Ischemic tolerance as an active and intrinsic neuroprotective mechanism

9.1. Introduction to ischemic tolerance

Ischemic tolerance, the phenomenon that occurs when a preconditioning sublethal insult induces protection against a subsequent severe ischemic insult, represents an intrinsic mechanism in which cells ‘learn’ to tolerate a stressful environment. Substantial literature has addressed ischemic tolerance in the myocardium, forming a strong foundation for the physiological relevance and inherent mechanisms of cellular protection via tolerance. Given the unique nature of the central nervous system – in terms of its vasculature, high level of organization, and cellular diversity – the mechanisms and application of cerebral ischemic tolerance probably diverge in critical aspects compared to models in other organ systems. Understanding the intrinsic neuroprotection afforded by sublethal preconditioning may yield both improved clinical understanding of the context of severe ischemia with prodromal transient ischemic attack and potential therapeutics based on a natural cellular response and environment. However, these and other applications are still in the distant future in the context of the nervous system, as our current knowledge of the mechanisms and effects of cerebral ischemic tolerance is still in its infancy.

Within this chapter, we will present data that supports cerebral ischemic tolerance as an active, endogenous process that typically involves energy-requiring alterations in gene expression, post-translational modification, and regulated signaling. Inducible ischemic tolerance serves as an experimental probe into the mechanism of endogenous neuroprotection, which, once better understood, may lead to more effective treatment of ischemic and other injuries to the nervous system. Instead of examining how neurons and glia die, ischemic tolerance also provides a unique opportunity to learn how cells induce survival in the face of stress. In this review, we present current thoughts on the features, mechanism, and potential applications of ischemic tolerance.

9.1.2. Ischemia-induced preconditioning and tolerance paradigms

Many models exist that have been found to induce ischemic tolerance, including non-ischemic preconditioning stimuli. While we will focus initially on the ischemic preconditioning models, non-ischemic models will be discussed later in the chapter. It is important to note, however, that the protection afforded by the many different paradigms of ischemic tolerance varies widely, ranging from protection for approximately one-third of the vulnerable population to nearly complete protection. In addition to the range of cellular protection, the different paradigms vary in the temporal nature of the observed effects, where many models represent transient neuroprotection. Thus, at its best, ischemic tolerance may represent a mechanism by which endogenous neuroprotective strategies are prophylactic against severe ischemia; at the least, ischemic tolerance may allow for an increased therapeutic window to intervene in post-ischemic treatment.

9.1.2.1. Ischemia induced in vivo

Ischemia, defined by lack of blood and oxygen supply to tissue, can be induced in the rodent brain by two primary models: global (forebrain) ischemia and focal ischemia. Both models have been used extensively to study the effects of preconditioning.

A large number of the studies focusing on ischemic preconditioning have employed global ischemic models, in which both preconditioning and subsequent prolonged ischemic events were induced by either permanent occlusion of the vertebral arteries combined with brief occlusion of bilateral common carotid arteries (4VO), or by bilateral carotid artery occlusion with systemic hypotension (2VO) (Kitagawa et al., 1990; Kato et al., 1991; Kirino et al., 1991; Tanaka et al., 2004; Ueda and Nowak, 2005). The primary differences between these two forebrain ischemia models lie in the method chosen to interfere with collateral circulation. Importantly, animal models vary widely in collateral blood flow. For example, the posterior communicating artery – largely responsible for providing collateral blood flow from the basilar artery to the middle and anterior cerebral arteries – is lacking in the gerbil but present in the rat. Thus, bilateral occlusion without hypotension in gerbils mimics the 4VO model in rats and induces forebrain ischemia. However, difficulties in inducing the global ischemia model reliably across and within species have been noted, and are reflected by conflicting reports in the literature, including in tolerance paradigms. Earlier studies induced global ischemia without careful control of brain temperature; upon later consideration, brain temperature was found to spontaneously decrease, causing inherent neuroprotection in some cases (Traystman, 2003). Therefore, in the evaluation of older ischemia and ischemic tolerance reports, potential effects of brain temperature must be taken into account. Perhaps the best controlled paradigm of global ischemia and preconditioning has come from measuring depolarization thresholds as an indicator of ischemic severity and using this indicator to optimize the global ischemic tolerance model conditions (Abe and Nowak, 2004; Halaby et al., 2004; Ueda and Nowak, 2005). However, despite the differences in the induction of global ischemia, ischemic tolerance has been identified under most experimental paradigms using these models.

Focal ischemia, induced by either transient or permanent occlusion of the middle cerebral artery (MCAO), has been a widely used model of cerebral ischemia in vivo. MCAO can be induced by intraluminal insertion of a filament through the internal carotid artery up to the juncture with the middle cerebral artery, effectively blocking blood flow into the surrounding MCAO territory. Removal of this filament allows for direct and immediate reperfusion, yielding infarct that is localized ipsilaterally, while the contralateral hemisphere serves as control. Focal ischemic tolerance induced by brief MCAO preconditioning has been well documented and most probably represents a better clinical presentation of prodromal transient ischemic attacks with subsequent stroke (Simon et al., 1993; Glazier et al., 1994; Matsushima and Hakim, 1995; Chen et al., 1996; Chen and Simon, 1997).

Interestingly, global ischemic preconditioning can also induce ischemic tolerance against focal ischemia (a global/focal paradigm) (Simon et al., 1993). Similarly, focal ischemic preconditioning induced ischemic tolerance against global ischemia (focal/global paradigm) (Glazier et al., 1994; Miyashita et al., 1994). Thus, ischemic preconditioning is a general phenomenon that can induce ischemic tolerance.

9.1.2.2. Ischemia induced in vitro

While in vivo models of cerebral ischemia maintain the systemic structures imperative to understanding disease states, in vitro models have provided an invaluable tool for identifying reproducible molecular and biochemical responses to stressors that can then be taken to more complex systems. Ischemia and tolerance paradigms can be mimicked in vitro in neuronal cultures or hippocampal slice preparations by oxygen– glucose deprivation, where hypoxic conditions are combined with depletion of glucose in the culture media, analogous to the complete loss of blood flow to tissue. Interestingly, ischemic tolerance has been reproduced in neuronal cultures, including cortical, hippocampal, and to a limited extent PC12 cultures, or in hippocampal slices preparations (Bruer et al., 1997; Khaspekov et al., 1998; Grabb and Choi, 1999; Gonzalez-Zulueta et al., 2000; Tanaka et al., 2004). Not only was tolerance found to decrease cell death in in vitro models but evoked potentials were maintained in preconditioned slice preparations, as opposed to the short-term loss observed following severe oxygen–glucose deprivation alone (Badaut et al., 2005). This argues for the preservation of some characteristics of system functionality in culture. The in vitro models have proved useful for providing a basis to extend into the more complicated, technically challenging, costly, and ethically mindful whole-animal model systems.

9.1.3. Two time windows of ischemic tolerance

Similar to tolerance characterized in the myocardium, two time windows of ischemic tolerance have been detected in cerebral ischemic models (Fig. 9.1). The first window is rapid, occurring within minutes after preconditioning and lasting approximately 1–2 hours; the second window is delayed, starting 24 hours after preconditioning, peaking at 48 hours to 72 hours, and lasting up to 1 week (Chen and Simon, 1997; Kirino, 2002; Dirnagl et al., 2003; Perez-Pinzon, 2004). No protection has been found if severe ischemia is induced during the interval between the two tolerance windows (i.e. from 2 hours to 1 day following sublethal ischemia).

Fig. 9.1.

Tolerance windows. Following sublethal ischemia, two windows of tolerance occur – a rapid window (top) and a delayed window (bottom). In the rapid window, severe ischemia must be induced within several hours (hatched box) for tolerance to occur, and results in transient neuroprotection (broken line). The delayed window occurs after a minimum of 24 hours following the sublethal ischemia, and results in more sustained neuroprotection.

9.1.3.1. Rapid window of ischemic tolerance

The rapid window of ischemic tolerance was first reported in vulnerable neurons of rat hippocampi following global ischemia (Perez-Pinzon et al., 1997). A procedure of 2 min of 2VO followed by 30 min of reperfusion and then 10 min of 2VO resulted in reduction of ischemic cell death on histopathological examination 3 days later when compared to groups exposed to 10 min of 2VO only (control ischemia group). This neuroprotection appeared to be transient, as hippocampal cell death in the preconditioned group was not significantly different to that of the control ischemia group when examined after 7 days. However, the rapid window was not observed in all global tolerance paradigms. Using a global ischemia model in gerbils, no protection was observed in a rapid timeframe; however, as discussed below, a delayed tolerance was observed 24 hours following the conditioning event (Kato et al., 1991). The differences observed in the rapid onset of tolerance in global ischemia models could be due in part to the species variations, such as the profound differences in cerebral blood flow between rat and gerbil (i.e. gerbils lack the posterior communicating artery) that modifies the model, or by differences in brain temperature during ischemia (Traystman, 2003).

