essential for maintenance and physiological modulation of inhibitory strength. The trafficking of GABAARs to and from the membrane is altered during prolonged seizures such as in Status Epilepticus (SE) and has been suggested to contribute to benzodiazepine pharmacoresistance in patients with SE. However, the intracellular signalling mechanisms that cause this modification in GABAAR trafficking remain poorly understood. In this study, we investigate the surface stability of GABAARs during SE utilising the low Mg2+ model in hippocampal rat neurons. Live-cell imaging of super ecliptic pHluorin (SEP)-tagged α2 subunit containing GABAARs during low Mg2+ conditions reveals that the somatic surface receptor pool undergoes down-regulation dependent on N-methyl-d-aspartate receptor (NMDAR) activity. Analysis of the intracellular Ca2+ signal during low Mg2+ using the Ca2+-indicator Fluo4 shows that this reduction of surface GABAARs correlates well with the timeline of intracellular Ca2+ changes. Furthermore, we show that the activation of the phosphatase calcineurin was required for the decrease in surface GABAARs in neurons undergoing epileptiform activity. These results indicate that somatic modulation of GABAAR trafficking during epileptiform activity in vitro is mediated by calcineurin activation which is linked to changes in intracellular Ca2+ concentrations. These mechanisms could account for benzodiazepine pharmacoresistance and the maintenance of recurrent seizure activity, and reveal potential novel targets for the treatment of SE. This article is part of the Special Issue entitled ‘GABAergic Signaling in Health and Disease’.


Introduction
GABA A Receptors (GABA A Rs) are ligand-gated chloride permeable ion channels which mediate both phasic (synaptic) and tonic (extrasynaptic) inhibitory neurotransmission in the central nervous system (Jacob et al., 2008;Luscher et al., 2011). They assemble from five subunits, the composition of which determines the receptors functional and pharmacological properties and the specific location on the neuronal membrane (Luscher et al., 2011;Jacob et al., 2008). GABA A Rs containing the g2 subunit mediate synaptic transmission (in contrast to extrasynaptic receptors located away from the synapse) and are a target for benzodiazepines (Pritchett et al., 1989).
The enrichment of GABA A Rs in subcellular compartments such as the axon initial segment (AIS) has been reported for the a2 subunit and although both a1 and a2 subunits are found at the synapse in dendrites, a minority of GABA A Rs in the AIS contain the a1 subunit (Panzanelli et al., 2011;Brünig et al., 2002). GABA A Rs undergo dynamic movement within the cellular membrane. Lateral diffusion facilitates trafficking and assures the appropriate surface localisation of the receptor (Mukherjee et al., 2011), while trafficking to and from the membrane through exocytotic and endocytotic processes allows constant maintenance of the inhibitory synaptic receptor pool (Bogdanov et al., 2006;Kittler et al., 2000Kittler et al., , 2004. Altered neuronal activity causes surface GABA A Rs to undergo plasticity-induced trafficking changes. These are mediated by alterations in the activity of protein phosphatases and kinases which are linked to changes in intracellular Ca 2þ (Muir et al., 2010;Bannai et al., 2009;Saliba et al., 2012;Luscher et al., 2011;Jacob et al., 2008;Petrini et al., 2014). SE evolves rapidly and dynamically, manifesting as a prolonged and self-sustaining seizure with significant morbidity and mortality (Lothman, 1990;Dodrill and Wilensky, 1990;Sutter et al., 2013). This distinct condition can occur in patients with previous epilepsy or may occur de novo as a result of acute neurological disorders (Trinka et al., 2012). As SE evolves, the patient's response to treatment with benzodiazepines decreases progressively which rapidly results in benzodiazepine pharmacoresistance. This may lead to refractory SE, a pathological state in which seizures are not sopped by firstor second-line anticonvulsant therapies.
