Introduction – link of fetal malnutrition to later development of disease
It is becoming increasingly apparent that the in utero environment in which a fetus develops may have long-term effects on the subsequent health and survival of the individual.1,2The landmark cohort study of 300 000 men by Ravelli et al.3showed that exposure of expectant mothers to the Dutch famine of 1944–5 during the first half of pregnancy had resulted in significantly higher obesity rates in the resulting male children at the age of 19 years. Subsequent studies of English men demonstrated a relationship between low birthweight and the development later in life of cardiovascular disease4 and impaired glucose tolerance.5–8 Other studies of populations in the USA,9–11 Sweden,12 France,13,14 Norway15 and Finland16 have demonstrated a significant correlation between low birthweight and the later development of adult diseases. The associations between low birthweight and increased risk of coronary heart disease, stroke and type 2 diabetes mellitus (T2DM) remain strong, even after adjusting for lifestyle factors (e.g. smoking, level of physical activity, occupation, income, dietary habits or childhood socioeconomic status) and occur independent of the current level of obesity or exercise.17,18
Low birthweight and insulin secretion
It remains controversial as to whether or not the adverse effects of fetal malnutrition on glucose homeostasis are mediated through programming of the fetal endocrine pancreas.1 Growth-restricted fetuses and newborns have been reported to have either a reduced population of pancreatic betacells19 or a normal percentage of pancreatic area occupied by beta-cells.20 Both of these studies were observational; morphometric analyses were not optimal and only a small number of fetuses/newborns were examined. It is likely that a significant proportion of, but not all, growth-restricted fetuses will have reduced beta-cell numbers. A more clinically relevant consideration is the impact of fetal growth restriction on beta-cell function.
Intrauterine growth-retarded (IUGR) fetuses have been found to exhibit lower insulin and glucose levels and a higher glucose–insulin ratio in the third trimester, as measured by cordocentesis.21 Two recent studies have shown that IUGR infants display decreased pancreatic beta-cell function, but increased insulin sensitivity at birth.22,23 Low birthweight has been associated with reduced insulin response after glucose ingestion in young non-diabetic men, whereas other studies have found no impact of low birthweight on insulin secretion.17,18 However, none of these earlier studies adjusted for the corresponding insulin sensitivity, which has a profound impact upon insulin secretion. Therefore, Jensen et al.24 measured insulin secretion and insulin sensitivity in a well-matched Caucasian population of 19-year-old glucose-tolerant men with birthweights of either below the 10th percentile [small for gestational age (SGA)] or between the 50th and 75th percentiles (controls). To eliminate the major confounders such as ‘diabetes genes’, none of the participants had a family history of diabetes, hypertension or ischaemic heart disease. There was no difference between the groups with regard to current weight, body mass index (BMI), body composition or lipid profile. When controlled for insulin sensitivity, insulin secretion was reduced by 30%. Insulin sensitivity, however, was normal in the SGA patients. The investigators hypothesized that defects in insulin secretion may precede defects in insulin action and that once SGA individuals accumulate body fat, they will develop insulin resistance.24
Genetics versus environment
Several epidemiological and metabolic studies of twins and first-degree relatives of patients with T2DM have demonstrated an important genetic component of diabetes.25–28 The association between low birthweight and the risk of T2DM in some studies could theoretically be explained by a genetically determined reduced fetal growth rate. In other words, the genotype responsible for T2DM may itself cause retarded fetal growth in utero. This forms the basis for the fetal insulin hypothesis, which suggests that genetically determined insulin resistance could result in low insulin-mediated fetal growth in utero as well as insulin resistance in childhood and adulthood.29 Insulin is one of the major growth factors in fetal life, and monogenic disorders that affect fetal insulin secretion or fetal insulin resistance also affect fetal growth. Mutations in the gene encoding glucokinase have been identified that result in low birthweight and maturity-onset diabetes in the young.30,31
