Address correspondence and reprint requests to Prof. K. S. Rao at Department of Biochemistry, University of Hyderabad, Hyderabad 500046, India. E-mail: firstname.lastname@example.org
Abstract: Apoptosis is now recognized as a normal feature in the development of the nervous system and may also play a role in neurodegenerative diseases and aging. This phenomenon has been investigated intensively during the last 6-7 years, and the progress made in this field is reviewed here. Besides a few in vivo studies, a variety of neuronal preparations from various parts of the brain, the majority of which were primary cultures, and some cell lines have been investigated. Several apoptosis-inducing agents have been identified, and these include lack of neurotrophic support, neurotransmitters, neurotoxicants, modulators of protein phosphorylation and calcium homeostasis, DNA-damaging agents, oxidative stress, nitric oxide, and ceramides. The precise signaling cascade is not well established, and there are lacunae in many suggested pathways. However, it appears certain that the Bcl family of proteins is involved in the apoptotic pathway, and these proteins in turn affect the processing of interleukin-1β converting enzyme (ICE)/caspases. The available evidence suggests that there may be several apoptotic pathways that may depend on the cell type and the inducing agent, and most of the pathways may converge at the ICE/caspases step.
The life history of all neurons consists of several discrete neurogenetic stages, including induction, differentiation, proliferation, migration, and formation of axonal pathways and synaptic connections, that eventually confer a specific physiological function to each neuron. However, in many parts of the CNS and PNS, roughly half of the neurons undergo an additional stage in which regression leading ultimately to death occurs. This relatively large loss of neurons is a common feature in many types of neurons (motor, sensory, interneurons, autonomic, etc.), occurs in all vertebrates, and appears to have evolved as an adaptive mechanism during development of the nervous system (Oppenheim, 1991). It is known now that such large-scale cell death occurs in oligodendrocytes also (Barres et al., 1992). The general explanation for this phenomenon is that the survival of developing vertebrate neurons depends on specific neurotrophic factors secreted by the target cells that the neurons innervate. There is excellent experimental proof for the above hypothesis and several neurotrophic factors called “neurotrophins,” like nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF), and neurotrophins (NT)-3/4/5 have been shown to promote the survival of specific developing neurons. In addition, the developing neurons also require signals from other neurons that innervate them, some require specific hormones, and perhaps some require signals from neighboring glial cells as well (Raff et al., 1993). Thus, the survival of neurons depends on a complex interplay of several factors, and any imbalance in these inputs may lead to cell death. The beneficial effect of this mechanism is that although many types of neurons are produced in excess, only a portion of them get sufficient neurotrophic support for their survival, and the rest die facilitating appropriate neuron-target cell invervation.
The cell death phenomenon, apart from being an important feature in the development of the nervous system, appears also to be a cause for many neurodegenerative diseases. Many neurological diseases are characterized by the gradual loss of specific sets of neurons and result in disorders of movement and CNS function. Such diseases are Parkinson’s disease, amyotrophic lateral sclerosis, retinitis pigmentosa, several forms of cerebellar degeneration, spinal muscular atrophy, and Alzheimer’s disease (Thompson, 1995). Aging, which may be considered as the terminal phase in the development of an organism, may also be a manifestation of specific neuronal loss by cell death. Several factors, such as oxidative stress, excitatory toxicity, inappropriate Ca2+ homeostasis, mitochondrial dysfunction, and lack of sufficient neurotrophic factors, have all been shown to contribute to the above-mentioned disorders. The available evidence indicates that the survival of neurons and their death are highly regulated and dynamic events that depend on a number of internal and external factors, with vital consequences to the well-being of an organism.
Kerr et al. (1972) described two types of cell death. One, called “cell necrosis,” results from injury and causes inflammation. The other, called “apoptosis,” is the normal developmental type with several distinct characteristics. From studies on a variety of organisms and cell lines, many distinguishing features of apoptosis have been discovered. It is interesting that membrane sphingolipids are now believed to play a role in the transduction of the apoptotic signal. However, information on the specific signals that neurons receive, their transduction pathways, the genes that are activated, and the gene products that are formed specifically in apoptosis of the nervous system is inconclusive. But there has been a surge of research activity in this field in recent years. In this article, the current knowledge of the molecular biology of the “apoptotic phenomenon” is described first, followed by a review of the recent findings (up to April 1999) that are specific to the cells of the nervous system.
APOPTOSIS: DEFINITION, CHARACTERISTICS, AND METHODS OF ASSAY
Kerr et al. (1972) described the morphological changes that occur during developmental cell death as cell shrinkage, membrane blebbing, chromatin condensation, and DNA fragmentation. These changes contrast with those found during cell death due to toxic substances, trauma, and ischemia, where cells and their organelles tend to swell and rupture in a process called “necrosis.” From these findings, it was postulated that cells have the ability to self-destruct by activation of an intrinsic cellular suicide program when they are no longer needed or have become seriously damaged. This normal physiologically relevant process is described as “apoptosis.” The morphological changes that occur in necrosis and in apoptosis are well documented in different systems, and the available evidence is summarized in Table 1. The idea that animal cells commit suicide through a built-in program is now accepted, and the term programmed cell death (PCD) has been coined, which gives a mechanistic meaning to the original term of apoptosis (McConkey and Orrenius, 1994). PCD now generally denotes any cell death that is mediated by the intracellular death program, no matter what initiates it and whether or not it displays all of the characteristic features of apoptosis, whereas the term apoptosis retains its historical strength in describing a phenomenon with a set of characteristic features (Table 1).
Table 1. Comparative features of necrosis and apoptosis
Loss of cellular homeostasis
No initial major changes
Altered membrane permeability
Not seen, at least initially
Potassium loss; sodium entry; fall in membrane potential
No sodium influx; no change in potassium concentration in cell
Swelling of all cytoplasmic compartments
Destruction of mitochondria and other organelles
Generally intact organelles
Protruberances from cell surface separate to form apoptotic bodies
The recognition of apoptosis as a normal physiological process was because of the genetic studies in the nematode Caenorhabditis elegans. A set of genes called ced genes have been shown to be required for efficient disposal of the dead cell corpse, whereas another member of the ced family, the ced-3, is essential for apoptosis to occur (Ellis et al., 1991; Yuan et al., 1993). It was found that ced-3 encodes a cysteine protease that is homologous to interleukin-1β converting enzyme (ICE), a mammalian cysteine protease that cleaves the proinflammatory cytokine interleukin-1β from its precursor protein. At least 11 members of the CED/ICE family of proteases have now been recognized in humans and implicated in apoptosis (Chinnaiyan and Dixit, 1996; Kuida et al., 1996). These enzymes, called “caspases,” are cysteine proteases that cleave certain proteins after specific aspartic acid residues, exist as zymogens, and are activated through self-cleavage (Thornberry and Lazebnik, 1998); some activate others, thus acting in a proteolytic cascade (Nagata, 1997), eventually leading to the death of the cell.