To unravel the role of benzodiazepine pharmacoresistance associated with SE patients, studies have addressed whether the trafficking of GABA A Rs to and from the cellular membrane is altered during models of SE (Naylor et al., 2005;Goodkin et al., 2005;Blair et al., 2004). Interestingly, it has been suggested that GABA A Rs are subjected to subunit-specific trafficking during prolonged depolarisation. GABA A Rs containing the synaptic subunits b2/3 and g2 undergo internalisation whereas those containing the extrasynaptic d subunit remain unchanged (Goodkin et al., 2008).
Despite recent studies, the temporal dynamics of GABA A R trafficking have not been investigated using live-cell imaging. Moreover, whether endocytosis occurs preferentially in distinct compartments such as dendrites or soma remains unclear. It is not known which molecular pathways underlie this subunit-specific trafficking of GABA A Rs. Furthermore, it remains to be determined whether Ca 2þ and its intracellular signalling cascades play a significant role in the modulation of GABAergic inhibition during SE.
To address the molecular mechanisms underlying altered GABA A R trafficking during SE, we used a live-cell imaging approach to examine the surface stability of GABA A Rs in hippocampal neurons. We induced prolonged epileptiform bursting activity in vitro by exposing neurons to artificial cerebrospinal fluid (aCSF) lacking Mg 2þ (Mangan and Kapur, 2004;Sombati and Delorenzo, 1995). Using this model, we show a decrease in somatic surface GABA A Rs that is dependent on NMDAR activity and the Ca 2þ -dependent phosphatase, calcineurin. Furthermore, we show that epileptiform activity alters intracellular Ca 2þ concentrations, which correlates with the decrease of GABA A Rs from the surface possibly contributing to pathological signalling during SE.

Constructs
The N-terminally tagged GABA A a 2 -SEP DNA was a kind gift from S. Moss (Tufts University, Cambridge, MA) and has been described previously (Tretter et al., 2008).

Cell culture and transfection
All animal experiments were carried out in accordance with the U.K. Animals (Scientific Procedures) Act, 1986. All efforts were made to minimise animal suffering and to reduce the number of animals used. Dissected hippocampi of P0 rat pups or E18 embryos were immediately placed in ice-cold dissection buffer (HBSS (Invitrogen)) and washed once. Using trypsin (0.25%) tissue was digested for 10 min before trituration in~2 ml of attachment medium. Neurons were plated onto poly-Llysine (Sigma) coated coverslips (500 mg/ml). For nucleofection, hippocampal neurons were nucleofected with GABA A a 2SEP plasmid DNA. Neurons were centrifuged and the cell pellet was resuspended in 100 ml transfection buffer (135 mM KCL, 10 mM HEPES-pH 7.3, 2 mM MgCl 2 , 5 mM EGTA, 0.2 mM CaCl 2 ) and transfected using a single cuvette AMAXA system (Lonza, programme O-003 or AK-009). Neurons were left to develop at 37 C and 95% O 2 , 5% CO 2 in maintenance medium [Neurobasal (Invitrogen), B27 Supplement (Invitrogen), 0.6% Glucose (Sigma), 2 mM Glutamine (Invitrogen) and PenicillineStreptomycin] for 14e21 DIV before imaging.

