In mammals, including man, epidemiological and experimental studies have shown that a range of environmental factors acting during critical periods of early development can alter the phenotype of the offspring. The processes by which environmental conditions lead to permanent changes in the structure and function of mammalian tissues is known as developmental programming (Barker, 2001). The resulting phenotypical diversity depends on epigenetics, the permanent alteration in tissue structure and function without a change in DNA sequence that may be inherited transgenerationally (Lucas, 1991; Reik, 2007). In invertebrates and lower order vertebrates, hormones, such as thyroxine, leptin, androgens and ecdysteroids, have an important role in determining phenotype by regulating metamorphosis in response to environmental and other developmental signals (Tata, 1996; Groothuis et al. 2005; De Loof, 2008). Much less is known about the role of hormones in developmental programming in mammals, although androgens control the phenotype that develops from the sex chromosomes inherited at conception (Welsh et al. 2008). Recent studies have also shown that inappropriate exposure to glucocorticoids during intrauterine development affects the ensuing adult phenotype and leads to abnormalities in a wide range of physiological systems much later in life (Seckl, 2004; Moritz et al. 2005; Fowden et al. 2006a). This review examines the potential role of hormones as epigenetic signals in determining the phenotypical outcome of environmental cues acting during intrauterine development. It considers the role of hormones as environmental signals, their effects on phenotypical development and the molecular mechanisms by which hormones programme tissue development in utero, with particular emphasis on the actions of glucocorticoids in rats and sheep.
In mammals, including man, epidemiological and experimental studies have shown that a range of environmental factors acting during critical periods of early development can alter adult phenotype. Hormones have an important role in these epigenetic modifications and can signal the type, severity and duration of the environmental cue to the developing feto-placental tissues. They affect development of these tissues both directly and indirectly by changes in placental phenotype. They act to alter gene expression, hence the protein abundance in a wide range of different tissues, which has functional consequences for many physiological systems both before and after birth. By producing an epigenome specific to the prevailing condition in utero, hormones act as epigenetic signals in developmental programming, with important implications for adult health and disease. This review examines the role of hormones as epigenetic signals by considering their responses to environmental cues, their effects on phenotypical development and the molecular mechanisms by which they programme feto-placental development, with particular emphasis on the glucocorticoids.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Hormones as environmental signals
Fetal hormone concentrations are known to change in response to a range of internal and external environmental cues, such as nutrition, oxygen availability, temperature, photoperiod and unfamiliar surroundings (Fowden & Forhead, 2001). These hormonal changes have four principal origins. Firstly, they may be due to hormone secretion by the fetal endocrine glands per se. The fetal pancreas, thyroid, pituitary and adrenal glands are functional from early in gestation, although they may operate with different sensitivities and set points from the adult glands (Fowden et al. 2005). Secondly, hormones may be derived from the placental tissues. The placenta produces a range of steroid and other hormones that are released into the umbilical and uterine circulations (Challis et al. 2002). Thirdly, lipophilic hormones may be derived directly from the mother by transplacental transfer depending on the materno-fetal concentration gradient for diffusion and the permeability of the placental barrier (Sibley et al. 1997). Finally, changes in fetal hormone levels can occur by activation of precursor molecules or by inactivation of active hormones through metabolism within fetal and/or placental tissues. The endocrine milieu in utero, therefore, reflects environmental conditions in the mother, placenta and fetus and is a composite of hormones derived from all three of these compartments.
In general, environmental conditions favourable for fetal growth increase the concentrations of anabolic hormones, such as insulin, insulin-like growth factors (IGFs) and the thyroid hormones, and lower concentrations of catabolic hormones, such as cortisol and the catecholamines (Fowden & Forhead, 2001). Conversely, adverse conditions that constrain fetal growth tend to raise catabolic and lower anabolic hormone concentrations in the fetal circulation. Fetal hormones can also act as specific as well as general signals of environmental change (Fletcher et al. 2000a,b, 2003; Forhead et al. 2002b). For example, the fetal endocrine milieu observed during hypoglycaemia differs from that seen in response to hypoxaemia alone or to combined hypoxaemia and hypoglycaemia during conditions such as placental insufficiency (Fowden & Forhead, 2001). The pattern of fetal endocrine changes, therefore, signals particular facets of the environment to the developing fetus (Fowden et al. 2006a).
Signals of nutrient availability In fetal sheep, concentrations of insulin and IGF-I rise with increasing fetal concentrations of glucose over the normal range of values induced by variations in maternal nutritional state (Fig. 1). Since insulin does not cross the placenta (see Klieger et al. 2008), the rise in plasma insulin is an endocrine response of the fetal β-cells to rising glucose levels (see Fowden & Hill, 2001). In contrast, the increment in fetal plasma IGF-I probably reflects overspill of IGF-I produced by a number of different fetal tissues, since IGF-I is primarily a paracrine growth factor in utero (Fowden, 2003). The positive correlation between plasma IGF-I and glucose availability in the fetus may, therefore, be an indirect consequence of the increased anabolic drive at the tissue level rather than a systemic signal causing tissue accretion, as occurs in newborn and juvenile animals (Gluckman & Pinal, 2003). Insulin and IGF-I, therefore, act as signals of nutrient plenty at the cellular level, and both promote tissue accretion in line with substrate availability in the fetus.
In contrast to the anabolic hormones, concentrations of cortisol and prostaglandin E2 (PGE2) rise as fetal glucose levels decline (Fig. 1). The increment in fetal plasma PGE2 is placental in origin and directly related to the increased placental production of PGE2 induced by hypoglycaemia and the fall in uteroplacental glucose consumption (Fowden et al. 1987, 1994; Whittle et al. 2001). In contrast, the increment in fetal plasma cortisol concentrations may have fetal and maternal origins depending on gestational age. However, by late gestation, fetal cortisol levels exceed those of the mother at the end of a 48 h fast, which shows that cortisol is secreted by the fetal adrenal cortex in these circumstances (Fowden et al. 1998b). In part, this is due to direct activation of the fetal hypothalamic–pituitary–adrenal (HPA) axis by hypoglycaemia but it may also reflect the rise in circulating levels of PGE2, an eicosanoid known to stimulate adrenal cortisol secretion in utero (Liggins et al. 1982). There are also increases in the concentrations of adrenaline and noradrenaline in fetal sheep when hypoglycaemia is severe (Fowden et al 1998b; Fowden & Forhead, 2007). Catabolic hormones, therefore, act as signals of nutrient insufficiency and reduce fetal growth rate when substrate availability for tissue accretion is limited. Their growth-inhibitory actions are both direct and indirect, via antagonism of the growth-promoting effects of the anabolic hormones (Fowden, 1980; Fowden et al. 1996; Bassett & Hanson, 2000).
