Neuropharmacology – Special Issue on Cerebral Ischemia Mechanisms of Ischemic Brain Damage – Review Article

Abstract

Each year in the United States approximately 700,000 individuals are afflicted with a stroke and currently there are almost 2 million survivors of stroke living in the US with prolonged disability. In China 1.5 million people die from stroke each year and in developed nations stroke is the third leading cause of death, only surpassed by heart disease and cancer. Brain injury following stroke results from the complex interplay of multiple pathways including excitotoxicity, acidotoxicity, ionic imbalance, oxidative/nitrative stress, inflammation, apoptosis and peri-infarct depolarization. Here, we discuss the underlying pathophysiology of this devastating disease and reveal the intertwined pathways that are the target of therapeutic intervention.

Introduction

Each year in the United States approximately 700,000 individuals are afflicted with a stroke. Currently there are approximately 2 million survivors of stroke living in the US with prolonged disability, many unable to work or resume personal relationships. In China 1.5 million people die from stroke each year and in developed nations stroke is the third leading cause of death, only surpassed by heart disease and cancer. In the US health care costs reach 62 billion dollars annually. The economic, social and psychological costs of stroke are substantial (Fisher and Bogousslavsky 1998; Pancioli, Broderick et al. 1998; Stephenson 1998; Caplan 2000; Rosamond, Flegal et al. 2007; Flynn et al, this issue). There are very few treatments for stroke and the development of new therapeutics is imperative. Here, we discuss the underlying pathophysiology of this devastating disease and reveal the intertwined pathways that are the target of therapeutic intervention.

Stroke Pathophysiology

Stroke can be subdivided into two categories, ischemic and hemorrhagic. Ischemic strokes are more prevalent than hemorrhagic making up approximately 87% of all cases and have been the target of most drug trials (Rosamond, Flegal et al. 2007). A thrombosis, an embolism or systemic hypo-perfusion, all of which result in a restriction of blood flow to the brain, can cause an ischemic stroke, which results in insufficient oxygen and glucose delivery to support cellular homoestasis. This elicits multiple processes that lead to cell death: excitotoxicity, acidotoxicity and ionic imbalance, oxidative/nitrative stress, inflammation, apoptosis and peri-infarct depolarization (Gonzalez, Hirsch et al. 2006).

Each of the above pathophysiological processes has a distinct time frame, some occurring over minutes, others over hours and days. These processes share overlapping and redundant features and cause injury to neurons, glia and endothelial cells. Within the core of the ischemic territory, where blood flow is most severely restricted, excitotoxic and necrotic cell death occurs within minutes. In the periphery of the ischemic area, where collateral blood flow can buffer the full effects of the stroke, the degree of ischemia and the timing of reperfusion determine the fate of individual cells. In this ischemic penumbra cell death occurs less rapidly via active cell death mechanisms such as apoptosis. Targeting these mechanisms provides therapeutic opportunity (Gonzalez, Hirsch et al. 2006).

Excitotoxcity, Acidotoxicity and Ionic Imbalance

The human brain comprises 2% of body weight but requires 20% of total oxygen consumption (Edvinsson and Krause 2002). The brain requires this large amount of oxygen to generate sufficient ATP by oxidative phosphorylation to maintain and restore ionic gradients. One estimate suggests that the Na⁺/K⁺ATPase found on the plasma membrane of neurons, consumes 70% of the energy supplied to the brain (Edvinsson and Krause 2002). This ion pump maintains the high intracellular K⁺ concentration and the low intracellular Na⁺ concentration necessary for the propagation of action potentials. After global ischemia, mitochondrial inhibition of ATP synthesis leads to ATP being consumed within two minutes, this causes neuronal plasma membrane depolarization, release of potassium into the extracellular space and entry of sodium into cells (Caplan 2000). Energy failure also prevents the plasma membrane Ca²⁺ ATPase from maintaining the very low concentrations of calcium that are normally present within each cell.

The extracellular calcium concentration is approximately 1.2mM and most cellular processes regulated by calcium have a Km value in the range of 0.1 to 1μM. During ischemia intracellular calcium levels rise to 50-100μM, activating many, if not all calcium dependent proteases, lipases and DNAses (Edvinsson and Krause 2002). Activation of these enzymes causes many cells in the ischemic core to die from simple catabolism. Because no ATP is available for the re-synthesis of cellular constituents these catabolic enzymes cause the necrosis of essential cellular structures.

