The severity of SLC1A2-associated neurodevelopmental disorders correlates with transporter dysfunction

Abstract

Background: Excitatory amino acid transporter 2 (EAAT2) is the predominant glutamate transporter and a key mediator of excitatory neurotransmission in the human brain. Here we present a cohort of 18 individuals harbouring 13 different SLC1A2 variants, who all present with neurodevelopmental impairment with variable symptoms and disease severities, and we delineate the impact of these variants on EAAT2 function.

Methods: The consequences of nine novel missense SLC1A2 variants for expression, transport and anion channel properties of EAAT2 expressed in mammalian cells were characterized by confocal microscopy, enzyme-linked immunosorbent and [³H]-D-aspartate uptake assays, and electrophysiological recordings.

Findings: Ten of the 13 SLC1A2 variants mediated significant changes to EAAT2 expression and/or function. These molecular phenotypes were classified into three categories: overall loss-of-function (F249Sfs∗17, A432D, A439V, c.1421+1G>C), mild gain-of-anion-channel function (I276S, G360A), and mixed loss-of-transport/gain-of-anion-channel function (G82R, L85R, L85P, P289R). In contrast, L37P, H542R and I546T did not mediate significant changes to EAAT2 expression or function. Although specific clinical outcomes in individuals carrying variants within each category varied somewhat, the three categories overall translated into distinct clinical phenotypes in terms of phenotypic traits and severity.

Interpretation: The observed associations between functional effects and clinical phenotypes produced by these variants offer valuable insights for future predictions of progression and severity of SLC1A2-associated neurodevelopmental disorders. Furthermore, these associations between variant-induced changes in EAAT2 function and phenotypic traits could assist in tailoring personalized treatments of these disorders.

Funding: This work was funded by the German Ministry of Education and Research and by the Lundbeck Foundation.

Keywords: Epilepsy; Excitatory amino acid transporter 2 (EAAT2); Genetic variants; Patch clamp; Phenotyping; SLC1A2; Transporter uptake.

Fig. 1

Gene structure of SLC1A2 and the positions of the SLC1A2 missense variants in the EAAT2 protein. (a) Genomic assembly of SLC1A2 with positions of missense base exchanges within exons (I-XI). The gene structure was extracted from SLC1A2 sequence information by ensemble.org (GRCh38.p14, ENST00000278379.9). Base pair numbering indicates starting regions of exons downstream to the gene starting site (denoted as #). (b) Extracellular view of the 3-dimensional structure of the EAAT2 trimer derived from pdb structure 7XR4 (www.rcsb.org).^, The dashed lines indicate the limits for each monomer. (c, d) 2-dimensional transmembrane topologies of the fully (c) or truncated (d) EAAT2 monomers with indicated positions of the SLC1A2 missense variants. Regions forming the trimerization and the transport domains in the EAAT2 monomer are shown in blue and red, respectively. Multiple sequence alignment of stretches from the five human EAAT subtypes covering the nearest amino acid residues surrounding the mutated residues in sequential order. Regions within helices are highlighted by colour, as in b and c. Conserved residues are given in bold and indicated with asterisks. (a, c) Variants that were functionally characterized in previous studies^,, , are indicated with a hash.

Fig. 2

Expression properties exhibited by WT EAAT2 and nine EAAT2 variants in confocal microscopy. (a) Representative confocal microscopy images of HEK293T cells expressing WT or variant EAAT2 fused to mYFP. (b) Confocal microscopy images of HEK293T cells co-expressing WT EAAT2 fused to mCherry and variants L85R-mYFP, A432D-mYFP or F249Sfs∗17-mYFP.

