Chapter 18
Programmed Cell Death and Neurotrophic Factors
Major hallmarks of embryonic development include the generation of new cells, the growth of tissues and organs, and the gradual but inexorable acquisition of new cellular and morphological properties (phenotypes). Accordingly, in the past, a major focus of developmental neurobiologists has been the study of progressive events, such as those described in several other chapters in this section. In this context, the concept of significant regressive events, such as programmed cell death (PCD), occurring during development, initially was considered counterintuitive (Oppenheim, 1981; 1991). Subsequently, however, it has been demonstrated that cell loss and other regressive events, including synapse elimination (Chapter 19), are the rule rather than the exception. In most developing tissues that have been examined in multicellular organisms, substantial cell loss occurs. As we discuss later, even a significant proportion of adult-generated neurons in the hippocampus and olfactory system also undergo normal PCD (Gage, Kempermann, & Song, 2008). Genetic programs that result in cell death have been suggested to be a default pathway for all cells. In this view neurons would escape this default fate only by receiving the appropriate receptor-mediated survival signals (e.g., neurotrophic factors, NTFs) that induce changes in the molecular and biochemical events required for cell survival and neuronal differentiation. In the absence of these ligands, the unbound receptors may activate death pathways and have been called “dependence-receptors.” Neurons that express dependence receptors require—that is, are dependent upon—the presence of ligands for their anti-apoptotic role in promoting survival; however, in the absence of ligands, these neurons do not die by default, but rather the unbound receptors are actively pro-apoptotic and trigger cell death. Aberrations in the mechanisms regulating the balance between cell proliferation and cell elimination can lead to pathologies of tissue growth and differentiation by reduced cell death (e.g., cancer) or excessive neuronal death (e.g., neurodegenerative disease). Mechanisms that regulate the balance between cell production and death serve to determine the ultimate number and maintenance of neurons in the nervous system. Genes regulating these mechanisms provide an underlying substrate for evolutionary changes in brain size and structure and ultimately in adaptive behavior.
Both progressive and regressive events during development are regulated by intercellular signals. Somewhat surprisingly, many of the same intercellular signals (e.g., NTFs) that contribute to the regulation of progressive events during nervous system development also contribute to the control of regressive events. Neurotrophic factors are now appreciated for their role in both survival- and nonsurvival-related activities in both developing and mature nervous systems where they play important roles in the regulation of activity-dependent structural and functional plasticity of the nervous system. Accordingly, we include here a review of both survival and nonsurvival functions of NTFs.
Because neurons may die for a variety of reasons and in many different situations, it is important to describe the type of cell death observed most often in the developing nervous system. Although the loss of cells during normal development has been called by many different names (normal cell death, spontaneous cell death, naturally occurring cell death, apoptosis, and developmental cell death), we prefer the term programmed cell death (PCD). PCD is defined as the spatially and temporally reproducible and species-specific loss of large numbers of individual cells both during development and, in many tissues and organs, throughout life, as adult cells die and are replaced by precursor or stem cells as a means of maintaining stable structure and function (homeostasis). Accidental, injury-induced, pathological, and disease-related forms of cell death are not included even though we recognize that the biochemical and molecular mechanisms used to kill cells in these situations may overlap with those involved in developmental PCD and in some cases, like PCD, the death may even be genetically controlled. This definition of PCD also makes no a priori assumptions about either the morphological or the biochemical pathways by which cells die or the stimuli that trigger cell death. The use of the word “programmed” refers to the reproducible, spatiotemporally species-specific occurrence of cell loss and is not meant to imply that the cell loss is genetically predetermined, inherited from precursor cells, or inevitable. Finally, and as we discuss in more detail later, PCD is not synonymous with the term apoptosis, which refers only to one specific, albeit common, mode of developmental cell death.
Cell Death and the Neurotrophic Hypothesis
Early embryological studies of the interactions between developing neurons and their peripheral targets laid the foundation for the discovery of cell death and the first NTF (Cowan, 2001). These studies formed the conceptual framework for the neurotrophic (“nerve feeding”) hypothesis (Fig. 18.1). By 1949, Viktor Hamburger and Rita Levi-Montalcini had provided compelling evidence to support the significance of normal embryonic neuronal death and postulated that target-derived signals act to regulate the number of neurons that survive embryonic development. In subsequent studies, they and their colleagues identified a specific protein, nerve growth factor (NGF), that influenced development of the same populations of neurons (sensory and sympathetic) that their earlier studies had suggested were regulated by target-derived signals.
Figure 18.1 The neurotrophic hypoth- esis postulates that developing neurons survive (green neurons) only when they successfully compete for target-derived trophic molecules (blue) that are internalized at the nerve terminal and are transported retrogradely along the axon (arrow) to the cell body. Neurons that fail to obtain a sufficient flow of trophic molecules from the target die by programmed cell death (black neurons with condensed chromatin).
Modified and reproduced with permission from Barde (1989).
By 1960, the preparation of specific antibodies that block NGF activity allowed Stanley Cohen, Levi-Montalcini, and co-investigators to demonstrate the almost total degeneration of sympathetic ganglia in vivo following the specific deprivation of NGF activity. Only with the evidence from this “immunosympathectomy” (the deletion of the sympathetic system by antibody treatment) was NGF considered to be an endogenous survival or maintenance factor for these neurons. Three decades after the original discovery that an unknown chemical substance produced by tumor cells affects the development of sensory and sympathetic neurons, it was finally recognized that NGF was the hypothetical target-derived trophic signal first postulated by Hamburger and Levi-Montalcini in 1949 to be involved in regulating the number of surviving neurons in these populations during normal development. Historically, then, the appreciation of the role of PCD in the developing nervous system and the discovery of the first target-derived NTF (NGF) were inextricably linked; together they provided the foundation for the neurotrophic hypothesis, and both owe their origin to earlier studies of interactions between developing neurons and their synaptic targets (Oppenheim, 1991).
Despite the significance of this conceptual advance and the singular role of the neurotrophic hypothesis in guiding research in this field for over half a century, new evidence indicates that the neurotrophic hypothesis is in need of revision in order to encompass the increased appreciation that the cell–cell interactions that regulate neuronal survival and death are more complex than originally thought. It is now clear, for example, that the efferent targets of neurons are only one of several sources of survival-promoting NTFs and that NGF is only one of many diverse NTFs that regulate development and survival.
The Origins of Programmed Cell Death and its Widespread Occurrence in the Developing Nervous System
PCD has been identified in plants, several species of unicellular eukaryotes, including yeast, as well as in prokaryotes such as bacteria, one of the oldest forms of life on earth (Ameisen, 2004). One likely explanation for the evolutionary origins of PCD in eukaryotes was the appropriation of cell death–associated genes from bacterial mitochondria by a process of endosymbiosis. Based on the principle of genetic pleiotropy, it seems likely that under the right conditions genes required for vital functions such as metabolism, proliferation, and differentiation can also serve pro- and anti-cell death functions and that this pleiotropy played a major role in their later selection in unicellular eukaryotes and metazoans. Even some of the core genetic machinery currently employed for regulating PCD in mammals, including neuronal PCD, has retained a degree of pleiotropy. Both caspases and Bcl-2 family members mediate a number of nondeath functions in both neuronal and nonneuronal cells (Jiao & Li, 2011).
After being incorporated into the eukaryotic genome, this core cell death machinery expanded in complexity over time to include the diversity found in extant multicellular organisms. Once the pro- and anti-apoptotic machinery was in place in the genome of multicellular animals, it could then be co-opted to mediate the many diverse roles now served by PCD, including different functions of PCD in developing neurons. Finally, a word of explanation is needed regarding the seemingly counterintuitive existence of PCD in unicellular organisms—that is, how can the death of individual organisms be beneficial? In fact, several adaptive roles of PCD in these organisms have been reported (Nedelcu, Driscoll, Durand, Herron, & Raskidi, 2010).
Although a systematic taxonomic survey of all representative species with a nervous system has not been done, nonetheless the available evidence is consistent with the idea that some PCD of developing neurons occurs in all such organisms. For many invertebrate and vertebrate species, the PCD of neurons involves virtually all regions and cell types in the central and peripheral nervous system. Motoneurons, sensory neurons, autonomic neurons, and both long projection and local circuit interneurons in the brain and spinal cord all undergo restricted periods of PCD. Although the magnitude of neuronal cell death varies from population to population, as many as one-half or more of all cells in a population will die during development (Fig. 18.2). In some special cases, such as the loss of transient neuronal structures during insect and amphibian metamorphosis (e.g., the Rohon-Beard sensory neurons in fish and frogs), most or all cells die. Cell death in the nervous system thus clearly occurs on a very large scale, indicating that it plays fundamental and essential roles in normal development.
Figure 18.2 Changes in the number of developing lumbar spinal motoneurons in the chick embryo (in red) compared to the number of dying motoneurons (in blue) during the first half of the 21 day incubation period. Approximately 50 to 60% of these motoneurons undergo PCD between embryonic day (E)6 and E12.
Studies of cell death in the nervous system have focused on the loss of developing postmitotic neurons as they form synaptic connections with targets and afferents. However, extensive PCD also occurs during neurulation and in mitotically active neural progenitor cells, as well as in postmitotic but undifferentiated neurons in the early neural tube (Fig. 18.3). Accordingly, PCD in the nervous system is not limited to any particular stage of development. The loss of cells at different developmental stages probably serves distinct functions and may be mediated by different mechanisms. PCD also occurs in central and peripheral glial cells. For example, like neurons, myelin-forming oligodendrocytes in the optic nerve and Schwann cells in peripheral nerves are overproduced, and many die by PCD. Their loss is thought to reflect a competition for axon-derived trophic signals, the end result of which is the survival of an appropriate number of glial cells for optimal myelination of the available axons (Fig. 18.4).
Figure 18.3 Schematic illustration of some key steps in neuronal development. Neurons undergoing PCD (
) are observed during neurogenesis in the ventricular zone, during migration and while establishing synaptic contacts. Schwann cells in developing nerves also undergo PCD. 
Represents peripheral glial (Schwann) cells; 
represents CNS glia (astrocytes, oligodendrocytes); 
represents surviving, differentiating neurons (e.g., motoneurons whose targets are skeletal muscle and other neurons in the CNS, with neuronal targets).
Figure 18.4 The PCD of glial cells is regulated by axon-derived signals. Glial cells that fail to compete successfully for these signals undergo PCD.
Because PCD is the normal differentiated or terminal fate of many developing cells, commitment to this fate occurs in much the same way as the phenotypic fate of cells destined to survive in the embryo. Developmental biologists have identified two major ways in which the commitment of a cell to a particular differentiated phenotype occurs. The first mechanism, intrinsic or autonomous specification, involves the segregation of critical cytoplasmic molecules by asymmetric cell division; cell death thus is programmed into the lineages that generate somatic cells. The second mechanism of commitment involves extrinsic signals from other cells and is called conditional specification; initially, the cells have the potential to follow more than one path of differentiation, but as development proceeds, signals from other cells act to gradually limit and specify cell fate. Although all organisms use a combination of autonomous and conditional developmental strategies, as a general rule invertebrates are more likely to utilize autonomous specification, whereas most vertebrates exhibit conditional specification. Despite this distinction, as described later, many of the genetic and molecular pathways for PCD are remarkably similar in invertebrates and vertebrates. In addition to these two major pathways for fate determination, as we discuss below, there is increasing evidence for a third pathway involving stochasticity: random fluctuations (noise) in gene expression (fate determination by the “roll of the dice”).
Summary
The PCD of developing neurons appears to occur in virtually all vertebrate and invertebrate species. More generally, PCD also occurs in many different cells and tissues of unicellular and multicellular organisms and thus may have arisen very early during evolution. With few exceptions, PCD occurs in virtually all types of developing neurons and can take place at stages of development from the time of proliferation until the establishment of synaptic connections. Developing glial cells also exhibit PCD. Cell death is the terminal phenotypic fate of subpopulations of developing neuronal and glial cells and, like other cell fate decisions, is controlled by both intrinsic and extrinsic signals.
Functions of Neuronal Programmed Cell Death
Why does PCD occur? This is a reasonable question to ask because the loss of large numbers of developing neurons is counterintuitive. Why should embryos invest precious resources in generating cells and tissues only to later cast many of these aside? A satisfactory answer to this apparent paradox requires an evolutionary perspective that addresses two central aspects of the problem. First, how did the biochemical machinery (the cell death program) needed to actively kill cells arise? Second, why, in many developing tissues, are more cells generated than are apparently needed?
The first question was addressed in the previous section; however, understanding the evolution of the biochemical cell death program does not help answer the second question of why there is often a massive overproduction of neurons during development that are later eliminated by PCD. Two explanations have been offered (Oppenheim, 1991). First, each case of PCD may have evolved to serve a distinct biological function. For example, according to this view, in the case of spinal motoneurons, natural selection is thought to be directly responsible for both the overproduction and the subsequent death of neurons as a means for creating an optimal (adaptive) level of functional muscle innervation (e.g., systems-matching).
The second view is that the overproduction of neurons (or other cells) is an inevitable outcome of the imprecise kinetics of proliferation of precursor cells. Once too many (or too few) cells are generated and the pro- and anti-apoptotic machinery is in place, natural selection then acts via regulated cell survival and death to mediate a variety of different adaptive needs. For example, following the loss or absence of essential survival signals, the death of excess cells could be accomplished easily by co-opting the cell death machinery that evolved to kill cells in early eukaryotes. In reality, both of the proposed mechanisms may occur (Table 18.1). Many of the circumstances in which the PCD of neurons occurs are thought to mediate distinct adaptive functions, as summarized in Table 18.1. In many of these examples, the production of excess neurons provides a substrate on which differential survival or PCD can then act to meet a variety of putative adaptive needs. Similarly, if fewer numbers of cells are generated, then compensation, by reducing the extent of cell death, provides a means for ensuring, for example, that a sufficient number of neurons are still available for optimal target innervation. As new genes are identified that regulate PCD, and as we gain a better understanding of how cellular and molecular signals control cell death and survival, there will be increased opportunities for preventing PCD in select neuronal populations in vivo and directly assessing whether its occurrence is, in fact, adaptive.
