1 INTRODUCTION

Epileptic encephalopathies (EEs) are severe epilepsy syndromes characterized by early onset, multiple seizure types, and abundant epileptiform electroencephalography (EEG) activity combined with developmental slowing or even regression. Recently, mutations in SLC1A2 (solute carrier protein member 1A2), the gene encoding the l-glutamate transporter EAAT2 (excitatory amino acid transporter 2), were associated with EE in three different studies.1-3 All patients exhibited severe developmental and EE, with up to 50 seizures daily under anti-epileptic therapy, profound global delay, and progressive brain atrophy.

EAAT2 is the major l-glutamate transporter of mature astrocytes4-6 and is also expressed in the presynaptic nerve terminal.7, 8 Its genetic ablation in mice results in lethal spontaneous seizures likely caused by increased extracellular l-glutamate concentrations due to reduced glial l-glutamate uptake.9 However, EAAT2 is not only a secondary active glutamate transporter but also an anion channel.10-12 Anion channel dysfunction in the related EAAT1 was recently shown to contribute to the pathophysiology of episodic ataxia 6.13-16

We here study the functional consequences of three SLC1A2 variants associated with EE: c.244G>A (p.Gly82Arg/G82R), c.254T>C (p.Leu85Pro/L85P), and c.866C>G (p.Pro289Arg/P289R) in hEAAT2 (NM_004171.3)1-3 through whole-cell patch clamping after heterologous expression in mammalian cells. All mutations were identified as de novo and pathogenic. G82R has been identified in three unrelated individuals, whereas L85P and P289R were found each in one individual.1, 2 All tested mutations increased anion currents and decreased l-glutamate uptake. Two mutations severely affected the size selectivity of the EAAT2 anion channel, permitting diffusion-mediated l-glutamate release.

2 MATERIAL AND METHODS 2.1 Heterologous expression of wild-type and mutant EAATs

The coding region of human EAAT2 (kindly provided by Dr. M. Hediger, Universität Bern, Switzerland) was subcloned into pRcCMV using polymerase chain reaction (PCR)–based strategies. Point mutations and fusions to monomeric yellow fluorescent protein (mYFP), were generated by PCR-based techniques. Transient transfection of HEK293T cells using the Ca3(PO4)2 technique was performed as described previously.17

2.2 Knock-down of LRCC8A in the HEK293T cells line

Knock-down of human LRRC8A (gene encoding for: leucine-rich repeat-containing 8 VRAC subunit A) in the embryonic kidney cell line HEK293T was performed with Crispr/Cas technique (Appendix S1).18

2.3 Electrophysiology

Standard whole-cell patch clamp recordings were performed using an Axopatch 200B amplifier (Molecular Devices) as described previously described.19, 20 Cells were clamped to 0 mV or −70 mV for at least 5 s between test sweeps. Junction potentials were corrected a priori. Standard bath solution contained (in mM): 140 NaCl, 4 KCl, 2 CaCl2, 1 MgCl2, 10 HEPES, ±0.5 Na-glutamate, pH 7.4, and standard pipette solutions (in mM): 115 KCl or 115 NaCl, 2 MgCl2, 5 EGTA, 10 HEPES, pH 7.4. In all experiments with K+ in the pipette solution, K+-outward currents were blocked by 5 mM TEA-Cl in the bath solution. Glutamate transport currents were calculated by subtracting currents before and after addition of 5 mM external l-glutamate after substitution of Cl− with d-gluconate in internal and external solutions. In the experiments with d-gluconate–based internal and external solutions, the bath solution contained 140 Na-gluconate, 2 Ca-gluconate, 1 Mg-gluconate, 5 TEA-Cl, 10 HEPES/NaOH, pH 7.4, and the intracellular solution 115 K-gluconate, 2 Mg-gluconate, 5 EGTA, 10 HEPES/KOH, pH 7.4. Because some of the disease-causing mutations increased the d-gluconate permeability, we additionally measured uptake currents as l-glutamate-dependent current increases at the calculated d-gluconate reversal potential (−5.8 mV). d-gluconate and l-glutamate permeabilities were tested either with symmetric d-gluconate–based solutions (140 Na-gluconate/±5 Na-glutamateext and 115 K-gluconateint at pH 7.4) or in bi-ionic measurements (140 NaCl/±0.5 Na-glutamateext, and 115 Na-gluconateint or 115 Na-glutamateint at pH 7.4).

