Adiposity in early life

The recent finding that obese individuals possess approximately 30% more fat cells than lean counterparts indicates that it is not only the accumulation of excess lipid within existing adipocytes that contributes to obesity⁽Reference Spalding, Arner and Westermark⁴⁾. What is even more striking about this finding is that the difference in fat cell number is apparent from the earliest age that measurements were made, i.e. approximately 5 years. Taken together these findings strongly suggest that the foundations for excessive adiposity are laid down during infancy and as such any therapeutic strategies aimed at preventing later obesity need to be similarly targeted. These findings are in contrast to previous work, primarily in the rat, that indicates that as there are so many fat cells present in all individuals this increase in fat cell numbers makes very little contribution to total fat mass⁽Reference Garaulet, Hernandez-Morante and Lujan⁵⁾, a concept that now needs re-examining.

In large animals, including sheep and pigs, the postnatal period represents the main critical window of fat growth, although the growth differs in relation to the anatomical location⁽Reference Symonds and Lomax⁶⁾. For example, in the neonatal pig fat is only present in the subcutaneous region⁽Reference Mostyn, Litten and Perkins⁷⁾ and is rarely found around internal organs, whilst in newborn sheep it is only present around the kidneys and heart⁽Reference Alexander and Bell⁸⁾. In both species fat distribution changes substantially over the first few weeks of life, as fat is the most rapidly-growing organ through this period and is soon found throughout the body⁽Reference Clarke, Buss and Juniper⁹⁾. On the other hand, in term infants of normal body weight fat is present around the internal organs as well as subcutaneously, and it is the latter depot that grows substantially after birth⁽Reference McCance and Widdowson¹⁰⁾. An appreciable amount of this fat can then be mobilised during the weaning period, although this process has not been well documented. There is, however, increasing evidence that excess fat deposition can occur at this time⁽Reference Ibanez, Suarez and Lopez-Bermejo¹¹⁾ and is very likely to have adverse long-term consequences.

Metabolic precursors for fetal adipose tissue development and fetal activity

The major metabolic substrate in fat accumulation in the fetus is glucose, although in the sheep acetate also makes an important contribution⁽Reference Symonds, Lomax and Kenward¹²⁾. In terms of other potential substrates for adipogenesis in the fetus fructose could be included, as it is the most abundant carbohydrate in the fetal circulation through gestation⁽Reference McGowan, Aldoretta and Hay^13, Reference Jauniaux, Hempstock and Teng¹⁴⁾ (Table 1). The function of fructose in the fetus has remained elusive to date as it only contributes to approximately 20% of oxidative requirements⁽Reference McGowan, Aldoretta and Hay¹³⁾. A further reason why attention needs to be focused on the role of fructose in early life is the substantial increase in its availability within Western-style diets because of the approximately 20-fold increase in the use of high-fructose maize syrup over the past 15 years⁽Reference Apovian¹⁵⁾.

Table 1. Summary of the plasma concentrations (mm) of the primary carbohydrates in the fetal circulation

Changes in the fetal carbohydrate environment may not only impact on organogenesis but also on fetal behaviour, as glucose is the primary determinant of fetal breathing movements and as such is an indirect indicator of activity⁽Reference Fowden, Harding and Ralph^16, Reference Symonds, Bird and Clarke¹⁷⁾. It could well be the case that subtle changes in early behavioural patterns may set the differences in activity, as defined by non-exercise activity and thermogenesis, that have been shown to have the potential to contribute to a gradual accumulation of adipose tissue through adulthood⁽Reference Levine, Lanningham-Foster and McCrady¹⁸⁾. At the same time in the fetus the entrainment of breathing movements with sleep state and swallowing⁽Reference Trahair and Harding^19, Reference Ross and Nijland²⁰⁾ may have the potential to influence development of the digestive system and even gut microflora. To date, this aspect of energy homeostasis has been completely ignored, but it could be important given the pivotal role that secretion of gut hormones such as oxyntomodulin and peptide tyrosine–tyrosine (both gut peptides released from intestinal enteroendocrine cells in response to a meal) have in the regulation food intake in adults⁽Reference Chaudhri, Small and Bloom^21, Reference Chaudhri, Wynne and Bloom²²⁾. The extent to which gut microflora may also differ between infants that are exclusively breast-fed or formula-fed remains to be established, although it is notable that the latter consume more food earlier in life and this intake is positively correlated with weight gain⁽Reference Ong, Emmett and Noble²³⁾.