Membrane depolarization also leads to neurotransmitter release, with the release of the excitatory neurotransmitter glutamate playing a critical role in ischemic pathology. A large concentration gradient of glutamate is maintained across the plasma membrane by sodium dependent glutamate transporters located on presynaptic and postsynaptic membranes. The synaptic glutamate concentration is in the micromolar range, whereas the cytosolic concentration of glutamate is approximately 10mM (Hsu 1998). Membrane depolarization and accumulation of sodium inside cells during ischemia causes reversal of glutamate transporters and allows glutamate to exit cells along its concentration gradient.

The effect of an increase in synaptic glutamate concentration is the activation of N-methyl-D-aspartate (NMDA) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors. NMDA receptors are calcium permeable and the opening of these channels leads to further membrane depolarization and greater calcium influx, exacerbating intracellular calcium overload (excitotoxicity) (Olney 1969). AMPA receptors are not normally calcium permeable by virtue of their GluR2 subunit, however, this subunit is reduced after ischemia increasing the calcium permeability of these receptors by up to 18 fold, allowing AMPA receptors to contribute to delayed calcium dependent cell death (Liu, Liao et al. 2006; Peng, Zhong et al. 2006). Blocking glutamate binding sites on NMDA and AMPA receptors has repeatedly been shown to provide robust neuroprotection in models of focal ischemia, where receptor blockade is thought to prevent calcium entry from reaching a toxic threshold (Hsu 1998).

Metabotropic glutamate (mGlu) receptors also contribute to excitotoxicity. MGlu receptors are G-protein coupled receptors that modulate excitatory synaptic transmission. There are 8 subtypes divided into 3 groups. The group I mGlu receptors (mGlu1 and mGlu5) are predominantly found at the postsynaptic membrane of glutamatergic synapses where they increase neuronal excitability by modulating NMDA and AMPA receptors. Evidence that these receptors enhance the induction and progression of excitotoxic neuronal death is provided by the finding that pharmacologic blockade of group I mGlu receptors provides neuroprotection in in vitro and in vivo models of ischemia (Bruno, Battaglia et al. 2001).

MGlu1 receptors are expressed in GABAergic neurons where their activation suppresses GABA release. Therefore mGlu1 antagonists may be neuroprotective by enhancing the release of the inhibitory neurotransmitter GABA. MGlu5 receptors are physically and functionally connected to NMDA receptors. Therefore, mGlu5 antagonists may limit excitotoxicity by reducing NMDA receptor activation (Pellegrini-Giampietro 2003).

Group II (mGlu2 and mGlu3) and group III (mGlu4, mGlu6, mGlu7, mGlu8) mGlu receptors are predominantly found at presynaptic terminals where they inhibit the release of glutamate. Group II receptors are also widely expressed on astrocytes. Both group II and group III receptor agonists have been found to be neuroprotective by limiting the induction of excitotoxicty, however, group II receptor agonists also confer neuroprotection by increasing production of neurotrophic factors such as nerve growth factor and TGF-β in astrocytes (Bruno, Battaglia et al. 2001).

Concurrent to the induction of excitotoxicity, calcium overload is further exacerbated by acidosis, one of the hallmark neurochemical elements of the anaerobic metabolism of ischemia. Hyperglycemia increases lactate in the ischemic environment further depressing pH. Dissociated protons activate sodium selective acid sensing ion channels (ASICs) that are permeable to calcium. There are 4 ASIC genes encoding 6 polypeptides (ASIC1a, ASIC1b, ASIC2a, ASIC2b, ASIC3 and ASIC4), each of which has a distinct pH sensitivity. The pH_0.5 of the calcium permeable homomeric ASIC1a channel is 6.2 and because the pH in ischemic brain falls to 6.0 to 6.5 it is likely that in ischemia this channel opens and allows further calcium entry into the cell (acidotoxicity) (Simon 2006). This phenomenon is glutamate independent, thus it is not prevented by administration of NMDA antagonists. However, administration of the selective ASIC1a blocker PcTx1 can prevent ASIC1a activation and has been shown to reduce lesion volume in experimental stroke (Pignataro, Simon et al. 2007).

