HIF-1α Signaling Upstream of NKX2.5 Is Required for Cardiac Development in Xenopus*

HIF-1α is originally identified as a transcription factor that activates gene expression in response to hypoxia. In metazoans, HIF-1α functions as a master regulator of oxygen homeostasis and regulates adaptive responses to change in oxygen tension during embryogenesis, tissue ischemia, and tumorigenesis. Because Hif-1α-deficient mice exhibit a number of developmental defects, the precise role of HIF-1α in early cardiac morphogenesis has been uncertain. Therefore, to clarify the role of HIF-1α in heart development, we investigated the effect of knockdown of HIF-1α in Xenopus embryos using antisense morpholino oligonucleotide microinjection techniques. Knockdown of HIF-1α resulted in defects of cardiogenesis. Whole mount in situ hybridization for cardiac troponin I (cTnI) showed the two separated populations of cardiomyocytes, which is indicative of cardia bifida, in HIF-1α-depleted embryos. Furthermore, the depletion of HIF-1α led to the reduction in cTnI expression, suggesting the correlation between HIF-1α and cardiac differentiation. We further examined the expression of several heart markers, nkx2.5, gata4, tbx5, bmp4, hand1, and hand2 in HIF-1α-depleted embryos. Among them, the expression of nkx2.5 was significantly reduced. Luciferase reporter assay using the Nkx2.5 promoter showed that knockdown of HIF-1α decreased its promoter activity. The cardiac abnormality in the HIF-1α-depleted embryo was restored with co-injection of nkx2.5 mRNA. Collectively, these findings reveal that HIF-1α-regulated nkx2.5 expression is required for heart development in Xenopus.

Hypoxia inducible factor-1 (HIF-1) 2 is a transcriptional factor that has the fundamental function in mammalian develop-ment and homeostasis, and is considered to become a target for therapy in cancer and cardiac ischemia (1). In response to localized tissue hypoxia, HIF-1 is activated to regulate the transcription of hypoxia-inducible genes that mediate angiogenesis, erythropoiesis, vasodilation, and anaerobic metabolism (2). Under normoxic conditions, the ␣ subunit of HIF-1 (HIF-1␣) undergoes rapid decay via a ubiqutin-proteasome degradation pathway involving the von Hippel-Lindau tumor suppressor gene product (pVHL) (3)(4)(5). Hypoxia prevents ␣ subunit from ubiquitination, thereby allowing HIF-1␣ to escape from proteolysis, to dimerize with constitutively expressed HIF-1␤ (6), and to translocate into the nucleus with HIF-1␤.
In the early development of mammals, the natural progression of organogenesis involves hypoxia. Diffusion of oxygen in the embryo is limited by its size shortly after gastrulation. In turn, molecular responses to oxygen gradients by HIF-1 are responsible for proper differentiation and maintenance of the cardiovascular system (7). Null mutation of either Hif-1␣ or Hif-1␤ in mice leads to midgestational lethality of embryos, accompanied with phenotypes that include defects in the vasculature, placenta, and heart at embryonic day 10.5 (E10.5) (2,8,9). Thus, improper response to low oxygen in the embryo leads to abnormal cardiovascular morphogenesis. It is suggested that hypoxia may control the formation and function of the hemangioblasts (7), bipotential precursors for both endothelial cells and hematopoietic cells in the differentiation model using embryonic stem cell (10).
In this study, we examine the role of HIF-1␣ for cardiac development in Xenopus using microinjection techniques. Gain or loss of HIF-1␣ in the experiment reveals the contribution of HIF-1␣ to development of endothelial cells and hematopoietic cells, and the requirement of HIF-1␣ for cardiac differentiation, respectively. We further show that HIF-1␣ regulates the promoter activity of Nkx2.5 that is essential for cardiomyocyte differentiation and proper distribution. Decrease in HIF-1␣ during cardiac morphogenesis results in cardia bifida caused by the disturbance of cardiomyocyte distribution. These data indicate the requirements of HIF-1␣ for Xenopus heart development.
A constitutively active form of Xenopus hif-1␣ (P558G) (13) was developed using the QuikChange mutagenesis protocol (Stratagene) with oligos 5Ј-gatgttggcagggtatattcctat-3Ј and its complementary sequence. After sequence verification, the Bsu36I-NheI fragment containing the desired mutation was exchanged with the corresponding region of wild-type hif-1␣ in pCS2ϩ. A constitutively active form of mouse Hif-1␣ (P577G) was similarly developed by mutagenesis with oligos 5Ј-gatgctggctggctatatcccaat-3Ј and its complementary sequence.
