Neuroprotective effects of bcl-2 overexpression ... - Wiley Online Library

Journal of Neurochemistry, 2002, 83, 914–923

Neuroprotective effects of bcl-2 overexpression in hippocampal cultures: interactions with pathways of oxidative damage Sarah Howard,* Clement Bottino,* Sheila Brooke,* Elise Cheng,* Rona G. Giffard ,à and Robert Sapolsky*,à,§ *Department of Biological Sciences, Stanford University, Stanford, California, USA Departments of Anesthesiology, àNeurosurgery and §Neurology and Neurological Sciences, Stanford University School of Medicine, Stanford, California, USA

Abstract Overexpression of bcl-2 protects neurons from numerous necrotic insults, both in vitro and in vivo. While the bulk of such protection is thought to arise from Bcl-2 blocking cytochrome c release from mitochondria, thereby blocking apoptosis, the protein can target other steps in apoptosis, and can protect against necrotic cell death as well. There is evidence that these additional actions may be antioxidant in nature, in that Bcl-2 has been reported to protect against generators of reactive oxygen species (ROS), to increase antioxidant defenses and to decrease levels of ROS and of oxidative damage. Despite this, there are also reports arguing against either the occurrence, or the importance of these antioxidant actions. We have examined these issues in neuron-enriched primary hippocampal cultures, with overexpression of bcl-2 driven by a herpes simplex virus amplicon: (i) Bcl-2 protected strongly against glutamate, whose toxicity is at least partially ROS-dependent. Such protection involved reduction in mitochondrially derived

superoxide. Despite that, Bcl-2 had no effect on levels of lipid peroxidation, which is thought to be the primary locus of glutamate-induced oxidative damage; (ii) Bcl-2 was also mildly protective against the pro-oxidant adriamycin. However, it did so without reducing levels of superoxide, hydrogen peroxide or lipid peroxidation; (iii) Bcl-2 protected against permanent anoxia, an insult likely to involve little to no ROS generation. These findings suggest that Bcl-2 can have antioxidant actions that may nonetheless not be central to its protective effects, can protect against an ROS generator without targeting steps specific to oxidative biochemistry, and can protect in the absence of ROS generation. Thus, the antioxidant actions of Bcl-2 are neither necessary nor sufficient to explain its protective actions against these insults in hippocampal neurons. Keywords: apoptosis, bcl-2, necrosis, neurotoxicity, oxygen radicals, reactive oxygen species. J. Neurochem. (2002) 83, 914–923.

An extensive work demonstrates the capacity of Bcl-2 to block cell death in numerous cell types, including both neurons and glia (Green and Reed 1998). The protein is thought to heterodimerize with pro-apoptotic proteins such as BAX, thereby impeding the release of cytochrome c from mitochondria. Release of cytochrome c is critical to the activation of caspases and the execution of programmed cell death. However, Bcl-2 has a variety of other actions within cells and, as the most explicit example of this, Bcl-2 can also block apoptosis following cytochrome c release (Rosse et al. 1998). These other actions include Bcl-2 enhancing mitochondrial calcium uptake (Murphy et al. 1996) and decreasing nuclear calcium accumulation (Marin et al. 1996), forming ion channels in mitochondria (Green and Reed 1998), and causing the translocation of kinases to mitochondria (Wang et al. 1996). Moreover, Bcl-2 is capable of blocking instances of

necrotic, as well as apoptotic cell death (Kane et al. 1993, in neural cell lines; Yang et al. 1998, in the substantia nigra; Papadopoulos et al. 1998, in cortical astrocytes). Thus, Bcl-2 appears to have salutary effects other than preventing cytochrome c release. It has long been postulated

914

Resubmitted manuscript received August 13, 2002; accepted August 19, 2002. Address correspondence and reprint requests to Robert Sapolsky, Department of Biological Sciences, Gilbert Laboratory MC 5020, Stanford University, Stanford, CA 94305–5020, USA. E-mail: [email protected] Abbreviations used: ABTS, 2,3-azino-bis(ethylbezothiazoline-6-sulfonic) acid; DCF, dichlorodihydrofluorescein diacetate; DMSO, dimethyl sulphoxide; HE, dihydroethidium; MEM, a modified minimum essential medium; MOI, multiplicity of infection; PBS, phosphate-buffered saline; ROS, reactive oxygen species.

2002 International Society for Neurochemistry, Journal of Neurochemistry, 83, 914–923

Neuroprotective effects of bcl-2 overexpression

that some of these additional effects occur in the oxidative realm. Such protection could involve Bcl-2 acting as a classical antioxidant [i.e. quenching reactive oxygen species (ROS)]. However, there is little evidence for this (Lee et al. 2001). In addition, Bcl-2 may decrease ROS generation, increase the level/activity of antioxidants and/or protecting targets of oxidative damage. Overexpression of bcl-2 attenuates the cell death caused by numerous insults whose toxicities depend heavily upon ROS generation (such as adriamycin, paraquat, hydrogen peroxide, 6-OHDA, MPTP; see below). Moreover, many of these instances of protection are accompanied by a reduction in ROS accumulation (Kane et al. 1993, insult of glutathione depletion in neural cell lines; Lawrence et al. 1996, adriamycin in primary hippocampal cultures; Papadopoulos et al. 1998, aglycemia in cortical astrocyte cultures). In addition, Bcl-2 decreases oxidative damage (Kane et al. 1993, glutathione depletion in neural cell lines; Myers et al. 1995, cyanide/aglycemia in hypothalamic tumor lines; Bruce-Keller et al. 1998, hydrogen peroxide and amyloid b-peptide in PC12 cells; Giardino et al. 1996, hyperglycemia in peripheral tissue; Lee et al. 2001, hydrogen peroxide in teratocarcinoma and neuroblastoma cell lines). Studies also report that Bcl-2 can increase the activity or levels of antioxidants (Kane et al. 1993, in neural cell line; Ellerby et al. 1996, in a hypothalamic cell line; Papadopoulos et al. 1998, in cultured cortical astrocytes; Steinman 1995, in peripheral tissue; Voehringer et al. 1998, in peripheral cell lines; Lee et al. 2001, in a peripheral and a neuroblastoma cell line). In one scenario revolving around the putative antioxidant effects of Bcl-2, the protein is thought to protect mitochondrial membranes from peroxidative damage (Bruce-Keller et al. 1998). Another study focuses on the capacity of Bcl-2 to maintain mitochondrial potential and decrease mitochondrial ROS production (Green and Reed 1998). Thus, it is clearly that Bcl-2 can exert significant antioxidant actions. However, there are findings which suggest that this is not always the case: 1. In a number of cases in peripheral tissue, Bcl-2 prevents cell death under circumstances which are highly unlikely to involve ROS (for example, growth factor deprivation under anaerobic conditions; (Jacobson and Raff 1995; Shimizu et al. 1995). 2. In at least some reports, Bcl-2 does not prevent the cell death induced by the pro-oxidant 6-OHDA (Oh et al. 1995, 1998, in a dopaminergic cell line; Yamada et al. 1999, in the substantia nigra for example of protection against 6-OHDA). 3. In one instance where Bcl-2 reduced both the ROS accumulation and cell death caused by an insult, the reduction in ROS accumulation was shown to be irrelevant to the sparing from death (Gardner et al. 1997, in fibroblasts), and could only account for part of the protection observed (Papadopoulos et al. 1998).

