Nine-day-old Long-Evans black-hooded rat pups of both sexes were placed in a stereotaxic frame adapted for newborns (Kopf) under isofluorane anaesthesia. The skull was opened using a surgical blade, and 0.15 μl of saline solution (0.9% NaCl, pH 7.4) containing 18,5 nmol of N-methyl-D-aspartate (NMDA) (Sigma, M-3262, Germany) were injected into the right sensorimotor cortex at the level of the coronal suture (2 mm lateral from bregma and at 0.5 mm of depth). Control animals received an injection of 0.15 μl of vehicle saline solution. After suture, pups were placed in a thermal pad and maintained at normothermia for 2 hours before being returned to their mothers. Experimental animal work was conducted according to Spanish regulations, in agreement with European Union directives. This experimental procedure was approved by the ethical commission of the Autonomous University of Barcelona. All efforts were made to minimize animal suffering in every step.

Intact immature brains from P9, P10, P12, and P16 rats of both sexes where used for the analysis of Cu/Zn SOD expression under physiological conditions (2–4 animals per time). Moreover, at 2, 4, and 10 hours, and 1, 3, 5 and 7 days after NMDA (3–4 animals per time) or saline injection (2 animals per time), rats were anaesthetized and perfused intracardially with 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). Brains were post-fixed in the same fixative for 2 hours and sunk in a 30% sucrose solution before being frozen with dry CO₂. Coronal sections (30-μm-thick) were obtained using a Leitz cryostat. Free-floating parallel sections were treated with 10% foetal calf serum in Tris-buffered saline (TBS: 0.05 M Trizma base containing 150 mM of NaCl, pH 7.4) +1% triton X-100 for 1 hour, and incubated overnight at 4°C with sheep polyclonal anti-Cu/Zn SOD (574597, Calbiochem, Darmstad, Germany) (1:300) in the same solution. Afterwards, sections were rinsed and incubated for 1 hour at room temperature with Cy3-conjugated anti-sheep secondary antibody (AP147C, Chemicon, California, USA) (1:150). As negative controls, sections were incubated in media lacking the primary antibody.

Double labelling was carried out in order to identify Cu/Zn SOD-expressing cells in sections previously immunoreacted for Cu/Zn SOD as reported above.

For neuronal identification, sections were further incubated with a monoclonal anti-NeuN antibody (MAB377, Chemicon, California, USA) (1:1000) and Cy2-conjugated anti-mouse secondary antibody (PA-42002, Amersham Pharmacia Biotech, England) (1:1000).

For astroglial labelling, sections were incubated with a polyclonal anti-GFAP antibody (Z-0334, Dakopatts, Denmark) (1:1800), and immunostaining was visualized with Cy2-conjugated anti-rabbit secondary antibody (PA-42004, Amersham Pharmacia Biotech, England) (1:1000).

As a microglial marker, sections were processed for double staining with tomato lectin histochemistry by incubating with the biotinylated lectin from Lycopersicon esculentum (tomato) (Sigma, L-9389, Germany) diluted to 6 μg/ml, followed by Cy2-conjugated streptavidin (PA-42000, Amersham Pharmacia Biotech, England) (1:1000).

Selected sections of all double labelling techniques were incubated for 5 min with a 0.00125 μg/ml solution of 4, 6-diamino-2-phenylindole (DAPI) in TBS. Double-stained sections were analyzed using a Nikon Eclipse E600 epifluorescence microscope and a LEICA TCS SP2 AOBS confocal microscope.

