Caspase-3 activation and DNA fragmentation in primary hippocampal neurons following glutamate excitotoxicity

Molecular Brain Research 94 (2001) 25–34 www.elsevier.com / locate / bres

Research report

Caspase-3 activation and DNA fragmentation in primary hippocampal neurons following glutamate excitotoxicity Stephan Brecht a

1 ,a ,

*, Mathias Gelderblom 1 ,a , Anu Srinivasan b , Kirsten Mielke a , Galina Dityateva c , Thomas Herdegen a

¨ Pharmakologie, Christian-Albrechts-Universitat ¨ , Hospitalstraße 4, 24105 Kiel, Germany Institut f ur b IDUN Pharmaceuticals, San Diego, CA 92037, USA c ¨ Molekulare Neurobiologie, 20246 Hamburg, Germany Zentrum f ur Accepted 26 June 2001

Abstract Excitotoxic glutamate CNS stimulation can result in neuronal cell death. Contributing mechanisms and markers of cell death are the activation of caspase-3 and DNA fragmentation. It remains to be resolved to which extent both cellular reactions overlap and / or indicate different processes of neurodegeneration. In this study, mixed neuronal cultures from newborn mice pubs (0–24 h) were stimulated with glutamate, and the co-localization of active caspase-3 and DNA fragmentation was investigated by immunocytochemistry and the TUNEL nick-end labelling. In untreated cultures, 8% scattered neurons (marked by MAP-2) displayed activated caspase-3 at different morphological stages of degeneration. TUNEL staining was detected in 5% of cell nuclei including GFAP-positive astrocytes. However, co-localization of active caspase-3 with TUNEL was less than 2%. After glutamate stimulation (125 mM), the majority of neurons was dying between 12 and 24 h. The absolute number of active caspase-3 neurons increased only moderately but in relation of surviving neurons after 24 h from 8 to 36% (125 mM), to 53% (250 mM) or to 32% (500 mM). TUNEL staining also increased after 24 h following glutamate treatment to 37% but the co-localization with active caspase-3 remained at the basal low level of 2%. In our system, glutamate-mediated excitotoxicity effects the DNA fragmentation and caspase-3 activation. Co-localization of both parameters, however, is very poor. Active caspase-3 in the absence of TUNEL indicates a dynamic degenerative process, whereas TUNEL marks the end stage of severe irreversible cell damage regardless to the origin of the cell.  2001 Elsevier Science B.V. All rights reserved. Theme: Neurotransmitters, modulators, transporters and receptors Topic: Excitatory amino acids: excitotoxicity Keywords: Cell death; GFAP; MAP-2; TUNEL

1. Introduction Neuronal cell death occurs under various physiological [52,47] and pathophysiological [12] conditions such as stroke [29] or neurodegenerative disease. Apoptotic cell death is defined as an active process resulting in characteristic morphological and metabolic features. As marker for (apoptotic) cell death, TUNEL (terminal transferase

*Corresponding author. Tel.: 149-431-597-3524; fax: 149-431-5973522. E-mail address: [email protected] (S. Brecht). 1 Both authors contributed equally to this work.

dUTP nick end labelling) is widely used to detect characteristic DNA fragmentation [18]. However, there is evidence that TUNEL is not specific for apoptosis and internucleosomal DNA fragmentation was also seen in cells with a morphology typical for necrosis or mixed types of cell death [5,17,35,39]. TUNEL indicates late stages of oxidative DNA damage [8] in apoptosis and necrosis, which occur not before many hours or days after the onset of the degenerative stimulus [34,56]. Another characteristic of apoptosis is the activation of caspase-3, a cysteine aspartate protease and mammalian homologue of ced3 from Caenorhabditis elegans [27]. The inactive 32kDa caspase-3 precursor protein CPP32 is activated by cleavage into an active 17-kDa and an inactive 20-kDa fragment. Upstream signal pathways of caspase-3 include

0006-8993 / 01 / $ – see front matter  2001 Elsevier Science B.V. All rights reserved. PII: S0006-8993( 01 )02767-6

26

S. Brecht et al. / Molecular Brain Research 94 (2001) 25 – 34

the damage of mitochondria with subsequent cytochrome C release [46,31], activation of caspase 11 or granzyme B [9,10] and activation of apoptosis inducing factor (AIF) [49]. The release of mitochondrial cytochrome C is supposed to trigger caspase-3 activation prior to DNA fragmentation [43], but caspase-3 can also be activated in the absence of mitochondrial stress [31]. In general, enzyme expression and / or activities are commonly used to detect cell death in vivo and in cell culture. Caspase-3 de novo synthesis has been described following cerebral ischemia in rats [36] and man [30] and its activation is essentially involved in the apoptotic cell death cascade following spinal cord injury [46], axotomy [6] or kainic acid treatment [16]. But the intracellular localization of active caspase-3 can vary. It was detected (i) in nuclei [16], (ii) in the cytoplasm of neurons and microglia [53], (iii) in cytoplasmatic granula following cerebral ischemia [34] and (iv) in neuronal clumps of the substantia nigra [24]. The importance of active caspase-3 for cell death is the attenuation of neuronal cell death by pharmacological inhibition following rodent cerebral ischemia [7,13,23,15] or axon transection [6] even if these inhibitors are not specific for caspase-3 and block the activity of other proteases such as cathepsin B [44]. Furthermore, the co-localization of TUNEL and active caspase-3 is also an issue under controversial discussion. Following cerebral ischemia, the majority of cytoplasmatic active caspase-3 (60–80%) was found in co-localization with TUNEL nuclei, whereas the majority of TUNEL nuclei was devoid of active caspase-3 labelling [34]. In contradiction, a strong link between active caspase-3 and TUNEL or even a stronger prominence of active caspase-3 than TUNEL in hypoxic neurons has been observed [56]. Reasons for these conflicting findings might be a different sequence of apoptotic events depending on different cell types or pathophysiological processes and the unspecific labeling of TUNEL in different or mixed forms of cell death as introduced above. Given the importance of active caspase-3 and TUNEL, we aimed to investigate the intracellular localization of active caspase-3 and the co-localization with the DNA fragmentation marker TUNEL. Primary hippocampal mixed cultures of postnatal murine pubs were stimulated with 125, 250 and 500 mM glutamate, resulting in sustained excitotoxicity as it occurs following epileptic seizures or cerebral ischemia [29]. Our findings demonstrate cytoplasmatic or total neuron active caspase-3 in untreated and glutamate stimulated neurons depending on the morphological changes of early or late degeneration. TUNEL staining was also found in untreated cultures but is hardly detectable within a cell-like morphology. Whereas TUNEL was inducible in neurons and in astroglia, active caspase-3 was restricted to neurons. TUNEL, however, was poorly co-localized with active caspase-3. Finally, TUNEL-positive isolated DNA clumps without remaining cytoplasm persist in culture dishes and by staining necrosis may lead to false-positive results of neuronal apoptosis.

