Gephyrin is a scaffolding protein important for the postsynaptic clustering of inhibitory neurotransmitter receptors. Here, we investigated the properties of gephyrin scaffolds at γ-aminobutyric acid- (GABA-)ergic synapses in organotypic entorhino-hippocampal cultures prepared from a transgenic mouse line, which expresses green fluorescent protein-tagged gephyrin under the control of the Thy1.2 promoter. Fluorescence recovery after photobleaching revealed a developmental stabilization of postsynaptic gephyrin clusters concomitant with an increase in cluster size and synaptic strength between 1 and 4 weeks in vitro. Prolonged treatment of the slice cultures with diazepam or a GABAA receptor antagonist disclosed a homeostatic regulation of both inhibitory synaptic strength and gephyrin cluster size and stability in 4-weeks-old cultures, whereas at 1 week in vitro, the same drug treatments modulated GABAergic postsynapse and gephyrin cluster properties following a Hebbian mode of synaptic plasticity. Our data are consistent with a model in which the postnatal maturation of the hippocampal network endows CA1 pyramidal neurons with the ability to homeostatically adjust the strength of their inhibitory postsynapses to afferent GABAergic drive by regulating gephyrin scaffold properties.
The proper balance of excitation and inhibition is a prerequisite for normal functioning of the nervous system and adequate behavioral responses (Luria 1932). Activity-induced changes in the efficacy of selected synapses that disturb this balance, such as long-term potentiation or long-term depression, must hence be compensated by homeostatic regulatory mechanisms that maintain network stability (reviewed in Pozo and Goda 2010; Turrigiano 2011). Similarly, during development changes in the number and strength of synapses on a given neuron require the homeostatic adaptation of excitability and synaptic efficacy to keep firing rates in the proper range. In central neurons, a slow adaptive mechanism termed homeostatic synaptic scaling has been shown to globally adjust the properties of excitatory synapses to persistent changes in activity, mainly by postsynaptically increasing or decreasing α-amino-3-hydroxy-5-methyl-4-isoxazolpropionic acid (AMPA) receptor numbers (for review see Turrigiano 2011).
At inhibitory synapses, evidence for adaptive changes in synaptic strength has also emerged (Turrigiano 2011). There, the scaffolding protein gephyrin is known to be important for the clustering of glycine and γ-aminobutyric acidA (GABAA) receptors (GABAARs; Essrich et al. 1998; Feng et al. 1998; Kneussel et al. 1999). Gephyrin is thought to form a hexagonal lattice beneath the postsynaptic membrane (Sola et al. 2004; Saiyed et al. 2007; Herweg and Schwarz 2012) and to determine receptor densities at, and hence the efficacy of, inhibitory synapses by interacting with specific receptor subunits (Meyer et al. 1995; Tretter et al. 2011). Single molecule tracking studies indicate that extrasynaptic glycine receptors and GABAARs diffuse freely in the neuronal plasma membrane but upon entering synaptic sites become less mobile due to receptor trapping by the postsynaptic gephyrin scaffold (Meier et al. 2001; Dahan et al. 2003; Jacob et al. 2005; Levi et al. 2008; Lüscher et al. 2011). Both the synaptic receptors and clustered gephyrin are subject to continuous exchange due to dissociation from and reassociation to the postsynapse (Maas et al. 2006; Calamai et al. 2009). The size and properties of gephyrin clusters are regulated by neuronal activity (Kirsch and Betz 1998; Bannai et al. 2009; Maas et al. 2009) and phosphorylation reactions (Bausen et al. 2010; Tyagarajan et al. 2011) and depend on interactions with the cytoskeleton (Kirsch and Betz 1995; Bausen et al. 2006; Charrier et al. 2006).
In this study, we used entorhino-hippocampal slice cultures prepared from mice expressing green flluorescent protein gephyrin (GFP-gephyrin) under the control of the Thy1.2 promoter to study the dynamics of gephyrin clusters at different postnatal stages by determining fluorescence recovery after photobleaching (FRAP). We find that synaptic gephyrin clusters are stabilized during developmental maturation in parallel with an increase in gephyrin cluster size and synaptic strength, and that in young slice cultures gephyrin cluster size and stability are increased by GABAAR activity. In contrast, in mature slice cultures, gephyrin cluster size and stability are regulated inversely by GABAAR activity. These results disclose a novel step in the postnatal maturation of the inhibitory hippocampal network, which endows CA1 pyramidal neurons with the capacity to change inhibitory synaptic strength in a homeostatic manner by regulating the dynamics of gephyrin scaffolds.
Materials and Methods
Generation of Thy1-EGFPgephyrin Mice
Transgenic mice expressing GFP-gephyrin under the control of the Thy1.2 promoter were generated following a previously described strategy (Caroni 1997; Feng et al. 2000). The pEGFP-C2-gephyrin plasmid encoding an enhanced green fluorescent protein (EGFP) fusion protein of rat gephyrin cDNA p1 (Prior et al. 1992) along with the 800-bp fragment of gephyrin's 3′ untranslated region (3′ UTR) has been described previously (Fuhrmann et al. 2002). A NheI–SacII fragment containing EGFPgephyrin obtained from this plasmid was inserted into the NheI–SacII sites of a modified mouse Thy1.2 expression cassette (Caroni 1997) harboring a multiple cloning site (kindly provided by Dr G. O'Sullivan, Max-Planck Institute for Brain Research, Frankfurt, Germany; Fig. 1A). The resulting Thy1-EGFPgephyrin transgene was purified as a NotI-PvuI fragment (Fig. 1A) by passage over a Sepharose 4B column (GE Healthcare, Freiburg, Germany), and the fractions collected were analyzed by agarose gel electrophoresis. The fraction containing the NotI-PvuI 10.5-kb transgene was ethanol precipitated, resuspended in injection buffer (10 mM Tris/0.1 mM ethylenediamine tetra-acetic acid, 100 mM NaCl, pH 7.4) at a concentration of 30 ng/μL, and injected by the Transgenic Facility of the University of Heidelberg into the pronuclei of E0.5 zygotes collected from superovulated C57BL/6 female zygote donors followed by transfer to pseudopregnant C57BL/6 recipients.
