Fetal programming hypothesis

In 1977, the association between the quality of early-life environment and the future risk of disease in later life was first proposed by Forsdahl⁽ Reference Forsdahl ^{16 )}, who discovered that infant mortality rate was positively associated with an increased risk of CVD in middle age. Subsequently, Barker et al. ⁽ Reference Barker, Winter and Osmond ^{17 )} found an inverse relationship between birth weight and increased CVD mortality in 1989. The effects of early-life nutrition on the increased risk of metabolic diseases were most clearly shown in studies of the Dutch famine during the winter of 1944. These studies showed that individuals whose mothers were exposed to the famine in the late stage of gestation had lower birth weight and an increased risk of obesity, CVD, insulin resistance and hypertension in later life compared with unexposed individuals⁽ Reference Painter, Roseboom and Bleker ^{18 )}. What is more, apart from nutrition deficiency in early life, overnutrition is also associated with a significantly increased susceptibility to metabolic diseases such as obesity, hypertension, atherosclerosis, insulin resistance and diabetes mellitus⁽ Reference Curhan, Willett and Rimm ^{19 )}. It can involve some organs related to glucose metabolism, including liver, pancreas, adipose tissue and skeletal muscle through different mechanisms (Fig. 3). Nutrition during early life has long-term phenotypic effects. This idea was first proposed by professor Barker and Hales in the 1990s, known as ‘fetal programming hypothesis’⁽ Reference Hales, Barker and Clark ^{20 )}. After more than 20-year long-term research, it has now been accepted worldwide.

Fig. 3 Fetal programming hypothesis in relation to glucose metabolism. Nutrition during early life has long-term phenotypic effects along with ageing, such as impaired glucose metabolism. This idea was first proposed by professor Barker and Hales in the 1990s, known as ‘fetal programming hypothesis’. It has been widely accepted that the essence of this hypothesis is epigenetic mechanism, such as DNA methylation. It can involve some organs related to glucose metabolism, including liver, pancreas, adipose tissue and skeletal muscle through different mechanisms. A colour version of this figure can be found online at http://www.journals.cambridge.org/bjn.

Early-life nutrition and altered epigenetic modifications

Epigenetic events are crucial for early development. However, it can be influenced by some environmental factors, potentially programming the genome for later adverse health outcomes. Gestation is the critical time window for maternal nutrition to affect the offspring. Early-life nutrition can induce persistent DNA methylation. More specifically, both undernutrition and overnutrition will bring about epigenetic modifications during early life, including embryonic development and the neonatal period. A number of clinical studies have shown that people born during the famine in The Netherlands still exhibited a lower methylation level of insulin-like growth factor 2 (IGF2) after 60 years, compared with those who were not exposed to prenatal famine⁽ Reference Tobi, Lumey and Talens ^{21 )}. It indicates that epigenetic modifications of some special genes affected by restricted periconceptional maternal nutrition can be persistent for decades. Another study has shown that children whose mothers took folic acid had 4·5 % higher methylation in the IGF2 gene differentially methylated region (DMR) than those whose mothers did not take folic acid. It also indicates an inverse independent association between IGF2 DMR and birth weight, which means that periconceptional folic acid use is associated with epigenetic changes in IGF2 in the child that may affect the intra-uterine programming of growth and development with consequences for health and disease throughout life⁽ Reference Steegers-Theunissen, Obermann-Borst and Kremer ^{22 )}. A recent cohort study has demonstrated that retinoid X receptor-α (RXRA) methylation had an independent association with sex-adjusted childhood fat mass, which could explain more than 25 % of the variance in childhood adiposity. Higher methylation of RXRA was associated with lower maternal carbohydrate intake in early pregnancy. These studies suggested that prenatal development is a substantial risk of metabolic diseases, such as diabetes, obesity and the metabolic syndrome⁽ Reference Godfrey, Sheppard and Gluckman ^{23 )}.

