Author contribution: T.M.: conception and design, financial support, administrative support, provision of study material or patients, collection and/or assembly of data, data analysis and interpretation, and manuscript writing; T.S.: administrative support, collection and/or assembly of data, and data analysis and interpretation; K.I.: provision of study material or patients; and K.N.: conception and design, financial support, administrative support, data analysis and interpretation, and final approval of manuscript. T.M. and T.S. contributed equally to this study.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS December 29, 2011.
Oxygen levels in tissues including the embryonic brain are lower than those in the atmosphere. We reported previously that Notch signal activation induces demethylation of astrocytic genes, conferring astrocyte differentiation ability on midgestational neural precursor cells (mgNPCs). Here, we show that the oxygen sensor hypoxia-inducible factor 1α (HIF1α) plays a critical role in astrocytic gene demethylation in mgNPCs by cooperating with the Notch signaling pathway. Expression of constitutively active HIF1α and a hyperoxic environment, respectively, promoted and impeded astrocyte differentiation in the developing brain. Our findings suggest that hypoxia contributes to the appropriate scheduling of mgNPC fate determination. STEM CELLS2012;30:561–569
The mammalian embryonic brain contains multipotent neural precursor cells (NPCs) that can self-renew and give rise to all three major cell types in the nervous system, that is, neurons, astrocytes, and oligodendrocytes. NPC fate determination is regulated by the collaboration between cell-external cues, from molecules such as cytokines and growth factors, and cell-internal epigenetic programs .
We and others have shown that members of the interleukin-6 (IL-6) family of cytokines, including leukemia inhibitory factor (LIF), efficiently induce astrocyte differentiation of late-gestational (lg) NPCs by activating the janus kinase (JAK)-signal transducer and activator of transcription (STAT) signaling pathway [2, 3]. In contrast to lgNPCs, midgestational (mg) NPCs differentiate only into neurons because the promoters of astrocytic genes, such as that encoding the typical astrocytic marker glial fibrillary acidic protein (GFAP), are hypermethylated at this stage [4, 5]. These findings indicate that DNA methylation is a critical determinant for the acquisition of astrocyte differentiation ability by NPCs . Notch receptors and their ligands, molecules best known for influencing cell fate decisions through direct cell-cell contact, participate in a wide variety of biological processes, including fate determination of NPCs. As gestation proceeds, mgNPCs generate committed neuronal precursors (or young neurons) expressing Notch ligands, and these cells activate Notch signaling in neighboring NPCs. Notch signal activation induces the demethylation of astrocytic gene promoters in NPCs, which explains, at least in part, how the neuronal-to-glial cell fate switch occurs in NPCs during brain development [1, 6, 7].
Oxygen is essential for cell survival as the final electron acceptor in the electron transport chain of aerobic respiration. Sensing and controlling oxygen localization are of extreme importance because normal metabolic processes generate radical intermediates. When oxygen levels exceed the capacity of endogenous antioxidant systems, radicals including reactive oxygen species attack cellular components such as nucleic acids, the sulfhydryl groups of proteins, or the unsaturated fatty acid moieties of phospholipids .
Hypoxia-inducible factors (HIFs) are known to play critical roles as molecular sensors of oxygen tension. HIF1α activity in cells is controlled by oxygen levels in multiple ways, such as transcriptional regulation and hydroxylation-dependent binding by Von Hippel-Lindau tumor suppressor protein followed by proteasomal degradation [9–11]. A HIF1 complex composed of HIF1α and HIF1β (also known as ARNT) binds to the hypoxia response element and activates the transcription of target genes such as inhibitor of differentiation 2 , insulin-like growth factor binding protein, p21, phosphoglycerate kinase 1, and vascular endothelial growth factor . HIF1α has also been reported to be crucial for normal brain development . Moreover, HIFs may act cooperatively with other signaling molecules, such as Notch  and mammalian target of rapamycin [16, 17], thereby influencing a wide range of processes including tumor malignancy and NPC growth, maintenance and differentiation [11, 18-20].
