Differential Glutamate Dehydrogenase (GDH) Activity Profile in Patients with Temporal Lobe Epilepsy


Address correspondence and reprint requests to Dr. J.C.K. Lai at Campus Box 8334, Department of Pharmaceutical Sciences, College of Pharmacy, Idaho State University, Pocatello, ID 83209, U.S.A. E-mail –lai@otc.isu.edu

Present address of Dr. Malthankar-Phatak: 14–1-10, Center for Neural Recovery and Rehabilitation Research, Helen Hayes Hospital/NY State Department of Health, West Haverstraw, NY 10993, U.S.A.


Summary: Purpose: Pathophysiologic mechanisms underlying temporal lobe epilepsy (TLE) are still poorly understood. One major hypothesis links alterations in energy metabolism to glutamate excitotoxicity associated with seizures in TLE. The purpose of this study was to determine whether changes in the activities of enzymes critical in energy and neurotransmitter metabolism contributed to the alterations in metabolic status leading to the excitotoxic effects of glutamate.

Methods: Activities of four key enzymes involved in energy metabolism and glutamate cycling in the brain [aspartate aminotransferase (AAT), citrate synthase (CS), glutamate dehydrogenase (GDH), and lactate dehydrogenase (LDH)] were measured in anterolateral temporal neocortical and hippocampal tissues obtained from three different groups of medically intractable epilepsy patients having either mesial, paradoxical, or mass lesion–associated temporal lobe epilepsy (MTLE, PTLE, MaTLE), respectively.

Results: We found that GDH activity was significantly decreased in the temporal cortex mainly in the MTLE group. A similar trend was recognized in the hippocampus of the MTLE. In all three patient groups, GDH activity was considerably lower, and AAT and LDH activities were higher in cortex of MTLE as compared with the corresponding activities in hippocampus (p < 0.05). In the MTLE cortex and hippocampus, GDH activities were negatively correlated with the duration since the first intractable seizure.

Conclusions: Our results support the hypothesis suggesting major alteration in GDH activity mainly in the MTLE group. It is proposed that significant alterations in the enzyme activities may be contributing to decreased metabolism of glutamate, leading to its accumulation.

The pathophysiologic mechanisms underlying temporal lobe epilepsy (TLE), one common form of intractable epilepsy (1), are poorly understood. Increased levels of extracellular glutamate, an excitatory neurotransmitter, characterize hippocampal seizure foci in TLE. Extracellular glutamate in patients with TLE with hippocampal sclerosis (MTLE) (2,3) begins to increase in the epileptogenic hippocampus just before the onset of a seizure and remain elevated for the duration of the seizure (3). Elevated extracellular glutamate also is observed in spiking cortex during seizures (4,5) and could contribute to increased cerebral excitability and ongoing toxicity (6,7).

Both glucose metabolism and blood flow are drastically decreased interictally in epileptogenic brain regions (8). The phosphocreatine-to-adenosine triphosphate (PCr/ATP) ratios and N-acetyl-aspartate to choline or creatine [NAA/(Ch or Cr)] ratios also are significantly decreased not only in areas with or without significant neuronal loss (9,10). Thus one may argue that a fundamental derangement of the various components of brain oxidative metabolism exists in TLE (8–10).

In the glutamate/glutamine cycle, the major steps that require energy are loading, transport, and reprocessing of vesicles (11,12); transport of glutamate into glia (13); and glutamine synthesis by glutamine synthetase in astrocytes (14,15). In an anesthetized rat, changes in the rate of glutamate/glutamine cycling are in almost an ∼ 1:1 relationship with changes in the rate of cortical neuronal glucose oxidation (16), suggesting that the glutamate/glutamine cycling is tightly coupled to brain energy metabolism. Consequently, any alterations in brain glucose oxidative metabolism will lead to changes in glutamate metabolism. Furthermore, in the surgically resected hippocampus of TLE patients, the rate of glutamate/glutamine cycling is very low, the lowest being in the patients with hippocampal sclerosis (17).