The recently described T2DM susceptibility gene, transcription factor 7-like 2 (TCF7L2), confers a risk-allele frequency of approximately 30%.32 Studies of non-diabetic subjects show that TCF7L2 diabetesrisk genotypes alter insulin secretion.33–35 A large study of 24 053 subjects combined from six studies demonstrated that TCF7L2 is the first T2DM gene to be reproducibly associated with altered birthweight. Each maternal copy of the T allele at re7903146 increased offspring birthweight by 30 g, and the investigators suggest that the most likely mechanism is reduced maternal insulin secretion resulting in maternal hyperglycaemia and increased insulin-mediated fetal growth.36
Recent genetic studies suggest that increased susceptibility to T2DM in patients who are born SGA also results from the combination of genetic factors and an unfavourable fetal environment. Polymorphisms of peroxisome proliferator-activated receptor gamma 2 (PPARγ2), a gene involved in the development and metabolic function of adipose tissue, has been shown to modulate the susceptibility of subjects who are born SGA to developing insulin resistance later in life.37,38 The polymorphism is associated with a higher risk of T2DM if only birthweight is reduced.37,38 There is obviously a close relationship between genes and environment; not only can maternal gene expression alter the fetal environment, but the maternal intrauterine environment can also affect fetal gene expression, and both influence birthweight.
Association between obesity in pregnancy and later obesity in the offspring
Obesity is one of the most pervasive and burdensome public health problems in modern times. The steady increase in overweight reproductive-age women correlates with increases in rates of childhood and infant obesity. A possible link between abnormal intrauterine environment and abnormal growth and development of offspring must be considered. Maternal obesity significantly increases fetal and neonatal adiposity in the human; thus, enhanced adipocyte development per se must play an important role in the genesis of obesity in the offspring.39
A number of epidemiological studies40,41 have shown that there is a direct relationship between birthweight and BMI in childhood and in adult life. In the US Growing Up Today study,42 a cohort study of over 14 000 adolescents, a 1-kg increment in birthweight in full-term infants was associated with an approximately 50% increase in the risk of being overweight at age 9–14 years. When adjusted for maternal BMI, the increase in risk remained significantly elevated at 30%. A study of Danish military conscripts40 showed that even after controlling for birth length and maternal factors, BMI at age 18–26 years was strongly correlated with birthweight.
What can animal models tell us?
Animal models have a normal genetic background on which environmental effects during gestation or early post-natal life can be tested for their role in inducing disease later in life. Ontogeny of beta-cell development in the rodent approximates that observed in the human.43,44 The most commonly used animal models for IUGR are calorific or protein restriction, glucocorticoid administration or induction of uteroplacental insufficiency in the pregnant rodent. In the rat, maternal dietary protein restriction (approximately 40–50% of normal intake, termed LP) throughout gestation and lactation has been reported to alter insulin secretory capacity and reduce beta-cell mass through a reduction in beta-cell proliferation rate and an increase in apoptosis.45–53 Expression of pancreatic and duodenal homeobox 1 (Pdx1), a homeodomain-containing transcription factor that regulates early development of both endocrine and exocrine pancreas, and later differentiation and function of cells,54 is also reduced in islets from pups of LP mothers.55 In adulthood, rats born to LP mothers still have reductions in beta-cell mass and insulin secretion and display glucose intolerance, but usually not overt diabetes.45,46,53 In old age, LP offspring develop fasting hyperglycaemia associated with insulin resistance.56–60
