Developmental programming of obesity in mammals


Corresponding author P. D. Taylor: Maternal & Fetal Research Unit, KCL Division of Reproduction & Endocrinology, 10th Floor North Wing, St Thomas' Hospital, London SE1 7EH, UK. Email:


Converging lines of evidence from epidemiological studies and animal models now indicate that the origins of obesity and related metabolic disorders lie not only in the interaction between genes and traditional adult risk factors, such as unbalanced diet and physical inactivity, but also in the interplay between genes and the embryonic, fetal and early postnatal environment. Whilst studies in man initially focused on the relationship between low birth weight and risk of adult obesity and metabolic syndrome, evidence is also growing to suggest that increased birth weight and/or adiposity at birth can also lead to increased risk for childhood and adult obesity. Hence, there appears to be increased risk of obesity at both ends of the birth weight spectrum. Animal models, including both under- and overnutrition in pregnancy and lactation lend increasing support to the developmental origins of obesity. This review focuses upon the influence of the maternal nutritional and hormonal environment in pregnancy in permanently programming appetite and energy expenditure and the hormonal, neuronal and autocrine mechanisms that contribute to the maintenance of energy balance in the offspring. We discuss the potential maternal programming ‘vectors’ and the molecular mechanisms that may lead to persistent pathophysiological changes resulting in subsequent disease. The perinatal environment, which appears to programme subsequent obesity, provides a potential therapeutic target, and work in this field will readily translate into improved interventional strategies to stem the growing epidemic of obesity, a disease which, once manifest, has proven particularly resistant to treatment.

The pandemic of adult and childhood obesity

The global epidemic of obesity and associated diseases is having a major impact on human morbidity, mortality and quality of life, and is a major drain on healthcare resources. In the USA, estimates suggest that one-third of adult women are clinically obese (American College of Obstetricians and Gynecologists, 2005). In the UK, recent figures from the Department of Health based on the Health Survey for England 2003 predict that by 2010, one in three adults will be obese, whilst 27% of girls (11–15 years old) will be obese approaching child-bearing age (Department of Health, 2006). Obesity in pregnancy represents a special problem not only because of adverse effects on maternal health and pregnancy outcome (Scialli, 2006) but because of growing evidence for persistent and deleterious effects on the developing child. In the USA, 18–35% of pregnant women are clinically obese (Ehrenberg et al. 2004); similarly, estimates from the UK suggest that 19% of pregnant women are obese at the time of registering for primary care (Kanagalingam et al. 2005). With all the attendant complications of obese pregnancy (hypertensive disorders, gestational diabetes, thromboembolic events and Caesarean section), the increased cost of prenatal care alone in obese women can be as much as 16-fold compared with normal weight subjects (Galtier-Dereure et al. 2000), and the added cost of neonatal intensive care for infants compromised by these disorders also needs to be considered. However, it is the potential social and economic cost in terms of the health of future generations that may present the biggest burden.

The demographic shift of populations towards a more obese phenotype in a relatively short period, just one or two generations, argues against a major genetic contribution in favour of environmental or epigenetic mechanisms. Recent evidence suggests that the prevention of childhood and adult obesity may need to begin even before conception (Magarey et al. 2003; Danielzik et al. 2004; Salsberry & Reagan, 2005). This review highlights the evidence that nutritional and environmental imbalances in utero and early postnatal life may make a significant contribution to the obesity epidemic.

The developmental origins of health and disease hypothesis

The idea that maternal well-being may have important implications for the future health of the baby is not new, and social and geographical inequalities in population health have been the subject of debate since Victorian times. However, it was not until the epidemiological studies of Anders Forsdahl in Norway that a causative link was suggested between early life environmental factors and subsequent disease (Forsdahl, 1977). Years later, David Barker and colleagues in the UK suggested that poverty, poor nutrition and the general health of mothers produced not only high rates of infant mortality but also a lifelong risk of coronary heart disease (Barker & Osmond, 1986). Then followed the studies of UK cohorts, which detailed the relationship between low birthweight and subsequent adult cardiovascular disease (CVD), which led Barker and colleagues to hypothesize that adverse environmental factors in early life cause disruption of normal growth and development, leading to a more susceptible adult phenotype prone to CVD (Barker et al. 1990, 1993; Barker, 1995). Studies in other countries then followed (Yajnik et al. 1995; Ravelli et al. 1999; Roseboom et al. 2000; Yajnik, 2000), and the ‘developmental origins of health and disease’ (DOHaD) hypothesis grew in credibility and strength.

