Glutamine synthetase becomes nitrated and its activity is reduced during repetitive seizure activity in the pentylentetrazole model of epilepsy


  • Naoki Nitta and Christophe Heinrich equally contibuted to this work.

Address correspondence to Hans-J. Bidmon, C. & O. Vogt Institute for Brain Research, University St. 1, D-40225 Düsseldorf, Germany. E-mail:


Purpose: The astrocyte-specific glutamine synthetase (GS) plays a key role in glutamate recycling and Gamma-aminobutyric acid (GABA) metabolism. Changes in the expression or activity of GS have been proposed to contribute to epileptogenesis. The mechanisms or how and where GS may contribute to epilepsy is still a matter of discussion. Here we asked the question whether brain regions, which show an astrocytic stress response respond with alterations of GS.

Methods: Biochemical and histological alterations of GS, HSP-27, and GFAP were studied after pentylenetetrazole-induced repetitive epileptic seizures (PIRS) in rats using a topographical quantification of the GS-immunoreactivity (GSIR) in relation to the focal heat shock response (HSR). Saline-treated rats served as controls and rats treated by the GS-inhibitor, L-methionine-sulfoximine (MSO) served as a positive control.

Results: No changes in the amount of GSIR and GS-protein occurred during PIRS. A significant reduction of GSIR was observed by histochemistry (in situ) and in native (nonheated) protein extracts of MSO-treated rats. In rats affected by PIRS, GS-activity showed a significant, region-specific reduction in association with a nitration of the enzyme.

Discussion: These results show that neither PIRS nor GS-inhibition reduced the amount of GS protein, but that MSO interferes with antibody binding to native GS. PIRS resulted in a focal increase of astrocytic stress response, whereas MSO caused a widespread, homogeneous astrocytic HSR independent from quantitative changes of GS content. In rats with PIRS the regions showing a strong glial HSR, respond with reduced GS-activity and GS-nitration, which all together are clear indicators of a nitrosative stress response.

The neurotransmitters glutamate and GABA are the main excitatory and inhibitory neurotransmitters of the central nervous system (CNS)(Petroff, 2002), respectively. Both glutamate and GABA play key roles in the pathophysiology of epilepsy (Jefferys & Traub, 1997; Hertz & Zielke, 2004: De Lanerolle & Lee, 2005). Glutamine, the product of the astrocytically localized glutamine synthetase (GS), represents the precursor for the synthesis of glutamate and GABA (Cavus et al., 2005; Tian et al., 2005: Schoesboe & Waagepetersen, 2007). Furthermore, GS is essential for the elimination and recycling of glutamate, which has been taken up via glutamate transporters (GLT-1) from the synaptic cleft and extracellular space by perisynaptic astrocytes (Derouiche & Frotscher, 1991; Rauen &  Wiessner, 2000; Volterra & Meldolesi, 2005). The glutamate-glutamine cycle is interlinked with the glutathione and glutathione peroxidase system,  indicating its involvement in the cerebral defense mechanisms against oxidative and nitrosative stress (Dringen et al., 1999; Knorpp et al., 2006: Re et al., 2006). Taken together, GS holds a key position for neurotransmission. Disturbances of GS at the level of its expression or activity may, therefore, lead to pathological alterations (Robinson, 2001). Further aspects of the role of the glutamate-glutamine cycle have been extensively studied in various pathophysiological conditions including epilepsy (Eloqayli et al., 2003, 2004; Melo et al., 2005). Considering the new insights regarding the active role of glia in setting the stage for neurotransmission (Schipke & Kettenmann, 2004; Hertz & Zielke, 2004; Volterra &  Meldolesi, 2005; Deitmer et al., 2006; ZurNieden & Deitmer, 2006), glial cells also represent an important target for pharmacological interventions in epilepsy.

Recent reports on the content and the activity of GS from cerebral tissues surgically removed from patients suffering from intractable epilepsy suggest side-specific responses. While Petroff et al. (2002); Eid et al. (2004); and Van der Hel et al. (2005) found decreasing levels of GS after hippocampal sclerosis, Steffens et al. (2005) could not demonstrate changes in GS content in the epileptic cortex. In vivo spectroscopy showed an increase of both glutamine and glutamate in the thalamus in epileptic patients (Helms et al., 2006). In addition, lower amounts of astrocytic GS have been reported from rats, which develop genetic absence epilepsy (Dutuit et al., 2000). These data indicate variable responses of GS in pathologically altered cerebral tissues.

The majority of GS is found in astrocytes (Martinez-Hernandez et al., 1977; Norenberg, 1979), which become affected by seizure activity both in animals and humans (Erdamar et al., 2000; Bidmon et al., 2004, 2005). Here we asked two questions, (1) whether and how GS may become affected during the early stages of seizure activity and (2) whether the astrocytes, which respond with a focal or region-specific HSR, are those which become affected. To address these questions, we used the established pentylenetetrazole (PTZ)-model of pentylenetetrazole-induced repetitive epileptic seizures (PIRS) (for details see, Caspers & Speckmann, 1972; Rauca et al., 2004; Bidmon et al., 2005) which in contrast to the kainic acid- and pilocarpin model does not cause cell death during the initial seizure episodes (Valente et al., 2004). As a positive control and to evaluate our immunohistological and biochemical methods the L-methionine-sulfoximine (MSO) model of direct GS-inhibition was used (Fischer & Kolousek, 1959; Hevor, 1996; Bernard-Helary et al., 2000; Böttcher et al., 2003).

Here we report that PIRS activity causes no decrease in the amount of GS protein, but results in GS-nitration accompanied by partial GS inhibition. In addition, we show that GS inhibition by MSO causes a significant reduction of GS immunoreactivity in situ as well as in native protein extracts probably by an effect of epitope masking.