Positive offspring were identified by polymerase chain reaction (PCR) of tail biopsies using a Thy1.2-specific 5′ primer (Thy1F1: 5′ TCTGAGTGGCAAAGGACCTTAGG 3′) and an EGFP-specific 3′ primer (EGFPrev: 5′ GTCCATGCCGAGAGTGATC 3′) to amplify a 940-bp fragment (Fig. 1B). Mice positive by PCR were further screened for germline integration of the transgene by backcrossing them with C57BL/6 wildtype (wt) mice, and two mice were identified as founders. All experiments described here were performed on progeny from founder 1 (Thy1-EGFPgephyrin transgenic line 1).
Brain homogenates were prepared as described (Kneussel et al. 1999), separated on 10% (w/v) SDS–polyacrylamide gels (30 μg protein/lane), transferred to nitrocellulose (Schleicher & Schüll, Dassel, Germany), and probed with different primary antibodies. The following primary antibodies were used: anti-gephyrin (mouse, BD Biosciences, Heidelberg, Germany, 1:500), anti-GFP (rabbit, Invitrogen, Karlsruhe, Germany; 1:1000), anti-GABAAR, γ2 subunit (rabbit, Alomone Laboratories, Jerusalem, Israel; 1:200), anti-glycine receptor (GlyR) α-subunits (mouse, mAb4a; 1:2000; Kirsch and Betz 1993), anti-vesicular inhibitory amino acid transporter (VIAAT; rabbit, Synaptic Systems GmbH, Goettingen, Germany; 1:1000), anti-β-actin (mouse, Sigma-Aldrich, Taufkirchen, Germany; 1:1000), followed by incubation with secondary antibodies conjugated to horseradish peroxidase (Promega, Madison, WI, USA; 1:10,000). Bound antibody was visualized using an enhanced chemiluminescence detection system (Perbio Science, Bonn, Germany).
Immunostaining on Cryostat Sections
For immunostainings with the gephyrin-specific antibody mAb7a (mouse, Synaptic Systems GmbH; 1:500; Feng et al. 1998; Kirsch and Betz 1998), mice were deeply anesthetized and decapitated. The brains were immediately removed and frozen on dry ice. Coronal hippocampal cryostat sections (14 μm) were fixed with 4% (w/v) paraformaldehyde (PFA) for 10 min at 4 °C, permeabilized with 0.3% (w/v) Triton X-100, and incubated overnight at 4 °C with the primary antibody at appropriate dilutions in phosphate buffered saline (PBS)/10% goat serum and for 1 h with a secondary antibody (Alexa 546, Invitrogen; 1:1000). For double immunostaining with the antibody recognizing gephyrin (mAb7a, 1:500) and an antibody specific for VIAAT (affinity-purified rabbit polyclonal, Synaptic Systems GmbH; 1:500), mice were deeply anesthetized and decapitated. The brains were immediately removed and incubated overnight at 4 °C in 50 mL fixative containing 4% (w/v) PFA in 0.1 M phosphate buffer, pH 7.4. Coronal 20-μm sections were prepared from frozen tissue and were mounted on SuperFrost Plus slides (Menzel GmbH, Braunschweig, Germany). Sections were then postfixed for 10 min with 4% (w/v) PFA at 4 °C and preincubated for 30 min at 95 °C in sodium citrate (SC) buffer (10 mM SC, 0.05% (v/v) Tween-20, pH 8.0) for antigen retrieval. Sections were permeabilized with 0.3% (w/v) Triton X-100, followed by immunostaining with the primary and secondary antibodies (Alexa 488 and Alexa 546, Invitrogen; 1:1000) as described above.
Preparation of Slice Cultures
Entorhino-hippocampal slice cultures were prepared from heterozygous Thy1-EGFPgephyrin mice at postnatal days 4 and 5 in agreement with the German law on the use of laboratory animals using a published protocol (Del Turco and Deller 2007). No attempt was made to distinguish between sexes in these experiments. Cultivation medium contained 50% (v/v) minimum essential medium, 25% (v/v) basal medium eagle, 25% (v/v) heat-inactivated normal horse serum, 25 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) buffer solution, 0.15% (w/v) bicarbonate, 0.65% (w/v) glucose, 0.1 mg/mL streptomycin, 100 U/mL penicillin, and 2 mM glutamax. The pH was adjusted to 7.3, and the medium was replaced every second or third day. All slice cultures were kept in a humidified atmosphere with 5% (v/v) CO2 at 35 °C.
Live Cell Confocal Imaging and FRAP Analysis
FRAP experiments were performed with a Zeiss LSM Exciter confocal microscope equipped with an Acousto Optic Tunable Filter (AOTF) at 35 °C. Individual GFP-gephyrin clusters in the stratum radiatum of area CA1 were visualized using a ×40 water immersion objective lens (0.8 NA; Zeiss, Jena, Germany) and ×2 scan zoom, with the pinhole diameter set at 1 Airy Unit. Image stacks (7 sections) were taken at an ideal Nyquest rate. Imaging parameters were initially optimized to minimize bleaching by the imaging procedure itself. Following 15 min baseline registration, selected GFP-gephyrin clusters in the middle plane of the stack were completely bleached (<5% of initial fluorescence) using the bleaching function of Zeiss Zen Software (AOTF-controlled Argon laser 488 nm; 100% transmission; 100 bleach iterations), and FRAP was then followed for 80 min (Δt = 5 min).
