Vulnerability to excitotoxic stimuli of cultured rat hippocampal neurons containing the calcium-binding proteins calretinin and calbindin D28K

BRAIN RESEARCH ELSEVIER

Brain Research 648 (1994) 109-120

Research Report

Vulnerability to excitotoxic stimuli of cultured rat hippocampal neurons containing the calcium-binding proteins calretinin and calbindin D28K V. M6ckel 1, G. Fischer * Pharma Division, Preclinical Research, F. Hoffmann-La Roche, PRPN, 4002 Basel, Switzerland Accepted 1 March 1994

Abstract

Rat embryonic hippocampal neurons cultured on astrocyte feeder-layers were sensitive to different excitotoxic stimuli after 10-12 DIV. Almost all neurons (~ 95%) died within 20 h after a transient exposure for 10 min to 50 p,M glutamate, a continuous exposure to either 25 tzM NMDA or 250 /~M kainate or after a 15-min deprivation of glucose and oxygen. Dizocilpine at 10/xM protected neurons against the glutamate- and NMDA-mediated toxicity as well as against 30 min glucose and oxygen deprivation. However, it failed to protect against kainate toxicity and prolonged glucose/oxygen deprivation (60 min). An additional treatment with CNQX (100 ~.M) protected neurons even under the latter two conditions. This indicates that the vast majority of neurons was sensitive to different excitotoxic stimuli acting through different types of glutamate receptors leading to calcium overload of the cells which might be the common denominator of triggering cell death under these conditions. Expression of calcium-binding proteins, such as calbindin D28K or calretinin, might increase the intracellular calcium buffer capacity of neurons, thus, rendering them more resistant to calcium overload. Therefore, we analysed whether neurons expressing these calcium-binding proteins would survive these toxic stimuli. Indeed, a small population of the neurons (3-5%) survived, including a subpopulation of calretinin-positive but not calbindin D28K-positive neurons. This implies that the expression of calcium-binding proteins per se does not render neurons more resistant towards these excitotoxic stimuli. Moreover, most of the surviving calretinin-positive neurons showed morphological damage as indicated by loss of neurites. When cytotoxicity due to calcium overload was induced by an exposure of the cells to the calcium ionophore 4-bromo-A23187 rather than by activation of glutamate receptors, calretinin-positive cells were found not to be significantly more resistant than the vast majority of neurons. This may indicate that the lower sensitivity of a subpopulation of calretinin-positive neurons to excitotoxic stimuli may be due to a lower expression of glutamate receptors.

Key words: Excitotoxicity; Calbindin D28K; Calretinin; Hippocampal cell culture

1. Introduction

Neuronal degeneration in stroke, brain trauma and severe epileptic seizures seems to be triggered in part by excitotoxic overstimulation leading to overload of the cells with calcium (for a review, see Ref. 1). Depending on the stimulation conditions, different entry routes for calcium may dominate. Excitotoxic overstimulation can be mimicked in primary neuronal cell cultures by exposure of the cells to different types of glutamate receptor agonists. When excitotoxicity is induced by brief exposure to high concentrations of

* Corresponding author. Fax: (41) (61) 688-1720. 1 Present address: G6decke, Freiburg, Germany. 0006-8993/94/$07.00 © 1994 Elsevier Science B.V. All rights reserved SSDI 0 0 0 6 - 8 9 9 3 ( 9 4 ) 0 0 2 7 2 - E

either glutamate or N M D A , neuronal degeneration can be completely prevented by N M D A receptor antagonists [3] but not by blockers of voltage-operated calcium channels [38], indicating that N M D A receptors are the main entry route for calcium under these conditions, and, recently, a close correlation between calcium entry via N M D A receptors and cell death could be shown [37]. When excitotoxicity is induced by prolonged incubation with low concentrations of AMPA, kainate or quinolinate, the latter being a weak agonist of the N M D A receptor [20], neuronal degeneration is not only blocked by inhibition of the respective glutamate receptors but also attenuated by blockers of voltage-operated calcium channels, indicating that calcium entry via calcium channels gains importance under these conditions [38]. In addition, glutamate neuro-

110

V. M6ckel, G. Fischer~Brain Research 648 (1994) 109-120

toxicity may be attenuated also by dantrolene, an inhibitor of intracellular calcium release [9]. Glutamate-mediated neurotoxicity also dominates neuronal degeneration in cortical cell cultures induced by combined glucose/oxygen deprivation since N M D A receptor antagonists protect neurons against short-term deprivation (up to 30 min) and, when combined with A M P A / k a i n a t e receptor antagonists, even against prolonged deprivations of up to 100 min [13,19]. At least for short-term deprivation, a close correlation between calcium influx and neuronal cell death could be shown recently [13]. Neurons expressing calcium-binding proteins, such as parvalbumin, calbindin D28 K o r calretinin, may be expected to have intrinsic buffer capacitiy for calcium, a buffer which may render them more resistant to excitotoxic stimuli resulting in calcium overload. However, reported data on this issue are quite controversial. Using rather immature hippocampal cell cultures Mattson et al. [25] found calbindin D28K-positive neurons to be more resistant to glutamate stimulation than other neurons. Parvalbuminpositive neurons in cortical cell cultures were reported to be resistant to N M D A receptor-mediated toxicity but vulnerable against non-NMDA receptor-mediated toxicity [39]. As for seizure-induced neurodegeneration in rats, less affected hippocampal areas were found to be rich in neurons containing calbindin D28K- or parvalbumin-positive cells [34]. However, in a model of global forebrain ischemia in rats, pyramidal cells in C A 1 expressing calbindin D28 K w e r e reported to be among the more vulnerable neurons whereas parvalburain-containing cells were more resistant [10]. Calbindin D2sK-positive neurons may be among the preferentially degenerating neurons in Alzheimer's disease, Parkinson's disease and Hungtington's disease, at least in some brain areas [17]. The same may be true in aged rats [4]. Furthermore, a subpopulation of the calretinin-positive neurons in the hippocampus is among the earliest degenerating neurons after transient global ischemia [11]. Therefore, a simple correlation between the expression of calcium-binding proteins and the resistance to neuronal degeneration under various conditions is not obvious. Because of their localization in different brain areas, the local environment, including patterns of excitatory/inhibitory inputs, may differ substantially for subpopulations of neurons expressing calcium-binding proteins. Local differences rather than intrinsic properties, such as absence or presence of calcium-binding proteins, may dominate neuronal degeneration under various conditions, including ischemic insults, and may mask possible favorable effects of intracellular calcium-buffering systems. Selective vulnerability in different hippocampal areas may, thus, reflect differences in local environment, intrinsic properties of neuronal subpopulations or both [31]. Because of the random distribution of cells in mono-

layer cell cultures, local differences are likely to be minimized. An intrinsic favorable effect of intracellular calcium-buffering systems against excitotoxic stimuli should be well recognized under these conditions. Therefore, we analysed the vulnerability of cultured hippocampal neurons to defined excitotoxic stimuli with special emphasis on subpopulations of neurons expressing the calcium-binding proteins calbindin D2s ~ and calretinin.