Table 18.1 Some Possible Functions of Developmental PCD in the Nervous Systema
aFor references, see Buss, Sun, and Oppenheim (2006).
Summary
The biochemical and molecular pathways that regulate cell death and survival arose early in the evolution of animal life. Once this cellular capacity arose, however, it is likely that it was co-opted to serve a variety of biological functions. In the nervous system, some major functions include establishing optimal levels of connectivity between neuronal populations, eliminating aberrant cells or connections, regulating the size of progenitor populations, and serving transient functional or other needs of immature animals.
Modes of Cell Death in Developing Neurons
The specific morphological appearance exhibited by degenerating neurons can provide insight into the cellular and molecular mechanisms by which the cells are destroyed. Historically, pathologists were the first to be interested in this issue, and they focused on distinguishing different kinds of cell and tissue degeneration following disease, injury, and trauma (Clarke & Clarke, 1996). Over 125 years ago, the term necrosis was coined to describe what today comprises the major form of accidental or pathological degeneration. The pathological necrotic death of neurons following injury usually involves the degeneration of groups of contiguous cells in a region that initiates an inflammatory response that can be discerned easily in tissue sections (Fig. 18.5). At about the same time that necrosis was first described, however, another form of cell degeneration, called spontaneous cell death, was observed in normal adult mammals. Spontaneous cell death (now called PCD) was thought to provide a means for counterbalancing mitosis in adult tissues in which the turnover of cells normally occurs. PCD typically involves the sporadic loss of individual cells in a population that often degenerate by a different mode from necrosis, one that does not involve inflammation. Early steps in the spontaneous PCD cascade leading up to when histological signs of frank degeneration first occur may take many hours or days. Once that point is reached, however, the degenerative process is rapid, with individual cells dying and being removed in minutes or a few hours. The loss of thousands of neurons over several days is the consequence of many rapid individual cell deaths that at any moment in time represent only a small minority (~1%) of all the cells in the population. For this reason, historically, the occurrence and magnitude of even massive PCD can (and often did) go unnoticed.
Figure 18.5 Spinal motoneurons in the chick embryo. (A) Ventral horn from a control embryo. Note that only two cells (asterisks) are undergoing apoptotic PCD; the others appear normal (arrows). (B) Ventral horn from an embryo following treatment with an excitotoxin. Most neurons are undergoing a necrotic type of cell death (asterisks) whereas some appear normal (arrows). Apoptotic (C) and necrotic motoneurons (D) in the chick embryo spinal cord as seen with an electron microscope.
Two different strategies can be used to quantify neuron death: One can determine the total number of neurons at different stages and quantify the decrease (Fig. 18.2), or one can determine an increase in the number of dying (pyknotic) cells during the period of PCD and determine the timing and (with less precision) the extent of cell death (Fig. 18.5). The counting of neurons originally was done by counting profiles in thin sections and application of correction factors to account for multiple profiles per particle; a particle is defined as an identifiable object such as a cell, nucleus, or nucleolus (Clarke & Oppenheim, 1995); such counting is now increasingly replaced by stereological methods that count particles in thicker sections, a method that is less affected by confounding differences in particle size but that still presents challenges due to other sources of potential errors if not executed properly (Schmitz & Hof, 2005).
Cell degeneration in adult and developing tissues has often been dichotomized into death by either apoptosis or necrosis, a distinction based initially on morphological differences (Fig. 18.6). Apoptosis is a Greek word indicating the seasonal piecemeal dropping of leaves from a tree and originally was coined to describe all forms of cell death that share certain morphological characteristics (Wyllie, Kerr, & Currie, 1980). Cells dying by apoptosis shrink in size, and the nuclear chromatin condenses and becomes pyknotic, whereas the cell membrane and cytoplasmic organelles tend to remain relatively intact. Eventually the cytoplasm and nucleus break up into membrane-bound apoptotic bodies that are phagocytized either by professional phagocytes (macrophages) or by healthy adjacent cells (e.g., glia that serve as transient phagocytes). Phagocytes recognize dying cells by their expression of death-related cell surface signals. In contrast, necrosis involves an initial swelling of the cell, only modest condensation of chromatin, cytoplasmic vacuolization, breakdown of organelles, and rupture of the cell membrane allowing the release of cellular contents (causing inflammation), followed by shrinkage and loss of nuclear chromatin. Because necrotic cell death elicits an inflammatory response in adjacent tissue, macrophages derived from the immune system attack and phagocytize cellular debris. In contrast, cell death by apoptosis usually involves individual cells that are engulfed and phagocytized before they can release their cellular contents or induce an inflammatory response in adjacent tissue.
Figure 18.6 Schematic representation of cellular changes during necrotic cell death and during three of the most common types of PCD observed at the ultrastructural level. Key features of each type of cell death are indicated by the red arrows. Only type 1 PCD meets most of the criteria for defining apoptosis. The cells on the right marked P represent phagocytic cells engulfing necrotic cell corpses and apoptotic bodies. Phagocytosis also occurs in the other types of PCD but is not shown.
Another feature that has been used to distinguish between apoptotic and necrotic cell death is the occurrence during apoptosis of a specific form of chromosomal DNA fragmentation and degradation that is mediated by DNA-specific proteases. DNA digestion occurs at internucleosomal sites, producing small, double-stranded fragments of DNA that migrate in a ladder pattern in multiples of 180–200 bp after electrophoresis in agarose gels. This form of DNA fragmentation can also be visualized in tissue sections by a technique that labels the double-stranded DNA breaks associated with apoptosis. DNA fragmentation suggests that apoptosis is an active process undertaken by the cell, whereas necrosis is a passive process resulting from extracellular damage. However, a variety of toxic and traumatic stimuli, previously thought to induce necrosis, can also induce morphological signs of apoptosis and may be associated with changes in PCD- or apoptosis-associated genes (Bredesen, Rao, & Mehlen, 2006).
Within the context of the developing nervous system, the dichotomy of apoptosis versus necrosis is an oversimplification. Developing cells undergoing normal PCD as well as neurons dying following injury should be categorized by operational definitions that use morphological, genetic, and biochemical criteria. Studies of the developing nervous system at the ultrastructural level provided an initial characterization of neuronal death noting that there were features of dying cells that appeared to indicate that all neurons do not die by identical mechanisms (Clarke & Clarke, 1996). “Nuclear” (Type I) neuronal death featured a dissociation of polyribosomes to free ribosomes, Golgi fragmentation to numerous vesicles, swollen mitochondria, and condensation of nuclear chromatin; the cell takes on a reduced size and crenulated appearance, neurotubules and neurofilaments are grouped together, and lysosomes appear normal. This type of neuronal degeneration is consistent with apoptosis.
There are other types of neuronal death during development, however, that do not fall under the categories of apoptosis or necrosis. In “cytoplasmic” (Type III) degeneration, the cisternae of Golgi, nuclear envelope, and rough endoplasmic reticulum are swollen, together with a vacuolization of mitochondria. Polyribosomes appear normal, and while the nuclear chromatin shows some condensation, the nucleus appears relatively normal. There is another type of cell death referred to as autophagy or Type II cell death that is characterized by the pres- ence of abundant lysosomes and autophagic vacuoles. The specific morphological events that a neuron undergoes during PCD may depend on the type of neuron, its state of differentiation, and the stimuli that induce cell death. For example, ciliary ganglion cells predominantly exhibit features of cytoplasmic cell death during PCD; however, when peripherally deprived of targets, the majority of ciliary cells exhibit features of nuclear or apoptotic cell death (Oppenheim, 1981). The PCD of developing vertebrate neurons provides a striking example of the problems encountered in attempting to rigidly classify the pathway of degeneration as being either necrotic or apoptotic. Importantly, evaluation of the morphological features of PCD can further the understanding of the biochemical pathways of PCD and the roles of the specific genes involved.
Summary
Degenerating cells are often categorized into two classes: death by apoptosis or death by necrosis. However, the occurrence of certain features of apoptosis in neurons following injury and the occurrence of several types of PCD other than apoptosis indicate that a more accurate means of identifying and defining distinct forms of cell death should now be employed.
The Mode of Neuronal Cell Death Reflects the Activation of Distinct Biochemical and Molecular Mechanisms
As described previously, the normal death of cells in the developing nervous system has long been thought to be regulated by competition for NTFs (the Neurotrophic Hypothesis). Investigators thought that the doomed neurons, lacking sufficient amounts of an NTF to sustain normal metabolic events, passively degenerated by a process analogous to starvation. Beginning in the late 1980s, several lines of evidence forced a reappraisal of the view that cell death in the nervous system is a passive process and led to the demonstration that for many types of neurons, PCD is regulated by the interaction of specific genetic programs that either inhibit or induce degeneration.
PCD of some nonneuronal cells was known previously to be a metabolically active, ATP-dependent process. In the nervous system, inhibition of RNA or protein synthesis could also prevent the death of sympathetic ganglion cells deprived of NGF in vitro and motoneuron and sensory neuron PCD in vivo. Additionally, in the early 1980s, genetic mutations in the nematode worm Caeno-rhabditis elegans that prevent PCD had been described, thereby providing further evidence that neuronal death is a genetically regulated, metabolically active process. Considerable progress has been made since the early 1990s in identifying cell death–associated genes and their pathways of action. Due to historical precedent, these have been classified as pro- and anti-apoptotic genes; however, it is important to point out that in many instances the genes involved may actually induce cell death with morphological features of both apoptotic and nonapoptotic degeneration (see the previous section).
Genetic studies of cell death in C. elegans first demonstrated that PCD is regulated by specific genes. Because the genetics, morphology, and cell lineages in C. elegans have been so well defined, this organism provides a particularly informative and powerful model for analyzing the molecular genetics of PCD (Horvitz, 2003). Of the approximately 1000 somatic cells generated (of which 302 are neurons and 56 are glial cells), 131 undergo embryonic PCD, and most of these are mitotically active neuron precursors. In each individual, the same cells die at specific times in development, and these corpses are then engulfed and degraded by neighboring cells. Despite the enormous evolutionary gap that separates the appearance of worms, flies, and vertebrates, significant homology exists in the structure and function of specific cell death pathways between these taxonomic groups.
As summarized in Figure 18.7, there is a sequential cascade of genetically regulated steps involved in PCD. Upstream of the actual execution of the death process, transcription factors specify certain cell types for death, while sparing others. In C. elegans, manifestation of the death fate requires expression of two pro-apoptotic genes, ced–3 and ced–4 (ced, cell death), which together mediate the actual breakdown of cellular constituents. Prevention of cell death induced by ced-3 and ced-4 can occur by activation of the anti-apoptotic gene ced-9. Separate genes, such as nuclease-1 (nuc-1), are required for the degradation of DNA in dying cells, and several additional genes are involved in the recognition, engulfment, and processing of dead cellular corpses by phagocytes. Evidence for the involvement of this genetic cascade in cell death and survival in C. elegans comes from several different approaches, most notably from genetic studies of loss-of-function mutants. For example, in the absence of ced-9, many cells that normally survive die, whereas in the absence of ced-3 or ced-4, all PCD is prevented.
Figure 18.7 (A) Schematic representation of the major steps in the developmental PCD pathway of neurons in the nematode worm C. elegans. Both fly and mammalian homologues have been identified supporting the hypothesis that mechanisms of PCD are evolutionarily conserved. The mammalian homologues are listed in italics below the name of the C. elegans gene. (B) Some of the major events involved in mammalian PCD are illustrated. Healthy cells (top) receive signals to survive or lack signals to die (on left). Cell death mediators are present in healthy cells, but their location and/or association with regulators prevent activation of the cell death pathways. In cells that receive appropriate signals to die (bottom), cell death–specific pathways are activated. These pathways involve a permeabilization of the mitochondria membrane and release of factors that either directly or indirectly activate cell death–specific events. In neurons that fail to obtain NTF support (bottom), the pro-apoptotic gene Bax interacts with and inhibits the anti-apoptotic gene Bcl–2 in mitochondria. This results in the release from mitochondria of cytochrome-c, which forms a complex with Apaf-1 and caspase-9 that in turn activates downstream caspases such as caspase-3 that ultimately directly or indirectly degrade diverse nuclear and cytoplasmic targets. These degradative changes are what define apoptosis and result in eventual engulfment and phagocytosis of the apoptotic cell. In some situations two additional molecules released from mitochondria (along with cytochrome-c) are the pro-apoptotic proteins AIF, which can degrade the nucleus independent of caspases, and Smac/Diablo, which can inhibit IAP and promote the apoptotic pathway via caspase-9 and caspase-3. Although not shown here, in some situations developing neurons undergoing PCD activate cell cycle proteins that also serve a signaling function required for apoptosis (Copani et al., 2001). It is critical to note that all neurons do not use the identical pathways for cell death. For example, cell cycle proteins can mediate the death of sensory neurons; however, they do not play a role in motoneuron PCD (Taylor et al., 2003. Mol Cell Neurosci. 24(2):323–39).