Diet and physical activity in childhood

It is not only the consumption of excess or inappropriate amounts of carbohydrate by the mother during pregnancy that needs to be considered but also the offspring's energy intake during the weaning period and through childhood. A striking change in dietary patterns of children between 1950 and the early 1990s is that although total energy consumption has not changed, a very different pattern of carbohydrate consumption has occurred, primarily as a result of greater sugar consumption (Fig. 1)⁽Reference Prynne, Paul and Price²⁴⁾. This pattern of change has been accompanied by a dramatic change in the types of drinks consumed, with tea being replaced by soft drinks and juices. In addition, the total consumption of soft drinks is now more than double the amount of tea consumed by children in the 1950s (Fig. 1). In both children and adolescents increased soft drink consumption is further related to raised salt intake⁽Reference He, Marrero and MacGregor²⁵⁾, thus also impacting on blood pressure control⁽Reference He and MacGregor²⁶⁾. The adverse outcome in terms of overall energy balance may be even worse in children with the rise in sedentary behaviour such as watching television⁽Reference Wiecha, Peterson and Ludwig^27, Reference Danner²⁸⁾. It is now recognised that the earlier a child adopts a pattern of increased television viewing then the more difficult it is to break this type of behaviour, although it can be dissociated from changes in overall physical activity⁽Reference Ekelund, Brage and Froberg^29, Reference Reilly³⁰⁾. In addition, there is a positive relationship between television viewing and consumption of dietary components advertised, including soft drinks⁽Reference Wiecha, Peterson and Ludwig²⁷⁾.

Fig. 1. Summary of the main dietary changes in children between 1950 and 1993. (A), Complete diet; (B) individual components of the diet ((□), tea; (▪), soft drinks and juices). NSHD, National Survey of Health and Development; NDNS, National Diet and Nutrition Survey. (Based on Prynne et al. ⁽Reference Prynne, Paul and Price²⁴⁾.)

The impact of changes in early carbohydrate consumption is complex, as early exposure to sucrose can actually reduce the motivation to acquire sucrose in young mice⁽Reference Frazier, Mason and Zhuang³¹⁾. However, early sucrose exposure subsequently promotes weight gain in adulthood when free access is allowed to palatable energy-dense foods. Whilst it is recognised that the sugar content of some soft drinks may have been reduced in recent years by replacement with artificial sweeteners, it does not necessarily improve overall health. Importantly, it could even contribute to a further increase in the total carbohydrate consumption, as studies in adult rodents indicate that artificial chemicals such as saccharin are more addictive than drugs such as cocaine⁽Reference Lenoir, Serre and Cantin³²⁾.

Genetic versus dietary causes of obesity and type 2 diabetes

Until very recently there have been no consistent reports of gene variants that influence the risk of obesity as well as predisposing to type 2 diabetes. Recent publications utilising genome-wide association studies based on a very large number of cohorts have, however, shown strong relationships between a common variant of the fat mass and obesity-associated (termed FTO) gene and BMI that is prevalent from as early as 7 years of age⁽Reference Frayling, Timpson and Weedon³³⁾. Indeed, this variant predisposes individuals to diabetes through an effect on body mass⁽Reference Frayling, Timpson and Weedon³³⁾, a relationship that has been confirmed in a separate study based on a European cohort⁽Reference Dina, Meyre and Gallina³⁴⁾, although not in Chinese populations⁽Reference Li, Wu and Loos³⁵⁾ or Oceanic populations⁽Reference Ohashi, Naka and Kimura³⁶⁾. Relative cohort size is, however, critical in determining such associations, as a minimum of 12 000 subjects are needed to show a strong relationship between FTO and BMI⁽Reference Freathy, Timpson and Lawlor³⁷⁾.

The potential divergence in response to variations in the FTO gene is further indicated by its widespread tissue distribution⁽Reference Frayling^38–Reference Gerken, Girard and Tung⁴⁰⁾, with gene expression being greatest in the hypothalamus⁽Reference Gerken, Girard and Tung⁴⁰⁾. Moreover, in both mice⁽Reference Gerken, Girard and Tung⁴⁰⁾ and rats⁽Reference Fredriksson, Hagglund and Olszewski⁴¹⁾ hypothalamic FTO mRNA abundance is modulated by nutritional status, but the response is species dependent; in mice starvation down regulates FTO mRNA transcription⁽Reference Gerken, Girard and Tung⁴⁰⁾ whereas in rats the opposite effect is seen⁽Reference Fredriksson, Hagglund and Olszewski⁴¹⁾. A number of genetic mouse models, however, indicate rapid dietary-induced obesity after consumption from 12 weeks of age of a high-fat diet in which energy intake from fat is increased 6-fold (and as such is not physiological); commencing at 6 weeks of age has no effect on FTO gene expression in either the hypothalamus or a range of peripheral tissues⁽Reference Stratigopoulos, Padilla and Leduc⁴²⁾. Clearly, more information is required in relation to the regulation of the FTO gene from a developmental point of view and how it responds to an environment of excess rather than insufficient energy supply. In this context low physical activity may amplify its association with obesity⁽Reference Andreasen, Stender-Petersen and Mogensen⁴³⁾, although to date not all studies confirm this proposal⁽Reference Speakman, Rance and Johnstone⁴⁴⁾.

