Author contributions: E.K. and N.K.: conception and design, collection and assembly of data, data analysis and interpretation, and manuscript writing; M.A.: conception and design and collection and assembly of data; H.H., K.A., H. Togari: conception and design; H. Takebayashi: provision of study material; K.I.: provision of study material; K. Sobue: conception and design and administrative support; K. Sawamoto: conception and design, financial support, data analysis and interpretation, and manuscript writing.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS August 13, 2012.
Perinatal hypoxia-ischemia (HI) frequently causes white-matter injury, leading to severe neurological deficits and mortality, and only limited therapeutic options exist. The white matter of animal models and human patients with HI-induced brain injury contains increased numbers of oligodendrocyte progenitor cells (OPCs). However, the origin and fates of these OPCs and their potential to repair injured white matter remain unclear. Here, using cell-type- and region-specific genetic labeling methods in a mouse HI model, we characterized the Olig2-expressing OPCs. We found that after HI, Olig2+ cells increased in the posterior part of the subventricular zone (pSVZ) and migrated into the injured white matter. However, their oligodendrocytic differentiation efficiency was severely compromised compared with the OPCs in normal tissue, indicating the need for an intervention to promote their differentiation. Erythropoietin (EPO) treatment is a promising candidate, but it has detrimental effects that preclude its clinical use for brain injury. We found that long-term postinjury treatment with a nonerythropoietic derivative of EPO, asialo-erythropoietin, promoted the maturation of pSVZ-derived OPCs and the recovery of neurological function, without affecting hematopoiesis. These results demonstrate the limitation and potential of endogenous OPCs in the pSVZ as a therapeutic target for treating neonatal white-matter injury. STEM Cells2012;30:2234–2247
In spite of improvements in neonatal intensive care, cerebral palsy still affects 1–2 newborns per 1,000 . Cerebral palsy patients frequently have white-matter injury  caused by intrauterine events such as infection  and hypoxia/ischemia (HI) . Although early treatment with neuroprotective drugs prevents white-matter injury in animal models [5–8], prediction of the injury is difficult in prenatal and early postnatal brains in humans. Therefore, effective therapeutic interventions are still needed for the treatment of perinatally injured white matter. One of the most promising strategies for restoring injured white matter is to promote the generation of new myelin-forming oligodendrocytes .
Cell transplantation-based regenerative therapies using stem cells, including embryonic stem cells and induced pluripotent stem cells, can improve neural functioning in animal models of various neurological diseases [10, 11]. However, the application of this approach to the infant brain carries a life-long increased risk of adverse events, such as tumorigenesis. Nonetheless, neural stem cells (NSCs) in the subventricular zone (SVZ), located at the lateral walls of the lateral ventricles, have been investigated as an endogenous cell source for new neurons and glial cells generated in the postnatal brain [12, 13]. Cell proliferation in the SVZ is upregulated in response to various pathological conditions, including ischemic stroke and neurodegenerative diseases [11, 14, 15]. The immature new neurons born in the SVZ are characterized by their prominent migration capacity. These cells migrate rapidly through a pathway known as the rostral migratory stream in rodents, toward the olfactory bulb [16–18]. We previously reported that these endogenous NSCs in the SVZ also have the regenerative capacity to produce new neurons that migrate toward the injured area [19–21].
NSCs in the SVZ also produce oligodendrocyte progenitor cells (OPCs). The SVZ-derived OPCs migrate toward the corpus callosum (CC) overlying the SVZ, where they differentiate into mature oligodendrocytes that form myelin [22–24]. In the brains of patients with multiple sclerosis and rodent models of demyelination, the production of new OPCs in the SVZ is significantly increased [25, 26], suggesting that the SVZ might be a source for new oligodendrocytes after white-matter injury. However, the precise fates and function of the SVZ-derived OPCs after white-matter injury remain largely unknown. Furthermore, although several interventions have been reported to accelerate remyelination in animal models of myelin diseases [27–33], effective therapies for neonatal HI-injured white matter that target the SVZ-derived OPCs and their progenies have not been developed.
Erythropoietin (EPO), a clinically used erythropoietic factor, has various beneficial effects on the central nervous system (CNS), such as neuroprotection and the promotion of neurogenesis [34–37]. Although EPO also stimulates the proliferation of neuronal and OPCs after HI in neonatal rats [38, 39] and accelerates the maturation of OPCs in vitro , the specific effects of EPO on SVZ-derived OPCs have not been determined. More importantly, although these desirable functions of EPO are potentially useful for the regeneration of injured white matter, its potent erythropoietic and procoagulant activities, which lead to polycythemia and thrombosis, preclude its long-term use for the treatment of neurological diseases [41, 42].