Glutamate dehydrogenase (GDH) catalyzes a reversible reaction, mediating the interconversion of glutamate from α-ketoglutarate and vice versa (18). Alternatively, glutamate, in the presence of oxaloacetate, is converted to aspartate and α-ketoglutarate by aspartate aminotransferase (AAT) (18). Both α-ketoglutarate and aspartate enter the tricarboxylic acid (TCA) cycle and are subsequently metabolized to CO2: the electron transport and oxidative phosphorylation associated with the fluxes of the TCA cycle and the respiratory chain lead to ATP synthesis. Although both GDH and AAT are the key enzymes that link glutamate/glutamine cycling to the TCA cycle, some studies suggest that GDH, rather than AAT, might be more responsible for the entry of glutamate into the TCA cycle (14). Thus some workers have argued that astrocytes can use glutamate as a substrate for energy metabolism (14) and particularly when the extracellular glutamate levels increase (19).

Alteration in brain energy metabolism is implicated in the pathophysiologic mechanisms underlying glutamate excitotoxicity associated with seizure disorders. Our hypothesis is that alterations in GDH and AAT activities may contribute to accumulation of extracellular glutamate in TLE patients. We therefore studied activities of four key enzymes that are critical components of brain glutamate and energy metabolism to determine their pathophysiologic role in TLE. A key feature of this study is the use of tissue samples derived from patients during neurosurgery, thereby bypassing the problems associated with the use of autopsy samples, which are known to be complicated with postmortem artifacts (20).



All substrates, enzymes, and other chemicals were purchased from Sigma (St. Louis, MO, U.S.A.).

Tissue collection

The hippocampus and anterolateral temporal neocortex were removed for seizure control from patients with medically intractable TLE according to procedures previously described (21), at the Yale–New Haven Hospital Epilepsy Surgery Program. As the tissue was removed by excision rather than by suction, it was histologically intact and in good condition. After removal, the tissue was immediately frozen on dry ice and then stored at −80°C until further use. All resections were performed for therapeutic reasons. Informed consent was obtained from all patients before they underwent surgery. The patient characteristics, clinical pathology, and medical history are briefly summarized in Table 1. The foci of the seizures in the patients that underwent the surgery were established by using techniques routinely employed for this purpose, including MRI and EEG. The localizations of the seizure focus obtained by using EEG and MRI findings and postsurgical outcomes for the patients are summarized in Table 2. The samples collected during surgery from the subjects in this study were “good” (i.e., histologically intact) mesial and lateral temporal specimens, as confirmed by using Nissl and immunostains for neuropeptide Y, substance P, and somatostatin, as previously described (22–28). Neuronal cell counts were determined on the tissue samples from all patients, as previously described (29).

Table 1. Patient demographics
Case no.ClassificationGenderAge at surgery (yr)Years since first unprovoked seizureAEDs at surgeryPathology
  1. AEDs, antiepileptic drugs; VPA, valproate; CBZ, carbamazepine; LTG, lamotrigine; PHT, phenytoin; OXC, oxcarbazepine; GBP, gabapentin; Li, lithium; Hipp, hippocampal; GDH, glutamate dehydrogenase; AAT, aspartate amino transferase; CS, citrate synthase; MTLE, mesial temporal lobe epilepsy; PTLE, paradoxical temporal lobe epilepsy; MaTLE, mass lesion–associated temporal lobe epilepsy.