As the number of genome-wide association studies increase both in number and scope there is an ever-increasing number of genes and loci that appear to be related to type 2 diabetes⁽Reference Scott, Mohlke and Bonnycastle^45, Reference Zeggini, Weedon and Lindgren⁴⁶⁾, although their predictive value for the population remains in doubt⁽Reference Lango, Palmer and Morris⁴⁷⁾. This outcome is not unexpected given the substantial number of changes in organ function that accompany type 2 diabetes, of which the main defects relate to impaired function of the pancreas together with insulin resistance⁽Reference Elbers, Onland-Moret and Franke⁴⁸⁾. To gain further insights into their relevance for glucose homeostasis such findings need to be related to changes in dietary preference and/or tissue function. For example, one genetic predisposition to sugar consumption relates to a variant of GLUT2 that is accompanied in both young and older adults by a greater intake, including that of fructose, if consumed as sweetened beverages⁽Reference Eny, Wolever and Fontaine-Bisson⁴⁹⁾ (summarised in Fig. 2). This finding is not unexpected as GLUT2 in the brain has been proposed to regulate glucose sensing, thereby having a similar function to that shown in pancreatic β-cells as well as having a primary role in regulating glucose homeostasis⁽Reference Marty, Dallaporta and Thorens⁵⁰⁾. Within the human cohorts examined, however, there is no relationship between excess sugar intake and an overall increase in food intake or in BMI⁽Reference Eny, Wolever and Fontaine-Bisson⁴⁹⁾. These observations are partly in contrast to findings from knock-out studies in which mice lacking GLUT2 eat more of a novel powdered diet, at least in the short term, but show no difference in body weight⁽Reference Bady, Marty and Dallaporta⁵¹⁾. It may be that in the human subject it is not until an imbalanced dietary pattern is attained that any pathological outcomes are manifest, such as when a dietary intake attained in which >50% total energy intake is derived from one macronutrient, be that carbohydrate or fat⁽Reference Marti, Martinez-Gonzalez and Martinez⁵²⁾.

Fig. 2. Summary of the effect of a gene variant for GLUT2 on carbohydrate intake in the adult. (A) Intake (g/d) of sucrose (▪), fructose ( ) and glucose (□) by Thr/Thr and Thr/Ile+Ile/Ile variants. The difference between the variants for total intake of the three sugars was significant (P<0·05). (B) Consumption (no. of servings per d) of sweetened beverages and sweets for the Thr/Thr (□) and Thr/Ile+Ile/Ile (▪) variants. Values are means with their standard errors represented by vertical bars. The difference between the variants was significant for sweetened beverages (P=0·002) and for sweets (P=0·0004). (Based on Eny et al. ⁽Reference Eny, Wolever and Fontaine-Bisson⁴⁹⁾.)

One pivotal factor determining diet and lifestyle from the time of conception is social class. The substantial impact that affluence has on, for example, the levels of breast-feeding and healthy adult eating patterns, as well as the proportion of the childhood population that is obese, is illustrated in Fig. 3⁽⁵³⁾. This factor is clearly one that may be implicated in epigenetic regulation of gene action⁽Reference Szyf, McGowan and Meaney⁵⁴⁾, which has gained substantial interest in relation to obesity and diabetes from mouse studies⁽Reference Waterland and Jirtle⁵⁵⁾. The translational relevance of such studies to human subjects must, however, be viewed with considerable caution, as there are far more imprinted genes in the mouse, which may make it unsuitable for disease pathway comparisons with larger mammals⁽Reference Jirtle and Skinner⁵⁶⁾.

Fig. 3. Summary of the potential influence of social class on the incidence of breast-feeding, adult healthy eating (defined as the daily consumption of five or more portions of fruit and vegetables) and childhood obesity. Percentages of the population are given in relation to the best five (□) and worst five (▪) performing areas of the UK⁽⁵³⁾.

