The Developmental Origins of Health and Disease hypothesis and nutritional programming

The Developmental Origins of Health and Disease hypothesis, which was initially based on epidemiological approaches, is now receiving steady back-up and boosted support through intensive genetically founded experimental research in both cell or animal models and through human genome-related findings⁽Reference Waterland and Michels⁹⁾.

The pioneer finding that poor fetal and infant growth are associated with a higher risk of suffering metabolic syndrome features in adulthood has allowed to sustain the thrifty phenotype hypothesis, which points out that early inadequate nutrition induces insulin resistance⁽Reference Hales and Barker¹⁰⁾. Additional studies conducted in a British Cohort⁽Reference Parsons, Power and Manor¹¹⁾, Danish subjects born between 1936 and 1983⁽Reference Rugholm, Baker and Olsen¹²⁾, Dutch famine or Leningrad siege affected children⁽Reference Silveira, Portella and Goldani¹³⁾ have been consistent with the possibility of developmentally mediated origin of obesity⁽Reference Oken and Gillman¹⁴⁾.

Furthermore, the National Children Study, the Southampton's Women Survey, the Viva project and the Behavioural Perinatology Research Program together with other more recently launched investigations such as the Avon Longitudinal Study of Parents and Children⁽Reference Fraser, Tilling and Macdonald-Wallis¹⁵⁾ are contributing to identify critical processes underlying the interactions between retarded fetal growth and development with adiposity-related outcomes and obesity features⁽Reference Wadhwa, Buss and Entringer¹⁶⁾.

Indeed, nutritional programming may explain the predisposition of some individuals to suffer non-communicable diseases in adulthood linked to in utero-deprived or infant development⁽Reference Bruce and Hanson¹⁷⁾ (see Fig. 1.). Inheritance-oriented investigations concerning gene–nutrient interactions on energy homoeostasis processes and metabolic cell functions is extending to all clinically chronic relevant diseases such as diabetes and cardiovascular events as well as to obesity and associated inflammatory features⁽Reference Stenvinkel, Karimi and Johansson^18, Reference Tamashiro and Moran¹⁹⁾. Also, some studies have identified that inflammation and transient infections could be not only a comorbidity of obesity and diabetes⁽Reference Kolb and Mandrup-Poulsen²⁰⁾ but also an aetiological agent⁽Reference Moreno-Aliaga, Campion and Milagro^21, Reference Van den Berghe²²⁾.

The fetal or developmentally programmed genesis of adult sickness hypothesis settled that environmental factors and maternal lifestyles, particularly adverse nutritional disturbances, proceed in early life to drive the risks for the onset of metabolic diseases and excessive weight gain in later life stages⁽Reference Freeman²³⁾. Indeed, maternal nutrition can programme gene expression patterns to the embryo that persist into adulthood and may contribute to the appearance of typical metabolic syndrome features such as hypertension, insulin resistance, hyperlipaemia and abdominal obesity⁽Reference Symonds, Sebert and Hyatt²⁴⁾. The parental conditions and lifestyles, which may involve maternal size/obesity, famine at perinatal periods, the use of nutritional supplements, alcohol or drug abuse as well as the administration of therapeutical agents in this critical period may alter specific processes with an impact on embryonic, placental and fetal growth, organogenesis or regulatory set points for system functions affecting adiposity, where inflammatory and immunologically mediated processes may be involved⁽Reference Campion, Milagro and Martinez²⁵⁾. Interestingly, some of these epigenetically mediated signals may be not permanent but transient, which is of interest not only for prevention, but also as a target for developing future therapeutic focus⁽Reference Herrera, Keildson and Lindgren²⁶⁾.

Unfavourable environmental cues coming from the mother such as psychological stress, infection, over or undernutrition, smoking, neuroendocrine disruptors, trauma or diseases are signalled inputs negatively affecting the embryo, fetus or neonate⁽Reference Langley-Evans and McMullen²⁷⁾. The adaptive responses may involve growth stunting or tissue remodelling with an impact on physiological functions and metabolism being the trade-off and increased risk in later life⁽Reference Lillycrop³⁾. The characteristics of the programmed outcomes depend upon the insult or stimuli, as well as on the timing of the exposure⁽Reference Oliver, Jaquiery and Bloomfield²⁸⁾.