Terminal dUTP Nick End Labelling (TUNEL) staining for detection of DNA fragmentation was performed on parallel sections mounted on slides. Tissue sections were rinsed in Tris buffer (10 mM, pH 8) and EDTA (5 mM) and then incubated in the same buffer plus Proteinase K (20 μg/ml) for 15 min. at room temperature. After several washes with EDTA (5 mM), sections were incubated for 10 min. in TdT buffer (Tris 30 mM, 140 mM Sodium Cacodilate, 1 mM Cobalt chloride, pH 7.7). Sections were then incubated in TdT buffer plus 0.161 Units/ μl TdT enzyme (Terminal Transferase, 3333566 Roche, Manheim, Germany) and 0.0161 nmol/ μl of biotin-16-dUTP (1093070, Roche, Manheim, Germany) for 30 min. at 37°C. The reaction was stopped by washing the sections in citrate buffer (300 mM sodium chloride, 30 mM sodium citrate, 5 mM EDTA). After several washes with TBS, sections were incubated with avidin-peroxidase (P-0364, Dakopatts, Denmark) (1:400) for another hour at room temperature. Finally, the peroxidase reaction product was visualised in 100 ml of Tris buffer containing 50 mg of 3'-diaminobenzidine and 0.01% of hydrogen peroxide.

Three or four NMDA-injected and two intact and saline-injected animals for each survival time were decapitated, and the complete injected cortex quickly extracted, chopped and frozen in liquid nitrogen. Samples were resuspended in Tris/HCl (50 mM, pH7.8), EDTA (1 mM), DTT (1 mM), PMSF (100 μg/ml), pepstatin A (2 μg/ml), leupeptin (2 μg/ml), trypsin inhibitor (10 μg/ml), benzamidine (0.2 mM) and submitted to mechanical dissociation. Total protein concentration was measured by the bicinchoninic acid method and equal quantities of protein were run on a 15% SDS-polyacrilamide gel electrophoresis (SDS-PAGE). Electrotransfered protein samples to polyvinylidene fluoride (PVDF) membranes were incubated overnight at 4°C with TBS+0.3% Tween 20 and 5% non-fat milk, and for 2 hours at room temperature with sheep polyclonal anti-Cu/Zn SOD (574597, Calbiochem, Darmstad, Germany)(1:1000). Membranes were rinsed and incubated in biotinylated anti-sheep antibody (1:1000) (RPN-1025, Amersham Pharmacia Biotech, England), avidin-peroxidase (1:2000) (P0364, Dakopatts, Denmark) and finally in the chemiluminiscent substrate SuperSignal West Pico (PIERCE) combined with exposure on Hyperfilm ECL (Amersham).

Semi-quantitative estimation of western blot protein signals were performed by measuring band integral intensity with analySIS^® software after high resolution scanning.

All results are expressed as mean ± standard error mean (SEM). ANOVA followed by Fisher's PLSD post-hoc test was used to determine significant differences (p<0.05) after western blot densitometry.

As there is no available description of Cu/Zn SOD cell-specific expression in the immature brain, we began our study with this general description. No changes in the distribution pattern and cellular identity of Cu/Zn-expressing cells were found between P9 to P16 and thus will be described in general. The immature rat brain showed widespread immunoreactivity for Cu/Zn SOD, mainly located in neurons. The most intense staining was observed in pyramidal neurons located in cortical layers V, III and II (Fig. 1A, B), in the pyriform cortex, and in the pyramidal neurons of the olfactory tubercle. Sub-plate neurons, a transient cell population located beneath the cortical plate 39 were also among the most intense cells (Fig 4A). In addition, hippocampal CA pyramidal neurons and some inter neurons (Fig. 1C), as well as dentate gyrus granular and sub-granular neurons, medial septum neurons, and hypothalamic neurons also displayed intense staining. There was also more diffuse staining in many other neuronal populations as for example in various thalamic nuclei, substantia nigra (Fig. 1D), and striatum. Most of the immunoreactivity was mainly observed in neuronal soma although less intense staining was also present in the nuclei, sparing the nucleolus.

