Activation of metabotropic GABA receptors increases the energy barrier for vesicle fusion

Metabotropic signalling through γ-aminobutyric acid B receptors (GABA_BR) has been implicated in a number of physiological and pathophysiological processes, such as hippocampal rhythmic activity (Scanziani, 2000), seizure disorders and anxiety and depression (Cryan and Kaupmann, 2005). On the cellular level, GABA_BRs are found to be both postsynaptically and presynaptically localized. On the postsynaptic side, they mediate the slow component of the inhibitory postsynaptic potential through activation of K⁺ conductances (Luscher et al., 1997; Newberry and Nicoll, 1984). On the presynaptic side, they can act as auto- or hetero-receptors on GABAergic and glutamatergic terminals, respectively. In both cases activation of presynaptic GABA_BRs leads to a decrease in transmitter release (Bowery, 2006). The reduction of transmitter release is for the most part achieved by the closing of voltage-dependent Ca²⁺ channels (VDCCs) (Dunlap and Fischbach, 1981). This conversion of VDCCs into the ‘reluctantly opening state’ occurs by G-protein βγ-subunit-mediated signal transduction in a membrane-delimited pathway, as has been directly demonstrated for the calyx of Held (Kajikawa et al., 2001; Takahashi et al., 1998).

However, there are also indications that GABA_BRs can modulate transmitter release independently of VDCCs. In hippocampal neuronal cell cultures, it has been shown that both adenosine receptor and GABA_BR activation can reduce the frequency of action potential (AP)-independent miniature excitatory postsynaptic currents (mEPSCs) in pyramidal neurons (Scanziani et al., 1992; Scholz and Miller, 1992). Furthermore, these receptors can inhibit artificially induced transmitter release evoked by ionomycin and α-latrotoxin, which occurs independently of VDCCs (Capogna et al., 1996). It has been argued that this form of GABA_BR-mediated inhibition acts ‘downstream’ of VDCCs, but a possible molecular mechanism for this modulation in the mammalian central nervous system has so far remained elusive. Studies in the lamprey spinal cord provided evidence that the SNARE complex member SNAP-25 is important for presynaptic serotinergic modulation of transmission (Blackmer et al., 2001). These authors showed that this modulation was largely independent of VDCCs but involved a direct G_βγ-subunit-mediated pathway acting on the SNAP-25 C-terminus (Gerachshenko et al., 2005).

In the present study, we investigated the mechanism by which GABA_BRs modulate synaptic transmission at hippocampal pyramidal neurons. We hypothesized that a direct interference of GABA_BR signalling with transmitter release at the level of the release machinery would increase the energy barrier for vesicle fusion (Rosenmund and Stevens, 1996). We found that GABA_BRs indeed affect transmission through two G-protein dependent pathways: besides the classical pathway involving reduction of Ca²⁺-influx through VDCCs, a second, downstream pathway leads to an increase of the energy barrier for vesicle fusion. Notably, this form of inhibition does not require the C-terminus of the SNARE complex member SNAP-25, as it is not affected by Botulinum neurotoxin-A (BoNT-A) treatment.

In order to assess the involvement of VDCCs in GABA_BR-mediated modulation of transmission, we monitored AP-triggered presynaptic Ca²⁺ transients in the Schaffer-collateral pathway and recorded field excitatory postsynaptic potentials (fEPSPs) in area CA1 in acute hippocampal slice preparations (Fig. 1A). GABA_BR activation with baclofen (3–30 μM) profoundly inhibited fEPSP amplitudes. At the same time, baclofen reduced the presynaptic Ca²⁺ signals by a maximum of 28%. In a parallel set of experiments, we reduced the Ca²⁺ concentration in the external solution stepwise and monitored presynaptic Ca²⁺ signals or postsynaptic fEPSPs, while CaCl₂ was replaced by equimolar concentrations of MgCl₂. A data fit revealed a power function for transmitter release at CA1 synapses with an exponent of 4.1. We correlated postsynaptic responses and presynaptic Ca²⁺ signals from the baclofen experiments with the power function and found that the baclofen effect on synaptic transmission was greater than what would be expected if the drug was acting exclusively on VDCCs (Fig. 1A2). We also recorded mEPSCs from CA1 pyramidal neurons in both standard (2.5 mM) extracellular Ca²⁺, as well as in Ca²⁺-free conditions (i.e. 0 mM Ca²⁺ and EGTA in the extracellular solution, data not shown). In both conditions, baclofen reduced the frequency of mEPSCs to a similar degree (reduction in 2.5 mM Ca²⁺, 26.9±2.6%, n=6, P<0.01; in 0 mM Ca²⁺, 28.5±3.9%, n=7, P<0.01), whereas the mEPSC amplitude remained unchanged. In summary, the data from the slice experiments indicate that GABA_BR activation has an additional effect downstream of the modulation of VDCCs.