1MTLEF2925   VPA, CBZHipp. sclerosis
2MTLEF2927   CBZHipp. sclerosis
3MTLEF2424   CBZHipp. sclerosis
4MTLEM4843   CBZHipp. sclerosis
5MTLEM5140   CBZ, LTGHipp. sclerosis
6MTLEM2825   VPA, LTGHipp. sclerosis
7PTLEF29 2   PHT, VPANormal
8PTLEF41 6   LTG, CBZNormal
9PTLEM3810   LTG, PHTNormal
10PTLEM46 9   OXCNormal
11PTLEM3820   VPA, GBP, PHTNormal
12MaTLEF 7 6   VPA, CBZGanglioma
13MaTLEF5720   LTG, CBZGanglioma
14MaTLEM6214   CBZ, PHTOligodendroglioma
15MaTLEM17 1.5NoneFocal fibrosis
Table 2. Seizure foci localization based on EEG and postsurgical outcome in patients
Case no.Seizure foci localization on scalp EEGMRI findingsSeizure frequency after surgery (per mo)Seizure description after surgery
  1. R, Right; L, Left; MCA, middle cerebral artery; CP, drooling, difficulty with speech, loss of contact, rolling around, body twitching, lip smacking; GTC, urinary incontinence, increased salivation, spitting, falls; AED, antiepileptic drug.

1UnavailableL mesial temporal sclerosisCP, three/moCP (Drooling, difficulty with speech, loss of contact
2Bilateral anterior temporal ictal and interictalAtrophy of right hemisphere, including amygdala and hippocampus; increased T2 signal of R hippocampus0Seizures stop with no aura (no AED)
3L hemisphere interictal and ictalL hemiatrophy involving amygdala, basal ganglia and thalamus; L hippocampal atrophy with signal changesSP, one eventSeizures stop with no aura (requires AED)
4UnavailableL hemiatrophy due to MCA infarction; L hippocampal sclerosis0Seizures stop with no aura (requires AED)
5R temporal interictal, ictal unlocalizedR hippocampal sclerosis0Seizures stop with no aura (requires AED)
6Interictal unlocalized, ictal L hemisphereL temporal lobe atrophy0Seizures stop with no aura (requires AED)
7L posterior temporal interictal, ictal unlocalizedRounded hyperintense mass, L amygdala, extending to the hippocampal headUnknown frequencyReported seizures daily, clustering around menses
8Right temporal interictal and ictalBilateral hippocampal atrophy0Reported spells of visual blackout lasting ∼5 s, involving both eyes with pain; ≤15 times per day
9Interictal unlocalized, ictal left temporalL hippocampal atrophy, no signal changesCP, one every 2 wkNocturnal seizures only, one every 2 weeks; seizures not observed but patient wakes up and “feels funny,” tired, irritable, and emotionally labile, with crying spells typical of postictal states
10L temporal interictal and ictalR temporal lobe lesion, adjacent to hippocampal head and amygdalaCP, one/6 mo, CP + GTC, two/6 moSeizures decreased
11R anterior temporal interictal, right hemisphere ictalR parietal lobe abnormality; mild L hippocampal atrophy; no signal changes0Seizures stop with no aura (requires AED)
12UnavailableTumor, L hippocampus0Seizures stop with no aura (requires AED)
13T temporal interictal, R hemisphere ictalHyperintense abnormality of R amygdala and perihippocampal gyrus, extending into hippocampal head and inferior aspect of basal ganglia0Seizures stop with no aura (requires AED)
14R temporal interictal and ictalHyperintense abnormality, R amygdala and parahippocampal gyrus; normal hippocampus0Seizures stop with no aura (requires AED)
15UnavailableCystic lesion, R parahippocampal gyrus0Seizures stop with no aura (no AED)

The specimens used in this study were assigned to one of three groups based on postsurgical neuropathologic (e.g., neuronal cell counts), immunohistochemical (e.g., localization of neuropeptides and proteins), and electrophysiological characteristics previously described (22–29). MTLE is characterized by classic mesial temporal sclerosis, histologically observed as a loss of >50% of the dentate granule cells and pyramidal neurons in the CA fields of the hippocampus, differential loss and sprouting of interneurons that express various neuropeptides in the dentate area, and where the hippocampus is the focus of the seizures in most cases as shown by electrophysiologic data (22–28). A second group, called paradoxical TLE or PTLE, has no hippocampal sclerosis or mass lesion, no history of injury or medial temporal atrophy, no neuropathologically identifiable mass lesion (thus distinguishing them from the MaTLE group), and no observed anatomic reorganization of the hippocampus (22,28). The third group is called mass lesion–associated TLE or MaTLE (28). MaTLE patients have an extrahippocampal temporal lobe mass but without any notable anatomic reorganization of the hippocampus (22,28). The patients are classified into three TLE groups based on postsurgical pathological and immunohistochemical findings from the resected tissue and not based on presurgical MRI findings. Thus for example, although the presurgical MRI findings for patient 6 and patient 9 are similar, they are classified into different TLE groups, as patient 6 differs from patient 9 in the absence of hippocampal sclerosis and immunohistochemical reorganization (Tables 1 and 2). For detailed discussion of TLE classifications, see reference 22.