Earlier studies showed that one of the biochemical hallmarks of apoptosis is DNA cleavage at internucleosomal linker regions, resulting in ladder formation of DNA of 180-200 bp or multiples thereof on agarose gel electrophoresis (McConkey et al., 1988; Compton, 1992). The reason for the ladder type of fragmentation has been attributed to the activities of Ca2+/Mg2+-activated endonucleases (Kyprianou et al., 1988), DNase I (Arends et al., 1990; Peitsch et al., 1993), or DNase II (Barry and Eastmann, 1993). There is another suspicion that some form of endo-exonuclease, identified as a major activity in nuclei of mammalian cells, may be responsible for the internucleosomal breakage of DNA during apoptosis (Fraser et al., 1993). It is now becoming apparent, however, that morphological characteristics of apoptosis are not always associated with the ladder-type DNA fragmentation and that it is probably an epiphenomenon (Cohen et al., 1992; Falcieri et al., 1993; Zakeri et al., 1993; Samata et al., 1995). This contention is supported by the observation of Fukuda et al. (1993) and Collins et al. (1992) that ladder-type DNA fragmentation is also found in some cells dying of necrosis. It therefore appears that DNA fragmentation analysis cannot be the sole criterion, but simultaneous morphological assessment must also be done for identifying apoptosis.
Many parameters have been used to examine cell death. These include the dye exclusion test (trypan blue, eosin Y, ethidium bromide, propidium iodide, or acridine orange), which denotes membrane integrity, phase-contrast microscopy for various morphological characteristics described above, and the ladder pattern of fragmented DNA. Apoptosis can also be followed by the TUNEL (terminal deoxynucleotidyl transferase-mediated dUTP nick end-labeling) method for which an optimized protocol is available (Negoescu et al., 1996). The TUNEL labeling reflects only fragmented DNA, no mater how and why the DNA is fragmented, and thus lacks specificity for apoptosis. Another way to detect apoptosis is to monitor the degradation of poly(ADP-ribose) polymerase, which is cleaved by an ICE-like protease or a caspase using appropriate antibodies to one or more of the cleaved fragments (Duriez and Shah, 1997). Apoptosis can also be detected through membrane phospholipid changes. Phosphatidylserine is located on the inner leaf-let of the plasma membrane in nonapoptotic cells, and it is translocated to the extracellular side when apoptosis sets in. It is a late apoptotic event crucial for microglial recognition and phagocytosis, as shown in hippocampal HN2-5 cells (Adayev et al., 1998). The appearance of phosphatidylserine on the external side can be detected with annexin V, a protein that binds to phosphatidylserine with high affinity (Holtsberg et al., 1998). Recently, a method for the in situ immunodetection of activated caspase-3 in apoptotic neurons was reported that is based on a polyclonal antiserum that specifically recognizes the P18 subunit of caspase formed during apoptosis, but not the zymogen (Srinivasan et al., 1998). It is useful to demonstrate that inhibitors of protein synthesis and nucleases abolish the observed apoptotic characteristics, as it is now clear that macromolecular synthesis is necessary at some point of time after the induction of apoptosis to precipitate all its characteristic features.
EXPERIMENTAL SYSTEMS AND DETECTION OF APOPTOSIS IN CELLS OF THE NERVOUS SYSTEM
To determine the causative factors and to delineate the mechanistic details of the apoptotic phenomenon in the nervous system, a variety of cell preparations have been used (listed with references in Table 2). In most of the studies, primary cultures of cells from various regions of the brain were used. The most explored preparations were cultures of cortical and cerebellar cells, as well as sympathetic neurons (isolated from superior cervical ganglion) and hippocampal neurons. However, a few studies with cultured hypothalamic, striatal, and retinal neurons were also reported. In addition, many cell lines have been used, and in some studies cells derived from transgenic animals and cultured were used to address specific questions. In this laboratory, freshly isolated neuronal and glial cell suspensions were used to examine the apoptotic phenomenon (Mandavilli and Rao, 1996). There are also reports in which apoptosis during brain development (Blaschke et al., 1996) or as a consequence of ionizing radiation (Borovitskaya et al., 1996; Ferrer et al., 1996), infection of recombinant HIV-1 gp120 (Bagetta et al., 1995), systemic administration of kainate (Gillardon et al., 1995; Van Lookern Campagne et al., 1995), or blockade of N-methyl-D-aspartate (NMDA) glutamate receptors during early neonatal life (Ikonomidou et al., 1999) was studied in whole animals. A cell-free system of neuronal apoptosis that shows all the established characteristics has also been reported (Ellerby et al., 1997).
Table 2. Nervous system cell preparations used in studies on apoptosis
Primary cultures of cortical neurons, oligodendrocytes, and astrocytes
The detection of apoptosis was done usually by monitoring internucleosomal DNA fragmentation and by microscopic observation of the formation of pyknotic nuclei with propidium iodide or Hoechst dy (Brugg et al., 1996) or by 4,6-diaminodiphenyl-2-phenylindole (Nardi et al., 1997) or acridine orange staining (Mandavilli and Rao, 1996). In a few cases, neuronal viability (Du et al., 1997), the TUNEL method (Bagetta et al., 1995; Eves et al., 1996; Honma et al., 1996), or the DNA polymerase 1-mediated biotin-dATP nick translation (PANT) method (Chen et al., 1997) was used to assess apoptosis. A method for the quantification of DNA fragmentation using Hoechst fluorescence was reported (Wiesner and Dawson, 1996). In an interesting study on the spatial and temporal distribution of dying cells in the murine embryonic cortex during development, an in situ end-labeling technique called “ISEL” that identifies fragmentated nuclear DNA in dying cells was used (Blaschke et al., 1996).
INDUCTION OF APOPTOSIS
A variety of compounds and experimental conditions have been found to induce apoptosis in cells of the nervous system, and these may be classified as shown in Table 3.
Table 3. Inducers of apoptosis in cells of the nervous system
Withdrawal of NGF from culture medium
Chick embryo sensory neurons
Withdrawal of serum from culture medium
Cerebellar granule cells
Proliferating cell line of neuronal origin
Cell line GT1-7
SV-40 T-immortalized rat hippocampal neuronal cell line
Cerebellar granule cells
Glial-derived neurotrophic protein S100β
Basic fibroblast growth factor
Chick embryonic neural retinal cells
SV-40 T-immortalized rat hippocampal neuronal cell line
Changing from a high K+ (25 mM) to a low K+ (5 mM) medium
Cerebellar granule cells
Addition of modulators of protein phosphorylation
Protein kinase inhibitor: staurosporine
Embryonic chick cerebral neurons
Cerebellar granule neurons
Embryonic astrocyte culture
Neuroblastoma × dorsal root ganglion hybrid cell line
Human neuroblastoma SH-SY5Y cell line GT1-7
Protein phosphatase inhibitor: okadaic acid
Neuroblastoma cell lines TR-14 and NT2-N
DNA-damaging agents and nucleosides
Topoisomerase-1 inhibitor: camptothecin
Topoisomerase-2 inhibitors: teniposide and mitoxantrone
Changes in the intracellular Ca2+ levels serve as important signals in the cell’s physiology. In line with this concept, treatment of cultured neurons (Zhong et al., 1993; Takei et al., 1995, 1999) and NCB-20 cells (Joseph et al., 1993) with the Ca2+ ionophores A23187 or ionomycin resulted in apoptosis. The endoplasmic reticular Ca2+ -ATPase inhibitor thapsigargin also produced apoptosis in a hypothalamic cell line (Wei et al., 1998).
