Gephyrin is a major scaffolding protein concentrated at most if not all GABAergic postsynaptic elements (reviewed in refs. 81, 82). Gephyrin is also concentrated at inhibitory glycinergic postsynaptic sites, and in fact was first identified as a protein that copurified with the glycine receptor. While gephyrin directly binds to the large intracellular loop of the glycine receptor P subunit83, direct binding to GABAA receptor subunits has not been found. Gephyrin contains multiple sites of alternative splicing, and different splice variants exhibit different subcellular localizations in transfected cell lines84,85. Some of the gephyrin variants aggregate in a filamentous-type pattern. Gephyrin binds tubulin, probably via a short internal region with 60-80% identity with the repeat motifs of MAP2 and tau mediating tubulin polymerization and microtubule binding85. Surprisingly, the primary sequence of gephyrin suggests that it evolved by insertion of a novel sequence into genes encoding molybdenum cofactor biosynthetic enzymes. Indeed, gephyrin retains molybdenum cofactor biosynthetic activity86, and mice lacking gephyrin are defective in molybdoenzyme activity in non-neural tissues87. Thus gephyrin appears to have dual unrelated functions.
The role of gephyrin in localizing GABAa receptors to the postsynaptic site has been probed with antisense oligonucleotide and mouse knockout experiments32,88-90. Both types of experiments revealed a partial dependence of GABA receptor synaptic clustering on gephyrin, to varying degrees. While synaptic clustering of GABAA receptor subunits a2, a3, P2/3, and y2 was partially dependent on gephyrin, clustering of a1 and a5 was completely gephyrin independent89,90. For a2 and Y2, dendrite surface levels were unaffected by loss of gephyrin, indicating a role for gephyrin in synaptic trapping or anchoring but not in trafficking receptors to the cell surface. Functionally, there was a decrease in amplitude but no change in frequency of mIPSCs in hippocampal neurons lacking gephyrin compared with wild type90. Developmental compensation in the knockout is always a possibility, but the complete absence of glycine receptor clustering in both spinal cord and hippocampal neurons from the gephyrin knockout indicates that other proteins cannot compensate for this aspect of gephyrin's function87,90. Indeed, the knockout mice die within one day of birth due to an apparent failure of glycinergic transmission87. Taken together, these results indicate the existence of a major gephyrin-independent pathway for synaptic clustering of GABAA receptors.
In contrast, the localization of gephyrin itself to the postsynaptic domain is highly dependent on GABAA receptors. Knockout of the Y2 subunit, present in >80% of GABAA receptors, results in a dramatic reduction in synaptic clustering of a subunits and gephyrin and in mIPSC frequency32. Rescue experiments in primary neuron cultures from these mice indicate that both the major cytoplasmic loop and TM4 of y2 are necessary for recruiting gephyrin to the synapse50.
An idea emerging from this evidence and the growing list of gephyrin interacting proteins is that gephyrin may function as a scaffolding protein to recruit components in addition to or other than GABAA receptors to the postsynaptic domain. In addition to the glycine receptor P subunit and tubulin, gephyrin can bind to the GDP-GTP exchange factor collybistin91, the actin regulatory protein profilin92, the translation regulator rapamycin and FKBP12 target protein RAFT193, the motor associated protein dynein light chain94, and the GABAA receptor binding protein GABARAP60. Most of these proteins are not specifically concentrated at GABAergic postsynaptic elements, but they may be recruited for signaling purposes and/or be involved in trafficking of gephyrin (e.g., ref. 95).
A family of proteins that functions at both GABA and glutamate synapses is the cadherin cell adhesion molecules and catenin binding partners. Cadherins are a large family of calcium-dependent homophilic cell adhesion molecules that link to F-actin and to signaling pathways via catenins. In hippocampal cultures, n-cadherin and P-catenin are among the earliest components of developing GABAergic and glutamatergic synapses96. While n-cadherin is lost from GABA synapses as they mature, P-catenin is retained, suggesting that some other cadherin may be present. A dominant-negative form of n-cadherin had a strong but transient effect on disrupting both GABA and glutamate synapse assembly early but not late in development of hippocampal cultures97,98. These results suggest that n-cadherin normally contributes to early stage GABA synapse development but it is not absolutely required. The cadherin-related y-protocadherins may also contribute to inhibitory synapse formation in some brain regions, in spinal cord but not hippocampus (refs. 99,100 and Chapter 10).