Fig. 3

Expression properties exhibited by WT EAAT2 and nine EAAT2 variants in an ELISA. Cell surface and total expression levels displayed by HA-tagged versions of WT and variant EAAT2 transiently expressed in COS-7 cells in an ELISA. (a) Expression levels displayed by HA-tagged WT and variant EAAT2 in cells transfected with three different cDNA amounts (20, 67 and 200 ng EAAT2 cDNA of a total cDNA quantity of 200 ng per 48w-well). Data are given as mean ± S.E.M. in % of the cell surface expression level of WT HA-EAAT2 (200 ng cDNA/48w-well) and are based on three independent experiments (n = 3) performed in triplicate. The mean cell surface expression levels of WT HA-EAAT2 at the three different cDNA transfection levels are indicated with grey hatched lines. Statistically significant differences between WT HA-EAAT2 vs. variant HA-EAAT2 expression at each of the three different transfection levels are indicated (black for cell surface expression, grey for total expression). (b) Expression levels displayed by HA-tagged L85R, A432D and F249Sfs∗17 EAAT2 expressed in cells on their own (100 ng variant HA-EAAT2 cDNA and 100 ng “empty” pCDNA3.1 per 48w-well) or together with (untagged) WT EAAT2 (100 ng variant HA-EAAT2 cDNA and 100 ng WT EAAT2 cDNA per 48w-well). Data are given as mean ± S.E.M. in % of the cell surface expression level of the HA-tagged variant on its own (indicated with a grey hatched line) and are based on three independent experiments (n = 3) performed in triplicate. Statistically significant differences between HA-EAAT2 variant expression on its own vs. co-expressed with untagged WT EAAT2 are indicated. (c) Expression levels displayed by HA-tagged WT EAAT2 co-expressed with untagged WT or variant EAAT2 in cells transfected with two different cDNA amounts (25:25 ng and 100:100 ng WT HA-EAAT2:untagged EAAT2 cDNA of a total cDNA quantity of 200 ng per 48w-well). Data are given as mean ± S.E.M. in % of the cell surface expression level of WT HA-EAAT2 co-expressed with WT EAAT2 (100 ng:100 ng cDNA/48w-well) and are based on four independent experiments (n = 4) performed in triplicate. The mean cell surface expression levels of WT HA-EAAT2 co-expressed with untagged WT EAAT2 at the two different cDNA transfection levels are indicated with grey hatched lines. Statistical analysis did not identify any statistically significant differences between WT HA-EAAT2 expression when co-expressed with WT EAAT2 vs. when co-expressed with variant EAAT2. (a–c) Statistical analyses were performed using one-way ANOVA followed up by Tukeys multiple comparisons post hoc test. All statistically significant differences are indicated by asterisks and P values, with significance levels: ∗P ≤ 0.05/∗∗P ≤ 0.01/∗∗∗P ≤ 0.001/∗∗∗∗P ≤ 0.0001.

Fig. 4

Transport properties exhibited by WT EAAT2 and nine EAAT2 variants. The functional properties of WT and variant transporters fused to YFP transiently expressed in COS-7 cells were characterized in a [³H]-D-Asp uptake assay. (a) Saturation D-Asp uptake exhibited by “homozygous” WT and variants in the [³H]-D-Asp uptake assay in cells transfected with 100 ng EAAT2 cDNA/96w-well. Data are from representative individual experiments and are given as mean ± S.D. in % of the fitted V_max exhibited by WT EAAT2 on the same 96-well plate. The average K_m and V_max values for D-Asp at WT and variant EAAT2 are given in Table 1. (b) Plot of the D-Asp V_max values of “homozygous” WT and variant EAAT2 (mean ± S.D., normalized to WT EAAT2 D-Asp V_max, Table 1) against the cell surface expression levels of HA-tagged WT and variant EAAT2 (mean ± S.D., normalized to WT EAAT2 cell surface expression, Fig. 3a). The dashed line indicates the extrapolated mean values of D-Asp uptake starting from zero expression and the grey shaded area indicates the extrapolated S.D. for the uptake at every expression level. Data in the plot are from Table 1 and Fig. 3a. (c) Concentration-inhibition relationships exhibited by D-Asp (left) and L-Glu (right) at “homozygous” WT and variant EAAT2 in the [³H]-D-Asp uptake assay. Data are from representative individual experiments and are given as means ± S.D. in % of the specific [³H]-D-Asp uptake in the absence of test compound. The average K_i values for D-Asp and L-Glu in their competition inhibition of [³H]-D-Asp uptake at WT and variant EAAT2 are given in Table 1. (d) Saturation D-Asp uptake exhibited by “heterozygous” WT/variant EAAT2 combinations in the [³H]-D-Asp uptake assay in cells (co-transfected with 50 ng WT EAAT2 cDNA and 50 ng WT or variant EAAT2 cDNA per 96w-well). Data for the WT::WT (100 ng WT EAAT2 cDNA/96w-well; open squares, black solid line) and WT:vector (50 ng WT EAAT2 cDNA/96w-well; full/open squares, black dashed line) controls are included. Data are from a representative individual experiment and are given as mean ± S.D. in % of the fitted V_max exhibited by “homozygous” WT EAAT2 (100 ng cDNA/96-w well). The average K_m and V_max values for D-Asp at the various combinations are given in Table 1.

