RESUMEN
Pronounced differences in neurotransmitter release from a given presynaptic neuron, depending on the synaptic target, are among the most intriguing features of cortical networks. Hippocampal pyramidal cells (PCs) release glutamate with low probability to somatostatin expressing oriens-lacunosum-moleculare (O-LM) interneurons (INs), and the postsynaptic responses show robust short-term facilitation, whereas the release from the same presynaptic axons onto fast-spiking INs (FSINs) is ~10-fold higher and the excitatory postsynaptic currents (EPSCs) display depression. The mechanisms underlying these vastly different synaptic behaviors have not been conclusively identified. Here, we applied a combined functional, pharmacological, and modeling approach to address whether the main difference lies in the action potential-evoked fusion or else in upstream priming processes of synaptic vesicles (SVs). A sequential two-step SV priming model was fitted to the peak amplitudes of unitary EPSCs recorded in response to complex trains of presynaptic stimuli in acute hippocampal slices of adult mice. At PC-FSIN connections, the fusion probability (Pfusion) of well-primed SVs is 0.6, and 44% of docked SVs are in a fusion-competent state. At PC-O-LM synapses, Pfusion is only 40% lower (0.36), whereas the fraction of well-primed SVs is 6.5-fold smaller. Pharmacological enhancement of fusion by 4-AP and priming by PDBU was recaptured by the model with a selective increase of Pfusion and the fraction of well-primed SVs, respectively. Our results demonstrate that the low fidelity of transmission at PC-O-LM synapses can be explained by a low occupancy of the release sites by well-primed SVs.
Asunto(s)
Neurotransmisores , Vesículas Sinápticas , Animales , Vesículas Sinápticas/metabolismo , Vesículas Sinápticas/fisiología , Ratones , Neurotransmisores/metabolismo , Hipocampo/metabolismo , Hipocampo/fisiología , Potenciales Postsinápticos Excitadores/fisiología , Transmisión Sináptica/fisiología , Interneuronas/metabolismo , Interneuronas/fisiología , Células Piramidales/metabolismo , Células Piramidales/fisiología , Sinapsis/metabolismo , Sinapsis/fisiología , Modelos NeurológicosRESUMEN
A stunning example of synaptic diversity is the postsynaptic target cell-type-dependent difference in synaptic efficacy in cortical networks. Here, we show that CA1 pyramidal cell (PC) to fast spiking interneuron (FSIN) connections have 10-fold larger release probability (Pv) than those on oriens lacunosum-moleculare (O-LM) interneurons. Freeze-fracture immunolabeling revealed that different nano-topologies and coupling distances between Ca2+ channels and release sites (RSs) are not responsible for the distinct Pv. Although [Ca2+] transients are 40% larger in FSINs innervating boutons, when [Ca2+] entry is matched in the two bouton populations, EPSCs in O-LM cells are still 7-fold smaller. However, application of a phorbol ester analog resulted in a â¼2.5-fold larger augmentation at PC - O-LM compared to PC - FSIN synapses, suggesting incomplete docking or priming of vesicles. Similar densities of docked vesicles rule out distinct RS occupancies and demonstrate that incompletely primed, but docked, vesicles limit the output of PC - O-LM synapses.
Asunto(s)
Hipocampo , Vesículas Sinápticas , Hipocampo/fisiología , Sinapsis/fisiología , Interneuronas/fisiología , ProbabilidadRESUMEN
The molecular mechanisms underlying the diversity of cortical glutamatergic synapses are still incompletely understood. Here, we tested the hypothesis that presynaptic active zones (AZs) are constructed from molecularly uniform, independent release sites (RSs), the number of which scales linearly with the AZ size. Paired recordings between hippocampal CA1 pyramidal cells and fast-spiking interneurons in acute slices from adult mice followed by quantal analysis demonstrate large variability in the number of RSs (N) at these connections. High-resolution molecular analysis of functionally characterized synapses reveals variability in the content of one of the key vesicle priming factors - Munc13-1 - in AZs that possess the same N. Replica immunolabeling also shows a threefold variability in the total Munc13-1 content of AZs of identical size and a fourfold variability in the size and density of Munc13-1 clusters within the AZs. Our results provide evidence for quantitative molecular heterogeneity of RSs and support a model in which the AZ is built up from variable numbers of molecularly heterogeneous, but independent RSs.