2. Materials and methods 2.1. Astrocyte feeder-layers Cerebellar glial precursor cells were obtained as described [5]. In short, the cerebellar tissue from 3-4 day-old mice (NMRI strain) was incubated for 10 min in 1% trypsin (Amimed). After addition of Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (Sigma) the tissue was triturated in 0.05% DNase with a fire*polished Pasteur pipette and centrifuged 5 rain at 100× g. The pellet was resuspended in culture medium and the cells were cultured in 250-ml flasks until confluency was reached. Cells were subcultured on poly-D-lysine-coated glass coverslips (15 mm diameter, 100,000 cells/cm 2) placed in 24 multiwell plates (Nunc) as described [5]. Cells were cultured in a chemically defined medium using D M E M supplemented with 10 nM epidermal growth factor, 10 /~g/ml insulin, 100 ~ g / m l transferrin (all from Boehringer Mannheim) and 1 m g / m l bovine serum albumin (Serva) as described [5]. Differentiation of glial precursor cells was induced by replacing the defined medium by D M E M supplemented with 10% fetal calf serum and 2 days later by DMEM supplemented with 10% horse serum (Sigma). These astrocytes were used as feeder-layers for co-cultures with hippocampal cells within 1 week.

2.2. Neuronal cell cultures Hippocampi from 17-18-day-old rat embryos (Roro spf 120) were dissected and meninges carefully removed. The collected tissue samples were washed with Ca2+-free controlled salt solution (CSS; see below) and incubated for 10 min in a mixture of dispase (grade II, 24 U / m l ) and 0.0125% DNase (Boehringer) at room temperature. The tissue was mechanically dissociated in 0.05% DNase by trituration with a fire-polished Pasteur pipette. The resulting cell suspension was centrifuged for 5 rain at 100× g. The pellet was resuspended in DMEM supplemented with 10% horse serum and the cells were seeded on astrocyte feeder-layers in 0.5 ml culture medium. Cells could be cultured with an inital density of 75,000 cells/em 2 up to 3 weeks. One-fifth of the medium was replaced after 7 DIV by fresh DMEM supplemented with 10% horse serum and 10 # M cytosine arabinoside to inhibit proliferation of non-neuronal cells. Cultivation was performed at 37°C under humid atmosphere containing 5% CO2.

2.3. Toxicity experiments Exposure to glutamate receptor agonists: the culture medium was replaced by prewarmed CSS with the following composition (in mM): NaC1, 120; KC1, 5.4; MgCl 2, 0.8; CaCl2. 1.8; Tris-HCl, 7.5; NaHCO3, 17.5; glucose, 15 and phenol red 10 mg/liter. After a second change of the medium, cells were exposed to 50 p~M glutamate for 10 min in the incubator. The addition of 25 p~g/ml propidium iodide was optional and not toxic by its own. After washing twice with CSS, cells

V. M6ckel, G. Fischer/Brain Research 648 (1994) 109-120 were either fixed as described below or cultured for ~ 20 h in DMEM supplemented with 0.5% horse serum in the CO 2 incubator. For permanent exposure (20 h) to 50/~M glutamate, 25 ~M NMDA or 250 ~M kainate the substances were dissolved in DMEM supplemented with 0.5% horse serum.

2.4. Combined glucose / oxygen deprivation Cultures were washed with glucose-free CSS (glucose replaced by sorbitol) and placed in an anaerobic chamber ( < 0.3% 02, 85% N 2, 10% H a and 5% COe). The cells were washed twice and incubated in glucose and oxygen-free CSS containing drugs to be tested. Deprivation intervals of 15, 30 or 60 min were terminated by replacing the medium with oxygenated DMEM supplemented with 0.5% horse serum and drugs. Thereafter, cultures were incubated for 20 h under normal culture conditions. In some experiments, 25 /zg/ml propidium iodide was added during the final 5 min of recovery and after two washes cultures were fixed for immunocytochemical stainings (see below).

111

Incubations with primary antibodies were performed at 4°C for at least 2 h using monoclonal anti-microtubule-associated protein (antiMAP 2, 1:2500, Sigma), monoclonal G5 and polyclonal R1 anticalbindin D2sK (diluted 1:5000 and 1:200, respectively), polyclonal anti-calretinin (diluted 1 : 500) and monoclonal anti-parvalbumin (diluted 1:500; Sigma). Parvalbumin-positive neurons could not be detected even when glutaraldehyde was omitted from the fixation solution. The antisera to calbindin D2sK were a generous gift of W. Hunziker (Hoffmann-La Roche) and were characterized in a recent report [28] whereas the antiserum to calretinin was kindly provided by M. Celio (Fribourg, Switzerland). It was characterized recently [32]. Secondary antibodies (swine anti-rabbit FITC or TRITC, goat anti-mouse FITC or TRITC) were used with a dilution of 1:50 dissolved in 0.1 M phosphate buffer. Unspecific binding of these antibodies was found to be negligible. Cultures were mounted in 90% glycerol with 0.1 M phosphate buffer and 2.5% of potassium iodide.

