Introduction

Traumatic brain injury (TBI) is a significant cause of morbidity and mortality [1]. TBI is caused by primary and secondary brain damage resulting in loss of neurons, astrocytic gliosis, and microglial activation which all culminate in complex neurological disorders [2]. Although primary brain damage is caused by mechanical damage during initial impact and is considered irreversible, secondary damage results from delayed neurochemical process and intracellular signaling pathways, which is reversible. Increases in neuronal damage during the chronic phase of TBI have been implicated in secondary brain damage and poor outcome [3,4,5,6].

Apoptosis is defined as programmed cell death mediated by mitochondria in particular, and result in internucleosomal DNA fragmentation, detectable by in situ using the terminal deoxynucleotidyl transferase-mediated dUTP nick end-labeling (TUNEL) [7]. Since apoptotic neuronal death is a mechanism to remove unnecessary neurons with minimal activation of immune system, neuronal apoptosis following a traumatic insult could represent a physiological and protective response to damage. However, transgenic mice overexpressing an anti-apoptotic protein showed a significant reduction in cortical and hippocampal damage following TBI [8, 9]. Therefore, excessive activation of apoptosis-related pathway could be harmful, especially in pathological conditions. Accumulated preclinical studies have revealed that neuronal cell death was observed in the pericontusional region and the hippocampus [10,11,12]. Interestingly, using the moderate lateral fluid-percussion brain injury model, apoptotic cells in the injured cortex was noted at as early as 24 h, whereas in the hippocampus and the thalamus, apoptotic response was delayed, peaking at 48 h and 2 weeks after injury, respectively [13]. In contrast, using controlled cortical impact (CCI) model, neuronal apoptotic cells were the most apparent in the contusional region and hippocampus at between 24 and 48 h after injury [13,14,15]. Furthermore, the functional outcome was correlated with severity of the injury [13,14,15]. Therefore, the type, extent, and spatiotemporal distribution of neuronal apoptosis could be related to injury type and severity (Table 1). This pericontusional and hippocampal apoptosis was also confirmed in post-traumatic human brain tissue [16,17,18,19]. Here we review accumulated evidence for apoptosis following TBI.

Table 1 Spatiotemporal distribution of neuronal cell death in preclinical and clinical study
Full size table

Intrinsic Pathway of Neuronal Apoptosis

Mitochondria trigger a variety of apoptotic signaling pathways, via interactions among the bcl-2 family proteins, such as cytochrome c, apoptosis-inducing factor (AIF), endonuclease G (Endo G), and second mitochondria-derived activator of caspase (Smac), to release pro-apoptotic proteins from the intermembrane of mitochondria, which result in apoptosis [20]. This pathway is called the intrinsic pathway.

Ca2+ Channel–Mediated Apoptosis

TBI induces depletion of energy leading to a loss of membrane potential and depolarization of neurons. Subsequently, voltage-gated Ca2+channels are activated and excitatory amino acids are released into the extracellular space [21]. Intracellular Ca2+ concentration increases and initiates cytoplasmic and nuclear events, involving the intrinsic apoptotic pathway [22]. An immediate response to increased intracellular Ca2+ is the activation of neutral proteases, calpain, which have been implicated in the cleaving of Bcl-2 interaction domain (bid) and its truncated active form (tBID) in ischemic stroke [23, 24]. However, in vitro studies suggest that activation of calpain could be associated with necrosis rather than apoptosis in the setting of TBI, although calpain activation may induce apoptosis to some extent [25, 26]. Elevated intracellular Ca2+ is also associated with activation of the caspase gene family, leading to the induction of apoptosis [27]. Ca2+-dependent endonucleases are also activated following TBI, causing DNA damage in the form of the orderly chromatin cleavage patterns which are typical of apoptosis [28]. However, since therapeutic strategies aimed at inhibiting absolute intracellular Ca2+ elevations, such as voltage-gated calcium channels, were efficient in a clinical setting [28], further research should target intracellular Ca2+ signaling pathways and Ca2+ homeostasis following TBI.