Total calorific restriction during the last week of pregnancy and throughout lactation also reduces beta-cell mass and impairs insulin secretion in the offspring.61,62 When maternal undernutrition is prolonged until weaning and normal nutrition is given to the offspring from weaning onwards, growth restriction and beta-cell mass reduction persists into adulthood.62
An ovine model of IUGR induced by placental insufficiency (heat-induced) results in a significant reduction in beta-cell mass in fetuses near term (0.9 of gestation) from decreased rates of beta-cell proliferation and neogenesis.64 Plasma insulin concentrations in IUGR fetuses are lower at baseline and glucose-stimulated insulin secretion is impaired. Similar deficits occur with arginine-stimulated insulin secretion. A deficiency in islet glucose metabolism also occurs in the rate of islet glucose oxidation at maximal stimulatory glucose concentrations. Thus, pancreatic islets from nutritionally deprived IUGR fetuses caused by chronic placental insufficiency have impaired insulin secretion caused by reduced glucose-stimulated glucose oxidation rates, insulin biosynthesis and insulin content. This impaired glucose-stimulated insulin secretion (GSIS) occurs despite an increased fractional rate of insulin release from a greater proportion of releasable insulin as a result of diminished insulin stores.65
To extend these experimental studies of growth restriction, we developed a model of IUGR that restricts fetal growth in the rat.66–68 Growth-restricted fetal rats have critical features of a metabolic profile characteristic of growth-retarded human fetuses: decreased levels of glucose, insulin, insulin-like growth factor 1 (IGFI), amino acids and oxygen.66–70 Birthweights of IUGR animals are significantly lower than those of controls until approximately 7 weeks of age, when IUGR rats catch up with controls. Between 7 and 10 weeks of age, the growth of IUGR rats accelerates and surpasses that of controls, and by 26 weeks of age, IUGR rats are obese.67 No significant differences are observed in blood glucose and plasma insulin levels at 1 week of age. Between 7 and 10 weeks of age, however, IUGR rats develop mild fasting hyperglycaemia and hyperinsulinaemia. IUGR animals are glucose intolerant and insulin resistant at an early age. First-phase insulin secretion in response to glucose is also impaired early in life in IUGR rats, before the onset of hyperglycaemia. There are no significant differences in beta-cell mass, islet size or pancreatic weight between IUGR and control animals at 1 and 7 weeks of age. In 15-week-old IUGR rats, however, the relative beta-cell mass is 50% that of controls, and by 26 weeks of age, beta-cell mass is less than one-third that of controls. This loss of beta-cell mass is accompanied by a reduction in Pdx1 expression that is greater than that in beta-cell mass.71 By 6 months of age, IUGR rats develop diabetes with a phenotype remarkably similar to that observed in humans with T2DM: progressive dysfunction in insulin secretion and insulin action.67 Thus, despite different animal models of IUGR, these studies support the hypothesis that an abnormal intrauterine milieu can induce permanent changes in beta-cell function after birth and can lead to T2DM in adulthood.
Animal model of obesity in pregnancy induces obesity in the offspring
Several investigators have used animal models of high-fat or Western-style diet-induced obesity (a diet high in fat and carbohydrate content) and have shown that maternal overnutrition induces increased adiposity and permanent changes in metabolism in the offspring.72–84 Using a similar Western diet, female Sprague–Dawley rats were started on the designated diet at 4 weeks of age. The rats were mated at 12–14 weeks of age, and all resulting pups were weaned onto a control diet.85 Offspring of dams fed the Western diet had significantly increased adiposity as early as 2 weeks of age; they also showed impaired glucose tolerance compared with offspring of dams fed a control diet. Inflammation and oxidative stress were increased in preimplantation embryos, fetuses and newborns of obese dams. Oxidative stress was correlated with increased expression of pro-adipogenic and lipogenic genes in fat tissue and in preimplantation embryos. These results suggest that obesity is programmed as early as the preimplantation stage of development.
