GABA_A receptors, gephyrin and homeostatic synaptic plasticity

Abstract

Homeostatic synaptic plasticity describes the changes in synapse gain and function that occur in response to global changes in neuronal activity to maintain the stability of neuronal networks. In this review, we argue that a coordinated regulation of excitatory and inhibitory synaptic transmission is essential for maintaining CNS function while allowing both global and local changes in synaptic strength and connectivity. Therefore, we postulate that homeostatic synaptic plasticity depends on signalling cascades regulating in parallel the efficacy of glutamatergic and GABAergic transmission. Since neurotransmitter receptors interact closely with scaffolding proteins in the postsynaptic density, this coordinated regulation of excitatory and inhibitory synaptic transmission probably involves posttranslational modifications of scaffolding proteins, which in turn modulate local synaptic function. Here we review the current state of knowledge on the regulation of GABA_A receptors and their main scaffolding protein gephyrin by posttranslational modifications; we outline future lines of research that might contribute to furthering our understanding of the molecular mechanisms regulating GABAergic synapse function and homeostatic plasticity.

Jean-Marc Fritschy studied biology at the University of Geneva and received his PhD in Neuroanatomy at the University of Lausanne, Switzerland. He stayed four years at the Dept of Neuroscience, Johns Hopkins University, Baltimore, MD, before joining the Institute of Pharmacology and Toxicology of the University of Zurich in 1991. He works on the functional organization of GABAergic neurotransmission, with a focus on the significance of GABA_A receptor molecular heterogeneity and his group is also interested in the mechanisms of GABAergic synapse formation and plasticity, on the role of GABA_A receptors in the regulation of adult neurogenesis, and on the identification of neuronal circuits mediating anxiety. He has also published numerous studies of the pathophysiology of temporal lobe epilepsy and is currently investigating the interactions between epilepsy and neuroinflammation. Since 2008 he has been co-Editor-in-Chief of European Journal of Neuroscience, the journal of the Federation of European Neuroscience Societies.

Introduction

Maintaining a balance between excitatory and inhibitory synaptic transmission is essential for long-term stability and function of neuronal networks. Global changes in network activity are counteracted by homeostatic compensation changing synaptic gain and adapting intrinsic neuronal properties. Likewise, local changes in the strength or number of excitatory synapses induce adaptations to inhibitory synapses with the aim of maintaining a level of network activity within preset limits (Turrigiano, 2007). This homeostatic plasticity is essential to ensure overall stability of neuronal networks while allowing local changes to synaptic strength and connections. Thus, even under pathological conditions, such as epileptic seizures, the strength of inhibitory transmission is enhanced in an effort to confine the overly increased excitatory drive (Nusser et al. 1998). Perturbations of synaptic homeostasis will cause profound alterations of brain function, and unsurprisingly, this has been linked with several pathological conditions, including epilepsy, anxiety and developmental neuropsychiatric disorders (Dani et al. 2005; Fritschy, 2008; Südhof, 2008). Of equal significance, the delayed therapeutic onset of psychoactive drugs, typically observed in patients treated for mood disorders, anxiety, or schizophrenia, might well reflect homeostatic adaptations of neuronal circuits to compensate for the chronic effects that these drugs have on synaptic transmission.

Homeostatic synaptic plasticity probably involves iterative synapse formation and removal, as well as dynamic regulation of synaptic strength, either presynaptically (e.g. regulation of transmitter release) and/or postsynaptically (e.g. posttranslational regulation of receptor function and trafficking; adaptation of downstream signal transduction; changes in synaptic structure) to counteract any global change in network activity (Kilman et al. 2002; Nelson & Turrigiano, 2008; Bannai et al. 2009). These changes are relayed by specific signalling cascades onto postsynaptic protein scaffolds, which serve as a crucial hub for holding functionally related proteins in close proximity, to produce a coordinated adaptation of synaptic strength and function.

So far, considerable knowledge has been gained about the regulation of glutamatergic synapses, whose postsynaptic density (PSD) can be isolated biochemically and visualized morphologically on dendritic spines (Boeckers, 2006). These synapses are characterized by a core scaffolding protein, PSD-95, which forms a complex assembly of proteins interacting via specific PDZ domains (Kim & Sheng, 2004). Studies of bidirectional synaptic plasticity (long-term potentiation and long-term depression), a hallmark of learning and memory, have uncovered rapid but long-lasting changes in glutamatergic synaptic function triggered by activity-dependent processes. These changes involve the activation of protein phosphorylation networks that regulate PSD-95 and signalling complexes to achieve the required adaptations of the glutamatergic synapses. For example, a recent study has revealed that PSD-95 phosphorylation by Ca²⁺–calmodulin-dependent protein kinase 2 (CaMKII) regulates its stability and trafficking and underlies changes in spine morphology that accompany LTP (Steiner et al. 2008).