Miniature inhibitory postsynaptic currents (mIPSCs) were recorded from CA1 pyramidal cells at 35 °C as previously described (Jedlicka et al. 2011). The bath solution (artificial cerebrospinal fluid [ACSF]) contained 126 mM NaCl, 2.5 mM KCl, 26 mM NaHCO3, 1.25 mM NaH2PO4, 2 mM CaCl2, 2 mM MgCl2, and 10 mM glucose. Patch pipettes contained 140 mM CsCl, 2 mM MgCl2, 10 mM HEPES, 0.3% (w/v) biocytin, and 10 μM Alexa 568 (pH 7.30 with CsOH, 295 mOsm with sucrose) and had a resistance of 6–10 MΩ. Data were recorded at a holding potential of −80 mV in the presence of 10 μM D-AP5, 10 μM CNQX, and 0.5 μM tetrodotoxin (TTX) and digitized at a sampling rate of 10 kHz. Series resistance was monitored every 2–3 min, and the recordings were discarded if the series resistance or the leak current reached 30 MΩ or 150pA, respectively. Cell-attached recordings (Mason et al. 2005; Perkins 2006) were performed in 0.5 μM TTX with recording pipettes filled with bath solution to qualitatively assess the effects of GABA in young (5 days in vitro [DIV]) and mature (30 DIV) CA1 pyramidal neurons. Puff application of GABA (100 μM in ACSF), glutamate (100 μM in ACSF), or ACSF was performed via a patch pipette positioned in the stratum radiatum close to the neuron of interest.
Immunostaining of Slice Cultures
Upon fixation in 4% (w/v) PFA and 4% (w/v) sucrose in PBS for 15 min, cultures were washed thoroughly in PBS, permeabilized for 1 h with 0.3% (v/v) Triton X-100, 4% (v/v) normal goat serum in PBS, and blocked for 2 h with 10% (v/v) normal goat serum in PBS. Cultures were incubated overnight at 4 °C with the affinity-purified polyclonal rabbit anti-VIAAT antibody (cytoplasmic domain, Synaptic Systems GmbH; 1:500) in PBS and 10% (v/v) goat serum, and after washing for 1 h with secondary antibodies (appropriate Alexa-coupled secondary antibodies, Invitrogen; 1:500). Cultures were then washed thoroughly in PBS, coverslipped in anti-fading mounting medium (DAKO Fluoromount) and analyzed using a Nikon Eclipse C1si confocal laser-scanning microscope equipped with a ×60 oil immersion objective (NA 1.3, Nikon; ×4 scan zoom).
Quantification and Statistics
Experimental data were evaluated by investigators blind to experimental conditions. The Image-J software package (http://rsb.info.nih.gov/ij) was used to analyze immunostainings and FRAP experiments. Immunostainings were analyzed as described (Bas Orth et al. 2005). FRAP of individual GFP-gephyrin clusters was corrected for background and bleaching by the imaging procedure itself. Values were normalized to prebleach fluorescence and to the first time point after bleaching. Values were expressed as the percentage of mean prebleach fluorescence (i.e. averaged corrected fluorescence of baseline recordings; Δt = 5 min) and averaged per culture, that is, per independent experiment. FRAP data were fitted using the curve fitting toolbox of MATLAB version 7.5 (The MathWorks Inc., Natick, MA, United States of America) with the following bi-exponential equation:4). Due to difficulties to reliably determine the often rather small Pf values observed in the pharmacological experiments presented in Figures 5 and 6, here statistical evaluations were based on comparisons of the averaged FRAP values sampled between 65 and 80 min. Apposition of gephyrin and VIAAT clusters was assessed manually on 3D-image stacks using the Zeiss LSM image browser by scrolling through the stack. Electrophysiological data were analyzed using pClamp 10.2 (Axon Instruments, United States of America) and MiniAnalysis (Synaptosoft, United States of America) software. Events (300–400) were analyzed per recorded neuron. Since a normal distribution of our data could not be assured, statistical comparisons in this study were made using the non-parametric Wilcoxon Mann–Whitney test. The Kruskal–Wallis test and the Wilcoxon Mann–Whitney test followed by Bonferroni's correction for multiple comparisons were used in some cases (Figs 1G, 5A–D, and 6A–D). P-values <0.05 were considered significant. All values represent means ± standard error of the mean. In the figures, *P < 0.05, **P < 0.01, and ***P < 0.001; not significant differences are indicated with “n.s.”
Confocal image stacks were exported as 2D projections and stored as TIFF files. Figures were prepared using the Photoshop CS2 graphics software (Adobe, San Jose, CA, United States of America). Image brightness and contrast were adjusted.
Generation and Characterization of Thy1-EGFPgephyrin Transgenic Mice
To allow visualization of gephyrin dynamics in living neurons, we generated mice in which the neuronal p1 splice variant of rat gephyrin (Prior et al. 1992) tagged with EGFP is expressed under the control of the neuron-specific Thy1.2 promoter (Fig. 1A,B; Caroni 1997; Feng et al. 2000). Thy1-EGFPgephyrin mice appeared phenotypically normal and showed no obvious alterations in development and fertility. Expression of GFP-gephyrin was readily detected in the hippocampus, neocortex (layer II, V, and VI), brain stem (caudal medulla), and spinal cord but barely seen in cerebellar neurons (Fig. 1D). Western blot analysis of brain homogenates from Thy1-EGFPgephyrin mice with antibodies against EGFP and gephyrin revealed a band of approximately 125 kDa corresponding to GFP-gephyrin (green arrow in Fig. 1C) that was not seen in age-matched wt animals which contained only the endogeneous 98 kDa gephyrin band (red asterisk in Fig. 1C). The expression levels of endogenous gephyrin and other postsynaptic (γ2 subunit of GABAAR, α subunits of GlyRs) or presynaptic (VIAAT) proteins were comparable among genotypes with β-actin as a loading control (Fig. 1C, right panel).
