Programming research: where are we and where do we go from here?

ABSTRACT

Convincing evidence has accumulated to show that both pre- and postnatal nutrition preprogram long-term health, well-being, and performance until adulthood and old age. There is a very large potential in the application of this knowledge to promote public health. One of the prerequisites for translational application is to strengthen the scientific evidence. More extensive knowledge is needed (eg, on effect sizes of early life programming in contemporary populations, on specific nutritional exposures, on sensitive time periods in early life, on precise underlying mechanisms, and on potential effect differences in subgroups characterized by, eg, genetic predisposition or sex). Future programming research should aim at filling the existing gaps in scientific knowledge, consider the entire lifespan, address socioeconomic issues, and foster innovation. Research should aim at results suitable for translational application (eg, by leading to health-promoting policies and evidence-based dietary recommendations in the perinatal period). International collaboration and a close research partnership of academia, industry, and small and medium enterprises may strengthen research and innovative potential enhancing the likelihood of translational application. The scientific know-how and methodology available today allow us to take major steps forward in the near future; hence, research on nutritional programming deserves high priority.

INTRODUCTION

Convincing evidence has accumulated to show that both pre- and postnatal nutrition preprogram long-term health, well-being, and performance until adulthood and old age (1, 2). This conclusion is supported by 3 separate lines of evidence: 1) lifetime experimental studies in animals, 2) historical and prospective observational studies in humans, and 3) experimental, hypothesis-testing trials in humans with long-term follow-up. A very large potential exists for application of this knowledge to preventive approaches, to improve health and well-being, to reduce costs for health care and social services, and to enhance the productivity and wealth of societies. The preventive potential is particularly important in view of rapid demographic changes in many societies. For example, the rapidly aging population in Europe, along with a low birth rate, is anticipated to greatly change the population demographics by 2060 (3). A marked decline in the working age population (ie, aged 16–65 y) and hence in the productive workforce, along with a large increase in the number of retired people (ie, aged >65 y), is predicted (Figure 1). This will result in a dramatic change from the current ratio of 4:1 of people of working age to people of retirement age, to only 2:1 by 2060. Therefore, a key challenge is to increase not only people’s lifespan but also the number of healthy and productive years, and to limit the burden of age-related, noncommunicable diseases. However, whether this goal can be reached is questionable given the dramatic increase in such diseases, in particular obesity and diabetes (4).

FIGURE 1.

Predicted demographic change in the European Union from 2008 to 2060 for women and men. Compared with 2008, in 2060 the proportion of the population at the retirement age of >65 y will have increased markedly, whereas the proportion of people of working age of ≤65 y will have decreased. Adapted from reference 3.

The lifetime burden from noncommunicable diseases can be reduced by nutritional approaches in the perinatal period, which therefore are an important element of health promotion, with potentially very large benefits. A solid science base is a prerequisite for translational application of programming knowledge to public health promotion. The effects of early life programming in contemporary populations with regard to long-term health and burden of adult disease need to be characterized in detail, along with the specific nutritional exposures that lead to programming in the preconceptual period, pregnancy, infancy, and early childhood. The precise mechanisms by which optimal nutrition can achieve programming effects on development, metabolic diseases, and health risk in later life must be explored. The potential effect differences in various subgroups of the population characterized, for example, by genetic predisposition or sex, need to be understood. In this manuscript, we aimed to review current knowledge, identify some key issues that need to be resolved, and describe opportunities for future research.

MECHANISMS OF EARLY NUTRITION PROGRAMMING

The precise mechanisms that underlie how early nutrition can cause programming of later health are unknown, but are thought to be associated with altered development of organ structure or persistent alteration at the cellular level (5). Some proposed mechanisms include

• Epigenetic memory: transcriptional modification (eg, DNA-binding proteins, histone acetylation, CpG methylation to 5-methyl-cytosine, X chromosome inactivation)

• Induction of altered organ structure [vascularization, innervation, juxtaposition (eg, position of hepatocyte, endothelial, and Kupffer cells during organogenesis may permanently modify metabolism), reduced nephron number]

• Alteration of cell number (hyperplasia, hypertrophy)

• Clonal selection (disproportionate growth of cells that proliferate rapidly under specific metabolic conditions)

• Metabolic differentiation (eg, hepatocellular polyploidization associated with enhanced metabolic activity)