Materials and Methods

Male Wistar rats (220–250 g) were placed in individual cages mounted onto a recording system in order to register intensity and duration of seizure activity (Bidmon et al., 2005). In addition, seizure activity was scored by visual inspection. The observed individual variations with regard to the seizure response occurred in the duration of the single seizure episodes, which varied between 40 min and 180 min among individuals as well as in the first appearance of a full tonic–clonic seizure, which occurred in some individuals directly after the first PTZ injection whereas others showed the full response during the second or even the third injection and thereafter. Therefore, with the term “strong responders” we refer to individuals, which showed generalized tonic–clonic seizures from the beginning of the treatment. As previously shown, seizure activity resulted in a delayed focal glial HSP-27 response within distinct cerebral regions whereas MSO caused a more immediate glial HSP-27 response between 8 and 24 h. The rats were divided into three groups (saline, n = 12 MSO, n = 14 PTZ, n = 17). The first group was injected (i.p.) every other day with saline, whereas the third group received injections with 40 mg PTZ/kg (Caspers & Speckmann, 1972; Bidmon et al., 2005). The second group (positive control) was injected (i.p) with a single dose of 100 mg MSO/kg dissolved in saline. Rats of groups 1 (saline) and 3 (PTZ) were sacrificed after 14 days while group 2 (MSO) was sacrificed 24  h after treatment.

Six rats of each group described above were anesthetized with sodium pentobarbital and fixed by cardiac perfusion using ice-cold physiological saline followed by Zamboni's fixative for immunohistology. The remaining rats of each group were perfused with saline only, in order to clear the blood. Afterwards the whole cerebral cortex or piriform-entorhinal cortex and hippocampus were immediately dissected from these brains and frozen over liquid nitrogen for biochemical analysis. All procedures were conducted in accordance with the German Animal Welfare Act, and approved by the responsible governmental agency.

The brains designated for immunohistology were cryoprotected in 25% sucrose, frozen in isopentane at −40°C and 50 μm thick sections were cut using a Frigocut (Leica, Bensheim, Germany). For the immunohistological localization of the investigated proteins (HSP-27, GFAP) in alternate sections of the brains, we followed our previously detailed protocols (Bidmon et al., 2004, 2005). Additionally, here we used a monoclonal primary antibody directed toward GS (Transduction Laboratories, Lexington, KY, U.S.A.) at a final dilution of 1:100. Double- or triple-staining was used to prove whether GS was restricted to astrocytes, HSP-27 was colocalized with GS, or whether tyrosine nitration occurs in GS and/or HSP-27 positive astrocytes. Polyclonal GFAP antibody (Sigma, Deisenhofen, Germany) or HSP-27 were used according to Bidmon et al. (2005) in combination with monoclonal GS antibody. For triple staining of GS, HSP-27, and nitrotyrosine, we used normal donkey serum (Dianova, Hamburg,  Germany) to block nonspecific binding, monoclonal GS (mouse), polyclonal HSP-27 (rabbit) and for the detection of nitrotyrosine an antibody raised in goat against nitrotyrosine (diluted 1:500, Biosciences Pharmingen, San Jose, CA, U.S.A.) was used. As secondary antibodies we used Cy-5 coupled anti-goat-IgG, Cy-3 coupled anti-rabbit-IgG, and FITC-linked antimouse-IgG at a final dilution of 1:150 (Dianova); all of which had been produced in donkey.

In order to determine the region-specific activity of GS within the hippocampal formation, we added two groups (one PTZ-group, n = 8; one saline control group, n = 6) of rats, which were treated similarly as above. These rats were decapitated and the hippocampal formation and piriform-entorhinal cortex were dissected. The piriform-entorhinal cortex was directly frozen on dry-ice whereas the hippocampi were rinsed in oxygenated artificial cerebrospinal fluid (ACSF) containing (in mmol/L) NaCl 124, KCl 4, CaCl2 1.0, NaH2PO4 1.24, MgSO4 1.3, NaHCO3 26, and glucose 10, which was constantly equilibrated with 5% CO2 in O2, stabilizing pH at 7.35–7.4 at 28°C. The hippocampi were cut into slices (500 μm thick), which were stored in constantly equilibrated ACSF. These hippocampal slices were then immediately placed into a dissection dish and the molecular and granular layers of the dentate gyrus were dissected using a dissecting microscope. The dissected samples from all slices of one individual were pooled in an ice-cold cup (Eppendorf,  Hamburg,  Germany), frozen and stored for the evaluation of GS activity (see below). Protein extracts from the piriform-entorhinal cortices of these rats were used together with those from three additional rats treated with MSO for western blots (WBs) for GFAP (Fig. 5).

Figure 5.

 Representative WB prepared from the piriform-entorhinal cortices of control, PTZ-, and MSO-treated rats showing the amounts of immunoreactive GFAP in relation to GAPDH (loading control). There is no indication for increased amounts of GFAP after PTZ- or MSO-induced seizures.

Quantitative analysis

For the quantitative analysis of the immunohistochemical data optical density (OD) values were measured (Losem-Heinrichs et al., 2005). For the OD determination in the desired cerebral regions 6–8 sections from each brain (1 hemisphere) were scanned together with glass standards (Zeiss, Jena, Germany) of defined optical densities using a KS 400 Image analyzer (Zeiss). Corresponding sections, which had been immunohistochemically processed without the primary antibody served to determine the background values. Within the images of the scanned sections, regions of interest were defined as outlined in Fig. 2H. Basically, we first measured the whole cortex and the whole hippocampus as one region respectively, and later parcellated them into several subdivisions for which the  optical densities were individually determined in order to allow a detailed comparison of individual brain compartments among the groups. Mean values were calculated per treatment group and per cortical or hippocampal compartment and the data were plotted and analyzed by means of an ANOVA (repeated measures design). The significance level was set at α = 0.05. For further details refer to Gladilin et al. (2000).

Figure 2.