Bcl-2 Family Proteins and Mitochondria

Mitochondria are critical to the apoptotic pathway by releasing pro-apoptotic factors from its intermembrane space into the cytoplasm. Bcl-2 family protein, stored in the mitochondrial intermembrane, is a principal regulator of mitochondrial membrane integrity and function. The Bcl-2 protein family is classified into three groups by structural homology: the anti-apoptotic proteins such as Bcl-2, Bcl-XL, and Bcl-w; the pro-apoptotic proteins such as Bax; and the BH3-only proteins including Bad, Bid, Bim, Noxa, and p53-upregulated modulator of apoptosis (PUMA) [20]. Two main theories have been described to explain Bcl-2 protein family interaction: the “direct model” and the “hierarchy model.” In the former model, anti-apoptotic proteins can inhibit pro-apoptotic proteins. This interaction is disrupted by BH3-only proteins, resulting in activation of pro-apoptotic proteins and apoptosis [29]. In the latter model, BH3-only proteins are divided into two subgroups: activators including Bim, PUMA, and truncated Bid (tBid), and inactivators including the other BH3-only proteins. Activator BH3-only proteins are inhibited by anti-apoptotic proteins. Inactivator BH3-only proteins inhibit this interaction, resulting in activation of activator BH3-only protein leading apoptosis [29]. The Bcl-2 family plays an important role in TBI. BH3-only proteins including Bad [30], Bim [31], Noxa [32], and PUMA [33] contribute to apoptotic cell death after neuronal damage. Increased expression of Bax has been observed in the nucleus of apoptotic cells following experimental TBI, while increased expression of Bcl-2 was observed in the neuron that survives following experimental TBI and in the damaged brain of human [10, 16].

Bcl-2 Family Downstream Interactions

Proteins in the intermembrane space, including cytochrome c [34, 35], second mitochondria-derived activator of caspase (Smac) [36], endonuclease G (Endo G), and AIF [23], are released and interact with each other after TBI. Then their interaction results in the release of the pro-apoptotic proteins [37]. Cytochrome c interacts with procaspase-9 and apoptotic protein-activating factor-1 (apaf-1), and forms the apoptosome, resulting in activation of procaspase-9 [38, 39]. Caspase-9 activates procaspase-3, then caspase-3 degrades inhibitors of caspase-activated DNase, leading to DNA fragmentation and apoptosis. Caspase-3 can also activate other enzymes after TBI which repair damaged DNA, such as poly (ADPribose) polymerase (PARP) [40]. Despite being involved in necrosis and apoptosis, 89- and 21-kDa fragments of PARP are cleaved by caspases and are related to apoptosis after cerebral ischemia [41, 42].

Smac is also involved in caspase activation. After TBI, Smac released from mitochondria binds to and neutralizes the effect of the X chromosome–linked inhibitor-of-apoptosis protein, resulting in further apoptosis [36].

Apoptosis Independent of Caspase

Accumulated studies show the importance of the caspase-independent pathways. Following TBI, apoptosis-inducing factor (AIF), a flavoprotein with NADH oxidase that resides in the mitochondrial intermembrane space, is released into the cytosol via membrane permeabilization. AIF then translocates to the nucleus and induces apoptosis [43]. This apoptotic pathway is also independent of cytochrome c, Apaf-1, and caspases [44, 45]. Apart from these, PARP-1, cyclophilin A, and HSP-70 are involved in regulation of AIF release from mitochondria and translocation to the nucleus. Inhibition of PARP-1 has neuroprotective effects after TBI [46]. cyclophilins are a family of peptidylprolyl cis-trans isomerases [47]. Cyclophilin A participates in the nuclear translocation of AIF from the cytosol to the nucleus and facilitates the chromatolytic effects of AIF [48]. The heat shock proteins of the HSP70 family have neuroprotective effect through their chaperone function [49]. The binding of HSP70 to Apaf-1 and AIF antagonizes their pro-apoptotic effects by inhibiting the formation of the apoptosome and nuclear translocation of AIF [50], respectively. HSP70 overexpression attenuates ischemic brain injury by sequestering AIF [51] and also by inhibiting the caspase-dependent pathway [49]. In contrast to caspase-dependent cell death, AIF-mediated apoptosis can occur under impaired bioenergetics status; this can readily be seen in the core lesion following cerebral ischemia [44]. In fact, in energy-depleted states, the mitochondria are more likely to release AIF, which results in caspase-independent apoptosis [52, 53]. Therefore, AIF-induced apoptosis may be more common in severe TBI where impaired bioenergetic conditions are more likely. Conversely in mild TBI, where mitochondrial energy utilization is less of an issue, caspase-mediated apoptosis is more likely. Endo G is also known to translocate to the nucleus, causing DNA fragmentation in rodent focal cerebral ischemia model, which may be applicable to TBI injury as well [54] (Fig. 1).

