Genetic Variation in Neuronal Glutamate Transport Genes and Associations with Post-Traumatic Seizure

Abstract

Objective

Post-traumatic seizures (PTS) commonly occur following severe traumatic brain injury (sTBI). Risk factors for PTS have been identified, but variability in who develops PTS remains. Excitotoxicity may influence epileptogenesis following sTBI. Glutamate transporters manage glutamate levels and excitatory neurotransmission, and they have been associated with both epilepsy and TBI. Therefore, we aimed to determine if genetic variation in neuronal glutamate transporter genes is associated with accelerated epileptogenesis and increased PTS risk after sTBI.

Methods

Individuals (N=253), 18-75yrs with sTBI, were assessed for genetic relationships with PTS. SNPs within SLC1A1 and SLC1A6 were assayed. Kaplan-Meier estimates and log-rank statistics were used to compare seizure rates from injury to 3yrs post-injury for SNPs by genotype. Hazard ratios were estimated using Cox proportional hazards regression for SNPs significant in Kaplan-Meier analyses adjusting for known PTS risk factors.

Results

32 tagging SNPs were examined (SLC1A1: n=28, SLC1A6: n=4). 49 (19.37%) subjects had PTS. Of these, 18 (36.7%) seized within 7days, and 31 (63.3%) seized between 8d-3yrs post-TBI. Correcting for multiple comparisons, genotypes at SNP rs10974620 (SLC1A1) were significantly associated with time-to-first seizure across the full 3yr follow-up (seizure rates: 77.1% minor allele homozygotes, 24.8% heterozygotes, 16.6% major allele homozygotes; p=0.001). When follow-up began day 2, genotypes at SNP rs7858819 (SLC1A1) were significantly associated with PTS risk (seizure rates: 52.7% minor allele homozygotes, 11.8% heterozygotes, 21.1% major allele homozygotes; p=0.002). Adjusting for covariates, rs10974620 remained significant (p=0.017, minor allele versus major allele homozygotes HR: 3.4, 95%CI: 1.3-9.3). rs7858819 also remained significant in adjusted models (p=0.023, minor allele versus major allele homozygotes HR: 3.4, 95%CI: 1.1-10.5).

Significance

Variations within SLC1A1 are associated with risk of epileptogenesis following sTBI. Future studies need to confirm findings, but variation within neuronal glutamate transporter genes may represent a possible pharmaceutical target for PTS prevention and treatment.

Introduction

Traumatic brain injury (TBI) represents an ever-growing public health problem. In the United States, over 2.5 million TBIs occur annually; of these, approximately 300,000 result in hospitalization or death¹. TBI is a major cause of morbidity and mortality, and TBI is increasingly recognized as a disease with many associated chronic health conditions. Individuals with severe TBI (sTBI) have significantly shorter life-spans versus demographically similar, non-TBI, populations². Recent data also show individuals with sTBI are 50 times more likely to die of seizure than age, sex, and racially similar populations².

Post-traumatic seizures (PTS), defined as any seizure occurring after TBI, are classified based on time-to-first seizure relative to injury: immediate (<24hours), early (1-7days), and late (>7days post-injury)³. Temporal classification cut-offs are attributed to differences in causal pathology and risk of seizure recurrence^{4; 5}. PTS incidence varies widely across adult TBI studies, likely due to differences in study design, population, and PTS definitions. In predominantly closed-head injury populations, incidence of immediate/early PTS and late PTS range from 1-12% and 4-19%, respectively^6-10. PTS risk factors like injury severity, specific intracranial pathologies, and patient characteristics have been identified^6-10. Yet, significant variability regarding who develops PTS remains. Evidence concerning factors affecting time-to-first seizure and the process of epileptogenesis after TBI is limited. Factors affecting time-to-first seizure can provide information on potential mechanisms associated with epileptogenesis. Previously, using time-to-event analysis, we reported variation in adenosine regulatory and IL-1b genes as associated with time to first PTS^{11; 12}. Pathological mechanisms involving other secondary injury cascades, such as excitotoxicity, are likely contributors to epileptogenesis and may contribute to increase PTS risk chronically after injury⁴.

