Neuronal over-stimulation by glutamate accumulation in the synaptic cleft has been repeatedly involved in acute neurodegenerative processes. The resulting excitotoxic-ity is considered as an initial trigger of neuronal PCDs in stroke (and associated experimental hypoxia-ischemia models), epileptic seizures, and traumatic brain injuries (Friberg and Wieloch 2002). However, it is important to keep in mind that the mechanisms underlying these acute neuronal injuries occurring in vivo are not identical to those underlying excitotoxicity. Indeed, death-inducing signals are much more complex in acute neuronal injuries occurring in vivo than is the simple application of glutamate in vitro (experimentally used to model excitotoxicity). For example, in vivo, leukocyte invasion and local production of cytokines (which are absent in vitro) are known to considerably contribute to excitotoxicity resulting from the excessive accumulation of glutamate during cerebral ischemia [for review, see Lipton (1999)].
According to a generally accepted view, glutamate excitotoxicity is mainly mediated by overactivation of NMDA receptors. Kainate receptors are also activated, but to a lesser extent (Cheung et al. 2005). The activation of the relevant receptor-associated ion channels leads to excessive Ca2+ influx. The intracellular Ca2+ level is further increased by NMDA-related Ca2+-induced Ca2+ release (Alford et al. 1993) from intracellular stores (ER and mitochondria) and allows for the subsequent cal-pain activation (Amadoro et al. 2006). It has also been shown that NMDA-driven increase in intracellular Ca2+ may be generated by activation of second messenger-producing enzymes, like neuronal nitric oxide synthase (Zhu et al. 2004). All these mechanisms converge on the uncompensated increase of intracellular Ca2+ level, which then triggers depolarization of mitochondrial membrane, loss of DYm, and mPT (Dubinsky and Levi 1998). These mitochondrial alterations are in turn associated with increased ROS generation [Dykens (1994) and Dugan et al. (1995);
reviewed in van Wijk and Hageman (2005)]. DNA damage resulting from ROS production triggers PARP-1 overactivation and a subsequent drop in ATP/ADP level down to 30% of the physiological level (Budd and Nicholls 1996). AIF then translocates from mitochondria to the nucleus in a PARP-1-dependent manner [i.e. it is abolished in PARP-1 knockout mice (Yu et al. 2002; Wang et al. 2003)]. Consistently, AIF-deficient neurons are resistant to PARP-1 dependent cell death mediated by PARs (Yu et al. 2006), thereby demonstrating that both effectors function along the same biochemical pathway.
AIF involvement in excitotoxicity has been consistently reported in more detail by a number of studies. In the original study by Klein and collaborators, CGN neurons derived from AIF hypomorphic Hq mice and treated with high glutamate concentrations (mM range) were more sensitive to cell death induction than neurons derived from their wild-type counterparts (Klein et al. 2002). In contrast, Cheung and colleagues found that, compared to wild-type neurons, cultured cerebellar neurons from Hq mice were protected from glutamate (100 |M) cell death induction (Cheung et al. 2005). These authors further reported that cortical neurons derived from Hq mice and treated with selective agonists of different types of glutamate receptors were less vulnerable to NMDA- and kainate-, but not AMPA, -induced cell death than wild-type neurons (Cheung et al. 2005). Furthermore, Hq mice were more resistant to excitotoxicity in vivo since they displayed reduced neuronal injury in the ischemic penumbra after middle cerebral artery occlusion (Culmsee et al. 2005), as well as decreased hippocampal damage resulting from kainic acid-induced seizures (Cheung et al. 2005). These findings, combined with in vitro data demonstrating a significant reduction of cell death by siRNA-mediated AIF down regulation in glutamate- and oxygen-glucose deprivation neuronal injury models (Culmsee et al. 2005), pointed to AIF as a major mediator of glutamate excitotoxicity.