Amyloid β-protein is a 39-42-amino acid peptide that is the primary component of plaques in Alzheimer’s disease, and it is believed to contribute to neurodegeneration. This peptide activates an apoptotic pathway in cultured cortical (Loo et al., 1993; Estus et al., 1997; Imaizumi et al., 1999) and hippocampal neurons (Jordán et al., 1997b; Guo et al., 1999). Apoptosis may play a role in AIDS pathogenesis of the immune system; intracerebroventricular injection of recombinant HIV-1 gp 120 resulted in apoptotic cells in the neocortex in rats, but not in the hippocampus (Bagetta et al., 1995, 1996). Macrophages, which can phagocytose dead cells, may contribute to the death of viable neurons when they invade brain lesions. Macrophage conditioned medium significantly increased the percentage of apoptotic cells of hippocampus neurons in culture, and relatively large and stable secreted macrophage proteins are responsible for this effect (Flavin et al., 1997).
There is compelling evidence that oxidative stress produced by a variety of conditions could lead to neuronal apoptosis. In the rat brain, middle cerebral artery occlusion and reperfusion resulted in damaged nuclear DNA and accumulated DNA strand breaks, suggesting apoptosis in cerebral neurons (Chen et al., 1997). Similarly, the CA1 pyramidal cells in rat brain were shown to be vulnerable to apoptosis caused by hypoxic-ischemic insult (Walton et al., 1997). Significantly, hypoxia (Singh et al., 1996) or higher oxygen content (50% oxygen atmosphere) (Enokido et al., 1993) led to apoptosis in hippocampal cells. Hydrogen peroxide induced apoptosis in cultures of human neuroblastoma cells (Zhang et al., 1997), and oxidative stress due to glutamate-induced glutathione depletion caused apoptosis in cortical neurons (Ratan et al., 1994). 3-Hydroxykynurenine is an endogenous neurotoxin, and its level increases in several neurodegenerative disorders. 3-Hydroxykynurenine induced apoptosis in cultured striatal neurons through the generation of intracellular oxidative stress (Okuda et al., 1998). Inhibition of superoxide dismutase (SOD-1), with either antisense oligonucleotides or diethyldithiocarbamate in spinal cord organotypic cultures resulted in the apoptotic degeneration of spinal neurons, including motoneurons (Rothstein et al., 1994).
Nitric oxide (NO)
NO is now known to be a signaling molecule with many functions. Treatment of a human neuroblastoma cell line, SH-SY5Y, with NO donors such as sodium nitroprusside and S-nitroso-N-acetylpenicillamine led to apoptosis (Kamoshima et al., 1997; Uehara et al., 1999). Cytokines induce neuronal injury presumably via an apoptotic mechanism triggered by NO. Thus, treatment of primary cultures of mixed neuronal/glial cells with interferon-γ plus interleukin-1β induced a high output of NO accompanied by neuronal loss with apoptotic characteristics (S. Hu et al., 1997).
A number of lipids trigger apoptosis in cultured neurons and in some cell lines. Among these are ceramide, retinoic acid, lysophosphatidic acid, phosphatidylinositol 3-phosphate, and cholestanol. Addition of acetylsphingosine (C2-ceramide), hexylsphingosine (C6-ceramide), as well as native ceramide, led to apoptosis in cultured mesencephalic neurons (Brugg et al., 1996), cerebellar granule cells (Manev and Cagnoli, 1997), astrocytes (Mangoura and Dawson, 1998), hippocampal neurons (Schwarz and Futerman, 1997), and in the neuronal cell line GT1-7 (Srinivasan et al., 1996). All trans-retinoic acid, a classic morphogen, induced apoptosis in the embryonic stem cell line P19 (Okazawa et al., 1996) and in carcinoma cells PCC7-Mz1 (Herget et al., 1998). Lysophosphatidic acid at very low concentration (1 μM) induced apoptosis in cultured hippocampal neurons (Holtsberg et al., 1998). Apoptosis was induced in a neuroblastoma × dorsal root ganglion hybrid cell line transfected to overexpress κ-opioid receptors (Dawson et al., 1997) and in cerebellar granule cells (Shimoke et al., 1999) with phosphatidylinositol 3-kinase inhibitors, wortmannin, and LY294002. Recently, cholestenol was reported to be apoptogenic to cerebellar neurons (Inoue et al., 1999).
Ultraviolet and x-ray irradiation were used as inducers of apoptosis in cultured cerebellar neurons (Enokido et al., 1997) and hippocampal neurons (Jordán et al., 1997a), respectively. In rats aged 3 days, nuclear DNA fragmentation, an indicator of apoptosis, was observed within 15 min following γ irradiation (Ferrer et al., 1996). Similarly, γ-irradiated 17-day-old fetal rat brain tissue showed nuclear condensation and DNA fragmentation within 5 h (Borovitskaya et al., 1996). It appears that cerebellar granule cells are highly sensitive to γ radiation during the first 2 weeks of postnatal life in mice and undergo apoptosis in vivo in response to this radiation (Wood and Youle, 1995). Such an apoptotic effect of γ radiation was also reported with cultured cerebellar neurons (Enokido et al., 1996b) and hippocampal neurons (Johnson et al., 1998).
It is plausible that the mechanism of action of some of the neurotoxicants may involve activation of apoptotic pathways in neurons. Thus, 1-methyl-4-phenylpyridinium (MPP+) causes apoptosis in cultures of cerebellar granule neurons (Dipasquale et al., 1991; Du et al., 1997). This compound is the active metabolite derived by the action of monoamine oxidase on 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), which is a selective neurotoxin of the nigrostriatal neurons, and it is widely used to create animal models of Parkinson’s disease (Kopin and Markey, 1988). The apoptotic effect was found in the substantia nigra cells in mice treated with MPTP also (Tatton et al., 1994). The widely abused amphetamine drug 3,4-methylenedioxymethamphetamine (MDMA, also called “ecstacy”) causes hallucination, panic, and psychosis, and it is cytotoxic to serotonergic neurons. MDMA induces apoptosis in human serotonergic neurons, JAR cells specifically (Simantov and Tauber, 1997).
Cocaine, but not its major metabolites, produced death in fetal mouse brain cocultures with the characteristics of apoptosis (Nassogne et al., 1997). Primary cultures of cerebellar granule cells undergo apoptosis in low K+ medium, and this is prevented by NMDA. Significantly, ethanol inhibits the trophic effect of NMDA and thus promotes apoptosis in the cerebellar neurons (Bhave and Hoffman, 1997). Similarly, ethanol promotes cell death in cerebellar granule neurons by inhibiting the antiapoptotic action of insulin-like growth factor-1 (Zhang et al., 1998).
Methyl mercury, a well established neurotoxicant known to cause Minamata disease, induced apoptosis in primary cultures of cerebellar neurons at low concentrations (up to 0.3 μM), but led to cell death in a nonapoptotic manner at higher concentrations (Kunimoto, 1994).
