Prenatal Steroids and Metabolic Dysfunction: Lessons from Sheep.

Environmental exposure to EDCs during critical periods of organ development has been postulated as an important contributor to the increased incidence of metabolic diseases in the last decades ( 100 , 101 ). One such EDC is BPA, a carbon-based compound that is widely used in the manufacture of epoxy resins and polycarbonate plastics. Studies have reported the ubiquitous presence of BPA in water, air, and dust, as well as in most human fluids, including the maternal circulation and placental fluids ( 102 – 104 ). BPA has estrogenic and antiandrogenic activities and has the potential to program the development of obesity, insulin resistance, and cardiovascular diseases ( 101 , 105 ). Prenatal BPA treatment from GD 30 to 90 (the same window used in studies with native steroids) resulted in lower birth weight without altering gestational length of female offspring ( 106 ). The effects on body weight were associated with differences in height and chest circumference of the newborn female lambs. Although changes in the IGF1 system appear to play a role in the IUGR seen in prenatal T-treated sheep ( 24 , 31 ), circulating concentrations of free IGF1 or insulin were not altered in the prenatal BPA-treated offspring ( 106 ). In spite of this, as in prenatal T-treated sheep, growth rate tended to be higher between two and four months of age in prenatal BPA-treated sheep, indicative of compensatory growth during early postnatal development ( 106 ). A dose-response study (0.05, 0.5, or 5 mg/kg/day) focusing on the offspring’s metabolic health revealed that prenatal exposure to BPA at levels similar to those found in humans reduced insulin sensitivity in the female offspring during prepubertal age ( 107 ). The changes in glucose-insulin homeostasis were characterized by an increase in fasting glucose concentrations and a reduction in the insulin sensitivity index during an intravenous glucose tolerance test ( 107 ). During the postpubertal period (13 months of age), perturbations in insulin sensitivity were even more evident in these prenatal BPA-treated females, manifesting as increased mean cumulative insulin response and elevated cumulative insulin/glucose ratio ( 107 ). Acute insulin response to glucose administration was also higher in all BPA groups compared with control females. Although the longitudinal characterization of insulin sensitivity in this model is limited, these findings suggest that insulin resistance is established early in life (prepubertally) in prenatal BPA-treated sheep and deteriorates further as animals approach adulthood. In addition, prenatal BPA treatment negated the effects of postnatal overfeeding on insulin sensitivity, suggesting a complex interaction between prenatal BPA exposure and postnatal diet. These observations in the ovine model are of important translational relevance, considering that the BPA doses used in these studies yielded circulating levels similar to those reported in humans, and are corroborated by findings in rodent models ( 49 ) that also demonstrate that prenatal BPA exposure is associated with adverse metabolic outcomes. Perturbations in adipose tissue distribution, adipocyte morphometry, and adipose tissue inflammation and oxidative stress contribute to the development of adverse metabolic outcomes, such as obesity and insulin resistance, during adult life. Additionally, environmental insults occurring postnatally (second-hit) have the potential to exacerbate a preexisting condition caused by the first insult during prenatal development (first-hit). Thus, conditions such as postnatal excessive weight gain and/or exposure to EDCs can amplify or unmask adverse outcomes programmed prenatally ( 108 , 109 ). When subjected to postnatal overfeeding, female sheep prenatally exposed to BPA manifested an increased ratio of visceral to subcutaneous fat compared with overfed control females, whereas no BPA-specific effects were observed in maintenance-fed animals ( 107 ). Given the association of visceral adipose tissue and negative metabolic outcomes ( 110 ), the extent to which increased visceral adiposity would ultimately amplify the severity of insulin resistance in prenatal BPA-treated, postnatally overfed animals is unclear. Similarly, although the adipocyte size distribution was shifted toward a larger area and diameter in the visceral adipose tissue of control overfed and BPA-treated sheep, highlighting the obesogenic potential of BPA ( 107 ), overfeeding did not exacerbate the effects of prenatal BPA on adipocyte size. In the subcutaneous adipose tissue, prenatal BPA exposure also increased the expression of CD68, suggesting increased macrophage infiltration ( 107 ). Collectively, these findings in the