Asunto(s)
Proteínas del Tejido Nervioso/metabolismo , Terminales Presinápticos/metabolismo , Animales , Región CA1 Hipocampal/metabolismo , Región CA1 Hipocampal/fisiología , Electrofisiología , Femenino , Técnica del Anticuerpo Fluorescente , Técnica de Fractura por Congelación , Masculino , Ratones , Ratones Endogámicos C57BL , Proteínas del Tejido Nervioso/fisiología , Terminales Presinápticos/fisiología , Sinapsis/metabolismo , Sinapsis/fisiologíaRESUMEN
Selective distribution of proteins in presynaptic active zones (AZs) is a prerequisite for generating postsynaptic target cell type-specific differences in presynaptic vesicle release probability (Pv) and short-term plasticity, a characteristic feature of cortical pyramidal cells (PCs). In the hippocampus of rodents, somatostatin and mGluR1α expressing interneurons (mGluR1α+ INs) receive small, facilitating excitatory postsynaptic currents (EPSCs) from PCs and express Elfn1 that trans-synaptically recruits mGluR7 into the presynaptic AZ of PC axons. Here we show that Elfn1 also has a role in the selective recruitment of Munc13-2, a synaptic vesicle priming and docking protein, to PC AZs that innervate mGluR1α+ INs. In Elfn1 knock-out mice, unitary EPSCs (uEPSCs) in mGluR1α+ INs have threefold larger amplitudes with less pronounced short-term facilitation, which might be the consequence of the loss of either mGluR7 or Munc13-2 or both. Conditional genetic deletion of Munc13-2 from CA1 PCs results in the loss of Munc13-2, but not mGluR7 from the AZs, and has no effect on the amplitude of uEPSCs and leaves the characteristic short-term facilitation intact at PC to mGluR1α+ IN connection. Our results demonstrate that Munc13-1 alone is capable of imposing low Pv at PC to mGluR1α+ IN synapses and Munc13-2 has yet an unknown role in this synapse.
RESUMEN
Elucidating the molecular mechanisms underlying the functional diversity of synapses requires a high-resolution, sensitive, diffusion-free, quantitative localization method that allows the determination of many proteins in functionally characterized individual synapses. Array tomography permits the quantitative analysis of single synapses but has limited sensitivity, and its application to functionally characterized synapses is challenging. Here, we aim to overcome these limitations by searching the parameter space of different fixation, resin, embedding, etching, retrieval, and elution conditions. Our optimizations reveal that etching epoxy-resin-embedded ultrathin sections with Na-ethanolate and treating them with SDS dramatically increase the labeling efficiency of synaptic proteins. We also demonstrate that this method is ideal for the molecular characterization of individual synapses following paired recordings, two-photon [Ca2+] or glutamate-sensor (iGluSnFR) imaging. This method fills a missing gap in the toolbox of molecular and cellular neuroscience, helping us to reveal how molecular heterogeneity leads to diversity in function.
Asunto(s)
Inmunohistoquímica/métodos , Sinapsis/metabolismo , Sinapsis/fisiología , Animales , Resinas Epoxi/química , Ácido Glutámico/química , Humanos , Microscopía Confocal/métodosRESUMEN
CCK-expressing interneurons (CCK+INs) are crucial for controlling hippocampal activity. We found two firing phenotypes of CCK+INs in rat hippocampal CA3 area; either possessing a previously undetected membrane potential-dependent firing or regular firing phenotype, due to different low-voltage-activated potassium currents. These different excitability properties destine the two types for distinct functions, because the former is essentially silenced during realistic 8-15 Hz oscillations. By contrast, the general intrinsic excitability, morphology and gene-profiles of the two types were surprisingly similar. Even the expression of Kv4.3 channels were comparable, despite evidences showing that Kv4.3-mediated currents underlie the distinct firing properties. Instead, the firing phenotypes were correlated with the presence of distinct isoforms of Kv4 auxiliary subunits (KChIP1 vs. KChIP4e and DPP6S). Our results reveal the underlying mechanisms of two previously unknown types of CCK+INs and demonstrate that alternative splicing of few genes, which may be viewed as a minor change in the cells' whole transcriptome, can determine cell-type identity.