Chromatin structure, deoxyribonucleic acid methylation and gene expression
Epigenetic modifications of the genome provide a mechanism that allows the stable propagation of gene expression from one generation of cells to the next. Epigenetic states can be modified by environmental factors, which may contribute to the development of abnormal phenotypes. There are at least two distinct mechanisms through which epigenetic information can be inherited: histone modifications and deoxyribonucleic acid (DNA) methylation.86,87
In eukaryotes, the nucleosome is formed when DNA is wrapped around an octameric complex of two molecules of each of the four histones: H2A, H2B, H3 and H4. The amino termini of histones can be modified by acetylation, methylation, SUMOylation, phosphorylation, glycosylation and adenosine diphosphate (ADP) ribosylation. The most common histone modifications involve acetylation and methylation of lysine residues in the amino termini of H3 and H4. Increased acetylation induces transcription activation, whereas decreased acetylation usually induces transcription repression. Methylation of histones, on the other hand, is associated with both transcription repression and activation.86,87 Moreover, lysine residues can be mono-, di- or trimethylated in vivo, providing an additional mechanism of regulation.86,87
The second class of epigenetic regulation is DNA methylation, in which a cytosine base is modified by a DNA methyltransferase at the C5 position of cytosine, a reaction carried out by various members of a single family of enzymes.87 Approximately 70% of cytosine-phosphate-guanine (CpG) dinucleotides in human DNA are constitutively methylated, whereas most of the unmethylated CpGs are located in CpG islands. CpG islands are CG-rich sequences located near coding sequences and serve as promoters for their associated genes. Approximately half of mammalian genes have CpG islands.87 The methylation status of CpG islands within promoter sequences works as an essential regulatory element by modifying the binding affinity of transcription factors to DNA binding sites. In normal cells, most CpG islands remain unmethylated; however, in circumstances such as cancer88–90 and oxidative stress they can become methylated de novo. This aberrant methylation is accompanied by local changes in histone modification and chromatin structure, such that the CpG island and its embedded promoter take on a repressed conformation that is incompatible with gene transcription. It is not known why particular CpG islands are susceptible to aberrant methylation.
Deoxyribonucleic acid methylation is commonly associated with gene silencing and contributes to X-chromosomal inactivation, genomic imprinting, as well as transcriptional regulation of tissue-specific genes during cellular differentiation.91–93 It is not known why some genes are able to undergo aberrant DNA methylation; however, a study by Feltus et al.94 suggests that there is a ‘DNA sequence signature associated with aberrant methylation’. Of major significance to T2DM is their finding that Pdx1, a pancreatic homeobox transcription factor, was one of only 15 genes (of 1749 examined) with CpG islands within the promoter that were methylation susceptible (which was induced by over-expression of a DNA methyltransferase). This study demonstrates that genes essential to pancreatic development, such as Pdx1, are susceptible to epigenetic modifications, which could ultimately affect gene expression.
Histone methylation can influence DNA methylation patterns and vice versa.91 For example, methylation of lysine 9 on histone 3 (H3) promotes DNA methylation, while CpG methylation stimulates methylation of lysine 9 on H3.92 Recent evidence indicates that this dual relationship between histone methylation and DNA methylation might be accomplished by direct interactions between histone and DNA methyltransferases.91 Thus, chromatin modifications induced by adverse stimuli are self-reinforcing and can propagate.
Maternal nutritional supplementation and epigenetic modifications in the offspring
The role of environmental regulation of epigenetic phenomena in the offspring has been established by experiments performed in Agouti mice.95 Wildtype expression of the Agouti protein results in a phenotypic brown coat colour in the mouse. In this mouse model, an endogenous retrovirus-like transposon sequence is inserted close to the gene coding for the Agouti protein. An unmethylated retrotransposon promoter overrides the wild-type Agouti promoter, resulting in ectopic Agouti expression and a yellow coat colour. A methylated retrotransposon is silenced and results in a wild-type Agouti (brown) coat. Wolff et al.96 have investigated whether or not maternal diet can alter the phenotype of the Agouti mouse and found that when pregnant females are fed a diet supplemented with methyl donors, a larger proportion of the offspring have a wild-type Agouti coat colour compared with the offspring of mothers fed a standard diet. These studies indicate that the maternal methyl donor diet leads to increased methylation of the offspring’s retrotransposon. Methylation silences the offspring’s retrotransposon, allowing the wild-type Agouti promoter to be expressed, thus resulting in a mouse with a wild-type (brown) coat colour. These results suggest that maternal nutritional environmental exposure can change the stable expression of genes in the offspring through an epigenetic modification that takes place in utero.