No specific Cu/Zn SOD immunoreactivity was detected in parenchymal grey matter GFAP-positive astrocytes (Fig 2A). However, co-expression of Cu/Zn SOD and GFAP was observed in scattered tanycytes of the third ventricle wall (Fig 2B), in corpus callosum, and in astrocytes forming the glia limitans (Fig 2C). Most ciliated ependymal cells expressed Cu/Zn SOD (Fig. 2B). Moreover, besides the ependymal cells described above, co-localization confocal studies with tomato lectin, which specifically detects microglial cells as well as endothelium 40, did not show overlap with Cu/Zn SOD staining (Fig 2D). No consistently Cu/Zn SOD-labelled cells were observed in white matter tracts, which indicated that oligodendrocytes at this location do not seem to express this enzyme.

Finally, western blots showed a significant developmental increase in the total amount of brain Cu/Zn SOD between P9 and P16 (Fig. 3B), evidenced as a single band of 16 KDa corresponding to the expected molecular mass of the enzyme monomer.

Control intracortical saline injection caused a local and transient increase in Cu/Zn SOD immunoreactivity mainly in cortical neurons (data not shown) 10 hours later surrounding the injection site. A trend towards a rapid and transient increase could also be observed by western blots (Fig 3B).

As shown before, intracortical NMDA administration is a model of excitotoxic damage that triggers rapid neuronal death and tissue injury, which expands rostro-caudally and includes part of the cortex, corpus callosum, dorsal striatum, septum and rostral hippocampus 3341.

In contrast to saline injected animals, as observed both by western blot tendencies (Fig. 3) and immunohistochemistry (Fig. 4, 5) at 2–4 hours after NMDA injection, the lesioned neural parenchyma showed a drastic downregulation of total Cu/Zn SOD immunoreactivity in neuronal cells, which returned to basal levels 24 hours after the lesion and increased significantly thereafter due to expression in astroglial cells. Interestingly, the early decrease in neuronal Cu/Zn SOD immunoreactivity at 2–4 hours (Fig. 4A–B) was accompanied by a slight condensation of nuclear chromatin (Fig. 4C–D) and changes in NeuN neuronal marker (Fig. 4F–G), in the absence of apparent signs of neuronal degeneration or TUNEL staining (Fig. 4H). At later time points, from 10 hours, degenerating neurons still showed downregulated Cu/Zn SOD immunoreactivity, but displayed condensed nuclei, nuclear blebbing (Fig. 4E) and also DNA fragmentation observed by TUNEL staining (Fig. 4I). Downregulation of Cu/Zn SOD and signs of neuronal degeneration were observed from 10 hours in primary cortical degenerating areas and from 1 day post-lesion in secondary degenerating regions such as the cortical, striatal and hippocampal penumbra (Fig. 5A). However, the total Cu/Zn SOD enzyme level increased from 1 day post-lesion, reaching a maximum induction after 3–7 days (Fig 3B and 5B), due to astroglial upregulation of the enzyme. Cu/Zn SOD was induced in the soma and proximal projections of the most hypertrophied and most intensely GFAP-immunopositive astroglial cells of the whole degenerative area including the white matter (Fig. 5C), but not in slightly activated astrocytes with lower GFAP immunoreactivity. This elevated expression continued at 5 and 7 days post-lesion when most astrocytes of the degenerative area were strongly hypertrophic and formed the glial scar (Fig. 5D). Only scattered reactive microglial cells expressed Cu/Zn SOD at 3 days after lesion, but most reactive microglial cells of the lesioned parenchyma remained negative.

It is well known that total brain levels of antioxidant enzymes vary throughout life 324142. Accordingly, brain glutathione peroxidase and Mn SOD were reported to increase during the first month of postnatal rat life, and to continue increasing slightly during adulthood and aging 2843444546. Catalase has been reported to peak around the first week of life and then decline to reach a plateau by the first month 4344. Regarding Cu/Zn SOD, it is known that its levels increase rapidly after birth, peaking around the second postnatal week, in agreement with our western blot studies. Later on, Cu/Zn SOD decreases slightly to reach adult levels, but increases slightly again in aging 424445. Thus, it seems clear that immature rat brain has a different balance of antioxidant enzymes, which reach the adult overall pattern only after the first month of life.

