|Home | About | Journals | Submit | Contact Us | Français|
Cerebral cortical neuron degeneration occurs in brain disorders manifesting throughout life, but the mechanisms are understood poorly. We used cultured embryonic mouse cortical neurons and an in vivo mouse model to study mechanisms of DNA damaged-induced apoptosis in immature and differentiated neurons. p53 drives apoptosis of immature and differentiated cortical neurons through its rapid and prominent activation stimulated by DNA strand breaks induced by topoisomerase-I and -II inhibition. Blocking p53-DNA transactivation with α-pifithrin protects immature neurons; blocking p53-mitochondrial functions with μ-pifithrin protects differentiated neurons. Mitochondrial death proteins are upregulated in apoptotic immature and differentiated neurons and have nonredundant proapoptotic functions; Bak is more dominant than Bax in differentiated neurons. p53 phosphorylation is mediated by ataxia telangiectasia mutated (ATM) kinase. ATM inactivation is antiapoptotic, particularly in differentiated neurons, whereas inhibition of c-Abl protects immature neurons but not differentiated neurons. Cell death protein expression patterns in mouse forebrain are mostly similar to cultured neurons. DNA damage induces prominent p53 activation and apoptosis in cerebral cortex in vivo. Thus, DNA strand breaks in cortical neurons induce rapid p53-mediated apoptosis through actions of upstream ATM and c-Abl kinases and downstream mitochondrial death proteins. This molecular network operates through variations depending on neuron maturity.
Degeneration of neurons in the cerebral cortex occurs in many human disorders manifesting acutely or chronically throughout life and causes or contributes to mental retardation, memory decline, dementia, and other functional impairments. In newborns and adults, the cerebral cortex is a major site of neurodegeneration after hypoxia-ischemia, stroke, and trauma (Martin et al. 1998; Zhang et al. 2005; Martin 2008). Cerebral cortical neuron loss occurs in Alzheimer's disease (AD), Parkinson's disease (PD), amyotrophic lateral sclerosis (ALS), Huntington's disease (HD), acquired immunodeficiency syndrome, and neurodevelopmental disorders (Hedreen et al. 1991; Gomez-Isla et al. 1997; Martin 2001; Johnston 2004; Martin et al. 2005; McArthur et al. 2005). The degeneration of neocortical neurons in AD is regionally specific (Gomez-Isla et al. 1997) and might be instigated by Aβ protein toxicity and synapse loss (Sze et al. 1997; Zhang et al. 2002). Loss of pyramidal neurons in the pre-supplementary motor cortex is found in PD (MacDonald and Halliday 2002). Upper motor neurons are cortical neurons; substantial loss of the pyramidal Betz cells occurs in ALS (Nihei et al. 1993; Shaikh and Martin 2002). In HD, neurons in layer 6 of frontal cortex degenerate (Hedreen et al. 1991). The protection of neocortical neurons against degeneration is critical because postnatal regeneration of these neurons in primates through neurogenesis is very limited (Rakic 2002; Breunig et al. 2007). It is therefore necessary to identify the signal transduction networks and specific molecules that mediate death of cerebrocortical neurons because these molecules can become specific targets for the discovery of drugs with potential to block neurodegeneration in many human disorders occurring during development and adulthood.
Apoptosis, a form of programmed cell death, driven by the tumor suppressor p53 is involved in the degeneration of cerebral cortical neurons in a variety of cell and animal models of neuronal death and in some human neurological disorders (Morrison et al. 2003; Martin et al. 2005; Zhang et al. 2005; Northington et al. 2007). During neurodevelopment, telencephalic neural precursor cells with genotoxic stress undergo apoptosis requiring p53 (D'Sa-Eipper et al. 2001). Cortical neuron degeneration occurring after hypoxia-ischemia, stroke, and trauma has a considerable apoptotic component (Hou and MacManus 2002; Zhang et al. 2005); furthermore, in human newborns with hypoxic-ischemic insults due to complications during parturition, neocortical neuron death is associated with oxidative stress and p53 activation (Martin 2008). In adult postmortem human brain, p53 is found in subsets of neocortical and hippocampal neurons in AD (De la Monte et al. 1997; Martin et al. 2005) and in motor cortical neurons in ALS (Martin 2000). p53 and the proapoptotic Bcl-2 family member Bax can mediate toxicity of Aβ to cultured human primary cortical neurons (Zhang et al. 2002), possibly by direct Aβ fibril activation of the p53 promoter (Ohyagi et al. 2005). Part of the activation sequence for p53-mediated apoptosis in cells is the accumulation of DNA damage (Jeffers et al. 2003).
DNA damage has been implicated in the mechanisms of neurodegeneration in age-related disease, cerebral ischemia, and brain trauma. For example, DNA single-strand breaks (SSBs) in neocortex of people with AD are increased 2-fold compared with control subjects (Mullaart et al. 1990), and many cortical neurons in cases of AD show DNA-SSBs and DNA double-strand breaks (DSBs) in situ (Adamec et al. 1999). The neurotoxicity of Aβ in cultured human cortical neurons involves DNA damage triggered by reactive oxygen species (Zhang et al. 2002). Aβ might also exert effects in the AD brain by triggering DNA synthesis in forebrain neurons (Yang et al. 2001). After focal ischemia, DNA-SSBs accumulate very quickly, within 15 min, in cerebral cortex at levels 20-fold higher than baseline (Huang et al. 2000). These DNA lesions could contribute to delayed neuronal death (Chen et al. 1997) involving DNA synthesis prior to cell death and mitotic catastrophe (Kuan et al. 2004). Similarly, neurons accumulate DNA-SSBs after cortical trauma (Clark et al. 2001). In the developing brain, migrating newborn neurons can sustain oxidative DNA damage and, if not repaired, undergo apoptosis (Narasimhaiah et al. 2005).
Nevertheless, the role of DNA damage as a definitive upstream initiator of postmitotic neuron degeneration or a consequence of the degeneration is vague. In this study we examined the mechanisms of DNA damage-triggered apoptosis of cortical neurons in a cell culture model with confirmed DNA strand breaks and developed a new mouse model of topoisomerase poison-induced apoptosis of cortical neurons. Exposure of neurons to topoiosmerase inhibitors, such as camptothecin (CPT) or etoposide (ETOP), is described frequently as a DNA-damaging stimulus (Park et al. 1998; Keramaris et al. 2000). It is believed, although hitherto unconfirmed, that CPT and ETOP trigger apoptosis in neurons through DNA damage (Endokido et al. 1996; Morris and Geller 1996; Park et al. 1998), but the mechanisms of CPT and ETOP toxicity and cell death in neurons have not been identified clearly, and comparisons in cell death signaling pathways in immature and well-differentiated neurons have not been studied fully (Lesuisse and Martin 2002a, 2002b). We tested the hypotheses, using a combination of gene-disrupted mice and small-molecule drugs, that cerebral cortical neurons at different stages of differentiation have different propensities for DNA damage accumulation and utilize different upstream and downstream molecular signaling pathways in the mediation of cell death in response to DNA damage. This work reveals that the therapeutic protection of neurons in the cerebral cortex of the immature and mature brain could be based on different key molecular targets.
To study the effects of DNA damage in postmitotic cells we used primarily a neuronal cell culture model. This approach has many advantages. An essentially pure neuron system is studied without appreciable contamination of cycling cells, such as glia. Also, neurons from the same source can be maintained and differentiated identically over long-term to examine cell death signaling mechanisms in immature cortical neurons or mature, synaptically integrated cortical neurons (Lesuisse and Martin 2002a). The immature and differentiated states of these neurons have been defined previously based on morphology, ultrastructure, and the levels of synaptic and cytoskeletal proteins, glutamate receptors, and metabolic proteins (Lesuisse and Martin 2002a). Furthermore, embryonic cortical neurons can be harvested from mice with targeted gene deletions to identify specific molecules that regulate apoptosis. Timed-pregnant wild-type mice (C57BL/6 strain, The Jackson Laboratory, Bar Harbor, ME) and mice with targeted gene disruptions of p53, bax, bak, or ATM (ataxia telangiectasia mutated) (The Jackson Laboratory) were used. Mice with homozygous null deletion of p53 (B6.129S2-Trp53tm1Tyj; stock number 002101) or bak (B6.129-Bak1tm1Thsn; stock number 004183) or with heterozygous null deletion of bax (B6.129X1-Baxtm1Sjk; stock number 002994) or ATM (129S6/SvEvTac-Atmtm1Awb; stock number 002753) were purchased as breeder pairs to establish colonies of these mice for the generation of primary embryonic cortical neuron cultures. Animal care was provided in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.
