Dual Requirement for Gephyrin in Glycine Receptor Clustering and Molybdoenzyme Activity

See allHide authors and affiliations

Science  13 Nov 1998:
Vol. 282, Issue 5392, pp. 1321-1324
DOI: 10.1126/science.282.5392.1321


Glycine receptors are anchored at inhibitory chemical synapses by a cytoplasmic protein, gephyrin. Molecular cloning revealed the similarity of gephyrin to prokaryotic and invertebrate proteins essential for synthesizing a cofactor required for activity of molybdoenzymes. Gene targeting in mice showed that gephyrin is required both for synaptic clustering of glycine receptors in spinal cord and for molybdoenzyme activity in nonneural tissues. The mutant phenotype resembled that of humans with hereditary molybdenum cofactor deficiency and hyperekplexia (a failure of inhibitory neurotransmission), suggesting that gephyrin function may be impaired in both diseases.

The main inhibitory inputs to spinal cord and brain-stem motoneurons use glycine as a neurotransmitter (1). The α and β transmembrane subunits of glycine receptors (GlyRs) from spinal cord copurify with gephyrin, a 93-kD cytoplasmic protein (2). Gephyrin binds to the β subunit of the GlyR and to tubulin, thereby linking GlyRs to the cytoskeleton (3). This interaction appears to be important for the accumulation of GlyRs at synapses, because GlyRs are precisely colocalized with gephyrin at synapses in the brain and spinal cord, gephyrin aggregates GlyRs when coexpressed with them in heterologous cells, and attenuation of gephyrin synthesis with antisense oligonucleotides prevents clustering of GlyRs at synaptic sites on cultured spinal neurons (4–6). Molecular cloning of gephyrin (7) revealed unexpected similarity to three Escherichia coli proteins (moeA, moaB, and mog), a Drosophila melanogaster protein (cinnamon), and an Arabidopsis thaliana protein (cnx1), all of which are involved in the synthesis of a molybdenum-containing cofactor essential for the activity of molybdoenzymes (8). This conservation (Fig. 1A) suggested that genes of the same bacterial operon may have been joined during evolution to form a multidomain protein that gained a novel function.

Figure 1

Structure of gephyrin and generation of geph–/– mice. (A) Gephyrin shows similarity with proteins from E. coli (mog, moaB, and moeA),Drosophila (cinnamon), and Arabidopsis (cnx1) that have been implicated in molybdenum cofactor metabolism. MoaB/mog-like and moeA-like regions are indicated by hatched and open bars, respectively. (B) Targeting strategy. Wild-typegeph gene (top), targeting vector (middle), and mutant locus (bottom) are shown. Sites of primers for polymerase chain reaction (PCR) genotyping (arrowheads) and external probe for Southern blot (DNA) analysis are indicated. E, Eco RI; Ea, Eag I; EV, Eco RV; H, Hind III; P, Pst I; S, Sac I; NEO, neomycin resistance gene; TK, thymidine kinase. (C) Southern blot analysis of wild-type and two successfully targeted ES cell clones. (D) PCR analysis of genomic DNA from wild-type (+/+), heterozygous (+/–), and homozygous (–/–) littermates. (E) Northern blot analysis of gephyrin mRNA from brain, using a full-length cDNA as probe. (F) Protein immunoblot analysis of gephyrin immunoreactivity from brain. The light band in all lanes represents nonspecific reactivity.

We disrupted the gephyrin gene by deleting exon 1 and upstream sequences responsible for initiating transcription and translation (9, 10). The mutant allele (Fig. 1B) was transferred to embryonic stem (ES) cells by electroporation, and two successfully targeted clones gave rise to germ-line chimeras (Fig. 1, C and D). Heterozygous offspring (geph+/–) were phenotypically normal. Homozygous mutants (geph–/–) were born in expected numbers, despite lacking detectable gephyrin mRNA (Fig. 1E) and protein (Fig. 1F). However, all geph–/–mice died within 1 day of birth. Thus, gephyrin was dispensable for embryonic development but essential for postnatal survival.

