ABSTRACT
The inhibitory glycine receptor (GlyR) is a ligandgated chloride channel protein found at many synapses of the mammalian central nervous system. During development, distinct isoforms of the GlyR are generated by the sequential expression of different a subunit variants. The appearance of adult-type GlyRs in spinal cord is accompanied by the accumulation of a 93×103Mr receptor-associated peripheral membrane protein. The latter has been localized at the cytoplasmic face of glycinergic postsynaptic membranes and is thought to anchor GlyRs beneath glycinergic nerve terminals. The 93×103Mr protein binds with high affinity to polymerized tubulin, suggesting that it functions as a receptor-microtubule linking component. Our data suggest that the interaction of developmentally regulated receptor isoforms with specialized microtubule-associated proteins represents a crucial step in the assembly of postsynaptic receptor matrices.
INTRODUCTION
Each neuron in the mammalian brain carries up to thousands of postsynaptic membrane specializations. These postsynaptic sites are characterized by receptor proteins, which mediate signal transduction upon binding of neurotransmitter released from the apposed nerve terminal. At present, little is known about the mechanisms involved in the selective localization of neurotransmitter receptors in postsynaptic membrane areas. This report describes some characteristics of receptor regulation at glycinergic synapses in the developing mammalian central nervous system (CNS).
The amino acid glycine is a major inhibitory neurotransmitter in the vertebrate CNS (Aprison and Daly, 1978). Glycine-mediated inhibition of neuronal activity results from activation of the inhibitory glycine receptor (GlyR), a ligand-gated chloride channel in spinal cord and certain brain regions (reviewed by Langosch et al. 1990; Betz, 1991). The GlyR has been purified from mammalian spinal cord and represents a pentameric protein composed of ligand-binding subunits of 48×103Mr (α) and homologous polypeptides of 58 × 103Mr(β) (Pfeiffer et al. 1982; Graham et al. 1985; Langosch et al. 1988). The primary structures of these GlyR subunits have been determined by Cdna sequencing and shown to share a common transmembrane topology and significant sequence homology with nicotinic acetylcholine and GABAA receptor proteins (Grenningloh et al. 1987, 1990a). All these receptors, therefore, are thought to constitute members of a superfamily of ligandgated ion-channel proteins that evolved from a common ancestral polypeptide (Betz, 1990).
Biochemical and molecular cloning techniques indicate considerable heterogeneity of GlyRs during development (Betz, 1991). Moreover, a 93×103Mr protein co-purifying with GlyR subunits upon affinity chromatography has been implicated in the synaptic topology of GlyRs (Triller et al. 1985; Schmitt et al. 1987). Here we summarize data which suggest mechanistic models for GlyR localization during development.
Developmental heterogeneity of the GlyR
Evidence for subtype heterogeneity of the GlyR was first detected in studies of rodent spinal cord development, in which a neonatal isoform prevalent at birth was shown to differ from the adult GlyR in pharmacological, immunological and biochemical properties (Becker et al. 1988). This neonatal receptor shows only low affinity for the glycinergic antagonist strychnine and contains an a subunit of 49×103Mr. Within 2-3 weeks after birth, the neonatal GlyR in spinal cord is completely replaced by the adult-type receptor.
Further evidence for developmental heterogeneity of the GlyR came from expression studies (Akagi and Miledi, 1988). Injection of poly(A)+ RNA isolated from rat brain or spinal cord into Xenopus oocytes caused expression of glycine-gated chloride channels in the oocyte membrane. By sedimentation on sucrose density gradients, two classes of GlyR mRNAs could be separated, both of which gave rise to functional channels when expressed in oocytes. A rapidly sedimenting heavy mRNA species was abundant in spinal cord of adult rats, whereas fractions from neonatal spinal cord and adult cerebral cortex contained a low molecular weight GlyR mRNA.
Molecular cloning studies in our laboratory now have revealed that the developmental heterogeneity of GlyRs observed at both the protein and mRNA expression level reflects the existence of several GlyR α subunit genes. By low-stringency screening of cDNA and genomic libraries, clones encoding different variants of the originally described GlyR α subunit (now termed α1) have been isolated. Two different α2 cDNAs, α2 (Grenningloh et al. 19906; Kuhse et al. 1991) and α2* (Kuhse et al. 1990a), as well as an α3 cDNA (Kuhse et al. 19906) were found to be differentially expressed in rats and humans. Moreover, genomic sequences encoding a fourth variant, α4, have been isolated from mouse (Y. Maulet, B. Matzenbach, and H. Betz, unpublished observations). All these a sequences display a high degree of amino acid identity and correspond to GlyR proteins, whose expression is under distinct temporal and regional control. The α2 polypeptides represent ligand-binding subunits of neonatal GlyRs in spinal cord and forebrain, whereas a3 receptors appear mainly to be expressed in the postnatal cerebellum (Malosio et al. 1991). The location of α4 GlyRs is presently unknown; in adult rodent brain,α4 mRNA is found only at very low levels.