Two types of induction of apoptosis were reported in the rat CNS-derived neuroblastoma B50 and B104 cells. In both cell types, apoptosis could be induced by prolonged treatment with dibutryl cyclic AMP after differentiation and by cycloheximide treatment in undifferentiated proliferating cells. Whereas chromatin condensation and DNA fragmentation were observed with dibutyryl cyclic AMP, only chromatin condensation, but not DNA fragmentation, was observed with cycloheximide (Honma et al., 1996). A covalent dimer of interleukin-2 formed by the in vitro action of nerve-derived transglutaminase is known to be cytotoxic to rat brain oligodendrocytes. This compound induces apoptosis in cultured oligodendrocytes (Eizenberg et al., 1995). Colchicine, a microtubule-disrupting agent and known to be neurotoxic, induced apoptosis in cerebellar granule cells (Gorman et al., 1999).
MOLECULAR MECHANISM OF APOPTOSIS
The importance of apoptosis in regulating the number and types of cells in various regions of the developing CNS and PNS is now well accepted. It was shown recently that during the period of cerebral cortical neurogenesis in murine species, cells undergoing apoptosis were rare at embryonic day 10, but by embryonic day 14, 70% of cortical cells were dying, and at embryonic day 18 only 50% of cells were undergoing apoptosis (Blaschke et al., 1996). In the adult cortex, only a few cells were found to be dying. Significantly, the majority of cells undergoing apoptosis were observed in zones of cell proliferation rather than in regions of postmitotic neurons (Blaschke et al., 1996). A role for apoptosis in the myelinating Schwann cells was also reported; 50% of the neonatal Schwann cells cultured in axon-free medium underwent rapid apoptosis, which was prevented by the addition of Neu differentiation factors or by increasing the adhesion of Schwann cells to substratum, suggesting that the number of Schwann cells in myelinating nerves is regulated by apoptosis and depends on the survival signals from axons (Nakao et al., 1997).
The exact mechanism of apoptosis, i.e., the cascade of events starting from the detection of the signal at the cell surface to the events that occur in the nucleus, has not been established yet. However, many events that occur at the cell surface and intracellularly during apoptosis in the nervous system have been reported. Singh et al. (1996) noted that during apoptosis in a clonal hippocampal cell line (HN2-5) triggered by hypoxia or nutrient deprivation, a significant increase (10-20%) in the proportion of saturated fatty acid side chains in some, but not all, membrane phospholipids occurs. It is suggested that the selective change in phospholipid fatty acid composition may alter membrane fluidity, cause leakiness of lysosomal and nuclear membranes, and lead to DNA fragmentation. This line of investigation needs to be pursued.
The neurotrophic factors (NGF, BDNF, NT-3/4/5, CDF, CNTF, and some interleukins) play a vital role in the development of the nervous system and in neuronal survival. Lack of neurotrophic support leads to apoptosis in a variety of neurons, but the mechanism is unknown. The neurotrophins bind to two types of receptors at the cell surface: the tyrosine kinase type (P140trk receptors) and the neurotrophin type (P75NTR receptor). Whereas binding to P140trk is responsible for differentiation and survival, binding to P75NTR is believed to mediate apoptosis. In fact, it was shown recently that NGF (but not BDNF or NT-3) induces apoptosis in P75NTR-expressing human neuroblastoma SK-N-Mc cells (Kuner and Hertel, 1998). The intracellular domain of P75NTR contains a motif similar to death domains of the tumor necrosis factor receptor family. But, unlike the death domains of the tumor necrosis factor receptor, the P75NTR intracellular domain does not self-associate in solution. Its surface area is devoid of charged residues, suggesting it to be a potential site for interaction with other downstream targets (Liepinsh et al., 1997). Although the proximal signaling events that occur with tumor necrosis factor receptor leading to the activation of caspases and transcription factors are better understood (Ashkenazi and Dixit, 1998), similar information on P75NTR is lacking and requires further study. In contrast, a role for the suicide receptor Fas/Apo (apoptosis)-1 in cerebral cortical development was reported recently, and some of the downstream signaling factors resulting in caspase and transcriptional factor activation were detected (Cheema et al., 1999). There is, however, considerable evidence for the activation of certain genes and the involvement of the gene products in the apoptotic process; prominent among them are the proteins of the Bcl family, p53 and the ICE/caspases, and the available information is summarized in Fig. 1.
The products of the bcl-2 and bcl-xL protooncogenes are regulators of neuronal apoptosis and substitute for the neurotrophic factors in primary neuronal cells, as well as in established neuronal cell lines (Garcia et al., 1992; Allsopp et al., 1993; Mah et al., 1993; Zhong et al., 1993; Martinou et al., 1994; Sato et al., 1994; Gonzalez-Garcia et al., 1995; Greenlund et al., 1995b). The Bcl-2 protein is widely distributed in many parts of the CNS during embryonic development, but declines with age. In the PNS, neurons and supporting cells of sympathetic and sensory ganglia retain this protein throughout life (Merry et al., 1994). Transgenic mice that were made to overexpress human Bcl-2 proteins in their neurons showed reduced apoptosis and hypertrophy of the nervous system (Martinou et al., 1994; Farlie et al., 1995). In the adult human brain, two Bcl-2 mRNA species were identified in neurons, but not in glia, and the content of the Bcl-2 mRNA species was not decreased in patients with Alzheimer’s and Parkinson’s diseases, implying that components other than Bcl-2 may also modulate apoptosis (Vyas et al., 1997). During development of the brain in mice, the Bcl-2α and Bcl-2β mRNA levels were highest on embryonic day 15, some two to three times higher than adult levels (Abe-Dohmae et al., 1993), suggesting a role for Bcl-2 in neurogenesis. If sympathetic neurons are maintained in culture for several weeks, they lose their trophic factor dependence with time, but Bcl-2 expression is not required for the development of trophic factor independence in mature sympathetic neurons (Greenlund et al., 1995b). In contrast to reduced expression of Bcl-2 in the adult CNS, Bcl-xL and Bcl-xβ were expressed in both embryonic and adult neurons; microinjection of their cDNAs into primary sympathetic neurons prevented their death due to nerve growth factor withdrawal. Thus, Bcl-x proteins appear to regulate apoptosis in the adult CNS (Gonzalez-Garcia et al., 1995). However, Bcl-x-deficient mice died at embryonic day 13 due to extensive apoptosis, indicating that Bcl-x expression is necessary during early development also (Motoyama et al., 1995). Experiments on radiation-elicited apoptosis showed an increased ratio of Bax (a Bcl-2-related protein) to Bcl-2 + Bcl-xL, indicating that the ratio of the effector to repressor determines the susceptibility of the cells (Borovitskaya et al., 1996). Similarly, in cultured rat sympathetic neurons, growth factor starvation down-regulates Bcl-2 expression and treatment with Bax antisense oligodeoxynucleotides promotes neuronal survival in the threshold situation of insufficient trophic support (Gillardon et al., 1996). Induction of apoptosis with kainate (Gillardon et al., 1995) or retinoic acid (Okazawa et al., 1996) resulted in down-regulation of Bcl-2 and up-regulation of Bax. In contrast estrogen, which is neuroprotective, increased the expression of Bcl-xL in cultured hippocampal neurons (Pike, 1999).
