Specific Local Cardiovascular Changes of N^ɛ-(Carboxymethyl)lysine, Vascular Endothelial Growth Factor, and Smad2 in the Developing Embryos Coincide With Maternal Diabetes–Induced Congenital Heart Defects

The Uppsala Regional Ethical Committee on Animal Experiments approved the research protocol including all experimental procedures involving animals. We used 3-month-old rats from a local colony of outbred U (Uppsala) Sprague-Dawley rats, previously noted for a high rate of congenital malformations in response to maternal diabetes (17,18) as well as the L and B inbred rat strains derived from the U and H rats (a Sprague-Dawley rat strain more resistant to diabetes-induced congenital malformations), respectively. All rats were fed a commercial pellet diet (AB Analycen, Lidköping, Sweden) and had free access to food and tap water. They were maintained at an ambient temperature of 22°C with a 12-h light-dark cycle. Introduction of diabetes in female rats was performed by injecting 40 mg/kg streptozotocin (Sigma-Aldrich, Uppsala, Sweden) into the tail vein 1 week before mating commences, i.e., 1–3 weeks before conception. A state of manifest diabetes (MD) was confirmed 1 week after the injection by the presence of a blood glucose level exceeding 20 mmol/l (Medisense Glucose Sensor; Abbot Scandinavia AB, Solna, Sweden). The induced level of hyperglycemia was similar between rats of the susceptible and resistant strains. Diabetic and nondiabetic (ND) females were caged with males overnight, and conception was verified by the presence of sperm in a vaginal smear. The day of a positive vaginal smear was designated gestational day 0. Pregnancy was interrupted on gestational day 13, 14, and 16. Embryos and fetuses were fixed in 4% PBS paraformaldehyde, subsequently dehydrated in graded ethanol, transferred to xylene, and finally embedded in paraffin. Cardiac morphology was analyzed in E16 embryos from diabetic (MD) L rats (n = 19) and E16 embryos from B-MD dams (n = 16). Data were compared with the U strain embryos, which we have studied before on both E16 and E18 (18). A total of 24 E13 embryos from MD and ND dams were used for the analysis of CML and MG adduct formation and detection of phosphorylated Smad2, being L-ND (n = 4), L-MD (n = 4), B-ND (n = 4), B-MD (n = 4), U-ND (n = 4), and U-MD (n = 4). At E14, we studied 13 embryos of L-ND (n = 2), L-MD (n = 2), U-ND (n = 6), and U-MD (n = 3).

Embryos were serially sectioned at 5 μm. Alternated sections were rehydrated and antigen retrieval was performed using the microwave to enhance staining. Sections were incubated for 1 h at room temperature with a 1:500 dilution of the anti-CML antibody and 1:5 dilution of the antibody directed against MG adducts (28,29). Subsequently, sections were washed in PBS and incubated for 60 min with a 1:200 diluted biotinylated horse anti-mouse antibody (ABC Kit; Vector Laboratories, Burlingame, CA) and incubated with ABC according to the recommendations of the manufacturer.

Immunohistochemistry was performed after the protocol described above, although no microwave enhancement was performed, and the incubation of the sections, with a 1:1,000 dilution of the phosphorylated Smad2 antibody (provided by Prof. Dr. P ten Dijke, Leiden University Medical Centre, Leiden, the Netherlands), was preceded by a blocking step using 10% normal goat serum in PBS.

Consecutive transverse sections of 5 μm were rehydrated and post-fixed in 4% paraformaldehyde for 20 min. Subsequently, they were treated with 20 μg/ml Proteinase K (Promega, Leiden, the Netherlands) and 4% paraformaldehyde and by 0.1 mol/l triethanolamine-HCl (pH 8.0) + 0.25% acetic anhydride. After dehydration, sections were hybridized overnight at 55°C using sense and anti-sense ³⁵S-radiolabeled VEGF-A RNA probes that were transcribed using a 451-bp EcoRI/BamHI cDNA clone encoding for the mouse VEGF-120 isoform (pVEGF2; provided by Dr. G. Breier, University of Technology, Dresden, Germany). The following morning, the sections were washed and then coated with Ilford G5 emulsion (Ilfort, Mobberly, U.K.), in which they were kept at 4°C for 10 days. Finally, the sections were developed in Kodak D19 (Kodak, Chalon s. Saonne, France) solution for 4 min, washed, and fixated in Ilford Hypam (Ilfort), counterstained with hematoxylin, and mounted with Pertex (Histolab Products, Goteborg, Sweden). The sections were examined by light microscopy using dark field optics.

