A point mutation in the ion conduction pore of AMPA receptor GRIA3 causes dramatically perturbed sleep patterns as well as intellectual disability

Abstract

The discovery of genetic variants influencing sleep patterns can shed light on the physiological processes underlying sleep. As part of a large clinical sequencing project, WGS500, we sequenced a family in which the two male children had severe developmental delay and a dramatically disturbed sleep-wake cycle, with very long wake and sleep durations, reaching up to 106-h awake and 48-h asleep. The most likely causal variant identified was a novel missense variant in the X-linked GRIA3 gene, which has been implicated in intellectual disability. GRIA3 encodes GluA3, a subunit of AMPA-type ionotropic glutamate receptors (AMPARs). The mutation (A653T) falls within the highly conserved transmembrane domain of the ion channel gate, immediately adjacent to the analogous residue in the Grid2 (glutamate receptor) gene, which is mutated in the mouse neurobehavioral mutant, Lurcher. In vitro, the GRIA3(A653T) mutation stabilizes the channel in a closed conformation, in contrast to Lurcher. We introduced the orthologous mutation into a mouse strain by CRISPR-Cas9 mutagenesis and found that hemizygous mutants displayed significant differences in the structure of their activity and sleep compared to wild-type littermates. Typically, mice are polyphasic, exhibiting multiple sleep bouts of sleep several minutes long within a 24-h period. The Gria3A653T mouse showed significantly fewer brief bouts of activity and sleep than the wild-types. Furthermore, Gria3A653T mice showed enhanced period lengthening under constant light compared to wild-type mice, suggesting an increased sensitivity to light. Our results suggest a role for GluA3 channel activity in the regulation of sleep behavior in both mice and humans.

Figure 1

Actograms for two brothers with a GRIA3(A653T) mutation. (A) Periods of sleep for the two brothers (labelled as 1 and 2 to match periodograms in (B) displayed over 27–28 days, with hourly resolution, displayed as a double-plotted actograms. Each row represents two 24-h periods, with each day plotted twice, first in the second half of one row and then in the first half of the next row. Black indicates periods of reported sleep. The numbers on the Y-axis refer to the days of the record. (B) Chi-Square periodograms of sleep from the data in panel A shows multiple peaks at 56-90h, with a smaller peak at 24 h in brother 2 (the younger brother). Qp is the Chi-square statistic for a given period (a measure of power of rhythmicity at that period). The diagonal line represents the level above which results would be considered significant (P < 0.01).

Figure 2

The A653T GRIA3 mutation relative to the Lurcher mutation. (A) Structure of a prototypical AMPAR, specifically a GluA2 homotetramer (PDB ID 3KG2) (66). Pore proximal and pore distal subunits are colored in dark and light blue respectively. The layered AMPAR architecture consists of the extracellular amino-terminal domain (ATD) and the ligand binding domain (LBD), followed by the trans-membrane domain (TMD) and intracellular “tails”, disordered and not present in the structure. Residue A621 (3KG2 numbering), equivalent to A653 in GluA3, is colored in red (side chain atoms shown as spheres). (B) Sequence alignment of transmembrane helix III (M3) from different human (h) and mouse (m) glutamate receptor proteins. The mutated alanine A653 (red) falls within a highly conserved transmembrane motif that lines the ion channel pore. In GluD2, encoding a different glutamate receptor subunit, the neighbouring alanine residue (cyan) is mutated to threonine in the Lurcher mouse (A654T). (C) Side view of the four-fold symmetric channel pore, with A653 indicated in red. For clarity, two subunits have been removed. (D) Top view of the channel gate, with A653 in red and A654 in cyan. (E) Top view of the channel gate, with threonine residues modelled at the alanine sites, to illustrate the potential steric impact of the mutations. While A654T points towards a neighboring subunit, favoring pore opening, A653T leads to pore occlusion.

Figure 3

The GluA3 A653T mutation does not interfere with channel trafficking. (A) Raw FACS profiles demonstrating membrane localization of GluA3 (left panel) and GluA3(A653T) (right panel) in transfected HEK293T cells (duplicate experiments shown by green and orange traces, untransfected and unstained cells shown by red traces and untransfected and stained cells shown by blue traces). Geometric mean expression values are shown in the histogram and were compared with a t-test on the log of the observations; error bars: standard error of the mean (S.E.M.). (B) Immunohistochemical staining of Cos7 cells transfected with GluA3 (left panel) and GluA3(A653T) (right panel) showing membrane localization of the receptor.