Peri-infarct Depolarizations

Cortical spreading depression (CSD) is a self-propagating wave of electrochemical activity that progresses through cortical tissue in intact brain. CSD causes sustained (1-5 minutes) cellular depolarization, depressed neuro-electrical activity, increased glutamate release and loss of membrane ionic gradients (Gonzalez, Leroux et al. 1992). Peri-infarct depolarizations (PIDs) are spontaneous waves of depolarization with all of the characteristic features of CSD that propagate through the penumbra following focal stroke. PIDs may be caused by the release of potassium and excitatory amino acids from the ischemic core. Although CSD in the normally perfused brain does not lead to cell death, recurrent PIDs in the ischemic brain are associated with increased ischemic injury.

Repeated depolarization in the penumbra may mediate tissue damage by allowing calcium to accumulate within neurons. A critical threshold of calcium could be reached in the case of PID due to the compromised energy supply of the tissue, thus causing damage in the case of PID but without evidence of lasting damage in the case of CSD.

PIDs are known to occur in mouse, rat and cat stroke models, where the incidence and duration of spreading depression correlates with infarct maturation (Gill, Andine et al. 1992; Strong, Smith et al. 2000). Recently Fabricius and colleagues demonstrated the existence of PIDs in the acutely injured human brain, which suggests that inhibition of spreading depression using a therapeutic approach such as hypothermia or glutamate receptor antagonism could be an important strategy to limit development of ischemic injury within the penumbra (Chen, Chopp et al. 1993; Fabricius, Fuhr et al. 2006).

Oxidative and Nitrative Stress

High levels of intracellular Ca²⁺, Na⁺ and ADP cause mitochondria to produce deleterious levels of reactive oxygen species. Unlike other organs the brain is especially vulnerable to reactive oxygen species due to neurons having relatively low levels of endogenous antioxidants (Coyle and Puttfarcken 1993). Overly abundant oxygen radicals cause the destruction of cellular macromolecules and participate in signaling mechanisms that result in apoptotic cell death (Halliwell 1994; Sugawara and Chan 2003). Ischemia activates nitric oxide synthase (NOS) and increases the generation of nitric oxide (NO), which combines with superoxide to produce peroxynitrite, a potent oxidant. The production of NO and oxidative stress is also linked to over-activation of poly(ADP-ribose)polymerase-1 (PARP-1), a DNA repair enzyme. In response to DNA strand breaks PARP-1 catalyzes the transformation of β-nicotinamide adenine dinucleotide (NAD+) into nicotinamide (NA) and long polymers of poly(ADP-ribose). When PARP-1 is over-activated it depletes cells of NAD+, impairing NAD+ dependent processes such as anaerobic glycolysis and mitochondrial respiration, which leads to ATP starvation, energy failure and neuronal death (Gonzalez, Hirsch et al. 2006).

Following reperfusion there is a surge in production of superoxide, NO and peroxynitrate. Formation of these radicals in the vicinity of blood vessels plays an important role in reperfusion-induced injury. These radicals activate matrix metalloproteases (MMPs), which degrade collagen and laminins in the basal lamina, which disrupts the integrity of the vascular wall and increases blood brain barrier (BBB) permeability. Oxidative and nitrative stress also triggers recruitment and migration of neutrophils and other leukocytes to the cerebral vasculature, which release enzymes that further increase basal lamina degradation and vascular permeability. These events can lead to parenchymal hemorrhage, vasogenic brain edema and neutrophil infiltration into the brain (Crack and Taylor 2005).

Thrombolytic therapy has a 3 hour time window of efficacy. Part of the reason for this limited time window is that the surge in production of free radicals associated with delayed reperfusion brings a second wave of oxidative and nitrative stress that increases the risk of brain hemorrhage and edema. Combined administration of thrombolytic agents with free radical or peroxynitrite scavengers, or NO synthase inhibitors is a potential strategy for reducing reperfusion-induced injury and extending the time window available for thrombolysis (Crack and Taylor 2005).