Embryo Manipulation and Microinjection-Fertilized Xenopus laevis embryos were generated by standard techniques and staged, according to Nieuwkoop and Faber (14). For blastomere injection, regularly cleaving embryos were selected. The synthetic mRNA encoding hif-1␣ (P558G) (750 pg) was microinjected into one of the two vegetal-dorsal blastomeres of the selected embryos at the 8-cell stage, leaving the other side as control. Knockdown was carried out using MO (40 ng) similarly to mRNA injection. For the double injection to an embryo, MO was microinjected into both of the vegetal-dorsal blastomeres at 40 ng/blastomere. For lineage tracing, mRNA encoding green fluorescent protein (GFP) (500 pg) was co-injected with mRNA or MOs into the blastomere. During and after injection, embryos were maintained in 3% Ficoll in Steinberg's solution at 20°C. After incubation overnight, embryos were separated to two groups (right or left side group) according to the position of the lineage tracer, GFP, using fluorescent microscopy. Then embryos were transferred to Steinberg's solution containing 10 units/ml penicillin and 10 g/ml streptomycin, and collected at stage 33/34.
In Vitro Translation and Transcription System-The DNA consisting of 5Ј-UTR (102 bp), N-terminal of hif-1␣ (300 bp) (BC043769), and venus, a variant of GFP (DQ092360), was inserted into BamHI and EcoRI sites of the pCS2ϩ vector (5ЈUTR-HIF(N)-Venus). In vitro transcription and translation of 5ЈUTR-HIF(N)-Venus was performed by using an in vitro transcription and translation-coupled system (TNT TM Coupled Reticulocyto Lysate System; Promega), according to the manufacturer's protocol with the following modifications: in a 25-l reaction, 0.5 g of 5ЈUTR-HIF(N)-Venus and various amounts of hif-1␣ MO (0, 2, and 10 M final concentration) were added to the TNT reaction mixture and incubated at 30°C for 90 min. Five microliters from the reaction mixture were subjected to 10% SDS-PAGE followed by immunoblot analysis with anti-GFP antibody, kindly provided by Dr. N. Mochizuki (National Cardiovascular Center Research Institute).
Immunoprecipitation of Endogenous HIF-1␣-Total cell lysates were prepared from Xenopus embryos microinjected with hif-1␣ MO in one of two vegetal-dorsal blastomeres at the 8-cell stage. At stage 20, 10 embryos were lysed in 800 l of immunoprecipitation buffer (20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 10 mM MgCl 2 , 1% Triton X-100, 1 mM EGTA, 1 mM EDTA, 1 mM dithiothreitol and protease inhibitor mixture (Roche)). After incubation with gentle agitation for 30 min at 4°C, the extracts were centrifuged at 10,000 ϫ g for 15 min. 400 g of supernatants were used for immunoprecipitation. After pre-treatment with normal mouse serum IgG (Santa Cruz, Santa Cruz, CA) and protein G-Sepharose (GE Healthcare) for 1 h, the supernatants were incubated with shaking at 4°C in the presence of 2 g of anti-HIF-1a monoclonal antibody (H1alpha67, Novus Biologicals, Littleton, CO) or normal mouse IgG overnight prior to the addition of 10 l of protein G-Sepharose for 1 h. Beads were washed in immunoprecipitation buffer three times. Eluted samples with 2ϫ SDS sample buffer were subjected to immunoblot analysis using the same anti-HIF-1␣ antibody.
Whole Mount in Situ Hybridization and Histological Analysis-Whole mount in situ hybridization was performed as described previously (15) using digoxigenin-labeled antisense RNA probes transcribed with a DIG RNA Labeling kit (Roche). Positive signals were visualized using BM purple (Roche). For the detection of two probes (tie2 and ␣-globin) in the same sample, one and then the other was synthesized with DIG-11-UTP and fluorescein-UTP, respectively (Roche). Signals were visualized using BM purple for one (tie2) and with DAB (Vector Laboratories, Burlingame, CA) for the other (␣-globin) (16). Embryos were dehydrated and embedded in paraffin after whole mount in situ hybridization. Transverse sections were made at 7-m intervals and counterstained with eosin.