2. Material and methods

2.1. Cell cultures To obtain mixed neuron–glia co-cultures as described before [11] the hippocampi of newborn C57BL6 mice pubs (up to 24 h postpartal) were dissected under sterile conditions in 1-mm pieces in ice-cold dissection solution (Hanks with albumin 3 mg / ml, MgSO 4 1.4 mg / ml) after removal of the leptomeniges. After washing and rewarming the hippocampal slices for 5 min in digestion solution (NaCl, 8 mg / ml, KCl, 37 mg / ml, Na 2 HPO 4 , 0.99 mg / ml, Hepes, 5.95 mg / ml, NaHCO 3 , 0.35 mg / ml mixed with trypsin, 3.3 mg / ml, DNAse, 0.83 mg / ml), the hippocampal slices were incubated in dissection solution with trypsin inhibitor for 5 and 3 min, and finally with horse serum (Sigma, 200 ml / ml dissection solution) for 10 min. After tissue homogenization and 2-fold centrifugation (15 min, 800 rpm, 48C) of 5 ml cell suspension in dissection solution, cells were counted after identification with trypan blue. Subsequently, cells (150 000 / cm 2 ) were plated on poly-L-lysine- (100 mg / ml) and laminin (40 mg / ml)coated plastic coverslips in MEM with D-glucose (5 mg / ml), transferrin (0.1 mg / ml), insulin (25 mg / ml), glutamax (4 mM), gentamicin 5 mg / ml and 10% heat-inactivated horse serum at 378C, gassed with 95% air and 5% CO 2 . After 2 days in culture, we reduced the content of horse serum to 5% and added 1% B27-Supplement (Gibco) and 5 mM cytosine-b-arabinofuranoside (Sigma). This procedure results in a mixed neuronal culture containing 70–80% neurons. The remaining cells were identified as astrocytes with very few single cells being neither positive for MAP-2 nor for GFAP (,1%).

2.2. Glutamate excitotoxicity Five-day-old cultures were incubated with 125, 250 or 500 mM glutamate freshly made from a 5 mM stock solution for 3, 12 or 24 h in the above used medium, however, without serum and B27-Supplement to avoid the influence of antioxidants. Cells were daily assessed by phase-contrast microscopy. For immunocytochemistry coverslips were fixed for 15 min with 4% paraformaldehyde at 378C in phosphate-buffered sodium chloride.

2.3. Immunocytochemistry and DNA fragmentation After fixation, cells were incubated with an anti-MAP-2 antibody (Roche, mouse monoclonal, 1:10 000) to label neurons or with an anti-GFAP antibody (Roche, mouse monoclonal, 1:4000) to identify astrocytes or anti-active caspase-3 antibody (rabbit polyclonal, 1:8000). The antibody against the p17 fragment of cleaved and activated caspase-3 was raised by Dr. Anu Srinivasan (Idun Pharmaceuticals, USA) and has been characterized previously [1,34,47]. For double labelling, MAP-2 staining followed either active caspase-3 or TUNEL immunocytochemistry.

S. Brecht et al. / Molecular Brain Research 94 (2001) 25 – 34

Coverslips were washed in 0.1% Triton X-100 in phosphate-buffered sodium chloride for 2 min, incubated with the primary antibody overnight and treated with biotinylated second antibody and avidin–peroxidase complex (Vector Kit, Burlingame, USA). As chromogen for visualization, we used either 3,3-diaminobenzidine (Sigma) resulting in a brown signal or for double labelling Vector SG (Vector) resulting in a blue labelling. For controls we stained only the secondary antibodies against rabbit or mouse IgG under omission of the primary antibody; this procedure did not result in a labelling signal. For detection of DNA fragmentation, the TUNEL [18] components from Roche were used according to manufacturers instructions with terminal dUTP-transferase (TdT) in a final concentration of 200 units / ml and visualization by peroxidase reaction with Vector SG chromogen. For double labelling TUNEL was stained prior to MAP-2 or GFAP but after active caspase-3 immunocytochemistry. For control of TUNEL specificity, cells were either treated prior with DNAse (positive control) or the enzyme TdT was omitted (negative control). The results from controls confirmed the specificity of TUNEL staining.