Mouse cortical neuron cultures were prepared as described previously (Lesuisse and Martin 2002a, 2002b). The Animal Care and Use Committee of the Johns Hopkins University School of Medicine approved the animal protocol. Deeply anesthetized (isoflurane:oxygen:nitrous oxide, 1:33:66) pregnant dams underwent cesarean section at 15–16 days gestation for harvesting of mouse embryos. Under a surgical microscope, the brains were dissected from the embryos and the cerebral cortices were isolated carefully and stripped of membranes. The cortices were dissociated by treatment (20 min at 37 °C) with 0.25% trypsin (Invitrogen, Carlsbad, CA) followed by triturating with a fire-polished Pasteur pipette. Cortical tissue was pooled from wild-type, p53−/−, and bak−/− null embryos for their respective experiments. For homozygous matings ~16 pregnant mice were used for each genotype. Cortical tissues from bax and ATM gene deficient mouse embryos were not pooled because they were derived by mating heterozygous mice (e.g., embryos could be bax+/+, bax+/−, or bax−/−). For heterozygous matings ~18 pregnant mice were used. Cortical tissue (both hemispheres) from each of these embryos was digested and plated individually while each embryo was genotyped by PCR (according to the protocol and designated primers from The Jackson Laboratory). Cells (~106) were plated onto 35 mm tissue culture dishes coated with 33 μg/ml poly-D-Lysine (Sigma, St Louis, MO) and human fibronectin (Sigma) or onto poly-D-Lysine and laminin coated 12 mm glass coverslips (BD Biosciences, Bedford, MA). The cells were plated in Neurobasal medium (Invitrogen) supplemented with B27 (Invitrogen), 25 μM β-mercaptoethanol (Invitrogen) and streptomycin/penicillin (Invitrogen). Three days after plating, 50% of the medium was changed and subsequently changed every 5 days. This cell culture is nearly neuron-pure because glial contamination of the cell culture is very low (Lesuisse and Martin 2002a). Nevertheless, the cultures prepared in this series of experiments were screened for glial contamination and found to be low as well (<3% oligodendrocytes, <2% astrocytes, <1% microglia), consistent with the inability of serum-free/low glutamine media to support glial cells. Most published in vitro studies of mechanisms of neuronal apoptosis triggered presumably by DNA damage and mediated by p53 have used cultures of primary embryonic or newborn neurons at 2–4 days in vitro (DIV) (Xiang et al. 1998; Keramaris et al. 2003). DIV2–DIV4 neurons are very immature and undifferentiated neurons with very few established synapses which emerge typically at about DIV5 but do not become robust until about DIV10, and then synaptic levels achieve maturity at about DIV15 (Lesuisse and Martin 2002a). In this study we compared apoptotic responses of immature and undifferentiated DIV5 neurons to more mature, highly differentiated DIV25 neurons. For every experiment that was done at different in vitro ages, the neurons in the different differentiation groups were prepared from the same original batches of dissociated cerebral cortices, so that appropriate comparisons could be made (Lesuisse and Martin 2002b). Each experimental data set reported was done at least in triplicate.
To study the molecular regulation of cell death in cortical neurons, DNA topoisomerases were inhibited with CPT or ETOP. These 2 poisons were used because they cause DNA damage through different mechanisms (Nitiss and Wang 1996). CPT and ETOP inhibit DNA topoisomerase-I (Topo-I) and DNA topoisomerase-II (Topo-II), respectively (Nitiss and Wang, 1996). Topo-I and Topo-II are nuclear enzymes that modify DNA topology. Topo-I catalyzes DNA single-strand cleavage and reunion of the phosphodiester backbone to allow relaxation of supercoiling in response to a torsional stress generated by replication proteins at replication forks during DNA synthesis or transcriptional proteins during RNA synthesis. Topo-II is a DNA gyrase that cuts simultaneously both strands of the DNA helix, allowing another DNA duplex to pass through the break, followed by religation. CPT and ETOP poisoning of non-neural cells is believed to involve stabilization of Topo-I–DNA complexes or Topo-II–DNA complexes, respectively, and the trapping these enzymes in an intermediate state to disallow religation of the DNA backbone and thus causing the formation of DNA strand breaks (Nitiss and Wang 1996; Nieves-Neira and Pommier 1999). Topo-I is the only know target of CPT, whereas Topo-II is the only known target of ETOP (Bendixen et al. 1990). In cycling cells, CPT and ETOP are thought to be S-phase–specific antineoplastic agents. CPT induces initially the formation of DNA-SSBs, whereas ETOP induces DNA-DSBs (Nitiss and Wang 1996). CPT can also cause premature termination of transcription and depletion of cellular Topo-I in non-neural cells through its ubiquitination and proteasomal degradation (Bendixen et al. 1990; Nieves-Neira et al. 1999). Their mechanisms of action in postmitotic cells are not as well characterized as in cycling cells. The actions of topoisomerase drugs in postmitotic neural cells could be different from those occurring in cycling cells as illustrated by the different actions of taxol in neuronal cells and non-neuronal cells (Figueroa-Masot et al. 2001). CPT induces apoptosis in immature mixed neuron-astrocyte cultures (Morris and Geller 1996) and in relatively pure immature neuron cultures (Lesuisse and Martin 2002b). ETOP induces apoptosis in cultured embryonic rat brain neurons (Nakajima et al. 1994) and mouse postnatal day 15–16 cerebellar neurons (Endokido et al. 1996). Thus, their S-phase–specific mechanisms of action in postmitotic cells can be questioned or the immature neuronal cultures used are not postmitotic. CPT (Sigma) was dissolved in double-distilled water supplemented with sodium hydroxide and heated at 55 °C for 30 min to dissolve completely the drug at a final concentration of 50 mM. ETOP (Sigma) was dissolved in 4% methyl-β-cyclodextrin (Sigma) in PBS. CPT and ETOP were further diluted in Neurobasal medium. Depending on the particular experimental paradigm, at DIV5 or DIV25, mouse cortical neurons were treated with varying concentrations of CPT or ETOP at 1, 10, or 100 μM for 1, 2, or 4 h or for 4, 8, 15, or 24 h. Low-dose experiments with 0.5 μM CPT or ETOP were conducted over 24, 48, and 72 h.
Pharmacological manipulations can be done on cultured neurons without major delivery and bioavailability concerns, except solubility and cell permeability of drug. Several compounds were used to examine mechanisms of neuronal cell death and to test for neuroprotective actions, including inhibitors of caspase-3, p53, Bax channel, the voltage-dependent anion channel (VDAC) that is a major component of the mitochondrial permeability transition pore (Bernardi et al. 2006), ATM, and c-Abl (Table 1). The justification for using each of these drugs is given in the results. The drugs were prepared as 10 mM stock solutions in dimethyl sulfoxide or water and then diluted in media and used at final concentrations of 1–100 μM in cultures. The concentrations used were based on published literature on the application of these drugs in cell systems and on empirical titration experiments done on neurons (data not shown). Cells were pretreated for 2 h with drug or vehicle before exposure to CPT or ETOP.
For quantification of living neuron densities, cells were viewed under phase contrast optics using a Nikon Eclipse TS100 microscope (40× objective) equipped with an Infinity camera to capture digital images from 6 nonoverlapping microscopic fields. For quantification of neuronal apoptosis, fixed cultures were stained with a nuclear dye (Hoechst 33258), digital images were captured from 6 nonoverlapping microscopic fields, and apoptotic nuclei were counted as before (Lesuisse and Martin 2002b).
Cell lysates were made from control and CPT-treated neurons for Western blot analyses. DIV5 and DIV25 cortical neuron cultures (3 different platings) were treated with 10 μM CPT and cells were harvested 1, 2, 4, 8, 15, and 24 h later. Cortical neuron cultures were lysed in TNE buffer (10 mM Tris–HCl, pH 7.4, 150 mM NaCl, and 5 mM ethylenediaminetetraacetic acid) containing protease inhibitors (1 mM phenylmethanesulphonylfluoride (PMSF), 10 μg/ml leupeptin, and 10 μg/ml pepstatin A) and detergents (2% sodium dodecyl sulfate [SDS], 1% deoxycholate, and 1% NP-40), and sonicated for 15 s. Because most of the primary antibodies used were generated against human proteins, samples of human cerebral cortex were used as positive controls in some western blot experiments. The human brain autopsy tissue was obtained from the Human Brain Resource Center at Johns Hopkins as described previously (Martin, 2000). Protein concentrations of each homogenate were determined by protein assay (Pierce, Rockford, IL). Homogenates were diluted further in Laemmli sample buffer.