To investigate whether GABA_BRs trigger a signalling cascade by acting directly on the release machinery, we used autaptic cultures of hippocampal pyramidal neurons. Combined with techniques that allow for rapid application of hypertonic solutions, this system enabled us to study the efficiency of vesicle exocytosis in presence and absence of GABA_BR activation (Basu et al., 2007; Gerber et al., 2008), while bypassing Ca²⁺ triggering of neurotransmitter release (Rosenmund and Stevens, 1996). We first determined the sensitivity of autaptic EPSCs to baclofen treatment, which gave similar effects to those from the fEPSP recordings in hippocampal slices (Fig. 1B). Next, we applied pulses of baclofen (30 μM) to investigate the kinetics of the GABA_BR effect on neurotransmission (Fig. 1C). A reduction of mEPSC frequency was already present within the first second of baclofen application, which is similar to the outward current by postsynaptic G-protein gated inward rectifier (GIRK) channels. The speed of this effect renders phosphorylation-dependent processes rather unlikely, as it displays comparable latencies to the postsynaptic activation of GIRK channels, which is mediated by G_βγ subunits.

We also found that GABA_BR activation decreased mEPSC frequency in Ca²⁺-free (+1 mM EGTA) conditions to a similar extent as in standard extracellular Ca²⁺, further indicating a signalling cascade acting downstream of VDCCs (Fig. 1D, Table 1). It has recently been shown that activation of G-protein βγ-subunits alters vesicles fusion properties and thereby changes the kinetics and amplitudes of EPSCs (Photowala et al., 2006). In our recordings, we did not detect differences in the kinetics or amplitudes of mEPSC following application of baclofen (Table 1), which suggests that GABA_BR-mediated presynaptic inhibition does not alter the fusion mode of synaptic vesicles. We confirmed that this effect involves G_i or G_o proteins, as preincubation of cultures with 0.5 μg/ml pertussis toxin (PTX) abolished baclofen-mediated inhibition of EPSCs (Fig. 1E) and mEPSCs (Fig. 1F). In contrast to earlier reports indicating that presynaptic GABA effects are independent of G_i or G_o (Dutar and Nicoll, 1988; Thompson and Gahwiler, 1992), these results argue for a G_i or G_o-dependent effect of GABA_BRs on transmitter release, both at the level of VDCCs and further downstream.

To gain further insight into the mechanism of VDCC-independent modulation of transmission by GABA_BRs, we used pulsed application of hypertonic solutions onto autaptic cultures. Application of 500 mM sucrose can be used to test the readily releasable pool (RRP) of vesicles (Rosenmund and Stevens, 1996), whereas the sensitivity of the stimulated release upon application of intermediate hypertonicity probes the energy barrier for vesicle fusion (Basu et al., 2007; Gerber et al., 2008). Application of 30 μM baclofen did not alter the size of the RRP [Fig. 2A, 500 mM control, 16.3(±2.2)×10³ compared with + baclofen, 15.4(±2.1)×10³ vesicles; n=17, P>0.05], whereas responses to 250 mM sucrose were significantly reduced by baclofen [Fig. 2B, 250 mM control, 5.9(±0.8)×10³ vesicles compared with + baclofen: 3.9(±0.6)×10³; n=17, P<0.001]. Consequently, the fraction of vesicles released by 250 mM sucrose significantly decreased from 38.8(±3.5)% to 26(±2.5)% (values normalized to the RRP size determined with 500 mM sucrose), which amounts to a reduction of ~33%. We verified the specificity of the baclofen effect on hypertonicity-evoked release by performing sucrose applications in neurons preincubated with 2 μM CGP 55845, a high-affinity GABA_BR antagonist. In presence of the antagonist, inhibition of EPSCs by 30 μM baclofen was essentially abolished (supplementary material Fig. S1A). CGP 55845 also strongly reduced the inhibitory effect of baclofen on release evoked by 250 mM sucrose [Fig. 2C1–C2, 250 mM control, 1.09(±0.26) nC compared with + baclofen, 0.97(±0.22) nC; n=11, P<0.05]. The decrease in CGP 55845 was significantly smaller than the effect of baclofen seen in absence of the GABA_BR antagonist (Fig. 2D, P<0.001, unpaired two-sided Student's t-test), confirming that the baclofen-induced increase of the energy barrier for vesicle fusion is specifically mediated by activation of GABA_BRs.