27

and over 70% neurons as identified by MAP-2. Only very few single cells were immunonegative for both markers. The comparison of MAP-2-labelled neurons with the phase-contrast pattern of unlabelled cells indicated no loss of cells during the immunocytochemical procedure. Five days after plating the cells, a dense network of neurons identified by MAP-2 immunocytochemistry was settling on or in close proximity of scattered astroglia. The density of these neurons (6467 per vision field) was set as 100%. After 3 h of 125 mM glutamate stimulation the number decreased to 89% and after 12 h to significant 75%. At the end of the observation period after 24 h, only 55% MAP-2-positive neurons were observed (Fig. 1). These surviving neurons, however, displayed morphological signs of stress after 12 and 24 h such as shorter neurites and a rough cell membrane in comparison with untreated cultures (Fig. 2). Starting 3 h after glutamate exposure, the neurons lost their MAP-2 immunolabelling undergoing cell death. Some neurons displayed a faint MAP-2 immunoreactivity (IR) which persisted in the distal neurites whereas the cell body was almost immunonegative for MAP-2 IR.

2.4. Cell counting and statistical analysis 3.2. Dose dependency of glutamate effects Cells were counted in the phase contrast microscope (Olympus) and marked areas were identified after immunocytochemistry to quantify a putative cell loss during the fixation and staining procedure. For final immunocytochemical counting the mean and the standard error of the mean were calculated for 10 representative light-microscopic areas per coverslip (Leica DHS, magnification, 3400). Evaluation of the levels of significance was performed by Students t-test with P, 0.05. Each experiment per timepoint and per glutamate concentration was repeated 4 times. Only distinct immunopositive signals were counted. Neurons with intact continuous neurites / axons and a round / oval smooth cell body were regarded as being alive and mainly intact, whereas cells with fragmentation of neurites / axons and a rough / fragmented / vacuolated cell body were considered as dying / dead. Active caspase-3 was counted as proportion of the neurons with MAP-2 IR at the investigated timepoints (basal, 3, 12 and 24 h), whereas TUNEL IR was counted in the proportion of the initial total cell number in untreated cultures. Pictures were achieved with a JVC monochip analogue video camera and Leica Qwin software.

The total number of surviving neurons was dose dependent following application of 125, 250 and 500 mM glutamate. Twenty-four hours after 500 mM glutamate, 2164 MAP-2-positive neurons (33%) were found with the majority in close proximity to astroglial cells and display signs of severe cell stress such as short ending neurites and swollen, vacuolated perikarya. The decline in the total number of surviving MAP-2 neurons after 24 h is quantified in Fig. 2D. The counted numbers of active caspase-3labelled neurons after 24 h were 36% following 125 mM glutamate, 53% following 250 mM and 32% after 500 mM glutamate exposure (Fig. 3B). All morphological stages of neurodegeneration in active caspase-3-labelled neurons

3. Results

3.1. MAP-2 labelling and cell loss by glutamate stimulation Our culture conditions resulted in a mixed cell population with less then 30% of GFAP-positive astroglia cells

Fig. 1. Loss of neurons as indicated by the decrease of MAP-2 immunoreactivity in mixed hippocampal neuronal culture following 125 mM glutamate treatment (P,0.05, Students t-test).

28

S. Brecht et al. / Molecular Brain Research 94 (2001) 25 – 34

Fig. 2. Morphological development of cell stress in MAP-2-positive neurons (magnification, 3400). (A) Untreated cultures. (B) Neurons morphology with short dendrites and vacuolated cytoplasm following 24 h of 500 mM glutamate treatment. (C) Some neurons (white arrow) in close proximity to astrocytes (black arrow) tolerate the glutamate treatment. (D) Quantification of MAP-2 loss in cell culture after 24 h of different doses of glutamate treatment. The average number of neurons in untreated cultures was set as 100% (P,0.05, Students t-test).

Fig. 3. (A) Proportion of neurons with active caspase-3 immunoreactivity over time following 125 mM glutamate stimulation. After 24 h the maximum of 36% was observed. (B) Active caspase-3 immunoreactivity raises dose-dependent 24 h after glutamate treatment. The maximum proportion is seen following 250 mM glutamate (P,0.05, Students t-test).

S. Brecht et al. / Molecular Brain Research 94 (2001) 25 – 34

(for morphology see next section) were constant in their relative proportion after all three glutamate doses.

3.3. Caspase-3 activation following stimulation with 125 m M glutamate In untreated cultures, 8% of neurons were labelled for active caspase-3. The labelled cells displayed all morphological stages of degeneration. We found cells with minor symptoms of cell stress such as membrane irregularities and small vacuoles but otherwise intact neuronal morphology (9%), neurons with degenerating neurites (14%) and cell clumps with the appearance of cell debris (77%) (Fig. 4A–D). The coincidence of these morphological stages indicates a dynamic progress of neuronal degeneration in cultures even in the absence of intentional stimulation. Scattered caspase-3 activation was always

29

restricted to neurons and was not accumulated in coherent clusters within the culture. In consequence, neurons with active caspase-3 IR were located in proximity to intact MAP-2 neurons (Fig. 4A,B). In rather healthy active caspase-3 neurons, the labelling was found exclusively in the cytoplasm with immunonegative nuclei, whereas more damaged neurons were completely labelled. Three hours following 125 mM of glutamate, the proportion of caspase3-labelled neurons compared to total MAP-2 neurons did not significantly change compared to controls (each 8% of total neurons with 562 labelled neurons per vision field). Both, the slight increase of the total number and the decrease in MAP-2 IR raised the proportion of active caspase-3 neurons after 12 h to significant 24%, and to 36% after 24 h. The total numbers of active caspase-3 neurons did not increase substantially (up to 1262 neurons per vision field after 24 h). Over the time, the proportion of

Fig. 4. Different stages of neurodegeneration observed by labelling of active caspase-3 immunoreactivity in neurons stimulated for 24 h with 125 mM glutamate (magnification, 3400). (A) Beginning degeneration in isolated neurons surrounded by of unlabelled MAP-2-positive neurons. The nucleus is immunonegative. (B) Dendrites begin to degenerate and the cell body is homogeneously labelled for active caspase-3. (C) Dendrites have almost disappeared and the remaining perikaryon forms a clump. (D) A final active caspase-3-labelled cell clump remains from a previous neuron. All these stages can be observed in parallel in treated and untreated cultures.

