γ-Aminobutyric acid transporter and GABA_A receptor mechanisms in Slc6a1^+/A288V and Slc6a1^+/S295L mice associated with developmental and epileptic encephalopathies

Abstract

We have previously characterized the molecular mechanisms for variants in γ-aminobutyric acid transporter 1-encoding solute carrier family 6-member 1 (SLC6A1) in vitro and concluded that a partial or complete loss of γ-aminobutyric acid uptake due to impaired protein trafficking is the primary aetiology. Impairment of γ-aminobutyric acid transporter 1 function could cause compensatory changes in the expression of γ-aminobutyric acid receptors, which, in turn, modify disease pathophysiology and phenotype. Here we used different approaches including radioactive ³H γ-aminobutyric acid uptake in cells and synaptosomes, immunohistochemistry and confocal microscopy as well as brain slice surface protein biotinylation to characterize Slc6a1^+/A288V and Slc6a1^+/S295L mice, representative of a partial or a complete loss of function of SLC6A1 mutations, respectively. We employed the γ-aminobutyric acid transporter 1-specific inhibitor [³H]tiagabine binding and GABA_A receptor subunit-specific radioligand binding to profile the γ-aminobutyric acid transporter 1 and GABA_A receptor expression in major brain regions such as cortex, cerebellum, hippocampus and thalamus. We also determined the total and surface expression of γ-aminobutyric acid transporter 1, γ-aminobutyric acid transporter 3 and expression of GABA_A receptor in the major brain regions in the knockin mice. We found that γ-aminobutyric acid transporter 1 protein was markedly reduced in cortex, hippocampus, thalamus and cerebellum in both mutant mouse lines. Consistent with the findings of reduced γ-aminobutyric acid uptake for both γ-aminobutyric acid transporter 1(A288V) and γ-aminobutyric acid transporter 1(S295L), both the total and the γ-aminobutyric acid transporter 1-mediated ³H γ-aminobutyric acid reuptake was reduced. We found that γ-aminobutyric acid transporter 3 is only abundantly expressed in the thalamus and there was no compensatory increase of γ-aminobutyric acid transporter 3 in either of the mutant mouse lines. γ-Aminobutyric acid transporter 1 was reduced in both somatic regions and nonsomatic regions in both mouse models, in which a ring-like structure was identified only in the Slc6a1^+/A288V mouse, suggesting more γ-aminobutyric acid transporter 1 retention inside endoplasmic reticulum in the Slc6a1^+/A288V mouse. The [³H]tiagabine binding was similar in both mouse models despite the difference in γ-aminobutyric acid uptake function and γ-aminobutyric acid transporter 1 protein expression for both mutations. There were no differences in GABA_A receptor subtype expression, except for a small increase in the expression of α5 subunits of GABA_A receptor in the hippocampus of Slc6a1^S295L homozygous mice, suggesting a potential interaction between the expression of this GABA_A receptor subtype and the mutant γ-aminobutyric acid transporter 1. The study provides the first comprehensive characterization of the SLC6A1 mutations in vivo in two representative mouse models. Because both γ-aminobutyric acid transporter 1 and GABA_A receptors are targets for anti-seizure medications, the findings from this study can help guide tailored treatment options based on the expression and function of γ-aminobutyric acid transporter 1 and GABA_A receptor in SLC6A1 mutation-mediated neurodevelopmental and epileptic encephalopathies.

Graphical abstract

Open in new tabDownload slide

Introduction

γ-Aminobutyric acid (GABA) transporter 1 (GAT-1) is one of the main GABA transporters in the brain responsible for the reuptake of GABA from the synaptic cleft into neurons and astrocytes, thereby terminating neurotransmission. GAT-1 is a member of the solute carrier 6 (SLC6) protein superfamily and is encoded by the SLC6A1 gene. SLC6A1 variants have been associated with a wide range of neurodevelopmental disorders including seizures, autism and learning disorders, with myoclonic atonic epilepsy and impaired cognition being two prominent features.^1-3 Previously, we have characterized the impact of different SLC6A1 variants on GAT-1 function and trafficking in vitro.^4-7 Using cell models, including human patient-induced pluripotent stem cells, patient-derived neurons and astrocytes, we ascertained that a partial or a complete loss of function of GAT-1 due to impaired protein trafficking is a common mechanism across GAT-1 variants associated with various clinical phenotypes.⁸ As GAT-1 is expressed in both neurons and astrocytes, and possibly in other cell types such as microglia and oligodendrocytes,⁹ we characterized the impact of the SLC6A1 variants in multiple cell types. We identified the common principles underlying the impaired trafficking and function of the variants in both neurons and astrocytes.⁸ Our in vitro work suggested that the variants caused reduced function of GABA uptake due to reduced trafficking and expression of membrane GAT-1.

