Cell Death Mechanisms and Modulation in Traumatic Brain Injury

Introduction 

Neuronal cell death is required for normal development of the CNS, as well as for removal of dysfunctional cells in pathological conditions including trauma. Excessive neuronal cell loss, however, underlies acute and chronic neurodegenerative disorders.1 Conceptually, trauma-induced neuronal cell loss has been characterized as either primary or secondary, with primary referring immediate cell death related to physical disruption of membranes and secondary referring to delayed cell death in surrounding or distant regions. Secondary neuronal cell death results from both physiological and biochemical changes induced by the insult.

Although cell death itself can be readily defined as the irreversible loss of integrated cellular activities or terminal disruption of key cellular substructures (or both), it has been more difficult to develop a generally acceptable taxonomy of cell death. Criteria such as type of inducer, specific molecular mechanisms, energy dependence, and morphological changes have all been used to classify cell death, leading to widespread confusion. Moreover, failure to cite critical prior work in other fields has resulted in both controversy and a tendency to oversimplify. For example, neuronal cell death has been commonly delineated into two distinct categories: necrosis, which is considered a passive process associated with loss of ionic homeostasis, failure of membrane integrity, and organelle and cell swelling, and apoptosis, often reflecting an energy-dependent process characterized by cytoplasmic and nuclear condensation and fragmentation, diminished cell volume, and relative preservation of organellar structure. The limitations of such nomenclature become readily apparent when apoptotic-type mechanisms are associated with necrotic-type morphology.2 In fact, these two patterns of cell death often coexist, and intermediate morphological forms have been identified, leading some to propose the term aponecrosis.3

The History of Cell Death: Morphological-Based Classification 

The word necrosis, which was used by the Greco-Roman physician Galen, originates from the ancient Greek word nekros, meaning corpse.4 In the mid-19th century, Rokitansky5 and Virchow6 used the term ‘necrosis’ to refer to macroscopic phenomena reflecting advanced tissue breakdown. By the end of the 19th century, use of microscopy techniques revealed the connection between macroscopic necrosis and cell death.4 Various changes observed in the cellular nucleus after ischemia or injury were described by terms such as ‘pyknosis’ (chromatin condensation), ‘karyolysis’ (nuclear disappearance), ‘karyorrhexis’ (nuclear fragmentation) and ‘chromatin margination’ (chromatin condensation on the inside of the nuclear membrane).7, 8, 9, 10 Cell death was also observed under physiological conditions, such as regressing ovarian follicles, with reference to pyknotic chromatin as well as cell fractionation into smaller bodies, a process called chromatolysis.11 Chromatolysis was also identified in the lactating mammary gland12 and in breast cancer tissue.13 In the early 1950s, chromatolysis was rediscovered in physiological cell death during embryo development, with the key detail that mitochondria appeared largely unaffected.14

Starting in the 1970s, Kerr, Wyllie, and colleagues in a series of seminal articles15, 16, 17, 18 described a conserved cell death process that occurs in development, physiological tissue turnover, and various pathological conditions. This process was characterized by ultrastructural features, such as condensation of the cytoplasm and chromatin, cell shrinkage, formation of chromatin balls, normal organelles, and fragmentation of cells by budding of membrane enclosed fragments (i.e., apoptotic bodies) and the absence of an inflammatory response. Initially, they called this form of cell death “shrinkage necrosis” but later coined the term ‘apoptosis,' from the ancient Greek words apo (from) and ptosis (falling), to reflect the falling leaves of autumn. Although ‘apoptosis’ and ‘chromatolysis’ were evidently different names for a similar process, Kerr, Wyllie, and colleagues were apparently not aware of the prior literature. Wyllie19 also described a “necrotic cell death” phenotype, which was present in severely injured tissues and suggested the dichotomy of apoptotic cell death and necrotic cell death.