Signals of oxygen availability Similar relationships are observed between the partial pressure of O2 of arterial blood and the plasma concentrations of insulin, IGF-I, cortisol and PGE2 in fetal sheep during late gestation, with increases in plasma insulin and IGF-I with increasing O2 availability and inverse correlations between blood partial pressure of O2 and plasma concentrations of cortisol and PGE2 (Iwamoto et al. 1992; Jenkin et al. 2001; Fletcher et al. 2003). Fetal hypoxaemia also increases the concentration of several other hormones with endocrine, metabolic and vasoactive effects in utero, including adrenaline, noradrenaline, ACTH, leptin, arginine vasopressin (AVP), angiotensin II, adenosine and neuropeptide Y (Giussani et al. 1994; Fletcher et al. 2000a,b, 2003; Forhead et al. 2002b). These hormones, therefore, act as signals of O2 availability and, by altering substrate availability and blood flow to specific tissues, change the pattern of intrauterine development, with consequences for tissue function both before and after birth (Giussani et al. 1994; Fowden et al. 2006a). For example, the rise in catecholamines in response to hypoxaemia helps to maintain an oxygen and glucose supply to key tissues, such as the brain, adrenals and placenta, but at the same time restricts growth of other, insulin-sensitive, peripheral tissues, such as adipose tissue and skeletal muscle, by inhibiting insulin secretion by the fetal β-cells (Fowden, 1980; Giussani et al. 1994; Bassett & Hanson, 2000).
Signals of maternal stress Since the placenta is permeable to lipophilic hormones, such as cortisol, stressful conditions in the mother can lead to increased fetal corticosteroid concentrations by transplacental transfer of maternal cortisol without direct activation of the fetal HPA axis. In rats, sheep and horses, fetal cortisol concentrations have been shown to parallel maternal concentrations closely during stressful conditions, such as maternal saline injection, noise, restraint, high temperature, unfamiliar surroundings and social mixing, that have little, if any, effect on fetal availability of glucose or oxygen, normally involved in stimulating the fetal HPA axis (Hennessy et al. 1982; Ward & Weisz, 1984; Barbazanges et al. 1996; Gardner et al. 1997; Fowden et al. 2008b). Indeed, in rats, exposure to maternal glucocorticoids has been shown to reduce fetal growth and account for the programming of metabolic and cardiovascular function in the adult offspring of mothers stressed during pregnancy by restraint and low-protein diets (Barbazanges et al. 1996; Gardner et al. 1997). Maternal glucocorticoids may also contribute to the rise in fetal glucocorticoid concentrations observed in response to maternal undernutrition and dietary manipulations, particularly at periods of gestation before the fetal HPA axis becomes responsive to hypoglycaemia (Gardner et al. 1997; Fowden et al. 2008b). Glucocorticoids can, therefore, act as transplacental signals of maternal stress with phenotypical consequences for the offspring.
In normal conditions, fetal and placental tissues are prevented from overexposure to the higher maternal glucocorticoid concentrations by the placental activity of 11β-hydroxysteroid dehydrogenase type 2 (11βHSD2), the enzyme that converts active glucocorticoids to their inactive keto-metabolites (Seckl, 2001). However, overexposure to glucocorticoids and adverse conditions during pregnancy, such as undernutrition, reduce activity of this enzyme and, thereby, increase the bioavailability of the active hormone within the placental tissues and fetal circulation (see Fowden et al. 2008a). In the fetus, the effects of transplacentally acquired and endogenously secreted glucocorticoids are determined by the activities of the different 11βHSD isoforms in the individual tissues, which control the balance between biologically active and inactive glucocorticoids (Seckl, 2001; O’Regan et al. 2002). In turn, these enzyme activities can be influenced by the fetal endocrine milieu, in particular, by the concentrations of oestrogen, progesterone and cortisol, in a manner specific to the tissue and stage of development (Sun et al. 1998; Clarke et al. 2002). The transplacental glucocorticoid signal of maternal stress may, therefore, be exaggerated or ameliorated in particular feto-placental tissues depending on the other endocrine signals operating at the time of the insult.
Signals of gestational age and maturation Hormonal responses to environmental challenges change with increasing gestational age as the fetal endocrine glands become progressively more responsive to stimuli towards term (Fowden & Forhead, 2004). For instance, in fetal sheep, the increment in plasma cortisol in response to both hypoglycaemia and hypoxaemia increases as the fetal HPA axis is activated during the last 10–20 days before birth (Fig. 2). Similarly, there are increases in the fetal pancreatic α- and β-cell responses to glucose and amino acids between mid and late gestation in sheep, horses and pigs (see Fowden & Hill, 2001). The magnitude of the hormonal response to specific environmental cues, therefore, signals proximity to delivery and induces an appropriate adaptive response for the stage of development of the fetus. In addition, basal concentrations of many hormones rise during the prepartum period in preparation for extrauterine life. In the last 10–15 days before birth, there are increases in the concentrations of cortisol, PGE2, IGF-I, leptin, tri-iodothyronine (T3) and adrenaline, even in well-nourished, normoxic, unstressed animals (Barnes et al. 1978; Thomas et al. 1978; Fowden et al. 1987; Li et al. 1996; Forhead et al. 2002b; Fowden & Forhead, 2007). In part, these ontogenic changes reflect alterations in the set point and sensitivity of the endocrine axes, particularly of the HPA axis, which initiate the onset of labour and are essential for neonatal survival in most species (see Silver, 1990). Indeed, the cortisol surge towards term is responsible for the ontogenic increases in the plasma concentrations of T3, leptin, IGF-I and adrenaline in the sheep fetus (Sensky et al. 1994; Li et al. 1996; Forhead et al. 2002b; Fowden & Forhead, 2007). Ontogenic changes in hormones, such as leptin, are also an index of the rising fetal adiposity essential for a successful transition from parenteral to enteral nutrition at birth (Forhead & Fowden, 2009). The prepartum changes in basal and evoked hormone concentrations in the fetus, therefore, signal the degree of tissue maturation and the chances of survival should delivery occur.