By comparison, little is known about molecular mechanisms underlying compensatory adaptations at GABAergic synapses. GABA_A receptors (GABA_ARs), which mediate the bulk of fast synaptic inhibition in the CNS, are regulated by phosphorylation mechanisms, affecting both their functional properties and their cell surface mobility and trafficking (Jacob et al. 2008). Their synaptic distribution is intimately linked with that of gephyrin, the main scaffolding protein of inhibitory postsynaptic sites (Fritschy et al. 2008). However, the regulation of the GABAergic PSD has received little attention so far, notably because gephyrin has long been considered as a mere cytoskeletal anchoring protein, despite being a phospho-protein (Langosch et al. 1992) and a substrate for calpain (Kawasaki et al. 1997). Therefore, in analogy to PSD-95, studying the effects of gephyrin phosphorylation might reveal key aspects of GABAergic synapse regulation in homeostatic plasticity.

Mechanistically, changes in excitability involving glutamatergic synapses have to be paralleled by corresponding changes at GABAergic synapses to avoid hyperexcitation (or silencing) of the network. Here, we hypothesize that synaptic homeostasis depends on signalling cascades regulating in parallel the efficacy of glutamatergic and GABAergic transmission. According to this view, these signals might converge onto postsynaptic protein scaffolds and regulate synaptic function by means of posttranslational modifications of specific target proteins. Such convergence is apparent in the concerted changes in synaptic function produced by specific protein kinases and phosphatases. An alternative (or complementary view) is that synaptic homeostasis might depend also on multifunctional proteins that regulate intrinsic neuronal properties along with excitatory and inhibitory neurotransmission. In particular, given the importance of Cl⁻ gradients in determining GABA_AR function (Blaesse et al. 2009), a potential role of chloride transporters in synaptic homeostasis appears as an attractive possibility.

The aim of this review is to discuss evidence in support of this hypothesis of a concerted regulation of glutamatergic and GABAergic synapse strength in homeostatic synaptic plasticity.

GABA_A receptor structure and regulation

GABA_ARs belong to the family of ligand-gated ion channels and form multimeric GABA-gated channels selectively permeable to Cl⁻ and HCO₃⁻ ions. They are encoded by a large family of subunit genes (α1–6, β1–3, γ1–3, δ, ɛ, π, θ, ρ1–3), whose differential assembly into receptor complexes gives rise to multiple GABA_A receptor subtypes differing in functional and pharmacological properties (Sieghart & Ernst, 2005). This heterogeneity is a major facet of their regulation, since distinct GABA_A receptor subtypes are localized at different subcellular sites (e.g. synaptic/extrasynaptic) and are integrated in specific neuronal circuits subserving distinct brain functions and pharmacological roles (Fritschy & Brünig, 2003). While the γ2 subunit is well known for its role in postsynaptic clustering of GABA_ARs (Essrich et al. 1998; Li et al. 2005), the mechanisms that specify extrasynaptic GABA_A receptor localization, notably of α4- and a subpopulation of α5- and α6-GABA_ARs are not known. However, analysis of α1-subunit null mice suggested that the two pools of receptors are not interchangeable, as extrasynaptic α4-GABA_ARs did not substitute for missing synaptic α1-GABA_ARs, as shown in thalamic neurons (Peden et al. 2007). Therefore, it is unlikely that homeostatic compensations of synaptic function involve a major redistribution of synaptic and extrasynaptic GABA_AR subtypes.

The subunit heterogeneity of GABA_ARs also provides considerable room for subtype-specific regulation of receptor function and trafficking by posttranslational modification (Jacob et al. 2008; Arancibia-Cárcamo & Kittler, 2009). Thus, subunit-specific phosphorylation of GABA_AR by CaMKII, notably on the β2 or β3 subunit, produces major effects on their functional properties in a synapse-dependent manner (Houston et al. 2008). In turn, phosphorylation of the γ2 subunit regulates GABA_AR cell surface expression by modulating binding to AP2 and clathrin-dependent internalization (Kittler et al. 2008). The significance of the γ2 subunit for intracellular and membrane trafficking on GABA_ARs is also underscored by its palmitoylation (Keller et al. 2004; Rathenberg et al. 2004) and its interaction with calcium-modulating cyclophilin ligand, which is a determinant for the recycling of endocytosed GABA_ARs (Yuan et al. 2008). In the context of homeostatic synaptic plasticity, the multiple mechanisms involved in the regulation of GABA_A receptor expression and postsynaptic clustering provide a framework for differential activation by specific signalling pathways.