Preconditioning paradigms of focal ischemia induced by MCAO also revealed a rapid window of ischemic tolerance in which sublethal ischemia was followed within 1 h by a more severe insult (Stagliano et al., 1999; Atochin et al., 2003). In contrast to the transient nature of the protection observed in the 2VO tolerance model, MCAO-induced tolerance occurring after a rapid window decreased the ischemic infarct, even when assessed 7 days following prolonged MCAO compared to ischemic controls (Stagliano et al., 1999; Nakamura et al., 2002; Atochin et al., 2003). Supporting the observations of an early, rapid window of ischemic tolerance, cortical cultures were found to be protected by preconditioning oxygen–glucose deprivation events occurring 1 hour, but not 4 h, prior to the lethal oxygen–glucose deprivation event (Meller et al., 2006). These results suggest that a rapid window of ischemic tolerance exists at least in certain study paradigms and that, within in this time frame, acute neuronal injury may be delayed, lessened, or potentially prevented.

9.1.3.2. Delayed window of ischemic tolerance

The majority of the literature on ischemic tolerance focuses on a prolonged window (i.e. severe ischemia induced 1–7 days following ischemic preconditioning). In global ischemia models, the window of delayed ischemic tolerance lasted 1–7 days and afforded significant neuroprotection following 1 week survival (Kitagawa et al., 1990. 1991; Kato et al., 1991), as opposed to the transient protection afforded by the rapid window in the global ischemia models discussed above. Although significant neuroprotection has been observed in global ischemia preconditioning models, this protection may not be as permanent as originally speculated. Using a more controlled model of determining ischemic severity based on depolarization thresholds, Nowak and his colleagues determined that, although preconditioning 2 days before severe ischemia resulted in significant protection, this protection lessened when assessed at 14 days of survival (Ueda and Nowak, 2005). However, the results still suggest that the therapeutic window following severe focal ischemia (analogous to human stroke) may be extended when induced consequent to a delayed preconditioning window.

Similarly, focal ischemic preconditioning 2–7 days prior to subsequent MCAO induced ischemic tolerance (Matsushima and Hakim, 1995; Chen et al., 1996; Kitagawa et al., 1996; Chen and Simon, 1997). Interestingly, the preconditioning MCAO effect could be attained either as a single event of sublethal ischemia or divided into several shorter events. The tolerance paradigm of focal ischemic preconditioning followed by delayed severe ischemia may be more clinically relevant to human disease.

Evidence of a second, later window of tolerance has also been demonstrated in vitro. As mentioned above, cortical cultures preconditioned with oxygen–glucose deprivation were not protected from severe oxygen-glucose deprivation initiated 4 h following the preconditioning; however, these cultures were protected from oxygen–glucose deprivation initiated 24–48 h following the preconditioning event (Meller et al., 2005, 2006).

Evidence thus far indicates that neuronal survival in the delayed window is more robust and sustained than that observed in the rapid window. The possibilities contributing to the discrepancies between the rapid and delayed windows of tolerance are probably multiple. The requirement for de novo protein synthesis, mitochondrial protection, and increased oxidant buffering capacity have all been implicated in delayed tolerance but have been found lacking in rapid tolerance (Barone et al., 1998; Dave et al., 2001; Perez-Pinzon et al., 2002; Danielisova et al., 2005). Bearing in mind that the delayed window of tolerance yields sustained neuroprotection, we will focus the majority of the discussion in the mechanisms of ischemic tolerance to those observed 1–7 days following the preconditioning event.

9.2. Mechanism of ischemic preconditioning

Many ischemic tolerance models are preconditioned with non-ischemic inducers, such as hypoxia, spreading depression, hibernation, inflammation, oxidative stress, and epilepsy; we will discuss the effects of these paradigms in more detail later in the chapter. However, in order to better understand the mechanisms induced by different preconditioning paradigms, we will first take the opportunity to introduce in sufficient detail common mechanisms that have been found to contribute to the tolerance phenomenon induced by ischemic preconditioning.

The mechanisms of ischemic tolerance induced by brief ischemia appear complex, and our current understanding of it is fragmentary at best. Accumulating evidence indicates that it is an active process, involving unique changes in cellular gene expression, metabolic and signaling pathways and enzymatic activation. Interestingly, physiological factors that had been postulated to account for ischemic tolerance, such as alterations in cerebral blood flow or blood oxygen or blood glucose levels, were not significantly altered between ischemic preconditioned and control groups either before or during the initial phase of the subsequent ischemic insults (Matsushima and Hakim, 1995; Wiegand et al., 1999; Alkayed et al., 2002; Atochin et al., 2003). Recently, late phase increases in cerebral blood flow recovery were observed during severe ischemic insults in tolerant paradigms (Nakamura et al., 2005; Zhao and Nowak, 2006), likely as a reflection of the preservation of viable tissue. While tolerance paradigms correlate with a reduction in pathology, parallel pathways to protection are likely to be induced on both the molecular and biochemical, as well as cellular, levels.

In this section, we present current data on the biochemical and molecular mechanisms suggested to induce ischemic tolerance, focusing primarily on ischemia/ischemia paradigms (i.e. conditioning with a sublethal ischemic insult prior to a lethal ischemic insult). With these mechanisms in mind, the following section of this chapter will provide an overview of the various cross-conditioning paradigms and the potential mechanisms that may be responsible for these phenomena.

9.2.1. Alterations of gene expression

Induction of ischemic preconditioning in neuronal models has been associated with the robust alteration of expression of gene products, as evidenced both by genome-wide DNA arrays and directed gene expression studies. Here, we will first describe the overall genomic trends and second focus primarily on specific transcription factors implicated in the induction of tolerance.

In tolerance studies with focal cerebral ischemia, either ischemic preconditioning alone or severe ischemia alone caused pronounced upregulation of total gene expression; in contrast, severe ischemia combined with preconditioning downregulated overall transcription (Stenzel-Poore et al., 2003, 2004; Dhodda et al., 2004). Logical trends can be delineated when surveying the functional groups of altered gene products (Table 9.1 Stenzel-Poore et al., 2003). Following preconditioning alone, upregulated genes included mainly those involved in cellular metabolism and cell cycle regulation. Although the transcriptional response to damaging ischemia alone is also upregulated, the involved genes were mainly related to inflammation response, host defense, and tissue repair. Conversely, delayed suppression of a subset of genes occurred following either preconditioning alone or in combination with severe ischemia. The suppressed genes were functionally related, tending toward roles in metabolism, ion transport, and cell cycle control. Thus, the genomic response to lethal ischemia following ischemic preconditioning appears to be specifically reprogrammed compared with either ischemic preconditioning alone or severe ischemia alone, with a tendency to reduce cellular activity, including energy use and ion channel activities. This idea has been supported by limited direct observations; for example, potassium channel activity was suppressed after brief non-injurious oxygen–glucose deprivation (Lin et al., 2004).

Table 9.1.

Genomic trends following preconditioning and ischemia in brain

	Stress response	Metabolism	Transport	Cell cycle/death regulation
Preconditioning	↑	↑↓	↑↑↓	↑↑↓
Sever ischemia	↑↑↑↑	↑↑↑	↑↓	↑
Preconditioning + severe ischemia	↑↓	↑↓↓↓	↓↓	↑↓↓

Increased mRNA expression may be grouped based on known functions of the associated proteins.

Ischemic preconditioning, severe ischemia, and severe ischemia preceded by ischemic preconditioning yield trends in predicted functional groups. Source: adapted from Stenzel-Poore et al., 2003.

Many of the altered gene products, such as signal transducers, cell death and survival effectors, chaperones, antioxidants, and ion channels, will be mentioned specifically elsewhere in the chapter. However, marked alterations in activity and expression of transcription factors have also been identified following many preconditioning paradigms (Kawahara et al., 2004; Stenzel-Poore et al., 2004), suggesting that preconditioning may regulate groups of gene products via activation of common transcription factors. Several key transcription factors modulated following ischemic preconditioning are discussed below.

The cyclic adenosine monophosphate (cAMP) response element binding protein (CREB) often has been characterized as a key transcriptional factor for neuronal survival (Walton and Dragunow, 2000). Following ischemic preconditioning in several models, a robust and sustained activation of CREB was observed, evidenced by phosphorylation at Ser113 and promoter activity at the cAMP response element (Mabuchi et al., 2001; Nakajima et al., 2002; Hara et al., 2003; Lee et al., 2004; Meller et al., 2005). Interestingly, in several of these models, pre-treatment with a cAMP response element decoy oligonucleotide or CREB antisense oligonucleotides abolished the preconditioning-induced protection (Hara et al., 2003; Lee et al., 2004). Once activated, CREB led to upregulation of Bcl-2 (Pugazhenthi et al., 2000; Meller et al., 2005), an important gene product in the protection against cell death.

Both increased DNA-binding activity and nuclear translocation of nuclear factor (NF)-κB, a key regulator of apoptosis and inflammation, were demonstrated following preconditioning (Blondeau et al., 2001). Inhibition of NF-κB transactivation, by use of both pharmacological inhibitors and decoy oligonucleotides, abrogated the neuroprotection induced by ischemic preconditioning (Blondeau et al., 2001). NF-κB trans-activation has been linked to the upregulation of several key survival regulators, such as superoxide dismutase (SOD) or the inhibitor of apoptosis proteins (IAPs; Mattson and Meffert, 2006), both of which have also been demonstrated to be increased in ischemic tolerance models (Toyoda and Lee, 1997; Tanaka et al., 2004).