30

S. Brecht et al. / Molecular Brain Research 94 (2001) 25 – 34

active caspase-3 neurons found in different morphological stages of neurodegeneration did not change. Again, severely damaged neurons labelled by active caspase-3 were found in the closest neighbourhood of intact looking neurons negative for active caspase-3 IR. In contrast to isolated active caspase-3-labelled neurons, the MAP-2labelled neurons maintained their neurite network.

3.4. TUNEL labelling In our hands, TUNEL staining exclusively labels round, segmented nuclei with occasional blebbing. TUNEL signals were distributed in single non-coherent cells all over the coverslip. Like with active caspase-3 labelling, TUNEL also appears in 5% of cells (363 signals with TUNEL IR) in untreated cultures indicating that dissociation and cultivation provoke DNA fragmentation in neurons. In contrast to active caspase-3 labelling, TUNEL-stained nuclei cannot be clearly identified as neurons because

numerous structures with TUNEL-positive signals did not reveal an identifiable cell body. After glutamate stimulation, the total number of TUNEL nuclei increased after 3 h to 7.8% (562 of all former 7168 cells per field), after 12 h to significant 18% (1263 cells) up to 38% (2765 cells per vision field after 24 h). We could differentiate between dark and light blue-labelled nuclei. Light-blue nuclei were rare and had a distinct cytoplasm left that could be identified by GFAP or MAP-2 labelling and both cell types appeared severely stressed and were dedicated to die. The vast majority of TUNEL nuclei (94%) appeared as dark dots without any identifiable cytoplasm (Fig. 5A), suggesting that intense TUNEL labelling detects only nucleic acid and / or chromatin debris from dead cells. Counting TUNEL-IR as proportion of the total MAP2IR at each investigated timepoint (which decreases to 55% 24 h after glutamate), as was done for active caspase-3, the level increased to even 75% after 24 h and exceeded the number of active caspase-3-labelled neurons.

Fig. 5. (A) The vast majority of TUNEL-positive DNA are seen as dark round dots without obvious structures of a cell body (arrow). In close proximity MAP-2-positive neurons are seen without TUNEL signal. (B) Less than 2% of all TUNEL-stained nuclei show a co-localization with neurons labelled by MAP-2. (C) Less than 2% of all TUNEL signals are co-localized with active caspase-3 in advanced degeneration. (D) Few GFAP-positive astrocytes here with vacuolated cytoplasm can be labelled with TUNEL following glutamate treatment (magnification, 3400 in A and 31000 in B–D).

S. Brecht et al. / Molecular Brain Research 94 (2001) 25 – 34

3.5. Co-localization studies ( TUNEL, active caspase-3, MAP-2 and GFAP) Co-localization was investigated by double labelling of active caspase-3 with MAP-2, the astroglial marker GFAP and TUNEL. Additionally, TUNEL was double labelled with MAP-2 or GFAP. Active caspase-3 IR was exclusively seen in neurons but not in astrocytes. Neurons positive for active caspase-3 tend to loose the immunosignal for MAP-2 already at very early stages, when the neuronal phenotype can be easily identified by morphology. Surprisingly, TUNEL and active caspase-3, both parameters commonly used to detect apoptotic cell death, are poorly co-localized (Fig. 5A). Less than 2% of all TUNEL signals (neurons and astrocytes) were found in MAP-2-labelled neurons with signs of severe stress (Fig. 5B). The ratio of co-localization of active caspase-3 IR with TUNEL neurons was less than 2%. Some neurons showing co-localization of active caspase and TUNEL staining displayed also morphological signs of severe stress such as edge standing nucleus and vacuoles indicating an advanced stage of cell death (Fig. 5C). The determination of TUNEL colocalization was complicated by the advanced cell degeneration. In contrast to active caspase-3, TUNEL-positive DNA fragmentation can also be detected in GFAP-positive astroglia (Fig. 5D) and this pool of apoptotic TUNEL nuclei in GFAP astroglia contributed to 1% of the total TUNEL count. However, TUNEL IR cells in the absence of a characteristic cytoplasm could not be determined as neurons or astrocytes. Therefore, the increase in TUNEL-IR was measured in comparison to the basal 7168 cells per vision field in untreated cultures.

4. Discussion In this study, we have analyzed the co-incidence of internucleosomal DNA fragmentation visualized by TUNEL and caspase-3 activation in mixed hippocampal cultures of neurons and astrocytes stimulated with excitotoxic glutamate. Scattered activated caspase-3 IR was present in 8% of neurons in untreated cultures and this number increased to 36% after 24 h of 125 mM glutamate stimulation. TUNEL was found in 5% of neurons or astrocytes in untreated cultures and raised to 38% after 24 h of 125 mM glutamate stimulation. Surprisingly, the co-localization of TUNEL and active caspase-3 is very poor indicating two apoptotic mechanisms different in time and / or cell population or two different events of neuronal / cellular degeneration.