What remains to be established is whether AIF mediates all excitotoxic cell deaths exclusively by an apoptosis-like PCD pathway (see section "Apoptosis-Like PCD"). Indeed, the possibility exists that AIF contributes to excitotoxic neuronal death also by (a yet undiscovered) programmed necrosis. The main argument supporting the latter hypothesis concerns the Bax-insensitive component of excitotoxic neuronal death. Thus, although excitotoxic neuronal death appears to be mostly independent of Bax, genetic ablation of Bax does confer a partial protection to neurons injured with glutamate or its agonist NMDA (Cheung et al. 2005; Dargusch et al. 2001). This raises the possibility of AIF involvement in programmed necrosis that might occur either as a primary or secondary response, after the initiation of the death process by apoptosis-like PCD. In this latter case, ROS generated along the excitotoxicity-induced apoptosis-like PCD may trigger DNA damage and shift to secondary induction of programmed necrosis (Fig. 3.2). Clarifying these points is critical, since it is increasingly clear that excitotoxic cell death actually involves a mixture of distinct cell death programmes, and that apoptosis-like PCD is certainly not the only PCD involved. The additional cell death outcomes that have been involved in excitotoxicity are AMPA receptor-mediated apoptosis (Cheung et al. 2005; Wang et al. 2004), and also kainate receptor-mediated autophagic PCD (Wang et al. 2008) and necrosis (Chihab et al. 1998).
Fig. 3.2 A model for a putative contribution of different PCD outcomes to acute neuronal injuries. All known acute neuronal injuries might trigger concomitantly different types of PCD, but their proportion might vary according to the initial lesion (epileptic seizure: yellow circle; stroke: blue circle; ischemia: pink circle). Excitotoxicity is represented by the yellow circle as is epilepsy because, among the selected injuiries, excitotoxicity probably overlaps the best with epileptic seizures. The latter are associated with excessive neuronal network activity related to noncompensated liberation of glutamate and glutamate receptor over-stimulation (Ben-Ari 2001) leading to cell death (Fujikawa et al. 2000a). Ischemia and stroke (where cell death is triggered by excess synaptic liberation of glutamate) also involve an excitotoxic component but they comprise additional triggers like instantaneous unavailability of glucose (more specifically related to ischemia) or fast ROS over-load (more specifically related to reperfusion in stroke). Indeed, because glucose is physiologically the only metabolite substrate that is rapidly transported across the blood-brain-barrier, in ischemia where blood supply is insufficient there is a rapid energy failure that is related to glutamate accumulation in the synaptic cleft (Camacho and Massieu 2006). Decline in ATP production due to the insufficient supply of glucose might activate defence mechanisms such as autophagy, to compensate for glucose absence by providing metabolic sources from degraded cell constituents. However, excessive autophagy might in turn trigger autophagic PCD (Adhami et al. 2007), meaning that autophagic PCD might be more involved in ischemia than, for example, in epilepsy. Similarly, during the reperfusion phase of stroke, there is a massive and rapid ROS production (Saito et al. 2005) which is, among other factors, related to DNA damage and PARP-1 activation (Skaper 2003). The latter has been involved in apoptosis-like PCD and in programmed necrosis (see section "PARP-1"), thus both of them might contribute more to stroke-induced neuronal death than to the type of death seen in other acute injuries
The reported involvement of (accidental) necrosis in excitotoxicity (and adjacent necrotic morphology of dying neurons) should motivate additional studies aimed at the assessment of the putative programmed character of this cell death outcome. Additional argument for the relevance of programmed necrosis in excitotoxicity is the involvement of effectors such as calpain and PARP-1, which in addition to AIF, has been demonstrated to contribute to neuronal apoptosis-like PCD in excitotoxic-ity, and are also shared by recently characterized programmed necrosis (Moubarak et al. 2007). The experimental evidence which supports this view comes from the data obtained after combined disabling of apoptosis (by Apaf-1 ablation) and apoptosis-like PCD (by siRNA-mediated AIF downregulation), which still leaves about 30% of neurons unprotected against the excitotoxic death (Cheung et al. 2005). The reported Bax independency of excitotoxic neuronal death observed in that study (Cheung et al. 2005) appears at first sight incompatible with programmed necrosis, since this PCD displays Bax-dependency in MEFs (Moubarak et al. 2007). However, in the previous study, Bax dependency was assessed in the paradigm that the excitotoxic insult was applied for a relatively short time period (1 h), then removed and cell death assessed only 24 h later. It is thus possible that initial, transient Bax activation, which would argue for the involvement of programmed necrosis, remained unnoticed. The involvement of Bax in excitotoxicity should be definitely re-examined during the early steps of cell death triggering in order to answer the question of a possible involvement of programmed necrosis in this type of neuronal death. In this light, very recently it was reported that |-calpain plays a crucial role in the release of AIF in neuronal ischemic injury, thus confirming that the same scenario as the one we proposed for programmed necrosis in MEFs is also functional in neurons (Cao et al. 2007).
Was this article helpful?