Understanding the maternal regulation of fetal development and programming involves a knowledge about the genome, baseline maternal body composition, dietary and metabolic status, uteroplacental blood flow and substrate transfer that may condition the nutrient balance and fetal malnutrition by inducing hypoxaemia or metabolic changes (cortisol, insulin and nutrient oxidation) and altering body composition in the newborns⁽Reference Godfrey and Barker²⁹⁾. In this context, the term developmental plasticity has been used to define the aptitude of a unique genotype to produce a variable phenotype in response to changing environmental circumstances, even though they may be neither adaptive nor prognostic as stated by the developmental programming theory⁽Reference McMillen, MacLaughlin and Muhlhausler³⁰⁾, but struggling for resources which may envisage or forecast future metabolic scenarios in an effort to tune gene expression to generate a better adapted phenotype to the predictable afterwards environment⁽Reference Godfrey, Lillycrop and Burdge³¹⁾.

A number of mammal's models have been developed to examine the potential processes and mechanisms involved in perinatal programming that depend on nutrition⁽Reference Mostyn and Symonds³²⁾. The phenomena and manifestations ascribed to early nutritional programming have been explained through different mechanisms such as the involvement of the adipose tissue, the participation of different hormones and endocrine systems, enzymes, transcription factors and signalling mediators (glucocorticoids, insulin, PPAR family, adipokines, etc.), the regulation of specific nutrient-related metabolic pathways (lipogenesis, glyconeogenesis, etc.), the control of neural networks affecting the appetite system (specific orexigenic/anorexigenic neuropeptides and the hypothalamic–pituitary–adrenal axis) and the up/down feedback concerning the gene expression machinery and epigenetic marks⁽Reference McMillen and Robinson^33, Reference McMullen and Mostyn³⁴⁾.

The question if later obesity is induced in utero or early after birth has been repeatedly addressed and researched⁽Reference Vickers^35, Reference Kemp, Kallapur and Jobe³⁶⁾. Thus, it seems that glucose, insulin and leptin coming from the mother's blood or taken from the breast milk are of relevance for the persistent programming of food intake control at least in the rodent⁽Reference McMillen, Adam and Muhlhausler³⁷⁾ as well as for obesity-related peptides and hormones such as insulin⁽Reference Hietaniemi, Malo and Jokela³⁸⁾. Furthermore, hormonal and metabolic signals acting during the perinatal period could alter the structure and functions concerning the fat–brain axis or adipogenic genes in the adipose tissue that regulates the energy balance during later life⁽Reference McMillen, Rattanatray and Duffield³⁹⁾. Alternatively, adverse intrauterine exposures may produce long-term changes in mRNA levels leading to a thrifty phenotype with changes affecting liver, muscle and renal anatomy and physiology⁽Reference Oken and Gillman¹⁴⁾ as well as long-lasting changes in mitochondria that can be associated with obesity and insulin resistance in later life⁽Reference Park, Kim and Kim⁴⁰⁾. Other mechanisms that have been investigated in different animal models in order to clarify the role of the perinatal feeding on tissue structure have revealed an important impact of the maternal diet on proliferation and differentiation processes in the pancreas and in the brain involving an overexposure to glucocorticoids mediated by a reduced activity of the placental 11-β hydroxysteroid dehydrogenase⁽Reference McMullen and Mostyn³⁴⁾. Also, a maternal obese condition influences fetal growth and body composition with implications in the future offspring health depending on the genetic background, the intrauterine metabolic environment and the generated maternal metabolites⁽Reference Freeman²³⁾. Thus, a mild maternal overnutrition led to increased adiposity, glucose intolerance and altered brain appetite regulators in offspring⁽Reference Rajia, Chen and Morris⁴¹⁾, while food-deprived dams may transfer to the offspring patterns of increased hepatic gluconeogenesis, enhanced release and impaired oxidation of fatty acids from adipocytes, resistance to ketosis and changes in glucose uptake mediated by an increased insulin receptor expression⁽Reference Hales and Barker¹⁰⁾. On the other hand, intrauterine growth restriction due to perinatal impaired uteroplacental function or nutrient deficiency has been linked to lower leptin, normal or lower adiponeptin and higher ghrelin and visfatin levels, while contradictory results have been reported concerning apelin and resistin and other pro-inflammatory markers such as TNFα and IL-6⁽Reference Briana and Malamitsi-Puchner⁴²⁾. Interestingly, maternal perinatal undernutrition attenuates T-cell function in adult male rat offspring⁽Reference Badr and Mohany⁴³⁾.