Protein (20 μg) from neuronal cultures were resolved on gradient SDS polyacrylamide gels (4–20%) and transferred to nitrocellulose filter membranes by electroblotting. Nitrocellulose membranes were blocked in 5% (w/v) nonfat dry milk in phosphate-buffered saline (PBS). After overnight incubation with the primary antibodies (Table 2) in PBS supplemented with 5% nonfat dry milk and 0.05% (v/v) Tween-20, membranes were rinsed and then incubated with peroxidase-conjugated secondary antibody. Immunoreactive proteins were visualized on x-ray film using enhanced chemiluminescence (Pierce).
Immunoprecipitation and western blot analysis of cell lysates from CPT-treated and control neurons were used to identify p53 and c-Abl as targets of ATM/ATM-Rad3–related kinase (ATR). Briefly, 100 μg of total lysate protein was immunoprecipitated using 5 μg of mouse monoclonal antibody to p53 or c-Abl (Table 1, Santa Cruz, Biotechnology, Santa Cruz, CA). After immunocapture with immobilized protein A/G (Pierce) the samples were subjected to SDS-polyacrylamide gel electrophoresis and electroelution onto nitrocellulose membrane for the detection of phospho-ATM/ATR substrate proteins (Table 2, Cell Signaling Technology, Danvers, MA). The efficiency of immunocapture was estimated by immunoblotting to be ~55% of total p53 or c-Abl.
Immunofluorescence was used to visualize cleaved caspase-3 and phospho-p53 in control and CPT-treated cortical neurons. Cells were fixed in 4% paraformaldehyde/4% sucrose in PBS (4 °C, 20 min) and then with methanol (4 °C, 10 min), and then permeabilized in 0.2% Triton X-100, 10% normal goat serum (NGS) in PBS (4 °C, 10 min), incubated overnight at 4 °C in 10% NGS in PBS with primary antibody to cleaved caspase-3 (D175, Cell Signaling, Technology) or phospho-p53 (Cell Signaling). Antibody binding was visualized with Alexa-488- or Cascade blue-conjugated goat anti-rabbit (Invitrogen) diluted at 1:600 in PBS. p53 immunostained cells were also dual-labeled with an antibody to manganese superoxide dismutase (Assay Design, An Arbor, MI) as a mitochondrial marker. Immunofluorescence preparations with viewed with a Nikon TS100 microscope and digital images were captured with a Nikon Infinity CCD camera or they were viewed using a Zeiss LSM confocal microscope.
Cortical neurons were lysed in cell lysis buffer, microcentrifuged (10000 × g), and supernatants were evaluated for the activity of caspases that recognize the DEVD sequence using a colorimetric assay kit based on the detection of p-nitroaniline after cleavage from DEVD (Chemicon, Temecula, CA). Absorbance was read spectrophotometrically at 405 nm. Human recombinant active caspase-3 (Chemicon) was used as a positive control. Treatment of samples with caspase-3 inhibitor (Ac-DEVD-CHO) was the negative control.
Crude lysates of cortical neurons were evaluated for Topo-I activity using an assay kit that is highly specific for Topo-I activity with no contribution of Topo-II (TopoGEN, Port Orange, FL). This assay is based on the ability of Topo-I to catalyze ATP-independent relaxation of negatively supercoiled plasmid DNA. Relaxation assays were carried out with 2.5–5 μg cortical neuron lysate protein in a final volume of 25 μL in Topo-I reaction buffer (100 mM Tris-Cl, pH 7.9, 1.5 M NaCl, 1% bovine serum albumin, 1 mM spermidine, 50% glycerol). Supercoiled plasmid DNA (0.25 μg) was used as substrate. Reactions were terminated with 5 μL (per 20 μL reaction volume) of stop buffer (5% sarkosyl, 0.0025% bromophenol blue, 25% glycerol). Reaction products were analyzed on 1% native agarose gels.
There are many different forms of DNA damage (Lindahl 1993), so we used the comet assay to profile, using specific pH conditions, DNA-SSBs and DNA-DSBs directly in mouse cortical neurons. This assay is exquisitely sensitive, detecting one strand-break per 2 × 1010 Daltons of DNA (Kindzelskii and Petty 1999), and quantitative for measuring early damage to genomic DNA of eukaryotic cells on a single-cell basis (Tice et al. 2000). We found previously that oxidative stress by hydrogen peroxide, nitric oxide donors, and peroxynitrite all induce DNA-SSBs in primary motor neurons (Martin and Liu 2002). Cortical neurons at DIV5 and DIV25 were exposed to CPT, ETOP, or vehicle and then collected at 4 h for DNA-SSB and DNA-DSB determinations using alkaline and neutral comet assays, respectively, essentially as described (Martin and Liu 2002) but with some modifications, including the use of CometSlides (Trevigen, Gaithersburg, MD) for embedding of ~1000 neurons per slide and SYBR Green staining of DNA instead of ethidium bromide staining after gel electrophoresis. Slides were viewed (excitation 425–500 nm) with a Zeiss Axiophot epifluorescence and at least 10 images (10–20 nuclei per image)/slide were captured at a magnification of 200×. Two Cometslides were used for each condition. With this excitation wavelength DNA-bound SYBR Green fluoresces green (Fig. 1G). In normal neurons the fluorescence is confined mostly to the nucleoid (nucleus after assay) because the DNA is undamaged and supercoiled and cannot migrate far in the electric field. In neurons with DNA damage, the DNA is unwound by the alkali solution for SSB detection, or neutral solution for DSB detection, and the negatively charged DNA fragments are released from the nucleus and migrate toward the anode. The length of DNA extrusion from the nucleoid is proportional to the amount of accumulated DNA-SSBs or DNA-DSBs (Martin and Liu 2002). The percentage of neurons yielding nucleoids with comets relative to the total number of neurons was determined from digital images.
As a first step to authenticate our observations made in cell culture as events that occur in brain, we studied the expression of selected cell death proteins in the developing mouse brain. Deeply anesthetized wild-type C57BL/6 mice were killed by decapitation on the day of birth (P0) and at postnatal day 5 (P5), 10 (P10), 15 (P15), and 20 (P20). The fresh forebrains were used to prepare subcellular fractions that were analyzed by Western blotting as described (Martin et al. 2003). In this developmental series, nuclear fractions were used to evaluate expression of phospho-53 and mitochondrial-enriched fractions were used to evaluate Bax, Bak, Noxa, and Puma. The subcellular fractionation method has been validated (Martin et al. 2003). Subcellular fractions of human cerebral cortex gray matter (Martin 2000) were used as positive or negative controls because the primary antibodies were raised against human protein sequences.
To further define the relevance of observations in a cell culture model system as events that occur in injured brain, we used CPT in an in vivo setting. C57BL/6 mouse dams with litters were used. The day of delivery was designated as P0. Pups with dams were housed in a laboratory animal suite with an ambient temperature of 23 °C, 12-h light/dark cycle, and ad libitum access to food and water. The Animal Care and Use Committee of the Johns Hopkins University School of Medicine approved the animal protocol. CPT was injected directly into the brain of wild-type and bak−/− mouse pups at P5. The CPT was the same stock as that used for cultured neuron experiments. Pups were deeply anesthetized (isofluorane/nitrous oxide/oxygen) and were placed in a neonatal small animal stereotaxic apparatus (Stoelting, Wood Dale, IL). Surgical anesthesia was maintained with isoflurane and nitrous oxide. Body temperature was maintained with external warming. The scalp was cleaned with betadine and 70% isopropyl alcohol prior to making a midsagittal incision. A small burr hole in the skull over the right hemisphere was made with a fine tip dental drill bit. CPT (10 μM, 1 μL total volume) or an equal volume of PBS was injected stereotaxically with a 10-μL Hamilton syringe (Hamilton Company, Reno, NV) into the lateral ventricle. CPT was delivered over 1 min; the needle was left in place for 3 min before it was withdrawn slowly to prevent leakage of drug and cortical trauma. The craniotomy was unsealed, the scalp was sutured, and the pup was returned to the dam.
CPT-treated and vehicle-treated mouse brains were analyzed for biochemical and histological changes indicative of apoptosis. CPT-treated and control mice were killed at 8 h (n =6) or 16 h (n = 6). Brains were removed for western blot analysis of phospho-p53, Bax, and Bak. Another cohort of mice was killed by an overdose of sodium pentobarbital followed by perfusion-fixation with aldehydes at 24 h after injection of CPT (n = 6) or vehicle (n = 6). Brains were harvested, cryoprotected, and cut serially into 40-μm-thick sections using a freezing microtome (American Optical Corporation, Buffalo, NY). To assess effects of CPT in brain, every 10th section was mounted on adhesive-coated glass microscope slides (Fisher, Pittsburgh, PA), dried, dehydrated with alcohols, defatted with xylenes, and stained with cresyl violet to observe apoptotic profiles.