Northern blot analysis, amplification by polymerase chain reaction (PGR) and in situ hybridization have been used to monitor the accumulation of GlyR subunit transcripts during CNS development. All presently available data indicate high levels of α2 and β transcripts in embryos and neonatal animals, whereas α1 and α3 sequences become significantly expressed only at about 2 weeks after birth (reviewed by Betz, 1991). This is consistent with the biochemically demonstrated exchange of GlyR isoforms in spinal cord during the first two postnatal weeks and suggests that adult glycinergic synapses contain GlyRs built of α1 or, to a much lesser extent, α3 polypeptides. Interestingly, the different α subunit variants show high sequence divergence in their predicted intracellular domains. This may relate to a differential regulation of their cell surface distribution by intracellular receptor-associated proteins.
Gephyrin, a 93×103Mr glycine receptor-associated protein, is a putative receptor-microtubule linker
GlyR solubilized from rat, pig or mouse spinal cord membranes co-purifies with a 93×103Mr protein upon affinity chromatography (Pfeiffer et al. 1982; Graham et al. 1985). Immuno-electron microscopy using selective monoclonal antibodies (Pfeiffer et al. 1984) indicates that this polypeptide decorates the cytoplasmic face of the glycinergic postsynaptic membrane (Triller et al. 1985; Altschuler et al. 1986). Moreover, in biochemical studies the 93×103Mr protein behaves as a typical peripheral membrane component, that can be extracted by alkaline pH or acylation (Schmitt et al. 1987). It therefore was proposed that this polypeptide may anchor the GlyR in the postsynaptic membrane by interaction with cytoskeletal proteins or structural components of postsynaptic densities. Consistent with this view, the increase of 93×103Mr protein immunoreactivity in developing spinal cord parallels the ontogenetic appearance of the adult GlyR isoform (Becker et al. 1988).
Sequencing of 93×103Afr protein cDNAs (P. Prior, B. Schmitt, G. Grenningloh, I. Pribilla, G. Multhamp, K. Beyreuther, Y. Maulet, P. Werner, D. Langosch, J. Kirsch and H. Betz, unpublished observations) revealed some similarities in the amino acid composition of the deduced protein sequence to that of microtubule-associated protein 5 (MAP5). We therefore investigated whether the 93 ×103Mr protein may bind to tubulin (Kirsch et al. 1991). Indeed, significant amounts of tubulin were found to copurify with the 93×103Mr protein upon fractionation of GlyR polypeptides. Also, tubulin binds to the isolated and immobilized 93 ×103Mr protein in an overlay procedure. Finally, the 93 ×103Mr protein co-assembles with polymerized tubulin or microtubules through repeated cycles of polymerization and depolymerization and mediates sedimentation of GlyR α and β subunits with polymerized tubulin. This interaction displays high affinity (KD≈2.5 nM), significant cooperativity (Hill coefficient≈2.1) and approaches a stoichiometry of about 1:4 under saturating conditions (Kirsch et al. 1991). Similar cooperative characteristics and stoichiometries of tubulin binding have been reported for the interaction of MAP2 and MAP5 with polymerized tubulin. The 93×103Mr protein therefore may be tentatively classified as a novel type of tubulin-associated protein, which serves as a receptor-cytoskeleton linking protein. Based on these considerations, we propose the name gephyrin (from greek gephyra, bridge) for this polypeptide to indicate its presumed function (P. Prior et al., unpublished work).
Isolation of gephyrin cDNAs has unraveled considerable heterogeneity resulting from alternative splicing (P. Prior et al., unpublished data). The functional significance of this variability is not known. We speculate however, that a set of gephyrin variants generated from a common pre-mRNA may be implicated in different membrane protein-microtubule interactions. Moreover, Northern analysis and PCR amplification indicate that gephyrin transcripts are present in many rat tissues. Thus, gephyrin may be important in determining the topography of different cell surface specializations.
Conclusions and perspectives
The data summarized here suggest that the localization of GlyRs to postsynaptic membrane areas is a multistep process during synaptogenesis. First, a neonatal (or embryonic) GlyR isoform is replaced by adult-type receptors, which characterize postsynaptic membrane areas in the mature CNS (Triller et al. 1985). This involves both induction of al (and a3), and repression of a2, gene transcription. Second, the increased synthesis of gephyrin, a 93×103Mr GlyR-associated protein, is proposed to be crucial for anchoring GlyRs at postsynaptic membranes by interaction with subsynaptic tubulin. Consistent with this view, the presence of ‘membrane-bound’ tubulin in brain synaptosomes and postsynaptic densities has been reported by several investigators (see review by Stephens, 1986). Moreover, a recent immunohistochemical study indicates the existence of cold-stable acetylated microtubules in the subsynaptic region of the sarcoplasm underlying the neuromuscular junction (Jasmin et al. 1990). Although the functional and structural relevance of tubulin as a component of the postsynaptic complex is still debated (Stephens, 1986), submembraneous tubulin arrays may be a general feature of postsynaptic specializations and provide a scaffold for anchoring receptors under nerve terminals. If so, induction of ordered microtubular structures under ingrowing nerve terminals may constitute a primary event in postsynaptic membrane formation.
ACKNOWLEDGEMENTS
This work was supported by Deutsche Forschungsgemeinschaft (Leibniz-Programm and Schwerpunkt ‘Funktionelle Domânen’), Fonds der Chemischen Industrie and the Minerva Programm of the Max-Planck-Gesellschaft. We thank S. Wartha for expert secretarial assistance.