Impact of maternal hyperglycemia on cardiac development: Insights from animal models

Abstract

Congenital heart disease (CHD) is the leading cause of birth-defect related death in infants and is a global pediatric health concern. While the genetic causes of CHD have become increasingly recognized with advances in genome sequencing technologies, the etiology for the majority of cases of CHD are unknown. The maternal environment during embryogenesis has a profound impact on cardiac development, and numerous environmental factors are associated with an elevated risk of CHD. Maternal diabetes (matDM) is associated with up to a five-fold increased risk of having an infant with CHD. The rising prevalence of diabetes mellitus has led to a growing interest in the use of experimental diabetic models to elucidate mechanisms underlying this associated risk for CHD. The purpose of this review is to provide a comprehensive summary of rodent models that are being used to investigate alterations in cardiac developmental pathways when exposed to a maternal diabetic setting and to summarize the key findings from these models. The majority of studies in the field have utilized the chemically induced model of matDM, but recent advances have also been made using diet-based and genetic models. Each model provides an opportunity to investigate unique aspects of matDM and are invaluable for a comprehensive understanding of the molecular and cellular mechanisms underlying matDM-associated CHD.

1. INTRODUCTION

Congenital heart disease (CHD) is the most prevalent type of birth defect, affecting approximately 1 in every 100 live births throughout the world (Hoffman & Kaplan, 2002; Oster et al., 2013; Wu, He, & Shao, 2020). The cardiac malformations range in severity, but 25% of CHD cases are critical, defined as those that will require surgical or catheter-based intervention within their first year of life. Although breakthrough medical advances have prolonged the lifespan of patients with CHD, long-term complications compromise the overall quality of life in a significant subset of cases, and 5% of non-critical CHD patients do not survive beyond 18 years of age (Leirgul et al., 2014). Despite the breadth of research focused on understanding the molecular mechanisms underlying CHD, limitations in translational approaches have hindered efforts to devise strategies for the prevention of CHD.

The etiology of CHD is complex, involving both hereditary and non-hereditary risk factors (Jenkins et al., 2007; Pierpont et al., 2018). An intricate genetic program regulates the dynamic stages of cardiac development (Srivastava & Olson, 2000; Gelb & Chung, 2014; Wu et al., 2020). Consequently, any perturbation in the cardiac genetic network can lead to abnormal development of the heart. While chromosomal aneuploidy and copy number variation are well established to be associated with CHD, advances in genomic technologies have led to the discovery of numerous pathogenic variants in cardiac genes in patients with CHD, including de novo mutations and rare inherited variants (Zaidi et al., 2013; Homsy et al., 2015; Sifrim et al., 2016; Jin et al., 2017; Pierpont et al., 2018; Gifford et al., 2019). Damaging mutations have been identified in genes encoding transcription factors, ligands and receptors of cell signaling pathways, structural proteins, epigenetic modifiers and laterality pathway proteins (Homsy et al., 2015; Sifrim et al., 2016; Jin et al., 2017). The striking genetic similarity between mice and men has allowed researchers to develop specific mouse models to validate CHD causing genetic variants, gaining additional insights into the molecular mechanisms underlying the genetic etiology of CHD (Moon, 2008; Majumdar, Yasuhara, & Garg, 2021).

Although the genetic contributors in the etiology of CHD are critically important, epidemiological studies also point to the role of environmental teratogens. While 30% of CHD cases have a definite underlying genetic cause, environmental risk factors can account for a significant subset of CHD while the remaining cases of unknown etiology are proposed to be the result of hitherto undefined gene-environment interactions (Nora, 1968; Jenkins et al., 2007; Hinton, 2013; Kalisch-Smith, Ved, & Sparrow, 2020). In fact, up to 57% of hypoplastic left heart syndrome (HLHS) and 37% of tetralogy of Fallot cases can be attributed to exposure to modifiable risk factors during pregnancy (Wilson, Loffredo, Correa-Villasenor, & Ferencz, 1998; Simeone et al., 2016). The maternal environment during embryogenesis is highly modifiable and vulnerable to xenobiotics that have the potential to disrupt cardiac morphogenesis (Kalisch-Smith et al., 2020). These include prenatal exposure to (1) heavy metals such as lead, (2) high doses of alcohol, (3) retinoic acid and vitamin A, and (4) therapeutic agents including thalidomide, antidepressants (such as lithium), and anticonvulsants (such as carbamazepine and valproic acid) (Lammer et al., 1985; Meador, Reynolds, Crean, Fahrbach, & Probst, 2008; Liu et al., 2015; J. Yang et al., 2015; Ou et al., 2017; Patorno et al., 2017). In contrast, the presence of a maternal disease or condition has a profound impact on cardiac development. Evidence of this is given by the increased risk of CHD in infants of mothers suffering from (1) diabetes mellitus (2) rubella infection (3) hyperthermia (4) hypoxia, (5) obesity (6) phenylketonuria or (7) nutritional deficiencies in folic acid, vitamin B3 or vitamin D (Jenkins et al., 2007; Oster, Riehle-Colarusso, & Correa, 2010; Hinton, 2013; Simeone et al., 2015; Simeone et al., 2016; Kalisch-Smith et al., 2020; Yang et al., 2021). A global rise in the prevalence of diabetes mellitus, particularly in women of childbearing age, has made maternal diabetes mellitus an increasingly pervasive risk factor for CHD, demanding the attention of researchers investigating the multifactorial etiology of CHD (Mayer-Davis et al., 2017; Ogurtsova et al., 2017).

Maternal diabetes mellitus (matDM) encompasses (1) diabetes mellitus type 1, which is an autoimmune disorder leading to insufficient insulin production, (2) diabetes mellitus type 2, in which the body loses its ability to respond to endogenous levels of insulin, and (3) gestational diabetes, a transient hyperglycemic state during pregnancy, manifesting in the second to third trimester. An association between diabetic mothers and birth defects has been suspected since the nineteenth century, but population studies investigating the rate of cardiac malformations from diabetic pregnancies did not crop up until the late twentieth to early twenty-first century (Pedersen, Tygstrup, & Pedersen, 1964; Åberg, Westbom, & Källén, 2001). The risk of CHD in infants from diabetic pregnancies is increased up to five-fold compared to non-diabetic controls, particularly in those affected by pre-gestational type 1 and type 2 diabetes (Loffredo, Wilson, & Ferencz, 2001; Wren, Birrell, & Hawthorne, 2003; Abu-Sulaiman & Subaih, 2004; Yang, Cummings, O’connell, & Jangaard, 2006; Ullmo et al., 2007; Liu et al., 2013; Øyen et al., 2016). Although the incidence of CHD from type 1 and type 2 matDM are similar, there appears to be increased predilection for specific types of CHD depending on the type of exposure (Liu et al., 2013). An association between gestational diabetes with cardiac structural defects has also been reported, though the risk is less than that with pre-gestational diabetes. However, a high incidence of hypertrophic cardiomyopathy, characterized by thickening of the cardiac walls, has been reported with exposure to gestational diabetes (Loffredo et al., 2001; Dervisoglu, Kosecik, & Kumbasar, 2018; Wu et al., 2020).