In this work, using retroviral labeling and genetic fate mapping approaches in a mouse model of HI-induced neonatal white-matter injury, we studied progenitor cells expressing Olig2, a basic helix-loop-helix transcription factor essential for specifying oligodendrocyte-lineage in the postnatal brain , in the posterior part of the SVZ (pSVZ). The progenies of the Olig2+ cells migrated rapidly and became widely distributed in the injured CC, but they did not differentiate efficiently into mature oligodendrocytes. Long-term treatment after the HI injury with asialo-erythropoietin (AEPO), a derivative of EPO with lower erythropoietic activity , increased the number of SVZ-derived mature oligodendrocytes in the injured CC and induced long-lasting neurological improvement with no obvious effect on blood count.
MATERIALS AND METHODS
Wild-type (WT) ICR mice were purchased from SLC (Shizuoka, Japan). Olig2 knockin mice expressing Cre-ER under control of the Olig2 promoter (Olig2CreERTM mice)  were mated with Rosa26ECFP mice  to obtain double-transgenic mice. All animal studies were approved by the Animal Research Committee, Graduate School of Medical Sciences, Nagoya City University and were conducted in accordance with the Institutional Laboratory Animal Care and Use Committee.
HI was induced in P5 mice. The right common carotid artery was cauterized as described previously [47, 48], followed by a 1-hour recovery period, and then by systemic hypoxia. The protocol is detailed in the supporting Information Materials and Methods.
Immunohistochemistry was performed as described previously [18, 49]. Detailed conditions are in the supporting Information Materials and Methods.
Mice were injected intraperitoneally with bromodeoxyuridine (BrdU, 50 mg/kg, Sigma) in phosphate buffered saline (PBS) to label newly born cells. The BrdU labeling protocol is detailed in the Supporting Information Materials and Methods.
Tamoxifen Preparation and Injection
Tamoxifen (Sigma) was dissolved in a dimethyl sulfoxide/ethanol/sesame oil (4:6:90) mixture at 10 mg/ml as reported previously [50, 51]. The mice were given a single injection of tamoxifen (200 μg/g body weight) intraperitoneally at P9.
Each animal received an injection of 1 μl of replication-incompetent retroviruses into the right pSVZ at P5. The stereotaxic coordinates were 2.0 mm anterior and 2.3 mm lateral to lambda, and 2.0 mm deep. The retrovirus preparation protocol is in the Supporting Information Materials and Methods.
Recombinant human EPO was gifted by Chugai Pharmaceutical Co., Ltd. and desialylated with neuraminidase as described previously . For the in vivo experiments, AEPO was diluted to 8 μg/ml with PBS, then intraperitoneally injected into mice once a day for either 3 or 14 consecutive days starting at P10. For the in vitro experiments, AEPO was diluted with culture medium to 0.0008, 0.008, or 0.08 μg/ml before use.
OPC cultures were prepared as previously described [49, 52]. Detailed protocols are in the Supporting Information Materials and Methods.
EPO Receptor Knockdown
Two EPO receptor (EPOR) stealth small interfering RNAs (siRNAs) (MSS274172, MSS236416) and a negative control siRNA were purchased from Invitrogen (Carlsbad, CA, http://www.invitrogen.com). Isolated OPCs on coverslips were transfected with siRNA, as described previously  on days 4, 5, and 7. AEPO (0.0008 ng/ml) was added to the culture on day 5, then the cells were fixed with 4% paraformaldehyde (PFA) on day 9.
The foot fault test  was modified and performed during postnatal weeks 4, 6, and 8. The protocol is in the Supporting Information Materials and Methods.
Image Processing and Quantification
The cell numbers in brain sections were counted in every sixth coronal section, including the pSVZ (1–3 total sections), except for the analysis of anterior part of SVZ (aSVZ) in Figure 1. Methodological details are described in the Supporting Information Materials and Methods.
All data were analyzed using Prism software (Graphpad) and are shown as the mean ± SEM. Multiple group comparisons were performed by ANOVA followed by Bonferroni's procedure. The two-tailed paired t test, unpaired t test, and Mann–Whitney U test were used for pairwise comparisons. p <.05 was considered significant.