Enzyme assays

The tissue wet weight was recorded, and an ∼10% (wt/vol) homogenate was prepared with a Dounce homogenizer by using isolation medium (32 mM sucrose, 5 mM HEPES, 5 mM EDTA, pH adjusted to 7.4 by using Tris) on ice. AAT (EC, citrate synthase (EC, GDH (EC, and lactate dehydrogenase (EC activities were assayed as described by Clark and Lai (30). Enzymatic activities were determined at 25°C by using a Perkin Elmer (400 Bio Series, Boston, MA, U.S.A.) spectrophotometer. All assays were performed at least in triplicate, by using sufficient Triton X-100 as a detergent in the reaction mixture to allow the detection of maximal enzymatic activity.

Protein determination was performed by using the BCA protocol (Pierce, Rockford, IL, U.S.A.) in a 96-well microtiter plate reader with the wavelength set at 562 nm.

Statistical analyses

The patients' brain tissue samples were numerically coded. The enzymatic activities in each tissue sample were determined without prior knowledge of the patients' diagnostic subdivision into the three TLE groups (i.e., MTLE, PTLE, and MaTLE). Thus enzymatic data were collected without any observer bias. Once the enzymatic data collection was completed, the patients' classification into the TLE groups was decoded, and the following statistical analyses of the data were then performed. The patients were classified as described earlier into MTLE (n = 6), PTLE (n = 5), and MaTLE (n = 4) groups.

The activities of all enzymes were compared between three groups by using one-way analysis of variance (ANOVA), and the Tukey test was used for post hoc analysis. Activities also were compared between the two brain regions—cortex and hippocampus—by using the independent sample t-test for each group. The predictability of patients being assigned to their respective TLE group by using enzyme activities was assessed by using multiple discriminant analysis (MDA). Ratios of the activities of all other enzymes versus that of citrate synthase (CS) were calculated and compared between three groups in both regions, with the one-way ANOVA and the post hoc Tukey test. Correlations between enzyme-activity levels and other parameters such as neuronal densities and the durations between first unprovoked seizures and surgical resections were tested by using Spearman's rho (nonparametric) correlation analyses. A priori significance level was set at 0.05 for this study.


Patient characteristics

The tissue samples from patients in this study were classified into three groups: six MTLE samples, five PTLE samples, and four MaTLE samples. The neuronal densities in hippocampus assessed on an adjacent tissue section from the same patients (29) confirmed that the patterns of cell loss for these tissue specimens conformed to those previously described (29). The average percentage of neurons remaining across all hippocampal fields for the MTLE group was 37.4 ± 11.0 (mean ± SD); PTLE, 74.7 ± 5.0; and MaTLE, 73.5 ± 3.4. Immunohistochemical analyses for the neuropeptides [i.e., neuropeptide Y (NPY), somatostatin (SOM), substance P (SP), and dynorphin] also were as previously described for these groups of hippocampi (22–28). Neuron density counts were not done on the cortical specimens, but a qualitative analysis showed no discernible difference in neuronal density across the three groups. Their immunohistochemical labeling patterns for the neuropeptides NPY, SOM, SP, and γ-aminobutyric acid (GABA) were also similar to each other and in normal cortex (data not shown).

