The impact of high-salt exposure on cardiovascular development in the early chick embryo

The cardiovascular system is the first system to emerge during embryonic development because it is responsible for distributing the blood supply to various tissues according to their requirements. The cardiovascular system is composed of the heart and the vascular system, which derive from both the embryonic mesoderm and the extraembryonic yolk sac. During early embryonic development, the heart initially forms in an embryonic disc as a simple paired heart tube inside the forming pericardial cavity. In this period, heart formation undergoes a series of transformations, starting from the formation of a straight heart tube deriving from the fusion of bilateral cardiomyogenesis cells in the primary/secondary heart fields on the embryonic midline (Yutzey and Kirby, 2002) and followed by the right looping of the heart tube and septation (Linask, 2003). Endoderm-derived signals such as bone morphogenetic protein (BMP), fibroblast growth factor (FGF) and Wnt antagonists are essential for precardiac mesoderm cells to differentiate into mature cardiomyocytes (Nakajima et al., 2009; Schultheiss et al., 1995) during cardiomyogenesis. Nkx2.5, GATA factors, myocardin and Tbx20 are well-known transcription factors that characterize and induce cardiogenic differentiation (Brand, 2003). Therefore, Nkx2.5 and BMP2 could be considered common cardiogenic factors.

Vascular development occurs in many regions as the mesoderm differentiates into blood islands. The blood islands then contribute to both the blood vessel epithelium and fetal red blood cells. Eventually, these blood islands join together to form the primary blood plexus that connects with the forming heart tube. Angiogenesis is thought to remodel and expand the vascular plexus through endothelial sprouting and intussusceptive microvascular growth (Patan, 2004). Located in the extra-embryonic region, the yolk sac membrane (YSM) is the first site where blood vessels and angioblasts develop. Its physiological function is to provide nutrition to the developing embryo. The chick chorioallantoic membrane (CAM) is a highly vascularized membrane underneath the inner surface of the eggshell. It is formed by the fusion of the chorionic membrane and the allantois during embryonic development. Both the YSM and CAM are excellent models for studying angiogenesis because they are easy to access and manipulate. There is an extensive literature about the genetic background, and the molecular and cellular mechanisms responsible for blood vessel formation during embryonic development. Fibroblast growth factor-2 (FGF2) and vascular endothelial growth factor (VEGF) play an important role in the angiogenic expansion of the early network (Marinaccio et al., 2013). Hypoxia activates hypoxia inducible factor-1 and 2-alpha (HIF-1α and HIF-2α) and could induce vessels to branch towards the hypoxic tissue (Pugh and Ratcliffe, 2003). Reactive oxygen species (ROS), as a component of oxidative phosphorylation, also play an important role in regulation of HIF-1α (Hagen, 2012), VEGF (Xia et al., 2007), FGF2 (Wang et al., 2015b), cardiovascular system (Brown and Griendling, 2015) and embryonic development (Dennery, 2007).

Chick embryos can easily be exposed to salt solutions, making them good models for analyzing the effects of hyperosmolarity on development. The early chick embryos are quite sensitive to external physiochemical compounds. In contrast to the growing mammalian embryo, salt exposure to the chick embryo is a pure high-salt osmolality model; in contrast to the situation in mammals, salt cannot be excreted by the chicken embryo, nor is there a prominent activation of a salt- and osmolality-regulating system such as the anti-diuretic hormone (ADH) system and the renin–angiotensin–aldosterone system (RAAS). This represents one kind of strength and a weakness of the model. In our previous study, we demonstrated that chick embryos exposed to excess salt have impaired retina and lens development (Chen et al., 2014) and induced neural tube defects (Jin et al., 2015). Here, we investigate the adverse impact of high-salt-induced hyperosmolarity on cardiovascular development using early chick embryos.

