Mitochondrial Mechanisms of Neuronal Cell Death: Potential Therapeutics

Abstract

Mitochondria lie at the crossroads of neuronal survival and cell death. They play important roles in cellular bioenergetics, control intracellular Ca²⁺ homeostasis, and participate in key metabolic pathways. Mutations in genes involved in mitochondrial quality control cause a myriad of neurodegenerative diseases. Mitochondria have evolved strategies to kill cells when they are not able to continue their vital functions. This review provides an overview of the role of mitochondria in neurologic disease and the cell death pathways that are mediated through mitochondria, including their role in accidental cell death, the regulated cell death pathways of apoptosis and parthanatos, and programmed cell death. It details the current state of parthanatic cell death and discusses potential therapeutic strategies targeting initiators and effectors of mitochondrial-mediated cell death in neurologic disorders.

INTRODUCTION

Neurologic diseases are a leading cause of disability and death in the developed world. The economic costs are correspondingly high. Alzheimer’s Disease International indicates that “for 2010 the global cost of dementia … [was] $604 billion – about 1% of world gross domestic product” (1). Other brain disorders have similarly high economic costs. In the United States in 2008, the cost of stroke was estimated to be $65.5 billion (2), and the cost of Parkinson’s disease (PD) was estimated to be $14.4 billion (3). Thus, the global burden of neurologic diseases is huge. Accordingly, preventing and treating neurologic disorders is of paramount importance, as it will improve the lives of many people in a myriad of ways.

Many neurologic disorders including stroke, Alzheimer’s disease (AD), PD, multiple sclerosis (MS), and others are characterized by the death of neurons. Understanding and interfering with these neuronal cell death pathways will have a huge impact on the quality of life of patients who suffer from these disorders. Unfortunately, the underlying mechanisms of neuronal damage that occur in neurodegenerative diseases, stroke, and other neurologic disorders are not fully understood. Mitochondria orchestrate neuronal cell death and survival. This review provides an overview of mitochondrial-mediated mechanisms of neuronal cell death and strategies for treatment of acute and chronic neurologic diseases.

MITOCHONDRIA

Mitochondria are intracellular organelles composed of two bilayers containing lipid and protein: the external outer mitochondrial membrane and the inner mitochondrial membrane. These two membranes enclose the intermembrane space and matrix. Mitochondria play critical roles as the energy generators of the cell through oxidative phosphorylation and in the electron transfer in the electron transport chain through a series of electron carriers in the inner mitochondrial membrane leading to the production of ATP (4) (Figure 1). In addition to their important role in cellular bioenergetics, mitochondria participate in the synthesis of iron-sulfur clusters (5, 6), and they control intracellular Ca²⁺ homeostasis. Mitochondria also participate in the Krebs cycle and fatty acid metabolism via enzymes in the matrix (7). Reactive oxygen species (ROS) including the hydroxyl radical (^•OH) and the superoxide anion (O₂^•−) are produced via mitochondrial bioenergetics. Mitochondrial-generated O₂^•− is detoxified by manganese superoxide dismutase and generates hydrogen peroxide (H₂O₂) (Figure 1). Glutathione peroxidase detoxifies H₂O₂. Failure to detoxify ROS can cause cellular injury.

Figure 1.

Mitochondria play vital cellular functions including oxidative phosphorylation through the Krebs cycle and the electron transport chain. Abbreviations: ADP, adenosine diphosphate; ANT, adenine nucleotide transporter; ATP, adenosine triphosphate; CoQ10, coenzyme Q 10; CytC, cytochrome C; G-6-P, glucose-6-phosphate; ETC, electron transport chain; GSH, glutathione; H₂O₂, hydrogen peroxide; mV, millivolt; ΔΨ, mitochondrial membrane potential; NAD⁺, nicotinamide adenine dinucleotide; NADP⁺, nicotinamide adenine dinucleotide phosphate; O₂, oxygen; P_i, phosphate; PIC, phosphate channel; O₂^•−, superoxide anion; SOD2, superoxide dismutase 2; TRX, thioredoxin; I, complex I; II, complex II; III, complex III; IV, complex IV.