915

4. Finally, there has been the suggestion that Bcl-2 can act as a pro-oxidant. One report involving bcl-2 overexpression in Escherichia coli, and in a B-cell line (Steinman 1995), and one involving cultured cortical astrocytes (Papadopoulos et al. 1998) showed that a primary effect of Bcl-2 was to increase ROS levels, and the increase in antioxidant levels could be viewed as a secondary compensation. Relatively few of the reports in this confusing literature are derived from studies of the nervous system, which is particularly vulnerable to ROS and in which Bcl-2 can be highly protective from various insults. Because of this, we have examined these issues in primary hippocampal cultures overexpressing bcl-2. We observe that: (i) Bcl-2 can reduce the neurotoxicity of an excitotoxic insult while having inconsistent effects upon indices of oxidative stress; (ii) Bcl-2 can reduce the neurotoxicity of a pro-oxidant while having no effect on indices of oxidative stress; (iii) Bcl-2 can spare neurons from an insult whose toxicity is not likely to involve the generation of ROS. Collectively, these findings argue against the importance of an antioxidant role for Bcl-2 in its neuroprotective actions in cultured hippocampal neurons.

Materials and methods Hippocampal cell cultures Tissue culture methods were described previously (Brooke et al. 1997) Briefly, hippocampus from 18-day-old fetal rats were removed, dissociated with papain, filtered through an 80-lm cell strainer, and resuspended in a modified minimum essential medium (MEM; UCSF Tissue Culture Facility, San Francisco, CA, USA) containing 25 mM glucose and 10% horse serum (Hyclone, Logan, UT, USA). Cells were plated at a density of 30 000/cm2 on poly-Dlysine-treated 96-well plates for the toxicity and superoxide studies, on 24-well plates for the lipid peroxidation studies, and on 48-well plates for the anoxia and anoxia/aglycemia studies. Cells were used after 10–12 days. Cells used in glutamate experiments were treated with 10 lM cytosine arabinoside on day 3 in culture, which increased the percentage of neurons to approximately 70–80%. Bcl-2 overexpression A modified herpes simplex virus was used to deliver plasmids to hippocampal neurons in culture. A bipromoter plasmid, pa22b gala4bcl-2, containing bcl-2 and the lacZ reporter gene under a4 and a22 promoters, respectively, was used to overexpress bcl-2. The control plasmid, pa4bgal, contained the a4 promoter and lac Z reporter gene. Both plasmids also contained the oriS and a sequences required for replication and packaging. Construction of plasmids was described previously (Lawrence et al. 1996). Vectors were generated by transfection of plasmids into E5 cells using lipofectamine (Gibco-BRL, Gaithersburg, MD, USA) and superinfecting 24 h later with the helper virus d120 at a multiplicity of infection (MOI) of 0.03 (Ho 1994). E5 cells stably transformed with the a4 gene allowed d120, which lacks a4, propagation (DeLuca et al. 1985). Amplicons and d120 helper virus were titered


916 S. Howard et al.

on Vero and E5 cells, respectively, and both were in the range of 0.5–3 · 107 infectious particles/mL. In all experiments, hippocampal cultures were infected approximately 12 h prior to experimental treatment. Under these conditions, 63% of neurons are infected, as are 11% of glia. Solutions For all experiments, working solutions were prepared in MEM Eagle media (UCSF Tissue Culture Facility). A 1-M glutamate stock solution in water was prepared prior to each experiment and further diluted in MEM Eagle media. A 2-mM adriamycin (also known as doxorubicin hydrochloride; Sigma, St Louis, MO, USA) stock solution was kept at 4C and diluted in MEM Eagle media for use. EDTA (Sigma) was diluted from a 100-mM stock for use. Toxicity studies Glutamate or adriamycin was added directly to tissue culture wells to the final concentrations of 5, 10, 15, 20 lM and 10, 30, 40, 50 and 60 lM, respectively. After 24 h, cells were fixed with cold methanol and analyzed according to a published method (Brooke et al. 1997). Briefly, cells were blocked with 5% milk in phosphate-buffered saline (PBS), followed by immunocytochemistry with a neuron specific primary antibody against MAP-2 (Sigma) at a dilution of 1 : 1000 in 5% milk in PBS. Following incubation with a ratadsorbed biotinylated secondary anti-IgG antibody (Vector, Burlingame, CA, USA), cells were treated with avidin-bound horseradish peroxidase (ABC reagent, Vector). Finally, 2,3-azino-bis(ethylbezothiazoline-6-sulfonic) acid (ABTS) was added according to manufacturer’s instructions (Vector), producing a color change in proportion to amount of MAP-2 present. Absorbance at 405 nM wavelength was read on a plate reader. During analysis, blanks (wells treated with cold media to kill neurons), were subtracted from all values and data were expressed as percentage survival according to comparison with control wells that received no insult. Hydroethidine studies Generation of intracellular superoxide was determined according to fluorescence of ethidium as a result of oxidation of hydroethidine [also known as dihydroethidium (HE); Molecular Probes, Eugene, OR, USA; Lagrange et al. 1994]. HE, 10 lg/lL in dimethyl sulphoxide (DMSO), was stored under nitrogen at )80C. Experimental treatments were added in 10–20 lL aliquots directly to tissue culture wells. Final concentrations of treatments were 40 lM adriamycin with and without 1.5 mM EDTA, and 10 and 20 lM glutamate. HE (16 lM) or DMSO vehicle was added at the same time. Thirty minutes later, media were aspirated and replaced with PBS containing 1% Triton-X. Fluorescence with excitation 480 nm and emission 590 nm was read on a fluorescence plate reader. Blanks without HE were subtracted from readings. A Pierce assay was done in order to standardize for the amount of protein. Data were expressed as percentage of control on each plate. For the rotenone experiments, cells were pretreated with a final concentration of 10 lM rotenone (Sigma) in MEM, 40 min prior to experimental treatment. DCF studies Dichlorodihydrofluorescein diacetate (DCF; Molecular Probes) fluoresces upon oxidation by hydrogen peroxide (Lebel et al.