Unfortunately, cadherins and catenins as yet have been little studied specifically in the context of GABAergic synapses. Characterization of the cadherins present in GABAergic pathways in vivo would be a good first step. Detailed analysis of GABAergic synapses in the catenin knockout mice would also be interesting. An a-N-catenin knockout mouse survives but exhibits defects in spine maturation and stability101; whether GABA synapses are less stable is not known. Deletion of P-catenin specifically in pyramidal neurons resulted in dispersal of synaptic vesicles and a reduced reserve pool102; whether P-catenin is required in a similar way for presynaptic organization at GABAergic synapses is not known. The dominant-negative n-cadherin did not disrupt the peri-somatic proximo-distal density gradient of GABAergic synapses along dendrites in hippocampal culture98, and other studies suggest the involvement of neurofascin adhesion molecules in the peri-somatic targeting of basket cell GABAergic axons103. Analyses of knockout mice indicate a contribution of other Ig domain family cell adhesion molecules such as L1, and extracellular matrix molecules such as tenascin-R, to proper development of GABAergic synapses104,105. A key question related to these other cell adhesion protein families is whether they function directly at the GABAergic synapse or perhaps indirectly via a role in earlier developmental events.
8. THE DYSTROPHIN GLYCOPROTEIN COMPLEX (DGC)
Originally identified in relation to muscular dystrophy, the DGC has been found as a major component of mature inhibitory GABAergic synapses. Mutations in dystrophin are a major etiology of Duchenne and Becker muscular dystrophy106. In addition, the finding that some patients with Duchenne muscular dystrophy display severe mental retardation suggests that dystrophin may have an important role in the CNS. Dystrophin is a large actin-binding protein primarily expressed in muscle and is also present in several isoforms in brain. Dystrophin binds the transmembrane protein dystroglycan, which binds laminin, agrin, and perlecan, thus forming a link from the cytoskeleton to the extracellular matrix107. Synthesized from a single precursor, dystroglycan is proteolytically cleaved to produce a and P subunits. a-Dystroglycan exhibits extensive and tissue-specific glycosylation, and defects in its glycosylation are thought to underlie Fukuyama congenital muscular dystrophy108. Links to additional intracellular signaling molecules are provided via binding of P-dystroglycan to Grb-2, and dystrophin to syntrophins and dystrobrevin.
A role for the DGC at GABAergic synapses was first suggested by Kneusel et al.109 who found dystrophin immunoreactivity in brain extensively colocalized with gephyrin and GABAA receptor subunits a1 and a2. Furthermore, the number and the size of GABAA receptor clusters was significantly reduced in mdx mice deficient in long dystrophin isoforms, although gephyrin clustering was unaltered. Further studies in hippocampal neuron cultures showed that dystrophin, a- and P-dystroglycan all concentrate at a subset of GABAergic synapses late in development110. Since targeted deletion of dystroglycan is early embryonic lethal in mice, due to its role in basement membrane formation, a conditional knockout was generated. Cultured neurons lacking dystroglycan still clustered gephyrin and GABAA receptors opposite GABAergic terminals, with no apparent defects. Moreover, brain-specific deletion of dystroglycan in vivo leads to major defects in cell migration and hippocampal long-term potentiation, but the mice survive and hippocampal slices exhibit normal field potentials in response to stimulation111. These studies indicate that the DGC is not essential for GABAergic synapse formation, including synaptic recruitment of gephyrin and GABA receptors, but suggest that the DGC may be involved in stabilization or other signaling at the mature GABAergic synapse. The finding that dystroglycan in brain is largely complexed with a-neurexin112,113 further supports some role in trans--synaptic maintenance or signaling.
While P-neurexin binding to neuroligin-2 can recruit gephyrin and GABAA receptors and may represent an early step in the formation of functional GABAergic synapses (see below), these interactions do not recruit the DGC47. The DGC is recruited to GABA synapses by a mechanism independent of gephyrin and of GABAA receptor via a signal unique to GABA but not glutamate axons110,114. How the observed interaction of dystroglycan with a-neurexins113 fits into the recruitment or function of the DGC at GABA synapses is not yet clear.
Was this article helpful?