Identification of the DNA sequences of the major cell death genes ced-3, ced-4, and ced-9 in the worm in the 1980s and early 1990s resulted in the subsequent discovery of vertebrate and insect homologues that serve similar functions (Fig. 18.7). In mammals, ced-3 is represented by a large family of related proteases called caspases. Caspases are cysteine proteases that cleave after aspartate residues, and 12 distinct caspases have been cloned in human. In cell death, initiator pro-caspases are autolytically cleaved, and they then cleave effector pro-caspases. The effector caspases act on specific substrates that result in cell demise. Ced-4 is represented by a single vertebrate homologue, Apaf-1 (apoptosis protease activating factor), that is required for caspase activation. The survival-promoting function of ced-9 originally was represented by a single vertebrate homologue, bcl-2 (B-cell lymphoma-related gene), but subsequently other bcl-2 family members have been identified with similar anti-apoptotic functions (e.g., bcl-x), as well as other bcl-2 family members that are pro-apoptotic (e.g., bax, bak, bim). Loss-of-function mutations in mice by targeted gene deletion (gene knockout) of many of these vertebrate homologues have confirmed their role as important regulators of PCD.
To understand cell death pathways, it is useful to consider the process of cell death in terms of signals that (i) induce death, (ii) regulate the intracellular process of cell death, and (iii) are involved in the removal of the corpse. NTFs act via membrane receptors resulting in intracellular signaling that either inhibits the expression or activity of pro-apoptotic genes or induces the expression or activity of anti-apoptotic gene products. On the other hand, as discussed below, more recent evidence indicates that pro-neurotrophins are capable of inducing cell death, suggesting that the availability of pro-survival versus pro-death signals can dictate the neuron’s fate. Additionally, Fas-Fas ligand interaction has also been shown to regulate neuronal PCD.
Availability of NTFs induces or maintains activation of the PLCg, PKA, PI3-K/AKT, PKC, and ERK1/2 signal transduction pathways (see below). These pathways utilize both transcriptional and nontranscriptional mechanisms to regulate neuronal survival. For example, PI-3K/Akt has been reported to phosphorylate the pro-apoptotic molecule, Bad, promoting its association with 14-3-3 and preventing inactivation of pro-survival molecules, Bcl-2 and Bcl-x. Activation of these pathways also influences gene expression by activating transcription factors—namely cAMP responsive element binding protein (CREB)—which can result in transcription of anti-apoptotic genes like Bcl-2 and inhibitor of apoptosis proteins (IAPS). On the other hand, the JNK signal transduction pathway is activated in some neurons following NTF deprivation or Fas-Fas ligand interaction. JNK activation results in the activation of BH-3 proteins, regulation of Smac release from the mitochondria, and caspase activation.
Following inactivation of trophic receptors, and/or activation of p75NTR or Fas-Fas ligand, a series of intracellular events occurs that leads to a rapid and efficient dismantling of the cell. A central integrator of cell death in vertebrates is the mitochondrion (Fig. 18.7). Both pro-survival and pro-apoptotic Bcl-2 proteins are critical for regulating mitochondrial membrane permeability. Permeabilization of the mitochondrial membrane results in increased reactive oxygen species, reduced ATP production, and release of the pro-apoptotic factors cytochrome c, Smac, and apoptosis inducing factor (AIF). Cytochrome c in the cytoplasm binds to Apaf-1, revealing Apaf-1’s caspase recruitment domain (CARD), and in the presence of ATP forms a heptamer. Pro-caspase 9 has a high affinity for the CARD and binds at the heptamer. This complex is referred to as the apoptosome. The increased local concentration of pro-caspase 9 results in its auto-cleavage, producing active caspase 9. Caspase 9 then cleaves pro-caspase 3 to the active form of caspase 3 that subsequently activates, inactivates, or destroys specific nuclear and cytoplasmic substrates, resulting in rapid cell degeneration. The activation of caspases as well as the availability of substrates may dictate the morphological changes associated with neuronal death (see previous section).
Many developing neurons appear to share the core cell death program described above, but there is also increasing evidence that alternative, nonapoptotic pathways may exist for mediating the normal PCD of some neurons. For example, when components of the apoptotic machinery are genetically deleted, cell death occurs by a morphologically distinct nonapoptotic death (Kroemer et al., 2009). Type II cell death or autophagy is characterized by the accumulation of double membrane–bound vesicles that fuse with lysosomes. Interestingly, Bcl-2 proteins and caspase activation appear to have regulatory roles in autophagy, suggesting that the different types of cell death utilize common mechanisms. Autophagy is also a critical nondeath event in maintaining homeostasis in many tissues, including the nervous system, and its role in normal cell death is still being debated. Another example of non-apoptotic cell death is necroptosis, a type of programmed necrosis recently characterized (Kroemer et al., 2009). Many of the stimuli that induce apoptosis instead activate necroptosis when caspases are not available. Morphological features of this type of death include organelle swelling, rapid mitochondrial dysfunction, plasma membrane permeabilization, and lack of nuclear fragmentation.
While Bcl-2 proteins and caspases are critical mediators of cell death, these factors also have important non–cell death–related roles. For example, Bcl-2 proteins are also involved in regulating mitochondrial fission and fusion independent of cell death. Caspases also have non–cell death roles, including proliferation, differentiation, and synaptic function (Fuchs & Steller, 2011). For example, caspase 3 is required for long-term depression and AMPA receptor internalization in hippocampal neurons.
The final stage of cell death is the removal of the corpse. This event is as critical as the other steps involved in the cell death process because ineffective removal of dying cells can result in apoptotic cells becoming necrotic and initiating inflammation or an autoimmune response. While nonprofessional phagocytes (e.g., ependymal or Schwann cells) can accomplish this task, professional phagocytes such as macrophages and microglia are most efficient at recognizing and removing dying cells (Mallat, Marín-Teva, & Chéret, 2005).
During early CNS development, microglia are usually not present in large numbers in the area of PCD. Some signal must be sent by the dying cells to recruit the phagocytes into the area. The phospholipid lysophosphatidylcholine, ATP, and UTP are macrophage chemotactic factors shown to be released by dying cells (Lauber et al., 2003; Elliott et al., 2009). Once the phagocytes are recruited to the area, they must be able to distinguish between healthy and dying cells. A series of evolutionarily conserved genes are involved in the expression of recognition markers on dying cells, expression of the receptors on phagocytic cells, as well as for those factors necessary for the processing of the engulfed apoptotic cells (Kinchen & Ravichandran, 2007).
Regulation of cell death is a critical issue in neurons because with only rare exceptions, postmitotic neurons are irreplaceable. One of these regulatory mechanisms is referred to as “competence to die.” As neurons differentiate, they develop a dependence on NTF support. At this stage, in order to undergo PCD, developing neurons require not only the release of cytochrome c from mitochondria and caspase activation, but in addition, they require the loss of NTF support. In the presence of trophic support, the x-lined inhibitor of apoptosis protein (XIAP) serves as a brake on caspase activation when cytochrome c is accidentally released; loss of NTF support removes this brake, thereby creating competence-to-die (Potts, Singh, Knezek, Thompson, & Deshinukh, 2003). Adult neurons also need to be protected from accidental cell death. As postmitotic neurons further mature past the period of naturally occurring cell death, they lose or reduce their dependence on NTFs such that they can now survive following NTF deprivation. This involves a block in the translocation of the pro-apoptotic Bax protein from the cytoplasm to the mitochondria, a failure of cytochrome c release and inhibition of apoptosis (Putcha, Deshmukh, & Johnson, 2000).
Summary
PCD is a metabolically active process that involves specific genetic pathways necessary for the cascade of events leading to degeneration. Several genes in the PCD pathway were first identified in C. elegans and homologues have been found in insects and vertebrates. The mitochondrion is a central integrator of cell death signaling. Although most developing neurons appear to share a common core PCD pathway, morphological and biochemical evidence also indicates the presence of alternatives to the core PCD machinery. One consequence of the increased evolutionary complexity in the regulation of neuronal PCD is the availability of additional checks and balances before a cell passes the irreversible point of commitment to die. Because virtually all neurons are postmitotic and neuron numbers finite, multiple, relatively fail-safe, mechanisms are required so that accidental death occurs only rarely.
Nerve Growth Factor: the Prototype Target-Derived Neuronal Survival Factor
Following the discovery of NGF by Levi-Montalcini, Cohen, and Hamburger in the 1950s and 1960s, and its possible role in the survival of developing sympathetic neurons, the characterization of the functional role of NGF became the archetypical model for the investigation of other putative NTFs. Analysis of the function(s) of a putative NTF involved several steps. Typically, the biological activities and target cell specificity of putative factors were first investigated in vitro. Primary cultures of neurons dissociated from peripheral ganglia were ideal for these early assays. Analysis of the developmental expression of specific NTFs and their corresponding receptors has been used to determine if both the ligand and the receptor are normally present at the appropriate time and place in vivo for the putative factor to serve in the regulation of a specific subpopulation of neurons.
In gain-of-function approaches, treatment of embryos with excess exogenous NTF has been used to determine if the survival or differentiation of responsive neurons is restricted by either the production or the access to a limited quantity of endogenous factor. Finally, in loss-of-function approaches, methods that inhibit or prevent the function of specific NTFs or their receptors have allowed investigators to determine whether the perturbation of endogenous NTF signaling alters normal development. These NTF/NTF receptor deprivation experiments have included treatment with activity-blocking antibodies that prevent trophic signaling, treatment with soluble receptor-derived antagonists that compete for and adsorb endogenous ligands, and the generation of transgenic mice with null mutations of either the factors or their receptors.
Sympathetic and sensory ganglia removed from developing animals have been shown to produce a dense halo of neurite outgrowth when treated with NGF. In fact, this “axonal halo” assay—not a neuronal survival assay—was the original biological activity first used to purify and characterize NGF as an NTF. NGF was shown to be required for the survival of dissociated sympathetic and some sensory neurons when they were grown in the absence of nonneuronal cells. This demonstrated that NGF could prevent cell death by the direct activation of receptors on isolated neurons. When developing embryos were treated with excess exogenous NGF, sympathetic and sensory ganglia were enlarged significantly, and neurite growth from these neurons increased markedly. In addition, ganglia in NGF-treated embryos contain many more neurons than normal because PCD had been prevented. The soma was enlarged significantly by NGF treatment, and the dendritic arbors of sympathetic neurons were more complex. These studies indicated that the supply or access to endogenous NGF in sympathetic and sensory targets was likely to be rate limiting for the survival and growth of these dependent populations.
The most convincing evidence that NGF is required for neuron survival has been gained from NGF deprivation experiments. Embryos treated with antibodies that selectively block NGF activity as well as the null mutation of either NGF or its trkA receptor in transgenic mice have both confirmed that sympathetic as well as some sensory neurons require NGF for survival (Fig. 18.8). NGF was localized to the peripheral targets of these neurons at the time of their normal innervation, consistent with its role as a target-derived survival factor. Further, the level of NGF synthesis was correlated with the density of target innervation, and the NGF receptor, trkA, was localized to dependent afferent neurons at the times and places appropriate for regulating normal PCD. The localization of NGF synthesis in sympathetic targets and the loss of these afferent neurons with NGF deprivation firmly established NGF as a prototype target-derived NTF required for the survival of sympathetic neurons and a subset of sensory neurons.
Figure 18.8 Phenotypic alterations in sensory/motor pathways caused by null mutations in NGF/trkA and NT-3/trkC. In the dorsal root ganglia (DRG) of normal mice (A), small-diameter (red), medium-diameter (green), and large-diameter (blue) neurons are present. Large-diameter neurons innervate muscle spindles and other proprioceptive end organs and have axon terminations in the lowest laminae of the dorsal horn and in the ventral horn. These neurons are lost when NT-3 or trkC is absent (compare A and B). Many of the small-diameter neurons innervate skin, respond to temperature and pain, and have terminations in the dorsal-most laminae of the dorsal horn. These neurons are lost when NGF or trkA is absent (compare A and C). DRG neurons indicated in green are neurotrophin-independent mechanoreceptive neurons (peripheral projections not shown).
Modified and redrawn with permission from Snider (1994).
One prediction of the neurotrophic hypothesis is that access to NGF in only the distant target region is adequate to support the survival of the remote cell body. This idea has been tested by Robert Campenot and colleagues in vitro. NGF has been applied selectively only to local axon terminals in a three-compartment tissue culture chamber to determine the long- and short-range effects of NGF treatment. NGF-dependent neuronal cell bodies in the central chamber survived when only their terminals were treated with the factor, indicating that target-derived NGF available only to neurites can generate and retrogradely transport the signal required for cell body survival. Peripheral processes were lost rapidly and selectively in outer chambers where NGF was withdrawn but were maintained and grew in the outer chambers where NGF was added. This important demonstration illustrates the capacity of target-derived NGF to have both direct long distance effects on the survival of neurons and direct local effects on the growth, maintenance, and sprouting of axonal branches. In addition, recent studies show that retrogradely transported NGF controls the assembly of afferent synapses on cell bodies and dendrites. NGF-responsive neurons possess both a “ligand-specific” receptor, trkA, and a “common” receptor, p75NTR (Fig. 18.9) that binds all neurotrophins with a similar affinity. Many of the biological activities of NGF have been attributed to the ligand-induced transduction of trkA.
Figure 18.9 Models of NGF, the catalytic (full-length) TrkA receptor, and p75NTR and its binding partners (sortilin, Nogo receptor, and Lingo 1). Note the ability of one ligand (NGF dimer) to bring together two trk receptor molecules to initiate signaling. p75NTR lacks a cytosolic kinase domain, but associates with multiple partners (possibly not including direct interactions with trk receptors) to assemble signaling platforms. Note that the NGF dimer binds to a p75NTR monomer in an opposite orientation compared to TrkA (yellow arrows). C and N, C and N termini of NGF; Ig-C1, Ig-C2, C-terminal immunoglobulin ligand binding domains 1, 2; LRR, leucine-rich repeat; CRD 1–4, cysteine-rich domains 1–4.
Modified from Barker (2004) and Wehrman et al., 2007 (Neuron 53, 25-38) with permission.