Developmental programming of adulthood obesity

Childhood and adult obesity are amongst the cardiovascular risk factors now considered to be ‘programmed’ by early life and, perhaps counter-intuitively, babies subjected either to early life nutritional deprivation or to an early environment over-rich in nutrients appear to be at risk. Supportive evidence includes the observation of a ‘U-shaped’ curve which relates birthweight to risk of adult obesity (Curhan et al. 1996).

The earliest studies relating early life undernutrition to later development of obesity were those from victims of the Dutch Hunger Winter 1944-5, which showed that females born to women exposed to famine in early gestation were more likely to be obese (increased body mass index (BMI) and waist circumference) in later life (Ravelli et al. 1976, 1999). These studies suggested that perturbations of central endocrine regulatory systems established in early gestation may contribute to the development of obesity in later life. This was followed by studies from UK cohorts showing associations between low birthweight with a raised BMI in adulthood (Law et al. 1992; Phillips et al. 2000; Sayer et al. 2004). In response to the criticism that BMI may be a relatively poor surrogate index of obesity (BMI is more an index of heaviness than fatness), Kensara et al. (2005) have shown recently in one of the UK cohorts that adulthood adiposity as assessed by whole body dual energy X-ray absorptiometry (DEXA) scanning also relates to low birthweight (Kensara et al. 2005). A study of 447 monozygotic female twin pairs (18–34 years of age) from the East Flanders Prospective Twin Survey also showed a negative association between birth weight and unhealthy body composition characteristics, including waist-to-hip ratio, skin-fold thickness and reduced muscle mass, but interestingly not BMI (see above). Monozygotic twin studies of this kind control for genetic and maternal factors which may confound the relationship between birth weight and adult adiposity (Loos et al. 2001).

Babies which are born small and then show rapid catch-up growth have in a recent systematic review been shown to be more obese in later life (Ong, 2006). Some authors consider that catch-up growth in the first few weeks of postnatal life is particularly disadvantageous (Stettler et al. 2003, 2005), whereas others suggest that low birthweight children who grow excessively in later childhood are also particularly at risk of later obesity (Rolland-Cachera et al. 1984; Ong et al. 2000). Lucas, Singhal and colleagues, predominantly from studies in premature low birth weight infants, suggest that rapid catch-up growth in early infancy, associated with the feeding of formula milk rich in macronutrients, is particularly deleterious for future metabolic and cardiovascular health in adolescence (Lucas, 1990; Singhal et al. 2002, 2003a, 2004; Singhal, 2006), an observation recently strengthened by the studies in USA populations by Stettler and colleagues which showed an increase in adult overweight status associated with formula feeding and rapid weight gain in the first week of life owing to overfeeding (Stettler et al. 2003, 2005). A number of recent meta-analyses clearly show a greater propensity to adulthood obesity amongst infants who were formula-fed as opposed to breast-fed (Harder et al. 2005; Owen et al. 2005).