The pGL4.10 -9kb Nkx2.5 promoter (Nkx2.5 promoter-Luc) (100 pg) and pRL-null (2 pg) (an internal control) were injected into one side of the vegetal-dorsal blastomeres at the 8-cell stage. In the same way, the promoter/reporter constructs were co-injected with 40 ng of CMO or hif-1␣ MO1. Embryos injected with luciferase reporter constructs were collected at stage 19 -20 and homogenized in passive lysis buffer using 30 l of buffer/embryo. Samples were left on ice for 10 min followed by centrifugation (10,000 ϫ g) for 5 min at 4°C to pellet debris. Luciferase assays were performed with 10 l of clear supernatant and 100 l of firefly or Renilla luciferase substrate using a Lumat LB 9507 luminometer (Berthold Technology, Bad Wildbad, Germany). For each sample, 15 pools containing 5 embryos each were used. The experiment was repeated three times.
Visualization of Nkx2.5 Promoter Activity in Vivo-cDNA encoding Venus was replaced with that encoding Luciferase in pGL4.10[luc2] (Promega) with the NcoI and XhoI sites (pGL4.10-Venus). The mouse 9kb-Nkx2.5 promoter was cleaved out from the Nkx2.5 promoter-Luc with KpnI and NcoI, and ligated into same site of pGL4.10-Venus. The Nkx2.5 promoter-Venus (100 pg) was injected into one of the two vegetal-dorsal blastomeres with CMO or hif-1␣ MO1. The expression pattern and the activity of the Nkx2.5 promoter were analyzed with green fluorescence at stage 20.
Statistical Analysis-Statistical significance was evaluated by one-way analysis of variance followed by Student's t test for unpaired values. Differences with a p value less than 0.05 were considered significant.

Expression Pattern of Xenopus HIF-1␣ during Development-
We compared the sequence of Xenopus hif-1␣ cDNA that we cloned with those of human (Fig. 1A). The deduced amino acid sequence of Xenopus hif-1␣ showed high identity with human. The domains of basic helix-loop-helix, Per-Arnt-Sim (PAS), N-terminal transactivation domain, and the C-terminal transactivation domain were conserved in the predicted Xenopus HIF-1␣.
To examine whether hif-1␣ is expressed in Xenopus embryos, the expression of hif-1␣ mRNA was analyzed by RT-PCR. The hif-1␣ mRNA was detected in fertilized eggs and in embryos from stages 1 to 25 (Fig. 1B). Furthermore, to examine the spatial expression of hif-1␣ mRNA in embryos at later stages, we performed whole mount in situ hybridization analysis. At mid-neurula (stage 23) the transcripts were detected along the dorsal axis in the brain and somites (Fig. 1C, panel a). At approximately stage 28, when the first cardiac differentiation markers are expressed, immediately prior to fusion of the heart tube, hif-1␣ expression was observed in the heart primordia (Fig. 1C, panel b). In tailbud embryos (stage 33/34), hif-1␣ continued to be expressed in the branchial arches, somites, and heart (Fig. 1C, panels c, e, and f) as well as in the ventral blood island (VBI) (Fig. 1C, panel d), eyes, and kidneys (Fig. 1C, panel c), similar to mouse and human.
Role of HIF-1␣ in Vascular Development-HIF-1␣ is known to activate the transcription of genes encoding angiogenic growth factors such as vascular endothelial growth factor, which is secreted by hypoxic cells and stimulates endothelial cells to promote angiogenesis (17). To understand the role of HIF-1␣ for vascular formation, we examined whether HIF-1␣ affects the distribution of primitive blood and endothelial cells in developing embryos. For the gain of function experiment, the constitutively active form of hif-1␣ (P558G) mRNA was injected into one of the two vegetal-dorsal blastomeres of embryos at the 8-cell stage (18,19). The other un-injected vegetal-dorsal blastomere of the embryos served as an internal control for the injected. Co-injection of GFP mRNA enabled us to identify the side that was derived from the blastomere injected with mRNAs. The embryos were harvested at tailbud stage 32, and analyzed by whole mount double in situ hybridization for tie-2, a specific marker of differentiating vascular cells (20), and ␣-globin, a hematopoietic cell marker. Overexpression of HIF-1␣ (P558G) augmented the formation of the vascular vitelline network (VVN) meshwork with an increase in endothelial cell number, whereas the increased HIF-1␣ level had little effect on the formation of the VBI ( Fig. 2A, panels e-h).