4.1. Caspase-3 activation Our results confirm previous findings that caspase-3

31

activation is involved in cell death in hippocampal mixed neuronal cultures treated with glutamate [14], where in the same culture active caspase-3 was absent in astrocytes. The issue of exclusive activation of caspase-3 in neurons [16,20] or its additional presence in astroglia or microglia [4,30,38,53] is still controversially discussed. Our postpartal hippocampal culture consists almost exclusively of neurons and astroglia. We consider the caspase-3 activation in neurons following glutamate treatment to be a specific neuronal reaction. The specificity of the active caspase-3 antibody has been proven elsewhere [1,34,47]. In untreated cell cultures, 8% of all neurons displayed active caspase-3IR, indicating a basal rate of degeneration. This proportion increased in a time- and dose-dependent manner following glutamate treatment with a maximum after 24 h, the end of our observation period. Our observed time range is within a similar range as described after cerebral ischemia in vivo by an activity assay and by immunohistochemistry [34]. The intracellular localization is another issue of controversial discussion. Over time, the entity of the cell dissolves and attribution of immunocytochemical signals to a specific cellular compartment becomes rather difficult. We found a clear dynamic progress in intracellular localization: cytoplasmatic-IR with immunonegative nuclei at early stages and whole cell labelling at later stages, when neurons have already lost their MAP-2 IR and neurites have degenerated, confirming previous data [51,56]. Once the cell entity is lost, the remaining degenerate proteins and the debris fall apart with decline of immunoreactive structures. This breakdown explains the moderate increase in the total number of active caspase-3-IR in the neurons after glutamate stimulation. The rate of protein degeneration could also explain the stable proportions of the different morphological stages during caspase-3 activation and cell death. Superfusion of the culture dishes by glutamate should affect all neurons. Importantly, only single neurons, but not clusters or coherent populations, undergo cell death with presence of active caspase-3 involvement. Several reasons might account for this finding. (1) The hippocampal neurons display an individual vulnerability: sensitive CA1 neurons versus more resistant CA2 / 4 neurons or dentate gyrus neurons; these subpopulations might also react in a different manner under cell culture conditions [3]. (2) Neurons in close proximity to astrocytes tolerate glutamate better compared to neurons without astrocytic company. After 500 mM glutamate the remaining neurons were almost exclusively in close relation to astrocytes. This indicates a protective mechanism by astrocyte presence and was also described in other experimental setups [41]. (3) Independent of active caspase-3, other proteases like calpain [42,40,45] or cathepsin D [22,32] may be responsible for the degradation of neurons stimulated by glutamate. This idea is supported by our finding that the extent of MAP-2 loss is higher than the increase of active caspase-3-IR. With 500 mM glutamate, the more rapid

32

S. Brecht et al. / Molecular Brain Research 94 (2001) 25 – 34

onset of neuronal death might be insufficient to trigger caspase-3 activation. In our studies, the focus was put on the ‘low dosage’ of 125 mM glutamate since a protracted cell degeneration allows the analysis of active processes in contrast to accelerated onset of neuronal death as seen with the 500 mM glutamate treatment.

4.2. TUNEL DNA fragmentation can be marked by TUNEL staining and is commonly used to detect apoptotic cell death, even if its specificity was recently questioned [8] since TUNEL was found in neurons with necrotic morphology following kainic acid treatment [17] and cortical contusion [35]. In our hands, TUNEL staining seems to be specific for cell death since intact MAP-2-labelled neurons are immunonegative. We also observed some TUNEL nuclei with the characteristic blebbings as the typical morphological feature for apoptotic cell death in vitro and in vivo. As with active caspase-3, the non-coherent TUNEL signals were distributed randomly. TUNEL is present in neurons [34,21], microglia [19,28,50] and astrocytes [26]. We can detect TUNEL-stained nuclei in neurons and in very few astrocytes. The co-localization of TUNEL with MAP-2 or GFAP, however, is very poor with less than 2% of all TUNEL nuclei being detected in MAP-2- or GFAP-labelled cells even under the use of high concentrations of terminal dUTP transferase (250 units / ml) [25]. Astrocytes are considered to be more resistant to glutamate toxicity [2,37], but we could also show that a few distinct GFAPlabelled astrocytes displayed severe cell stress and contributed to the pool of total TUNEL-positive cells. In fact, most TUNEL-labelled nuclei were probably seen as debris of neurons (because more than 70% of all cells in culture are neurons) almost without any cytoplasm left (even in phase contrast microscopy) or in severely damaged cells in the absence of MAP-2 or GFAP-IR. TUNEL was seen in 5% of untreated cells, again indicating the vulnerability of postnatal neurons in vitro. Since the vast majority of TUNEL signals could not be differentiated in the original cell type, the numbers were calculated in relation to all cells of untreated cultures. A slight increase in TUNEL staining following glutamate stimulation to 7.8% is already obvious after 3 h with 125 mM glutamate. In our experiment, the total number of TUNEL signals exceeded that of active caspase-3 by far from 12 h following glutamate treatment. Reasons for that might be that DNA fragmentation is also caspase-3 independent and that the chromatin leftover has probably a longer half-life than an active protein once the cell is dedicated to die or neurons simply die by necrosis.