Fertilized chicken eggs were obtained from the Avian Farm of the South China Agriculture University. The eggs were incubated in a humidified incubator (Yiheng Instrument, Shanghai, China) at 38°C and 70% humidity. After 36 h of incubation, the chick embryos were treated as previously described (Chen et al., 2014; Jin et al., 2015): in brief, 2 ml albumen was removed from an egg and the egg was then injected with 500 µl of three different concentrations of NaCl. For the control group, we injected 0.7% NaCl (calculated final osmolality of the egg was 240 mosm l⁻¹); for the 300 mosm l⁻¹ group, we injected 16.85% NaCl (calculated final osmolality of the egg, 300 mosm l⁻¹); for the vitamin C (VC) rescue experiment (Li et al., 2014), we administered VC (0.5 mg egg⁻¹) after injection of 16.85% NaCl. The treated embryos were incubated for a further 72 h. For treatment of the CAM, we injected either simple saline (500 µl of 0.7% salt solution) or high-salt solution (500 µl of 16.85% salt solution) to the blunt side of the egg at day 6. The treated embryos were incubated for a further 48 h. The embryos were firstly determined to be alive or dead by the presence/absence of a heartbeat. Surviving embryos were harvested for assessment of other parameters. All dead embryos were excluded from further analysis.

Hematoxylin and Eosin staining (H&E staining) was performed on the corresponding transverse embryonic sections [including embryonic day (E)4.5 chick embryos, E4.5 YSM and E8.0 CAM] according to standard protocols. The thickness of the ventricular walls and the length and number of blood vessels were then measured. Ventricular wall thickness was measured across the transverse plane of the chick ventricular wall from the outer surface to the cardiac lumen at five different points. The trabecular muscle thickness was measured similarly. Blood vessel length and number were measured on the transverse sections of chick YSM. To assess blood vessel density, the blood vessels around the hearts were photographed using a stereomicroscope (Olympus MVX10) and an OPTPRO 2007 image acquisition system. The blood vessel plexus in the same location in the hearts was quantified using an IPP 5.0 image analysis program.

Chick embryos were harvested after high-salt exposure and fixed in 4% paraformaldehyde (PFA) overnight at 4°C. Whole-mount chick embryo immunostaining was performed using MF20 or phospho-histone H3 (PH3). Briefly, the embryos were washed in phosphate-buffered saline (PBS) and incubated with a primary monoclonal antibody mixture raised against MF20 (1:200, DSHB) or PH3 (1:400, Santa Cruz Biotechnology) overnight at 4°C on a shaker. After extensive rinsing in PBS, the embryos were incubated with Alexa Fluor 555 anti-mouse IgG or anti-rabbit IgG secondary antibody (1:1000, Invitrogen, USA), respectively, at 4°C overnight on a shaker. All embryos were later counterstained with DAPI (1:1000, Invitrogen) at room temperature for 1 h. All immunofluorescence staining was performed in replicate from at least 5–6 embryos

High-salinity culture [1.125% NaCl:albumen (1:1); calculated final osmolality of the culture, 300 mosm l⁻¹] was applied to one side of stage HH3 onwards chick embryos during early chicken (EC) culture as previously described (He et al., 2013,, 2014; Wang et al., 2015a). The other side of embryos treated with simple saline served as the control [0.7% NaCl:albumen (1:1); calculated final osmolality of the culture, 300 mosm l⁻¹] (Chapman et al., 2001). Whole-mount in situ hybridization of chick embryos was performed according to standard in situ hybridization protocols (Henrique et al., 1995). Digoxigenin-labeled probes were synthesized against VE-cadherin (Yang et al., 2008). The whole-mount stained embryos were photographed and frozen sections were prepared by sectioning at a thickness of 15–20 μm on a cryostat microtome (Leica CM1900).

BrdU stock solution (10 mmol l⁻¹, 10 µl) was administered to high-salt treatment or control eggs/chick embryos 4 h before the embryos reached 4.5 days. Transverse sections of E4.5 chick hearts were stained with a monoclonal antibody against BrdU according to the manufacturer's instructions (Roche). Cell counts were performed from a defined area (visual field at 20× magnification). We counted BrdU-positive cells and the total cell number to calculate the proliferation index (BrdU-positive cells/total cell number).

The hearts of E4.5 chicken embryos were fixed in Bouin's solution and paraffin embedded. Sections (4 µm) were deparaffinized and stained with an in situ cell death detection kit (Roche) according to the manufacturer's instructions. The extent of cell death was quantified by counting TUNEL⁺ cells on consecutive transverse sections of treated and untreated E4.5 chick embryonic hearts and YSM. We counted the TUNEL-positive cells and the total cell number to calculate the apoptotic index (TUNEL-positive cells/total cell number).

