Skip to main content Accessibility help
×
Home

Information:

  • Access
  • Cited by 26

Actions:

      • Send article to Kindle

        To send this article to your Kindle, first ensure no-reply@cambridge.org is added to your Approved Personal Document E-mail List under your Personal Document Settings on the Manage Your Content and Devices page of your Amazon account. Then enter the ‘name’ part of your Kindle email address below. Find out more about sending to your Kindle. Find out more about sending to your Kindle.

        Note you can select to send to either the @free.kindle.com or @kindle.com variations. ‘@free.kindle.com’ emails are free but can only be sent to your device when it is connected to wi-fi. ‘@kindle.com’ emails can be delivered even when you are not connected to wi-fi, but note that service fees apply.

        Find out more about the Kindle Personal Document Service.

        Nutritional models of the developmental programming of adult health and disease
        Available formats
        ×

        Send article to Dropbox

        To send this article to your Dropbox account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you use this feature, you will be asked to authorise Cambridge Core to connect with your <service> account. Find out more about sending content to Dropbox.

        Nutritional models of the developmental programming of adult health and disease
        Available formats
        ×

        Send article to Google Drive

        To send this article to your Google Drive account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you use this feature, you will be asked to authorise Cambridge Core to connect with your <service> account. Find out more about sending content to Google Drive.

        Nutritional models of the developmental programming of adult health and disease
        Available formats
        ×
Export citation

Abstract

The ability to not only replicate but also extend the findings from both historical epidemiological studies and contemporary cohorts of the developmental programming of later disease are critical if the mechanisms by which early diet impacts on later disease are to be fully understood. To date, a plethora of models have been established, with the range including global changes in dietary input, imbalanced diets and diets deficient in single nutrients. Key factors in translating these findings to the human situation are the pronounced differences in the relative growth and development between large and small mammals from the time of conception through pregnancy, lactation and weaning. This disparity is reflected in the very different nutritional requirements between species and the substantial divergence between rodents and large animals in the ontogeny of many of the organ systems that are nutritionally regulated. For example, hypothalamic circuitry is much more developed in species with a long gestation and offspring are born with a mature hypothalamic–pituitary axis in sheep and man compared with mice and rats. Similarly, nephron number is established towards the end of gestation in large mammals compared with the lactational period in rats. These types of differences will impact on the ability of individual and combined nutritional interventions to reset developmental processes, and may be further compounded by the gender of a fetus. The challenge for future work in this exciting and dynamic area of research is to utilise these marked comparative differences to generate imaginative nutritional interventions in order to improve the viability, health and well-being of the offspring.

Research into ‘the developmental origins of adult health and disease’ has made considerable progress over the past two decades, but a clear consensus on the exact nutrients involved and their mechanisms remains to be established. This problem relates, in part, to the long-term developmental time frame in which changes in metabolism and cardiovascular function occur. At the same time, the discipline has had to contend with the dual challenges of integrating findings from lifetime epidemiological studies that have largely focused on birth weight and its relationship (or otherwise) with adult disease and the ability to incorporate these results into appropriate nutritional intervention studies using animal models. It is now established that there are many potential influences on offspring outcome, including maternal body composition, age and parity, genetic constitution and macro- and micronutrient intake and handling, size, shape and number of offspring, their gender and type of lactational diet. Thus, it is not unexpected that as knowledge of the intricacies involved expands, a substantive number of publications have, in retrospect, been misleading because of inappropriate statistical analysis and/or mismatches in the different groups (of which some key examples will be summarised).

In the present brief review an attempt will be made to provide an overview of some of the major challenges for both human and animal studies, in conjunction with optimal experimental paradigms that may be utilised in future research aimed at elucidating the precise mechanisms by which changes in the maternal diet during reproduction can impact on the lifetime health of resulting offspring. Particular emphasis will be given to the applicability of the animal model that has been utilised and how similarities in the reproductive process may enable its best use in future examinations of the nutritional programming of adult health and disease.