Hypoxic-Ischemic White-Matter Injury Promotes the Production of OPCs in the Neonatal pSVZ
To investigate the oligodendrogenic potential of the neonatal SVZ after white-matter injury, we used a mouse model of neonatal HI, which mainly damages myelin formation in the CC and its adjacent tissues [47, 48, 54] (Fig. 1A, 1B) and increases cell proliferation in the underlying SVZ (Supporting Information Fig. 1). To study the production and distribution of new OPCs in the SVZ in response to HI, mice were intraperitoneally injected with BrdU (50 mg/kg) 5 days after HI (P10), and BrdU-labeled cells that expressed Olig2, an oligodendrocyte lineage marker, were quantified 6 hours (P10), 3 days (P13), and 7 days (P17) later (Fig. 1C). Given that different SVZ regions are reported to contain heterogeneous progenitor cells with distinct potentials [16, 55, 56], we compared the responses between aSVZ and pSVZ, divided at the anterior border of the hippocampus. The density of Ki-67+ proliferating cells increased significantly in both the aSVZ and pSVZ after HI compared with that in intact mice at the same age (Supporting Information Fig. 1A, 1C). In contrast, the density of BrdU+Olig2+ cells at P10 increased significantly in the pSVZ (Fig. 1D, top, 1E) but not in the aSVZ (Supporting Information Fig. 1D). In addition, the percentage of BrdU+ cells that was Olig2+ oligodendrocyte-lineage cells was significantly higher in the pSVZ than in the aSVZ in both groups of mice (intact: pSVZ 37.5% ± 3.2%, aSVZ 14.1% ± 0.9%, HI: pSVZ 47.2% ± 3.8%, aSVZ 27.5% ± 4.6%). An HI-induced increase in BrdU+Olig2+ cells in the ipsilateral pSVZ was transiently observed at P10, but not at P13 or P17 (Fig. 1E). In contrast, in the injured CC, the increased density of BrdU+Olig2+ cells, compared with that in the CC of intact mice, persisted throughout the experimental period (Fig. 1F). These regionally and temporally distinct BrdU+Olig2+ cell responses suggested that HI stimulates the production of OPCs in the pSVZ, and that these OPCs migrate into the injured CC.
To compare the HI-induced OPC production with that in the normal brain at the early postnatal stage, we counted the number of Olig2+ cells labeled with BrdU that was injected 2 hours before sacrifice at P3, P5, P7, P10, P13, and P17 (Supporting Information Fig. 2). In both the aSVZ and pSVZ, the highest density of BrdU+ proliferating Olig2+ cells was observed at P3, and it decreased rapidly thereafter. Taken together, these results indicate that the OPC production for normal brain development is downregulated in the first postnatal week but that it can be transiently reupregulated by HI in the pSVZ.
HI Increases the Number of pSVZ-Derived OPCs Migrating into the Injured CC
To specifically label proliferating pSVZ cells, we stereotaxically injected a DsRed-encoding retrovirus (CAG-DsRed-Express) into the pSVZ 2 hours before HI and studied the distribution of labeled cells in the CC 8 days later (P13) (Fig. 1G). The majority of DsRed-labeled cells migrated into the lateral CC (Fig. 1H). In contrast, only a small number of labeled cells appeared in the olfactory bulb or cerebral cortex in either the injured or intact hemisphere.
To identify oligodendrocyte lineage cells in this virus-labeled population in the CC, the brain sections were immunostained for Olig2. The majority of the DsRed+ cells in the CC of the injected hemisphere at P13 were Olig2+ in both the intact and HI-exposed group (n = 3, for each group, HI: 72.2% ± 4.9%, intact: 81.0% ± 8.3%, p >.05). These findings were quite different from those of previous studies of the aSVZ, from which most of the retrovirus-labeled cells migrate anteriorly and differentiate into neurons in the OB in the intact brain [16, 57, 58]. Notably, significantly more DsRed+Olig2+ cells were observed in the injured CC than in the intact one (Fig. 1I, 1J). Taken together, these results provide evidence that OPCs in the pSVZ migrate laterally into the CC and that HI injury significantly increases the number of these cells. These findings raise the possibility that the pSVZ is a potential source of new OPCs for oligodendrogenesis after HI injury.