 Examples of immunostained frontal sections of rat brains used for measurements of optical densities (OD) to quantify the immunoreactivity for GS (A–C), GFAP (D), and HSP-27 (E–G). In H the regions for which the OD-values as shown in Fig. 5 and Tables 1–2 were determined are marked (RS, retrosplenial cortex). Note that GS inhibition resulted in lower staining intensity for GS (C) and a more homogenous HSP-27 distribution (E) compared to repeated PTZ-treatment (B, F). GFAP immunoreactivity remained stable under all conditions as shown here for a MSO-treated rat. Bar: 1 mm.

Western and dot-blot analysis

After preparation of the respective brain area, tissue samples were homogenized with a Ultraturrax (Janke & Kunkel, Staufen, Germany) at 4°C in lysis buffer, 209  mmol/L Tris/HCl buffer (pH 7.4) containing 1% Triton X-100, 140  mmol/L NaCl, 1  mmol/L EDTA, 1  mmol/L EGTA, 10  mmol/L NaF, 10  mmol/L Na-pyrophosphate, 1  mmol/L sodium vanadate, 20 mmol/l ß-glycerophosphate and protease inhibitor mixture (Roche Applied Science, Mannhein, Germany).

The homogenized lysates were centrifuged three times at 20,000 gat 4°C. Resulting pellets were discarded and the supernatant were used for further processing. Protein concentration of the supernatant was estimated by using the BioRad protein assay (BioRad, Munich, Germany). For dot blot (DB) analysis equal amounts of proteins in a volume of 2 μl were applied to a nitrocellulose membrane and allowed to dry for 10 min. Detection of GS under denaturating conditions was achieved by adding equal amounts of 2×gel loading buffer (pH 6.8) containing about 4% sodium dodecyl sulfate (SDS) and 200  mmol/L dithiothreitol to the protein samples before dot-blot analysis.

For western-blot analysis samples were heated to 95°C for 5 min in 2×SDS buffer and subjected to gel electrophoresis on 10% gels. Following electrophoresis, gels were equilibrated with transfer buffer (39   mmol/L glycine, 48  mmol/L Tris-HCl, 0.03% SDS, 20% methanol). Proteins were transferred to nitrocellulose membranes using a semidry transfer apparatus (Pharmacia, Freiburg, Germany). DB and WB were blocked in 5% bovine serum albumine solubilized in 20  mmol/L Tris/HCl pH 7.5 containing 150  mmol/L NaCl and 0.1% Tween 20 and then incubated for 2  h with antiserum against GS (monoclonal antibody, 1:5,000), GFAP (monoclonal antibody 1:5,000, Sigma), or glyceraldehyde-3-phosphate-dehydrogenase (monoclonal antibody, 1:5,000). Following washing and incubation with horseradish peroxidase-coupled anti-mouse-IgG antibody (Biorad) diluted 1:10,000 at room temperature for 2  h, blots were washed again and developed using western lightning chemiluminescent detection (PerkinElmer, Boston, MA, U.S.A.).

For immunoprecipitation, brain homogenates containing defined protein amounts were incubated with 1 μg anti-3-nitrotyrosine antibody (Upstate Biotechnology, Lake Placid, NY, U.S.A.). The immune  complexes were collected using protein A/G agarose (Santa Cruz Biotechnology, Santa Cruz, CA, U.S.A.) and then subjected to SDS-polyacrylamide gel electrophoresis (PAGE) on a 10% gel. GS was detected among the tyrosine-nitrated proteins using a monoclonal anti-GS antibody (Transduction Laboratories). Densitometric analysis was performed with the Kodak Image Station 4000MM using Kodak MI software (Eastman Kodak, Rochester, NY, U.S.A.).

GS activity

GS activity was assayed according to the γ-glutamyl transferase reaction (Webb & Brown, 1976). Aliquots (100 μl) of cortex (piriform-entorhinal), whole hippocampus or microscopically dissected dentate gyrus (molecular and granular layers) homogenates were incubated with 900 μl of reaction mixture at 37°C. Reaction  mixture consisted of 60  mmol/L L-glutamine, 15  mmol/L hydroxylamine-HCl, 20  mmol/L Na-arsenite, 0.4  mmol/L adenosine diphosphate, 3  mmol/L MnCl2, and 60  mmol/L imidazole-HCl buffer, pH 6.8, in a final volume of 1 ml. The reaction was terminated by adding a 1 ml stop-solution containing 0.2  mmol/L trichloacetic acid, 0.67 mol/L HCl, and 0.37 mol/L FeCl3. The solution was cleared by centrifugation at 20,000 g at 4°C, and the glutamyl hydroxamate formed was determined in supernatant at 500 nm. A standard row was created by the use of γ-glutamyl hydroxamate (Sigma). Results from multiple independent experiments (Figs. 4 and 8) were expressed as the mean ± SEM and were analyzed by a Student t-test (Görg et al., 2005).

Figure 4.

 Western blot (WB) prepared from cerebral cortex of control rats and from rats 24  h after inhibition of GS by MSO showing that HSP-27 is strongly induced by MSO injection, whereas no changes are seen in glyceraldehyde-3-phosphate-dehydrogenase (GAPDH), which served as loading control (A). The level of GS inhibition for cerebral cortex and hippocampus 24  h after MSO injection as determined in three individual experiments (B).

Figure 8.

 After repeated PTZ-induced seizures a significant reduction in GS activity occurs in piriform-entorhinal cortex and in the granular and molecular layers of the dentate gyrus, whereas in total hippocampus a nonsignificant reduction is indicated. Note that in saline-treated controls very similar levels of GS activity existed in all regions.