In view of the rising birth weight and the incidence of large for gestational age (LGA) babies in developed countries (Surkan et al. 2004), the prevalence of obesity in pregnancy and the association with gestational diabetes, there is increasing interest in the potentially detrimental influence of a maternal hypernutritional status and raised birth weight on the risk of disease in childhood and beyond (Catalano, 2003; Catalano et al. 2003; Oken & Gillman, 2003; Ehrenberg et al. 2004). Suggested by studies showing a U- or J-shaped relationship between birth weight and later obesity (Curhan et al. 1996; Rogers, 2005), children of obese women who are at increased risk of diabetes in pregnancy are themselves more likely to become overweight and develop insulin resistance in later life (Dorner & Plagemann, 1994). The 25–36% increase in maternal BMI over the last decade recorded in the Swedish birth registry has translated to an approximately 25% increase in the incidence of LGA babies (Surkan et al. 2004). Pre-pregnancy obesity exerts an early tendency to overweight which is perpetuated as the child grows (Salsberry & Reagan, 2005), and high normal weight status in childhood further predicts overweight in adulthood (Baird et al. 2005; Field et al. 2005). Whilst this may represent in part a genetically inherited disorder, a comprehensive study of over 150 000 women reported in the Lancet recently that even moderate weight gain between successive pregnancies resulted in significant increase in adverse pregnancy outcomes, including LGA birth (Villamor & Cnattingius, 2006). Moreover, sibling pair studies from the Pima Arizonian Indians, in which siblings were discordant for maternal diabetes, strongly suggest an offspring diabetic and obesogenic trait acquired in pregnancy (Dabelea et al. 2000; Dabelea & Pettitt, 2001). Similarly, increased risk of metabolic syndrome has been reported among children who were born LGA to mothers with either gestational diabetes or obesity (Boney et al. 2005). Maternal hyperglycaemia is usually thought of as a predominant driving influence in overweight birth, but this and another recent study in diabetic pregnancies suggest that prepregnancy maternal body composition and hitherto unrecognized factors may have equally important influences in pregnancies complicated by insulin-requiring gestational or type 2 diabetes (Sacks et al. 2006).

High birth weight is undoubtedly associated with a higher adult BMI, but this may be explained by increased muscle mass rather than fat mass (Singhal et al. 2003b), and contemporary studies are now addressing this by detailed measurement of body composition in the child in relation to maternal obesity and gestational diabetes. Thus, Catalano and colleagues have shown that children of mothers with gestational diabetes have a greater neonatal fat mass than children of the same birth weight whose mothers did not have diabetes in pregnancy (Catalano et al. 2003). The association between maternal obesity with or without gestational diabetes and the body composition of the developing child through to adulthood requires better definition and, although currently underway in some large cohorts, will take many years to elucidate, but given the current obesity epidemic will have important public health implications for the perpetuating cycle of obesity and related disorders.

Experimental models in animals

Prospective investigation of human cohorts, although of enormous value in determining associations between maternal nutritional status and offspring outcome, is complex, expensive and confounded by the influence of uncontrollable variables of genetic and environmental origin. The use of models in experimental animals has provided a supplementary approach that has proven very informative. It appears that the developing cardiovascular system and pathways of glucose and fat metabolism are particularly prone to perturbation in nutritional imbalance, although increasingly other biological systems and pathways are implicated as interest has intensified. The focus on maternal and fetal undernutrition in the human cohorts is reflected in the abundance of reports in experimental animals concerned with the consequence of maternal nutrient deprivation and, hitherto, a relative paucity of those addressing effects of overnutrition.

Models of maternal undernutrition Experimental studies in animals support the concept that maternal undernutrition during critical periods can programme adipocyte metabolism and fat mass to give rise to later obesity, especially when challenged postnatally with a hypernutritional diet (Bispham et al. 2005; Budge et al. 2005; Stocker et al. 2005), as predicted by the thrifty phenotype hypothesis (Hales & Barker, 2001). Low birth weight sheep have a higher relative fat mass as neonates compared with higher birth weight offspring (Greenwood et al. 1998), and fetal sheep also show increased fat deposition when exposed in utero to maternal nutrient restriction in early pregnancy, during the period of maximum placental growth (days 28–80 of gestation), suggesting that the timing and duration of nutrient restriction in pregnancy is crucial to the programming of fat mass (Budge et al. 2004, 2005; Symonds et al. 2004). Altered adipocyte function was also suggested in this model by the increased mRNA expression of uncoupling protein 2 (UCP2) and peroxisome proliferator-activated receptors (PPAR-α), which are both intimately involved in fat metabolism. There is also evidence that a protein-restricted (50%) diet in utero programmes susceptibility to obesity in adult rats and mice (Ozanne et al. 2004; Zambrano et al. 2006), although often not without a postnatal dietary challenge (Bellinger et al. 2006). However, the increase in the nutritional plane between pre- and postnatal environment appears significant, even in the transition from low protein in utero to a normal diet postnatally (Ozanne et al. 2004; Zambrano et al. 2006). Moderate (50%) and severe (70%) prenatal caloric restriction (i.e. 30% of ad libitum intake) is also associated with greater fat deposition in rodents when presented with a hypercaloric or high fat diet (Vickers et al. 2000, 2003; Kind et al. 2003). Interestingly, a 50% restriction of normal vitamin intake in pregnancy has been reported to increase body fat content in adult offspring, associated with raised triglycerides and lower lean body mass, and may therefore contribute to the altered body composition reported in severe global dietary restriction studies (Venu et al. 2004). One also cannot exclude a role for prenatal stress as a programming vector in starvation models of this severity (Weinstock, 2001). Prenatal stress (chronic variable stress) in mice experienced mid- to late gestation produced long-term effects on body weight in offspring from stress-sensitive dams, with male offspring being 15% heavier as adults at 6 months (Mueller & Bale, 2006).