In this study, we focused on how oxygen tension affects the DNA methylation status of astrocytic genes in mgNPCs. Oxygen levels in the microenvironment around NPCs in the embryonic brain at midgestation were comparable to those seen under hypoxic culture conditions (2-5% O2). Bisulfite sequencing revealed that these conditions promoted demethylation of the gfap promoter. This hypoxia-induced demethylation was mediated by cooperation between HIF1α and the Notch signaling pathway. Furthermore, ectopic expression of a constitutively active form of HIF1α in the embryonic brain induced precocious astrocyte differentiation of NPCs. By contrast, when embryos developed in hyperoxic conditions, astrocyte differentiation of NPCs was suppressed. These findings suggest that oxygen levels in the embryonic brain play a critical role in fine-tuning the timing of NPC fate switching during development.
MATERIALS AND METHODS
mgNPCs were prepared from telencephalons of E11.5 embryos and cultured as described previously . E15.5 ventricular zone (VZ) containing NPCs was manually separated from other parts of the brain using a hand-made microknife (Supporting Information Fig. S1). As described previously , NPCs were cultured with basic fibroblast growth factor (bFGF; 1 × 106 cells per dish) in poly-L-ornithine/fibronectin-coated 6-cm culture dishes. Hand-made chambers were used to obtain atmospheres of different oxygen levels . Notch signal activation was inhibited with the γ-secretase inhibitor N-[N-(3,5-difluorophenacetyl-L-alanyl)]-S-phenylglycine t-butyl ester (DAPT; Calbiochem-MERCK, Darmstadt, Germany, www.merckgroup. com, 10 μM).
Genomic DNA was extracted from NPCs and subjected to bisulfite sequencing as previously described . Specific DNA fragments were amplified by polymerase chain reaction (PCR) using primers described previously [5, 7]. The PCR products were cloned into pT7Blue vector (Novagen, Darmstadt, Germany, www.merckgroup.com/en/index.html), and 10-20 randomly picked clones were sequenced. Each experiment was performed at least tree times.
A mouse HIF1α-targeting short hairpin RNA (shRNA) sequence  was cloned into pLLX vector . Lentiviruses were pseudotyped with the vesicular somatitis virus–G envelope and concentrated by centrifugation as previously described . HIF1α mRNA degradation was confirmed by quantitative PCR (qPCR). All the recombinant DNA experiments in this manuscript followed the guidelines by Ministry of Education, Culture, Sports, Science and Technology of Japan, which conform to the National Institutes of Health Guidelines.
Cells were fixed with 4% paraformaldehyde and processed for immunostaining as described . The following primary antibodies were used: chick anti-green fluorescent protein (anti-GFP; 1:500, Aves Labs, Tigard, OR, www.aveslab.com), rat anti-hemagglutinating (anti-HA) (1:500, Roche Applied Science, Indianapolis, IN, www.roche-applied-science.com), rat anti-bromodeoxyuridine (1:250, AbD Serotec, Raleigh, NC, www.abdserotec.com) chicken anti-Nestin (1:1000, Aves Labs), mouse anti-Map2ab (1:500, Sigma, St. Louis, MO, www.sigmaaldrich.com), and mouse anti-GFAP (1:500, Sigma). Secondary antibodies were Alexa488-conjugated goat anti-rat IgG (1:500), Alexa488-conjugated goat anti-chick IgY (1:500), Alexa555-conjugated goat anti-rabbit IgG (1:500), Alexa555-conjugated goat anti-rat IgG (1:500), Alexa555-conjugated goat anti-mouse IgG (1:500), or Alexa647-conjugated goat anti-mouse IgG (1:500, Invitrogen, Carlsbad, CA, www.invitrogen.com). Nuclei were stained using bisbenzimide H33258 fluorochrome trihydrochloride (Hoechst; Nacalai Tesque, Kyoto, Japan, www.nacalai.co.jp). All experiments were independently replicated at least three times. A Hypoxyprobe-1 Plus kit (Hypoxyprobe-Millipore, Billerica, MA, www. millipore.com) was used, following the manufacturer's protocol, to determine the hypoxicity of cultured cells and the developing brain. After culturing NPCs for 4 days in 2%, 5%, or 21% O2, pimonidazole-HCl was added to the culture medium for 1.5 hours and the cells were then fixed. For brain sections, pimonidazole-HCl was injected intraperitoneally into pregnant mice (E15.5). Embryos were fixed 2 hours later with 4% paraformaldehyde and the brains were cryosectioned at 20-μm intervals. Pimonidazole adducts were detected with a fluorescein isothiocyanate–conjugated specific antibody supplied with the kit (1:500). Stained sections were visualized with a confocal microscope (Fluoview FV10i, Olympus, Tokyo, Japan, www.olympus.co.jp) or a fluorescence microscope (Zeiss Axiovert 200M, Zeiss, Jena, Germany, www.zeiss.com).