Similar to our previous study, rats that had received repeated PTZ-injections for two weeks (Bidmon et al., 2005) showed a strong upregulation of HSP-27 in astrocytes and blood vessels located mainly in the cerebral cortex and the hippocampus (Figs. 1–3, Tables 1 and 2). The most severely affected cortical regions were the piriform and entorhinal cortex. Rats treated with MSO (positive control) showed, concomitant with a significant inhibition of GS, a rapid induction of HSP-27 (Fig. 4, Tables 1 and 2). None of these treatments caused a significant focal or general increase of GFAP immunoreactive astrocytes as shown topographically (Figs. 1–3, Tables 1 and 2) or by WBs (Fig. 5). In control rats as well as in PTZ- or MSO-treated rats, GS-immunoreactivity (GSIR) in the cerebral cortex and  hippocampus was confined to astrocytes as shown by colocalization with GFAP (Fig. 1A–C). Neuronal somata appear as darker weakly stained regions between the GS containing astrocytes (Fig. 1A–L), since they are only partly overlayed by GS containing glial processes. There are four major differences observed in rats treated either with PTZ or with MSO: (1) repeated PTZ treatment induces seizures in all treated individuals within minutes, but with differing severity and duration, whereas MSO as used here caused transient seizures in some rats after 8–24  h; (2) following 2 weeks of PIRS, a slight but significant reduction in GS-activity was detectable in the piriform-entorhinal cortex and dentate gyrus, whereas MSO-treated rats had a reduced GS-activity, which was only 20–30% of that of controls (Fig. 4), (3) PTZ-injections of 40 mg/kg body weight had to be repeated for up to 2 weeks in order to induce glial HSP-27 induction, whereas MSO resulted in a widespread astrocytic HSP-27 induction within 24  h in all treated individuals (Figs. 1G, J; 2E–F; and Bidmon et al., 2005), (4) after PIRS, the HSP-27 induction appeared focal, whereas MSO treatment resulted in a more general astrocytic increase of HSP-27 (Figs. 1–3), showing that MSO affects GS in all astrocytes similarly (Figs. 2 and 3). However, the histological comparison of HSP-27 induction in both PTZ- and MSO-treated rats allowed a direct comparison of the affected astrocytes. As seen in Fig. 1D and 1E, control rats did not stain for HSP-27 in the cortex, but showed a uniform strong labeling with GS (green, Fig. 1D and 1F), whereas rats treated with MSO showed a widespread distribution of various degrees of HSP-27 in astrocytes (Fig. 1G) and a comparatively lower staining intensity for GS (green in 1G and 1I), although HSP-27 positive astrocytes still showed immunoreactivity for GS (Fig. 1H and 1I). This in situ immunocytological analysis indicated that the reduction of GSIR was most pronounced in fine, distal astrocytic processes when compared to control or PTZ-treated rats (Fig. 1E, 1H, and 1K). In sections of rats with PIRS all astrocytes showed normal intensity of GSIR, and HSP-27 positive astrocytic foci of various sizes were present, e.g., in the entorhinal cortex (Figs. 1J and 2). It is, however, noteworthy to mention that GSIR remained normal in those astrocytes, which were specifically affected by seizures as evidenced by their concomitant specific HSP-27 immunoreactivity (Fig. 1J, 1K and 1L). This fluorescence-immunocytological data, though not directly quantifiable, indicated that MSO had resulted in a reduction of the immunoreactivity for GS and an increase in HSP-27. Furthermore, it was also supported by the results obtained for whole brain slices, which had been alternately stained for a single epitope such as GS, HSP-27, or GFAP (Fig. 2). Sections from rats with PIRS showed a similar immunoreactivity for GS as controls (Fig. 2A and 2B), whereas a much weaker GSIR was seen in sections from MSO-treated rats (Fig. 2C). PTZ treatment or MSO treatment did not affect the immunoreactivity for GFAP, which remained normal (Figs. 2D and 5; Bidmon et al. 2005).

Figure 1.

 (A–C), Colocalization of GS (green) and GFAP (red) in hippocampal astrocytes of the lower fascia dentata. Note that GS outlines the whole cytoplasm of the astrocyte, whereas GFAP remains restricted to the cytoskeletal fibers (examples marked by white arrowheads). The larger dark (unstained) structures refer to neuronal somata. In entorhinal cortex of saline-treated rats only GSIR is detectable (D, overview; E–F high magnification) while no HSP-27 (red) is present (F), whereas in MSO-treated (G–I) and PTZ-treated (J–L) rats GS (green) and HSP-27 (red) are colocalized in various astrocytes. Note that in MSO-treated cortex many more HSP-27 positive astrocytes are present (G, overview) than after treatment with PTZ (J, overview), and that the relative staining intensity for GS is reduced when compared to control (D–E) and to PTZ-treatment (J–K). After treatment with MSO the reduction of GSIR seems to be more pronounced in distal processes of astrocytes (H). Bars: 100 μm (D, G, J); 15 μm (A–C, E–F, H–I, K–L).

Figure 3.

 Fingerprints showing the changes observed for the OD-values of the immunoreactivity for GS (A–B), HSP-27 (C–D), and GFAP (E–F) in controls (g1), after MSO treatment (g2) and after repeated PTZ-induced seizures (g3). For standard deviation see the corresponding values ( Tables  1A–3B). Note the stronger reduction of GSIR in piriform-entorhinal cortex (Pir+Ent in A) compared to the exceptionally high induction of HSP-27 (B) after GS inhibition. Similar to our previous finding for PTZ-treated rats (Bidmon et al. 2005) GFAP immunoreactivity remained also stable in MSO-treated rats when compared to controls indicating that both treatments cause gliosis. CA1, hippocampal area CA1; DG, dentate gyrus; infra, infragranular cortical layers; par1, primary somatosensory cortex area 1; RS, retrosplenial cortex, supra, supragranular cortical layers.