Oligodendrocytic Differentiation of OPCs Is Disturbed in the HI-Injured CC
To investigate the fate of the OPCs that are newly generated following HI, we used Olig2CreERTM knockin mice  crossed to Rosa26ECFP transgenic mice . Olig2CreERTM; Rosa26ECFP mice express enhanced cyan fluorescent protein (CFP) under control of the Olig2 promoter after Cre-mediated excision of a loxP-flanked stop codon, allowing us to follow the fate of Olig2-expressing cells and their progenies. HI was induced at P5, and tamoxifen was injected into the Olig2CreERTM; Rosa26ECFP mice at P9 to induce Cre-mediated recombination in the proliferating OPCs (Fig. 2A). On the day after the injection, more than 90% of the CFP+ cells expressed Olig2 protein, as detected by immunohistochemistry, in both the HI-injured and intact CC (HI: 93.9%, intact: 90.4%) (Fig. 2B). Only a small population of the CFP+ cells expressed the astrocytic marker glial fibrillary acidic protein (GFAP) or the neuronal progenitor marker doublecortin (Dcx) at P13, P17, and P24, regardless of HI treatment (Supporting Information Fig. 3). Two weeks after the injection, some of the CFP+ cells expressed a myelin marker, myelin basic protein (MBP) (Fig. 2C), suggesting that they had differentiated into myelin-forming functional oligodendrocytes. Thus, even after HI, this procedure efficiently labeled the Olig2+ cell lineage.
Using this protocol, we then compared the fate of the CFP+ cells between the injured and intact CC at P13, P17, and P24 (Fig. 2A). The damaged CC contained more CFP+ cells than the intact CC at all three time points (Fig. 2D). Since various insults can affect the fate of the OPCs [43, 50, 51, 59], we next examined whether the CFP-labeled Olig2+ cells that were increased after HI could differentiate normally into oligodendrocytes in the lesion. Brain sections were immunostained for adenomatous polyposis coli (APC) (Fig. 2E), which is expressed in mature oligodendrocytes, and NG2 (Fig. 2G), which is expressed in OPCs and downregulated upon maturation. Between P13 and P17, the percentage of CFP+ cells that expressed APC had significantly increased, and the percentage expressing NG2 had decreased; these changes were comparable in the ipsilateral and contralateral CC of HI-exposed mice (Fig. 2F, 2H). However, at P24, while nearly 70% of the CFP+ cells in the contralateral CC expressed APC (Fig. 2E, top, filled arrowheads, 2F), only about half of the CFP+ cells in the ipsilateral CC expressed it (Fig. 2E, bottom, unfilled arrowheads, and F). Consistent with this observation, the percentage of NG2+CFP+ cells in the ipsilateral CC was significantly higher than in the contralateral CC (Fig. 2G, arrows, and 2H). These results indicate that Olig2+ cells in the neonatal brain can proliferate in response to HI insult, but their oligodendrocytic differentiation in the injured CC is impaired.
AEPO Promotes the Differentiation of Cultured OPCs into Mature Oligodendrocytes via the EPO Receptor
The increase in immature Olig2+ cells and their progenies observed in the injured CC (Fig. 2) prompted us to search for a drug that would promote their differentiation into mature oligodendrocytes. We chose AEPO, a derivative of recombinant human EPO but with lower erythropoietic activity , since EPO accelerates oligodendrocytic maturation in vitro . To test whether AEPO directly promotes the differentiation of OPCs, we used a purified OPC culture described previously [49, 52]. The OPCs were cultured in medium containing AEPO (0.0008, 0.008, and 0.08 ng/ml) or vehicle for 4 days (Fig. 3A), and then their viability, proliferation, and differentiation were assessed by immunostaining. AEPO did not enhance the differentiation of OPCs at high cell density (5.0 × 104 cells per well in 24-well plates) that was induced by the withdrawal of growth factors, a well-established technique for stimulating OPC differentiation in vitro (data not shown). Therefore, for the control group, we chose a low-density culture condition (1.5 × 104 cells per well in 24-well plates), in which all but a small fraction of the cells (which differentiated into oligodendrocytes) were maintained in an immature state.
AEPO treatment did not affect the overall cell density (Fig. 3B), the percentage of cleaved caspase-3-expressing apoptotic cells (Fig. 3C), or the percentage of Ki67+ proliferating cells (Fig. 3D) in these cultures, however, the percentage of 2′,3′-cyclic nucleotide 3′-phosphodiesterase (CNPase)+ mature oligodendrocytes increased significantly in cultures treated with 0.0008 and 0.008, but not 0.08 ng/ml AEPO (Fig. 3E), as previously reported . Moreover, CNPase+ cells with extensively arborized processes extending from the cell bodies, a typical morphology of mature oligodendrocytes, were observed in the AEPO-treated groups (Fig. 3A, right) but not in the control group (Fig. 3A, left). These findings suggested that AEPO promotes the differentiation of cultured OPCs into mature oligodendrocytes.