Glutamate is the most prominent excitatory neurotransmitter in the human brain. In response to the primary TBI, there is an immediate release of glutamate into the extracellular space and ion channel activation¹³. These phenomena can lead to neuronal depolarization, disrupted cellular metabolism, and excitotoxic glutamate levels⁵. Excitotoxicity may lead to neuronal and astrocytic swelling, mitochondrial damage, cell death, and immediate/early PTS. Seizures can cause over-activation of excitatory amino acid receptors, inducing calcium dependent production of nitric oxide and reactive oxygen species and free radical damage to DNA and cellular membranes⁴. Moreover, after experimental TBI, studies show reduced regional glutamate transporter expression that is maintained chronically post-injury and is modifiable with levetiracetam¹⁴. These observations suggest decreased glutamate clearance, and low-level excitoxicity, is an ongoing mechanism of TBI pathology and contributor to epileptogenesis. Antecedent immediate/early seizure activity may, along with altered glutamate transporter expression, perpetuate excitoxicity and cell death and contribute to epileptogenesis¹⁵.

Genetic variation within glutamate transporter genes may predispose individuals to excitotoxicity after TBI. There are five glutamate transporters in the human central nervous system (CNS), encoded by separate genes. Of these, the SLC1A1 and SLC1A6 genes encode the neuronal glutamate transporters, excitatory amino acid transporters (EAAT) 3 and 4. EAAT3/4 contribute to, but are not the main transporters responsible for extracellular glutamate uptake in most brain regions. However, in brain regions where astrocyte expression is limited, neuronal glutamate transporters may play a more dominant role in glutamate clearance. EAAT4 expression is limited to Purkinje cells, while EAAT3 is expressed on and within multiple neuron types in various regions¹⁶. In addition to glutamate signal termination, studies suggest EAAT3 is vital to glutathione and GABA synthesis^17-19. However, disruption of these functions could increase seizure susceptibility through excitoxicity, decreased antioxidant reserves, or decreased inhibitory neurotransmission. Previous research demonstrates variation within SLC1A1 and SLC1A6, and augmented EAAT3/4 expression, is associated with multiple neurological conditions including multiple sclerosis, schizophrenia, and epilepsy¹⁸. Neuronal glutamate transporter associations with multiple neurological disorders reflect the importance of the excitatory/inhibitory balance and suggest multiple variants may contribute to individual phenotypes and pathologies.

Therefore, we hypothesized genetic variation in neuronal glutamate transporter genes SLC1A1 and SLC1A6 would be associated with epileptogenesis, measured as differences in time-to-first seizure, following sTBI. Additionally, we hypothesized different variants would be associated with PTS in sub-components of a 3-year time frame.

Methods

Study Design and Population

Individuals were recruited to participate in a larger study assessing genetic relationships with TBI outcomes. Patients ages 18-75 presenting consecutively to a Level 1 trauma center with sTBI (Glasgow Coma Scale≤8), with positive head CT findings and requiring extra-ventricular drainage catheter placement for intracranial pressure management, were screened. Patients were excluded if they had penetrating head injury, prolonged cardiac or respiratory arrest prior to admission, or legal proxy consent could not be obtained. To remove genetic effects of population stratification, analyses were limited to individuals listed as white by self/proxy-report (n=25 participants excluded). Individuals with a premorbid seizure history (n=5) were also excluded due to inability to attribute seizure to injury or pre-existing pathology, leaving 253 individuals analyzed. The University of Pittsburgh Institutional Review Board approved all informed consent and study procedures.

Critical Care Management of Severe TBI

All patients were admitted to the neuro-intensive care unit and received treatment consistent with The Guidelines for the Management of Severe Head Injury²⁰. Generally, patients with sTBI received PTS prophylaxis for 1 week.