Human umbilical vascular endothelial cells (HUVECs) were cultured in a humidified incubator at 37°C and 5% CO₂ inside 96-well plates. The cells (1×10⁶ cells ml⁻¹) were maintained in DMEM (Gibco, USA) plus 10% fetal bovine serum (Gibco, Australia). 1 ml of 16.85% NaCl (high osmolarity) or 1 ml 0.7% NaCl (control) were added to the 100 ml HUVEC cultures. Levels of intracellular ROS were determined using a non-fluorescent dye DCF-DA (2′,7′-dichlorodihydrofluorescein diacetate) (Sigma-Aldrich), which is oxidized by ROS generated by cells into a fluorescent dye DCF (2′,7′-dichlorofluorescin). The control and high-salt-treated HUVECs were incubated in the presence of 10 μmol l⁻¹ DCF-DA for 20 min. The fluorescence intensity was measured using a BD FACSAria.

Total RNA was isolated from chick heart and YSM tissues using TRIzol (Invitrogen, USA) according to the manufacturer's instructions. First-strand cDNA was synthesized to a final volume of 25 μl using the SuperScript RIII first-strand synthesis system (Invitrogen, USA). Following reverse transcription, PCR amplification of the cDNA was performed as described previously (Dugaiczyk et al., 1983; Maroto et al., 1997). The primers were as follows: PPIA, TGACAAGGTGCCCATAACAG and GCGTAAAGTCACCACCCTGA; cyclin D1, TCGGTGTCCTACTTCAAGTG and GGAGTTGTCGGTGTAAATGC (Scott-Drechsel et al., 2013); VMHC, GCTACAAACACCAAGCAGAG and TCTTATATCTGGGAGCCAGG (Schlueter et al., 2006); Nkx2.5, GGATCCCTCCTCGTTGCTCTCG and CCTTGACACGCCGCTCTGACTT (Schlueter et al., 2006); GATA4, CCACACGACCACAACCACACTCTG and AAAGCTTCAGGGCTGAAATTGCAG (Zhang et al., 2003); BMP2, AGCGTCAAGCGAAACACAAACAG and GGGCAACAATCCAGTCATTCCAC (Endo et al., 2012); HIF2, CCGAGCGTGACTTCTTCATGAGG and GCTCATCGTCAACAGGTGTGGCT (Larger et al., 2004); FGF2, TTCTTCCTGCGCATCAAC and GGATAGCTTTCTGTCCAG (Larger et al., 2004); VEGFR, GGTCGCATGAACATGAAGAA and TTGGTAGGGTTTGTAAGGAC; SOD1, AGGAGTGGCAGAAGTAG and CACGGAAGAGCAAGTA; SOD2, CTTCCTGACCTGCCTTAC and CGTCCCTGCTCCTTATT; Gpx, GCCACCTCCATCTACGAC and TGCTCCTTCAGCCACTTC. PCR reactions were performed in a Bio-Rad S1000TM thermal cycler (Bio-Rad, USA). The final reaction volume was 50 μl, composed of 1 μl first-strand cDNA, 25 μmol l⁻¹ forward primer, 25 μmol l⁻¹ reverse primer, 10 μl PrimeSTARTM Buffer (Mg²⁺ plus), 4 μl dNTPs mixture (TaKaRa, Japan), 0.5 μl PrimeSTARTM HS DNA Polymerase (2.5 U μl⁻¹; TaKaRa, Japan), and RNase-free water to 50 μl. The cDNAs were amplified for 30 cycles. One round of amplification was performed at 94°C for 30 s, 58°C for 30 s and 72°C for 30 s. The PCR products (20 μl) were resolved in 1% agarose gels (Biowest, Spain) in 1× TAE buffer (0.04 mol l⁻¹ Tris acetate and 0.001 mol l⁻¹ EDTA) and 10,000× GeneGreen Nucleic Acid Dye (Tiangen, China) solution. The resolved products were visualized with a transilluminator (Syngene, UK), and photographs were captured using a computer-assisted gel documentation system (Syngene). The housekeeping gene PPIA (peptidylprolyl isomerase A) was run in parallel to confirm that equal amounts of RNA were used in each reaction. The ratio between intensity of the fluorescently stained bands corresponding to genes and PPIA was calculated to quantify the level of the transcripts for those mRNAs (De Simone et al., 2005; Wang et al., 2015a). The RT-PCR result was representative of four independent experiments.