Models of maternal overnutrition Relatively few studies have investigated long-term consequences of a maternal hypernutritional status during pregnancy or lactation on development of obesity in the offspring (Armitage et al. 2004). One of the earliest, in the baboon, showed that overfeeding in the preweaning period permanently increased adiposity through fat cell hypertrophy (Lewis et al. 1986); a gender-dependent effect in the females only. Guo & Jen (1995) reported increases in body weight and adiposity (fat pad weight) in weaned offspring of rats fed a 40% fat (by weight) diet during pregnancy and lactation. The increased rates of cannibalism recorded in this study are a common but often unreported feature of fat feeding studies (Taylor et al. 2003) and may relate to a failure of lactation associated with very high fat diets. We have also reported increased body weight and a twofold increase in the visceral fad depot weight in adult offspring of fat-fed dams (24% fat by weight; lard) from 6 months of age (Khan et al. 2003; Khan et al. 2004; Taylor et al. 2005). An additional postnatal dietary challenge of continued fat feeding into adulthood results in increased body weight at 6 months in males only, associated with a significant increase in gross energy intake, but with no significant increase in adiposity, suggesting a ‘predictive adaptive response’ to the maternal high fat diet (Khan et al. 2004). Programmed changes in offspring body weight and adiposity to a range of different maternal dietary fat interventions have been reported, including 10% fish oil (Amusquivar et al. 2000), 2% linoleic acid (Bee, 2000), altered ratios of linoleic to α-linolenic fatty acids in which Soybean oil (n-6/n-3) or linseed oil (n-3) were used as the additives (Korotkova et al. 2002, 2005).

Developmental programming in the suckling period Evidence in rats from cross-fostering studies in our obesogenic models of fat feeding (Khan et al. 2005) and other models of postnatal overfeeding (Plagemann et al. 1999a; Schmidt et al. 2001; Plagemann, 2005) clearly shows that the suckling period (postnatal day 0–21) is critical for developmental programming of increased adiposity, hyperleptinaemia and hypertension in adult offspring fed a normal diet after weaning. Cross-fostering of obesity-resistant pups to genetically obese dams results in a diet-induced increase in adiposity in later life (Gorski et al. 2006). Overfeeding during the suckling period in rodents, by rearing them in small litters, produces hyperphagia and obesity (Schmidt et al. 2001), and Srinivasan et al. (2003) have shown that artificial rearing of rat pups on a high carbohydrate diet during the suckling period leads to increased adiposity in the offspring and that this phenotype is transferred to the second generation. Plagemann and colleagues have shown that maternal gestational diabetes, which effectively produces a similar overnutrition through hyperglycaemia, also programmes adult obesity (Plagemann et al. 1999a; Franke et al. 2005) and that transient neonatal hyperinsulinaemia during critical postnatal periods is responsible for permanent programming of the obese phenotype from early adulthood (Harder et al. 1999). There is also evidence for a similar effect of leptin in mice (Yura et al. 2005), although evidence is somewhat contradictory in rats (Schmidt et al. 2001). Leptin administration to either protein-restricted dams (Stocker et al. 2005) or neonates from calorie-restricted dams (Vickers et al. 2005) prevents the development of an obese phenotype in the adult offspring. Clearly the timing and duration of exposure is critical, but both insulin and leptin, perhaps the two most essential component signals in regulation of energy balance, may also play a role in early life programming of energy homeostasis.

Mechanisms mediating the developmental programming of obesity

Programming of obesity could occur by permanent alteration of one or more relevant pathways during early development. Given that obesity is fundamentally a disorder of energy balance, in which energy intake exceeds energy expenditure, these pathways may include appetite regulation and altered energy expenditure, including altered tissue metabolism and physical activity.