Enlightenment of monozygotic twins

Since epigenetic mechanisms are partly under genetic control⁽ Reference Chong and Whitelaw ^{24 )}, the role of epigenetic factors in disease aetiology could be confounded by genetic factors in unrelated individuals. Monozygotic (MZ) twins are well matched on their DNA sequence, thus providing an ideal model for epigenetic research. Type 1 diabetes mellitus shows an approximately 40 % concordant rate in MZ twins, suggesting the role of environmental factors and epigenetic modifications in the aetiology of the disease. Stefan et al. ⁽ Reference Stefan, Zhang and Concepcion ^{25 )} identified eighty-eight CpG sites displaying significant methylation changes in type 1 diabetes mellitus-discordant MZ twin pairs. Using a well-matched MZ twin cohort, Zhao et al. ⁽ Reference Zhao, Goldberg and Bremner ^{26 )} demonstrated that global DNA methylation in peripheral blood leucocytes was significantly associated with insulin resistance. On average, a 10 % increase in global DNA methylation was associated with a 4·55-unit increase in the homeostasis model assessment (HOMA) value. Some observational human and experimental animal studies have indicated that low birth weight was associated with an increased risk of metabolic diseases, such as obesity and T2DM. Although genetic factors might contribute to this association, studies of MZ twins showed that the twin who was lighter at birth had a more adverse metabolic profile in adulthood compared with its genetically identical co-twin, who was heavier at birth⁽ Reference Grunnet, Vielwerth and Vaag ^{27 )}. A population of twelve MZ twin pairs discordant for T2DM recognised a considerable non-genetic contribution to glucose metabolic disturbances in genetically identical individuals⁽ Reference Vaag, Henriksen and Madsbad ^{28 )}. An exploratory approach showed that the site of umbilical cord insertion into the placenta in monochorionic twins has the strongest positive association with methylation in all IGF2/H19 DMR⁽ Reference Loke, Galati and Morley ^{29 )}, which was crucial for prenatal growth. In addition, the epigenetic state at this locus was environmentally labile. A study of 53- to 80-year-old MZ twin pairs discordant for T2DM found that increased variation in repetitive LINE1 DNA sequence methylation was associated with more phenotypic dissimilarity such as BMI and 2 h plasma glucose. T2DM-related genes, including PPARGC1α encoding PPARγ co-activator 1α (PGC-1α) in muscle and hepatocyte nuclear factor-4α (HNF4α) in adipose tissue, have increased methylation in T2DM individuals⁽ Reference Ribel-Madsen, Fraga and Jacobsen ^{30 )}. A recent study on peri/postnatal epigenetic twins that contained a longitudinal cohort of 250 pairs of twins and their mothers confirmed that twins exhibit a wide range of epigenetic discordance at birth, and that within-pair birth-weight discordance correlated with epigenetic discordance in genes associated with glucose and lipid metabolism⁽ Reference Loke, Novakovic and Ollikainen ^{31 )}. Therefore, all these lines of evidence support an epigenetic mechanism for the DOHaD and fetal programming hypothesis.

Role of the placenta in developmental programming in the context of obesity and diabetes

The placenta performs multiple functions required for bringing forth new life. It is not only a very important organ to provide nutrients for the developing fetus, it also exchanges a wide range of endocrine signals, cytokines and growth factors with the mother and the fetus, thereby regulating intra-uterine development. It is also well known that the placenta have a high and prolific expression of imprinted genes, and recent research have suggested that epigenetic modifications such as DNA methylation in imprinting and non-imprinting genes can have an important impact on the health of the offspring. Modified patterns of placental DNA methylation have been reported in human pregnancies. For example, it has been shown that the maternally imprinted mesoderm specific transcript (MEST) gene is hypomethylated in the placenta obtained from the newborns of mothers with gestational diabetes mellitus (GDM). In addition, decreased blood MEST methylation was also observed in adults with morbid obesity, suggesting that intra-uterine exposure to GDM has long-lasting effects on the epigenome of the offspring⁽ Reference El Hajj, Pliushch and Schneider ^{32 )}. Another study showed that maternal glycaemia was associated with DNA methylation changes in leptin and adiponectin genes, suggesting that these epigenetic adaptations are probably functional and may have impacts on the long-term regulation of newborns’ energy metabolism⁽ Reference Houde, Hivert and Bouchard ^{33 )}.