The mean age at surgery for the three groups was not statistically different: MTLE, 34.8 ± 11.5 years; PTLE, 38.4 ± 6.2 years; and MaTLE, 35.7 ± 27 years (mean ± SD; Table 1). Likewise, the gender representation across groups was similar (Table 1). However, the number of years between the first spontaneous seizure and surgical resection was longer in the MTLE patients (mean ± SD, 30.6 ± 8.5) compared with that in the PTLE patients (9.4 ± 6.7; p < 0.05) and MaTLE patients (10.4 ± 8.2; p < 0.05) (Table 1). The antiepileptic drugs (AEDs) used by the subjects before surgery (Table 1) were unlikely to affect the levels of enzyme activity because (a) the mechanisms of action of phenytoin (PHT), carbamazepine (CBZ), oxcarbazepine (OXC), and lamotrigine (LTG) are similar: they inhibit voltage-dependent sodium channels; (b) those patients taking two drugs were receiving either another sodium channel–acting drug or a drug that acts on the GABA system in approximately equal distribution across the three groups; (c) LTG has an additional mechanism—inhibiting excitatory neurotransmitter release—but was equally distributed in two patients of each of the three groups; and (d) the AEDs were administered with no known bias, because none is considered more useful in any of these kinds of epilepsy.

Data for seizure-foci localization were available for all but four patients. Based on scalp EEG intensive monitoring, no substantial difference was found in the degree of localization to the medial temporal region in the MTLE and PTLE groups (Table 2). As reported previously (21), the postsurgical outcome for the MTLE and MaTLE groups was good, with a small number of complex partial seizures seen only in one patient in the MTLE group. In the PTLE group, only in one patient (patient 11) were seizures clearly controlled. In another (patient 8), spells of seizure-like behavior were reported, but in phase I monitoring, such behavior was not associated with electrographic seizure activity (Table 2).

Enzyme activities in the cerebral cortex of the three groups of TLE patients

GDH activity in the MTLE group was most significantly affected (p = 0.001), being >50% lower than that in the PTLE and MaTLE groups (Table 3; Fig. 1A). The AAT, CS, and LDH activities in the MTLE group were not significantly different from the corresponding activities in the other two groups (Table 3); nevertheless, all three enzymes (i.e., AAT, LDH, and CS) showed a trend of slightly higher activity in the MTLE group than in the other two groups.

Table 3. Specific activities of enzymes in cortex and hippocampus in three groups of TLE patients
  1. Activities are reported for individual patients in the respective groups (U/mg protein, mean ± SEM). Activities for each patient were in turn average of at least three separate determinations.

  2. GDH, glutamate dehydrogenase; AAT, aspartate amino transferase; CS, citrate synthase; MTLE, mesial temporal lobe epilepsy; PTLE, paradoxical temporal lobe epilepsy; MaTLE, mass lesion–associated temporal lobe epilepsy.

  3. aRepresents statistically significant from the other two groups of TLE (p = 0.001).

  4. bRepresents statistically significant as compared with corresponding values in hippocampus (p ≤ 0.05).

MTLE0.05 0.2030.4660.0530.1010.1550.3430.059
0.0280.1390.4090.0530.0650.09 0.2090.039
0.0490.2590.4220.0760.1030.1230.29 0.074
0.0360.2080.3720.06 0.0830.0960.2410.039
0.0420.0920.4050.04 0.0490.08 0.1920.034
 Mean ± SD0.042a± 0.0080.218b± 0.110.424b± 0.0370.055 ± 0.0120.079 ± 0.0210.111 ± 0.0270.253 ± 0.0550.049 ± 0.006
0.1040.22 0.4770.05 0.09 0.1050.3140.041
0.1040.1080.32 0.0350.1470.1810.44 0.057
 Mean ± SD0.094 ± 0.0090.151 ± 0.0230.373 ± 0.130.043 ± 0.0060.106 ± 0.0250.128 ± 0.0540.345 ± 0.060.043 ± 0.008
MaTLE0.1240.2720.52 0.0650.1050.1770.3830.046
0.0640.0630.2050.0250.98 0.1430.2740.031
 Mean ± SD0.092 ± 0.0240.174 ± 0.0850.328 ± 0.1390.049 ± 0.0170.111 ± 0.0150.145 ± 0.0250.341 ± 0.0890.043 ± 0.008
Figure 1.