female sheep not only demonstrate a risk of metabolic perturbations during postnatal life following prenatal exposure to BPA at human-relevant levels but also highlight the complicated interaction that prevails between prenatal BPA and postnatal overfeeding/obesity. Prenatal BPA treatment was also found to induce oxidative stress ( 111 ), a forerunner of adult metabolic disease ( 112 ). Using oxidized tyrosine moieties to study pathways involved in oxidative stress, Veiga-Lopez et al. ( 111 ) found that prenatal BPA treatment increases 3-nitrotyrosine (a marker for nitrosative stress) in GD-90 female fetuses, whereas a positive trend was observed at GD 65. Moreover, 3-nitrotyrosine was also elevated in both plasma and adipose tissue in prenatal BPA-treated sheep during adult life, demonstrating long-term effects of prenatal BPA exposure ( 111 ). This may also reflect the cumulative effects of BPA throughout the window of exposure (GD 30 to GD 90), because BPA has been found to accumulate in adipose tissue ( 113 ). Altogether, these observations in the ovine model demonstrating the induction of systemic and tissue-level nitrosative stress support a causal link between prenatal BPA and postnatal oxidative stress, because associations between BPA exposure and nitrosative stress have also been observed in human studies ( 111 ). Our studies addressing the interaction of prenatal BPA with postnatal overfeeding on the cardiovascular system in sheep found prenatal BPA had no effect on blood pressure or morphometric measures but increased the expression of atrial natriuretic peptide, a marker of left-ventricular hypertrophy frequently associated with systemic hypertension, suggestive of the impending risk of adverse left-ventricular programming by BPA ( 114 ). Moreover, prenatal BPA exposure reduced collagen expression in the right ventricle, which could result in altered systolic function ( 114 ). Interestingly, whereas postnatal overfeeding resulted in a marked increase in diastolic blood pressure, prenatal BPA treatment prevented the increase in blood pressure and increase in left-ventricular surface area caused by overfeeding, indicative of complex interactions in the manifestation of prenatally programmed events relative to the postnatal environment ( 114 ). Collectively, these observations suggest that prenatal exposure to BPA alone, while having no evident effect on cardiac structure, may induce changes at the molecular level that could reduce compliance and prevent compensatory responses to postnatal overfeeding in sheep ( 114 ). Assessment of transcriptional changes by RNA-Seq analysis of the myocardium to better understand the paradoxical interaction between prenatal BPA and postnatal overfeeding revealed transcriptional dysregulations in both the prenatal BPA and postnatal overfed groups. Although the biological pathways they affected were remarkably similar, the genes that were affected in these two groups differed considerably ( 115 ). Furthermore, similar to the effects on cardiovascular function, the combined treatment failed to produce a synergistic response; on the contrary, prenatal BPA exposure partially prevented the effects of postnatal overfeeding ( 115 ). Additional studies addressing whether these unexpected observations are adaptive or maladaptive are needed to help us better understand the interactive effects of prenatal BPA and postnatal obesity on adult cardiovascular function. Public pressure has led to a shift away from BPA in several consumer products, resulting in a marked increase in the use of other bisphenolic chemicals, such as bisphenol S (BPS) ( 116 ), another environmental steroid mimic recently demonstrated to act as an obesogen ( 117 ). Of concern, BPS has been detected in 70% of human urine samples ( 118 , 119 ) and fetal cord blood ( 120 ). Recent studies in sheep have shown that mid-gestation (GD 30–100) exposure to BPS reduced maternal levels of pregnancy-associated glycoproteins ( 121 ) secreted by the trophoblast layer of the placenta and considered to be biomarkers of pregnancy health in ruminants ( 122 ). Gestational BPS treatment also reduced the maternal progesterone, supportive of impaired placental endocrine function ( 121 ). Importantly, BPS was shown to cross the ovine placental barrier and reach the fetus in high concentrations ( 121 ). At the fetal level, gestational BPS treatment altered gene expression of steroid receptors in preadipocytes and impaired terminal differentiation of adipocytes in male but not female fetuses ( 123 ), indicative of sex-specific effects of gestational BPS exposure. Although these findings demonstrate important effects of BPS on placental and fetal development, in the absence of information on the postnatal repercussions of gestational