Signals of other external environmental cues In addition to responding to internal environmental cues, fetal hormone concentrations also signal features of the external environment, such as day length and the dark–light cycle (see Seron-Ferre et al. 2007). In sheep, diurnal rhythms have been observed in the concentrations of cortisol, ACTH, prolactin and PGE2 in fetal plasma during late gestation (McMillen et al. 1987; Fowden et al. 1987). Similar diurnal rhythms have been observed for cortisol in fetal horses and human infants and for fetal dehydroepiandrosterone in non-human primates (Cudd et al. 1995; Seron-Ferre et al. 2007). These fetal endocrine rhythms are related to the dark–light cycle and have two potential Zeitgebers, melatonin and glucose availability. In fetal horses and sheep, the diurnal cortisol and PGE2 rhythms closely follow the variations in fetal glucose concentrations caused by the once or twice daily feeding regime of the mother (McMillen et al. 1987; Cudd et al. 1995). When food is made more freely available, the daily fluctuations in fetal glucose, PGE2 and cortisol concentrations are much less pronounced (Slater & Mellor, 1981; Fowden et al. 1994). In contrast, the fetal prolactin rhythm appears to parallel the fetal melatonin concentrations which, in turn, mirror maternal values, since melatonin crosses the placenta readily (Houghton et al. 1993). Indeed, manipulating fetal melatonin concentrations by maternal pinealectomy, by exogenous infusion or by varying maternal light exposure has been shown to alter fetal prolactin concentrations, in part, by actions through the fetal suprachiasmatic nucleus (SCN) and hypothalamus (see Seron-Ferre et al. 2007). Since melatonin receptors are present in a number of fetal tissues, including the SCN and adrenal glands (Torres-Farfan et al, 2004), maternal melatonin may be the entraining signal for several of the diurnal endocrine and behavioural rhythms observed in utero as well as for the seasonal variation in fetal prolactin concentrations that occurs with changes in day length (Bassett et al. 1988; Torres-Farfan et al. 2004; Seron-Ferre et al. 2007). Hormones can, therefore, act as signals of photoperiod and alter phenotypical development to maximize chances of survival in the ambient conditions prevailing at the time of birth. This may explain the seasonal variation in pregnancy duration and neonatal coat length in species, such as the horse, which have a wide window for delivery spanning winter to summer months (Davies et al. 2002).
Hormones and development of phenotype
In mammals, there are two main mechanisms by which hormones can affect the phenotype of the offspring. Firstly, they may influence the morphology and structure of the placenta, which, in turn, controls the fetal nutrient supply and the bioavailability of specific hormones critical for intrauterine development (Fowden et al. 2006c). Secondly, hormones in the fetal circulation may act directly on the fetal tissues to alter cell growth and differentiation, with consequences for their function much later in life (Fowden et al. 1998a, 2006a). The phenotypic consequences of prenatal hormone exposure may not always be evident at birth but may emerge later in life when the demand on the tissue increases with puberty, pregnancy, ageing or a postnatal environmental insult.
Placental phenotype The functional capacity of the placenta to supply nutrients to the fetus depends on its size, morphology, blood flow, transporter abundance and its rate of nutrient consumption (Fowden et al. 2006c). Hormones have been shown to affect all these factors (Table 1). In particular, IGF-I and glucocorticoids have major effects on the placental capacity for nutrient transfer. In sheep and rats, glucocorticoids affect placental size, gross and cellular morphology, blood flow, glucose utilization and gene expression for nutrient transporters and hormones, such as leptin and key somatomammotrophins (Table 1). Similarly, the IGFs alter placental growth and morphology and the clearance of glucose and amino acids as well as placental expression of the glucose transporters (GLUTs) and system A family of amino acid transporters (see Forbes & Westwood, 2008). Indeed, placental amino acid transport and amino acid transporter abundance appear to be particularly sensitive to endocrine manipulations during late gestation and are known to be modified by cortisol, IGF-I, IGF-II, growth hormone (GH), leptin and angiotensin II (Table 1). In part, the actions of IGF-I and glucocorticoids on placental phenotype may be mediated via the concomitant down-regulation of placental expression of the Igf2 gene (Ain et al. 2005; Sferruzzi-Perri et al. 2007). This gene has an important role in placental development, and deletion of its placenta-specific transcript in mice causes growth retardation of the placenta (Constância et al. 2002), in association with specific morphological abnormalities and up-regulation of system A transporter abundance and activity in the mutant relative to the wild-type placenta (Sibley et al. 2004; Constância et al. 2005; Coan et al. 2008).