GABA, GABA receptors and GABA transporters work in concert to maintain neuronal homeostasis. This homeostasis is essential for normal brain development as mutations in both GAT-1 encoding SLC6A1 and GABA_A receptors (GABA_ARs) are major aetiologies for developmental and epileptic encephalopathies.^3,10,11 We have demonstrated that GAT-1 haploinsufficiency is the major mechanism for SLC6A1 variant-mediated disorders based on GABA uptake activity from a large cohort of variants.⁸ We concluded that a common partial or complete loss of function of the mutant protein underlies disease pathophysiology despite enormous phenotypical heterogeneity.^5,6,8 Previous work on Slc6a1 knockout mice indicated that GAT-1 activity regulates both tonic and phasic GABA_AR-mediated neurotransmission.¹² Brain slice recordings on cortical layers II–III¹³ or hippocampal pyramidal neurons^12-14 indicated that GABAergic tonic current is increased in both brain regions in the Slc6a1 knockout mice. This study also demonstrated that the frequency of GABAergic miniature inhibitory post-synaptic currents is reduced in hippocampal pyramidal neurons.¹² Moreover, deficiency of the GABA_AR α1 subunit in Gabra1 knockout mice causes increased ambient GABA concentration and GABAergic tonic current, suggesting a complex relationship between GABA_ARs and transporter function in maintaining neuronal homeostasis.¹⁵ Together, this suggests that adaptive changes in the expression of GABA_ARs and GAT-1 occur when there is altered activity in either the receptor or the transporter. Altered GABA_AR function is a primary mechanism for both genetic and acquired epilepsy^16-20 and GABA_ARs are major targets for anti-seizure drugs.^21-25 GABA_AR activity can modify the phenotype of epilepsy caused by other non-GABA genes,^26,27 suggesting their critical role in maintaining a balance in neuronal excitation and inhibition. Understanding adaptive changes in GABA_AR expression and function in GAT-1-deficient mice could provide critical insights into how to best leverage GABA_AR activity to mitigate the patho-mechanisms caused by GAT-1 deficiency.

Since the molecular changes in SLC6A1 variants have not previously been studied in vivo, the current study was undertaken to characterize two knockin mouse lines, Slc6a1^+/A288V and Slc6a1^+/S295L, bearing variants representative of partial or complete loss of GABA uptake function due to impaired GAT-1 protein stability and trafficking.⁸ The SLC6A1(A288V) variant is a reoccurring variant identified in multiple patients, while SLC6A1(S295L) is a de novo mutation. The SLC6A1(A288V) variant is associated with various epilepsy syndromes, autism and neurodevelopmental delay, while the SLC6A1(S295L) variant is associated with the absence of epilepsy and neurodevelopmental delay. We used biochemical approaches, confocal microscopy, crude live synaptosome GABA uptake and autoradiography to characterize the expression and function of GAT-1 and the expression of GABA_AR subunits. This study, in combination with the previous characterizations of multiple SLC6A1 variants in cell models,^2,4-8,28 provides further understanding of the pathophysiology underlying disorders mediated by SLC6A1 variants.

Materials and methods

SLC6A1 variants information

The patient variants A288V and S295L were selected from the lab complementary deoxyribonucleic acid (cDNA) library which was built and described in our previous studies.^1-3,8  SLC6A1(A288V) variant is representative of a partial loss of function, while SLC6A1(S295L) for a complete loss of GABA uptake due to impaired protein stability and trafficking, which is a common mechanism for SLC6A1 pathological variants.

SLC6A1 variants knockin mouse models

Both variants SLC6A1(A288V) and SLC6A1(S295L) have been characterized in vitro in our previous study.⁸ The SLC6A1(A288V) line was generated in collaboration with UConn Health, and the SLC6A1(S295L) line was generated by Shanghai Model Organisms (Shanghai Model Organisms Center, Inc. Cat. NO. NM-KI-190014). Both mouse lines were developed with the clustered regularly interspaced short palindromic repeats and CRISPR-associatedprotein 9 (CRISPR-CAS9) global knockin approach. Both mouse lines have been maintained in the C57BL/6J mice (Jax Stock #000664). Both mouse models displayed increased 5–7 Hz spike-wave discharges, reminiscent of the absence of seizure in patients bearing the mutations^4,28 and having a normal life span. Heterozygous mice were generated by breeding heterozygous mice with wild-type mice, while homozygous mice were generated by breeding heterozygous mice. All experiments involving mice were conducted under the guidelines of Association for Assessment and Accreditation of Laboratory Animal Care and approved by the Vanderbilt University Institutional Animal Care and Use Committee committee. Both male and female mice were included in the study. The mice used for all experiments in this study were between 1 and 2 months old and were back-crossed for at least eight generations (F8). The comparison was made between the wild-type and the heterozygous mice. The homozygous mice were only included in some experiments for reference as they were not always available, possibly due to embryonic lethality or reduced viability of the homozygote. All experiments in mice have been approved by the Institutional Animal Care and Use Committee.