exposure to BPS, a key requirement for determining if these adaptive changes culminate in adult pathology, the adult metabolic risks associated with these early developmental changes induced by prenatal BPS exposure remain unclear. Methoxychlor (MXC) is a synthetic organochlorine with estrogenic and antiandrogenic activity that was used as a pesticide in the United States until 2003. Although its use has been banned for over a decade, MXC is a persistent chemical that still exists in the environment and in the circulation of men and women ( 124 ). Mid-gestational MXC treatment resulted in high concentrations in the fetus, indicating it crosses the placental barrier ( 106 ). Although it did not impair fetal growth, gestational MXC treatment had long-term effects on several reproductive endocrine parameters in the female sheep ( 106 ). Considering MXC impairs metabolic function in rodents ( 125 ), it will be important in the future to characterize the metabolic outcomes associated with gestational MXC exposure in the ovine model.

Although prenatal exposure to excess androgens, glucocorticoids, and EDCs with steroidogenic activity can disrupt the developmental trajectory of the fetus, culminating in adult disease, there are multiple similarities in the outcomes programmed by these different prenatal insults ( Table 1 ). For example, changes in placental function and reduced fetal growth are common consequences of prenatal T ( 32 , 36 ) and glucocorticoid excess in sheep ( 75 ). Moreover, cardiometabolic derangements, such as insulin resistance, increased adiposity, and hypertension, are mutual phenotypic traits programmed by prenatal exposure to different steroid hormones and steroid mimics. The similarity in these phenotypic outcomes suggests that common mechanisms may be involved. Depending on the steroid hormone/mimic studied, organ system affected, and window of exposure, several mechanisms have been proposed. It is becoming increasingly evident that distinct prenatal insults can induce fetomaternal responses that converge on mutual mechanistic pathways to reprogram the adult phenotype of the offspring ( Figure 3 ) ( 3 ). Steroid hormones such as androgens, estrogens, and glucocorticoids, as well as metabolic hormones, such as insulin and leptin, regulate several key physiological processes during fetal development and are potential targets of perinatal insults ( 126 – 130 ). Animal studies have shown that different prenatal insults, such as maternal undernutrition ( 2 , 131 ), a condition that increases glucocorticoid levels, as well as gestational exposure to T excess ( 25 ), increase maternal insulin and leptin concentrations. Moreover, both gestational undernutrition and stress increase not only maternal glucocorticoids but also maternal androgens ( 132 – 134 ). Similarly, prenatal T treatment increases not only androgen levels in the fetomaternal circulation but also maternal insulin and fetal estrogen levels ( 25 ). Studies employing cotreatment with T and androgen antagonists or insulin sensitizers suggest that the programming effects of T are mediated via the androgen receptor and insulin signaling, pointing to the potential involvement of other pathways, such as conversion of T into estradiol and activation of estrogen receptors. Findings that prenatal exposure to BPA, an EDC with estrogenic activity, replicates several phenotypic alterations seen in prenatal T-treated sheep provide evidence for fetal programming via estrogenic pathways. Therefore, endocrine alterations (e.g., hyperandrogenism and hyperinsulinemia) during gestation may serve as common mechanisms by which different prenatal insults can program similar adverse adult outcomes in the offspring. Epigenetic modifications, such as DNA methylation, histone modification, and differential expression of noncoding RNAs, are emerging as important mechanisms of fetal programming ( 135 ). Because epigenetic modifications are inherently plastic and can accumulate over time, prenatal insults can have marked effects on this process, resulting in altered expression of several genes controlling the normal trajectory of fetal development ( 136 , 137 ). For instance, prenatal dexamethasone treatment alters the DNA methylome of the hypothalamus, liver, and adrenal glands in rodents ( 138 – 140 ). Similarly, prenatal T excess alters the DNA methylation patterns in both infant and adult visceral adipose tissues in female nonhuman primates ( 141 ). Notably, epigenetic alterations appear to be important not only in programming the phenotype of the offspring but also in mediating transgenerational inheritance of traits ( 142 ). Oxidative stress, a deleterious process that leads to damage of proteins, lipids, and DNA, is another important mechanism believed to underlie