The Amira V3.1 software package (Template Graphics Software, San Diego, CA) was used to make three-dimensional reconstructions of day 13 embryos. For the reconstruction, we used 5-μm sections every 25 μm. Via manual drawing, the cardiovascular lumen and CML⁺ areas were color-labeled and rendered for three-dimensional visualization. Embryos were processed under comparable conditions described above, allowing for a direct comparison of the reconstructions.

To identify the potential of AGEs in the development of diabetes-induced CHD, we analyzed the presence and localization of CML and MG adducts in E13 and E14 embryos from the malformation-prone L and U rat strains and the resistant B rats with (MD) and without (ND) diabetes.

At E13, CML was identified in all the MD offspring of the malformation-susceptible rats L-MD and U-MD (Fig. 1). The CML⁺ area was identified in the PAAs (Fig. 1,B and D). More precisely, it overlapped with segments derived from the fourth and sixth PAA and extended just proximal to the connection to the subclavian arteries. This we have shown in the three-dimensional reconstruction of the developing cardiovascular system of an E13 L-MD embryo (Fig. 2,B–D) and its schematic representation indicating the various PAA segments (Fig. 2,A). Formation of CML also occurred in the vessel wall of the ascending aorta and pulmonary trunk, where staining was more intense in the aortic wall than in the pulmonary trunk (compare Fig. 2,C to Fig. 2,D). Furthermore, CML⁺ cells were identified in the outflow tract myocardium bordering the cushions of the outflow tract (Fig. 2,B, blue). CML accumulated in the cytoplasm rather than in the extracellular matrix. No CML could be detected in any of the E13 offspring of the L-ND or in MD or ND embryos derived from the malformation-resistant B strain (Fig. 3). In embryos of U-ND rats, we observed a weak CML staining (Fig. 1). Using immunohistochemical analysis, no MG adducts could be detected in any of the embryos at this stage.

At E14, CML was detected in all embryos in the proximal outflow tract myocardium, in the ascending aorta, all PAAs, the subclavian arteries, the dorsal aorta, the pulmonary trunk, and the pulmonary arteries. CML was also present in the cushions of the outflow tract, where it was increased in the MD offspring of the U and L rats (Fig. 4,B, D, and F), compared with that in ND embryos or MD embryos of the resistant B rats (Fig. 4 A, C, and E). Again no MG adducts could be detected using immunohistochemical analysis.

As hyperglycemia influences VEGF expression and alterations in the amount of VEGF can disturb cardiovascular development, we performed VEGF in situ hybridizations in embryos from ND and MD dams. At E13, a high expression of VEGF was seen in all myocardial cells aligning the endocardial cushions of both the inflow tract and the outflow tract of the heart, the myocardial cells in the ventricular septum, and the fibrous tissue protruding into the base of the atrial septum, the so-called spina vestibuli. At this stage of development, VEGF expression was similar in both ND and MD embryos of L, U, or B rats. E14 ND embryos and the MD embryos of the resistant B strain showed a VEGF mRNA expression pattern similar to that seen on E13. However, a diabetes-induced decrease in VEGF expression was observed in five of six MD embryos derived from the malformation-prone L and U strain. This decrease in VEGF could be observed in all areas that normally have a high VEGF expression level (Fig. 5).

Because both glycation products and VEGF influence the Smad2 signaling, we analyzed active Smad2 using an antibody against phosphorylated Smad2. At E13, phosphorylated Smad2 was increased in the CML⁺ positive vessel wall of the ascending aorta and pulmonary trunk as well as in the myocardium of the outflow tract (Fig. 6,D) in L and U MD embryos, compared with either the surrounding CML⁻ cells or to the same area from MD-B embryos (Fig. 6 B) or ND offspring. In the PAAs of all embryos, the level of phosphorylated Smad2 was relatively high compared with other arteries without a clear distinction for CML⁺ or CML⁻ areas.