Fig. 1
figure 1

Signaling cascade of apoptosis in traumatic brain injury (TBI). Extrinsic pathway (red arrows); following TBI, Fas ligand binds to Fas death receptors (FasR) and triggers the recruitment of the death domain protein (DD), which binds to procaspase-8. This complex, death-inducing signaling complex (DISC), activates caspase-8. Activated caspase-8 either results in the cleaving of Bid to truncated active form of Bid (tBid), which integrates the different death pathways at the mitochondrial checkpoint of apoptosis and activates caspase-3. tBid interacts with Bax, which is inhibited by ant-iapoptotic Bcl-2 family proteins, Bcl-2 or Bcl-xL. Interaction of tBid with Bax leads to the opening of mitochondrial transition pores, subsequently releasing cytochrome c (Cytc) or second mitochondria-derived activator of caspase (Smac). Once released into the cytosol, Cytc binds with apoptotic protein-activating factor-1 (Apaf-1) and procaspase-9 to form an apoptosome, which activates caspase-9, leading to caspase-3-dependent cell death. This pathway also interacts microglia activation. Intrinsic pathway (yellow arrows): TBI elevates cytosolic calcium concentration and activates calpains and mediates cleavage of Bid to Bid. At the mitochondrial intermembrane, tBid interacts with apoptotic proteins and induces caspase-3-dependent pathway. In contrast, apoptosis-inducing factor (AIF) or endonuclease G (Endo G) translocates to the nucleus where it mediates DNA damage and cell death in a caspase-independent manner. In addition, nuclear pathways of neuronal apoptosis are activated in response to DNA damage through activation of p53, which interacts with bcl-2 family proteins and precedes Cytc release. Mitogen-activated protein kinases (MAPK) and Akt pathway are also inactivate Bad or Bim, resulting in apoptosis. Furthermore, TBI generate superoxide anions, which also causes DNA damage. Both extrinsic and intrinsic pathway interact via Bid and cause DNA damage leading to apoptosis

Full size image

Upstream of the Intrinsic Pathway

Akt is a serine-threonine kinase and is involved in the apoptotic signaling pathway as a major downstream target of PI3-K. The phosphorylation of Akt inactivates Bad following TBI in mice, resulting in inactivation of the apoptotic pathway [55]. Akt also phosphorylates procaspase-9 and caspase-9 and inhibits their activity, resulting in the inactivation of apoptotic pathway in an ischemic stroke model as well [56].

Following TBI in a rodent model, significant mitogen-activated protein kinase (MAPK) is observed which results in the inhibition of pro-apoptotic proteins, such as Bad and Bim, and subsequent reduction in contusion volume and apoptosis [57]. In addition, since a number of Bcl-2 family proteins are regulated by p53, a well-known tumor suppressor and transcription factor induced by cellular damage, it is not surprising that p53 can also activate a number of intrinsic apoptotic pathways. In rodent models of global cerebral ischemia, p53 translocates to mitochondria, interacts with Bcl-XL, and induces caspase-dependent apoptosis [32].

Extrinsic Pathway of Neuronal Apoptosis

Extrinsic mechanisms of apoptosis involve cell surface receptors present on multiple cell types, including neurons, and is also referred to as the “death receptor pathway.” The binding of extracellular tumor necrosis factor (TNF) to TNF receptor or extracellular FasL to the Fas receptor triggers recruitment of the cytoplasmic death domain protein (DD), resulting in the death-inducing signaling complex (DISC) [58]. Interestingly, TNF α and Fas knockout mice have significantly smaller brain lesion and favorable memory performance after controlled cortical impact (CCI). These findings emphasize the contribution of death receptors in cell death after TBI [59].

Cross-talk Between Intrinsic and Extrinsic Pathway

The extrinsic pathway receives extracellular signals and transduces them to intracellular signals, and the Fas pathway is one of those extracellular signals. Both Fas and Fas ligand (FasL) proteins are upregulated after brain injury [60, 61]. Fas, the death domain (DD), and procaspase-8 form a protein complex that is referred to as the DISC. DISC activates procaspase-8, similar to procaspase-9 activation by the apoptosome. Caspase-8 activation is followed by activation of caspase-3 and caspase-10 after cerebral ischemia [61]. Bid is a key molecule for cross-talk between extrinsic and intrinsic pathway, which is truncated by caspase-8, translocates to mitochondria, interacts with other Bcl-2 family proteins, and cytochrome c release followed by apoptotic cell death [62].

Apoptosis and Oxidative Stress

It has been thought that mitochondria are the primary source of reactive oxygen species (ROS) involved in ischemia-induced apoptosis, which can be seen in impaired energy metabolism after severe TBI. Previous studies have demonstrated that a variety of stimuli, including hypoxia, excitotoxicity, and Ca2+ influx after cerebral ischemia, produce mitochondrial ROS [63]. ROS is a trigger of the release of cytochrome c and other pro-apoptotic proteins from the mitochondria into the neuronal cytosol, leading to apoptosis and defective gene expression after stroke [35, 64]. Thus, increasing evidence suggests that oxidative stress and apoptosis are closely linked phenomena in the pathophysiology of ischemic stroke.