Asunto(s)
Región CA3 Hipocampal/citología , Colecistoquinina/metabolismo , Interneuronas , Canales de Potasio Shal , Animales , Células Cultivadas , Interneuronas/química , Interneuronas/clasificación , Interneuronas/metabolismo , Potenciales de la Membrana/fisiología , Fenotipo , Subunidades de Proteína/química , Subunidades de Proteína/genética , Subunidades de Proteína/metabolismo , Ratas , Ratas Wistar , Canales de Potasio Shal/química , Canales de Potasio Shal/genética , Canales de Potasio Shal/metabolismo , Transcriptoma/genéticaRESUMEN
The nanoscale topographical arrangement of voltage-gated calcium channels (VGCC) and synaptic vesicles (SVs) determines synaptic strength and plasticity, but whether distinct spatial distributions underpin diversity of synaptic function is unknown. We performed single bouton Ca2+ imaging, Ca2+ chelator competition, immunogold electron microscopic (EM) localization of VGCCs and the active zone (AZ) protein Munc13-1, at two cerebellar synapses. Unexpectedly, we found that weak synapses exhibited 3-fold more VGCCs than strong synapses, while the coupling distance was 5-fold longer. Reaction-diffusion modeling could explain both functional and structural data with two strikingly different nanotopographical motifs: strong synapses are composed of SVs that are tightly coupled (â¼10 nm) to VGCC clusters, whereas at weak synapses VGCCs were excluded from the vicinity (â¼50 nm) of docked vesicles. The distinct VGCC-SV topographical motifs also confer differential sensitivity to neuromodulation. Thus, VGCC-SV arrangements are not canonical, and their diversity could underlie functional heterogeneity across CNS synapses.
Asunto(s)
Canales de Calcio/metabolismo , Sinapsis/metabolismo , Transmisión Sináptica/fisiología , Vesículas Sinápticas/metabolismo , Animales , Ratones , Ratones Endogámicos BALB C , Ratones Endogámicos C57BLRESUMEN
KEY POINTS: The amplitude of unitary, single action potential-evoked [Ca2+ ] transients negatively correlates with GCaMP6f expression, but displays large variability among hippocampal pyramidal cells with similarly low expression levels. The summation of fluorescence signals is frequency-dependent, supralinear and also shows remarkable cell-to-cell variability. The main source of spike inference error is variability in the peak amplitude, and not in the decay or supralinearity. We developed two procedures to estimate the peak amplitudes of unitary [Ca2+ ] transients and show that spike inference performed with MLspike using these unitary amplitude estimates in weakly GCaMP6f-expressing cells results in error rates of â¼5%. ABSTRACT: Investigating neuronal activity using genetically encoded Ca2+ indicators in behaving animals is hampered by inaccuracies in spike inference from fluorescent tracers. Here we combine two-photon [Ca2+ ] imaging with cell-attached recordings, followed by post hoc determination of the expression level of GCaMP6f, to explore how it affects the amplitude, kinetics and temporal summation of somatic [Ca2+ ] transients in mouse hippocampal pyramidal cells (PCs). The amplitude of unitary [Ca2+ ] transients (evoked by a single action potential) negatively correlates with GCaMP6f expression, but displays large variability even among PCs with similarly low expression levels. The summation of fluorescence signals is frequency-dependent, supralinear and also shows remarkable cell-to-cell variability. We performed experimental data-based simulations and found that spike inference error rates using MLspike depend strongly on unitary peak amplitudes and GCaMP6f expression levels. We provide simple methods for estimating the unitary [Ca2+ ] transients in individual weakly GCaMP6f-expressing PCs, with which we achieve spike inference error rates of â¼5%.
Asunto(s)
Calcio/fisiología , Hipocampo/fisiología , Proteínas Luminiscentes/fisiología , Células Piramidales/fisiología , Animales , Señalización del Calcio , Masculino , RatonesRESUMEN
Research over the past half a century has revealed remarkable diversity among chemical synapses of the CNS. The structural, functional and molecular diversity of synapses was mainly concluded from studying different synapses in distinct brain regions and preparations. It is not surprising that synapses made by molecularly distinct pre-synaptic and post-synaptic cells display different morphological and functional properties with distinct underlying molecular mechanisms. However, synapses made by a single presynaptic cell onto distinct types of postsynaptic cells, or distinct presynaptic inputs onto a single postsynaptic cell, also show remarkable heterogeneity. Here, by reviewing recent experiments, I suggest that robust functional diversity can be achieved by building synapses from the same molecules, but using different numbers, densities and nanoscale arrangements.