Altered appetite control The pivotal role of the hypothalamus in the control of food intake is well recognized, but only relatively recently was it realized that it may play an essential role in developmental programming associated with neonatal exposure to a hypercalorific diet (Cripps et al. 2005; McMillen et al. 2005, 2006; Mulhausler et al. 2005; Plagemann, 2005, 2006). Hypothalamic nuclei continue to differentiate until day 20 of postnatal life in rodents (Grove et al. 2005); this period is therefore critical to the study of the expression of key regulatory hypothalamic neuropeptides and receptors, the expression of which may be permanently ‘programmed’ by maternal and fetal dietary-related factors. Whilst investigation of postnatal neuronal development in these models is of direct relevance to altricial species, it is also pertinent to man, since extensive neuronal development is likely to occur in the suckling period in precocial species also, in which CNS and hypothalamic maturation begins in utero but continues in early postnatal life (Grove et al. 2005).

Overfeeding of neonatal rat pups by rearing them in small litters (4 pups) leads to an increase in offspring adiposity (Bassett & Craig, 1988; Plagemann et al. 1999b; Davidowa & Plagemann, 2004). Plagemann's group and others (Velkoska et al. 2005) have shown that overfed neonatal rats, reared in small litters, develop hyperphagia, accelerated fat deposition and body weight gain, in association with peripheral hyperleptinaemia and central leptin resistance at the level of the arcuate nucleus (ARC; Schmidt et al. 2001). These studies suggest that hypothalamic ‘malprogramming’ is likely to occur in the suckling period (Plagemann, 2005). Kozak et al. (2000) have also directly implicated the hypothalamus in programming of obesity by showing that adult offspring of dams fed a highly fat-rich diet (55% margarine) have an exaggerated feeding response to injection of neuropeptide Y (NPY) into the lateral brain ventricle, eating twice as much as control animals (Kozak et al. 2000). Prenatal overnutrition in sheep also results in altered appetite regulation in the early postnatal period, with impaired signalling of the appetite-inhibiting cocaine and amphetamine-regulated transcript (CART; Muhlhausler et al. 2005, 2006).

A central role for leptin in hypothalamic programming has also been suggested by a study in which subcutaneous administration of leptin on postnatal days 3–13 reversed the hyperphagia and obesity in adult offspring of rats subjected to prenatal undernutrition (Vickers et al. 2005), although similar investigations have not been carried out in hypernutritional models. Conversely, Yura and colleagues showed that prenatal exposure to maternal undernutrition resulted in a premature leptin surge in mice at 8–10 days versus 16 days in control animals, which was associated with adult obesity. Mimicking this premature leptin surge in control animals by exogenous leptin treatment also produced an obese adult phenotype (Yura et al. 2005). A ‘landmark’ report has proposed a potential mechanism. In an investigation of the developing neonatal rat hypothalamus, leptin was found to promote neuronal outgrowth from the ARC to the paraventricular nucleus (PVN) during the lactation period, thus potentially ‘hard-wiring’ the hypothalamic appetite regulatory system. Leptin appears to affect differentially the two opposing pathways controlling energy intake, favouring the development of appetite stimulatory NPY and agouti-related protein (AgRP) projections from the ARC to the PVN over neurite outgrowth from appetite inhibitory α-melanocyte-stimulating hormone (α-MSH)-containing neurones (derived from pro-opiomelanocortin, POMC; Bouret et al. 2004; Bouret & Simerly, 2006; Horvath & Bruning, 2006). Paradoxically, in the genetically leptin deficient (ob/ob) mice, lack of leptin in the neonatal hypothalamus preferentially suppresses outgrowth of NPY and AgRP fibres to the PVN, which might be expected to hardwire an anorexigenic hypothalamus, rather than producing a hyperphagic phenotype. However, it is likely that the NPY- and AgRP-containing fibres that fail to project to the PVN remain in the ARC, where they may exert their orexogenic effects through inhibition of POMC-containing neurones (Horvath & Bruning, 2006), effectively inhibiting appetite suppression.