We have shown that in the postnatal brain the majority of Cu/Zn SOD immunoreactivity is observed in neuronal somas with less intense staining present in the nuclei, sparing the nucleoli, as has been reported earlier for adult animals 12131415161747. Accordingly, even though changes in Cu/Zn SOD levels occur in postnatal development, the cell type distribution of this enzyme does not change. Neurons are the brain cells with the highest oxygen consumption, constantly submitted to oxidative stress as it is suggested by O₂^-. mediated oxidation of hydroethidine 9, or by the presence of nitrotyrosine 4849. Accordingly, it is not surprising that high levels of Cu/Zn SOD are observed in a wide number of neuronal populations, specifically in the large neurons with high energetic requirements, like pyramidal neurons, or catecholaminergic neurons which are submitted to elevated oxidative stress due to catecholamine metabolism 50. Very recently, a site for the pro-inflammatory transcription factor NFκB has been reported at the human Cu/Zn SOD promoter, which can induce its expression 51. Accordingly, in the CNS, neurons are the main cells that display constitutively activated NFκB 5253, and its activity is required for their survival 54. In addition, the other important cellular superoxide dismutase, the Mn SOD, is also enriched in subsets of neurons 55, although different from those enriched in Cu/Zn SOD.

No immunoreactivity for Cu/Zn SOD was observed in the GFAP-expressing astrocyte population of the neural parenchyma in the immature brain, as has been previously reported for adult brain 1314151747, and despite the fact that immature astrocytes differ from adult astrocytes. This lack of Cu/Zn SOD expression contrasts with the higher tolerance of astrocytes to oxidative stress 4856. Accordingly, and in comparison with neuronal cells, astrocytes in culture have been shown to have higher levels of glutathione and the lipophilic antioxidant vitamin E 57, and they are capable of synthesizing their own glutathione from cysteine 58, mechanisms that probably confer upon astrocytes an elevated antioxidant status and increased resistance towards oxidative stress, despite the absence of detectable levels of Cu/Zn SOD. Noteworthy, Cu/Zn SOD immunoreactivity is only found in glia limitans astrocytes and in GFAP-positive tanycytes, which could be attributed to the contact of these astrocytes with non-CNS molecules, which are also known to induce the expression of several other activation markers observed previously in these regions, like GFAP overexpression, vimentin or metallothioneins 41. Regarding microglial cells, neither resting ramified microglia nor amoeboid microglial cells found in the immature brain showed Cu/Zn SOD immunoreactivity, as has been shown for adult resting microglial cells 121747. In addition, we were unable to find consistent immunoreactivity for Cu/Zn SOD in white matter oligodendrocytes of immature brain, as has been described in the adult 121747. The lack of this enzyme could help explain the well known high sensitivity of oligodendrocytes to excitotoxicity and oxidative stress 59, especially in the immature brain where white matter injury is thought to be the underlying mechanism of brain damage 38.

After an excitotoxic injury to the postnatal brain, we have observed a dramatic and rapid neuronal downregulation of Cu/Zn SOD in the NMDA injection site, which is early evident 2–4 hours after injection in neurons that only show slight and very early signs of degeneration and are negative for TUNEL staining. In fact we have previously shown that these neurons display, 10 hours after NMDA injection, NFκB activation and COX2 upregulation, suggesting that they are still active and functional 5360. The Cu/Zn SOD downregulation also coincides with neuronal nitration 48 suggesting endogenous O₂^-./peroxynitrite formation at these very early time points in compromised neurons. Although to our knowledge this is the first study describing the expression of Cu/Zn SOD after immature brain damage, studies on adult brain injury have also shown, as early as 4 hours after transient cerebral ischemia, Cu/Zn SOD immunoreactivity downregulation in striatum and cortex 37 and in neurons of the hippocampal CA region 13. Moreover, 24 hours after injection of kainate to adult rat hippocampus, Cu/Zn SOD immunoreactivity is also downregulated in neurons of CA, despite an absence of apparent neuronal degeneration at that time-point 14. Accordingly, it is known that superoxide scavengers protect from injury at these early time points 232661. Although the mechanism whereby this rapid downregulation occurs is not clear, it appears that it could be mediated by oxidative stress, as PC12 cells treated with H₂O₂ rapidly (after 4 hours) downregulate Cu/Zn SOD 51. Interestingly, hypothermia, the most powerful neuroprotective strategy known, not only inhibits the rapid downregulation of Cu/Zn SOD after a traumatic brain injury but also induces its over-expression 36. Most surprisingly, this effect is specific for Cu/Zn SOD, and in fact hypothermia induces a less significant upregulation of other antioxidant enzymes such as catalase and glutathione peroxidase in comparison with non-hypothermic brain.