2. 7. Morphological evaluation of cellular damage 2.5. Quantification of cell death Cell damage was quantified by measurement of lactic acid dehydrogenase (LDH) in the culture supernatant using a Biomek 1000 Automated Laboratory Workstation (Beckman Instruments) as described [21]. Experiments were performed in quadruplicates. S.D. was usually < 20% of the mean value. To compare values of different experiments, the difference between protected controls and cytotoxic-treated cultures without added drugs was taken as 100% damage or 0% protection. For control cultures in toxicity experiments (exposure to glutamate receptor agonists, the calcium ionophore 4-bromo A23187 or glucose/oxygen deprivation), 1 ~M dizocilpine was added during the incubation period.

2.6. Immunocytochemistry Cultures were fixed for 20 min with prewarmed 4% paraformaldehyde and 0.1% glutaraldehyde dissolved in 0.1 M phosphate buffer. Cells were permeabilized with 0.5% Triton X-100 in the presence of 5% normal goat and 5% normal swine serum for lh.

Immunocytochemically stained cultures were examined and photographed using a Zeiss Axiovert 105 M microscope. Labelled cells were photographed with a tri-X-pan film (Kodak, 400 ASA). Quantitative analysis was performed at a magnification of 200× for counting of cell bodies or 400× for morphometric analysis of primary neurites. Examined microscopic fields (up to 50) were located in a transect across the diameter of each coverslip or randomly distributed. Cells were counted in at least three independent culture batches. From each batch one or two coverslips were used for morphological analysis.

2.8. Drugs Ten mM stock solutions in 50% DMSO of dizolcipine or CNQX (Tocris Neuramin) were diluted with the appropriate buffers. The calcium ionophore 4-bromo-A23187 (Sigma) was dissolved in DMSO and diluted to a stock solution of 200 ~M. The final concentration of DMSO during the experiments ranged from 0.05-0.5% and was not toxic by its own. All other drugs were from Sigma.

Table 1 Immunocytochemical characterization of neurons surviving different toxic treatments Surviving neurons (% of control) Treatments

MAP 2positive

Calbindinpositive

Calretininpositive

Calbindin/ calretinin-negative

Glutamate (50 p.M, 10 min, 20 h recovery) NMDA (25 ~M, 20 h) Kainate (250/zM, 20 h) Glucose/oxygen deprivation (15-30 min, 20 h recovery) Glucose/oxygen deprivation (60 min, 20 h recovery) 4-Bromo A23187 (10 p,M, 30 min; 1/xM dizocilpine, 20 h recovery)

5 + 2 (481) 5 + 1 (625) 8 + 3 (520) 6 + 3 (1377) 3 _+ 2 (166) 7:1:6 (763)

Not detectable <1(10) 3 _+ 3 (29) Not detectable Not detectable 9 _+ 11 (125)

54 + 21 (286) 59_+ 2(273) 75 + 20 (331) 63 + 21 (722) 35 + 25 (102) 28 _+ 13 (156)

1 1 3 2 1 5

Cultures were used for all experiments after 10-12 DIV. Cells were either exposed for 10 min to 50 p,M glutamate followed by a recovery period of 20 h or for 20 h to 250/zM kainate or 25/~M NMDA. Combined glucose/oxygen deprivation (15 or 30 min) was followed by a recovery period of 20 h. Incubations with 10/zM 4-bromo A 23187 for 30 min were followed by a recovery period of 20 h and NMDA receptor activation was inhibited by addition of 1 /xM dizocilpine during exposure and recovery period. After these treatments, cultures were fixed and stained. In total 3-10 coverslips derived from at least three independent culture batches were analysed for every experimental set up. In all cases, double-stainings (MAP 2 and calbindin D28K; MAP 2 and calretinin or calbindin D2sK and calretinin) were analysed. Total numbers of analysed cells are given in parenthesis. Number of microscopic fields analysed ranged from 69 to 379. % values for calbindin D28K/calretinin-negative cells represent difference between MAP 2-positive cells and calbindin D2sK- as well as calretinin-positive cells. All values are expressed as % of control cultures.

V. M6ckel, G. Fischer/Brain Research 648 (1994) 109-120

112


A

B

120

(.~ 1 2 o ,<

-7-

t'~ 100

~t'~JlOO,

80 60.

~'~, uJ

20.

-r-

-r

•

- 7

50 IIM glutamate -Idizocilpine

-7-

~

.

250 ~M kainate + -Idizocilpine CNQX GLY

25 pM NMDA -Idizocilpine

--i--

-Idizocilpine CNQX GLY

Fig. 1. Excitotoxicity induced by glutamate, NMDA and kainate in mature hippocampal cell cultures at 11-14 DIV. Excitotoxic stimulation was performed as described in detail. A: cells were exposed for 20 h to 50/zM glutamate or 25 izM NMDA with or without addition of 10 izM dizocilpine. B: cells were exposed to 250/zM kainate for 20 h in presence or absence of 10/zM dizocilpine and/or 100 ~M CNQX as well as 100 IzM glycine (GLY) as indicated. Neuronal cell damage was quantified by measurements of LDH in cell-culture supernatants. Mean _+S.D. values were calculated from at least three experiments each performed in quadruplicates. Control values for glutamate and kainate exposure alone were set as maximal damage (100%). Deviations within different experiments were < 15% from mean value.

In contrast, most of t h e m o r e m a t u r e n e u r o n s after 1 1 - 1 3 D I V died within the r e c o v e r y p e r i o d of 20 h, w h e n they w e r e e x p o s e d for 10 m i n to 50 ~ M glutamate. T h e analysis o f the p r o p i d i u m i o d i d e - s t a i n i n g and t h e i m m u n o c y t o c h e m i c a l staining for M A P 2 (Table 1) r e v e a l e d that ~ 5 % of the n e u r o n s survived this t r a n s i e n t excitotoxic g l u t a m a t e stimulation. A l m o s t all n e u r o n s survived w h e n dizocilpine (10 ~ M ) was present during t h e s h o r t - t e r m (10 min) a n d e v e n the

3. Results

3.1. Glutamate neurotoxicity W h e n relatively i m m a t u r e h i p p o c a m p a l cultures after 6 D I V w e r e e x p o s e d to 50 t z M g l u t a m a t e for 30 min, almost all n e u r o n s ( > 95%) survived the recovery p e r i o d of 20 h. H o w e v e r , most o f t h e n e u r o n s died d u r i n g a 6-h i n c u b a t i o n in 5 0 0 / ~ M g l u t a m a t e .