Finally, epigenetic marks affecting a number of genes regulating energy metabolism, adipogenesis or inflammatory processes are providing new clues to understand the relationships between nutritional programming and obesity in adulthood⁽Reference Campion, Milagro and Martinez^25, Reference Lillycrop and Burdge⁴⁴⁾.

Despite the general acceptance of the Developmental Origins of Health and Disease hypothesis, the terms of such proposal have not always been demonstrated in epidemiological surveys, with some inconsistent results being reported in animal models⁽Reference Langley-Evans and McMullen²⁷⁾. A systematic error in interpreting experimental data and a publication bias due to missing information are not ruled out.

Animal models and epigenetic regulation

There are a number of experimental interventions in animals concerning the role of nutrition in pregnancy and lactation including energetic deprivation, protein restriction and excess fat feeding⁽Reference McMullen and Mostyn³⁴⁾. Such investigations have proven a cluster of disorders affecting energy efficiency as well as the impairment of different metabolic pathways and adverse predisposition for suffering CVD, glucose intolerance and obesity in the offspring⁽Reference Simmons⁸⁾ and unfavourable inflammatory interactions⁽Reference Marcos-Gomez, Bustos and Prieto⁴⁵⁾.

A loss of diurnal variation in heart rate and blood pressure in adulthood has been described when maternal undernutrition is followed by postnatal overnutrition in rodents, while hyperphagia resulting from perturbed development of the hypothalamic circuitry devoted to food intake control may contribute to overweight and developmental changes in fat-cell precursors⁽Reference Remacle, Bieswal and Bol⁴⁶⁾. On the other hand, maternal obesity has an effect on pancreatic β-cells inducing a higher risk of diabetes⁽Reference Han, Xu and Epstein⁴⁷⁾. The mitochondrial DNA content of the liver and skeletal muscle was reduced in fetal and early postnatal undernourished animals even when balanced nutrition was provided after weaning, which was accompanied by a decrease in mitochondrial DNA-encoded gene expression indicating that poor nutrition in early life causes long-lasting changes in mitochondria that may contribute to the development of insulin resistance in later life⁽Reference Park, Kim and Kim⁴⁰⁾. The adverse effects of an in utero low protein dietary intake have been associated with diabetes⁽Reference Sparre, Reusens and Cherif⁴⁸⁾, increased systolic blood pressure⁽Reference Langley-Evans, Phillips and Jackson⁴⁹⁾, altered glucose tolerance⁽Reference Langley-Evans, Phillips and Jackson⁴⁹⁾, hyperisulaemia and reduced insulin signalling protein expression later in life⁽Reference Fernandez-Twinn, Wayman and Ekizoglou⁵⁰⁾. Also, a reduced maternal protein consumption during pregnancy and lactation has window-of-exposure and sex-specific effects on offspring growth, adiposity, appetite, glucose utilisation and circulating leptin⁽Reference Zambrano, Bautista and Deas⁵¹⁾. A programming of hepatic insulin-sensitive enzymes has been found in the offspring of rat dams fed a protein-restricted diet where glucokinase activity decreased (approximately 50%), whereas phosphoenolpyruvate carboxykinase activity increased (approximately 100%) with parallel changes in gene expression in both enzymes⁽Reference Desai, Byrne and Zhang⁵²⁾. Indeed, the understanding of genetic and epigenetic factors in human nutrition and health is helping to translate basic biology into clinical applications⁽Reference Marti, Martinez-Gonzalez and Martinez^1, Reference Stover and Caudill⁵³⁾ concerning perinatal nutrition and disease in adulthood.