Demographic and Injury Related Data

Demographic and injury related data were documented at study enrollment. Intracranial pathology type was separated into seven categories using ICD-9 classification derived from radiological findings. These categories were dichotomized by injury types (present or absent) and were not mutually exclusive. Admission GCS was used to establish study eligibility, but the best GCS score during the first 24 hours after admission was used as a covariate in analyses. Injury severity score (ISS) is an overall body injury measure extracted from medical records, with a maximum score of 75 based on survivability of injuries within and across body regions²¹. Medical records were reviewed for antiepileptic drug use during acute care.

Single Nucleotide Polymorphism Selection

Tagging single nucleotide polymorphisms (SNPs) for SLC1A1 and SLC1A6 were evaluated based on data available from the National Center for Biotechnology Information, HapMap Build 36. SNPs, with minor allele frequency (MAF) ≥20% and pairwise r² ≥80% with respect to other known SNPs in the genes, were selected to optimize heterozygosity and to facilitate analysis of common variants among unrelated individuals. Identified SNPs captured variability in the genes including 1000 bases 5' upstream into the promoter region.

DNA Extraction and Genotyping

DNA was extracted from cerebrospinal fluid (CSF), collected via passive drainage, using Qiamp DNA extraction protocol (Qiagen) or from whole blood using a salting out procedure²². DNA samples were genotyped using iPLEX Gold SNP Assay (Sequenom). Double-masked genotype assignments were made for each SNP, and discrepancies were addressed using raw data or re-genotyping. Assays included blind duplicates for quality assessment. Genotypes for SNPs representing variability within SLC1A1 and SLC1A6 were evaluated. All SNPs were evaluated for Hardy-Weinberg Equilibrium (HWE), MAF, and linkage disequilibrium (LD) specific to the study population using Haploview²³ (Figure 1).

Figure 1.

Haploview generated gene map displaying linkage disequilibrium (D’) for SNPs located on SLC1A1 (panel A) and SLC1A6 (panel C). Deeper red colors are indicative of greater D’ values. Panel B shows a magnified view of SNPs on SLC1A1 shown to be associated with time to first seizure in the current analyses (19=rs10974620, 20=rs10815020, 21=rs7858819, 24=rs301430).

Outcome Measure: Post-Traumatic Seizure

Time-to-first seizure following TBI was the primary outcome of interest. PTS status was obtained by retrospective review of all electronic inpatient and outpatient medical records available from our medical center. Date of first seizure was determined by ambulance and/or emergency room report, inpatient progress or nursing note, EEG report, patient history, and discharge or transfer summaries. Medical record notation referring to convulsions, seizures, status epilepticus, or seizure disorder was considered evidence of seizure occurrence. Date of death was also extracted from medical records or from social security death data (http://www.ssa.gov/sitemap.htm). All participants were followed until date of first seizure or date of death. Follow-up was censored at 3 years post-injury.

Statistical Analysis

Analyses were completed using SAS-9.4 (Carry NC) and R-3.0.3. All genotyped participants meeting eligibility criteria were included in analyses. Demographic and injury characteristics were compared between individuals who did (versus did not) seize at different time-points post-injury, using chi-square and Kruskal Wallis tests as appropriate. Individuals with documented seizures were separated into groups based on time of first seizure (i.e. immediate, early, late). Due to small sample sizes, immediate and early groups were collapsed for comparison of demographic and injury characteristics.

Among individuals with seizures, Chi-square analyses were conducted, using Fisher’s exact test when appropriate, to determine if genotype frequencies differed by time of first seizure (i.e. immediate, early, late). Immediate and early seizures were again combined and compared to late seizure.

Time-to-event analyses were used to address the primary hypothesis regarding genetic variation and post-TBI epileptogenesis over a 3-year time period. Due to LD, i.e. correlation among selected SNPs, the effective number of tests conducted was smaller than the number of SNPs screened. The minimum number of effective tests (M_eff) was calculated using methods based on eigenvalues^{24; 25}. The M_eff was calculated for SLC1A1 and SLC1A6 independently, and results were summed. A Bonferroni correction was then applied to the original α=0.05 using the total M_eff as the number of independent tests for subsequent time-to-event analyses. SNPs that were statistically significant after multiple comparison correction were further evaluated using Cox regression.