|Hormone||Species||Route of administration||Placental effects||References|
|Leptin||Human||In vitro||↑ Amino acid transport and transporter expression||Jansson et al. (2003)|
|Angiotensin II||Human||In vitro||↓ MeAIB transport, ↓ 11βHSD2||Shibata et al. (2006); Lanz et al. (2003)|
|Growth hormone||Human||In vitro||↑ Glucose uptake, ↓ MeAIB uptake||Ericsson et al. (2005)|
|Sheep||In vivo to mother||↑ Urea clearance, ↑ Uteroplacental weight||Harding et al. (1997); Wallace et al. (2004)|
|IGF-I||Guinea-pig||In vivo to mother||↑ Weight, ↑ glucose and MeAIB uptake, ↑ SNAT2 gene expression||Sferruzzi-Perri et al. (2007)|
|Sheep||In vivo to mother||↑ Lactate production, ↑ amino acid uptake||Liu et al. (1994)|
|Human||In vivo to fetus||↓ Lactate production, ↓ glucose and MeAIB clearance||Bloomfield et al. (2002)|
|In vitro||↑ Amino acid uptake, ↓ apoptosis, ↑ differentiation||Kniss et al. (1994); Miller et al. (2005)|
|IGF-II||Guinea-pig||In vivo to mother||↑ Volume and surface area of labyrinth||Sferruzzi-Perri et al. (2006)|
|Human||In vitro||↓ Apoptosis, ↑ proliferation||Miller et al. (2005)|
|T3||Rat||In vivo to mother||↓ Glycogen content, ↑ aminotransferase activity||Shafrir et al. 1994|
|Glucocorticoids||Mouse||In vivo to mother||↓ Weight, ↑ apoptosis, ↑ aquaporin, ↓ cyclin expression||Baisden et al. (2007)|
|Rat||In vivo to mother||↓ Weight, ↓Igf2 & prolactin gene family expression||Ain et al. (2005)|
|↑ GLUT1 & GLUT3 expression||Langdown & Sugden (2001)|
|↓ Leptin expression, ↑ Leptin receptor ObR-S, ↓ Ob-Rb||Sugden et al. (2001); Smith & Waddell (2002)|
|↓ Labyrinthine VEGF expression, ↓ PPARγ expression||Hewitt et al. (2006a,b)|
|↑ PEPCK & aminotransferase activities||Shafrir et al. (1994)|
|Sheep||In vivo to mother||↓ Weight. Altered placentome type distribution, ↓ BNC number||Jensen et al. (2002); Braun et al. (2007)|
|↑ Umbilical blood flow||Jellyman et al. (2004a)|
|In vivo to fetus||↓ Glucose transfer, ↑ Glucose consumption||Ward et al. (2004)|
|↓ Placentome eversion, ↓ BNC number||Ward et al. (2006)|
|↓ Glutamate uptake||Timmerman et al. (2001)|
|↓ Deiodinase type III||Forhead et al. (2006, 2007)|
|Monkey||In vivo to mother||↓ Weight, ↓ eNOS protein expression||Johnson et al. (1979); Aida et al. (2004)|
|Human||In vitro||↑ Amino acid transport, ↑ SNAT2 transporter gene expression||Jones et al. (2006)|
|↑ Fibronectin expression, ↑ AT1 receptor expression||Lee et al. (2004); Lanz et al. (2003)|
|↑ PGE2 production||Mirazi et al. (2004)|
Hormones also modify the endocrine function of the placenta by altering the synthesis and metabolism of placental and systemic hormones. This influences the transplacental passage and bioavailability of hormones in the fetal and maternal circulations, with consequences for feto-placental development and the partitioning of maternal nutrients to the fetus (Table 1). The glucocorticoids, in particular, are known to influence placental synthesis of PGE2, PGF2α, placental lactogen, leptin, corticotrophin-releasing hormone (CRH), oestrogens, progesterone and other progestagens depending on the species (Rossdale et al. 1992; Fowden et al. 1994; Sugden et al. 2001; Challis et al. 2002; Braun et al. 2007). In sheep, they also regulate their own metabolism by down-regulating placental 11βHSD2 activity and, thereby, increasing feto-placental exposure to active cortisol (Clarke et al. 2002). Furthermore, cortisol regulates metabolism of thyroid hormones and PGs to their inactive forms by controlling deiodinase and prostaglandin dehydrogenase activities in ovine and human placenta (Fowden et al. 1994; Whittle et al. 2001; Forhead et al. 2006, 2007). In turn, availability of the primary PGs regulates activity of the different 11βHSD isoforms in the placenta and fetal membranes, further increasing bioavailability of active glucocorticoids (Seckl, 2001; Challis et al. 2002). There are, therefore, complex, paracrine interactions between cortisol, CRH, PGs and sex steroids within these tissues that regulate placental hormone secretion and myometrial contractile activity. Photoperiodic signals, such as melatonin, may also influence these endocrine interactions, since melatonin receptors are present in the placenta (Torres-Farfan et al. 2004; Iwasaki et al. 2005; Seron-Ferre et al. 2007). In addition to these immediate effects, hormonally induced changes in placental endocrine function may also have longer term phenotypical consequences by altering mammary gland development and the metabolic preparations of the mother for lactation, which will influence the nutrition and growth of the offspring postnatally (Ward et al. 2002; Braun et al. 2007).
The effects of the hormones, such as IGF-I and the glucocorticoids, on the placenta appear to depend, in part, on the route of administration. For instance, in sheep, IGF-I administration to the mother increases placental lactate production and amino acid uptake, while fetal treatment decreases placental lactate production and amino acid clearance (Liu et al. 1994; Bloomfield et al. 2002). Similarly, umbilical blood flow increases in response to glucocorticoid treatment of the ewe but not of the fetus directly (Jellyman et al. 2004a; Ward et al. 2004). These differential effects of fetal and maternal treatment may be due, in part, to differing receptor abundance on the two sides of the placenta or to differences in the other endocrine and metabolic responses to treatment between the mother and fetus. However, not all responses differ with the route of administration. For example, in sheep, both fetal and maternal glucocorticoid treatment decrease binucleate cell number and prevent placentome eversion during late gestation (Ward et al. 2002; Braun et al. 2007). Hormones, therefore, affect placental phenotype, both directly and indirectly via altering other key regulators of placental development, such as haemodynamics and nutrient availability. However, few studies have examined whether the placental effects of hormone manipulation persist after cessation of treatment to permanently programme the transport and endocrine functions of the placenta.