Geph–/– neonates appeared externally normal but failed to suckle and never produced the vocalizations characteristic of normal neonates. In response to mild tactile stimuli, control mice flailed their limbs, whereas geph–/– littermates assumed a rigid, hyperextended posture (Fig. 2, A and B). The mutants became increasingly hyperresponsive to tactile stimuli and exhibited apnea (difficulty breathing) by 12 hours after birth. These symptoms are consistent with impairment of inhibitory glycinergic inputs to motoneurons. To test this possibility, we stained spinal cord sections of geph–/– mice and littermate controls with antibodies to synaptic components (11). In controls, GlyRs and gephyrin are colocalized at inhibitory synapses on motoneuronal somata and primary dendrites (4). In homozygous mutants, synaptic boutons were numerous, but no gephyrin was detected, and GlyRs were diffusely distributed (Fig. 3A). Likewise, gephyrin and GlyRs were coclustered on subsets of neurons in the brainstem and hypothalamus of control neonates, but gephyrin was undetected and GlyRs were diffusely localized in geph–/– littermates (Fig. 3B). High levels of GlyR mRNA and protein persisted in geph–/– brains (Fig. 3, D and E), indicating that gephyrin was required for GlyR aggregation rather than GlyR synthesis. No difference was detected between mutants and controls in the size or distribution of glutamate receptor clusters, in the synaptic localization of their putative clustering proteins of the PSD-95/SAP-90 family (12), or in the overall morphology of the spinal cord and brain (Fig. 3, A and C). Thus, defects were specific for GlyRs.

Figure 2

Geph+/−(A), geph−/− (B), and geph+/+ (C) mice approximately 8 hours after birth. Littermates in (A) and (B) were touched gently before photography, to show the rigid, hyperextended posture of geph−/− mice. Also note that the homozygote has no milk in its stomach. (C) Injection of a wild-type neonate with strychnine (1.4 μg/g of body weight) phenocopies the characteristic geph−/− posture.

Figure 3

Disrupted clustering of GlyRs in geph–/– mutant neonates. (A) Cryostat sections are shown of spinal cord stained with two antibodies to gephyrin that recognize distinct epitopes (5a and 7a), and with antibodies to the GlyR α1 subunit, the glutamate receptor GluR1 subunit, a conserved domain on several members of the PSD-95/SAP-90 family (PDZ), and the synaptic vesicle proteins SV2 and synaptophysin (SN). Gephyrin and GlyRs cocluster on motoneuronal somata and proximal dendrites in controls. Insets show localization of these proteins at synaptic sites by double labeling (green, gephyrin or GlyR; red, synaptophysin; yellow, overlap). Gephyrin is absent and GlyRs are diffusely distributed in mutants, but synapses are present. GluR1 and PSD-95–like proteins, shown in a double-labeled section, are clustered in both +/– and –/– mice. Bar, 10 μm. (B) Confocal images of GlyR immunoreactivity in hypothalamic neurons from +/+ and –/– mice. Inset shows a portion of the mutant neuron at increased gain, to demonstrate diffusely distributed GlyRs. (C) Geph–/– spinal cord exhibits normal overall morphology, revealed by staining with hematoxylin and eosin. (D) Northern blot analysis of GlyR α1 subunit mRNA in homozygous mutants and littermate controls. (E) Protein immunoblot analysis of GlyR immunoreactivity in mutant and controls.