1990). DCF was stored at 4 mM in DMSO at – 80C. Experimental treatments were added in 10–20 lL aliquots with final concentrations of 40 lM adriamycin with and without 1.5 mM EDTA. DCF (20 lM) or DMSO vehicle was added at the same time. Fifteen minutes later, media were aspirated and replaced with PBS containing 1% Triton-X. Fluorescence was read on a plate reader with excitation 480 nm and emission 520 nm. Blanks, without DCF, were subtracted from readings. A Pierce assay was used to determine amounts of protein for standardization. Data were expressed as a percentage of control on each plate. Because DCF is pH-sensitive and glutamate decreases intracellular pH, attempts to detect glutamate-induced increases in hydrogen peroxide were not useful. Lipid peroxidation studies Lipid peroxidation was determined by measuring the loss of fluorescence due to peroxidation of the naturally fluorescent fatty acid, cis-parinaric acid (Molecular Probes; Kuypers et al. 1987). Hippocampal cultures were incubated with 10 lM cis-parinaric acid or 90% EtOH vehicle for 1 h prior to addition of 10 lL insult or vehicle. Final conditions were 10 lM glutamate or 40 lM adriamycin with and without 1.5 mM EDTA. Two hours later, cells were scraped from tissue culture wells and suspended in PBS bubbled with nitrogen. Fluorescence was measured on a Perkin-Elmer LS50B spectrometer (Perkin-Elmer, Foster City, CA, USA) with excitation 312 nm and emission 414 nm. Blanks containing no cis-parinaric acid were subtracted from readings and data were expressed as percent of control. Because cisparinaric acid is light sensitive, all manipulations were performed in the dark. Anoxia studies Experimental and control cells received two media changes with a balanced salt solution. The experimental solution contained NaCl 116 mM, CaCl2 1.8 mM, MgSO4 0.8 mM, KCl 5.4 mM, NaH2PO4 1 mM, glucose 5.5 mM, NaHCO3 14.7 mM, and HEPES 10 mM. The control solution was the same except it lacked HEPES and had a higher concentration of NaHCO3 (27 mM). Experimental cells were transferred to an anoxia chamber (Sheldon Manufacturing, Cornelius, OR, USA) in an atmosphere of 5% CO2, 5% H2 and 90% N2, where the media was changed twice to the deoxygenated salt solution lacking glucose. Control plates received the same treatment with solution incubated in the 37C incubator with 5% CO2. Experimental plates were incubated at 37C in the anoxia chamber (with N2 gas). After 5–6 h, survival was assessed by the ABTS assay described. Data analysis and statistics For all studies, data were expressed as a percentage increase above Bcl-2 or bgal control after comparison by t-test to determine that Bcl-2 and bgal controls were not significantly different from each other. For toxicity studies, two-way ANOVA followed by Tukey posthoc test was used to determine difference between bgal and Bcl-2 groups. For cPnA, HE and DCF studies, t-test was used to compare insult group with Bcl-2 versus insult group with bgal. In the HE study with two concentrations of glutamate, a 2-way ANOVA followed by Tukey’s post-hoc test was used to compare the bgal group versus the Bcl-2 group. For the anoxia and anoxia/aglycemia



917

studies, percentage survival with bgal was compared with percentage survival with Bcl-2 by t-test. For all statistics, significance was set at p < 0.05, and data are presented as mean ± SE.

Results

As would be expected, glutamate was neurotoxic to neuronenriched hippocampal cultures in a dose-dependent manner [Fig. 1; the extent of toxicity did not differ from mockinfected cultures exposed to equivalent amounts of glutamate (data not shown)]. In support of the picture of excitotoxininduced neuron death involving ROS accumulation and damage, glutamate also caused a significant accumulation of superoxide (Fig. 2) and a significant amount of lipid

Fig. 1 Glutamate-induced neurotoxicity in neuron-enriched hippocampal cultures treated with either control vector (m) or Bcl-2 (d). Bcl-2 caused a significant decrease in neurotoxicity (p < 0.02, F ¼ 6.04, d.f. ¼ 1/113, two-way ANOVA). n ¼ averaged 10/data point, derived from three different weekly culture preparations.

Fig. 2 Effects of glutamate and of differing viral vectors on superoxide accumulation in neuron-enriched hippocampal cultures. Cultures were exposed to indicated concentrations of glutamate and either the control (bgal) or bcl-2-expressing vector. Increasing glutamate concentrations produced increasing superoxide accumulation in control cultures (p < 0.02 by one-way ANOVA), but not in Bcl-2-treated cultures (n.s). When exposed to 10 lM glutamate, Bcl-2-treated cultures had significantly less accumulation than did control cultures (*p < 0.05, Tukey test following two-way ANOVA). n ¼ 22–23/group, derived from five different weekly culture preparations.