Maternal energy restriction

Protein restriction is frequently used as a model for maternal malnutrition. A low-protein (LP) diet is associated with impaired fetal growth, the development of obesity, insulin resistance and diabetes in the offspring⁽ Reference Bocock and Aagaard-Tillery ^{34 )}. Many epigenetic modifications have been reported in offspring exposed to a maternal LP diet. One of the first studies showing the association between nutritional imbalances during intra-uterine development and epigenetic modifications was that feeding a LP diet (9 % casein v. 18 % casein) to rats during pregnancy resulted in global DNA hypermethylation in the liver of the fetuses⁽ Reference Rees, Hay and Brown ^{35 )}. Recent studies confirmed that maternal LP feeding during pregnancy might also result in locus-specific changes in DNA methylation. For example, feeding pregnant rats a protein-restricted diet induced hypomethylation of the glucocorticoid receptor (GR) and PPARα promoters in the liver of juvenile and adult offspring⁽ Reference Burdge, Slater-Jefferies and Torrens ^{36 )}. Also in Wistar rats, methylation of the PPARα gene was 20·6 % lower, with expression being 10·5-fold higher, and that of the GR gene was 22·8 % lower, with expression being 200 % higher in pups of dams that were fed a protein restriction diet throughout pregnancy⁽ Reference Lillycrop, Phillips and Jackson ^{37 )}. The HNF4α gene has been implicated in the aetiology of T2DM. Sandovici et al. ⁽ Reference Sandovici, Smith and Nitert ^{38 )} observed that maternal LP diet during pregnancy and lactation could lead to progressive epigenetic silencing of the entire HNF4α locus in rat pancreatic islets, which weakened the promoter–enhancer interaction and resulted in a permanent reduction in HNF4α expression. Maternal protein undernutrition during pregnancy and lactation in Balb/c mice affected the balance between food intake and energy expenditure in adults. Moreover, this nutritional stress resulted in the removal of methyls at CpG sites located in the promoter of leptin, which was associated with DNA hypomethylation of the leptin promoter in adipose tissue⁽ Reference Jousse, Parry and Lambert-Langlais ^{39 )}. Interestingly, the DNA methylation of the LP diet is not restricted to rodents. Male offspring of maternal LP diet of Meishan sows during pregnancy and lactation demonstrated higher GR binding to the mitochondrial DNA promoter, which was accompanied by lower cytosine methylation and hydroxymethylation on mitochondrial DNA promoter⁽ Reference Vo, Revesz and Sohi ^{40 )}. Similarly, piglets with maternal LP diet during pregnancy and lactation showed significantly lower body weight and liver weight at weaning. Hepatic activation of 3-hydroxy-3-methylglutaryl-CoA reductase (HMGCR) gene transcription in LP piglets was associated with promoter hypomethylation⁽ Reference Cong, Jia and Li ^{41 )}.