It is interesting that bcl-2 rescues only sensory neurons that depend on the NGF family of neurotrophic factors for their survival, but not ciliary neurons that depend on CNTF, indicating that at least two death pathways exist in neurons that can be distinguished by their susceptibility to bcl-2 (Allsopp et al., 1993). The apoptotic pathways also appear to be different in undifferentiated and differentiated neurons; Bcl-2 or Bcl-xL can prolong survival of the differentiated neurons only (Eves et al., 1996). Similar evidence for multiple pathways for apoptosis exists in glial cells also (Llanos et al., 1998). Bcl-2 suppresses death of cells due to serum or growth factor withdrawal and also due to the Ca2+ ionophore A23187, glucose withdrawal, membrane peroxidation, and free-radical-induced damage, suggesting that it acts at a central step in apoptosis (Zhong et al., 1993). Recent reports suggest that Bcl-2 acts upstream of ICE/CED-3 proteases and inhibits apoptosis by preventing their posttranslational activation. This has been demonstrated in the neuronal cell line GT1-7 and in cells overexpressing Bcl-2. In these cells, Bcl-2 prevented the apoptosis-induced processing of pro-Nedd 2 to the cleaved form, Nedd, a member of the ICE/CED family (Srinivasan et al., 1996).
It is known that dimerization is involved in Bcl-2 function (Oltvai and Korsmeyer, 1994; Sedlak et al., 1995; Yang et al., 1995), but it is not clear whether apoptosis is prevented by homodimers like Bcl-2/Bcl-2 (Oltvai et al., 1993) or heterodimerization with related factors such as Bax is required (Sedlak et al., 1995). The expression of interacting factors in vivo such as Bax, Bag-1, Bak, or Bad has not been well studied in the nervous system. However, Bax mRNA is highly expressed in rat and mouse brain (Oltvai et al., 1993; Gillardon et al., 1994), and Bak mRNA was found in adult, but not in fetal, brain (Kiefer et al., 1995). It is likely that the expression of the various members of the Bcl-2 family has tissue- and cell-type specificity. In fact, in global ischemia, Bax protein was expressed in hippocampal CA1 neurons before delayed death, whereas Bcl-2 protein was not expressed. In contrast, Bcl-2 protein expression, but not that of Bax, was increased in CA3 neurons, which are less susceptible to ischemic injury. Both proteins were expressed in the dentate gyrus. This study supports the hypothesis that Bax, either independently or by forming heterodimers with gene family members other than Bcl-2, contributes to delayed death in these vulnerable neurons (Chen et al., 1996). Recently, two novel proteins, DP5 and Diva, which regulate apoptosis, were reported. When apoptosis was induced in cortical neurons with amyloid β-protein, DP5, a product of the DP5 gene, was expressed. It binds to Bcl-xL and promotes apoptosis (Imaizumi et al., 1999). The expression of Diva, a member of the Bcl-2 family of proteins, promotes apoptosis in sensory neurons by inhibiting the binding of Bcl-xL to Apaf-1, an adapter molecule that activates caspase-9 (Inohara et al., 1998). From the many studies described above, it is evident that Bcl-2 and Bcl-xL are antiapoptotic, whereas Bax, DP5, and Diva are proapoptotic. It appears that dimerization of these proteins is a necessary event, but it is not known yet which dimers cause apoptosis. Although there is evidence suggesting that these proteins act upstream of caspases, the steps in between in the pathway are yet to be established. Similarly, the immediate element upstream of Bcl-2 proteins in the pathway has not been identified so far.
Immediate early genes
There is now evidence for the activation of “immediate early genes” during apoptosis in neural cells. Enhanced expression of the transcriptional factors c-Jun and c-Fos (Estus et al., 1994, 1997; Anderson et al., 1995; Ferrer et al., 1996; Oo et al., 1999), increased levels of c-Jun mRNA (Miller and Johnson, 1996), and phosphorylation of c-Jun on its N-terminal transactivation domain indicating the involvement of AP-1 (Ham et al., 1995; Oo et al., 1999) were observed in neural apoptosis. Activation of p38 and c-Jun N-terminal kinase (JNK) group of mitogen-activated protein kinases was reported in apoptosis induced by glutamate (Kawasaki et al., 1997), as well as ischemia (Ozawa et al., 1999). Mice lacking the JNK-3 gene are resistant to kainic acid-induced apoptosis and show reduced levels of phosphorylated c-Jun and AP-1 transcription factor complex (Yang et al., 1997). The level of transcription factors binding to 12-O-tetradecanoylphorbol 13-acetate-responsive element increase when excitotoxic apoptosis is induced with NMDA, glutamate, or quisqualate (Finiels et al., 1995). Ref-1 is a bifunctional protein implicated in the transcriptional regulation of AP-1 and DNA repair. Loss of Ref-1 protein expression was shown to precede DNA fragmentation in apoptotic neurons (Walton et al., 1997). In contrast, nuclear factor-κB/Rel transcription factors contribute to NGF-dependent neuronal survival (Maggirwar et al., 1998). Thus, expression of c-Jun and c-Fos proteins, their phosphorylation, and formation of the AP-1 factor appear to be essential for neuronal apoptosis. It is most likely that AP-1 and nuclear factor-κB, which coordinate gene expression, determine commitment of neurons to apoptosis.
p53 and DNA strand breaks
The p53 tumor suppressor gene has been implicated in neuronal apoptosis. Primary cultures of rat neurons and oligodendrocytes constitutively express p53 protein (Eizenberg et al., 1996). During maturation of neurons in vitro, the subcellular localization of p53 changes. It is found in the cytoplasm in mature cells, whereas in the differentiating cells it localizes in the nucleus. Apparently, upon receiving appropriate signals, p53 moves into the nucleus and plays a regulatory role in directing primary neurons toward either differentiation or apoptosis (Eizenberg et al., 1996). When apoptosis was induced in cultured oligodendrocytes, a shift in location of p53 from the cytoplasm to the nucleus was observed, and infection of the cells with a recombinant retrovirus encoding a C-terminal p53 miniprotein (p53DD), an inhibitor of wild-type p53 activity, protected them from apoptosis (Eizenberg et al., 1995). Overexpression of p53 with a recombinant adenovirus vector carrying wild-type p53 gene was sufficient to induce apoptosis in sympathetic (Slack et al., 1996) and hippocampal neurons (Jordan et al., 1997a). Cerebellar neurons in culture from p53 null mice, but not from the wild type, were protected from cytosine arabinoside toxicity (Enokido et al., 1996a). Apparently, cytosine arabinoside induces apoptosis in a p53-dependent but JNK-independent manner (Anderson and Tolkovsky, 1999). DNA-damaging agents like ionizing radiation, etopside, or bleomycin also induced apoptosis in wild-type but not in p53 null mutant mice cerebellar neurons, indicating that p53 is required for apoptotic death due to DNA strand breaks (Enokido et al., 1996b). There is evidence that damage to nuclear DNA caused by neuronal ischemia leads to accumulation of unrepaired DNA single-strand breaks and triggers apoptosis (Chen et al., 1997). It appears that neurons undergo apoptosis by two pathways: one p53-dependent and the other p53-independent. Ionizing radiation induces apoptosis by a p53-dependent pathway, whereas methylazoxymethanol, a genotoxic agent, causes apoptosis by a p53-independent pathway and without showing DNA fragmentation (Wood and Youle, 1995; Johnson et al., 1998). Thus, it is clear that p53 plays a significant role in apoptosis following DNA damage. It is also an essential component of the two apoptotic signal pathways: the NGF/TrkA survival cascade and the P75NTR apoptotic signal cascade and functions downstream of JNK and upstream of Bax (Aloyz et al., 1998).