Contradictory results have been reported regarding the toxicity of O₂^-. after hypoxic/ischemic injury to the immature brain. Whereas slightly worsened neuropathological outcome was observed in transgenic mice overexpressing Cu/Zn SOD and submitted to severe hypoxia/ischemia 27, several antioxidant molecules including SOD mimetics as O₂^-. dismuting metalloporphyrins were shown to be neuroprotective 26. One hypothesis is that Cu/Zn SOD transgenic mice produce excess H₂O₂ that in the immature brain is not cleared by the upregulation of the glutathione peroxidase as has been reported for adult animals 2862. However, an additional explanation would be that as the postnatal brain express increased amounts of the NMDA receptor 63, whose activation leads to increased O₂^-. generation 3435, an initial enhanced oxidative stress would occur. Taken together, this data suggest that in the immature brain subjected to ischemia/excitotoxicity, neurons at the lesion zone are very early submitted to an elevated oxidative stress, and therefore Cu/Zn SOD downregulation contributes to the further amplification of cell damage and neuronal cell death.

Although Cu/Zn SOD expression remains very low in compromised neurons, a return to normal expression levels is seen by 24 hours after lesion, and an increase in total enzyme level is observed later on. This secondary Cu/Zn SOD induction is due to upregulation in reactive hypertrophic astrocytes within the lesion site. We have shown in previous studies that these reactive astrocytes display activated NFκB from 10 hours after the excitotoxic lesion 53, which could contribute to the induction of the Cu/Zn SOD observed here. We have also shown that the hypertrophic Cu/Zn SOD-overexpressing astrocytes are heavily nitrated, and also display metallothionein I-II expression, suggesting an elevated rate of oxidative stress in this particular group of cells 48. Our findings in immature brain are in accordance with studies of adult brain damage that have shown induction of Cu/Zn SOD and Mn SOD in astrocytes several days after focal ischemia 37 excitotoxicity 1418, or in Alzheimer's disease and Down's Syndrome 16. In addition to the upregulated enzymatic antioxidant defences, astrocytes have been shown to produce their own glutathione and also provide neurons with cysteine, a rate-limiting precursor in neuronal glutathione synthesis 58. Thus, astrocytes seem to be the main cell type increasing the total antioxidant capabilities in the nervous tissue after a lesion, which can in addition explain their elevated resistance to cell death after an injury. In this sense, previous studies showed that Cu/Zn SOD-overexpressing astrocytes have increased resistance to oxidative damage 64 and attenuated oxidative inhibition of glutamate uptake 65, allowing for a maintenance of their physiological functions after a lesion.

Regarding microglial cells, it is somehow surprising that only a reduced number of reactive ameboid microglial cells express Cu/Zn SOD, and that this occurs very transiently, as in some circumstances activated microglia produce large amounts of oxygen radicals after a lesion including O₂^-. 66.

We believe that the results presented here highlight the importance of in vivo cell-localization studies for antioxidants, as some of the reactive species like O₂^-. do not diffuse across cell membranes and thus will most probably react within the cell where it is formed.