SLC4A10 mutation causes a neurological disorder associated with impaired GABAergic transmission

Abstract

SLC4A10 is a plasma-membrane bound transporter that utilizes the Na+ gradient to drive cellular HCO3- uptake, thus mediating acid extrusion. In the mammalian brain, SLC4A10 is expressed in principal neurons and interneurons, as well as in epithelial cells of the choroid plexus, the organ regulating the production of CSF. Using next generation sequencing on samples from five unrelated families encompassing nine affected individuals, we show that biallelic SLC4A10 loss-of-function variants cause a clinically recognizable neurodevelopmental disorder in humans. The cardinal clinical features of the condition include hypotonia in infancy, delayed psychomotor development across all domains and intellectual impairment. Affected individuals commonly display traits associated with autistic spectrum disorder including anxiety, hyperactivity and stereotyped movements. In two cases isolated episodes of seizures were reported in the first few years of life, and a further affected child displayed bitemporal epileptogenic discharges on EEG without overt clinical seizures. While occipitofrontal circumference was reported to be normal at birth, progressive postnatal microcephaly evolved in 7 out of 10 affected individuals. Neuroradiological features included a relative preservation of brain volume compared to occipitofrontal circumference, characteristic narrow sometimes 'slit-like' lateral ventricles and corpus callosum abnormalities. Slc4a10 -/- mice, deficient for SLC4A10, also display small lateral brain ventricles and mild behavioural abnormalities including delayed habituation and alterations in the two-object novel object recognition task. Collapsed brain ventricles in both Slc4a10-/- mice and affected individuals suggest an important role of SLC4A10 in the production of the CSF. However, it is notable that despite diverse roles of the CSF in the developing and adult brain, the cortex of Slc4a10-/- mice appears grossly intact. Co-staining with synaptic markers revealed that in neurons, SLC4A10 localizes to inhibitory, but not excitatory, presynapses. These findings are supported by our functional studies, which show the release of the inhibitory neurotransmitter GABA is compromised in Slc4a10-/- mice, while the release of the excitatory neurotransmitter glutamate is preserved. Manipulation of intracellular pH partially rescues GABA release. Together our studies define a novel neurodevelopmental disorder associated with biallelic pathogenic variants in SLC4A10 and highlight the importance of further analyses of the consequences of SLC4A10 loss-of-function for brain development, synaptic transmission and network properties.

Keywords: NBCN2; NCBE; acid-base; gamma aminobutyric acid; intellectual disability.

Figure 1

Family pedigrees and biallelic SLC4A10 variants. (A) Simplified family pedigrees for individuals affected with SLC4A10-related neurodevelopmental disorder, showing autosomal recessive segregation of SLC4A10 variants. Co-segregation confirmed in other family members as indicated, in each case the plus symbol indicating variant allele and the minus sign indicating wild-type allele. (B) Simplified SLC4A10 exon structure (NM_001178015.2) showing location of the multi-exon deletion identified in Individuals III:1 and III:2 (Family 1) and the splicing variant (c.2863-2A>C). Only a part of the large, non-coding, UTR exon 27 is shown. (C) Simplified SLC4A10 protein structure (Q5DTL9-1) showing location of missense (yellow) and predicted loss-of-function (red) variants in relation to the predicted domain architecture (Pfam domains: https://www.ebi.ac.UK/interpro/) of SLC4A10. Cytoplasmic = band 3 cytoplasmic domain (PF07565); HCO₃⁻ transporter = HCO₃⁻ transporter family (PF00955). (D) Multi-species alignments of SLC4A10, showing each of the missense variants identified in this study. aa = amino acids.

Figure 2

Neuroimaging from affected individuals with biallelic SLC4A10 variants. (A, E and I) Family 1, Subject III:1. T₂-weighted axial (A), T₁-weighted sagittal (E) and T₁-weighted coronal (I) MRI images of the patient at age 5 years. (B, F, J) Family 1, Subject III:2. T₂-weighted axial (B), T₁-weighted sagittal (F) and T₁-weighted coronal (J) MRI images of the patient at age 4 years. (C, G and K) Family 2, Subject II:1. T₂-weighted axial (C), T₂-weighted sagittal (G) and T₂-weighted coronal (K) MRI images of the patient at10 months of age. (D, H and L) Family 3, Subject II:3. T₂-weighted axial (D), T₁-weighted sagittal (H) and T₂-weighted coronal (L) MRI images at 1 year 2 months of age. In all cases lateral ventricles are small (A–D, arrowhead) with normal fourth ventricle (E–H), posterior fossa and external CSF spaces. In Family 1, the corpus callosum is dysmorphic, appearing thickened and flattened (E and F). This is associated with an unusual configuration of the fornix and septum pellucidum especially in Family 1, Subject III:1 (E, arrowhead). In Families 2 and 3 it is hypoplastic (G and H). Myelination is complete or adequate for age in all cases. Normal MRI brain images for comparison are available at https://www.imaios.com/en/e-Anatomy/Brain/Brain-MRI-in-axial-slices (adult) and https://radiopaedia.org/cases/normal-mri-head-3-years-old-1 (3-year-old child).