Furthermore, we found that the peak vesicular release rate was significantly reduced in the presence of baclofen for both 500 and 250 mM sucrose applications [Fig. 3A,B, values normalized to 500 mM control, 500 mM sucrose + baclofen, 80.5(±3.9)%, n=17, P<0.0001; 250 mM control, 13.1(±1.1)% and 250 mM sucrose + baclofen, 7.0(±0.6)%, n=17, P<0.0001]. In addition, the onset kinetics were significantly slower upon GABA_BR activation [Fig. 3C1, 500 mM control, 0.52(±0.01) seconds compared with + baclofen, 0.64(±0.02) seconds, n=17, P<0.0001; 250 mM control, 0.91(±0.06) seconds compared with + baclofen: 1.2(±0.10) seconds, n=17, P<0.001]. In line with these results, the decay kinetics of the sucrose-evoked current transients were slower for 500 mM sucrose [Fig. 3C1, 500 mM control 0.42(±0.03) seconds + baclofen 0.51(±0.02) seconds; n=17, P<0.001]. As expected, co-application of the GABA_BR antagonist CGP 55845 blocked the decelerating effect of baclofen on hypertonicity-evoked release for both 500 and 250 mM sucrose (supplementary material Fig. S1C). In summary, these findings indicate that GABA_BR signalling directly affects vesicle release, independently of VDCCs. As the RRP size is unaltered, it rather seems that the fusion process itself is affected.

The above data strongly argue for a scenario in which GABA_BR activation affects the release machinery in a way that increases the energy demand for overcoming the fusion barrier for vesicle release. Which specific components of the release apparatus might be affected by GABA_BR activation? Considering the literature, a probable candidate would be a member of the proteins forming the SNARE complex, specifically SNAP-25 (Blackmer et al., 2001; Gerachshenko et al., 2005). We incubated cultures with BoNT-A, which cleaves nine C-terminal amino acids from SNAP-25 and could thereby block SNAP-25-dependent effects of baclofen. To achieve full cleavage of the protein in neuronal cell culture, a 48-hour incubation with 500 ng/ml BoNT-A was required, as observed by western blotting (Fig. 4A). Cleavage of SNAP-25 in autaptic neurons resulted in complete silencing of transmission with respect to both AP-evoked and spontaneous release (Fig. 4B,C). Moreover, incubation with BoNT-A also prevented vesicle release with hyperosmotic solutions (Fig. 4B). In order to test the possible block of GABA_BR-mediated reduction of vesicle release by SNAP-25 cleavage, we used 2 μM ionomycin to restore release by introducing artificial Ca²⁺-permeable pores in BoNT-A-treated cultures (Capogna et al., 1997). Application of ionomycin led to a steady inward current with multiple superimposed release events (Fig. 4C). All ionomycin-induced currents were completely blocked by 10 μM NBQX [an 2-amino-3-(5-methyl-3-oxo-1,2-oxazol-4-yl)propanoic acid (AMPA)-receptor antagonist], confirming that this procedure evoked no other conductances except for AMPA-receptor-mediated currents (Fig. 4C, Fig. 5A). We calculated the total charge of release over 4 seconds before, during and after baclofen application and found that baclofen reduced the ionomycin-evoked charge by 27% in control cultures [Fig. 4D,E, untreated ionomycin + baclofen, 73.4(±6.5)% of control; n=11, P<0.05] and by 38% in BoNT-A treated cultures [Fig. 4D,E, BoNT-A-treated ionomycin + baclofen, 61.5(±5.7)% of control; n=9, P<0.05]. This effect was not significantly different between the two conditions (P=0.2). Therefore, we conclude that GABA_BR activation increases the barrier for vesicle fusion independently of SNAP-25. This set of experiments was performed in the continuous presence of SCH 23390 (20 μM) in order to block postsynaptic GIRK channels activated by GABA_BRs (supplementary material Fig. S2). The effectiveness of SCH 23390 was verified in all experiments. In another series of experiments, we applied baclofen in the presence and absence of 10 μM NBQX instead of SCH 23390. This enabled us to subtract the baclofen-induced GIRK channel currents from the ionomycin experiments (Fig. 5A–C). In addition, baclofen reduced the ionomycin-induced transmitter release in control [Fig. 5B,D, 66.6(±7.2)%, n=9, P<0.001], as well as in BoNT-A-treated cultures [Fig. 5C,D: 75.6(±10.8)%, n=10, P<0.05], further corroborating our evidence for a SNAP-25-independent mode of action for GABA_BRs.

