Hypoxia-inducible Factor-1α Inhibits Self-renewal of Mouse Embryonic Stem Cells in Vitro via Negative Regulation of the Leukemia Inhibitory Factor-STAT3 Pathway*

During mammalian embryogenesis, the early embryo grows in a relatively hypoxic environment due to a restricted supply of oxygen. The molecular mechanisms underlying modulation of self-renewal and differentiation of mouse embryonic stem cells (mESCs) under such hypoxic conditions remain to be established. Here, we show that hypoxia inhibits mESC self-renewal and induces early differentiation in vitro, even in the presence of leukemia inhibitory factor (LIF). These effects are mediated by down-regulation of the LIF-STAT3 signaling pathway. Under conditions of hypoxia, hypoxia-inducible factor-1α (HIF-1α) suppresses transcription of LIF-specific receptor (LIFR) by directly binding to the reverse hypoxia-responsive element located in the LIFR promoter. Ectopic expression and small interference RNA knockdown of HIF-1α verified the inhibitory effect on LIFR transcription. Our findings collectively suggest that hypoxia-induced in vitro differentiation of mESCs is triggered, at least in part, by the HIF-1α-mediated suppression of LIF-STAT3 signaling.

Analysis of mESCs 2 isolated from the inner cell mass of preimplantation embryos has aided in elucidating the early molecular events and mechanisms responsible for maintenance of ESC pluripotency (1). In contrast to human ESCs (hESCs), selfrenewal of mESCs can be sustained indefinitely in feeder-free conditions if supplemented with LIF (2). mESC lines can be derived directly from embryos in the presence of LIF, indicating that this factor is specifically required for the maintenance of pluripotency. Intracellular signaling cascades initiated by LIF are mediated through the heterodimerization of LIFR and glycoprotein 130 (gp130). This complex activates Janus-associated tyrosine kinase and the signal transducer and activator of transcription 3 (STAT3) signaling pathway. Activation of STAT3 by LIF is crucial in mediating self-renewal signals in mESCs (3,4).
During mammalian development, oxygen needed for cellular metabolism in the early embryo is supplied by simple diffusion from fluids within the oviduct and uterus. Thus, the early embryo develops in a relatively hypoxic environment until the onset of vascularization after implantation (5). Actually, oxygen tension in the mammalian reproductive tract is reported to be less than half the atmospheric oxygen tension (6). Low oxygen tension triggers a wide range of cellular events centered on the regulation of hypoxia-inducible factor-1 (HIF-1) (7), which consists of a common ␤ subunit (HIF-1␤) and an oxygen-sensitive ␣ subunit (HIF-1␣). Under normoxia, HIF-1␣ is rapidly degraded via the von Hippel-Lindau protein-mediated ubiquitin-proteasome pathway (8). Although the role and biological relevance of HIF-1 during murine embryonic development have been established (9 -11), our understanding of how HIF-1 affects the early differentiation of mESCs at the molecular level is beginning to emerge.
Our previous study demonstrated the presence of hypoxic regions during normal mouse development (12), which prompted us to study the role of hypoxia in mESC self-renewal or differentiation during in vitro culture. In the present work, we have cultured feeder-free mESCs together with LIF under low O 2 tension (1% O 2 ). Our data show that activated HIF-1␣ under these conditions induces mESC differentiation via suppression of the LIF-STAT3 signaling pathway. Accordingly, our results propose the important roles of HIF-1␣ in regulating the early differentiation of mESCs.
Cell Proliferation Assay-CCE cells were seeded in gelatincoated 48-well plates at a density of 2 ϫ 10 4 cells per well and cultured under indicated conditions. The proliferation rates of the cells were then measured using a Non-Radioactive Cell Proliferation Assay Kit (Promega, Madison, WI) according to the manufacturer's instructions. Briefly, 50 l of freshly mixed tetrazolium/phenazine methosulfate was added, and the cells were incubated for 2 h to allow color development. The absorbance at 490 nm was measured to indicate the number of viable proliferating cells.