Defects in mitochondrial respiration have been implicated in many neurodegenerative diseases (8). Although defects in oxidative phosphorylation and the accompanying ROS generation contribute to neuronal cell death and neurodegenerative diseases, these defects may not be the primary driver of pathology. Instead, defects in oxidative phosphorylation and the accompanying ROS are downstream consequences of primary defects that set in motion mitochondrial dysfunction that ultimately contributes to neurodegeneration.

Mitochondrial Quality Control

To maintain oxidative phosphorylation and other important mitochondrial functions, quality control systems maintain the integrity of mitochondria (9). Mitochondria are in a constant state of flux to meet cellular metabolic demands. Fusion and fission dictate the size of mitochondria. Damaged mitochondria must be identified, segregated, and eliminated, and new mitochondria must be synthesized (Figure 2). In the nervous system, owing to long cellular processes, active axonal and dendritic transport mechanisms also participate in the quality control (Figure 2). Damaged mitochondria are segregated through dynamin-related protein (DRP1)-mediated fission and eliminated through autophagy (mitophagy). Mitochondria also combine through fusion to maintain their intrinsic functions. A fastidious quality control system is important because defective mitochondria would lead to calcium dysregulation, loss of energy, oxidative damage, and cell death from activation of intrinsic mitochondrial cell death machinery (9).

Figure 2.

The life cycle of mitochondria. Mitochondrial quality control is a tightly controlled, multistep process to meet the metabolic demands of the cell. ❶ Biogenesis coordinates the synthesis of new mitochondria. ❷ Fusion leads to joining of mitochondria ❸ whereas fission functions in part to segregate damaged mitochondria, which are engulfed by the autophagosome and degraded via ❹ mitophagy. In the nervous system, mitochondria are ❺ transported to the distal ends of axons and dendrites. Abbreviations: mtDNA, mitochondrial DNA; OXPHO, oxidative phosphorylation.

Mutations in key regulators of mitochondrial quality controls cause neurodegenerative diseases. Mutations in PTEN-induced putative kinase 1 (PINK1) and parkin, which play roles in all aspects of mitochondrial control, cause autosomal recessive PD (10–12). Failure to eliminate damaged, ROS-generating mitochondria through mitophagy causes inflammasome activation, which can be further exacerbated by defects in mitochondrial biogenesis (9, 13, 14). Mutations in the mitochondrial DNA polymerase γ, which lead to defects in mitochondrial DNA replication, can contribute to some cases of PD. Charcot-Marie-Tooth disease type 2A is caused by mutations in mitofusin 2 (MFN2), and mutations in optic atrophy 1 and 3 (OPA1 and 3) cause optic atrophy and are linked to defects in mitochondrial fusion (15). MFN1/2 and OPA1 are dynamin-like GTPases localized to the outer and inner mitochondrial membranes, where they coordinate fusion. Mutations in DRP1, another dynamin-like GTPase, cause severe infantile neurodegeneration (15). Other mutations in mitochondrial genes or regulators of mitochondrial function are linked to neurologic disorders (for reviews, see 8, 15).

Mitochondria and Cell Death

Owing to the essential functions of mitochondria in energy bioenergetics and other important and critical metabolic functions, mitochondria have evolved strategies to kill cells that are not able to continue their vital functions. They use different effectors to kill cells. Several cell death pathways have been identified and are segregated into accidental, regulated, and programmed cell death (16). Accidental cell death, as described by Galluzzi et al. (16), is uncontrolled cell death due to extreme physical, chemical or mechanical stimuli and manifests morphologically with necrotic features. Regulated cell death can be influenced, at least to some extent, by specific pharmacologic or genetic interventions and exhibits both necrotic and apoptotic features. Programmed cell death is a regulated cell death program that can occur as part of a developmental program or to preserve physiologic adult tissue homeostasis and is apoptotic morphologically (16). Mitochondria play roles in all three forms of cell death. In accidental cell death, intracellular calcium levels increase owing to complete disruption of mitochondrial function and energy failure. Ultimately, calcium-dependent catabolic enzyme activation is the coup de grâce.