All quantitative data are expressed as mean ± standard deviation. Statistical analysis among groups was performed with one-way analysis of variance (ANOVA) or 2-way ANOVA followed by the Duncan's multiple range test, with P < 0.05 considered statistically different.
We determined if embryonic cortical neurons cultured to different stages of differentiation (Lesuisse and Martin 2002a) have different responses to CPT, which is presumed to cause DNA damage in neurons. Cortical neurons at DIV5 (immature and undifferentiated) and DIV25 (mature and differentiated) were treated with 1, 10, or 100 μM CPT or with vehicle and then evaluated for apoptosis after 24 h as gauged by nuclear morphology with Hoechst 33258 staining (Fig. 1A). All concentrations of CPT-induced apoptosis in cortical neurons with a similar light microscopic morphology (Fig. 1A, left) as described before (Lesuisse and Martin 2002b), and the amount of apoptosis was dose-dependent (Fig. 1A, right). At the 10 μM dose, DIV5 neurons were more sensitive than DIV25 neurons, as evidenced by the greater percentage of apoptotic neurons (Fig. 1A, right, 2-way ANOVA). Western blotting revealed that differentiated neurons have higher levels of Topo-I, detected at ~120 kDa, than immature neurons and that 10 μM CPT treatment induced rapidly a loss of Topo-I in both DIV5 and DIV25 neurons (Fig. 1B). The loss of Topo-I protein seen at 4 h of 10 μM CPT treatment was paralleled by a loss of functional enzyme activity as seen by decreased formation of relaxed plasmid DNA topoisomers (Fig. 1C).
Immunocytochemistry on fixed DIV5 and DIV25 cortical neurons treated with 10 μM CPT revealed numerous neurons immunopositive for cleaved caspase-3, but very few neurons in control cultures displayed immunoreactivity for cleaved caspase-3 (Fig. 1D). In keeping with the immunolabeling, both DIV5 and DIV25 neurons treated with 10 μM CPT showed a time-related increase in caspase-3 enzyme activity (Fig. 1E). At both stages of differentiation, caspase-3 activity was increased significantly at 2 hof CPT exposure, and 3- to 4-fold increases were observed at 8 h. DIV5 neurons had a greater increase in caspase-3 activity compared with DIV25 neurons (Fig. 1E). The effect of increased caspase-3 activity in mediating apoptosis in this system was determined by pretreating neurons with a novel small-molecule non-peptide inhibitor of caspase-3, shown previously to potently and selectively block caspase-3 in vitro and in vivo (Chen et al. 2006). CH95 promoted significant decreases in the amount of neuronal apoptosis, resulting in ~38% and ~42% neuroprotection in both immature and differentiated neurons, respectively (Fig. 1F).
It has been assumed that CPT triggers apoptosis in neurons by inducing DNA strand breaks (Park et al. 1998; Stefanis et al. 1999) as in cycling non-neuronal cells (Nieves-Neira and Pommier 1999), but this has not been confirmed in neurons. To understand the upstream mechanism through which CPT triggers apoptosis in immature and mature cortical neurons, we evaluated neurons for the presence of DNA strand breaks using alkaline and neutral comet assays, which detect selectively DNA-SSBs or DNA-DSBs under these conditions, respectively (Martin and Liu 2002). In both DIV5 and DIV25 cortical neurons both DNA-SSBs and -DSBs accumulated significantly with 4 h of CPT exposure (Fig. 1G-I). Counts of comet profiles in microgels revealed dose-dependent accumulation of cortical neurons harboring DNA-SSBs (Fig. 1H). The accumulation of DNA-DSBs was not CPT-dose-related (Fig. 1I). Immature neurons accumulated more DNA-SSBs compared with differentiated neurons after 10 μM CPT treatment (Fig. 1H). There was no difference between DIV5 and DIV25 neurons with the 50 μM CPT (Fig. 1H), and the accumulation of DNA-DSBs in immature and differentiated neurons was similar (Fig. 1I).
The accumulation of DNA damage leads to p53 activation through phosphorylation and subsequent apoptosis in cycling cells (Giaccia and Kastan 1998). We therefore evaluated total cell lysates of CPT-treated cortical neurons at different stages of maturation for the presence of p53 activation as witnessed by its phosphorylation at an N-terminal serine residue (Fig. 2A,B). This antibody specifically detects human p53 and mouse p53 only when serine-15 or serine-18, respectively, is phosphorylated; serine-15 in human p53 and serine-18 in mouse p53 are homologous with regard to p53 function (Chao et al. 2000). In well-maintained control cultures of mouse cortical neurons at DIV5 and DIV25, the level of active p53 is very low (Fig. 2A,B). Both DIV5 and DIV25 neurons had markedly higher levels of phospho-p53 compared with age-matched controls after exposure to 10 μM CPT. We evaluated first a 4- to 24-h time course and found the accumulation of phosphorylated p53 already at 4 h of CPT exposure (Fig. 2A). DIV5 neurons mounted a sharp ~80-fold increase in p53 phosphorylation with 4 and 8 h exposure to CPT, which then dissipated to a 20-fold increase above control (Fig. 2A, right). In DIV25 neurons there was also a high increase (~25-fold) in p53 phosphorylation at 4 h of CPT exposure, but then the level of phosphorylated p53 declined progressively, although remaining significantly above control levels (Fig. 2A, right). We then evaluated CPT-treated DIV5 neurons over a 1- to 4-h period (Fig. 2B), using lower amounts of protein loading in an effort to detect more subtle gradations in p53 phosphorylation in this more collapsed time course compared with the earlier experiment (Fig. 2A). There was a significant increase in p53 phosphorylation at 1 hour of treatment, and then phosphorylation further increased progressively to achieve ~50-fold higher levels above control (Fig. 2B, right). These increases in phospho-p53 occurred in the absence of elevations in the overall levels of p53 as detected with pan-p53 antibody (Fig. 2A,B). This immunoreactive band is known to be p53 because it is absent in p53−/− mouse brain (Martin et al. 2003) and in primary embryonic neuron cultures from these mice (data not shown). Immunocytochemistry on fixed neurons treated with CPT for 4 h showed numerous DIV5 and DIV25 neurons immunoreactive for phospho-p53 (Fig. 2C). Cultures under baseline conditions or vehicle treatment had very few neurons immunoreactive for phospho-p53 (Fig. 2C, lower left inset), consistent with the low level of phospho-p53 detected in controls by western blotting (Fig. 2A,B). After CPT treatment, DIV25 neurons were different from DIV5 neurons in that they showed cytoplasmic and nuclear phospho-p53 accumulation (Fig. 2C, right), but DIV5 neurons primarily had nuclear accumulation of phospho-p53 (Fig. 2C, left). Some cytoplasmic p53 immunoreactivity was localized to mitochondria in CPT-treated differentiated DIV25 neurons (Fig. 2C, right inset).
To determine if p53 participates causally in the mechanisms of cortical neuron apoptosis and to evaluate possible differential molecular requirements for cell death depending on neuron differentiation, we used genetic and pharmacological manipulations. Cortical neuron cultures were prepared from wild-type and p53−/− mouse embryos, maintained for 5 or 25 days, and treated with 10 μM CPT for 24 h. The apoptotic response of p53−/− DIV5 neurons was blocked nearly completely, whereas the apoptotic response of p53−/− DIV25 neurons was blocked by about 50% (Fig. 2D). In another experimental paradigm, wild-type p53+/+ cortical neurons were treated with 2 inhibitors of p53 that have different mechanisms of action (Fig. 2E): cyclic-α-pifithrin (α-PFT) that blocks reversibly p53-dependent transactivation of p53-dependent responsive genes in non-neuronal cells (Komarov et al. 1999) or μ-PFT that blocks p53 interaction with Bcl-xL and Bcl-2 and inhibits selectively p53 translocation to mitochondria without affecting the transactivation function of p53 at the level of the genome in non-neuronal cells (Strom et al. 2006). In DIV5 p53+/+ neurons, the level of apoptosis was reduced from ~87% to 35% of neurons with α-PFT treatment (Fig. 2E). μ-PFT was much less effective than α-PFT in reducing the apoptosis in DIV5 p53+/+ neurons, although still having significant effect (Fig. 2E). Treatment of cortical neurons with α-PFT and μ-PFT did not provide significant antiapoptotic action over α-PFT alone in DIV5 neurons. In contrast, in DIV25 p53+/+ neurons μ-PFT was significantly more effective than α-PFT in blocking apoptosis (Fig. 2E). As control experiments for drug specificity, treatment experiments with α-PFT and μ-PFT were repeated in p53−/− neuron cultures, no additional neuroprotection was observed over that provided by the null mutations (data not shown).