The biological activity of EPO in the brain is generally mediated by the EPOR . To examine whether the effect of AEPO on OPC differentiation was also mediated by EPOR, we knocked down EPOR using siRNAs (MSS236416 or MSS274172) in cultured OPCs before AEPO treatment (0.0008 ng/ml). The effects of these siRNAs were confirmed by immunocytochemistry and quantitative reverse transcription-polymerase chain reaction (Q-RT-PCR). Immunocytochemical analysis showed a drastic reduction in EPOR immunoreactivity on the EPOR-siRNA-treated cells (Fig. 3F, bottom), but not on the control-siRNA-treated ones (Fig. 3F, top), as previously reported . In addition, Q-RT-PCR showed that the EPOR mRNA levels were effectively suppressed in the OPC culture with the EPOR-siRNA (Supporting Information Fig. 4). AEPO treatment significantly increased the percentage of CNPase+ oligodendrocytes with a mature morphology in the control, nontargeted siRNA-transfected OPCs, however, this effect was not observed in the EPOR siRNA-transfected group (Fig. 3G, 3H). Taken together, these results indicate that AEPO treatment promotes the oligodendrocytic differentiation of OPCs via EPOR.
Long-Term AEPO Treatment Promotes OPC Maturation in the HI-Injured CC
We next assessed whether the AEPO treatment could also improve the maturation defects of Olig2+ cell progenies observed in the HI model mice (Fig. 2). First, we investigated the expression of EPOR in the oligodendrocyte-lineage cells in the CC after HI injury by immunohistochemistry, because it is controversial whether EPOR is expressed in mouse OPCs and oligodendrocytes . In HI-injured Olig2CreERTM; Rosa26ECFP mice treated with tamoxifen (at P9), EPOR expression was observed in 53.4% ± 4.0% (n = 3) of the CFP+ cells in the regions around the lateral ventricle, including the SVZ and CC, at P17 (Fig. 4A). We also identified the major cell types expressing EPOR in this region by immunostaining for Nestin (neural progenitor marker), Dcx, or GFAP as well as EPOR. In the HI-injured CC, CFP+ oligodendrocyte lineage cells represented the largest population of EPOR-expressing cells (n = 3, for each group, CFP+: 32.0% ± 3.4%, Nestin+: 16.2% ± 3.5%, Dcx+: 6.7% ± 1.4%, GFAP+: 24.7% ± 3.0), suggesting that AEPO treatment could directly affect the OPCs after HI.
To target the OPCs and their progenies involved in restoring the HI-injured CC, we injected AEPO (80 ng/g body weight/day) intraperitoneally into the Olig2CreERTM; Rosa26ECFP mice 5 days after HI, once a day for three consecutive days or for two consecutive weeks, and quantified the expression of NG2 and APC in CFP+ cells in the injured CC at P24 (Fig. 4B). In the 2-week AEPO-treated group, the percentage of CFP+ cells that remained OPCs decreased significantly (Fig. 4D), and the percentage that differentiated into mature oligodendrocytes increased significantly (Fig. 4C, arrows indicate CFP+APC+ cells) (Fig. 4E) in the injured CC compared with the control group. Furthermore, the density of CFP+APC+ cells was elevated about 1.5-fold by the 2-week AEPO treatment (Fig. 4C, 4F). These effects were not observed in the 3-day AEPO-treated group (Fig. 4D, 4E, 4F). These results indicate that long-term (2-week) but not short-term treatment with AEPO beginning 5 days after the insult improved the defective differentiation of OPCs into mature oligodendrocytes in the HI-injured CC.
Since EPO is reported to have neuroprotective effects [34, 36, 62, 63], we also examined whether AEPO had an antiapoptotic effect on developing OPCs and oligodendrocytes (Fig. 4G). The number of cleaved caspase 3-positive apoptotic CFP+ cells without expressing APC (CFP+APC− cells) in the injured CC at P17 was significantly decreased by the AEPO treatment (n = 5, PBS: 1920.0 ± 647 cells per mm3, AEPO: 560 ± 136.4 cells per mm3, p <.05) (Fig. 4H). Only a few mature CFP+ cells (CFP+APC+ cells) underwent apoptosis, which was not affected by the AEPO treatment, suggesting that AEPO decreased the number of apoptotic OPCs, but not mature oligodendrocytes, in this HI model.
We next compared the adverse effects of long-term daily treatment with AEPO versus EPO in mice. After a 2-week treatment, the erythrocyte and reticulocyte count, hemoglobin concentration, and hematocrit values were significantly elevated in the EPO-treated group. In contrast, the AEPO-treated group did not show any detectable changes in these values (Fig. 4I), indicating that the long-term AEPO treatment is safe in neonatal animals.