Cell cycle and topoisomerases
There is now considerable evidence suggesting that apoptosis in postmitotic neurons may be an aborted attempt to reenter the mitotic cycle. Okadaic acid, a specific protein phosphatase inhibitor, induces apoptosis in human neuroblastoma cell lines. It increases the cells in S phase, leads to the formation of cyclins B1 and D1, markers of the cell cycle, and forces the cells to enter the mitotic cycle, although they fail to complete the cycle (Nuydens et al., 1998). In sympathetic neurons deprived of NGF, a selective expression of cyclin D1 was observed in dying neurons (Freeman et al., 1994). Similarly, cyclin D1 levels increased in apoptosis induced by kainate (Giardina et al., 1998), as well as low K+ (Boutillier et al., 1999), in cerebellar neurons. Besides an increase in cyclin D1 levels, stimulation of cyclin D1-dependent kinase was noticed in neuronal apoptosis, and the cyclin D-dependent kinase inhibitor p161NK4 protected them from apoptotic cell death, suggesting that activation of endogenous cyclin D1-dependent kinase is essential for neuronal apoptosis (Kranenburg et al., 1996).
Camptothecin, a DNA topoisomerase-1 inhibitor, induces apoptosis in neurons (Morris and Geller, 1996). This phenomenon is suppressed by G1/S cell-cycle blockers, as well as by the cyclin-dependent kinase inhibitors, but not by DNA synthesis inhibitors. It appears that neuronal apoptosis caused by camptothecin involves signaling pathways that control the cell cycle (Park et al., 1997). All the arabinoside nucleosides (adenine, cytosine, guanine, and thymine) when combined with 5-fluorodeoxyuridine induce apoptosis in sympathetic neurons cultured in the presence of NGF, presumably by causing double-strand breaks. Similarly, topoisomerase II inhibitors, teniposide and mitoxanthrone, which form double-strand breaks, also induce apoptosis (Tomkins et al., 1994). This would mean that double-strand breaks could be a mechanism of induction of apoptosis.
Ever since the cell death gene ced-3 in C. elegans was identified as the prototype of the ICE protease family, several ICE proteases have been discovered and implicated in apoptosis. Of these, the CPP32/Yama/apopain exhibits the highest similarity to CED-3 in sequence homology and substrate specificity. In CPP-32-deficient mice, brain development was profoundly affected, resulting in a variety of hyperplasias and disorganized cell deployment, indicating decreased apoptosis. Thus, CPP32 plays a critical role during morphogenetic cell death in mammalian brain (Kuida et al., 1996). An ICE-related protease (IRP) from rat brain that shows strong sequence homology to human CPP-32 was cloned. In situ hybridization histochemistry showed that this IRP mRNA is expressed in neuron-enriched regions of developing brain, but profoundly down-regulated in adult brain. In cultured cerebellar granule neurons undergoing apoptosis, an overexpression of IRP mRNA was observed, and simultaneously cleavage of the putative death substrate poly(ADP-ribose) polymerase was also noted, suggesting that transcriptional activation of IRP is involved in the mechanism of apoptosis (Ni et al., 1997). Treatment with bocaspartyl(O-methyl)-fluoromethyl ketone (BAF), a cell-permeable inhibitor of ICE proteases, blocked apoptosis in sympathetic neurons, but did not prevent the fall in protein synthesis or increase in the expression of c-Jun, c-Fos, and other mRNAs, implying that ICE proteases function downstream of these events (Deshmukh et al., 1996). Similarly, the ICE protease inhibitor benzyloxycarbonyl-Asp-CH2OC(O)-2,6-dichlorobenzene (Z-D-DCB), as well as the calpain inhibitors 3-(4-iodophenyl)-2-mercapto-(Z)-2-propenoic acid (PD 150606) and N-acetyl-Leu-Leu-Met-aldehyde (calpain inhibitor II) protect cultured rat cerebellar granule neurons from apoptosis, but staurosporine-induced apoptosis in neuroblastoma was prevented only by Z-D-DCB and not by calpain inhibitors (Nath et al., 1996). Peptide inhibitors of ICE proteases also prevented apoptosis in motoneurons in vivo, as well as in vitro (Milligan et al., 1995). On the other hand, the ICE protease inhibitor, benzyloxycarbonyl-Val-Ala-Asp-fluoromethyl ketone (ZVAD-fmk) prevents staurosporine-induced apoptosis in cerebellar neurons, but not that induced by K+ deprivation (Taylor et al., 1997). Therefore, it appears that ICE proteases play a general role in neuronal apoptosis, although different enzymes may be involved, depending on the inducing agent, and calpain may be cell-specific.
A number of selective inhibitors were used to differentiate the various caspases. The viral caspase inhibitor p35 and ZVAD-fmk are broad-spectrum inhibitors and protected apoptosis induced with staurosporine (McCarthy et al., 1997) and lysophosphatidic acid (Holtsberg et al., 1998), respectively. N-Acetyl-Asp-Glu-Val-Asp-fluoromethyl ketone (DEVD-fmk)-sensitive caspase, which is different from CPP32 and Nedd2, mediates the low K+-induced apoptosis (D’Mello et al., 1998). The specific involvement of CPP32-like caspase in MPP+-mediated apoptosis was shown through its tetrapeptide inhibitor, acetyl DEVD-aldehyde (Du et al., 1997). Thus, it is clear that ICE-like caspases are usually involved in the mechanism of neuronal apoptosis, but different enzymes appear to be triggered, depending on the initial apoptotic signal. However, radiation-induced neuronal death may occur in a caspase-independent manner (Johnson et al., 1998).
Oxidative stress, SOD, and cytochrome c
Oxygen stress, SOD, and cytochrome c play a role in the apoptotic signaling pathway. Chronic inhibition of Cu/Zn-SOD with either antisense oligodeoxynucleotides or diethyldithiocarbamate in spinal cord organotypic cultures resulted in apoptotic degeneration, which was prevented by the antioxidant, N-acetylcysteine (Rothstein et al., 1994). In transgenic mice overexpressing SOD, there is a chronic prooxidant state with reduced levels of glutathione and altered Ca2+ homeostasis that exacerbates kainic acid-induced apoptosis (Bar-Peled et al., 1996). Mn(III) tetrakis(benzoic acid) porphyrin, a SOD mimic, protects neurons from staurosporine-induced apoptosis (Patel, 1998). A high (50%) oxygen atmosphere (Enokido and Hatanaka, 1993) and glutathione depletion due to oxidative stress (Ratan et al., 1994) are causative factors for neuronal apoptosis. High oxygen generates reactive oxygen species (ROS), enhances glutamate release, and activates caspases (Ishikawa et al., 1999). Primary hippocampal neurons from presenilin-1 gene knock-in mice showed increased superoxide production and caspase activation when treated with amyloid β-peptide (Guo et al., 1999). Amylin treatment of cortical neurons results in induction of oxidative stress genes, as well as transcriptional factors (Tucker et al., 1998). The role of transcription factors in mediating apoptosis due to increased levels of ROS has been reviewed recently (Tong et al., 1998). ROS accumulate in NGF-deprived sympathetic neurons (Greenlund et al., 1995a) and in 3-hydroxykynurenine-induced apoptosis in striatal neurons (Okuda et al., 1998) and act as signals. In NMDA-induced apoptosis, ROS formation was blocked completely by caspase inhibitors, suggesting that their site of action is probably downstream of caspases (Tenneti et al., 1998).