Live-cell imaging
Live-cell imaging was performed on an upright Olympus microscope (BX51WI) coupled to an EM-CCD camera (Ixon, Andor). Cells were imaged with a waterimmersion 60Â objective (Olympus). Excitation was provided by an X-cite 120Q light source (Lumen Dynamics). Appropriate filters were used (in nm): Excitation: 470/40; Emission: 525/50; Dichroic: 495, long pass. The image pixel scale was calculated by dividing the camera pixel size (16 mm) by the lens magnification (60Â) yielding a pixel size of 0.27 mm. Before constant perfusion with a ColeeParmer Master-Flex pump (~4 ml/min), aCSF (126 mM NaCl, 24 mM NaHCO 3 , 10 mM D-Glucose, 2.5 mM KCL, 2 mM CaCl 2 , 1 mM MgCl, 1 mM NaH 2 PO 4 , 5 mM Sodium Pyruvate) was pre-equilibrated for 20 min with 95% O 2 and 5% CO 2 to establish a pH of 7.4. Temperature of the waterbath was constantly measured using a digital Thermometer (Hanna Instruments) and maintained at 37 C. Any focus drift was corrected manually. Protocols were adapted to achieve minimal bleaching conditions. Imaging of SEP-tagged GABA A Rs was done for 60 min at a rate of one frame every 20 s (180 frames, 48.8 ms exposure, no averaging). For imaging of intracellular Ca 2þ using fluo4 (1 mM, Molecular Probes, Invitrogen) hippocampal neurons were incubated for 30 min at 37 C. After washing twice, fluo4-imaging was done for 60 min (720 frames, 5 ms exposure, no averaging) at 1 frame every 5 s.

Cell-attached recording
Cell-attached recordings were made on transfected hippocampal neurons at 13 DIV using an Axopatch 200B amplifier (Molecular Devices) and pClamp software. Cells were visualised using an upright Olympus BX50WI microscope equipped with a 40Â water-immersion objective and infrared optics. Recording electrodes were pulled from standard-walled borosilicate glass capillaries (Warner Instruments) and filled with aCSF. Gigaseal cell attached recordings were made in voltage-clamp mode at À70 mV; the cells were constantly perfused with aCSF. To block currents during recording NBQX disodium salt (20 mM, Abcam) and dAPV (D-(À)-2-Amino-5phosphonopentanoic acid, 25 mM, TOCRIS) were added to the perfusion solution.

Low Mg 2þ and drug treatments
To induce epileptiform bursting activity, aCSF without Mg 2þ but 2 mM glycine (126 mM NaCl, 24 mM NaHCO 3 , 10 mM D-Glucose, 2.5 mM KCL, 2 mM CaCl 2 , 1 mM NaH 2 PO 4 , 5 mM sodium pyruvate, 2 mM glycine) was used (Blair et al., 2004). We confirmed previous studies (Robinson et al., 1993;Mangan and Kapur, 2004;DeLorenzo et al., 1998) that low Mg 2þ results in cellular burst spiking that is dependent upon glutamateric transmission (Sup. Fig.2). Moreover, action potentials were associated with post-synaptic currents, indicating that the bursting was the result of network activity (Sup. Fig. 2). The mitochondrial substrate sodium pyruvate was supplemented to reduce neuronal death (Kovac et al., 2012). Transfected hippocampal neurons were perfused with control aCSF for 3.3 min (10 frames, baseline), followed by either low Mg 2þ treatment or continued perfusion with aCSF with Mg 2þ (control) for 60 min. To block NMDAR activity during low Mg 2þ treatment, the NMDAR blocker dAPV (25 mM, TOCRIS) was used continuously throughout the low Mg 2þ treatment without preincubation. For low Mg 2þ /NMDA treatment, NMDA (30 mM, TOCRIS) was added to the low Mg 2þ medium and applied continuously for 60 min. To block the activity of the Ca 2þ dependent phosphatase calcineurin, cells were pre-incubated with calcineurin autoinhibitory peptide (Terada et al., 2003) (50 mM, Calbiochem) for 25 min at 37 C and imaged (without application of the peptide during perfusion) either during control or low Mg 2þ treatment (Muir et al., 2010).