Summary
The modern study of neuronal cell death began with investigations of how synaptic targets of sensory and motor neurons regulate their development. Viktor Hamburger and Rita Levi-Montalcini, beginning in the 1930s, ultimately showed that targets promote the survival and maintenance of innervating neurons. This notion, in turn, provided a conceptual framework for the discovery of the first target-derived NTF, nerve growth factor. From these beginnings, the neurotrophic hypothesis was formulated: neurons compete for limiting amounts of target-derived survival promoting (trophic) agents during development. NGF was established as the prototypical target-derived NTF.
The Neurotrophin Family
Only a few subpopulations of peripheral neurons, including sympathetic and some sensory neurons, are exclusively dependent on NGF for survival during development. The survival of CNS neurons is largely unchanged following the null mutation of NGF. Other survival factors are now known to regulate neuron survival elsewhere in the nervous system. Extracts made from a number of tissues, as well as media containing proteins secreted by a variety of cultured neuronal and nonneuronal cells, have been shown to support the survival of many different classes of neurons that are not NGF dependent and that do not express the NGF receptor trkA, suggesting that additional NTFs exist. Because NTFs are made in extremely low quantities, the biochemical isolation of NGF-related molecules using conventional protein purification methods proved to be difficult.
A significant breakthrough occurred with the purification of a second NGF-related NTF by Yves Barde and colleagues in 1982. Unlike NGF, which was purified several hundredfold from an extraordinarily rich biological source (salivary gland) unrelated to the nervous system, endogenous brain-derived neurotrophic factor (BDNF) was purified several millionfold from adult pig brains. Each kilogram of starting material yielded only a microgram of factor that over time was eventually sequenced and cloned to produce recombinant factor. The molecular cloning and expression of BDNF opened the door for an accelerated period of research on NGF-related factors. When the protein structure of BDNF was compared with NGF, they were both found to encode homodimers of small, very basic secreted peptide ligands with an amino acid homology of approximately 50%. Using polymerase chain reaction primers prepared from homologous domains to search for other related proteins, investigators rapidly identified additional neurotrophin family members in multiple species (Lewin & Barde, 1996). Described as neurotrophins or nerve feeding factors (i.e., NT-3 and NT-4/5, NT-6), these additional proteins were cloned and sequenced without the requirement of exhaustive protein purification.
Each neurotrophin family member is synthesized as an approximately 250 amino acid precursor (pro-neurotrophin) that is processed into a roughly 120 amino acid protomer. The precursor was initially thought to be functionally irrelevant, but it turned out to have important death-signaling capabilities, as described below. Homologous regions of the several different family members are concentrated in six hydrophobic domains containing cysteine residues. The linkage formed by each homodimer ligand utilizes these regions to form a “cysteine knot” that maintains the twin protomers in juxtaposition. The secreted dimer appears as a symmetrical twin with variable regions containing basic amino acid residues exposed on the surface (Fig. 18.9). Because all family members share this core structure, they are remarkably similar, with three-dimensional symmetry around two axes. The symmetry of this twin structure allows the neurotrophin ligand to activate trk receptors by binding separate receptor molecules together in the membrane, permitting docking of additional intracellular signaling molecules, and initiation of signal transduction cascades (Fig. 18.9). The exposed outer regions that vary among neurotrophin family members are responsible for receptor-binding specificity.
Summary
The purification, molecular cloning, and expression of BDNF opened a floodgate of research on an NGF-related family of NTFs called neurotrophins or nerve feeding factors. Each neurotrophin family member is synthesized as a pro-neurotrophin that is processed into a homodimer with a conserved region containing a cysteine knot in the core of the molecule. The secreted factor is a symmetrical twin with duplicate sites used for bivalent receptor binding and formation of signaling platforms.
Neurotrophin Receptors
NGF binds to a relatively small number of very high-affinity binding sites and a second set of about 10-fold more abundant, but lower affinity, binding sites at higher concentrations (Roux & Barker, 2002). A 75-kDa protein (p75) was purified and cloned first. This transmembrane glycoprotein with extracellular cysteine repeat motifs shares structural homology with the tumor necrosis factor receptor family. The cytoplasmic domain of p75 lacks the kinase domain present in most growth factor receptors for intracellular signal transduction, but it can associate with several other signaling proteins (discussed later) (Fig. 18.9). When expressed in fibroblasts, this receptor has low-affinity NGF-binding properties (ligand binding is rapidly on and off) and therefore also has been called the low-affinity NGF receptor. This name has proven to be a misnomer, since other neurotrophin family members also bind p75 with a similar affinity. It is therefore more appropriately named the “common” neurotrophin receptor (p75NTR).
A major breakthrough in the characterization of the NGF receptors came with the fortuitous discovery and cloning of an oncogene identified in a human colon cancer. The sequence of this 140-kDa transmembrane protein contained a cytoplasmic kinase common to many growth factor receptors. The corresponding proto-oncogene was named trk (pronounced “track”—for tropomyosin-related kinase). It rapidly was appreciated as a member of the tyrosine kinase-containing receptor superfamily with trk mRNA expression localized to NGF-responsive neurons.
Low-stringency screening of cDNA libraries with the original trk proto-oncogene probes led to the discovery of other related neurotrophin receptors. The NGF binding receptor was called trkA, whereas two additional 145-kDa members of this protein family were named trkB and trkC (Fig. 18.10). Expression of trkA conferred specific high-affinity NGF binding and NGF-induced receptor phosphorylation. NGF-signaling properties have been examined extensively in the NGF-responsive pheochromocytoma (PC12) cell line, derived from adrenal medullary cells. Mutant PC12 cell lines that have lost their capacity to respond to NGF contain many p75NTR receptors but lack trkA. Transfection of these mutant cells with trkA restores their biological responses to NGF treatment. The most convincing evidence for the necessity of trkA comes from the analysis of transgenic mice lacking functional trkA receptors. As expected, these mice have a phenotype that is similar to that of transgenic animals that have a null mutation for NGF (Table 18.2).
Figure 18.10 Ligand binding preferences of neurotrophins for each member of the trk receptor family. Not shown are the truncated (kinase deleted) isoforms of trkB and trkC. Other isoforms containing inserts and deletions also exist, providing a wide variety of receptors. NT-4 is also named NT-4/5. K = tyrosine kinase.
Table 18.2 Percentage of Neurons Lost in Neurotrophic Factor or Receptor Deficient Micea
aFor references to original studies, see Huang and Reichardt (2001), Airaksinen and Saarma (2002), von Bartheld and Fritzsch (2006).
bLosses reported by Klein et al., 1993 (Cell 75, 113–122) could not be replicated.
cLosses of gamma motoneurons only; Kucera et al., 1995 (Neuroscience 69, 321–330).
The trkB receptor is activated specifically by low concentrations of BDNF or NT-4/5 and, to a lesser extent, by higher concentrations of NT-3. NT-3 activates the trkC receptor most effectively. All trk receptors contain three leucine-rich motifs, two cysteine clusters, and two immunoglobulin-like motifs in the extracellular region, a transmembrane domain, and a tyrosine kinase domain in the cytosolic region (Fig. 18.9). The unusual combination of extracellular motifs makes up the ligand-binding region and places this family in a novel class of tyrosine kinase receptors. The region of highest sequence homology among family members and other growth factor receptors is in the kinase domain.
Receptor isoforms resulting from splice variants of trk mRNA transcripts exist for each family member. A variety of both full-length and kinase-deleted or truncated receptors are widely expressed on neurons throughout the nervous system. Truncated receptors, which also are expressed on glial cells, can bind and internalize their cognate ligand and are capable of limited signaling, but they cannot initiate the phosphorylation events required for mainstream receptor signal transduction (see below). As a result, the distribution and membrane concentration of truncated receptors could potentially modulate neurotrophin activity.
Although trk receptors account for many of the biological responses of neurons to neurotrophins, p75NTR can modify trk ligand binding and neurotrophin specificity, it associates with additional signaling partners, and can initiate pathways for intracellular signaling independent of trk receptors (Fig. 18.9). Several mechanisms have been proposed to account for an accessory role of p75NTR. The fast on and fast off kinetics of p75NTR could maintain and increase the local concentration of neurotrophins at the membrane surface and thereby increase the access of trk receptors to ligands. The p75NTR does not appear to form a transient heterodimer with trk receptors, and NGF’s binding sites overlap in a way that prevents them from binding simultaneously to p75NTR and trkA in a 2:2:2 complex (Fig. 18.9). Despite lack of evidence for a direct p75NTR/trk interaction, p75NTR may still “hand off” the factor for trk binding (Roux & Barker, 2002).
In addition to enhancing NGF binding to and activation of trkA, p75NTR initiates NGF responses in cells that lack trkA. NGF binding to p75NTR can cause activation and nuclear translocation of a transcription factor, nuclear factor kappa B (NFkB), that promotes cell survival. However, p75NTR is related structurally to members of the tumor necrosis factor receptor (TNFR) family, many of which regulate the onset of cell death programs in the immune system and, for that reason, are termed death receptors. The cytoplasmic domain of p75NTR contains a “death domain” sequence similar to active sequences found in the TNFR family, and the p75NTR mechanisms of inducing PCD appear to be shared with other death receptors.
In neurons that express p75NTR but not trkA, NGF or proNGF induce cell death via p75NTR binding (Roux & Barker, 2002). Once bound, p75NTR may activate a PCD signaling pathway via the Jun kinase cascade. The Jun kinase cascade activates the transcription factor p53 with gene targets that include the pro-apoptotic gene BAX. It should be noted that NGF and other trophic factors can be utilized in different cells or in different epochs of time to induce many diverse biological activities. It is noteworthy that under specific circumstances (such as the absence of trk activity), a factor originally identified for the capacity to prevent PCD instead is utilized to execute programs promoting neuronal death.
Neurotrophin receptors were generally considered to be inactive (“neutral” in terms of neuronal survival) unless bound by their ligand. However, as noted above, recent work has shown that several receptors, including trkA and trkC, actively transduce a negative (apoptotic) signal in the absence of the ligand, and thus behave as “dependence receptors” (Mehlen, 2010; Nikoletopoulou et al., 2010).
Summary
Many neurons have both specific and common binding sites for neurotrophins. Biological responses primarily are associated with high-affinity binding and rapid phosphorylation signaling events. All neurotrophins bind p75NTR, the common neurotrophin receptor. p75NTR lacks a cytoplasmic kinase domain but can independently initiate signaling, can associate with additional signaling partners, and can enhance signaling through trkA. There are three tyrosine receptor kinase (trk) family members: trkA, trkB, and trkC. Each receptor binds one or more members of the neurotrophin family. Splice variants of trks result in isoforms that include truncated receptors with reduced signaling capabilities. p75NTR is related to the TNFR family of death receptors and contains a cytosolic “death domain.” Activation of p75NTR, especially by pro-neurotrophins, may serve to kill neurons or other cell types through well-established signaling pathways used to promote PCD.
Secretion and Axonal Transport of Neurotrophins and Pro-Neurotrophins
Processing and Secretion of Neurotrophins
Neurotrophins are expressed as larger precursors that are cleaved by furins and prohormone convertases. The extent and location (intra- or extracellular) of cleavage differs between cell types. The precursor forms were believed for a long time to be of little functional significance, and it was thought that the mature neurotrophins were the only important moiety that was secreted. Research by Barbara Hempstead and her colleagues in the early 2000s demonstrated that many cells secrete pro-neurotrophins and that the release of both pro-neurotrophins and mature neurotrophins contributes to the spectrum of physiological functions. Pro-neurotrophins bind with higher affinity to p75NTR than mature neurotrophins and appear to be the “preferred” ligand for p75NTR. Whether endogenous pro-neurotrophins bind with lower affinity to trk receptors is still controversial. To activate different signaling pathways, the neurotrophin-bound p75NTR can associate with several signaling partners, including Nogo receptor (a GPI-linked protein), sortilin (a member of the VPS10 family), and LINGO-1 (involved in myelin-based growth inhibition; Fig. 18.9). This complex is involved in regulating neurite outgrowth in response to myelin-derived proteins such as Nogo, MAG, and OMgp. Formation of these different platforms may explain the multiple effects of p75NTR in different cell types and contexts. Sortilin also facilitates anterograde axonal transport of trk receptors to the axon terminus, which is required for subsequent retrograde neurotrophin signaling.
The proteolytic cleavage of pro-neurotrophins, and thus the local balance and concentrations of pro- and mature neurotrophins, provides a mechanism to regulate the direction of action of neurotrophins in a wide range of cellular processes (Lu, Pang, & Woo, 2005). Recent work has shown that pro-neurotrophins can be released from neurons, apparently leading to elimination of either competing p75NTR-expressing neurons or neurites, in what may be considered a molecular correlate of neuronal competition (Deppmann et al., 2008; Singh et al., 2008).
Neurotrophic Factors and Synaptic Plasticity
The classic neurotrophic hypothesis postulates that NTFs, including neurotrophins, are released by postsynaptic target cells and bind to presynaptic receptors for internalization, retrograde axonal transport, and activation of signal transduction cascades after arrival at the neuronal soma (Fig. 18.1). Yet, NTF receptors are present on both post- and presynaptic sites of the synaptic cleft. In vitro studies have shown that NTFs, and in particular BDNF, have rapid effects on synaptic transmission (McAllister, Katz, & Lo, 1999; Poo, 2001). Numerous studies have confirmed that NTFs have important functions in synaptic plasticity, meaning that these “synaptotrophins” regulate the functional strength of synaptic transmission including long-term potentiation (LTP), an experience-dependent, long-lasting increase in chemical strength of neurotransmission that is a crucial mechanism in learning and memory (Table 18.3).