Kaplan-Meier curves were used to estimate seizure rates at three years post-injury, considering the full follow-up period (i.e. time of injury through 3 years post-TBI), for individual SNPs by genotype, and rates were compared using the log-rank statistic. Cox proportional hazards regression was used to estimate hazard ratios (HR) for SNP genotypes that demonstrated significantly different Kaplan-Meier estimated PTS rates based on Bonferroni corrected p-values derived from log-rank statistics. Cox regression models were then adjusted for demographic and injury characteristics that differed significantly across seizure groups (no seizure, immediate/early, late seizure). Proportionality assumptions were examined for all variables. All time-to-event analyses were repeated 1) where immediate seizures were removed by beginning the follow-up period day 2 post-injury (individuals with seizure or expiring before day 2 excluded) and 2) where both early and immediate seizures were removed by beginning the follow-up period on day 8 post-injury (individuals with seizure or expiring before day 8 excluded), to specifically examine late PTS.

For each follow-up timeframe, a gene risk score (GRS) was created using all SNPs that were nominally significant (p<0.05) in Kaplan-Meier analyses in order to explore possible additive effects of having multiple risk genotypes. The number of SNPs for which an individual was homozygous for the minor allele (risk genotype for all SNPs based on Kaplan-Meier results) was summed. Subsequent GRSs were then analyzed for associations with time-to-first seizure in their respective follow-up period.

Power analyses were completed, using an alpha threshold corrected for multiple comparisons and based on a minor allele homozygosity rate of 5%, to detect a significant difference by genotype between two survival curves over a range of effect sizes using the log-rank test.

Results

We identified and genotyped 31 SNPs from SLC1A1 and 4 SNPs from SLC1A6. Three SNPs on SLC1A1 failed to genotype using the iPLEX Gold SNP Assay (rs10815018, rs12378107, rs7864309), and no data pertaining to these SNPs were included in analyses. All other SNPs were in HWE. Therefore, we examined 28 SNPs from SLC1A1 and 4 from SLC1A6 (Supplemental Table 1). M_eff calculations indicated 16 independent tests (14 and 2 for SLC1A1 and SCL1A6, respectively), resulting in a Bonferroni adjusted significance level of 0.003.

Two hundred fifty-three individuals met all inclusion criteria and were genotyped. Similar to other studies involving sTBI, our study population was predominantly men (79.5%), and the average age was 35.3 years old. The majority of individuals had sTBI as determined by best in 24hr GCS score (91.7%); the average best in 24hr GCS score for the total cohort was 6 (Table 1).

Table 1.

Population and Injury Characteristics by Seizure Status N (%) unless noted

	No Seizure	Immediate/Early Seizure	Late Seizure	P value*
Sample Size	204 (80.6)	18 (7.1)	31 (12.3)	---
Age at Injury, mean (SD)	35.40 (15.71)	37.17 (16.57)	33.65 (13.47)	0.825
Sex, males	163 (79.9)	14 (77.8)	24 (77.4)	0.935
Admission GCS				0.543
Severe (3-8)	187 (91.7)	17 (94.4)	28 (90.3)
Moderate (9-12)	16 (7.8)	1 (5.6)	2 (6.5)
Mild (13-15)	1 (0.5)	0 (0)	1 (3.2)
ISS, mean (SD)	35.87 (9.69)	34.94 (8.21)	31.7 (7.8)	0.057
Received Acute Seizure Prophylaxis	192 (94.1)	18 (100)	31 (100)	0.422
Depressed Skull Fracture	28 (13.7)	3 (16.7)	10 (32.3)	0.039
Subdural Hematoma	119 (58.3)	13 (72.2)	26 (83.9)	0.013
Subarachnoid Hemorrhage	140 (68.6)	12 (66.7)	20 (64.5)	0.874
Diffuse Axonal Injury	65 (31.9)	5 (27.8)	8 (25.8)	0.814
Epidural Hemorrhage	27 (13.2)	4 (22.2)	7 (22.6)	0.226
Contusion	101 (49.5)	8 (44.4)	14 (45.2)	0.863
Intraventricular Hemorrhage	66 (32.4)	5 (27.8)	8 (25.8)	0.788
Intracerebral Hemorrhage	73 (35.8)	7 (38.9)	10 (32.3)	0.880