fetal programming ( 143 ). Gestational insults associated with IUGR, low birth weight, and adult disease, such as prenatal T and glucocorticoid excess, increase oxidative stress ( 47 , 144 ). Prenatal BPA treatment, similar to native steroids, impairs fetal growth and promotes systemic and tissue-specific oxidative stress ( 111 ). Moreover, IUGR has been shown to promote low-grade inflammation in the adult offspring, which may serve as a potential mediator linking reduced fetal growth and the development of chronic disease ( 145 ). Therefore, both oxidative stress and inflammation have been postulated as important mechanisms underlying adverse cardiometabolic outcomes associated with prenatal steroid excess and EDC exposures ( 146 ). The observation that maternal antioxidant supplementation decreases oxidative stress and improves metabolic function in the offspring of a rodent model of gestational obesity further indicates that inflammation and oxidative stress are likely contributors to the programming of adverse metabolic outcomes ( 147 ). Finally, early postnatal perturbations resulting from in utero exposure to steroid excess, such as endocrine imbalances and metabolic disruptions, themselves may play a role in unmasking and/or amplifying the defects programmed prenatally. For instance, postnatal excessive weight gain exacerbates the metabolic derangements programmed by prenatal T excess ( 38 ). Furthermore, postnatal treatment with an insulin sensitizer not only restores insulin sensitivity but also improves other metabolic parameters in adult sheep prenatally exposed to T excess ( 47 ), suggesting that postnatal imbalances in glucose-insulin homeostasis are required for the full manifestation of the disease phenotype. Thus, it is apparent that although prenatal exposure to steroid excess has important organizational effects, postnatal perturbations in endocrine and metabolic function have activational effects impacting the development, manifestation, and/or severity of the disease phenotype in the offspring ( 3 , 14 ). Epidemiological data and studies in animal models, including the ones in the sheep models discussed above, demonstrate that in utero exposure to steroid excess or steroid mimics can impact the offspring’s cardiometabolic health. However, the scarce information on the effects of such insults on the fetomaternal environment and limited longitudinal phenotyping studies hinder our understanding of the windows of susceptibility and the potential short- and long-term effects on relevant organ systems. Moreover, the interaction between prenatal steroid excess and resulting early postnatal perturbations on the development of cardiometabolic diseases remains poorly understood. Thus, the sheep model represents a valuable tool to investigate the long-term effects of prenatal exposure to native steroids and EDCs on the offspring’s health, the cellular and molecular mechanisms involved, and testing of novel intervention strategies. Animal models with a developmental trajectory similar to humans, such as sheep, are also important research tools to investigate the transmission of phenotypic traits to subsequent generations. Studies in different animal models have shown that prenatal exposure to steroid excess results in multigenerational modifications in the epigenome and cardiometabolic phenotype ( 91 , 148 – 150 ). Future studies should focus on distinguishing the role of epigenetic inheritance from the vertical transmission of endocrine perturbations, such as hyperandrogenism and hyperinsulinemia, which may facilitate perpetuation of pathologic outcomes to subsequent generations. Because prenatal steroid excess can impact multiple physiological systems, future research should also gain a better understanding of the interplay between these systems in the development of metabolic dysfunctions. For example, prenatal T excess impacts both the reproductive and metabolic systems, and reproductive perturbations (e.g., functional hyperandrogenism) impair metabolic function, whereas metabolic imbalances (e.g., insulin resistance and hyperinsulinemia) can impact reproductive function, thus forming a vicious cycle ( 3 ). Consequently, interventions targeting multiple organ systems may be needed to prevent the manifestation of adverse outcomes programmed prenatally. Observations in women with PCOS support this premise, in which combined treatment with an androgen antagonist and an insulin sensitizer yields additive benefits in improving metabolic function compared with monotherapies ( 151 ). Finally, the identification of common mediators activated by prenatal exposure to different steroids and/or EDCs in models such as sheep can help identify early biomarkers and potential therapeutic targets for intervention.