Fig. 5

L85R, A432D and A439V EAAT2 variants cause loss-of-anion-channel or robust gain-of-anion-channel function. (a) Representative whole-cell patch clamp recordings from HEK293T cells expressing WT, L85R or A439V EAAT2 and from cells co-expressing WT with L85R or A432D EAAT2 (WT::L85R, WT::A432D, 1:1 cDNA ratios) in the absence (control, top) or in the presence of 0.5 mM L -Glu (bottom). (b, c) Mean whole-cell current-voltage relationships in the absence (b) or presence (c) of 0.5 mM L-Glu (−150 mV: WT: wo L-Glu: −21.4 ± 4, with L-Glu: −62.1 ± 11.1 pA·pF⁻¹, n = 10/9; L85R: wo L-Glu: −14 ± 5.8; with L-Glu: −7.2 ± 2.1 pA·pF⁻¹, n = 9/8; A439V: wo L-Glu: −22.6 ± 4.6, with L-Glu: −15.2 ± 1.4 pA·pF⁻¹, n = 11/5; WT::L85R: (−150 mV: wo L-Glu: −390.8 ± 125.2, with L-Glu: 131.1 ± 27, pA·pF⁻¹, n = 10/10); WT::A432D: wo L-Glu −33.9 ± 3.9; with L-Glu: −62,4 ± 11,8 pA·pF⁻¹, n = 14/13). (d) Statistical analysis of whole cell current densities at −150 mV for WT and variants using one-way ANOVAs for each condition and Holm-Sidak post hoc testing (^#the value for comparison of WT to WT::A432D is shown for a single comparison with a two-sample two-tailed t-test with P = 2.92743 × 10⁻⁴). Box plots show upper and lower quartiles with data medians and all error bars and whiskers span 95% confidence intervals of current amplitudes.

Fig. 6

L85R variant renders the EAAT2 anion channel permeable for L-Glu. (a) Representative whole-cell recordings of HEK293T cells expressing WT EAAT2 alone (top) and co-expressing WT and L85R EAAT2 (bottom). (b) Current-voltage relationships of gluconate currents from WT EAAT2 and WT co-expressed together with L85R under ionic conditions as in a. (c) Whole-cell recordings of HEK293T cells co-expressing WT and L85R EAAT2 at indicated ion compositions with external Na-gluconate and internal K-gluconate with Na⁺-L-Glu_int (top) or K⁺-L-Glu_int (bottom). (d) Current-voltage relationships recorded from WT and L85R under biionic conditions (insert) with internal L-Glu and external Cl⁻ show prominent inward currents for L85R EAAT2, but not for WT EAAT2. The reversal potentials indicate that L85R anion channels are permeable for L-Glu (P_L-Glu/Cl = 0.095 ± 0.013 with Na⁺_int, n = 5 and 0.103 ± 0.042 with K⁺_int, n = 6). Supplementing the external solution with 0.5 mM L-Glu did not modify the observed reversal potentials (P > 0.55^Na+int/K+int). Errors represent 95% confidence intervals of current amplitudes or relative permeabilities L-Glu⁻/Cl⁻.

Fig. 7

I276S and G360A variants increase EAAT2 anion currents. (a) Representative whole-cell patch clamp recordings from HEK293T cells expressing I276S or G360A EAAT2 in the absence (control, top) or in the presence of L-Glu (bottom) (b) Mean current-voltage relationships from whole cell recordings as shown in Fig. 6a in the absence (b) or presence (c) of 0.5 mM L-Glu (I276S: wo L-Glu: −27.5 ± 5.7 pA·pF⁻¹, with L-Glu: 93.2 ± 16.5 pA·pF⁻¹, n = 15/14; G360A: wo L-Glu: −27.2 ± 4.5 pA·pF⁻¹, with L-Glu: −103.9 ± 19.5 pA·pF⁻¹, n = 15/10). (d) Statistical analysis of whole cell current densities at −150 mV for WT and variants using one-way ANOVAs for each condition and Holm-Sidak post hoc testing. Box plots show upper and lower quartiles with data medians, and all error bars and whiskers span 95% confidence intervals of current amplitudes.