Figure 1.

Scatterplots of the activities of GDH versus CS in the (A) anterolateral temporal neocortical and (B) hippocampal tissues of temporal lobe epilepsy patients. GDH, glutamate dehydrogenase; CS, citrate synthase; MTLE, mesial temporal lobe epilepsy; PTLE, paradoxical temporal lobe epilepsy; MaTLE, mass lesion–associated temporal lobe epilepsy. All points represent a mean of at least three different measurements for each patient.

Figure 1.

Figure 1.

Scatterplots of the activities of GDH versus CS in the (A) anterolateral temporal neocortical and (B) hippocampal tissues of temporal lobe epilepsy patients. GDH, glutamate dehydrogenase; CS, citrate synthase; MTLE, mesial temporal lobe epilepsy; PTLE, paradoxical temporal lobe epilepsy; MaTLE, mass lesion–associated temporal lobe epilepsy. All points represent a mean of at least three different measurements for each patient.

We used CS as a mitochondrial marker to reflect the mitochondrial integrity. Ratios of GDH and AAT versus CS were separately calculated to “normalize” putative changes in GDH or AAT activities relative to changes due to mitochondrial damage/dysfunction, as indicated by CS. Compatible with the variations of GDH activity alone, the GDH/CS ratio (∼63% decrease, p < 0.001) in the MTLE group (mean ± SD; 0.78 ± 0.08) was significantly different from the corresponding ratios in the PTLE (2.22 ± 0.2) and the MaTLE groups (1.98 ± 0.21).

We further investigated the feasibility of using enzyme activity as a marker to delineate the patients according to the three-group classification of TLE. Because the classification of patients in three TLE groups is done postsurgically by using time-consuming immunostaining techniques, GDH activity could be used as a quick preliminary marker for assigning the patients to different TLE groups. We therefore performed MDA on the data and found that GDH activity in cerebral cortex could most certainly be used as a marker for the MTLE group (p < 0.0001). The prediction rate with GDH as a discriminant factor was 100% accurate for the MTLE group and was 60% accurate for all three groups taken together. The variability (40%) is mainly due to the similar GDH activities in the PTLE and MaTLE groups. The other enzymes investigated did not turn out to be useful to categorize patients according to the three-group classification of TLE. A correlation analysis (Spearman's rho) of GDH activity and duration between the first unprovoked seizure and surgical resection showed a significant negative correlation (r=−0.696, p < 0.01).

Enzyme activities in the hippocampus of the three groups of TLE patients

The neuronal cell densities in the hippocampi of the MTLE group were 50% lower than those of the PTLE and MaTLE groups. Surprisingly, none of the enzymes investigated in the hippocampus showed significant differences when the three TLE groups were compared against each other. GDH activity was lower in the MTLE group than in the other two groups, but this difference did not reach statistical significance (p = 0.071), probably because of low sample size (Table 3; Fig. 1B). Interestingly, the GDH/CS ratio was significantly lower in the MTLE group (1.627 ± 0.114, mean ± SD) than those in the PTLE (2.487 ± 0.18) and MaTLE groups (2.61 ± 0.21) (p < 0.05). Discriminant analysis showed that none of the enzyme activities in hippocampus was very useful in categorizing patients according to the three-group classification of TLE.

A correlation analysis of enzyme activities against the percentage mean density of neurons across all regions of the hippocampus showed a significant negative correlation for GDH activities in the MTLE group, but not in the other groups, by Spearman's rho correlation analysis (r=−0.9; p < 0.05). A correlation analysis of GDH activity and the duration between first unprovoked seizure and surgical resection also showed a significant negative correlation (r=−0.57; p < 0.05).