In regulated and programmed cell death, mitochondria contribute to these processes by the release of proteins that activate intrinsic cell death programs. The morphologic classification of cell death (apoptosis versus necrosis) has been replaced by quantifiable biochemical assessments (17). Mitochondria play important roles in intrinsic apoptosis and parthanatos and less important roles in other forms of regulated cell death (16). For instance, mitochondria are not required for necroptosis (18). The final step in all forms of cell death is the inability to meet the energy and metabolic demands of the cell. For further insight into nonmitochondrial-mediated and/or-initiated forms of cell death, the reader is referred to Reference 16.

Intrinsic apoptosis is initiated by a variety of stressors that lead to activation of the initiator caspase 9 followed by proteolytic maturation of the executioner caspase 3 (Figure 3). This process is set in motion by mitochondrial outer membrane permeability (MOMP) in a B cell lymphoma 2 (Bcl-2)-associated X protein (BAX)- and Bcl-2 homologous antagonist/killer (BAK)-dependent manner, which releases the intrinsic intermembrane space mitochondrial protein cytochrome C (CytC). In combination with the apoptotic peptidase-activating factor 1 and deoxy-ATP, CytC forms the apoptosome that activates the initiator caspase 9 (19–21) followed by activation of caspase 3 and downstream processes such as activation of DNAases and substrate proteins that orchestrate the apoptotic cell death program (16) (Figure 3). Activation of apoptosis stimulating fragment or tumor necrosis factor receptors or other death receptors can lead to mitochondrial-mediated cell death through cleavage of caspase 8 and activation of BID [BH3 (Bcl-2 homology 3) interacting-domain death agonist], which causes MOMP (22). The cellular signals regulating both intrinsic and extrinsic apoptosis have been reviewed extensively (see 23–25).

Figure 3.

Mitochondrial-regulated cell death. Key players in apoptosis (left) and parthanatos (right) are highlighted. Abbreviations: AIF, apoptosis-inducing factor; AIMP2, aminoacyl-tRNA synthetase complex interacting multifunctional protein-2; AKT, protein kinase B; ANT, adenine nucleotide transporter; APAF1, apoptotic peptidase-activating factor 1; ARH3, ADP-ribosyl-acceptor hydrolase 3; ATP, adenosine triphosphate; BAK, Bcl-2 homologous antagonist/killer; BAX, Bcl-2-associated X; Bcl-2, B cell lymphoma 2; BID, BH3 (Bcl-2 homology 3) interacting-domain death agonist; CyPD, cyclophilin D; CytC, cytochrome C; Glc, glucose; G-6-P, glucose-6-phosphate; HK, hexokinase; Ψ, mitochondrial membrane potential; MOMP, mitochondrial outer membrane permeability; MPT, mitochondrial permeability transition; NAD⁺, nicotinamide adenine dinucleotide; PAAN, parthanatos AIF-associated nuclease; PAR, poly (ADP-ribose); PARG, PAR glycohydrolase; PARP, PAR polymerase; PI3K, phosphatidylinositol-3-kinase; p53, tumor suppressor p53; tBID, truncated BID.

Oxidative stress– or calcium overload–mediated (26) mitochondrial membrane permeabilization leads to permeability transition pore complex (PTPC) opening, which causes mitochondrial permeability transition (MPT). Mitochondria can accumulate large amounts of calcium through the calcium uniporter under pathologic conditions in which intracellular calcium concentrations increase (27). Osmotic swelling and rupture of the mitochondrial membranes and release of mitochondrial intermembrane proteins such as CytC, apoptosis-inducing factor (AIF), and the matrix protein endonuclease G occur after complete MPT (26). Cell death induced by MPT can occur during oxidative stress or ischemia-reperfusion injury, and unlike MOMP, MPT killing of cells does not require caspase activation (28). MPT can switch from low- to high-conductance states (29), and prolonged high conductance leads to mitochondrial collapse (30). Gradations of MPT can occur because mild MPT can be reversed by inhibitors of cyclophilin D (CyPD) (31). When a small fraction of mitochondria undergo MPT, they can be eliminated by mitophagy.