Serum-deprived cerebellar granule neurons show impaired mitochondrial function and increased oxidation during apoptosis, and glutamate antagonists provide neuroprotection, implying that oxidative stress precedes glutamate receptor activation (Atabay et al., 1996). A steady decrease in mitochondrial function occurs in low K+-induced apoptosis in cerebellar neurons (Nardi et al., 1997), whereas inhibitors of mitochondrial function enhance the dopamine-induced apoptotic death of striatal neurons (McLaughlin et al., 1998). A reduction in mitochondrial membrane potential was noticed in lysophosphatidic acid-induced apoptosis in hippocampal neurons (Holtsberg et al., 1998). Release of cytochrome c from mitochondria was found in cerebellar neurons when apoptosis was induced with MPP+ (Du et al., 1997) or colchicine (Gorman et al., 1999). It appears therefore that neurotoxins like MPP+, as well as glutamate and dopamine, induce free radical formation and disrupt mitochondrial membrane potential, which promotes cytochrome c release, an apoptogenic protein, and ultimately lead to activation of caspases. However, release of cytochrome c does not appear to be sufficient to induce apoptosis, and a further event independent of macromolecular synthesis and Bax function is necessary (Deshmukh and Johnson, 1998).
The production of NO is a key event in certain apoptotic signaling pathways. Treatment of mixed neuronal/glial cells in culture with cytokines results in a high output of NO and apoptotic death of the neurons, which is prevented by the NO synthase inhibitor N-monomethyl-L-arginine (S. Hu et al., 1997). Similarly, when astrocyte-neuron cocultures are treated with S100β, a glial-derived neurotrophin, inducible NO synthase is activated, leading to apoptosis (J. Hu et al., 1997). Serotonergic neurons when treated with 3,4-methyl-enedioxymethamphetamine undergo apoptosis with the formation of NO (Simantov and Tauber, 1997). NO donors (S-nitrosocysteine or S-nitroso-N-acetylpencillamine) induce apoptosis in cultured cerebellar granule cells. This effect involves tyrosine nitration, Ca2+ influx through NMDA-gated channels, and breakdown of actin filaments and microtubules, resulting in apoptosis (Bonfoco et al., 1996). These reports clearly indicate the involvement of NO in certain apoptotic signaling pathways. Recent reports suggest that NO releases cytochrome c (Uehara et al., 1999), and the apoptotic mechanism includes caspase-3 activation after down-regulation of Bcl-2 and up-regulation of Bax protein levels (Tamatani et al., 1998).
Calcium and cAMP
There is convincing evidence showing that the intracellular [Ca2+] plays a crucial role in the apoptotic phenomenon. Ca2+ is necessary for the activation of endonucleases, and a specific Ca2+-dependent endonuclease, believed to play a role in apoptosis, was identified in thymocytes (Gaido and Cidlowski, 1991). A Ca2+-regulated protease has been implicated in staurosporine-induced apoptosis in hippocampal neurons (Jordán et al., 1997b), and these neurons overexpressing calbindin D28K are less susceptible to apoptosis (Prehn et al., 1997). The Ca2+ ionophores, A23187 (Joseph et al., 1993) and ionomycin (Takei et al., 1995) and the selective inhibitor of endoplasmic reticular Ca2+-ATPase thapsigargin (Wei et al., 1998) elevate intracellular [Ca2+] and cause apoptosis. However, in low K+-induced apoptosis in cerebellar granule cells, a decrease in intracellular [Ca2+] was observed, and the L-type channel agonists enhanced the survival of these neurons (Galli et al., 1995). The intracellular [Ca2+] increased with time in dorsal root ganglion neurons in culture, and simultaneously their dependence for survival on NGF decreased. If the intracellular [Ca2+] is lowered by reducing external [Ca2+], they again became dependent on NGF, suggesting that there is a relationship between intracellular [Ca2+] and their vulnerability to apoptosis (Tong et al., 1996). Thus, it appears that there is a narrow range of intracellular [Ca2+] that is physiologically acceptable, and a variation from it on either side will result in the cell’s commitment to apoptosis. Increased intracellular [Ca2+] leads to mitochondrial membrane depolarization and generation of ROS in apoptotic cells (Kruman and Mattson, 1999; Mason et al., 1999). The further events in Ca2+-evoked apoptosis may include activation of calcineurin, dephosphorylation of Bad in cytosol, and its translocation to mitochondria, where it will dimerize with a member of the Bcl family of proteins and thus promote apoptosis (Wang et al., 1999). Supporting this mechanism, it was observed that Bcl-2 inhibits Ca2+-evoked apoptosis (Wei et al., 1998; Kruman and Mattson, 1999). It may be concluded that Ca2+ acts essentially upstream of the Bcl group of proteins.
Cyclic AMP is a recognized second messenger and mediates many cellular responses. Cyclic AMP was shown to inhibit apoptosis of cerebellar granule cells in a low [K+] medium (D’Mello et al., 1993). Similarly, dibutyryl cyclic AMP attenuates ethanol-induced apoptosis in endorphin neurons in fetal hypothalamic cell cultures (De et al., 1994), but the mechanism of the modulatory effect of cyclic nucleotides is completely unknown. It is possible that cyclic AMP plays a role in apoptosis via protein kinases that phosphorylate the Bad and other proapoptotic members and acts in concert with Ca2+.
Ceramide and phosphatidylinositol 3-phosphates
There are many reports indicating that sphingolipids are involved in the regulation of cell growth, differentiation, and apoptosis in neurons. Ceramide is required for survival and dendritic differentiation of Purkinje cells (Furuya et al., 1998). Glucosylceramide sustains axonal growth, and its short-chain analogues promote formation of minor processes, whereas a higher concentration of ceramide or sphingomyelinase induces apoptosis (Schwarz and Futerman, 1997). In embryonic chick cerebral neuron cultures, staurosporine induces apoptosis via the hydrolysis of sphingomyelin to ceramide. N-Acetylsphingosine (C2-ceramide), a soluble analogue of ceramide, mimics the effects of staurosporine, and the ceramidase inhibitor oleoylethanolamine enhances the apoptotic effect, suggesting a crucial role for ceramide in the apoptotic signal pathway (Wiesner and Dawson, 1996). Similarly, apoptosis could be induced with native ceramide or C2- or C6-ceramide in dopaminergic and other neurons of mesencephalon in primary cultures (Brugg et al., 1996). A mouse neuroblastoma × dorsal root ganglion hybrid cell line transfected to overexpress κ-opioid receptors (F-11K7) and chronically treated with a κ agonist, showed enhanced susceptibility to apoptosis by staurosporine or the phosphatidylinositol 3-kinase inhibitors, wortmannin and LY294002. In both these instances, the apoptotic signal was mediated through ceramide (Dawson et al., 1997).
