Developmental programming of CVD risk by in utero exposure to PE

Developmental programming of chronic diseases is now an established paradigm. The link between early life environmental factors and later life disease susceptibility was initially proposed by David Barker who in early 1990s showed that restricted growth in fetal life was associated with mortality from CVD.^{Reference Barker18,Reference Barker19} This concept that exposure to unfavourable conditions in utero at times of critical organ development may lead to later life disease risk was originally called the Barker Hypothesis/fetal origins of disease hypothesis and now the Developmental Origins of Health and Disease hypothesis.^{Reference Barker20} While most of the initial work on this topic linked intrauterine undernutrition with later life CVD and metabolic disease risk, subsequent epidemiological studies have shown that numerous intrauterine exposures including major pregnancy complications (PE, intrauterine growth restriction [IUGR], spontaneous preterm birth and gestational diabetes mellitus), maternal obesity and smoking during pregnancy and exposure to environmental chemicals can each trigger propensity for a myriad of cardiovascular, metabolic, immunological, reproductive and neurodevelopmental disorders in the offspring.^{Reference Murphy, Cohn and Loria21}

Depending on the severity of the insult during critical periods of rapid growth and development, permanent tissue adjustments can occur which lead to long-term functional changes in vital organs. Although PE is a multifactorial disorder, several pathophysiological mechanisms including angiogenic imbalance, excessive inflammation, ischaemia/perfusion and imbalances in the renin angiotensin system are all implicated in its pathogenesis.^{Reference Andraweera, Dekker and Roberts22,Reference Stojanovska, Scherjon and Plosch23} During a pregnancy complicated by PE, the placenta releases circulating factors, including sFlt and sEng,^{Reference Staff24,Reference Redman and Sargent25} into the maternal circulation as a result of syncytiotrophoblast stress due to a variety of insults including placental ischemia/hypoxia, contributing to excessive inflammation and generation of reactive oxygen species.^{Reference Brennan, Morton and Davidge26-Reference Kao, Morton, Quon, Reyes, Lopez-Jaramillo and Davidge29} Consistent with a causal role for these placental factors in programming of progeny CVD, overexpression of sFlt in a mouse model of PE mimicking the placenta of a pregnancy affected by PE results in elevated systolic and diastolic blood pressure in male offspring.^{Reference Lu, Bytautiene and Tamayo30} Higher than normal levels of these placental circulating factors lead to placental oxidative stress, inflammation and lipid peroxidation^{Reference Aouache, Biquard, Vaiman and Miralles31,Reference Myatt, Rosenfield, Eis, Brockman, Greer and Lyall32} which can have permanent effects on the susceptibility of that fetus to CVD later in life. Increased oxidative stress and placental oxidative DNA damage are associated with fetal growth restriction in offspring from pregnancies complicated by PE,^{Reference Fujimaki, Watanabe, Mori, Kimura, Shinohara and Wakatsuki33} and there is evidence of oxidative stress and DNA damage in cord blood from offspring of PE-affected pregnancies.^{Reference Hilali, Kocyigit and Demir34} Exposure to hypoxia in a rat model of PE leads to impaired placental function, and offspring exhibit inflammation and atherosclerosis.^{Reference Wang, Huang, Lu, Lin and Ferrari35-Reference Zhang, Zhu and Chen37}

The impact of perinatal exposure to inflammation also has various impacts on the fetus, including changes to the fetal immune system^{Reference Santner-Nanan, Straubinger and Hsu38,Reference Hu, Eviston and Hsu39} (e.g. lower abundance of regulatory T [Treg] cells) which persist into early childhood.^{Reference Hu, Eviston and Hsu39} Exposure to intrauterine inflammation as a result of placental dysfunction may lead to increased inflammation and reduced capacity to dampen inflammation due to a deficiency in Treg cells in later life. Such immune programming may contribute to progeny CVD, since inflammation is central in the pathogenesis of atherosclerosis and CVD.^{Reference Hu, Eviston and Hsu39,Reference Nguyen Maria, Wallace Megan, Pepe, Menheniott, Moss Timothy and Burgner40}