Fig. 3 Effects of glutamate and of differing viral vectors on lipid peroxidation in neuron-enriched hippocampal cultures. Cultures were exposed to 10 lM glutamate and were infected with either control or bcl-2-expressing vector. Glutamate caused a highly significant increase in lipid peroxidation, regardless of vector treatment (*p < 0.001, comparing bgal/0 glutamate with bgal/10 glutamate, or Bcl-2/0 glutamate with Bcl-2/10 glutamate; Tukey’s post-hoc test following two-way ANOVA). Viral vector treatment, however, had no effect on the extent of lipid peroxidation (n.s., by two-way ANOVA). n ¼ 23–28/ data point, taken from five different weekly culture preparations.

peroxidation (Fig. 3). Such superoxide appeared to be derived from mitochondria. As evidence, treatment of cultures with rotenone, which blocks mitochondrial superoxide production (Sensi et al. 1999; Saybasili et al. 2001), blocked glutamate-induced superoxide accumulation (superoxide accumulation above baseline induced by 10 lM glutamate: 35% ± 9; p < 0.01 by t-test, compared with 0 glutamate; superoxide accumulation above baseline induced by 10 lM glutamate plus 10 lM rotenone: 15% ± 6; n.s. as compared with 0 glutamate plus rotenone). In agreement with a prior report (Lawrence et al. 1996), overexpression of bcl-2 decreased the neurotoxicity of glutamate, causing an approximate doubling of the LD50 (Fig. 1). We then explored whether bcl-2 overexpression attenuated the ROS-related effects of glutamate. We observed that Bcl-2 completely blocked the superoxide accumulation induced by 10 lM glutamate, a concentration at which Bcl-2 also decreased the neurotoxicity (Fig. 2; superoxide production was maximal at this time point). In addition, Bcl-2 caused a trend towards decreased superoxide accumulation at 20 lM glutamate, a concentration at which Bcl-2 was not protective. Surprisingly, despite this effect, bcl-2 overexpression had no effect on the extent of lipid peroxidation induced by 10 lM glutamate (Fig. 3). We then examined the effects of bcl-2 overexpression on the actions of the ROS generator, adriamycin. Bcl-2 significantly decreased adriamycin-induced neurotoxicity, causing an approximate 35% increase in the LD50 (Fig. 4). As would be expected, adriamycin markedly increased superoxide accumulation (Fig. 5; note that the scale on the x-axis differs from Fig. 2), hydrogen peroxide accumulation (Fig. 6; as



Fig. 4 Adriamycin-induced neurotoxicity in hippocampal cultures treated with either control vector (m) or Bcl-2 (d). Bcl-2 caused a significant decrease in neurotoxicity (p < 0.001, F ¼ 22.68, d.f. ¼ 1/76, two-way ANOVA). n ¼ averaged 6/data point, taken from two different weekly culture preparations.

noted in the Methods section, similar measures were not carried out with glutamate, because the DCF assay is disrupted by the pH changes caused by the excitotoxin), and lipid peroxidation (Fig. 7). Despite the protective effects of Bcl-2 against adriamycin neurotoxicity, we observed no effects of overexpression on these ROS-related endpoints. Bcl-2 did not decrease adriamycin-induced superoxide accumulation (Fig. 5); as a positive control, under those same conditions, such accumulation was significantly blunted by the calciumchelating effects of EDTA. Of note, the positive effects of Bcl-2 on glutamate-induced superoxide accumulation were demonstrable at a single time point (Fig. 2). The failure of Bcl-2 to decrease adriamycin-induced superoxide accumulation was not due to having chosen the wrong single time point, as this was demonstrable over a range of times (Fig. 5b). Similarly, Bcl-2 had no effect on adriamycininduced hydrogen peroxide accumulation, under conditions where EDTA was effective (Fig. 6). Moreover, overexpression had no effect on the extent of lipid peroxidation under conditions where EDTA was protective (Fig. 7). We then examined whether Bcl-2 could protect against a necrotic insult whose damaging effects were unlikely to involve the generation of ROS. Specifically, we tested a model of permanent anoxia. Five to six hours of anoxia was significantly damaging to neurons in cultures transfected with control vector (Fig. 8). In contrast, Bcl-2 overexpression provided complete protection. Discussion

As discussed, the role of Bcl-2 in preventing neuron death is more complicated than the protein solely preventing apoptosis by antagonizing the actions of BAX. The interactions of Bcl-2 with the mitochondrial membrane are likely to be pivotal to its larger role. While this appears to be central to

Fig. 5 (a) Left: Effects of adriamycin and of differing viral vectors on superoxide accumulation in hippocampal cultures. Cultures were exposed to 40 lM adriamycin and either the control or bcl-2-expressing vector. Adriamycin caused a significant increase in superoxide accumulation, regardless of vector treatment (***p < 0.001, comparing bgal/ 0 adriamycin with bgal/40 adriamycin, or Bcl-2/0 adriamycin with Bcl-2/ 40 adriamycin; Tukey’s post-hoc test following two-way ANOVA). Viral vector treatment, however, had no effect on superoxide accumulation (n.s., by two-way ANOVA). Right: Effects of adriamycin and of EDTA on superoxide accumulation in hippocampal cultures. Cultures were exposed to 40 lM adriamycin with or without 1.5 mM EDTA. Adriamycin caused a significant accumulation of superoxide generation in the absence of EDTA (p < 0.01, when compared with control; Tukey’s post-hoc test), while treatment with EDTA prevented such adriamycininduced accumulation (*p < 0.05, Tuvkey’s post-hoc test). For unknown reasons, adriamycin was not as effective at increasing superoxide generation in the EDTA experiment as in the vector experiment; (p < 0.05, when compared with 40 lM adriamycin, bgal). n ¼ 12–13/group, from two weekly culture preparations. (b) The effects of adriamycin and indicated vectors on superoxide accumulation at 10, 20, 30 and 40 min Adriamycin significantly increased accumulation regardless of vector, but vector treatment did not alter accumulation (n.s). n ¼ 12–15/data point. d, bgal; s, bgal + 40 lM adriamycin; ., bcl-2; ,, bcl-2 + 40 lM adriamycin.



919

Fig. 6 Left: Effects of adriamycin and of differing viral vectors on hydrogen peroxide generation accumulation (as measured with DCF fluorescence) in hippocampal cultures. Cultures were exposed to 40 lM adriamycin and either the control or Bcl-2-expressing vector. Adriamycin caused a significant increase in DCF fluorescence, regardless of vector treatment **p < 0.01, ***p < 0.001, comparing bgal/0 adriamycin with bgal/40 adriamycin, or Bcl-2/0 adriamycin with Bcl-2/40 adriamycin, respectively; Tukey’s post-hoc test following two-way ANOVA). Viral

vector treatment, however, had no effect on superoxide accumulation (n.s., by two-way ANOVA). Right: Effects of adriamycin and of EDTA on hydrogen peroxide generation in hippocampal cultures. Cultures were exposed to 40 lM adriamycin with or without 1.5 mM EDTA. Adriamycin caused significant generation in the absence of EDTA (**p < 0.01, when compared to control; Tukey’s post-hoc test), while treatment with EDTA prevented such adriamycin-induced accumulation. n ¼ 6–19/ group, from three different weekly cultures.