Fetal phenotype Surgical ablation of fetal endocrine glands has shown that specific hormones are required for normal growth and development in utero. In particular, deficiency of the pancreatic, pituitary, adrenal or thyroid hormones leads to fetal growth retardation and/or abnormalities in individual fetal tissues, such as the brain, liver, lungs and skeletal muscle (see Fowden et al. 1998b, 2006b). In many instances, replacement of the deficient hormone by exogenous infusion restores the normal pattern of intrauterine growth (Fowden & Forhead, 2004). Similarly, infusion of hormones, such as insulin, T4, leptin and glucocorticoids, directly into normal fetuses during late gestation alters development of particular fetal tissues, such as the liver, fat and skeletal muscle, with potential consequences for tissue growth and metabolism after birth (Barnes et al. 1978; Fowden & Silver, 1995; Mostyn et al. 2003; Forhead et al. 2008). These changes in fetal development are often accompanied by alterations in fetal substrate utilization, which, in some instances, occur without any detectable change in the placental capacity to supply macronutrients (Fowden & Forhead, 2007). Hormones can, therefore, influence fetal phenotype, in part, by direct actions on the fetal tissues.
Of the hormones involved in regulating normal fetal growth, the glucocorticoids are the most likely to have a major role in prenatal tissue programming because they are responsive to most environmental challenges (Figs 1 and 2) and affect both cell accretion and differentiation in utero (Fowden et al. 1998a). They also influence all the tissue and organ systems known to be at increased risk of adult pathophysiology after suboptimal intrauterine conditions (Fowden et al. 2006a). The changes in development induced by fetal overexposure to glucocorticoids at inappropriate periods of gestation are both morphological and functional (Table 2). In the lung, glucocorticoids alter the ratio of type I to type II pneumocytes, the alveolar density and the thickness of the alveolar wall, while in the gut, they influence enterocyte migration, muscle thickness and the length and density of the villi (see Kitterman et al. 1981; Trahair & Sangild, 1997). Often, the morphological changes induced by early glucocorticoid exposure mimic those seen in response to the natural prepartum rise in fetal glucocorticoid concentrations, but this early differentiation may occur at the expense of total cell number, with consequences for the capacity of the tissue to respond to subsequent challenges, such as pregnancy and ageing (Table 2). Functionally, glucocorticoids are known to induce changes in the cellular abundance of receptors, ion channels, transporters, enzymes, growth factors and intracellular signalling molecules in a range of different fetal tissues (Table 2), which lead to integrated physiological changes at the systems level (Fowden et al. 2005, 2006a). For instance, the rise in fetal blood pressure induced by glucocorticoid administration in late gestation reflects the combined structural and functional changes in the fetal heart, kidneys, blood vessels and brain (Fowden et al. 1998b; Moritz et al. 2005; Fletcher et al. 2005). Premature activation of the glucocorticoid-induced switch from cell proliferation to differentiation can also lead to changes in cell number, type or function that have little apparent systemic effect until a subsequent challenge unmasks the defects, often at a later stage of development.
|Tissue||Period of development||Consequences of exposure||References|
|Liver||Prenatal||↓ Relative weight||Mosier et al. (1982); LaBorde et al. (1992)|
|↑ G6Pase, ↑ PEPCK||Fowden et al. (1993)|
|↑ Aspartate aminotransferase, ↑ pyruvate carboxylase||Fowden et al. (1993)|
|↑ Argininosuccuate lipase, ↑ 11βHSD1||Renouf et al. (1995); Sloboda et al. (2002)|
|↑ GH & prolactin receptors||Li et al. (1996); Phillips et al. (1997)|
|↓ AT1 receptors||Segar et al. (1995)|
|↑ IGF-I, ↓ IGF-II gene expression||Li et al. (1996), (1998)|
|↑ CBG, ↑ IGFBP, ↑ erythropoietin||Lim et al. (1996)|
|↓ Angiotensinogen gene expression||Olson et al. (1991); Segar et al. (1995)|
|↑ Deiodinase type 1||Forhead et al. (2006, 2007)|
|↑ Glycogen deposition||Barnes et al. (1978); Klepac, (1985); Franko et al. (2007)|
|Postnatal||↑ PEPCK, ↑ G6Pase, ↑ GR||Nyirenda et al. (1998); Sloboda et al. (2005)|
|↓ CBG||Sloboda et al. (2005)|
|Δ Angiotensinogen, Δ renin||O’Regan et al. (2004)|
|Lungs||Prenatal||↓ Relative weight||LaBorde et al. (1992)|
|↓ Alveolar density||Willet et al. (2001)|
|↑ Type I, ↓ type II pneumocytes||Crone et al. (1983)|
|↑ Thinning of alveolar wall||Polglaise et al. (2007)|
|↑ ACE, ↑ iNOS and eNOS||Forhead et al. (2000); Arima et al. (2008)|
|↑ Tropoelastin expression||Pierce et al. (1995)|
|↑ Surfactant proteins||Kitterman et al. (1981); Tan et al. (1999)|
|↑ Adrenoreceptors||Barker et al. (1990)|
|↑ Fatty acid synthase||Xu & Rooney, (1997)|
|↓ Glycogen content||Barnes et al. (1978)|
|Postnatal||↓ Adrenoreceptors||Kudlac et al. (1990)|
|Kidney||Prenatal||↓ Relative weight||Mosier et al. (1982)|
|↑ ACE||Forhead et al. (2000)|
|↓ Erythropoietin expression||Lim et al. (1996)|
|↓ AT1 and AT2 receptors, ↓ Renin expression||Segar et al. (1995); Massmann et al. (2006)|
|↑ Na+–H+ exchanger (NHE3)||Guillery et al. (1995)|
|↑ Epithelial Na+ channels||Nakamura et al. (2002)|
|↓ Nephrogenesis||Moritz et al. (2002)|
|↑ Deiodinase type I, ↑ deiodinase type III||Forhead et al. (2006, 2007)|