AP-1 constitutes a group of dimeric bZIP proteins, regulated by the mitogen-activated protein kinase (MAPK) signaling cascades (Shaulian and Karin, 2002). Following ischemic preconditioning, a rapid increase in AP-1 binding activity was observed within 3 hours, whereas severe ischemia demonstrated a delay in AP-1 binding (Kapinya et al, 2000). While the temporal characteristics of AP-1 DNA binding activity correlated with a possible rapid response during sublethal ischemia, the difference in response timing of AP-1 transactivation between severe and preconditioning ischemia has not been demonstrated to have a functional impact on the induction of tolerance versus a cell stress response.

Unlike transcriptional factors discussed above, p53 transactivation has been historically associated with promotion of cell death. While many transcription factors were found to be acutely upregulated following preconditioning, a subset were downregulated 24 h following the sublethal event, or the combined preconditioning/severe ischemic paradigm (Stenzel-Poore et al., 2004). p53 expression levels, when measured directly at the mRNA and protein (immunocytochemical) levels, were upregulated following either severe or sublethal ischemia alone (Tomasevic et al., 1999). However, following the preconditioning/severe ischemic paradigm, p53 expression levels decreased (Tomasevic et al., 1999; Mocanu and Yellon, 2003). Expression levels alone of transcription factors do not necessary translate to alteration of function, and p53 in particular is prone to very controlled regulation at multiple post-translational levels. The actual transactivational activity of p53 following preconditioning paradigms compared to ischemia alone has yet to be assessed.

Genomic analysis in ischemic preconditioning has provided a great deal of information regarding potential regulators of tolerance. However, the increased mRNA products may not necessarily be translated into proteins, particularly in the light of the inhibition of general protein synthesis following ischemia (Furuta et al., 1993). For example, at 24 hours after ischemic preconditioning, DNA array indicated increased expression of mRNAs that theoretically code for 30 proteins, but further protein arrays indicated that only five of these protein products were actually increased in expression (Dhodda et al., 2004). Thus, combined analysis of mRNA and protein expression levels may help to define processes and regulation of genomic responses to ischemic tolerance.

Overall, ischemic preconditioning has been demonstrated to induce distinct alterations in overall gene expression. These alterations may be in consequence to modified activation of transcription factors, affecting the transcription of gene families and coordinating functional responses. The future challenge in addressing the regulation of ischemic tolerance by altered gene expression will be to delineate how the precise regulations of the target genes function systemically.

9.2.2. Heat shock proteins

In addition to de novo gene expression, maintenance and correct folding of pre-existing proteins can severely influence the response of a cell to an altered environment. Heat shock proteins (hsps) are highly conserved among different species, functioning as molecular chaperones to ensure proper folding and glycosylation of cellular proteins. As part of the stress response, heat shock proteins are thought to serve primarily to buffer sudden accumulation of denatured proteins or protein aggregation under the stress condition and thus improve cell survival (Chen and Simon, 1997; Kirino, 2002). The roles of heat shock proteins in ischemic tolerance have been intensively studied but still remain controversial.

Early gene expression studies revealed increased expression of heat shock proteins following preconditioning paradigms. The expression of hsp 70 was found to be increased in vulnerable neurons following global ischemic preconditioning (Kirino et al., 1991; Liu et al., 1993; Nishi et al., 1993) and focal ischemic preconditioning (Chen et al., 1996; Dhodda et al., 2004). The time course of hsp 70 expression after preconditioning correlated with ischemic tolerance (Ohtsuki et al., 1993; Chen et al., 1996; Dhodda et al., 2004). Hsp27 was also demonstrated to be consistently upregulated in ischemic tolerance paradigms, primarily localized in astrocytes (Currie et al., 2000; Valentim et al., 2001, 2003; Dhodda et al., 2004; Nishino and Nowak, 2004).

However, the relevance of the expression of hsp 70 and hsp 27 in tolerance paradigms has been widely debated. Use of the anti-inflammatory agent quercetin was correlated with a loss of hsp 70 immunostaining and loss of tolerance, and infusion of an anti-hsp 70 antibody decreased neuroprotection in a global ischemia tolerance model (Nakata et al., 1993). More recent studies have presented strong evidence that hsp 70 may not be necessary for induction of ischemic tolerance. Using DC thresholds to accurately determine ischemic thresholds required for tolerance, hsp 70 mRNA was found to remain unchanged under conditions that still improved neuroprotection by tolerance (Abe and Nowak, 2000, 2004). Thus, under some experimental conditions, the induction of hsp 70 may be an epiphenomenon to a more severe preconditioning stimulus than the stimulus required for tolerance.

9.2.3. Endoplasmic reticulum stress

The endoplasmic reticulum is most characterized as the site of protein synthesis, folding, and many post-translational modifications. When large amounts of improperly folded proteins accumulate in the endoplasmic reticulum because of disruption of normal cellular homeostasis, a cellular stress response is activated (Rutkowski and Kaufman, 2004). This response includes the depression of general protein synthesis, increased production of endoplasmic reticulum chaperone proteins such as BIP, and enhanced ubiquitin-proteasomal function. If the stress overwhelms the response, however, cell death machinery is initiated (DeGracia and Montie, 2004; Rutkowski and Kaufman, 2004).

Cerebral ischemia is one major cause of endoplasmic reticulum stress in neurons (Chen et al., 1996; Ito et al., 2001; DeGracia and Montie, 2004; Zhang et al., 2004), contributing to glucose depletion, hypoxia, calcium-storage exhaustion, oxidative damage, and protein aggregation (Hu et al., 2000; Hayashi et al., 2003b; Kumar et al., 2003; Tajiri et al., 2004). Ischemic preconditioning also induced endoplasmic reticulum stress, indicated by the phosphorylation of PERK and eIF2α, transient protein synthesis inhibition and BIP upregulation (Chen et al., 1996; Hayashi et al., 2003a; Garcia et al., 2004). However, the recovery phase of depressed protein synthesis was shorter when severe ischemia was preceded by preconditioning compared to severe ischemia alone (Furuta et al., 1993; Nakagomi et al., 1993). Additionally, ischemic preconditioning greatly reduced protein aggregation in vulnerable neurons after global ischemia (Liu et al., 2005). These results suggest that, under tolerance paradigms, transient endoplasmic reticulum stress may lead to an increased capacity to buffer more severe endoplasmic reticulum stress. However, thus far these are simply observations; the possible consequences of specific signaling via endoplasmic reticulum stress in the induction of tolerance are currently unknown.

9.2.4. Cell death signaling

9.2.4.1. Caspases

The so-called ‘intrinsic’ cell death pathway is best characterized by the convergence of a cell death signal on to the mitochondria, leading to the release of certain mitochondrial proteins such as cytochrome c and Smac/DIABLO. Once in the cytosol, cytochrome c, along with dATP and apoptotic protease activating factor (Apaf)-1, form the apoptosome, which in turn recruits and activates caspase-9. In its active formation, caspase-9 further cleaves and activates effector caspase-3, typically resulting in a further amplification of the cell death signal and the execution phase (Graham and Chen, 2001; Liou et al., 2003; Zhang et al., 2004). Although other caspase pathways can initially activate caspase-3, the feedback and amplification via the intrinsic pathway has often been observed in cell death models. Caspase-3 activation has been found to be a major effector of ischemia induced cell death in the brain (Chen et al., 1998; Namura et al., 1998).

Ironically, caspase-3 was also activated following preconditioning alone (Garnier et al., 2003; McLaughlin et al., 2003; Tanaka et al., 2004). Cleaved caspase-3 fragments and caspase-3 activity were detec table 6– 24 hours following ischemic preconditioning in brain and in in vitro ischemia models (Garnier et al., 2003; McLaughlin et al., 2003). In these settings, activated caspase-3 did not induce apoptosis; on the contrary, it appeared to contribute to induction of ischemic tolerance. Treatment with a caspase-3 inhibitor led to the loss of the neuroprotective effect of preconditioning (Garnier et al., 2003; McLaughlin et al., 2003). Taken at face value, this may appear as a contradiction. However, recent work has elucidated cellular mechanisms geared toward balancing the activity of activated caspase-3. Specifically, proteins termed ‘inhibitor of apoptosis’ (IAPs) effectively bind to catalytically active caspases, particularly caspase-3, and suppress caspase activity (Shi, 2004). Conversely, Smac/DIABLO has been identified as a mitochondrial protein, which, when released into the cytosol, binds to IAPs, resulting in the disinhibition of active caspases and leading to execution of cell death (Du et al., 2000). Consistent with this mechanism of caspase activity modulation, severe ischemia induced increased caspase-3 activity and increased IAP expression and translocation of Smac/DIABLO into the cytosol (Tanaka et al., 2004). Severe ischemia preceded by preconditioning, however, led to increased caspase-3 activity and IAP expression but inhibited Smac/DIABLO translocation (Tanaka et al., 2004). Thus, following severe ischemia, caspase activity is allowed to proceed by Smac/DIABLO interference with IAP-mediated caspase inhibition, whereas following ischemia with preconditioning paradigm, IAPs effectively inhibit or control caspase activity.