While in vitro culture of mammalian cells has provided useful insight on the dysregulation of molecular processes in response to hyperglycemia, matDM represents an abnormal metabolic state that has pleiotropic effects. Additionally, the development of the four-chambered mammalian heart requires complex interactions between several cell lineages, each of which likely have a unique response to the diabetic milieu. While there are several existing animal models that are being used to investigate diabetes mellitus and multiple models are utilized in the study of cardiac development, rodents are the primary animal model that has been used to study cardiac development in a diabetic setting (Rees & Alcolado, 2005; Jawerbaum & White, 2010; Shafrir, 2010; Tu & Chi, 2012; Vilches-Moure, 2019; Majumdar, Yasuhara, & Garg, 2021). Here, we will review the different rodent models of matDM-associated CHD, summarize their salient characteristics, and provide a comprehensive overview of key findings generated from these studies.

2. CARDIAC DEVELOPMENT

Prior to delineating the complex mechanisms underlying matDM-associated CHD, it is important to understand the cellular and molecular pathways that regulate cardiac development. Here, we will provide an overview of the key morphogenetic events comprising cardiac development and refer the readers to comprehensive, detailed reviews (Kelly, Buckingham, & Moorman, 2014; Meilhac & Buckingham, 2018; Christoffells & Jensen, 2020; Buijtendijik, Barnett, van den Hoff, 2020). In the mammalian embryo, the heart is the first organ to form, receiving contributions from multiple cell lineages under tight spatial and temporal regulation (Schleich, Abdulla, Summers, & Houyel, 2013). The distinct stages of cardiac development have been elucidated and documented in several animal models but for the purpose of our study, we will describe the timeline for cardiac morphogenesis in mice. Briefly, cardiogenic specification begins at embryonic day (E)7.5 of mouse development as the first heart field cells (FHF), residing in the anterior plate mesoderm lateral to the head forming region, receive inductive FGF and BMP signaling from the adjacent endoderm and organize themselves into the cardiac crescent (Figure 1a). Second heart field cells (SHF), arising from the pharyngeal mesoderm and marked by expression of Islet-1, follow suit as they become committed to a cardiogenic fate (Kelly et al., 2014; Meilhac & Buckingham, 2018). At ~E8.0, the FHF, closely followed by SHF, migrates to the ventral midline and fuse together to form a single linear beating heart tube, composed of an inner endocardial layer and an outer myocardium, separated by an acellular cardiac jelly (Figure 1b). As the migrating cardiac progenitor cells are added along the length of the tube, it begins to loop in a rightward direction bringing the inlet segment, comprising the common atria and atrioventricular canal, above the ballooning future left ventricle and the common outflow tract is positioned above the future right ventricle (Figure 1c). Highly conserved cardiac transcription factors Nkx2.5, Gata4/5/6, Tbx5/20, Mef2b/c and Hand1/2 are expressed in the cardiac progenitor cells, establishing a cardiogenic gene network that drives downstream lineage restriction and cardiac cell fate determination (Olson, 2006). The FHF population primarily contribute to the left ventricle and the most primitive parts of the atria while SHF gives rise to right ventricle, outflow tract and majority of the atria. At this point, left-right asymmetry is established in the heart and at E9.0, the looped heart begins to segment into four distinct chambers (Figure 1d). Epicardial cells, derived from the epicardium, an extracardiac structure located dorsal to the heart tube, begin to outline the myocardial surface of the developing heart. At E9.5, endothelial to mesenchymal transition (EMT) is initiated at the atrioventricular (AV) canal under the coordinated influence of multiple signaling pathways including TGFβ, Notch, Wnt, and VEGF signaling (Nakano, Nakano, Smith, & Palpant, 2016). Endocardial cells invade the accumulated cardiac jelly at this region and transform into mesenchymal cells, ultimately forming the endocardial cushions. The AV endocardial cushions undergo remodeling to give rise to the atrioventricular valves, the atrioventricular septum and the membranous portion of the ventricular septum. The SHF cells also contribute to the atrioventricular mesenchymal complex, giving rise to the dorsal mesenchymal protrusion (DMP) (Snarr et al., 2007). The third component of the AV septal complex includes endocardial-derived mesenchymal cap that arises from the primary atrial septum and fuses with DMP to drive atrial and AV septation (Deepe et al., 2020). However, the origin and molecular mechanisms that govern mesenchymal cap formation is still under investigation.

Figure 1: Stages of cardiac morphogenesis.

Summary of key stages of cardiac development in human and mice. (a) The first heart fied (FHF) and second heart field (SHF) organize themselves into a crescent like structure (E7.5). (b) FHF/SHF migrate to the midline and fuse to form linear heart tube (E8.0). (c) Heart tube undergoes rightward looping, segmenting the heart intro atria, ventricles, outflow tract (OFT). Endocardial cushions form via endothelial to mesenchymal transformation in the atrioventricular canal (AVC) and OFT, as well as neural crest cell migration to the OFT (E10.5). Epicardial- derived cells also migrate into the myocardium to give rise to several cell types. (d) Mature heart forms as complete septation of the heart leads to the four distinct chambers and septation of OFT results in aorta (Ao) and pulmonary artery (PA). Endocardial cushion remodeling gives rise to atrioventricular valves and semilunar valves. RA = Right atrium, LA = Left atrium, RV= right ventricle, LV = Left ventricle, IVS = interventricular septum. Created with Biorender.com.

Endocardial cushions also develop in the OFT starting E10.5. However, OFT cushions require contributions from migrating neural crest cells, arising from the pharyngeal arches and marked by Pax3 expression (Conway, Henderson, & Copp, 1997; Miyagawa-Tomita, Arima, & Kurihara, 2016). The OFT cushions form the aorticopulmonary septum, separating it into the aorta and pulmonary artery, while also giving rise to the semilunar valves (Chang et al., 2004; Kodo et al., 2017; Stefanovic, Etchevers, & Zaffran, 2021). A subset of epicardial cells migrate into the myocardium and differentiate into coronary endothelial cells, vascular smooth muscle cells, and fibroblasts (Chong et al., 2011; Wu, Dong, Regan, Su, & Majesky, 2013; Cano et al., 2016; Simões & Riley, 2018).

Key genes responsible for the highly orchestrated actions of cardiac progenitor cells have now been identified and many of the molecular processes underlying cardiac development have been described elsewhere (Kathiriya, Nora, & Bruneau, 2015; Paige, Plonowska, Xu, & Wu, 2015, Meilhac & Buckingham, 2018; Christoffels & Jensen, 2020). However, understanding how these processes become dysregulated to give rise to CHD is of utmost importance, as the heterogeneity of CHD represents perturbation of phenotype specific cardiac developmental pathways.