Altered adiposity and adipocyte metabolism Adipogenesis, which begins in utero and accelerates in neonatal life, is a prime candidate for developmental programming. In humans, following a brief respite in fat deposition in infancy, it accelerates rapidly again at about 6 years of age, and it may be relevant that premature onset of this adipose tissue mass (before 5.5 years of age) in childhood is associated with increased adult obesity (Eriksson et al. 2002). It is unclear how maternal influences on fetal adipogenesis may determine the timing of the ‘adiposity rebound’ but there is evidence for the programming of adipocyte morphology and metabolism. Unlike many tissues, adipose tissue has the potential for unlimited growth but, importantly, diet-induced increases in fat cell number are apparently irreversible (Faust et al. 1978; Corbett et al. 1986). A direct influence on offspring adipocyte hyperplasia or hypertrophy by maternal hypernutritional diet per se might be anticipated, since glucose is the primary metabolic precursor for lipid synthesis; indeed, direct infusion of glucose into the ovine fetus is accompanied by a parallel increase in fat mass (Stevens et al. 1990), but whether this persists into adult life is not established. Adipocyte hypertrophy also occurs in weanling rats from dams fed a highly palatable diet in pregnancy and lactation (Bayol et al. 2005). Persistent alteration in expression of any of the many proteins which influence adipocyte development and lipolysis, e.g. peroxisome proliferator-activated receptor-γ (PPAR-γ), could result in a permanent influence on adipocyte proliferation and hypertrophy.

Altered expression of adipocyte proteins has already been reported in response to maternal undernutrition, e.g. adipocytes of prenatally nutrient-restricted lambs show increased expression of 11β-hydroxysteroid dehydrogenase type-1 (11β-HSD-1); which could lead to increased cortisol exposure and proliferation of adidocytes(Reynolds et al. 2001). More recently, early nutrient restriction in sheep has been reported to increase expression of both 11β-HSD-1 and glucocorticoid receptor (Gnanalingham et al. 2005). In human studies, 11β-HSD-1 expression in subcutaneous fat correlated with BMI, suggesting a potential therapeutic target for selective antagonists in man (Wake & Walker, 2004). Programming of PPARs peroxisome proliferator-activated receptors-α-γ (PPAR-α) has been reported in liver of offspring prenatally exposed to a protein-deficient diet and was related to altered methylation status (Lillycrop et al. 2005); these are therefore potentially important molecular targets in developmental programming of obesity and largely underexplored.

Altered physical activity Few studies to date have investigated activity of offspring prenatally exposed to maternal obesity or a maternal hypernutritional status; however, severe maternal undernutrition in pregnancy followed by a richer nutritional plane postnatally results in offspring with programmed hyperphagia and reduced locomotor activity, associated with profound obesity (Vickers et al. 2000, 2003). Interestingly, a programming model in which mice were fed a high polyunsaturated fat diet throughout pregnancy resulted in offspring with increased locomotor activity in a swim test (Raygada et al. 1998). Conversely, we have reported in a rat model that a maternal diet rich in saturated fats derived from animal lard programmes reduced locomotor activity, as assessed by radio telemetric recording in the animals' home cage (Khan et al. 2003). It appears, therefore, that the fatty acid composition of the maternal diet may be crucial in programming activity levels, hence energy expenditure.

Altered metabolic rate Effects of maternal diet on offspring basal metabolic rate are largely unexplored, and hitherto only attempted in undernutrition models. Yura and colleagues have reported blunted diet-induced thermogenesis in adult offspring from a mouse model of maternal nutrient restriction in pregnancy (Yura et al. 2005) and reported reduced oxygen consumption and carbon dioxide production compared with control animals, when maintained on a high fat diet. Administration of leptin during pregnancy and lactation in protein-restricted dams produces offspring with an enhanced metabolic rate that are resistant to diet-induced obesity (Stocker et al. 2004). In sheep, maternal protein restriction programmes abnormal thyroid function in the offspring, which will influence basal metabolic rate (BMR; Rae et al. 2002). In addition to altered fat mass, adipose tissue mitochondrial function is also altered in fetal sheep in this model and related to altered thermogenesis (Symonds et al. 2004).

Molecular mechanisms that may underlie the developmental programming of obesity

Any plausible molecular mechanism for developmental programming of an obese adult phenotype must explain how very early environmental stress can set in train persistent molecular changes that will cause pervasive damaging affects at a later date.