In addition to protein restriction, some other models of maternal energy restriction such as nutrient restriction can also affect the epigenome and the metabolism of the offspring. In one study, it has been shown that periconceptional undernutrition at least 5 months before conception to days 6 and 7 after conception was associated with a decrease in the adrenal mRNA expression of IGF2 and decreased methylation in the proximal CCCTC-binding factor (CTCF)-binding site in the DMR of the IGF2/H19 gene in the adrenal gland in offspring sheep, and it is associated with decreased expression of IGF2 in offspring along with an increase in the cortisol stress response in females⁽ Reference Zhang, Rattanatray and MacLaughlin ^{42 )}. In the fetal liver of sheep, maternal periconceptional restriction of vitamin B₁₂, folate, and methionine altered DNA methylation of 4 % of the 1400 CpG islands examined, and more than half of these changes were specific to male offspring⁽ Reference Sinclair, Allegrucci and Singh ^{43 )}. Bilateral uterine artery ligation of pregnant rats has been used to generate a model of fetal growth restriction. A genome-wide survey of male Sprague–Dawley rat offspring showed altered DNA methylation of approximately 1400 loci that occurred predominantly in intergenic regions in pancreatic islets. Moreover, one gene in particular, GTP cyclohydrolase 1 (Gch1), showed almost a 3-fold reduction in mRNA expression, associated with hypermethylation at a conserved intergenic site < 45 kb upstream of the gene itself⁽ Reference Thompson, Fazzari and Niu ^{44 )}. More importantly, these changes in DNA methylation may remain stable until adulthood, thus providing a molecular basis for DOHaD.

Maternal energy excess

Maternal energy restriction (LP diets) is not the only experimental model that suggests that maternal diet influences the epigenome of offspring during development. Nowadays, both the incidence of obesity during pregnancy and GDM are increasing along with improved living conditions. Therefore, scientists are increasingly concerned about the effect of overnutrition on human health. There is increasing evidence that energy excess in maternal diet is associated with an altered metabolic phenotype. Recently, studies have shown that maternal high-fat diet might modify DNA methylation and gene expression in the offspring. Maternal high-fat diet during gestation was associated with global and gene-specific promoter DNA hypomethylation, such as the dopamine reuptake transporter, the μ-opioid receptor (MOR) and preproenkephalin in the brain from the first-generation (F₁) offspring of C57BL/6J × DBA/2J hybrids. These changes might affect feeding behaviour, which can increase obesity and obesity-associated risk for the metabolic syndrome⁽ Reference Vucetic, Kimmel and Totoki ^{45 )}. During the postnatal period of the same model, the offspring of maternal high-fat diet showed increased binding of methyl CpG-binding protein 2 (MeCP2) to the MOR promoter in reward-related regions of the brain, which can repress the transcription of the MOR gene⁽ Reference Vucetic, Kimmel and Reyes ^{46 )}. A recent study has shown that the offspring whose mothers fed on a high-fat diet during late gestation has increased weight accrual and food intake, exhibiting insulin resistance and hyperlipidaemia. Furthermore, the offspring mice emerged with increased methylation of adiponectin and leptin receptor, and decreased methylation of leptin genes⁽ Reference Khalyfa, Carreras and Hakim ^{47 )}. Modified patterns of placental DNA methylation have also been reported in animal models exposed to an altered maternal environment. In one study, it has been shown that exposure to a high-fat diet during pregnancy triggered global and gene-specific DNA methylation in mouse placenta alterations in a manner dependent on the sex of the conceptus⁽ Reference Gallou-Kabani, Gabory and Tost ^{48 )}. In Wistar rats, it has also been demonstrated that maternal intake of essential fatty acids during pregnancy affected the global levels of placental DNA methylation⁽ Reference Kulkarni, Dangat and Kale ^{49 )}.

Therefore, it is currently unknown whether these genes may contribute to the development of insulin resistance or diabetes in later life. It is certainly interesting to further explore the potential implication and epigenetic mechanisms of these genes on glucose metabolism.