|↑ MR and GR gene expression||Hantzis et al. (2002)|
|Postnatal||↑ Relative weight, ↓ nephron number||Ortiz et al. (2003); Wintour et al. (2003)|
|↑ Type I and III collagen||Wintour et al. (2003)|
|↓ 11βHSD2||Wyrwoll et al. (2007)|
|↑ Na+–K+-ATPase, ↑ renin, ↑ ACE, ↑GR||Wyrwoll et al. (2007)|
|↑ Na+–H+ exchanger (NHE3)||Dagan et al. (2007)|
|Δ Superoxide production (vessel dependent)||Roghair et al. (2008)|
|Skeletal muscle||Prenatal||↑ Glycogen content||Barnes et al. (1978)|
|↓ GLUT1 and 4||Gray et al. (2006)|
|↓ Protein kinase ζ||Jellyman et al. (2008)|
|Postnatal||↑ GLUT4, ↑ PPARγ, ↑ PPARγC1a||Wyrwoll et al. (2008)|
|↓ UCP3, ↓ glycogen content, ↓ GR||Cleasby et al. (2003)|
|Heart and blood vessels||Prenatal||↓ & ↑ Relative heart weight||Mosier et al. (1982); Lumbers et al. (2005)|
|↓ Calreticulin & ↑ calsequestrin expression||Langdown et al. (2003)|
|↑ SCN1b and SCN5a expression||Fahmi et al. (2004)|
|↓ GLUT1, ↑ angiotensinogen gene expression||Lumbers et al. (2005)|
|↑ Glycogen content||Barnes et al. (1978)|
|Postnatal||↑ Relative heart weight||Woods & Weeks (2005)|
|↑ Calreticulin protein||Langdown et al. (2003)|
|↑ GLUT1 protein, ↑ protein kinase B and C||Langdown et al. (2001a)|
|↓ UCP2 and 3 protein||Langdown et al. (2001b)|
|Adipose tissue||Prenatal||↑ Leptin expression, ↑ UCP1 protein||Forhead et al. (2002); Mostyn et al. (2003)|
|Postnatal||↑ Fat deposition, ↓ lipoprotein lipase, ↑ GR||Cleasby et al. (2003)|
In addition to altering the normal pattern of tissue development, early exposure to glucocorticoids changes fetal responses to adverse conditions in utero both during the period of exposure and thereafter. In sheep, the metabolic, endocrine and cardiovascular responses to hypoxaemia are altered by administration of dexamethasone to either the mother or the fetus (Fletcher et al. 2003; Jellyman et al. 2004a,b). Glucocorticoids prolong the bradycardic response, increase the femoral vasoconstrictor response and enhance the increments in plasma neuropeptide Y, glucose and lactate to hypoxaemia during the period of steroid exposure (Fletcher et al. 2000b, 2003; Jellyman et al. 2005). Some but not all of these effects persist after treatment (Fletcher et al. 2003; Jellyman et al. 2005). Other effects develop only after dexamethasone has cleared from the circulation. For instance, the catecholamine and AVP responses to hypoxaemia are suppressed during dexamethasone treatment but are enhanced after treatment relative to the responses in saline-treated control animals (Fletcher et al. 2000b, 2004). Similarly, the fetal pituitary–adrenocortical response to hypoxaemia is abolished during dexamethasone treatment but is two- to threefold greater than control values 48 h after treatment ceases, in association with down-regulation of glucocorticoid receptors (GR) in the fetal pituitary (Fletcher et al. 2004; Jellyman et al. 2004b). In addition, pretreatment of the sheep fetus with cortisol increases pulmonary growth and liquid accumulation in response to subsequent obstruction of the fetal trachea (Boland et al. 1997). Prior exposure to glucocorticoids, therefore, alters the adaptive responses to subsequent environmental challenges, which may contribute to the tissue programming long after the original stimulus that raised fetal glucocorticoid concentrations.
Although glucocorticoids appear to have a central role in developmental programming, the fetal phenotypical outcome of environmental challenges reflects the actions of multiple hormones in utero. Indeed, the actions of the glucocorticoids in modifying fetal development are often mediated through other hormones and growth factors (Fowden & Forhead, 2004). Cortisol has been shown to suppress IGF2 gene expression in fetal ovine liver, adrenal and skeletal muscle, which, in turn, may affect proliferation, clonal selection, differentiation and apoptotic remodelling of the cell populations within these tissues (see Fowden & Hill, 2001; Fowden & Forhead, 2004). Glucocorticoids also stimulate the onset of GH-dependent production of endocrine IGF-I in fetal ovine hepatocytes close to term (Li et al. 1996, 1999). Premature activation of this maturational process could have widespread effects on fetal growth, particularly since GH concentrations are high in utero (Gluckman & Pinal, 2003). In part, these effects of glucocorticoids on tissue IGF-I and IGF-II synthesis are mediated via the thyroid hormones and the glucocorticoid-induced changes in tissue deiodinase activity (Forhead et al. 2002a,b, 2006, 2007). Thyroid hormones also modify the effects of glucocorticoids on gluconeogenic enzyme activities in the liver and angiotensin-converting enzyme activity in the lungs and kidneys of fetal sheep (Fowden et al. 2001; Forhead & Fowden, 2002). In addition, glucocorticoids alter the secretion and/or bioactivity of leptin, adrenomedullary hormones, the renin–angiotensin system and pancreatic hormones, either by effects on development of the endocrine cells or by changes in receptor abundance in the target tissues (Blondeau et al. 2001; Forhead et al. 2002a; Moritz et al. 2002; O'Connor et al. 2007). Since all these hormones are sensitive to environmental conditions in utero and many have direct effects on fetal growth and metabolism (Fowden & Forhead, 2004), they are likely to make a significant contribution, along with the glucocorticoids, to the epigenetic modifications in tissue structure and function responsible for developmental programming.