While more controversial, caspase-3 may also contribute to ischemic tolerance via cleavage and subsequent inactivation of poly(ADP-ribose) polymerase (PARP) following preconditioning (Garnier et al., 2003). Although PARP has been described as contributing to DNA repair, its activation may also lead to the depletion of cellular NAD⁺ and ATP levels (Skaper, 2003). Activation of PARP has been identified following cerebral ischemia, and inhibition of PARP activity was found to be protective (Komjati et al., 2004; Chiarugi, 2005a; Ikeda et al., 2005; Zhang et al., 2005). Consistent with this, caspase-sized cleavage and inactivation of PARP were found following ischemic preconditioning (Garnier et al., 2003). Therefore, the possibility exists that – while IAPs suppresses caspase-3 activity enough to deter cell death following preconditioning paradigms – limited caspase-3 activity may target PARP for cleavage and inactivation, thereby reducing energy depletion.

While other caspase or proteolytic pathways may be altered following ischemic tolerance paradigms, their roles and regulation are less clear. Once thought to be a fairly simple cell death pathway, caspase activation and regulation has been revealed as a delightfully complex and dynamic network. Future studies on the alterations of caspase-signaling pathways in tolerance paradigms will hopefully reveal novel insights into the role of caspases in living, as well as dying, cells.

9.2.4.2. Bcl-2 family

Bcl-2 has been demonstrated to play a critical role in the regulation of cell death pathways converging on the mitochondria. The Bcl-2 family includes both promoters (e.g. Bax, Bid, or Bad) and suppressors of cell death (e.g. Bcl-2 or Bc1-xL). The regulation of these family members depends in part on the stoichiometry, localization, and interaction of the different members. Particularly in the case of the pro-cell-death Bcl-2 family members, activity can be suppressed by compartmentalization to the cytosol via ‘tethering’ or protein conformation. For example, the pro-cell death molecules Bax and Bad were found to be held in the cytosol by binding protein 14–3–3 (Wang et al., 1999; Nomura et al., 2003). Following cerebral ischemia, 14–3–3 was phosphorylated, causing the release of Bax, which translocated to the mitochondria and induced cytochrome c release (Gao et al., 2005). Under normal physiological conditions, Bid remains in the cytosol; however, when cleaved via caspase or calpain activation following ischemia, truncated Bid (tBid) may translocate to the mitochondria, destabilizing Bcl-2 and resulting in cytochrome c release and activation of the intrinsic cell death pathway (Plesnila et al., 2001; Yin et al., 2002).

In models of cerebral ischemic tolerance, the expression of Bcl-2 was upregulated following preconditioning (Nakatsuka et al., 2000; Shimizu et al., 2001; Meller et al., 2005), and bcl-2 antisense treatment decreased the neuroprotective effects of ischemic preconditioning (Shimizu et al., 2001). Consistent with upregulation of Bcl-2, cytochrome c release was suppressed following lethal ischemia in preconditioned brain (Nakatsuka et al., 2000; Tanaka et al., 2004). The role for other members of the Bcl-2 family has not been investigated, with conflicting reports of induction of Bcl-xL and Bax expression following preconditioning (Nakatsuka et al., 2000; Wu et al., 2003). However, perhaps more relevant than the protein expression levels, determination of the intracellular localization, dimerization, and conformation of the Bcl-2 family members may offer new insight into the functions of this family in the induction of ischemic tolerance.

9.2.4.3. Kinase signaling

Kinase signaling pathways have been repeatedly demonstrated to mediate signal transduction leading to cell survival, death, differentiation, and mitogenesis in a variety of models and cell types. Typically, kinase signaling pathways are built on a hierarchical system, where upstream kinases signal via phosphorylation on to preferred subordinate kinases. The result of the signaling cascade leads to the activation of a downstream effector-like kinase, which alters non-kinase protein function via phosphorylation, leading to alterations in transcription factors, proteases, and adaptor proteins. The kinase cascades themselves are regulated both positively, via phosphorylation by upstream kinases, and negatively, via dephosphorylation by phosphatases.

The MAPK pathways have been implicated in the regulation of tolerance. Extracellular-regulated kinase (ERK) and c-jun N-terminal kinase (JNK) are often characterized to have diametrically opposed functions, where ERK has been typically associated with survival or differentiation, and JNK with stress and death. Consistent with this, phosphorylated JNK was found to decrease in preconditioning paradigms, whereas phosphorylated ERK increased, compared to severe ischemia alone (Gu et al., 2001; Miao et al., 2005). Additionally, inhibition of the upstream kinase to ERK decreased the effects of preconditioning (Gu et al., 2001; Meller et al., 2005), indicating that MAPK pathways may participate in the induction of ischemic tolerance in brain.

Protein kinase C (PKC), with its many isoforms, has also been implicated in the neuroprotection afforded by tolerance, primarily argued to occur as a downstream event to activation of either the N-methyl d-aspartate (NMDA) receptor or adenosine A₁ receptor, with regulation by the MAPK pathway (Lange-Asschenfeldt et al., 2004). Variations in the localization of different PKC isoforms have been noted following sublethal ischemia (Kurkinen et al., 2001b) but the functional significance of this finding was not directly ascertained. Furthermore, the actual involvement of PKC isoforms in the induction of ischemia-induced tolerance in the brain is widely contested. Interestingly, isoform-selective PKC activation does appear to correlate with many cross-conditioning paradigms, such as spreading depression (Kurkinen et al., 2001a).

Akt, activated by its upstream kinase, phosphoinositol-3-kinase (PI3K), has been described to play a critical role in the suppression of cell death and may contribute to ischemic tolerance. Several studies have found that the expression level of phosphorylated Akt was sustained at higher levels following ischemic tolerance paradigms compared to transient activation following severe ischemia alone, and that inhibitors of PI3K decreased neuroprotection of preconditioning (Nakajima et al., 2004). However, other reports have found decreased levels of phosphorylated Akt in tolerance models and no effect of PI3K inhibitors on the efficacy of preconditioning to induce protection (Namura et al., 2000; Meller et al., 2005). Significant differences existed in the ischemic paradigms used, perhaps contributing to the inconsistencies.

Although these studies indicate that kinases may be involved in ischemic tolerance, the mechanisms and regulation are currently unexplored. Within the past few years, powerful molecular tools have been exploited to further reveal the complex regulation and effects of kinase signaling. Particularly in the context of their role as rapid signal transducers, the effects of preconditioning – both in the delayed and rapid windows – on the control of kinases via phosphorylation and dephosphorylation may yield interesting insights.

9.2.5. Nitric oxide

The mechanism of nitric oxide (NO) in inducing ischemic tolerance remains unclear, partly because of the nature of NO itself as both a signaling molecule activating multiple pathways, and as a neurotoxin in and of itself. NO is produced in the brain by three different synthases – eNOS, iNOS and nNOS – all of which have been implicated in neuronal protection or toxicity following ischemia. The synthase generating the production of NO appears to influence the functional consequences of NO under ischemic conditions (Warner et al., 2004). However, much of the functions of NO may also be dependent on its redox state, and have recently been identified as spatially and temporally controlled (Hess et al., 2005). While NO itself can act as a free radical generating reactive oxygen species and exacerbating oxidative stress, signaling by NO via S-nitrosylation of target proteins can modify the function of proteins related to ischemic toxicity. Given the multifarious nature of NO synthesis and signaling, involvement of NO in ischemic preconditioning is likely to be diverse (Huang, 2004).

In tolerance models, eNOS, nNOS and iNOS have all been implicated in the development of ischemic tolerance under various conditions and settings, primarily stemming from data employing pharmacological inhibitors or knockout models (Nandagopal et al., 2001; Kawahara et al., 2004). iNOS inhibition (by aminoguanidine, a semi-selective iNOS inhibitor) blocked protection from MCAO by global ischemic preconditioning, a finding confirmed by the observation that ischemic tolerance could not be induced in iNOS knockout mice (Cho et al., 2005). Ischemic preconditioning also failed to protect against MCAO in eNOS and nNOS knockout mice (Atochin et al., 2003). Use of a nonselective NOS inhibitor blocked induction of ischemic tolerance in neonatal hypoxia-ischemia (Gidday et al., 1999). Thus, although the mechanism is still unclear, NO and its synthesizing enzymes appear to contribute to the induction of ischemic tolerance.

The events surrounding NO synthesis, including initiation and downstream effects, are more widely debated. In ischemic tolerance settings, nNOS-mediated NO synthesis has been found to increase ERK activity via Ras/Raf signaling (Gonzalez-Zulueta et al., 2000; Nandagopal et al., 2001), a signaling pathway also found to be necessary to induce tolerance in some tolerance models (Gu et al., 2001; Meller et al., 2005). Given that ERK activation is highly involved in directed gene product synthesis through activation of transcription factors, the activation of the Ras/Raf signaling cascade may be an example of NO involvement in ischemic tolerance via altered gene expression. However, the same pathway has also been shown to activate PI3K and Akt, which, along with ERK, also have targets that post-translationally modify function with rapid effects on cell death or survival, such as the phosphorylation of Bad and other components of cell death pathways (Amaravadi and Thompson, 2005). Therefore, the activation of the Ras/Raf signaling cascade by NO may be a mechanism of NO involvement in a multipronged approach to the induction of tolerance.