3. CONGENITAL HEART DISEASE AND THE ASSOCIATION WITH MATERNAL DIABETES

The intricate nature of cardiac morphogenesis necessitates that the molecular processes regulating cellular migration, induction, and patterning are tightly regulated and any disruption can lead to CHD. Accordingly, malformations may occur in all structures of the heart, but most can be classified into the following subtypes: 1) cardiac septation defects, 2) conotruncal heart defects, 3) obstructive defects of the right and left ventricular outflow tract and 4) laterality defects (left-right abnormalities) (Garg, 2006). Defects in cardiac septation account for nearly 50% of CHD cases, including atrial septal defects (ASD), ventricular septal defects (VSD) and atrioventricular septal defects (AVSD). Defects in the OFT are found in 20–30% of all CHD and include persistent truncus arteriosus (PTA; lack of OFT septation into aorta and pulmonary artery), transposition of the great arteries (TGA), tetralogy of Fallot (ToF), and double outlet right ventricle (DORV). Obstructive lesions include pulmonary valve stenosis and hypoplastic right heart syndrome which affect right-sided cardiac structures and bicuspid aortic valve, aortic valve stenosis and hypoplastic left heart syndrome which affects structures on the left-side of the heart. Abnormal cardiac looping presents itself as laterality defects and is often associated with heterotaxy syndrome.

A range of cardiac malformations is reported in infants exposed to matDM. Meta-analyses have been carried out by several groups and have identified the increased likelihood of specific CHD phenotypes from exposure to maternal pre-gestational diabetes mellitus, including TGA, PTA, heterotaxy, VSD and ASD (Lisowski et al., 2010; Hoang, Marengo, Mitchell, Canfield, & Agopian, 2017). The incidence of CHD from pre-gestational matDM type 1 and type 2 are similar, but type 1 has the highest risk for conotruncal malformations and AVSD while type 2 is more associated with heterotaxy and left ventricular outflow tract obstructive malformations (Liu et al., 2013). In fact, mild elevations in serum glucose during pregnancy, even without clinical diagnosis of diabetes mellitus, has been associated with tetralogy of Fallot, although this association was not found with other cardiac phenotypes such as TGA (Priest, Yang, Reaven, Knowles, & Shaw, 2015). In contrast, the lower rate of CHD in infants exposed to gestational diabetes suggests that timing of hyperglycemic insult during cardiac development is crucial for the manifestation of CHD, as gestational diabetes is observed after the critical period of cardiac morphogenesis (Helle et al., 2018).

The diabetic environment in utero represents a complex biochemical scene. Abnormal glucose and inositol metabolism as well as dysregulated arachidonic acid signaling are hallmarks of diabetic pathogenesis and the developing fetus may have a unique molecular response to each of these biochemical changes (Goldman et al., 1985; Hod, Star, Passonneau, Unterman, & Freinkel, 1986; Zabihi & Loeken, 2010). In addition, diabetic dyslipidemia and deranged maternal lipid profiles have also been associated with CHD (Chapman et al., 2011; Smedts et al., 2012). However, the principal presentation of all forms of diabetes mellitus is hyperglycemia and studies from across the field have pointed to maternal hyperglycemia as the primary teratogen for matDM-associated CHD. This is corroborated by the association of CHD in infants of women with poor glycemic indexes even without clinical presentation of diabetes mellitus and in animal models of matDM-associated CHD with elevated glucose levels (Table 1) (Priest et al., 2015). In matDM, the insulin-independent glucose transporter, GLUT1, transports maternal glucose across the placenta to the fetal circulation leading to elevated intrauterine glucose levels that potentially alter fetal cardiac developmental processes (Hay Jr, 2006). Although a diverse range of animal models have been used to tease apart the molecular mechanisms underlying cardiac development and CHD, investigating cardiac development in a diabetic setting with hyperglycemic exposure requires certain special considerations. In the subsequent section, we have described the different rodent models of matDM where cardiac developmental pathways have been found to be dysregulated in the exposed progeny.

Table 1:

Different protocols used for streptozotocin (STZ) induced maternal type 1 diabetes in rodents

Rodent strain	STZ injection protocol	Diabetic blood glucose cut-off (mmol/L)	Reference(s)
Swiss Albino mice	i.p injection; 75 mg/kg body weight; 3 days	>16	(Kumar, Dheen, & Tay, 2007), (Lin, Cai, Zhang, & Chen, 2018)
	i.p injection; 200 mg/kg body weight; single dose	>16.65	(Hachisuga et al., 2015)
FVB mice	i.p injection; 100 mg/kg body weight; 3 days Subcutaneously implanted insulin pellets given; not removed throughout pregnancy	N/A	(Morgan, Relaix, Sandell, & Loeken, 2008)
	i.p injection; 100 mg/kg body weight; 2 days	>13.9	(Bohuslavova, Skvorova, Sedmera, Semenza, & Pavlinkova, 2013), (Pavlinkova, Salbaum, & Kappen, 2009), (Cerychova et al., 2018)
	i.p. injection; 60 mg/kg body weight; 5 days; Subcutaneously implanted insulin pellets given; not removed throughout pregnancy	>17	(Zhao et al., 2016)
C57BL/6J	i.p injection; 75 mg/kg body weight; 3 days	>11.11	(Basu et al., 2017),
	i.p injection; 75 mg/kg body weight; 2 days Insulin pellets given post IP which was removed at E5.5	>13.9	(Wang, Fisher, Zhong, Wu, & Yang, 2015), (Wang, Reece, & Yang, 2015), (Wang, Wu, Quon, Li, & Yang, 2015)
	i.p injection; 80 mg/kg body weight; 3 days	>11.00	(Moazzen et al., 2014)
	i.p injection; 50 mg/kg body weight; 5 days	>11.00	(Saiyin et al., 2019)
Sprague Dawley rats	i.v. injection (tail vein); 40 mg/ kg; single dose	>20	(Wentzel, Gäreskog, & Eriksson, 2008)
	i.p injection; 35 mg/kg body weight; single dose Additional 1–4 doses of 15 mg/kg body STZ injected when necessary	>15	(Lehtoranta et al., 2013)

4. RODENT MODELS OF MATERNAL DIABETES MELLITUS

In order to recapitulate the placental transfer of elevated maternal glucose into the developing fetus, a mammalian model must be used to study matDM-associated CHD. The similarity of embryonic development between human and rodents have made them a vital research tool, and this is particularly relevant for cardiac development where the four-chambered heart in the rodent allows for study of the spectrum of cardiac malformations found in humans (Krishnan et al., 2014). Over the years, a variety of diabetic rodent models have been described in order to study its pathophysiology and molecular mechanisms with a goal of identifying novel therapies (Foglia, Ibarra, & Cortés, 1967; Rees & Alcolado, 2005; Cefalu, 2006; Kiss et al., 2009). Diabetes mellitus can be induced in rodents via surgical pancreatectomy, treatment with diabetogenic chemicals or diets, and genetic manipulation of the insulin receptor. Additionally, rodent strains that develop spontaneous autoimmune disorders leading to insulin deficiency have also been outbred to study the pathogenesis of diabetes mellitus. However, not all models of diabetes mellitus are suitable for studying the effect of matDM on organogenesis, as the severe diabetes can negatively impact the animal’s fertility. In this section, we focus on two rodent models that primarily have been used to study matDM-associated CHD, the chemically induced model of matDM type 1 and a high-fat diet based model of matDM type 2, and summarize the key pathogenic mechanisms identified in each.

4.1. Chemical induction of maternal type 1 diabetes mellitus

Administration of diabetogenic toxins is the most widely used strategy to generate diabetic rodent models. Among the several diabetogenic agents identified, streptozotocin (STZ) and alloxan are the two most commonly used chemicals. Although both compounds have a similar mode of action and have been used to study matDM-associated congenital malformations, only STZ has been used to investigate matDM-associated CHD (Lenzen, 2008; Radenković, Stojanović, & Prostran, 2016).