^*p-value for chi-square and Kruskal-Wallis tests comparing 3 groups

Overall, 49 individuals (19.4%) developed PTS. Of these, 12 (24.5%) seized within 24 hours, 6 (12.2%) seized within the first 7 days, and 31 (63.3%) seized between 8d and 3yrs post-injury. Depressed skull fracture and subdural hematoma (SDH) occurred more frequently among individuals who seized compared to those who did not seize (Table 1). Among individuals who seized, there were no differences in genotype frequencies between individuals seizing immediately and early versus those seizing late (data not shown).

SDH and depressed skull fracture frequencies differed significantly by seizure status. SDH was included as a covariate in all Cox regression models. Since depressed skull fracture did not meet proportionality assumptions, models were stratified by presence/absence of depressed skull fracture. All SNPs met the assumption of proportionality.

Time-to-Event Across Full Follow-up Period

We found significant differences in seizure rates by genotype for rs10974620 (p=0.001), located on SLC1A1, when assessing the full follow-up period (Table 2). 24.8% of major allele homozygous (CC), 16.6% of heterozygous, and 77.1% of minor allele homozygous (GG) individuals had a seizure during follow-up. Among those with seizure, the average time-to-first seizure was twice that for heterozygous and major allele homozygous individuals versus minor allele homozygotes (786, 794, 384 days, respectively). Three additional SNPs in SLC1A1 (rs10815020, rs7858819, and rs301430) were nominally associated with seizure rates. For each SNP, minor allele homozygotes had the highest risk of seizures 3yrs post-injury and the shortest time-to-first seizure (Figure 2).

Table 2.

SNPs in SLC1A1 with Significantly Different Seizure Rates Determined by Comparison of Kaplan Meier Curves using Log Rank Statistic

	Full Follow-Up		No Immediate Events		No Immediate or Early Events
	3Yr Seizure Rate (%)	P-value	3Yr Seizure Rate %	P-value	3Yr Seizure Rate %	P-value
rs10974620		0.001		0.004		0.044
CC	24.8		20.3		19.7
GC	16.6		12.0		10.7
GG	77.1		71.4		66.7
rs10815020		0.007		0.050		0.134
AA	25.8		22.6		21.9
AG	21.1		14.9		13.8
GG	58.4		48.0		42.9
rs7858819		0.035		0.002		0.009
CC	23.7		21.1		20.5
CT	21.2		11.8		10.6
TT	56.4		52.7		47.5
rs301430		0.033		0.018		0.072
TT	25.8		22.1		20.7
CT	18.9		12.3		12.3
CC	49.3		42.5		38.1

All genotypes in order of major allele homozygous, heterozygous, minor allele homozygous

Figure 2.

Kaplan Meier estimates for time to first seizure by SLC1A1 SNP rs10974620 genotypes for full follow-up (Time of Injury to Three Years).

In both unadjusted and multivariable adjusted Cox regression, we found significant differences in seizure risk by SNP rs10974620 genotypes (p=0.004 and p=0.017, respectively). Individuals homozygous for the minor allele had a significantly higher hazard of seizure (unadjusted HR=4.08; adjusted HR=3.43) versus major allele homozygotes (Table 3). There was no significant difference in hazards between heterozygotes and individuals homozygous for the major allele. In adjusted models, SDH was significantly associated with increased seizure risk (HR=2.36, p=0.016).

Table 3.

Results From Unadjusted and Adjusted Cox Proportional Hazards Regression Models for Two SNPs in SLC1A1

Model	Hazard Ratio	95% Confidence Interval	P-Value
Unadjusted Models
rs10974620 Ref = CC			0.004
CG	0.67	0.34 – 1.34
GG	4.08	1.58 – 10.55
rs7858819 Ref = CC			0.005
CT	0.51	0.22 – 1.18
TT	3.90	1.35 – 11.31
Adjusted Models*
rs10974620 Ref = CC			0.017
CG	0.68	0.34 – 1.35
GG	3.43	1.26 – 9.34
rs7858819 Ref = CC			0.023
CT	0.56	0.24 – 1.32
TT	3.39	1.10 – 10.46

rs10974620 from full follow-up model;

rs7858819 from model beginning day 2 post-injury

^*Adjusted for subdural hematoma, stratified by depressed skull fracture

No SNPs on SLC1A6 had significantly different time-to-first seizure based on Kaplan Meier analyses, and were not further included in Cox models.