Inflammation

Inflammation contributes to stroke related brain injury. However, the effect of individual components of the inflammatory cascade can be beneficial depending on the stage of tissue injury, the magnitude of the response and whether the inflammatory component also activates neuroprotective pathways (Bruce, Boling et al. 1996; Nawashiro, Brenner et al. 2000; Zhang, Zhang et al. 2000). The inflammatory response is a composite process that involves many different cell types, inflammatory mediators and extracellular receptors. Below, the inflammatory response to stroke is sub-divided into the cellular response, the cytokine response and the response to toll-like receptor (TLR) activation.

i.) The Cellular Inflammatory Response

Stroke causes neutrophilia, lymphocytopenia and an increase in the number of circulating monocytes (Ross, Hurn et al. 2007). Although the increase in neutrophil and monocyte number is likely to contribute to ischemic damage, this change is also evident in patients that have experienced a transient ischemic attack (TIA), which indicates that these cellular elements alone are insufficient by themselves to mediate damage (Ross, Hurn et al. 2007). Instead, access to the brain may be the key determinant of whether this change contributes to ischemic injury.

Neutrophils accumulate in the brain as early as 30 minutes after permanent middle cerebral artery occlusion (MCAO). Transmigration is mediated by three classes of cell adhesion molecules: selectins, integrins, and immunoglobulins, the expression of which is regulated both intracellularly and by cytokine signaling (Huang, Upadhyay et al. 2006). The recruitment of neutrophils to ischemic brain begins with neutrophil rolling on activated endothelial blood vessel walls, mediated by selectins, followed by neutrophil activation and adherence, mediated by integrins and immunoglobins. When adhered to cerebral blood vessel walls, neutrophils transmigrate into the cerebral parenchyma, a process facilitated by blood brain barrier (BBB) disruption. The recruitment of neutrophils can obstruct the microcirculation and prevent complete restoration of cerebral blood flow after reperfusion. This blockage may cause further tissue damage after ischemia and is described as the ischemic no-reflow phenomenon (Huang, Upadhyay et al. 2006).

Once neutrophils penetrate into ischemic brain they cause tissue damage by releasing oxygen-free radicals and proteolytic enzymes. Neutrophil depletion, inhibition of neutrophil adhesion, and inhibition of neutrophil function are all strategies that have been shown to reduce infarct volume and improve outcome. For example, protein kinase C has been shown to play a significant role in neutrophil adhesion, degranulation, and superoxide generation. Mice that are deficient for protein kinase C have diminished infarct volumes when subjected to transient cerebral ischemia (Chou, Choi et al. 2004; Huang, Upadhyay et al. 2006).

Lymphocytes also appear to be responsible for mediating damage in response to brain ischemia. Although lymphocytes are ordinarily excluded from the central nervous system (CNS), they appear within 24 hours in post ischemic brain (Schroeter, Jander et al. 1994). While the mechanism producing their infiltration into the brain remains unclear, it is likely that BBB disruption plays a major role in the influx, either by directly allowing free lymphocyte movement, or by leakage of brain antigens resulting in the transmigration of activated lymphocytes. It has been shown recently, using severe combined immunodeficiency (SCID) mice deficient for T and B lymphocytes, that these cells contribute to development of the lesion in the cortex. In these mice Hurn et al find that following focal cerebral ischemia, striatal infarction is not altered, suggesting that the core of an evolving infarct is not protected by a lack of T and B lymphocytes, while cortical infarct volume is reduced by as much as 40% (Hurn, Subramanian et al. 2007).

The infiltration of bone marrow derived cells into the ischemic brain persists for weeks following stroke, and while the initial infiltration leads to worsening of tissue damage and exacerbation of neurological deficits, subsequent aspects of the infiltration are beneficial. For example, the phagocytosis of debris and the release of cytokines that promote glial scar formation are critical for effective wound healing.

ii.) The Cytokine Inflammatory Response

Cytokines and chemokines contribute to stroke related brain injury (Gong, Qin et al. 1998). During ischemia, cytokines, such as IL-1, IL-6, TNF-α, TGF-β and chemokines such as CINC and MCP-1 are produced by a variety of activated cell types, including endothelial cells, microglia, neurons, platelets, leukocytes, and fibroblasts (Huang, Upadhyay et al. 2006).