GAT-1 cDNA constructs

The plasmid cDNA encoding enhanced yellow fluorescent protein (YFP)-tagged rat GAT-1 has been previously described.^4,8

Subcellular fractionation and isolation of live synaptosomes

The procedures of subcellular fractionation were modified from previous studies.^17,29

Live brain slice biotinylation

The protocol was based on our previous studies.^6,11,17

Radioactive ³H-labelled GABA uptake assay

The radioactive ³H-labelled GABA uptake assay in HEK293T cells was modified from the protocol used in our previous studies.^5,7,8 The protocols for GABA reuptake assay in live synaptosomes isolated from mouse forebrain were modified from the GABA uptake protocol on HEK293T cells.⁸ The GAT-1 inhibitors Cl-966 and NNC-711 were applied in cells transfected with the wild-type GAT-1 to validate if the detected radioactivity is specific, and in some cases, GAT-3 inhibitor SNAP5114 was also included to make sure the radioactive counts measured were GAT-1 specific.

Autoradiography of GAT-1 and GABA_ARs

The brains of Slc6a1^+/A288V and Slc6a1^+/S295L mice at 1–2 months of age were dissected, quickly frozen and shipped to Roche. The samples were subject to autoradiography using specific radioligands for GAT-1 and various GABA_AR subtypes using standard protocols.^30,31

Confocal microscopy and image acquisition

Live cell confocal microscopy was performed using an inverted Zeiss laser scanning microscope (Model 510) with a 63 × 1.4 NA oil immersion lens, ×2 to ×2.5 zoom and multi-track excitation.⁹ Cells were seeded on glass-bottom imaging dishes at the density of 1–2 × 10⁵ cells per dish and co-transfected with 0.5 µg of the wild-type or the mutant GAT-1 cDNAs. The endoplasmic reticulum (ER) marker ER^CFP cDNAs were co-transfected with the wild-type or the mutant GAT-1 cDNAs with a total of 1 µg cDNAs were transfected per 35-mm glass-bottomed culture dish with polyethylenimine. All images were obtained from live cells with single confocal sections averaged from 8 to 16 times to reduce noise.

Statistical analysis

Data were expressed as mean ± SEM. Proteins were quantified by Odyssey software, and data were normalized to loading controls and then to wild-type subunit proteins, which were arbitrarily taken as 1 in each experiment. Fluorescence intensities from confocal microscopy experiments were determined using MetaMorph imaging software, and the measurements were carried out in ImageJ as modified from a previous description.^8,9,21,25 For statistical significance, we used one-way or two-way analysis of variance with Newman–Keuls test or Student’s unpaired t-test. In some cases, a one-sample t-test, unpaired t-test or paired t-test was performed (GraphPad Prism, La Jolla, CA, USA), and statistical significance was defined as P < 0.05.

Results

SLC6A1(A288V) and SLC6A1(S295L) are two representative variants with impaired protein trafficking and function

We have selected two representative variants, SLC6A1(A288V) and SLC6A1(S295L), to understand the impact of SLC6A1 variants in mice. SLC6A1(A288V) is a partial loss-of-function variant, while SLC6A1(S295L) is a complete loss-of-function variant.⁸ As demonstrated in the previous study, both variants exhibit compromised transporter activity in vitro due to a protein trafficking defect.⁸ In the GAT-1 protein, A288V is located at the third extracellular loop, while S295L is located at the sixth transmembrane domain (Fig. 1A). Other known SLC6A1 variants are distributed in various locations and domains of the encoded GAT-1 protein (Fig. 1A). Both Slc6a1^+/A288V and Slc6a1^+/S295L global knockin mice were generated with CRISPR/CAS9 technology. Sequencing shows the variant nucleotide 863C>T in Slc6a1^+/A288V and 884C>T in Slc6a1^+/S295L mouse models (Fig. 1B). The mutant mice were identified through automated genotyping using the Transnetyx service.

To evaluate the impact of the variant on GAT-1 protein function, we determined the GABA uptake of both variants in HEK293T cells co-transfected with a mixture of the wild-type and the mutant GAT-1^YFP cDNAs at a 1:1 ratio for 48 h. Compared with the cells transfected with the wild-type alone, both heterozygous and homozygous mutations had reduced GABA uptake function as measured by the high-throughput [³H]GABA uptake assay (Fig. 1C) (0.648 ± 0.04 for A288V ‘het’; 0.602 ± 0.038 for S295L ‘het’; 0.327 ± 0.038 for A288V ‘hom’; 0.021 ± 0.005 S295L ‘hom’ versus the wild-type which was taken as 1). However, the level of GABA uptake of the heterozygous mutations was higher than the cells expressing the wild-type treated with GAT-1 inhibitor Cl-966 (50 µM) (0.238 ± 0.022) and NNC-711 (35 µM) (0.353 ± 0.018), suggesting the contribution of GABA uptake function from the wild-type allele.