Three genomic targets have been involved in the modulation of the gene expression changes in relation to disease susceptibility: epigenetic marks at the promoter regions of some epiobesogenes, transposable elements that lie adjacent to genes with metastable epialleles, and the regulation of imprinted genes⁽Reference Jirtle and Skinner⁵⁴⁾. In this context, epigenetic studies are contributing to unravel some putatively hidden phenomena that are not being explained by the accomplishment of the Human Genome Project in relation to obesity⁽Reference Rodenhiser and Mann⁵⁵⁾. In the last few years, different examples of dynamic changes in DNA methylation patterns, histone covalent modifications and the involvement of non-coding RNA due to the restriction or supplementation with different nutrients have been reported⁽Reference vel Szic, Ndlovu and Haegeman^4, Reference Campion, Milagro and Martinez^25, Reference Ford, Ions and Alatawi⁵⁶⁾, but also in relation to obesity⁽Reference Marti and Ordovas⁵⁷⁾. Thus, the methylation pattern of the leptin promoter in adipocytes is affected by a high fat intake in rats following an inverse trend to body weight changes⁽Reference Milagro, Campion and Garcia-Diaz⁵⁸⁾, while weight gain induced by an isoenergetic pair-fed high-fat diet produces a nutriepigenetic outcome on fatty acid synthase and NADH dehydrogenase ubiquinone 1β subcomplex subunit 6 gene promoters⁽Reference Lomba, Martinez and Garcia-Diaz⁵⁹⁾.

Interestingly, maternal supraphysiological methyl group (folate, cobalamine, choline and betaine) or genistein supply throughout pregnancy modifies DNA methylation of some key metabolic genes, with implications in adiposity⁽Reference Dolinoy, Weidman and Waterland^60, Reference Waterland, Travisano and Tahiliani⁶¹⁾, while protein restriction of pregnant rats induces DNA hypomethylation in the glucocorticoid receptor and PPARα genes in the liver of the newborns, which was prevented by folic acid supplementation⁽Reference Lillycrop, Phillips and Jackson⁶²⁾. Also, it has been found that a chronic high-fat diet in fathers epigenetically programmes β-cell dysfunction in female rat offspring⁽Reference Ng, Lin and Laybutt⁶³⁾. These findings are in agreement with previous observations suggesting that transgenerational epigenetic inheritance may be sex-dependent for specific traits⁽Reference Wu and Suzuki^64, Reference Cordero, Gomez-Uriz and Milagro⁶⁵⁾.

In addition to changes in methylation patterns, epigenetic transfer may involve histone modifications and miRNA-mediated mechanisms⁽Reference Sinclair, Lea and Rees⁶⁶⁾. Thus, an energy-dense maternal diet driving to obesity epigenetically impairs fetal chromatin structure in primates via covalent modifications on histones⁽Reference Aagaard-Tillery, Grove and Bishop⁶⁷⁾, while a role for miRNA in the alternative expression of insulin-like growth factor 2 in fetal livers from high-fat fed dams has been reported⁽Reference Zhang, Zhang and Didelot^68, Reference Heerwagen, Miller and Barbour⁶⁹⁾.

All these data and experiments strongly suggest that epigenetic mechanisms may be boosted or impaired by dietary and environmental factors in the gestating mother and could be involved in obesity susceptibility in the offspring⁽Reference Waterland and Michels^9, Reference Kemp, Kallapur and Jobe³⁶⁾.

Obesity, inflammation and epigenetics

Inflammation is a protective complex biological response mounted by tissues to combat injurious stimuli in order to maintain cell homoeostasis⁽Reference Medzhitov⁷⁰⁾, which include host defence, tissue remodelling and metabolic changes, and involve multiple mechanisms such as the contribution of immune cells (recruitment and activation of leucocytes, granulocytes, monocytes, B–T-lymphocytes and dendritic cells), the involvement of different mediators (IL, TNFα, leptin, adipokines, etc.) or the regulation of signalling pathways (insulin, glucose, lipids, etc.) and eventually the epigenetic regulation of the expression of some related genes⁽Reference vel Szic, Ndlovu and Haegeman^4, Reference Entringer, Buss and Wadhwa⁷¹⁾. Indeed, many important risk factors for obesity (overnutrition, low dietary fibre intake, sedentary lifestyles, sleep debts, neuroendocrine status or genetic make-up) have been found to be implicated in local or low-grade systemic inflammation⁽Reference Kolb and Mandrup-Poulsen²⁰⁾. Thus, an excessive adiposity has been considered either as a cause or as a consequence of chronic inflammatory disorders⁽Reference Moreno-Aliaga, Campion and Milagro²¹⁾ and epimutations⁽Reference Baple, Poole and Mansour⁷²⁾. Interestingly, inflammatory signalling has been identified as mediator of epigenetic modulation in tissue-specific chronic inflammation⁽Reference Backdahl, Bushell and Beck⁷³⁾.