2.4 Confocal microscopy and biochemical characterization of EAAT fusion proteins

For confocal imaging HEK293T cells expressing mYFP-hEAAT2 fusion proteins were plated on poly-l-lysine coated coverslips 24 h after transfection.16 Images were acquired with a Zeiss LSM780 inverted microscope using a 63×/1.40 NA oil immersion objective in phosphate-buffered saline at room temperature 48 h after transfection. YFP was excited with an argon laser (488 nm), and emission was detected between 543 and 594 nm. Images were analyzed with FIJI image analysis software.21

For sodium-dodecyl sulfate polyacrylamide gel electrphoresis (SDS-PAGE) cells were solubilized in 0.4% dodecyl-maltoside for 30 min on ice, transferred 12 h, and then centrifuged for 35 min at 15,781 g at 4°C).16 The concentration of the protein in the supernatant was determined with the bicinchoninic acid assay kit (BCA assay kit, Abcam, ab102536). Expression levels and glycosylation states were estimated as described (Appendix S1).16, 22

2.5 Statistical methods

Data were analyzed with a combination of Clampfit (Molecular Devices), SigmaPlot 12.3 (Jandel Scientific), OriginPro 2018G (OriginLab), and Excel (Microsoft Corp.) programs. Current-voltage relationships were constructed by plotting macroscopic current amplitudes determined 150 ms after the voltage step vs the membrane potential. All data are given as means ± 95% confidence intervals (CIs) unless otherwise stated. For statistical analysis of data, one-way or two-way analyses of variance (ANOVAs), or Kruskal-Wallis tests were used with Holm-Sidak post hoc tests with significance levels *p ≤ .05/**p ≤ .01/***p ≤ .001, or Dunn's post hoc testing with a threshold of #p = .05. The data that support the findings of this study are openly available in GitHub at https://github.com/peterkovermann/epileptic_encephalopathy_1.

3 RESULTS 3.1 EE-associated SLC1A2 mutations are located in proximity to the EAAT anion channel

EAAT glutamate transport is based on a large-scale rotational-translational movement of the substrate-harboring transport domain,23-25 and the EAAT anion channel is transiently formed in intermediate positions at the interface between trimerization and transport domains.11, 26, 27 Two disease-associated mutations (G82, P289) exchange residues within the trimerization domain that are conserved across mammalian EAATs as well as in mammalian neutral amino acid transporters and bacterial homologs. The remaining one, L85, is only conserved within mammalian SLC1 proteins (Figure 1A). Figure 1B depicts the localization of the three EE-associated mutations, G82R, L85P, and P289R, in an open anion channel EAAT2 model, based on the channel conformation of the prokaryotic sodium-aspartate symporter from Pyrococcus horikoshii (GltPh), generated.11 L85 is homologous to F50 in GltPh, which projects into the open anion pore in this model,11 and the substitution of L85 with a negatively charged residue (L85D) was shown to generate partially cation-permeable EAAT2 anion channels.11 G82 and L85 are both parts of the second transmembrane helix so that R82 likely also projects into the anion conduction pathway. P289 is at the kink of transmembrane helix 5, close to G82 and L85, but not in direct contact with the aqueous conduction pathway. The corresponding mutation P290R has been extensively studied in hEAAT1 and shown to cause a pronounced gain-of-function of the EAAT1-associated anion channel.14, 15, 28