Following immunohistochemistry, whole-mount embryos were photographed using a stereo-fluorescent microscope (Olympus MVX10) associated with the Olympus software package Image-Pro Plus 7.0. The embryos were sectioned into 15 µm slices using a cryostat microtome (Leica CM1900) and then photographed using an epifluorescent microscope (Olympus LX51, Leica DM 4000B) with a CN4000 FISH Olympus software package.

The assays of blood islands density and blood vessel density (BVD) were performed according to the previous description (Chen et al., 2014; He et al., 2013). Data analyses and creation of charts were performed using a GraphPad Prism 5 software package (GraphPad Software, CA, USA). The data were presented as the mean±s.d. Statistical significances were determined using a non-parametric test. P<0.05 was considered to be significant.

In our previous experiments investigating the influence of high-salt exposure on eye and neural tube development (Chen et al., 2014), we found embryonic lethality. In this study, we carefully tracked the possible mechanisms underlying the embryonic lethality. To mimic a high-salt environment, 500 µl of 16.85% salt solution or simple saline (0.7%, control) was injected into fertilized eggs through the eggshell window (Fig. 1A). Thus, the final osmolarity surrounding the developing embryos was 300 mOsm in the high-salt group (N=30) and 240 mOsm in the control group (N=10). The treated eggs were incubated for 4.5 days. After 4.5 days, 20 out of 30 embryos were dead after the high-salt exposure (Fig. 1E).

The cardiovascular system is the first organ system to develop, as subsequent embryonic development relies on its physiological function. Severe embryonic heart defects could therefore cause embryonic death. We first compared the developing hearts of the control (240 mOsm, Fig. 1B) and high osmolarity (300 mOsm, Fig. 1C,D) embryos. We found two abnormal heart development phenotypes in E4.5 chick embryos in the high-salt group (Fig. 1C,D) compared with control embryos (Fig. 1B). In the control embryonic heart (N=10; Fig. 1B), we clearly saw the appearance of the fully looped heart, while either abnormal looping of heart tube (28%, N=29; Fig. 1C) or blood congestion (17%, N=29; Fig. 1D) occurred in the high-salt-exposed groups (the rate of embryo malformation was established by determining the ratio of abnormal embryo numbers to the total surviving embryo numbers). However, the incidence of heart malformation was not high (Fig. 1F; dead embryos were excluded from analysis). By viewing the transverse sections of these hearts, we also saw abnormal development of the atrium and ventricle (ventricle is still nearer to rostral part than atrium of chick embryo; Fig. 1C1–C3) or excessive numbers of blood cells in the heart chambers (Fig. 1D1–D3), compared with control heart sections (Fig. 1B1–B3).

Next, we measured cardiomyocyte proliferation and apoptosis with BrdU incorporation and TUNEL assays (Fig. 2). Compared with the control (Fig. 2B), the BrdU-positive cell number was higher in the high-salt-exposed ventricular walls (control=9.82±4.96%, N=11; high osmolarity=21.14±7.31%, N=7; P<0.01; Fig. 2E,G), suggesting that cell proliferation was activated in cardiomyocytes. This outcome is not due to reduced cardiomyocyte numbers in the high-salt-exposed ventricles, because the DAPI staining revealed more cells in the control ventricle walls than in the high-salt-exposed group (Fig. 2A,D). The mRNA expression of cyclin D1, a nuclear protein required for cell cycle progression in G1, was up-regulated nearly two-fold (control=0.56±0.04, high osmolarity=0.99±0.07, N=4; P<0.01; Fig. 2I), which also suggests that high osmolarity induced by high salt promoted cell proliferation in embryonic cardiomyocytes. Interestingly, TUNEL staining showed that cell apoptosis in the ventricular wall was also significantly promoted by high salt exposure (Fig. 2F) compared with the control (Fig. 2C) (control=22.68±3.57%, N=6; high osmolarity=37.85±5.72%, N=8; P<0.01; Fig. 2H).

