Apoptosis is important during normal brain development [30, 31]. Our previous studies have shown that apoptotic cell death after HI in the immature brain involves caspase-dependent and –independent pathways . Caspases are activated after HI, particularly in the immature brain  and caspase inhibition affords neuroprotection in the immature brain [14, 15, 32]. A causal role of AIF for neuronal cell death and brain injury following HI in the immature brain has been identified by using Hq mutant mice, where brain injury was reduced by 52.6% at 3 days after HI . In this study we demonstrate long-term neuroprotection in AIF-deficient mice after HI, where brain injury was reduced by as much as 63.5% 4 weeks after HI. This extends our earlier findings and further confirms that AIF plays a major role in the development of brain injury after HI. After HI, a large proportion of the dying neurons in the immature brain are immunopositive for both active caspase-3 and nuclear AIF, but there are also populations of neurons displaying only active caspase-3 or only nuclear AIF . We found that Hq mice displayed approximately 50% smaller infarct volumes than wild type mice. The caspase inhibitor Q-VD-OPh also produced approximately 50% smaller infarct volumes, and combining the two, by treating Hq mice with Q-VD-OPh, produced an additional 50% infarct volume reduction . Also in a model of traumatic brain injury in adult mice, increased protection was found when combining functional AIF reduction (as in cyclophilin A
mice) with caspase inhibition (boc-Asp-FMK) . Hence, caspase-dependent and caspase-independent (AIF-dependent) pathways appear to act, at least partly, in parallel.
Opposite its role in mediating apoptosis once it is released from mitochondria, AIF, as a flavoprotein, is essential for the maintenance of a fully functional complex I . In healthy cells, the physiological role of AIF in sustaining complex I-driven oxidative phosphorylation appears related to the local redox activity of AIF and is independent of its proapoptotic properties . Efforts to study the mitochondrial function of AIF have focused on the putative ability of AIF to regulate reactive oxygen species . Genetic mutant Hq mice with up to 80% reduction of mitochondrial AIF, display reduced levels of complex I and impaired assembly of complex I subunits . These mice exhibit mitochondrial respiratory chain diseases, such as cerebellar neurodegeneration with ataxia and progressive retinal degeneration. A recent study showed that the mitochondrial complex I contributes to oxidative injury during early reperfusion after HI in the neonatal mouse brain and that inhibition of complex I decreased the extent of HI injury . The Hq mutation, displaying reduced levels of AIF, but also reduced levels of for example catalase and complex I, renders the brain tissue more susceptible to oxidative stress [14, 37]. We did not assess the levels of oxidative stress in the SGZ in this study, but it would be interesting to evaluate the effects of an antioxidant agent, to see if the effects in the SGZ would be different from the effects in mature neurons.
Interestingly, we observed a wave of apoptotic cell death starting at the inner layers of the GCL and moving outwards, and at the same time increasing about 10-fold in numbers of dying cells from 4 h to 24 h after HI (Figure 4). Necrotic cell death, as judged by massive calcium influx leading to calpain-specific cleavage of fodrin to yield the 150 kDa breakdown product (FBDP), was only observed in the outer layers of the GCL, indicating that necrosis occurred only in fully differentiated neurons, not in stem and progenitor cells. Only a fraction of the BrdU-labeled cells in the inner layers of the GCL died (underwent apoptosis) after HI. This means that out of all the BrdU-labeled cells in the SGZ, born 1 or 2 days earlier and surviving until 4 h after HI, approximately 16% were TUNEL-positive and 7% were active caspase-3-positive (compare Figures 2B and 5B). In the Hq mice, approximately 70% fewer cells were dying (were TUNEL-positive) in the SGZ after HI, but the overall numbers of BrdU-labeled cells (dying and not dying) in the SGZ 4 h and 24 h after HI were not different between Hq and wild type mice (Figure 2B). The double-positive cells were fewer in the SGZ 24 h than 4 h after HI and no significant differences between Wt and Hq mice were observed (data not shown). As noted above, by this time point (24 h) the wave of cell death had spread outward and, consequently, fewer dying, newly generated cells could be detected in the inner layers (SGZ), even though the total number of dying cells was approximately 10 times higher in the entire GCL at this later time point. It is not clear why the apparently lower rate of dying cells in the SGZ in Hq mice did not lead to a difference in undifferentiated, BLBP-positive cells 4 weeks later (Figure 2A2B). Presumably, the AIF downregulation only protected neuronally committed progenitor cells and neurons, not undifferentiated stem cells. Also, as mentioned above, it was only a fraction of the BrdU-labeled cells that were TUNEL- or active caspase-3-positive, and only a fraction of these that was protected by the AIF deficiency. Nevertheless, the loss of a number of undifferentiated as well as more or less differentiated cells after HI leads to impaired growth of the DG, resulting in a smaller DG volume 5 weeks after HI, as shown earlier . Overall, we have shown that AIF plays an important role not only in the HI-induced death of mature neurons throughout the brain, but also in the HI-induced death of newborn cells and neuronal progenitors in the DG.
Neurogenesis can be induced by brain ischemia, indicating that regenerative mechanisms are activated by injurious stimuli. This could inspire hope for development of restorative therapies for neurological disorders and brain injuries. However, the majority of newborn cells appearing after HI do not survive beyond 3 weeks, as judged by BrdU-labeling [26, 27]. Since immature cells are more prone to undergo apoptosis than fully differentiated neurons, particularly in the immature brain, we propose that the continuous and massive decrease in the number of BrdU-labeled cells induced by an injury may be due to apoptosis-mediated cell death, which in turn reduces the restorative capacity. A previous study showed that stem and progenitor cells in the subventricular zone die after HI at least partly through caspase-3 and calpain activation . Under normal conditions in the adult hippocampus, the majority of newborn cells undergo death by apoptosis in the first 1 to 4 days of life, during the transition from amplifying neuroprogenitors to neuroblasts . In the present study, we followed the fate of cells born 1 or 2 days before HI, i.e. when they were changing from transient amplifying cells to neuroblasts. Reducing apoptosis through overexpression of bcl-2 under the neuron-specific enolase (NSE) promoter was found to double the rate of neurogenesis in the dentate gyrus as demonstrated by quantification of doublecortin-positive progenitor cells and BrdU/NeuN double-labeling. The effect of Bcl-2 was limited to the late phase of progenitor maturation, presumably correlating with the onset of NSE expression, as proliferation and early-phase progenitor cells were not affected and the increased level of neurogenesis led to a significantly higher total number of granule cells in the dentate gyrus . Pharmacological caspase inhibition could increase the number of surviving, seizure-induced newborn neurons . The role of AIF in the death of newborn cells after HI in the immature brain has not been investigated before. In this study, we found evidence for AIF-induced cell death both in the early phase of transient amplifying cells/early neuroblasts as well as in the later phase of mature neuronal progenitors and young neurons.