GFP-gephyrin expression may affect the clustering of endogenous gephyrin at developing postsynapses. Hence, sagittal brain sections from wt and Thy1-EGFPgephyrin mouse brains were counterstained with the gephyrin-specific antibody, mAb 7a, and gephyrin cluster densities, and cross-sectional areas were determined in the hippocampal formation (Fig. 1E–H). This revealed an overlap of EGFP-labeled and mAb 7a-immunoreactive clusters in the dendritic layers of the Thy1-EGFPgephyrin animals examined (Fig. 1E, F). The GFP-gephyrin clusters were synaptically localized, as revealed by their apposition to presynaptic VIAAT immunoreactivity, and their mean size in the stratum radiatum was not significantly different from that of mAb 7a-immunoreactive and GFP-negative clusters in brain sections prepared from either wt or Thy1-EGFPgephyrin animals (wt: 0.24 ± 0.01 μm2; Tg: GFP-gephyrin(−), 0.23 ± 0.01 μm2; and GFP-gephyrin(+), 0.25 ± 0.01 μm2; Fig. 1G). Also, the densities of mAb 7a-immunoreactive puncta per 1000 μm2 of the stratum radiatum did not significantly differ between genotypes (328 ± 29 in wt and 288 ± 21 in Tg; Fig. 1H). Thus, GFP-gephyrin expression had no apparent effect on gephyrin clustering in the dendritic layers of the hippocampal formation. However, we found GFP-gephyrin aggregates in the somata of neurons located in the stratum pyramidale and at the border between the stratum pyramidale and the oriens (arrow in Fig. 1E). The somatic aggregation of GFP-gephyrin could also be seen in neurons of the granule cell layer, in the hilus of the dentate gyrus, and in the layer V of the neocortex (data not shown). Similar gephyrin aggregates have been previously observed in mammalian cells (Meyer et al. 1995) and cultured neurons (Colin et al. 1996).
Developmental Maturation of GFP-Gephyrin Clusters and Synaptic Inhibition in Entorhino-Hippocampal Slice Cultures From Thy1-EGFPgephyrin Mice
To further examine the properties and dynamics of gephyrin clusters under close to in situ conditions, we used organotypic entorhino-hippocampal slice cultures prepared from brains of 4–5-days-old Thy1-EGFPgephyrin mice. These slice cultures are known to recapitulate the major steps of hippocampal development with excellent preservation of in vivo connectivity (Frotscher et al. 1995; Gahwiler et al. 1997). Prominent GFP-gephyrin expression was regularly seen in area CA1 of these organotypic cultures (Fig. 2A) both, after only 5–8 DIV (5–8; “week 1”), and ≥4 weeks in culture (DIV 30–33; “week 4”). At both, the young and the mature stage, individual GFP-gephyrin clusters were clearly discernable and mainly associated with dendritic shafts (Fig. 2D). Notably, most if not all of the GFP-gephyrin clusters were synaptically localized as revealed by co-immunolabeling for the presynaptic VIAAT (Fig. 2B). In 1- and 4-weeks-old cultures, about 92% (92.3 ± 0.7% and 92.2 ± 0.8%, respectively) of the GFP-gephyrin clusters were found to be apposed to VIAAT immunoreactivity in the CA1 stratum radiatum. This value is consistent with an almost exclusive localization of the GFP-gephyrin clusters at sites of nerve terminal contact and considerably higher than respective colocalization values determined in dissociated hippocampal neuron cultures (∼54–74%; Dobie and Craig 2011). We attribute this difference to the high anatomical integrity of the organotypic slice preparation.
Quantitative evaluations showed that the sizes of GFP-gephyrin clusters and VIAAT immunoreactive puncta increased during the differentiation of the slice cultures in vitro (GFP-gephyrin: 0.13 ± 0.01 μm2 at 1 week and 0.24 ± 0.01 μm2 at 4 weeks, P < 0.001; VIAAT: 0.20 ± 0.01 μm2 at 1 week and 0.28 ± 0.03 μm2 at 4 weeks, P < 0.05; Fig. 2C). In contrast, the corresponding cluster densities did not change significantly during this period (VIAAT cluster densities of 339 ± 37 per 1000 μm2 at 1 week and of 320 ± 15 per 1000 μm2 at 4 weeks; GFP-gephyrin cluster densities per μm dendrite of 0.23 ± 0.02 at 1 week and of 0.20 ± 0.01 at 4 weeks; Fig. 2D).
To evaluate the functional consequences of the changes in GFP-gephyrin cluster size and VIAAT immunoreactivity observed between 1 and 4 weeks in culture, individual CA1 pyramidal neurons were patch-clamped, and mIPSCs were recorded (Fig. 3). In these experiments, Alexa 568 containing patch pipettes were used. This made it possible to readily distinguish GFP-gephyrin(+) and GFP-gephyrin(−) neurons in confocal image stacks prior to recording (Fig. 3A). No significant differences in mIPSC amplitudes and inter-event intervals were found between GFP-gephyrin(+) and GFP-gephyrin(−) cells in both 1- and 4-weeks-old cultures (Fig. 3B, C). Apparently, transgenic expression of GFP-gephyrin did not affect the strength of inhibitory neurotransmission in cultured CA1 pyramidal cells from Thy1-EGFPgephyrin mice. However, at 4 weeks, a ∼50% increase in mean mIPSC amplitude was seen in CA1 pyramidal cells when compared with 1-week-old cultures (22.8 ± 1.4pA at 1 week, and 33.5 ± 1.2pA at 4 weeks; P < 0.001; Fig. 3D, E), and the mean frequency of the mIPSCs was about 10-fold higher (3.0 ± 0.2 Hz at 1 week and 32.7 ± 2.2 Hz at 4 weeks; P < 0.001). In addition, their average rise time was shortened (4.8 ± 0.15 ms at 1 week and 2.0 ± 0.04 ms at 4 weeks; P < 0.001), whereas decay times were not significantly different (13.5 ± 0.6 ms at 1 week and 12.6 ± 0.2 ms at 4 weeks). These results indicate an important strengthening of synaptic inhibition between 1 and 4 weeks in culture.