In the present study, we investigated the effects of GABA_BR activation on evoked and spontaneous release from glutamatergic hippocampal neurons and found that, in addition to the VDCC-mediated effects, the activation of presynaptic GABA_BRs leads to a substantial increase in the energy barrier for vesicle fusion. We furthermore demonstrate that, in contrast to other synapses studied previously, this modulation of the release machinery is independent of the C-terminus of SNAP-25.

What are the mechanistic consequences of a release modulation downstream of VDCCs? We addressed this question by challenging autaptic neurons with hypertonic solutions. Similar to high frequency ‘trains’ of APs, applications of 500 mM sucrose allow to estimate the number of vesicles in the RRP (Rosenmund and Stevens, 1996; Stevens and Williams, 2007; Moulder and Mennerick, 2005). In addition, hypertonic solutions have the advantage of providing a Ca²⁺-independent stimulus to induce transmitter release, which is indispensable when investigating VDCC-independent inhibition of transmitter release. Hypertonic solutions provide energy to vesicles primed for fusion, comparable to the effect of Ca²⁺-binding to synaptotagmin. When the tonicity at synapses increases, more primed vesicles reach the energy threshold for fusion. If the energy threshold for primed vesicles is increased by a G_i or G_o-dependent mechanism, responses to subsaturating tonicities, such as 250 mM added sucrose, will decrease in size relative to the maximal response induced by 500 mM sucrose. In addition, the onset of hypertonically induced transmitter release will be delayed and the maximal release rate will decrease. We found that this was indeed the case, as GABA_BR activation substantially slows the kinetics of sucrose-evoked fusion and reduces the total number of vesicles released by intermediate hypertonic stimuli. With similar methods, it was previously possible to identify mutations that change the energy barrier for fusion independently of the priming processes further upstream in the presynaptic vesicle cycle. For example, the gain-of-function mutation H567K in the vesicle priming factor Munc13-1 lowers the energy barrier for vesicle fusion, resulting in an increased sensitivity to sucrose challenge and increased spontaneous vesicle release, whereas it does not affect the RRP size (Basu et al., 2007). Conversely, complexin-deficient synapses have a normal RRP size, but show a comparable increase of the energy barrier and reduction in spontaneous release (Xue et al., 2010), as seen for GABA_BR activation in the current work. So, our results establish a role for a direct modulatory influence on GABA_BRs for the vesicular release machinery in the mammalian synapse.