We then determined the total GAT-1 expression and analysed the mature and immature species of the wild-type and the mutant GAT-1. The protein lysates were either undigested (U) or digested with endoglycosidase H, Endo-H (H), the enzyme that removes glycan attached to the protein inside the ER and then analysed by sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE). The wild-type GAT-1 had two strong bands (Band 1 and Band 2) run at 108 and 96 kDa and a faint band at 90 kDa (Band 3) (Fig. 1D and Supplementary Fig. 1). To demonstrate the GAT-1 antibody specificity, Chinese hamster ovary cells were used as a control because of the low-level expression of endogenous GAT-1 compared with the HEK293T cells. Compared with the wild-type, both GAT-1(A288V) and GAT-1(S295L) proteins had reduced amount of the mature form of GAT-1. The mature form of GAT-1(A288V) was reduced to ∼30% of the wild-type and was almost undetectable for GAT-1(S295L) (Fig. 1E). However, the immature form of GAT-1 conducted at 90 kDa was ∼3-fold greater than the wild-type for both A288V and S295L variants (Fig. 1E).

Slc6a1^+/A288V and Slc6a1^+/S295L mouse lines had reduced total GAT-1 protein expression, but no compensatory increase of GAT-3

We then analysed the cellular consequence of the mutant protein in vivo. Based on our extensive studies on GABA_AR gene mutations, the mutant protein is often subject to rapid degradation inside the ER. Consistent with our hypothesis and our previous study,³² biochemical studies demonstrated that total levels of GAT-1 in both Slc6a1^+/A288V and Slc6a1^+/S295L mice were reduced in multiple brain regions, including cortex, cerebellum, hippocampus and thalamus (Fig. 2A–D and Supplementary Fig. 1 for full-length gels) (A288V mice, cortex: 0.552 ± 0.024 for het and 0.186 ± 0.041 for hom, cerebellum: 0.546 ± 0.022 for het and 0.167 ± 0.033 for hom, hippocampus: 0.538 ± 0.025 for het and 0.190 ± 0.06 for hom and thalamus: 0.524 ± 0.024 for het and 0.217 ± 0.06 for hom). The GAT-1 expression in the homozygous Slc6a1^A288V/A288V mice was ∼20% of the wild-type, which was further reduced compared with the heterozygous mice. GAT-1 expression in the heterozygous lines was about half of the wild-type and was comparable in both mouse models. The homozygous Slc6a1^S295L/S295L mice had almost undetectable GAT-1, which may be a consequence of faster degradation of the GAT-1(S295L) protein. These results suggest less interference of the mutant Slc6a1 allele on the wild-type protein function in Slc6a1^+/S295L mice, compared with the Slc6a1^+/A288V mice.

GAT-3 is another major GABA transporter in the brain, which is abundantly expressed in the neonatal cortex prior to increased GAT-1 expression later in development.³³ It has been demonstrated that enhancing GAT-3 in thalamic astrocytes promotes resilience to brain injury in rodents, suggesting some overlapping function between GAT-1 and GAT-3.³⁴ We determined whether there is a compensatory increase of GAT-3 activity in Slc6a1^+/S295L and Slc6a1^+/A288V mice. Our data indicated that GAT-3 was abundantly expressed in the thalamus, and expression of GAT-3 was 4- to -7-fold higher in the thalamus than that in other brain regions (Fig. 2E and G and Supplementary Fig. 1 for full-length gels). There was no obvious compensatory increase of GAT-3 in all major brain regions (Fig. 2F and H). By contrast, we observed a small, but significant reduction of GAT-3 in the cortex of the homozygous Slc6a1^A288V/A288V mice (Fig. 2F). This suggests that the mutant GAT-1(A288V) could have a small negative impact on GAT-3 due to aberrant oligomerization or other unknown mechanisms.

Both GAT-1(A288V) and GAT-1(S295L) ‘heterozygous’ transporters exhibited increased retention in the ER in live cells

Our previous work demonstrated that different cell types have conserved post-translational protein modifications and ER surveillance.^8,35 We are not able to visually distinguish the mutant and wild-type GAT-1 in mice, so we thus utilized HEK293T cells and tagged the mutant proteins with YFP to determine if the ‘heterozygous’ expression of GAT-1(A288V) and GAT-1(S295L) transporters had increased ER retention. Cells were transfected with wild-type GAT-1 alone or the ‘heterozygous’ condition by mixing the wild-type, and the mutant GAT-1(A288V) or GAT-1(S295L) with the ECFP-ER marker (ER^CFP) at 1:1 ratio (1 µg for GAT-1^YFP:1 µg of ER^CFP cDNAs) for 48 h, after which confocal images were acquired in live cells (Fig. 3A). For both the GAT-1(A288V) and GAT-1(S295L) ‘heterozygous’ mutant conditions, the GAT-1^YFP fluorescence overlapping with ER^CFP fluorescence was increased compared with the cells expressing the wild-type (Fig. 3B) (19.29±2.77% for wt; 56.77±2.98% for A288V; 62.31±3.15% for S295L). Although the ER-retained GAT-1^YFP was higher in the mutant conditions, the total GAT-1^YFP signal in the whole field of the cells expressing the mutant GAT-1 was still lower than the wild-type (Fig. 3C) (65.30 ± 2.41 for wt; 45.73 ± 3.11 for A288V; 40.09 ± 2.56 for S295L), suggesting rapid ER-associated degradation of the mutant GAT-1 protein. It is of note that the ER marker signalling appeared to be higher and the size of the ER was larger in the cells expressing the mutant GAT-1 than in the cells expressing the wild-type GAT-1. This is likely related to the aberrant protein retention and degradation inside ER in the cells overexpressing the mutant GAT-1.