A

120

120

~1oo.

~100

80

~ 80

~ 60

~ 6o

0

0

N 40

~ 40

X

~ 20 9 30 min deprivation dizocilpine

N CNQX

~ 20

60 min deprivation dizocilpine CNQX

dizocilpine CNQX

Fig. 2. Neurotoxicity induced by combined glucose/oxygen deprivation in mature hippocampal cell cultures at 11 DIV. Deprivation period was 30 (A) or 60 min (B). Dizocilpine (10 ~M), CNQX (100/zM) or a combination of these substances was present throughout deprivation period and subsequent 20 h recovery. Neuronal cell damage was quantified by measurements of LDH in cell-culture supernatants. Mean _+S.D. values were calculated from at least three experiments each performed in quadruplicates. For further details, see legend of Fig. 1.

V. M6ckel, G. Fischer/Brain Research 648 (1994) 109-120

113

Fig. 3. lmmunocytochemical characterization of hippocampal cell cultures using antibodies against MAP 2, calbindin D2SK and calretinin. Phase-contrast micrographs of mature hippocampal cultures at 11 DIV are shown (A,B). Cultures were stained for MAP 2 (C) and calbindin D28K (E; polyclonal antibody) or for calbindin D28K (D; monoclonal antibody) and calretinin (F) using an immunofluorescent double-staining technique as described. Calbindin D28K- (D) and cairetinin-staining (F) did not overlap. A broad variation in staining intensities was visible, including weakly MAP 2-positive neurons which were brightly stained for calbindin DESK. Analysis of neuronal subpopulations expressing calbindin D28K or calretinin was restricted to cells which were stained for these antigens in soma and in primary dendrites. Scale bar, 50 ~m.

114

V.. M6ckel, G. Fischer~Brain Research 648 (1994) 109-120

long-term (20 h) exposure of the cultures to 50 /xM glutamate or to 25/zM NMDA (Fig. 1). Astrocytes were not damaged under the above conditions as evaluated by morphological observation and lack of labelling with propidium iodide. All subsequent experiments were performed with the more mature cell cultures (11-13 DIV) in which neurons were more sensitive to the excitotoxic stimulations as compared with the immature cell cultures after 6 DIV.

3.2. Kainate neurotoxicity Exposure of the cultures for 20 h to 250/xM kainate resulted in a wide-spread degeneration of neurons (Table 1). Dizocilpine (10 /xM) was found not to be protective under these conditions. In contrast, CNQX (100 /zM), together with glycine (100 /zM), protected ~ 80% of the ceils (Fig. 1). This concentration of glycine should block the action of CNQX at the glycine site of the NMDA receptor, thus, making its action more selective for the AMPA/kainate receptor [33]. An almost complete neuroprotection against kainate toxicity was achieved by the combination of 10 /xM dizocilpine, 100/.~M CNQX and 100/zM glycine (Fig. 1). As revealed by microscopic evaluation of propidium iodide-stained cultures astrocytes survived the longterm exposure to 250/zM kainate.

3.3. Combined glucose/oxygen deprivation When hippocampal cultures after 6 DIV were deprived of glucose and oxygen for 30 min, neurons did not die within a subsequent period of 20 h, indicating that these less mature cells were not yet sensitive to this stress. However, when cultures after 11-12 DIV were deprived of glucose and oxygen for > 15 min, most of the neurons (~ 94%) died within a subsequent incubation period of 20 h (Fig. 2A,B, Table 1). The small subpopulation of surviving neurons ( ~ 6%) was still alive after a prolonged subsequent recovery period of 72 h, indicating true survival rather than delayed neurodegeneration. As revealed by microscopic examination of propidium iodide-stained cultures, astrocytes survived even a 60-min period of glucose/oxygen deprivation. The neuroprotective effects of dizocilp'ine and CNQX were dependent on the duration of glucose/ oxygen deprivation. A complete protection against 30 min deprivation could be achieved with 10 /zM dizocilpine independent of whether the subsequent incubation period lasted 20 or 72 h. However, only ~ 55% protection could be obtained when the deprivation period was prolonged to 60 min (Fig. 2B). CNQX (100 tzM) produced ~ 60% protection when the deprivation

Table 2 Characterization of neuronal subpopulations by immunocytochemical staining Marker

N u m b e r of M A P 2-positive cells analyzed

% stained

Calbindin D28 K Calretinin

1303 1932

15+3 5 _+ 1

Double-stainings with anti-calbindin-28K or anti-calretinin and antiM A P 2 antibodies were performed. In separate experiments, it was shown that poly- and monoclonal anti-calbindin-28K antibodies stained same population of neurons. Coverslips from three to four independent culture batches after 11-14 DIV were used to count stained cells at a magnification of 200 × . Data were calculated as % of M A P 2-positive cells.

period was 30 min but proved to be almost inactive when the deprivation period was prolonged to 60 min. The addition of 100/~M glycine did not reduce CNQX effects under these conditions (not shown). After prolonged deprivation periods (60 min), a combined treatment of the cultures with dizocilpine (10 /~M) and CNQX (100 ~M) resulted in ~ 80% protection (Fig. 2B).

3.4. Immunocytochemical characterization of surviving neurons It was striking that a small subpopulation of neurons survived different excitotoxic stimuli as revealed by staining for MAP 2 (Table 1). To determine whether neurons expressing calcium-binding proteins were among the surviving cells, cultures were stained for parvalbumin, calbindin D28K and calretinin. Parvalbumin-expressing neurons could not be detected in control cultures. ~ 15% of the neurons identified with MAP 2-immunostaining were co-stained with antibodies directed against calbindin D2sK and ~ 5% with antibodies directed against calretinin (Table 2, Fig. 3). The morphology of both neuronal subpopulations was highly variable. The number of primary neurites varied from one to eight as shown in a histogram for calretinin-positive neurons in Fig. 6. In addition, the length of primary neurites as well as their branching pattern seemed to be highly variable. However, because of low staining intensities in distal parts of primary and secondary neurites, a more complex morphological analysis, including, e.g., mean length of neurites, could not be performed. Double-staining with mono- and polyclonal antibodies against calbindin D2sK revealed a complete overlap of the stainings whereas double-staining with monoclonal antibodies against calbindin D28K and polyclonal antibodies against calretinin showed no overlap, indicating that distinct neuronal subpopulations can be identified in this way. Immunocytochemical characterization was then per-

V. M6ckel, G. Fischer~Brain Research 648 (1994) 109-120

formed with cell cultures subjected to excitotoxic stimuli. As summarized in Table 1, calbindin D28K-positive neurons were no longer detectable after glutamate exposure (50 tzM for 10 min and a recovery period of

115

20 h) or after glucose/oxygen deprivation (15 min and a recovery of 20 h). In cultures continuously exposed to 25 ~M NMDA or 250 /xM kainate (Fig. 4), calbindin D28K-positive neurons rarely survived.