AEPO Treatment Increases the SVZ-Derived Mature Oligodendrocytes in the HI-Injured CC
To investigate whether the pSVZ-derived OPCs that increased in the injured CC at P13 (Fig. 1) underwent oligodendrogenesis, we analyzed their fates at later time points. CAG-DsRed retrovirus was stereotaxically injected into the pSVZ of Olig2CreERTM; Rosa26ECFP mice 2 hours before the exposure to HI at P5, then CFP expression in the Olig2+ cells was induced by tamoxifen treatment at P9. Twelve days after HI (P17), DsRed+ pSVZ-derived cells were observed widely distributed in the CC (Fig. 5A). Some of these cells coexpressed CFP in the intact CC, indicating that they were progenies of olig2+ cells derived from the pSVZ (Fig. 5B, arrow). However, the density and percentage of the DsRed+CFP+ cells that coexpressed APC (DsRed+CFP+APC+ cells) in the injured CC was significantly lower than in the intact CC at P17 and P24 (Fig. 5C, 5D).
We then tested whether long-term AEPO treatment could increase the pSVZ-derived mature oligodendrocytes in the injured CC. AEPO or PBS was intraperitoneally administered for 2 weeks into Olig2CreERTM; Rosa26ECFP mice treated with CAG-DsRed retrovirus and subjected to HI as described above, and then the DsRed+CFP+ cells coexpressing APC in the injured CC at P24 were quantified (Fig. 5E). The 2-week AEPO treatment significantly increased the density of pSVZ-derived mature oligodendrocytes (DsRed+CFP+APC+ cells) (Fig. 5F, arrows, 5G) and the percentage of APC expression (Fig. 5H) in the injured CC compared with the PBS-treated group. Taken together, these results indicate that the long-term AEPO treatment promoted oligodendrogenesis of the pSVZ-derived OPCs in the injured CC.
Long-Term AEPO Treatment Improves the Neurological Outcome at Later Periods
To assess the effect of AEPO treatment on the neurological deficits following HI, we used a modified foot fault test . Briefly, 4-, 6-, and 8-week-old mice that had undergone HI injury (at P5) were placed on elevated hexagonal grids of 26- or 40-mm diameter for 5 minutes, and the number of missteps, referred to as “foot faults,” of each limb was separately counted. The results were determined as the foot fault rate, which was calculated by dividing the number of foot faults made by the contralateral limbs by the number of total (bilateral) errors. The foot fault rate was determined for the fore limbs alone and for the fore limbs and hind limbs. When sham-operated mice walked on the grids, the foot faults occurred almost symmetrically, with a foot fault rate of approximately 0.5 (Fig. 6, gray circles). On the other hand, in the HI-operated mice treated with PBS, the foot fault rate was significantly increased for both the fore limb and the total (fore limb and hind limb) results at all time points (Fig. 6, black squares), suggesting that HI at P5 caused sensorimotor dysfunction in the contralateral limbs that persisted into later life. AEPO treatment for 2 weeks significantly improved this defect at all time points (Fig. 6, red triangles). Consistent with the results of our histological analyses, the 3-day AEPO treatment did not significantly affect the foot fault rate at any time point.
Finally, we tested the long-term effect of the AEPO treatment on myelin formation in the CC after HI. The brain sections of 8-week-old mice that had been treated with AEPO or PBS following neonatal HI were immunostained for MBP and an axonal marker, neurofilament. The ratio of the MBP+ area in the ipsilateral CC to that in the contralateral CC was significantly increased in the 2-week, but not in the 3-day AEPO-treated group compared with that in the PBS-treated group (Fig. 6E–6G). The MBP+ area in the contralateral CC of each group did not differ significantly from that of the sham-operated group. Close examination with higher-magnification images revealed a marked increase in myelinated axons in the HI-exposed CC of the 2-week AEPO-treated group (Fig. 6G). Taken together, these results show that long-term AEPO treatment, which promoted the regeneration of mature oligodendrocytes in the injured CC (Figs. 4 and 5), could also improve the neurological and histological outcomes for a long period after HI injury in mice.
pSVZ-Derived OPCs Undergo Oligodendrogenesis Following Neonatal HI-Induced White-Matter Injury
Since neonatal HI decreases OPCs and leads to serious disturbance of myelin formation in the white matter [64, 65], repopulating the OPCs is considered to be a promising strategy for treating this pathology. Previous studies indicate that OPCs also contribute to remyelination in demyelinating diseases [66–68]. Although the number of new OPCs at the lesion site is reported to increase after various kinds of white-matter injuries, the potential of the OPCs to undergo oligodendrogenesis after neonatal HI has not been demonstrated.