Epileptic encephalopathy (EE)–associated missense mutations affect human excitatory amino acid transporter 2 (hEAAT2) expression and subcellular trafficking. (A) Positions of EE-associated mutations p.Gly82Arg (NM_004171.3), p.Leu85Pro (NM_004171.3), and p.Pro289Arg (NM_004171.3) in hEAAT2 (trimerization/transport domains: blue/magenta). (B) Homology modeling–derived structure of hEAAT2 mapped onto the open channel structure from the archaeal aspartate transporter GltPh show proximity of affected residues to the anion pore.12 (C) Confocal images from HEK293T cells expressing WT and mutant hEAAT2 fused to yellow fluorescent protein (YFP, scale bars: 10 µm). (D) Relative surface fluorescence of cells expressing mutant hEAAT2-YFP. (E) Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) analyses from extracted YFP-fusion proteins. (F) Distributions of band intensities from SDS-PAGEs as shown in E. Proportions of glycosylation states in F were analyzed with one-way analysis of variance (ANOVA) for each set of bands separately with WT as control, and Holm-Sidak post hoc testing (*p ≤ .05, **p ≤ .01, ***p ≤ .001). Color code: WT/G82R/L85P/P289R: black/red/blue/green 3.2 EE-associated point mutations affect subcellular trafficking of hEAAT2

Disease-causing mutations often decrease protein expression/stability or modify the subcellular distribution of affected proteins. We expressed wild-type (WT) and mutant hEAAT2 as mYFP fusion proteins transiently in mammalian cells and studied their subcellular localization with confocal imaging as well as biochemical analysis.14, 22, 29 Confocal images showed almost exclusive insertion of WT and L85P mYFP-hEAAT2 into the surface membrane or in domains in proximity (Figure 1C). In contrast, the expression of G82R and P289R mYFP-hEAAT2 resulted in fluorescent staining of intracellular regions. Figure 1D depicts intensities of membrane-localized fluorescences of individual transfected cells expressing mYFP-tagged EAAT2 variants normalized to the mean membrane-localized fluorescences of cells expressing mYFP-tagged WT transporter.

Membrane protein glycosylation is required for exit from the endoplasmic reticulum (ER), and glycosylation levels thus permit quantification of ER exit for WT and mutant hEAAT2. In SDS-PAGEs mYFP-hEAAT2 appears as a triplicate of fluorescence bands, with molecular sizes between ~60 and 75 kDa. An earlier study demonstrated that the higher molecular-weight band disappears upon deglycosylation with PNGase, but not with EndoH,30 demonstrating that hEAAT2 is complex glycosylated. Whereas the majority of WT and L85P hEAAT2 was complex glycosylated, we observed reduced fractions of complex glycosylated protein for G82R and P289R mYFP-hEAAT2 (Figure 1E,F). After correcting fluorescence band intensities by transfection efficiencies (Appendix S1 and Figure S1A) we found that the tested EE variants reduced whole protein expression in HEK293T cells to a variable extent (Figure S1B). We conclude that G82R and P289R, but not L85P, impair ER exit and surface membrane expression of hEAAT2.

3.3 EE-associated SLC1A2 mutations reduce glutamate transport

EAAT2 is a high-capacity glutamate transporter that can generate l-glutamate uptake currents up to 1 nA in transfected mammalian cells.10, 29, 31, 32 To separate transport currents we substituted Cl− by d-gluconate, which is impermeant through WT EAAT channels.32 To measure the unspecific leak component in each experiment, currents were first measured without l-glutamate and then subtracted from current amplitudes obtained after perfusing 5 mM l-glutamate.10, 29 We realized that control currents measured in the absence of l-glutamate were largely increased in cells expressing G82R or L85P hEAAT2 (Figure 2A,B). Unlike in cells that express WT EAAT2, these currents do not represent background currents. They do not require l-glutamate and thus are likely caused by d-gluconate permeation through G82R and L85P hEAAT2 anion channels. Figure 2C,D depicts representative examples of subtracted current recordings and the voltage dependence of average amplitudes of such current components. WT EAATs transport each l-glutamate together with three Na+ and one H+, in exchange with one K+,33-35 resulting in pronouncedly voltage-dependent WT hEAAT2 transport currents. G82R, L85P, and P289R virtually abolished such currents (−150 mV; −316.4 ± 169.6 pA/−5.6 ± 7.6 pA/−35.4 ± 13.4 pA/−31.1 ± 27 pA, WT/G82R/L85P/P289R, ±CI, n = 10/9/7/7) (Figure 2C,D).