The C-terminus of the SNARE complex member SNAP-25 has been suggested to be the target for direct modulation of vesicular fusion by GPCRs, as it has been found to be directly affected, through G_βγ-subunits, in serotonin-mediated modulation in the spinal cord of lampreys (Blackmer et al., 2001; Gerachshenko et al., 2005). A similar mechanism has been proposed for noradrenaline (norepinephrine)-mediated modulation at glutamatergic fibres onto the central amygdala in rats, also arguing for SNAP-25 as a direct target of metabotropic signalling at mammalian synapses (Delaney et al., 2007). In the case of the lamprey spinal cord, synapse direct modulation was also reported to involve alterations in the opening of the vesicular fusion pore (Photowala et al., 2006; Schwartz et al., 2007). Mechanistically these studies suggested that the interaction of G-protein βγ-subunits with the C-terminus of SNAP-25 competes with synaptotagmin for the binding site to the SNARE complex, thereby hindering tight coupling of the Ca²⁺ sensor to the vesicle (Blackmer et al., 2005). On the basis of three lines of evidence, we rather exclude the SNAP-25 interaction site with synaptotagmin as a target for G-protein βγ-subunit-mediated direct inhibition of neurotransmitter release by GABA_BRs. First, we find a reduction of mEPSC frequency by GABA_BR activation in Ca²⁺-free extracellular solution, arguing against the involvement of the Ca²⁺ sensor. Second, we observe that GABA_BRs cause a reduction in sucrose-evoked transmitter release that is independent of Ca²⁺ (Rosenmund and Stevens, 1996) and synaptotagmin-1 (Xu et al., 2009). Finally, we show that cleavage of SNAP-25 by BoNT-A does not abolish the inhibitory effect of GABA_BRs on ionomycin-evoked release. A possible explanation for the discrepancy between our data and studies that did show an involvement of SNAP-25 is that metabotropic receptors might in fact utilize different pathways at distinct synapses and thereby possess diverse targets for modulation. The completeness of the BoNT-A block might also affect the interpretation of results. We have taken great care to achieve complete cleavage of SNAP-25, necessitating active restoration of release by agents such as ionomycin (Capogna et al., 1997). Therefore, we can exclude the involvement of the SNAP-25 C-terminus in direct inhibition by GABA_BRs at hippocampal synapses. However, our experiments do indicate a GABA_BR-mediated direct effect on the core release machinery that is involved in the very late steps of exocytosis. A study in the calyx of Held has demonstrated that GABA_BR activation can directly modulate vesicle priming, leading to a slower refilling of the RRP. This effect, however, was observed only after prolonged stimulation paradigms, and it involved a cAMP-dependent signalling cascade (Sakaba and Neher, 2003). Because we did not detect any changes in the RRP size, the docking and priming steps seem unimpaired by GABA_BR activation during low frequency activity in our preparation. So, direct inhibition of vesicle fusion acts after priming, at a very late step of transmitter release.

Here, we establish a role for a direct modulation of the release machinery during GABAergic inhibition of transmission at glutamatergic synapses, which adds to our understanding of metabotropic signalling. What functional relevance could be attributed to the increased energy barrier seen in our study? Inhibiting release while only slightly reducing Ca²⁺ influx might be advantageous in situations of higher frequency activity, during which a substantial Ca²⁺ influx is required for accelerated replenishing of the RRP (Wang and Kaczmarek, 1998). The exact target of the release machinery that is responsible for this modulation remains to be determined and will be the focus of future studies.

All experiments were performed according to the rules of Berlin authorities and the animal welfare committee of the Charité Berlin. For field potential, whole-cell and Ca²⁺-imaging experiments, sagittal slices (300 μm) were prepared from C57/BL6 mice at postnatal day 19–29. Animals were briefly anesthetized with isofluorane and decapitated. Brains were rapidly removed and chilled in cold solutions containing: 87 mM NaCl, 26 mM NaHCO₃, 75 mM sucrose, 25 mM glucose, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 0.5 mM CaCl₂ and 7 mM MgCl₂, saturated with 95% O₂ and 5% CO₂. Slices were cut on a vibratome (VT 1200S, Leica) and incubated at 34°C for 30 minutes. They were transferred into the recording solution containing: 119 mM NaCl, 26 mM NaHCO₃, 10 mM glucose, 2.5 mM KCl, 1 mM NaH₂PO₄, 2.5 mM CaCl₂ and 1.3 mM MgCl₂, equilibrated with 95% O₂ and 5% CO₂ at room temperature.