Although the exact molecular composition of the PTPC is not known, researchers do know it is a supramolecular multiprotein complex spanning the outer and inner mitochondrial membranes (Figure 3). It is composed, in part, of the integral protein of the inner mitochondrial membrane, the adenine nucleotide translocase (ANT), the outer mitochondrial membrane protein, the voltage-dependent anion channel (VDAC), and the matrix protein CyPD. The mitochondrial ATP synthase, another multiprotein complex, also known as F₁F₀-ATP synthase, that is responsible for the synthesis of ATP from dissipation of the chemiosmotic gradient (32), is emerging as a key component of the PTPC (33, 34). In addition, the PTPC also contains hexokinase (HK), creatine kinase, and several other modulatory enzymes (35). The rate-limiting step of glycolysis is controlled by HK, which in an ATP-dependent manner converts glucose to glucose-6-phosphate. HK’s tight association with VDAC provides ready access to mitochondrial-generated ATP. Disruption of the interaction of HK with VDAC can induce MPT (36). Only CyPD seems to be critical for MPT-mediated cell death, as loss of CyPD attenuates MPT-induced cell death (28), whereas ANT and VDAC are dispensable for MPT-induced cell death (26). Ultimately, complete MPT leads to cell death through energy collapse.

STROKE, NEURODEGENERATION, AND NEUROLOGIC DISORDERS

Disease-modifying therapies for neurodegenerative disorders do not exist or have very minimal benefits. Currently, no therapeutic strategies are available for treating neurologic injury due to stroke and other neurologic disorders. Thus, a clearer understanding of the pathologic processes that contribute to cell death in stroke, neurodegenerative diseases such as AD and PD, and neurologic disorders such as MS is essential. Inroads into developing effective disease-modifying therapies for neurologic disorders rest on gaining a detailed molecular understanding of the mechanisms of neuronal cell death. Having laid the foundation for why mitochondria play an important regulatory and contributory role to cell death, the remainder of this review highlights how neuronal cell death is orchestrated by this organelle. Cell death due to stroke is highlighted, as stroke is the leading neurologic cause of death and disability, and the basic molecular mechanisms of cell death that play important roles in neurologic injury due to stroke are likely to extrapolate to other neurologic disorders

Stroke

Ischemic neuronal injury that occurs following loss of oxygen and glucose to the brain leads initially to accidental cell death. Secondary pathologic processes follow the initial insult (Figure 4). Although prevention is paramount to treating stroke, once it occurs, both the control of these secondary processes and the restoration of blood supply are essential to limit ischemic neuronal damage. Advances in restoration of blood flow following stroke, including tissue plasminogen activator therapy (37) and mechanical thrombectomy (38), have led to effective therapies to restore blood flow, limiting neuronal and glial damage, which has changed medical practice. Despite the advances in restoring blood flow, substantial neuronal, glial, and neurovascular damage still occurs, particularly due to reperfusion injury of the penumbra (compromised but not necrotic brain tissue).

Figure 4.

Mechanisms of neuronal cell death during stroke. Neuronal cell death occurs in a four-step process involving initiation due to loss of cerebral blood flow; activation due to opening of a variety of calcium channels; atrophication, which involves the activation of calcium-dependent enzymes and oxidative and nitrosative stress, which leads to parthanatos; and, ultimately, death through large-scale DNA fragmentation and caspase activation. Tissue damage continues and is amplified through inflammatory mediators. Other causes of neuronal cell death may occur through similar mechanisms. Abbreviations: AIF, apoptosis-inducing factor; ASIC, acid-sensitive ion channel; ATP, adenosine triphosphate; [Ca²⁺]_i, intracellular calcium; CBF, cerebral blood flow; DNA, deoxyribonucleic acid; HK, hexokinase; iNOS, inducible nitric oxide synthase; [K⁺]_e, extracellular potassium; NAD⁺, nicotinamide adenine dinucleotide; NMDA, N-methyl-d-aspartate; NO, nitric oxide; nNOS, neuronal NO synthase; O₂, oxygen; O₂^•−, superoxide anion; ONOO⁻, peroxynitrite; PAAN, parthanatos AIF-associated nuclease; PAR, poly (ADP-ribose); PARP-1, PAR polymerase-1; TRPC, transient receptor potential channel; VSCC, voltage-sensitive cation channel.