STZ is a glucose analog in which the N-methyl-N-nitrosourea moiety is linked to the second carbon of a hexose. Following intraperitoneal or intravenous injection, STZ enters the pancreas via GLUT2 transporters, exerting cytotoxic effects on the insulin producing β-cells by DNA alkylation and generation of excess free radicals (Lenzen, 2008). Consequently, STZ results in insulin deficiency leading to a diabetes mellitus type 1-like state. A single high dose injection or multiple low doses of STZ can be administered to achieve a diabetic state (Table 1). Multiple doses of 60–100 mg/kg over 2–5 consecutive days is the most common protocol used for matDM mouse models in order to achieve stable hyperglycemia while maintaining fertility (Kumar, Dheen, & Tay, 2007; Morgan, Relaix, Sandell, & Loeken, 2008; Bohuslavova, Skvorova, Sedmera, Semenza, & Pavlinkova, 2013; Wang, Reece, & Yang, 2015; Basu et al., 2017). Hyperglycemia is detected within a week of injections. On the other hand, a single dose of 35–40 mg/kg is usually administered in rat models, though one study has reported using additional injections of 15 mg/kg to establish stable hyperglycemia (Wentzel, Gäreskog, & Eriksson, 2008; Lehtoranta et al., 2013). The cut-off for blood glucose of diabetic animal groups differs between studies, likely due to differences in injection protocols and these have been summarized in Table 1.

As liver and kidney cells also express GLUT2, a high dose of STZ can cause acute liver and kidney toxicity (Tesch & Allen, 2007). However, upon administration, STZ is rapidly metabolized in the liver, with a serum half-life of 10–15 minutes, and is quickly eliminated through renal excretion. As persistent hyperglycemia is established weeks prior to mating and tissue collection, any pathological changes in the fetus is attributed to the maternal diabetic state and not the toxic nature of STZ itself (Wu & Yan, 2015). However, STZ does have the potential to cross the placental barrier as observed when injections are performed during gestation in studies using rhesus monkeys and thus, treatment should be restricted to pre-gestational periods (Reynolds, Chez, Bhuyan, & Neil, 1974; Eleazu, Eleazu, Chukwuma, & Essien, 2013). Another challenge encountered with this model is the inability to achieve high pregnancy rates due to insulinemia and ovarian dysfunction, particularly in animals with severe hyperglycemia (Tesone et al., 1983). To mitigate this, studies describe the use of subcutaneously implanted insulin pellet until conception (Morgan et al., 2008; Wang, Fisher, Zhong, Wu, & Yang, 2015; Wang, Reece, et al., 2015; Zhao et al., 2016). The pellets are removed once the animal is confirmed to be pregnant, around E5.5, to allow hyperglycemia to be reestablished during the critical period of cardiac development (E7.5 onwards).

The incidence of cardiac defects in offspring exposed to STZ-induced maternal diabetes is relatively consistent across studies, ranging from 10–30% (Kumar, Yong, Dheen, Bay, & Tay, 2008; Wang, Wu, Quon, Li, & Yang, 2015; Basu et al., 2017). However, one study in which the diabetic group was not given insulin pellets, the incidence of CHD in the offspring rose to 58.1% (Moazzen et al., 2014). The CHD phenotypes observed include cardiac septation defects, including AVSD and VSD, and conotruncal heart defects including anomalies in septation of the cardiac outflow tract and DORV.

STZ provides a cost-effective method for generating multiple diabetic dams simultaneously as these injections do not require a high level of expertise and the mice can be treated as soon as they are of breeding age. The rapid onset of disease following treatment allows researchers to increase sample sizes while reducing the lag time required generating more diabetic females. As a result, the majority of work used to understand mechanisms underlying matDM-associated CHD have used the STZ-induced matDM type 1 model (Wang, Fisher, et al., 2015; Basu & Garg, 2018). These studies have identified numerous cellular processes and molecular pathways that are dysregulated in the developing heart of STZ-induced matDM exposed embryos and have been summarized in Table 2.

Table 2:

Dysregulated fetal cardiac gene and protein expression with exposure to maternal diabetes

Rodent strain of matDM	Altered fetal cardiac gene and protein	Reference(s)
(A) STZ induced matDM type 1
Swiss Albino mice	Bmp4, Msx1, Pax3	(Kumar et al., 2007)
	Affymetrix mouse genome 430 2.0 microarrays (GSE32078)	(Lin et al., 2018)
	Pitx2	(Hachisuga et al., 2015)
FVB mice	Pax3	(Morgan et al., 2008)
	Hif1α, Nkx2.5, Tbx5, Mef2C, α-SMA, Cx43, Nppa	(Bohuslavova et al., 2013)
	Affymetrix mouse 430 2.0 arrays (GSE9675)	(Pavlinkova et al., 2009)
	SOLid SAGE mRNA deep sequencing (PRJNA275285)	(Zhao et al., 2016)
	RNA-seq of diabetes exposed Hif1α+/− and wt offspring (GSE109633)	(Cerychova et al., 2018)
C57BL/6J mice	Notch1, Hey2, Bmp10, EfnB2, Nrg1, Jarid2, eNOS	(Basu et al., 2017)
	sFRP1, Dkk1, β-catenin, Islet1, Gja1, Versican, Wnt5a, NFAT2/4, Mrtf-b, Tpm1, Rcan1	(Wang, Fisher, et al., 2015)
	Gata4, Gata5, Vegfa	(Moazzen et al., 2014)
	TGFβ1, TGFβ3, TβRII, Smad2/3, Snai2, CTGF, GDF1	(Wang, Reece, et al., 2015)
	Cyclin D1, Cyclin D3, P21, P27, BMP4, p-Smad1/5/8, Nkx2.5, Gata4, Tbx5, p-ASK1, cleaved caspase 3/8, p-JNK1/2, p-C-JUN, p-ATF2, p-Elk1, p-FoxO3a, p-PERK, p-eIF2α, BiP, CHOP, XBP1	(Wang, Wu, et al., 2015)
	Gata4, Hif1α, Cyclin D1, Notch1, Snail1	(Saiyin et al., 2019)
Sprague-Dawley rats	Nppa, Nppb, Myh2, Myh3, Atp2a2, Kcnip2, Ucp2/3, Slc2a4, Egln3, Tnfrsf12a	(Lehtoranta et al., 2013)
	CuZnSOD, MnSOD, Gpx-1	(Wentzel et al., 2008)
(B) HFD induced matDM type 2
C57BL/6J	CHOP, BiP, Calnexin, PDIA, GRP94, p-IRE1α, p-eIF2α, p-PERK, XBP1, Cleaved caspase 3/8	(Wu et al., 2016)
(C) Direct glucose infusion during gestation
Sprague-Dawley rats	PI3-Akt, MAP-Kinase, JNK, ERK, p38	(Gordon et al., 2015)
(D) Genetic models of matDM
Ins2Akita/+	PGC-1α, PPAR-α, LPL, FAT/CD36, and FATP1	(Lindegaard & Nielsen, 2008)
	Tnnt2	(Nakano et al., 2017)

Cellular Processes

Oxidative stress.