For photodiode Ca²⁺ measurements (Gundlfinger et al., 2007; Regehr and Tank, 1991), slices were placed under an upright microscope (BX51WI, Olympus) equipped with an Olympus LumPlan FI 60× 0.9 NA water-immersion objective. Schaffer collaterals were locally labelled with the low-affinity Ca²⁺ indicator Magnesium Green AM (100 μM, Invitrogen) dissolved in 20% Pluronic/DMSO. Epifluorescence was measured outside of the loading spot at least 30 minutes after labelling. Signals were low-pass filtered at 1 kHz, digitized at 5 kHz and recorded with IGOR Pro (WaveMetrics). AP-triggered change in fluorescence intensity (ΔF) relative to the baseline intensity of fluorescence (F) was calculated as ΔF/F.

Field EPSPs (fEPSPs) were recorded with low-resistance pipettes filled with external solution in the stratum radiatum of CA1 using a Multiclamp 700A (Molecular Devices). Fibres were stimulated at 0.1 Hz. Data were filtered at 2 kHz and digitized at 5 kHz and recorded with IGOR. Extracellular CaCl₂ was replaced by equimolar concentrations of MgCl₂ to keep the total concentration of divalent cations constant.

Autaptic cultures were prepared as described previously (Bekkers and Stevens, 1991; Pyott and Rosenmund, 2002). For astrocyte precultures, cortices of newborn mice (postnatal day zero, P0) were digested with trypsin. Cells were grown in T75 flasks in BME medium (with 10% fetal calf serum, 1 mM Glutamax, 0.2% penicillin-streptomycin, 10 mM HEPES, 5 mM glucose and 2.5 μg/ml insulin; Invitrogen and Sigma) for 1 week. Growth permissive 200-μm spots of a 1:4 collagen and poly-D-lysine mixture were printed, using a custom-made stamp, on agarose-coated coverslips in six-well plates. 2×10⁴ astrocytes from precultures were seeded per well and grown until fully covering the microislands. Before plating hippocampal neurons (3×10³ cells per well), the medium was changed to Neurobasal A (with 2% B27, 0.2% penicillin-streptomycin; Invitrogen). To obtain neurons, P0 mouse hippocampi were digested using papain (Worthington, 20 units/ml, 1 mM L-cysteine, 0.5 mM EDTA in EBSS) for 60 minutes at 37°C.

Whole-cell voltage clamp recordings were performed with Axopatch 200A or Multiclamp 700B amplifiers using borosilicate electrodes (2–3 mΩ tip resistance) filled with: 146 mM K-gluconate, 17.8 mM Hepes, 1 mM EGTA, 4 mM Mg-ATP, 0.3 mM Na-GTP, 12 mM creatine phosphate and 50 units per ml creatine-phosphokinase (pH adjusted to 7.3 with KOH). Data were filtered at 2–5 kHz and digitized at 5–10 kHz. Fast applications of hypertonic solutions were performed using procedures described previously (Pyott and Rosenmund, 2002; Rosenmund et al., 1995). Sucrose was dissolved in extracellular solution containing: 140 mM NaCl, 2.4 mM KCl, 10 mM HEPES, 10 mM glucose, 2 mM CaCl₂ and 4 mM MgCl₂ (pH adjusted to 7.3 with NaOH, 300 mOsm).

To determine the efficacy of Botulinum neurotoxin A (BoNT-A), protein lysates of toxin-treated neuronal cultures were analyzed by western blotting using a monoclonal antibody directed against SNAP-25 (Synaptic Systems, Goettingen, Germany).

Data are given as means ± standard error. Significance was determined using paired Student's t-test or single-factor ANOVA with Bonferroni's post-hoc test where appropriate. The number of vesicles in the RRP was calculated by normalizing the charge of the 500 mM sucrose transient by the mean mEPSC charge.

We thank Anke Schönherr for excellent technical assistance. We would also like to thank Sarah Shoichet for help preparing the manuscript. D.S. is supported by grants from the DFG (SFB 618, SFB 665, GRK1123, EXC257), the Einstein Foundation and the BMBF (Bernstein Focus, Bernstein Centre for Computational Neuroscience); C.R. is supported by the DFG (EXC257).