The potency and specificity of these MOs were tested by an in vitro transcription and translation system. The translation of the HIF(N)-Venus protein was inhibited by hif-1␣ MO1 and hif-1␣ MO2, but not CMO (Fig. 2C). We examined the specificity of hif-1␣ MOs in vivo by co-injecting the 5ЈUTR-HIF(N)-Venus DNA with MOs into one of the two vegetal-dorsal blastomeres at the 8-cell stage embryos. Green fluorescence from HIF(N)-Venus was dramatically reduced by co-injection of hif-1␣ MOs compared with CMO (Fig. 2D). The inhibition of HIF(N)-Venus translation in vivo was confirmed by immunoblotting using anti-GFP antibody (Fig. 2E). These results indicate the specificity and efficiency of hif-1␣ MOs in vivo.
We further examined the effect of hif-1␣ on endogenous expression of HIF-1␣ (Fig. 2F). HIF-1␣ in embryos from those treated with hif-1␣ MOs was significantly reduced to 75% of that treated with CMO, although it was not completely decreased, because hif-1␣ MOs were injected into one of the two blastomeres at the 8-cell stage embryos. These results revealed that hif-1␣ MOs could be capable of knockdown of HIF-1␣ in embryos.
Injection of hif-1␣ MO1 caused a marked reduction in the expression of tie-2 and ␣-globin, specific markers of differentiating vascular cells and hematopoietic cells, respectively, in the VVN meshwork and VBI area ( Fig. 2A, panels i-l), indicating the decrease in endothelial cells and hematopoietic cells. These results imply that HIF-1␣ might be important for the generation of the common precursors of vascular endothelial cells and hematopoietic cells.
Depletion of HIF-1␣ Results in Cardia Bifida-Although hif-1␣ is expressed during cardiogenesis and in cultured embryonic ventricular myocytes (21), the role of HIF-1␣ in cardiac differentiation and development has been not well known. We performed knockdown of HIF-1␣ in Xenopus embryos by injecting hif-1␣ MO into one of the two vegetal-dorsal blastomeres. In this manner, the un-injected side serves as an internal control. Consistent with the previous report (22), these two vegetal-dorsal blastomeres contributed to heart and somitic tissues. Compared with control tadpoles at stage 42, HIF-1␣-depleted tadpoles showed cardiac malformations. They had small or undetectable hearts, although wild-type embryos have beating hearts (Fig. 3A). In addition, we could not observe the blood flow in most of the HIF-1␣-depleted tadpoles (Table 1).
To further investigate the role of HIF-1␣ in cardiac development, we examined the expression of a cardiac differentiation marker, cTnI (cardiac troponin I), at the tailbud stage (stage 33/34), at which the myocardium has formed a tube in all but the most anterior region. cTnI transcripts were restricted to two diffuse lateral patches in hif-1␣ MO-injected embryos (Fig.  3B, panels c and e), whereas it encompassed a large triangular area in the ventral midline in wild-type embryos (Fig. 3B, panel  a). The formation of the heart tube was not observed in transverse sections of hif-1␣ MO-injected embryos (Fig. 3B, panels d and f). This phenotype was attributable to the failure in migration of the bilateral heart precursors into the ventral midline and fusion of these cells, which is called cardia bifida. Cardia bifida was increased by hif-1␣ MO injection (hif-1␣ MO1, 82.7%; hif-1␣ MO2, 52.9%), compared with control MO injection (CMO, 14.7%) ( Table 2). The hif-1␣ MO1 was used for the following experiment as a representative of hif-1␣ MO. Quantitation of differentiated myocardium (marked by cTnI expression) showed that the area of the HIF-1␣-depleted side was smaller (or absent) than the control side. Remarkable reduction of cTnI expression was observed in 92% of embryos (supplementary Fig. S1 and supplementary Table 1). At the later stage (stage 42), the area of cTnI expression was smaller or not detected at the midline in the hif-1␣ MO1-injected embryos, compared with CMO-injected embryos (supplementary Fig.  S2, a-c). The transverse section more clearly showed that the pre-cardiac region was derived from the blastomere, where hif-1␣ MO injected was small or undetectable and lay asym-