Human models: historical and contemporary models of fetal programming

The majority of the early work conducted by David Barker and colleagues utilised data from historical cohorts born in the 1930s and benefitted from meticulous birth records, often kept by the same individual over long periods of time(1). This approach has enabled clear relationships between not only the size and shape of an infant at birth but also its accompanying placental mass and later hypertension to be established(1). Subsequently, the recruitment of long-term historical records from Finland has enabled longitudinal studies of infant growth to be related to adult insulin sensitivity(2). More recently, the use of nutritional interventions of preterm formula in randomised controlled studies has emphasised the impact of inappropriate growth in early infancy on later disease risk(3). However, with more contemporary studies and the rise in both childhood and adult obesity it is now apparent that the translation of findings from such studies into current lifestyle interventions is very difficult. This situation results, at least in part, from the complexity of the biological processes involved and the challenge of effecting lifestyle changes in activity and dietary intake. At the same time, the causes of the ongoing epidemic of obesity, and the predicted increase in associated renal diseases and CVD, are multifactorial(4) and may be either exacerbated, or reduced, by early dietary exposure(5).

One consistent theme that is apparent from both historical and contemporary studies is that changes in nutrition at specific stages of pregnancy can have very different outcomes(6). This finding is not unexpected as different organs have critical and precise developmental windows that may be compromised, or enhanced, and thereafter permanently set for the rest of that individual's life. Importantly, adaptations of this type appear not only to be dependent on the period in which the mother's diet is altered but also the diet to which she is rehabilitated(7). One fundamental consideration is that the self-limitation in food intake between early-to-mid gestation that occurs commonly as a result of the nausea affecting approximately 90% of pregnant women in the UK may be directly linked to Western diets(8). The extent to which this factor relates to changes in placental function and/or fetal growth remains less clear, but there is a need to match global pre- and postnatal nutritional requirements in order to avoid accelerated growth and the concomitant increased risk of later obesity and metabolic complications(9). Importantly, however, intergenerational acceleration mechanisms do not appear to make an important contribution to levels of raised childhood BMI within the population(10).

The lactational environment and postnatal development

A further area requiring consideration is the relationship between the maternal diet in late pregnancy, its impact on mammary gland development and milk production and whether the infant is breast-fed or formula-fed(11). The higher macronutrient content of formula feed compared with breast milk, in conjunction with its fixed composition throughout a feed (unlike breast milk, the composition of which changes with time), will impact on the nutrient supply to the infant. It is therefore necessary to consider not only short-term but also long-term advantages of breast-feeding in relation to appetite regulation(12). The type of lactational diet also impacts on other behavioural aspects, including sleep–awake activity cycles(13), so that extended breast-feeding may not only be beneficial in developing countries(14). Other confounding factors such as social class and smoking during pregnancy further determine postnatal diet(15).

In summary, the relationship between nutrient supply during key stages of development from the time of conception to weaning is highly complex and requires careful in-depth consideration. It is necessary to conduct detailed animal experiments in a range of species in order to elucidate the mechanisms involved, be they epigenetic or related processes(16).

Animal models of nutritional programming

The main animal models that have been utilised to date to investigate the impact of maternal diet on long-term programming have been rats and sheep(17). These animals obviously have very different developmental patterns in the relationship between placental and fetal growth, maturity at birth and milk composition(18). The advantage of utilising rats is their very short gestational length. However, the type of diet they normally consume in the wild is very different from that fed to housed laboratory animals in which semi-purified diets are the norm. Such diets provide substantially greater differences in nutrients to pregnant rats than to controls and, therefore, should be considered ‘pharmacological’ as opposed to ‘physiological’ as they are outside the normal distribution of wild-type diets. For example, high-fat diets utilised to date in rat studies contain four times as much fat when compared with control diets and are at risk of being deficient in micronutrients. Not surprisingly, when fed a diet so rich in fat, maternal food intake is reduced(19). In addition, rats exhibit coprophagia, which has a substantial effect on nutrient flux and the ability to experimentally manipulate the intake of specific nutrients.