4.3. Active caspase-3 and TUNEL Both, caspase-3 activation [33,16] and DNA fragmentation [16,55] contribute to cell death following excitotoxici-

ty. A causal relationship of both parameters in the progress of active cell death has been suggested [54]. We see in our experiments that the presence of active caspase-3 and TUNEL in the same neuron / cell is almost excluded. Only less then 2% of all active caspase-3 neurons were also TUNEL-positive and these neurons showed morphological signs of severe stress. We observe similar findings in vivo in kainic acid-treated mice, where active caspase 3 flanks the hippocampal area of TUNEL-positive DNA fragmentation with a poor co-localization. Reasons for this finding are either a very narrow time window of co-localization or two independent mechanisms for cell death. The latter mechanism is supported by the parallel increase in TUNEL and active caspase-3 following glutamate treatment without increasing the proportion of co-localization. The number of TUNEL exceeds the number of active caspase3-labelled cells, because TUNEL but not active caspase-3 can also be found in astroglia (in contrast to other investigators from in vivo experiments [48]). One reason for the long-lasting TUNEL-positive chromatin signal might be the missing macrophagic activity in our mixed cell culture to phagocytose chromatin debris. From the co-localization of TUNEL and active caspase3 we cannot conclude for sure which events come first on the way to glutamate triggered neuronal death — nevertheless the cell morphology suggests that caspase-3 activation starts in an intact neuron, whereas TUNEL stains irreversibly damaged neurons later on. Finally, we have to point to other models such as cerebral ischemia in vivo that reveal much higher levels of co-localization in another time range [34] reflecting the experimental particularity.

References [1] R.C. Armstrong, T.J. Aja, K.D. Hoang, S. Gaur, X. Bai, E.S. Alnemri, G. Litwack, D.S. Karanewsky, L.C. Fritz, K.J. Tomaselli, Activation of the CED3 / ICE-related protease CPP32 in cerebellar granule neurons undergoing apoptosis but not necrosis, J. Neurosci. 17 (1997) 553–562. [2] E. Aronica, E.A. van Vliet, O.A. Mayboroda, D. Troost, F.H. da Silva, J.A. Gorter, Upregulation of metabotropic glutamate receptor subtype mGluR3 and mGluR5 in reactive astrocytes in a rat model of mesial temporal lobe epilepsy, Eur. J. Neurosci. 12 (2000) 2333–2344. [3] A.J. Becker, F. Gillardon, I. Blumcke, D. Langendorfer, H. Beck, O.D. Wiestler, Differential regulation of apoptosis-related genes in resistant and vulnerable subfields of the rat epileptic hippocampus, Brain Res. Mol. Brain Res. 67 (1999) 172–176. [4] R. Beer, G. Franz, A. Srinivasan, R.L. Hayes, B.R. Pike, J.K. Newcomb, X. Zhao, E. Schmutzhard, W. Poewe, A. Kampfl, Temporal profile and cell subtype distribution of activated caspase-3 following experimental traumatic brain injury, J. Neurochem. 75 (2000) 1264–1273. [5] D. Borras, M. Pumarola, I. Ferrer, Neuronal nuclear DNA fragmentation in the aged canine brain: apoptosis or nuclear DNA fragility?, Acta Neuropathol. (Berlin) 99 (2000) 402–408. [6] P. Chaudhary, F. Ahmed, P. Quebada, S.C. Sharma, Caspase inhibitors block the retinal ganglion cell death following optic nerve transection, Brain Res. Mol. Brain Res. 67 (1999) 36–45.

S. Brecht et al. / Molecular Brain Research 94 (2001) 25 – 34 [7] Y. Cheng, M. Deshmukh, A. D’Costa, J.A. Demaro, J.M. Gidday, A. Shah, Y. Sun, M.F. Jacquin, E.M. Johnson, D.M. Holtzman, Caspase inhibitor affords neuroprotection with delayed administration in a rat model of neonatal hypoxic–ischemic brain injury, J. Clin. Invest. 101 (1998) 1992–1999. [8] J. Cui, E.H. Holmes, T.G. Greene, P.K. Liu, Oxidative DNA damage precedes DNA fragmentation after experimental stroke in rat brain, FASEB J. 14 (2000) 955–967. [9] A.J. Darmon, T.J. Ley, D.W. Nicholson, R.C. Bleackley, Cleavage of CPP32 by granzyme B represents a critical role for granzyme B in the induction of target cell DNA fragmentation, J. Biol. Chem. 271 (1996) 21709–21712. [10] A.J. Darmon, D.W. Nicholson, R.C. Bleackley, Activation of the apoptotic protease CPP32 by cytotoxic T-cell-derived granzyme B, Nature 377 (1995) 446–448. [11] A. Dityatev, G. Dityateva, M. Schachner, Synaptic strength as a function of post- versus presynaptic expression of the neural cell adhesion molecule NCAM, Neuron 26 (2000) 207–217. [12] R.C. Dodel, Y. Du, K.R. Bales, Z. Ling, P.M. Carvey, S.M. Paul, Caspase-3-like proteases and 6-hydroxydopamine induced neuronal cell death, Brain Res. Mol. Brain Res. 64 (1999) 141–148. [13] R.C. Dodel, Y. Du, K.R. Bales, Z.D. Ling, P.M. Carvey, S.M. Paul, Peptide inhibitors of caspase-3-like proteases attenuate 1-methyl-4phenylpyridinum-induced toxicity of cultured fetal rat mesencephalic dopamine neurons, Neuroscience 86 (1998) 701–707. [14] Y. Du, K.R. Bales, R.C. Dodel, E. Hamilton-Byrd, J.W. Horn, D.L. Czilli, L.K. Simmons, B. Ni, S.M. Paul, Activation of a caspase 3-related cysteine protease is required for glutamate-mediated apoptosis of cultured cerebellar granule neurons, Proc. Natl. Acad. Sci. USA 94 (1997) 11657–11662. [15] M. Endres, S. Namura, M. Shimizu-Sasamata, C. Waeber, L. Zhang, T. Gomez-Isla, B.T. Hyman, M.A. Moskowitz, Attenuation of delayed neuronal death after mild focal ischemia in mice by inhibition of the caspase family, J. Cereb. Blood Flow Metab. 18 (1998) 238–247. [16] C.J. Faherty, S. Xanthoudakis, R.J. Smeyne, Caspase-3-dependent neuronal death in the hippocampus following kainic acid treatment, Brain Res. Mol. Brain Res. 70 (1999) 159–163. [17] D.G. Fujikawa, S.S. Shinmei, B. Cai, Kainic acid-induced seizures produce necrotic, not apoptotic, neurons with internucleosomal DNA cleavage: implications for programmed cell death mechanisms, Neuroscience 98 (2000) 41–53. [18] Y. Gavrieli, Y. Sherman, S.A. Ben-Sasson, Identification of programmed cell death in situ via specific labeling of nuclear DNA fragmentation, J. Cell Biol. 119 (1992) 493–501. [19] J. Gehrmann, R.B. Banati, Microglial turnover in the injured CNS: activated microglia undergo delayed DNA fragmentation following peripheral nerve injury, J. Neuropathol. Exp. Neurol. 54 (1995) 680–688. [20] F. Gillardon, B. Bottiger, B. Schmitz, M. Zimmermann, K.A. Hossmann, Activation of CPP-32 protease in hippocampal neurons following ischemia and epilepsy, Brain Res. Mol. Brain Res. 50 (1997) 16–22. [21] D.M. Hermann, G. Mies, R. Hata, K.A. Hossmann, Microglial and astrocytic reactions prior to onset of thalamic cell death after traumatic lesion of the rat sensorimotor cortex, Acta Neuropathol. (Berlin) 99 (2000) 147–153. [22] M. Hetman, R.K. Filipkowski, W. Domagala, L. Kaczmarek, Elevated cathepsin D expression in kainate-evoked rat brain neurodegeneration, Exp. Neurol. 136 (1995) 53–63. [23] T. Himi, Y. Ishizaki, S. Murota, A caspase inhibitor blocks ischaemia-induced delayed neuronal death in the gerbil, Eur. J. Neurosci. 10 (1998) 777–781. [24] B.S. Jeon, N.G. Kholodilov, T.F. Oo, S.Y. Kim, K.J. Tomaselli, A. Srinivasan, L. Stefanis, R.E. Burke, Activation of caspase-3 in developmental models of programmed cell death in neurons of the substantia nigra, J. Neurochem. 73 (1999) 322–333.