O ¢

Fig. 4. Kainate-induced excitotoxicity in neurons expressing calbindin D28K or calretinin. Cultures were treated for 20 h with 250 jzM kainate. Phase-contrast micrographs are shown (A,B). Cultures were stained for M A P 2 (C,D) and calbindin D28 K (E) or calretinin (F) as described. Scale bar, 5 0 / z m .

116

V. Miickel, G. Fischer / Brain Research 648 (1994) 109-120

Fig. 5. Effect of 15 min glucose/oxygen deprivation for calretinin-positive neurons. Cultures were stained for M A P 2 (B) and calretinin (C) 20 h after deprivation period. Phase-contrast micrograph is shown in A. Subpopulations of calretinin-positive as well as -negative neurons survived this stimulus. Scale bar, 5 0 / z m .

In contrast to these results with calbindin D28K-positive neurons, a subpopulation of calretinin-positive neurons was found to be more resistant to these stimuli. ~ 50-70% of the calretinin-positive neurons survived exposures to glutamate (50/zM for 10 rain with a subsequent incubation period of 20 h), N M D A (25/zM for 20 h) or kainate (250/zM for 20 h) as revealed by counting calretinin-positive cells excluding propidium iodide (Table 1, Fig. 4). ~ 60% of calretinin-positive neurons survived a 15-30-min deprivation of oxygen and glucose with a recovery period of 20 h (Fig. 5) or even 72 h, indicating survival rather than delayed degeneration of these neurons. The percentage of surviving calretinin-positive neurons decreased to ~ 35% when the deprivation period was extended to 60 min. In all these cases, a small subpopulation of ~ 1 - 3 % of MAP 2-positive neurons survived these procedures (Table 1). This subpopulation did not express detectable levels of the calcium-binding proteins. We then analysed, whether the surviving calretininpositive neurons showed signs of morphological damage. For this purpose, the mean number and the histogram of primary neurites was examined for the subpopulation of calretinin-positive neurons (Fig. 6). The mean number of primary neurites was 3.4 _+ 0.2/cell in control cultures. When the cultures were exposed either to 250/~M kainate for 20 h or to glucose/oxygen deprivation for 30 min with a recovery period of 20 h, the mean number of primary neurites was reduced to 2.1 _ 0.2 and 2.0 + 0.3, respectively. This indicated that either most of the surviving calretinin-positive neurons were morphologically damaged or that calretinin-posi-

tive neurons with a more complex morphology (higher number of primary neurites) degenerated preferentially but not exclusively. Because of the faint staining in secondary and tertiary neurites, a more complex morphological analysis (mean length of neurites, degree of branching, etc.) could not be performed with these cultures. However, secondary and tertiary neurites were hardly visible after the kainate treatment or 6050-

A

40-

[]

CONTROL

•

30 rain deprivation

[]

30 mln deprivation/10 ~M MK-801

[]

20 h kainate 250 ~M

ca 30-

¢j 20.

10

.y2

3

4 5 6 no. of prlrnary neurltas

7

8

Fig. 6. Effect of kainate and combined glucose/oxygen deprivation on n u m b e r of primary neurites of calretinin-positive neurons, Hippocampal cell cultures were deprived for 30 min of glucose and oxygen, followed by a recovery phase of 20 h in absence or presence of 10 /,tM dizocilpine, or exposed for 20 h to 250 /,~M kainate. Afterwards, cultures were stained for calretinin. Relative n u m b e r of cells bearing one to eight primary neurites (abscissa) is given in comparison to control cultures. Note that neurons with a complex morphology have dissappeared after toxic treatments and that dizocilpine fully protects against glucose/oxygen deprivation. Mean 5: S.D. values were calculated from three different experiments in which at least 50 neurons were analysed.

v. M6ckel, G. Fischer/Brain Research 648 (1994) 109-120

after glucose/oxygen deprivation, suggesting loss of the majority of cellular processes. When the cultures were treated with 10 /zM dizocilpine throughout the 30-min glucose/oxygen deprivation as well as the recovery period, the morphology of calretinin-positive neurons was preserved as revealed by determination of the mean number (3.5 + 0.5) and the histogram of primary neurites (Fig. 6). A similar analysis of the small population of MAP 2-positive and calretinin-negative neurons that survived the different excitotoxic stimuli could not be performed because this neuronal subpopulation could not be identified in control cultures. However, some of these cells showed a complex morphology with obviously intact secondary and tertiary neurites after the kainate treatment (Fig. 4) or 15 min of glucose/oxygen deprivation (Fig. 5). 3.5. Response to calcium ionophore 4-bromo A23187

One reason for the apparent resistance to excitotoxic stimuli of a subpopulation of calretinin-positive neurons may be a lower expression of glutamate receptors in comparison to other more sensitive neurons. A stimulation of the apparently more resistant neurons by excitotoxic treatments would then result in a lower influx of calcium and, thus, a weaker stimulus. Therefore, the cultures were exposed in the presence of 10 /xM dizocilpine to the calcium ionophore 4-bromo A23187 (10 ~M for 30 min with a subsequent recovery period of 20 h) to induce a calcium overload independent from expression of glutamate receptors. Survival of cells was quantified by counting immunocytochemically stained cultures. Although there might be a tendency for a higher survival rate for calretinin-positive neuronal cell bodies (Table 1), the stained ceils showed even a more severe loss of neurites than those exposed to excitotoxic agonists because almost all cellular processes were lost. Astrocytes also degenerated under these conditions.

