Maternal voluntary exercise mitigates oxidative stress and incidence of congenital heart defects in pre‐gestational diabetes

Abstract Women with pre‐gestational diabetes have a higher risk of producing children with congenital heart defects (CHDs), caused predominantly by hyperglycemia‐induced oxidative stress. In this study, we evaluated if exercise during pregnancy could mitigate oxidative stress and reduce the incidence of CHDs in the offspring of diabetic mice. Female mice were treated with streptozotocin to induce pre‐gestational diabetes, then mated with healthy males to produce offspring. They were also given access to running wheels 1 week before mating and allowed to exercise voluntarily until E18.5. Heart morphology, gene expression, and oxidative stress were assessed in foetal hearts. Maternal voluntary exercise results in a significantly lower incidence of CHDs from 59.5% to 25%. Additionally, diabetes‐induced defects in coronary artery and capillary morphogenesis were also lower with exercise. Myocardial cell proliferation and epithelial‐mesenchymal transition at E12.5 was significantly lower with pre‐gestational diabetes which was mitigated with maternal exercise. Cardiac gene expression of Notch1, Snail1, Gata4 and Cyclin D1 was significantly higher in the embryos of diabetic mice that exercised compared to the non‐exercised group. Furthermore, maternal exercise produced lower reactive oxygen species (ROS) and oxidative stress in the foetal heart. In conclusion, maternal exercise mitigates ROS and oxidative damage in the foetal heart, and results in a lower incidence of CHDs in the offspring of pre‐gestational diabetes. Exercise may be an effective intervention to compliment clinical management and further minimize CHD risk in mothers with diabetes.


| INTRODUC TI ON
Congenital heart defects (CHDs) are structural abnormalities of the heart and coronary arteries that can arise when developmental pathways are disturbed. 1 An estimated 1%-5% of live births are affected by CHDs; however, this is believed to be a conservative estimate considering many defects can be asymptomatic at birth and remain undetected until adulthood when severe complications can arise. 2,3 Studies on CHD aetiology suggest that approximately 15% of cases are attributable to genetic causes, and several environmental factors have been identified. 4 Of these, maternal pre-gestational diabetes, either type 1 or type 2 diabetes mellitus diagnosed prior to conception, is a major risk factor with a 4-5 times increased risk of producing children with CHDs. [5][6][7] The prevalence of CHDs in both children and adults has increased by 11% and 57%, respectively, from 2000 to 2010. 8 Some of this increase may be attributable to the rapidly growing prevalence of diabetes in the general population. 9 This is concerning because CHDs have several associated co-morbidities including pulmonary hypertension, coronary artery disease, arrhythmia and sudden cardiac death. The total costs of treatment and management is a rising burden on the healthcare system. 3,10 For these reasons, it is imperative to expand our understanding of the pathogenesis of CHDs and discover viable strategies for prevention.
Maternal physiological adaptations induced by exercise may protect against adverse cardiovascular development in the offspring.
Remarkably, a recent study showed that voluntary exercise during pregnancy reduced the maternal age-related risk of CHDs in offspring with cardiac transcription factor Nkx2.5 haploinsufficiency. 11 It is currently unknown how maternal exercise modulates cardiovascular development. This study investigated the effects of maternal voluntary exercise on CHDs induced by pre-gestational diabetes in mice, and determined possible mechanisms of protection. High-glucose environments during development disrupt intracellular redox homoeostasis by stimulating reactive oxygen species (ROS) production to exceed levels that can be managed by endogenous antioxidants. 12,13 Oxidative stress can trigger changes in embryonic gene expression, cause oxidative damage to proteins and disrupt coordinated developmental processes. 14 Our recent studies showed that inhibition of oxidative stress using an antioxidant, N-acetylcysteine, resulted in significantly lower incidences of CHDs and coronary artery malformation in the offspring of mice with pre-gestational diabetes, suggesting an important role of oxidative stress in the pathogenesis of CHDs and coronary artery malformation. 13,15 Notably, exercise can stimulate the expression and activity of endogenous antioxidants, superoxide dismutase (SOD) and catalase, in aortic and cardiac tissues under diabetes-induced oxidative stress. 16 However, there is limited knowledge on whether maternal exercise can affect foetal oxidative stress in utero. Furthermore, it is not known if maternal exercise protects against CHDs induced by pregestational diabetes. We hypothesized that exercise would mitigate oxidative stress and produces lower incidences of CHDs and coronary artery malformation in the offspring of diabetic mothers. We tested this hypothesis using streptozotocin (STZ)-induced pre-gestational diabetes and voluntary wheel running in mice.