33

[25] L.L. Jones, R.B. Banati, M.B. Graeber, L. Bonfanti, G. Raivich, G.W. Kreutzberg, Population control of microglia: does apoptosis play a role?, J. Neurocytol. 26 (1997) 755–770. [26] T. Kohji, N. Tanuma, Y. Aikawa, Y. Kawazoe, Y. Suzuki, K. Kohyama, Y. Matsumoto, Interaction between apoptotic cells and reactive brain cells in the central nervous system of rats with autoimmune encephalomyelitis, J. Neuroimmunol 82 (1998) 168– 174. [27] S. Kumar, The apoptotic cysteine protease CPP32, Int. J. Biochem. Cell Biol. 29 (1997) 393–396. [28] Y. Li, P.M. Field, G. Raisman, Death of oligodendrocytes and microglial phagocytosis of myelin precede immigration of schwann cells into the spinal cord [In Process Citation], J. Neurocytol. 28 (1999) 417–427. [29] P. Lipton, Ischemic cell death in brain neurons, Physiol. Rev. 79 (1999) 1431–1568. [30] S. Love, R. Barber, G.K. Wilcock, Neuronal death in brain infarcts in man, Neuropathol. Appl. Neurobiol. 26 (2000) 55–66. [31] G. MacDonald, L. Shi, C. Van de Velde, J. Lieberman, A.H. Greenberg, Mitochondria-dependent and -independent regulation of Granzyme B-induced apoptosis, J. Exp. Med. 189 (1999) 131–144. [32] D. Mantle, G. Falkous, S. Ishiura, R.H. Perry, E.K. Perry, Comparison of cathepsin protease activities in brain tissue from normal cases and cases with Alzheimer’s disease, Lewy body dementia, Parkinson’s disease and Huntington’s disease, J. Neurol. Sci. 131 (1995) 65–70. [33] N. Marks, M.J. Berg, A. Guidotti, M. Saito, Activation of caspase-3 and apoptosis in cerebellar granule cells, J. Neurosci. Res. 52 (1998) 334–341. [34] S. Namura, J. Zhu, K. Fink, M. Endres, A. Srinivasan, K.J. Tomaselli, J. Yuan, M.A. Moskowitz, Activation and cleavage of caspase-3 in apoptosis induced by experimental cerebral ischemia, J. Neurosci. 18 (1998) 3659–3668. [35] J.K. Newcomb, X. Zhao, B.R. Pike, R.L. Hayes, Temporal profile of apoptotic-like changes in neurons and astrocytes following controlled cortical impact injury in the rat, Exp. Neurol. 158 (1999) 76–88. [36] B. Ni, X. Wu, Y. Su, D. Stephenson, E.B. Smalstig, J. Clemens, S.M. Paul, Transient global forebrain ischemia induces a prolonged expression of the caspase-3 mRNA in rat hippocampal CA1 pyramidal neurons, J. Cereb. Blood Flow Metab. 18 (1998) 248– 256. [37] F. Nicoletti, V. Bruno, M.V. Catania, G. Battaglia, A. Copani, G. Barbagallo, V. Cena, J. Sanchez-Prieto, P.F. Spano, M. Pizzi, GroupI metabotropic glutamate receptors: hypotheses to explain their dual role in neurotoxicity and neuroprotection, Neuropharmacology 38 (1999) 1477–1484. [38] T. Nishioku, K. Miyamoto, K. Murao, C. Hara, K. Yamamoto, H. Nakanishi, Involvement of caspase 3-like protease in methylmercury-induced apoptosis of primary cultured rat cerebral microglia, Brain Res. 871 (2000) 160–164. [39] K. Nishizaki, T. Yoshino, Y. Orita, S. Nomiya, Y. Masuda, TUNEL staining of inner ear structures may reflect autolysis, not apoptosis, Hear. Res. 130 (1999) 131–136. [40] R.A. Nixon, K.I. Saito, F. Grynspan, W.R. Griffin, S. Katayama, T. Honda, P.S. Mohan, T.B. Shea, M. Beermann, Calcium-activated neutral proteinase (calpain) system in aging and Alzheimer’s disease, Ann. NY Acad. Sci. 747 (1994) 77–91. [41] M. Ohgoh, M. Kimura, H. Ogura, K. Katayama, Y. Nishizawa, Apoptotic cell death of cultured cerebral cortical neurons induced by withdrawal of astroglial trophic support, Exp. Neurol. 149 (1998) 51–63. [42] W.Y. Ong, M.Q. Ren, J. Makjanic, T.M. Lim, F. Watt, A nuclear microscopic study of elemental changes in the rat hippocampus after kainate-induced neuronal injury, J. Neurochem. 72 (1999) 1574– 1579. [43] C. Sasaki, H. Kitagawa, W.R. Zhang, H. Warita, K. Sakai, K. Abe,