The stress-activated protein kinases are activated in response to a variety of stresses and lead to apoptosis. The process appears to be ceramide-dependent. The C2-ceramide-induced apoptosis in cerebellar granule neurons was prevented by inhibitors of protein kinases, isoquinoline sulfonamides, suggesting that ceramide may act through a protein kinase (Manev and Cagnoli, 1997). In fact, in chick embryo astrocytes in culture, staurosporine-induced apoptosis was associated with increased formation of ceramide and activation of 60-kDa serine/threonine protein kinase (PK60), and these events appear to be early steps in the signaling pathway (Mangoura and Dawson, 1998). In other systems, however, ceramide is believed to activate either a proline-directed 97-kDa kinase (Liu et al., 1994) or a phosphatase (Hannun, 1994) or protein kinase Cξ (Muller et al., 1995), leading to activation of ICE-like proteases, endonucleases, and DNA fragmentation. Although most reports show that an elevated level of ceramide causes apoptosis, Ito and Horigome (1995) observed that ceramide actually prevents neuronal PCD induced by NGF withdrawal in sympathetic neurons. Thus, it appears that a certain critical level of ceramide may be necessary for the cell’s survival, whereas excess levels are apoptotic.
There is evidence for the involvement of phosphatidylinositol 3-phosphates in the apoptotic and survival signal pathways. Insulin-like growth factor-1 (D’Mello et al., 1997) and several neurotrophins (Skaper et al., 1998) rescue cerebellar neurons from low K+-induced apoptosis. In these instances, there is a robust increase in phosphatidylinositol 3-kinase activity followed by suppression of activation of JNK and c-Jun expression (Shimoke et al., 1999). This implies that products of phosphatidylinositol 3-kinase act at the level of transcription factors and thus attenuate apoptosis. In fact, addition of phosphatidylinositol 3,4-bisphosphate and phosphatidylinositol 3,4,5-trisphosphate to the medium is sufficient to prevent neuronal death (Shimoke et al., 1998). However, the exact step at which these lipids act in the signal cascade remains to be elucidated.
Glyceraldehyde-3-phosphate dehydrogenase (GAPDH)
GAPDH is a key enzyme in glycolysis and a highly conserved protein. Besides its enzymic role, it has several other functions, which include microtubule-bundling facilitation (Huitorel and Pantaloni, 1985), protein kinase modulation (Kawamoto and Caswell, 1986), and activation of transcription in neurons (Morgenegg et al., 1986). Ishitani et al., (1996), for the first time, showed that the 38-kDa GAPDH associated with the particulate fraction is overexpressed in age-induced apoptosis in cerebellar granule cells, and this was prevented by pretreatment with cycloheximide and actinomycin D. The GAPDH mRNA level increased and GAPDH antisense oligodeoxyribonucleotide blocked the expression of GAPDH and apoptosis in this system, thus suggesting a direct role for this protein in age-induced apoptosis in cerebellar neurons. A similar overexpression of GAPDH was observed in cytosine arabinonucleoside-induced apoptosis in cultured neurons also (Ishitani and Chuang, 1996). Brain GAPDH is known to undergo ADP-ribosylation under conditions of NO production (Vaidyanathan et al., 1993). NO-induced cell death in human neuroblastoma SH-SY5Y involves ADP-ribosylation of GAPDH and poly(ADP-ribose) polymerase (Kamoshima et al., 1997). During apoptosis, GAPDH translocates to the nucleus (Saunders et al., 1999). Thus, there is evidence for a role for GAPDH in the apoptotic process, but the step at which it is involved in the signal cascade is not known.
It is well established that apoptosis is a normal feature in developing brain and may play a role in some neurodegenerative diseases and aging as well. Therefore, this phenomenon has attracted enormous attention during the past few years mainly to delineate the mechanistic details of the apoptotic pathway in the nervous system. A variety of neuronal preparations from various parts of the brain, the majority of which were primary cultures, and some cell lines were investigated. Several apoptosis-inducing agents were identified, and these include lack of neurotrophic support, neurotransmitters, neurotoxicants, modulators of protein phosphorylation and Ca2+ homeostasis, DNA-damaging agents, oxidative stress, NO, and membrane lipids. In most of these studies, apoptosis was established by internucleosomal DNA cleavage and the appearance of the DNA ladder, the hallmark of apoptosis. It is not clear which of the above-mentioned agents act as signal generators and which may only be present in the apoptotic signal cascade.
The signal pathway should consist of recognition elements at the cell surface (or intracellularly), events in or at the neuronal plasma membrane, modulation of cytosolic contents, and eventually, reactions that occur in the nucleus, resulting in internucleosomal DNA fragmentation. As yet, information on most of these steps is grossly inadequate, and the cascade of the apoptotic signal pathway is far from clear. From the available evidence, the broad outlines of the apoptotic pathway can be envisaged as shown in Fig. 1. Although it is clear that withdrawal of growth factors results in apoptosis, the activation or inactivation of the relevant receptors and events at the plasma membrane are yet to be identified. Also, whether these signals are coupled to sphingomyelinase or not is not established. But it appears certain that the apoptotic pathway involves the Bcl family of proteins, which in turn affect the processing of pro-ICE/caspases. Ceramides may also bring about apoptosis by interfering with the processing of pro-ICE/caspases, but there may be some intermediate steps, such as activation of protein kinases/phosphatases, that have not yet been investigated. There is evidence suggesting that Ca2+ and ROS act upstream of Bcl proteins. In another cascade, c-Jun, and c-Fos, and their phosphorylated products appear to be involved. These signaling pathways, as well as that of cytochrome c, seem to converge at the ICE/caspases step. There is evidence for the involvement of DNA double-strand breaks and p53 in the apoptotic signal pathway, but whether they act through caspases or through some other mechanism is not clear. Similarly, the role of cyclic AMP and cyclin-dependent kinases, NO, and GAPDH in the apoptotic pathway has not been investigated sufficiently. On the whole, it appears that there may be several apoptotic pathways that may depend on the cell type and the inducing agent, and most of the pathways may converge at the ICE/caspase step. However, the expression of caspases may be regulated by the antiapoptotic genes [inhibitor of apoptosis (IAP) family]. In a recent report, it was shown that the forced expression of IAPs in cerebellar granule neurons blocked K+-induced apoptosis with concomitant blockade of caspase activity (Simons et al., 1999). Further work on the role of the IAPs in developmental apoptosis would be rewarding. Finally, the activated caspase may recruit additional factors that trigger DNA fragmentation. In this context, it is noteworthy that a novel protein, designated as DNA fragmentation factor, was purified from HeLa cells and acts downstream of caspase (Liu et al., 1997). This factor is composed of two subunits, a 40-kDa nuclease and a 45-kDa inhibitor. During apoptosis, the nuclease is activated by a caspase-3-mediated cleavage of the inhibitor (Inohara et al., 1999). However, the existence of DNA fragmentation factor in the nervous system is yet to be demonstrated.
This work was supported by a grant from the Department of Science and Technology, Government of India (SP/SO/B-21/95). P.S.S. wishes to thank the Indian National Science Academy, New Delhi, for the Senior Scientist Fellowship. The authors are grateful to Dr. N. S. Raji for assistance in preparation of the manuscript.