Gephyrin structure and regulation

Gephyrin, a cytoplasmic protein initially isolated as a constituent of purified glycine receptor (GlyR) preparations, is now considered as the ‘master’ scaffolding molecule of inhibitory synapses. In neurons, gephyrin is selectively clustered to the PSD of glycinergic and GABAergic synapses, contributing to GlyR and GABA_AR clustering (Fritschy et al. 2008). The selective localization of gephyrin and its capacity to auto-oligomerize and form postsynaptic clusters underscores its role as a scaffolding molecule. This idea was strengthened further once the crystal structure of the N-terminal G-domain and the C-terminal E-domain of gephyrin showed that they form trimers and dimers in solution, respectively. The G- and E-domains of gephyrin are interconnected by a linker region, the C-domain, which is sensitive to proteolytic degradation and its structure remains elusive. A leading model based on these partial crystallography results suggested that gephyrin forms a hexagonal lattice by G-domain trimerization and E-domain dimerization linking GlyR (or GABA_AR) in the membrane with the cytoskeleton (Sola et al. 2004). This model is attractive because it explains how a protein scaffold can emerge in the absence of protein–protein interaction domains equivalent to the PDZ-domains of PSD-95. However, auto-aggregation of gephyrin in a hexagonal lattice does not provide much room for dynamic regulation of the postsynaptic scaffold, a prerequisite of homeostatic plasticity. An alternative model, based on observations that native gephyrin forms trimers in solution, suggests that the gephyrin scaffold is formed by regulated aggregation of trimeric units (Fritschy et al. 2008). Accordingly, in glycinergic synapses, a gephyrin trimer would be linked to a single GlyR, with each E-domain binding to the intracellular loop of a β-subunit. In this model, a key role for the C-domain, which contains most of the interacting sites with other proteins, is proposed for the regulation of gephyrin aggregation. However, while several functionally diverse proteins have been identified as gephyrin-binding partners (Fritschy et al. 2008), it is unclear how they regulate gephyrin aggregation, and by extension, contribute to the plasticity of inhibitory postsynaptic sites. Therefore, it will be essential to identify and characterize the protein–protein interaction sites on the C-domain to understand how trimeric units assemble in a regular scaffold.

The model of trimeric gephyrin units as basic building blocks of gephyrin scaffolds is well supported by evidence from GlyR tracking studies that a large fraction of extrasynaptic GlyR is bound to gephyrin (Ehrensperger et al. 2007) and that their cell surface dynamics are regulated by gephyrin clustering properties (Calamai et al. 2009). Further evidence for this model stems from mutagenesis studies, in which a surface-exposed loop, named L2, containing residues differing between vertebrate and plant or bacterial gephyrin homologues was found to be essential for regulating gephyrin clustering (Lardi-Studler et al. 2007). Mutations in this loop result either in impaired gephyrin aggregation (L2B) or formation of supernumerary postsynaptic gephyrin clusters (L2C) in cultured neurons. The latter observation is particularly intriguing, because it implies that regulation of gephyrin clustering is directly related to synapse formation. Furthermore, it identifies specific residues that play a key role in gephyrin aggregation at postsynaptic sites.

GABA_A receptor–gephyrin interactions

In striking contrast to AMPA and NMDA receptors, whose interaction with PSD-95 has been extensively analysed in the context of synaptic plasticity and homeostasis, the interaction between GABA_AR and gephyrin remains largely elusive. Direct binding has long been considered unlikely, in the absence of conclusive biochemical evidence. However, analysis of the cell surface dynamics of GABA_AR revealed that gephyrin negatively regulates their diffusion coefficient (Jacob et al. 2005), allowing the formation of postsynaptic GABA_AR clusters, implicating at least an indirect interaction. Recently, a possible direct interaction between gephyrin and α2-GABA_AR has been reported and postulated to play a key role in the targeting of these receptors to the axon initial segment of pyramidal cells (Tretter et al. 2008). Numerous studies have attempted to clarify the significance of gephyrin–GABA_A receptor interaction by up- or down-regulation experiments. Thus, overexpression of gephyrin has little or no effect on the size and number of gephyrin clusters and GABAergic synapses, but causes a significant reduction in glutamatergic PSD size and NMDA receptor density, suggestive of heterotypic effects on excitatory synapses (Yu & De Blas, 2008). Conversely, down-regulation of gephyrin expression by gene targeting or gene silencing leads to the rapid disappearance of postsynaptic GABA_A receptor clustering and loss of IPSCs (Essrich et al. 1998; Yu et al. 2007), without affecting the number of GABA_ARs at the cell surface (Jacob et al. 2005), suggesting lateral diffusion to extrasynaptic sites. Despite these effects, there is near consensus that the formation of postsynaptic GABA_A receptor clusters is gephyrin independent in developing neurons (Danglot et al. 2003).

