Disruption of a Structurally Important Extracellular Element in the Glycine Receptor Leads to Decreased Synaptic Integration and Signaling Resulting in Severe Startle Disease

Abstract

Functional impairments or trafficking defects of inhibitory glycine receptors (GlyRs) have been linked to human hyperekplexia/startle disease and autism spectrum disorders. We found that a lack of synaptic integration of GlyRs, together with disrupted receptor function, is responsible for a lethal startle phenotype in a novel spontaneous mouse mutant shaky, caused by a missense mutation, Q177K, located in the extracellular β8-β9 loop of the GlyR α1 subunit. Recently, structural data provided evidence that the flexibility of the β8-β9 loop is crucial for conformational transitions during opening and closing of the ion channel and represents a novel allosteric binding site in Cys-loop receptors. We identified the underlying neuropathological mechanisms in male and female shaky mice through a combination of protein biochemistry, immunocytochemistry, and both in vivo and in vitro electrophysiology. Increased expression of the mutant GlyR α1^Q177K subunit in vivo was not sufficient to compensate for a decrease in synaptic integration of α1^Q177Kβ GlyRs. The remaining synaptic heteromeric α1^Q177Kβ GlyRs had decreased current amplitudes with significantly faster decay times. This functional disruption reveals an important role for the GlyR α1 subunit β8-β9 loop in initiating rearrangements within the extracellular-transmembrane GlyR interface and that this structural element is vital for inhibitory GlyR function, signaling, and synaptic clustering.SIGNIFICANCE STATEMENT GlyR dysfunction underlies neuromotor deficits in startle disease and autism spectrum disorders. We describe an extracellular GlyR α1 subunit mutation (Q177K) in a novel mouse startle disease mutant shaky Structural data suggest that during signal transduction, large transitions of the β8-β9 loop occur in response to neurotransmitter binding. Disruption of the β8-β9 loop by the Q177K mutation results in a disruption of hydrogen bonds between Q177 and the ligand-binding residue R65. Functionally, the Q177K change resulted in decreased current amplitudes, altered desensitization decay time constants, and reduced GlyR clustering and synaptic strength. The GlyR β8-β9 loop is therefore an essential regulator of conformational rearrangements during ion channel opening and closing.

Keywords: fast decay; glycine receptor; hydrogen bond network; shaky; startle disease; β8–β9 loop.

Figure 1.

Shaky mice display a neuromotor phenotype and carry a Glra1 mutation. A, Comparison of hindfeet and hindlimb clenching behavior of wild-type Glra1^+/+, Glra1^sh/sh, and Glra1^ot/ot mice at P14 when mice were lifted by the tails, and progressive motor phenotype at P22 (bottom panels) with rigidity in forelimbs, hindlimbs, and tails. B, Impaired righting time of mutant Glra1^sh/sh mice. Time Glra1^sh/sh mice are required to get up at P14 to P21 compared with wild type, prolonged righting time at stage P22 to P28. Cutoff time was 10 min, and 2 of 10 mice older than 21 d timed out. Glra1^+/+, n = 7; Glra1^sh/sh, n = 10. C, Survival curves of Glra1^sh/sh, n = 9; Glra1^+/sh, n = 19; Glra1^+/+, n = 4 mice. Glra1^sh/sh mice die between weeks 3 and 6 of life. D, Survival curves of backcross experiments with GlyR mutant mouse lines spasmodic (spd) and oscillator (ot): note that heterozygous Glra1^sh/ot mice die 4 weeks after birth similar to Glra1^ot/ot mice and in contrast to Glra1^+/sh mice. Glra1^sh/spd, n = 10; Glra1^sh/sh, n = 9; Glra1^sh/ot, n = 10; Glra1^ot/ot, n = 10. E, Rotarod performance. Wild-type mice had no difficulties staying on the rotating rod for a maximum time of 300 s. Glra1^sh/sh mice were able to remain on the rod for only a few seconds. Glra1^+/+, n = 17; Glra1^sh/sh, n = 12. *p < 0.05; **p < 0.01; ***p < 0.001.

Figure 2.

Glra1^sh/sh mice display protein expression differences and reduced agonist binding. A, Transcript analysis of GlyR α1, α2, and β subunits, the scaffold protein gephyrin, the glycine transporter 1 (GlyT1), and β-actin (housekeeping gene) in spinal cord, brainstem, and cortex from Glra1^+/+ (marked +/+), Glra1^+/sh (+/sh), and Glra1^sh/sh (sh/sh) mice. B, GlyR protein expression profile in spinal cord, brainstem, and cortex at P21 in wild-type (+/+), heterozygous (+/sh), and homozygous animals (sh/sh). Gephyrin (geph) was detected at the appropriate molecular weight of 93 kDa; all GlyR α subunits were stained with a pan-α antibody (mAb4a, 48 kDa), GlyR α1 specifically with mAb2b (48 kDa); and β-actin was used as loading control (LC; 42 kDa). C, D, Quantification of GlyR α1 protein (C) and gephyrin (D) in spinal cord and brainstem normalized to β-actin in wild-type (Glra1^+/+) and shaky (Glra1^sh/sh; n = 6–10, *p < 0.05, n.s.). E, Strychnine–glycine competition in spinal cord tissue isolated from wild-type (pooled, n = 3) and homozygous shaky mice (pooled, n = 4), three independent experiments were performed. Thirty millimolar glycine bound to membrane preparations of tissue were displaced by [³H]strychnine using increasing concentrations of the radioactive antagonist (0–200 nm). Note the increase in K_D in membrane preparations from shaky mice.