Altered methylation status Persistent epigenetic changes to the methylation status of nuclear DNA (nDNA) have been proposed, and studies from several animal models support this hypothesis (Waterland & Jirtle, 2004; Lillycrop et al. 2005; Blewitt et al. 2006; Waterland, 2006). In addition to the potential neurotrophic action of leptin, which may programme appetite regulatory centres in the developing hypothalamus (Bouret et al. 2004), it is also possible that programmed alteration in methylation status of hypothalamic genes may also contribute to the programming of appetite and energy expenditure. Altered methylation status in very early embryonic development may also contribute to the obese phenotype associated with embryo transfer and cloning (Tamashiro et al. 2002; Sakai et al. 2005), which has implications for human IVF programmes.

Altered sympathetic innervation and selective leptin resistance Several developmental programming models also provide evidence for altered sympathetic control (Khan et al. 2003; Alexander et al. 2006), and specific programming of the neonatal hypothalamus by hyperleptinaemia may also contribute to the increased sympathetic tone that is thought to underlie obesity-related hypertension. In established obesity, with high circulating plasma leptin, selective leptin resistance at the level of the hypothalamus has been proposed to result, on the one hand, in the attenuation of the anorectic and weight-reducing actions of leptin, whilst on the other hand, a preservation of leptin's central pressor actions contributes to an elevation of blood pressure (Haynes, 2005; Rahmouni et al. 2005). Selectivity in leptin resistance may result in a reduced ability of leptin to activate signalling pathways in the appetite regulatory areas of the ARC, with preservation of leptin's sympatho-excitatory action in the cardiovascular-related ventromedial nucleus (of the hypothalamus) (VMN) and dorsomedial nucleus (of the hypothalamus) (DMN) (Marsh et al. 2003). Given the potential neurotrophic role of leptin during neonatal hypothalamic plasticity, it seems likely that maternal and/or neonatal hyperleptinaemia in the immediate postnatal period could programme selective leptin resistance and therefore a propensity to both obesity and obesity-related hypertension. Selective resistance may arise though altered leptin receptor expression/density but may also relate to divergent signalling pathways downstream from the leptin receptor, such as the signal transducer and activator of transcription 3 (STAT3), for example (Cheng et al. 2002; Howard et al. 2004).

Epigenetic programming of the mitochondrial and related nuclear genome The developmental characteristics of mitochondria are particularly suited for translating the early stress associated with developmental programming into cellular dysfunction observed in later life. Levels of mitochondrial DNA (mtDNA) are exquisitely sensitive to environmental stress, and a suboptimal environment can cause a reduction in the quantity and quality of mtDNA via increase in random mutation rates (Graziewicz et al. 2006). We and others have reported evidence for mitochondrial dysfunction in animal models of developmental programming (Song et al. 2001; Park et al. 2003; Taylor et al. 2005). Recent reports from human studies (Ruiz-Pesini et al. 2004; Wilson et al. 2004; Ritov et al. 2005) and from experimental models of obesity and diabetes (Lamson & Plaza, 2002; Wisloff et al. 2005) imply that altered mitochondrial function is at least contributory, if not causal, in the development of obesity-related disorders. Mutations in mtDNA which can be tolerated for many years before reaching a damaging threshold level (Chinnery et al. 2002) could therefore influence the long-term function of mitochondria with implications, in particular, for energy expenditure. In a recent study, rats were selectively bred for low or high intrinsic exercise capacity. After 11 generations, the low exercise group were obese, hypertensive and insulin resistant. Notably, these animals demonstrated low aerobic oxidative capacity, which was associated with reduced amounts of nuclear DNA encoded transcription factors required for mitochondrial biogenesis (Wisloff et al. 2005).

Uncoupling proteins (UCPs) Uncoupling proteins of the inner mitochondrial membrane act by uncoupling the mitochondrial respiratory chain involved in oxidative phosphorylation and ATP production. In brown adipose tissue of rodents, UCP-1 dissipates the energy of substrate oxidation as heat (Nicholls & Locke, 1984). UCP-2 is implicated in the control of energy regulation (Boss et al. 2000), whereas UCP-3, found in adipose tissue and skeletal muscle, is thought to be involved in fatty acid metabolism (Wang et al. 2003) and may protect against obesity by preventing accumulation of tryglycerides (Walder et al. 1998; Costford et al. 2006). Transgenic mice overexpressing UCP-3 in skeletal muscle show increased metabolic rate and marked reduction in adipose tissue mass (Clapham et al. 2000). Pharmacological manipulation of UCP expression and their modulators may therefore present potential therapeutic targets in the prevention of obesity in the future.