Epigenetic regulation of gene expression in fetal growth restriction
A number of studies suggest that uteroplacental insufficiency, a common cause of IUGR, induces epigenetic modifications in offspring.97–100 Epigenetic modifications affecting processes important to glucose regulation and insulin secretion, characteristics essential to the pathophysiology of T2DM, have been described in the IUGR liver, pancreatic beta-cells and muscle.97–100
Chromatin remodelling in the beta-cells of intrauterine growth-retarded rats
Pancreatic and duodenal homeobox 1 is a homeodomain-containing transcription factor that plays a critical role in the early development of both the endocrine and exocrine pancreas, and in the later differentiation and function of the beta-cells. As early as 24 hours after the onset of growth restriction, Pdx1 messenger ribonucleic acid (mRNA) levels are reduced by more than 50% in IUGR fetal rats. Suppression of Pdx1 expression persists after birth and progressively declines in the IUGR animal, implicating an epigenetic mechanism.
Changes in histone acetylation are the first epigenetic modification found in beta-cells of IUGR animals. Islets isolated from IUGR fetuses show a significant decrease in H3 and H4 acetylation at the proximal promoter of Pdx1.99 These changes in H3 and H4 acetylation are associated with a loss of binding of the upstream stimulatory factor 1(USF1) to the proximal promoter of Pdx1.99 USF1 is a critical activator of Pdx1 transcription and its decreased binding markedly decreases Pdx1 transcription.101,102 After birth, histone deacetylation progresses and is followed by a marked decrease in histone 3 lysine 4 (H3K4) trimethylation and a significant increase in dimethylation of H3K9 in IUGR islets.99 H3K4 trimethylation is usually associated with active gene transcription, whereas H3K9 dimethylation is usually a repressive chromatin mark. Progression of these histone modifications parallels the progressive decrease in Pdx1 expression that manifests as a deterioration in glucose homeostasis and increased oxidative stress in the ageing IUGR animals.99 Nevertheless, at 2 weeks of age, the silencing histone modifications in the IUGR pup are responsible for suppression of Pdx1 expression since there is no appreciable methylation of CpG islands in mice at this age.99 Reversal of histone deacetylation in IUGR islets at 2 weeks of age is sufficient to nearly normalize Pdx1 mRNA levels permanently, perhaps owing to active beta-cell replication present in the neonatal rodent.99
In IUGR, Pdx1 is first silenced as a result of recruitment of corepressors, including histone deacetylase 1 (HDAC1) and mSin3A.99 These repressors catalyse histone deacetylation. Binding of these deacetylases facilitates loss of trimethylation of H3K4, further repressing Pdx1 expression.99 We found that inhibition of HDAC activity by trichostatin A (TSA) treatment normalizes H3K4me3 levels at Pdx1 in IUGR islets.99 These data suggest that the association of HDAC1 at Pdx1 in IUGR islets probably serves as a platform for the recruitment of a demethylase, which catalyses demethylation of H3K4.
The molecular mechanism responsible for DNA methylation in IUGR islets is probably dependent on the methylation status of lysine 9 on H3 (H3K9). Previous studies have shown that changes in methylation of H3K9 precede changes in DNA methylation.103,104 It has also been suggested that DNA methyltransferases may act only on chromatin that is methylated at H3K9.105 Histone methyltransferases, specifically DNA methyltransferase 3A (DNMT3A) and DNA methyltransferase 3B (DNMT3B), bind to DNA methylases, thereby initiating DNA methylation.105
These results demonstrate that IUGR induces a selfpropagating epigenetic cycle in which the mSin3A/HDAC complex is first recruited to the Pdx1 promoter, histone tails are subjected to deacetylation and Pdx1 transcription is repressed. At the neonatal stage, this epigenetic process is reversible and may define an important developmental window for therapeutic approaches. However, as dimethylated H3K9 accumulates, DNMT3A is recruited to the promoter and initiates de novo DNA methylation, which locks in the silenced state in the IUGR adult pancreas, resulting in diabetes.