Time-to-Event Removing Immediate Seizures

When follow-up began on post-injury day 2 (i.e. only early and late seizures included), there were significant differences in the 3-year seizure rates by genotype for SLC1A1 SNP rs7858819 (p=0.002). Minor allele homozygous (TT) individuals had the highest seizure rates (52.7%) versus heterozygotes (11.8%) and major allele homozygotes (CC; 21.1%) (Figure 3). Among those with seizure, minor allele homozygous individuals had the shortest time-to-first seizure (270 days) versus major allele and heterozygotes (520 and 937 days, respectively). There were nominal differences by genotype for SLC1A1 SNP rs10974620 (p=0.004) (Table 2). In Cox regression, seizure risk for SNP rs7858819 minor allele homozygous individuals was significantly greater versus major allele homozygotes (HR=3.9, p=0.005). After adjusting for SDH and stratifying by depressed skull fracture status, rs7858819 genotype effects were attenuated, but remained significantly associated with seizure risk (HR=3.39, p=0.023; Table 3). In the adjusted model, SDH was also significantly associated with seizure (HR=3.11, p=0.013).

Figure 3.

Kaplan Meier estimates for time to first Seizure by SLC1A1 SNP rs7858819 genotypes for follow-up beginning day 2 post-injury to three years (individuals seizing or expiring before day 2 excluded).

Time-to-Event Examining Late Seizures Only

To determine if there were significant associations with genotypes in epileptogenesis of late PTS, analysis was restricted to follow-up beginning on post-injury day 8. Associations with genotypes at SNPs rs10974620 (p=0.044) and rs7858819 (p=0.009) were nominally significant (Table 2).

Gene Risk Scores

Across the full follow-up, Kaplan-Meier estimates differed significantly for individuals with no risk genotypes, one, or more than one risk genotypes (3yr seizure rates: 16.7, 45.5, 42.9%, respectively; log-rank p-value <0.001). Results were similar when a gene risk score was calculated for follow-up beginning on day 2. 3yr seizure rates were 12.8, 33,3, 33.3% for individuals with 0, 1, or >1 risk genotype (log-rank p-value=0.005). However, there were no significant differences between individuals having 1 risk genotype versus those with >1risk genotype (data not shown).

Power Estimates

The study was powered to detect only large effect sizes. Based on our data, we had 37% power to detect a HR=3.0, 77% power to detect a HR=4.0, and 95% power to detect a HR=5.0.

Discussion

Genetic variation and changes in neuronal glutamate transporter expression are associated with seizure and epilepsy. TBI results in decreased glutamate transporter expression¹⁴, perpetuating ongoing excitotoxic injury after TBI and facilitating a pro-epileptogenic environment. However, it remains unclear if genetic variation in neuronal glutamate transporters affects epileptogenesis or seizure development following TBI. Thus, we examined associations between SCL1A1 and SLC1A6 genetic variation and epileptogenesis, measured by time-to-first seizure, among individuals with sTBI.

We found genetic variation in SLC1A1, but not SLC1A6, was associated with reduced time-to-first seizure and increased seizure risk up to 3-years post-injury. Individuals homozygous (GG) for the SLC1A1 SNP rs10974620 minor allele had significantly higher seizure risk over this period, even after adjusting for relevant covariates. Individuals homozygous (TT) for the SLC1A1 SNP rs7858819 minor allele also had greater PTS risk in multivariable models when follow-up began day 2 post-injury. Both SNPs were nominally associated with other time periods characterized. We found no significant differences in seizure risk when comparing individuals with 1 risk genotype to those with >1 risk genotype in each timeframe, suggesting no additive genetic effects. High LD among genotypes for SNPs significantly associated with seizure risk within our study population, ~2,600bp from one another within the same intron, is one possible explanation for this finding (Figure 1). Possible differences in SNP associations with PTS over time indicates future work should focus on if/how genetic variants within SLC1A1 influence temporally dynamic PTS risk post-injury.