Epileptic encephalopathy (EE)–associated mutations impair human excitatory amino acid transporter 2 (hEAAT2) -mediated l-glutamate transport. (A) Representative whole-cell current recordings from HEK293T cells expressing wild-type (WT) or the pathogenic variants G82R, L85P, and P289R hEAAT2. In these experiments Cl− is replaced by d-gluconate−, which is impermeant through WT EAAT channels. (B) Mean current-voltage relationships of d-gluconate currents (±CI) for all tested variants. (C) Representative l-glutamate uptake currents from WT and mutant hEAAT2 in the presence of 5 mM l-glutamate. (D) Mean uptake current-voltage relationships (±CI) for WT and mutant hEAAT2. Color code: WT/G82R/L85P/P289R: black/red/blue/green

To prevent possible contamination of l-glutamate uptake currents with d-gluconate currents through mutant hEAAT2 anion channels, we additionally measured l-glutamate–sensitive currents at the d-gluconate reversal potential, at which any change in channel activity will not modify the current amplitude. Again, we observed drastic reductions in l-glutamate uptake currents for all tested mutant hEAAT2 (Figure S2). We conclude that EE-associated SLC1A2 mutations impair hEAAT2 l-glutamate uptake.

3.4 Pore mutations G82R and L85P modify anion channel function of hEAAT2

We next studied the functional consequences of G82R and L85P on hEAAT2 anion channels through whole-cell patch clamp in transfected HEK293T cells. In these experiments, solutions were adjusted to salt concentrations that resemble physiological solutions. Cells were internally dialyzed with K+-based solutions to permit all physiologically occurring transitions in the l-glutamate uptake cycle,36 and Cl− was used as the main anion in the internal and the external solution. Experiments were performed under symmetric Cl− distributions to increase hEAAT2 anion currents in the physiological voltage range. Figure 3A shows representative whole-cell current responses to voltage steps between −150 mV and +105 mV from HEK293T cells expressing WT hEAAT2 before and after application of 0.5 mM l-glutamate. WT hEAAT2 currents were very small in the absence of l-glutamate, and extracellular substrate application (0.5 mM l-glutamate) elicits an inwardly rectifying current component that consists mostly of l-glutamate uptake currents37 (Figure 3A,B, −150 mV: wo/w L-glutamate: −113.9 ± 34.3/−532.9 ± 145.2 pA, ±CI, n = 12/15). Cells expressing G82R hEAAT2 exhibited currents that change almost linearly with voltage with large current amplitudes without external l-glutamate. Application of l-glutamate did not increase, but rather slightly decreased current amplitudes over the whole voltage range (−150 mV: wo/w l-glutamate: −1742.8 ± 317/−1501.1 ± 582.9 pA, ±CI, n = 11/11) (Figure 3A,B). L85P hEAAT2 currents were also much larger than WT currents in l-glutamate-free media; however, l-glutamate augmented currents and generated time- and voltage-dependent changes in amplitude currents (−150 mV: wo/w l-glutamate: −2481.8 ± 803.3/−3408.6 ± 900 pA, ±CI, n = 12/12) (Figure 3A,B). Figure 3B compares current voltage-relationships from whole cell currents for WT, G82R, and L85P hEAAT2 in the absence and the presence of external l-glutamate. Because both mutations reduce surface membrane expression levels, this pronounced increase of current amplitudes cannot be due to changes in expression levels. We next correlated single-cell fluorescence amplitudes from mYFP-EAAT2 expressing HEK293T cells with anion channel amplitudes at −150 mV38, 39 (Figure 3C,D and Appendix S1). If the observed anion currents are conducted by mutant hEAAT2, a clear correlation between whole-cell current amplitude and the number of expressed fusion proteins, that is, the whole-cell fluorescence, is expected. We found that current amplitudes indeed increase linearly with fluorescence levels (R2adj,−G82R = .91 and R2adj,−L85P = .93) (Figure 3C,D).