Appreciable placental growth necessarily continues up to term in the rat to meet the much higher protein demands for fetal growth compared with that of man or sheep(20). In large mammals, however, the maximal period of placental growth is early in pregnancy in order to meet the greatly increased nutrient requirements of the fetus in late gestation when its growth is exponential(21). Furthermore, rats produce large litters whereas sheep and man only normally produce one (or two) offspring of comparable birth weight per pregnancy.

Comparative differences in organ development

Another substantive difference between rats and man (and rats and sheep) is that the development and maturation of the hypothalamic–pituitary system occurs after birth during the neonatal period in rats compared with during late gestation in sheep and man. Thus, any nutritional intervention that stops at birth in the rat will have substantially different outcomes compared with those extending into the neonatal period(9). Other organs that continue to develop postnatally in the rat, compared with prenatally in sheep and man, include the kidney and brown adipose tissue(6) (Fig. 1). For example, in those species born with a mature hypothalamic–pituitary axis after a long gestation, such as in sheep and man, the majority of the appetite-controlling networks appear to be in place by mid- to late gestation(22). This position is clearly in contrast to that in small mammals in which there is substantial development during the lactational period as the hypothalamic axis matures. It is thus not unexpected that leptin administration has such a marked effect on this process(23) and related processes(24, 25), which are not seen when given to the late-gestation fetus or neonatal sheep(26, 27). Indeed, there is no postnatal leptin surge in either sheep(28) or man(29).

Fig. 1. Summary of the ontogenic differences in development and maturation of the hypothalamus and kidney in relation to pre- and postnatal growth trajectories between large (including man) and small animals.

Nephron development occurs primarily after birth, as does brown fat, which in sheep and man was originally thought to be completely lost in the postnatal period(30), but is retained throughout the life cycle in the rodent. As such, brown adipose tissue has a fundamental impact on the energy balance of adult rats because it uniquely possesses uncoupling protein 1 and is thus able to rapidly generate large amounts of heat(31). Indeed, brown fat has a pivotal role in the regulation of thermogenesis, with a thermogenic potential of 300 W/g whilst all other tissues generate approximately 1 W/g(32). Not unexpectedly, loss of uncoupling protein 1 is accompanied by an increase in fat mass, making it difficult to dissociate the dietary and related effects on changes in fat mass from adaptations in uncoupling protein 1 in this species(33). Recently, indirect measurements in patients with cancer suggest that brown fat may be retained in small amounts in these individuals(34). Its location is, however, primarily in the neck region(35, 36) and whether it has any functional importance in energy balance remains to be established.

Methodological considerations in the interpretation of metabolic programming

There have been two major problems with rat studies utilised in the assessment of long-term cardiovascular outcomes. First, in many studies blood pressure has only been measured using the tail-cuff technique in restrained and heated animals during the day when they are normally inactive, being nocturnal animals (for example, see Langley-Evans(37); Kwong et al.(38)). It is recognised that this approach can give very different results when compared with blood pressure measurements obtained using telemetry(39); as such, it is only recommended for use in obviously hypertensive animals in which the technique was originally validated(40). Thus, in the majority of studies in which it has been used in this area the tail-cuff technique is not informative in relation to the modest differences in blood pressure recorded in such normotensive rats. This factor may explain why in more recent studies the offspring born to dams fed a low-protein high-carbohydrate diet through pregnancy show either no difference or a reduction in blood pressure when measured using either telemetry or an indwelling arterial catheter(41, 42). Comparable findings are seen in offspring born to dams in which food intake is reduced by 50% through pregnancy compared with controls(43). A further confounding factor in many of these early studies has been a failure to consider the dam as the experimental unit rather than her offspring(38), which is now recommended as standard practice(44). There also appears to be a marked divergence in the long-term outcomes between genders in rats that is primarily linked to the faster, as well as continued, growth of males compared with females(45).

Despite the fact that sheep are ruminants, they have proved invaluable in enabling the understanding of the nutritional and endocrine regulation of placento–fetal development. As in man, the primary metabolic substrate for fetal metabolism is glucose, for which GLUT 1 is the main placental regulator(46). Glucose is thus transported across the placenta by active diffusion determined by its concentration in maternal blood(47). In addition, not only does kidney development show a very similar ontogeny in sheep and man, but the distribution of total nephrons across the adult population is comparable(48). It is also feasible to obtain very consistent blood pressure recordings in the offspring using arterial cannulation whilst the animal is standing freely with continual access to its diet(49). At the same time, there is no discernable difference in blood pressure control or glucose regulation between genders when measured in intact adult sheep(48, 50).