4. Discussion

The vast majoritiy of mature neurons in the rat hippocampal cultures used in these studies was highly sensitive to different excitotoxic stimuli. They died after exposure to glutamate, kainate and combined glucose/oxygen deprivation. Similar findings were reported for cortical neurons cultured under comparable conditions [2,7,13] as well as for hippocampal neurons [27,30]. However, the high vulnerability to short-term oxygen/glucose deprivation (15-30 min) was striking. Whereas in the present hippocampal cultures almost all neurons died during the subsequent incubation period other cell cultures might be less vulnerable be-

117

cause, even after 40 min of oxygen/glucose deprivation, only a subpopulation of neurons degenerated in cortical cultures [e.g. 13]. To examine a possible protective effect of different glutamate receptor antagonists, only high concentrations of these compounds were used, such as 10 /~M dizocilpine or 100 ~M CNQX, which should exert maximal protective effects [14,19]. As expected, dizocilpine protected neurons against short-term glutamate neurotoxicity (10 min exposure to 50 /zM glutamate, not shown), long-term glutamate exposure (50 /~M for 20 h) as well as short-term glucose/oxygen deprivation, indicating that activation of NMDA receptors dominated neurotoxicity under these conditions. Similar results were reported for cortical cultures [13,14,29]. No protection was obtained with dizocilpine against high concentrations of kainate (250 /xM for 20 h) whereas CNQX partly protected neurons. Dizocilpine blocks NMDA receptors but not AMPA/kainate receptors [6] whereas CNQX is a mixed antagonist at AMPA/kainate receptors as well as the glycine-site of NMDA receptors [33]. This strongly suggests that AMPA/kainate receptors were sufficiently activated by high kainate concentrations to predominantly mediate neurotoxicity as reported for cortical cell cultures [22]. Complete neuroprotection against high concentrations of kainate was obtained only by a combined treatment with dizocilpine and CNQX. This implies that kainate excitotoxicity is partly mediated by the release of endogenous glutamate that then activates NMDA receptors. A similar protection by a combined treatment with dizolcipine and CNQX was obtained against long-term deprivation of glucose/oxygen. This indicates that, in contrast to short-term deprivation, activation of different types of glutamate receptors contributes to neurodegeneration under long-term glucose/oxygen deprivations [13,19]. Taken together, the data clearly indicate that the vast majority of neurons expressed sufficient glutamate receptors, including NMDA as well as AMPA receptors, to be highly vulnerable to the different excitotoxic stimuli. A striking finding was the survival of a small percentage of neurons exposed to these different toxic stimuli thought to result in calcium overload triggered by stimulation of excitatory amino acid receptors. Degeneration of neurons in acute glutamte toxicity as well as short-term oxygen/glucose deprivation is closely related to calcium influx through NMDA receptors [13,37]. Calcium-binding proteins, such as parvalbumin, calbindin D28K and calretinin, were reported to be intracellular calcium buffer systems, although their function is not yet clear [16]. Expression of calciumbinding proteins may, therefore, be expected to render neurons more resistant to excitotoxic stimuli although a simple relationship between expression of calcium-

118

v. M~ckel, G. Fischer~Brain Research 648 (1994) 109-120

binding proteins and resistance to excitotoxic stimuli is not obvious in vivo [e.g. 10,11,12]. An immunocytochemical analysis of control cultures revealed that ~ 5% of all neurons expressed calbindin D2s K and ~ 5% of all neurons calretinin as indicated by an intensive immunostaining with the respective antibodies. This is in agreement with in vivo findings in hippocampal slices of the rat in which more calbindin D28K- than calretinin-positive neurons were found [18]. An overlap of calbindin D28K- and calretinin-staining could not be detected, indicating that different subpopulations of neurons expressed these two calcium-binding proteins. The same is described for immunocytochemical staining of brain slices [26]. In the hippocampus, calbindin D28K is known to be present in GABAergic nonpyramidal cells [36] as well as in CA 1 pyramidal cells and granule cells of the dentate gyrus. Whereas calbindin D28K-containing non-pyramidal cells are resistant to ischemia and epilepsy the calbindin D2s K pyramidal neurons are among the more vulnerable neurons [10,12]. In the hippocampus, exclusively non-pyramidal neurons express calretinin. This population includes GABAergic neurons of a spiny type with two to three thick primary dendrites and a spine-free type with three to six primary dendrites [15] and may be some glutamatergic neurons [35]. Recently, it could be shown that the spiny type neurons, located in hilus and CA 3 and receiving input from mossy fibers are among the neurons which degenerate first after transient global ischemia, well before degeneration of CA 1 pyramidal neurons [11]. They are also highly vulnerable to kainate toxicity [24]. Using various excitotoxic stimuli, including glutamate and kainate as well as glucose/oxygen deprivation, calbindin D2sK-positive neurons were found not to be more resistant than the vast majority of other neurons, indicating that calbindin De8 K expression per se does not render neurons more resistant towards excitotoxic stimuli. There is no indication that the disappearance of calbindin D2sK-positive neurons might have been an artefact due to a transient loss of immunoreactivity for calbindin D28K because calbindin D28K-positive neurons were also not detected after prolonged recovery periods of 48 to 72 h under various conditions. One explanation for the vulnerability of calbindin D28K-positive neurons might be that, under our culture conditions, the calbindin DesK-positive neurons are primarily the more vulnerable pyramidal cells (see above) and that the more resistant non-pyramidal GABAergic cells did not develop as it was the case for parvalbumin-positive GABAergic interneurons. However, other non-pyramidal neurons, e.g., calretininpositive cells, were present. Contradicting results with respect to vulnerability of calbindin D28K-positive neurons in vitro were reported by Mattson et al. [25]. However, in these studies less