Epileptic encephalopathy (EE)–associated missense mutations increase human excitatory amino acid transporter 2 (hEAAT2) -associated anion currents. (A) Representative whole-cell current recordings of wild-type (WT) and mutant hEAAT2 expressed in HEK293T cells under ionic conditions, which permit transitions between all transporter conformations of hEAAT2 (in mM: ±0.5 l-glutamate−/140 Na+Cl−ext and 115 K+Cl−int, see inset), before and after application of 0.5 mM l-glutamate. (B) Mean current-voltage relationships of WT and mutant hEAAT2 whole-cell currents in the absence (open symbols) or in the presence (closed symbols) of 0.5 mM l-glutamate. (C) Representative whole-cell current recordings of WT and mutant hEAAT2 in the presence of internal and external Na+Cl−. (D) Correlation between current amplitudes at −150 mV and yellow fluorescent protein fluorescence amplitudes obtained at the same cell before establishing the whole-cell mode. The lines represent linear regressions (R2adj, G82R = .91, R2adj, L85P = .93). Data in B are provided as means ± CI. Correlation of whole-cell currents and single cell fluorescence was tested with linear regression and calculation of adjusted R2 values (R2adj). The Y-axis intercept was manually set to the current amplitude of untransfected cells under the same ionic conditions (22.8 ± 10.4 pA, n = 5). Color code: WT/G82R/L85P: black/red/blue

Heterologous expression sometimes causes upregulation of endogenous proteins, and volume-activated ion channels (VRACs) seem to be particularly sensitive to such maneuvers.40 To exclude upregulation of VRACs as a potential source of the large anion currents in cells expressing G82R or L85P hEAAT2, we expressed EAAT2 variants in a HEK293T-derived cell, in which a major subunit of the voltage regulated anion channel (VRAC/ LRCC8A, leucine-rich repeat protein 8A)41 has been knocked out by Crispr-Cas techniques,18 using the same experimental parameters as in Figure 3A,B. Figure 4 shows representative currents and mean current-voltage relationships from G82R and L85P hEAAT2 expressed in KO LRCC8A HEK293T cells. We found no differences between anion channel current amplitudes in HEK293T expressing VRAC (dashed lines) or in modified cells (symbols, −150 mV, −2219.9 ± 218.2/−1814.4 ± 274.5 pA, G82R /L85P, ±CI, n = 11/11) in the absence of l-glutamate (Figure 4A,B), and perfusion of these cells with external l-glutamate resulted in similar changes in both cell types (−150 mV, 0.5 mM l-glutamate; −1821.8 ± 218.2/3268.5 ± 402.8 pA, G82R/L85P, ±CI, n = 9/11) (Figure 4A,B).

G82R and L85P human excitatory amino acid transporter 2 (hEAAT2) anion currents in HEK293T cells, in which the gene coding for leucine-rich repeat protein containing 8a (LRCC8A) has been genetically removed. (A, B) Representative whole-cell current traces (left column) with corresponding mean current-voltage relationships in the absence (open symbols) and presence (closed symbols) of 0.5 mM l-glutamate (right column, n = 11/9, (A) for G82R and L85P (n = 11/11, (B) hEAAT2 transporters expressed in KO LRCC8A HEK293T cells. Mean currents (±CI) of this variants expressed in unmodified HEK293T cell line are provided as small circles (in mM: ±0.5 Na+-L-glutamate−/140 Na+Cl−ext, 115 K+Cl−int). Color code: G82R/L85P: red/blue