Researchers have proposed many pathways that may contribute to cell death in the penumbra following stroke, but the more extensively investigated mediators include glutamate excitotoxicity, which drives the increases in intracellular calcium, production of oxygen free radicals, and nitric oxide (NO) that set cellular death mechanisms in motion (39–42). In addition, other routes of calcium entry through acid-sensing channels or transient receptor potential channels play important roles in neuronal injury (for reviews, see 43–45). Inappropriate activation of other transmitter second-messenger pathways owing to the initial ischemic insult may also contribute to the neuronal injury (39–42, 46). Accompanying the neuronal damage is activation of neuroinflammatory pathways, which contribute to damage and late regenerative processes (47).

Excitotoxicity

Glutamate excitotoxicity plays a role in ischemia/stroke, seizures, Huntington’s disease (HD), PD, AD, amyotrophic lateral sclerosis, hepatic encephalopathy, MS, traumatic brain injury, and metabolic disorders of the brain (39–41, 48–51). Thus, understanding the molecular mechanisms by which glutamate excitotoxicity kills neurons has relevance to many disorders of the nervous system.

Glutamate binds to four major types of receptors: N-methyl-d-aspartate (NMDA) receptors, α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors, metabotropic receptors, and kainate receptors, all of which transduce glutamate’s action in the nervous system (52). Ischemia-induced glutamate excitotoxicity and neuronal damage is mediated in large part through NMDA receptor activation (39, 41, 50). Evidence in support of targeting glutamate excitotoxicity in stroke comes from recent work with postsynaptic density-95 (PSD-95) inhibitors, which uncouple the NMDA receptor from its signaling cascade. PSD-95 inhibitors show unequivocal protection in rodent brains (53–56), high-order brains of old-world primates (57, 58), and humans (59). These recent exciting results have reinvigorated the search for novel neuroprotectants to treat acute ischemic stroke in humans (60–62). In addition, there is tremendous interest in interfering with nonglutamate receptor–mediated calcium entry through blockade of acid-sensing channels (63) and transient receptor potential channels (64), particularly because these channels seem to play a role in the later stages of calcium entry following ischemic injury in the brain.

Nitric Oxide and Poly (ADP-ribose) Polymerase-1

Glutamate excitotoxicity is mediated and initiated predominantly by glutamate activation of NMDA receptors, intracellular calcium flux, NO, and free radicals (Figure 4) (65–67). Neuronal cell death due to a short pulse (5 min) of glutamate receptor activation causes cell death 24 h later (68). This form of cell death, called delayed or rapidly triggered neurotoxicity, is dependent on NMDA receptor activation, calcium influx, and NO (65–67). Extrasynaptic NMDA receptors account for most of the toxicity associated with NMDA receptor activation, whereas synaptic NMDA receptor activation leads to activation of cell survival pathways (69).

Treatment with NO synthase (NOS) inhibitors, disruption of the interaction between the NMDA receptor and neuronal NOS (nNOS), or gene deletion of nNOS prevent cell death in cell culture, slice models, and in vivo models of stroke (53, 66, 67, 70–74). Overproduction of NO generated by NOS leads to the generation of peroxynitrite (ONOO⁻) (75) through the reaction of mitochondrial-generated O₂^•− with NO (76). ONOO⁻ modification of DNA and the subsequent DNA nicks trigger the next step in the excitotoxic death program by activating the DNA damage-sensing enzyme poly (ADP-ribose) (PAR) polymerase-1 (PARP-1) (77–79). Cell demise from PARP-1 overactivation has been attributed to depletion of cellular energy; release of the mitochondrial death effector AIF from mitochondria; and production of excess PAR polymer, a novel death signal (46, 80–88).

Apoptosis

Glutamate excitotoxicity can induce apoptosis by increasing intracellular calcium and mitochondrial calcium overload and triggering MOMP or MPT (89). Oxidative stress can also lead to apoptosis through MOMP or MPT. MOMP leads to CytC release and activation of intrinsic apoptosis. Inflammation can induce both intrinsic and extrinsic apoptosis (90). Many studies in experimental animal and culture models of neurodegenerative and neurologic disorders, as well as human postmortem tissue examination, have revealed caspase activation and markers of apoptosis, raising the possibility that interference with both intrinsic and extrinsic apoptotic cascades could be neuroprotective in neurologic disorders (for reviews, see 91–93). Numerous other stressors or signals can lead to apoptosis in neurons. For instance, in response to DNA damage, tumor suppressor p53 inhibits the antiapoptotic protein Bcl-2 (94). The lack of neurotrophic support, which promotes cell survival through kinase signaling cascades such as the phosphatidylinositol-3-kinase and protein kinase B pathways, is the classic neuronal trigger of apoptosis, which plays important roles during neuronal development (93). Lack of neurotrophic support may also play a role in some neurodegenerative and neurologic disorders. Although apoptotic mechanisms have dominated the discourse on cell death, most of the cell death that occurs during stroke is nonapoptotic (95).