Table 18.3 Nonsurvival Functions of Neurotrophic Factorsa
| Function | Examples of Relevant Trophic Factors and Cytokines |
| Cell fate decisions (stem cell/neuronal precursor derivatives) | BDNF, NGF, EGF, NT-3, CNTF, LIF, oncostatin M, CT-1 |
| Axon guidance | NGF, BDNF, HGF, GDNF, neurturin |
| Neurite growth, branching | BDNF, NT-3, NT-4, CNTF, CT-1, HGF, GDNF, LIF, IGF-I |
| Myelination | IGF-1, BMPs, NT-3, BDNF, GDNF, CNTF |
| Dendrite development | BDNF, NT-3, NGF, GDNF |
| Synapse development | BDNF, NT-3, GDNF, neurturin, NT-4, CNTF |
| Synapse stabilization and plasticity | BDNF, NT-3, GDNF, FGFs, HGF, VEGF, CNTF, LIF, NGF, IGF-I |
| Long-term potentiation, learning, and memory | BDNF, NT-3, NGF, NT-4, IGF-I, interleukins |
| Pain physiology and pathophysiology | NGF, BDNF, GDNF, artemin, NT-3 |
| Neurotransmitter phenotype | NGF, LIF, FGFs, GDNF, CNTF, BDNF, NT-3, IGF-I |
| Ion channel regulation/function | BDNF, interleukins |
| Neurological disease and neurodegeneration | |
| Seizure, epilepsy | BDNF, NGF, NT-3, VEGF, GDNF, LIF, CT-1, oncostatin M, CNTF |
| Stress, ischemia | NGF, NT-3, BDNF, IGF-I, VEGF, GDNF, HGF, BMPs |
| Traumatic injury, axonal regeneration, inflammation, stress | GDNF, IGF-I, LIF, CNTF, oncostatin M, CT-1, NGF |
| Psychiatric disease: depression | BDNF, NGF, FGFs, IGF-I, GDNF |
aFor references, see Snider (1994), McAllister et al. (1999), Huang and Reichardt (2001), Poo (2001), Chao (2003), Zweifel et al. (2005).
The regulation of secretion of neurotrophins is particularly important for mechanisms of synaptic plasticity (Lessmann, Gottmann, & Malcangio, 2003). Neurotrophins can be released by constitutive modes (from cell bodies and processes) and by activity-dependent modes of secretion (from dendrites and axons). The pattern of activity, such as high-frequency stimulation, is crucial for the release. LTP is relatively synapse-specific; only active and successfully transmitting synapses are strengthened, so the regulated release of neurotrophins and subsequent signaling should be restricted to active synapses. Depolarization and cAMP-dependent pathways are thought to regulate the locally restricted release of neurotrophins such as BDNF. This may contribute to the local “tagging” or marking of the active synapses destined for strengthening. Trafficking of BDNF mRNA within the neuron might also contribute to synapse-specific plasticity, if the mRNA is “captured” for local translation at active synapses. Interestingly, a naturally occurring polymorphism of the BDNF gene (a Val-Met substitution in the 5’pro-region of BDNF) gives rise to a reduced capacity for activity-dependent (but not constitutive) secretion of BDNF, causing abnormal hippocampal function and some learning and memory deficits in humans (Egan et al., 2003).
Endogenous neurotrophins may be derived from either pre- or postsynaptic sites, and they may act on either side of the synaptic cleft in paracrine or autocrine modes to implement functional and structural changes in synaptic plasticity. Trophic responses can also be modified by the recruitment of receptors to surface membranes from internal storage sites. Furthermore, after release and binding of newly synthesized neurotrophins, they can be recycled, and the internalized neurotrophin is not necessarily degraded immediately after the initial binding and internalization steps. Recycled neurotrophins have been shown to play functionally significant roles in LTP.
The Endosome Signaling Hypothesis of Retrograde Axonal Signaling
The neurotrophic hypothesis postulates retrograde axonal movement of trophic signals. Although evidence for retrograde transport of exogenous neurotrophins (NGF) was established in the 1970s, the precise mechanisms of axonal transport of trophic signals from the nerve terminal to the neuronal soma remained largely elusive. The prevailing hypothesis is that neurotrophins bind presynaptic surface membrane receptors on nerve terminals, are internalized, and form a ligand-receptor complex with the neurotrophin inside transport vesicles, leaving the trk receptor with its kinase domain on the outside of the vesicle or endosome (Fig. 18.11). The trk receptors may remain “active” (phosphorylated) during axonal transport along microtubules, because the ligand remains bound to the receptor within the vesicle. When the endosome, with the ligand-receptor complex attached, arrives at the neuronal soma, the phosphorylated trk receptor initiates a signal transduction cascade that results in changes of gene expression. Signaling endosomes can be isolated from axons; trk receptors and other signaling proteins interact at the level of the axon; trks bind to the dynein light chain subunit (part of the dynein motor complex) directly, and thus partake in the axonal transport machinery (Fig. 18.11). Furthermore, axon-derived endosomes that reach the soma may generate a different signal (activate a different ERK) than those generated by binding to cell surface receptors on the soma.
Figure 18.11 The signaling endosome. The internalized neurotrophin (NT)–receptor (Trk) complex is transported retrogradely from the nerve terminal to the cell body. The internalized vesicle serves as a platform for docking of adaptor proteins (light blue) that elicit signal transduction cascades via kinases (red); molecular motors and regulators (green) attach for transport along microtubules (brown). The arrow indicates the direction of transport toward the minus end of microtubules. The main signal transduction cascades are via the phospholipase C-γ (PLCγ), Raf-MAPK-ERK, and PI3K signaling pathways (red lightning symbol). For details of signal transduction, see legend to Figure 18.12. Rap1 and Rab5 are small G-proteins. Other abbreviations: Akt, v-akt murine thymoma viral oncogene homologue (protein kinase B); ARMS, ankyrin-rch membrane spanning protein; B-Raf, v-raf murine sarcoma viral oncogene homologue B1; EEA1, early endosome antigen 1; ERK, extracellular signal-regulated kinase; GAB1, GRB2 (growth factor receptor bound protein 2)-associated binding protein1; HAP, huntingtin-associated protein 1; htt, huntingtin; MEK, MAPK (mitogen-activated protein kinase); NT, neurotrophin; PI3K, phosphatidylinositol 3-kinase; Shc, Src homology 2 domain-containing transforming protein C; Trk, trk tyrosine kinase receptor. For details and references, see Zweifel, Kuruvilla, and Ginty (2005) and Gauthier et al. (2004).
However, not all trophic signals from the nerve terminal that arrive at the soma may require the neurotrophin to be physically transported in a ligand-receptor complex. In addition to retrograde axonal transport from nerve terminals to the soma, trophic signals also are transported anterogradely along axons (possibly even a quantitatively more important route) from soma to nerve terminals, mediated by kinesin motors, followed by activity-dependent release of neurotrophins from axon terminals.
Summary
Expression and secretion of pro-neurotrophins has significant physiological functions in the regulation of cell death. Pro-neurotrophins bind preferentially to the p75NTR. The p75NTR associates with several different signaling partners to form multiple signaling platforms that include sortilin and Nogo receptors. Secretion of neurotrophins requires activity-dependent stimulation patterns that may specifically strengthen active synapses. Neurotrophins are transported retrogradely along axons in signaling endosomes, but they can also be transported anterogradely for release from axon terminals. Trk receptors are expressed both pre- and postsynaptically, forming paracrine as well as autocrine loops of synaptic signaling.
Signal Transduction Through TRK Receptors
Neurotrophin binding to trk receptors at the cell surface causes the formation of receptor dimers and coactivation of their tyrosine kinase activity. The homodimeric structure of the factors allows each bivalent ligand to bring two separate receptor molecules into close proximity. Aggregated receptors phosphorylate each other on specific tyrosine substrates within intracellular domains. The generation of phosphotyrosine residues in turn activates the receptor kinase and further catalyzes the formation of large signaling complexes through the recruitment of adaptor proteins. These proteins link the activated receptor kinase with intracellular signaling pathways shared by other growth factors (Ip & Yancopoulos, 1996). Once activated, the receptor initiates intracellular signals both locally in the cytoplasm and by a series of enzymatic cascades that eventually produce changes in gene transcription within the nucleus (Segal & Greenberg, 1996).
Receptor signal transduction involves multiple signaling pathways that can differ between individual neurons or even in the same neurons depending on recent events. Thus, the response of neurons to trophic factor stimuli is dependent on the current intracellular status of the cell in a dynamic fashion. The molecular components of these pathways are so well conserved that many of the signaling proteins are interchangeable among invertebrate and vertebrate species.
Three major pathways have been identified to implement trk signal transduction events that mediate survival as well as many other cellular responses to neurotrophins (Fig. 18.12). They are (1) the phospholipase C (PLC-g) pathway, (2) Ras-MAP kinase pathway, and (3) phosphatidylinositol-3 kinase (PI3K) pathway. The latter two pathways begin with adaptor proteins that contain a structural motif, the src homology domain 2 (SH2), which specifically recognizes the phosphotyrosine residue and flanking sequences. PLC-g activity generates two distinct second messenger signals: inositol triphosphate (IP3) and diacylglycerol (DAG). IP3 rapidly releases intracellular Ca2 sequestered in local membrane compartments. This signal initiates the activity of local Ca2 dependent enzymes (protein kinases and phosphatases). Similarly, DAG regulates the activity of DAG-dependent enzymes. Both Ras and PI-3 kinase pathways are engaged via adaptor and associated linker or extender proteins.
Figure 18.12 Trk and p75NTR signaling pathways. Neurotrophin (NT) dimers bind two trk receptor monomers or one p75NTR to activate survival (1–3) or apoptotic (4) signaling pathways. Ligand binding initiates trk receptor transduction by the phosphorylation (indicated by small red dot) in the cytoplasmic domains. The activated Trk kinase docks adaptor and linker proteins (light blue), which engage signaling cascades often containing multiple kinases (red). The three principal signaling pathways illustrated are the (1) phospholipase C pathway, (2) Ras-MAP kinase pathway; and (3) PI-3 kinase pathway. These pathways lead to nuclear translocation of transcription factors (green), such as CREB and NFκB, and ultimately regulation of gene expression. The binding of neurotrophins or proneurotrophins to p75NTR can activate BAD via the JNK cascade and eventually engages, via release of effectors from mitochondria, caspases involved in apoptosis (see also Fig. 18.7). There are several venues for crosstalk between the survival- and pro-apoptotic signaling cascades as indicated. Abbreviations not already explained in Legend 18.11: BAD, bcl2-antagonist of cell death; Bcl2, B-cell lymphoma 2; Ca, calcium; CREB, cyclic AMP response element binding protein; Cyto C, cytochrome C; DAG, diacylglycerol; JNK, C-Jun N-terminal kinases; NFκB, nuclear factor-kappa B; NRAGE, neurotrophin receptor interacting MAGE (melanoma antigen gene expression) homolog; NRIF, neurotrophin receptor interacting factor; p75NTR, p75 neurotrophin receptor; (pro)NT, pro-neurotrophin; Raf, Ras, and RhoA, small GTPases; RSK, ribosomal S6 kinase. For details, see text, and Huang and Reichardt (2001), Chao (2003), and Barker (2004).
One of the best known Ras-dependent signaling pathways involves the ERK family of MAP kinases. This cascade is composed of serine/threonine kinases that are serially phosphorylated and activated. The initial member of this cascade is Raf, which phosphorylates and activates MEK, which phosphorylates the MAP kinases (ERK1 and ERK2). Once activated, these MAP kinases in turn phosphorylate a number of cytoplasmic and nuclear effectors. Importantly, activation of MAP kinases and their substrate, the protein kinase RSK, results in the phosphorylation of a number of transcription factors, including CREB (cAMP responsive element-binding protein), which controls expression of immediate-early genes (i.e., c-fos and c-jun) and delayed response genes. Once activated, transcription factors cause rapid and long-lasting changes in gene expression regulating cell survival, axonal and dendritic growth, neuronal differentiation, synaptic potentiation, and plasticity. In addition, the requirement on Ras-dependent pathways for biological responses varies among neuron types.
Neurotrophic factor receptor activation of the PI-3 kinase pathway is essential for the normal survival of many neurons. The phosphatidyl inositides made by PI-3 kinase regulate in part the activity of AKT (protein kinase B). This important protein kinase plays a critical role in controlling the biological activity of several regulatory proteins that govern normal PCD (see Fig. 18.12 and previous sections). Trk receptor activation of both the PI-3 kinase and the Ras-ERK pathways suppresses the capacity of activated p75NTR to induce cell death programs via Jun kinase. p75NTR may therefore induce neuronal death when appropriate neurotrophin binding to trk receptors is absent. Interactions between the multiple trk receptor signaling pathways in different cell phenotypes and under a variety of dynamic receptor signaling conditions provide a rich assortment of interrelated mechanisms to govern the PCD machinery.
Summary
Three tyrosine receptor kinases and trk family members—trkA, trkB, and trkC—transduce neurotrophin signals. Signaling pathways used by the trk family members are shared with those activated by many other growth factor receptors. Neurotrophin binding to trk causes receptor dimerization and phosphorylation of cytoplasmic tyrosine residues. Phosphotyrosines recruit cytosolic adaptor proteins that couple the activated receptor with intracellular signaling pathways. The biological response of the cell to a neurotrophic factor is dependent on the dynamic status of the pathway that can vary with recent cell history. Three of the well-known signaling pathways are PLC-g, Ras-ERK kinase, and PI-3K. Normal PCD is governed in many neurons by PI-3 kinase activation of the protein kinase Akt. Trophic factor deprivation of dependence receptors can activate pro-apoptotic mechanisms that promote cell death.