Numerous studies have implicated oxidative stress as the primary mechanism underlying diabetic embryopathy. In fact, hyperglycemia mediated oxidative stress is capable of inducing downstream molecular dysregulation that may contribute to CHD (Zabihi & Loeken, 2010). Evidence of increased oxidative stress is consistently reported in fetal hearts of diabetic pregnancies in rodents (Morgan et al., 2008; Wang, Fisher, et al., 2015; Wang, Reece, et al., 2015; Moazzen et al., 2014; Basu et al., 2017). In Sprague Dawley rats, activities of antioxidant scavenging enzymes superoxide dismutase (SOD1), catalase and glutathione peroxidase are lower in embryos from diabetic dams compared to untreated controls (Sivan, Lee, Wu, & Reece, 1997). Furthermore, decreased expression of glutathione peroxide was found specifically in the heart of malformed embryos compared to non-malformed embryos of the same diabetic litter (Wentzel et al., 2008). In diabetic C57BL/6 mice, supplementation with antioxidant N-acetyl cysteine (NAC) in drinking water of pre-gestational diabetic mice reduced the incidence of CHD from 58.1% to 16.3%. Fetal hearts from NAC-treated diabetic mice had increased total glutathione and GSH levels, decreased reactive oxygen species (ROS), increased cell proliferation, decreased apoptosis, and restored expression of key cardiac developmental genes, Gata5, Gata4 and Vegfa, to normal untreated levels (Moazzen et al., 2014). However, a significant rescue of VSD in offspring from NAC-treated diabetic dams compared to non-NAC treated diabetic group was not recapitulated by another group but this may have been related to the severity of maternal hyperglycemia (Basu et al., 2017). Diabetic Swiss Albino mice that were given zinc supplements, acting as cofactors for antioxidant enzymes, also had reduced incidence of CHD, from 9.61% to 2.17%, and decreased oxidative stress in the embryonic hearts exposed to matDM (Kumar et al., 2012). Strikingly, maternal voluntary exercise of diabetic dams was able to reduce the incidence of CHD in embryos from 59.5% to 25%, with the maternally exercised embryos having significantly lower ROS levels and higher expression of Notch1, Snai1, Gata4, and Cyclin D1 compared to non-exercised controls (Saiyin et al., 2019). Consistent with this, genetic overexpression of antioxidant enzyme, SOD1, in mouse embryos exposed to matDM showed 100% rescue of CHD phenotypes, decreased oxidative stress, cellular apoptosis and restored the impaired TGFβ and Wnt signaling observed in fetal hearts exposed to matDM (Wang, Fisher, et al., 2015; Wang, Reece, et al., 2015). Additionally, administration of antimycin A, used to generate oxidative stress at day 7.5 of gestation in Pax3-reporter mice, showed defective cardiac neural crest cell migration, leading to OFT defects, similar to that observed in matDM (Morgan et al., 2008). When the oxidative stress responsive apoptotic gene, Ask1, was deleted in matDM exposed embryos, expression of cardiac transcription factors Bmp4, Nkx2.5, Gata5 and Smad1/5/8 phosphorylation was restored to normal levels (Wang, Wu, et al., 2015). Thioredoxin, another antioxidant, has been shown to rescue neural tube defects in embryos exposed to matDM via ASK1-FoxO3a-caspase 8 pathway, but the effect on CHD has not been reported (Yang et al., 2013). However, a major caveat of the oxidative stress hypothesis is that most studies depend on analysis of ROS at later stages of cardiac development, ranging from E12.5 – E14.5, at which point cardiac cell lineages are committed and well underway to form specific structures of the heart. In fact, during early proliferative stages of cardiac development, oxidative phosphorylation, the major source of ROS, is not operational and only begins to take effect after E9.5 (Mackler, Grace, & Duncan, 1971; Cox & Gunberg, 1972; Hom et al., 2011). Therefore, it is possible that hyperglycemia-associated oxidative stress is not responsible for the early molecular disruption required to induce a subset of CHD and further investigation is required to determine its effect on cardiac morphogenetic processes which occur between E9.5 – E12.5 (e.g neural crest migration, EMT, cardiac cushion remodeling).

Cell apoptosis and proliferation.

The numerous morphogenetic events that occur during cardiac development require the coordinated regulation of cell proliferation and apoptosis (Fisher, Langille, & Srivastava, 2000). As the atrial and ventricular chambers enlarge throughout development, the cells in these regions undergo high rate of proliferation while also allowing a small number of cells to undergo apoptosis. In contrast, in the endocardial cushions of the OFT and atrioventricular canal, programmed cell death and proliferation are tightly balanced as these structures undergo remodeling to form the cardiac valves, the aorticopulmonary septum and the membranous ventricular septum (Fisher et al., 2000). MatDM exposure consistently leads to decreased proliferation and increased apoptosis in both rat and mouse fetal hearts, which can be rescued with antioxidant supplementation, as described above. Studies have found increased number of apoptotic cells and decreased cell proliferation specifically in the endocardial cushions and the ventricular myocardium of embryos exposed to matDM (Kumar et al., 2007). In rats, there is downregulation of Cited2 in the hearts of matDM-exposed embryos, which leads to abnormal apoptosis of cardiomyocytes (Su et al., 2016). In another study, expression of a cardiomyocyte proliferation marker, Tnfrsf12a, was observed to be upregulated and a high cellular turnover was identified in fetal rat hearts exposed to matDM (Lehtoranta et al., 2013). Consistent with this, transcriptomic analysis of E13.5 and E15.5 hearts from diabetes-exposed embryos yielded apoptosis as one of the top dysregulated pathways (N. Lin, Cai, Zhang, & Chen, 2018). As mentioned above, loss of the apoptotic gene Ask1 in matDM exposed embryos led to rescue of deficits in cell proliferation along with normalization of cardiac gene expression including Bmp4, Nkx2.5, Gata5 (Wang, Wu, et al., 2015). Taken together, it appears that oxidative stress mediates improper cardiac cell proliferation and apoptosis during exposure to matDM. However, the partial rescue of CHD phenotypes despite restoration of normal apoptosis and proliferation suggests other molecular mechanisms are also at play in mediating matDM-induced CHD.

Molecular Pathways

TGFβ signaling.

TGFβ signaling is crucial during early cardiac development, playing a significant role in cardiac OFT remodeling and ventricular septum formation (Sanford et al., 1997; Compton, Potash, Brown, & Barnett, 2007). BMP4, a member of the TGFβ superfamily and a signaling molecule for EMT, is downregulated in the ventricular myocardium and the cardiomyocytes overlying endocardial cushions of E13.5 embryos from diabetic mice (Kumar et al., 2007). Additionally, Msx1, known to regulate Bmp4, is downregulated in the endocardial cushions of these embryos (Han, Yang, Farrington, & Muneoka, 2003; Kumar et al., 2007). Conflicting with this, mRNA levels of Bmp4 were not changed in E11.5 hearts exposed to matDM compared to control hearts, while others have shown that matDM exposure leads to downregulation of TGFβ ligands (tgfb1 and tgfb3) and phosphorylation levels of downstream effectors (TBRII and Smad2/3) in E12.5 murine hearts (Moazzen et al., 2014; Wang, Reece, et al., 2015). SOD1 overexpression in embryos from diabetic mice was able to restore the expression levels of TGFβ signaling targets Snai2, Ctgf and Gdf1. It appears that dysregulation of certain components of TGFβ signaling, such as Bmp4, is specific to the developmental stage, which may explain effects on specific morphogenetic events during cardiac development.