| Animals
This study was performed in accordance with the guidelines of the Canadian Council on Animal Care and approved by the Animal Care Committee at Western University, Canada (Protocol #2016-099). All efforts were made to minimize animal suffering and minimize the number of animals used.
Female C57BL/6 mice were purchased from Jackson Laboratory (Bar Harbor, Maine). Animals had access to standard chow and water ad libitum, and were housed in 12 hours light/dark cycle. At 8 weeks old, mice were given intraperitoneal injections of STZ (Sigma) at a dose of 50 mg/kg body weight/day for five consecutive days. 17 Control mice received saline injections. One week post-injection, non-fasting blood glucose was measured using the One Touch Ultra2 glucose meter (LifeScan) to determine diabetic status. Mice with blood glucose levels exceeding 11 mmol/L were considered diabetic. Diabetic and control female mice were then placed individually in a cage with or without a freespinning running wheel (Mouse running wheel, 5.1 × 10.2 cm, Columbus Instruments, Columbus, OH) for one week to allow acclimatization.
The females were then bred with healthy 10-12 weeks old WT males overnight in cages with no running wheel, and returned to their original cage in the morning. The presence of a vaginal plug indicated embryonic day (E) 0.5. The number of wheel turns per day was recorded during the entire pregnancy using the Multi-Device Channel Interface software (Columbus Instruments, Columbus, OH). The distance run per day was calculated by multiplying the number of running wheel rotations in 24 hours by the circumference of the wheel. Non-fasting blood glucose measurements were taken at the same time every morning to monitor glycemic levels throughout pregnancy from E0.5 to E18.5. There were four groups of offspring: control, exercise control, as well as offspring of maternal diabetes (OMD) with and without maternal exercise.

| Histological analysis of CHD
Foetal samples were collected on E18.5, then fixed in 4% paraformaldehyde for 18 hours at 4°C. Samples were then dehydrated in ethanol and embedded in paraffin to be sectioned into 5 µm thick serial sections using a Leica RM2255 microtome. Transverse sections began at the level of the thymus (above the aortic arch) and continued until the apex of the heart. To diagnose CHDs, sections were stained in hematoxylin and eosin (H/E), and examined under a light microscope (Observer D1, Zeiss). 13,17

| Immunohistochemistry
Foetal heart sections were subjected to antigen retrieval in citric acid buffer (0.01 M, pH 6.0) for 12 minutes at 94°C in a microwave

| AMIRA 3D reconstruction
AMIRA® software (Template Graphics Software) was used to reconstruct E18.5 sections of the heart into a three-dimensional (3D) model. Sections stained with α-smooth muscle actin were used so that coronary vasculature could be visualized and reconstructed.
Images were taken at each 25-µm section of the heart. The compact myocardium and coronary arteries were labelled manually in each section. AMIRA calculated the volume of the labelled components and these values were used to calculate the ratio of coronary artery volume to myocardial volume. 15

| Western blot analysis
Snap frozen E14.5 ventricles were used to measure levels of phosphorylated eNOS (p-eNOS) protein by Western blotting. Protein was isolated in PhosphoSafe buffer (Fisher Scientific) and lysed by sonication. Forty-five micrograms of total protein per well were separated by 8% SDS-PAGE, then transferred onto PVDF membrane.
The membrane was blocked for 1 hour in TBST with 5% skim milk, then bands were tagged using anti-phospho-NOS3 primary antibody (1:1000; Santa Cruz). Additionally, membranes were also blotted for total eNOS using anti-NOS3 primary antibody (1:1000; Santa Cruz), and anti-α-actinin antibody (1:1000; Sigma) which served as the loading control. Horse radish peroxidase-conjugated secondary antibodies (1:2500; Bio-Rad) were used to detect protein bands using the enhanced chemiluminescence method. Signal intensity was

| Statistical analysis
Data are presented as means ± SEM. Statistical analysis was performed using two-way ANOVA followed by Tukey's multiple comparisons test (GraphPad Prism, version 6.0). A chi-squared test was used to analyse the incidence of CHD. A P < 0.05 was considered statistically significant.

| Exercise and blood glucose levels monitored during gestation
Prior to STZ injection, there was no difference in blood glucose levels between the four groups ( Figure 1A). For both exercised and non-exercised diabetic mice, non-fasting blood glucose levels were significantly higher at E0.5 compared to controls (P < 0.05). As pregnancy progressed towards E18.5, blood glucose levels gradually increased for the diabetic groups. There was no significant difference in the hyperglycemic state of mice that exercised compared to mice that did not exercise at any time point in gestation. Diabetic mice ran approximately 2-5 km per day in the first 9 days of gestation, after which the levels of exercise gradually declined as the dams grew larger ( Figure 1B). Mean litter size was significantly smaller for diabetic dams with no exercise, compared to control ( Figure 1C, P < 0.05). Furthermore, litters from diabetic dams without exercise had a trend of higher numbers of absorbed offspring ( Figure 1D). There was no significant difference in maternal body weight at E0.5 among the four groups ( Figure 1E).