Multiple Cytokines and Growth Factors Implement Neurotrophic Activities
Several Cytokines Mediate Cell Interactions within the Nervous System
We have emphasized neurotrophins as an example of the activities of NTFs in the developing and mature nervous system. Yet there are several other important families of NTFs. Often NTF families utilize receptors with co-receptor complexes (e.g., neurotrophins—trk, p75NTR; and GDNF family ligands—ret and GFRs). In many tissues, intercellular induction factors are important for governing the proliferation and differentiation of both embryonic and adult stem cells. In addition, these factors also play an important role in the response of tissues to trauma, inflammation, infection, or tumor growth. In 1974, Stanley Cohen proposed that both lymphocyte-derived and nonlymphocyte-derived chemotactic and migration inhibitory factors be grouped into families of cytokines (“cell movement factors”). More recently, this term has been adopted as a general umbrella for many families of secreted proteins that mediate diverse biological responses, including changes in the immune system (interleukins), tumor cytotoxicity (tumor necrosis factors), and inhibition of viral replication or cell growth (interferons) (Table 18.4).
Table 18.4 Cytokine and Growth Factor Families
| Family | Representative Members | Original Biological Activities |
| Neurotrophins | NGF, BDNF, NT-3, NT-4/5, NT-6 | Neuronal survival and differentiation |
| Neuropoietic cytokines | CNTF, LIF, CT-1, CLC/CLF, oncostatin M | Survival of ciliary neurons and motoneurons, leukemia inhibitory activity, increased cholinergic properties |
| Tissue growth factors | TGF-α, TGF-β, FGFs, IGF-Iα, | Cell proliferation and differentiation in diverse tissues and organs |
| GDNF family | IGF-Ib, IGF-II, EGF, PDGF GDNF, neurturin, artemin, persephin | Neuronal survival and dopaminergic cell differentiation, and morphogens in kidney and sperm development |
| Interleukins | IL-1α, IL-1β, IL-2 through IL-15 | Immunoregulation, diverse activities in the immune system |
| Tumor necrosis factors | TNF-α, TNF-β | Tumor cytotoxicity |
| Chemokines | MCAF, MGSA, RANTES, NAP-1, NAP-2, MIP-1 | Leukocyte chemotaxis and cell activation |
| Colony-stimulating factors | G-CSF, M-CSF, GM-CSF | Hematopoietic cell proliferation and differentiation |
| Interferons | IFN-α, IFN-β, IFN-γ | Inhibition of viral replication, cell growth, or immunoregulation |
| CDNF/MANF family | CDNF, MANF | Survival of dopaminergic neurons |
Many cytokines originally were named according to the particular biological activity that was utilized for their isolation, only to be later rediscovered or identified as important mediators of other physiological processes. For example, some factors were isolated on the basis of their ability to enhance the survival of specific populations of neurons isolated in vitro. These include ciliary neurotrophic factor (CNTF); glial cell line–derived neurotrophic factor (GDNF); and the other GDNF family members, neurturin, persephin, and artemin. CNTF is a member of a broader family of neuropoietic cytokines, including leukemia inhibitory factor (LIF), oncostatin M, cardiotrophin-like cytokine/cytokine-like factor (CLC/CLF) and cardiotrophin-1 (CT-1), that share a common three-dimensional structure and receptor subunits. CNTF originally was isolated and named because it supports the survival of neurons cultured from the parasympathetic ciliary ganglion.
The GDNF family primarily supports enteric neurons, dopaminergic neurons, and some motoneurons. Other proteins that originally were identified as mitogens or chemotactic factors in nonneuronal tissues also have been shown to affect either the survival or the differentiation of neurons, including the fibroblast growth factors (FGFs), insulin-like growth factors (IGFs), and hepatocyte growth factor (HGF). More recent additions to this list include cerebral dopamine neurotrophic factor (CDNF) and mesencephalic astrocyte-derived neurotrophic factor (MANF) (Lindholm & Saarma, 2010). The first neuronal function identified for LIF was the induction of cholinergic properties in cultured sympathetic neurons, but it also supports the survival of several classes of neurons and induces neural precursors to become astrocytes.
LIF, which has a number of actions in the immune system and other nonneuronal tissues, shares receptor subunits with the CNTF family and therefore can mimic both the survival and the cholinergic differentiation activities of CNTF observed in culture. Many cytokines important for the development or maintenance of other organs and tissues are also widely expressed within the nervous system. Their roles in the nervous system, however, remain to be defined. Likewise, factors first recognized as neuronal survival factors have mitogenic properties for either nonneuronal cells or neuronal precursors. As a result, these pleiotropic factors are grouped into families based on their protein sequences and receptor usage rather than on their biological properties. Table 18.4 summarizes major known ligands, receptors, and biological functions for cytokines and growth factor families.
Neurotrophic Factors Have Multiple Activities
Neurotrophic factors that prevent neuronal death during development appear to have many other important biological activities, including effects on cell proliferation, migration, differentiation, axonal growth and sprouting, alterations in dendritic arbors, and synaptic plasticity of the nervous system (Table 18.3). Frequently, different populations of neurons respond to the same factor in distinct ways, and the same neuron may respond differently to the same factor at different developmental stages. Variations in neuronal responses to the same factor appear to depend not only on modifications of trophic receptors and their binding partners, but also on potential differences in the intracellular context of downstream signaling pathways. The distinctive activities of an NTF on different cells or during different epochs of development reflect the intrinsic properties of differentiating neurons or a dynamic change in a trophic response due to alterations in neural activity or other signaling events.
As described earlier, neurotrophins have been shown repeatedly to play fundamental roles in the survival of many peripheral neuronal populations (Table 18.2; Fig. 18.8). However, relatively few changes in the number of neurons have been observed in the CNS of neurotrophin or trk receptor null mutant mice. This observation was surprising because significant cell death occurs in the CNS, and NTFs and their receptors are widely expressed in the CNS during this period. The complexity and number of synaptic relationships established by CNS neurons compared to those established by PNS neurons may partially explain why CNS neurons are not as sensitive to the loss of a single neurotrophin or neurotrophin receptor (Fig. 18.13). The trophic support to CNS neurons is likely to arise from multiple families of NTFs with possible synergistic and/or compensatory effects.
Figure 18.13 Possible sources of trophic support for peripheral (PNS) and central (CNS) neurons. (A) Many peripheral neurons such as sympathetic and bipolar sensory neurons (left side) have only two sources of support: one in the periphery (Target #1) for retrograde support and one from afferents (AFF #1) or the central target (for bipolar sensory neurons). Glial cells (gray) may also provide trophic factors. In contrast, central neurons (right side) receive synaptic input from many different types of neurons (AFF #1, 2, and 3), which may serve as a source of anterograde trophic support. Central neurons may also project to several different targets (Targets #1, 2, and 3), which each may provide retrograde trophic support. We list trophic factors that have been demonstrated to be either anterogradely or retrogradely transported along axons, with question marks when transport is suspected, but not proven. Adapted, with permission, from Snider (1994) and von Bartheld, Wang, and Butowt (2001). (B) Schematic illustration of different sources of potential trophic signals acting on one specific cell population (motor neurons in the spinal cord). Motor neurons (green cell bodies) can receive trophic support from a number of different sources. Axon terminals of the motor neurons have access to diffusible muscle-derived (1) or extracellular matrix-associated (2) trophic factors. Schwann cells (3) in the peripheral nerve or ventral root are another source of trophic support. Glial cells (4) in the spinal cord (astrocytes and/or oligodendrocytes) may also influence motor neuron survival. Motor neurons receive afferent input from several sources, including descending fibers from the brain, spinal interneurons, and axons from dorsal root ganglion (DRG) neurons (5), which can supply trophic support. Finally, motor neurons may influence their own growth and survival via autocrine or paracrine trophic support (6), as well as respond to trophic support provided by circulating hormones (7).
The traditional view of the neurotrophic hypothesis has been that trophic support is derived from target tissues, but other sources of trophic support are now recognized (Figs. 18.13 and 18.14). Interpretation of mRNA localization studies is complicated by the fact that NTFs may not simply be released locally in the region of the cell body, but may also (or instead) be delivered to distant regions of the nervous system by anterograde or retrograde axonal transport. Thus, although the original neurotrophic hypothesis that neurons depend on target-derived trophic factors still holds true for many neurons during critical periods of development (including sympathetic neuron dependence for target-derived NGF), other sources of trophic support appear to play an important functional role in development, as briefly described earlier. Additionally, as described below, neuronal survival not only depends on competing successfully for trophic support but also on avoiding those cell–cell interactions that can actively kill cells, another example of how the traditional view of the neurotrophic hypothesis has had to be modified.
Figure 18.14 During embryonic/fetal development motoneurons establish functional synaptic connections with skeletal muscles via the neuromuscular junction (NMJ). This normal synaptic transmission activates skeletal muscles resulting in embryonic movements. If left unperturbed, this muscle activity reduces the production of (or access to) muscle-derived NTFs resulting in the PCD of a subset of motoneurons (MNs) (A). By contrast, if muscle activity is blocked, either genetically or pharmacologically, causing paralysis, the PCD of MNs is prevented (B). In addition to the role of synaptic activity at neuronal targets (e.g., the NMJ) in regulating PCD, afferent synapses also play a role in regulating PCD. A model is illustrated that integrates afferent- and target-derived influences on the choice of neuronal death or survival (C). In the developing chick embryo ciliary ganglion (CG, blue cells), young CG neurons are very heterogeneous with respect to the number of α7nAChRs that regulate calcium influx (e.g., b vs c) and they are already innervated by afferent cholinergic axons from the accessory oculomotor nucleus (gray terminals). There is also very little of the endogenous chicken prostate stem cell antigen (chPSCA), a prototoxin that blocks α7nACHR activation (b). Thus, young neurons with excessive α7nAChRs have a higher probability of PCD. As the developing neurons grow out axons that reach their targets in the eye, they become exposed to target-derived NTFs such as CNTF, GDNF, activin as well as other as yet unidentified factors. The influence of these factors facilitates maturation of the neurons, enhancing calcium buffering as well as inducing the expression of chPSCA. The end results are mature neurons (a), stable synapses, a population of silenced nAChRs together with active nAChRs at the stabilized synapses (Part C courtesy of Rae Nishi; see Hruska and Nishi, 2007 (J. Neurosci. 27, 11501-11509)).
Finally, although beyond the scope of the present chapter, neurotrophins and other NTFs and their receptors also are distributed throughout the mature brain, and NTFs can alter neural activity by rapid and long-term changes in synaptic transmission. Neurotrophins and other NTFs also play a role in modulating long-term changes in functional and anatomical plasticity in the developing and mature brain by altering long-term potentiation, synaptic connectivity, development of adult-generated neurons, and responses to stress, inflammation, and trauma (Table 18.3).
Summary
There are multiple families of NTFs. CNTF and LIF belong to a neuropoietic cytokine family. The expression and biological properties of these cytokines distinguish them from neurotrophins. They possess widespread neurotrophic activity for many different neuronal and nonneuronal populations in vitro and facilitate the cholinergic differentiation of sympathetic and motor neurons. The GDNF family (GDNF, neurturin, artemin, and persephin) is related to the TGF-β gene family and has effects on enteric, dopaminergic, and motor neurons. FGFs may play important roles during development or after injury. NTFs have different functions during different developmental stages: After early effects on proliferation and phenotype determination (stem cells), NTFs influence neuronal migration, survival, differentiation, and dendritic, as well as axonal growth. At later stages, NTFs regulate synaptic competition and plasticity, LTP, responses to injury, regeneration, and adult neurogenesis. The activity of multiple trophic factors from multiple sources is integrated to achieve a functional circuitry.
Programmed Cell Death is Regulated by Interactions with Targets, Afferents, and Nonneuronal Cells
The PCD of vertebrate neurons and their precursors can occur at any stage of neuronal development from neurogenesis and neurulation to the time of establishment of synaptic connections with targets and afferents, and can involve mitotically active cells and migrating neurons, as well as undifferentiated and immature postmitotic cells (Fig. 18.3). However, the most common and historically the best studied type of PCD of neurons involves postmitotic, differentiating cells that die while establishing synaptic connections with other neurons and target cells (Oppenheim, 1991).
Because massive neuronal death and its regulation were first clearly recognized in early studies of neuron-target interactions (Oppenheim, 1981; Cowan, 2001), the role of targets in controlling PCD historically has received the most attention. However, as discussed in the previous section, signals derived from afferent inputs, as well as from nonneuronal cells such as central and peripheral glia and endocrine glands (e.g., steroid hormones), have now been shown to be possible sources of trophic regulation of cell death and survival (Fig. 18.13).
Studies of the regulation of vertebrate neuronal PCD have shown that targets are critically involved in regulating how many postmitotic cells in the innervating population survive or die; in this way the final number of surviving neurons can be maintained at an optimal level despite individual differences in target size. Complete or partial deletion of targets reduces survival, whereas increasing the size or number of available targets results in increased survival (Oppenheim, 1991).
As noted in a previous section, if too few neurons are generated by proliferation in individual animals, the extent of PCD in that population may be curtailed so as to preserve an optimal quantitative relationship between neurons and their synaptic targets (DeMarco Garcia & Jessell, 2008). Although it is thought that the relationship between neuronal survival and the availability of synaptic targets is proportional and linear (this has been called numerical-, quantitative-, size-, or systems- matching), there may also be other ways in which the number of neurons that innervate a target can be regulated. For example, signals from glial cells along axon pathways may modulate neuron numbers prior to target innervation. Spinal motoneurons provide one of the best examples of how systems-matching is regulated by a competition for target-derived signals (NTFs) that promote survival (the Neurotrophic Hypothesis).
However, before discussing the example of motoneurons, it is first necessary to discuss the principle of competition. Competition at the cellular level is a key principle in the conceptualization of neuronal death and survival. It occurs when individual cells with different properties interact with each other, directly or indirectly, and results in the survival of the “stronger” cell and the death of the “weaker” cell. However, despite its long history, neither the origins of nor the characteristics themselves that define weak versus strong neurons are well understood (Oppenheim, 1991).