Animal studies provide mechanistic evidence that each of these exposures programs offspring blood pressure. The strength of most of the animal models is that they avoid the confounding factors of shared genetics and shared lifestyles and focus specifically on in utero exposure. The models of PE in which programming of progeny cardiometabolic outcomes has been most extensively characterised are in rodents. These include the surgically induced reduced uterine perfusion pressure (RUPP) rat, and PE induced by maternal treatment during pregnancy with hypoxia, the nitric oxide synthase (NOS) inhibitor L-NAME or angiotensin II autoantibodies. Postnatal outcomes have also been well characterised in progeny of mice where elevated sFlt1 was achieved by adenoviral transfection of dams during pregnancy. In each of these models, progeny are lighter at birth but catch up in body weight to that of controls during or before adulthood.^{Reference Zhang, Zhu and Chen37,Reference Alexander41-Reference McDonnold, Tamayo and Kechichian51} Development of obesity is variable, with obesity reported in adult female, but not male, RUPP rats with ageing,^{Reference Intapad, Tull and Brown52,Reference Intapad, Dasinger, Fahling, Backstrom and Alexander53} but normal body and central fat abundance in adult progeny of sFlt1-transduced mice of both sexes.^{Reference Bytautiene, Tamayo and Kechichian54} Likewise, effects of in utero exposure to a preeclamptic phenotype on postnatal glucose tolerance are sex- and age-dependent in the RUPP rat,^{Reference Intapad, Dasinger, Fahling, Backstrom and Alexander53,Reference Intapad, Dasinger, Brown, Fahling, Esters and Alexander55} L-NAME-treated rat^{Reference Butruille, Mayeur and Moitrot56} and sFlt1-transduced mice^{Reference McDonnold, Tamayo and Kechichian51} models of PE. Progeny cardiac dysfunction is common to all of these models, although with some variation in phenotype and different sets of outcomes studied in each model. Hypertension at baseline and/or in response to stress or adrenaline has been reported in progeny of the RUPP rat,^{Reference Anderson, Lopez, Zimmer and Benoit10,Reference Alexander41-Reference Dasinger, Intapad, Backstrom, Carter and Alexander45,Reference Intapad, Tull and Brown52,Reference Payne, Alexander and Khalil57} hypoxic rat^{Reference Tang, Zhu and Xia47,Reference Liu, Liu and Shi48,Reference Peyronnet, Dalmaz and Ehrstrom58,Reference Wang, Li, Huang and Wang59} and in males but not females in the sFlt1-transduced mice model of PE.^{Reference Lu, Bytautiene and Tamayo30,Reference Bytautiene, Tamayo and Kechichian54} Interestingly, progeny of the L-NAME-treated rat are normotensive as young and mature adults.^{Reference Herraiz, Pellicer and Serra50,Reference Witlin, Gangula, Thompson and Yallampalli60} Greater arterial vasoconstriction in response to stimuli, greater renal sympathetic nerve activity, contributions from circulating sex steroids and impaired vasodilation are all implicated as mechanisms for hypertension in progeny of the RUPP rat.^{Reference Anderson, Lopez, Zimmer and Benoit10,Reference Alexander, Hendon, Ferril and Dwyer42,Reference Ojeda, Grigore and Yanes43,Reference Intapad, Tull and Brown52,Reference Payne, Alexander and Khalil57} Similarly, in progeny of hypoxic rat dams, phenylephrine-induced vasoconstriction is greater, and both endothelium-dependent and endothelium-independent vasodilatation are reduced compared to control progeny.^{Reference Giussani, Camm and Niu36,Reference Tang, Zhu and Xia47,Reference Liu, Liu and Shi48,Reference Allison, Kaandorp and Kane61,Reference Tang, Li and Chen62}