Wnt signaling.

Wnt signaling is vital to cardiac development and deletion of both canonical and non-canonical effectors of Wnt pathway lead to OFT and ventricular septation defects in mice, consistent with the CHD phenotypes observed in offspring exposed to matDM (Lin et al., 2007; Schleiffarth et al., 2007; Yu, Ye, Guo, & Nathans, 2012). Wnt antagonists, sFRP1 and Dkk1 are found to be increased while phosphorylation of DVL2 and GSK3B is reduced in embryonic hearts exposed to matDM, suggesting inactivation of canonical Wnt signaling (Wang, Fisher, et al., 2015). Additionally, noncanonical Wnt5a, restricted to the OFT, is found to be downregulated in matDM-exposed embryos compared to controls. SOD1 overexpression in matDM-exposed embryos is able to reduce the levels of sFRP1 and Dkk1 and activate downstream Wnt targets Islet 1, Gja1 and Versican as well as restoring Wnt5a expression to normal levels observed in non-diabetic embryos (Wang, Fisher, et al., 2015).

Notch and Nitric oxide signaling.

Notch signaling in the endocardium plays a key role in regulating EMT and ventricular trabeculation and mutations in the Notch signaling pathway have been associated with human congenital heart defects (Timmerman et al., 2004; Garg et al., 2005; High & Epstein, 2008; Monte-Nieto et al., 2018). Interestingly, Notch1^+/− embryos in mice exposed to matDM have a significantly higher incidence of CHD (85.7%) compared to their wildtype littermates (22.2%) and non-diabetic Notch1^+/− embryos (0%) (Basu et al., 2017). Additionally, downstream Notch targets Hey2, Nrg1, EfnB2 and Bmp10 are found to be downregulated in embryonic hearts exposed to matDM. Expression of Jarid2, a component of the polycomb repressive complex (PRC2), was found to be upregulated with matDM exposure and enrichment of JARID2 was found on Notch1 regulatory sequence upstream of the promoter region, providing evidence supporting an epigenetic mechanism underlying downregulation of Notch signaling in matDM exposed embryonic hearts (Basu et al., 2017). Another study showed reduced Notch1 signaling in ex vivo embryo cultured in hyperglycemia leading to defects in left-right axis formation (Hachisuga et al., 2015). Furthermore, a genetic interaction has been described between endothelial nitric oxide synthase (Nos3) and Notch1, in which Nos3^−/−;Notch^+/− mice exhibited a range of conotruncal and semilunar valve malformations (Bosse et al., 2013; Koenig et al., 2016). Interestingly, decreased nitric oxide bioavailability, which is found with oxidative stress, and reduced Nos3 expression are found in embryonic hearts exposed to maternal diabetes, whereas NOS2 expression is found to be elevated (Kumar et al., 2008; Weng, Li, Reece, & Yang, 2012; Basu et al., 2017). Treatment of diabetic dams with sapropterin, a cofactor of NOS3, was able to reduce the incidence of CHDs from 59% to 27% and fetal hearts from treated group had increased NOS3 dimerization and lowered ROS (Engineer et al., 2018).

Hypoxia signaling.

Primarily mediated by HIF-1, hypoxia signaling plays a significant role in proper OFT remodeling (Sugishita, Leifer, Agani, Watanabe, & Fisher, 2004). Transcriptomic analysis on E10.5 embryos exposed to matDM revealed upregulation of 20 genes regulated by Hif1α, suggesting that dysregulation in hypoxia signaling could underlie some of the cardiac malformations observed in diabetic embryopathy (Pavlinkova, Salbaum, & Kappen, 2009). However, haploinsufficiency of Hif1α in matDM-exposed embryos did not lead to an increased incidence of CHD when compared to wildtype littermates, and there was increased expression of Nkx2.5, Tbx5, Nppa, Cx43, and Mef2c (Bohuslavova et al., 2013). Of note, Hif1α^+/− offspring exposed to matDM developed left ventricular dysfunction at 12 weeks of age with decreased fractional shortening and structural remodeling of the myocardium (Cerychova et al., 2018). Differential gene expression analysis following transcriptional profiling revealed that 53% of differentially expressed genes were direct or predicted targets of Hif1α and were enriched in gene ontology categories of heart development, response to stress, ECM organization, apoptosis, cell proliferation and angiogenesis. These studies suggest that haploinsufficiency of Hif1α increases the predisposition of offspring exposed to matDM to cardiovascular disease, though not necessarily congenital cardiac malformations.

Other molecular pathways.

Transcriptomic profiling of matDM-exposed mouse hearts at different developmental stages revealed dysregulation of numerous genes that encode a variety of proteins which function as transcription factors, chromatin regulators, signal transduction components, cell surface receptors, extracellular matrix adhesion molecules, and cytoskeletal assembly factors (Kumar et al., 2007; Pavlinkova et al., 2009; Vijaya et al., 2013; Basu & Garg, 2018). Other implicated genes are involved in the regulation of RNA-binding, transport, metabolism, metal-ion homeostasis and cell cycle/apoptosis (Pavlinkova et al., 2009; Vijaya et al., 2013; Cerychova et al., 2018). Among this diverse group of genes, the dysregulation of molecular pathways specifically involved in endocardial cushion development and neural crest migration are consistently identified in matDM-exposed embryonic hearts by several investigative groups. Downregulation of Bmp4, Msx1 and Pax3 in diabetic hearts leads to impaired EMT and defects in endocardial cushion remodeling and neural crest migration that may result in conotruncal and cardiac septal defects (Phelan, Ito, & Loeken, 1997; Morgan et al., 2008; Vijaya et al., 2013). SOLid SAGE mRNA deep sequencing on the anterior half (developing heart and brain) of E8.5 and the thoracic segment (including the heart) of E9.5 diabetic and control also found Pax3 to be significantly downregulated in response to matDM (Zhao et al., 2016). MatDM disrupts Notch, Tgfβ and Wnt signaling, all of which are involved in EMT and endocardial cushion remodeling. Furthermore, in vitro culture of mouse endocardial cushion explants has revealed that hyperglycemia is sufficient to disrupt EMT and cell migration (Kumar et al., 2007). As many of the cardiac defects arising from matDM exposure appear to be derived from abnormal endocardial cushion formation and remodeling, it is imperative that future studies in animal models elucidate the precise mechanisms underlying the sensitivity of the endocardium and the endocardial cushions to hyperglycemic insults. MatDM also impairs left-right (L-R) axis formation during embryogenesis which often leads to CHDs associated with heterotaxy. In fact, E8.5 embryos exposed to severe hyperglycemia (>27 mmol/L) had high penetrance of heart looping defects and impaired expression of the L-R patterning gene, Pitx2c (Hachisuga et al., 2015). Upstream of this, hyperglycemia affects Notch signaling in the node leading to loss of asymmetric expression of Nodal in the LPM and disruption in subsequent L-R axis formation resulting in specific types of CHD including TGA and DORV. As these processes occur prior to the embryo’s switch to aerobic respiration, oxidative stress is less likely to be the cause of these defects and it is imperative to investigate molecular changes that occur earlier in development to elucidate the causal mechanism of matDM-induced CHD, as discussed above. Lastly, it should be noted that some of the cellular and molecular changes that are described above are analyzed at later embryonic stages (after E12.5), and therefore may be reflective of dying embryos as opposed to the maternal hyperglycemic environment.