Fig. 7 Left: Effects of adriamycin and of differing viral vectors on lipid peroxidation in hippocampal cultures. Cultures were exposed to 40 lM adriamycin and either the control or bcl-2-expressing vector. Adriamycin caused a significant increase in lipid peroxidation, regardless of vector treatment (***p < 0.001, comparing bgal/0 adriamycin with bgal/40 adriamycin, or Bcl-2/0 adriamycin with Bcl-2/ 40 adriamycin; Tukey’s post-hoc test following two-way ANOVA). Viral vector treatment, however, had no effect on superoxide accumula-

tion (n.s., by two-way ANOVA). Right: Effects of adriamycin and of EDTA on lipid peroxidation in hippocampal cultures. Cultures were exposed to 40 lM adriamycin with or without 1.5 mM EDTA. Adriamycin caused significant lipid peroxidation in the absence of EDTA (**p < 0.01, when compared with control; Tukey’s post-hoc test), while treatment with EDTA caused a significant diminution of this adriamycin effect (*p < 0.05). n ¼ 18–22, from five different weekly culture preparations.

Bcl-2 preventing BAX-induced release of cytochrome c from the mitochondria, the maintenance of mitochondrial function is also likely to help explain the protein’s capacity to protect

against necrotic cell death (Kane et al. 1993; Papadopoulos et al. 1998; Yang et al. 1998). This involvement of Bcl-2 in mitochondrial function has also prompted explorations of its



Fig. 8 Percentage neuron loss in cultures made permanently anoxic and treated with control vector expressing bgal alone, or experimental vector expressing Bcl-2. ***p < 0.001, by t-test. n ¼ 20–22/group, from four different weekly culture preparations.

potential role as an antioxidant. In an early version of this view, the proximity of Bcl-2 to the mitochondria suggested that it would be preferentially poised to quench ROS released from mitochondria (Hockenberry et al. 1993; this view predated the discovery of Bcl–2 interactions with BAX). More recently, there has emerged the view that Bcl-2, while not necessarily functioning as a classical ROS quencher, may decrease ROS production by mitochondria, secondary to its capacity to preserve both mitochondrial potential and function of the electron transport chain. Furthermore, Bcl-2 can decrease peroxidative damage to mitochondrial membrane (Bruce-Keller et al. 1998). As reviewed, the evidence of Bcl-2 attenuating oxidative damage, and for such actions being critical to its overall protective effects, is equivocal, especially in the nervous system. Our data suggest that Bcl-2’s protective effects in hippocampal neurons need not be heavily dependent on its capacity to protect against the generation and/or proximal consequences of ROS, in contrast to our prior findings in astrocytes (Papadopoulos et al. 1998). Glutamate neurotoxicity We initially observed Bcl-2 to protect against glutamatergic excitotoxicity in hippocampal cultures. This protection involved an approximate doubling of the LD50, such that at its most efficacious, Bcl-2 decreased neuron death more than 50% (at 10 lM glutamate). Given that herpes viral vectors have a strong preference for infecting neurons over glia in primary cultures (Ho et al. 1995) and that these studies utilized neuron-enriched cultures, the protective effects of Bcl-2 were overwhelmingly likely to be due to direct actions within infected neurons, rather than secondary to glial effects. This should be contrasted with our previous observation that selective bcl-2 overexpression in astrocytes cocultured with wild-type neurons does afford protection from combined oxygen glucose deprivation, an injury largely dependent on the activation of glutamate receptors (Xu et al. 1999).

These protective Bcl-2 effects agree with prior reports showing neuroprotection by Bcl-2 against excitotoxins, hypoglycemia and adriamycin in primary cultures derived from a number of brain regions (Jia et al. 1996; Lawrence et al. 1996; McLaughlin et al. 2000; Tamatani et al. 2000) and against in vivo models of excitotoxicity, hypoxia–ischemia, ROS generators, or mechanical trauma (Linnik et al. 1995; Lawrence et al. 1996, 1997; Antonawich et al. 1999; Yamada et al. 1999; Phillips et al. 2000; Shimazaki et al. 2000). Moreover, we observed that such protection was accompanied by a complete block of glutamate-induced superoxide accumulation at the time point where such accumulation is maximal post-glutamate. Mitochondria appear to be a major source of such accumulation during excitotoxic insults (Dugan et al. 1995), probably secondary to the disruption of mitochondrial potential; supporting this, we observed that rotenone, which blocks mitochondrial superoxide production, blocked the effects of glutamate on this endpoint. Thus, the Bcl-2 effect is commensurate with the broadly protective array of effects of Bcl-2 in mitochondria [nonetheless, in this particular culture system, we observe that Bcl-2 does not alter glutamate-induced declines in mitochondrial potential (manuscript in preparation)]. As an alternative or additional mechanism, Bcl-2 can increase the activity or levels of antioxidants, or optimize their subcellular distribution (Kane et al. 1993; Steinman 1995; Ellerby et al. 1996; Papadopoulos et al. 1998; Voehringer et al. 1998; Lee et al. 2001); insofar as this involves an increase in SOD activity, this should lead to a reduction in superoxide accumulation. Finally, the effects of Bcl-2 on availability of substrates such as GSH or GSSG are unknown, although the glutamatergic excitotoxin kainic acid does not alter levels of either in this culture system (Patel et al. 2002). Glutamate-induced superoxide production appears to be an important contributor to the subsequent neuron death (Dugan et al. 1995) and administration or overexpression of a variety of antioxidants (SOD, vitamin E, 21-amino steroids, oxypurinol, glutathione peroxidase) can decrease glutamatergic injury in the brain (Acosta et al. 1987; Monyer et al. 1990; Chan et al. 1991; Lin and Phillis 1991; Lafon-Cazal et al. 1993). Therefore, the decreased accumulation could help explain the neuroprotective actions of Bcl-2. Thus, while there is little evidence that Bcl-2 is acting in this case as a classical quencher of ROS, some of its protective functions may revolve around it indirectly decreasing ROS accumulation. Our data also indicate that the glutamate-induced superoxide accumulation does not necessarily lead to peroxidative damage to lipid membranes, as Bcl-2 reduced the former without altering the latter. Excitotoxin-induced peroxidative damage to cell membranes probably reflects ROS generation in the cytosol and membrane itself. This would typically be secondary to calcium-induced activation of xanthine oxidase, phospholipase A2, and NOS, and the generation of hydrogen peroxide, hydroxyl radicals, and peroxynitrites. In addition,



