Epilepsy and Exacerbation of Brain Injury in Mice Lacking the Glutamate Transporter GLT-1

See allHide authors and affiliations

Science  13 Jun 1997:
Vol. 276, Issue 5319, pp. 1699-1702
DOI: 10.1126/science.276.5319.1699


Extracellular levels of the excitatory neurotransmitter glutamate in the nervous system are maintained by transporters that actively remove glutamate from the extracellular space. Homozygous mice deficient in GLT-1, a widely distributed astrocytic glutamate transporter, show lethal spontaneous seizures and increased susceptibility to acute cortical injury. These effects can be attributed to elevated levels of residual glutamate in the brains of these mice.

The extracellular concentration of the excitatory neurotransmitter l-glutamate in the mammalian central nervous system must be kept low to ensure a high signal-to-noise ratio during synaptic activation and to prevent neuronal damage from excessive activation of glutamate receptors (1). This control is achieved by high-affinity, Na+-dependent glutamate transporters in the plasma membrane of neurons and surrounding glial cells (2). The failure or reversal of these transporters may contribute to cellular damage in stroke, trauma, Alzheimer's disease, amyotrophic lateral sclerosis, and Huntington's disease (3). Four subtypes of glutamate transporters have been defined by differences in sequence, pharmacology, tissue distribution, and channel-like properties: GLAST, GLT-1, EAAC1, and EAAT4 (4). EAAC1 and EAAT4 are selectively localized to neurons, whereas GLT-1 and GLAST are astroglial transporters (5). However, the roles of glutamate transporter subtypes in synaptic transmission and neurotoxicity are not known because subtype-specific inhibitors are not available. We therefore generated mice that lack GLT-1, using homologous recombination.

To disrupt the mouse gene encoding GLT-1 in E14 embryonic stem (ES) cells by homologous recombination, we constructed a targeting vector in which the exon encoding the putative third transmembrane segment was replaced with the neomycin resistance gene (Fig. 1A). Four targeted clones were identified from 144 G418- and gancyclovir (GANC)–resistant clones by Southern (DNA) blotting with 5′-flanking and 3′-flanking probes (Fig. 1A). Two mutant clones were separately injected into C57BL/6J blastocysts to produce chimeric animals. Heterozygous animals were identified by Southern blotting and were bred with each other to obtain homozygous animals, which showed the proper structure of the GLT-1 gene by Southern analysis (Fig. 1B). Northern (RNA) blotting showed that brains from homozygous mutant mice contain a hybridizable transcript that is similar in size to the wild-type GLT-1 mRNA (Fig. 1C). Reverse transcription–polymerase chain reaction experiments, followed by Southern blot analysis with an exon-specific probe corresponding to the disrupted region, indicated that the transcript derived from the mutated allele lacks exon 4 (6). However, protein immunoblot analysis (Fig. 1D) detected no GLT-1 protein in the brains of the mutant mice (7). By in situ hybridization with oligonucleotide probes, expression of the four glutamate transporter subtypes was examined in the brains of wild-type and mutant mice (Fig. 1, E to L) (8). Hybridization signals for GLT-1 disappeared completely from the mutant brain (Fig. 1J). The levels of the other glutamate transporter subtype mRNAs were not appreciably affected by mutation. Glutamate uptake in cortical crude synaptosomes of mutant mice was decreased to 5.8% of that in synaptosomes from wild-type mice [mean ± SEM (three animals per group) for the Michaelis constant (Km) (at micromolar concentration) and the maximum uptake velocity (Vmax) (in picomoles per minute per milligram of protein) are, respectively: wild type, 46.1 ± 4.2, 19,716.2 ± 7280.2; mutant, 22.1 ± 0.46, 1139.0 ± 432.1], which suggests that GLT-1 is responsible for the greatest proportion of cerebral glutamate transport (9).