The molecular mechanisms proposed include acute or persistently altered changes in gene expression through a variety of epigenetic pathways. During in utero or early postnatal development, short-term changes through environmental influences could permanently change organ development at a time of extreme vulnerability or “plasticity.” For example, experimental studies and human observations have shown that a reduction in nutrient and oxygen supply differentially affects the growth and development of organs and tissues. Organs affected include the lungs, kidney, gut, and liver. Additionally, clinical and experimental studies provide evidence of developmental changes in the homeostatic set points for many hormones and of alterations in tissue sensitivity to these hormones. Alterations of the fetal hypothalamic-pituitary-adrenal axis, central mechanisms that control energy balance and sympathoadrenal responses, are likely to be important mechanisms by which developmental exposures affect the subsequent responses of the offspring to stressful challenges. One of the keys to understanding how these changes are brought about is to establish whether the placenta plays a facilitatory or a protective role in the face of nutritional challenge. As a necessary prelude, one needs to establish which maternal exposures are modified by the placenta and how, and to determine what the vulnerable fetus actually experiences. Only then can we begin to unravel the pathways to in utero programming, which will lead to successful interventions in the mother (6).

Although, conceptually, epigenetic modification provides a framework for understanding how differences in the early environment can lead to permanent changes in metabolism and therefore long-term health risks, much work still has to be done to unravel the specific postepigenetic modifications involved in different disease processes. Given the enormous methodologic challenges, uncertain biological implications, and evidence of involvement of other processes, we conclude that there are inevitable limitations to focusing primarily on molecular markers of epigenetic processes.

EXPERIMENTAL STUDIES: STRENGTHS AND LIMITATIONS

A vast number of experimental studies that have used both small and large animal models together with either global changes in dietary intake or manipulation of individual macro- and micronutrients have shown the full extent to which normal developmental processes can be reset with potentially lifetime consequences (7, 8). These studies have shown that either a reduction or an increase in nutrient supply during the reproductive period can reset a range of physiologic control mechanisms, which include growth, metabolism, appetite control, cardiovascular function, and so forth. The large range of outcomes appears to be much more common in small animal models (9), which may be explained in part by the fact that the control animals are maintained within a confined environment. They are well fed, have raised fat stores, and thus become hypertensive (10). Consequently, the restriction of caloric intake to an amount that is closer to the animal’s food intake in its natural (outdoor) environment, irrespective of the in utero environment, results in raised activity, a lower body weight, less fat, and a lower blood pressure. Ultimately, this can extend the lifespan from 2.4 to 4 y (10, 11). Indeed, there is limited evidence from experiments involving both rats and sheep that a general reduction in food intake either throughout pregnancy or at defined stages of gestation can result in young adult offspring with lower blood pressure (12, 13). These findings emphasize the caution that must be applied to findings from developing experimental studies before they can be readily translated to the human situation.

ANIMAL MODELS OF MATERNAL OBESITY: DO THEY REFLECT THE MECHANISMS SEEN IN OVERWEIGHT WOMEN?

The current major concern to both health agencies and research funding bodies is the rise in the incidence of adult and childhood obesity (14). An increasing number of animal models are now being established in which the mother is made to become rapidly obese before becoming pregnant. In rodents this may be a result of feeding them a high-fat and/or high simple sugar–low polysaccharide and/or low-protein diet (15), whereas in sheep this is accomplished by feeding sheep ad libitum rather than to requirements (16). Although human studies show that birth weight is raised in overweight mothers, this does not appear to be the case in either small or large animal models of excess caloric intake (16, 17). It may be that the metabolic changes with these types of models do not really reflect the changes seen in obese women, who would not be expected to show the remarkable magnitude of hyperinsulinemia (eg, a 5-fold increase in plasma insulin up to 25 μU/mL) that is maintained through gestation (16). Indeed, the relative contributions of prepregnancy weight compared with weight gain through gestation remain largely unexplored. Interestingly, allowing pregnant sheep to eat ad libitum from midpregnancy increases birth weight but does not have a major effect on maternal fat mass (18). Irrespective of these nutritional and related factors it is clear that simply changing maternal diet around the time of conception or during gestation will not provide a simple solution to the potential long-term adverse outcomes of being born to a mother who is consuming too much food (19).

ADIPOSE TISSUE AND ITS CONTRIBUTION TO EXCESS BODY WEIGHT IN LATER LIFE: A PREVENTATIVE ROLE FOR BROWN ADIPOSE TISSUE?