mature hippocampal cell cultures were used after 6 DIV and exposed for 6 h to 5 0 0 / z M glutamate. In our hands, not only calbindin D28K-positive neurons but also the vast majority of these less mature neurons survived different stimuli, including combined glucose/ oxygen deprivation for 30 min which were fatal for more mature neurons. Similar findings for glutamate excitotoxicity were described for immature vs. mature neuronal cultures [8,23], suggesting that sensitivity towards excitotoxicity is fully expressed only in differentiated neurons. In contrast to calbindin D28K-positive neurons, a subpopulation of calretinin-positive neurons ( ~ 5070%) survived the different excitotoxic stimuli. It might well be that the degenerating calretinin-positive neurons represent the spiny type GABAergic interneurons in CA 3 and hilus being very sensitive to damage induced by transient global ischemia [11]. Most of the surviving calretinin-positive neurons were morphologically damaged as shown by the loss of secondary and tertiary neurites. The latter finding together with the loss of ~ 30-50% of the whole population of calretinin-positive neurons indicates that the expression of calretinin per se did not protect neurons against these stimuli. Treatment with dizocilpine during glutamate excitotoxicity or combined glucose/oxygen deprivation protected sensitive neurons, as shown by L D H measurements and preservation of neurites. Both calbindin D28K- and calretinin-positive neurons were among the protected neurons, indicating that under these conditions N M D A receptor activation was the dominant excitotoxic stimulus also for these subpopulations. One reason for the apparent lower sensitivity of a subpopulation of calretinin-positive neurons to excitotoxic stimuli may be that the majoritiy of these cells expressed less glutamate receptors than the majoritiy of other neurons, thus, making them less sensitive to toxicity mediated via activation of these receptors. This argument is supported by toxicity experiments with the calcium ionophore 4-bromo A23187 in the presence of dizocilpine. In these experiments, the increase in intracellular calcium should not be influenced significantly by the activation of glutamate receptors. Calretinincontaining neurons were not found to be significantly more resistant than calbindin D2sK-positive neurons and most of the neurons died under the experimental conditions. In addition, the small percentage of surviving cell bodies had lost almost all cellular processes. However, another explanation might be that different calcium entry routes affect cells in different ways as shown for cultured spinal cord neurons [37]. It can not be excluded that increases in calcium by activation of glutamate receptors can be buffered more efficiently by calretinin than increases mediated via treatment of cells with the calcium ionophore. Taken together, these results strongly suggest that

V. M6ckel, G. Fischer / Brain Research 648 (1994) 109-120

calcium-binding proteins per se do not render most of the neurons significantly more resistant towards the excitotoxic stimuli used which are thought to mimic excitotoxicity in stroke and brain trauma. However, whether a significant protection can be obtained to more subtle stimuli can not be deduced from our experiments.

[10]

[11]

[12]

Acknowledgements The authors highly appreciate the excellent technical assistence of V. Graf and helpful discussions with Professor W. Haefely.

[13]

[14]

Abbreviations AMPA CNQX CSS DIV DMEM FITC LDH MAP 2 MK-801 NMDA TRITC

a-amino-3 -hydroxy-5 -methyl-4-isoxazolpropionate 6-cyano-7-nitroquinoxaline-2,3-dione controlled salt solution days in vitro Dulbecco's modified Eagle's medium fluorescein-isothiocyanate lactate dehydrogenase microtubule-associated protein 2 dizocilpine N-methyl-D-aspartate tetramethylrhodamin-isothiocyanate

[15]

[16] [17]

[18]

[19]

References [1] Choi, D.W., Cerebral hypoxia: some new approaches and unanswered questions, 3. Neurosci., 10 (1990) 2493-2501. [2] Choi, D.W., Koh, J.-Y. and Peters, S., Pharmacology of glutamate neurotoxicity in cortical cell culture: attenuation by NMDA antagonists, J. Neurosci., 8 (1988) 185-196. [3] Choi, D.W., Peters, S. and Visesekul, V., Dextrorphan and levopharnol selectively block N-methyl-o-aspartate receptor mediated neurotoxicity on cortical neurons, J. Pharm. Exp. Ther., 242 (1987) 713-719. [4] Dutar, P., Potier, B., Lamour, Y., Emson, P.C. and Senut, M.C., Loss of calbindin-28K immunoreactivity in hippocampal slices from aged rats: a role for calcium? Ear. J. Neurosci., 3 (1991) 839-849. [5] Fischer, G., Leutz, A. and Schachner, M., Cultivation of immature astrocytes of mouse cerebellum in a serum-free, hormonally defined medium. Appearance of the mature astrocytic phenotype after addition of serum, Neurosci. Lett., 29 (1982) 297-302. [6] Foster, A.C. and Wong, E.H.F., The novel anticonvulsant MK801 binds to the activated state of the N-methyl-o-aspartate receptor in rat brain, Br. J. PharmacoL, 91 (1987) 403-409. [7] Frandsen, A. and Schousboe, A., Time and concentration dependency of the toxicity of excitatory amino acids on cerebral neurons in primary culture, Neurochem. Int., 10 (1987) 583-891. [8] Frandsen, A. and Schousboe, A., Development of excitatory amino acid induced cytotoxicity in cultured neurons, Int. J. DeL'. Neurosci., 8 (1990) 209-216. [9] Frandsen, A. and Schousboe, A., Dantrolene prevents glutamate cytotoxicity and Ca 2+ release from intracellular stores in

[20] [21]

[22]

[23]

[24]

[25]

[26]

[27]