AP Staining-Self-renewal of mESCs was determined by morphological assessment and alkaline phosphatase (AP) staining using a diagnostic kit (Sigma). The ratios of AP-positive colonies were scored.
RT-PCR Analysis-Total RNA was extracted form cultured cells using TRIzol reagent (Invitrogen), and quantified with a spectrophotometer (NanoDrop, Nyxor Biotech). First strand cDNA was synthesized using Moloney murine leukemia virus reverse transcriptase (Promega) with 5 g of each DNA-free total RNA sample and oligo(dT) 15 , according to the manufacturer's instructions. cDNA (1 l) was amplified by PCR using the ExTaq DNA polymerase kit (Takara). Information for specific primers used for RT-PCR analysis is given in supplemental Table S1.
FACS Analysis-For cell cycle analysis, cells were detached with trypsin-EDTA and fixed overnight in ice-cold 70% ethanol. Fixed cells were washed with phosphate-buffered saline and incubated with 0.1 g/l RNase A and 50 g/ml propidium iodide at 37°C for 30 min. Quantitative analysis was performed using a BD Biosciences FACSCalibur TM flow cytometer, Cell-Quest Pro software package (BD Biosciences), and ModFit software (Verity Software House).
Immunofluorescence-Cells were fixed in 4% paraformaldehyde for 20 min at room temperature, washed gently, and blocked with 0.1% Triton X-100, 1% bovine serum albumin, 10% goat serum in phosphate-buffered saline at room temperature for 30 min. Next, cells were incubated with LIFR antibody (Santa Cruz Biotechnology) overnight at 4°C. After three washed in phosphate-buffered saline, cells were incubated with secondary fluorescein isothiocyanate-conjugated IgG (Pierce). Nuclear counterstaining was performed using Hoechst (Sigma). Fluorescence staining was visualized using a fluorescence microscope (Carl Zeiss).
ChIP Analysis-ChIP was performed with the ChIP assay kit (Upstate), according to the manufacturer's protocol. The PCR primers, 5Ј-GAGCTGAGAGGTCCCCTCA-3Ј and 5Ј-GCCA-GCCTTGCAAGGTCA-3Ј, were designed to amplify a region of the LIFR promoter (Ϫ806 to Ϫ497) harboring at least two rHREs (Ϫ739 to Ϫ735 and Ϫ553 to Ϫ549).
Luciferase Assay-A partial genomic DNA sequence encompassing the promoter region of LIFR bearing three rHREs was obtained by genomic PCR and cloned into an enhancerless luciferase pGL3 promoter vector (Promega). CCE cells were plated at a density of 2 ϫ 10 5 cells per well of a 6-well plate and transfected with various combinations of effector plasmids. After transfection, luciferase assays were performed using the Luciferase Assay System kit (Promega) and a luminometer (Turner Design). The relative luciferase activity was normalized to relative light units/␤-galactosidase activity.
Immunohistochemistry-Tissue sections from mouse embryonic liver and retina were quenched with hydrogen peroxidase, blocked with serum-free protein blocker (DAKO), and incubated overnight at 4°C with HIF-1␣ (Novus) and LIFRspecific antibodies (Santa Cruz Biotechnology). Biotinylated anti-mouse/rabbit IgG (DAKO) was used to label the bound primary antibodies. Immunoreactivity was detected with streptavidin-conjugated peroxidase (DAKO) and diaminobenzidine (DAKO) as a chromogen. Sections were counterstained with Mayer's hematoxylin (Sigma).
Statistical Analysis-Data are expressed as means Ϯ S.D., and analysis was performed using the Student's t test.