Neuronal Apoptosis Driven by the Other Cell Types

Microglia is the primary immune cell in the CNS and maintains CNS homeostasis by engulfing and clearing neurons that die as a result of apoptosis; they also participate in brain development and neuronal plasticity [65,66,67]. After TBI, microglia drive neuronal apoptosis via the release of superoxide anion [68], nerve growth factor [69], and tumor necrosis factor (TNF) [70]. However, the role of microglia in TBI remains to be elucidated. Wang et al. demonstrated that microglial depletion significantly reduced neuronal apoptosis following moderate fluid percussion injury in rodents [71]. On the other hand, Bennett et al. [72] demonstrated that inhibition of microglia did not contribute to acute axon degeneration after multiple concussive injury. Hanlon et al. [73] demonstrated that microglia depletion with clodronate caused an increase in neurodegeneration in a rat CC model, possibly due to a decrease in the clearance of dying cells. Therefore, the impact of the microglia may vary depending on the type of injured cells and TBI models. Astrocytes are highly connected via gap junctions that allow for intercellular exchange of metabolites, ions, and small molecules. Lin et al. [74] demonstrated that gap junctions propagate intercellular signals, which cause cell injury induced by intracellular calcium overload, oxidative stress, and metabolic impairment in cerebral ischemia. Connexin 43 (Cx43) is the mostly studied gap junction protein highly expressed on astrocytes [75, 76]. In vivo and in vitro study suggested that remote hippocampal apoptosis was associated with spread of apoptotic signal through the astrocytic Cx43 hemi-channels in rodent TBI model [75,76,77]. Surprisingly, heterozygotic Cx43 null mice, which express 50% of the normal Cx43, showed large lesions compared with wild-type mice in focal cerebral ischemia in rodents [78]. Therefore, implication of Cx43 in neuronal damage spread has remained unknown.

Apoptosis in the Chronic Phase of TBI

There are many common features among neurodegenerative diseases, including inflammatory and neurovascular pathologies associated with irregular accumulation of tau protein [79], blood-brain barrier permeability [80], abnormal angiogenesis, and apoptosis [81]. Although limited evidence exists, the chronic sequelae of TBI might also share these pathologies [82]. Furthermore, there is a pathological link between TBI and other neurodegenerative diseases, such as Alzheimer’s and Parkinson’s diseases. All the aforementioned diseases have some abnormal proteinaceous accumulation, whether it be amyloid β, tau protein, or both, suggesting a shared spectrum between them [83]. Previous preclinical studies in rodents exposed to CCI demonstrate neuronal degeneration and total brain atrophy that occurs over at least 1 year [81, 84]. Another study demonstrated upregulation of cleaved-caspase-3 in the white matter of the corpus callosum 3 months after CCI in rats, suggesting an association between chronic neurodegeneration and apoptotic cell death [85]. Furthermore, in CCI-exposed rats, caspase-3 was co-localized with tau accumulation, indicating a possible correlation between apoptosis and tau during the chronic phase of TBI [86]. Further study may allow us to understand association of apoptosis with the chronic sequelae of TBI.

Apoptosis and Mild TBI

Emerging evidence suggests that repeated mild TBI may have a great cumulative effect on brain functions, leading to chronic persistent neurological symptoms due to the neurodegeneration [87,88,89]. In mild TBI, cellular metabolism is disturbed and microstructural damage occurs and results in both biochemical and vascular autoregulation abnormalities [90]. Although cell death is not typically observed acutely after a single mild TBI, some studies indicated a potential for neuronal cell death through apoptosis in repeated mild TBI [91, 92]. Microarray analysis in rodent brains exposed to repeated mild TBI at 17 weeks showed significantly altered the expression level of 87 genes which are involved in apoptosis, stress response, metabolism, and synaptic plasticity, suggesting an association with chronic neuronal degeneration [89]. Furthermore, repeated, mild TBI led to a significant reduction in neuronal cells and substantially increased microglial activation in ipsilateral and contralateral hippocampi at 28 days after fluid percussion injury in rats [93]. Thus, sustained activation of microglia may also contribute to the chronic spread of cell damage via apoptosis.