Production of IL-1 is increased after permanent or transient cerebral ischemia in microglia, astrocytes, and neurons. The exact role of IL-1 in propagating tissue damage is unclear, although possible deleterious effects of IL-1 include fever, arachidonic acid release, enhancement of NMDA mediated excitotoxicity, and stimulation of nitric oxide synthesis (Huang, Upadhyay et al. 2006). An additional role of IL-1 may be recruitment and adhesion of neutrophils. IL-1 has been shown to cause up-regulation of E-selectin, ICAM-1, ICAM-2, and VCAM-1 on cerebral endothelial cells and the induction of such adhesion molecules may explain why elevated IL-1 levels after ischemia increases neutrophil infiltration (Yamasaki, Matsuo et al. 1997; Huang, Upadhyay et al. 2006). That the effects of IL-1 are deleterious was demonstrated by Garcia and Relton who showed that administration of recombinant IL-1 receptor antagonist reduces the severity of neurologic deficits and tissue necrosis in rats subjected to permanent MCAO (Garcia, Liu et al. 1995; Relton JK, Martin D et al. 1996; Huang, Wang et al. 2003). Insight into the deleterious effect of IL-1 also comes from data in rats administered recombinant human IL-1β via intracerebroventricular (icv) injection after MCAO. These rats show increased brain edema and lesion size, as well as an increased influx of neutrophils (Yamasaki Y, Shozuhara H et al. 1994; Yamasaki, Matsuura et al. 1995; Huang, Upadhyay et al. 2006).

In rats subjected to permanent MCAO, IL-6 mRNA expression is up-regulated as early as 3 hours after occlusion, peaks at 12 hours, and continues for at least 24 hours (Wang, Yue et al. 1995). The biological activity of IL-6 overlaps with those of IL-1, and data from human studies suggest a proinflammatory role for IL-6 in stroke. Peripheral blood levels of IL-6 are higher in stroke patients and detectable within a few hours of stroke onset and higher CSF and serum levels of IL-6 correlate with larger infarct size and poorer clinical outcome (Tarkowski, Rosengren et al. 1995; Huang, Upadhyay et al. 2006). However, IL-6 also has anti-inflammatory properties due to its ability to induce IL-1 receptor antagonist synthesis (Schindler, Mancilla et al. 1990; Relton JK, Martin D et al. 1996). Thus, it is unclear whether the overall effect of IL-6 is beneficial or detrimental in the context of stroke.

Up-regulation of TNF-α mRNA parallels that of IL-1 and IL-6 mRNA within the first hours after ischemia (Huang, Upadhyay et al. 2006). Both experimental and human data indicate a positive correlation between TNF-α and the extent of ischemic injury. For example, a study of 24 patients with ischemic stroke found that CSF levels of TNF-α were markedly increased within 24 hours of ischemic stroke and that the levels of CSF and serum TNF-α were positively correlated with infarct volume (Zaremba, Skrobanksi et al. 2001). Like IL-1, TNF-α induces adhesion molecule expression in cerebral endothelial cells and promotes neutrophil accumulation and transmigration. In addition TNF-α stimulates acute-phase protein production, disrupts the blood-brain barrier and stimulates the induction of other inflammatory mediators. Administration of TNF-α neutralizing antibody reduces brain injury after focal ischemia in rats and TNF-α targeted therapies hold great promise as a stroke treatment (Nawashiro, Martin et al. 1997; Nawashiro, Martin et al. 1997).

Growing evidence suggests that TGF-β plays a neuroprotective role in the pathogenesis of stroke. In rodent models of cerebral ischemia, increased expression of TGF-β mRNA is demonstrated in ischemic tissues as early as 1 to 6 hours after the ischemic event and remains elevated for up to 15 to 21 days (Wiessner, Gehrmann et al. 1993). The effects of TGF-β upon stroke volume have been studied with intracarotid and icv administration and TGF-β has been found to be neuroprotective if administered before or after the ischemic insult (McNeill, Williams et al. 1994; Huang, Upadhyay et al. 2006).

It is likely that the neuroprotective effect of TGF-β is the concerted result of the activation of several neuroprotective pathways. TGF-β1 has a concentration-dependent protective effect against neuronal injury caused by glutamate excitotoxicity in vitro, and recent evidence indicates that administration of a TGF-β1 blocking agent increases the extent of excitotoxic lesions after focal cerebral ischemia (Ruocco, Nicole et al. 1999; Huang, Upadhyay et al. 2006). Additionally, intracarotid administration of TGF-β has been shown to reduce the number of circulating neutrophils, which may ameliorate the post ischemic no-reflow state (Mori, Del Zoppo et al. 1992). Data also exists to support a role for TGF-β in diminishing ischemia-induced endothelial dysfunction (Lefer, Ma et al. 1993).