The motor defects seen in geph–/– mice occurred earlier and were more severe than those observed in mutant mice that lack the GlyR α1 subunit or have reduced levels of the GlyR β subunit (13). This difference might be due to a more drastic reduction of synaptic GlyR levels in geph–/– mice, but the sequence similarity noted above (7, 8) raised the possibility that gephyrin might also be required for molybdenum metabolism. In fact, humans born with autosomal recessive molybdenum cofactor deficiency exhibit severe neurological defects that resemble those seen in geph–/– mice, such as hypertonicity, myoclonus, and difficulty in feeding (14). Molybdenum cofactor deficiency is diagnosed in humans by demonstrating coordinate loss of activity of two distinct molybdenum-containing enzymes, sulfite oxidase and xanthine dehydrogenase, which are ubiquitously expressed (14). We readily detected sulfite oxidase activity (15) in livers of control neonates, but not in livers of geph–/–mice (Fig. 4A). Xanthine dehydrogenase is not present in liver at birth even in controls, but is expressed in intestine. Xanthine dehydrogenase is also abundant in milk (16), so we avoided this source of contamination by assaying fetal intestine (17) and detected no activity in the homozygous mutants (Fig. 4B). As a control, we assayed the activity of an unrelated metabolic enzyme, lactate dehydrogenase, and the abundance of sulfite oxidase mRNA (18). Neither of these parameters differed among genotypes (Fig. 4, C and D). Thus, gephyrin appears to be essential for molybdenum cofactor biosynthesis in mice.

Figure 4

Activities of two molybdoenzymes, sulfite oxidase (A) and xanthine dehydrogenase (B), and a control enzyme, lactic dehydrogenase (LDH) (C), in neonatal liver [(A) and open bars in (C)] and embryonic day-18 intestine [(B) and dark bars in (C)] of geph–/–, geph+/–, and geph+/+ littermates. (D) Northern blot analysis of sulfite oxidase mRNA. Molybdoenzyme activities were undetected in geph–/– mice, but LDH levels and abundance of sulfite oxidase RNA did not vary significantly among genotypes.

In view of these defects in molybdoenzyme activity, we considered the possibility that the neurological symptoms of geph–/–mice were secondary to their metabolic disorder rather than to disruption of glycinergic synapses. Two findings favor this possibility. First, patients with mutations in the sulfite oxidase gene are symptomatically similar to patients that lack molybdenum cofactor, suggesting that neurological defects in both groups result from sulfite toxicity (19). Second, glycinergic transmission may be excitatory rather than inhibitory in embryos (20). The developmental stage at which activation of GlyR becomes inhibitory is around birth in rats (21) but is unknown in mice. If it were postnatal, interference with glycinergic transmission might have a calming rather than a stimulatory effect on motor behavior, and the observed hyperresponsiveness of geph–/– mice could reflect sulfite toxicity. To address these issues, we injected neonatal mice with strychnine, a specific antagonist of GlyR (21). Like gephyrin mutants, strychnine-intoxicated neonates assumed a rigid, hyperextended posture in response to mild tactile stimuli (Fig. 2C). Thus, glycinergic transmission appeared to be predominantly inhibitory at birth, and blockade of glycinergic transmission in the absence of interference with molybdenum metabolism phenocopied the motor symptoms of gephyrin deficiency. Motor defects in geph–/–mice—and by implication, in molybdenum cofactor–deficient humans—may therefore result from both impaired inhibitory neurotransmission and impaired molybdoenzyme activity.

Our results demonstrate that gephyrin is essential for the synaptic clustering of GlyRs in vivo. Gephyrin may play a role at inhibitory synapses similar to that played by the clustering protein rapsyn at neuro- muscular junctions (22). A second role of gephyrin is in synthesis of molybdenum cofactor. Homologous proteins have been described in other phyla (8), but no components of the molybdenum cofactor synthetic pathway have been identified in vertebrates. This peculiar pleiotropy raises four interesting possibilities. First, mutations in the gephyrin gene may underlie some cases of autosomal recessive human molybdenum cofactor deficiency. Second, some neurological symptoms now attributed to molybdoenzyme inactivity in humans may actually reflect a lack of receptor accumulation at inhibitory synapses. Third, although some cases of hyperekplexia (startle disease or stiff baby syndrome) are due to mutations in the GlyR (23), others might result from mutations of gephyrin, in which case some of the symptoms could reflect molybdenum insufficiency. Finally, activation or aggregation of GlyRs might modulate the ability of gephyrin to promote molybdopterin biosynthesis, thus resulting in a functional link between molybdoenzymes and inhibitory neurotransmission.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: sanesj{at}


View Abstract

Navigate This Article