Figure 3.

Developmental expression of the GlyR α1 subunit in shaky mice and after backcross into the mouse line oscillator. A, Spinal cord (sc), brainstem (bs), and cortex (cx; negative control) were analyzed at developmental stages P0, P7, P14, and P28 until severe symptoms in homozygous shaky mice compared with wild-type mice were observed. The GlyR α1 subunit was stained with the monoclonal mAb2b antibody at 48 kDa, with β-actin was used as a loading control (LC; 42 kDa). α1LY represents a lysate control following transfection of GlyR α1 into HEK293 cells and was used as a positive control. B, GlyR expression in P21 mouse tissue after backcross of shaky into the oscillator line. Sc, bs, and cx of mice carrying genotypes Glra1^+/+, heterozygous Glra1^+/sh, Glra1^+/ot, Glra1^sh/ot, and homozygous Glra1^ot/ot were analyzed for the expression of GlyR α1 subunit (48 kDa). Note that homozygous oscillator mice Glra1^ot/ot lack α1 and Glra1^sh/ot as well as Glra1^+/ot reveal reduced GlyR α1 levels. Again, β-actin served as an LC (42 kDa). C, D, Quantification of GlyR α (pan-α antibody; C) and GlyR α1 protein (D) from spinal cord preparations of Glra1^+/+, heterozygous Glra1^+/sh, Glra1^+/ot, Glra1^sh/ot animals, and homozygous oscillator Glra1^ot/ot normalized to β-actin. The expression level of wild-type animals Glra1^+/+ were set to 1 (100%), n = 4, *p < 0.05. n.s. = nonsignificant.

Figure 4.

GlyR α1^Q177K results in a lack of synaptic integration. A–C, Immunohistological stainings of spinal cord tissue from Glra1^+/+ compared with homozygous Glra1^sh/sh animals. Spinal cord slides (9 μm) were stained for the GlyR α1 subunit with the monoclonal antibody mAb2b together with presynaptic markers synapsin (syn; A) vesicular transporter (VGAT; B) and the postsynaptic marker gephyrin (geph; C). Ventral or dorsal horns are marked by a white dotted line. DAPI was used to stain nuclei. Right panels represent enlarged images of each staining. Note that there is less colocalization of GlyR α1 and gephyrin in homozygous shaky mice (white arrows). D, Spinal cord neuronal cultures from E13 embryos with genotypes Glra1^+/+ or Glra1^sh/sh were differentiated for 3 weeks in culture and stained for α1 (mAb2a) and gephyrin. An upregulation of GlyR α1 in shaky neurons was observed, but less colocalization in synaptic clusters with gephyrin (white arrows). Right panels represent enlarged dendrites of spinal cord neurons, which are marked by white boxes. E, Quantification of synaptic clusters in neurons from Glra1^+/+ and Glra1^sh/sh, n = 11 from two independent experiments. The relative expression of gephyrin and GlyR α1 is shown (*p < 0.05, n.s.), comparison of expression levels between Glra1^+/+ and Glra1^sh/sh and within each group.

Figure 5.

GlyR α1^Q177K leads to reduced agonist potency and faster decay times. A, Expression of wild-type and mutant heteromeric receptor complexes following cotransfection with the GlyR β subunit in HEK293 cells. Biotinylation assays were used to discriminate between surface and whole-cell protein. Left, Quantification of wild-type and mutant GlyR α1 protein with or without the GlyR β-subunit from whole-cell and surface pools normalized to pan-cadherin; data were obtained from at least three independent sets of experiments (n = 3–6). Right, Western blot analysis of biotinylation assays. The monoclonal pan-α antibody was used for recognition of the GlyR α1 subunit (48 kDa), the membrane protein pan-cadherin served as a loading control (LC; 135 kDa). UT, Untransfected cells; MOCK, GFP-transfected cells. B, Functional parameters from whole-cell recordings of transfected HEK293 cells for α1β and α1^Q177Kβ heteromeric receptor configurations, with current amplitudes at 1 mm (α1β, n = 9; α1^Q177Kβ, n = 16) and 100 μm glycine (α1β, n = 5; α1^Q177Kβ, n = 5); 1 mm (α1β, n = 11; α1^Q177Kβ, n = 9) and 100 μm β-alanine (α1β, n = 6; α1^Q177Kβ, n = 6); and 1 mm (α1β, n = 6; α1^Q177Kβ, n = 5) and 100 μm taurine (α1β, n = 5; α1^Q177Kβ, n = 6). C, Ligand-binding potencies (EC₅₀) determined by current measurements of transfected cells expressing the heteromeric receptor configurations α1β and α1^Q177Kβ according to the adult in vivo receptor configuration at seven different glycine, β-alanine, and taurine concentrations (0.3–3.000 μm, n = 5 for each receptor configuration and for the agonist used). The 1 mm concentration of the agonist glycine or the partial agonists β-alanine and taurine were used to determine I_max values. D, Decay times for desensitization. The decay currents were calculated in the presence of agonist (1 s). Right, Representative scaled current traces with α1β (n = 5, black) and α1^Q177Kβ (n = 6, red). E, Faster decay of shaky channels expressed in artificial synapses. Spontaneous IPSCs of α1β (black) and α1^Q177Kβ (red) shown at two different time scales, 20 s and 100 ms. Quantification of decay time constants of α1β (n = 10) and α1^Q177Kβ (n = 9) in artificial synapses/ enlarged scaled view of α1β (black) and α1^Q177Kβ (red). p < 0.05; **p < 0.01; ***p < 0.001.