Slc6a1^+/A288V and Slc6a1^+/S295L mice had reduced GAT-1 puncta and total GAT-1 in mouse brain cortex

To understand the cellular fate of ER-retained mutant GAT-1 in the knockin mice, we determined the expression and subcellular localization of GAT-1 in the cortex of both knockin mouse models. We focused on layers V–VI of the somatosensory cortex because of the involvement of corticothalamic circuitry in seizure generation and our previous characterization of this brain region for other epilepsy mouse models.¹⁷ The brain sections from 1-month-old wild-type and heterozygous (het) knockin littermates were blocked, processed and immunostained with anti-GAT-1 antibody. Like GABA_AR subunits, GAT-1 exhibited punctate expression in the cortex (Fig. 4A, green). GAT-1 puncta were distributed around cellular nuclei as marked by the nuclei marker, To-pro-3 (blue). Compared with the wild-type littermates, both Slc6a1^+/A288V and Slc6a1^+/S295L mice had reduced total GAT-1 signal (Fig. 4A and B and Supplementary Fig. 2). Based on our protocols for analysing GABA_AR subunits,¹⁷ the fluorescence intensities of the whole field (Fig. 4B), non-somatic (Fig. 4C) or somatic (Fig. 4D) regions were measured. The fluorescence intensities of GAT-1 of the whole field, non-somatic and somatic regions were reduced for both knockin mice (whole field A288V mice: 56.22 ± 2.526% for wt and 32.33±1.8% for het; S295L mice: 59.10 ± 2.58% for wt and 30.40±2.53% S295L). A similar profile was observed for GAT-1 fluorescence intensity in non-somatic regions. However, for somatic regions, the magnitude of GAT-1 protein reduction was smaller in Slc6a1^+/A288V mice than in Slc6a1^+/S295L mice (somatic region A288V mice: 53.45 ± 2.33 for wt and 43.82 ± 2.65% for het; S295L mice: 57.30 ± 3.22 for wt and 22.7 ± 2.43 S295L). It is worth noting that a ring-like GAT-1 fluorescence signal (Fig. 4A) was identified immediately around the cell nucleus, suggesting that some GAT-1(A288V) was retained inside the ER, as identified in the HEK293T cells. This was not detectable in the Slc6a1^+/S295L mice more likely because GAT-1(S295L) was degraded faster.

Both Slc6a1^+/A288V and Slc6a1^+/S295L mice had reduced GABA uptake in the live synaptosomes

To understand the functional consequence of reduced GAT-1 expression in the knockin mice, we then determined GABA uptake in live synaptosomes. We first isolated crude synaptosomes from forebrain samples, containing cerebral cortex, hippocampus and thalamus of wild-type and mutant heterozygous mice between 1 and 2 months of age. The live crude synaptosomes were incubated with preincubation solutions containing [³H]GABA without or with Cl-966 (50 µM), NNC-711 (35 µM) or SNAP5114 (30 µM) for 30 min and then counted on a liquid scintillator with QuantaSmart. The GABA uptake in the heterozygous mice was normalized to gender-matched littermates in each experiment. The GABA uptake activity was dramatically reduced with treatment with GAT-1 inhibitors Cl-966 and NNC-711. GABA uptake was reduced to a lesser extent when treated with the GAT-3 inhibitor SNAP5114, suggesting that GABA uptake activity in synaptosomes isolated from the forebrain is primarily mediated by GAT-1 (Fig. 5A) (17.17 ± 2.83% for Cl-966, 25 ± 2.96% for NNC-711 and 76.67 ± 2.08% for SNAP5114). We then compared the GAT-1-mediated GABA uptake level of the mutant mice with the wild-type littermates for both mouse lines. SNAP5114 was applied to inhibit GAT-3 activity during the uptake. The counts per minute (CPM) counts were lower in synaptosomes treated with SNAP5114, suggesting GAT-3 activity was inhibited (Fig. 5C). It is interesting that Slc6a1^+/A288V and Slc6a1^+/S295L mice had a comparable level of GABA reuptake either for raw CPM counts (Fig. 5C) or the ratio normalized to littermates (Fig. 5D). The GABA uptake activity mediated by GAT-1 in forebrain crude synaptosomes was reduced in both mouse lines (A288V mice: 6115 ± 321.8 for wt and 3175 ± 225.7 for het; S295L mice: 6326 ± 634.7 for wt and 3322 ± 274 CPM for het).