One of the challenges in the epigenomics field is identifying and characterising the epigenetic marks and those stimuli modulating the expression of some specific genes (epiobesogenes) in pathways involving obesity/body weight homoeostasis and energy balance processes such as adipogenesis, inflammation, appetite, insulin signalling, thermogenesis or macronutrient turnover⁽Reference Campion, Milagro and Martinez²⁵⁾. Indeed, a bioinformatics analysis of promoter regions for the search of epigenetic biomarkers of obesity, has identified putative methylation-prone sequence patterns in several obesity-related genes such as fibroblast growth factor 2, phosphatase and tensin homolog deleted on chromosome 10, cyclin-dependent kinase inhibitor 1A and oestrogen receptor 1, implicated in adipogenesis, suppressor of cytokine signalling 1/suppressor of cytokine signalling 3, in inflammation and cytochrome c oxidase 7A1 lipoprotein lipase, caveolin 1 and insulin-like growth factor binding protein 3, in intermediate metabolism and insulin signalling⁽Reference Campion, Milagro and Martinez⁷⁴⁾. The characterisation of those individuals which at an early age could present changes in the methylation profiles of specific genes could help to predict their susceptibility to later develop obesity, which may make it possible to prevent and follow-up its progress, as well as to research and develop newer therapeutic approaches⁽Reference Feinberg, Irizarry and Fradin⁷⁵⁾. Thus, from approximately 760 human genes under putative epigenetic regulation, about 20% of them, defined as epiobesogenes, could be associated with obesity⁽Reference Campion, Milagro and Martinez^74, Reference Martinez, Milagro and Campion⁷⁶⁾. The knowledge of the modification of their methylation patterns due to different dietary factors, age, inflammation or some of the physiological aspects surrounding overweight, could be crucial to investigate the role of these mechanisms in the prevention, onset and therapy of obesity. Also, the reversibility/stability of the epigenetic code and the involvement of specific enzymes (methylases, acetylases, etc.) is of interest⁽Reference Rodenhiser and Mann⁵⁵⁾.

Epigenetically mediated signal-specific inflammatory mechanisms may operate through transcription factors (NF-κB family), kinases (IκB kinase-related kinases, salt-inducible kinase, protein kinase A, phosphoinositide 3-kinase, serine/threonine protein kinase, etc.), the endoplasmic reticulum (Ca), biochemical activation of DNA methyltransferases and histone modifier enzymes (histone deacetylase/histone acetyltransferase, histamine methyltransferase, sirtuin, etc.), changes in cellular pools of acetyl-CoA, NAD or methyl donors, all of them sensitive to oxidative stress, hyper- or hypoglycaemia and fatty acids load, and activated or inhibited by overnutrition⁽Reference Heerwagen, Miller and Barbour^69, Reference Hotamisligil⁷⁷⁾. Furthermore, in addition to epigenetic processes involving methylation marks and histone modifications, also non-coding RNA are susceptible to dynamic inflammatory control⁽Reference vel Szic, Ndlovu and Haegeman⁴⁾.