RESULTS
Hypoxia Induces Early Differentiation of mESCs-In view of the limited information available on how mESCs respond to low oxygen tension, we analyzed the effect of hypoxia (1% O 2 ) on the proliferation and morphology of mESCs. Proliferation of CCE cells revealed that hypoxia in the presence of LIF inhibited stem cell growth in a time-dependent manner compared with normoxia ( Fig. 1A, right panel). It is well known that LIF acts as a distinct factor for enhancing the survival and proliferation of mESCS, thus enabling the long term propagation of undifferentiated cells (13). Consistently with this, LIF-withdrawn cells under normoxia showed a more reduced proliferation rate (Fig.  1A, right panel). Furthermore, under normoxia (21% O 2 ) in the presence of LIF, CCE cells displayed typical compact colony morphology, a fundamental characteristic of undifferentiation (Fig. 1A). In contrast, upon LIF withdrawal, cells became largely flattened and sharpen at the end of cells, a phenotype characteristic of differentiation. Interestingly, when cultured under hypoxia, even in the presence of LIF, cells grew as monolayers with fibroblast-like morphology comparable to that of differentiated cells (Fig. 1A). In parallel with these morphological changes, AP staining revealed a significant decrease in the proportion of undifferentiated colonies under hypoxia (Fig. 1B). Based on these results, RT-PCR analysis was conducted to assess the expression of ESC marker genes. Under normoxic condition in the presence of LIF (up to 48 h), there was no difference in expression level of ESC markers. Thus, we selected 48-h normoxic sample as an end-time point control to compare with that of hypoxic samples. Compared with normoxia, downregulation of Oct4, Fgf4, and Rex-1, representing mESC pluri-potency, and up-regulation of Fgf5 associated with early differentiation were observed for 24-to 48-h hypoxia in the presence of LIF, consistent with the expression patterns of markers observed in LIF-withdrawn CCE cells (Fig. 1C). This differentiation pattern was confirmed in other mESC lines, such as R1 and E14TG2a (supplemental Fig. S1).
The results indicate that mESCs, cultivated under hypoxia in the presence of LIF, are committed to differentiate at the expense of reduced proliferation and self-renewal activity.
Apoptosis, a genetic program leading to cell death, is evident in the inner cell mass of late pre-implantation embryos and epiblastic cores of early post-implantation embryos in vivo (14). Additionally, mESCs are irreversibly committed to differentiation upon LIF withdrawal, and parts of the cells undergo apoptosis during in vitro differentiation (15). To verify that apoptosis is implicated in hypoxiainduced differentiation of mESCs, the cell cycle pattern was assessed by FACS. As shown in Fig. 1D, mESCs cultivated under normoxia displayed a unique cell cycle pattern composed largely of cells in the S-phase. However, the proportion of cells in the G 0 /G 1 phase was elevated under hypoxia for 24 h. Furthermore, an increased proportion of apoptotic cells in the sub G 1 phase was detected at 48 h under hypoxia (7.9%), comparable to that of LIF-withdrawn cells (10.9%). Consistently, proteolytic activation of caspase-3 and cleavage of its substrate, poly(ADPribose) polymerase, were elevated under hypoxia, indicating that mESCs respond to low O 2 tension through a change in G 1 phase arrest and subsequent apoptosis, even in the presence of LIF (Fig. 1E). These results suggest that hypoxia induces early differentiation of mESCs accompanying apoptosis, similar to that frequently observed upon LIF withdrawal.
Hypoxia Down-regulates the LIF-STAT3 Signaling in mESCs-STAT3 activation is important for the maintenance of pluripotent mESCs. Notably, phosphorylation at tyrosine 705 is crucial for self-renewal activity of mESCs (4). To confirm the inhibitory role of hypoxia in LIF-mediated mESC self-renewal, we initially examined phosphorylated STAT3 (Tyr-705), a target of LIF signaling, after depletion of LIF and serum. Phosphorylation of STAT3 was rapidly induced by LIF under normoxic conditions, whereas hypoxia markedly impaired the ability of LIF to activate STAT3 ( Fig. 2A). Thus, hypoxia is likely to be implicated in regulation of the LIF-STAT3 signaling pathway of mESCs. To characterize the mechanism by which hypoxia attenuates the phosphorylated STAT3 level, we investigated whether LIFR, an upstream signaling receptor involved in STAT3 activation, is down-regulated by hypoxia. Western blot analysis revealed a marked time-dependent decrease in LIFR protein expression under hypoxia in CCE cells that had been adapted to grow in serum-containing medium supplemented with LIF. The decrease was consistent with the reduction in phosphorylated STAT3 levels (Fig. 2B). We additionally analyzed c-Myc expression under the same conditions. This oncoprotein was recently identified as a functional target of STAT3 in mESC self-renewal (16). Upon down-regulation of LIF-STAT3 signaling under hypoxia, there was a concomitant drop in the level of c-Myc (Fig. 2B). The results imply that hypoxia disrupts LIF-STAT3 signaling, which is crucial for the maintenance of mESC self-renewal.