It has become apparent recently that the increase in fat cell numbers with obesity can be established very early in life (20, 21) because the pool of white adipocyte precursors is determined around the time of birth (22). These landmark discoveries mean that the adipocyte research agenda is now focused on very early life events. Adipose tissue in the newborn comprises both brown and white adipocytes for which the relative recruitment depends on the maturity and thermal environment around the time of birth (23). Brown adipose tissue (BAT) is one factor that has been established as having a fundamental effect on energy homeostasis in rodents (24). It possesses the unique mitochondrial uncoupling protein 1, which enables the rapid generation of heat by nonshivering thermogenesis (25) and is retained throughout the life cycle in rodents (26). In large mammals such as sheep and humans, the abundance of BAT decreases through the postnatal period, but BAT has recently been shown to be present in adults (27). Both acute cold exposure (27, 28) and the prevailing environmental temperature can determine BAT activity in adult humans (29). These findings generate the exciting possibility that BAT function can be manipulated to prevent obesity in humans in a similar manner to its manipulation in mice. These studies show that the absence of BAT leads to obesity (30), whereas increased uncoupling protein 1 is protective against weight gain (31). If the origins of BAT are determined in early life (32), then this offers the potential for increasing its deposition at this time.

THE EFFECT OF EARLY-LIFE DIET ON LATER ADIPOSE TISSUE DYSFUNCTION

The extent to which early-life dietary interventions can be made to target the growth and development of white, as opposed to brown, adipocytes is also important. It is the central obesity that characterizes the metabolic syndrome, and metabolic complications specifically relate to excess white adipose tissue deposition (33). The magnitude of insulin resistance within this depot appears to be enhanced when nutritionally programmed offspring are raised within an obesogenic environment (34). At the same time, they become hypertensive (35), which suggests an exacerbated blood pressure response to obesity (13). These offspring, as adults, also exhibit increased ectopic lipid deposition in the left ventricle that is accompanied by a resetting of cardiac energy metabolism (36). A striking feature of this adaptation is that control subjects, despite being equally obese, had equally low levels of myocardial ectopic lipid as lean animals. Myocardial lipid infiltration is normally related to current diet, aging, and diastolic dysfunction and, clinically, is often seen in patients whose cardiac failure is associated with diabetes or obesity (37, 38). The potential to prevent this type of adverse complication may thus reside in ensuring that physical activity does not become limited in either childhood or adulthood. As such, this type of intervention may need to be targeted depending on social class and prevailing diet (19).

FEATURES OF OBSERVATIONAL EPIDEMIOLOGY RELEVANT FOR DEVELOPMENTAL PROGRAMMING RESEARCH

Observational epidemiology harvests the experience that lies hidden in the population regarding causality of diseases, such as relations between exposure and health outcomes. The great challenge is to be able to obtain this information in the most valid and cost-efficient way. The tradition and nature of observational epidemiology is to measure and collect as many data as possible. This approach to data collection is both a strength and a weakness compared with the more rigorous procedures adopted, for example, in conducting randomized controlled trials (RCTs). Many scientific advances are directly attributable to observational epidemiology because it is the unexpected associations that will advance our understanding, and it may be in just one corner of the data that information (the “hidden population experience”) brings us new and important knowledge regarding the exposure–disease relation. However, early researchers of the intrauterine origin of diseases were criticized for this strategy. The seminal, and at that time highly controversial findings from the late 1980s and early 1990s on associations between perinatal variables and the risk of chronic diseases such as ischemic heart disease and breast cancer many years later (39,40), were criticized for being the result of data “dredging”. However, such phenomena have now been observed in a multitude of other settings and these observations represent major milestones in the field of developmental programming.

Nonetheless, one has to be aware of the limitations of such a strategy, which uses highly exploratory data analyses that also result in false signals. This has led to critique and self-criticism in the field of epidemiology (41, 42). Among the lessons learned from this process is that, whenever possible, one should use large studies, in which random error is less likely to occur. In addition, one should establish fora with open discussions, and establish mechanisms whereby one can do parallel analyses or confirmatory analyses of ≥2 data sets, preferably before publishing unexpected findings. Indeed, this is something that has been practiced in the Early Nutrition Programming Project (www.early-nutrition.org) across the Danish and Norwegian birth cohort studies (43–47).