Cells commonly adopt different phenotypes by virtue of either their lineage (autonomous specification) or cell–cell interactions (conditional specification). In some cases, however, cells may initially differ stochastically, independent of either lineage or cell–cell interactions, by the random generation of genetic noise and a means to amplify and stabilize the noise-generated differences between cells (Losick & Desplan, 2008). Although speculative, it seems plausible that the generation, amplification, and stabilization of genetic noise could provide a separate mechanism whereby neurons become “stronger” (winners) by virtue of, for example, their expression of NTF receptors or of other features that confer a competitive edge over “weaker” cells (the losers).
In the case of avian spinal motoneurons that innervate limb skeletal muscle, the number of neurons that survive the period of PCD bears a 1 : 1 relationship with the number of primary myotubes present in individual muscle precursors during the period of cell death, rather than being correlated with the final number of myotubes or myofibers (i.e., muscle size) present after the cessation of cell death. Accordingly, in this situation, motoneuron numbers are controlled, in part, by signals that are limited by the number of primary myotubes available. Although for many populations of neurons the essential factors provided by targets that mediate neuronal survival are not known, extrapolation from what is known for sensory, sympathetic, and motor neurons suggests that specific target-derived NTFs are involved. The PCD of CNS neurons in the avian isthmo-optic nucleus and neurons in the mammalian thalamus and substantia nigra are controlled by many of the same mechanisms involved in the PCD of peripheral neurons, including a need for target- and afferent-derived signals. As discussed previously, the survival of myelin-forming glial cells also involves competition, in this case, competition for trophic signals derived from axons. In both neurons and glia, the final outcome of this competitive process (Figs. 18.3 and 18.4) is the survival of optimal numbers of cells for innervation (neurons) or myelination (glia).
Motoneurons (and some other neuronal populations as well, including retinal ganglion cells, isthmo-optic neurons, and ciliary ganglion neurons) have another interesting property in that their target dependency appears to be regulated by physiological synaptic interactions with their targets (Oppenheim, 1991). For example, following the formation of initial synaptic contacts between motoneurons and target muscles, the initiation of synaptic transmission activates the muscle and results in overt embryonic movements. Chronic blockade of this activity during the cell death period with specific drugs or toxins that cause paralysis, or in genetic mutants lacking synaptic transmission, prevents the death of all motoneurons (Fig. 18.14).
Because PCD is enhanced after the pharmacological blockade of afferent synaptic activity, the functional input provided by afferents also appears to be of fundamental importance in this situation. Functional afferent input may act to regulate the survival of both developing and adult-generated postsynaptic neurons by several different mechanisms:
1. Neurotransmitter release results in depolarization by afferent signals that can alter intracellular calcium levels in postsynaptic cells, which in turn can independently modulate survival.
2. Afferent activity can regulate the expression of NTFs and their receptors in postsynaptic cells.
3. The activity-dependent release of NTFs from terminals of afferent axons or from adjacent glial cells may also provide survival signals to postsynaptic cells.
In addition to the survival-promoting role of afferent-derived signaling, in some neuronal populations afferents may also have the counterintuitive role of promoting either death or survival depending, in part, on the extent to which receptor-mediated afferent signals change calcium influx: high influx promotes death, and low or moderate influx promotes survival. Because these neurons may also depend upon target-derived NTF signaling for survival, when taken together this kind of evidence indicates that the original version of the neurotrophic hypothesis is no longer sufficient for understanding the complexity of the cell–cell interactions involved in regulating the survival and death of developing neurons (Fig. 18.14). Competitive differences between neurons in a population may involve success in competing for survival-promoting signals as well as competitive success in avoiding death-promoting signals (Fig. 18.15) and these signals may include activity-independent signals derived from targets, afferents, and nonneuronal cells. In the final analysis, the various target- and afferent-derived extrinsic signals influence survival (or death) by modulating intracellular signaling pathways that regulate the expression of pro- and anti-apoptotic gene products (Figs. 18.7 and 18.12).
Figure 18.15 In (A) a model for developmental competition based on studies of developing sympathetic neurons is illustrated: (1) Before target innervation neurons (green) are modestly responsive to NGF; (2) upon target innervation and exposure to NGF, levels of TrkA, then BDNF and NT-4 (not shown) are increased; (3) induction of p75NTR expression, as well as differential sensitization of neurons, by modulation of NGF-TrkA signal strength and duration. (4) BDNF and NT-4 (apoptotic cues) kill neurons with low NGF-TrkA signaling; neurons with high NGF-TrkA signaling are resistant; (5) selection and neuronal death (Adapted from Deppmann et al., 2008, Science 320, 369-373). In (B) a model of neurodegeneration involving the selective loss of developing neurons and their axons is illustrated. Following NTF deprivation the transmembrane amyloid-ß precursor protein (APP) is routinely cleaved by ß- and γ-secretase enzymes to generate both amyloid-ß peptide (Aß) and the amino-terminal portion of APP (N-APP). The soluble N-APP may then undergo further processing before binding to the death receptor DR6. DR6 responds to interaction with N-APP by engaging and activating key mediators of apoptotic cell death: caspase-3 (C3) in the neuronal cell body and caspase-6 (C6) in axons. Caspase-6 can also cleave APP near the ß-secretase target site, potentially contributing to the formation and/or amplification of this apoptotic circuit (adapted from Nicholson, 2009, Nature 457, 970-971).
Although the cellular mechanisms are not yet well established, the survival and death of newly generated neurons in the adult hippocampus (Fig. 18.16) and olfactory system provide another example of how afferent activity can regulate these events. For example, learning, stress, motor activity, sensory stimuli, and other environmental influences appear to control granule cell numbers in the dentate gyrus of the hippocampus by activity-dependent modulation of neuronal survival, and odor experience modulates the survival of adult-generated olfactory neurons. Similar to developing neurons during embryonic and postnatal stages, the default state of developing adult-generated neurons is PCD and their survival, differentiation, and incorporation into functional circuits requires synaptic input and NTF signaling (Gage et al., 2008). Although there is still debate and controversy over whether and if so how adult-generated neurons contribute to brain functions that mediate learning, memory, and neural plasticity, increasing evidence supports the role of adult neurogenesis in specific behavioral adaptations.
Figure 18.16 Schematic illustration of the hippocampus (A) including the dentate gyrus, a portion of which is shown enlarged in B. (B) The dentate gyrus is comprised of several layers of differentiated granule cells (light gray circles and dark cell with dendrites on left). Stem cells in lower subgranule layer of dentate gyrus undergo cell division and a subset of these migrate into the granule layer and differentiate (cell with dendrites on right) whereas others migrate and begin differentiation but then undergo PCD (small dark cells with dashed axons and dendrites). Modified and redrawn from Kempermann (2006).
Summary
During development, the fate of a cell, including the decision to live or die, can be determined by intrinsic cell-autonomous mechanisms or by extrinsic signals derived from cell–cell interactions. The survival of most developing neurons is likely to be dependent on a variety of signals, including multiple NTFs derived from diverse sources, that serve to maintain survival and regulate differentiation in complex ways that reflect the specific requirements of neurons at each step in their development. Targets and afferents provide critical survival-promoting signals that prevent PCD. One major class of such signals are NTFs. Neurons are thought to compete for limiting amounts of these NTFs. For many populations of neurons, synaptic transmission and physiological activity also plays an important role in regulating survival. Accordingly, both NTFs and electrical activity/synaptic transmission are important for the regulation of PCD and are closely linked functionally.
The Role of Trophic Factors and Programmed Cell Death in Neuropathology
The role of trophic factors and the widespread occurrence of PCD during normal development indicate that cell loss, together with cell production (proliferation), is a fundamental mechanism for controlling final cell numbers in many tissues. During development trophic factors promote neuronal development, survival, and function. Therefore, it is not surprising that much research has been devoted to exploring their use as therapeutic agents to protect neurons in neuropathology (Lanni, Stanga, & Govoni, 2010). Trophic factor administration has been explored in animal models of many disorders of the nervous system, including the neurodegenerative diseases such as Parkinson’s Disease (PD) and Amyotrophic Lateral Sclerosis (ALS) (Rangasamy, Soderstrom, Barkay, & Kordower, 2010; Gould & Oppenheim, 2011).
In ALS, motoneurons become dysfunctional and die. Many trophic factors such as IGF-1, BDNF, GDNF, and CNTF that promote motoneuron survival during development have been administered to the ALS mouse model with positive results. These encouraging results of preclinical studies where administration of trophic factors promoted neuronal survival and delayed disease progression led to numerous clinical trials in ALS patients. Unfortunately, the results from the animal model studies did not translate into positive results in clinical trials. The failure of the clinical trials, however, does not warrant dismissal of the therapeutic potential of trophic factors, but rather brings to light the complexity of developing effective therapeutics for neurodegenerative diseases, including trial design, delivery mechanisms, dose and cellular targets.
The dysregulation of normal PCD can be maladaptive and pathological. Developmental, genetic, or congenital neurological defects may be caused by perturbations of PCD. For example, one proposed mechanism underlying autism is that there is a delay or absence of PCD (Courchesne et al., 2011). In the mature CNS, cell death exhibiting morphological characteristics and mechanistic components of PCD is a hallmark of many neuropathological conditions (Table 18.5). Neuropathology is a key example of why morphological characteristics of dying cells must be interpreted with caution. While DNA fragmentation and damage are observed in dying cells in pathological conditions, rarely are cells with morphological features of apoptosis observed; rather, dying cells appear to exhibit a combination of features characteristic of PCD types II, III, and necrotic cell death discussed above.
Table 18.5 Neuropathological Conditions in which Aberrant Cell Death Is Implicateda
| Developmental Disorders autism Down’s syndrome medullobastomas and neuroblastomas fetal alcohol syndrome sudden infant death syndrome spinal muscular atrophy leukodystrophies neonatal hypoxia-ischemia Triplet Repeat Disorders Huntington’s Disease spinal bulbar muscular atrophy Machado-Joseph disease (SCA 3) spinocerebellar ataxia (SCA) 1, 2, 6 fragile X syndrome Friedreich’s ataxia myotonic dystrophy Psychiatric Disorders schizophrenia bipolar disorder depression Trauma and Acquired Disorders prion diseases neuropathology of HIV infection multiple sclerosis gliobastomas epilepsy subarachnoid hemorrhage Wernicke’s encephalopathy spinal cord injury traumatic brain injury and ischemia/stroke Neurodegenerative Diseases Alzheimer’s amyotrophic lateral sclerosis Parkinson’s Other diabetic retinopathy glaucoma chronic pain Hirschsprung’s disease |
Courchesne et al., (2007). Neuron. 56(2):399–413; Courchesne et al., (2011), JAMA 306:2001–2010 Sawa (2001). J. Neural Trans. Suppl. 61:361–71 Johnson et al., (2009). Apoptosis 14:424–438 Cortese et al., (2006). Neurotoxicol Teratol 28:597–606 Sparks and Hunsaker (2002). Childs Nerv Syst 18:568–592 Lorson et al., (2010). Hum. Mol. Gen. 19:R111–8. Southwood and Gow (2001). Microsc Res Tech 52:700–8 Northington et al., (2011). Ann Neurol 69:743–758 Timchenko and Caskey (1999). Cell Mol Life Sci 55:1432–1447 Jarskog et al., (2005). Prog Neuropsychopharm Bio Psychiatry 29:846–858. Hu et al., (2011). Prog Neuropsychopharm Bio Psychiatry 35:11–7. McKernan et al., (2009). Prog. Neurobiol. 88:246–63. Liberski et al., (2008). Folia Neuropathol. 46:1–25. Sabri et al., (2003). Brain Pathol. 13:84–94. Artemiadis and Anagonostouli (2010). Eur. Neurol. 63:65–72. Krakstad and Chekenya (2010). Mol. Cancer 9:135-147. Friedman (2010). Neuroscientist 16:244-252/ Hasegawa et al., (2011). Acta Neurochir Suppl: 110:43-8. Hazell (2009). Neurochem Int 55:129-135 Rowland et al., (2008). Neurosurg Focus 25:E2 Engel et al., (2011). J. Cerebral Blood Flow and Metabolism 31:1196–1210 Ribe et al., (2008). Biochem J 415:165–182 Traub et al., (2011). Curr Neurol Neurosci Rep. 11:67–77; Pedrini et al., (2010). Hum. Mol. Genet 19:2974–86 Vives-Bauza and Przedborski (2011). Trends Mol Med 17:158–65 Barber et al., (2011). Invest Ophthalmol Vis Sci 52:1156–1163 Danesh-Meyer (2011). Curr Opin Opthalmol 22:78–86. Apkarian and Scholz (2006). Ann Neurol 69:13–21. Uesaka et al., (2007). Development 134:2171–81. |
aEnhanced or reduced cell death has been associated with many developmental and neuropathological conditions listed in the table. Also provided are review articles or reports describing the aspects of cell death identified in each disorder.
Autophagy is thought to be an adaptive response that can promote survival; however, dysregulation of autophagy is also proposed to contribute to neurodegenerative disorders (Banerjee, Beal, & Thomas, 2010). Although the morphological features may be different, many pathways of developmental cell death (see Fig. 18.7) can be re-activated in mature cells following insult or injury. As with developmental PCD, the mitochondria appear to have a central role in initiating cell death in epilepsy, Alzheimer’s disease (AD), PD, ALS, and stroke (Martin, 2010). The Bcl-2 proteins that have key roles in mediating release of pro-death factors from the mitochondria have altered expression or interactions in neuropathologies. Caspase activation has also been observed in postmortem tissue from AD, PD, ALS, and stroke patients.