The developing fetus is particularly vulnerable to changes in the intrauterine environment during prenatal life. Although a few of these changes may be favorable for the healthy development and survival of the fetus, most of the physiological adaptations occurring during prenatal development in response to insults are detrimental and contribute to the development of adverse health outcomes postnatally. A myriad of animal and epidemiological studies have demonstrated that maternal exposure to adverse conditions during gestation owing to nutritional deficits/excess, disease conditions, stress, drug abuse, and/or environmental exposure to endocrine-disrupting chemicals (EDCs) can alter the uterine environment, culminating in the development of adult disease in the offspring ( 1 – 4 ). Therefore, improper lifestyle choices made by the mother during pregnancy pose a threat to the normal development of the fetus ( Figure 1 ). Barker’s ( 5 ) developmental origins of health and disease hypothesis has gained considerable attention during the past decades, especially after epidemiological data from the 1944–1945 Dutch famine demonstrated that gestational malnutrition is associated with a significant increase in the incidence of cardiovascular and metabolic diseases in the offspring ( 6 ). These findings, corroborated by subsequent preclinical and clinical studies ( 5 , 7 , 8 ), indicate that the prenatal period, a period in which organogenesis and tissue differentiation occur through a tightly controlled and timed process, is a critical susceptibility window for programming the offspring’s phenotype. More recently, it has become clear that the mechanisms underlying the fetal origins of adult disease involve reprogramming of the epigenome by environmental factors ( 9 , 10 ). This reprogramming is possible during prenatal life owing to the plasticity that allows the developing fetus to adopt a phenotype that best suits the environment. In the context of reprogramming, steroid hormones are major players, as they are involved in modulating normal cell differentiation and organ system development during fetal life ( 11 – 13 ). As such, improper exposure to steroids or steroid mimics during critical periods of development can disrupt the normal developmental trajectory of organ systems, culminating in adult disease. The reproductive and metabolic systems are particularly susceptible to the deleterious effects of steroid excess. The growing fetus can be exposed to inappropriate levels of steroid hormones owing to disease conditions, such as congenital adrenal hyperplasia and polycystic ovary syndrome (PCOS); continued maternal use of contraceptive steroids; maternal use of anabolic steroids; and environmental exposure to EDCs with steroidogenic activity, such as bisphenol A (BPA), diethylstilbestrol, and trenbolone ( 13 – 16 ). The recent increased incidence of estrogen-sensitive cancers, endometriosis, male genital abnormalities, reduced semen quality, and precocious puberty in girls illustrates the looming problem ( 17 , 18 ). Animal models provide unparalleled resources to study the effects of perinatal exposure to steroid excess. Most studies have used mice, rats, sheep, and nonhuman primates as research models, and comparative aspects of these different models have been reviewed previously ( 19 , 20 ). This review focuses primarily on research performed in the sheep, a model in which longitudinal studies focusing on multiple developmental time points have been carried out extensively. There are several advantages of using sheep for endocrine research. Sheep are amenable to different experimental procedures and interventions; their large size allows the performance of detailed and repetitive hormonal profiling and in vivo sampling and measurement of hypothalamic neuropeptides. Because sheep are domesticated, they are kept in a natural environment and not susceptible to the stress effects associated with caging. Moreover, as a precocial species, the developmental trajectory of several organs, such as brain and pancreas, follows a similar pattern as in humans ( 21 ) and hence is of translational relevance.