Down-regulation of LIFR in hypoxic mESCs was further confirmed by monitoring its expression patterns at the cellular level. Analysis of cell surface expression of LIFR assessed by immunostaining revealed that mESCs formed a compact colony and abundantly expressed LIFR in the cell membrane under normoxia, whereas LIFR expression was diminished under hypoxia (Fig. 2C). This result clearly suggests that hypoxia down-regulates the expression of LIFR, an upstream signaling receptor involved in STAT3 activation. In accordance with the previous report (17), retinoic acid, a well known differentiation factor, also inhibited LIFR expression in our culture conditions (Fig. 2D), further supporting that hypoxia induces mESC differentiation through down-regulation of LIFR.
HIF-1␣ Decreases LIFR Transcription by Binding to rHREs of LIFR Promoter-To determine the mechanism by which hypoxia down-regulates LIF-STAT3 signaling and whether O 2 tension reversibly modulates LIFR expression, we evaluated the effects of hypoxia and reoxygenation on LIFR expression. The levels of LIFR protein and its mRNA transcript were dramatically reduced under hypoxia and gradually restored during reoxygenation (Fig. 3A). In contrast, HIF-1␣ was stabilized under hypoxia, but rapidly degraded during reoxygenation, as determined by Western blot analysis (Fig. 3A). These results support the possible involvement of HIF-1␣ in down-regulation of LIFR under hypoxia.
We then attempted to elucidate the mechanisms by which HIF-1␣ could regulate LIFR transcription. HIF-1␣ is a major transcription factor that activates the transcription of several genes involved in oxygen homeostasis, including vascular endothelial growth factor and erythropoietin (EPO), by binding to specific hypoxia-responsive elements (HREs) in their promoters (18). An earlier report (19) shows that HIF-1␣ acts as a transcriptional repressor of certain genes by direct binding to specific reverse HREs (rHREs), HREs on the antisense strand. Interestingly, the mouse LIFR promoter contains three potential rHREs (DNA consensus 5Ј-TGCAC-3Ј located at positions Ϫ894 to Ϫ890, Ϫ739 to Ϫ735, and Ϫ553 to Ϫ549 from the translation start site) (20) (Fig. 3B). In view of this finding, we investigated the possibility of LIFR as a direct transcriptional target of HIF-1␣. EMSAs were performed to examine the binding of HIF-1␣ to the three putative rHREs of the LIFR promoter.
EMSA data confirmed that all three potential rHREs act as binding sites for HIF-1␣. Notably, HIF-1␣ binding to rHRE2 was stronger than that to rHRE1 or rHRE3 sites, possibly due to the differences in adjacent sequences (Fig. 3C, left panel). Binding of HIF-1␣ to rHREs was abrogated or supershifted upon preincubation with the competitor or HIF-1␣ antibody, respectively, which ensured the specificity of these interactions (Fig.  3C, right panel). We additionally performed the ChIP assay to verify HIF-1␣ interactions with rHREs in mESCs. Promoter occupancy was investigated using specific primers spanning two rHREs (Ϫ806 to Ϫ497) of LIFR. In CCE and R1 cells, the specific binding of HIF-1␣ to the LIFR promoter was observed under hypoxia (Fig. 3D).