Animal Procedures and Electroporation
All aspects of animal care and treatment were conducted according to the guidelines of the Experimental Animal Care Committee of Nara Institute of Science and Technology. The surgical procedures performed on pregnant ICR mice and embryo manipulations in utero were conducted as previously described . E11.5 pregnant mice were deeply anesthetized by intraperitoneal injection with sodium pentobarbitone (50 μg/g b.wt.). HA–constitutively active HIF1α (HA-caHIF1α) cDNA was cloned into the EcoRI site of pCAGGS vector. After the uterus was exposed, approximately 1-2 μl of plasmid solution (1 μg/μl in phosphate-buffered saline) was injected into the lateral ventricle of the telencephalon with a glass micropipette. The embryos were held with the tips of a tweezers-type electrode with a diameter of 5 mm (CUY650-P3; Tokiwa Science, Fukuoka, Japan), and five electronic pulses (27V, 50 ms, at intervals of 950 ms) were given to each embryo with an electroporator (CUY21SC; Tokiwa Science). The embryos were reinserted into the abdominal cavity and the abdominal wall was closed with surgical sutures.
Exposure to High-Oxygen Atmosphere and Tissue Oxygen Measurement
The animal cage was placed in a hyperoxic oxygen chamber (Terucom, Kanagawa, Japan, www.terucom.co.jp). Oxygen tension was measured using a portable oxygen meter (Terucom) in the same chamber. E11.5 pregnant mice were exposed to 60% O2 for 2 days, and the oxygen tension was then raised to 80% (the upper nonlethal concentration limit) and maintained at that level until fixation at E17.5.
Local tissue oxygen tension corresponding to each atmospheric oxygen level was measured with an OxyLab pO2 monitor (Oxford Optronix, Oxford, UK, www.oxford-optronix. com)  (Supporting Information Fig. S4Q). The probe was guided to the E15.5 brain using a 20-gauge needle with a plastic canula (Terumo, Tokyo, Japan, www.terumo.co.jp).
It is well known that lgNPCs but not mgNPCs differentiate into astrocytes in response to stimulation with IL-6 family cytokines [2-5, 7]. When we cultured lgNPCs prepared from the VZ of mouse telencephalon at E15.5 (Supporting Information Fig. S1), we observed their LIF-induced astrocyte differentiation, as judged by the expression of the astrocytic marker GFAP (Fig. 1C, 26.8 ± 6.5%). By contrast, hardly any LIF-treated E11.5 mgNPCs differentiated into astrocytes (Fig. 1A). Interestingly, E11.5 mgNPCs cultured in vitro for 4 days (nominally corresponding to E15.5) did undergo astrocyte differentiation (Fig. 1B, 12.2 ± 1.9%), albeit to a lesser extent than E15.5 lgNPCs.
Since an inverse correlation exists between the potential of NPCs to express gfap and the methylation status of the gfap promoter, which includes a STAT3-binding site [5, 27], we examined whether the in vitro culture conditions of mgNPCs delayed the demethylation of the promoter compared to its demethylation in vivo. Bisulfite sequencing for the gfap promoter of E11.5, E11.5 + 4-day culture in vitro, and E15.5 NPCs revealed that this was indeed the case (Fig. 1D, 1E). These data indicate that the in vitro culture conditions retard the demethylation of the astrocytic gene promoter in NPCs.