CHARACTERIZATION OF STUDIES IN THE FIELD OF DEVELOPMENTAL PROGRAMMING

After the above, more general, words of caution, we return to the 2 questions, where are we and where do we go? Is it possible to characterize the studies that have been made so far, and those needed in the future?

The first studies leaned on crude growth measures that were related to simple mortality or morbidity measures (39, 40). A further series of studies applied different strategies. Many focused on specifying the exposures that were believed to be involved, especially diet in pregnancy (48, 49). There are now a multitude of studies in Europe and other places that have assessed dietary intake in pregnancy, several of which have received funding from the European Commission. Others refined the postnatal growth exposure measures more to develop postnatal trajectories and related those to morbidity measures such as diabetes (50), and some studies even established intrauterine growth trajectories that have made and continue to make important scientific advances.

The next wave of studies applied a multitude of approaches. One approach has been to establish prospective cohorts with assessment of diet before pregnancy, a design that represents immense logistical challenges, but which seem to have been overcome in the Southampton Women’s Survey (51). Another approach has ensured the collection of biomaterials for assessment of genetic and metabolic markers or of environmental toxicants as was done in some European Commission–funded consortia (eg, project NewGeneris: Newborns and Genotoxic exposure risks). A further approach has been to perform physiologic investigations in children to explore biomarkers of health before disease manifestation [eg, ultrasound measurements of the carotid arteries to identify early signs of atherosclerosis (52), or DXA scans to identify children with reduced bone mineral density (53)].

Several cohorts have collected genetic materials, for example, to examine gene–nutrient interactions or to explore the epigenetic effects of early nutritional exposures. A separate line has been to study monozygous twins. In this model, variation in birth weight between genetically identical twins is linked to variation in later health (eg, insulin resistance in adulthood) and any association observed is attributed to environmental factors acting during fetal life (54). One important approach was to make studies larger, such as was accomplished in the Danish National Birth Cohort (55, 56) and the Norwegian Mother Child Cohort (57, 58). Each of these studies recruited ≈100,000 pregnancies. A specific feature of these 2 particular studies is that they share many methodologies (eg, estimation of maternal dietary intake). Thus, joint data evaluation is feasible, which provides greater statistical power to allow the exploration of true morbidity measures in the offspring (eg, asthma, cancer, childhood diabetes), rather than only child growth measures such as body mass index (BMI; in kg/m²). The large size of these cohorts also means that focused studies based on narrowly selected subcohorts can be conducted.

FUTURE OBSERVATIONAL STUDIES IN THE FIELD OF DEVELOPMENTAL PROGRAMMING

No clear answer can be given to the question as to where we go now in longitudinal epidemiology. We have seen that methodologic approaches grow like branches on a tree, simultaneously and in many different directions, and this multitude and heterogeneity of approaches will ensure that we will learn more in the years to come with regard to the highly multifaceted and complicated process of developmental programming. It may even be counterproductive to decide on or advise on the best study designs for future studies; this is up to the creativity of the individual researchers and their groups. However, some desirable features can be emphasized. Large studies that ensure high statistical power should be aimed for. In new data collections, a minimum of compatibility with other studies should be sought. More studies with preconceptional dietary assessments should be attempted. Focused follow-up studies from existing larger studies are highly desirable. As regards the latter, in the Danish National Birth Cohort, between ≈6000 and 20,000 children will become 15 y of age between 2013 and 2016. It therefore represents a unique sampling frame from which we can select children with particular exposure characteristics that represent potential programming effects (eg, high and low folate intake in pregnancy, high and low prepregnancy BMI or gestational weigh gain, and maternal gestational diabetes).

HUMAN INTERVENTION TRIALS

Human intervention trials are essential to firmly establish cause-and-effect relationships, to show the suitability and safety of interventions, and to provide a firm evidence base for policy decision and implementation of action. The strongest evidence is derived from well-designed and well-conducted RCTs. However, clinical trials that meet current scientific quality standards and that are large enough to provide sufficient statistical power to show lasting effects over several years of follow-up are demanding, labor intensive, and costly. We estimate that the evaluation of a novel dietary approach in pregnancy or infancy, with the collection of the preclinical information necessary to start a clinical trial, and the execution of a large-scale clinical trial, will typically have a minimum cost of between 1 and 5 million Euro. Therefore, prioritization is mandatory to invest current limited resources in the most promising human trials.