How do these epigenetic events lead to diabetes?
Targeted homozygous disruption of Pdx1 in mice results in pancreatic agenesis, and homozygous mutations yield a similar phenotype in humans.106 Milder reductions in Pdx1 protein levels, as occur in the Pdx+/− mice, allow for the development of a normal mass of beta-cells,106 but result in the impairment of several events in glucose-stimulated insulin secretion.106 These results indicate that Pdx1 plays a critical role in the normal function of beta-cells106 in addition to its role in beta-cell lineage development. This may be the reason why humans with heterozygous missense mutations in PDX1 exhibit early- and late-onset forms of T2DM.106
The discovery of a critical developmental stage during which aberrant epigenetic modifications may be reversed represents a therapeutic window for the use of novel agents that could prevent common diseases with late-onset phenotypes. T2DM is one such disease, and predisposed individuals could be treated with agents that normalize the epigenetic programming of key genes, thus providing protection against development of the adult diabetic phenotype.
Genome-wide deoxyribonucleic acid methylation is disrupted in intrauterine growth-retarded islets
Epigenetic modifications are not confined to the Pdx1 locus in the IUGR rat. We mapped DNA methylation across approximately 1 000 000 loci using the Hpall tiny fragment enrichment by ligation-mediated PCR (HELP) assay.107 Comparison of IUGR with normal rats at 7 weeks of age, prior to the onset of diabetes, revealed changes in DNA methylation at a number of novel loci, not limited to canonical CpG islands or promoters. We found that IUGR in the rat causes consistent and non-random changes in cytosine methylation, affecting < 1% of HpaII sites in the genome in the islet. The majority of these changes take place not at promoters but at intergenic sequences, many of which are evolutionarily conserved. Furthermore, some of these loci are in proximity to genes manifesting concordant changes in gene expression and are enriched near genes that regulate processes that are markedly impaired in IUGR islets (e.g. vascularization, proliferation, insulin secretion and cell death).
The studies described above clearly show that environmental effects can induce epigenetic alterations, ultimately affecting expression of key genes linked to the development of T2DM including genes critical for pancreatic development and beta-cell function, peripheral glucose uptake, and insulin resistance and atherosclerosis. Recent progress in understanding epigenetic programming of gene function has led to the development of novel therapeutic agents with epigenetic targets in diseases such as cancer. Understanding the role of developmental programming of genes crucial to the development of T2DM may unveil a critical window during which epigenetic therapeutic agents could be used as a means to prevent the later development of a disease. Prior to the use of such therapeutic agents, there remains much to be learned about the programming of the epigenetic code, especially on a genome-wide scale. Much of the recent progress in understanding epigenetic phenomena is directly attributable to technologies that allow researchers to pinpoint the genomic location of proteins that package and regulate access to the DNA. The advent of DNA microarrays and inexpensive DNA sequencing has allowed many of those technologies to be applied to the whole genome. It is now possible that epigenetic profiling of CpG islands in the human genome can be used as a tool to identify genomic loci that are susceptible to DNA methylation. Aberrant methylation may then be used as a biomarker for disease. The genome-wide mapping of histone modifications by chromatin immunoprecipitation (ChIP)-chip and ChIP-seq has led to important insights regarding the mechanism of transcriptional and epigenetic memory, and how different chromatin states are propagated through the genome in yeast and in mammalian cells.108,109 Although Chip-seq experiments are currently being performed in human tissue, obstacles such as intrinsic human epigenetic variability (including age-related changes) and tissue-specific epigenetic variability must be characterized and mapped in the healthy, non-diseased state before this information can be applied to diseases such as T2DM. Eventually, genome-wide epigenetic characterization will lead to specific therapies with epigenetic targets and will also allow monitoring of genome-wide epigenetic consequences of these therapies once they are applied.