The concept of epigenetic regulation is gradually being recognised as an important factor in inflammatory-related events such as obesity, diabetes and CVD⁽Reference Wierda, Geutskens and Jukema⁷⁸⁾. Gene silencing in severe systemic inflammation has been associated with the reprogramming of acute pro-inflammatory genes, the intervention of NF-κB and the compartmentalisation of the epigenetic process⁽Reference McCall and Yoza⁷⁹⁾. In this context, the impact of inflammation on global DNA methylation has been demonstrated in chronic kidney disease, while an epigenetic regulation of high-glucose-induced pro-inflammatory cytokine production in monocytes has been described for the polyphenol curcumin involving NF-κB⁽Reference Yun, Jialal and Devaraj⁸⁰⁾. On the other hand, histone deacetylase inhibitors are emerging as possible epigenetic modulators of gene expression controlling the inflammatory response in some circumstances⁽Reference Faraco, Pittelli and Cavone⁸¹⁾. Also, an increased expression of DNA methyltransferase3A in obese adipose tissue has been reported⁽Reference Kamei, Suganami and Ehara⁸²⁾, while the methylation of polycomb target genes may be mediated by inflammation⁽Reference Hahn, Hahn and Lee⁸³⁾. Furthermore, redox modulation of chromatin remodelling may have an effect on histone acetylation/deacetylation modulating the expression of pro-inflammatory genes⁽Reference Rahman, Marwick and Kirkham⁸⁴⁾. Finally, enhanced levels of miRNA125b are associated with increased expression of specific inflammatory genes in db/db mice⁽Reference Villeneuve, Kato and Reddy⁸⁵⁾. A relevant locus-specific DNA methylation affecting inflammatory processes has been reported for at least the following genes: leptin, superoxide dismutase, glucocorticoid receptor, PPAR, TNFα, endothelial nitric oxide synthase and inducible nitric oxide synthase and hypoxia-inducible factor, and the understanding of their epigenetic machinery will contribute to the management of inflammation and associated disorders including obesity⁽Reference vel Szic, Ndlovu and Haegeman⁴⁾.

Obesity-associated adipose tissue enlargement is often associated with an elevated secretion of pro-inflammatory adipokines such as leptin and cytokines such as TNFα, whose epigenetic regulation has emerged as a potentially important determinant of gene expression⁽Reference Hermsdorff, Puchau and Zulet⁸⁶⁾. Thus, at baseline, obese women with better response to an energy-restricted dietary intervention designed to induce weight loss showed lower promoter methylation levels of leptin and TNFα than the non-responder group, which suggests that leptin and TNFα methylation levels could be used as epigenetic biomarkers concerning the response to a low-energy diet⁽Reference Cordero, Campion and Milagro⁸⁷⁾. Indeed, the methylation profile could help to predict the susceptibility to weight loss as well as to some comorbidities such as hypertension or type 2 diabetes⁽Reference Cordero, Campion and Milagro^87, Reference Campion, Milagro and Goyenechea⁸⁸⁾. Additional investigations concerning the interactions between obesity and inflammation under an epigenetic perspective have allowed identification of different CpG sites from Wilms tumour 1 and ATP10A genes, which show a differential response as a result of a hypoenergetic-diet-induced weight loss in human subjects by altering DNA methylation status of these specific genes indirectly related with inflammatory processes, suggesting that baseline DNA methylation patterns may be used as prognostic epigenetic markers that could help to predict weight loss⁽Reference Milagro, Campion and Cordero⁸⁹⁾.

Summing up, there is growing evidence suggesting that interindividual differences in obesity susceptibility depend not only on the DNA sequence (genetics) but also on epigenetic factors affecting gene expression such as DNA methylation, covalent histone modifications, chromatin folding and the regulatory actions of miRNA and polycomb complexes, in which inflammatory phenomena may be involved. Thus, epigenetics is providing novel insights into cellular identity, stem cell flexibility, tissue regeneration, tumorigenesis and aging and to understand monozygotic twin differences and interestingly the onset of chronic diseases in the adult such as obesity. The following aims are presently pursued in the ground of obesity and epigenomics: the early identification of epigenetic biomarkers concerned in individual's disease susceptibility and the description of weight lowering protocols for tailored dietary treatments/advice to avoid/neutralise likely adverse epigenomic events. Other questions that remain to be answered are to understand the regulation of epigenomic phenomena, the period(s) for intervention, the key nutritional factors and doses, which will allow diagnosis and prognosis implementation and will facilitate preventive/curative strategies in a personalised ‘epigenomically’ based manner to combat obesity.