TABLE 2 Frequency of cardia bifida in embryos injected with antisense morpholino against HIF-1␣
Xenopus embryos at 8-cell stage were microinjected with morpholino oligonucleotides (CMO, n ϭ 34; hif-1␣ MO1, n ϭ 75; hif-1␣ MO2, n ϭ 68). All embryos were assayed for expression of cTnI using whole mount in situ hybridization at stage 33/34. The frequency of cardia bifida was increased by hif-1␣ MOs injection compared with CMO injection (refer to Fig. 3B).  APRIL 25, 2008 • VOLUME 283 • NUMBER 17 metrically (supplementary Fig. S2, d-f). Injection of hif-1␣ MO1 into both of the two vegetal-dorsal blastomeres of the 8-cell stage also induced cardia bifida. 3 cTnI expression was reduced in both of these two pre-cardiac regions. These results suggest that suppression of cardiomyocyte differentiation before the migration into midline might become a cause of cardia bifida in the HIF-1␣-depleted embryos. Cardiac Gene Expression in HIF-1␣-depleted Embryos-To analyze the molecular basis of the defect in heart development of HIF-1␣-depleted embryos, we examined the expression of five transcriptional factors (nkx2.5, gata4, tbx5, hand1, and hand2) and one inductive molecule (bmp4). BMP, a member of the transforming growth factor-␤ superfamily, is necessary to direct the specification and terminal differentiation of cardiomyocytes by regulating the expression of early cardiac transcription factors, Nkx2.5 and Gata4 (23,24). Nkx2.5 is the earliest marker for cardiac region specification, and regulates the transcription of genes required for cardiac differentiation (25). At the early stage of heart development, the expression of Nkx2.5 is regulated by BMP4 and GATA4, the GATA zincfinger transcription factors (12). The basic helix-loop-helix transcription factors (Hand1, Hand2) and T-box transcription factor (Tbx5) are considered to be markers of later steps of cardiac development (26 -30).

HIF-1␣ Regulates Cardiac Development
In situ hybridization analyses revealed that the expression of these heart markers except bmp4 was dramatically reduced at the side derived from the blastomere where hif-1␣ MO1 was injected (Fig. 4). bmp4 expression was reduced in a few embryos (13%) ( Table 3). The frequency of the reduced expression of gata4, hand1, hand2, and tbx5 at the region was 66.7, 43.1, 43.9, and 50.0%, respectively. Notably, nkx2.5 was reduced in most of the embryos (93.7%). Reduction of tbx5, hand1, and hand2 expression appeared to be attributable to the inhibition of cardiomyocyte differentiation at an earlier stage, as revealed by the reduced expression of gata4 and nkx2.5. These findings, therefore, prompted us to examine whether HIF-1␣ regulates the expression of gata4 and nkx2.5 during early heart development.
Thus, we tried to investigate HIF-1␣-regulated nkx2.5 gene expression in Xenopus embryos. The expression profile of nkx2.5 during Xenopus development was examined. The nkx2.5 expression started at early neural plate stage 13 and remained throughout tail-bud stage 25, with a peak in neural tube stage 17-21, as assessed by RT-PCR (Fig. 5B). We considered that the embryos at stage 19 -21 were suitable for Nkx2.5 promoter assay according to the result of nkx2.5 expression.
Because it has been reported that several regulatory elements are conserved in the Nkx2.5 promoters of human, mouse, and Xenopus (31) and that structural domains are conserved in HIF-1␣ among them (Fig. 1A), we assumed that the mouse Nkx2.5 promoter could reflect the endogenous Xenopus HIF-1␣ function in embryos. To visualize Nkx2.5 promoter activity in embryos, we constructed the plasmid containing Nkx2.5 promoter-Venus. It allowed us to visualize Nkx2.5 promoter activity by yellow fluorescence in developing embryos. Nkx2.5 promoter activation was indeed observed around the cardiac primordial region of embryos at stage 20 (Fig. 5C, panels a and b), consistent with that found in transgenic mice harboring the transgene, Ϫ9 kb of Nkx2.5 5Ј flanking sequence linked to the reporter gene (12).  Table 3.