Comparison of enzyme activities between cortex and hippocampus

In the MTLE group, three of the four enzymes investigated showed significant differences between cortex and hippocampus. GDH activity in the cortex was found to be 46.8% lower than that in the hippocampus (p < 0.005), whereas LDH (p < 0.00001) and AAT (p = 0.043) activities in the cortex were 96.4% and 67.6% higher, respectively, than the corresponding activities in the hippocampus (Table 3). These differences in enzymatic activities suggest significant alterations in the cortical metabolism in MTLE. These results are especially significant, considering that no observable difference was found in the neuronal density in cortex, whereas that in the hippocampus was 50% lower in the MTLE group as compared with those in the PTLE and MaTLE groups. No significant difference was seen between the enzyme activities in cortical and hippocampal tissues obtained from the PTLE and MaTLE patients.


The data showed that changes in metabolic enzyme activities, particularly GDH, were seen both in the temporal neocortex and hippocampus of patients with TLE associated with hippocampal sclerosis (MTLE) rather than the PTLE and MaTLE groups.

Enzyme activity changes in temporal cortex

Among the enzymes studied, the clearest difference was in GDH activity in anterolateral temporal neocortex, which was the most altered in the MTLE group, being >50% lower than that in the PTLE and MaTLE groups. Moreover, GDH activity in cortex can be used as a marker to distinguish the MTLE patients from the other two groups of TLE patients. These observations are important because to date, the hippocampus has been the main focus of most investigations in TLE, in spite of imaging studies showing that the interictal hypometabolic zone associated with a hippocampal seizure focus extends beyond the confines of the hippocampus into adjacent neocortex (31,32). This has remained a puzzling observation. Anatomically, no differences were found in the neocortical specimens of the MTLE group compared with the PTLE or MaTLE group—no neuron loss detectable by visual inspection, and no detectable changes in the distribution of GABA-, somatostatin-, neuropeptide Y-, or substance P-containing interneurons (22–28). However, as discussed in Results, the difference in enzymatic activities we found cannot be attributed to the AEDs taken by the subjects before surgery.

Because GDH is largely a mitochondrial enzyme (18), the possibility that the reduction of GDH activity may be due to mitochondrial damage or dysfunction must be considered. CS is a general marker of mitochondrial integrity: any decrease in CS activity can be inferred to result from direct or indirect mitochondrial damage. That CS activity was not different between the three groups suggests that the decrease in GDH activity was not because of damage or alterations in mitochondrial number due to cell death.

One explanation for the lower GDH activity in the cortex of MTLE patients than that in the other groups may be that the interval between the first unprovoked seizure and surgical resection of the tissue was significantly longer in the MTLE group than that in the PTLE or MaTLE group (Table 1). The negative correlation between GDH activity and duration of epilepsy only for the MTLE group suggests that the changes in GDH activities in the temporal cortex might be associated with propagation of seizures into the cortex over a longer period rather than the cause of seizure initiation.

That AAT and LDH activities in the cortex were higher than those in the hippocampus of MTLE patients further supports the idea that the metabolic derangement in MTLE extends beyond hippocampus and might be involved in seizure propagation rather than initiation. AAT catalyzes the reversible reaction: glutamate + oxaloacetate ⇔ aspartate +α-ketoglutarate. In MTLE group, the AAT activity might contribute to the increased excitability because of increased aspartate (another excitatory neurotransmitter) formation from glutamate (Fig. 2). The increased LDH activity in the MTLE cortex suggests a slight shift reflecting increased glycolytic metabolism, possibly an attempt to generate sufficient ATP for extracellular glutamate removal by astrocytes.

Figure 2.

Biochemical pathways showing the relation between energy metabolism and glutamate metabolism.