NO-mediated alteration of cell death signaling has also been argued to be a direct and rapid means of neuroprotection (Lipton, 1999). NO generation leads to increased S-nitrosylation of target cysteine residues in proteins (Lipton, 1999). In one example, NO was found to attenuate the activity of the NMDA receptor by nitrosylating Cys-399 in its NR2A subunit (Kim et al., 1999; Choi et al.,, 2000), thus potentially decreasing glutamate-stimulated, NMDA-receptor-mediated calcium influx to a subsequent challenge. Furthermore, activity of caspases has been found to be altered by direct nitrosylation, thus implying that NO may directly target and modulate cell death enzymes (Tenneti et al., 1997; Lipton, 1999). However, these possibilities have not been directly investigated in ischemic tolerance settings.

9.2.6. Oxidative stress

While normal physiological functions produce low-level basal oxidative stress, stimuli, such as ischemia, increase this stress to levels that can overwhelm cellular defense mechanisms. Enzymes (e.g. SOD or catalase) and low-molecular-weight molecules (e.g. glutathione, ascorbic acid, or vitamin E) comprise two major classes of endogenous mechanisms to neutralize reactive oxygen species. Both antioxidant classes have been found to increase in ischemic tolerance paradigms.

As will be discussed further under the cross-conditioning models, the induction of the antioxidant defenses may actually be dependent on transient sublethal oxidative stress. The more abstract theory is that low-level oxidative stress itself may be required for induction of tolerant pathways, in that the ‘charging’ of a cell to arm itself with an increased supply of antioxidants or DNA repairing capacity may contribute to the ability of a cell to cope with additional stress. Supporting this hypothesis, SOD and catalase expression levels and activity were demonstrated to be latently upregulated in neurons following ischemic preconditioning stimuli (Kato et al., 1995b; Danielisova et al., 2005), as well as several other cross-conditioning paradigms. Furthermore, a separate study demonstrated that the neuroprotective effects of preconditioning were diminished by pharmacological free radical scavengers and antioxidants (Puisieux et al., 2004). It is worthwhile to mention, however, that while preconditioning was blocked by antioxidants, the same study did not find evidence of increased SOD or glutathione peroxidase (Puisieux et al., 2004). However, given the multiple cellular antioxidant defenses, as well as the variability between ischemic conditioning paradigms, different oxidant-sensitive pathways may be elicited. More studies to detect whether free radical scavengers abrogate preconditioning-mediated neuroprotection may be warranted.

In addition to the larger enzymes described above, smaller antioxidants (e.g. uric acid) have also been found to increase rapidly following preconditioning (Glantz et al., 2005). Because these small molecules do not require new protein synthesis to respond quickly to oxidative stress, the mobilization of small antioxidants may decrease reactive oxygen species accumulated during subsequent severe ischemia, even within a rapid window of tolerance (Glantz et al., 2005). Additionally, metal scavenging enzymes were upregulated following ischemic preconditioning. Metallothioneins, which act to buffer both reactive oxygen species and zinc levels, were increased following sublethal ischemia (Dhodda et al., 2004). However, studies thus far have only described the expression, not function, of these molecules in inducing ischemic tolerance.

While much of the focus of antioxidant defenses has been based on neuronal capacity, the role of astrocytes in scavenging reactive oxygen species may factor into neuroprotection in ischemia and ischemic tolerance. Astrocytes have been found to often contain higher levels of antioxidants, such as metallothioneins, and were found to be able release glutathione into the extracellular space (Dringen and Hirrlinger, 2003), perhaps allowing buffering of oxidants outside of the neuron itself. Furthermore, co-cultures of neurons with astrocytes increased the glutathione content in neurons, presumably by increased release precursors necessary for glutathione synthesis (Dringen et al., 1999). However, these relationships between astrocytes and neurons have not been directly investigated in ischemic preconditioning paradigms.

9.2.7. DNA repair capacity

Oxidative DNA damage is a prominent feature of ischemic brain injury and has been detected in neurons and astrocytes following cerebral ischemia. If left unchecked, the damage eventually leads to DNA fragmentation and triggers neuronal death (Liu et al., 1996; Chen et al., 1997; Lan et al., 2003). DNA damage may occur via direct attack by reactive oxygen species, forming hydroxyl radical-modified bases, apurinic/apyrimidinic abasic sites, single-strand break, and double-strand breaks (Barzilai and Yamamoto, 2004). These modifications of DNA lead to gene loss of function, decreased transcription, and increased need for energy to induce repair (Janssen et al., 1993; Cui et al., 1999; Yu et al., 2003). Additionally, the presence of either apurinic/apyrimidinic abasic sites or DNA strand breaks can lead to activation of the transcription factor p53, which then contributes to cell death by upregulating several pro-cell death molecules, including Bax, PUMA and Noxa (Oda et al., 2000; Nakano and Vousden, 2001; Mihara et al., 2003). The accumulation of DNA single-stranded breaks may also activate PARP signaling pathways, leading to NAD⁺ and ATP depletion of apoptosis-inducing factor-dependent cell death (reviewed by D’Amours et al., 1999).

The predominant mechanism of repair for oxidative DNA damage is base excision repair. During this process, the damaged DNA bases are first probed by DNA glycosylases, which then cut the N-glycosylic bond to remove the damaged base; this allows an AP endonuclease to excise the remaining deoxyribose. DNA polymerase then adds the correct base into the remaining gap, which is then ligated to the pre-existing bases by DNA ligases (Krokan et al., 2000; Lan et al., 2003; Li et al., 2006). Base excision DNA repair may also be mediated by the DNA-dependent protein kinase (DNA-PK), comprised of a catalytic subunit and several DNA binding subunits (Ku).

Ischemic preconditioning has been shown to activate the base excision repair pathway, as indicated by significantly increased enzymatic activities of 8-oxodG glycosylases, AP endonuclease, and DNA polymerase-β (Li et al., 2006). Increases in base excision repair enzyme expression and activity occurred only where ischemic tolerance was conferred. In contrast, in anatomical areas where ischemic tolerance was not observed, no increase in base excision repair activity was observed. Increased mitochondrial base excision repair activity was also demonstrated following brief, sublethal ischemia (Chen et al., 2003), indicating that it is possible that mitochondrial base excision repair may also be an element in preconditioning paradigms. Upregulation of the Ku subunits of DNA-PK was noted following preconditioning ischemia; the expression pattern overlapped with the onset of tolerance (Sugawara et al., 2001).

Taken together, ischemic preconditioning may reduce oxidative DNA damage by increasing certain DNA repair activity after ischemia, and therefore reduce subsequent DNA-fragmentation, energy depletion, and p53 activation.

9.2.8. Adenosine accumulation

Adenosine, a metabolic product of ATP, was found to accumulate during cerebral ischemia, presumably playing a neuroprotective role. The physiological functions of extracellular adenosine are highly transient, as adenosine undergoes rapid reuptake and metabolism. Extracellular adenosine acts directly on four subgroups of receptors – A₁,A_2A, A_2b, and A₃ – which confer diverse effects on synaptic transmission (Pearson et al., 2003). However, the overall physiological effects of A₁ activity have been often correlated with a reduction in synaptic activity (Pugliese et al., 2003); this reduction has been hypothesized to curb the spread of excitotoxicity. Furthermore, adenosine was found to indirectly activate the ATP-sensitive potassium (K_ATP) channel, perhaps through the activation of the A₁ receptor (Heurteaux et al., 1995). The opening of potassium channels has been associated with overall cellular hyperpolarization, although much of this work is in non-neuronal cells (Fujiwara et al., 1987; Yamada and Inagaki, 2005). Interestingly, these channels are localized both on the cellular surface as well as the mitochondrial membrane

The role of adenosine in cerebral ischemic tolerance has been supported by two lines of evidence. First, application of adenosine uptake inhibitors or adenosine A₁ receptor agonists have been found to induce ischemic tolerance (Von Lubitz et al., 1994; Kawahara et al., 1998); secondly, antagonists of the adenosine A₁ receptor blocked ischemic tolerance induced by ischemic preconditioning (Hiraide et al., 2001; Nakamura et al., 2002; Pugliese et al., 2003; Yoshida et al., 2004). Similarly, inhibition of mitochondrial K_ATP channels also blocked the effects of preconditioning in the brain when administered before severe ischemia (Yoshida et al., 2004). Pointing to the differences in physiological effects of adenosine, A₂ and A₃ receptors were found either to worsen subsequent ischemia or have no effect on the induction of tolerance (Pugliese et al., 2003).

Methods of quantifying ischemic preconditioning by virtue of depolarization have resulted in higher stringency and reproducibility of the model (Sorimachi and Nowak, 2004). Using this sort of quantification of intensity of preconditioning, more recent studies have not found effects of A₁ inhibitors, A₁ agonists, or a K_ATP channel inhibitor on the induction of ischemic tolerance (Sorimachi and Nowak, 2004). However, this study used a general K_ATP channel inhibitor. Other data have demonstrated that mitochondrial K_ATP channel openers establish preconditioning that is blocked by mitochondrial, but not general, K_ATP channel inhibitors (Kis et al., 2003). Thus, once thought to have been proved to unequivocally mediate induction of cerebral tolerance, the role of adenosine and its downstream effects may now need to be re-examined.