Image analysis
Intensity analysis of specific regions of interests (ROIs: background, soma, diffuse, clustered) was done in ImageJ 1.43u which allowed the export of raw data to MatlabR2008a Software. Image correction was done in ImageJ software using the plugin StackReg macro (Th evenaz et al., 1998) which corrects for drift in (x,y). Inverted average intensity projection was done in ImageJ by using the Z-stack application from frame 1e10 (0e3 min) and frame 30e60 (10e20 min). Analysis of SEP-imaging raw data was done using Matlab Software through a custom designed code. Background was subtracted from each frame, fluorescence intensity was normalised to the baseline (average value of t ¼ 0e3.33 min) and averaged for each experimental group. The fluorescence intensity values for each specific ROI were analysed in individual loops which allowed separate analysis. Furthermore, the standard error of mean (SEM) was calculated for each time-point and the mean normalised values including error bars were plotted against time. Ca 2þ imaging was analysed using ImageJ. Fluorescence intensity in the soma was extracted from one ROI per cell. Baseline for 60 min Ca 2þ -imaging was the average of the first 10 frames (t ¼ 0e50 s), which corresponds to control conditions.

Statistical analysis
All experiments were performed on neurons from at least three individual preparations. The software GraphPad Prism was used for statistical tests and to generate bar charts. Data sets were tested to determine if they were normally distributed (KS normality test) before undertaking further statistical analysis. For low Mg 2þ only and low Mg 2þ NMDA treatments, p values were determined using a Student's t test (two-tailed). Repeat measures ANOVA (for normally distributed data) or Friedman test was used to analyse significance of low Mg 2þ induced effects during Ca 2þ -imaging, low Mg 2þ /dAPV and low Mg 2þ /CAIP experiments since there were more than two experimental groups. Appropriate post-hoc tests such as Tukey's for normally distributed data or Dunn's multiple comparison for nonnormally distributed data were used. Values are given as mean ± SEM. Error bars represent SEM.

Surface stability of somatic GABA A Rs in hippocampal neurons is altered during Low Mg 2þ treatment
To examine the influence of SE on GABA A R stability and clustering in vitro, we mimicked the characteristic repetitive epileptiform bursting activity of SE by removal of Mg 2þ from the extracellular medium of cultured hippocampal rat neurons transfected with SEP-tagged GABA A R a 2 subunit (a 2SEP ) (Sombati and Delorenzo, 1995). Surface GABA A Rs were imaged via the SEP-tag on the a 2 subunit (Muir et al., 2010) (which allows visualisation through high fluorescence in neutral pH, Sup. Fig.1) for 60 min.
a 2SEP fluorescence was analysed in 3 distinct regions of interests (ROIs): soma, diffuse (extrasynaptic compartment in dendrites) and clusters. At t ¼ 20 min, somatic fluorescence of a 2SEP -containing GABA A Rs was significantly decreased (control F/F 0 : 0.997 ± 0.02, low Mg 2þ F/F 0 : 0.85 ± 0.05; p ¼ 0.02) indicating that internalisation of GABA A Rs at the somatic level increases during low Mg 2þ treatment (Fig. 1C,F). This could account for a decrease in hippocampal GABAergic inhibition during epileptiform activity. However, a 2SEPfluorescence intensity at the soma was not found to be significantly changed at t ¼ 60 min (control F/F 0 : 0.92 ± 0.03, low Mg 2þ F/F 0 : 1.01 ± 0.04; p ¼ 0.09) suggesting a biphasic regulation of surface GABA A Rs during low Mg 2þ treatment (Fig. 1C,F'). Interestingly, a 2SEP -GABA A R clusters (t ¼ 20 min; control F/F 0 : 1.02 ± 0.02 low Mg 2þ F/F 0 : 0.96 ± 0.07; p ¼ 0.36) and diffuse (t ¼ 20 min; control F/ F 0 : 1.01 ± 0.02, low Mg 2þ F/F 0 : 0.95 ± 0.05; p ¼ 0.28) fluorescence intensity in the neuronal dendrites during low Mg 2þ treatment showed only a minor, non-significant decrease. Our data thus indicates compartmental specificity of low Mg 2þ induced decrease of GABA A Rs from the surface (Fig. 1D,E), with GABA A Rs primarily endocytosed from the cell soma surface.