Also in the 1970s, Schweichel and Merker20 proposed a classification of developmental cell death based on lysosomes as a key ultrastructural feature: type 1 involving lysosomes of phagocytic cells (heterophagocytosis); type 2, involving the lysosomes of the dying cells themselves (autophagocytosis); and type 3, with no lysosomal involvement. In 1990, Clarke21 revised the Schweichel and Merker classification to include not only lysosomes but also ultrastructural changes affecting the nuclei or chromatin, cytoplasm, organelles, and cell budding or blebbing.

Each distinct cell death type shared a conserved set of ultrastructural features and was thought to reflect distinct mechanisms of cell demise. Type 1 is essentially apoptosis, wherein the apoptotic bodies are engulfed or destroyed by lysosomes of neighboring cells or macrophages. Type 2 is autophagic degeneration, marked by presence of autophagic vacuoles that destroy the cytosol and organelles of the cell. Unlike apoptosis, type 2 in general does not feature prominent nuclear pyknosis but shows dilation of organelles and is generally observed in regions undergoing massive degeneration20 (although exceptions are known in each case21). Most developmental cell death can be classified as type 1 or 2 in this nomenclature.

Type 3A (nonlysosomal vesiculate degradation) features prominent swelling of the organelles, presence of lacunar spaces in the cytosol and cellular disintegration without phagocytosis. Type 3B (cytoplasmic degeneration) starts with dilation of organelles and vacuolation of the cytosol, followed by nuclear degeneration by karyolysis, but the cell is phagocytosed without fragmentation. Although less common, some forms of developmental cell death reflect features of more than one type or lack a critical component.21 Type 3B developmental cell death shows ultrastructural characteristics (mitochondrial swelling, dilation of organelles, late karyolysis) similar to the necrotic cell death phenotype reported in severely injured tissues,19 further blurring the separation between physiological and pathological cell death.

Majno and Joris4 also disagree with the apoptosis versus necrosis dichotomy and argue the importance of identifying the point of no return in cell death. In their view, past this stage the cell should be considered dead, even if active processes such as cell dismantling continue. Others prefer a later threshold for death and define it as the irreversible loss of integrated cellular activity,22 which may correspond to the rupture of cell membranes and lysis of the cell, cell fragmentation in separate bodies, or cell engulfment by other cells.23 Majno and Joris24 interpret necrosis as the presence of dead tissue surrounded by living tissue. For them, necrosis is not a form of cell death, but instead reflects the secondary changes that occur after cell death, being the cellular equivalent of postmortem decay. Taking ischemia as a cell death inducer, Majno and Joris24 observe that cell swelling is one of the earliest observed changes and propose the term ‘oncosis’ to describe these prelethal processes leading to cell death (i.e., point of no return).

In fact, however, the term ‘oncosis’ (which derives from the ancient Greek word for swelling and is related to onkos, a mass or bulk) was already used much earlier to describe osteocyte cell death during rickets and during cartilage growth.25 Oncosis and type 3 cell death are virtually identical in their defining elements of cellular morphology. Majno and Joris,24 as well as others,26 propose limiting the term ‘necrosis’ to denote the late phases of cell degradation, irrespective of the mechanisms of the cell death. Levin et al.27 suggest adding a descriptor to mark the type of necrosis, for example apoptotic necrosis or oncotic necrosis. They argue further that only morphologic or ultrastructural analysis can inform the cell death type, and observe that other frequently used criteria are nonspecific; for example, apoptosis can affect large masses of cells and can be accompanied by inflammation28 and, alternatively, oncosis can occur in single cells.

Remarkably, few articles provide precise descriptions of cell death based on ultrastructural changes and distinguish between prelethal cell death processes (i.e., apoptosis, autophagy, or oncosis) and postlethal degradation (i.e., necrosis). In part, this may reflect the smaller number of studies using electron microcopy–based criteria, leading many authors to use the simplistic apoptosis versus necrosis dichotomy, or to focus on mechanisms rather than morphology to define cell death. Kroemer et al.23 suggest that authors refrain from making definitive pronouncements regarding cell death type and instead provide descriptive terms when presenting results (e.g., “caspase-3-positive cell death” instead of “apoptosis”).