Slc6a1^+/A288V and Slc6a1^+/S295L heterozygous mouse lines showed reduced binding of a GAT-1 radioligand [3H]tiagabine in autoradiography

Tiagabine is a known GAT-1 inhibitor. [³H]tiagabine radioligand binding is an established method for evaluating GAT-1 expression. We determined GAT-1 expression levels by [³H]tiagabine autoradiography. Ten µm thick cryo-sections of fresh frozen brains from wild-type and heterozygous mice (aged 1–2 months) of the Slc6a1^+/A288V and Slc6a1^+/S295L lines were incubated with [³H]tiagabine (5.2 nM) to quantify total radioligand binding (TB) to GAT-1 (Fig. 6A and B). Nonspecific binding (NSB) was defined by co-incubation with 10 µM NNC-711 and all values were expressed as fmol specific binding (SB) per mg protein (SB = TB − NSB). Regions of interest were delineated for cortex, hippocampus, thalamus and cerebellum. In all brain regions, binding was reduced by 42–51% in heterozygous mice compared with wild-type controls in Slc6a1^+/A288V or Slc6a1^+/S295L mice (Fig. 6C and D). Interestingly, both mouse lines had similar levels of radioligand binding. This is in contrast with different mutant protein amount of GAT-1(A288V) and GAT-1(S295L) when expressed in HEK293T cells but is in line with the total GAT-1 expression in mutant mice (Fig. 2A–D). Based on biochemistry data, the mutant GAT-1 in the Slc6a1^S295L/S295L mice was undetectable, while Slc6a^A288V/A288V homozygous mice had ∼30% of protein expression and GABA uptake function in the synaptosome (data not shown). This points to a potential differential dominant negative suppression of the mutant GAT-1 protein due to the complex interaction between the wild-type and the mutant protein products in vivo, especially in Slc6a^+/A288V mice.

Slc6a1^+/A288V and Slc6a1^+/S295L heterozygous and homozygous mice had unaltered GABA_AR subunit protein expression in cortex, cerebellum, hippocampus and thalamus

The altered GAT-1 structure compromises the GABA reuptake activity in the brain. Consequently, excessive GABA due to GAT-1 deficiency can impact both GABAergic tonic and phasic currents. GABA_ARs are a major target for epilepsy treatment development and many existing anti-seizure drugs work by enhancing GABA_ARs activity. We have profiled the expression of major GABA_AR subunits in different brain regions: cortex (cor), cerebellum (cb), hippocampus (hip) and thalamus (thal) from the wild-type (wt), heterozygous (het) and homozygous (hom) mice of both Slc6a1^+/A288V and Slc6a1^+/S295L mice. We included γ2 and δ subunits because γ2-containing GABA_ARs mediate predominantly phasic currents while δ-containing GABA_ARs mediate tonic currents. In the 1–2 months old mice, there was no difference of total expression of the GABA_AR α1, δ and γ2 subunits in the brain of the wild-type and the heterozygous and homozygous mutant mice (Fig. 7A–D).

GABA_AR α5 subunit was increased at cell surface in the hippocampus of the Slc6a1^+/S295L mice

Consistent with the previous study on GABA_AR distribution in the rat brain,³⁶ the α5 subunit was abundantly expressed in the hippocampus, with a lower level of expression in the cortex and thalamus and a minimal level of expression in the cerebellum, while the expression of the α1 and γ2 subunits was abundant in all brain regions (Fig. 7A and B and Supplementary Fig. 4 for full-length blots).²⁹ By contrast, the δ subunit was the most abundant in the cerebellum with a modest expression in the cortex and thalamus, but a minimal expression in the hippocampus. The expression pattern of α5 and δ subunits was opposite. The total expression of α5 subunit was increased by ∼20% in the hippocampus of the Slc6a1^S295L/S295L homozygous mice compared with wild-type littermates (Fig. 7F), but this increase of α5 subunit was not observed in the age-matched heterozygous Slc6a1^+/S295L mice (Fig. 7F), Slc6a1^+/A288V mice (Fig. 7E) and the Slc6a1^A288V/A288V homozygous mice (Fig. 7F). The increased α5 subunit expression in the hippocampus of the Slc6a1^S295L/S295L mice could be a compensatory mechanism due to GAT-1 deficiency.