TABLE 3 Summary of reduction of cardiogenesis-related transcription factor in HIF-1␣-depleted embryos
Xenopus embryos at 8-cell stage were microinjected with hif-1␣ morpholino (hif-1␣ MO1) in one of the two blastomeres. The expression of transcription factors was analyzed in the side derived from the blastomere where MO was injected. The frequency of reduced expression of the transcription factors was analyzed at stage 33/34 according to the results from Fig. 4. We furthermore, examined the effect of hif-1␣ MO on the nkx2.5 transcriptional activity in Xenopus embryo. Nkx2.5 promoter-Luc or pGL4.10 vector was injected into one of the two vegetal-dorsal blastomeres of the 8-cell stage Xenopus embryos. Significant activation of the luciferase was observed in Nkx2.5 promoter Luc-injected embryos (Fig. 5D). Of importance, luciferase activity was decreased in the presence of hif-1␣ MO1 (Fig.  5E). To confirm this finding in vivo, we injected Nkx2.5 promoter-Venus with hif-1␣ MO and found that the fluorescence around the precardiac region was suppressed by depletion of HIF-1␣ (Fig. 5C, panel d). Consistently, RT-PCR analysis revealed the reduction of Nkx2.5 mRNA expression in hif-1␣ MO-injected embryos at stage 20 (Fig. 5F).
We also assumed that GATA4 was the other candidate factor downstream of HIF-1␣. In the luciferase assay using the Gata4 reporter gene, transcription of gata4 was reduced in HIF-1␣depleted Xenopus embryos (supplemental Fig. 3). However, it is not clear whether gata4 would have the responsibility for the phenotype caused by depletion of HIF-1␣. Although we tried to rescue cardia bifida by injecting gata4 mRNA, injection of gata4 mRNA induced severe gastrulation defects leading to death. 3

DISCUSSION
We identified HIF-1␣ as a regulator of cardiac development, as well as endothelial and blood cell differentiation using Xenopus embryos. Overexpression of HIF-1␣ induced the fine structure of the vascular VVN meshwork. In contrast, knockdown of HIF-1␣ resulted in the decrease in the number of both endothelial cells and blood cells. These results clearly indicate that HIF-1␣ controls the endothelial cell number and the maturation of blood vessels. In addition, HIF-1␣-depleted embryos exhibited the inhibition of cardiac differentiation and showed cardia bifida. A previous study reports that HIF mediates generation of mesoderm and that blood and vascular progenitor cells are crucial for early embryonic development (17). Several recent studies reveal that cardiomyocytes, endothelial cells, and smooth muscle cells arise from a common progenitor (32)(33)(34). We demonstrated that loss of HIF-1␣ affects the development of endothelial cells/blood cells and cardiomyocytes, suggesting that HIF-1␣ commonly regulates progenitors originated from the mesoderm.
We reported here that depletion of HIF-1␣ becomes the cause of cardia bifida in Xenopus. During cardiac development, splanchnic mesoderm symmetrically located on either side of the midline is committed to form cardiac tissue (35). The paired cardiogenic tissue migrates ventrally until it fuses in the ventral midline of the embryo (36,37). Following fusion, the heart tube loops to the right, converting the initial anterior-posterior pattern into left-right symmetry. Subsequently, it remodels into a multichambered heart, which is required for a functional organ (38). Disruption of one of these processes in vertebrates, failure of bilateral symmetrical heart primordial to fuse, has been shown in cardia bifida. This phenotype is known to be caused by defects in cardiac differentiation, endoderm differentiation, extracellular matrix deposition, or migration of the primordial heart. For instance, mice showing cardia bifida are caused by mutations of Gata4 (39), Mesp1 (40), furin (41), N-cadherin (42), or fibronectin (43). In Xenopus embryos, cardia bifida is caused by disruption of BMP signaling using an antagonist (SMAD6) or a dominant negative mutant, truncated BMP receptor (24,44). It is also caused by the inhibition of GATA6 (45) or WNT11 (46), non-canonical WNT signaling.