| Incidence of CHDs was lower in offspring of exercised mice with pre-gestational diabetes
The incidence of CHDs in E18.5 OMD without exercise was 59.5% ( Table 2). Male OMD had a significantly higher incidence of CHDs (72.2%) than females (47.4%), and maternal exercise resulted in lower incidences of CHDs to 29.4% and 13.3% for male and female OMD, respectively (P < 0.05, Figure 1F). The commonly diagnosed defects were ASD (43.2%), VSD (10.8%), atrioventricular septal defect (AVSD), double outlet right ventricle (DORV) and stenosis of pulmonary and aortic valves ( Table 2, Figure 1G-M). Overall, the offspring of diabetic mice that exercised had a significantly lower incidence of CHDs (25%). The incidence of ASD was also lower (15.6%), while VSD, DORV or AVSD was not present in the exercised group (Table 2). In the OMD group, one foetus had a hypoplastic left heart syndrome (HLHS) and a hypoplastic right heart syndrome (HRHS), which feature hypoplasia of LV, RV, aorta and pulmonary artery with an ASD, patent ductus arteriosus and atresia of mitral and tricuspid valves ( Figure 1N,O). Interestingly, OMD with exercise also had an HLHS but without HRHS (Table 2).
Representative images of pulmonary valves (Figure 2A-D) and aortic valves ( Figure 2E-H) at E18.5 show poorly re-modelled valves in the OMD group without exercise. Offspring from this group had significantly thicker pulmonary and aortic valves, and smaller aortic luminal diameters compared to those of controls, which was prevented with maternal exercise (2I-K).

| Effect of maternal exercise on coronary artery and capillary malformations in the offspring of diabetic mice
Coronary artery malformation was evaluated in E18.5 foetal hearts using immunostaining with α-smooth muscle actin antibody, which labelled smooth muscle cells of coronary arteries (representative images Figure 3A,B). The coronary artery diameter was determined in the ventricular myocardium at the four chamber view by measuring the widest luminal diameter in three sections. Hearts collected from diabetic mice that did not exercise had significantly smaller coronary artery diameter ( Figure 3E). The relative coronary artery abundance was quantified by averaging the number of positively stained ventricular arteries per section in three sections of the heart. The OMD group had significantly fewer coronary arteries per section compared to both controls ( Figure 3F). Three-dimensional reconstruction of the coronary vasculature and myocardium also showed less branching, as well as narrower and shorter coronary arteries in the OMD group compared to other groups ( Figure 3C). The ratio of the coronary artery volume to the myocardial volume was significantly smaller in this group ( Figure 3G). These abnormalities of coronary artery development in foetal hearts from diabetic mothers were prevented with maternal exercise. Finally, capillaries visualized by immunostaining with lectin-1 showed evidence of impaired capillary development in the offspring of diabetic mice ( Figure 3D). Capillary abundance was quantified in three sections at the widest four chamber view of the heart. In both the right and left ventricular myocardium, the number of capillaries normalized to the area of myocardium was significantly lower in the OMD group compared to respective controls, and this defect was prevented with maternal exercise ( Figure 3H).

| Effects of maternal exercise on cell proliferation and epicardial EMT in the developing embryonic heart
Cell proliferation and epicardial EMT are required for proper morphogenesis of the heart and coronary vasculature. 19,20 Proliferating cells were marked by immunostaining E12.5 hearts with pHH3 antibody ( Figure 4A-C), as histone H3 is specifically phosphorylated during mitosis. 21 Likewise, cells undergoing epicardial EMT were labelled by immunostaining with anti-Wt1 antibody ( Figure 4D).
Embryonic hearts from diabetic mice were smaller overall and ap-

| Effects of maternal exercise on gene expression and oxidative stress in embryonic hearts
To investigate molecular pathways regulating cardiogenesis and vasculogenesis during embryonic development, real-time RT-PCR was used to measure relative gene expression levels of Gata4, a critical factor for heart development, and Hif-1α, involved in Bmp10 ( Figure 5F-H). To evaluate whether maternal exercise also had an impact on foetal enzyme function, eNOS phosphorylation was measured using Western blotting ( Figure 5I). The ratio of p-eNOS to total eNOS was significantly lower in the OMD embryos compared to other groups. Moreover, the ratio was significantly higher in both exercise groups compared to the control group ( Figure 5J). There was no significant difference in relative levels of total eNOS ( Figure 5K).
Lastly, to determine the effects of maternal exercise on redox homoeostasis in heart development, markers for oxidative damage were evaluated at E14.5. DHE probe was used to measure superoxide levels, and immunostaining with anti 4-hydroxynonenal (4-HNE) antibody was performed to visualize lipid peroxidation ( Figure 6A,C).
Images were taken at a constant exposure in three sections and the fluorescence intensity was quantified by densitometry. There were significantly higher levels of oxidative stress markers (superoxide and 4-HNE) in the OMD group without exercise, which was normalized in the OMD group with maternal exercise (Figure 6B,D).