Neuronal and glial progenitors are actively produced in the neonatal SVZ. While a higher level of neurogenesis by neuronal progenitors is observed in the anterior part of the SVZ , OPCs are enriched in the pSVZ, and these OPCs migrate into the white matter for oligodendrogenesis [24, 26, 69]. Severe ischemia causing extensive neuronal degeneration accelerates the production of neuronal progenitors in the neonatal SVZ; these progenitors migrate into and around the injured area [70–72]. In this study, using stereotaxic retrovirus injection combined with the genetic labeling of Olig2+ cells and their progenies, we demonstrated that the pSVZ plays a distinctive role as a source of OPCs for oligodendrogenesis in the injured white matter following neonatal HI. Interestingly, similar to the SVZ-derived neuronal progenitors migrating in the injured striatum , the pSVZ-derived Olig2+ cells and their progenies migrated extensively within the CC toward the injured area. Although it is not clear how much the pSVZ-derived OPCs contribute to the repair after white-matter injury, such a prominent migration capacity suggests that the pSVZ-derived new oligodendrocyte-lineage cells could support this regeneration process.
After reaching their destinations, the OPCs need to differentiate into mature oligodendrocytes to repair the injured white matter. By fate mapping experiments with Olig2CreERTM; Rosa26ECFP mice, we demonstrated that the Olig2+ cells, especially those derived from the pSVZ, did not efficiently differentiate into mature oligodendrocytes in the injured CC. This finding may account for the previous observation that OPCs increase in and around the injured white matter, without appearing to differentiate into mature oligodendrocytes, in human patients with chronic demyelination diseases  and neonatal white-matter injury .
Olig2, a basic-helix-loop-helix transcription factor critical for oligodendrocyte lineage development, is expressed in very early oligodendrocyte progenitors through later stages of their differentiation in the neonatal brain, including the SVZ. In addition, Olig2 is also reported to be expressed in other cell types, including neuronal progenitors in the embryonic brain [45, 75], activated astrocytes in the injured brain , and in mature oligodendrocytes . However, even 4 days after a single injection of tamoxifen, most of the CFP-labeled cells did not express APC, Dcx, or GFAP, suggesting that recombination was induced almost exclusively in the OPCs. Notably, about 2 weeks after the injection of tamoxifen, more than 80% of the labeled cells were still positive for a marker of oligodendrocytic lineage cells such as NG2 or APC, even in the injured CC (Fig. 2F, 2H, adding the NG2-positive and APC-positive CFP+ cells at P24). Therefore, although we cannot rule out the possibility that some minor population (<10%) of these CFP+ cells were not derived from the oligodendrocytic lineage, our labeling method was sufficiently specific and suitable for studying the fates of OPCs in the neonatal HI model. Moreover, the Olig2CreERTM; Rosa26ECFP mice showed efficient labeling of OPCs, which enabled the double-labeling experiment that was performed with retrovirus injection into the pSVZ (Fig. 5).
We further confirmed our results in WT mice labeled with BrdU 5 days after HI, in which newly generated Olig2+ cells could be identified by Olig2-BrdU double immunostaining. As expected, 2 weeks later, in spite of an increase in the number of BrdU+Olig2+ cells, the number of BrdU+APC+ mature oligodendrocytes was significantly decreased in the injured CC compared with the contralateral CC (n = 3, *, p <.05, ipsilateral: 2,803 cells per mm3, contralateral: 5,761 cells per mm3). Thus, we conclude that the defective oligodendrocyte differentiation observed in the genetically labeled cells reflects the nature of the majority, if not all, of the OPCs in our neonatal HI model.
Promotion of Oligodendrogenesis in the HI-Injured White Matter Using AEPO
The presence of immature cells in the injured white matter indicates the need for interventions to enhance their differentiation into mature oligodendrocytes. Agents that promote remyelination and/or the maturation of OPCs, which include LINGO-1 antagonist [78, 79], brain-derived neurotrophic factor (BDNF) [80, 81], and glatiramer acetate [82, 83], may be useful, but their efficiency and safety need to be extensively verified before their clinical application as neonatal therapy. We therefore studied the pro-oligodendrogenic effects of AEPO, a nonerythropoietic derivative of the clinically used drug, EPO. Peripherally administered EPO penetrates the blood-brain barrier  and protects the CNS tissues in animal models of various insults [34, 36, 48, 63, 85-88], as well as in neonatal patients with hypoxic-ischemic encephalopathy . EPO also accelerates the proliferation of neuronal progenitor cells [35, 38] and OPCs [39, 90] in vivo, and the maturation of rat OPCs in vitro . On the other hand, recent clinical studies indicate that EPO has undesirable side effects, especially thrombotic complications [91, 92]. Thus, in spite of the successful results of preclinical studies, the risk of thrombosis due to the erythropoietic activity of EPO has precluded its long-term use in clinical applications for treating neonatal white-matter injury.