3.2. Activity of NMDA receptors induces the down-regulation of somatic GABA A receptors from the surface during Low Mg 2þ treatment Low extracellular Mg 2þ induces epileptiform activity which is abolished by application of the NMDAR antagonist dAPV Collingridge, 1985, 1987;Tancredi et al., 1990;Albowitz et al., 1997;Westerhoff et al., 1995;Guly as-Kov acs et al., 2002;Mangan and Kapur, 2004). Therefore, we tested whether inhibition of NMDAR activity during low Mg 2þ treatment blocks the somatic downregulation of GABA A Rs (Fig. 2). Low Mg 2þ alone induced a significant decrease of a 2SEP -GABA A R fluorescence intensity at t ¼ 20 min whereas this loss was inhibited by the co-application of dAPV (t ¼ 20 min; control F/F 0 : 0.97 ± 0.05, low Mg 2þ F/F 0 : 0.65 ± 0.03 (p < 0.001), low Mg 2þ /dAPV F/F 0 : 0.96 ± 0.07 (p > 0.05); one-way ANOVA test, Tukey's multiple comparison post test), confirming that the down regulation of surface GABA A Rs was dependent on the activation of NMDARs (Fig. 2C,C').

Epileptiform activity evokes intracellular Ca 2þ changes that correspond to the temporal dynamics of somatic surface GABA A R decrease
To further explore the mechanisms of NMDAR-driven decrease in surface GABA A Rs during low Mg 2þ treatment, we applied the fluorescent Ca 2þ indicator fluo4 in low Mg 2þ treated hippocampal neurons. This allowed us to investigate intracellular Ca 2þ transients evoked by low Mg 2þ treatment. Hippocampal neurons perfused with control aCSF exhibit small Ca 2þ transients reflecting spontaneous activity, whereas low Mg 2þ perfusion significantly altered intracellular Ca 2þ throughout the timeline of 60 min (Fig. 3B,B'). Fluo4 imaging reported intracellular Ca 2þ increases rapidly upon early perfusion with low Mg 2þ (10e20 min F/F 0 : 375.8 ± 28.4; p < 0.001, Friedman test and Dunn's multiple comparison post test) and at t ¼ 60 min (F/F 0 : 215.2 ± 28.8; p < 0.05, Friedman test and Dunn's multiple comparison post test) in comparison to baseline (t ¼ 100e150 s; F/F 0 : 104.5 ± 4.4), (Fig. 3C). This indicates that low Mg 2þ treatment induces an intracellular Ca 2þ rise, which is likely to be caused by activation of NMDARs. Interestingly, intracellular Ca 2þ concentration drops significantly during the timeline (10e20 min, F/F 0 : 375.8 ± 28.4; 50e60 min, F/F 0 : 215.2 ± 28.8; p < 0.05, Friedman test and Dunn's multiple comparison post test) showing that intracellular Ca 2þ concentration undergoes alteration on a similar timescale to that of somatic surface GABA A R decrease (Fig. 3C).