Asunto(s)
Encéfalo/citología , Neurotransmisores/metabolismo , Terminales Presinápticos/fisiología , Sinapsis/fisiología , Animales , Calcio/metabolismo , Terminales Presinápticos/ultraestructura , Membranas Sinápticas/fisiologíaRESUMEN
Nanoscale distribution of molecules within small subcellular compartments of neurons critically influences their functional roles. Although, numerous ways of analyzing the spatial arrangement of proteins have been described, a thorough comparison of their effectiveness is missing. Here we present an open source software, GoldExt, with a plethora of measures for quantification of the nanoscale distribution of proteins in subcellular compartments (e.g. synapses) of nerve cells. First, we compared the ability of five different measures to distinguish artificial uniform and clustered patterns from random point patterns. Then, the performance of a set of clustering algorithms was evaluated on simulated datasets with predefined number of clusters. Finally, we applied the best performing methods to experimental data, and analyzed the nanoscale distribution of different pre- and postsynaptic proteins, revealing random, uniform and clustered sub-synaptic distribution patterns. Our results reveal that application of a single measure is sufficient to distinguish between different distributions.
Asunto(s)
Algoritmos , Hipocampo/metabolismo , Proteínas del Tejido Nervioso/metabolismo , Neuronas/metabolismo , Programas Informáticos , Sinapsis/metabolismo , Animales , HumanosRESUMEN
Target cell type-dependent differences in presynaptic release probability (Pr ) and short-term plasticity are intriguing features of cortical microcircuits that increase the computational power of neuronal networks. Here, we tested the hypothesis that different voltage-gated Ca2+ channel densities in presynaptic active zones (AZs) underlie different Pr values. Two-photon Ca2+ imaging, triple immunofluorescent labeling, and 3D electron microscopic (EM) reconstruction of rat CA3 pyramidal cell axon terminals revealed â¼1.7-1.9 times higher Ca2+ inflow per AZ area in high Pr boutons synapsing onto parvalbumin-positive interneurons (INs) than in low Pr boutons synapsing onto mGluR1α-positive INs. EM replica immunogold labeling, however, demonstrated only 1.15 times larger Cav2.1 and Cav2.2 subunit densities in high Pr AZs. Our results indicate target cell type-specific modulation of voltage-gated Ca2+ channel function or different subunit composition as possible mechanisms underlying the functional differences. In addition, high Pr synapses are also characterized by a higher density of docked vesicles, suggesting that a concerted action of these mechanisms underlies the functional differences.SIGNIFICANCE STATEMENT Target cell type-dependent variability in presynaptic properties is an intriguing feature of cortical synapses. When a single cortical pyramidal cell establishes a synapse onto a somatostatin-expressing interneuron (IN), the synapse releases glutamate with low probability, whereas the next bouton of the same axon has high release probability when its postsynaptic target is a parvalbumin-expressing IN. Here, we used combined molecular, imaging, and anatomical approaches to investigate the mechanisms underlying these differences. Our functional experiments implied an approximately twofold larger Ca2+ channel density in high release probability boutons, whereas freeze-fracture immunolocalization demonstrated only a 15% difference in Ca2+ channel subunit densities. Our results point toward a postsynaptic target cell type-dependent regulation of Ca2+ channel function or different subunit composition as the underlying mechanism.