9.2.9. Inflammation

Following ischemia – both sublethal and severe – evidence of inflammation has been observed. In particular, increased expression levels and release of the soluble form of tumor necrosis factor (TNF-α) was detected following ischemic preconditioning (Romera et al., 2004). Arguing for a role beyond an epiphenomenon, ischemic tolerance has been found to be induced by the application of TNF-α) itself (Nawashiro et al., 1997), and application of a TNF-α) antibody or antisense during ischemic preconditioning blocked neuroprotection against subsequent severe ischemia (Romera et al., 2004). Additionally, increased activity of the upstream convertase of TNF, TACE, was demonstrated following ischemic preconditioning, leading to increased formation of active TNF-α)

The consequences of TNF-α) induction are poorly understood; under severe ischemic conditions, both detrimental and protective roles have been argued (Barone and Feuerstein, 1999). Application of TNF-α) antisense or a TNF-α) inactivating antibody during in vitro ischemic preconditioning prevented both the decreased glutamate release and the increased expression of the neuronal excitatory amino acid transporter 3 (EAAT3) normally observed following preconditioning (Romera et al., 2004), although the physiological relevance of glutamate levels during tolerance paradigms has been widely debated. Although TNF-α) may be involved in ischemic tolerance, the mechanism remains unclear.

9.2.10. Observations with glial connections

While much of the attention in central nervous disorders has focused on the dysfunction and alterations of the neuron itself, a significant literature is currently emerging highlighting the role of the glia in mediating pathogenesis and neuroprotection. Although the mechanistic function of glia in ischemic tolerance paradigms is virtually unexplored, significant advances in models of severe ischemia alone have suggested that glia, in particular microglia and astroglia, may be worthwhile targets for investigation in preconditioning paradigms.

As the major type of cells involved in the inflammatory response to brain injury, microglia are essential for the elimination of debris of damaged cells and tissues (Dirnagl et al., 2003; Kariko et al., 2004). Microglia were found to be activated following severe ischemia (Kato et al., 1995a). However, following both rapid and delayed tolerance paradigms, reduced numbers of microglia were present during the phase of protection (3 days following severe ischemia) in vulnerable areas (Kato et al., 1995a; Perez-Pinzon et al., 1999; Liu et al., 2001). The question of whether suppression of microglia is an active process or is simply correlated with less severe injury remains to be determined.

It is tempting to speculate on the role of astroglia in ischemic tolerance. As an extensive network of cells linked electrochemically by gap junctions, astrocytes have been found to have complex and dynamic communication with both neurons and vascular cells. By forming connections with neurons and microvasculature, astrocytes are effective sensors and regulators of environmental changes, maintaining ion, glutamate, and water homeostasis, supplying energy substrates, guiding axonal repairs, scavenging free radicals, and releasing growth factors and cytokines. As discussed above, these activities are also altered or induced following ischemia and ischemic preconditioning. For example, observations of increased mRNA of trophic factors, such as brain-derived neurotrophic factor (BDNF), nerve growth factor (NGF) and epidermal growth factor (EGF), have been reported following ischemic preconditioning (Truettner et al., 2002; Naylor et al., 2005). Furthermore, conditioned media from ischemic astrocytes was protective against oxygen– glucose deprivation in neurons (Ruscher et al., 2002). The protective effects of the glial conditioned media were blocked in the presence of an erythropoietin receptor antagonist, suggesting the erythropoietin is released by astrocytes and induces protection via neuronal receptors (Ruscher et al., 2002). Given that astrocytes are a major source of trophic factors in the brain, it is possible that glia may play a protective role via trophic support, although this has not been tested directly. Thus, studies incorporating the contributions of astrocytes to the induction of tolerance will probably yield an exciting and dynamic depth to our understanding of the effects of ischemic preconditioning.

9.2.11. Summary

The preceding sections have briefly outlined several of the potential mechanisms leading to the induction of ischemic tolerance (Fig. 9.2). However, the interaction between many of these observations remains obscure. While one could imagine a direct connection between low-level oxidative stress attacking DNA and subsequent increased repair capacity in addition to antioxidant defenses, interactions with kinase signaling pathways or energetic resources are less clear. Kinase signaling is highly sensitive to oxidative stress via inactivation of phosphatases (Chiarugi, 2005b) and switching to lactate as an energy source may result in decreased basal oxidative stress generated from metabolism. However, these ideas are only hypothetical and have not been directly examined in ischemic tolerance paradigms. Furthermore, intercellular communication and extraneuronal conditions are parameters that are only beginning to be investigated in ischemic conditions alone; it will be interesting to see what directed studies in tolerance paradigms will yield.

Fig. 9.2.

Several mechanisms involved in ischemic tolerance. (A) A normal neuron. (B) Following sublethal ischemia alone, the accumulation of adenosine and activation of A1 receptor may stimulate K_ATP channels. The expression of chaperone proteins and antioxidants is upregulated, increasing cellular antistress activity. Small-scale DNA damage induces the expression and activation of DNA repair enzymes; the antiapoptotic members of Bcl-2 family and IAPS are upregulated. (C) After severe ischemia alone, excitotoxicity causes overload of intracellular calcium through the NMDA receptor. Unfolded, misfolded, and damaged proteins aggregate within cells and endoplasmic reticulum stress occurs, free radicals accumulate, leading to prominent DNA damage. Proapoptotic Bcl-2 family members overwhelm antiapoptotic members, resulting in mitochondrial release of cytochrome c, SMAC/Diablo, and subsequent caspase-3 activation. (D Following severe ischemia preceded by ischemic preconditioning, upregulated chaperones may attenuate the protein aggregation and endoplasmic reticulum stress induced by ischemia, and the increased level of antioxidants decreases oxidative damage. Similarly, the increased DNA repair capacity keeps DNA intact. While caspase-3 is activated, its activity is suppressed by IAPs, and SMAC/Diablo is retained in the mitochondria.

The mechanisms of neuroprotection afforded by ischemic preconditioning are unclear but reveal interesting trends – increased buffering capacity for dealing with oxidative stress and DNA damage, suppression of cell death signaling, and induction of protective gene products. Are all these characteristics of ischemic preconditioning required? Could there exist alternative pathways? As we shall discuss in the following section, evidence from cross-conditioning paradigms implies that there are potentially several common or redundant mechanisms that lead to ischemic tolerance.

9.3. The phenomenon of cross-preconditioning

Cross-preconditioning refers to prior exposure to a noxious stress other than ischemia that confers ischemic tolerance. It is not surprising that cross-preconditioning exists in the brain, considering that many CNS disorders may induce common components of molecular pathways in response to environmental stimuli. Brief episodes of various stresses can trigger similar tolerance pathways that counteract the initiation of cell death pathways by subsequent ischemia. Evidence of such cross-preconditioning has been demonstrated by hyperthermia and hypothermia, hypoxia, cortical spreading depression, hibernation, inflammation, oxidative stress, inhibition of glucose metabolism, low dose thrombin, epilepsy, and traumatic brain injury (Table 9.2) (Chen and Simon, 1997; Kirino, 2002; Dirnagl et al., 2003). While none of these stressors impinge on the cell in exactly the same fashion, we will discuss some common themes related to the mechanisms described above, as well as mentioning some less well understood but still interesting paradigms.

Table 9.2.

Cross-conditioning stimuli which induce ischemic tolerance

Preconditioning stimulus		Tolerance to	Examples in the literature
Ischemia	Focal	Focal, global	Glazier et al., 1994; Miyashita et al., 1994; Matsushima and Hakim, 1995; Chen et al., 1996
	Global	Focal, global	Kitagawa et al., 1990; Kato et al., 1991; Simon et al., 1993
	OGD*	OGD	Bruer et al., 1997; Khaspekov et al., 1998
Temperature	Hyperthermia	Global	Chopp et al., 1989; Nowak et al., 1990
	Hypothermia	Focal, global	Nishio et al., 1999, 2000; Yunoki et al., 2002
	Hibernation	OGD	Frerichs and Hallenbeck, 1998; Drew et al., 2002
Hypoxia	Hypoxia	H/I, focal	Vannucci et al., 1998; Gidday et al., 1999; Bernaudin et al., 2002b; Prass et al., 2003
Cortical spreading depression	Cortical spreading depression	Focal, global	Kobayashi et al., 1995; Matsushima et al., 1996
Inflammation	LPS	Focal	Bordet et al., 2000; Puisieux et al., 2000; Furuya et al., 2005
	TNF-α	Focal, OGD	Nawashiro et al., 1997; Liu et al., 2000
	IL-1β	Global	Ohtsuki et al., 1996
Chemical	DEDC	Global	Ohtsuki et al., 1992
	3-NPA	Focal, global, OGD	Wiegand et al., 1999; Sugino et al., 2000; Kis et al., 2003
	Diazoxide	OGD	Kis et al., 2003
Others	Epilepsy	Global	Plamondon et al., 1999
	Trauma	Global	Takahata and Shimoji, 1986; Otori et al., 2004

^*OGD represents in vitro models.

3-NPA, 3-nitropropionic acid; DEDC, diethyldithiocarbamate; IL, interleukin; LPS, lipopolysaccharide; OGD, oxygen–glucose deprivation; TNF, tumor necrosis factor.