GABA has been shown to depolarize cultured neurons at embryonic and newborn stages (Ben-Ari et al. 2007). To exclude that GABA responses are depolarizing at the developmental stages analyzed here, we recorded responses to GABA-puffs in the cell-attached mode from CA1 pyramidal neurons at 5 and 30 DIV (Fig. 3F). These experiments revealed that GABA is hyperpolarizing in both 1- and 4-weeks-old slice cultures that were prepared at postnatal day p4.
Gephyrin Clusters are Stabilized During Synapse Maturation
The findings described above showed that the functional maturation of inhibitory synapses coincides with an increase in mean gephyrin cluster size. To unravel whether this increase may reflect a change in the exchange rate of gephyrin, we determined the FRAP of GFP-gephyrin clusters in 1- and 4-weeks-old slice cultures (Fig. 4). Following a 15-min baseline recording, individual GFP-gephyrin clusters were completely bleached (<5% of initial fluorescence intensity). The reappearance of GFP fluorescence was then followed for up to 80 min at 5-min intervals (Fig. 4). These experiments disclosed a significant difference in the FRAP of GFP-gephyrin clusters depending on the developmental stage analyzed (P < 0.01 for averaged 65–80 min values; Fig. 4B). In 1-week-old cultures, 30% of the averaged prebleach GFP-gephyrin fluorescence recovered within ∼5 min, and up to 80% within 80 min, after photobleaching. In contrast, in 4-weeks-old cultures, 30% recovery was reached only ∼30 min after the bleach, and after 80 min, <40% of the prebleach signal was detected. When applying a bi-exponential fitting procedure to the FRAP curves presented in Figure 4B, 2 kinetically distinct components could be distinguished, a fast one with a recovery time constant of about 1–3 min (τf of 2.2 ± 0.9 min at 1 week and of 0.9 ± 0.4 min at 4 weeks; P = 0.22) and a slower one with a constant ranging between 40 and 70 min (τs of 68 ± 18 min at 1 week, and of 43 ± 11 min at 4 weeks; P = 0.42; Fig. 4C). Notably, although the τf and τs values were not significantly different at the 2 stages analyzed, the relative contributions of the fast and slow components differed significantly in that the relative fraction of the fast component was lower (Pf-norm of 28 ± 4.9% at 1 week, and of 12 ± 3.1% at 4 weeks; P < 0.05), and of the slow component larger (Ps-norm of 72 ± 4.9% at 1 week, and of 88 ± 3.1% at 4 weeks; P < 0.05), at 4 weeks when compared with 1 week. More importantly, when the respective stationary fractions were calculated, a >10-fold increase to 43 ± 8.9% was found at 4 weeks over 2.8 ± 6.4% at 1 week (P < 0.05; Fig. 4C). Together, these results showed that the re-incorporation of GFP-gephyrin into individual clusters is slowed during postnatal development. We attribute this slowing to a reduced turnover, that is, the stabilization of postsynaptic gephyrin scaffolds (see Discussion).
To examine whether the slowed FRAP of the GFP-gephyrin clusters observed in the mature slice cultures is a consequence of cluster growth, the recovery rates of equally sized GFP-gephyrin clusters were compared between 1- and 4-weeks-old cultures (Fig. 4D). Unexpectedly, the difference in the FRAP rate seen between the 2 developmental stages was still preserved; GFP-gephyrin clusters in 1-week-old cultures recovered faster than clusters of equal size in 4-weeks-old cultures (P < 0.05 for averaged 65–80 min values). Moreover, comparison of clusters of significantly different sizes within each of the 2 age groups disclosed no difference in FRAP of small versus large GFP-gephyrin clusters (data not shown). Thus, gephyrin cluster size does not determine gephyrin re-incorporation rates but additional mechanisms, such as diffusional constraints resulting from modification reactions or interactions with the underlying cytoskeleton or other postsynaptic proteins, must underlie the developmental slowing of fluorescence recovery observed here.
Prolonged Diazepam Treatment Upregulates Both Inhibitory Synaptic Strength and Gephyrin Clustering in Young Slice Cultures
Different lines of evidence indicate that neuronal activity regulates the dynamics and/or transport of gephyrin in primary hippocampal neurons (Bannai et al. 2009; Maas et al. 2009). Since presynaptic changes at inhibitory synapses were detected between 1- and 4-weeks-old slice cultures (see VIAAT cluster size in Fig. 2C, and mIPSC frequency in Fig. 3E), we reasoned that changes in inhibitory drive, that is, in postsynaptic GABAAR activity, might affect gephyrin clustering at inhibitory postsynapses of CA1 pyramidal neurons. We therefore treated 1-weeks-old slice cultures prepared from the Thy1-EGFPgephyrin mice with 1 μM diazepam to chronically increase GABAAR activity (Fig. 5A, B). Electrophysiological recordings immediately after the addition of diazepam (acute treatment) revealed a significant increase in the mean mIPSC amplitude (P < 0.01; Fig. 5A). This increase in mIPSC amplitude was still present after 12 h of drug treatment (recordings in the presence of diazepam; P < 0.01) and partially persisted after diazepam washout (P < 0.05; Fig. 5A). Apparently, the CA1 pyramidal neurons in the 1-week-old cultures reacted to prolonged diazepam potentiation of GABAergic inhibition with an increase in inhibitory synaptic strength.