In terms of the physical conditions surrounding cells, one of the biggest differences between in vitro and in vivo environments is oxygen tension. The atmosphere contains 21% O2 (160 mmHg), whereas interstitial oxygen concentration ranges from 1 to 5% (7-40 mmHg) in mammalian tissues including the embryonic brain [11, 20]. Although hypoxia is generally considered a pathological phenomenon, mammalian embryos develop naturally in a mildly hypoxic environment . To determine the tissue oxygen tension in the embryonic brain, where NPCs reside, we used the chemical reagent pimonidazole. Pimonidazole is reductively activated in hypoxic cells and forms stable adducts with sulfhydryl groups in amino acids, at around or below 10 mmHg, which can be detected with specific antibodies [29, 30]. Pimonidazole adducts were weakly but clearly detected in NPCs cultured in 2% and 5% O2 conditions, respectively, but not in NPCs cultured in 21% O2 (Fig. 2A–2C). Pimonidazole adducts were also abundant in midgestational embryonic brain, indicating that oxygen levels there are low, particularly in the NPC-containing VZ (Fig. 2D–2G).
From the above findings, we hypothesized that culturing mgNPCs at a low oxygen level might accelerate their differentiation into astrocytes. To test this idea, mgNPCs were cultured in 2% or 21% O2 atmospheres for 4 days in the presence of bFGF, and subsequently stimulated with LIF for an additional 4 days. As shown in Figure 3A–3C, a higher proportion of mgNPCs cultured in the presence of LIF under 2% O2 than under 21% O2 became GFAP-positive astrocytes. Reflecting the promotion of astrocyte differentiation, NPC (Nestin) and neuron (Map2ab) marker-positive cell numbers were lower under 2% O2 than under 21% O2 (Supporting Information Fig. S2A). The proliferation of cells in our normoxic and hypoxic culture was similar as judged by BrdU staining (Supporting Information Fig. S2B). Furthermore, in the 2% O2 condition, significantly fewer CpG sites in the gfap promoter were methylated than in the 21% O2 condition (Fig. 3D); this was also the case for CpG sites in the promoter of another astrocytic marker, S100β (Supporting Information Fig. S3A). Consistent with the lower methylation, S100β expression in hypoxic culture in the presence of LIF was higher than that in normoxic condition (Supporting Information Fig. S3B). These results suggest that low oxygen levels facilitate the DNA demethylation of astrocytic genes in mgNPCs.
We have shown previously that Notch signal activation induces the demethylation of astrocytic genes in mgNPCs . Intriguingly, hairy and enhancer of split 1 (Hes1) and Hes5, two known targets of Notch signaling, were markedly upregulated in the 2% O2 culture condition (Fig. 4A), indicating that Notch signaling is more active at lower O2 levels. Therefore, we examined whether low oxygen levels enhance the astrocyte differentiation of mgNPCs via Notch signal activation. An inhibitor of Notch signal activation, DAPT, was added to the culture medium during the first 4-day expansion phase of mgNPCs with bFGF in the hypoxic condition. The expression of Hes1 and Hes5 was dramatically reduced, confirming that hypoxia-induced elevation of Notch signaling was inhibited by DAPT (Supporting Information Fig. S3C). Furthermore, as shown in Figure 4B and 4C, the astrogenic potential of NPCs was almost completely suppressed by DAPT, and the enhanced demethylation of the gfap and S100β promoters observed in the hypoxic condition was no longer seen after DAPT treatment (Fig. 4D and Supporting Information Fig. S3D).
Activation of the Notch signaling pathway in mgNPCs induces expression of the transcription factor nuclear factor IA (NFIA), which leads to demethylation of astrocyte-specific genes including gfap : NFIA binds to astrocytic gene promoters and induces dissociation of DNA methyltransferase 1 (DNMT1) from the promoters, resulting in their demethylation. Therefore, we examined the expression of Nfia and found that it increased dramatically under the hypoxic condition (Supporting Information Fig. S4A). Moreover, we did not observe astrocyte differentiation of NFIA-deficient NPCs even when cultured with LIF under the hypoxic condition (Supporting Information Fig. S4B). Taken together, these findings suggest that activation of the Notch-NFIA cascade is the mechanism whereby hypoxia induces demethylation of the gfap promoter.