superoxide can be generated in the cytosol by oxygenases, and thus could readily contribute to oxidative damage to the cell membrane. However, our rotenone data suggest that the superoxide generated by glutamate in our insult model is predominately mitochondrial in origin. Thus, a change in superoxide generation from mitochondria is not likely to impact the endpoint of lipid peroxidation in the cell membrane. It has been speculated that Bcl-2 can protect mitochondrial membrane from lipid peroxidation (BruceKeller et al. 1998). Such peroxidation could be generated by the superoxide and be diminished by Bcl-2 in the present case; however, any such peroxidation would be in amounts below the level of detection in our assay. Our data also demonstrate that in this model, glutamateinduced lipid peroxidation is not likely to be playing an obligatory role in the neurotoxicity, insofar as bcl-2 overexpression reduced neurotoxicity without altering the extent of peroxidation. This is in contrast to the situation in ischemic brain injury, in which lipid peroxidation is thought to play a more central role in the neuron death (Traystman et al. 1991; Chan 1996; Liu et al. 1998). However, it is quite plausible that the peroxidative damage could impair functional recovery in surviving neurons. In support of this, under a number of circumstances, bcl-2 overexpression can spare neurons from insult-induced death, but not from insult-induced dysfunction (McLaughlin et al. 2000; Dumas et al. 2000). Adriamycin neurotoxicity Bcl-2 can decrease the toxicity of a number of insults that are heavily or entirely oxidative in nature, such as adriamycin, paraquat, hydrogen peroxide, or 6-OHDA (Oh et al. 1995; Lawrence et al. 1996; Lezoualc’h et al. l996; Marin et al. 1996; Bruce-Keller et al. 1998; ; Hochman et al. 1998; Yang et al. 1998Yamada et al. 1999; Luc Cadet et al. 2000). We examined whether Bcl-2 could protect against an insult that is overwhelmingly oxidative in nature, and if any such protection arose as a result of reducing ROS accumulation or oxidative damage. We utilized adriamycin (doxorubicin), a potent pro-oxidant commonly used in the treatment of malignant tumors, which is toxic to cultured neurons in the lM range. As expected, the neurotoxin generated a considerable oxidative challenge; at its approximate LD50 (40 lM), adriamycin caused a 200% increase in superoxide accumulation (in contrast, glutamate, at its LD50 of 10 lM, caused a 15% increase). We then observed that bcl-2 overexpression decreased adriamycin neurotoxicity, although to a lesser extent than against glutamate. Despite these neuroprotective effects, Bcl-2 had no effect on the accumulation of superoxide, hydrogen peroxide at a range of time points, or on the extent of lipid peroxidation. This dissociation between protecting from ROS while not decreasing a measure of oxidative damage is reminiscent of the finding that in a peripheral cell

921

line, Bcl-2 decreased hydrogen peroxide toxicity without decreasing oxidative damage to lipids, DNA or protein (Lee et al. 2001). As discussed, the ability of Bcl-2 to lessen glutamate-induced superoxide accumulation is, most parsimoniously, a consequence of the protein’s actions at mitochondria. Commensurate with this, we observe that Bcl-2 also blocks cytochrome c release from mitochondria in this insult model (manuscript in preparation). The lack of an effect of Bcl-2 against the far greater superoxide accumulation induced by adriamycin suggests either that (i) the superoxide is derived from mitochondria but exceeds Bcl-2’s capacity to constrain such accumulation, and/or (ii) the superoxide is predominately derived from non-mitochondrial sites not subject to Bcl-2’s effects. Recent work has emphasized the potential role of nitrosylative rather than oxidative damage in cell death. Along these lines, in a case where Bcl-2 protected against hydrogen peroxide without decreasing oxidative damage, it decreased 3-nitrotyrosine levels (Lee et al. 2001). Thus, the same may hold in the present study. ROS accumulation is thought to be one of the signals initiating injury-induced apoptosis (e.g. the translocation of BAX to the mitochondria). Our data suggest that Bcl-2 reduces adriamycin neurotoxicity by one of two routes. First, it may protect downstream of the oxidative realm, with blocking of cytochrome c release being the most implicated, but not sole, possible mechanism. Second, it is currently not known whether adriamycin causes nitrosylative damage and whether such damage can also initiate apoptosis. If so, Bcl-2 might be blocking the nitrosylation pathway. Thus, while protecting against an ROS generator, such protection may not be centered in the oxidative realm, a point emphasized previously (Oh et al. 1995). Anoxic neurotoxicity Glutamatergic excitotoxicity represents a model in which ROS generation is likely to contribute at least somewhat to damage, while adriamycin toxicity is overwhelmingly oxidative in nature. Anoxia, in contrast, represents an insult in which ROS play a minimal role, if any, in the neuron death. As noted, Bcl-2 protects against insults under anaerobic conditions in peripheral cell types (Jacobson and Raff 1995; Shimizu et al. 1995), and this has been strongly interpreted as evidence against Bcl-2’s protective actions being solely antioxidant in nature. We observed that Bcl-2 overexpression blocked the toxicity induced by permanent anoxia in these cultures. Broadly, this suggests the same non-oxidative facets of Bcl-2 actions within the CNS. In conclusion, these data suggest a mixed picture concerning the antioxidant actions of Bcl-2. The glutamate data suggest that, while Bcl-2 can have antioxidant actions, they may not impact one of the major endpoints of oxidative damage. The adriamycin data, moreover, suggest that, while Bcl-2 can protect against a classical ROS



generator, such protection may arise from actions either lateral to, or downstream of, specific oxidative events. Finally, the anoxia data show that Bcl-2 can protect against an insult likely to have little or no elements of ROS generation. Thus, in this model system, the antioxidant actions of Bcl-2 may not be either necessary or sufficient to explain its protective actions. Acknowledgements Funding was provided by NIH PO1 NS27520 (RG and RS), a TRDRP State of California grant (RS), the International Anesthesia Research Society (RG) and a URO Grant (CB). Technical assistance was provided by Martin Brown, Nick Denko, Adrian Dunn, Pedram Ghafourifar, Mark Mattson, Stefano Sensi and John Weiss.