To disclose whether changes in gephyrin dynamics accompany the diazepam-induced increase of mIPSC amplitudes observed in the 1-week-old slice cultures, we next performed FRAP experiments. Figure 5B shows that the FRAP of GFP-gephyrin clusters in these cultures was slowed, and the cluster size increased, after 12 h of diazepam treatment when compared with untreated cultures. These differences were significant, as indicated by comparing mean cluster sizes (P < 0.01) and averaged FRAP values sampled between 65 and 80 min (P < 0.05). Thus, the pharmacological potentiation of GABAARs caused changes in GFP-gephyrin re-incorporation that partially mimicked the differences observed between 1- and 4-weeks-old slice cultures.
To further corroborate these results, GABAARs were blocked in 1-week-old slice cultures by the GABAAR antagonist SR95531 (10 μM; Fig. 5C,D). In addition, TTX (1 μM) was added to eliminate effects resulting from changes in network activity, that is, to prevent an increase in excitatory drive due to disinhibition. While GFP-gephyrin cluster FRAP and size were not significantly affected by 12-h SR95531 + TTX (Fig. 5C, D), a 24-h treatment with SR95531 + TTX caused a significant increase in FRAP (P < 0.05 for averaged 65–80 min values; Fig. 5C) and a significant decrease in GFP-gephyrin cluster size (P < 0.05; Fig. 5D). Control experiments showed that incubations with TTX alone for 12 and 24 h did not significantly affect the fluorescence recovery of GFP-gephyrin (Supplementary Fig. S1A), although some tendency toward changed FRAP was observed during the early recovery phase. Similarly, the mean GFP-gephyrin cluster size was unaltered after prolonged TTX treatment (Supplementary Fig. S1A), suggesting that network activity does not play a major role in gephyrin cluster regulation in young slice cultures, at least within the 12–24-h treatment period used here. Taken together, these results disclosed a regulation of gephyrin cluster properties by GABAAR activity in 1-week-old slice cultures that is consistent with a Hebbian model of synaptic plasticity, according to which more active synapses are strengthened (Hebb 1949).
In Mature Slice Cultures GABAA Receptor Activity Regulates Synaptic Strength and Gephyrin Cluster Properties in a Homeostatic Manner
We next investigated whether a comparable regulation of postsynaptic gephyrin clusters by GABAAR activity is observed in mature slice cultures (Fig. 6). Acute treatment with diazepam resulted in a significant increase in the mean mIPSC amplitude of mature CA1 pyramidal neurons, as seen in young cultures (P < 0.01; Fig. 6A). However, after 12-h incubation with diazepam, this increase in amplitude was not observed when recording mIPSCs in the continued presence of diazepam. Moreover, the subsequent washout of diazepam revealed reduced mIPSC amplitudes (P < 0.05; Fig. 6A). Thus, the CA1 pyramidal neurons in 4-weeks-old slice cultures reacted to the prolonged diazepam potentiation of GABAergic inhibition with a compensatory decrease in inhibitory synaptic strength. This result is in line with previous reports (Poisbeau et al. 1997; Li et al. 2000) and concurs with the concept of homeostatic synaptic plasticity in response to chronic perturbations of synaptic activity (Pozo and Goda 2010; Turrigiano 2011).
To test whether changes in gephyrin exchange rate underlie the diazepam-induced reduction of mIPSC amplitudes, we also performed FRAP experiments on 4-weeks-old slice cultures. Figure 6B shows that a 12-h treatment with diazepam significantly accelerated the fluorescence recovery of GFP-gephyrin clusters when compared with untreated control cultures (P < 0.05). In addition, a significant reduction in GFP-gephyrin cluster size was found (P < 0.01; inset in Fig. 6B). These results suggested that GABAAR activity regulates gephyrin clustering at mature inhibitory postsynapses in a homeostatic manner.
To further corroborate a homeostatic regulatory mechanism, GABAARs were blocked by SR95531 (10 μM) + TTX (1 μM). In contrast to what was observed in 1-week-old cultures, at 4 weeks incubation with SR95531 + TTX for 12 h resulted in a significant reduction in GFP-gephyrin FRAP when compared with untreated control cultures (P < 0.05 for averaged 65–80 min values; Fig. 6C). This result is consistent with a compensatory increase in the fraction of stationary gephyrin upon the prolonged blockade of GABAAR activity. Similar to the results obtained with 1-week-old cultures, treatment with TTX alone for either 12 or 24 h did not significantly affect GFP-gephyrin FRAP and mean GFP-gephyrin cluster size (Supplementary Fig. S1B). Interestingly, GFP-gephyrin cluster size was also not significantly affected by 12-h SR95531 + TTX (Fig. 6D), indicating that changes in gephyrin exchange rate may precede changes in cluster size. In line with this interpretation, a 24-h treatment with SR95531 + TTX produced not only a further decrease in fluorescence recovery when compared with the 12-h treatment but also a highly significant increase in GFP-gephyrin cluster size (P < 0.001; Fig. 6D). Thus, GABAAR activity inversely regulates gephyrin cluster properties at mature inhibitory postsynapses, a mechanism which may serve to homeostatically adjust the strength of inhibitory postsynaptic currents to afferent GABAergic drive.