To further investigate early heart development, transverse sections from control and high-salt-exposed hearts were generated at the level across the atrium and ventricle (Fig. 3A,B). The measurement of the thickness of the ventricular walls (control=55.87±5.18 μm, N=10; high osmolarity=31.94±8.71 μm, N=14; P<0.001) and the trabecular muscles (control=22.62±3.23 μm, N=10; high osmolarity=14.57±2.75 μm, N=14; P<0.001) showed that both were reduced in high-salt-exposed hearts (Fig. 3D) compared with control hearts (Fig. 3C,G,H). Because the MF20 monoclonal antibody recognizes the myosin II heavy chain in cardiac and skeletal muscles, we used it to examine cardiomyocyte differentiation following high-salt treatment. MF20 was expressed in nearly all the ventricular wall and trabecular muscles in the control hearts (Fig. 3E,E1). However, only some of the ventricular wall and trabecular muscles were MF20 positive in the high-salt-exposed hearts (Fig. 3F,F1), suggesting that cardiomyocyte differentiation was affected by high-salt exposure. We next investigated expression of crucial genes during heart tube development, including VMHC, Nkx2.5, GATA4 and BMP2 using PT-PCR. VMHC, Nkx2.5 and GATA4 were down-regulated and BMP2 was not affected by high-salt exposure (VMHC/PPIA: control=0.88±0.08, high osmolarity=0.41±0.11, P<0.05; Nkx2.5/PPIA: control=0.46±0.04, high osmolarity=0.08±0.02, P<0.05; GATA4/PPIA: control=0.55±0.07, high osmolarity=0.20±0.07, P<0.05; BMP2/PPIA: control=0.99±0.38, high osmolarity=1.36±0.67, P>0.05; N=4; Fig. 3I). This experimental evidence indicated that high-salt exposure negatively influenced cardiomyocyte differentiation during heart formation.

We next asked whether angiogenesis was normal in a high-salt environment. High-salt solution was administered to embryos and the chick YSM model in the extraembryonic area opaca was studied, as it is the earliest region of angiogenesis and can be easily manipulated. We examined the development of blood plexuses at the same localization in yolk sac membranes (YSMs) from control (Fig. 4A) and high-salt-exposed (Fig. 4B) embryos. The blood vessel density was lower in the presence of high salt than in the control (control=9.02±1.02%, N=9; high osmolarity=5.38±0.96%, N=20; P<0.001; Fig. 4G). Blood plexus development in the yolk sac membrane occurs not only in two dimensions, but also protrudes into the yolk. Thus, we compared the blood vessels in transverse sections from the same locations of the YSM and found that both the number and diameter of blood vessels in the control YSM sections (Fig. 4C,E) were obviously greater than in the YSM of high-salt-exposed (Fig. 4D,F) embryos (diameter: control=809.91±112.41 μm, N=10; high osmolarity=518.63±128.61 μm, N=20; number: control=6.40±1.51, N=10; high osmolarity=3.47±1.31, N=19; P<0.001; Fig. 4H,I). These data suggested that high-salt exposure inhibited angiogenesis in YSMs.

We next studied whether cell proliferation (PH3) or apoptosis (TUNEL) contributed to the decreased angiogenesis induced by high-salt exposure. The PH3 immunofluorescent staining was performed in serial sections as in Fig. 4 for both control (Fig. 5A) and high-salt-exposed embryos (Fig. 5B). There were fewer PH3-positive cells in high-salt-exposed embryos (Fig. 5D,F) than in control embryos (Fig. 5C,E). The difference in PH3-positive cells was not due to absolute cell number differences in sections, as it was divided by the total cell number stained by DAPI (control=70.01±16.08%, N=10; high osmolarity=51.69±14.37%, N=8; P<0.01; Fig. 5G). Because Cyclin D1 regulates cell cycle progression, we determined the expression of Cyclin D1 mRNA using RT-PCR. Cyclin D1 was down-regulated in the presence of high salt (control=0.98±0.12, high osmolarity=0.26±0.10, N=4; P<0.05; Fig. 5H), further indicating that cell proliferation was repressed by exposure to high salt. However, as observed in heart formation, cell apoptosis as labeled by TUNEL staining was enhanced in the presence of high salt (control=23.29±16.08, high osmolarity=53.99±14.37, N=8, respectively; P<0.05; Fig. 5I–K).