Figure 3

Slc4a10 ^−/− mice show behavioural abnormalities in the two-object novel object recognition task and display grossly intact cortical architecture. (A) The recognition of the novel object is altered in knockout (KO) mice. Top: Illustration of the two-object novel object recognition (NOR) test. Bottom: During the NOR test the exploration time, the number of visits for the old and the new, and the duration of these visits were quantified. A difference score (time exploring novel object − time exploring familiar object) and the discrimination ratio (time exploring the novel versus the familiar object) was calculated (nine mice per genotype, bootstrap t-test; *P < 0.05; **P < 0.01, ***P < 0.001). A negative value for the difference score or a value smaller 0.5 (dashed lines) for the discrimination ratio suggest that exploration is altered. (B) Top view of dissected brains from 12-month-old Slc4a10 wild-type (WT) and KO mouse. The weight of perfused and fixed brains of KO mice was smaller compared to WT (n = 5 mice per genotype; bootstrap t-test; **P < 0.01). Scale bar = 2 mm. (C) The gross architecture of the somatosensory cortex appeared intact in Slc4a10 KO mice. Sagittal brain sections from 2-month-old Slc4a10 WT and KO mice were stained for the pan neuronal marker NeuN and neurons counted layer wise (n = 3 mice per genotype; GEE model using normal errors identity link and independent working correlation matrix). Scale bar = 75 µm. Quantitative data are presented as mean + standard error of the mean (SEM). n.s. = not significant.

Figure 4

Localization of SLC4A10 to GABAergic presynapses. Slc4a10 wild-type (WT) mouse brain sections. Scale bars = 20 µm, enhanced view of merged marker images also shown (boxed areas). (A) VGLUT1, a marker of excitatory presynaptic terminals, rarely co-localizes with SLC4A10 in the CA1 region of the hippocampus (green: SLC4A10, red: VGLUT1). (B) SLC4A10 and VGAT, a marker for GABAergic presynapses, co-localize in the CA1 region of the hippocampus (green: SLC4A10, red: VGAT). (C) Quantitative analysis of co-localization of SLC4A10 with either VGLUT1 or VGAT and calculation of Pearson correlation coefficients between these data in the CA1 region of the hippocampus (VGLUT1 n = 28 and VGAT n = 39 images each, bootstrap t-test; ***P < 0.001).

Figure 5

SLC4A10 acts on presynaptic pH_i to promote GABA release in CA1 pyramidal neurons. Glutamatergic transmission is not impaired in CA1 neurons of Slc4a10^−/− mice (A–C). (A) Representative miniature excitatory postsynaptic current (mEPSC) recordings of pyramidal neurons from Slc4a10 wild-type (WT) and knockout (KO) mice. (B) Averaged mEPSCs show that the kinetics of mEPSCs are not affected by disruption of Slc4a10. (C) Cumulative plots and bar charts of different mEPSC properties. No significant differences were detected in mEPSC frequency, amplitude or kinetics (n = 9/14; bootstrap t-test). n.s. = not significant. (D–G) The miniature inhibitory postsynaptic currents (mIPSC) frequency is diminished in Slc4a10^−/− mice in the presence of bicarbonate. (D) Representative recordings of ongoing mIPSC activity in pyramidal neurons from Slc4a10 WT and KO mice as well of pyramidal neurons from Slc4a10 KO mice in the presence of 20 mM trimethylamine chloride (TriMA). (E) Averaged mIPSC recordings of pyramidal neurons from Slc4a10 WT and KO mice to illustrate kinetics and amplitude. (F) Cumulative plots and bar charts of mIPSC properties (n = 12/19/11; bootstrap F-test with post hoc analysis: *P < 0.05; **P < 0.01; ***P < 0.001; n.s. = not significant). While no differences in the mean amplitudes of mIPSCs were observed, the frequency of mIPSCs was significantly diminished in cells derived from Slc4a10 KO mice but could be partially rescued by application of TriMA. Diminished τ_decay and half-width of averaged mIPSCs in pyramidal neurons from Slc4a10 KO mice compared with WT in bicarbonate-buffered artificial CSF were not affected by TriMA. (G) In HEPES-buffered nominally bicarbonate-free solution mIPSC frequencies and kinetics did not differ between genotypes (n = 14/12; bootstrap t-test: *P < 0.05; **P < 0.01; ***P < 0.001). Quantitative data are shown as mean + standard error of the mean (SEM).