We determined the surface expression of α5 subunit in both Slc6a1^+/A288V and Slc6a1^+/S295L mouse lines with live brain slice biotinylation as the surface expression can more accurately reflect the function of a given protein. We found an increase of α5 subunit at the cell surface in the Slc6a1^+/S295L mice (homozygous mice were not assessed). However, there was no difference in α5 subunit expression between wild-type and Slc6a1^+/A288V mice (Fig. 7). (A288V: 0.993 ± 0.067 for cortex and 0.975 ± 0.053 for cb, 0.945 ± 0.063 for hippocampus and 0.975 ± 0.038 for thalamus; S295L: 1.082 ± 0.06 for cortex and 0.944 ± 0.071 for cb, 1.294 ± 0.093 for hippocampus and 0.936 ± 0.061 for thalamus and Supplementary Fig. 3 for the expression level in cerebellum and thalamus.) This may suggest differential compensatory mechanisms due to differential mutational effects in vivo. Overall, the expression of most GABA_AR subunits analysed in the heterozygous and homozygous mice in both mutant lines had a similar profile as wild-type mice except the increased α5 subunit in the hippocampus of the Slc6a1^+/S295L (at the cell surface) and the Slc6a1 ^S295L/S295L (in total hippocampus) mice (Fig. 7G–J and Supplementary Fig. 4 for full-length blots).

Slc6a1^+/A288V and Slc6a1^+/S295Lheterozygous mouse lines showed unaltered radioligand binding for γ2-subunit and α5-subunit containing GABA_ARs

Small molecules targeting γ2 or α5 subunit-containing GABA_ARs could be utilized as a pharmacological treatment option to restore GABAergic neurotransmission in carriers of SLC6A1 variants. We used radioligand binding at these GABA_AR subtypes to assess their expression pattern in the Slc6a1^+/A288V and Slc6a1^+/S295L mouse lines. We used [³H]flumazenil (1 nM) for α1, α2, α3 and α5/γ2-containing GABA_AR subtypes and [³H]L-655 708 (2 nM) to visualize the α5/γ2-containing GABA_ARs. Brain sections were incubated with either [³H]L-655 708 (2 nM) or [³H]flumazenil (1 nM) to quantify TB to α5- and γ2-containing GABA_AR populations, respectively. NSB was defined by co-incubation with 10 µM flunitrazepam and all values were expressed as SB per mg protein (SB = TB − NSB). Regions of interest were delineated for cortex, hippocampus, thalamus and cerebellum. In all brain regions, the binding of both radioligands was unchanged in the Slc6a1^+/A288V (Fig. 8A and C) and Slc6a1^+/S295L (Fig. 8B and D) heterozygous mice compared with the wild-type controls.

Discussion

Investigations into the cellular mechanisms of GAT-1 function have provided valuable insights into how GAT-1 activity modulates neuronal homeostasis in disorders mediated by SLC6A1 variants.^9,37,38 This study characterized the spatial distribution of GAT-1 in the brains of wild-type and SLC6A1 variants in mice and showed that GAT-1 expression was abundant in major brain regions, including cortex, cerebellum, hippocampus and thalamus in wild-type littermates but was markedly reduced in these regions in both Slc6a1^+/A288V and Slc6a1^+/S295L knockin mice. The mutant GAT-1 protein was detected in the Slc6a1^A288V/A288V homozygous mice, but not in the Slc6a1^S295L/S295L homozygous mice, suggesting that the GAT-1(S295L) protein experiences a faster degradation inside the ER. This relationship was corroborated by the Endo-H digestion that discriminated the mature versus immature form of the GAT-1 protein (Fig. 1E). The GAT-1 protein expression data from HEK 293T cells and the knockin mice suggests that GAT-1(A288V) is more stable than the GAT-1(S295L) protein.

The mutant GAT-1(A288V) protein when expressed under homozygous conditions had ∼30% GABA uptake of the wild-type in HEK293T cells and in mice (data not shown), suggesting that the GAT-1(A288V) protein is functional if it can be trafficked to the cell surface. The mutant GAT-1(S295L) protein had negligible GABA uptake in HEK293T cells and in mice when expressed under homozygous conditions (data not shown). This is consistent with the larger magnitude of reduction of GAT-1(S295L) in HEK293T cells when expressed as a homozygous mutant. In the heterozygous condition, the mutant GAT-1(A288V) protein may oligomerize with the wild-type allele instead of being disposed of quickly. Nonetheless, the total GAT-1 expression and GABA uptake function were comparable in both heterozygous mouse lines. This suggests that net GAT-1 is not a simple addition of GAT-1 produced half by wild-type and half by mutant alleles, and there is an interaction of mutant and wild-type proteins, especially in vivo. We identified that both GAT-1(A288V) and GAT-1(S295L) were more likely to be retained inside the ER in cell models. Considering the mutant GAT-1(A288V) protein is partially functional, the GAT-1(A288V) may have a more complex interaction with the partnering proteins than the GAT-1(S295L) protein in vivo. This merits more detailed investigation with combined approaches from in vitro and in vivo models.

Since GAT-1 and GAT-3 are two major GABA transporters that contribute to GABA uptake in the brain and play a prominent role in modulating tonic and phasic GABA_AR-mediated inhibition,³¹ we also compared GAT-3 expression in the brains of wild-type and SLC6A1 variants in mice. We identified that GAT-3 is much more abundant in the thalamus than in other brain regions in both the wild-type and the mutant mice. Our data indicate that at 1–2 months of age, there was no compensatory increase of GAT-3 expression and function between the wild-type and the heterozygous mutant mice in both Slc6a1^+/A288V and Slc6a1^+/S295L mice. However, it is unknown if there is any change of GAT-3 in Slc6a1^+/A288V or Slc6a1^+/S295L mice at other ages, especially older ages, which merits further investigation in the SLC6A1 mutation knockin mouse models at different developmental stages.