Mismatched offspring and later outcomes?

Another confounding factor that has become apparent in a number of recent studies using sheep is a mismatch between the numbers of singletons and twins between nutritional groups. This mismatch has introduced erroneous conclusions, which is not unexpected given the pronounced differences in outcomes of human newborns that are also dependent on fetal number(51). Appropriate statistical correction (as in the case of rat studies) for this factor is therefore necessary(44). Some of the more prominent examples of such mismatch between control and intervention groups are summarised in Table 1. Indeed, this mismatch is further confounded by using both twins from the same mother for whom the lactational environment will be further altered. In contrast, when one twin is reared alone by its mother, its growth rate matches that of a singleton, as does its later blood pressure control and glucose homeostasis(50, 52).

Table 1. Summary of studies into nutritional programming in which outcome measures are potentially confounded by a mismatch between groups in their composition of offspring from singleton and twin pregnancies

It is now clear that future studies in the area will need to take into account not only changes in diet through pregnancy but also the relative contributions of gender, litter size, type and duration of lactation as well as weaning diets. These considerations will undoubtedly increase the time necessary to conduct such studies as well as their cost, but it is imperative that with the plethora of nutritional models already established these factors are optimised in order to elucidate the main pathways involved rather than pursuing models that may have little translational relevance(53).

Conclusion

In summary, appropriate nutritional interventions are now available that can examine relevant short- and long-term outcomes and determine how contemporary diets impact on fat deposition, metabolic homeostasis and cardiovascular control in a consistent and validated manner. The completion of such studies may enable the determination of the optimum nutrition in terms of quantity and quality that may ultimately improve the health of future generations.

Acknowledgements

The authors declare no conflict of interest and acknowledge the support of the British Heart Foundation and the European Union Sixth Framework Programme for Research and Technical Development of the European Community – The Early Nutrition Programming Project (FOOD-CT-2005-007036) in their research.