Adult phenotype Hormone exposure in utero also affects adult phenotype. In particular, exposure to abnormal concentrations of glucocorticoids and sex steroids during fetal life leads to structural and functional changes in a wide range of adult tissues (Table 2). These result in abnormalites in cardiovascular, metabolic, endocrine and reproductive function in the adult that can lead to clinical disorders and diseases, such as coronary heart disease, infertility and type 2 diabetes, particularly as physiological systems deteriorate with age (Seckl, 2004; Moritz et al. 2005; Fowden et al. 2006a). In part, the adult phenotypical consequences of prenatal hormone exposure track from fetal life. Morphological changes induced in utero by glucocorticoid overexposure, such as clonal selection, reduced cell number and lower relative organ size, tend to persist after birth (Table 2). Similarly, some of the functional changes observed at the cellular level in fetal tissues after glucocorticoid treatment are also present in the tissues of the adult offspring (Table 2). Increases in the hepatic glucogenic capacity induced in utero by maternal glucocorticoid administration late in gestation persist after birth and are associated with glucose intolerance in the adult offspring (Nyirenda et al. 1998, 2006; Franko, 2007). In some tissues, maternal glucocorticoid administration has opposite effects in the fetus and adult offspring (Table 2). In sheep, for instance, prenatal glucocorticoid overexposure reduces hepatic production of angiotensinogen and corticosteroid-binding globulin in the fetus but enhances their production in adult liver in a sex-linked manner (Olson et al. 1991; Lim et al. 1996; O’Regan et al. 2004; Sloboda et al. 2005). Some of the consequences of prenatal hormone exposure do not become apparent until after birth. The hypertensive effects of maternal glucocorticoid treatment at 20% of gestation are not seen in the fetus near term but are observed postnatally from as early as 4 months of postnatal age (Dodic et al. 1999, 2002; Moritz et al. 2002). Prenatal glucocorticoid exposure also induces insulin resistance in the adult offspring but not in utero (Ward et al. 2004; De Blasio et al. 2007). In particular, the reproductive consequences of early glucocorticoid or androgen exposure are only evident after puberty, when many of the sex-linked differences in developmental programming appear for the first time with the onset of gonadal steroidogenesis (Welsh et al. 2008; Grigore et al. 2008). Conditions during postnatal life may, therefore, influence the adult phenotype that develops in response to prenatal hormone exposure.
The phenotypical outcome of prenatal glucocorticoid administration also depends, in part, on gestational age at treatment and whether exposure is to natural or synthetic glucocorticoids (Fowden et al. 1998a; Moritz et al. 2005). In rats, hepatic phosphoenolpyruvate caboxykinase (PEPCK) expression is upregulated in the adult offspring after maternal dexamethasone treatment in the third but not the first or second week of pregnancy (Nyirenda et al. 2006). Similarly, in sheep, maternal treatment with synthetic glucocorticoids early in gestation leads to adult hypertension but not glucose tolerance, whereas treatment later in gestation has the opposite effects in the adult offspring (Gatford et al. 2000; Moss et al. 2001). The adult hypertensive effect of early prenatal glucocorticoid treatment is seen with both natural and synthetic glucocorticoids, although it is due to increased cardiac output in animals treated with dexamethasone and to increased peripheral resistance in those receiving cortisol (Dodic et al. 2002; Moritz et al. 2005). In contrast, treatment with cortisol early in gestation leads to fasting hyperglycaemia and hyperinsulinaemia in the adult offspring, whereas treatment with dexamethasone does not (De Blasio et al. 2007). This may reflect ontogenic differences in receptor abundance in different fetal tissues at the time of steroid exposure, since natural and synthetic glucocorticoids differ in their affinities for the GR, mineralocorticoid and other steroid receptors. However, even in the same tissue, glucocorticoids can have different actions on specific cell functions depending on gestational age. In fetal ovine liver, cortisol increases both glucose-6-phosphatase (G6Pase) and fructose diphosphatase activity at 130 days, but only G6Pase activity at 115 days of gestation (Fig. 3). At the time of organogenesis early in gestation, the consequences of glucocorticoid exposure may be teratogenic and lead to gross morphological abnormalities. In sheep, maternal dexamethasone administration at 20% of gestation, when the fetal metanephric kidney is developing, reduces nephron number in the offspring both at term and in adulthood, although this is only associated with hypertension in the adult (Moritz et al. 2002; Wintour et al. 2003). In contrast, maternal glucocorticoid administration late in gestation, after nephrogenesis is complete, elevates fetal blood pressure but has no apparent effect on blood pressure of the adult offspring (Bennet et al. 1999; Moss et al. 2001; Jellyman et al. 2004a). In late gestation, the consequences of inappropriate glucocorticoid exposure are due, in part, to premature activation of the maturational events normally induced by the natural increment in fetal glucocorticoid concentrations towards term (Fowden et al. 1998a). Changes in prepartum tissue maturation in relation to the timing of delivery may, therefore, have consequences for the set point and sensitivity of key physiological systems during the perinatal period that can persist, revert or amplify in later life.
Like fetal phenotype, the adult phenotypical outcome of suboptimal intrauterine conditions depends on multiple hormones acting both before and after birth. Prenatal exposure to androgens alters pituitary gonadotrophin production and gonadal steroidogenesis in the offspring from puberty onwards (Page et al. 2001; Manikkam et al. 2008; Welsh et al. 2008). In turn, the changes in sex steroid concentrations contribute to the alterations in growth rate, sexual behaviour, metabolism and functioning of the cardiovascular and other physiological systems in the adult (Smith & Waddell, 2000; Page et al. 2001; Fowden et al. 2006a; Pedrana et al. 2008). Similarly, inappropriate glucocorticoid exposure in utero alters the functioning of all the hypothalamic–pituitary endocrine axes (O’Regan et al. 2001; Slone-Wilcoxon & Redei, 2004; Theogaraj et al. 2005; Sloboda et al. 2007; Welsh et al. 2008). It also influences functioning of the endocrine pancreas and the renin–angiotensin system as well as the endocrine axes regulating appetite and energy balance in the adult offspring (Remacle et al. 2007; Taylor & Poston, 2007; Wyrwoll et al. 2007, 2008). Indeed, the changes in cell structure and function observed in adult tissues after prenatal glucocorticoid administration may be secondary to altered postnatal glucocorticoid concentrations rather than a direct consequence of the fetal endocrine milieu per se. Postnatal hormone concentrations can, therefore, act as epigenetic signals of the prenatal developmental environment and may be the primary cause of the postnatal physiological abnormalities associated with prenatal hormone overexposure. Programming the set point and sensitivity of the endocrine axes in this way may provide a mechanism for transmitting the memory of early developmental events to tissues much later in life, long after the original environmental challenge.