| D ISCUSS I ON
Finding effective strategies to modify disease risk is the focus of preventative healthcare, which has become increasingly relevant as the prevalence of chronic disease continues to rise. Risk factors such as pre-gestational diabetes have the potential to be controlled through pre-natal interventions. The aim of this study was to evaluate maternal exercise as an intervention to counteract the damaging effects of pre-gestational diabetes on heart development, as well as to highlight potential pathways that might lead to these benefits. The results of this study demonstrate that maternal exercise effectively Streptozotocin has been used in past studies to induce congenital malformations in animal models of type 1 diabetes. 13,15,22 To avoid unintended teratogenic effects, STZ was administered 1 week prior to mating. The drug is rapidly metabolized and has a plasma half-life of 5 minutes in mice, so it is expected to have been completely eliminated from the maternal environment by the time mating occurred. 23 No significant difference was found in running distance between control and diabetic mice. There was also no significant difference in the non-fasting blood glucose levels of Studies have shown that oxidative stress is a major driver of CHD pathogenesis. 13,15,34,35 In our study, E14.5 embryonic hearts from the offspring of diabetic mice had significantly higher levels of superoxide and lipid peroxidation, an indicator of oxidative damage. 36 Conversely, with maternal exercise, levels were normalized to control levels despite the diabetic dams having comparable hyperglycemic status. These results demonstrate that maternal exercise may improve the ability to maintain redox balance in in the In adults, exercise is known to transiently elevate ROS production, but paradoxically lead to improved redox homoeostasis capacity by up-regulating antioxidant defences. [37][38][39] A recent study by Chung et al. 40 found that maternal exercise during pregnancy lead to increased mitochondrial enzymatic activity and ATP production, while decreasing hydrogen peroxide levels in foetal mouse hearts.
However, whether maternal exercise may lead to up-regulation of endogenous antioxidants and/or, reduced ROS production due to improved mitochondrial efficiency in our model remains to be studied. Notably, eNOS-derived NO promotes Notch signalling, cell proliferation, EMT and differentiation during heart development, and has been implicated to play a role in ROS handling. 35,41,42 In the pres- One limitation of this study is that the effects of exercise were not evaluated also using a model of type 2 diabetes, which is the more common type of diabetes in patients and could be examined in future studies. However, of note, large meta-analyses have found that although women with type 2 diabetes achieve better glycemic control during pregnancy, the rate of malformations is not significantly different between type 1 and type 2 diabetes. 44 There is evidence that glycemic status alone is not the most important factor in diabetic teratogenesis. Serum from diabetic animals with normalized glucose levels were found to be teratogenic in in vitro models. 45 Also, numerous studies have shown that treatment with antioxidants alone are able to significantly reduce malformation rates in in vivo models, despite not having effects on hyperglycemic status. 13,15,34,35 While the present study did not show a direct causal relationship between reducing oxidative stress and beneficial effects of exercise, prior studies strongly indicate that oxidative stress is a key teratogenic component leading to diabetic embryopathy. 13,15,34 To conclude, the results of our study indicate that maternal exercise protects against diabetes-induced changes to embryonic gene expression, cell proliferation and EMT in embryonic hearts, possibly due to mitigation of oxidative stress. Future studies are required to provide more conclusive evidence to elucidate the molecular mechanisms by which maternal exercise benefits CHD pathogenesis. Studies in humans could be conducted to further evaluate the efficacy of maternal exercise in reducing CHD risk in pre-gestational diabetes. Blood glucose levels were high in our animal model because diabetes was untreated, whereas better glycemic control is expected in pregnant women receiving prenatal care. Unfortunately, although the risks of birth defects due to hyperglycemia have become well known and modern prenatal care emphasizes optimal glucose control before conception, the incidence of CHDs in the offspring of diabetic women has not improved over time. 5 Therefore, there is potential to further improve outcomes by modifying other teratogenic factors, in addition to maintaining glycemic control. Exercise may be a simple, yet effective intervention to supplement insulin therapy and minimize risks of birth defects in mothers with diabetes.

ACK N OWLED G EM ENTS
None. This work was performed in partial fulfilment of the requirements for Tana Saiyin for the Master of Science (MSc) degree in Physiology and Pharmacology, University of Western Ontario, London, Ontario, Canada.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.