Progeny heart structure and function are also impaired in several of these models. For example, cardiac and cardiomyocyte size are normal at 3 months of age in young male adults whose mothers were exposed to intermittent hypoxia from early pregnancy (10% O₂ for 3 h/day from GD7 to 21)^{Reference Wang, Li, Huang and Wang59}. At 5 months of age, hypertensive adult male progeny from these dams exhibit cardiac hypertrophy, with normal cardiomyocyte size but greater left ventricular collagen expression of collagen I and III.^{Reference Wang, Li, Huang and Wang59} Furthermore, lesions suggestive of developing atherosclerosis are present in the aortas of adult male progeny at 5 months of age of dams housed continuously in 10.5% O₂ from GD5-delivery,^{Reference Zhang, Zhu and Chen37} and in aged adult male and female progeny at 16 months of age who were exposed to hypoxia in utero 10% O₂ for 3 h/day from GD7 to 21.^{Reference Wang, Huang, Lu, Lin and Ferrari35} Although their relative heart weights are normal, male progeny of dams treated with angiotensin-II-like auto-antibodies (which bind and activate the angiotensin II-type I receptor) have enlarged cardiomyocytes, with areas of disorganisation, necrotis and apoptosis evident in their left ventricles as young adults at 5 months postnatal age.^{Reference Jin, Zhang and Yang49} The function of isolated hearts from these progeny in a Langendorff apparatus is normal under basal conditions, but they have impaired recovery and larger infarcts following ischaemia-reperfusion.^{Reference Jin, Zhang and Yang49} Cardiac hypertrophy is also evident throughout the first month of postnatal life in progeny of rats treated with L-NAME before and throughout pregnancy.^{Reference Martini and Mandarim-de-Lacerda63}

The fact that hypertension and impaired metabolism occur in progeny in these models of induced PE provides direct evidence of causality for the exposure. Nevertheless, these models lack the initiating processes that lead to spontaneous PE in humans. Interpretation of the programming mechanisms in progeny exposed to PE in utero and evaluation of potential maternal or postnatal interventions also need to consider the stage of cardiovascular development when in utero exposure occurs. For example, the switch from cardiomyocyte proliferation to hypertrophy occurs postnatally in rodents but before birth in other species including humans.^{Reference Lock, Tellam and Botting64} A recent report that progeny of preeclamptic baboons exhibit elevated systolic blood pressure during a dietary salt loading challenge^{Reference Yeung, Sunderland and Lind65} provides direct evidence of cardiovascular programming by in utero exposure to PE in species that are more mature at birth, although studies with greater animal numbers are needed to assess any sex differences.

When combined with epidemiological evidence from human studies, these suggest several plausible mechanisms for the hypertensive phenotype seen in offspring of preeclamptic pregnancies.^{Reference Davis, Newton and Lewandowski11} Several studies have shown reduced numbers of nephrons and cardiomyocytes in offspring born with low birthweight or born preterm, two pregnancy complications that often co-exist with PE.^{Reference Luyckx, Bertram and Brenner66} A reduction in the number of nephrons contributes to a reduction in the rate of renal ultrafiltration that affects the circulating blood volume leading to increased blood pressure.^{Reference Keller, Zimmer, Mall, Ritz and Amann67} Altered nephron number is associated with glomerular hypertrophy and reduced renal vascular dilation contributing to risk of hypertension.^{Reference Richter, Briffa, Moritz, Wlodek and Hryciw68,Reference Marchand and Langley-Evans69} Alterations in the renin–angiotensin–aldosterone system (RAAS) including upregulation or downregulation of specific receptors within the kidney, blood vessels and the placenta have also been implicated in the development of renal dysfunction.^{Reference Moritz, Cuffe and Wilson70} Emerging evidence suggests that epigenetic processes which are strongly influenced by the environment may play a key role in developmental programming of adult CVD.