Next, we determined the effects of HIF-1␣ on transcriptional activity of the LIFR promoter containing three rHREs (LIFR-rHREs) by luciferase reporter gene assays. Hypoxia induced a significant decrease in the reporter gene activity in cells transfected with LIFR-rHREs. Co-transfection with HIF-1␣/1␤ vectors decreased luciferase activity under both normoxic and hypoxic conditions (Fig. 3E, left panel), supporting the involvement of HIF-1␣ in suppressing LIFR promoter activity through binding to rHREs. In a parallel experiment performed with a reporter vector containing four HREs of EPO (EPO-HREs), there was a significant increase in the reporter gene activity in cells cotransfected with HIF-1␣/1␤ (Fig. 3E, right panel). Therefore, it appears that HIF-1 decreases LIFR promoter activity by binding to LIFR-rHREs, whereas enhances EPO promoter activity by binding to EPO-HREs. It remains possible that the interaction with other transcriptional cofactors dictated by the promoter architecture allows the opposite transcriptional activity of HIF-1␣.
HIF-1␣ Is Involved in Early Differentiation of mESCs by Regulating the LIF-STAT3 Signaling-To confirm the hypoxic responsiveness of LIFR, the experiment was repeated with other mESC lines, including J1, R1, and E14TG2a. LIFR was consistently down-regulated under hypoxia, which was associated with increased stability of HIF-1␣ (Fig. 4A). We further addressed whether forced expression of HIF-1␣ triggers early differentiation of mESCs under normoxia. Overexpression of TM-HIF-1␣ vector, which contains a triple mutation of the hydroxylatable residues (21), resulted in more stable expression of HIF-1␣, and this led to marked decrease in LIFR expression in J1 cells (Fig. 4B, top panel). Also, overexpression of wild-type and TM-HIF-1␣ into J1 cells induced the appearance of differentiated cells (Fig. 4B, middle panel), and the expression of germ layer marker genes such as brachyury (mesoderm) and GATA-4 (endoderm) (Fig. 4B, bottom panel). In addition, siRNA knockdown of HIF-1␣ under hypoxia led to a substantial recovery in LIFR and phosphorylated STAT3 relative to the control (Fig. 4C). Although these effects were not very high, knockdown of HIF-1␣ under hypoxia increased the proportion of undifferentiated colony determined by AP activity, suggesting that HIF-1␣ is involved in early differentiation of mESCs at least in part (Fig. 4D). Furthermore, treatment with L-mimosine, a prolylhydroxylase inhibitor stabilizing HIF-1␣ (22), decreased LIFR-rHRE reporter gene activity and LIFR protein expression (supplemental Fig. S2A). Consistently, L-mimosine-treated cells underwent differentiation with characteristic morphological changes and marker gene expression (supplemental Fig. S2B). These results strongly suggest that HIF-1␣ induces early differentiation of mESCs by down-regulating LIFR expression.
To examine whether LIFR expression is affected by an elevated HIF-1␣ protein level in vivo, we examined the expression patterns of LIFR and HIF-1␣ in mouse embryonic liver, normal FIGURE 3. HIF-1␣ binds to rHREs of the LIFR promoter under hypoxic conditions. A, CCE cells were exposed to normoxia (N), hypoxia (H), or hypoxia for 6 h followed by reoxygenation for 5 and 30 min (5m and 30m), and 2 and 6 h (2h and 6h), respectively. At each time point, Western blot (top panel) and RT-PCR (bottom panel) analyses were conducted. B, upstream sequence of the mouse LIFR exon 1 promoter (GenBank TM accession number AF014933). Nucleotides are numbered relative to the translation initiation site, and three rHREs are marked by squares. C, EMSA was performed using nuclear extracts from CCE cells exposed to normoxia or hypoxia for 6 h. Competitor, unlabeled oligonucleotide; NS, nonspecific; SS, supershift. D, ChIP assay, soluble chromatin was prepared from CCE and R1 cells exposed to normoxia (N) or hypoxia for 6 h (6H) or 18 h (18H). retina, and ischemic retina by immunostaining. Immunoreactivity of HIF-1␣ was evident in the embryonic liver at E13.5-16.5 stages, however, weak or no expression of LIFR was observed in HIF-1␣-expressing regions of embryonic liver (supplemental Fig. S3A). Furthermore, in normal retinal tissues (CL-), LIFR was primarily identified in the retinal ganglion cell layer (GCL) and inner nuclear layer (INL) (supplemental Fig.  S3B). After ischemia treatment, LIFR was not detected through all retinal layers at 24-h post-ischemia (supplemental Fig. S3B). However, HIF-1␣ displayed enhanced expression in tissues surrounding blood vessels of the GCL and upper and lower limits of the INL at 24 h in post-ischemic retina, but not in normal retina (supplemental Fig. S3B). This finding is consistent with the previously reported patterns of HIF-1␣ expression in ischemic retina (23). Taken together, these results indicate that HIF-1␣ and LIFR expression are reciprocally regulated in mouse tissues in vivo, lending support to our proposed model of association between HIF-1␣ and LIFR.