Figure 1

Targeted disruption of the mouse gene encoding GLT-1. (A) Homologous recombination resulted in replacement of the 0.8-kb fragment from the GLT-1 gene, including 224 bp of exon 4 with the neomycin resistance gene (neo). Closed boxes E2 through E4, exons 2 through 4; tk, herpes simplex virus thymidine kinase gene. Restriction sites are as follows: Apa I (A), Eco RV (RV), Kpn I (K), and Not I (N). The 5′- and 3′-flanking probes generated the 9.8-kb Kpn I–Apa I and the 5.0-kb Eco RV fragments from the wild-type GLT-1 gene, respectively, and the 15.2-kb Kpn I–Apa I and the 9.1-kb Eco RV fragments from the properly disrupted gene, respectively. (B) Southern blot analyses of genomic DNA from tails of the wild-type (+/+), heterozygous (+/−), and homozygous mutant mice (−/−). Kpn I–Apa I–digested DNAs and Eco RV–digested DNAs were hybridized with the 5′-flanking probe (left) and 3′-flanking probe (right), respectively. (C) Northern blot analysis of cerebral RNAs from the wild-type (+/+), heterozygous (+/−), and homozygous mutant (−/−) mice. (D) Protein immunoblot analysis of cerebral membrane fractions with antibody to GLT-1. (E through L) In situ hybridization analysis of the parasagittal brain sections of the wild-type [(E) through (H)] and mutant [(I) through (L)] mice at the second postnatal month with oligonucleotide probes specific for the GLAST [(E) and (I)], GLT-1 [(F) and (J)], EAAC1 [(G) and (K)], and EAAT4 [(H) and (L)] mRNAs. Arrows indicate the hippocampal CA1 region in the mutant, which lacks the GLAST and EAAC1 mRNAs. Scale bars, 2 mm.

Mice heterozygous for GLT-1 were indistinguishable from wild-type mice. Homozygous mice were born from heterozygous crosses at the frequency predicted by Mendelian ratios (n = 312): 24.4% wild type, 50.6% heterozygous, and 25.0% homozygous. The body weight and general appearance of the homozygous mice were normal at birth, but homozygous mice gained weight more slowly than did wild-type mice (Fig.2A) and tended to die prematurely (50.0% survival after 6 weeks) (Fig. 2B). Deaths were not preceded by any noticeable health problems. Postmortem analyses revealed no evidence of hemorrhage, infarction, or ischemia that might be associated with cardiovascular failure or stroke and no gross abnormalities in skeletal muscles or visceral organs. However, continuous videotape monitoring of small groups of mice revealed the occurrence of spontaneous epileptic seizures with behavioral patterns similar to those of N -methyl-d-aspartate (NMDA)–induced seizures (10), characterized by explosive running followed by maintained opisthotonus and straub tail (Fig. 2C). In most cases, these mice died within a few minutes of seizure onset. A subset of mutant mice quickly recovered and resumed apparently normal behavior but developed additional seizures and died later.

Figure 2

 Phenotypes of GLT-1 mutant mice. (A) Body weight (mean ± SEM) of homozygous (−/−) (n = 50) and wild-type (+/+) (n = 58) mice. (B) Percentage of postnatal survival of wild-type (+/+) (n = 76) and homozygous (−/−) (n =78) mice. (C) Opisthotonus-like posture in a GLT-1 mutant mouse at postnatal day 35 (P35). (D) EEG of wild-type (+/+) and homozygous (−/−) mice after PTZ administration. The high-voltage sharp wave bursts are underlined.

To confirm profound hyperexcitability in mutant mice, we compared the electroencephalogram (EEG) patterns of wild-type and homozygous mutant mice treated with the convulsant agent pentylenetetrazole (PTZ), a γ-aminobutyric acid receptor antagonist (11). After a single injection of PTZ at a subconvulsive dose of 30 mg per kilogram of body weight (30 mg/kg), high-voltage sharp wave bursts unaccompanied by behavioral changes were detected in mutant mice, whereas no epileptiform discharges were observed in wild-type mice (Fig. 2D). GLT-1 maps to the central region of mouse chromosome 2 (12), near the region of the quantitative trait locus EL-2 of the mouse epilepsy strain (13).

To study the role of GLT-1 at central glutamatergic synapses, we performed electrophysiological analysis in the hippocampal CA1 pyramidal neurons (14). In CA1 hippocampal pyramidal neurons, stimulation of the afferent fibers led to the generation of an excitatory postsynaptic current (EPSC) that contained both the slow NMDA and fast non-NMDA components. No significant difference in the time course of non-NMDA receptor–mediated EPSCs recorded at −90 mV was found between the wild-type [decay time constant, 11.5 ± 0.8 ms (mean ± SEM); n = 8] and mutant (12.5 ± 0.6 ms; n = 7) slices (Fig. 3, A and C). There was no clear difference in the decay time course of NMDA receptor–mediated EPSCs between the wild-type (55.7 ± 1.9 ms; n = 11) and mutant mice (51.4 ± 2.2 ms; n = 9) (Fig. 3, B and D). These results indicate that GLT-1 does not determine the decay rate of EPSCs in the hippocampus and are consistent with the observation that glutamate transporter blockers have no effect on the decay of EPSCs in the hippocampus (15).