119

cultured cerebral cortical neurons, J. Neurochem., 56 (1991) 1075-1078. Freund, T.F., Buzsaki, G., Leon, A., Baimbridge, K.G. and Somogyi, P., Relationship of neuronal vulnerability and calcium binding protein immunoreactivity in ischemia, Exp. Brain Res., 83 (1990) 55-66. Freund, T.F. and Magloczky, Z., Early degeneration of calretinin-containing neurons in the rat hippocampus after ischemia, Neuroscience, 56 (1993) 581-596. Freund, T.F., Ylinen, A., Miettinen, R., Pitkauen, A., Lahtinen, H., Baimbridge, K.G. and Riekkinen, P.J., Pattern of neuronal death in the rat hippocampus after status epilepticus. Relationship to calcium binding protein content and ischemic vulnerability, Brain Res. Bull., 28 (1991) 27-38. Goldberg, M.P. and Choi, D.W., Combined oxygen and glucose deprivation in cortical cell cultures: calcium-dependent and calcium-independent mechanisms of neuronal injury, J. Neurosci., 13 (1993) 3510-3524. Goldberg, M.P., Visekul; V. and Choi, D.W., Phencyclidine receptor ligands attenuate cortical neuronal injury after Nmethyl-D-aspartate exposure to hypoxia, J. Pharm. Exp. Ther., 245 (1988) 1081-1087. Gulyas, A.I., Miettinen, R., Jacobowitz, D.M. and Freund, T.F., Calretinin is present in non-pyramidal cells of the rat hippocampus-I. A new type of neuron specifically associated with the mossy fibre system, Neuroscience, 48 (1992) 1-27. Heizmann, C.W. and Hunziker, W., Intracellular calcium-binding proteins: more sites than insights, TIBS, 16 (1991) 98-103. Iacopino, M.A. and Christakos, S., Specific reduction of calcium-binding protein (28-kilodalton calbindin-D) gene expression in aging and neurodegenerative diseases, Proc. NatL Acad. Sci. USA, 87 (1990) 4078-4082. Jacobowitz, D.M. and Winsky, L., Immunocytochemical localization of calretinin in the forebrain of the rat, J. Cornp. Neurol., 304 (1991) 198-218. Kaku, D.A., Goldberg, M.P. and Choi, D.W., Antagonism of non-NMDA receptors augments the neuroprotective effect of NMDA receptor blockade in cortical cultures subjected to prolonged deprivation of oxygen and glucose, Brain Res., 554 (1991) 344-347. Kim, J.P. and Choi, D.W., Quinolinate neurotoxicity in cortical cell culture, Neuroscience, 23 (1987) 423-432. Klingman, J.G., Harley, D.M. and Choi, D.W., Automated determination of excitatory amino acid neurotoxicity in cortical culture, J. Neurosci. Methods, 31 (1990) 47-51. Koh, J.Y. and Choi, D.W., Selective blockade of non-NMDA receptors does not block rapidly triggered glutamate-induced neuronal death, Brain Res., 548 (1991) 318-321. K611er, H., Siebler, M., Schmalenbach, C. and Miiller, H.-W., GABA and glutamate receptor development of cultured neurons from rat hippocampus, septal region, and neocortex, Synapse, 5 (1990) 59-64. Magloczky, Zs. and Freund, T.F., Selective neuronal death in the contralateral hippocampus following unilateral kainate injections into the CA3 subfield, Neuroscience, 56 (1993) 317-336. Mattson, M.P., Rychlik, B., Chu, C. and Cristakos, S., Evidence for calcium-reducing and excito-protective roles for the calcium-binding protein calbindin-D28k in cultured hippocampal neurons, Neuron, 6 (1991) 41-51. Miettinen, R., Gulyas, A.I., Baimbridge, K.G., Jacobowitz, D.M. and Freund, T.F., Calretinin is present in non-pyramidal cells of the rat hippocampus - II. Co-existence with other calcium binding proteins and GABA, Neuroscience, 48 (1992) 29-43. Pauwels, P.J., van Assouw, H.P. and Leysen, J.E., Attenuation of neurotoxicity following anoxia or glutamate receptor activation in EGF- and hippocampal extract-treated neuronal cultures, Cell. Signal., 1 (1989) 45-54.

120

V. M6ckel, G. Fischer~Brain Research 648 (1994) 109-120

[28] Pinol, M.R., K~igi, U., Heizmann, C.W., Vogel, B., Sequier, J.-M., Haas, W. and Hunziker, W., Poly- and monoclonal antibodies against recombinant rat brain calbindin D28-K were produced to map its selective distribution in the central nervous system, J. Neurochem., 54 (1990) 1827-1833. [29] Priestley, T., Horne, A.L., McKernan, R.M. and Kemp, J.A., The effect of NMDA receptor glycine site antagonists on hypoxia-induced neurodegeneration of rat cortical cell cultures, Brain Res., 531 (1990) 183-188. [30] Rothman, S., Synaptic release of excitatory amino acid neurotransmitter mediates anoxic neuronal death, J. Neurosci., 4 (1984) 1884-1891. [31] Schmidt-Kastner, R. and Freund, T.F., Selective vulnerability of the hippocampus in brain ischemia, Neuroscience, 40 (1991) 599-636. [32] Schwaller, B., Buchwald, P., Bliimcke, I., Celio, M.R. and Hunziker, W., Characterization of a polyclonal antiserum against the purified human recombinant calcium binding protein calretinin, Cell Calcium, submitted. [33] Sheardown, M.J., Nielsen, E.O., Hansen, A.J. Jacobsen, P. and Honor6, T., 2,3-Dihydroxy-6-nitro-7-sulfamoyl-benzo (F) quinoxaline: a neuroprotectant for cerebral ischemia, Science, 247 (1990) 571-574.

[34] Sloviter, R.S., Calcium-binding protein (calbindin-D28k) and parvalbumin immunocytochemistry: localization in the rat hippocampus with specific reference to the selective vulnerablility of hippocampal neurons to seizure activity, J. Comp. Neurol., 280 (1989) 183-196. [35] Soriano, E. and Frotscher, M., First evidence for glutamtaergic nonpyramidal neurons in the rat hippocampus: golgi impregnation combined with postembedding immunocytochemistry, Suppl. Eur. J. Neurosci., 5 (1992) 2178. [36] Toth, K. and Freund, T.F., Calbindin D28K-containing nonpyramidal cells in the rat hippocampus: their immunoreactivity for GABA and projection to the medial septum, Neuroscience, 49 (1992) 793-805. [37] Tymianski, M., Charlton, M.P., Carlen, P.L. and Tator, Source specificity of early calcium neurotoxicity in cultured embryonic spinal neurons, J. Neurosci., 13 (1993) 2085-2104. [38] Weiss, J.H., Hartley, D.M., Koh, J. and Choi, D.W., The calcium channel blocker nifedipine attenuates slow excitatory amino acid neurotoxicity, Science, 247 (1990) 1474-1477. [39] Weiss, J.H., Koh, J.-Y., Baimbridge, K.G. and Choi, D.W., Cortical neurons containing somatostatin- or parvalbumin-like immunoreactivity are atypically vulnerable to excitotoxic injury in vitro, Neurology, 40 (1990) 1288-1292.