Figure 4

GluA3 with the A653T mutation displays reduced sensitivity to partial agonist kainate. (A) Example traces of current responses of GluA3 homomers or GluA3(A653T) homomers; outside-out patches from HEK293T cells stably expressing γ-2, voltage-clamped at -60 mV) to 10 mM l-glutamate (black) and 500 μM kainate (grey). (B) Summary graph for recordings shown in (A). The ratio of peak responses to kainate and glutamate (KA/Glu) was 50 ± 7% (n = 20) and 7 ± 1% (n = 20) for GluA3 and GluA3(A653T), respectively, and 28 ± 4% (n = 14) and 6 ± 2% (n = 13) for GluA2/GluA3 heteromers and GluA2/GluA3(A653T) heteromers, respectively. ****P < 0.0001 (unpaired t-test with Welch`s correction). (C) Concentration-response characteristics of GluA3 homomers or GluA3(A653T) homomers (in the presence of TARP gamma 2). Peak current responses to 100 ms application of 0.001–30 mM l-glutamate were fitted with a three-parameter dose-response curve (see Methods for details), yielding EC50 values of 0.53 mM and 1.65 mM for GluA3 homomers and GluA3(A653T) homomers, respectively.

Figure 5

Generation of the Gria3^A653T mouse model. (A) Bottom panel shows the wild-type Gria3 sequence with the position of the target amino-acid shown in bold. The protospacer target used for the CRISPR-Cas9 nuclease design is highlighted in grey. Top panel shows detail of the ssODN used to introduced the desired amino-acid change, along with silent mutations (underlined) to mark the recombined allele with an AluI restriction site and mutate the Protospacer Adjacent Motif. (B) Top panel shows the genotyping amplicon obtained from the Gria3^A653T and wild-type Gria3 alleles with the position of the AluI restriction enzymes marker. Lower panel shows an example of the digestion of the amplicon obtained from heterozygous Gria3^A653T/WT (A653T/WT) female, hemizygous Gria3^A653T (A653T/Y) and wild-type (WT/Y) male mice. (C) Top panel shows the genomic structure of Gria3 with the position of the primers used for RT-PCR analysis marked. Middle panel shows the pattern of AluI restriction sites in the amplicons obtained from wild-type and A653T cDNA. Lower panel shows the AluI digestion products obtained following ampicon digestion obtained from hemizygous Gria3^A653T (A653T/Y) and wild-type (WT/Y) male mice. (D) Micrographs of Nissl stained sagittal brain sections from wild-type and hemizygous Gria3^A653T mice in regions of the brain where Gria3 is highly expressed. Scale bars: 200 μm.

Figure 6

Gria3 ^A653T mice show impaired synaptic transmission. (A) Example traces of AMPAR mEPSC recordings from cerebellar Purkinje cells of wild type (WT) and Gria3^A653T mice. (B) Average mEPSC amplitude per cell (WT: 13.44 ± 1.06 (n = 10 cells); Gria3^A653T: 11.09 ± 0.876 (n = 9); unpaired t-test, P = 0.111). (C) Cumulative frequency distributions of a representative subset of mEPSC amplitudes show a reduction in large amplitude mEPSCs in Gria3^A653T mice. Frequency distributions, when fitted with a skewed normal curve (not depicted), are significantly different between populations (F-test, P = 0.0001, F(3, 36) = 9.18). (D) mEPSC frequency was significantly lower in Gria3^A653T mice when compared to WT (WT: 0.37 ± 0.03 (n = 10); Gria3^A653T: 0.23 ± 0.05 (n = 9); unpaired t-test, P = 0.0467).

Figure 7

Activity and sleep in Gria3^A653T mice and wild-type littermates. (A) The number of bouts of activity in Gria3^A653T mice and wild-type (WT) mice (mean ± S.E.M.), separated into the average activity within the bout (intensity, upper panel) and into bout-lengths (lower panel) of <1 min, 1–10 min, 10 min to 1 h and > 1 h (mean ± S.E.M.). *P = 0.0218 for bouts of activity <1 min, 2-way RM ANOVA, with Bonferroni correction. (B) The number of bouts of sleep in Gria3^A653T mice and wild-type littermates, separated into lengths of <1 min, 1–10 min, 10 min to 1 h and > 1 h (mean ± S.E.M.). **P = 0.0049 for sleep bouts <1 min, 2-way RM ANOVA, with Bonferroni correction. (C) The 24 h patterns of activity (upper panel) and sleep (hourly mean ± S.E.M.), from 7 days of 12:12 L:D (n = 6 mice for each genotype in all panels). (D) The circadian rhythmicity in constant darkness of wild-type and Gria3^A653T mice (mean ± S.E.M.).

Figure 8

Actigraphy in constant lighting conditions suggests increased sensitivity of the retino-hypothalamic tract in Gria3^A653T hemizygous male mice. (A) Average Tau values from the chi-square periodograms of all mice in constant darkness and three different levels of environmental illumination (mean ± S.E.M.). **P = 0.0032 for 100 Lux, 2-way RM ANOVA, with Bonferroni correction (n = 6 mice for each genotype). (B) Average daily activity levels of all mice in 12:12 Light:Dark, constant darkness and three different levels of environmental illumination (mean ± S.E.M.). Individual values for each mouse were adjusted based on the internal Tau under that lighting condition before averaging (24/Tau x daily average activity). See Supplementary Material, Figure S2 for actograms showing activity patterns.