4.2. Diet-based models of maternal type 2 diabetes mellitus

High Fat Diet

Although the majority of molecular findings discovered in matDM-associated CHD have been generated from STZ-treated diabetic rodents, there remains concern over the pleotropic effects of STZ and its overall teratogenicity. Additionally, insulin deficiency is a key characteristic of type 1 diabetes mellitus and these models do not give insight into mechanisms relevant to maternal type 2 diabetes, which is similarly associated with CHD. The high-fat diet (HFD) is a common strategy used to generate obesity models, and many of these animals recapitulate the pathological effects of human diabetes mellitus type 2 (Winzell & Ahrén, 2004). However, it is only in recent years that diet-induced diabetic models have been used to study the effect on cardiac development and CHD (Wu et al., 2015; Wu et al., 2016).

A high-fat diet (HFD) in rodents comprises 20% protein, 20% carbohydrate and 60% fat as opposed to 20% protein, 70% carbohydrates and 10% fat in normal rodent diet (Winzell & Ahrén, 2004). After 15 weeks on HFD, female mice exhibit characteristic features of type 2 diabetes mellitus including high fasting glucose levels, hyperinsulinemia, glucose intolerance and insulin resistance. At this point, female diabetic mice are mated with non-diabetic males to generate embryos exposed to matDM (Wu et al., 2016).

In mice maintained on a 15 week HFD prior to mating, CHD was found at an incidence of 17% (Wu et al., 2016). The two cardiac phenotypes observed were PTA and VSD. Interestingly, no CHD was observed when HFD was started at E0.5 and maintained throughout pregnancy. Additionally, when mice were fed HFD for only 4 weeks prior to breeding, no CHD was observed in the offspring (Schulkey et al., 2015). A gene-environment interaction has been reported between HFD and Cited2 where Cited2^+/− embryos exposed to maternal HFD had much more severe cardiac defects with left-right patterning defects compared to Cited2^+/− from control diets (Bentham et al., 2010). Interestingly, the dams were fed HFD for 8 weeks prior to breeding and though they were significantly heavier, they were only mildly hyperglycemic compared to controls fed on normal diets, making it difficult to detangle any interaction between matDM and Cited2. Taken together, it appears that the duration of HFD is critical to ensure that matDM-associated CHD is found in the offspring to recapitulate the human condition.

The molecular mechanisms underlying matDM type 2-associated CHD is not as well studied as the STZ model. Even in this limited pool of studies, oxidative stress appears to be the major manifestation. Increased ROS was found in E12.5 hearts exposed to matDM type 2 (Wu et al., 2016). The endoplasmic reticulum (ER) stress markers, CHOP, BiP, Calnexin, PDIA, GRP94, p-IRE1a, p-eIF2a, p-PERK, were elevated in embryonic hearts exposed to matDM type 2. Additionally, increased levels of apoptotic markers, caspase 8 and caspase 3, were found, and significantly higher numbers of apoptotic cells were found in the endocardial cushion, ventricular myocardium and OFT in developing hearts from matDM type 2 (Figure 2). Again, investigation of ROS levels during earlier stages of cardiac morphogenesis (before E12.5) is required prior to concluding that oxidative stress is the cause of matDM–induced CHD.

Figure 2: Fetal cellular and molecular pathways dysregulated in response to matDM.

Both STZ-induced matDM type 1 and High fat diet (HFD)-induced matDM type 2 leads to increased oxidative stress in the fetal hearts. In STZ-induced matDM type 1, oxidative stress is proposed to bring about dysregulation of cardiac transcription factors (TFs) and several molecular and cellular pathways including endoplasmic reticulum stress (ER), cellular apoptosis and proliferation, transforming growth factor – beta (TGFβ) signaling, Wnt signaling, Notch signaling, epigenetic changes and neural crest cell migration. In fetal hearts exposed to HFD induced matDM type 2, oxidative stress leads to ER stress and abnormal apoptosis and proliferation in the fetal hearts. Genetic models of matDM are not well studied but dysregulation in cardiomyocyte proliferation and lipid uptake has been identified in fetal hearts in offspring of Ins2^Akita/+. Created with Biorender.com.

While it has not been fully utilized, HFD-induced matDM type 2 is an excellent model for studying the effect of frank hyperglycemia on cardiac development, without the confounding effects of insulin-deficiency and STZ toxicity. A drawback is the requirement of at least 15 weeks of a HFD to establish the diabetic state prior to breeding with males, which increases study time significantly. Despite this, the model provides an important tool for comparing mechanisms underlying maternal diabetes type 1 and type 2-associated CHD, the latter becoming more prevalent in younger people worldwide (Imperatore et al., 2012).

5. OTHER MODELS OF MATERNAL DIABETES MELLITUS

Although the STZ-induced matDM type 1 and HFD induced matDM type 2 rodent models are the most extensively used, there are several other diabetic models that may be used to study the effect of matDM on cardiac development. Direct glucose infusion into the fetal blood circulation is an approach that has been used to mimic the effect of maternal gestational diabetes (Gordon et al., 2015). Sprague Dawley rats injected with high glucose gave birth to infants displaying overgrowth of the ventricular septum, a phenotype reminiscent of the cardiac hypertrophy observed in gestational diabetes (Gordon et al., 2015). These hearts had upregulated expression of mitogenic marker, pJNK, compared to non-HG exposed embryos, although no changes were observed in the expression level of other mitogens such as AKT, pERK or P38.

Genetic models for both diabetes mellitus type 1 and type 2 have been used to investigate complications from diabetic pregnancies but their use in studying the effect on cardiac development and concurrent dysregulation of molecular pathways is limited (Jawerbaum & White, 2010). Rodent strains that develop diabetes due to spontaneous autoimmune disorders leading to destruction of insulin producing B cells (NOD mice, BB rats, Cohen rats) are an established model of diabetes mellitus type 1. Approximately 50% of fetuses from non-obese diabetic (NOD) mice exhibit cardiovascular anomalies and 75% have heterotaxy of the heart and other organs (Otani, Tanaka, Tatewaki, Naora, & Yoneyama, 1991; Morishima, Yasui, Ando, Nakazawa, & Takao, 1996; Yoon & Jun, 2001). Biobreeding (BB) rats, outbred from Wistar rats, have severe hyperglycemia and require insulin therapy for survival. Embryos from these animals exhibit larger hearts compared to non-diabetic Wistar rat strains and withdrawal of insulin for 2 days at different gestational stages leads to severe congenital malformations including CHD, particularly when treatment is interrupted immediately before or during organogenesis (Eriksson, Bone, Turnbull, & Baird, 1989). Diabetes sensitive Cohen rats develop overt type 2 diabetes mellitus when fed a high sucrose-low copper diet (HSD), and 47% of their embryos lack a heartbeat with reduced total SOD1 levels compared to offspring from normal fed diets (Zangen, Yaffe, Shechtman, Zangen, & Ornoy, 2002).