DISCUSSION
Low oxygen tension is known to induce angiogenesis and glycolysis for cell growth and survival and leads to growth arrest and apoptosis (24). Previously, Carmeliet et al. (25) suggested that hypoxia reduces stem cell proliferation and increases apoptosis through HIF-1␣. Accordantly, a recent report (26) suggested that ESC apoptosis is found to be dependent on the cel- lular differentiation and on the duration and degree of hypoxia. In this context, it is interesting to note that low O 2 tension inhibits differentiation of hESCs (27) and human marrow stromal cells (28). Notably, LIFR expression is low or absent in hESCs, indicating that the LIF-STAT3 signaling pathway does not appear to be involved in self-renewal of hESCs (29).
Our results suggest that the molecular milieu of mESC differentiation under low oxygen tension involves HIF-1-mediated suppression of the LIF-STAT3 signaling pathway in vitro (Fig. 4E). Under normoxia, mESCs retain self-renewal activity through STAT3 activation in the presence of LIF. However, under hypoxia, stabilized HIF-1␣ binds to rHREs of the LIFR promoter, represses LIFR expression, and attenuates STAT3 phosphorylation. Consequently, loss of LIF-STAT3 signaling inhibits self-renewal and induces early differentiation of mESCs in our in vitro culture. Consistent with this, it was reported that the number of AP-positive colonies was reduced in lifr Ϫ/Ϫ mESCs, compared with the wild-type mESCs in vitro (30), supporting the important role of LIF signaling in mESC proliferation and self-renewal.
LIF and LIFR mRNAs are expressed in the mouse blastocyst in a reciprocal pattern between the trophectoderm and the inner cell mass (31), suggesting their involvement in regulation of pluripotent cell survival in vivo. Nonetheless, an alternative, intrinsic determinant that is redundant to LIF signaling could be involved in the maintenance of inner cell mass. Recent reports suggest that Nanog maintains mESC self-renewal independently of the LIF-STAT3 pathway (32,33). However, there appears to be cross-talk between Nanog and the LIF-STAT3 pathway to promote maximal self-renewal efficiency of mESCs (32,33). In addition, STAT3 activated by LIF regenerates pluripotent mESCs from early mesoderm-specified progenitor cells via up-regulation of Nanog expression (34). These findings suggest that such intrinsic and extrinsic factors responsible for mESC pluripotency could act cooperatively in vivo. Therefore, it will be a crucial task to clarify the functional relevance of these interactions in mouse embryo. At the same time, it remains to be clarified whether such alternative signaling pathways are regulated by O 2 tension in vitro and in vivo.
In summary, present data collectively provide insights into the mechanisms by which hypoxia inhibits self-renewal and induces early differentiation of mESCs via suppression of the LIF-STAT3 signaling in vitro. Moreover, we provide evidence that HIF-1␣ acts as a negative regulator of the LIF-STAT3 signaling, thereby establishing a possible link between early differentiation of mESCs and low O 2 tension.