Parthanatos

Because apoptosis may not be the major form of cell death in stroke, what other mechanisms might be at play in the death of neurons due to stroke? The term parthanatos was coined from Thanatos, the personification of death in Greek mythology, to describe cell death initiated by the PAR polymer (Figure 2) (16, 17, 46, 80, 84, 88). Parthanatos is distinct biochemically from apoptosis and necrosis. Biochemically, it is characterized by rapid PARP-1 activation followed by PAR accumulation, early NAD⁺ depletion, and mitochondrial AIF translocation to the nucleus. Ultimately, late caspase activation and loss of cellular ATP occur (46, 80, 81, 84–88). In stroke, both PARP-1 and PARP-2 contribute to injury via AIF (96).

DNA damage is the traditional activator of PARP-1. DNA breaks are detected by PARP-1 via its zinc-finger domain and are followed by synthesis of PAR, which acts as a signaling scaffold and participates in the orchestration of the DNA repair response (97, 98). Other PARPs, including PARP-2 and PARP-3, contribute to DNA repair (97, 98). Excessive activation of the PARPs leads to parthanatos. Mitochondria play a key role in parthanatos, as mitochondrial-generated ROS due to mitochondrial calcium uptake are required for PARP-1 activation (99) and mitochondria house the death effector AIF (100). Bursts of O₂^•− production that occur during transient MPTs (101) could contribute to PARP-1 activation through formation of ONOO⁻ (102). Other mechanisms of activation of PARP-1 may not require DNA damage and expand the ways in which parthanatos can be initiated. For instance, in PD, aminoacyl-tRNA synthetase complex interacting multifunctional protein-2 activates PARP-1 directly and contributes to the degeneration of dopamine neurons (103). Glyceraldehyde-3-phosphate dehydrogenase can promote PARP-1 activation in the setting of oxidative and nitrosative stress (104).

Shrunken and condensed nuclei are the hallmark morphologic features of parthanatos. Shortly after activation of PARP-1 in parthanatos, cells become propidium iodide positive. Caspase inhibitors cannot prevent parthanatos (46, 81, 85, 87, 88). Experimental studies indicate that over-activity of parthanatos plays a prominent role in neuronal injury in stroke (105–107). Parthanatos also plays an important role in cell damage following trauma, ischemia-reperfusion of the retina, PD, AD, HD, and the experimental autoimmune encephalomyelitis model of MS (for reviews, see 42, 46, 83, 108).

Once AIF enters the nucleus, chromatinolysis and large-scale DNA degradation happens, followed by cell death. The identity of the AIF-associated nuclease has remained elusive. Recently the macrophage migration inhibitory factor (MIF) was identified as a PARP-1–dependent AIF-associated nuclease (PAAN) (109). Molecular modeling indicates that the MIF trimer possesses the same topologic structure as the PD-D/E(X)K nuclease superfamily, in which it has both 3′ exonuclease and endonuclease activity (110, 111). MIF was shown to be a Mg²⁺/Ca²⁺-dependent nuclease that is carried into the nucleus by AIF after PARP-1 activation. Prevention of the translocation of MIF from the cytosol to the nucleus by disruption of the AIF and MIF protein-protein interaction protects against glutamate excitotoxicity and stroke. Moreover, knockout of MIF and a nuclease–deficient MIF prevents glutamate excitotoxicity and stroke to an extent similar to that afforded by deletion or inhibition of PARP-1. Identification of MIF as the elusive PAAN opens up a new therapeutic avenue to explore as a potential disease-modifying therapy in neurologic disorders. Inhibition of MIF’s nuclease activity may have substantial advantages over inhibition of PARP, particularly in chronic neurodegenerative diseases, as it would avoid impairment of the DNA repair response by inhibition of PARP.