Increased expression of CINC and MCP-1 mRNA is detected in the brain of rats as early as 6 hours after permanent MCAO, reaches a maximal level at 12 hours and is decreased by 24 hours (Minami and Satoh 2003). CINC and MCP-1 expression attracts neutrophils to ischemic tissue, evidence for which comes from rodent experiments using transient ischemia, where CINC and MCP-1 are detectable in cerebral tissue before neutrophil infiltration (Huang, Upadhyay et al. 2006). In rats, administration of anti-CINC antibody decreases cerebral edema and infarction, which further supports a role for CINC in mediating neutrophil infiltration and demonstrates another therapeutic opportunity (Yamasaki, Matsuo et al. 1997).

iii.) The Role of TLRs in the Inflammatory Response

TLRs function as a first-line defense against pathogen invasion. TLRs recognize pathogen-associated molecules such as the bacterial cell wall components peptidoglycan (TLR2) and lipopolysaccharide (TLR4), as well as dsRNA (TLR3), ssRNA (TLR7), and nonmethylated Cytosine-Guanosine (CpG) DNA (TLR9). Upon activation TLRs, induce downstream signals that lead to cytokine and chemokine production, which initiates a localized inflammatory response. In the periphery, TLRs are expressed on B cells, dendritic cells, and macrophages. Within the CNS they are expressed on endothelial cells, microglia, astrocytes, oligodendrocytes, and neurons (Marsh and Stenzel-Poore 2007).

Endogenous molecules associated with tissue damage can also activate TLRs. Therefore, in addition to playing a role in pathogen detection and defense, TLRs function as sensors of tissue damage. Fibrinogen, heat shock proteins and components of the extracellular matrix activate TLR4, while host DNA and mRNA are ligands for TLR9 and TLR3, respectively. Several recent studies implicate TLR activation by endogenous ligands as a detrimental response to cerebral ischemia. Endogenous TLR ligands such as Hsp70 are upregulated in the brain following ischemia and mice lacking either TLR2 or TLR4 have significantly smaller infarcts than wild-type mice (Kinouchi, Sharp et al. 1993; Cao, Yang et al. 2007; Ziegler, Harhausen et al. 2007).

TLRs present a novel therapeutic target for stroke pretreatment. Tasaki et al show that administration of a low dose of the TLR4 agonist LPS, can induce tolerance to cerebral ischemia. LPS induced tolerance to brain ischemia has since been demonstrated in a mouse model of stroke and in a porcine model of deep hypothermic circulatory arrest (Rosenzweig, Lessov et al. 2004; Hickey, You et al. 2007). Tolerance induction appears to require a small inflammatory response to LPS, as it can be blocked by simultaneous administration of cyclohexamide, dexamethasone, and TNF-α inhibitors (Tasaki, Ruetzler et al. 1997; Bordet, Deplanque et al. 2000; Rosenzweig, Minami et al. 2007). Antecedent treatment with LPS also appears to confer protection by protecting against the cytotoxic effects of TNF-α following cerebral ischemia. Mice that have been pretreated with LPS show reduced TNF-α in the serum, decreased levels of TNFR1 and enhanced levels of neutralizing soluble TNFR1 following stroke (Rosenzweig, Minami et al. 2007). LPS induced tolerance is comparable to ischemic preconditioning, the paradigm of a brief non-injurious period of ischemia conferring neuroprotection against a subsequent bout of injurious ischemia. Both these phenomena appear to protect in part by reprogramming the genomic response to ischemic injury (Stenzel-Poore, Stevens et al. 2003).

Apoptosis

Mild ischemic injury preferentially induces cell death via an apoptotic-like mechanism rather than necrosis. Because the ischemic penumbra sustains milder injury and preserves ATP, apoptosis predominates in this region (Kerr 1965; Kerr, Wyllie et al. 1972; Gonzalez, Hirsch et al. 2006). Triggers of apoptosis include oxygen free radicals, death receptor ligation, DNA damage, protease activation and ionic imbalance.

The release of cytochrome c from the outer mitochondrial membrane plays a central role in mediating apoptosis in response to ischemia. Release of cytochrome c is caused by ionic imbalance and mitochondrial swelling or by formation of a pore in the outer mitochondrial membrane. The complex interplay of the Bcl-2 family of proteins either promotes (Bax, Bak, Bad, Bim, Bid) or prevents (Bcl-2, Bcl-XL, Bcl-w) pore formation. The pore is formed by oligomerization of Bax and/or Bak in the outer membrane, with Bax transcriptionally induced by p53, which in turn is activated by DNA damage. The anti-apoptotic Bcl-2 proteins, Bcl-2 and Bcl-XL, can form heterodimers with Bax, thereby preventing pore formation (Edwards and Dean 1977; Miyashita and Reed 1995; Hengartner 2000; Kroemer and Reed 2000; Adams and Cory 2001; Antonsson, Montessuit et al. 2001).