Asunto(s)
Canales de Calcio/metabolismo , Ácido Glutámico/metabolismo , Hipocampo/citología , Plasticidad Neuronal/fisiología , Neuronas/metabolismo , Terminales Presinápticos/metabolismo , Probabilidad , Potenciales de Acción/fisiología , Animales , Animales Recién Nacidos , Axones/metabolismo , Calcio/metabolismo , Canales de Calcio/ultraestructura , Ácido Glutámico/clasificación , Técnicas In Vitro , Lisina/análogos & derivados , Lisina/metabolismo , Masculino , Proteínas del Tejido Nervioso/metabolismo , Proteínas del Tejido Nervioso/ultraestructura , Parvalbúminas/metabolismo , Terminales Presinápticos/ultraestructura , Ratas , Ratas Wistar , Receptores de Glutamato Metabotrópico/metabolismo , Receptores de Glutamato Metabotrópico/ultraestructura , Ácido gamma-Aminobutírico/metabolismoRESUMEN
Classic theories link cognitive abilities to synaptic properties and human-specific biophysical features of synapses might contribute to the unparalleled performance of the human cerebral cortex. Paired recordings and multiple probability fluctuation analysis revealed similar quantal sizes, but 4-times more functional release sites in human pyramidal cell to fast-spiking interneuron connections compared to rats. These connections were mediated on average by three synaptic contacts in both species. Each presynaptic active zone (AZ) contains 6.2 release sites in human, but only 1.6 in rats. Electron microscopy (EM) and EM tomography showed that an AZ harbors 4 docked vesicles in human, but only a single one in rats. Consequently, a Katz's functional release site occupies ~0.012 µm(2) in the human presynaptic AZ and ~0.025 µm(2) in the rat. Our results reveal a robust difference in the biophysical properties of a well-defined synaptic connection of the cortical microcircuit of human and rodents.
Asunto(s)
Interneuronas/fisiología , Terminales Presinápticos/metabolismo , Células Piramidales/fisiología , Vesículas Sinápticas/metabolismo , Animales , Fenómenos Biofísicos , Tomografía con Microscopio Electrónico , Humanos , Microscopía Electrónica de Transmisión , Terminales Presinápticos/ultraestructura , Ratas , Vesículas Sinápticas/ultraestructuraRESUMEN
Hippocampal pyramidal cells (PCs) express many GABAAR subunit types and receive GABAergic inputs from distinct interneurons. Previous experiments revealed input-specific differences in α1 and α2 subunit densities in perisomatic synapses, suggesting distinct IPSC decay kinetics. However, IPSC decays evoked by axo-axonic, parvalbumin- or cholecystokinin-expressing basket cells were found to be similar. Using replica immunogold labeling, here we show that all CA1 PC somatic and AIS synapses contain the α1, α2, ß1, ß2, ß3 and γ2 subunits. In CA3 PCs, 90% of the perisomatic synapses are immunopositive for the α1 subunit and all synapses are positive for the remaining five subunits. Somatic synapses form unimodal distributions based on their immunoreactivity for these subunits. The α2 subunit densities in somatic synapses facing Cav2.1 (i.e. parvalbumin) or Cav2.2 (cholecystokinin) positive presynaptic active zones are comparable. We conclude that perisomatic synapses made by three distinct interneuron types have similar GABAA receptor subunit content.
Asunto(s)
Segmento Inicial del Axón/química , Hipocampo/citología , Interneuronas/química , Subunidades de Proteína/análisis , Células Piramidales/química , Receptores de GABA/química , Sinapsis/química , Animales , Inmunohistoquímica , Masculino , Ratas WistarRESUMEN
Encoding continuous sensory variables requires sustained synaptic signalling. At several sensory synapses, rapid vesicle supply is achieved via highly mobile vesicles and specialized ribbon structures, but how this is achieved at central synapses without ribbons is unclear. Here we examine vesicle mobility at excitatory cerebellar mossy fibre synapses which sustain transmission over a broad frequency bandwidth. Fluorescent recovery after photobleaching in slices from VGLUT1(Venus) knock-in mice reveal 75% of VGLUT1-containing vesicles have a high mobility, comparable to that at ribbon synapses. Experimentally constrained models establish hydrodynamic interactions and vesicle collisions are major determinants of vesicle mobility in crowded presynaptic terminals. Moreover, models incorporating 3D reconstructions of vesicle clouds near active zones (AZs) predict the measured releasable pool size and replenishment rate from the reserve pool. They also show that while vesicle reloading at AZs is not diffusion-limited at the onset of release, diffusion limits vesicle reloading during sustained high-frequency signalling.