Table 1.   Mean values of optical density ± S.D. of GS-, HSP-27- and GFAP-immunoreactivity for the whole cerebral cortex, retrosplenial (RS), piriform, and entorhinal cortices (Pir+Ent), as well as for Par I (Total, averaged over all cortical layers; supragranular and infragranular layers)Thumbnail image of
Table 2.   Mean values of optical density ± S.D. of GS-, HSP-27- and GFAP-immunoreactivity for the whole hippocampus and area CAI or the dentate gyrus (DG)Thumbnail image of

Compared to the repeatedly PTZ-treated rats (Fig. 2F), MSO treatment resulted in a more widespread uniform HSP-27 induction throughout the brain (Fig. 2E). In order to quantify the observed changes we measured the optical density (OD)-values in specific HSP-27-responsive brain regions as depicted in Fig. 2H. Here we first measured the OD-values for the whole cerebral cortex and the whole hippocampus (Fig. 3, Tables 1 and 2) and afterwards we selected those subdivisions which specifically responded with an increase in HSP-27 following PIRS in order to evaluate whether changes in GSIR may be  restricted to  regions in which astrocytes seem to be severely affected. As seen in Fig. 3A and 3B and Tables 1 and 2, no significant differences existed for GSIR of PTZ-treated rats (group 3) when compared to saline-treated controls (group 1) for the cortical and hippocampal amounts of GSIR, whereas the MSO-treated group (group 2) showed a highly significant decrease in the OD-values for GSIR. Furthermore, in the polar plots it becomes evident that it is a general decrease, since the value of decrease is similar for most cortical and hippocampal subdivisions with one exception, the piriform-entorhinal cortex, for which the level of decrease was 57% higher than that measured for the whole cortex (Fig. 3A). HSP-27 induction was also highest in the MSO-treated group (g2, Fig. 3C and 3D), because HSP-27 was increased much more homogeneously and in a larger number of astrocytes when compared to the brains from PTZ-treated rats, in which in most individual cases smaller foci of HSP-27 positive astrocytes were characteristic (Figs. 1J and 1K, 2E, 2F, and 2G). Furthermore, differences in the normal packing density of cortical astrocytes in HSP-27 positive sections are also clearly visible within the somatosensory cortex (Par1) when comparing the supragranular layers with the infragranular layers (Fig. 3C), since the packing density remains higher in the supragranular layers. That indicates that a gliosis, as seen in other neuropathological conditions such as acute ischemic foci and leading to almost similar amounts of astrocytes in all cortical layers (Losem-Heinrichs et al., 2005) has not occurred in the present model. The same is seen for the piriform-entorhinal cortices and the dentate gyrus of the hippocampal formation (Fig. 3D), which contain a homogeneous, dense packing of astrocytes (Zilles et al., 1991), showing that MSO affects astrocytes much more generally than during PIRS. That the piriform and entorhinal cortices and the dentate gyrus contain a higher amount of GFAP positive astrocytes is also shown in (Fig. 3E and 3F), but MSO treatment caused no significant changes in the OD-values for GFAP-immunoreactivity in any of the studied regions when  compared to controls, similar to the findings known for repeated PTZ-application (Bidmon et al., 2005). Taken together these results showed that during PIRS focal changes of cortical and hippocampal HSP-27 expression do not result in corresponding focal changes of GSIR, whereas the widespread reduction of cerebral GS inhibition by MSO caused global decrease in GSIR and a nonfocal global HSP-27 induction within 24  h.

In order to evaluate the specificity of our used GS-antibody and for the comparison of GS-protein levels between control and treatment groups, we first used standard WB protocols. In standard WB using heat-denatured amounts of protein from whole cerebral cortex and hippocampus no differences were obvious between all tested individuals from all treatment groups. Therefore, we were unable to reproduce our highly significant immunohistological findings using standard WBs. It was obvious that MSO-treated rat brains contained the same amount of GS-protein as controls (not shown). Only in WBs stained for HSP-27 were we able to reproduce the immunohistochemical finding that MSO treatment resulted in a pronounced HSP-27 induction when compared to controls (Figs. 1, 2, 3, and 4). In order to resolve these contradicting data, we used original native and heat-denatured protein from the cerebral cortex and hippocampus from saline and MSO-treated rats in a DB assay and in standard WBs (Fig. 6). These data clearly showed that only when using the native nonheated protein in DBs were we able to reproduce the findings in a way comparable to that seen by GSIR in situ. In a separate experiment the comparison of DBs using nonheated and heated protein and WBs was done also for cerebral cortex (not shown), piriform-entorhinal cortex, and hippocampus dissected from PTZ-treated rats for which no changes in the amount of GSIR could be established (Fig. 7). Taken together these data show that neither PIRS nor MSO treatment resulted in a loss of GS-protein in the studied brain regions. However, MSO binding to the GS reduced GSIR. Therefore, in native cerebral protein extracts from MSO-treated rats MSO interfered with antibody binding and this situation could be reversed by boiling the protein extracts.

Figure 6.

 Dot blots (DBs) prepared with native nonheated protein (A–B) and boiled protein (C–D) extracts obtained from control rats and MSO-treated rats for cerebral cortex and hippocampus compared to WBs from the same samples immunostained for GS (E–F) and GAPDH as loading control (G). Note that no reduction of GS immunoreactivity is indicated in heated protein extracts compared to the clear loss of GSIR after MSO treatment in nonheated protein extracts, thus showing comparable results to that seen by immunohistochemistry (Figs. 1–4).

Figure 7.

 DBs using native protein (A–B) and WBs (C–F) for GS and GAPDH (loading control) for control and PTZ-treated rats. Note that neither in hippocampus nor in piriform-entorhinal cortex are changes in the amount of GS detectable.