3.5 G82R and L85P render hEAAT2 anion channels glutamate permeable

For G82R and L85P hEAAT2, we observed currents well above background currents in cells intracellularly dialyzed and external perfused with d-gluconate-based solutions (Figure 2A). These currents demonstrate that d-gluconate can permeate through mutant hEAAT2 anion channels. G82R and L85P may thus even allow the passage of l-glutamate through hEAAT2 anion channels. We measured reversal potentials of mutant hEAAT2 currents in cells internally dialyzed with l-glutamate or d-gluconate-based solutions (Figure 5 and Figure S3) and perfused with physiological Cl−-based external solutions. To prevent possible contaminations with endogenous K+ channels and to block l-glutamate transport, currents were measured in K+-free internal and external solutions (in mM, ext: 144 NaCl, 1 MgCl2, 2 CaCl2, 10 HEPES/NaOH; int: 115 Na-gluconate/glutamate, 2 MgSO4, 5 EGTA, 10 HEPES/NaOH, pH 7.4/7.4). Because perfusion with Cl−-based external solutions might result in intracellular Cl− accumulation, cells were held constantly at −70 mV between voltage steps.

Epileptic encephalopathy (EE)–associated variants G82R and L85P render the human excitatory amino acid transporter 2 (hEAAT2) anion pore permeable to l-glutamate. (A) Representative G82R and L85P hEAAT2 whole-cell currents under ionic conditions, which prevent l-glutamate uptake (in mM: ±0.5 l-glutamate−ext/140 Na+Cl−ext and 115 Na+-l-glutamate−int). (B) Mean current voltage relationships of whole-cell currents from WT, G82R, or L85P hEAAT2. The gray-shaded region is magnified as inset. G82R and L85P hEAAT2-mediated inward currents correspond to outward permeation of internal l-glutamate. Data in B are provided as means ± CI. Color code: WT/G82R/L85P: black/red/blue

In cells dialyzed with Na-glutamate-based solution (in mM: 115 Na-glutamate, 2 MgSO4, 5 EGTA, 10 HEPES/NaOH), G82R and L85P, but not WT hEAAT2 conducted prominent inward currents (Figure 5). Without external l-glutamate, L85P and G82R hEAAT2 glutamate currents were significantly larger than WT hEAAT2 currents at positive and negative holding potentials (+60 mV:36.2 ± 13.8/395.4 ± 117.2/606 ± 158.5 pA; −150 mV: −24.3 ± 5.9/−162.1 ± 49.8/−316.4 ± 76.3 pA, n = 11/14/11, WT/G82R/L85P). Application of external l-glutamate (0.5 mM) left WT currents nearly unaffected, but increased L85P and decreased G82R inward and outward currents (+60 mV:42.1 ± 10.4/316.6 ± 112.7/861.2 ± 223 pA; −150 mV: −21.5 ± 7.5/−124.3 ± 49.7/−411.6 ± 97.3 pA, n = 9/14/11, WT/G82R/L85P). Reversal potentials from whole-cell currents during perfusion with external NaCl were negative (wo/w: 0.5 mM l-glutamateext, WT: −36.8 ± 9.5/−49.6 ± 16 mV, n = 11/9; G82R: −59.1 ± 7.3/−61.1 ± 6.8 mV, n = 14/14; L85P: −63.2 ± 4/−70 ± 3.4 mV, n = 19/11), indicating lower l-glutamate than Cl− permeability (Figure S3). We used the Goldman Hodgkin Katz equation to calculate relative permeabilities for G82R PGluPCl=0.151±0.064,N=14withoutexternalL-glutamateandPGluPCl=0.137±0.051,N=15withexternalL-glutamatePGluPCl=0.151±0.064,N=14withoutexternalL-glutamateandPGluPCl=0.137±0.051,N=15withexternalL-glutamate and L85P hEAAT2andPGluPCl=0.087±0.015,N=11withexternalL-glutamatePGluPCl=0.117±0.02,N=19withoutexternalL-glutamate Negligible inward currents demonstrated absent l-glutamate permeability in WT hEAAT2 (Figure 5B).