The role of epigenetics in the relationship between PE and increased risk of CVD in offspring

Epigenetics refers to modifications that result in gene expression changes, often through chromatin remodelling, without altering the underlying DNA sequence^{Reference Berdasco and Esteller71} These modifications include DNA methylation, histone modifications, non-coding RNA (ncRNA) and other modifications that can result in altered gene expression.^{Reference Berdasco and Esteller71} Epigenetic changes have been widely studied in developmental processes, especially their role in determining cell lineages, and also in disease states as they can mediate aberrant gene expression.^{Reference Berdasco and Esteller71} In addition, epigenetic modifications are routinely used for diagnostic and prognostic biomarkers of adverse outcomes. Epigenetic modifications are also altered in response to environmental exposures and can potentially indicate whether an adverse exposure early in life is associated with future chronic disease risk.^{Reference Berdasco and Esteller71}

The most widely studied epigenetic alteration is DNA methylation which is the addition of a methyl group to a cytosine catalysed by enzymes known as DNA methyltransferases. DNA methylation is involved in development, imprinting and X chromosome inactivation, as well as numerous other processes.^{Reference Berdasco and Esteller71} Hypermethylation at gene regulatory regions or promoter associated CpG islands is often associated with gene repression. Histone modifications refer to changes to the histone tails including acetylation, methylation or phosphorylation and together with DNA methylation impact chromatin packing and hence accessibility of transcription factors to the regulatory DNA sequences controlling gene expression. ncRNAs consist of short ncRNAs like microRNAs, piwiRNAs, snoRNAs or long non-coding RNAs (lncRNA) and can inhibit transcription and translation or alter protein trafficking and folding to name just a few of their actions.^{Reference Karlsson and Baccarelli72}

Epigenetic changes are considered likely mechanisms that link maternal PE with offspring CVD with the possibility of passing on the risk to subsequent generations, by germ-line inheritance of epigenetic marks.^{Reference Heindel73,Reference Hanson and Skinner74} Animal models, which allow for tight regulation and control of the adverse exposure, have repeatedly shown that adverse events in utero are associated with increased risk of chronic disease in offspring.^{Reference Heindel73,Reference Goyal, Limesand and Goyal75,Reference Hanson76} Similar studies in humans are difficult, not only due to the lack of control of ‘other’ life-time environmental exposures but also due to genetic predispositions and additionally access to the appropriate tissue for assessment. However, epidemiological studies do show evidence for changes in DNA methylation as potential mediators or biomarkers of in utero exposure to PE when assessed in the placenta^{Reference Apicella, Ruano, Mehats, Miralles and Vaiman77,Reference Bianco-Miotto, Mayne, Buckberry, Breen, Rodriguez Lopez and Roberts78} and cord blood.^{Reference Kazmi, Sharp and Reese79} Whether these epigenetic changes are associated with an increased risk of CVD in the offspring later in life is unknown as no studies have yet reported epigenetic alterations in the adult offspring born to women who experience PE. This is not surprising due to the length of follow-up required for these studies and also access to the right tissue sample for the epigenetic analyses. However, there are a few studies that demonstrate an association between PE exposure and epigenetic changes in offspring. In a 2015 study by Julian et al., DNA methylation in blood from young adult men born following hypertensive pregnancies (n = 5) compared to controls (n = 19) showed differential methylation in the genes SMOC2, ARID1B and CTRHC1.^{Reference Julian, Pedersen and Salmon80} Murray et al. reported that DNA methylation in cord blood leukocytes of ANRIL predicted pulse wave velocity and heart rate in the child at 9 years of age.^{Reference Murray, Bryant and Titcombe81} Therefore, with the appropriate longitudinal cohorts, in future, we may be able to determine whether epigenetic change induced by in utero exposure to PE underlies increased risk of future CVD risk in the offspring.