References Acosta D., Kass I. and Cottrell J. (1987) Effect of a-tocopherol and free radicals on anoxic damage in the rat hippocampal slice. Exp. Neurol. 97, 607–614. Antonawich F., Federoff H. and Davis J. (1999) Bcl-2 transduction, using a HSV ampolicon, protects hippocampal neurons from transient global ischemia. Exp. Neurol. 156, 130–136. Brooke S. M., Chan R., Howard S. and Sapolsky R. (1997) Endocrine modulation of the neurotoxicity of gp120: implications for AIDSrelated dementia complex. Proc. Natl Acad. Sci. USA 94, 9457– 9461. Bruce-Keller A., Begley J., Fu W., Butterfield D., Bredesen D., Hutchins J., Hensley K. and Mattson M. (1998) Bcl-2 protects isolated plasma and mitochondrial membranes against lipid peroxidation induced by hydrogen peroxide and amyloid b-peptide. J. Neurochem. 70, 31–38. Chan P. (1996) Role of oxidants in ischemic brain damage. Stroke 27, 1124–1129. Chan P., Yang G., Chen S., Carlson E. and Epstein C. (1991) Coldinduced brain edema and infarction are reduced in transgenic mice overexpressing CuZn-superoxide dismutase. Ann. Neurol. 29, 482– 489. DeLuca N., McCarthy A. and Schaffer P. (1985) Isolation and characterization of deletion mutants of herpes simplex virus type I in the gene encoding immediate-early regulatory protein ICP4. J. Virol. 56, 558–570. Dugan L., Sensi S. and Choi D. (1995) Mitochondrial production of reactive oxygen species in cortical neuron sfollowing exposure to N-methyl-D-aspartate. J. Neurosci. 15, 6377–6388. Dumas T., McLaughlin J., Ho D., Lawrence M. and Sapolsky R. (2000) Gene therapies that enhance hippocampal neuron survival after an excitotoxic insult are not equivalent in their ability to maintain synaptic transmission. Exp. Neurol. 166, 180–188. Ellerby L., Ellerby H., Park S., Hlleran A., Murphy a Fiskum G., Kane D., Testa M., Kayalar C. and Bredesen D. (1996) Shift of the cellular oxidation-reduction potential in neural cells expressing Bcl-2. J. Neurochem. 67, 1259–1266. Gardner A., Xy F., Fady C., Sarafian T., Tu Y. and Lichtenstein A. (1997) Evidence against the hypothesis that BCL-2 inhibits apoptosis thorugh an anti-oxidant effect. Cell Death Different. 4, 487–494. Giardino I., Edelstein D. and Brownlee M. (1996) BCL-2 expression or antioxidants prevent hyperglycemia-induced formation of intra-

cellular advanced glycation endproducts in bovine endothelial cells. J. Clin. Invest. 97, 1422–1428. Green D. and Reed J. (1998) Mitochondria and apoptosis. Science 281, 1309–1313. Ho D. Y. (1994) Amplicon-based herpes simplex virus vectors. Meth. Cell Biol. 43, 191–210. Ho D. Y., Fink S., Lawrence M., Meier T., Saydam T., Dash R. and Sapolsky R. (1995) Herpes simplex virus vector system: analysis of its in vivo and in vitro cytopathic effects. J. Neurosci. Meth. 57, 205–216. Hochman A., Sternin H., Gorodin S., Korsmeyer S., Ziv I., Melamed E. and Offen D. (1998) Enhanced oxidative stress and altered antioxidants in brains of Bcl-2-deficient mice. J. Neurochem. 71, 741– 748. Hockenberry D., Oltvai Z., Yin X., Milliman C. and Korsmeyer S. (1993) Bcl-2 functions in an antioxidant pathway to prevent apoptosis. Cell 75, 241–248. Jacobson M. and Raff M. (1995) Programmed cell death and Bcl-2 protection in very low oxygen. Nature 374, 814–816. Jia W., Wang Y., Qiang D., Tufaro F., Remington R. and Cynader M. (1996) A bcl-2 expressing viral vector protects cortical neurons from excitotoxictiy even when administered several hours after the toxic insult. Mol Brain Res. 42, 350–356. Kane D., Sarafian T., Anton R., Hahn H., Gralla E., Valentine J., Ord T. and Bredesen D. (1993) Bcl-2 inhibition of neural death: decreased generation of reactive oxygen species. Science 262, 1274–1277. Kuypers F., van den Berg J., Schalwijk B., Roelofsen B. and Op den Kamp J. (1987) Parinaric acid as a sensitive fluorescent probe for the determination of lipid peroxidation. Biochim. Biophys. Acta 921, 266–274. Lafon-Cazal M., Pietri S., Culcasi M. and Bockaert J. (1993) NMDAdependent superoxide production and neurotoxicity. Nature 364, 535–537. Lagrange P., Livertoux H., Grassiot M. and Minn A. (1994) Superoxide anion production during monoelectronic reduction of xenobiotics by preparations of rat brain cortex, microvessels, and choroid plexus. Free Rad. Biol. Med. 17, 355–359. Lawrence M., Ho D., Sun G., Steinberg G. and Sapolsky R. (1996) Overexpression of bcl-2 with herpes simplex virus vectors protects CNS neurons against neurological insults in vitro and in vivo. J. Neurosci. 16, 486–495. Lawrence M. S., Sun G., Ho D., McIntosh L., Kunis D., McLaughlin J., Sapolsky R. and Steinberg G. (1997) Herpes simplex viral vectors expressing Bcl-2 are neuroprotective when delivered following a stroke. J. Cereb. Blood Flow Metab. 17, 740–747. Lebel C., Ali S., McKee M. and Bondy S. (1990) Organometal-induced incrase in oxygen reactive species: the potential of dichlorofluorescein diacitate as an index of neurotoxic damage. Toxicol. Appl. Pharm. 104, 17–25. Lee M., Hun D., Marshall K., Ellerby L., Bredesen D., Jenner P. and Halliwell B. (2001) Effect of overexpression of bcl-2 on cellular oxidative damage, nitric oxide production, antioxidant defenses, and the proteasome. Free Rad. Biol. Med. 31, 1550–1558. Lezoualc’h F., Rupprecht R., Holsboer F. and Behl C. (1996) Bcl-2 prevents hippocampal cell death induced by the neuroleptic drug haloperidol. Brain Res. 738, 176–181. Lin Y. and Phillis J. (1991) Oxypurinol reduces focal ischemic brain injury in the rat. Neurosci. Lett. 126, 187–190. Linnik M., Zahos P., Geschwind M. and Federoff H. (1995) Expression of bcl-2 from a defective herpes simplex virus-1 vector limits neuronal death in focal cerebral ischemia. Stroke 26, 1670–1677/. Liu Y., Rosenthal R., Haywood Y., Miljkovic-Lolic M., Vanderhoek J. and Fiskum G. (1998) Normoxic ventilation ofollowing cardiac