HIF1s have been shown to play important roles in cellular adaptation to hypoxia [11, 13, 31]. HIF1α and HIF1β are expressed in the developing mouse brain , and HIF1α protein accumulates to a higher level in hypoxic than in normoxic conditions as a result of increased transcription and protein stabilization [33–35]. Consistent with those observations, we found that both HIF1α transcript and nuclear HIF1α protein levels in E11.5 mgNPCs were upregulated under the 2% O2 culture condition (Supporting Information Fig. S5A, S5B). Expression of the HIF1α target gene Id1 was also upregulated (Supporting Information Fig. S5C), as was observed in neuroblastoma cells .
To determine whether HIF1α contributes to the hypoxia-promoted astrogenic potential of mgNPCs, we suppressed HIF1α mRNA using a specific shRNA. mgNPCs were prepared from E11.5 telencephalon and were infected the following day with lentiviruses expressing HIF1α-shRNA, cultured under 2% O2 for 3 days, and then stimulated with LIF for a further 4 days. The level of HIF1α mRNA was reduced greatly in cells infected with HIF1α shRNA-expressing viruses (Supporting Information Fig. S6A). Moreover, HIF1α shRNA expression markedly diminished both the astrogenic potential of mgNPCs and the degree of demethylation in the gfap promoter (Fig. 5A–5G, and Supporting Information Fig. S6B).
caHIF1α is a hydroxylation-resistant, constitutively active mutant form of HIF1α in which two amino acids (prolines 402 and 564) are substituted with alanines . One-day in vitro-cultured E11.5 mgNPCs were infected with lentiviruses expressing HA-tagged caHIF1α (HA-caHIF1α) and cultured for a further day, and then subjected to LIF stimulation for 3 days under the normoxic condition. HA-caHIF1α protein was clearly detected in the nucleus (Fig. 5M). In contrast to control virus-infected mgNPCs (Fig. 5H–5K), cells transduced to express HA-caHIF1α became GFAP-positive astrocytes in response to LIF stimulation, even under the 21% O2 condition (Fig. 5L–5O). HA-caHIF1α expression also augmented Notch signal activation in these cells (Fig. 5P). Furthermore, HA-caHIF1α-induced astrocyte differentiation and Notch signal activation were both abolished by DAPT treatment (Fig. 5P and Supporting Information Fig. S6C–S6J). Consistent with the results of these in vitro experiments, ectopic expression of HA-caHIF1α in E11.5 brains led to precocious GFAP expression in NPCs (Supporting Information Fig. S6K–S6P). These experiments suggest that HIF1α promotes the astrogenic potential of mgNPCs by enhancing Notch signal activation, probably because hypoxia-stabilized HIF1α can form a complex with the Notch intracellular domain (NICD) to effectively induce the expression of Notch-target genes .
Given that the 21% O2 condition delayed the acquisition of astrocyte differentiation ability by NPCs in vitro, we next asked whether a hyperoxic environment has a similar effect on NPCs in vivo. To address this, pregnant mice were housed in a normoxic or hyperoxic chamber for 6 days (E11.5 to E17.5). To confirm that the embryonic brain was indeed under hyperoxia, we measured the local brain oxygen tension at E15.5 using an oxygen electrode. When the pregnant mice were in the normoxic (21% O2) and hyperoxic (80% O2) conditions, the oxygen tensions were 5.24 and 112.9 mmHg, respectively (Supporting Information Fig. S6Q), indicating that local brain oxygen tension was increased as a result of the atmospheric oxygen levels surrounding the pregnant mother mouse. After 6 days of hyperoxic housing, embryonic brains were then subjected to immunostaining and Western blotting. Interestingly, as shown in Figure 5Q–5S, exposure to high oxygen markedly decreased GFAP expression in the brains of embryos of mice housed in the hyperoxic condition.