Unfortunately, this useful suggestion has largely been ignored. Thus, the term ‘necrosis’ is commonly used to describe both type 3–oncotic cell death and the uncontrolled cell lysis often associated with tissue inflammation.29 Nonetheless, type 1–apoptosis and type 3–oncosis can be initiated by the same triggering events, with low intensity of the insult favoring apoptosis and high intensity of the insult pushing cells toward oncosis,30 or with high intracellular ATP levels favoring initiation of apoptosis and low intracellular ATP levels associated with nonapoptotic cell death.31 Moreover, data suggest that inhibition of one cell death pathway may redirect the cell toward other mechanisms and phenotypes of cell death.32

Another key concept is that of programmed cell death (PCD)—which is not equivalent to apoptosis. Programmed cell death was first described in the 1960s, even before apoptosis, in a developmental context as cell death that occurs at a predetermined time as a result of a genetic clock.33 Others view PCD as a cell death process that follows a fixed molecular pathway.34 Because all types of developmental cell death meet one or both criteria, they can all be considered variants of PCD. A more useful definition of PCD, inspired in part by the work of Sloviter,35 suggests that PCD encompasses all active cell deaths, in which cellular processes are required for pushing the cell to and beyond the point of no return. In this interpretation, PCD requires time to develop and involves cellular mechanisms that are potentially open to therapeutic intervention. In opposition to PCD, passive cell death results from overwhelming cell injuries (e.g., catastrophic cell damage) that instantaneously push the cell past the point of no return without involvement of active cell processes.36 Unfortunately, necrosis is often considered the paradigm of passive cell death, which incorrectly implies as untreatable many forms PCD with necrotic-type morphology, such as type-3–oncotic cell death.

Neuronal Cell Death in the CNS 

The study of neuronal cell death in the CNS is particularly challenging, because in bright-field microscopy most dying neurons appear shrunken, eosinophilic, and with pyknotic nuclei—regardless of the type of death.27 Even at the ultrastructural level, dying neurons do not well fit the cell death descriptions established in other tissues or during development.31 Thus, distinct types of neuronal cell death such as paraptosis37 have been described, featuring a requirement for gene expression, nonapoptotic morphology marked by vacuolization, and absence of caspase activation. Although it has some partial similarities with type 3B cell death (i.e., cytoplasmic degeneration),37 paraptosis is one of several forms of neuronal cell death with nonclassical morphological characteristics and with identified biochemical markers.

To attempt to clarify neuronal cell death phenotypes, studies led by Olney and Fujikawa have used ultrastructural features. In one of these studies, two types of neuronal cell death in the brain were identified, termed ‘physiologic’ cell death and ‘excitotoxic’ cell death.38 Physiologic cell death was initially characterized as a developmental death of neurons in the mammalian brain; its description is similar to type 1–apoptosis,39 although with some features that differ from apoptosis in other tissues.38 For example, in the CNS, although organelles initially appear normal except for mild mitochondrial swelling,39 mitochondrial degeneration and endoplasmic reticulum vacuolization become prominent after rupture of the nuclear membrane.38 This pattern of orderly morphological changes is found in neurons from various brain regions during development38 or in response to diverse injuries, such as ethanol or TBI.40

In contrast, excitotoxic cell death is characterized by rapid cytoplasmic changes, including marked cell swelling, as well as dilation of mitochondria and endoplasmic reticulum, followed by their rupture. Later, chromatin condensation in the nucleus results in small chromatin clumps that eventually consolidate to form a large irregular mass at the center of the nucleus. Subsequently, the cell membrane ruptures, but the nuclear membrane remains intact.40 Excitotoxic cell death closely resembles type 3B–oncosis cell death, except that excitotoxic cell death shows more intense and longer-lasting masses of clumped chromatin.

At the ultrastructural level, excitotoxic cell death is very different from apoptosis, but these two cell death processes are more difficult to distinguish under typical light microscopy.39 Excitotoxic cell death is detected in the first hours after TBI, mostly at the local lesion site, whereas PCD is seen later, and particularly at a distance from the site.38 These authors (i.e., the Olney and Fujikawa research group) concluded that physiological cell death and excitotoxic cell death are two fundamentally distinct neuronal cell death types, and that the activation of one path excludes the other. Notably, this interpretation conflicts with the continuum theory, the view that these death paradigms are extremes of the same process,41 and Fujikawa et al.42 contest the existence of a cell death continuum.