Whether the decreased GDH activity is the consequence of a change in gene expression, translation, or posttranslational modification is yet to be determined. In female Albino rats treated with monosodium glutamate for 1 year, a 45% reduction of the GDH activity in crude brain extracts was associated with a significant decrease in the GDH enzyme protein but no change in the GDH messenger RNA (mRNA) expression (33). These observations suggest posttranscriptional changes or degradation of GDH protein or both as probable molecular mechanism(s) underlying the decrease in GDH activity on prolonged exposure of the brain to glutamate (33). The propagation of seizure activity into the cortex may result in increased glutamate release there.

Further characterization of the role of GDH isoforms in TLE may better explain the consequences of decreased GDH activity and clarify the limits in normalizing enzyme activity in terms of tissue protein, an approach not taking into account the putative differential localization of GDH and its isoforms in neurons and astrocytes. Consequently, the data presented here may assume more pathophysiologic importance than that which the statistical analysis purportedly demonstrates.

Enzyme activity changes in the hippocampus

Hippocampal GDH activity tended to be lower in the MTLE group, but this difference was not statistically significant because of small sample sizes. Because astrocytes are a principal source of GDH (34–36) and significant gliosis accompanies neuronal loss in MTLE, GDH activity would be expected to increase. However, we may infer from our finding that even after normalizing for astrocyte density, the GDH activity per astrocyte may be lower than normal in the MTLE hippocampus. Thus in keeping with the negative correlation between neuronal densities and glial densities (J. H. Kim, unpublished data) in the MTLE group, neuronal densities were negatively correlated with GDH activity. The negative correlation between GDH activity and the duration between first unprovoked seizure and surgical resection suggests that the reduction in GDH is a progressive process in the hippocampus. Because this relation is not confined to the MTLE group, it may be a secondary effect of a factor associated with seizures. Extracellular glutamate levels may be the most likely factor. In vivo GDH levels are reduced in rats with a prolonged intake of monosodium glutamate (33). In astrocyte cultures, exogenous glutamate levels determine the metabolic fate of glutamate (19), with the amount of glutamate metabolized via the TCA cycle progressively increasing with an increase in extracellular glutamate. Thus in the MTLE hippocampus, the high extracellular glutamate levels (2–7,37) may favor glutamate oxidation via the TCA cycle and the formation of α-ketoglutarate (α-KG) and aspartate (18) (Fig. 2). The aspartate levels in the MTLE hippocampus have not been studied. In this “oxidation preferred” condition, if the GDH activity is reduced, the glutamate oxidation will be decreased, leading to accumulation of the glutamate in the mitochondria of astrocytes, and this may account for high glutamate levels found intracellularly in the MTLE hippocampus (17). Such a reduction in GDH activity would take on greater significance in light of the reduced glutamine synthetase activity in MTLE astrocytes (38) (Fig. 2). This study demonstrated an ∼40% decrease in GS levels in the MTLE hippocampus and suggested that this may play a significant role in the elevation of extracellular glutamate levels in MTLE.

These enzymatic changes in the MTLE hippocampus have significant implications for glutamate metabolism in this group of TLE patients. Our data, together with existing pathological data in TLE patients, such as decreased GS activity (38), higher levels of intracellular (14,39,40) and extracellular (2,3) glutamate, and in vivo hippocampal metabolic dysfunction (34–36), suggest that considerable metabolic derangement is present in the MTLE patients. Based on our data, we hypothesize that the reduction of GDH and GS activity contribute to the elevated levels of intra- and extracellular glutamate in MTLE patients, which are associated with the excitotoxicity induced by seizures in TLE. Our results strongly suggest significant alterations in glutamate metabolism in the cortex of MTLE patients as compared with PTLE and MaTLE, and provide an explanation for elevated extracellular glutamate in TLE. This is the first study to document the putative metabolic derangement induced by MTLE in the cortex.


Acknowledgment:  This project was supported by NIH grants P20 RR16454 (from the Idaho BRIN Program of the National Center for Research Resources), PO1 NS39092 (via a subcontract from Yale University under the NIH Program Project), and RO1 NS34813. G.H. M-P also thanks Idaho NIH-BRIN grant P20 RR16454 for a research fellowship.