3.6 P289R modifies opening of hEAAT2 anion channel, but not its selectivity

P289R hEAAT2 is homologous to P290R hEAAT1, which was the first disease-associated SLC1 mutation linked to neuronal hyperexcitability.13 The functional consequences of this proline-by-arginine exchange mutation have been studied in hEAAT1 in mammalian cells,13, 14 as well as in hEAAT3 (P259R) in Xenopus oocytes and mammalian cells.28 In hEAAT1 as well as in hEAAT3, exchange of proline by arginine increases anion channel open probabilities in the absence as well as in the presence of l-glutamate and causes time-dependent increases of anion currents upon hyperpolarizing voltage steps, a characteristic feature of such mutant transporters that can also be observed in Bergmann glia cells of the transgenic Slc1a3P290R/+ animals.15 We did not observe any effects of this mutation on EAAT1/EAAT3 anion channel selectivity or unitary current amplitudes.14

Figure 6A depicts representative current recordings from a HEK293T cell expressing P289R hEAAT2 under standard conditions before and after l-glutamate application. In the absence of l-glutamate, large mutant anion currents could be observed that were inwardly rectifying, activated upon hyperpolarizing voltage steps, and increased by external l-glutamate application. A comparison of current-voltage relationships from WT and P289R Cl− currents showed similar current amplitudes at hyperpolarizing potentials (−150 mV: −532.8 ± 74.1/−600.7 ± 174.9 (±CI, n = 12/15, WT/P289R) in the presence of external l-glutamate (−150 mV: −133.9 ± 34.3/−399.1 ± 132.1 pA (n = 12/15) (Figure 6B). However, WT currents predominantly represent electrogenic uptake currents, whereas P289R currents are almost exclusively Cl− currents, since the mutation abolishes l-glutamate transport currents (Figure 2). P289R hEAAT2-mediated d-gluconate currents were indistinguishable from WT (Figure 2), demonstrating unaltered anion channel selectivity of mutant transporters. The increased anion current amplitudes of P289R hEAAT2 at lower transporter numbers in the membrane demonstrate gain-of-function of P289R hEAAT2-associated anion channels, as observed previously in the episodic ataxia 6-associated P290R mutation in hEAAT1.14

Epileptic encephalopathy (EE)–associated mutation P289R changes the amplitude and the time and voltage dependence of human excitatory amino acid transporter (hEAAT2) anion currents. (A) Representative P289R hEAAT2 whole-cell currents under ionic conditions, which permit all transitions between transporter conformations (in mM: ±0.5 L-glutamate−/140 Na+Cl−ext and 115 K+Cl−int). (B) Mean current-voltage relationships for WT and P289R hEAAT2 without or with external l-glutamate (open/closed symbols, n = 12/15, WT/P289R). Data in B are provided as means ±CI. Color code: WT/P289R: black/green

4 DISCUSSION

The epileptic encephalopathies (or EEs) are a group of severe and often therapy-resistant early onset epilepsies.42, 43 Severe and frequent seizures, often of distinct generalized epilepsy types, cause cognitive impairment, likely by impairing networks involved in cognitive processing.42, 44 We studied the functional consequences of all SLC1A2 mutations that have been reported in patients with EE thus far.1, 2 We analyzed how G82R, L85P, and P289R affect glutamate transport, anion channel function, and subcellular localization and protein expression of human EAAT2 in a mammalian cell line as heterologous expression system. These experiments revealed gain-of-anion channel function and reduced l-glutamate transport capability as a major functional alteration for all three mutations. L85P hEAAT2 anion currents were larger than current amplitudes recorded in cells expressing G82R or P289R and much larger than for WT hEAAT2 (Figure 2). L85P and G82R permit permeation of l-glutamate and d-gluconate through hEAAT2 anion channels. Our functional results are in full agreement with current structural concepts about the EAAT anion channel pore, in which L85 functions as pore-forming residue projecting its side chain into the aqueous permeation pathway formed at the interface between the trimerization and the mobile transport domain.11, 26, 27 Substitution of leucine 85 by proline or glycine 82 by arginine might modify the dimensions of the anion channel pore and permit passage of d-gluconate and l-glutamate, which both have diameters too large to permeate through the WT hEAAT2 anion channel. The functional consequences of the L85P mutation have been recently addressed in another study that mostly quantified l-glutamate uptake by radiotracer experiments.45 This study reported decreased l-glutamate uptake in mammalian cells expressing L85P hEAAT2 and reduced WT uptake upon co-expression of L85P, suggesting a dominant-negative effect on l-glutamate transport. EAAT2 anion currents were not taken into consideration; however, L85P hEAAT2-mediated L-glutamate efflux can fully account for their experimental results.

