The main findings of this study pertain to the dramatic and irreversible suppression of subventricular zone neurogenesis and the loss of oligodendrocyte precursors following whole brain radiation. There was no significant recovery in the SVZ up to 15 months post radiation. The irreversible and progressive loss of proliferating cells in the SVZ could be due to a loss in stem cell numbers or the functional inactivation of the stem cell pool. Our data stand in marked contrast to data based on pharmacological suppression using antimitotic agents such as Ara-C 
. SVZ exposure to AraC leads to a dramatic but transient loss of all precursors (type A and C cells). The relatively quiescent SVZ astrocytes (type B cells) remain largely intact, reenter the cell cycle after washout of AraC and repopulate the entire SVZ within a week 
Previous studies have suggested that radiation can suppress proliferating SVZ cells for at least 3 months 
. Our data extend these observations for up to 15 months confirming that SVZ damage is truly irreversible. Interestingly, regions outside the SVZ such as the corpus callosum and the cortex are capable of at least partial recovery. An increase in proliferating cell numbers is seen in both regions within 2 weeks post radiation. This recovery approaches age-matched control cell numbers, particularly in the cortex, but is not sustained beyond 9 months post XRT. Such a transient proliferation response is compatible with the activation of neural precursors with limited self-renewal potential resulting in a transient recovery followed by exhaustion of the precursor pool. It also implicates the loss of long-term self-renewing stem cells or their inability to re-enter the cell cycle.
One particular feature in our experimental design was the shielding of the olfactory bulb from radiation exposure. The response in the OB is essentially tri-phasic. There is initial loss of neuroblasts coming in from the SVZ with delayed loss of glomerular calretinin-expressing neurons. A second phase involves a robust recovery characterized by neuroblast proliferation and leading to a successful and complete repopulation of the glomerular neurons by 6 months post XRT. This occurs despite continued suppression of proliferating cells and complete absence of DCX+cells in the SVZ and proximal RMS. This proliferative activity is initiated in the distal RMS, which was effectively shielded from high dose irradiation and is likely due to the ability of local neural precursors to self-renew and repair their niche independently of the SVZ
. Nonetheless, the recovery fails dramatically beyond one year, with complete exhaustion of proliferating doublecortin cells and significantly reduced calretinin neurons. We hypothesize that this result is due to the continued suppression of the SVZ and the lack of long-term renewing precursors in the RMS and OB. Therefore high dose radiation resulted in greater suppression of the quiescent SVZ stem cell compartment compared with the cycling progenitor populations outside the SVZ. Additional regional influences may also play a role since the same cell populations (BrdU/NG2) follow different kinetics depending on their location, with greater and permanent suppression experienced in the SVZ, compared to the cortex or callosum. An alternative explanation for this finding is a region or niche-dependent difference in stem cell or precursor origin.
Previous studies suggested that niche-dependent inhibition of stem cell function is responsible for the reduction in hippocampal neurogenesis observed after radiation 
. While the OB here was shielded from the direct effects of radiation, it could have been affected by a bystander effect 
. This phenomenon is considered an important mediator of the delayed effects of radiation and is typically transmitted via cytokine secretion or intercellular contacts such as gap junctions. In our study we cannot rule out that perturbations in the OB niche occurred in a delayed fashion. However this appears to be an unlikely interpretation of the data since the decrease in calretinin neurons is accompanied with a decrease in both total and proliferating doublecortin+neuroblasts. Our data strongly suggest that neuron loss in the OB is dependent on SVZ precursor cell loss rather than niche related changes in the OB.
Outside of the SVZ and hippocampal progenitor pools, the long-term effects of brain irradiation are poorly understood. In the normal brain, the majority of cycling cells are thought to be NG2-expressing oligodendrocyte progenitors. While there is data suggesting they may have multiple functional roles within the adult CNS
, NG2+cells are considered part of the oligodendrocyte lineage and are capable of giving rise to new oligodendrocytes under both normal and demyelinating conditions
. During differentiation, NG2 cells, often co-expressing PDGFRA, are gradually downregulated and cells enter a transitory pro-oligodendrocyte stage where they express the O4 antigen. As cells mature, they progressively lose expression of progenitor markers and acquire markers of mature oligodendrocytes, including MBP, MAG and CNP
. Other data show that NG2 cells may be recruited to a demyelinated area
but do not always contribute to efficient remyelination. These studies suggest that environmental factors play a significant modulatory role that may inhibit NG2 cell differentiation. The interpretation of our NG2 findings is complicated by the effect of aging whereby proliferating NG2 cells decrease steadily especially beyond a year of age. There are also regional differences, with the cortex and callosum exhibiting some recovery of NG2 proliferation following radiation, but not the SVZ. This could indicate context dependent alterations in NG2 cell behavior or fate. More recently lineage tracing studies have shown that the adult SVZ can contribute to oligodendrogenesis 
. Despite questions about the fate of NG2 cells and their pleomorphic role and in light of the concomitant loss of PDGFRA and O4 it is reasonable to conclude that the depletion of cycling NG2 cells contributes to the inability to remyelinate.