Counter-intuitively for a scaffolding protein, postsynaptic clustering of gephyrin is disrupted in the absence of GABA_ARs, as shown in vitro and in vivo (Essrich et al. 1998; Schweizer et al. 2003; Li et al. 2005; Kralic et al. 2006; Studer et al. 2006). This observation provides compelling arguments in favour of the view that gephyrin clustering is a regulated process, probably dependent on local synaptic activity, and might be a crucial mechanism to adjust the strength and gain of inhibitory transmission.

An avenue that has so far remained unexplored is the possibility that gephyrin phosphorylation might regulate GABA_A receptor binding and their postsynaptic localization or trafficking. Preliminary evidence is only available for GlyR, notably the demonstration that proline-directed phosphorylation of gephyrin can induce a conformational change favouring GlyR binding (Zita et al. 2007). Furthermore, since NMDA receptor activation reduces the surface mobility of GlyR in a Ca²⁺-dependent fashion, the authors of this study postulated that gephyrin or GlyR phosphorylation, possibly by PKC, might contribute to this effect (Lévi et al. 2008).

Regulation of postsynaptic scaffolds by protein phosphorylation

Protein kinases and phosphatases form intricate networks integrating multiple signals to regulate excitatory and inhibitory synapses. So far, most attention has been given to the respective neurotransmitter receptors (AMPA/NMDA and GABA_A receptors), and to a lesser extent to the regulation of excitatory PSD proteins (Kim & Sheng, 2004). Importantly, protein phosphorylation modulates not only protein–protein interactions within the PSD, but also the stability of target proteins through the activation of specific proteases, such as calpain-1. Therefore, bidirectional effects (increase or decrease of synapse strength) can be expected in dependence of local and global neuronal activity.

Here, we postulate that intracellular signalling cascades act in concert on the postsynaptic protein complexes of glutamatergic and GABAergic synapses to maintain the balance of excitation and inhibition. Such coordinated regulation would imply extensive convergence of signal transduction cascades acting on PSD-95 and gephyrin scaffolds. Therefore, it will be of primary interest to identify and characterize putative phosphorylation sites on gephyrin to determine their significance for the regulation of postsynaptic clustering, interaction with GABA_ARs (and GlyRs), and other proteins of the GABAergic PSD. Next, the protein kinases and phosphatases involved will have to be identified. Given the relevance of Ca²⁺-dependent mechanisms for intracellular signalling, a major open question to resolve in GABAergic synapses will be the identification of the source of calcium, and its relationship to GABA_A receptor function. However, evidence for cross-talk between metabotropic receptors and GABA_ARs, involving Ca²⁺-dependent phosphorylation mechanisms is accumulating rapidly (e.g. Chen et al. 2006; Janssen et al. 2009), indicating that multiple pathways probably contribute to the regulation of GABAergic transmission in different populations of neurons. Ultimately, it will be necessary to distinguish the respective roles of GABA_ARs versus gephyrin phosphorylation, and notably the possibility that the gephyrin scaffold serves to anchor functionally interrelated signalling proteins that mediate homeostatic synaptic plasticity.

Role of cation–chloride cotransporters

As a complementary view to the main hypothesis discussed in the previous sections, synaptic homeostasis might also depend to a large degree on the action of multifunctional proteins, such as KCC2, which exert multiple, independent effects on intrinsic neuronal properties, synapse formation and synaptic plasticity (Blaesse et al. 2009). Thus, KCC2 expression levels in developing neurons regulate the formation of GABAergic and glutamatergic synapses (Chudotvorova et al. 2005), the latter being independent of its cotransporter function but achieved by interactions with specific cytoskeletal proteins (Li et al. 2007). Deficiency in KCC2 affects dendritic spine morphology, in line with the selectively high concentration of KCC2 proteins in the vicinity of spines in mature neurons (Gulyás et al. 2001). Finally, by regulating intracellular Cl⁻ gradients, KCC2 has an immediate impact on the strength of GABA function, which might be set dynamically by phosphorylation-dependent cell surface trafficking (Lee et al. 2007). Therefore, rather than requiring the parallel or concerted action of signalling cascades onto distinct postsynaptic complexes, the multifunctionality of KCC2 offers another powerful and versatile mechanism of homeostatic synaptic plasticity that might complement the regulation of scaffolding proteins in GABAergic and glutamatergic PSDs. It will be of interest to determine whether these two principal pathways are interconnected, for example via an interaction between KCC2 and gephyrin (and/or PSD-95).

Glossary

Abbreviations

CaMKII

Ca²⁺–calmodulin-dependent protein kinase II

GABA_AR

GABA_A receptor

GlyR

glycine receptor

PSD

postsynaptic density

Author contributions

S.K.T. and J.M.F. have jointly written this review and have approved the final version.