AEPO was developed as a derivative of EPO with lower erythropoietic activity and a shorter plasma half-life . Our in vivo experiments clearly indicated that long-term treatment with AEPO that started 5 days after the insult promoted the maturation of Olig2+ cells and their progenies without affecting the number of erythrocytes (Fig. 4). In vitro, we did not observe significant maturation-promoting effects of AEPO at the standard cell density (5.0 × 104 cells per well in 24-well plates), suggesting that it was not very effective under pro-oligogenic conditions. However, in the lower cell-density condition (1.5 × 104 cells per well in 24-well plates), which maintained OPCs in an immature state, similar to the environment in the damaged white matter following HI, AEPO significantly increased the number of CNPase-positive cells. Importantly, this pro-oligodendrogenic effect of AEPO was most conspicuous in the pSVZ-derived cells (Fig. 5). Moreover, the long-term AEPO treatment caused long-lasting improvements in sensorimotor function and myelination (Fig. 6). A similar delayed treatment protocol might be applied to patients after the development of an HI injury has been diagnosed. Taken together, our findings in mice indicate that long-term postinjury AEPO treatment shows promise as a safe and efficient clinical intervention for neonatal white-matter injury.
The pro-oligodendrogenic effects of peripherally administered AEPO on OPCs could be mediated by direct and indirect mechanisms. For example, AEPO promotes angiogenesis, which may indirectly support the survival and/or differentiation of many types of brain cells [35, 38]. On the other hand, because EPO inhibits the apoptosis of erythrocyte progenitor cells through the activation of Bcl-xL , it is also possible that AEPO attenuated the apoptosis of OPCs after injury. Our in vitro experiments using purified OPCs indicated that AEPO directly promoted the maturation of OPCs without affecting their proliferation or survival (Fig. 3), as reported previously . In contrast, however, we found that in vivo, AEPO protected the OPCs from a long-lasting increase in apoptosis following neonatal HI (Fig. 4), in addition to its maturation-promoting effect. This discrepancy might be attributable to the difference in the OPCs' microenvironment: there was no ischemic insult, and the culture lacked activated immune cells as well as other glial cells. It is likely that the AEPO-induced increase in the number of mature oligodendrocytes observed in the HI-injured brain was due to these multiple effects of AEPO on multiple cell types.
How does AEPO act directly on OPCs? Our siRNA experiments in OPC culture with little contamination of other cell types indicated that AEPO's effect was mediated by EPOR expressed on the OPCs (Fig. 3). EPOR is involved in two types of receptors: the classical EPOR consisting of homodimeric EPORs, and a heteroreceptor consisting of a hetero-oligomeric complex of EPOR and the common-β chain receptor. However, the expression level of the common-β chain is extremely low or absent in the neonatal and adult brain . The homodimeric EPOR has a high affinity for AEPO . Furthermore, EPOR is widely distributed in the brain and increases in response to several brain insults [49, 85, 96], and OPCs were the largest population of EPOR-expressing cells in the pSVZ and the CC after HI in our study, suggesting that the effect of AEPO on OPCs is mediated by this receptor, at least in part. Further investigation is needed to determine the specific effect of EPOR on OPCs after HI in vivo.
In conclusion, our study indicates that the pSVZ-derived OPCs are a promising therapeutic target for the injured white matter after neonatal HI. Several weeks of AEPO treatment may be applicable to neonates with white matter diseases, such as periventricular leukomalacia, a main cause of cerebral palsy.
We thank the anonymous reviewers for their helpful comments on the manuscript, Dr. Jacqueline Trotter (Mainz University) for the gift of the rat monoclonal anti-AN2 antibody , Dr. Fred H. Gage (Salk Institute) for providing the CAG-GFP retrovirus through Addgene , and Chugai Pharmaceutical Co. Ltd. for the gift of the recombinant human erythropoietin. We also thank members of the Sawamoto laboratory for technical assistance and useful discussions. This work was supported by the Funding Program for the Next Generation World-Leading Researchers (to K. Sawamoto); a Grant-in-Aid for Young Scientists (S) (to K. Sawamoto); by the International Human Frontier Science Program Organization (to K. Sawamoto); by the Takeda Science Foundation (to K. Sawamoto); and by the Project for Realization of Regenerative Medicine from MEXT (to K. Sawamoto). H. Takebayashi is currently affiliated with Division of Neurobiology and Anatomy, Graduate School of Medical and Dental Sciences, Niigata University, Niigata, Niigata 951-8510, Japan.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.