This experiment suggested that apoptosis in immature and differentiated cortical neurons might have different levels of transcriptional dependence. This suspicion was confirmed by treating wild-type cortical neurons with a low concentration of cycloheximide (1 μg/mL) or actinomycin D (1 μg/mL) prior to CPT exposure. Both agents blocked the apoptosis in DIV5 neurons but not in DIV25 neurons (Fig. 2F).
The mechanisms of cortical neuron apoptosis downstream of p53 activation were analyzed (Fig. 3). We examined by Western blotting mediators of apoptosis in cell lysates of CPT-treated neurons harvested after different times of exposure. Multidomain (Bak and Bax) and BH3-only (Puma and Noxa) members of the Bcl-2 family were measured over a 4-h CPT exposure time course. Bak, detected at ~30 kDa, was increased markedly in CPT-treated cortical neurons, but the temporal patterns were different in DIV5 and DIV25 neurons. In DIV5 neurons there was a modest but significant increase in Bak at 1 h of CPT exposure and then a large ~4-fold spike in Bak levels at 2 h, followed by a diminution in the increase at 4 h (Fig. 3A,B). In DIV25 neurons baseline control Bak levels were significantly higher (>3-fold) than in control DIV5 neurons, and CPT-treated DIV25 neurons mounted a sustained elevation (~2-fold) in Bak compared with DIV25 controls (Fig. 3B). The specificity of the antibody to Bak was shown by the absence of the ~30 kDa band in tissue extracts of bak−/− mouse brain (Fig. 6B). Baseline levels of Bax, detected at ~23 kDa, were similar in control DIV5 and DIV25 cortical neurons (Fig. 3A,C), but DIV5 and DIV25 neurons differed in their Bax responses induced by CPT. This ~23 kDa band is absent in bax−/− mouse brain extracts (Martin et al. 2003). Bax was increased ~6-fold and ~4-fold after 1 h of CPT exposure in DIV5 and DIV25 neurons, respectively (Fig. 3A,C). By 2 h exposure, the Bax increase was dissipating in DIV5 neurons, although still elevated above control level, whereas in DIV25 neurons the Bax increase above control was more sustained than in DIV5 neurons (Fig. 3A,C). DIV5 and DIV25 neurons had different baseline levels of Puma, detected at ~20 kDa (this immunoreactive band is absent in Western blots of liver extracts of puma−/− mice, data not shown); Puma was significantly higher in control DIV25 neurons compared with DIV5 neurons (Fig. 3A,D). DIV5 and DIV25 neurons had strikingly different Puma responses after CPT. Puma levels were unchanged in DIV5 neurons at 1 h after CPT, but then levels increased progressively to achieve about a 5-fold elevation above control by 4 h after CPT (Fig. 3A,D). DIV25 neurons displayed an abrupt ~2-fold increase in Puma at 1 hour after CPT, and then levels declined progressively (Fig. 3A,D). The patterns of Noxa expression, detected at ~11 kDa, differed from the Puma patterns at baseline and in the responses induced by CPT (Fig. 3D,E). Noxa in control DIV5 neurons was greater than in control DIV25 neurons (Fig. 3A,E). DIV5 neurons displayed an abrupt almost 3-fold increase in Noxa at 1 and 2 h after CPT and then levels declined sharply (Fig. 3A,E). In DIV25 neurons, Noxa levels increased slowly compared with DIV5 neurons after CPT exposure but achieved an almost 5-fold elevation by 4 h (Fig. 3A,E).
We used genetic and small-molecule pharmacological manipulations to identify if Bax and Bak mediate cortical neuron apoptosis and to determine if their roles differ in neurons at different stages of differentiation. However, a caveat of our cell system is the possibility of selection of neurons inherently more resistant to stress. We first evaluated the baseline survival of bax- and bak-null neurons compared with wild-type neurons, ATM−/− neurons, and p53−/− neurons. Survival of bax-, bak-, and p53-null cortical neurons was significantly greater at DIV5 and DIV25 than wild-type neurons plated at similar densities and grown under identical conditions (Fig. 3F). ATM-null neurons were similar to wild-type neurons (Fig. 3F). Deletion of the bax or bak gene protected strongly against apoptosis in DIV5 and DIV25 neurons challenged with CPT (Fig. 3G). Generally, the neuroprotection was greater in DIV25 null cells. The apoptotic response was reconstituted in bak-null cells using a cell-permeable Bak-BH3 fusion peptide (Fig. 3G) that binds selectively to Bcl-xL and antagonizes its function in non-neuronal cells (Holinger et al. 1999). The reconstitution of apoptosis was achieved to greater levels in DIV25 neurons compared with DIV5 neurons (Fig. 3G). For controls, neurons were exposed to a Bak peptide with a mutated BH3 sequence where leucine-78 is replaced by alanine (Holinger et al. 1999). These neurons had a level of apoptosis similar to vehicle control (Fig. 3I). To confirm the direct role of Bax in driving apoptosis in both immature and mature cortical neurons, wild-type neurons were treated with cell-permeable Bax-channel blocker (Bombrun et al. 2003; Hetz et al. 2005) or Bax-inhibiting peptide (Sawada et al. 2003) prior to CPT exposure. To show drug/peptide specificity, treatment experiments with Bax-channel blocker and Bax-inhibiting peptide were done on bax−/− neuron cultures. No additional neuroprotection was observed over that provided by the null mutations (data not shown). Both of these Bax antagonists protect HeLa cells from apoptosis (Bombrun et al. 2003; Sawada et al. 2003) and both provided major neuroprotection in immature and mature bax+/+ neurons (Fig. 3H). A mutated analog of the Bax-inhibiting peptide harboring no apoptotic activity (Sawada et al. 2003) had no effect on CPT-induced neuronal apoptosis (Fig. 3H).
It has been proposed that Bax and Bak can interact with the mitochondrial outer membrane protein VDAC to cause apoptosis (Shimizu et al. 2000). To examine the contribution of the VDAC in Bak-mediated neuronal apoptosis, wild-type DIV5 and DIV25 were treated with a cell-permeable peptide containing the conserved N-terminal Bcl-2 homology domain (BH4) of Bcl-xL (Bcl-xL-BH4) that prevents apoptosis by binding to the VDAC and blocking its activity in non-neural cells (Shimizu et al. 2000). Bcl-xL-BH4 peptide protected robustly against Bak-mediated apoptosis in cortical neurons (Fig. 3I).
The mechanisms of cortical neuron apoptosis upstream of p53 activation were examined. Phosphorylation of p53 is a critical part of its activation sequence in response to DNA damage (Giaccia and Kastan 1998) and is mediated by about 5–10 different protein kinases, among them are ATM, ATR, and c-Abl (Kastan and Lim 2000). The ATM signaling network can be studied by tracking the phosphorylation of ATM and its substrates. To examine the overall phosphorylation status of proteins that are phosphorylated by ATM and the closely related kinase ATR, we did an ATM/ATR phosphorylated substrate screen of proteins from total cell lysates of wild-type cortical neurons treated with CPT for different times (Fig. 4A). In DIV5 neurons the phosphorylation of ATM/ATR substrate proteins, ranging from 17 kDa to 214 kDa, was increased significantly at 8 h (~150% of control) through 24 h (~250% of control) of CPT treatment (Fig. 4A). The pattern of ATM/ATR substrate protein phosphorylation was very different temporally and quantitatively in DIV25 neurons treated with CPT compared with DIV5 neurons treated similarly. In differentiated neurons, markedly increased (~300% of control) phosphorylation of ATM/ATR substrate proteins was seen at 4 h of CPT treatment, particularly a protein band at ~37 kDa (Fig. 4A). The phosphorylation of proteins in the range of 40 kDa to 200 kDa increased progressively from 4 h to 15 h in DIV25 neurons and then was dissipating at 24 h (Fig. 4A).