The blood vessels in the yolk sac membrane are directly derived from blood islands in the extraembryonic area opaca at an early developmental stage. We therefore determined whether the inhibitive effect of high-salt exposure occurred at the earlier stage of blood island formation using early chick embryos. To avoid deviations in individual embryonic development, we exposed one side of the embryo to high salt, while the other side acted as a control (Fig. 6A). VE-cadherin in situ hybridization was performed in these treated chick embryos (Fig. 6B) because VE-cadherin is specifically expressed in blood islands in early stage embryos. Blood islands were clearly labeled with VE-cadherin in the area opaca, so we could observe and compare the blood island density (Fig. 6C,D) and numbers in transverse sections (Fig. 6E,F). The blood island density in high-salt-exposed embryos was reduced compared with control embryos (control=43.18±4.74%, high osmolarity=31.59±3.32%, N=6 embryos, respectively, P<0.01; Fig. 6G). The number of blood islands was also reduced in the presence of high salt (control=7.00±1.63, high osmolarity=4.43±1.62, N=7 sections from three embryos, respectively, P<0.05; Fig. 6H). These data suggest that blood island formation was also inhibited by the exposure of high salt and that the inhibitive effect on angiogenesis by exposure to high salt occurs at an early stage.

The expression of well-known angiogenesis-related genes following high-salt treatment also supports these experimental observations. We measured the expression of HIF2, FGF2 and VEGFR mRNA using RT-PCR in YSM samples. HIF2 and FGF2, but not VEGFR, were down-regulated in the presence of high salt. A parallel test for PPIA confirmed that equal amounts of RNA were used in the RT reaction (HIF2/PPIA: control=0.89±0.10, high osmolarity=0.48±0.08, P<0.05; FGF2/PPIA: control=0.54±0.12, high osmolarity=0.31±0.11, P<0.05; VEGFR/PPIA: control=0.42±0.38, high osmolarity=0.41±0.35, P>0.05; N=4, respectively; Fig. 6I).

To confirm our observation of the high-salt effect on angiogenesis, we studied the CAM model, another well-known angiogenesis model. The embryos were exposed to a high-salt environment through the injection of either simple saline (500 µl of 0.7% salt solution) or high-salt solution (500 µl of 16.85% salt solution) from day 6 to day 8 (Fig. 7A). On day 8, we photographed and analyzed the CAM angiogenesis in the control (Fig. 7B) and high-salt-exposed (Fig. 7C) embryos. Blood vessel density was measured by appearance and analysis using specific software (control=10.01±0.60%, high osmolarity=7.23±0.58%, N=6 embryos, respectively, P<0.01; Fig. 7H). Moreover, the high-magnification images and the transverse sections from control and high-osmolarity groups showed that blood vessel density was reduced in the presence of high salt compared with the control (Fig. 7D–G). Altogether, the data from both angiogenesis models revealed that a high-salt environment significantly suppressed angiogenesis at different developmental stages.

In our previous studies, excess reactive oxygen species (ROS) generation played an important role in angiogenesis defects (Jin et al., 2013; Li et al., 2014; Wang et al., 2015b). Therefore, we next studied the effect of excess ROS generation. Our results showed that ROS production was significantly increased in HUVECs after high-salt treatment (control=32.24±20.30%, high osmolarity=71.15±8.41%; N=6 in control group and N=6 in high-salt group, P<0.01; Fig. 8A). Because they are crucial genes encoding antioxidant enzymes, GPx, SOD1 and SOD2 expression could indicate the ROS generation activity. Using RT-PCR assay, we detected the mRNA expression of GPx, SOD1 and SOD2 in the developing heart (Gpx/PPIA: control=0.38±0.11, high osmolarity=1.48±0.32; SOD1/PPIA: control=0.79±0.06, high osmolarity=1.68±0.14; SOD2/PPIA: control=0.80±0.09, high osmolarity=1.46±0.16; Fig. 8B) and yolk sac membrane (Gpx/PPIA: control=0.28±0.10, high osmolarity=0.74±0.27; SOD1/PPIA: control=0.69±0.11, high osmolarity=0.98±0.20; SOD2/PPIA: control=0.46±0.11, high osmolarity=0.85±0.23; Fig. 8C) tissues following the high-salt treatment.