We demonstrated that Slc6a1^+/A288V and Slc6a1^+/S295L mice had comparable levels of GABA uptake in crude live synaptosomes from the forebrain (cerebral cortex, hippocampus and thalamus). Previous work in the GAT-1 knockout mice demonstrated that GAT-1 activity was reduced in the heterozygous and the homozygous mice. Both GAT-1 inhibitors (CL-966 or NNC-711) and GAT-3 inhibitors (SNAP5114) reduced GABA uptake activity in the crude synaptosomes; however, the magnitude of reduction was higher in CL-966- or NNC-711-treated groups. This suggests that the GABA reuptake activity in the forebrain is primarily controlled by GAT-1, while the contribution of GAT-3 on GABA uptake is relatively minor. Both SLC6A1(A288V) and SLC6A1(S295L) mutations caused reduced GABA reuptake in the forebrain of the mutant mice.

Heterozygous GAT-1 knockout mice do not have seizures but have an abnormal behavioral phenotype, while both the Slc6a1^+/S295L and Slc6a1^+/A288V knockin mice have seizures.^28,32 This suggests a possible dominant negative effect from both mutations. In human patient-derived astrocytes, the wild-type allele was more likely to be retained inside the ER than expressed in the normal control astrocytes,³² suggesting the mutant allele may interact with the wild-type allele or the cellular environment in the patient cells is not as favorable as the wild-type. GAT-1 deficiency could cause increased extracellular GABA levels and tonic inhibition, tilting the brain to a pro-seizure state.¹³ We investigated the expression of GABA_AR subtypes in the brains of Slc6a1^+/S295L and Slc6a1^+/A288V knockin mice by biochemistry (wild-type versus heterozygous and homozygous) and autoradiography (wild-type versus heterozygous). Our biochemical data indicate that the α1 subunit of GABA_ARs was abundant in all major brain regions, including the cortex, cerebellum, hippocampus and thalamus, while the α5 subunit was highly expressed in the hippocampus with a minimal level of expression in the cerebellum. The γ2 subunit was abundant in multiple regions, but more abundant in the cortex and hippocampus. The δ subunit was mainly expressed in the cerebellum and thalamus with a very low level of expression in the hippocampus. These data are consistent with early studies on the expression of GABA_AR subunit messenger ribonucleic acid within in situ hybridization and immunohistochemistry.^39-41 GAT-1 is abundantly expressed in all major brain regions. There were no major adaptive changes of GABA_ARs in the heterozygous mice in both the biochemistry and autoradiography studies. However, a modest but significant increase of α5 subunit protein was observed in the Slc6a1^+/S295L heterozygous (surface expression) and homozygous (total expression) mutant mice. Homozygous mice were not assessed by autoradiography, so this finding would need to be followed up in future studies. Importantly, α5 GABA_ARs have been identified to contribute to dendritic spine maturation and excitatory synapse development.^42-44 Considering the life-long disease course, defective GAT-1 activity could cause increased extracellular GABA levels and cause adaptive changes in GABA receptors, including both GABA_A and GABA_B receptors. GABA_B receptor signalling could be augmented by GAT-1 or GAT-3 inhibition, thus potentiating absence-like epileptiform oscillations in the thalamus.⁴⁵ It has been demonstrated that GABA overspill could excessively activate peri-synaptic GABA_ARs, such as δ subunit-containing GABA_ARs, and desensitize the postsynaptic γ2-containing GABA_ARs.^13,46 Based on our extensive work in a large cohort of pathological SLC6A1 variants, partial or complete loss of GAT-1 function is a common mechanism which alters GABAergic signalling and consequently leads to different epilepsy phenotypes. Thus, these data suggest a molecular interplay between GAT-1 function and GABA_AR α5 subunit expression. Future research will be needed to understand the mechanistic role of this interaction in disorders mediated by SLC6A1 variants. In addition, investigation of the potential beneficial effect of GABA_AR α5 subtype selective ligands in Slc6a1 mutant mice may lead to therapeutic options for individuals with GAT-1 deficiency.

Supplementary material

Supplementary material is available at Brain Communications online.

Acknowledgements

Special thanks to Xinyu Gao and Dr Kirill Zavalin for figure reformatting. Imaging data were performed in part through the VUMC Cell Imaging Shared Resource.

Funding

The work was supported by the research grant National Institute of Neurological Disorders and Stroke (NINDS) NS121718 to K.J.Q. The work was also supported by research grants from SLC6A1 Connect and UCB Pharma company to K.J.Q.

Competing interests

The authors report no competing interests.

Data availability

The data supporting the findings of this study are available within the article and its Supplementary material.

References