Our data suggest that normal animals have the ability to maintain O4 levels in aging despite a decrease in cycling NG2 precursor cell numbers. In contrast irradiated animals are incapable of maintaining O4 levels either due to loss of NG2 precursors below a critical threshold or loss of the mechanism that controls O4 homeostasis. The robust recovery response of the NG2/BrdU+precursors to near normal levels argues against NG2 precursor cell loss as the primary reason for the inability to maintain O4 levels post radiation. However at late time points (beyond 1 year post XRT) the number of NG2 BrdU+cells may be below a potential critical threshold required for replenishment of the O4 pool. In normal animals, a relatively small number of cycling NG2 cells (30% of 3-month control animals) is sufficient to maintain O4 levels during aging while irradiated animals with similar NG2/BrdU cell numbers are unable to sustain O4 levels. The factors that control O4 levels for a given number of oligodendrocyte precursors are not known but may include cell autonomous or environmental factors that impact progression along the oligodendrocyte lineage. Alternatively NG2 progeny may not survive due to radiation-related mitotic cell death or to the perpetuation of cytokine cascades triggered by tissue response to XRT
The response of MBP expressing cells is unique among all the cell populations described here. In contrast to the oligodendrocyte precursor markers such as NG2/BrdU, PDGFRA or O4, MBP in irradiated animals was maintained at close to control levels until 9 months post radiation. However, beyond 1 year we observed a rapid decrease in MBP. Late onset demyelination after brain irradiation has been described in multiple species including humans
but the mechanism for this delayed response has remained unclear
. One possible explanation is a tight control of MBP levels despite a significant decrease in oligodendrocyte precursor cells. The lack of an initial MBP loss suggests that MBP producing cells are relatively resistant to the immediate effects of radiation presumably due to their highly differentiated nature. While the exact turn-over rate of mature MBP+cells is not known, the kinetics of MBP loss after radiation is compatible with a very slow turn over rate keeping MBP at near normal levels for up to 12 months. The loss of MBP levels beyond the putative MBP turn-over rate could not be compensated due to the lack of functional oligodendrocyte precursors. Alternatively, MBP turn-over rates may be faster and MBP levels actively maintained through proliferation and differentiation of the oligodendrocyte precursor cell compartment. In such a scenario loss of oligodendrocyte precursors below a critical threshold or an inability to maintain MBP homeostasis may trigger late onset MBP loss. Ultrastructural studies demonstrate clearly that MBP levels are not only downregulated but are associated with structural damage to the myelin sheath indicative of oligodendrocyte death or dysfunction. The failure of recovery could be due to the transmission of radiation-induced genetic instability over many cell divisions leading to delayed reproductive death of cells in the oligodendrocyte lineage 
Some authors have attributed demyelination to endothelial cell damage, ischemia and necrosis
. In fact, endothelial cells have been invoked as the primary target of radiation to the CNS as they are sensitive to acute radiation damage. However there is limited information about the long-term effects of radiation on endothelial cell numbers
. Here we report that endothelial cell numbers recover to near control levels within 2 weeks and remain within normal range for periods beyond onset of demyelination. In recent work Hopewell's group
demonstrated that radioprotection of endothelial cells against apoptosis reduces the risk of delayed radiation-induced necrosis but did not comment on the impact of radioprotection on demyelination. There are additional recent investigations that suggest that depletion of precursors is independent of damage to the vasculature.
. Here we demonstrate demyelination by radiographic and histological methods prior to the occurrence of vascular necrosis at a stage when endothelial cell numbers are close to normal levels. Furthermore, demyelination occurs in a diffuse histological pattern while necrotic events, observed several months after onset of demyelination, occur in focal areas, particularly in the corpus callosum and fornix. It is important to note that the study of the vascular system here is purely structural. Changes in endothelial cell function such as capillary permeability and status of VEGF pathways have not been investigated and may still play a role in facilitating demyelination
Interestingly our data in the rat model were further corroborated by the analysis of clinical specimens of human brain at early and late time points post radiation. An early loss of oligodendrocyte precursors, as evidenced by loss of O4 and PDGFRA expression, preceding demyelination and a near complete recovery of endothelial cell numbers supports the hypothesis that loss of oligodendrocyte precursors is a primary event. Additionally, electron micrographs of human and rat specimens at various time points after radiation support our findings by revealing a similar pattern of myelin sheath degradation over time post radiation with absence of significant axonal damage. This pattern of loss of lamellar compaction and subsequent vesiculation of myelin sheaths coupled with a moderate influx of lipid laden macrophages is consistent with pathological findings of primary demyelination. Importantly, neurofilament integrity and organization of the axoplasm appeared normal; staining for MBP and neurofilament proteins in late post radiation rat and human samples confirmed a loss of myelin with apparent preservation of axons.
In summary this study demonstrates permanent suppression of the SVZ stem cell compartment following radiation as well as an early and sustained loss of oligodendrocyte precursor cells with subsequent delayed demyelination. The detailed analysis of various cell populations over time reveals potential therapeutic windows that could target the recovery phases of neural precursors post injury prior to the occurrence of structural damage to the myelin sheaths. The rat model is validated by similar findings in human tissue. Based on this model, therapeutic strategies may be directed at reducing initial precursor cell loss or possibly at replacing the lost precursor cells via transplantation of primary or stem cell derived oligodendrocyte precursors as a means of preventing late radiation-induced demyelination.