Bad, Bim and Bid also influence pore formation, although the exact manner by which these molecules do so is unclear. Bim is bound to dynein and actin and is released by dissociation of the cytoskeleton. Upon mobilization, Bim translocates to the mitochondrial membrane and promotes the release of cytochrome c. Bid is present in a proform in the cytosol and is cleaved by caspase 8 after TNF/Fas receptor activation. Once cleaved, Bid also translocates to the mitochondrial membrane and promotes the release of cytochrome c. In its phosphorylated form Bad is bound to protein 14-3-3. Bad is dephosphorylated by calcineurin, a calcium dependent serine/threonine phosphatase. When Bad is dephosphorylated it is released from 14-3-3 and translocates to the mitochondria where it also promotes cytochrome c release. Unlike Bax and Bak, Bid, Bad and Bim do not have the ability to directly mediate cytochrome c release by forming a pore. Instead they appear to function as sensors of cell stress that may promote apoptosis by heterodimerizing and antagonizing the function of anti-apoptotic Bcl-2/Bcl-XL and/or activating the function of Bax and Bak (Edwards and Dean 1977; Hengartner 2000; Kroemer and Reed 2000; Adams and Cory 2001).

Cytochrome c release activates downstream caspases of the intrinsic pathway through formation of the apoptosome, a complex of dATP, cytochrome c, procaspase 9 and Apaf1. Effector caspases 3 and 7 then target substrates that dismantle the cell by cleaving homeostatic, cytoskeletal, repair, metabolic, and cell signaling proteins. These caspases also cause further DNA fragmentation by activating capase-activated deoxyribonuclease (CAD) by cleaving the inhibitor protein ICAD. Caspase activation can be modulated by protein inhibitors of apoptosis (IAP) and indirectly by secondary mitochondria-derived activator of caspase (Smac/Diablo).

Activation of the extrinsic pathway of death receptors can induce caspase activation independent of the release of cytochrome c. Death receptor ligation results in activation of caspase-8 and caspase-10, which in turn can activate effector caspase 3 (Namura, Zhu et al. 1998). Activation of death receptors such as Fas/CD95, TNFR1, and the TRAIL receptor is promoted by the TNF family of ligands, including FASL, TNF, LT-alpha, LT-beta, CD40L, LIGHT, RANKL, and TRAIL, which are released as part of the inflammatory response to ischemia (Zoppo 1997; Zoppo, Ginis et al. 2000).

Susceptibility to ischemia often correlates with expression of Bcl-2. For example in the hippocampus, basal Bcl-2 immunoreactivity is high in ischemia resistant pyramidal neurons of the CA3 region, but very low in the ischemia sensitive CA1 region (Chen, Graham et al. 1995). Bcl-2 expression is also high in the brainstem where autonomic function is often preserved following ischemia and low in the selectively vulnerable neurons of the cortex (Chen, Graham et al. 1995). This correlation of Bcl-2 expression and resistance to apoptosis may be explained by recent findings that suggest in addition to physically trapping pro-apototic proteins, Bcl-2 has other properties that enable it to attenuate cell death. For example, Ellerby et al find that Bcl-2 is sensitive to redox changes and has antioxidant properties during calcium stress (Ellerby, Ellerby et al. 1996). As Bcl-2 is found in the ER, the plasma membrane, and the nuclear membrane, additional Bcl-2 functions may be revealed.

Several experimental studies, have shown that inhibition of apoptosis reduces ischemic injury (Graham and Chen 2001). For example, activation of the extracellular signaling protein kinase (ERK) pathway and the phosphatidylinositol-3 (PI3) kinase pathway has been found to be neuroprotective. These pathways activate transcription factors such as CREB and NFκB related to cell survival and phosphorylate Bax and Bad, thereby preventing the release of cytochrome c. Additionally, caspase 3 inhibitors, gene deletions of Bid, the use of peptide inhibitors and viral vector mediated gene transfer of Bcl-2 and Bcl-XL are strategies that are known to be neuroprotective (Shinoura, Satou et al. 2000; Zhao, Yenari et al. 2003; Gonzalez, Hirsch et al. 2006; Guan, Pei et al. 2006; Guan QH, Pei et al. 2006).