We next examined directly the activation of ATM in cortical neurons. DNA strand breaks can cause rapid activation of ATM through autophosphorylation at serine-1981 (Kastan and Lim 2000). Baseline activation of ATM, seen as a protein band at ~350–370 kDa, was very low in immature and differentiated control neurons (Fig. 4B). Antibody specificity was confirmed by the absence of this band of immunoreactivity in ATM−/− neurons (data not shown) and the robust immunoreactivity detected in irradiated HeLa cells (Fig. 4B). In DIV5 neurons challenged with CPT, phospho-ATMSer1981 immunoreactivity was elevated transiently ~10-fold greater than control levels at 4 h but then was ~5-fold higher at 8 h, and subsequently levels dropped below control at 15 and 24 h (Fig. 4B). These increases in phospho-ATM occurred in the absence of significant elevations in the overall levels of ATM as detected with pan-ATM antibody. The temporal profile of ATM activation was different in DIV25 neurons. DIV25 neurons treated with CPT displayed phospho-ATMSer1981 levels that were elevated robustly ~10-fold above control level from 4 h through 15 h and levels were still ~6-fold greater than control at 24 h (Fig. 4B).
The non-receptor tyrosine protein tyrosine kinase c-Abl can directly phosphorylate p53, and ATM can mediate p53 phosphorylation through ATM phosphorylation of c-Abl (Nakagawa et al. 1999). To investigate the possible role of c-Abl in the p53 activation sequence in CPT-treated cortical neurons, we examined the phosphorylation of c-Abl at tyrosine residue 245 which is involved in activation of c-Abl kinase activity (Kharbanda et al. 1998). Differentiating and mature cortical neurons at baseline have high levels of c-Abl as detected at ~135 kDa with phosphorylation state-specific and phosphorylation-independent antibodies, but active c-Abl is higher in immature neurons (Fig. 4C). DIV5 neurons treated with CPT had sharply decreased (~62% of control) levels of phospho-c-AblTyr245 at 4 h and then subsequently a partial recovery of phospho-c-AblTyr245 levels by 8 h and thereafter but with levels still reduced significantly to ~80% of control (Fig. 4C). In contrast, DIV25 neurons treated with CPT had levels of phospho-c-AblTyr245 ~1.5-fold higher than control at 4, 8, and 15 h but then levels declined markedly to 36% of control by 24 h (Fig. 4C).
We next examined the relationship between ATM and specific substrate protein phosphorylation. p53 was immunoprecipitated from CPT-treated DIV5 and DIV25 cortical neuron lysates with a pan-p53 antibody, and then immunoprecipitated protein was Western blotted using ATM/ATR phosphosubstrate antibody. ATM/ATR phosphorylated-p53 was elevated above control levels at 4, 8, and 15 h in DIV5 neurons and at 4 and 8 h in DIV25 neurons (Fig. 4D). We were not able to immunoprecipitate c-Abl from CPT-treated DIV5 cortical neurons, mostly likely because of its reduced levels. However c-Abl in CPT-treated DIV25 neurons was an ATM/ATR phosphosubstrate and its phosphorylation was increased markedly with CPT treatment (Fig. 4D).
Cortical neuron apoptosis was studied in the presence of genetic and pharmacological inactivation of ATM or c-Abl. Neuron cultures were prepared from wild-type, ATM+/−, and ATM−/− mouse embryos, maintained for 5 or 25 days, and treated with 10 μM CPT for 24 h. The apoptotic response of ATM−/− DIV5 cortical neurons was attenuated modestly (~23% protection), whereas that of ATM+/− neurons at DIV5 was not affected (Fig. 4E). In contrast, DIV25 neurons with no ATM were protected robustly from apoptosis (~40%) and hemizygous neurons showed slight protection (Fig. 4E). The activation of p53 in response to CPT was attenuated in ATM−/− DIV25 neurons (Fig. 4E). Wild-type cortical neurons were then treated prior to CPT exposure with Ku-55933 or caffeine which are highly specific (Hickson et al. 2004) or less specific (Blasina et al. 1999) inhibitors of ATM kinase activity, respectively. Neither drug had significant effect on the amount of apoptosis in CPT-treated DIV5 neurons (Fig. 4F). In contrast, both Ku-55933 and caffeine afforded significant protection to CPT-treated DIV25 neurons (Fig. 4F). As control experiments for ATM drug specificity, treatment experiments with Ku-55933 and caffeine were repeated in ATM−/− neuron cultures but no additional actions were observed (data not shown).
c-Abl null neurons were unavailable for our experiments, but we used a highly selective inhibitor of c-Abl (Raina et al. 2002) to examine the involvement of c-Abl in cortical neuron apoptosis. CPT-induced apoptosis in DIV5 neurons was attenuated modestly when c-Abl was inhibited with STI571, but apoptosis was worsened in DIV25 neurons in the presence of c-Abl inhibition (Fig. 4F). The activation of p53 in response to CPT was prominently blocked in DIV5 neurons treated with STI571 (Fig. 4F).
To better identify possible differences in the vulnerabilities of immature neurons and differentiated neurons to DNA-damaging insults, DIV5 and DIV25 cortical neurons were treated with a submicromolar concentration of CPT for longer than 24 h (Fig. 5A). CPT at a concentration of 0.5 μM induced a time-dependent accumulation of apoptotic DIV5 and DIV25 cortical neurons. DIV25 neurons were about 50% less vulnerable than DIV5 neurons to the toxicity of CPT over 72 h (Fig. 5A). We used ETOP as a different toxin to determine if the greater vulnerability of immature neurons compared with more mature neurons is a general rule. ETOP at a concentration of 0.5 μM also induced a time-dependent accumulation of apoptotic DIV5 and DIV25 cortical neurons, and, once again, differentiated neurons were significantly less vulnerable than immature neurons (Fig. 5B). This low-dose, time-based experiment reveals that immature neurons are significantly more sensitive than differentiated neurons to DNA-damaging agents (Fig. 5A,B).
To affirm that the molecular mechanisms driving apoptosis in cortical neurons are related to DNA damage rather than being CPT-specific, we repeated some key experiments using ETOP to induced DNA damage and apoptosis in DIV5 and DIV25 cortical neurons.
DIV5 and DIV25 cortical neurons were treated with 1, 10, or 100 μM ETOP or with vehicle and then evaluated for apoptosis after 24 h (Fig. 5C). ETOP-induced apoptosis in cortical neurons at all concentrations (Fig. 5C) that was similar morphologically (data not shown) to CPT-induced apoptosis (Fig. 1A). The amount of ETOP-induced apoptosis in cortical neuron cultures was dose-dependent (Fig. 5C). DIV25 neurons were less sensitive than DIV5 neurons at the 10 and 100 μM doses (Fig. 5C).
ETOP-treated neurons were evaluated for the presence of DNA-DSBs using the neutral comet assay (Martin and Liu 2002). In both DIV5 and DIV25 cortical neurons DNA-DSBs accumulated significantly with 4 h of ETOP exposure (Fig. 5D). Comet counts revealed that immature neurons accumulate more DNA-DSBs compared with differentiated neurons after 10 μM ETOP exposure (Fig. 5D).
The mechanisms of ETOP-induced apoptosis of cortical neurons involving p53 and its downstream and upstream signaling pathways were analyzed, as done for CPT. Cortical neuron cultures were prepared from wild-type and p53−/− mouse embryos, maintained for 5 or 25 days, and treated with 10 μM ETOP for 24 h. Apoptosis of p53−/− DIV5 neurons was blocked nearly 90%, whereas the apoptotic response of p53−/− DIV25 neurons was blocked by about 40% (Fig. 5E). DIV5 and DIV25 neurons without bax or bak genes also showed attenuated apoptosis after exposure to ETOP (Fig. 5F). The apoptotic response in bak-null cells was also reconstituted using Bak–BH3 fusion peptide (Fig. 5F). Apoptosis of DIV5 and DIV25 cortical neurons without ATM−/− was attenuated significantly after ETOP exposure, whereas ATM+/− neurons were unaffected (Fig. 5G). Cortical neurons were then treated prior to ETOP exposure with ATM kinase inhibitors (Ku-55933 or caffeine). Both drugs had significant effects on the amount of apoptosis in ETOP-treated DIV5 neurons and DIV25 neurons (Fig. 5H). ETOP-induced apoptosis in DIV5 neurons was attenuated significantly by the c-Abl inhibitor STI571, but the amount of apoptosis was unaffected in DIV25 neurons in the presence of c-Abl inhibition (Fig. 5H).
Another caveat of our cell system is whether neuron maturation in culture is similar to neuron maturation in vivo. To begin to establish the relevancy of our findings on cultured cortical neurons to the intact brain setting, we analyzed the expression of selected cell death proteins in mouse forebrain during normal maturation and in response to CPT (Fig. 6). For phospho-p53, Bax, Bak, and Noxa, the general pattern was highest levels at P0 followed by progressive declines in levels as the forebrain matured postnatally (Fig. 6A), but they were still detectable in P20 forebrain. In contrast, Puma protein levels increased progressively in forebrain as maturation proceeded (Fig. 6A).