Based on ultrastructural cell changes, Fujikawa et al.43 describe two unique neuronal cell death phenotypes. One has features very similar to apoptosis; the other, excitotoxic cell death, shows ultrastructural changes similar to type 3B–oncosis. Unlike non-neuronal type 3B–oncotic cell death, however, neuronal excitotoxic cell death exhibits a condensed cytoplasm and nucleus (i.e., pyknosis). Fujikawa44 also emphasizes that few if any studies demonstrate typical ultrastructural features of neuronal apoptosis outside of neonatal animals. Moreover, he concludes that widely held concepts about the apoptosis-specificity of mechanisms and markers (such as TUNEL, internucleosomal DNA cleavage, and caspase activation) have no basis, because they can be detected and are required for execution of certain excitotoxic cell death. The latter observations indicate an active cell participation (PCD) in excitotoxicity. Therefore, characterization of cell death should be based on morphology—which reflects the summation of participating factors.

Cell Death Programs: Molecular Pathways that Participate in Cell Death 

Programmed cell death is a form of cell death executed by activation or inactivation of various molecular pathways, with multiple death phenotypes. Many varieties of PCD and associated molecular mechanisms have been identified in the CNS, as described in several excellent and comprehensive reviews.36, 45 Programmed cell death has been confirmed as a major cause of post-traumatic neuronal cell death and is associated with poorer prognosis in patients after TBI.46 Morphologically defined PCD includes apoptosis, autophagy, paraptosis, calcium-dependent death, and oncosis. The cell death mechanisms that mediate the specific PCD processes include, among many others, caspases and proapoptotic members of the Bcl-2 family (apoptosis), JNK and ATG orthologs (autophagy), ERK2 (paraptosis, a cell death associated with trophotoxicity), PARP/AIF (PARP/AIF-dependent death), calpains and cathepsis (calcium-dependent death), and JNK (oncosis).36, 45 Notably, few or none of these mechanisms are irreplaceably necessary for taking a cell past the point of no return. Most often, multiple mechanisms are simultaneously and redundantly activated, and in response to blocking any individual mechanism others can serve to execute the cell death. Several significant developments involving cell death mechanisms will be discussed.

Cooperation of Cell Death Programs 

Inhibition of caspase-dependent cell death pathways may not provide substantial neuroprotection, because of the associated activation of caspase-independent PCD pathways.83 In fact, there is growing support for the hypothesis that in every stress-induced model of cell death there may be multiple pathways involved84 and that the slower or less intense mechanisms become apparent primarily when the dominant mechanisms are attenuated.45 A key role is likely played by the cellular bioenergetic status. When this status is preserved, caspases are dominant and AIF-dependent cell death becomes important only after caspase activation has been blocked. Under bioenergetic deficient conditions, however, AIF may be the more predominant mechanism.

An important question is whether inhibition of the dominant pathway may actually initiate or enhance alternative pathways.3, 85, 86, 87, 88 Whether inhibition of the AIF pathway causes an enhanced caspase-dependent response has not been fully determined.89 There is, however, solid evidence that AIF and caspases act through parallel pathways, and that strategies that target both have potentially additive therapeutic effects.90 Autophagic and caspase-dependent PCD might also show a similar relationship.45

Cell Death Mechanisms: Implications for Treatment of TBI 

Challenges to successful treatment of TBI-induced neuronal cell death include the presence of a multitude of cell death pathways, which have both overlapping and distinct molecular mechanisms, and the short therapeutic windows for some types of neuronal cell death.91 Furthermore, in addition to neuron-specific pathways there is also neurotoxicity secondary to microglia-initiated inflammatory responses.92 These facts might explain, in part, the failure of clinical neuroprotection trials in TBI, which too often have only targeted a single cell death pathway or have modulated mechanisms with a relatively short therapeutic window.91