The placenta may mediate this future CVD risk in the offspring as several studies have shown altered DNA methylation in placentas from women who experience PE compared to healthy pregnancies (reviewed in ref.^{Reference Apicella, Ruano, Mehats, Miralles and Vaiman77}). The genes shown to have aberrant DNA methylation include those with a role in trophoblast proliferation, differentiation and invasion, as well as genes involved in vascular function and immunological processes.^{Reference Apicella, Ruano, Mehats, Miralles and Vaiman77} Epigenetic alterations in these critical genes may impact the function of the placenta during pregnancy resulting in an adverse environment for the fetus.^{Reference Tarrade, Panchenko, Junien and Gabory82} The fetus adapts to this adverse environment, resulting in epigenetic alterations that mediate future CVD risk. More work is required to define how PE impacts the intrauterine environment, what epigenetic alterations this induces in the fetus and if these epigenetic alterations are still evident in adult offspring. Dissecting out the epigenetic mechanism is made difficult in humans by the need for longitudinal cohorts as well as determining which tissue to assess.

The role of shared genetics in the relationship between PE and increased risk of CVD in offspring

The link between PE and CVD in offspring could also be explained by shared genetic factors that predispose individuals to vascular diseases that manifest at different time points during the life course.^{Reference Andraweera, Dekker, Arstall, Bianco-Miotto and Roberts83} The incidence of PE is higher among women who were born of a pregnancy complicated by PE,^{Reference Lie, Rasmussen, Brunborg, Gjessing, Lie-Nielsen and Irgens84} while the risk of fathering a pregnancy complicated by PE is also higher among men who previously fathered a pregnancy complicated by PE with a different partner^{Reference Lie, Rasmussen, Brunborg, Gjessing, Lie-Nielsen and Irgens84} and among men who were themselves the product of a pregnancy complicated by PE,^{Reference Esplin, Fausett and Fraser85} all suggesting that PE is at least in part heritable. Further evidence for a genetic mechanism is provided by studies that show that genetic variants that are known risk factors for CVD are more prevalent among women who experience PE, men who father pregnancies complicated by PE and among infants born of pregnancies complicated by PE compared to unaffected populations.^{Reference Andraweera, Dekker, Thompson and Roberts86-Reference Andraweera, Dekker, Thompson, North, McCowan and Roberts89} Genetics, therefore, may act as a confounding factor in the association between maternal PE and offspring CVD. The underlying mechanism for some offspring who experience CVD in adult life after exposure to PE in utero may not be the effect of PE but genetic susceptibility to CVD, and these individuals may experience CVD even if they had not been exposed to PE in utero. This theory is supported by the findings of genetic variants that are common in both PE and CVD from a study on genetic dissection of the 2q22 PE susceptibility locus.^{Reference Sitras, Fenton and Acharya90} This study identified four independent single nucleotide polymorphisms (SNPs) within four genes that were associated with PE in an Australian family cohort: lactase (LCT, rs2322659), low-density lipoprotein receptor-related protein 1B (LRP1B, rs35821928), rho family GTPase 3 (RND3, rs115015150) and grancalcin (GCA, rs17783344).^{Reference Johnson, Fitzpatrick and Dyer91} These four SNPs were also associated with cardiovascular risk factors in an independent cohort of Mexican American families.^{Reference Johnson, Fitzpatrick and Dyer91} An attempt to confirm the above findings in an independent Australian population-based cohort of mothers and their adolescents showed nominal associations with cardiovascular risk factors (weight, blood glucose and triglycerides) for all four SNPs.^{Reference Loset, Johnson and Melton92} These findings suggest the possibility that certain genetic variants may predispose individuals to different vascular diseases, with some women developing PE during pregnancy, some men fathering pregnancies complicated by PE and some developing CVD.