In contrast, genetic models of maternal gestational diabetes have also been reported. Ins2^Akita/+ mice have an autosomal dominant mutation in one copy of the insulin2 gene that prevents normal processing of pro-insulin, leading to accumulation of misfolded protein and subsequent ER stress (King, 2012). Fetal hearts collected from E16-E19 embryos from Ins2^Akita have downregulated expression of genes that govern cardiac lipid uptake (PGC-1α, PPAR-α, LPL, FAT/CD36, and FATP1) and the expression of these genes correlated with the maternal blood glucose concentration, suggesting maternal diabetes affects metabolic utilization of fetal hearts (Lindegaard & Nielsen, 2008). In fact, PGC-1α is associated with cardiac hypertrophy, as observed in offspring exposed to gestational diabetes (Lehman & Kelly, 2002; Wu et al., 2020). Similarly, Lepr^db/+ mice, which harbor a heterozygous loss-of-function mutation in the leptin receptor gene, are fertile (unlike their homozygous counterparts) and develop hyperphagia during gestation, gain weight and show impaired insulin and glucose tolerance (Jawerbaum & White, 2010). Offspring of these mice exhibit macrosomia, though specific effects on cardiac morphology have not been investigated. Finally, the transgenic Tet29 rat model, in a Sprague-Dawley background, is a novel genetic model of hyperglycemia in which the insulin receptor is knocked down via RNA interference upon doxycycline (DOX) administration (Kotnik et al., 2009). Loss of the insulin receptor leads to hyperglycemia and hyperinsulinemia in a highly controlled manner. Fetal hearts from these mice have lower expression of cardiomyocyte marker, Tnnt2, and increased number of apoptotic cells; however, the adult male offspring do not exhibit any overt symptoms and the effect on cardiac function has not been explored (Hillier et al., 2007; Nakano et al., 2017; Schütte et al., 2021).

The primary limitation in using genetic models arises from the increased time needed to generate diabetic females. The spontaneous diabetic models can take as long as 30 weeks before showing signs of diabetes mellitus. This often results in severe disease that is not conducive to viable pregnancies. In contrast, the Ins2^Akita+/− and Lep^db/+ may require multiple rounds of breeding to obtain the correct genotype. Another concern is that the lack of CHD phenotypes from these models might represent a lower hyperglycemic state that does not recapitulate the human diabetic uterine environment. However, more studies focusing on cardiac development using these models must be done before any conclusions can be made about their efficacy to study CHD.

6. CONCLUSIONS AND FUTURE PERSPECTIVES

The pathophysiology of matDM-associated CHD is complex and clinically relevant animal models are in use to define the underlying mechanisms. In this review, we have described numerous molecular pathways that have been implicated in matDM-associated CHD, and current evidence points to the central role of oxidative stress signaling as an initiating factor. The majority of support for this finding come from the STZ-induced matDM type 1 model, though molecular changes mediated by oxidative stress has been confirmed in HFD based matDM type 2 model and in the offspring of diabetic Cohen rats. However, CHDs arising from impaired L-R patterning cannot be explained by oxidative stress as these processes occur in a hypoxic environment prior to the switch to oxidative phosphorylation. Additionally, matDM results in dysregulation of multiple signaling pathways critical for cardiac development, including known CHD causing genes in humans. Growing evidence also points to the role of epigenetic modifiers in mediating the transcriptional changes in response to the maternal diabetic milieu in the STZ-induced matDM type 1 model. In contrast, genetic models of diabetes mellitus have garnered limited attention in studying matDM-associated CHD, perhaps due to the longer time it takes for the animals to exhibit overt diabetes and inconsistent CHD phenotypes in the offspring. Regardless, at least two genetic models of spontaneous autoimmune disorder, the NOD mice and Cohen rats fed with HSD, have high incidence of CHD in their offspring and warrant additional investigation.

Although maternal hyperglycemia is proposed to be the primary teratogen in matDM-associated CHD, the developing fetus is exposed to a complex diabetic milieu that is challenging to standardize experimentally. Rodent models are vital for capturing a holistic picture of the effect of matDM on cardiac development, but in order to study the specific teratogenicity of hyperglycemia, it is important to incorporate ex utero and in vitro techniques that allow a higher degree of control over confounding variables. Recently, a study described a novel protocol for ex utero culture of pre-gastrulation mouse embryos (E5.5) until late-stage organogenesis (E11) (Aguilera-Castrejon et al., 2021). This constitutes a powerful tool that can be utilized to investigate the dose and time specific effects of hyperglycemia on cardiogenesis. Additionally, the advent of human embryonic stem cells (h-ESC) and human induced pluripotent stem cells (h-iPSC) provides new clinically relevant models for studying the effect of hyperglycemia on developing cardiac cell populations. In fact, studies have focused on differentiating h-ESC into cardiac cell types under high glucose conditions in order to identify molecular pathways that may be dysregulated during human development with exposure to matDM (Nakano et al., 2017). This provides an exciting avenue as h-iPSC derived from CHD patients exposed to maternal diabetes can be differentiated into a variety of cardiac cell types to investigate differences in molecular and epigenetic signatures compared to non-matDM-exposed iPSCs. In fact, a recent study described a novel method of generating a heart forming organoid using h-iPSC, recapitulating the first step of human cardiogenesis in vitro (Drakhlis et al., 2021). As this methodology is further optimized, it could provide a more pertinent model to study the effect of hyperglycemia on human heart development.

Furthermore, animal models have provided ample evidence that susceptibility to maternal diabetes induced congenital malformations can be determined by the offspring’s genome, providing a basis for the gene-environment interaction model of CHD. As described above, haploinsufficiency of key cardiac developmental genes such as Notch1^+/− increases the predisposition for CHD in mouse embryos when exposed to maternal diabetes. This can be translated into human CHD cases where the presence of an otherwise “harmless” gene variant is sensitized by maternal diabetes exposure, causing it to behave as a CHD causing gene. In other words, maternal hyperglycemia may be disrupting the dosage, rather than the function, of key developmental genes, bringing them below the threshold required for proper development. In fact, mouse models have shown that maternal diabetes can disrupt expression of several key cardiac regulatory genes that are also found to be mutated in individuals with human CHD (e.g NOTCH1, NKX2.5, GATA4). If this is the case, genetic screening of developing fetuses from diabetic mothers could be employed to allow for targeted therapy and intervention to bring down the incidence of CHD in this population.

Successful elucidation of the mechanisms by which maternal hyperglycemia disrupts cardiac developmental pathways is crucial to developing targeted interventions. With the advance of high-throughput sequencing technologies, it is now possible to investigate the effect of maternal diabetes on embryo development at the single cell level. The animal models described in this review are indispensable for gaining a complete understanding of matDM-associated CHD. Furthermore, these models will be invaluable when testing potential therapeutic strategies as we move into an era of multi-omics research.