Excess ROS generation in tissues and cells is cytotoxic. However, it is not necessarily an essential component of the high-salt-induced cardiovascular development defects. To study this correlation, vitamin C, an antioxidant compound, was added together with high salt and development was observed. The embryonic mortality was reduced to some extent when the vitamin C was added (high osmolarity=70%, high osmolarity+Vc=50%, N=50 embryos in these two groups; Fig. 8D). The morphological observation of angiogenesis in the YSM also showed that addition of vitamin C (Fig. 8E) significantly rescued the reduction in blood vessel density in YSMs (control=9.21±0.89%, N=9; high osmolarity=5.07±1.34%, N=10; high osmolarity+Vc=7.59±2.02, N=8; P<0.05 between control and high osmolarity group; P<0.05 between high osmolarity+Vc and high osmolarity group; Fig. 8F–H).

We next determined the expression of VMHC, Nkx2.5 and GATA4 in heart tissue after the addition of vitamin C and found that the reduced expression after high-salt exposure was significantly rescued (VMHC/PPIA: control=1.22±0.11, high osmolarity=0.68±0.07, high osmolarity+Vc=0.90±0.04; Nkx2.5/PPIA: control=1.17±0.52, high osmolarity=0.29±0.20, high osmolarity+Vc=0.89±0.25; GATA4/PPIA: control=0.92±0.09, high osmolarity=0.34±0.19, high osmolarity+Vc=0.62±0.12; P<0.05; N=4, respectively; Fig. 8I). To assess angiogenesis-related gene expression, we measured HIF2 and FGF2 expression in YSM tissues after the addition of vitamin C using RT-PCR and found that the reduction in HIF2 and FGF2 expression induced by exposure to high salt was dramatically reversed in the rescued group (HIF2/PPIA: control=0.71±0.20, high osmolarity=0.43±0.05, high osmolarity+Vc=0.63±0.09; FGF2/PPIA: control=0.18±0.01, high osmolarity=0.09±0.03, high osmolarity+Vc=0.18±0.02; P<0.05; N=4, respectively; Fig. 8J). These experimental data suggest that excess ROS generation is involved in the high-salt-induced defects in cardiovascular development.

In our study, We firstly observed the heart phenotypes in chick embryos after 4.5 days of high-salt exposure (Fig. 1C,D). Abnormal heart tube looping (Fig. 1C) and blood congestion within the heart cavity were detected in surviving embryos (Fig. 1D). In their corresponding transverse sections, the normal overturn of the developing atrium and ventricle was not yet completed compared with controls (Fig. 1B1–C3), and red blood cells filled the heart chamber (Fig. 1D1–3). Meanwhile, cardiomyocyte cell proliferation and apoptosis were also disturbed by high-salt exposure (Fig. 2). The cardiac transcription factors Nkx2.5 and GATA4 play important roles in vertebrate cardiac development. Nkx2.5 and GATA4 were initially identified as the genes causing congenital heart disease (Granados-Riveron et al., 2012; Wang et al., 2007). Nkx2.5 and GATA4 expression was dramatically reduced by high-salt exposure (Fig. 3I), indicating that high levels of salt interfered with the expression of these crucial heart formation genes. It is probably because of this inhibition that the MF20 was not completely expressed, indicating suppression of cardiomyocyte differentiation (Fig. 3E,F).

The accomplishment of cardiovascular functions relies on an integrated heart–blood vessel system. The pathological cardiovascular development phenotypes may also result from defective vascular development. Therefore, we investigated angiogenesis in the presence of high salt using chick YSM and CAM models, which have been extensively employed for studying angiogenesis (He et al., 2014; Mitashev et al., 2001). Vessel formation in the YSM includes vasculogenesis at early stages and angiogenesis at later stages, including the specification of the migrating mesoderm-derived hemangioblasts, the primary and secondary sprouting of intersomitic vessels and arterial versus venous modeling (Baldessari and Mione, 2008).