Asunto(s)
Fibras Nerviosas/fisiología , Sinapsis/metabolismo , Vesículas Sinápticas/metabolismo , Animales , Ratones , Movimiento (Física) , Imagen ÓpticaRESUMEN
Cerebellar Golgi cells (GoCs) efficiently control the spiking activity of granule cells through GABAA receptor-mediated tonic and phasic inhibition. Recent experiments provided compelling evidence for the extensive interconnection of GoCs through electrical synapses, but their chemical inhibitory synaptic inputs are debated. Here, we investigated the GABAergic synaptic inputs of GoCs using in vitro electrophysiology and quantitative light microscopy (LM) and electron microscopy (EM). We characterized GABAA receptor-mediated IPSCs in GoCs and Lugaro cells (LuCs), and found that IPSCs in GoCs have lower frequencies, smaller amplitudes, and much slower decay kinetics. Pharmacological and LM immunolocalization experiments revealed that GoCs express α3, whereas LuCs express α1 subunit-containing GABAA receptors. The selective expression and clustered distribution of the α3 subunit in GoCs allowed the quantitative analysis of GABAergic synapses on their dendrites in the molecular layer (ML). EM and LM experiments in rats, and wild-type and GlyT2-GFP transgenic mice revealed that only one third of axon terminals establishing GABAergic synapses on GoC dendrites contain GlyT2, ruling out LuCs, globular cells, and any noncortical glycinergic inputs as major inhibitory sources. We also show that axon terminals of stellate/basket cells very rarely innervate GlyT2-GFP-expressing GoCs, indicating that only a minority of the inhibitory inputs to GoCs in the ML originates from local interneurons, and the majority of their inhibitory inputs exclusively releases GABA.
Asunto(s)
Cerebelo/citología , Neuronas GABAérgicas/citología , Interneuronas/citología , Inhibición Neural , Animales , Axones/efectos de los fármacos , Axones/metabolismo , Cerebelo/efectos de los fármacos , Cerebelo/fisiología , Dendritas/efectos de los fármacos , Dendritas/metabolismo , Femenino , Neuronas GABAérgicas/efectos de los fármacos , Neuronas GABAérgicas/fisiología , Interneuronas/efectos de los fármacos , Interneuronas/fisiología , Masculino , Ratones Endogámicos C57BL , Ratones Transgénicos , Inhibición Neural/fisiología , Vías Nerviosas/citología , Vías Nerviosas/efectos de los fármacos , Vías Nerviosas/fisiología , Ratas Wistar , Receptores de GABA-A/metabolismo , Ácido gamma-Aminobutírico/metabolismoRESUMEN
The strength and variability of electrical synaptic connections between GABAergic interneurons are key determinants of spike synchrony within neuronal networks. However, little is known about how electrical coupling strength is determined due to the inaccessibility of gap junctions on the dendritic tree. We investigated the properties of gap junctions in cerebellar interneurons by combining paired somato-somatic and somato-dendritic recordings, anatomical reconstructions, immunohistochemistry, electron microscopy, and modeling. By fitting detailed compartmental models of Golgi cells to their somato-dendritic voltage responses, we determined their passive electrical properties and the mean gap junction conductance (0.9 nS). Connexin36 immunofluorescence and freeze-fracture replica immunogold labeling revealed a large variability in gap junction size and that only 18% of the 340 channels are open in each plaque. Our results establish that the number of gap junctions per connection is the main determinant of both the strength and variability in electrical coupling between Golgi cells.
Asunto(s)
Cerebelo/citología , Sinapsis Eléctricas/fisiología , Uniones Comunicantes/fisiología , Interneuronas/fisiología , Animales , Conexinas/metabolismo , Dendritas/fisiología , Femenino , Masculino , Ratones , Proteína delta-6 de Union ComunicanteRESUMEN
The release of GABA from cholecystokinin-containing interneurons is modulated by type-1 cannabinoid receptors (CB1). Here we tested the hypothesis that the strength of CB1-mediated modulation of GABA release is related to the CB1 content of axon terminals. Basket cell boutons have on average 78% higher CB1 content than those of dendritic-layer-innervating (DLI) cells, a consequence of larger bouton surface and higher CB1 density. The CB1 antagonist AM251 caused a 54% increase in action potential-evoked [Ca(2+)] in boutons of basket cells, but not in DLI cells. However, the effect of AM251 did not correlate with CB1 immunoreactivity of individual boutons. Moreover, a CB1 agonist decreased [Ca(2+)] in a cell type- and CB1-content-independent manner. Replica immunogold labelling demonstrated the colocalization of CB1 with the Cav2.2 Ca(2+) channel subunit. Our data suggest that only a subpopulation of CB1s, within nanometre distances from their target Cav2.2 channels, are responsible for endocannabinoid-mediated modulation of GABA release.