In some of these diseases, specific gene mutations have been identified, and the identification of these mutations has facilitated the development of transgenic or cellular models of these diseases. Studies of these model systems indicate that the mutated proteins may directly or indirectly perturb inter- and intracellular mechanisms involved in the maintenance and survival of specific populations of neurons. For example, mutations in PTEN-inducing putative kinase 1 (PINK1) and Parkin are associated with some familial forms of PD, and these mutated proteins may function to modulate mitochondrial degradation. Mutations in copper, zinc superoxide dismutase 1 (SOD1) are associated with some forms of familial ALS. Misfolded mutant SOD1 has been shown to associate specifically with Bcl-2 in mitochondria, and this SOD1-Bcl-2 interaction appears to be toxic. In familial AD, mutations in the amyloid precursor protein (APP) genes have been identified. APP, amyloid-β, and the cytoskeletal protein tau are major contributors to the plaques and tangles observed in AD brains. Although the secretases appear to be the major proteases that modulate cleavage of APP, APP has also been shown to be a substrate for caspases, and inhibition of caspase activity resulted in decreased production of Aβ, a molecule thought to mediate neurotoxicity in AD. Tau is also a substrate for caspases, and the caspase-cleaved tau fragment is a component of the neurofibrillary tangles.
The examples above illustrate potential roles for mutated proteins in pathological cell death; however, other mutations may affect how damaged or dying cells are cleared. For example, mutations in progranulin resulting in decreased expression of the protein are associated with frontotemporal dementia. Interestingly, in C. elegans mutants lacking progranulin, there are fewer apoptotic corpses observed, consistent with the suggestion that the clearance of dying cells may contribute to disease processes (Kao et al., 2011).
The idea that aberrant cell death in the developing and adult nervous system may reflect a loss of normal control mechanisms for survival is appealing in that the increased understanding of the genetic regulation of PCD provides a potentially powerful and rational approach for the development of therapeutic treatment strategies. However, recent evidence questions the validity of the widely held assumption that it is the death of neurons that is the primary cause rather than the result of the pathophysiology in these diseases. There is growing evidence that neuronal and synaptic dysfunction that precedes any significant neuronal loss may be a major factor in the functional and behavioral pathology that characterizes many neurodegenerative diseases (Gould & Oppenheim, 2007; Saxena & Caroni, 2011). Therefore, identification of the initiating factors leading to cell dysfunction and death will be critical for the development of effective therapeutics.
Summary
Because PCD is primarily a developmental phenomenon, historically the major focus of investigation has been the normal biology of cell death in the embryo, fetus, and newborn. However, with the growing recognition that pathological neuronal cell death may share certain biochemical and molecular features with PCD, there is growing hope that a better understanding of PCD may reveal potential therapeutic strategies for the treatment of developmental neuropathologies, neurodegenerative disease, and neuronal loss following CNS trauma and disease.
References
1. Airaksinen MS, Saarma M. The GDNF family: Signaling, biological functions and therapeutic value. Nature Review Neuroscience. 2002;3:383–394.
2. Ameisen JC. Looking at death at the core of life in the light of evolution. Cell Death Differentiation. 2004;11:4–10.
3. Banerjee R, Beal MF, Thomas B. Autophagy in neurodegenerative disorders: pathogenic roles and therapeutic implications. Trends in Neuroscience. 2010;33:541–549.
4. Barde YA. Trophic factors and neuronal survival. Neuron. 1989;2:1525–1534.
5. Barker PA. p75NTR is positively promiscuous: novel partners and new insights. Neuron. 2004;42:529–533.
6. Bredesen DE, Rao RV, Mehlen P. Cell death in the nervous system. Nature. 2006;443:796–802.
7. Buss RR, Sun W, Oppenheim RW. Adaptive roles of programmed cell death during nervous system development. Annual Review of Neuroscience. 2006;29:7–35.
8. Chao MV. Neurotrophins and their receptors: A convergence point for many signaling pathways. Nature Review Neuroscience. 2003;4:299–309.
9. Clarke PGH, Oppenheim RW. Neuron death in vertebrate development: In vivo methods. Methods in Cell Biology. 1995;46:277–323.
10. Clarke PGH, Clarke S. Nineteenth century research on naturally occurring cell death, and related phenomena. Anatomy and Embryology. 1996;193:81–99.
11. Copani A, Uberti D, Sortino MA, Bruno V, Nicoletti F, Memo M. Activation of cell-cycle-associated proteins in neuronal death. Trends in Neuroscience. 2001;24:25–31.
12. Courchesne E, Mouton PR, Calhoun ME, et al. Neuron number and size in prefrontal cortex of children with autism. JAMA. 2011;306:2001–2010.
13. Cowan WM. Viktor Hamburger, and Rita Levi-Montalcini: The path to the discovery of nerve growth factor. Annual Review of Neuroscience. 2001;24:551–600.
14. DeMarco Garcia NV, Jessell TM. Early motor pool identity and muscle nerve trajectory defined by postmitotic restrictions in NKX6.1 activity. Neuron. 2008;57:217–231.
15. Deppmann CD, Mihalas S, Sharma N, Lonze BE, Niebur E, Ginty DD. A model for neuronal competition during development. Science. 2010;320:369–373.
16. Egan MF, Kojima M, Callicott JH, et al. The BDNF val66met polymorphism affects activity-dependent secretion of BDNF and human memory and hippocampal function. Cell. 2003;112:257–269.
17. Elliott MR, Chekeni FB, Trampont PC, et al. Nucleotides released by apoptotic cells act as a find-me signal to promote phagocytic clearance. Nature. 2009;461:282–286.
18. Fuchs Y, Steller H. Programmed cell death in animal development and disease. Cell. 2011;147 742–458.
19. Gage FH, Kempermann G, Song H, eds. Adult neurogenesis. Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press; 2008.
20. Gauthier LR, Charrin BC, Borrell-Pages M, et al. Huntingtin controls neurotrophic support and survival of neurons by enhancing BDNF vesicular transport along microtubules. Cell. 2004;118:127–138.
21. Gould TW, Oppenheim RW. Synaptic dysfunction in disease and following injury in the developing and adult nervous system: Caveats in the choice of therapeutic intervention. Neuroscience and Biobehavavioral Reviews. 2007;31:1073–1087.
22. Gould TW, Oppenheim RW. Motor neuron trophic factors: Therapeutic use in ALS?. Brain Research Reviews. 2011;67:1–39.
23. Horvitz HR. Worms, life and death (Nobel lecture). Chembiochem. 2003;4:697–711.
24. Huang EJ, Reichardt LF. Neurotrophins: Roles in neuronal development and function. Annual Review of Neuroscience. 2001;24:677–736.
25. Ip N, Yancopoulos G. The neurotrophins and CNTF: Two families of collaborative neurotrophic factors. Annual Review of Neuroscience. 1996;19:491–515.
26. Jiao S, Li Z. Non-apoptotic function of Bad and Bax in long-term depression of synaptic transmission. Neuron. 2011;70:758–772.
27. Kao AW, Eisenhut RJ, Martens LH, et al. A neurodegenerative disease mutation that accelerates the clearance of apoptotic cells. Proceedings of the National Academy of Science of United States of America. 2011;108:4441–4446.
28. Kempermann G. Adult neurogenesis: Stem cells and neuronal development in the adult brain New York: Oxford University Press; 2006.
29. Kinchen JM, Ravichandran KS. Journey to the grave: signaling events regulating removal of apoptotic cells. Journal of Cell Science. 2007;120:2143–2149.
30. Kroemer G, Galluzi L, Vandenbeele P, et al. Classification of cell death: recommendations of the nomenclature committee on cell death. Cell Death and Differentiation. 2009;16:3–11.
31. Lanni C, Stanga M, Govoni S. The expanding universe of neurotrophic factors: therapeutic potential in aging and age-associated disorders. Current Pharmaceutical Design. 2010;16:698–717.
32. Lauber K, Bohn E, Kröber SM, et al. Apoptotic cells induce migration of phagocytes via caspase-3-mediated release of a lipid attraction signal. Cell. 2003;113:717–730.
33. Lessmann V, Gottmann K, Malcangio M. Neurotrophin secretion: Current facts and future prospects. Progress in Neurobiology. 2003;69:341–374.
34. Lewin GR, Barde YA. Physiology of the neurotrophins. Annual Review of Neuroscience. 1996;19:289–317.
35. Lindholm P, Saarma M. Novel CDNF/MANF family of neurotrophic factors. Development Neurobiology. 2010;79:360–371.
36. Losick R, Desplan C. Stochasticity and cell fate. Science. 2008;320:65–68.
37. Lu B, Pang PT, Woo NH. The yin and yang of neurotrophin action. Nature Review Neuroscience. 2005;6:603–614.
38. Mallat M, Marín-Teva JL, Chéret C. Phagocytosis in the developing CNS: more than clearing the corpses. Current Opinion in Neurobiology. 2005;15:101–107.
39. Martin LJ. Mitochondrial pathobiology in Parkinson’s disease and amyotrophic lateral sclerosis. Journal of Alzheimers Disease. 2010;20(Suppl 2):S335–356.
40. McAllister AK, Katz LC, Lo DC. Neurotrophins and synaptic plasticity. Annual Review Neuroscience. 1999;22:295–318.
41. Mehlen P. Dependence receptors: the trophic theory revisited. Science Signaling. 2010;3:pe 47.
42. Nedelcu AM, Driscoll WW, Durand PM, Herron MD, Raskidi A. On the paradigm of altruistic suicide in the unicellular world. Evolution. 2010;65:3–20.
43. Nikoletopoulou V, Lickert H, Frade JM, et al. Neurotrophin receptors TrkA and TrkC cause neuronal death whereas TrkB does not. Nature. 2010;467:59–63.
44. Oppenheim RW. Neuronal cell death and some related regressive phenomena during neurogenesis. In: Cowan WM, ed. Studies in developmental neurobiology: Essays in honor of Viktor Hamburger. New York: Oxford; 1981;74–133.
45. Oppenheim RW. Cell death during development of the nervous system. Annual Review of Neuroscience. 1991;14:453–501.
46. Poo MM. Neurotrophins as synaptic modulators. Nature Review Neuroscience. 2001;2:24–32.
47. Potts RR, Singh S, Knezek M, Thompson CB, Deshinukh M. Critical function of XIAP in regulating caspase activation during sympathetic neuronal apoptosis. Journal of Cell Biology. 2003;163:789–799.
48. Putcha GV, Deshmukh M, Johnson EM. Inhibition of apoptotic signalling cascades causes loss of trophic factor dependence during neuronal maturation. Journal of Cell Biology. 2000;149:1011–1017.
49. Rangasamy SB, Soderstrom K, Barkay RAE, Kordower JH. Neurotrophic factor therapy for Parkinson’s Disease. Progress in Brain Research. 2010;184:237–264.
50. Roux PP, Barker PA. Neurotrophin signaling through the p75 neurotrophin receptor. Progress in Neurobiology. 2002;67:203–233.
51. Saxena S, Caroni P. Selective neuronal vulnerability in neurodegenerative diseases: from stressor thresholds to degeneration. Neuron. 2011;71:35–48.
52. Schmitz C, Hof PR. Design-based stereology in neuroscience. Neuroscience. 2005;130:813–831.
53. Segal R, Greenberg M. Intracellular signalling pathways activated by neurotrophic factors. Annual Review of Neuroscience. 1996;19:463–489.
54. Singh KK, Park KJ, Hong EJ, et al. Developmental axon pruning mediated by BDNF-p75NTR-dependent axon degeneration. Nature Neuroscience. 2008;11:649–658.
55. Snider WD. Functions of the neurotrophins during nervous system development: What the knockouts are teaching us. Cell. 1994;77:627–638.
56. von Bartheld CS, Fritzsch B. Comparative analysis of neurotrophin receptors and ligands in vertebrate neurons: Tools for evolutionary stability or changes in neural circuits?. Brain Behavior and Evolution. 2006;68:157–172.
57. von Bartheld CS, Wang X, Butowt R. Anterograde axonal transport, transcytosis, and recycling of neurotrophic factors: The concept of trophic currencies in neural networks. Molecular Neurobiology. 2001;24:1–28.
58. Wyllie AH, Kerr JE, Currie AR. Cell death: the significance of apoptosis. International Review of Cytology. 1980;68:251–306.
59. Zweifel LS, Kuruvilla R, Ginty DD. Functions and mechanisms of retrograde neurotrophin signaling. Nature Review Neuroscience. 2005;6:615–625.
Suggested Readings
1. Hamburger V. History of the discovery of neuronal death in embryos. Journal of Neurobiology. 1992;23:1116–1123.
2. Huang EJ, Reichardt LF. Neurotrophins: Roles in neuronal development and function. Annual Review of Neuroscience. 2001;24:677–736.
3. Jacobson M. Developmental Neurobiology New York: Plenum; 1991.
4. Jellinger KA. Basic mechanisms of neurodegeneration: a critical update. Journal of Cellular and Molecular Medicine. 2010;14:457–487.
5. Levi-Montalcini R. The nerve growth factor 35 years later. Science. 1987;237:1154–1162.
6. Oberst A, Bender C, Green DR. Living with death: The evolution of the mitochondrial pathway of apoptosis in animals. Cell Death and Differentiation. 2008;15:1139–1146.
7. Oppenheim RW. The concept of uptake and retrograde transport of neurotrophic molecules during development: History and present status. Neurochemical Research. 1996;21:769–777.
8. Oppenheim RW, Caldero J, Esquerda J, Gould T. Target-independent programmed cell death in the developing nervous system. In: Kalverboer AF, Gramsbergen A, eds. Brain, and behaviour in human development. Dordrecht, The Netherlands: Kluwer; 2001:343–407.
9. Purves D. Body and brain: A trophic theory of neuronal connections Cambridge, MA: Harvard; 1988.
10. Putcha GV, Johnson EM. Men are but worms: neuronal cell death in C Elegans and vertebrates. Cell Death and Diffierentiation. 2004;11:38–48.