The role of shared environment and lifestyle in the relationship between PE and increased risk of CVD in offspring

Similar to shared alleles, shared lifestyle factors between mother and offspring may also confound the association between maternal PE and offspring CVD. It is well known that suboptimal environmental and lifestyle conditions including unhealthy diets, lack of exercise, low socioeconomic status and low educational levels are all associated with both PE and CVD. Since these factors are very likely to be shared between mothers and their children, they may underlie the association between maternal PE and CVD in offspring in a manner similar to the potential role of similar genes. These unfavourable lifestyle factors are also proposed to act as ‘second hits’ for some individuals. Some who are ‘programmed’ for increased disease risk only develop disease when exposed to a ‘second hit’^{Reference Cheong, Wlodek, Moritz and Cuffe93} at a later stage in life. These ‘second hits’ are often adverse lifestyle factors including smoking, unhealthy diets and sedentary lifestyles,^{Reference Cheong, Wlodek, Moritz and Cuffe93} so shared adverse environment may contribute to CVD risk both directly and by providing a ‘second hit’ exposure to individuals who are susceptible as a consequence of their intrauterine exposures.

Maternal PE and offspring CVD: developmental programming, shared genes or shared lifestyle?

If the association between maternal PE and offspring CVD was solely due to developmental programming, adverse cardiovascular profiles would only be seen among those exposed to PE in utero and siblings born of normotensive pregnancies will be unaffected. On the other hand, if the association between maternal PE and offspring CVD is due to shared genetics and/or lifestyles, then offspring of the same parents from pregnancies unaffected by PE would demonstrate similar CVD risk to that of offspring of pregnancies complicated by PE. The HUNT study comprising three large population-based surveys (HUNT 1, n = 77,212; HUNT2, n = 65,215; HUNT 3, n = 50,807) conducted in Norway explored these questions by comparing CVD risk factors among siblings discordant for the exposure to hypertensive disease in pregnancy (n = 472 participants within 210 sibships).^{Reference Alsnes, Vatten and Fraser94} In this large study, offspring exposed in utero to maternal hypertension including PE and their siblings born after a normotensive pregnancy had similar adverse CVD risk profiles, suggesting that shared genes or lifestyle may account for the association, rather than an intrauterine effect. However, other studies suggest that the increased CVD risk in the offspring could be a long-term consequence of fetal exposure to PE. In support of this theory, offspring who were born after a pregnancy complicated by PE display marked vascular dysfunction (higher pulmonary artery pressure and lower flow-mediated dilation), but their siblings born after a normotensive pregnancy display normal vascular function.^{Reference Jayet, Rimoldi and Stuber13} In this study, the birthweight of offspring exposed to PE in utero was approximately 400 g lower than the control group. This suggests the possibility of an alternative explanation. Birthweight correlates linearly with nephron number in adults and children, with nephron number increasing by 257,426 per kg increase in birthweight^{Reference Hughson, Farris, Douglas-Denton, Hoy and Bertram95} and low nephron number being associated with higher blood pressure (reviewed in ref.^{Reference Luyckx, Bertram and Brenner66}). Effects of PE exposure in utero on postnatal blood pressure may therefore be in part mediated by effects of fetal growth restriction on renal function. Interpretation of associations between exposure to PE in utero and later life CVD becomes more complicated by the fact that offspring of pregnancies complicated by PE can also be born with low birthweight and prematurity. Low birthweight and prematurity are both associated with a congenital reduction in nephron number, raised blood pressure, proteinuria and chronic kidney disease in adulthood.^{Reference Luyckx, Bertram and Brenner66} Therefore, developmental programming as a consequence of these conditions may confound the associations between PE and offspring CVD risk. Low birthweight and prematurity both suggest exposure to a more severe placental disease and are mostly seen in early onset PE. Due to the interactions between genes, environment and potentially programming mechanisms, understanding the relative contribution of each of these mechanisms to the link between PE and CVD is complex and a combination of all three mechanisms appears the most plausible explanation.