Since there exists increasing evidence that GS becomes affected during the course of epilepsy, we measured the activity of GS in extracts from cortex and hippocampus of rats with repeated seizures. The results revealed no significant differences when compared to saline-treated controls. However, a slight reduction was indicated (not shown) as seen for the whole hippocampus (Fig. 8). In order to focus more on affected astrocytes, we repeated these measurements by dissecting only the piriform-entorhinal cortices and the molecular and granular layer of the dentate gyrus since they showed strongest packing densities for HSP-27 positive astrocytes (see, Figs. 2 and 3). In these piriform-entorhinal cortices and dentate gyri from saline-treated controls, the GS activity was nearly identical, showing that the changes in the methods of dissection between whole hippocampus, piriform-entorhinal cortex, or fascia dentata of the dentate gyrus described before did not cause changes in GS activity (Fig. 8). However, after PIRS a moderate but significant reduction of the GS activity had occurred in piriform-entorhinal cortex and the molecular and granular layer of the dentate gyrus (Fig. 8). Since it is known that GS activity is affected by nitration, which does not affect GSIR (Görg et al., 2005, 2007), we checked for nitrated proteins in WB and found several nitrated proteins, including GS (not shown). As revealed by immunoprecipitation, rats affected by PIRS showed significantly higher levels of nitrated GS when compared to control as well as to MSO-treated rats (Fig. 9). In addition, when the rat with the highest seizure score was compared with the one with the lowest seizure score, a clear difference was evident (Fig. 10). Furthermore, immunohistochemistry showed that in PTZ-treated rats many glial cells showed a colocalization of NO2Tyr-, HSP-27-, and GS-immunoreactive astrocytes as well as many neurons with NO2Tyr-immunoreactivity (Fig. 11).

Figure 9.

 After immunoprecipitation using GS-antibody, an increased amount of protein tyrosine nitration is detectable in extracts of cerebral (piriform-entorhinal) cortex obtained after repeated PTZ-induced seizures when compared to control and MSO-treatment.

Figure 10.

 It is indicated that tyrosine-nitration of GS is seizure dependent, when comparing the rat which represented a high-responder (PTZ-A, seizure activity) to that of a low-responder (PTZ-B).

Figure 11.

 Triple staining and 3D-reconstruction using Improvision software of entorhinal cortex from a PTZ-treated rat showing the distribution of GS (red) positive astrocytes and their colocalization with HSP-27 (green) and nitrotyrosine (blue). Note that tyrosine-nitration is not restricted to HSP-27 positive astrocytes, and that many neurons are also affected by tyrosine nitration. Grid square: 14,7 μm.


The present data clearly show that experimentally induced repeated seizures not only elicit an astrocytic stress response as evidenced by the focal induction of HSP-27 and other markers (Rauca et al., 2004; Bidmon et al., 2005; Dhir et al., 2007), but also affect GS by tyrosine nitration and partial inhibition, indicating that especially nitrosative stress affects glutamate catabolism within the glutamate-glutamine cycle. Furthermore, these astrocytic effects seem to be restricted to the so called epileptic circuitry (White, 2002) and regions affected by electrical stimulation of the amygdala (Pitkänen et al., 2007) and appear strongest in piriform-entorhinal cortex and the dentate gyrus, both regions to become severely effected by seizures (Schwabe et al., 2004; Freichel et al., 2006). However, seizure-related region-specific glial responses may not be always comparable with that known for neurons as previously discussed (Bidmon et al., 2004, 2005) or that seen during the onset of the first seizure episode (Koegh et al., 2005; Brevard et al., 2006). As mentioned in the introduction most previous work referred to changes in the amount of GS in epileptic regions. Therefore, we used here the PTZ model, which does not result in initial cell death (Valente et al., 2004; Bidmon et al., 2005) to address the question how seizure activity may induce early focal changes in the amount of astrocytic GS. However, since all of the described alterations of GS did not result in quantitative changes in the amount of GS, we had to take into account methodological aspects first by using the MSO model. The major differences between the PTZ model and the MSO model are the following: in the PTZ model, PTZ a GABA antagonist affects acute and directly neurotransmission at the level of the neurotransmitter receptor, which results in immediate seizure activity within seconds (Brevard et al., 2006). In contrast, MSO partially inhibits the synthesis of a precursor for the synthesis of glutamate and GABA and reduces ammonia detoxification. According to our experiments increasing levels of ammonia do not initiate seizures directly, since in our rat strain ammonia levels peaked 3.5 h post MSO injection, approximately 4–7 h before seizure onset, and they remained about 25% lower than in a rat model for acute hepatic encephalopathy in which the animals fully recover without showing seizures (Görg and Bidmon, unpublished). Furthermore, MSO-induced seizures may be caused by several complex mechanisms including cerebral energy and neurotransmitter metabolism as previously discussed (Hevor, 1996; Cloix & Hevor, 1998; Böttcher et al., 2003; Schoesboe & Waagepetersen, 2007). Therefore, in the MSO model a reduction of GS activity over time may cause disturbances at the neurotransmitter and/or adaptational threshold, where alterations at the neurotransmitter receptor level may accumulate over time and in a manner, which depends on the individual conditions of each individual. Because synapses are especially vulnerable to comparatively small cumulative changes in neurotransmitter content over time, these disturbances could render the system to a state of criticallity (Levina et al., 2007), which may be  responsible for late appearance of seizure activity. Because we had used MSO-treated animals in order to control for our applied methods, we would not further discuss the aspects of delayed MSO-induced seizures, especially since, due to its astrocyte toxicity (Nitsch et al., 1986), MSO cannot be applied repeatedly in a manner comparable to PTZ treatment.

Methodological considerations

Comparable to our previous report (Bidmon et al., 2005) experimentally induced repeated seizures caused focal  induction of glial HSP-27 indicating an astrocytic stress response. In contrast to our expectation this focal glial HSP-27 response did not result in a concomitant focal change of GSIR. This finding was not due to a lack of sensitivity of our applied immunohistological techniques,  because in MSO-treated animals (positive control) our antibody detected a significant widespread reduction of GSIR. The MSO-induced reduction of GSIR was comparable to observations reported for MSO-treated hyperammonemic rat brains (Tanigami et al., 2005). In addition, this positive control experiment revealed that the inhibition of GS elicits a more homogeneous astrocytic stress response in all individuals indicating that inhibition of GS is sufficient to elicit an astrocytic stress response within 24  h than in our rats affected by PIRS (Bidmon et al., 2005). However, the MSO-induced reduction of GSIR had occurred independently from changes in the total amount GS protein as shown for the heated and denatured protein using DBs and WBs. These contradicting findings provided clear evidence that the reduction of GSIR in our positive control did not reflect real changes in the amount of GS. As shown by DBs using nondenatured protein extracts the MSO-induced reduction of GS was reproducible indicating that MSO binding to GS obviously interfered with antibody binding.