Among the key criteria for allocation of a high level of priority to a clinical trial would be that a proposed intervention has the potential for a relevant benefit on an outcome of major importance in the population, has a high likelihood of translational application, is acceptable to the population, and is economically viable. In Europe, cardiovascular diseases induce the dominant burden of disease (23% of disability-adjusted life-years lost from noncommunicable disease) and premature mortality (52% of deaths) (59). Therefore, a certain proportion of human intervention trial approaches should be dedicated to potentially modifiable programming effects on the risks of cardiovascular disease and associated factors such as obesity, metabolic syndrome, hypertension, and others.

The long-term risk of obesity and associated disorders has been related not only to a variety of environmental and genetic factors, but also to fetal and postnatal nutrition and growth in a large number of research studies (60) (Figure 2). Among the variables of potential importance that should be considered are maternal prepregnancy weight and nutritional status, diet and weight changes in pregnancy, gestational diabetes, placental function, markers of fetal growth, breastfeeding and other aspects of infant feeding, and infant growth. The potential relevance of genetic variation, ethnicity and geographic background, environment, sex, physical activity, and smoking needs to be explored. More trials are needed in socioeconomically challenged populations, both in developed as well as in developing and threshold countries, and in populations with a high biological risk, such as pregnant women.

FIGURE 2.

Factors related to the development of obesity and associated disorders.

Clinical trials on nutritional interventions must achieve current methodologic standards (61, 62), which makes their execution both challenging and costly. However, it is more cost effective to invest in a trial with good standards and adequate statistical power that will allow firm conclusions than to design a low-cost approach that does not provide conclusive results. One very cost-effective way to evaluate the long-term effects of early diet is to follow up existing cohorts of subjects who had been exposed previously to randomized nutritional interventions in pregnancy, lactation, or infancy. Such open-label extensions of RCTs, with further follow-up of subjects participating in RCTs after the unblinding of the trial to explore potential longer-term effects of the intervention, are obviously subject to potential bias because of lack of blinding. In addition, an even more important threat to validity may be slippage in relation to the sample randomized in the preceding RCT (63).

In addition, new trials need to be established to evaluate new and refined interventions. Targets for intervention include total diet and dietary patterns, as well as specific modifications of macro- and micronutrient intake. New trials are also needed to confirm positive results from a previous trial in an independent second study.

One very encouraging example of the usefulness of performing a human intervention trial is the European Childhood Obesity Trial, in which 1678 healthy infants born at term in 5 European countries were enrolled to evaluate the potential of dietary modification in early weight gain. This question is of importance, because rapid weight gain in infancy and the first 2 y of life has been closely linked to a variety of adverse outcomes, which include overweight and obesity, visceral obesity, insulin resistance, markers of cardiovascular risk, and asthma (64–66). Our study concept was based on the observation that breastfeeding, which supplies less protein than conventional formulas, is associated with some 20% lower obesity at later ages (67, 68). The hypothesis was examined that higher protein intakes than those of conventional formulas enhance plasma and tissue levels of insulin-releasing amino acids and of insulin and insulin-like growth factor 1, and thereby increase weight gain and adipogenic activity (Figure 3). Therefore, formula-fed infants who participated were randomly assigned to receive, for at least the first year of life, infant and follow-on formulas with either a higher protein content or a lower protein content more similar to that provided by breastfeeding (66). Infants fully breastfed for ≥3 mo served as a reference group. The results obtained at 2 y of age show that early reduction of protein intake prevents excessive growth and normalizes mean weight for length and body mass achieved at 2 y, relative to healthy breastfed infants and to the international growth standards of the World Health Organization (70) (Figure 4). On the basis of the results of observational studies, the reduction of weight gain in the first 2 y achieved with the modified infant feeding approach can be predicted to reduce obesity risk by up to 13% at 14–16 y. A preliminary assessment of the health economic consequences predicts that this degree of obesity prevention could result in cost savings of >2 million Euros for each year’s birth cohort in a country such as Germany if only one-half of the infants born benefitted from exposure to formula with reduced protein content (Niels Straub, personal communication, 2011). This is a very encouraging result, because such modified dietary products for infants can be introduced easily on a population-wide basis without appreciable added costs or change in lifestyle habits. It therefore represents a very cost-effective preventive strategy.

FIGURE 3.