In preclinical studies, improved levels of neuroprotection have been obtained using therapeutic agents with multifunctional activities,93, 94 such as small cyclized dipeptides,95 progesterone,96 statins, and erythropoietin, among others. Treatment with various cyclic dipeptides significantly improved motor and cognitive recovery after TBI in both rat and mouse models, and attenuated both apoptotic and oncotic cell death in primary neurons.97 These agents limit mitochondria changes associated with cytochrome c release, decrease expression of secondary injury pathways such as cell cycle proteins and cathepsins or calpains, and increase expression of neuroprotective molecules, including brain-derived neurotrophic factor (BDNF) and heat shock proteins.97

In TBI models, progesterone treatment attenuates edema and inflammatory cytokines and limits neuronal loss by preventing mitochondrial changes, with improved functional outcomes.96 Erythropoietin provides neuroprotection after experimental TBI, likely reflecting its ability to stabilize mitochondrial function and to reduce inflammation and oxidative stress.98, 99 Statins have also been shown to significantly improve outcomes after experimental TBI, with effects likely due, at least in part, to their ability to activate Akt-dependent mitochondrial sparing pathways and to attenuate microglial activation.100, 101 What links these therapeutic strategies is the fact that each of the compounds shows pleiotropic actions, reflecting effects on multiple secondary injury pathways that are likely synergistic.98

A different strategy is to focus on therapies that specifically, and sometimes also simultaneously, target multiple PCD mechanisms. In this paradigm, multifunctional effects can be generated by multiple drug combinations, or by targeting single factors that modulate multiple secondary injury cascades (e.g., activation of the cell cycle,48, 49, 50 PARP-1,70, 102, 103 calpains,104 or HSP70), among others. Targeting processes such as autophagy represents a more complex issue, in light of studies indicating both cell death and neuroprotective activity for autophagy.105, 106 A schematic of several key mechanisms that mediate neuronal cell death after TBI is presented in FIG. 1, including both more mechanism-specific modulators and pluripotential agents with diverse but less clearly defined mechanisms of action.

View Large Image Download to PowerPoint

FIG. 1. Key molecular mechanisms of neuronal programmed cell death and their modulators. Green indicates more mechanism-specific modulators; yellow, pluripotential agents with diverse but less clearly defined mechanisms of action.

We suggest that improved therapeutic effects may result from treatment strategies that are directed at multiple specific targets and mechanisms of cell death, using either a combination of therapeutic agents or multifunctional (i.e., pluripotential) drug strategies—or with both approaches. Optimal targets should include both caspase-dependent and caspase-independent PCD. Because activation of these pathways occurs in parallel and peaks 1–3 days after injury, such strategies should be both additive or synergistic and should show a relatively wide therapeutic window. Also promising are therapies that target not only neuron-specific cell death mechanisms, but also block microglial-dependent neurotoxicity,92 or have actions on other cell types, such as oligodendroglia,47 astroglia,47 and endothelial cells.107

Conclusions 

The creation of a comprehensive and rational taxonomy of cell death is difficult because of the diversity of cell death types and the numerous molecular mechanisms involved. Moreover, a given phenotype—even with an ultrastructural requirement—may reflect different mechanisms, or the same mechanism, albeit with different expression level or intensity, may lead to different phenotypes. Nonetheless, we believe that the consistent application of several key concepts and strategies should help clarify the process.

Determination of a cell death phenotype should ideally require ultrastructural support. In the absence of ultrastructural details, a descriptive mode should be applied that includes the known morphological and mechanistic data (e.g., caspase-3-positive apoptosis). Morphology alone may be unable to address the question of programmed cell death versus passive cell death, considering that both forms of cell death share type 3–oncotic morphology. Furthermore, morphology as the final integrator of all molecular pathways involved in the cell death tends to reflect the dominant pathways and may mask the potential role of additional contributing processes. The latter issue can best be addressed by detailed mechanistic studies. It is increasingly evident that any PCD process may involve multiple, interdependent mechanisms. For these reasons, we suggest that the optimal therapeutic strategy to limit cell post-traumatic cell death is the use of combination or multipotential treatments that target multiple cell death pathways.