AIF binding to PAR is required for parthanatos both in vitro and in vivo (85). AIF’s interaction with the outer mitochondrial membrane is disrupted by PAR binding, which leads to AIF release from mitochondria. The PAR binding site on AIF is distinct from its DNA binding site. This PAR-dependent release of AIF occurs in a calpain-independent manner (101). An AIF PAR binding mutant (R588A, K589A, R592A) cannot be released from mitochondria and prevents parthanatos despite being fully capable of AIF’s other functions, including FAD and DNA binding, NADH oxidase activity, and AIF-mediated nuclear condensation. Thus, PAR polymer binding to AIF is required for execution and provides a mechanism for AIF-mediated cell death. The PAR-dependent releasable pool of AIF is on the outer membrane of mitochondria (112). This mitochondrial outer membrane localization enables AIF to bind PAR and accounts for AIF’s release prior to the mitochondrial membrane depolarization (85), which explains the later release of CytC in parthanatic cell death after MPT occurs (88). Calpain- and BAX-dependent mechanisms of mitochondrial AIF release have also been described (113).

One of the remaining mysteries in parthanatos is how the highly negatively charged PAR leaves the nucleus to translocate to mitochondria to induce the release of AIF. Multiple routes of delivery of PAR from the nucleus to mitochondria likely exist, including PAR-binding proteins such as histone 1.2, which is known to translocate from the nucleus to mitochondria following cellular injury (114). In addition, free PAR polymer, which is released from PARylated proteins by the action of PAR glycohydrolase (PARG), contributes to mitochondrial AIF release (115). This may be the main route of cytosolic PAR, as ADP-ribosyl-acceptor hydrolase 3 cleavage of PAR protects against parthanatos by lowering PAR levels in the cytoplasm that are created through the action of PARG (115). Future studies will be required to determine whether a PAR binding protein or PARylated protein such as histone 1.2 plays a role in carrying PAR out of the nucleus to mitochondria.

Like in other cell death pathways, there are endogenous inhibitors of parthanatos. PARG is an enzyme that removes PAR from PARylated proteins, terminating the actions of PARP modification of proteins in much the same way that phosphatases terminate the action of kinases (116). Overexpressing PARG reduces cell death, and lowering PARG levels exacerbates cell death induced by activators of PARP-1, including glutamate excitotoxicity and stroke, consistent with the notion that PAR is a cell death messenger (81). In addition, failure to degrade PAR leads to early embryonic lethality and increased sensitivity to cytotoxicity (117). The majority of Drosophila lacking PARG die in the larval stage, but 25% of Drosophila survive, accumulate PAR, and develop a progressive neurodegenerative disorder (118). This PAR-dependent cell death induced by the lowering of PARG levels seems to occur via AIF, as knockdown of AIF by RNA interference reduces the enhanced, chemotherapy-induced cell death due to lowering of PARG (119). Mutations of c6orf130, which interacts with terminal ADP-ribose protein glycohydrolase, cause severe neurodegeneration through impairment of the turnover and recycling of PAR (120).

Iduna (RNF146) is another endogenous inhibitor of parthanatos. Iduna protects against stroke and NMDA receptor–mediated excitotoxicity both in vitro and in vivo by interfering with PAR-dependent cell death (121). It is a PAR-dependent ubiquitin E3 ligase targeting proteins that are PARylated or bind PAR (122–124). Iduna is activated by binding PAR, which serves as an allosteric activation signal (125). It also protects against DNA damage, facilitates DNA repair, regulates cell survival in a PAR-dependent fashion (123), and regulates Wnt signaling (122, 124). Interestingly, Iduna is a gene that is induced by preconditioning (121). Considering the importance of this inhibitor in survival and DNA repair, it will be important to understand how its expression is regulated as well as how its E3 ligase PAR-dependent substrates are involved in cell death.

The mechanism of energetic collapse following PARP-1 activation has been clarified recently. Consumption of cellular NAD⁺ owing to PARylation was thought to be the underlying mechanism of PARP-1–induced energetic collapse (126, 127). Instead, PAR-dependent inhibition of HK leads to defects in glycolysis that account for the bioenergetics collapse (128, 129). PAR binds to HK and inhibits HK’s activity, which leads to inhibition of glycolysis and subsequent reductions in cellular NAD⁺ and ATP. Consistent with notion that PAR binding to HK causes defects in glycolysis, these PAR-dependent defects can be rescued by the mitochondrial substrates pyruvate and glutamine. Thus, the reduction in cellular NAD⁺ is not due to excessive PARylation, as researchers proposed originally (126, 127). These results are consistent with the idea that PAR is a death-signaling molecule (46, 81, 87) and suggest that direct interference of PAR polymer signaling may provide unique opportunities for preventing cell death following activation of PARP-1.