In this study, we revealed that the number, density and diameter of blood vessels were suppressed by exposure to high salt in both the YSM and CAM models (Figs 4 and 7). Because blood island formation occurs in the area opaca where the initial angiogenesis also occurs, we determined whether high-salt exposure affected blood island formation and vasculogenesis (Fig. 6). High-salt exposure suppressed blood island formation, suggesting that the high-salt-induced suppression of angiogenesis occurred at different stages of angiogenesis. Angiogenesis is a strictly regulated process, balancing stimulatory and inhibitory factors to control the correct development of blood vessels. Well-characterized angiogenic and angiostatic factors such as HIF-1, AP-1 and Sp-1 modulate angiogenesis by regulating VEGF expression (Josko and Mazurek, 2004). We measured cell proliferation and apoptosis using PH3 and TUNEL staining in the transverse sections of YSM blood vessels and showed that cell proliferation was suppressed and apoptosis was enhanced by high-salt exposure (Fig. 5A–J). The consequent reduction in cyclin D1 expression also supported the inactivation of the cell cycle by exposure to high salt (Fig. 5H). Fibroblast growth factor-2 (FGF2) is in the FGF family. FGF2 exerts its angiogenic activity by interacting with endothelial cell surface receptors including tyrosine kinase receptors, heparan sulfate proteoglycans and integrins (Presta et al., 2009). In this study, RT-PCR data showed that FGF2 expression was down-regulated by high-salt exposure (Fig. 6I). These data suggest that high-salt-suppressed angiogenesis might be due to the negative regulation of HIF2 and FGF2, but not VEGFR expression. What we need to keep in mind is that VEGF itself is not deemed to be efficient, although it does plays a crucial role in vascularization. Therefore, other growth factors/cytokines, including angiopoietins, SDF-1, PDGFb and FGF2 etc. are also involved in sprouting angiogenesis (Cao et al., 2003; Thurston et al., 1999). Further experiments are still required to address this question.

As a natural byproduct of aerobic metabolism, ROS play an important role in cell signaling and homeostasis. However, excess ROS might result in significant damage to cell structures (Devasagayam et al., 2004). Whether ROS will cause damage or act as a signaling molecule depends on the delicate balance between ROS production and elimination. It was reported that embryonic dysmorphogenesis is blocked by adding an antioxidant in vivo and in vitro, suggesting that excess ROS in embryonic tissues activate the teratogenic process of diabetic pregnancy (Forsberg et al., 1998). Similar results on excess ROS generation were also observed in our study (Fig. 8A). Moreover, the RT-PCR data indicated that GPx, SOD1 and SOD2 expression dramatically increased in both the heart tube and yolk sac membrane following high-salt treatment (Fig. 8B,C), suggesting that excess ROS were generated there. Meanwhile, the high embryonic mortality was partially rescued by the addition of vitamin C, an antioxidant (Fig. 8D). The partial rescue effect was also observed in blood vessel developmental morphology (Fig. 8E–G). To further investigate the effect of antioxidants in oxidation stress, we detected the expression of key genes in heart tube formation and angiogenesis using an RT-PCT assay and showed that VMHC, Nkx2.5 and GATA4 expression in the antioxidant group rebounded from the level of high-salt-suppressed gene expression in heart tubes (Fig. 8I). Similarly, HIF2 and FGF2 expression in the YSM of the antioxidant group was also rescued to some extent (Fig. 8J).

Altogether, our data suggest that the embryonic cardiovascular dysplasia due to high osmolarity exposure from high salt might result from defects in heart formation and angiogenesis, respectively. Moreover, the excess ROS generation due to high-salt exposure could interfere with expression of genes involved in heart development (Nkx2.5 and GATA4) and angiogenesis (HIF2 and FGF2). Consequently, the abnormal looping of heart tube and blood congestion in developing heart emerges, as illustrated in Fig. 9. Further experimental studies on signaling pathways are required to explore the precise molecular biological mechanisms.

We would like to thank Dr Zhi Huang for providing the HUVECs.