Therefore, our findings indicate that the binding of MSO to the active center of GS (Berlicki & Kafarski, 2006) and its phosphorylation either blocks the epitope-specific site for antibody binding, or it may cause conformational changes within the structure of GS, resulting in a loss of affinity for the antibody. Since no significant increase in GS nitration occurred in cerebral extracts of MSO-treated rats and nitration does not interfere with antibody binding (Görg et al., 2005), it could be ruled out that nitration of GS results in reduced antibody binding. Furthermore, reduction of antibody binding could be reversed by boiling the protein extracts as shown in the present study. Thus, these MSO-dependent changes are obviously heat sensitive (Ronzio & Meister, 1968; Rowe et al., 1969).

Since certain antiepileptic drugs inhibit GS activity probably by a similar mechanism as MSO (Fraser et al., 1999), these methodological findings have to be considered when studying GS contents in tissue samples obtained from patients with epilepsy by means of immunochemistry.

PTZ-induced seizures, nitration, and inhibition of GS

In contrast to the well-characterized focal induction of HSP-27 by PIRS (Bidmon et al., 2005), no changes in the immunoreactivity of GS were present in the most severely affected brain regions. Also on a topographical basis, there was no indication that those astrocytes, which showed a distinct strong HSP-27 response after PIRS had concomitantly changed their GS content. These histological findings were supported by the biochemical data obtained by WB and DB. In addition, these findings were also supported by data obtained for the pilocarpin model for which an extensive protein analysis had been performed which showed that a significant rise in HSP-27 protein is not accompanied by quantitative changes of GS (Greene et al., 2007).

Therefore, our data support the findings by Steffens et al. (2005) who reported no changes for GS in the temporal cortex resected from epileptic patients as well as those reported for the nonsclerotic hippocampus by Van der Hel et al. (2005). The data indicate, that in the PTZ-model of PIRS we were unable to evoke a reduction in the amount of GS as it had been reported for the human epileptic, sclerotic hippocampus from patients with mesial temporal lobe epilepsy (MTLE; Eid et al., 2004: Van der Hel et al., 2005). Even when the measurements were restricted to brain regions, which showed a high focal stress response no alterations in GSIR were found on a histochemical basis or by means of biochemistry detecting the native or heat denatured protein. According to these observations, it becomes evident that PIRS do not induce changes with the severity comparable to that found in sclerotic hippocampi of human MTLE patients, despite the fact that our PTZ-model showed a similar cortical and hippocampal induction of HSP-27 as known from epileptic patients (Bidmon et al., 2004, 2005). Therefore, the PTZ-model of tonic–clonic seizures (Brevard et al., 2006) may be a model for the study of the early processes in the nonsclerotic tissue at the  borderline between epileptic foci and normal tissue. This suggestion may be supported by findings in other animal models of epilepsy, showing weak or no changes of certain aspects of the glutamate-glutamine cycle which are not associated with an abnormal metabolism of astrocytes (Melo et al., 2005; Greene et al., 2007), thus causing in some models even an increase in glutamine (Shirayama et al., 2005), a finding also described for epileptic patients examined by means of 13C-magnetic resonance spectroscopy (Otsuki et al., 2005; Helms et al., 2006). In this regard GS content seems to be relatively insensitive to pure seizure activity as repetitively induced by PTZ (current study) or seen in nonsclerotic cerebral tissues of patients (van der Hel et al., 2005; Steffens et al., 2005). Certainly, there may also be region-specific differences between hippocampus and temporal cortex as could be concluded when referring to the findings of Steffens et al. (2005) and Eid et al. (2004), but in our experimentally treated rats neither PIRS nor MSO treatment provided clear differences in the responses observed for cortical and hippocampal astrocytes. Therefore, it seems to be understandable that a reduction of GS content may occur under severe conditions during which a loss of functional cells is noted. This could be due to a loss of neurons which may contain minor amounts of GS in there synaptic compartment (Dennis et al., 1980) or to a concomitant loss of astrocytes in regions where neurodegeneration leads to symptoms more related to a microinfarct-like pathology compared to that related to pure seizure activity. This loss of astrocytes may be overlooked, because under normal conditions there may have been many more astrocytes present than usually detected by staining for GFAP (Hajos & Zilles, 1995; Walz & Lang, 1998). Thus, a pathology-related increased GFAP expression (Hajos & Zilles, 1995) does not necessarily preclude a simultaneous death of astrocytes as well as of neurons in the affected regions, since the remaining astrocytes and previously GFAP-negative astrocytes increase their amount of GFAP in response to pathology. This hypothesis is also supported by the finding that after pilocarpin-induced status epilepticus a degeneration as well as partial regeneration of astrocytes has been noticed (Borges et al., 2006; Kang et al., 2006). In addition, Kang et al. (2006) reported that the newly generated astrocytes showed a reduced expression of GS and glutamate transporter. In our study neither MSO treatment nor PIRS caused a significant increase in GFAP immunoreactivity (Fig. 5) indicating that in our model alterations that may go along with gliosis are negligible.