Figure 3

Kinetics of glutamatergic synaptic responses and estimation of an increase in glutamate concentrations in the synaptic cleft in hippocampal slices of GLT-1 mutant mice. (A) The non-NMDA receptor–mediated EPSC in the wild-type mouse recorded from a CA1 pyramidal cell with whole-cell patch-clamp techniques at a holding membrane potential of −90 mV. The time constant of the decaying phase of the current in this cell was 12.4 ms. All the traces were an average of 10 consecutive responses. (B) The NMDA receptor–mediated EPSC in the wild-type mouse (control). The decay time constant was 50.1 ms. Application of 400 μMl-AP5, a low-affinity NMDA receptor antagonist, caused a large inhibition of the current (l-AP5). (C) The non-NMDA receptor–mediated EPSC in the mutant mouse. The decay time constant was 11.7 ms. (D) The NMDA receptor–mediated EPSC in the mutant mouse (control). The decay time constant was 49.2 ms.

We next estimated the peak concentration and time course of free glutamate in the synaptic cleft by analyzing the displacement of a rapidly dissociating NMDA receptor antagonist,l-2-amino-5-phosphonopentanoic acid (l-AP5), from NMDA receptors during synaptic transmission (16). The inhibition of NMDA EPSCs by 400 μM l-AP5 in the mutant slices (26.1 ± 4.6%; n = 6) was significantly less than in the wild-type slices (53.0 ± 4.0%; n = 9) (P < 0.001), implying that the peak concentration of synaptically released glutamate is increased in mutant mice and that glutamate remains elevated in the synaptic cleft for longer periods in mutant mice (Fig. 3, B and D). These results suggest that GLT-1 is an important determinant of the clearance of free glutamate from the synaptic cleft. A recent study suggests that the uptake rate of GLT-1 is significantly slower than the predicted time course of synaptically released glutamate (17). Thus, it is likely that binding to the transporter rather than uptake per se removes glutamate from the cleft. Our observation could be accounted for by the absence of glutamate binding sites.

Because excessive synaptic glutamate leads to neuronal degeneration, we performed systematic histological analysis of the entire brain (18). We observed selective neuronal degeneration in the hippocampal CA1 field in 7 of 22 homozygous mutant mice aged 4 to 8 weeks (Fig. 4, A through D). Differences in the selective destruction of hippocampal neurons among individuals likely reflect differences in the occurrence of spontaneous seizures. No obvious morphological abnormalities were found elsewhere (19). These results suggest that GLT-1 is essential for maintaining low extracellular glutamate concentrations and for preventing glutamate neurotoxicity. Because GLT-1 is distributed uniformly throughout the hippocampus, another mechanism that is responsible for the selective neuronal vulnerability in the hippocampal CA1 field of mutant mice may exist, a proposal that can be readily investigated in GLT-1 mutant mice.

Figure 4

Selective neuronal degeneration in the hippocampal CA1 of the GLT-1 mutant mouse (C and D) at P56 as compared to that in the wild-type mouse (A andB). (A through D) Nissl staining. Selective neuronal loss and residual nuclear debris in the hippocampal CA1 pyramidal layer of the mutant mouse are shown [arrows in (C)]. DG, dentate gyrus. Scale bar in (A), 100 μm; in (B), 10 μm.

Experimental induction of injury and ischemia is associated with a large, almost immediate increase in extracellular glutamate concentration (20). To examine the role of GLT-1 in the pathogenesis of acute brain injury, we compared edema development after cold-induced injury in wild-type and mutant mice (21, 22). The homozygous mutant mice were significantly more susceptible to edema than were the wild-type mice (Fig.5A). Edema was 68% greater in the homozygous mutant mice than in the wild-type mice (P < 0.05), and edema size was visibly larger in mutant mice than wild-type mice (Fig. 5B). Our results suggest that GLT-1 normally contributes to the prevention of acute glutamate neurotoxicity after trauma.

Figure 5

Wild-type and mutant mouse brains subjected to cold-induced injury to the left cerebral hemisphere. (A) Vasogenic edema of wild-type (+/+) and homozygous (−/−) littermates after cold-induced brain injury. (B) Comparison of edema formation between wild-type (+/+) (n = 9) and homozygous (−/−) (n = 9) mice after cold-induced injury to the left cerebral hemisphere. Values are mean ± SEM. Asterisk indicates P < 0.05 compared to edema index (20) of wild-type controls, with the use of the Tukey test.

Our results indicate that GLT-1 contributes to the maintenance of extracellular glutamate concentrations at low levels. Without its action, glutamate levels rise enough to cause epilepsy and cell death.

  • * To whom correspondence should be addressed. E-mail: tanaka{at}


Stay Connected to Science

Navigate This Article