Early protein hypothesis: a high dietary supply of protein in infancy may enhance plasma and tissue concentrations of insulinogenic amino acids, insulin, and insulin-like growth factor 1 (IGF1); weight gain during the first 2 y of life; and adipogenic activity. The long-term risk of obesity and associated disorders may therefore also be enhanced. Adapted from reference 69.

FIGURE 4.

BMI at the age of 2 y in children who were breastfed during the first months of life or who were randomly assigned to receive formula with lower or higher protein contents during the first year of life [z scores (±SD) relative to World Health Organization references]. The P value was based on ANOVA adjusted for baseline value. Adapted from reference 69.

It appears highly worthwhile to follow up this cohort of children further into school age, to explore the longer-term effects of the early dietary intervention on body size and body composition, as well as functional outcome markers. Moreover, this large, multicenter study which showed that dietary modification of infant growth trajectories had an expected long-term health benefit, should stimulate the exploration of more refined approaches to support optimal early growth and long-term health.

Open questions remain. One would wish to explore whether the observed effects can be reproduced in another study, and which dietary factors are of key importance (eg, the intake of total protein, milk protein, or specific protein components). Moreover, one would wish to characterize during which time period in early life the dietary modulation of weight gain is of appreciable importance. It would also be desirable to define specific biomarkers that reliably predict growth patterns and long-term outcome and hence would enable the exploration of different interventions without the need to perform difficult and expensive long-term studies, at least during an initial exploration.

CONCLUSIONS AND PERSPECTIVES

Major progress has been achieved in research on early nutrition programming, but further challenges lie ahead. In our view, programming research should now aim at filling the existing gaps in scientific knowledge, consider the entire lifespan, address socioeconomic issues, foster innovation, and aim at results that are suitable for translational application (eg, by leading to health-promoting policies and recommendations on diet in the perinatal period). Major unresolved questions include understanding the underlying mechanisms, determining sensitive time periods for specific programming effects, identifying nutrients and dietary interventions with optimal benefits, and investigating the interaction of diet and genetics as well as sex-specific effects in programming. It is also important to better understand which programming effects might be reversible, and to what extent. The further exploration of programming effects in humans will be markedly facilitated by the development of valid and specific biomarkers.

The field will likely benefit from a close research partnership between academia, industry, and small and medium enterprises, to strengthen research and innovative potential as well as to enhance the likelihood of translational application. International collaboration that extends beyond European borders is highly desirable, to pool scientific know-how and resources, to share methods and technology, to explore communalities and differences across populations, and to apply programming knowledge to also benefit disadvantaged populations. One platform for facilitating collaboration, disseminating information, and enhancing scientific interaction, and for strengthening the training of new investigators is the Early Nutrition Academy (www.early-nutrition.org), which has evolved from the Early Nutrition Programming Project. Such collaborative research approaches should ultimately lead to evidence-based conclusions and recommendations on the desirable diet before and during pregnancy, during lactation, and during infancy, to promote long-term health, well-being, and performance of mother and child, and preferably lead to easy-to-apply dietary advice and food products with enhanced health benefits. The opportunity is there, and scientific know-how and currently available methodology will allow us to achieve major steps forward in the near future. Therefore, research on nutritional programming deserves to get high priority.

We gratefully acknowledge the dedicated support of Susanne Höckel, Brigitte Brands, and Yuliya Sarmant and the invaluable contributions of Keith Godfrey, Southampton, and Lucilla Poston, London. Support from the Child Health Foundation, Munich, and the LMU innovative research priority project MC-Health (sub-project I) is also recognized.

The authors’ responsibilities were as follows—BK, MES, and SFO: preparation, revision, and final approval of the manuscript. None of the authors declared a conflict of interest according to the guidelines of the International Committee of Medical Journal Editors.

FOOTNOTES

2

Presented at the conference “The Power of Programming: Developmental Origins of Health and Disease,” held in Munich, Germany, 6–8 May 2010.

3

This manuscript does not necessarily reflect the views of the Commission of the European Communities and in no way anticipates future policy in this area.

4

Supported in part by the Commission of the European Communities, specific RTD Programme, “Quality of Life and Management of Living Resources,” within the 6th Framework Programme, contract no. 007036. Additional support was provided by the Child Health Foundation, Munich, and the LMU innovative research priority project MC-Health (sub-project I). BK is the recipient of a Freedom to Discover Award from the Bristol-Myers-Squibb Foundation, New York, NY.