Nutritional programming and epigenetic adaptations: responses to extreme dietary manipulations

The epigenome is most susceptible to dysregulation during those periods of life associated with the greatest changes in organ development and growth, i.e. embryogenesis, fetal and neonatal development, puberty and final old age⁽Reference Jirtle and Skinner⁵⁶⁾. For example, in the more-extreme challenge of uterine artery ligation in the late-gestation rat, which reduces uterine artery blood flow by approximately 50%⁽Reference Ogata, Bussey and Finley⁵⁷⁾, substantial intrauterine growth retardation occurs⁽Reference Wigglesworth⁵⁸⁾. When these offspring are cross-fostered onto control dams they show rapid catch-up growth, an adaptation that is likely to be as important in determining the subsequent adverse outcomes as the preceding growth restriction in utero ⁽Reference Symonds⁵⁹⁾. As adults these growth-manipulated offspring exhibit type 2 diabetes that is associated with progressive epigenetic silencing of the homeobox 1 transcription factor Pdx1 ⁽Reference Park, Stoffers and Nicholls⁶⁰⁾, which is critical for pancreatic β-cell function and development⁽Reference Park, Stoffers and Nicholls⁶⁰⁾. Nevertheless, during the neonatal period the reduction in Pdx1 expression could be reversed in vitro by inhibition of histone deacetylase action. Potential epigenetic effects may well be extended to a number of other tissues and cells and could include the regulation of the intracellular energy locus, as recently shown in vitro (using human neonatal skin fibroblasts), at least under conditions of zero or very high (i.e. 1000 mg/l) glucose concentration⁽Reference Murayama, Ohmori and Fujimura⁶¹⁾.

Further evidence that DNA methylation may have an important role in obesity comes from studies that have utilised the agouti mouse, in which a shift in coat colour from pseudo-agouti to a mottled and/or yellow appearance is accompanied by an increase in mature body weight to >40 g compared with approximately 30 g in controls⁽Reference Wolff, Kodell and Moore^62, Reference Morgan, Sutherland and Martin⁶³⁾. In this model exposure to a diet that is artificially high in methyl donors as a result of an approximately 9-fold increase in choline and folic acid, approximately three times more methionine and approximately sixty times more vitamin B₁₂ compared with control diets⁽Reference Wolff, Kodell and Moore^62, Reference Reeves⁶⁴⁾ results in a shift in the methylation pattern and thus a comparable change in the coat colour distribution⁽Reference Waterland and Jirtle⁵⁵⁾. This magnitude of change for each of these nutrients is extreme, especially when compared with the approximately doubling of folic acid that has accompanied dietary fortification in America⁽Reference Jacques, Selhub and Bostom⁶⁵⁾, although re-cycling of these nutrients is very different in rodents because they exhibit croprophagia⁽Reference Devlin, Bottiglieri and Domann⁶⁶⁾. However, as the majority of these mice (i.e. approximately 90%) show a coat colour that is mottled in appearance there is actually little, if any, effect on the relative body-weight distribution, with the majority of mice continuing to have a high body weight and thus increased fat mass⁽Reference Waterland and Jirtle⁵⁵⁾.

The recent finding that exposure of pregnant and lactating mice to the environmental oestrogen bisphenol A, which has been linked to later obesity and cancer⁽Reference Maffini, Rubin and Sonnenschein⁶⁷⁾, results in a very subtle shift in the number of yellow offspring but has little effect on the total number of heavy (and thus obese) mice⁽Reference Dolinoy, Huang and Jirtle⁶⁸⁾. Moreover, exactly the same shift in coat colour can be obtained by adding high quantities of methyl donors or the phyto-oestrogen genistein to the diet⁽Reference Dolinoy, Huang and Jirtle⁶⁸⁾. It therefore simply reflects the same effects both diets have in comparison with controls irrespective of the additional dietary challenge⁽Reference Waterland and Jirtle^55, Reference Dolinoy, Weidman and Waterland⁶⁹⁾. There is thus not convincing evidence of an ‘additional protective effect’ of a hypermethylating diet against the epigenetic response to bisphenol A per se⁽Reference Dolinoy, Huang and Jirtle⁶⁸⁾.

It has also recently been claimed that a similar magnitude of dietary methyl donor supplementation may prevent the shift in population distribution towards an increase in body weight⁽Reference Waterland, Travisano and Tahiliani⁷⁰⁾. On closer examination of the published results it is apparent that the only change in body-weight distribution is in the F3 generation, at which time in the unsupplemented group there is a shift from 54% to 72% of all animals being >50 g compared with no real change in the methyl-supplemented group, which remained at approximately 50%. This apparent absence of a change in body-weight distribution is only because there are approximately 20% more mice in the supplemented group that all appear to have a body weight <40 g. Given that in this strain of mice the females can have up to twice as much fat as males, and thus show a much wider distribution in body weight, this result could simply be an artifact of an unequal distribution of males and females between the two groups rather than being a direct response to the excess methyl donor intake that the animals have been exposed to for three generations. Clearly, the translation of such types of extreme nutritional interventions to the human population must be considered with the utmost caution.