A final common pathway?

Evidence from both epidemiological and animal studies now suggests that the programming of obesity can arise from environmental influences occurring in embryonic through to neonatal life and early childhood. Studies employing various animal models now suggest that a host of different hormonal and dietary insults in utero and in early postnatal life seem to converge on a common phenotype of hyperphagia, obesity with altered adipocyte function and altered physical activity. Whilst the programming of obesity is undoubtedly a multifactorial process, the diversity of models with a common end-point might suggest some common pathways. Certainly there appears to be a role for altered adipocyte development and glucocorticoid signalling, but it is tempting to speculate that the plasticity of the hypothalamus in late pregnancy and early postnatal life is key to the programming of appetite and metabolism towards establishing an elevated body weight set point, which may or may not be adjustable over time (Keesey & Hirvonen, 1997).

Public health messages and possible interventions

Obesity in pregnancy clearly carries with it not just increased risk for the pregnancy outcome, as was recently reported in a study of 150 000 Swedish women (Villamor & Cnattingius, 2006), but also risks for the future health of the child or, in public health terms, the health of the next generation. The long-term goal must be to reduce the incidence of obesity in pregnancy and increase public awareness as to the importance of a balanced diet before and during pregnancy. A recent American College of Obstetricians and Gynecologists committee on obesity in pregnancy advocated that obstetricians should provide preconception counselling and education about the possible complications and should encourage obese patients to undertake a weight reduction programme before attempting pregnancy (American College of Obstetricians and Gynecologists, 2005). There are currently no analogous guidelines in the UK. Remarkably, no studies have looked at prepregnancy weight loss as a preventative measure. Recent research, however, suggests that weight gain between pregnancies should also be avoided, and women should aim to regain their prepregnancy weight before conceiving again (Villamor & Cnattingius, 2006). The Public Affairs Committee of the Teratology Society similarly recommends counselling of women on appropriate caloric intake and exercise and education about childhood nutrition and breast-feeding (Scialli, 2006). Whilst breast-feeding is widely acknowledged to have beneficial effects on the prevention of obesity in the general population, as a cautionary note, there is evidence that prolonged duration of breast-feeding may promote obesity (Harder et al. 2005). Similarly, there is evidence that diabetic breast milk and milk from obese mothers may also be obesogenic to the developing child (Rodekamp et al. 2005). To avoid rapid growth in infancy, formula feeds and their delivery may need to be optimized. From a dietary perspective, future research and education should focus on the identification of maternal nutritional insults that are the vectors for programming, i.e. the role of fats versus carbohydrates and saturated versus unsaturated fats. Medically, the more effective management of maternal obesity and gestational diabetes, and specifically the control of maternal hyperglycaemia, hyperinsulinaemia and hyperleptinaemia both before and during pregnancy would contribute to the prevention of obesity programming. Future pharmacological interventions may include centrally acting drugs that inhibit food intake but also peripherally acting drugs that modify metabolism and energy balance, e.g. 11β-HSD-1 antagonists (Wang et al. 2006). PPAR-α and PPAR-γ agonists, including the glitazones, may improve insulin resistance (Guo & Tabrizchi, 2006) and improve mitochondrial biogenesis in adipose tissue (Bogacka et al. 2005), offering the potential of dual PPAR agonist therapies to treat obesity risk factors (Gervois et al. 2004). There is also potential for novel drugs targeting the signalling cascade involved with adipocyte differentiation (Rodriguez et al. 2006). However, until the safety and efficacy of such drugs can be demonstrated, dietary and lifestyle interventions provide the only recourse, especially in pregnancy. In terms of possible interventions in childhood, there is the potential for identification of children at risk by assessment of growth/BMI and early identification of infanthood adiposity but also for potential biomarkers of future obesity with early lifestyle and pharmacological intervention in those most at risk. Research in this field offers a real possibility of strategic intervention to prevent the swelling tide of obesity in future generations.