An additional caveat of our neuronal cell system is whether the molecular regulation of apoptosis in cultured neurons is representative of events occurring in brain. In P5 mice treated with CPT, p53 phosphorylation in cerebral cortex was elevated compared with control by 8 h after intraventricular CPT infusion and was increased further by 16 h (Fig. 6B). Bax and Bak were also increased in cerebral cortex after in vivo CPT treatment compared with vehicle treatment (Fig. 6B). The brains of mice treated with intraventricular CPT at P5 and killed 24 h later were atrophic, as revealed by dilation of the lateral ventricles and shrinkage of the forebrain, and had widespread apoptosis within the cerebral cortex (Fig. 6C). Importantly, the morphology of the cellular degeneration in cerebral cortex induced by CPT was entirely apoptotic, similar to that occurring in cultured cortical neurons treated with CPT. There was no evidence for necrotic cell death or, surprisingly, apoptosis-necrosis hybrid cell deaths that occur in other rodent models of cortical neuron degeneration induced by excitotoxicity or hypoxia-ischemia (Portera-Cailliau et al. 1997; Northington et al. 2007).
It is not known whether DNA damage initiates cell death in postmitotic cortical neurons and how this occurs. Our study is important because it implicates strongly the accumulation of DNA strand breaks as a major triggering event for neuronal apoptosis. We then delineated the molecular mechanisms of DNA damage-induced apoptosis upstream and downstream of p53 in cortical neurons at immature and differentiated states.
Cortical neurons accumulate high levels of DNA strand breaks within 4 h exposure to Topo inhibitors. It is likely that the strand breaks accumulate before 4 h because p53 phosphorylation is enhanced by 1 h. This outcome is not drug-specific and the apoptosis is most likely induced by the DNA strand breaks. These events occur in both immature DIV5 neurons and well-differentiated DIV25 neurons in a dose-dependent fashion. Immature neurons are more prone than mature neurons to accumulate DNA-SSBs. Maturity-related differences in vulnerability to DNA strand break accumulation could be due to acquisition of proficient DNA repair mechanisms in more mature neurons (Romero et al. 2003) or to differences in constitutive levels of topoisomerase isoforms in cortical neurons. Topo-I protein is depleted rapidly in cortical neurons exposed to CPT, similar to cycling cells (Nieves-Neira et al. 1999); thus, interference with RNA transcription could be involved also in cortical neuron apoptosis. The high vulnerability for postmitotic neurons to accumulate DNA-SSBs, apparently dissociated from S-phase DNA replication, could be related to their high level of RNA synthesis, and thus a greater dependence on Topo-I. The rapid accumulation of DNA strand breaks in this neuronal model places DNA damage as an activator of an ATM/p53-regulated apoptosis cascade.
Our conclusion that DNA damage is an initiator of apoptosis in cortical neurons is supported strongly by data showing that the ATM/c-Abl-p53 signaling network is activated. Scant work has been done on the upstream signal transduction mechanisms in neurons that link DNA damage to p53 and apoptosis. ATM and ATR are DNA damage sensor kinases that become activated in response to DNA damage; p53 and c-Abl are among the identified substrates (Kastan and Lim 2000). In non-neuronal human cells, ATM is activated by phosphorylation and is required for the phosphorylation of p53 at serine15 and serine20 (Dumaz et al. 2001). Neonatal mice without ATM have reduced p53-mediated brain neuron apoptosis in response to DNA damage from ionizing radiation (Lee et al. 2001), but responses of ATM-null brain neurons to other genotoxins are not known. We show that ATM is activated, and ATM/ATR protein substrates are phosphorylated during apoptosis in cortical neurons. Importantly, ATM-null neurons are protected from apoptosis, demonstrating a direct role for ATM in cortical neuron apoptosis. Moreover, inhibitors of ATM kinase activity are antiapoptotic in cortical neurons (Fig. 7). One earlier study has shown that ATM-null DIV14 cortical neurons are resistant to apoptosis induced by excitotoxicity (Macleod et al. 2003). Another study has shown that ATM phosphorylates p53 and drives apoptosis in DIV2 cortical neurons exposed to CPT (Keramaris et al. 2003). We know now that the activation of ATM and the requirement of ATM for DNA damaged-induced apoptosis are greater in differentiated neurons compared with immature neurons (Fig. 7).
DNA damage is also linked to p53 and apoptosis through the non-receptor tyrosine kinase c-Abl. c-Abl is activated by DNA damage and binds p53 causing increased transactivation function (Kharbanda et al. 1998). Phosphorylation of c-Abl by ATM and DNA-PK activates its kinase activity. Deletion of c-Abl and inhibition of c-Abl with STI571 can block apoptosis of non-neural cells (Raina et al. 2002). We show that c-Abl is activated strongly in mature neurons with DNA damage. However, c-Abl inhibition with STI571 is neuroprotective in immature neurons but worsens the degeneration in mature neurons. We conclude that c-Abl is a regulator of apoptosis and survival in cortical neurons and that this role depends on neuronal differentiation state (Fig. 7).
We show that p53 is a major regulator of cortical neuron apoptosis. CPT exposure induces substantial phosphorylation and nuclear localization of p53 in cultured neurons, consistent with the presence of DNA damage. p53 activation is more robust in immature neurons compared with mature neurons, in keeping with the greater accumulation of DNA-SSBs in immature neurons. The apoptosis induced by CPT and ETOP is blocked in neurons without p53 genes. The neuroprotection is greater in immature neurons compared with differentiated neurons. Inhibition of the DNA binding trans-activation function of p53 with α-PFT (Komarov et al. 1999) partially blocks apoptosis in immature and mature neurons (Fig. 7). Interestingly, blocking p53 interaction with Bcl-xL and Bcl-2 and inhibiting p53 translocation to mitochondria, without affecting the trans-activation function of p53 with μ-PFT (Strom et al. 2006), has very robust antiapoptotic effects in differentiated neurons (Fig. 7). Additional evidence that the nuclear trans-activation function of p53 is more important in immature neurons compared with differentiated neurons is the observation that inhibitors of RNA transcription and translation block apoptosis in immature neurons but not in differentiated neurons (Fig. 7). These data reveal the possibility that p53-mediated apoptosis in immature neurons is transcription-dependent, whereas p53-mediated apoptosis in differentiated neurons is transcription-independent (Fig. 7). p53 can mediate mitochondrial permeabilization through direct physical interaction with Bax and Bak (Chipuk et al. 2004). We conclude that p53 regulates DNA damage-induced apoptosis in both immature and mature cortical neurons, but p53 functions differently depending on neuron differentiation by having primarily a nuclear role in immature neurons and probably nuclear and mitochondrial roles in mature neurons (Fig. 7).
Earlier studies have shown that the tumor suppressor p53 can mediate neuronal death in cell culture. Overexpression of p53 causes neurodegeneration in cultures of sympathetic ganglion neurons (Slack et al. 1996), immature cerebellar granule neurons (Cregan et al. 1999), and cortical neurons (Xiang et al. 1996, 1998). CPT-induced death of DIV2–4 cortical neurons is blocked in cells deficient in p53 (Xiang et al. 1998). Another study has shown increased levels of p53 protein and increased p53 phosphorylation in DIV2 cortical neurons exposed to CPT (Keramaris et al. 2003). The involvement of p53 in neuronal apoptosis might be differential depending on cell type and cell maturity because the dependence of apoptosis on p53 in neurons might decline with maturation (Johnson et al. 1999), but this interpretation is confounded because the neuronal cultures were generated from different sources (embryonic and postnatal brains) with inherently different amenabilities to culture. A novel contribution of our work is the finding that immature neurons and differentiated mature neurons have some common molecular mechanisms for apoptosis with respect to p53 dependence, but they also exhibit key differences. Immature and mature cortical neurons do not undergo p53 activation equivalently, p53-mediated apoptosis in immature cortical neurons has greater transcriptional-translational dependence than in differentiated neurons, and p53 may function at mitochondria (thus bypassing transcriptional-translational dependence) in differentiated cortical neurons (Fig. 7).