The major finding of this study is that the developmental maturation of GABAergic postsynapses in an organotypic hippocampal slice preparation is characterized not only by a striking increase in inhibitory synaptic strength concomitantly with increases in gephyrin cluster size and stability, but in addition involves the acquisition of a mechanism that allows to homeostatically regulate gephyrin cluster properties in response to changes in GABAAR activity. In mature slice cultures, chronic diazepam potentiation of GABAARs resulted in a “down-scaling” of both gephyrin clusters and inhibitory transmission, while in young cultures, prolonged GABAARs activation caused the opposite changes. Our results disclose an interdependence between gephyrin scaffold properties and inhibitory synaptic strength. Furthermore, they suggest a yet unknown differentiation step in the postnatal maturation of the hippocampal network, which endows neurons with the ability to homeostatically adjust the strength of their inhibitory postsynapses in response to prolonged changes in afferent GABAergic drive.
To allow direct visualization of gephyrin clusters in individual living neurons, we generated transgenic mouse lines that express GFP-gephyrin under the control of the Thy1.2 promoter (Caroni 1997; Feng et al. 2000). In the Thy1-EGFPgephyrin line used here, GFP-gephyrin expression was observed in the neocortex, the hippocampus, brain stem, and spinal cord. Transgenic GFP-gephyrin expression did not detectably affect the morphology and functional properties of GABAergic synapses. Immunostaining of brain sections failed to reveal changes in the mean size and density of gephyrin clusters in GFP-gephyrin expressing mice when compared with non-expressing ones. Similarly, in patch-clamp recordings, we found no difference in mIPSC properties between GFP-gephyrin(+) and GFP-gephyrin(−) neurons. We assume this to reflect comparatively low levels of GFP-gephyrin expression that are within or close to inter-animal variance. Consistent with this view, the Thy1-EGFPgephyrin mice did not show gross behavioral abnormalities and bred normally.
For all in vitro experiments reported here, we used entorhino-hippocampal slice cultures prepared from the Thy1-EGFPgephyrin mice. In such organotypic cultures, neuronal networks are well-preserved (Frotscher et al. 1995; Gahwiler et al. 1997), and changes in synapse morphology and protein composition can be studied at high resolution under developmental as well as mature conditions (Li et al. 1995; Dailey and Smith 1996; Ziv and Smith 1996; Del Turco and Deller 2007). Here, patch-clamp recordings from CA1 pyramidal neurons in the slice cultures revealed that the frequency of GABAergic mIPSCs increased about 10-fold between 1 and 4 weeks in vitro. This likely reflects a presynaptic maturation step that may involve an increase in the vesicle pool and/or a higher vesicle release probability, since only the average size but not the density of VIAAT immunoreactive puncta increased during this period. Concomitantly, mean mIPSC amplitudes increased by about 50% together with a shortening of the average rise time and an 85% increase in mean GFP-gephyrin cluster area, whereas cluster density did not change. Our data are consistent with considerable growth of both pre- and postsynaptic compartments accompanying the functional maturation of GABAergic neurotransmission.
FRAP analysis disclosed that GFP-gephyrin clusters are dynamic structures at the 2 stages of in vitro development analyzed. The observed 30% recovery rates in the range of ∼5 to 30 min are in agreement with previous studies showing that gephyrin continuously enters and leaves postsynaptic sites (Hanus et al. 2006; Maas et al. 2006, 2009; Calamai et al. 2009). It should, however, be noted that the FRAP rates determined here are much longer than the average dwell time of GABAARs at the synapse as estimated by single-particle tracking, which in dissociated hippocampal neurons has been reported to be only a few seconds (Bannai et al. 2009). Such short residence times would support models in which inhibitory receptors associate with and dissociate from the postsynaptic gephyrin scaffold without attached gephyrin (Kirsch and Betz 1998; Bogdanov et al. 2006). However, FRAP experiments with pHluorin-tagged GABAAR subunits have disclosed considerably slower bulk recovery rates for synaptic GABAARs that lie within a similar time range as those determined here for GFP-gephyrin FRAP (Jacob et al. 2005). Also, an about 50% recovery of GABAergic mIPSCs after agonist-induced irreversible inhibition of synaptic GABAAR has been shown to require 10 min and proposed to depend on the dissociation of the blocked receptors from their postsynaptic anchoring sites as the limiting step in recovery (Thomas et al. 2005). These latter estimates of synaptic GABAAR mobility together with our GFP-gephyrin FRAP results are consistent with time-lapse imaging studies, suggesting that GABAARs enter and leave postsynaptic sites with gephyrin being bound (Maas et al. 2006).
Our developmental FRAP analysis revealed that the kinetics of GFP-gephyrin re-incorporation into individual clusters change between 1 and 4 weeks in culture. During this period of hippocampal neuron maturation, initial FRAP rates were slowed, and in the 4-weeks-old cultures, a major fraction of the bleached fluorescence failed to recover during our 80-min observation period. Apparently, the slowly exchanging “stationary” fraction of gephyrin within postsynaptic clusters increases during postnatal development. This can be attributed to reduced scaffold turnover, that is, a decrease in gephyrin dissociation and not association rate, since clusters grow whilst FRAP is slowing. In other words, synapse growth during postnatal development is associated with not only an enlargement of the postsynaptic gephyrin scaffold but also a substantial increase in its stability. Both changes may provide for more efficient receptor trapping and, thereby, enhance the efficiency of inhibitory neurotransmission. It should, however, be emphasized that, although being nicely correlated during development, gephyrin cluster size and stability are not strictly coupled. In both 1- and 4-weeks-old cultures, the FRAP rates of individual clusters proved to be independent of their size. Thus, the stationary fraction of gephyrin within a single cluster is not determined by cluster size. This excludes models of postsynaptic scaffold stabilization, in which the developmental increase in the stationary fraction of gephyrin would simply reflect gephyrin recruitment to and dissociation from postsynaptic clusters occurring only at the cluster periphery, that is, the increasing area-to-circumference ratio resulting from cluster growth. Clearly more specific mechanisms, such as selective protein interactions and/or modifications leading to a denser packing of the scaffold or its tighter interaction with the cytoskeleton must underlie the stabilization of gephyrin clusters occurring postnatally.