These results suggest that a hypoxic environment is important for the proper timing of astrocyte differentiation during embryonic development.
HIF1α is known to be an important factor in the response of various types of cells to hypoxic stress such as ischemia [11, 13] and to play a critical role in normal brain development . In the present study, we have shown that astrocytic genes are progressively demethylated in the mildly hypoxic environment of the developing brain (Figs. 1–3). We further suggest that this demethylation process is executed by a collaboration between HIF1α and Notch signal activation (Figs. 4, 5), probably through the formation of a complex between HIF1α and NICD, to effectively induce the expression of Notch-target genes . The augmentation of Notch signal activation under hypoxia led to higher expression of Nfia (Supporting Information Fig. S4A), which has been shown to induce dissociation of DNMT1 from astrocytic gene promoters, resulting in demethylation of the promoters . It has been suggested that HIFs act cooperatively with other signaling molecules, thereby influencing a wide range of biological processes including NPC maintenance and differentiation [11, 18-20]. Nevertheless, we report for the first time, to the best of our knowledge, that oxygen levels not only modify the behavior of transcription factors but also affect the epigenetic status of genes. The promotion of astrocytic gene demethylation by the hypoxic condition contributes to specifying the appropriate timing of the neural-to-glial cell fate switch of NPCs, by ensuring a proper balance between neurons and astrocytes that are generated during brain development.
We also found in this study that oxygen tension affects the DNA methylation status of astrocytic genes in mgNPCs via HIF-Notch signaling. Strikingly, hypoxia, HIF1α, Notch signaling and DNA methylation are all known to participate in the onset and/or progression of glioblastoma [12, 13, 15, 37], the most common and malignant type of brain tumor. Thus, a deeper understanding of glial cell-generating mechanisms, including astrocyte differentiation, may be of therapeutic interest.
As has been shown in a previous study  and the present work, DNA demethylation in the astrocytic gene promoters of NPCs is crucial for astrocyte differentiation, and the efficacy of astrocyte differentiation of NPCs is influenced by oxygen levels in the brain throughout development. Premature infants in neonatal intensive care units (NICUs) are often incubated under hyperoxic conditions to support their immature respiration. Oxygen concentration is strictly controlled in NICUs, since it has been shown that excess oxygen administration causes retinopathy of prematurity . Furthermore, a few studies report other risks of hyperoxia to the central nervous system of extremely premature infants, including defects in mental and psychomotor development [39, 40]. The present study suggests that such effects may be attributable to an imbalance in NPC differentiation caused by a high level of oxygen in the incubator, and identifying their underlying mechanism in future work could provide new approaches for clinical applications that address developmental abnormalities in the nervous system, particularly in the context of neonatal intensive care.
We showed that local oxygen concentration can affect the fate of NPCs through an epigenetic mechanism. Unraveling how actual microenvironmental oxygen levels in the embryo influence epigenetic gene regulation in NPCs will provide new aspects for the study of NPC regulation.
We thank T. Kobayashi for the hyperoxic oxygen chamber, H. Harada, M. Hiraoka, T. Takizawa, M. Namihira, Y. Bessho, T. Matsui, Y. Nakahata, and K. Semi for valuable discussions and technical advice, I. Smith for editing the manuscript, and M. Tano for secretarial assistance. This work was supported by a Grant-in-Aid for Young Scientists (Start-up), Grant-in-Aid for Scientific Research on Innovative Area: Neural Diversity and Neocortical Organization, Grant-in-Aid for Research Program of Innovative Cell Biology by Innovative Technology and by the NAIST Global COE Program (Frontier Biosciences: Strategies for survival and adaptation in a changing global environment) from the Ministry of Education, Culture, Sports, Science and Technology of Japan; Health Sciences Research Grants from the Ministry of Health, Labor and Welfare, Japan. T.M. is currently affiliated with Biological Sciences, Gunma Kokusai Academy Secondary School, Ota, Gunma, Japan.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.