A summary of the most salient features of mitochondria-dependent and mitochondria-independent activation of cell death is provided in figure 1.

Figure 1. Apoptosis in Stroke.

Cell death pathways relevant to apoptosis in cerebral ischemia. The release of cytochrome c (cyt c) from the mitochondria is mediated by the proapoptotic proteins Bax and/or Bak forming a pore in the mitochondrial membrane. Pore formation is facilitated by Bad and Bid. Calcium influx causes the dephosphorylation of Bad by calcineurin (CaN), which releases BAD from 14-3-3 and allows its translocation to the mitochondria. Calcium can also activate calpains, which can activate cathepsins that mediate the limited proteolysis of Bid, allowing truncated Bid (t-Bid) to translocate to the mitochondria. Caspase 8, which is activated by TNF receptor ligation, also mediates the limited proteolysis of Bid, allowing truncated Bid (t-Bid) to translocate to the mitochondria. Once present in the cytosol, cyt c forms the apoptosome complex by binding to Apaf-1 and procaspase 9. The apoptosome complex cleaves and activates caspase 3, which causes actin fragmentation, and endonuclease activation. Caspase 3 can also be activated by caspase 8. Apoptosis-inducing factor (AIF) can also be released from the pore created in the mitochondria, causing DNA degradation. DNA damage activates p53, which further increases Bax expression. The antiapoptotic proteins Bcl-XL and Bcl-2 prevent Bax mediated pore formation and cyt c release. Various survival factors also prevent pore formation and cytochrome C release by activation of Akt and ERK pathways (adapted from (Edvinsson and Krause 2002)).

Conclusion and Perspectives

Cell death following stroke results from the complex interplay of excitotoxicity, acidosis, inflammation, oxidative stress, peri-infarct depolarization and apoptosis. On the basis of the complexity of events in cerebral ischemia and the disappointing results from single agent trials, it may be unrealistic to expect that a single neuroprotective drug will demonstrate benefit in human stroke. In light of this complexity, it is likely that effective stroke therapy will require a combinatorial approach. This point is particularly salient when one considers that stroke is also a heterogeneous disorder that can be ischemic or hemorrhagic, involve small or large blood vessels and be exacerbated by hypotension, fever and hyperglycemia. Age, gender, racial background, comorbidity and concurrent medications also influence stroke; thus individual differences among patients undoubtedly influence the impact and temporal profile of ischemic injury.

Several experimental studies of combination drug therapy for stroke have shown that drugs that target different pathways can act together to have a synergistic or additive effect. For example tirilazad mesylate, a scavenger of free radicals, has been successfully combined with MK-801, a glutamate receptor antagonist. Tirilazad mesylate has also been successfully combined with insulin, which reduces hyperglycemia following acute stroke and diazepam, a GABA-ergic drug that inhibits nitric oxide formation in the brain. Nimodipine, a calcium channel blocker, has been successfully combined with MK-801. FGF, which strengthens antiapoptotic pathways in neurons has been successfully combined with citicoline, an essential intermediate in the biosynthetic pathway of neuronal membranes that inhibits membrane phosphlipases and can restore Na+/K+ ATPase function (Uematsu, Araki et al. 1991; Matsumoto, Scheller et al. 1993; Meden, Overgaard et al. 1993; Auer 1995; Lyden, Lonzo et al. 1995; Meden, Overgaard et al. 1996; Schabitz, Fuhai et al. 1999). Additional studies have shown that the effect of a single drug can be enhanced or the time window available for administration extended if applied in combination with hypothermia, a powerful therapy that targets multiple mechanisms of damage. For example, Matsumoto et al found synergistic effects of s-emopamil, which reduces brain edema, and nimopidine with hypothermia and Zhao et al used hypothermia to extend the time window available for Bcl-2 gene therapy (Zausinger, Scholler et al. 2003; Kollmar, Henninger et al. 2004; Zhao, Yenari et al. 2004; Zhao, Shimohata et al. 2005). Therefore, due to the extensive overlap that exists with the mechanisms of ischemic damage activated by stroke, combinatorial therapy may represent the most promising direction for future stroke research.