arrest reduces oxidation of brain lipids and improves neurological outcomes. Stroke 29, 1679–1686. Luc Cadet J., Harrington B. and Ordonez S. (2000) Bcl-2 overexpression attenuates dopamine-induced apoptosis in an immortalized neural cell line by suppressing the production of reactive oxygen species. Synapse 35, 228–235. McLaughlin J., Roozendaal B., Gupta A., Ajilore O., Dumas T., Hsieh J., Ho D., Lawrence M., McGaugh J. and Sapolsky R. (2000) Sparing neuronal function post-seizure with gene therapy. Proc. Natl Acad. Sci. USA 97, 12804–12809. Marin M., Fernandez A., Bick R., Brisbay S., Buja L., Snuggs M., McConkey D., Eschenbach A., Keating J. and McDonnell T. (1996) Apoptosis suppression by bcl-2 is correlated with the regulation of nuclear and cytosolic calcium. Oncogene 12, 2259– 2266. Monyer H., Hartley D. and Choi D. (1990) 21-Aminosteroids attenuate excitotoxic neuronal injury in cortical cell cultures. Neuron 5, 121– 128. Murphy A., Bredesen D., Cortopassi G., Wang E. and Fiskum G. (1996) Bcl-2 potentiates the maximal calcium uptake capacity of neural cell mitochondria. Proc. Natl Acad. Sci. USA 93, 9893–9898. Myers K., Fiskum G., Liu Y., Simmens S., Bredesen D. and Murphy A. (1995) Bcl-2 protects cells from cyanide/aglycemia-induced lipid oxidation, mitochondrial injury, and loss of viability. J. Neurochem. 65, 2432–2439. Oh J., Choi W., Kim J., Seo J., O’Malley K. and Oh Y. (1998) Overexpression onf HA-Bax but not bcl-2 or bcl-XL attenuates 6-OHDA-induced neuronal apoptosisl. Exp. Neurol. 154, 193– 200. Oh Y., Wong S., Moffat M. and O’Malley K. (1995) Overexpresion of bcl-2 attenuates MPP+, but not 6-OHDA, induced cell death in a dopaminergic neuronal cell line. Neurobiol. Dis. 2, 157–164. Papadopoulos M., Koumenis I. and Xu 1 Giffard R. (1998) Potentiation of murine astrocyte antioxidant defence by bcl-2: protection in part reflects elevated glutathione levels. Eur. J. Neurosci. 10, 1252– 1258. Patel R., McIntosh L., McLaughlin J., Brooke S., Nimon V. and Sapolsky R. (2002) Disruptive effects of glucocorticoids on glutathione peroxidase biochemistry in hippocampal cultures. J. Neurochem. 82, 118–125. Phillips R., Lawrence M., Ho D. and Sapolsky R. (2000) Limitations in the neuroprotective potential of gene therapy with Bcl-2. Brain Res. 859, 202–209.

923

Rosse T., Olivier R., Monney L., Rager M., Conus S., Fellay I., Jansen B. and Borner C. (1998) Bcl-2 prolongs cell survival after Baxinduced release of cytochrome c. Nature 391, 496–499. Saybasili H., Yuksel M., Haklar G. and Yalcin A. (2001) Effect of mitochondrial electron transport chain inhibitors on superoxide radical generation in rat hippocampal and striatal slices. Antioxidants Redox Signaling 3, 1099–1104. Sensi S., Yin H., Carriedo S., Rao S. and Weiss J. (1999) Preferential Zn2+ influx through Ca2+-permeable AMPA/kainite channels triggers prolonged mitochondrial superoxide production. Proc. Natl Acad. Sci. USA 96, 2414–2419. Shimazaki K., Uirabe M., Monahan J., Ozawa K. and Kawai N. (2000) AAV vector-mediated bcl-2 gene tranfser into post-ischemic gerbil brain in vivo: prospects for gene therapy of ischemia-induced neuronal death. Gene Ther. 7, 1244–1251. Shimizu S., Eguchi Y., Kosaka H., Kamiike W., Matsuda H. and Tsujimoto Y. (1995) Prevention of hypoxia-induced cell death by Bcl-2 and Bcl-Xl. Nature 374, 811–814. Steinman H. (1995) The Bcl-2 oncoprotein functions as a pro-oxidant. J. Biol. Chem. 270, 3487–3493. Tamatani M., Mitsuda N., Matsuzaki H., Okado H., Miyake S., Vitek M., Yamaguchi A. and Tohyama M. (2000) A pathway of neuronal apoptosis induced by hypoxia-reoxygenatrion: roles of NF-kB and Bcl-2. J. Neurochem. 75, 683–690. Traystman R., Kirsch J. and Koehler R. (1991) Oxygen radical mechanisms of brain injury following ischemia and reperfusion. J. Appl. Physiol. 71, 1185–1195. Voehringer D., McConkey D., McDonnell T., Brisbay S. and Meyn R. (1998) Bcl-2 expression causes redistribution of glutathione to the nucleus. Proc. Natl Acad. Sci. USA 95, 2956–2961. Wang H., Rapp U. and Reed J. (1996) Bcl-2 targets the protein kinase Raf-1 to mitochondria. Cell 87, 629–636. Xu L., Lee A. and Giffard R. (1999) Overexpression of bcl-2, bcl-x or hsp70 in murine cortical astrocytes reduces injury of co-cultured neurons. Neurosci. Lett. 277, 193–197. Yamada M., Oligino T., Mata M., Goss J. R., Glorioso J. C. and Fink D. J. (1999) HSV vector-mediated expression of Bcl-2 prevents 6-OHDA induced degeneration of neurons in the substantia nigra in vivo. Proc. Natl Acad. Sci. USA 96, 4078–4083. Yang L., Matthews R., Schulz J., Klockgether T., Liao A., Martinou J., Penney J., Hyman B. and Beal M. (1998) 1-Methyl-4-phenyl1,2,3,6-tetrahydropyride neurotoxicity is attenuated in mice overexpressing Bcl-2. J. Neurosci. 18, 8145–8152.