G82R and L85P are—to our knowledge—the first mutations that modify the size selectivity of EAAT anion channels. Glutamate is a potent neurotoxin, and a key physiological function of the EAATs is to accumulate l-glutamate over large concentration gradients and to ensure low resting l-glutamate concentration in the extracellular space. Such function requires a strict selectivity against l-glutamate of the EAAT anion channels. Under ischemic conditions, activation of volume-activated VRAC anion channels is known to permit l-glutamate efflux, causing excessive l-glutamate release and brain damage.46-48 Glutamate-permeable mutant EAAT2 anion channels could serve a similar pathological function in EE. EAAT2 is not only present in astrocytes, but also in presynaptic nerve terminals.7, 8 Because neurons lack significant glutamine-synthase activity, intracellular l-glutamate levels are larger in neurons than in astrocytes predicting higher l-glutamate effluxes through presynaptic mutant EAAT2. In both cases, extracellular glutamate homeostasis will be significantly altered, possibly causing neuronal death and brain atrophy via glutamate excitotoxicity.

SLC1A2-associated epileptic syndromes exhibit a clinical phenotype1, 2 that is far more severe than for many other EEs (please see the case reports of Guella et al1). All patients experienced epileptic seizures within the first weeks of life and showed significant developmental delay, and all exhibit cerebral, but not cerebellar, atrophy.1-3 Patients with the G82R mutation presented with prolonged focal motor seizures that started as left clonic leg jerks and later with bilateral tonic-clonic seizures.1 In case of L85P, continuous multifocal twitching, particularly in sleep, was observed at an age of 2 days. Later, generalized and focal myoclonic seizures occurred.2 The patient with the P289R mutation had myoclonic jerks and absence and gelastic seizures.1 As his case was reported, the patient was 6-years-old with 50 seizures daily on a combined treatment of rufinamide, levetiracetam, nitrazepam, and the ketogenic diet.1 The L85P mutation was associated with the almost complete absence of myelination of cerebrum, but normal myelination of the cerebellum and brainstem at 1 year, and at the age of 13 years, extreme supratectorial atrophy involving the cortex, deep gray matter, and white matter, with normal-sized brainstem and cerebellum.2 Whereas patients with G82R had progressive generalized atrophy of the cerebrum, MRI of the P289R patients revealed only an asymmetric brainstem of unlikely clinical significance at an age of 3 years. Because all patients with EE-associated SLC1A2 mutations have therapy-resistant seizures with early onset and high frequency, cerebral atrophy associated with G82R and L85P is likely due to the additional glutamate permeability of these two mutations. The more pronounced cerebral atrophy of patients with the L85P than with the G82R mutations might be explained by differences in intracellular trafficking. L85P only slightly affects ER exit and surface membrane insertion so that heterozygous patients with this variant may exhibit higher numbers of mutant transporters in the surface membrane than patients with the G82R mutation.

P289R does not increase d-gluconate permeability of hEAAT2 anion channels. However, it causes pronounced gain-of-function of hEAAT2 chloride currents, resulting in large increases of P289R hEAAT2 anion currents, despite strongly reduced surface expression of these mutant transporters (Figure 1). The functional consequences of P289R mutation on hEAAT2 transport functions are identical to those of the P290R on hEAAT1. The SLC1A3 mutation causes episodic ataxia type 6