Dispersion of clustered GABA A receptors is induced by NMDA receptor activation
It has been reported that the dispersal of surface GABA A R clusters in neuronal processes is regulated through Ca 2þ influx via NMDARs (Muir et al., 2010). Therefore we tested whether further increasing the activation of NMDARs by co-application of low Mg 2þ and the agonist NMDA would trigger dispersion of surface GABA A R clusters in proximal dendrites. Indeed, the activation of NMDARs with low Mg 2þ in addition to application of the agonist NMDA (low Mg 2þ /NMDA) caused a loss of a 2SEP GABA A R fluorescence intensity in dendritic clusters (Fig. 4A,B,B') at t ¼ 20 min (control F/F 0 : 1.001 ± 0.04, low Mg 2þ /NMDA F/F 0 : 0.799 ± 0.03; p < 0.01) suggesting that surface stability of GABA A Rs corresponds with the potency of NMDAR activation (Fig. 4C,F). Interestingly, during low Mg 2þ /NMDA perfusion diffuse (control F/F 0 : 1.02 ± 0.03, low Mg 2þ / NMDA F/F 0 : 0.95 ± 0.09, p ¼ 0.58) and total (data not shown) fluorescence intensity in neuronal processes remains unaltered at t ¼ 20 min (Fig. 4D,F). This indicates dispersion of surface GABA A R control (green, n ¼ 7 cells), low Mg 2þ (blue, n ¼ 8 cells) and low Mg 2þ /dAPV (dark blue, n ¼ 6 cells). (C') Summary of somatic F/F 0 at 20 min after low Mg 2þ treatment. Low Mg 2þ induces a significant decrease (p < 0.001) in somatic fluorescence intensity, which is inhibited by application of NMDAR blocker dAPV (p < 0.001). ***p < 0.001. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) upon low Mg 2þ /NMDA. Somatic GABA A R fluorescence intensity is significantly decreased at t ¼ 10 min after low Mg 2þ /NMDA (control F/F 0 : 1.00 ± 0.02, low Mg 2þ /NMDA F/F 0 : 0.77 ± 0.059, p < 0.01) treatment, however it is not significantly altered at t ¼ 20 min (control F/F 0 : 0.98 ± 0.03, low Mg 2þ /NMDA F/F 0 : 0.85 ± 0.11, p ¼ 0.34) (Fig. 4 E). Although the decrease in somatic surface GABA A Rs fluorescence intensity during low Mg 2þ /NMDA also occurs rapidly after treatment and is of similar size compared to low Mg 2þ only treatment, the biphasic recovery phase is shorter in the low Mg 2þ /NMDA treatment.

Calcineurin mediates the decrease of GABA A receptors from the surface during Low Mg 2þ induced bursting activity
We next investigated the signalling mechanisms involved in NMDAR mediated GABA A R surface decrease during epileptiform bursting activity. Calcineurin is implicated in activity-dependent regulation of GABAergic inhibition and hence could play an important role in Ca 2þ mediated signalling, we therefore analysed its role in GABA A R stability during low Mg 2þ bursting (Lu et al., 2000;Wang et al., 2003;Chen and Wong, 1995;Muir et al., 2010). Cells undergoing epileptiform activity showed a decrease in somatic GABA A R fluorescence intensity compared to control. We found that treating cells with a calcineurin autoinhibitory peptide did not significantly affect somatic GABA A R intensity (control F/F 0 : 1.07 ± 0.10, control/CAIP F/F 0 : 0.95 ± 0.03; p > 0.05, one-way ANOVA test and Tukey's multiple comparison post test) showing that calcineurin had no effect at t ¼ 20 min in control conditions (Fig. 5C,C'). However, blocking calcineurin activity, inhibited the low Mg 2þ -induced decrease of surface GABA A R at the soma at t ¼ 20 min significantly (low Mg 2þ F/F 0 : 0.75 ± 0.06, low Mg 2þ /CAIP F/F 0 : 1.043 ± 0.06; p < 0.05, one-way ANOVA test and Tukey's multiple comparison post test) (Fig. 5C,C'). These results suggest that calcineurin activation upon Ca 2þ influx through NMDARs is directly involved in the decrease of surface GABA A R triggered by epileptiform bursting activity.

Discussion
GABA A Rs are a target for a variety of drugs including benzodiazepines, which are of high clinical relevance for first-line treatment of SE. Therefore it is likely that the modulation of surface stability of GABA A Rs induces or supports benzodiazepine pharmacoresistance in patients with SE. To identify potential mechanisms facilitating GABA A R internalisation, we performed liveimaging on SEP-tagged GABA A Rs. The key finding of this study shows epileptiform activity induces activation of calcineurin leading to a decrease in the number of surface GABA A Rs in the soma. This activity-dependent alteration of inhibitory strength is mediated by activation of NMDARs and is parallelled by an increase in intracellular Ca 2þ which in turn is likely to activate calcineurin. These data identify a signalling mechanism underlying surface GABA A R decrease during SE and supports studies that report NMDAR activation regulates GABAergic inhibition (Muir et al., 2010;Bannai et al., 2009;Stelzer et al., 1987).