One possibility for the lack of associations between genetic variation in SLC1A6 and PTS is that unlike SLC1A1, where EAAT3 is expressed ubiquitously through the brain, SCL1A6 associated EAAT4 expression is limited primarily to cerebellar purkinje cells²⁶. Aside from some forms of myoclonic epilepsies²⁷, the cerebellum is not a brain region often associated with epileptogenesis. However, some reports suggest EAAT4 can be expressed in forebrain astrocytes²⁸, which can be down-regulated in an experimental model of TBI²⁹, lending some rationale for their evaluation with PTE risk.

SLC1A1 encodes EAAT3, is located on chromosome 9p24, and is ~97kb in length. SNPs rs10974620 and rs7858819 are both located within the second intron. We used the most recent Genome Reference Consortium data, GRCh38, from Utah residents with northern/western European ancestry (CEPH; http://hapmap.ncbi.nlm.nih.gov/citinghapmap.html.en) to explore gene regions around rs10974620 and rs7858819. In the CEPH population, little LD information for rs10974620 is available. However, rs7858819 may tag a region containing multiple functional variants as it is in LD with a 13.5kbp region extending from intron 2 into intron 5 that includes multiple missense polymorphisms.

EAAT3 terminates post-synaptic action and maintains physiological glutamate levels, but is not as critical for terminating glutamate signaling compared to glial glutamate transporters³⁰. EAAT3’s binding affinity and synaptic location suggests it has a more prominent role in glutamate signal termination in pathological conditions involving elevated extracellular glutamate levels³¹. EAAT3 also facilitates cysteine transport³², and thus, cysteine dependent glutathione (antioxidant) production¹⁹. EAAT3 is highly expressed on glutamatergic and GABAergic neurons¹⁶ in cortex, hippocampus, cerebellum, and basal ganglia³³. Regional expression suggests EAAT3 mediated glutamate transport supplies GABAergic neurons with intracellular glutamate required for GABA production. Using a similar cohort, our laboratory previously reported variation in the glutamic acid decarboxylase (GAD1) gene, responsible for synthesizing GABA from glutamate, as associated with increased late PTS risk³⁴. Therefore, disrupted EAAT3 function and/or expression may reduce antioxidant reserves and impair GABA production, subsequently increasing excitatory tone and facilitating epileptogenesis.

Multiple animal models have examined EAAC1 (EAAT3 rodent analog) in epilepsy. EAAC1 antisense treatment reduces EAAC1 availability and increases epileptogenesis in a dose dependent manner³⁵. Also, increased EAAC1 expression is required to manage elevated glutamate levels associated with seizures³⁶. Chemically induced epilepsy models show increased EAAC1 expression 24 hours after seizure, with lower EAAC1 levels associated with increased epilepsy susceptibility³⁷ and EAAC1 trafficking to the intracellular space during early epileptic activity³⁸. Taken together, low EAAC1 expression increases seizure susceptibility, and EAAC1 expression/location within the epileptic brain may influence glutamatergic mechanisms of epileptogenesis.

Clinical EAAT3 expression studies report individuals with temporal lobe epilepsy (TLE) have altered neuronal EAAT3 mRNA compared to controls^{36; 39}. Differences in EAAT3 immunoreactivity were associated with hippocampal sclerosis, with increased EAAT3 immunoreactivity occurring on granule cells from sclerotic regions³⁹. Conversely, among individuals with pharmacoresistent neocortical epilepsy, EAAT3 expression was decreased in epileptic regions⁴⁰.