A recent report by Xu et al. (130, p. 333ra48, 1) stated aptly that “translating neuroprotective treatment from discovery in cell and animal models to the clinic has proven challenging.” Recent advances in deriving human neurons from embryonic stem cells or inducible pluripotent stem cells provide an opportunity to validate processes identified in simpler model systems in human neurons (131–133). Human cortical neurons possessing a balanced excitatory and inhibitor network die in a NO- and parthanatos-dependent fashion when exposed to glutamate excitotoxicity or oxygen glucose deprivation. Moreover, inhibitors of PARP that are currently in clinical trials prevent neuronal cell death, providing further support for the notion that inhibition of the parthanatic death cascade may offer therapies that are disease modifying in human neurologic disorders (130).

PERSPECTIVE AND THERAPIES

Based on the central role mitochondria play in neuronal cell death, strategies that target mitochondrial-mediated cell death pathways have particular promise as neuroprotective therapies. Prevention of neurologic injury is likely to have the greatest impact, such as in trauma, stroke, and the restoration of blood flow in stroke. As advances in genomic manipulation such as CRISPR-Cas9 (clustered regularly interspaced short palindromic repeats-CRISPR associated 9) technology enable the correction of genetic defects, it may be possible to prevent neurodegenerative diseases that are due to mutations in causal genes (134). Owing to the daunting barrier or inefficiency of human brain gene transduction, including brain size and the blood-brain barrier, genetic correction may be limited to regions or areas that are readily accessible, such as the retina or small areas such as the substantia nigra that would be amenable to current viral and nanoparticle transduction approaches. Wide application of genetic correction in neurologic disease would require concomitant advances in gene transduction approaches.

Fortunately, as discussed here, many other approaches besides prevention and correction of genetic mutations could alter the course of neurologic disorders. Inhibitors of MPT and MOMP have been touted as cytoprotective agents (35). Targeting mitochondrial dynamics and mitophagy has received attention as a strategy to treat mitochondrial disorders and is the subject of recent reviews (9, 135). Areas of focus include efforts to enhance mitochondrial biogenesis, increase mitophagy, and control of the balance between fission and fusion.

Methods to interfere with mitochondrial-initiated regulated cell pathways include inhibition of caspases or PARP. Although evidence exists to implicate both caspases and PARP in neurologic disorders, the identification of PARP and caspase activity does not indicate causality or a beneficial effect through PARP or caspase inhibition (46, 136). Because multiple cell death pathways are initiated by mitochondrial dysfunction, simple inhibition of one path to cell death may not be sufficient. Moreover, inhibition of caspases or PARP, particularly in chronic neurologic disorders, could have long-term off-target consequences. Caspase or PARP inhibition in acute disorders such as stroke may avoid potential long-term complications. Inhibition of PARP has been particularly attractive since the first PARP inhibitor, olaparib, was approved by the US Food and Drug Administration for ovarian cancer (137). Caspase inhibitors have yet to meet this threshold. Several other PARP inhibitors are also in clinical trials for the treatment of different cancers (138, 139). It will be important to identify brain-penetrant PARP inhibitors with good safety profiles to repurpose and evaluate in neurologic disorders.

Recent advances in elucidating the molecular mechanisms of parthanatos also offer additional therapeutic targets, including interference with PAR signaling and inhibition of PAAN. Both targets are amenable to high-throughput screening, and agents targeting both could potentially avoid the long-term consequences of PARP inhibition.

In summary, research to improve our understanding of mitochondrial function and the cell death pathways orchestrated by mitochondria offers several interventions that hold the potential for disease-modifying therapies to treat neurologic disorders. Substantial work remains before we can identify safe and potent inhibitors of mitochondrial cell death pathways as well as agents that improve mitochondrial function that can be translated for use in humans.