It is known that internalisation of GABA A Rs containing a, g 2 or b 2/3 subunits is increased during epileptiform activity in vitro using low Mg 2þ or high KCl media (Blair et al., 2004;Goodkin et al., 2008Goodkin et al., , 2005. Furthermore, it has been demonstrated that GABA A Rs are internalised in chemoconvulsant models of SE in vivo (Naylor et al., 2005;Nishimura et al., 2005). Our data supports the hypothesis that GABA A R internalisation occurs during prolonged seizures, by demonstrating that GABA A Rs which contain the a 2 subunits are decreased during low Mg 2þ treatment. In the majority of cells this decrease in surface GABA A Rs was biphasic possibly indicating an adaptational switch in inhibitory strength. Interestingly, we observe that this effect occurs preferentially in the soma but less in dendrites. Compartmental internalisation of GABA A Rs during SE has not been reported and the mechanisms underlying this differential regulation are unknown, but there are a number of possibilities. Firstly, it remains to be investigated to what extent intracellular Ca 2þ buffering systems such as endoplasmic reticulum or mitochondria are contributing to surface stability of GABA A Rs in neurons undergoing epileptiform activity. Expression of signalling proteins in specific subcellular compartments could explain this effect. Secondly, since this study makes use of experiments based on overexpression of GABA A Rs, this could account for increased inhibition which could suppress a decrease of GABA A Rs in dendrites specifically. Omitting Mg 2þ in the extracellular medium of cultured hippocampal neurons triggers a sequence of events leading to neuronal death (Kovac et al., 2012;Yoon et al., 2010). To significantly reduce neuronal cell death, mitochondrial substrate sodium pyruvate was added in our experiments. Although we control for this substitution, it would be interesting to test whether mitochondrial ATP production affects GABA A R trafficking during prolonged seizures. Thirdly, correlating the amount of intracellular ATP to the trafficking of GABA A Rs could contribute to the understanding of compartmentalised trafficking of GABA A R during low Mg 2þ treatment.
A comprehensive body of literature describes that low Mg 2þ induced epileptiform activity is dependent on increased NMDAR activity Collingridge, 1985, 1987;Tancredi et al., 1990;Albowitz et al., 1997;Westerhoff et al., 1995;Guly as-Kov acs et al., 2002;Mangan and Kapur, 2004), therefore we tested whether the somatic surface GABA A R decrease was mediated by NMDAR activity. Our experiments confirm this hypothesis by reporting that the decrease in surface GABA A R was blocked by application of NMDAR blocker dAPV. To our knowledge this is the first study showing a direct regulation of GABA A Rs by NMDAR activation Significant loss of fluorescence in the clusters compared to control at 20 min following after low Mg 2þ treatment (p ¼ 0.0008). Diffuse fluorescence is not altered upon low Mg 2þ treatment at 20 min (p ¼ 0.36) after low Mg 2þ treatment. Diffuse fluorescence is unaltered following low Mg 2þ treatment (t ¼ 20 min; p ¼ 0.58) compared to control. (F') At 60 after low Mg 2þ treatment clustered fluorescence intensity is still significantly reduced (p < 0.001), diffuse (p ¼ 0.99) and somatic (p ¼ 0.63) fluorescence are not significantly altered. *p < 0.05, ***p < 0.001. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) during low Mg 2þ induced epileptiform activity. However, whether the same process underlies the regulation of GABA A Rs during epileptiform activity induced by different approaches (Goodkin et al., 2008;Laur en et al., 2005) remains to be determined. The observation that NMDAR activation and rises in internal calcium are common to all models of SE (Raza et al., 2004;Rice and DeLorenzo, 1998;Mazarati and Wasterlain, 1999) would suggest that this is a universal mechanism.