We propose here cardiac defects due to depletion of HIF-1␣ attributable to incomplete cardiomyocyte differentiation. Early cardiac gene expression is needed for the maintenance and activation of cardiac mesoderm differentiation (47). Depletion of HIF-1␣ resulted in the reduction of early cardiac gene expression in hif-1␣ MO-injected embryos (Fig. 4). On the other hand, embryos with the reduced wnt11-Related gene (wnt11-R) expression exhibited cardia bifida, but not the inhibition of cardiac differentiation (48). In that study, aberrant contact within the myocardial wall leads to defects in fusion of the heart tube. Comparing these results with our present results, we suggested that cardia bifida observed in HIF-1␣-depleted embryos may be due to the inhibition of cardiomyocyte differentiation in the one side derived from hif-1␣ MO-injected blastomere before the migration into midline, because several cardiac genes were down-regulated in the side depleted of HIF-1␣.
Nkx2.5, one of the earliest genes to be expressed specifically in cardiac precursor cells (49), is necessary for cardiac primordial differentiation and fusing into the single heart tube in Xenopus. Mutant embryos tinman, the divergent homeodomain homolog of Nkx2.5, fail to develop a heart (50,51). We noticed the reduced expression of nkx2.5 in HIF-1␣-depleted embryos. Interestingly, co-injection of Nkx2.5 mRNA prevented cardia bifida caused by depletion of HIF-1␣ (Fig. 6A). These results suggest that NKX2.5 is downstream of HIF-1␣ signaling. The Xenopus embryos injected with the dominant negative form of nkx2.5 exhibited un-fused cardiac primordial with reduced cTnI staining (52,53). Our HIF-1␣-depleted embryos exhibited a similar cTnI staining pattern. We further confirmed that NKX2.5 depletion with nkx2.5 MO inhibited the migration of heart primordial cells into the midline in Xenopus embryos (Fig. 6, B and C). We assume that NKX2.5 is required as transcriptional activators for the earliest stages of heart formation before cardiac primordial migrates into midline. However, in the mouse study, the targeted ablation of NKX2.5 does not inhibit the formation of the heart tube, although it blocks cardiac development at the stage of looping morphogenesis. This difference might be due to the redundancy in the mouse system, because NKX2.5 is critical for the initiation of cardiac cell fate decisions.
How does HIF-1␣ regulate the nkx2.5 gene? Regulation of the Nkx2.5 gene is highly complex, because there are multiple enhancers/repressor-binding sites in the regulatory regions of the mouse Nkx2.5 gene. Several consensus binding sites for GATA factors and SMADs have been functional in driving the expression of the mouse Nkx2.5 (26 -30). We suggest that HIF- to detect the precise migration of cardiomyocytes. Cardia bifida caused by depletion of HIF-1␣ was restored by co-injection of nkx2.5 mRNA. B, depletion of NKX2.5 results in cardia bifida in Xenopus embryos. CMO-injected embryos (a), nkx2.5 MO-injected embryos (b), and hif-1␣ MO1-injected embryos (c) were examined for cTnI by in situ hybridization at stage 33/34. NKX2.5-depleted embryos showed a similar staining pattern of cTnI to HIF-1␣-depleted embryos. Yellow arrowheads indicate the side derived from the blastomere where hif-1␣ MO1 was injected. C, the incidence of cardia bifida in nkx2.5 MO-injected embryos, hif-1␣ MO-injected embryos, or CMO-injected embryos.
1␣-regulated nkx2.5 expression is not mediated by BMP4, because the embryos injected with hif-1␣ MO showed no change in the expression of bmp4 (Fig. 4). In the present study, depletion of HIF-1␣ affected nkx2.5 expression by suppressing its promoter activity in developing Xenopus embryos. To identify the HIF-1␣ binding region in the promoter, we constructed a series of deletion constructs of the Nkx2.5 promoter. Deletion from Ϫ9.5 to Ϫ2.5 kbp did not affect luciferase activity, whereas deletion from Ϫ9.5 to Ϫ1.2 kbp eliminated the activity, suggesting that the region from Ϫ2.5 to Ϫ1.2 kbp might be responsible for HIF-1␣ signaling. 3 Although HIF-1␣ depletion inhibited gata4 expression (Fig. 4), we could not demonstrate the relationship between HIF-1␣ and GATA4 acting upstream of NKX2.5. Further studies are required to explore whether NKX2.5 is directly or indirectly regulated by HIF-1␣.
Understanding the individual modular steps in cardiac morphogenesis is particularly relevant to congenital heart disease, which usually involves defects in specific structural components of the developing heart. The present study would provide new information that regulation of NKX2.5 by HIF-1␣ is necessary for normal cardiac development.