9.3.1. Overview of cross-conditioning paradigms of ischemic tolerance

Tolerance to severe ischemic stress can be induced by non-ischemic preconditioning stimuli. While the end effect is neuroprotection following severe ischemia, cross-conditioning is distinct from prophylactic treatment in that the phenomenon describes the sublethal occurrence of a stressor. The inducers of tolerance typically relate to specific characteristics involved in ischemic preconditioning. Metabolic modulation, such as hibernation or caloric restrictions, may induce tolerance by alteration of blood flow or decreased cellular metabolism. Oxidative stressors, such as hypoxia or mild inflammation, may serve to strengthen the cellular defenses necessary to counter a severe insult. Mild alteration of similar anatomical structures, via epilepsy or cortical spreading depression, perhaps leads to a heightened cellular awareness of irregularities in the local environment. While this list is by no means comprehensive, and the mechanisms are poorly understood, it provides us with a base for understanding some common trends in the induction of ischemic tolerance by non-ischemic stressors.

9.3.2. Common trends

9.3.2.1. Inflammation buffering

The concept of increasing buffering capacity against a subsequent severe insult initially appears somewhat controversial. Both sublethal oxidative and inflammatory stressors are postulated to induce an antioxidation and anti-inflammatory response, positioning the cells to respond more quickly to subsequent oxidative or inflammatory stressors, such as ischemia. Inflammation is typically induced by molecules derived from damaged and dying cells and tissues. The release of such molecules stimulates the Toll-like receptors (TLR) on microglia/macrophage membranes, inducing the secretion of inflammatory cytokines such as TNF-α, interleukin (IL)-1β, and IL-6 (Kariko et al., 2004). Although the physiological function of inflammation is to eliminate damaged cells and tissues and to restore normal function, it may also exacerbate ischemic brain injury. Cerebral ischemia – severe or sublethal – has been found to induce an inflammatory response in the surrounding tissue (Han and Yenari, 2003).

Ischemic tolerance can be induced with proinflammatory agents as the preconditioning stimuli; however, the mechanism in the brain is obscure. Preconditioning with hypoxia, TNF-α or lipopolysaccharide, a bacterial endotoxin, induced increased ceramide levels but resulted in attenuation of subsequent severe ischemia (Nawashiro et al., 1997; Bordet et al., 2000; Liu et al., 2000; Furuya et al., 2001; Zimmermann et al., 2001). Ceramide, a signaling phospholipid activated by TNF-α, was found to be necessary in hypoxia-induced ischemic tolerance, as inhibition of ceramide synthase blocked neuroprotection (Liu et al., 2000). Furthermore, preconditioning with either ceramide or IL-1, both inflammatory signaling messengers, was capable of inducing ischemic tolerance (Ohtsuki et al., 1996; Liu et al., 2000). Cortical spreading depression was also found to transiently increase mRNA expression of the proinflammatory cytokines, TNF-α and IL-1β; cortical spreading depression has also been well characterized to induce ischemic tolerance (Jander et al., 2001). The inflammatory response following cortical spreading depression has been correlated with the activation of microglia (Gehrmann et al., 1993). Following cortical spreading depression, microglia were co-localized with IL-1β staining, suggesting glia as a source for inflammatory cytokine activation (Jander et al., 2001). Even though preconditioning itself often led to transient increase of cytokines, following severe ischemia preconditioned by either ischemia or 3-nitropropionic acid, the proinflammatory cytokines IL-1β and IL-6 were decreased at the mRNA expression level (Pera et al., 2004). This suggests a subsequent attenuation of the inflammatory response following severe ischemia.

It is unclear whether less damage occurs due to a heightened capacity to quench inflammation, or less inflammation occurs due to less damage. An interesting hypothesis was proposed that the decreased inflammation in tolerant ischemia could be due to negative feedback onto and disruption of the inflammatory pathway via signaling pathways, inactivated receptors, plasma membrane disruption, or anti-inflammatory molecules (Kariko et al., 2004). Analogous to the findings of oxidative stress inducing an increased oxidant buffering capacity, sublethal inflammation may be involved in tolerance induction.

9.3.2.2. Upregulation of antioxidants

Several examples of the upregulation of antioxidants have been observed in induction of ischemic tolerance. Consistently, preconditioning by oxidative stress (e.g. hypoxia) both inhibited subsequent severe ischemia and was associated with upregulated antioxidant enzymes, including metallothioneins and MnSOD (Bernaudin et al., 2002b; Arthur et al., 2004; Freiberger et al., 2006). The mitochondrial succinate dehydrogenase inhibitor 3-nitropropionic acid was capable of inducing ischemic tolerance, and increased free radical production prior to the onset of ischemia (Kis et al., 2003). Inhibition of SOD with diethyldithiocarbamate induced tolerance associated with upregulation of MnSOD (Ohtsuki et al., 1992). Diazoxide preconditioning, which stimulates mitochondrial K_ATP channel opening, also resulted in increased levels of reduced glutathione associated with ischemic tolerance (Kis et al., 2003). While the mechanisms of hibernation-induced tolerance are not well understood, it is worthwhile to mention that, in addition to decreased blood flow and altered gene expression, hibernation also was associated with increased levels of ascorbate, a small-molecule antioxidant (Drew et al., 2002). Similarly, hypoxic preconditioning was correlated with increased expression of thioredoxins, another subgroup of small-molecule antioxidants, as well as components necessary for the synthesis of glutathione (Stroev et al., 2004). Additionally, an endotoxin derivative, diphosphoryl lipid A, was also able to induce ischemic tolerance, associated with upregulation SOD (Toyoda et al., 2000). Induction of antioxidants, perhaps via low-level increases in reactive oxygen species, may thus represent a possible common mechanism shared among a variety of tolerance inducers.

9.3.2.3. Gene expression

Genomic expression patterns have not been widely compared across different inducers of ischemic tolerance. Hypoxic preconditioning induced the expression of a distinct set of genes. HIF-1α, a transcription factor strongly induced by hypoxia (Semenza, 2001), was found to be activated following hypoxic preconditioning paradigms in the neonatal rat (Bernaudin et al., 2002a; Wang et al., 2004). The effects of HIF-1α activation via synthesis of its downstream gene products could well be multifunctional, ranging from vascular alterations by expression of erythropoietin, tissue plasminogen activator and vascular endothelial growth factor (VEGF), to increased glucose transport and utilization by expression of GLUT-1 and phosphofructokinase (Bernaudin et al., 2002a). In hypoxic preconditioning paradigms in both the neonate and adult, HIF-1α-mediated molecules, such as erythropoietin, VEGF, GLUT-1, tissue plasminogen activator, adrenomedullin and prolyl 4-hydroxylase α, were increased at the mRNA level (Jones and Bergeron, 2001; Bernaudin et al., 2002a; Prass et al., 2003; Tang et al., 2006). HIF-1α pharmacological activators alone could mimic the protection of hypoxic preconditioning (Bergeron et al., 2000), while erythropoietin inhibitors abolished the effects of preconditioning (Ruscher et al., 2002; Prass et al., 2003). Although HIF-1α activation has been identified and implicated in neuroprotection afforded by the hypoxic preconditioned brain, its role in more general models of preconditioning has not been well addressed.

9.3.2.4. Heat shock protein induction

Hsp 70, described above as a major stress-induced chaperone, can be activated by a variety of sublethal conditions that also may be used to induce tolerance. Hyperthermia, the classic inducer of heat shock proteins, trauma, calorie restriction, and hypoglycemia have all been demonstrated to induce ischemic tolerance and were associated with the upregulation of hsp 70 (Chopp et al., 1989; Nowak et al., 1990). While hsp 70 expression can be induced by a multitude of factors, its presence is not necessary for induction of tolerance by all non-ischemic preconditioning. For example, preconditioning with either 3-nitropropionic acid (an irreversible inhibitor of succinate dehydrogenase) or the inflammatory agent lipopolysaccharide did not induce increased heat shock protein expression but were still potent inducers of ischemic tolerance (Puisieux et al., 2000; Kato et al., 2005). Therefore, while the upregulation of heat shock protein is associated with many inducers of tolerance, it does not appear to be necessary.

9.3.2.5. Kinase signaling for survival

As described earlier, ERK and PKC are commonly associated with cellular survival pathways. Consistent with this concept, several inducers of ischemic tolerance are associated with increased ERK or PKC activation. Cortical spreading depression induced both a transient increase in ERK phosphorylation and overexpression and phosphorylation of PKCδ (Kurkinen et al., 2001a; Chow et al., 2002; Horiguchi et al., 2005). The protection afforded by preconditioning with diazoxide against in vitro ischemia was found to both be correlated with and dependent on PKC, as inhibition of PKC with staurosporine blocked the preconditioning effect (Kis et al., 2003).

9.3.2.6. Physiological parameters

Blood flow and brain temperature are both physiological factors that have been proposed to modulate severity of stroke by suppression of metabolism and reduced stress during the reperfusion (i.e. injury) phase (Perez-Pinzon et al., 2005). However, cerebral blood flow has been found to be unaltered at acute phases following severe ischemia in tolerance paradigms (Chen et al., 1996; Barone et al., 1998; Dawson et al., 1999; Alkayed et al., 2002); increases in cerebral blood flow were observed at later timepoints, perhaps reflecting the recovery of blood flow into preserved tissue (Nakamura et al., 2005; Zhao and Nowak, 2006). In contrast to ischemic preconditioning, several studies using cross-conditioning paradigms have demonstrated that alteration of cerebral blood flow may be a component of induction of ischemic tolerance. Preconditioning with cortical spreading depression or localized cortical lesion decreased cerebral blood flow acutely following subsequent MCAO (Otori et al., 2003; Muramatsu et al., 2004). Lipopolysaccharide-induced ischemic tolerance was found to produce subtle alterations of the microvasculature in both the penumbra and ischemic core (Dawson et al., 1999), and possibly increased blood flow in the periinfarct area (Furuya et al., 2005), probably reflective of preservation of viable tissue. While other biochemical or molecular effects of both lipopolysaccharide and cortical spreading depression preconditioning may be more relevant for the induction of ischemic tolerance, vascular alterations may underlie metabolic effects on basal cellular activities. Thus, hypothermic temperatures likely lead to relatively long-lived molecular or biochemical alterations, even after a return to physiological temperature.