Molecular mechanisms by which hormones act on phenotype
There are a number of mechanisms by which hormones can alter gene expression to allow phenotypical diversity. Generally, these involve alterations to chromatin structure or to mRNA transcription and translation. Hormones, such as the glucocorticoids, have been shown to alter DNA methylation, particularly at the GR promotor (Robyr & Wolffe, 1998; Adcock et al. 2004). In cultured hepatocytes from fetal rats, dexamethasone induces stable DNA demethylation at the glucocorticoid response element of the tyrosine aminotransferase gene, which is associated with increased expression of this gene (Thomassin et al. 2001). Changes in methylation status induced in utero may persist after birth to alter adult gene expression, since demethylation of the GR promotor and increased GR gene expression have been observed in kidneys of adult rats overexposed to glucocorticoids during late gestation (Wyrwoll et al. 2007). Relatively little is known about the effects of hormones on histone modifications during early development, although histone acetylation is hormone sensitive in adult tissues and can be altered in fetal liver and brain by conditions, such as maternal dietary manipulation and uteroplacental insufficiency, which affect the fetal endocrine environment (Robyr & Wolffe, 1998; Fu et al. 2004; Ke et al. 2006; Aagaard-Tillery et al. 2008). Even less is known about the endocrine regulation of the DNA methyltransferases responsible for the maintenance and de novo methylation of DNA during development.
At the transcriptional level, hormones can alter gene expression by changes in transcription factor abundance, RNA stability and/or promotor usage and differential splicing. In fetal liver, glucocorticoids have been shown to influence promotor usage of several genes with multiple mRNA transcripts, including the GR, growth hormone receptor (GHR), IGF-I and Hnf4a genes (Li et al. 1996, 1998; McCormick et al. 2000; Nyirenda et al. 2006). This leads to alternative slicing of the exons and differences in the 5′ untranslated end of the mRNA transcripts, which may influence their translation. In sheep, cortisol at prepartum concentrations activates expression of the GHR gene from exon 1A, the adult specific promotor of the mRNA transcript responsible for coding the functional adult GHR (Fig. 4). Certainly, cortisol-induced up-regulation of this transcript coincides with the appearance of GH receptors in hepatocyte membranes from fetal sheep near term (Gluckman et al. 1983). Cortisol also activates expression of the adult GH-sensitive mRNA transcript of the IGF I gene in fetal ovine liver (Fig. 4), although whether this is a direct effect of cortisol or mediated via the increased GHR abundance remains unclear (Li et al. 1996). The effect of cortisol in down-regulating IGF2 gene transcription in fetal ovine liver is also specific to exon 7 of the gene (Li et al. 1998). In part, the glucocorticoid-induced changes in tissue IGF2 gene expression may be due to alterations in imprint status, since this gene is expressed only from the paternal allele in the fetus (Constância et al. 2004). In both human and ovine liver, the IGF2 gene switches from monoallelic to biallelic expression during the perinatal period in parallel with the fetal cortisol surge (Kalscheuer et al. 1993; McLaren & Montgomery, 1999). Similarly, in rats, silencing of the hepatic Igf2 gene occurs when corticosterone concentrations are rising most rapidly at the time of weaning (Senior et al. 1996). Whether these changes in Igf2 expression are due to altered expression of the H19-derived non-coding RNA or to changes in methylation at the differentially methylated region or imprinting control region of the Igf2-H19 locus remains unclear (Fowden et al. 2006b). Indeed, very few studies have investigated the specific effects of glucocorticoids, or any other hormones, on genomic imprinting.
In fetal rat lung in vitro, glucocorticoids increase expression of fatty acid synthase, a key enzyme in surfactant synthesis, by enhancing both the rate of transcription and the stability of the mRNA transcript (Xu & Rooney, 1997). Hormones, such as glucocorticoids, can also affect gene transcription by altering the abundance of key transcription factors, such as cfos, Activator protein 1, Transcription factor 1, Hepatocyte nuclear factor 4 alpha and CCAAT(C)/enhancer-binding protein delta, in the fetal tissues (Breed et al. 1997; Slotkin et al. 1998; Nyirenda et al. 2006). In rats, the persisting up-regulation of hepatic PEPCK activity associated with prenatal glucocorticoid overexposure is related, in part, to increased expression of two of its transcription factors, Tcf1 and Hnf4a (Nyirenda et al. 2006). Indeed, glucocorticoid-stimulated upregulation of these two transcription factors tracks from late fetal life and precedes any detectable increase in hepatic PEPCK enzyme activity (Nyirenda et al. 1998, 2006; Franko, 2007). Again, up-regulation of Hnf4a mRNA abundance in fetal rat liver by maternal dexamethasone administration is associated with a switch in promotor usage from the fetal to the adult isoform (Nyirenda et al. 2006). Hormones, therefore, operate at several molecular levels to produce an epigenome that reflects the environmental conditions prevailing during intrauterine development.
Hormones have an important role in modifying fetal development in relation to environmental conditions. They can signal the type, severity and duration of different environmental cues as well as the gestational age and maturity of the fetus at the onset of the challenge (Fig. 5). They affect development of fetal tissues both directly and indirectly by changes in placental phenotype (Fig. 5). They act to alter gene expression, hence protein abundance, in a wide range of fetal tissues, which has consequences for many physiological systems both before and after birth (Fig. 5). By producing an epigenome specific to the prevailing intrauterine conditions, hormones lead to phenotypical diversity, with potential health and evolutionary consequences (Gluckman et al. 2007; Uller, 2008). Hormones, therefore, act as epigenetic signals in developmental programming and allow transgenerational transmission of non-genomic factors important in developing the optimal phenotype for survival to reproductive age.
We would like to thank all the members of the Department of Physiology, Development and Neuroscience who have helped with these studies over the years. We are also indebted to the BBSRC, the Royal Society, the Wellcome Trust and the Horserace Betting Board for their funding.