Figure 6.

Shaky mice have impaired glycinergic synaptic transmission in PreBötC neurons. All recordings were made in whole-cell voltage-clamp mode in the presence of the ionotropic glutamate receptor antagonist KA (2 mm) and the GABA_A receptor antagonist bicuculline (10 μm). A, Traces from a wild-type PreBötC neuron illustrate the evoked IPSC (at stimulation intensity of 200 μA for 0.1 ms) and sensitivity to strychnine (10 μm). B, Input–output curves of evoked IPSCs in Glra1^+/+ (n = 24 from 12 mice) and Glra1^sh/sh (n = 25 from 12 mice) mice; comparison of Glra1^+/+ and Glra1^sh/sh at 100 μA: ***p = 3.198E-09; at 200 μA, ***p = 1.539E-12; at 300 μA, ***p = 3.20E-12; at 400 μA, ***p = 3.32E-12, t test). C, Traces from wild-type and Glra1^sh/sh neurons show postsynaptic current responses to glycine application (50 μm), recorded in TTX (1 μm). Note the increase in baseline noise with glycine in Glra1^sh/sh neuron, despite no shift in the holding current. Diagram on the right summarizes glycine-induced currents (Glra1^+/+, n = 7 from 3 mice; Glra1^sh/sh, n = 9 from 4 mice). D, Traces of spontaneously occurring mIPSCs, recorded in TTX (1 μm) from both genotypes, illustrate the reduced action potential-independent synaptic activity in Glra1^sh/sh mice. Averaged mIPSCs from the two neurons were superimposed and normalized, revealing the strong decrease in amplitude and acceleration in decay for Glra1^sh/sh mIPSCs. E, Quantitative analysis of spIPSCs (without TTX) and mIPSCs in PreBötC neurons demonstrates significant changes in properties and kinetics of spontaneously occurring events (spIPSCs: Glra1^+/+, n = 17 from 10 mice; Glra1^sh/sh, n = 21 from 12 mice; mIPSCs: Glra1^+/+, n = 6 from 3 mice; Glra1^sh/sh, n = 6 from 4 mice). *p < 0.05; **p < 0.01; ***p < 0.001.

Figure 7.

The shaky mutation does not affect GABAergic synaptic transmission in the PreBötC neurons. All recordings were made in the presence of KA (2 mm). The frequency of GABAergic spIPSCs was calculated by comparing and subtracting the spIPSCs before and after bicuculline (10 μm) application in the same neuron. PreBötC neurons from Glra1^sh/sh mice have GABAergic spIPSCs at a frequency of 0.76 ± 0.18 Hz (n = 12 from 9 mice), which is not different from that observed in Glra1^+/+ neurons (0.78 ± 0.19 Hz, n = 4 from 4 mice).

Figure 8.

Glra1^+/+ and Glra1^sh/sh do not differ in motoneuron numbers. Lumbar spinal cord of Glra1^+/+ and Glra1^sh/sh at P28 was cut in 12.5 μm slices, and the numbers of motoneurons were counted. Spinal cord slices were stained with cresyl violet, and motoneurons were determined due to their large cell body (white arrowheads). A, Representative pictures from Glra1^+/+ and Glra1^sh/sh slices with large cell bodies of motoneurons. B, Quantitative analysis of motoneuron numbers. The quantification was made using six to seven animals of each genotype (Glra1^+/+, n = 6; Glra1^sh/sh, n = 7). Error bars represent SDs.

Figure 9.

GlyR α1^Q177K disrupts hydrogen bonding with the key ligand binding residue R65. A, Cryo-EM model 3JAE with the wild-type GlyR α1 (α1^Q177, top right) and the modeled shaky mutation (α1^K177, bottom right). B, Top view of the homo-pentameric arrangement and position of Q177K. C, Glycine-binding pocket in the GlyR structure, where the glycine ligand is shown in magenta, and surrounding residues are labeled appropriately. D, Position of Q177 based on the cryo-EM model shown relative to the new position of R65 derived upon flexible fitting into the cryo-EM density using Flex-EM (Topf et al., 2008; Joseph et al., 2016). Interactions are indicated for Q177 with R65 from the glycine-binding pocket. E, The modeled Q177K substitution.