Interestingly, the direct activation of NMDARs with low Mg 2þ and NMDA induces a potent change in surface stability of GABA A R clusters. The mechanisms underlying this modulation need further investigation, however dephosphorylation of GABA A R g2 subunits could play a role (Muir et al., 2010). Our data suggest that potency of NMDAR activation correlates with the extent of GABA A R modification, and may explain the synergistic effect of benzodiazepines and NMDAR antagonists in the treatment of SE (Rice and DeLorenzo, 1999;Martin and Kapur, 2008).
Ca 2þ influx from the extracellular environment through NMDARs can alter the stability of inhibitory neurotransmitter receptors (Bannai et al., 2009;Muir et al., 2010). Bannai et al. showed that diffusion dynamics of GABA A Rs too are tuned by Ca 2þ entry from the extracellular space. Indeed, here we report an increase in GABA A R F/F 0 at t ¼ 20 min. Somatic fluorescence intensity of low Mg 2þ treated cells is significantly decreased compared to control at t ¼ 20 min (dark blue bar, p < 0.05). Treatment of low Mg 2þ perfused cells with a calcineurin autoinhibitory peptide prevents the change in fluorescence intensity (dark green bar, p < 0.05). Treatment with calcineurin autoinhibitory peptide alone does not significantly alter the fluorescence intensity of GABA A R a 2SEP (magenta bar, p > 0.05). *p < 0.05. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) intracellular Ca 2þ concentration during low Mg 2þ treatment. Moreover, Ca 2þ levels correlate with the temporal dynamics of the low Mg 2þ induced effect on GABA A Rs. This indicates that fast Ca 2þsignalling could indirectly be altering GABA A Rs surface stability. Although technically challenging, it would be of major interest to simultaneously dual record Ca 2þ dynamics and GABA A Rs surface stability to better address the relationship of low Mg 2þ induced spiking activity and GABA A R trafficking on a single cell level.
Surface stability of GABA A Rs is regulated by multiple processes which are facilitated by direct or indirect interaction with trafficking proteins (Marsden et al., 2007;Jacob et al., 2008). Our experiments further emphasise the relationship of indirect signalling via intracellular Ca 2þ sensing proteins to stabilise neurotransmitter receptors. Downstream effects of Ca 2þ sensing proteins such as calmodulin orchestrate a number of target proteins and can trigger selective effects on surface GABA A R stability. It is known that increased phosphorylation of GABA A Rs contributes to surface stability (Saliba et al., 2012). Terunuma et al. demonstrated deficits in GABA A R phosphorylation during SE mediated by protein kinase C (Terunuma et al., 2008). Opposingly, dephosphorylation induces declustering and increases the diffusion dynamics of GABA A Rs (Muir et al., 2010). Calcineurin has been shown to interact with GABA A Rs via the g 2 subunit and it modulates neuronal inhibition.
Interestingly, basal and maximal activity of calcineurin is increased (Kurz et al., 2001) and subcellular distribution is altered (Kurz et al., 2003) in SE in vivo. However, it has been poorly investigated whether calcineurin mediates the decrease in surface GABA A Rs (Wang et al., 2009). We identify calcineurin as a mediator of the decrease in surface GABA A R and therefore, provide a mechanism of inhibitory modulation during SE, and a potential target for therapy. Moreover, this study is the first to demonstrate a decrease in surface GABA A Rs by live-imaging in hippocampal neurons during SE. The identification of the underlying trafficking mechanism could account for resistance to benzodiazepines and could have additive effects on duration or frequency of seizure activity. Further research is needed to develop more effective therapeutic strategies against SE to which this study will contribute.