A central finding of the present study was the observation that gephyrin scaffolds are regulated inversely by afferent GABAergic synaptic activity in mature slice cultures, while in young cultures, such compensatory regulation was not observed. In 1-week-old cultures, GABAAR potentiation by diazepam for 12 h increased both the size and the stationary fraction of gephyrin clusters as well as mean mIPSC amplitudes, whereas a 24-h incubation with the GABAAR antagonist SR95531 caused a reduction in gephyrin cluster size and increased FRAP. These results obtained with the young slice cultures are in agreement with the activity dependence of inhibitory synapse maturation seen in GABAAR deficient mice (Patrizi et al. 2008) and in animals with a conditional knockdown of glutamate decarboxylase 67 (GAD-67) in basket interneurons (Chattopadhyaya et al. 2007). Thus, after 1-week in vitro enhanced synaptic activity led to larger and more stable inhibitory postsynapses, obeying the rules of synaptic strengthening proposed by Hebb (1949). In contrast, in 4-weeks-old cultures, the same diazepam treatment reduced gephyrin cluster size and increased fluorescence recovery together with a decrease in mean mIPSC amplitude. Inversely, when these 4-weeks-old cultures were treated with SR95531 for 24 h, both the size and the stability of the GFP-gephyrin clusters increased. These results indicate that in mature hippocampal neurons changes in GABAergic transmission affect the size and stability of gephyrin clusters in a way that adjusts synaptic strength in a homeostatic manner. This observation is consistent with recent electrophysiological and immunocytochemical data from dissociated hippocampal neurons that demonstrated homeostatic changes in mIPSC properties as well as in the size of GABAAR clusters and presynaptic GAD-65 puncta upon prolonged depolarization (see also Kilman et al. 2002; Hartman et al. 2006; Maffei et al. 2006; Rannals and Kapur 2011). Apparently, CA1 pyramidal neurons of entorhino-hippocampal slice cultures acquire postnatally the ability to adjust their inhibitory synaptic strength in a homeostatic manner. This mechanism, which involves the activity-dependent regulation of gephyrin scaffolds, may serve to adjust and maintain the stability of the inhibitory network following its postnatal maturation.
The molecular mechanisms that regulate the size and stability of gephyrin clusters at synaptic sites remain poorly understood. Different interacting proteins, for example, neuroligin 2 (Poulopoulos et al. 2009) and collybistin (Papadopoulos et al. 2007), have been implicated in gephyrin deposition and maintenance at inhibitory synapses. As discussed elsewhere (Kneussel and Betz 2000), synaptic signaling pathways triggered by Ca2+ influx (Kirsch and Betz 1998), such as Ca2+-dependent phosphorylation/dephosphorylation reactions (Bannai et al. 2009), may regulate inhibitory postsynaptic scaffold dynamics and thereby control GABAAR recruitment and internalization rates. Recently, abolition of gephyrin phosphorylation by glycogen synthase kinase 3β at serine residue 270 was reported to increase gephyrin cluster densities and mIPSC frequencies in dissociated hippocampal neurons (Tyagarajan et al. 2011). The same study also showed that the inhibition of the Ca2+-activated protease calpain-1 by calpastatin-enhanced gephyrin cluster number but not size. Calpain-1–mediated degradation of gephyrin is, however, unlikely to cause the reduction in gephyrin cluster size observed upon diazepam treatment of the 4-weeks-old slice cultures, since in our cultures, GABA currents were hyperpolarizing (for a recent review on excitatory effects of GABA see Bregestovski and Bernard 2012), and hence diazepam should decrease Ca2+ influx (e.g. Ben-Ari et al. 1989; Ganguly et al. 2001; Tyzio et al. 2007). We therefore propose that the homeostatic regulation of gephyrin cluster properties involves changes in gephyrin's oligomerization/dissociation rather than synthesis/degradation rates.
The maturation step through which cultured CA1 pyramidal neurons in organotypic slice cultures acquire the capacity to regulate inhibitory synaptic strength in a homeostatic manner remains to be determined. It is interesting to speculate that changes known to affect GABAergic neurotransmission during postnatal development, such as changes in KCC2 expression (Blaesse et al. 2009), could be involved in this process. A systematic comparison of the differences in mRNA and protein expression levels between 1- and 4-weeks-old slice cultures may allow to identify novel candidate regulatory molecules involved in the homeostatic plasticity of inhibitory postsynapses. The Thy1-EGFPgephyrin mice described here may be a useful system to address these and other questions in future work.
This work was supported by the Max-Planck Society and Deutsche Forschungsgemeinschaft (EXC 115 to H. B. and DFG DE 551/10-1 to T.D.). A.V. was supported by a Young Investigators Grant (Faculty of Medicine, Goethe-University Frankfurt) and S.R.A. was supported by the International Max-Planck Research School, Frankfurt.
We thank Dr Gregory O'Sullivan for help with initiating this project and comments on an earlier version of this manuscript, Dr Pico Caroni for kindly providing the mouse Thy1.2 expression cassette, Dr Jeong-Seop Rhee for help with the curve fitting analysis, Dr Gaby Schneider for assistance in statistics, and Charlotte Nolte-Uhl, Nadine Zahn, Belquis Nassim-Assir, and Ina Bartnik for excellent technical assistance. Conflict of Interest: None declared.