Apoptotic Biomarker in Serum and/or Cerebrospinal Fluid in Human

Several clinical studies have investigated apoptotic biomarkers in serum and/or cerebrospinal fluid (CSF) following TBI with different severities and time points (Table 2). Increase in the level of caspase-3 and caspase-9 is indicative of involvement of apoptotic pathway [94,95,96,97]. Levels of caspase-specific cleavage products such as caspase-cleaved cytokeratin-18 (CCCK-18), caspase-cleaved tau, and caspase-specific spectrin breakdown products (SBDP) 120 are also suggestive of cellular involvement of particular cell types and demonstrate injury mechanisms [98,99,100,101].

Table 2 Expressed apoptotic biomarkers in human serum and/or cerebrospinal fluid following traumatic brain injuries
Full size table

Apoptosis Following TBI as Therapeutic Targets

In the central nervous system, apoptosis is a mechanism by which the neurons that have not formed functional synaptic connections are removed. In addition, during development, apoptosis contributes to neuronal plasticity [102]. One of the features of apoptosis is minimal activation of the immune system over the course of cell death, and thus the surrounding cells are able to remain relatively unaffected. Therefore, neuronal apoptosis seems to be a reasonable mechanism to limit collateral damage following ischemic or traumatic brain injury. In pathological conditions, treatment with tetrapeptide caspase-3 inhibitor (z-DEVD-fmk) at 24 h or overexpression of the anti-apoptotic protein bcl-2 significantly reduced brain damage following CCI in rodents, although motor and cognitive deficits were not improved [9]. However, pretreatment with the caspase-3 inhibitor improved neurological outcome as well as reduced lesion size following FPI in rats [8, 103, 104]. Similarly, treatment with a pan-caspase inhibitor (z-VAD-fmk) after FPI improved functional outcome in rats by attenuating mitochondrial release of cytochrome c [105, 106]. Moreover, depletion of PARP and early treatment with inhibitors of PARP at 24 h after TBI also improved functional outcome by limiting caspase-independent pathway mediated by AIF following CCI in rodents [46, 107,108,109]. Therefore, anti-apoptotic treatment targeting multiple apoptotic pathways may have to be applied within appropriate therapeutic time window.

Furthermore, several anti-apoptotic drugs with multifunctional activities were evaluated in clinical studies [110]. Statins exert anti-inflammatory effects by limiting the production of inflammatory mediators, glial cell activation, and cerebral edema in the rodent TBI model [111,112,113]. They also decrease apoptosis after trauma and effectively alter the ratio of anti-apoptotic to apoptotic factors and improved the cognitive function [114]. A small, prospective, randomized, double-blind clinical trial showed a slight improvement in amnesia and disability following moderate to severe TBI [115, 116]. However, another prospective clinical study could not show clinical benefit to statin use at the time of moderate or severe TBI [117].

Progesterone attenuated glutamate excitotoxicity [118], membrane lipid peroxidation [119], and inflammation, and also reduced cognitive impairment and neuronal cell death after TBI in preclinical models [120,121,122]. Two phase 2 randomized controlled clinical trial also improved clinical outcome of moderate to severe TBI patients [123, 124]. However, despite the strong preclinical and early clinical data, a multinational placebo-controlled trial failed to show clinical benefit of progesterone following TBI [125].

Another potentially useful agent in TBI could be cyclosporine A, an immunosuppressive agent, which attenuates mitochondrial failure by binding cyclophilin D and stabilizing mitochondrial permeability [126]. Treatment with cyclosporine A reduced lesion size and improved outcome by preventing the apoptotic cascade–induced mitochondrial dysfunction, which was independent of its immunosuppressive function in a rodent TBI model [48, 127]. A prospective randomized, controlled, double-blinded study of cyclosporine A for severe TBI patients showed a decrease in the lactate to pyruvate ratio via cerebral microdialysis [128], suggesting a decrease in anaerobic metabolism. Since this agent has several disadvantages including biphasic drug response, shows poor BBB permeability, and prolonged use results in chronic immunosuppression in phase I clinical trials [129], further evaluation of this drug is necessary. Phase III clinical trials of cyclosporine A for TBI are in preparation [130]. Clinical trials of anti-apoptotic drugs for human TBI are summarized in Table 3.

Table 3 Clinical trial of anti-apoptic drugs for traumatic brain injury
Full size table

TBI involves a highly complex pathology characterized by multiple interacting secondary injury cascades. Therefore, using multiple models and species for preclinical screening should be attempted before clinical translation. Bidirectional translational research between preclinical and clinical investigators is warranted to identify novel signaling pathways, target, and modify them, in the hope of identifying pharmaceuticals that may one day improve TBI outcome.