Brain temperature was initially found to induce neuroprotection because of the observation that cerebral temperatures spontaneously decrease following global ischemia, leading to reduced infarct size (Traystman, 2003). Consistent with this observation, preconditioning by hypothermia or hibernation – both of which reduce cerebral temperature – lead to neuroprotection against subsequent severe ischemia (Nishio et al., 1999, 2000; Drew et al., 2001; Yunoki et al., 2002). However, the hypothermia associated with hibernation is not necessary for inducing tolerance, as organotypic slices from hibernating squirrels were still tolerant to severe ischemia, even at normothermic temperatures (Frerichs and Hallenbeck, 1998).

9.3.2.7. Lactate accumulation

Lactate is perhaps best characterized for its involvement in energetic states and cellular metabolism. However, increased concentrations of lactate in the brain is a common phenomenon following various types of preconditioning, including hypoxia, spreading depression, and glutamate (Scheller et al., 1992; Schurr et al., 2001). Once thought to be more of an epiphenomenon, recent discoveries have implicated lactate as a possible intercellular mediator of neuroprotection. Lactate transport inhibition by a monocarboxylate transporter inhibitor abolished neuroprotection induced by sublethal hypoxia or glutamate preconditioning, leading to the hypothesis that effective preconditioning requires higher levels of lactate to be shuttled intercellularly via monocarboxylate transporter (Schurr et al., 2001). A role for lactate may also exist in ischemic preconditioning. Lactate was able to support synaptic activity in the absence of glucose in preconditioned hippocampal slices, indicating that ischemic preconditioned neurons were able to employ lactate as an alternate energy substrate (Kitano et al., 2002). Thus, future studies may focus on a possible intercellular – perhaps involving astroglia – supply of lactate as an energy source to induce tolerance.

9.3.2.8. Glial involvement

As mentioned above, the involvement of glia in neurodegenerative diseases is relatively unexplored. As in ischemic preconditioning, microglial activation was decreased following TNF α- or lipopolysaccharide-induced or ischemic tolerance (Wang et al., 2002; Rosenzweig et al., 2004). Again, this reduction in activated microglia may be a reflection of the neuroprotection incurred by preconditioning but, in the cases of TNF-α and lipopolysaccharide preconditioning, microglial alterations may be particularly interesting. Both of these cross-conditioners are proinflammatory agents and would presumably increase microglial activation. The observation that sublethal proinflammatory stimuli lead to decreased microglial activation lends support that tolerance may occur due to an increased capacity to buffer inflammation.

In astrocytes, TNF-α preconditioning resulted in increased NF-κB transactivation. As discussed above, NF-κB transactivation was found to be integral to protection afforded by ischemic preconditioning (Blondeau et al., 2001). However, specific regulation of NF-κB transactivation was discovered following TNF-α preconditioning in astrocytes. Both MnSOD, a cellular antioxidant, and intercellular adhesion molecule (ICAM) are under the transcriptional control of NF-κB. While MnSOD, thought to be protective under cellular stress conditions, was transcribed in astrocytes exposed to TNF-α the transcription of ICAM, thought to be pathogenic under stress, was found to be inhibited in preconditioned astrocytes (Ginis et al., 2002). This subtle control of the differential effects of transactivation was identified as due to the phosphorylation of the p65 subunit of NF-κB, which allowed for the trans-activation of MnSOD but inhibited the association of NF-κB with the adaptor protein p300, consequently inhibiting the transactivation of ICAM (Ginis et al., 2002). Thus, cellular, as well as molecular, modulation of mechanisms may contribute to ischemic tolerance.

9.3.2.9. Summary

While identifying common pathways employed by cross-conditioning paradigms, it is important to consider that often these components represent only the fraction of molecules or mechanisms that have been directly studied. One could easily conceive of many more unknown, and perhaps more important, mechanisms leading to tolerance than those presented here. However, given the multifaceted nature of cross-conditioning, perhaps the most valuable message is that tolerance is likely to involve multiple pathways, with redundancies and alternative mechanisms linking preconditioning and subsequent severe stress.

9.4. Clinical implications of ischemic preconditioning and ischemic tolerance

9.4.1. Clinical evidence of ischemic tolerance

Ischemic tolerance in the heart, first identified experimentally over 20 years ago, has rapidly progressed to a mechanism for clinical therapeutics. Clinical evidence for the protective effect of ischemic preconditioning in myocardium has been identified in patients presenting with angina prior to myocardial infarction; these patients developed smaller infarct volume and improved survival rates when compared to patients without angina prior to myocardial infarction (Tomai et al., 1999). Percutaneous transluminal coronary angioplasty reduced anginal pain and ST segment changes on electrocardiogram during balloon inflation 5 min subsequent to a prior inflation (Deutsch et al., 1990). Furthermore, repeated brief intracoronary balloon inflations, with intervening reperfusion, can enhance tolerance to subsequent longer interruptions of myocardial blood flow by percutaneous transluminal coronary angioplasty (Yellon and Dana, 2000). Inducible ischemic tolerance in myocardium has become a clinical reality.

Retrospective studies of patients with pre-stroke transient ischemic attacks suggest that ischemic tolerance may also occur in the human brain (Weih et al., 1999; Moncayo et al., 2000; Wegener et al., 2004; Schaller, 2005). Transient ischemic attack, a brief episode of neurological dysfunction caused by focal brain or retinal ischemia, presents with clinical symptoms typically lasting less than an hour and without evidence of acute infarction on imaging. In several case-control studies, post-hoc statistical analyses indicated that ischemic stroke patients with pre-stroke transient ischemic attacks correlated with better clinical presentation and outcome when compared to stroke patients presenting without prior transient ischemic attack (Weih et al., 1999; Moncayo et al., 2000). Supporting these findings, magnetic resonance imaging data suggested that ischemic stroke patients with prior transient ischemic attack presented with reduced infarct volume compared to patients without prodromal transient ischemic attack (Wegener et al., 2004). Analogous to angina pectoris before myocardial infarction, prior transient ischemic attack was thus proposed to be associated with possible protection against ischemic insult. However, in one study designed to address the relationship between the characteristics of prior transient ischemic attack and subsequent stroke, duration or timing of the prior transient ischemic attack and neurological outcome were not correlated (Johnston, 2004), although a separate study found that transient ischemic attacks lasting between 10 and 20 minutes may be associated with improved outcome following anterior circulation stroke (Moncayo et al., 2000). As with many clinical case analyses, the research design may limit the generalization of the findings, particularly in the exclusion of morbidity, the neglect to account for antithrombolytic medication, the limitation of statistical power, or the use of incompletely documented case histories in retrospective studies (Johnston, 2004). The design of prospective, controlled clinical studies may lead to more credible conclusions. Furthermore, animal studies indicated that the neuroprotection afforded by ischemic preconditioning may be transient; therefore, careful analysis during the acute phase of recovery from severe stroke may reveal effects of prodromal transient ischemic attack. While the clinical relevance and application of ischemic preconditioning still remain somewhat elusive, better understanding of the molecular and biochemical mechanisms of ischemic tolerance may yield better parameters to analyze patient populations.

9.4.2. Increased therapeutic window

According to animal studies, ischemic preconditioning may lead to long-term, but not necessarily permanent, neuroprotection (Chen and Simon, 1997; Ueda and Nowak, 2005). As such, the therapeutic window following severe ischemia with prodromal transient ischemic attack may yield an interesting opportunity for the evaluation and application of pharmaceuticals targeted toward neuroprotection. While prospective clinical trials might benefit by including patient populations separated on the basis of prodromal transient ischemic attack, retrospective analysis of past clinical trials of drugs administered acutely following stroke may also yield an effect of drugs on populations who had reported a previous transient ischemic attack. Certainly, the concept of the clinical ramifications of ischemic tolerance is hypothetical, and remains to be addressed. However, the experimental evidence depicting a delay in neuronal death following tolerance lends momentum to the idea of clinical applicability in post-ischemic therapeutics.

9.5. Conclusion

Similar to ischemic tolerance identified in the myocardium, cerebral ischemic tolerance induced by sublethal injury presents a system by which the brain can induce its own neuroprotection. Despite the incomplete and possibly transient nature of the neuroprotection afforded by ischemic preconditioning, understanding the mechanisms underlying the cellular attempts at self preservation may provide more relevant targets for pharmacological therapies with fewer complications. While sublethal ischemia has been the classically studied inducer of ischemic tolerance, determining more specific inducers of tolerance could lead to the identification of targeted and safe sublethal injuries geared toward ischemic recovery or pretreatment.