However, as pointed out in the introduction, GS holds a key role within the glutamate-glutamine cycle, and finding that GS content is affected in most severely damaged, sclerotic foci calls for answers as to how these changes may start, probably in a way initially not affecting the amount of GS. The first indication that this may be the case came from the present results obtained by focusing on the “hot spots” of glial HSP-27 induction which could be dissected with reliable accuracy (the piriform-entorhinal cortex and hippocampal fascia dentata) and which showed a significant reduction in GS activity, thus supporting the observation that PIRS decreases glutamate turnover (Eloqayli et al., 2003). Since it was already known that the nitration of purified GS causes a loss of its activity without affecting its detectability by antibody binding (Görg et al., 2005) and that piriform-entorhinal cortex contains high amounts of neuronal nitric oxide synthase (Bidmon et al., 1997), we searched for nitrated proteins. Indeed, tyrosine-nitrated proteins were enriched in neurons and glial cells of the piriform-entorhinal cortex of rats with PIRS. Among the nitrated proteins, tyrosine-nitrated GS was significantly  increased. Although the amount of tyrosine-nitration was increased in PTZ-treated rats, the individuals with the lowest score for seizure intensity exhibited lower levels of tyrosine-nitrated GS in comparison to the individual with the highest seizure score, thus indicating that tyrosine-nitration of GS is related to seizures and not due to other pharmacological effects of PTZ. This is also supported by our positive control, since MSO-treated rats which had undergone only one seizure episode showed no significant increase of GS nitration when compared to saline-treated controls. Direct evidence for a seizure-dependent protein tyrosine nitration is also known from seizures induced by hyperbaric oxygen treatment (Chavko et al., 2003), also supporting the relation to seizure activity, whereas the neuronal protein tyrosine nitration during kainate-induced neurodegeneration in metallothionein-I + II-deficient mice (Carrasco et al., 2000) seems to be more associated with cell death. In addition, oxidative and nitrosative stress may have also affected neurons after PIRS (current study Fig. 11), probably resulting in nitration of manganese-dependent superoxide dismutase which has been shown to become affected after PIRS (Rauca et al., 2004) and which is sensitive to tyrosine-nitration (Bayir et al., 2007). Since in our PTZ-model 40 mg PTZ/kg did not result in neurodegeneration (Bidmon et al., 2005), astrocytic HSP-27 induction together with the observed protein tyrosine nitration of GS strongly indicates that seizures not only affect  neurons but also astrocytes directly by both oxidative and nitrosative stress. That explains their strong HSP-27 expression also known as a reliable marker for  oxidative stress as previously discussed above. Whether the observed partial inhibition of GS by tyrosine nitration has to be viewed as a contribution to astrocyte pathology or is part of an initially protective cell and tissue response remains to be revealed by future studies. Currently it has been shown in vivo (Shin et al., 2003) and in vitro for epileptic tissue slices, that the addition of glutamine increases seizures (Tani et al., 2007). Therefore, the reduction of GS activity could be part of an endogenous protective response by reducing glutamine levels as a precursor for neuronal glutamate synthesis. As suggested for peripheral tissues, the nitration of certain proteins may be part of a regulatory mechanism of their activity, since they can be reactivated by denitration due to increased denitrase activity known to occur also in the brain (Kuo et al., 1999; Kosenko et al., 2003; Görg et al., 2007). Because the glutamate-glutamine cycle and its fine tuning between neurons and glial cells is most important for normal cerebral activity and astrocytes represent essential constituents within that system (Tian et al., 2005; Volterra & Meldolesi, 2005), astrocytic impairment of glutamate metabolism may have severe consequences over time, also in neurons either by delayed glutamate clearance from extracellular space or by impairing the timely and sufficient glutamine supply for the synthesis of both glutamate and GABA. Concentrations of glutamate and glutamine can be determined either by a regulation of their synthesis and/or by their catabolism as well as through modulation of the time-dependent micro-compartmentalization among neurons, extracellular spaces, and glia and/or endothelial cells. Since it has been recognized that neurons as well as glial cells are not only able to respond to neurotransmitters (Had-Aissouni et al., 2002; Volterra & Meldolesi, 2005), but that both cell types are also able to modulate cerebral activity through neurotransmitter regulation, namely by glutamate storage and release (Chen et al., 2005) as well as by stimulation of neuronal glutamate release (Perea & Araque, 2007), it is conceivable that even a mild delay in astrocytic glutamate metabolism could affect normal  neurotransmission even independently of total transmitter concentrations within the whole tissue just by creating pathological distribution patterns of glutamate and glutamine in time. Therefore, we hypothesize that a pathology-related, adaptive regulatory mechanism, which involves the nitration of GS may stand at the beginning of a neuropathological cascade, which over time may transform the tissue in a way that it becomes epileptogenic. Neuronal degeneration and glial degeneration/proliferation and sclerosis may then develop during progressing epilepsy leading to the pathological alterations known from cerebral tissues resected from patients with epilepsy.


  • 1Repeated PTZ-induced seizures result in a focal and region-specific induction of HSP-27 in astrocytes but do not result in generalized or concomitant focal quantitative changes of GS content.
  • 2In regions such as piriform-entorhinal cortex and the fascia dentata of the hippocampus which are affected most severely by glial HSP-27 induction, a concomitant nitration and partial inhibition of GS occurred indicating that PTZ-induced seizures affect GS and thereby may alter Glu and GABA metabolism and/or catabolism in a region-specific manner within the epileptic circuitry. Therefore, focal glial HSP-27 induction may not only be a marker for an oxidative stress response, but could be viewed as glia-specific marker for the identification of affected astrocytes.
  • 3Glial GS should be viewed as target for therapeutic interventions.


The authors thank B. Herrenpoth, L. Igdalova, and C. Opfermann-Rüngeler for excellent technical assistance and artwork, and Dr. L. Berlicki, Department of Bioorganic Chemistry, Wroclaw University of Technology, Poland for helpful comments. This study was supported by the Deutsche Forschungsgemeinschaft and in part by SFB 575.

Conflict of interest: We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. None of the authors has any conflict of interest to disclose.