At E14, phosphorylation of Smad2 overlapped with the increased CML patterning, which in this stage of development was present in the cushions of the outflow tract of the heart (Fig. 6,F–H). In addition, phosphorylated Smad2 is higher in the cushions of the E14 MD offspring (Fig. 6,H) than in the ND offspring (Fig. 6 F).

The L and B strains are inbred rat strains of which the L strain is known to present with extra-cardiac malformations such as micrognatia, cleft palet, or skeletal malformations, whereas the B strain is more resistant to these malformations. Cardiovascular anomalies have not been described in these strains. It was not known if the diabetes-induced facial anomalies were linked to cardiovascular malformations. We have therefore studied cardiac morphology of E16 L-MD and B-MD embryos and compared the identified malformations with those we have previously characterized in the U strain embryos of E16 and E18 (18). In the inbred L rats, all MD fetuses had a CHD compared with 67% in the outbred U-MD offspring. Outflow tract anomalies were identified in 74% of the L-MD offspring compared with 63% in the U-MD offspring. Defects of the fourth PAA were seen in 47% of the L-MD and 22% of the U-MD embryos and sixth PAA defects in 21% of the L-MD and 11% of the U-MD embryos (Table 1). In the B-MD offspring resistant to the maternal diabetes–induced extra-cardiac malformations, the number of cardiovascular malformations is also considerably lower (Table 1). From these data, we can conclude that the L rat strain is susceptible to diabetes-induced CHD comparable to the U rat strain.

At E13, CML accumulation appeared in a small area of the developing outflow tract and in the PAAs in MD embryos of the susceptible L and U strains. At this developmental stage, the PAAs are still paired and their remodeling has not yet started (18). The addition of myocardium from the secondary heart field to the primary heart tube is almost completed, and outflow tract septation is in process (30). The cells positive for CML on E13 are known to originate from two cell populations. The most proximal CML⁺ cells are secondary heart field–derived cells that have migrated from the splanchnic mesoderm (31,32). The other CML⁺ cells in the ascending aorta and PAA are derived from the cardiac neural crest cells (32,,–35). Given the temporal spatial overlap between CML accumulation and the development of cardiovascular outflow and PAA anomalies, it can be postulated that both the addition of myocardium from the secondary heart field and the function of neural crest cells are hampered in diabetic pregnancies (17,18,36,–38).

The accumulation of CML and the increase in phosphorylated Smad2 at E13 is asymmetric, being higher on the aortic side than on the side of the pulmonary trunk. This is exactly opposite to the lacZ expression identified in the y96-Myf5-nLacZ-16 transgenic mice, which is predominantly seen on the side of the pulmonary trunk (39). From this, Bajolle et al. (39) hypothesize that deviations from the left/right asymmetry inhibits outflow tract rotation and result in anomalies such as a double outlet right ventricle. Analogous and asymmetric increase in CML accumulation might disturb the left/right signaling in MD embryos and subsequently the rotation of the outflow tract, also resulting in CHD, such as a double outlet right ventricle that is frequently seen in our MD offspring (18). Furthermore, embryos that are heterozygous for both a Smad2 and nodal mutation (40) show defects in left-right patterning, resulting in abnormal cardiac looping. This indicates that also the asymmetric increase in phosphorylated Smad2 (co-localizing with the asymmetric CML increase) in the E13 MD embryos is likely to play a role in diabetes-induced congenital heart malformations.

At E14, septation is almost completed and the process of PAA remodeling is still in progress. Although CML could be detected in the embryonic vasculature in both MD and ND offspring of all strains, increased CML was found in the cushions of the outflow tract in the MD offspring of the susceptible U and L dams. Abnormal cushions as seen on E14 could lead to congenital heart malformations such as Tetralogy of Fallot or double outlet right ventricle. We have previously identified these malformations in combination with hypoplastic cushions in the MD offspring of these rats (18), in which we now show an increase in CML.