34

[44]

[45]

[46]

[47]

[48]

[49]

[50]

S. Brecht et al. / Molecular Brain Research 94 (2001) 25 – 34 Temporal profile of cytochrome c and caspase-3 immunoreactivities and TUNEL staining after permanent middle cerebral artery occlusion in rats, Neurol. Res 22 (2000) 223–228. P. Schotte, W. Declercq, S. Van Huffel, P. Vandenabeele, R. Beyaert, Non-specific effects of methyl ketone peptide inhibitors of caspases, FEBS Lett. 442 (1999) 117–121. R. Siman, D. Bozyczko-Coyne, M.J. Savage, J.M. Roberts-Lewis, The calcium-activated protease calpain I and ischemia-induced neurodegeneration, Adv. Neurol. 71 (1996) 167–174. J.E. Springer, R.D. Azbill, P.E. Knapp, Activation of the caspase-3 apoptotic cascade in traumatic spinal cord, Nat. Med. 5 (1999) 943–946. A. Srinivasan, K.A. Roth, R.O. Sayers, K.S. Shindler, A.M. Wong, L.C. Fritz, K.J. Tomaselli, In situ immunodetection of activated caspase-3 in apoptotic neurons in the developing nervous system, Cell Death. Differ. 5 (1998) 1004–1016. J.H. Su, K.E. Nichol, T. Sitch, P. Sheu, C. Chubb, B.L. Miller, K.J. Tomaselli, R.C. Kim, C.W. Cotman, DNA damage and activated caspase-3 expression in neurons and astrocytes: evidence for apoptosis in frontotemporal dementia, Exp. Neurol. 163 (2000) 9–19. S.A. Susin, E. Daugas, L. Ravagnan, K. Samejima, N. Zamzami, M. Loeffler, P. Costantini, K.F. Ferri, T. Irinopoulou, M.C. Prevost, G. Brothers, T.W. Mak, J. Penninger, W.C. Earnshaw, G. Kroemer, Two distinct pathways leading to nuclear apoptosis, J. Exp. Med. 192 (2000) 571–580. N. Takai, H. Nakanishi, K. Tanabe, T. Nishioku, T. Sugiyama, M. Fujiwara, K. Yamamoto, Involvement of caspase-like proteinases in

[51]

[52]

[53]

[54]

[55]

[56]

apoptosis of neuronal PC12 cells and primary cultured microglia induced by 6-hydroxydopamine, J. Neurosci. Res 54 (1998) 214– 222. L. Tenneti, S.A. Lipton, Involvement of activated caspase-3-like proteases in N-methyl-D-aspartate-induced apoptosis in cerebrocortical neurons, J. Neurochem. 74 (2000) 134–142. K. Urase, E. Fujita, Y. Miho, Y. Kouroku, T. Mukasa, Y. Yagi, M.Y. Momoi, T. Momoi, Detection of activated caspase-3 (CPP32) in the vertebrate nervous system during development by a cleavage sitedirected antiserum, Brain Res. Dev. Brain Res. 111 (1998) 77–87. J.J. Velier, J.A. Ellison, K.K. Kikly, P.A. Spera, F.C. Barone, G.Z. Feuerstein, Caspase-8 and caspase-3 are expressed by different populations of cortical neurons undergoing delayed cell death after focal stroke in the rat, J. Neurosci. 19 (1999) 5932–5941. B.B. Wolf, M. Schuler, F. Echeverri, D.R. Green, Caspase-3 is the primary activator of apoptotic DNA fragmentation via DNA fragmentation factor-45 / inhibitor of caspase-activated DNase inactivation, J. Biol. Chem. 274 (1999) 30651–30656. D.D. Yang, C.Y. Kuan, A.J. Whitmarsh, M. Rincon, T.S. Zheng, R.J. Davis, P. Rakic, R.A. Flavell, Absence of excitotoxicity-induced apoptosis in the hippocampus of mice lacking the Jnk3 gene, Nature 389 (1997) 865–870. C. Zhu, X. Wang, H. Hagberg, K. Blomgren, Correlation between caspase-3 activation and three different markers of DNA damage in neonatal cerebral hypoxia-ischemia, J. Neurochem. 75 (2000) 819– 829.