Multiple candidate gene studies report SLC1A1 genetic variation associated with psychiatric conditions including post-traumatic stress disorder⁴¹, autism spectrum disorder⁴², and schizophrenia¹⁸. The most data regarding human SLC1A1 genetic variation and psychiatric disorders is reported with obsessive-compulsive disorder (OCD). Family based linkage studies and case-control association studies of unrelated individuals have reported SLC1A1 variation is associated with OCD diagnosis and age of onset^{43; 44}. SLC1A1 haplotypes associated with OCD include some SNPs examined within our current analysis^{43; 44}, but SNPs included and risk allele designations differ across studies. One study reported a three SNP haplotype, including rs301430 and rs7858819 C-alleles, was associated with OCD⁴⁴. Though not always drawing consistent conclusions regarding risk alleles, these studies provide evidence that SLC1A1 genetic variation is associated with pathological phenotypes.

In addition to activity-related expression changes, EAAT3/EAAC1 expression and trafficking can be modified via post-translational mechanisms and interaction with multiple kinases¹⁸, affecting glycosylation and phosphorylation sites important for transporter function and post-translation regulation¹⁶. EAAT3 also interacts with intracellular proteins for proper anchoring on cell membranes and EAAT3 trafficking¹⁸. Specific alleles may result in changes to amino acid residues or protein misfolding, disrupting these interactions and affecting membrane protein expression.

Additional studies are needed to examine differences in SLC1A1 expression among individuals with/without PTS. Further research regarding how genetic variation may effect translation, expression, and/or EAAT3 trafficking is essential to assess how pharmacological modulation may reduce PTS. Experimental TBI models that lead to PTS may also facilitate investigation of how glutamate concentration and reduced EAAC1 expression affects epileptogenesis, and if levetiracetam associated increases in post-TBI glutamate transporter expression¹⁴ can reduce epileptogenesis. Additionally, EAAT3’s function differs across neuro-developmental phases¹⁸. Thus, SLC1A1 genetic variation effects on PTE may vary across the age spectrum of individuals sustaining TBI. TBI-induced EAAT3 disruption may interact with genetic variation to impact neuroplasticity during the post-injury period.

Unlike EAAT3, EAAT1 and EAAT2 are located predominantly on glia and are the major transporters limiting the half-life of released glutamate in the extracellular space¹⁸. These transporters are also involved in glutamate recycling for neuronal vesicular release via the glutamate-glutamine cycle⁴⁵. Given the relative differences in neurobiological role, location, and function for these transporters, future work should assess whether glial transporters are independently associated with epileptogenesis post-TBI and how these transporters may work collectively with other candidate gene variants to affect excitatory and inhibitory pathways influencing PTS.

Our results represent novel insights regarding genetic SLC1A1 variation and PTS risk. However, our data show that with our sample size, PTS rates, and gene variant distributions, we were not adequately powered to detect a HR less than 4.0 when comparing time-to-event curves between genetic polymorphisms. Time-to-first seizure was classified based on an intensive medical record review of individuals cared for through the largest health care provider in the geographic region. However, seizure status and time-to-first seizure may have been misclassified due to missing data on healthcare provided outside this system. To minimize differences in allelic frequency by race and ancestry, we limited our analyses to individuals self-reporting race as white; however, residual population stratification may remain. We only included individuals with sTBI, and results may not generalize to less severe TBI populations. More studies are needed to replicate our findings and assess associations with functional SNPs within the LD block implicated in our study. Increasing our knowledge of genetic variants affecting PTS development may improve prognostic seizure models (In Revision), possibly enabling researchers and clinicians to assess more accurately the probability of individual PTS development. If validated, these results may represent an innate factor by which to identify individuals with increased PTS risk; also EAAT3 may be a potential therapeutic target for PTS prevention and treatment.

Key Points.

1. Genetic variation within neuronal glutamate transporters may effect epileptogenesis at different times post-TBI

2. Minor allele homozygous individuals at SNP rs10974620 (SLC1A1) were at an increased PTS risk from time-of-injury to 3yrs post-injury

3. Minor allele homozygous individuals at SNP rs7858819 (SLC1A1) had an increased early and late PTS risk from day 2 to 3yrs post injury.

4. Future work is needed to confirm findings; but genetic variation in excitatory/inhibitory pathways may affect PTS risk