CML is an AGE that can be formed on lysine residues in proteins by a glycoxidation reaction (41). In the embryos of our study, we failed to detect MG adducts, which are produced from a glycation reaction. This raises the question whether the CML accumulation indicates an increase in the formation of AGEs or an increase in oxidative stress. MG adducts–mediated formation of nonenzymatic glycation did not take place in sufficient amounts. It cannot be ruled out that other AGEs, which are not detected by the antibodies used in this study, may play a role in the formation of CHD. However, we favor the explanation that, in our model, the increase in CML can be used as a biomarker for oxidative damage, as has been suggested before (41,42). We base this opinion on the fact that antioxidant supplementation reduces the number and severity of the diabetes-induced CHD in the offspring of these rats (17). In addition, increasing the oxidative stress by injection of antimycin A in pregnant mice results in similar CHD, as seen in the offspring of diabetic mice (43), an effect that could be diminished by antioxidant supplementation. Furthermore, CHD was reduced by the addition of antioxidant in chicken embryos in which neural crest cells were exposed to elevated glucose (37). Next to these experimental models, it has been shown that women who gave birth to a child with a CHD have higher levels of biomarkers for oxidative stress (44). On the basis of these data, we are convinced that a local glucose-induced increase in oxidative stress, as pointed out by CML, is an important mechanism in the development of congenital malformations in the offspring from diabetic pregnancies.

Elevated glucose levels can disturb the cardiovascular development by influencing the expression of VEGF (45). The addition of high glucose concentrations to cultured mice embryos inhibits endothelial to mesenchymal transformation in the endocardial cushions by a reduction of VEGF levels (45). In the MD embryos, a clear overlap between CML accumulation, reduced VEGF expression, and hypoplastic endocardial outflow tract cushions was identified. We assume that a decrease in VEGF causes cushion hypoplasia and the related maldevelopment of the outflow tract (double outlet right ventricle and Tetralogy of Fallot). In addition to our results, decreased levels of VEGF were identified in cultured embryos and in vitro endocardial cushions after exposure to elevated glucose (23,45). This negative relation between diabetes and VEGF expression in the embryo that we and others found is opposite to data for adult endothelial cells (46,47).

At E14 in the endocardial cushions of the outflow tract of the heart, CML and phosphorylated Smad2 were increased, whereas VEGF expression was decreased. AGEs can induce Smad2 phosphorylation in vitro and in vivo (26), and therefore CML might be responsible for the raise in phosphorylated Smad2. However, VEGF has been reported to inhibit Smad2 phosphorylation (48); therefore, a reduced level of VEGF in these structures could also relate to the identified increase in Smad2 phosphorylation. Although endothelial to mesenchymal transformation can be increased by TGFβ/Smad2 signaling (49,50), the hypoplastic appearance of the outflow cushions contradicts this assumption. We assume that the lack of myocardial VEGF expression is more detrimental for cushion development than the stimulating effect one might expect from Smad2. Unfortunately, the pathways for activation and action of Smad2 are complex, involving interactions with multiple players, and at this time, it is not possible to fully elucidate them.

The incidence of full-blown CHD in fetuses (>E16) between the L and U strain clearly reveals their genetic predisposition for diabetes-induced congenital cardiovascular malformations. The high percentage of CHD in the MD offspring of the inbred L strain, for which rats were selected on the basis of extra-cardiac malformations, implies that there is a link between cardiovascular and noncardiac congenital malformations. The similarity between the number and type of malformations identified in the L-MD and U-MD fetuses justifies the use of the L strain next to the U strain for the identification of the pathways involved in diabetes-induced CHD.

In this research, we have provided a novel proof for a specific temporal-spatial pattern rather than a general increase in CML accumulation, suggesting a raise in oxidative stress. The increase in CML and phosphorylated Smad2 strikingly overlapped with those cardiovascular regions that are at high risk during development. Furthermore, the hyperglycemia-related decrease in VEGF was observed in the same regions. We postulate that local rather than general changes in CML accumulation, Smad2 phosphorylation, and VEGF expression are detrimental for the developing cardiovascular system in MD embryos.

This work was funded by the Netherlands Heart Foundation grants NHS2002B035 and NHS2006B074, the Ernfors Family Fund, the Swedish Diabetes Association, and the Swedish Research Council Grant 12X-7475.

This work was also funded by the Novo Nordisk Foundation. No other potential conflicts of interest relevant to this article were reported.

We would like to thank P ten Dijke for the gift of the antibody raised against phosphorylated Smad2. We also thank Jan Lens for graphics and layout.

Parts of this study were presented at the 4th International Symposium on Diabetes and Pregnancy, 29–31 March 2007, Istanbul, and at the 12th Annual Weinstein Cardiovascular Development Conference, 19–22 May 2005, Tucson, Arizona.