Fig. 8

L37P, H542R and I546T variants do not alter EAAT2 anion channel function. (a) Representative whole-cell patch clamp recordings from HEK293T cells expressing L37P, H542R or I546T EAAT2 in the absence (control, top) or in the presence of 0.5 mM L-Glu (bottom). (b, c) Mean current-voltage relationships from whole cell recordings as shown in Fig. 7a in the absence (b) or presence (c) of 0.5 mM L-Glu (L37P: wo L-Glu: −21.8 ± 5.5, with L-Glu: −62.9 ± 17.4 pA·pF⁻¹, n = 10/8; H542R: wo L-Glu: −18.3 ± 3.2, with L-Glu: −66.9 ± 12.2 pA·pF⁻¹, n = 10/9; I546T: wo L-Glu: −22.7 ± 6.9, with L-Glu: −61.8 ± 11.2 pA·pF⁻¹, n = 11/10). (d) Statistical analysis of whole cell current densities at −150 mV for WT and variants using one-way ANOVAs for each condition and Holm-Sidak post hoc testing. Box plots show upper and lower quartiles with data medians and all error bars and whiskers span 95% confidence intervals of current amplitudes.

Fig. 9

Correlation analyses of WT and variant EAAT2 functional properties. (a) Plot of mean anion current densities at −150 mV (from Figs. 5d, 7d and 8d) vs. normalized cell surface expression levels of the WT and variant transporters (from Fig. 3a) (b) Plot of mean anion current densities at −150 mV (from Figs. 5d, 7d and 8d) vs. the D-Asp V_max values for the WT and variant transporters (from Table 1). The dashed lines indicate the extrapolated mean values of X-axis values starting from zero and the grey shaded area indicates the extrapolated S.D. for the X-axis values for every level.

Supplementary figs1

Supplementary Fig. 1. Electrophysiological analyses of transport properties of loss-of-function EAAT2 variants F249Sfs∗17 and A432D co-expressed with WT EAAT2. (a) Representative confocal images from cells expressing WT EAAT2-mCherry together with EAAT2 variants F249Sfs∗17 or A432D fused to mYFP (scale bars: 10 μm). (b) Current-voltage relationships of L-Glu transport currents from cells expressing WT EAAT2-mCherry alone (black circles) or co-expressed with F249Sfs∗17-mYFP (crossed stars) or A432D-mYFP (green diamonds). For comparison the dashed line shows the mean current densities for WT EAAT2-mYFP (calculated from an earlier publication, available at https://github.com/peterkovermann/epileptic_encephalopathy_1). (c) Plots of transport current densities against mean mCherry fluorescence for WT-mCherry alone (black circles) and co-expressed with variants F249Sfs∗17 (left) and A432D (right). Data were fitted with solid (WT) or dashed (variants) lines originating at zero. The shaded areas depict the 95% confidence intervals for the fits. (d) Box plots compare current densities at −123 mV (left) and mean mCherry fluorescence (right) for all tested cells (one-way ANOVAs, with Holm-Sidak post hoc testing). Whiskers indicate 95% confidence levels, squares indicate the mean values. Due to non-normality of a single data set (fluorescence of WT::F249Sfs∗17) these data were additionally compared to WT EAAT2 with a Mann-Whitney U test, with equal result.

Supplementary figs2

Supplementary Fig. 2. Relations of WT EAAT2-current amplitudes from L-Glu uptake and anion currents observed with the permeable anions Cl⁻ or NO₃⁻. Current-voltage relationships of L-Glu transport currents (in mM: 140 Na-gluc_ext/115 K-gluc_int, solid black line, n = 14), and currents in the presence of the permeable anions Cl⁻ (140 NaCl_ext/115 KCl_int, dashed red line, n = 19) or NO₃⁻ (140 NaNO_3,ext/115KNO_3,int, dotted blue line, n = 9). Graph shows mean current-voltage relationships with 95% confidence intervals.

Supplementary figs3

Supplementary Fig. 3. Comparison of Cl⁻ current densities from WT, L85P and WT::L85R EAAT2. Boxplots show the distributions of Cl⁻ current densities with medians (lines) for cells expressing WT EAAT2 (black circles, n = 19), L85P EAAT2 (blue circles, n = 10), pr co-expressing WT and L85R EAAT2 (magenta, n = 11, one-way ANOVAs, with Holm-Sidak post hoc testing). Whiskers indicate 95% confidence levels.

Supplementary figs4

Supplementary Fig. 4. Correlation analyses of functional properties from WT, G82R, L85P, and P289R EAAT2 from a previously published study. (a, b) Plots of transport-activated anion currents at −150 mV of the recently characterized EAAT2 variants G82R, L85P, and P289R (denoted with #) vs. normalized cell surface expression (a) and vs. normalized transport currents (b). The plots are based on data shown in Kovermann et al. 2022, and raw data for analyses for these data are freely available at https://github.com/peterkovermann/epileptic_encephalopathy_1. The dashed lines indicate the extrapolated mean values of X-axis values starting from zero and the grey shaded area indicates the extrapolated S.D. for the X-axis values for every level.