Our study reveals several new aspects about the regulation apoptosis in cortical neurons by Bcl-2 family members. We demonstrate that Bax protein levels are increased several fold, surprisingly within one hour, in cultured cortical neurons treated with CPT; furthermore, both immature and differentiated neurons without Bax are protected strongly against DNA damage-induced apoptosis in vitro. Moreover, interruption of the Bax function with a Bax-channel blocker and a Bax-inhibiting peptide protects robustly wild-type immature and mature neurons, showing for the first time the antiapoptotic effects of small-molecules targeting Bax in cortical neurons at different stages of maturation (Fig. 7). We also show that Bax is upregulated in cerebral cortex after intracerebral exposure to CPT. The increase in the level of Bax protein is rapid and transient, thus it can be missed by studying an inappropriate time scale, and suggests that transcriptional mechanisms are engaged rapidly and/or there is an existing pool of Bax transcripts available for rapid translation. Previous reports have shown that CPT-induced death of cultured immature neurons involves proteins in the Bcl-2 family. Death of DIV1–2 or DIV4 cortical neurons induced by CPT is blocked in cells deficient in Bax (Xiang et al. 1998), and Bax-deficient cerebellar granule neurons in vitro are resistant to apoptosis induced by adenoviral overexpression of p53, indicating that Bax events are downstream of p53 (Cregan et al. 1999). The human bax gene promoter is activated directly by p53 (Miyashita and Reed, 1995), but there is some controversy about the role of p53 in direct transcriptional activation of the murine bax gene promoter and whether there are other p53-responsive binding sites in intronic sequences of the murine bax gene (Igata et al. 1999; Schmidt et al. 1999, Thornborrow et al. 2002). We find after CPT exposure an increase in Bax in cultured mouse neurons and in the mouse brain coinciding with the onset of p53 activation, suggesting that a commitment to apoptosis occurs very rapidly in cortical neurons in a p53-dependent manner, but we have not shown that p53 is directly causing the elevation in Bax.
We show for the first time that Bak, a close homologue of Bax, has a role in mediating apoptosis in cortical neurons. Bak protein is upregulated in cortical neurons undergoing apoptosis. Bak protein is also upregulated in mouse cerebral cortex during apoptosis in vivo. The transcriptional activity of p53 can enhance Bak expression (Pohl et al. 1999; Kannan et al. 2001). The operations of Bax and Bak in apoptotic immature and differentiated neurons appear to be different temporally based on their protein expression profiles. Neurons without Bak are protected against apoptosis; differentiated bak-null neurons show greater resistance than immature neurons. In addition, exposure of wild-type and bak-null cortical neurons to Bak–BH3 peptide induces apoptosis. Bak–BH3 mediated apoptosis is blocked fully by a fusion peptide that inhibits the VDAC (Shimizu et al. 2000), supporting an idea that Bak interacts with the mitochondrial permeability transition pore to induce its opening and apoptosis in cortical neurons. Such an interaction has been seen in rat liver and yeast mitochondria (Shimizu et al. 1999), and mitochondrial permeability transition pore inhibitors (e.g., cyclosporin A and metformin) can block apoptosis in cortical neurons (El-Mir et al. 2008). In contrast to Bax, which resides primarily in the cytosol (Hsu et al. 1997), Bak is a constitutive integral mitochondrial membrane protein (Wang et al. 2001), even in cells not exposed to death stimuli, and can interact with other resident or translocated mitochondrial proteins to alter its death permissibility function. Because Bax and Bak are both involved causally in the mechanisms for apoptosis of immature and differentiated cortical neurons (Fig. 7), the absence of either one of these proteins could affect the operation of the other, possibly through the formation of heteroligomeric complexes (Nechushtan et al. 2001) or by interaction with the VDAC (Shimizu et al. 2000) or other mitochondrial permeability transition pore components (Bernardi et al. 2006). Our finding suggests that multidomain proapoptotic Bcl-2 family members do not have completely redundant functions in cortical neurons.
BH3-only proapoptotic proteins can link DNA damage-induced p53 activation to mitochondria in non-neural cells (Jeffers et al. 2003), but this connection is not clear in neurons. We found that Puma and Noxa protein levels are elevated in cortical neurons undergoing DNA damage-induced apoptosis. Our Puma result is consistent with the upregulated Puma mRNA and protein in cytosine arabinoside-treated sympathetic ganglion neurons (Wyttenbach and Tolkovsky 2006) and in DIV6 postnatal cortical neuron cultures undergoing arsenite-induced apoptosis (Wong et al. 2005). Puma mRNA upregulation is also found in CPT-treated cerebellar granule neurons (Cregan et al. 2004). Increased Noxa mRNA expression occurs in DIV6 postnatal cortical neurons undergoing arsenite-induced apoptosis (Wong et al. 2005). Of greater conceptual importance is the observation made here that undifferentiated and differentiated neurons have distinct temporal profiles for Puma and Noxa protein upregulation; thus, the upregulation of some p53 target genes or the mobilization of their gene products is different in cortical neurons undergoing apoptosis depending on maturation state.
Our study shows that caspase-3 has a major participatory role in the mechanisms of DNA damage-induced apoptosis in cortical neurons that are undifferentiated and well-differentiated. Most related studies on cortical neurons have been done on immature, undifferentiated DIV1–4 neurons and have not identified definitively the roles of caspases in CPT-induced apoptosis. Protection against CPT-induced apoptosis has been seen with 100 μM caspase-3 inhibitor in rat embryonic cortical neurons at DIV1 (Stefanis et al. 1999) and mouse telencephalic neuron cultures at DIV4 (Johnson et al. 1999). Caspase inhibition protected partially against and delayed CPT-induced apoptosis in mouse embryonic cortical neurons at DIV2 (Keramaris et al. 2000); yet, other experiments have failed to block CPT-induced apoptosis of DIV2 neurons with caspase inhibitor (Park et al. 1998), and neuronal death was only delayed by 24 h in neurons without caspase-3 (Keramaris et al. 2000). We found previously a near complete block of CPT-induced apoptosis with a reversible peptide inhibitor of caspase-3 in DIV5 cortical neurons, but with differentiated cortical neurons the peptide inhibitor was much less effective, despite the CPT-induced presence of cleaved caspase-3 at both in vitro ages (Lesuisse and Martin 2002b). Here we show that a novel small-molecule, non-peptide caspase-3 inhibitor blocks significantly CPT-induced apoptosis in immature neurons as well as differentiated cortical neurons in vitro. This finding highlights a difference in neuroprotective actions of small-molecule versus peptide inhibitors of caspases. More importantly, this finding on caspase-3 demonstrates that the molecular mechanisms of DNA damage-induced cell death in cortical neurons converge downstream in the apoptotic pathway independent of neuron differentiation state.
The execution of apoptosis in neurons of cerebral cortex triggered by DNA damage is multifactorial involving at least 4 molecular networks (ATM/c-Abl protein kinases, p53, Bcl-2 multidoman and BH3-only family members, and caspases). Another level of complexity is differential molecular signaling in response to DNA damage depending upon neuronal differentiation state or age. The activation of p53 and coinciding neuronal apoptosis in cortical neurons with DNA damage is not merely an in vitro system-specific process, because similar molecular and morphological events occur in cerebral cortex after CPT treatment in mice.
DNA damage-induced, p53-mediated cell death possibly contributes to cortical neuron degeneration in human disorders (Martin 2001; Martin et al. 2005). In age-related diseases, if cortical neurons accumulate selectively DNA strand breaks as part of the mechanistically relevant pathobiology (Mullaart et al. 1990), then why is the neurodegeneration apparently slow, in light of evidence on mouse neurons showing that when DNA strand breaks accumulate, the p53-cell death response is rapid and robust? There is likely a balance between steady-state DNA strand break accumulation and DNA repair capacity. Inefficiency in DNA repair or enhanced DNA strand break formation could be acquired slowly over time (aging) in neurons with the amount of DNA damage accumulating initially at subthreshold levels for engaging cell death mechanisms. DNA damage could accumulate asynchronously in subsets of disease-vulnerable postmitotic neurons during aging that are eliminated individually slowly over time through p53-mediated mechanisms. Another possibility is that p53 activation can be independent of DNA damage (Ohyagi et al. 2005). Neocortical neurons need to be protected against age- and injury-related degeneration because endogenous postnatal genesis of these neurons in primates is rare (Rakic 2002; Breunig et al. 2007). Thus, strategies for neuroprotection in cerebral cortex will need to be age appropriate due to differences in cell death mechanisms in neurons at different stages of maturity (Fig. 7).
U.S. Public Health Service, National Institutes of Health, National Institute of Neurological Disorders, and Stroke (NS052098 and NS034100); and National Institute on Aging (AG016282).
We are grateful for the expert technical assistance of Ann Price and Yan Pan. L.J.M. thanks Dr Christian Lesuisse for his inspirational work with this neuronal culture model and Drs John Jeffers and Gerard Zambetti (St Jude Children's Research Hospital) for providing puma null mice. L.J.M. dedicates this manuscript to the memory of his father, Joseph G. Martin (born, June 10, 1926; died, December 21, 2007). Conflict of Interest: None declared.