Paternal malnutrition

Apart from the detrimental impacts of maternal malnutrition on abnormal glucose metabolism in offspring, recent studies have shown that paternal malnourished exposures could also affect the phenotype of the offspring. Paternal lifestyle and particular nutrition factors can affect spermatogenesis at the level of germ and sertoli cells⁽ Reference Sharpe ^{50 )} and the composition of the seminal fluid⁽ Reference Robertson ^{51 )}. Similarly, alterations in paternal diet are also associated with altered DNA methylation in the offspring. As shown by Ng et al. ⁽ Reference Ng, Lin and Laybutt ^{52 )}, paternal exposure to a high-fat diet could induce intergenerational transmission of impaired glucose tolerance and insulin homeostasis in their female offspring. The IL-13 receptor-α2 (Il13ra2) promoter was hypomethylated in female offspring after high fat feeding of their fathers⁽ Reference Ng, Lin and Laybutt ^{52 )}. Offspring of fathers fed on a LP diet in C57BL/6J mice exhibited elevated hepatic expression of many genes involved in glucose metabolism and lipid biosynthesis. Epigenomic bisulphite sequence analysis on DNA from the liver of the offspring revealed numerous modest (about 20 %) changes in cytosine methylation depending on the paternal diet, including a substantial increase in methylation at an intergenic CpG island 50 kb upstream of the PPARα gene⁽ Reference Carone, Fauquier and Habib ^{53 )}. A recent study has shown that paternal prediabetes increased the susceptibility of the offspring to diabetes through gametic epigenetic alterations⁽ Reference Wei, Yang and Wei ^{54 )}. All these results indicate that paternal diet can also induce impaired glucose metabolism in offspring through epigenetic modification mechanisms, especially DNA methylation.

Transgenerational epigenetic effects

Maternal malnutrition during gestation and lactation not only have lifelong effects on F₁ offspring, but can also impact the second-generation (F₂), despite the fact that there is no further dietary modification to F₁. The first direct evidence of transgenerational epigenetic inheritance in mammals came from studies in mouse nucleocytoplasmic hybrids, showing that the cytoplasmic ‘environment’ of the early embryonic genome induces persistent and heritable changes in gene expression and CpG methylation⁽ Reference Roemer, Reik and Dean ^{55 )}. Another study indicated that mice fed on a high-fat diet from adulthood developed T2DM themselves, which was reversible, but their F₁ and F₂ offspring developed T2DM irreversibly, without further high-fat diet feeding⁽ Reference Gniuli, Calcagno and Caristo ^{56 )}. Similar results on glucose metabolism have been reported in another animal study on offspring of mothers with GDM. Impaired glucose tolerance was found in the F₁-GDM mice, whose plasma glucose level was significantly increased, while the fasting insulin level of the F₂-GDM was significantly increased. The expression of the imprinted genes IGF2 and H19 was down-regulated in pancreatic islets, caused by abnormal methylation status of the DMR⁽ Reference Ding, Wang and Shu ^{57 )}. Emerging evidence suggests that cross-generation transmission of adverse dietary effects during gestation/lactation is relevant to changes in epigenetics. Burdge et al. ⁽ Reference Burdge, Hanson and Slater-Jefferies ^{58 )} observed that DNA methylation of hepatic PPARα in both F₁ and F₂ generations was reduced due to dietary protein restriction during gestation, despite that dietary protein restriction was only applied in one generation and no dietary modification was administered in their offspring. However, not all epigenetic changes are transmittable across generations. For example, diet-induced hypermethylation at agouti viable yellow is not inherited transgenerationally through the female⁽ Reference Waterland, Travisano and Tahiliani ^{59 )}. A recent study uniquely revealed that prediabetes could be inherited transgenerationally through the mammalian germ line by DNA methylation⁽ Reference Wei, Yang and Wei ^{54 )}. In summary, most studies revealed that altered epigenetic modifications during early life could be transmitted to the next generation.

In conclusion, studies of human cohorts and animal models demonstrated that developmental programming of adult diseases occurs at each nutrition spectrum during the early life, due to maternal energy deprivation, maternal nutritional excess and paternal malnutrition. Although the specific mechanisms leading to aberrant glucose metabolism in each situation are different, current evidence supports that epigenetic modifications, especially DNA methylation, may be an important molecular link between them, and this epigenome can be inherited from one generation to another.