References

1. Barker, DJP, Bull, AR, Osmond, C et al. (1990) Fetal and placental size and risk of hypertension in adult life. Br Med J 301, 259262.
2. Barker, D, Osmond, C, Forsén, T et al. (2005) Trajectories of growth among children who have coronary events as adults. N Engl J Med 353, 18021809.
3. Singhal, A (2005) Endothelial dysfunction: role in obesity-related disorders and the early origins of CVD. Proc Nutr Soc 64, 1522.
4. Keith, SW, Redden, DT, Katzmarzyk, PT et al. (2006) Putative contributors to the secular increase in obesity: exploring the roads less traveled. Int J Obesity (Lond) 30, 15851594.
5. Williams, P, Kurlak, LO, Perkins, A et al. (2007) Impaired renal function and hypertension accompany juvenile obesity: effect of prenatal diet. Kidney Int 72, 279289.
6. Symonds, ME, Stephenson, T, Gardner, DS et al. (2007) Long term effects of nutritional programming of the embryo and fetus: mechanisms and critical windows. Reprod Fertil Dev 19, 5363.
7. Reynolds, RM, Godfrey, KM, Barker, M et al. (2007) Stress responsiveness in adult life: influence of mother's diet in late pregnancy. J Clin Endocrinol Metab 92, 22082210.
8. Pepper, G & Roberts, S (2006) Rates of nausea and vomiting in pregnancy and dietary characteristics across populations. Proc R Soc Lond B Biol Sci 273, 26752679.
9. Symonds, ME (2007) Integration of physiological and molecular mechanisms of the developmental origins of adult disease: new concepts and insights. Proc Nutr Soc 66, 442450.
10. Davey Smith, G, Steer, C, Leary, S et al. (2007) Is there an intra-uterine influence on obesity? Evidence from parent-child associations in ALSPAC. Arch Dis Child 92, 876880.
11. Toschke, AM, Martin, RM, von Kries, R et al. (2007) Infant feeding method and obesity: body mass index and dual-energy X-ray absorptiometry measurements at 9–10 y of age from the Avon Longitudinal Study of Parents and Children (ALSPAC). Am J Clin Nutr 85, 15781585.
12. Sievers, E, Oldigs, HD, Santer, R et al. (2002) Feeding patterns in breast-fed and formula-fed infants. Ann Nutr Metab 46, 243248.
13. Lee, K (2000) Crying and behavior pattern in breast- and formula-fed infants. Early Hum Dev 58, 133140.
14. Heird, WC (2007) Progress in promoting breast-feeding, combating malnutrition, and composition and use of infant formula, 1981–2006. J Nutr 137, 499S502S.
15. Bogen, DL, Hanusa, BH & Whitaker, RC (2004) The effect of breast-feeding with and without formula use on the risk of obesity at 4 years of age. Obesity Res 12, 15271535.
16. Symonds, ME, Stephenson, T, Gardner, DS et al. (2009) Tissue specific adaptations to nutrient supply: more than just epigenetics? Adv Exp Med Biol 646 (In the Press).
17. McMillen, IC & Robinson, JS (2005) Developmental origins of the metabolic syndrome: Prediction, plasticity, and programming. Physiol Rev 85, 571633.
18. Prentice, AM & Prentice, A (1995) Evolutionary and environmental influences on lactation. Proc Nutr Soc 54, 391400.
19. Taylor, PD, McConnell, J, Khan, IY et al. (2005) Impaired glucose homeostasis and mitochondrial abnormalities in offspring of rats fed a fat-rich diet in pregnancy. Am J. Physiol Regul Integr Comp Physiol 288, R134R139.
20. Widdowson, EM (1950) Chemical composition of newly born animals. Nature 116, 626628.
21. Symonds, ME, Budge, H, Mostyn, A et al. (2006) Nutritional programming of foetal development: endocrine mediators and long-term outcomes for cardiovascular health. Curr Nutr Food Sci 2, 389398.
22. Mühlhäusler, BS, McMillen, IC, Rouzaud, G et al. (2004) Appetite regulatory neuropeptides are expressed in the sheep hypothalamus before birth. Neuroendocrinology 16, 502507.
23. Bouret, SG, Draper, SJ & Simerly, RB (2004) Trophic action of leptin on hypothalamic neurons that regulate feeding. Science 304, 108110.
24. Vickers, MH, Gluckman, PD, Coveny, AH et al. (2005) Neonatal leptin treatment reverses developmental programming. Endocrinology 146, 42114216.
25. Attig, L, Solomon, G, Ferezou, J et al. (2008) Early postnatal leptin blockage leads to a long-term leptin resistance and susceptibility to diet-induced obesity in rats. Int J Obesity (Lond) 32, 11531160.
26. Mostyn, A, Bispham, J, Pearce, S et al. (2002) Differential effects of leptin on thermoregulation and uncoupling protein abundance in the neonatal lamb. FASEB J 16, 14381440.
27. Yuen, BS, Owens, PC, Mühlhäusler, BS et al. (2003) Leptin alters the structural and functional characteristics of adipose tissue before birth. FASEB J 17, 11021104.
28. Bispham, J, Budge, H, Mostyn, A et al. (2002) Ambient temperature, maternal dexamethasone, and postnatal ontogeny of leptin in the neonatal lamb. Pediatr Res 52, 8590.
29. Matsuda, J, Yokota, I, Iida, M et al. (1999) Dynamic changes in serum leptin concentrations during the fetal and neonatal periods. Pediatr Res 45, 7175.
30. Lean, MEJ (1989) Brown adipose tissue in humans. Proc Nutr Soc 48, 243256.
31. Cannon, B & Nedergaard, J (2004) Brown adipose tissue: Function and significance. Physiol Rev 84, 277359.
32. Power, G Biology of temperature: the mammalian fetus (1989) J Dev Physiol 12, 295304.
33. Mostyn, A, Keisler, DH, Webb, R et al. (2001) The role of leptin in the transition from fetus to neonate. Proc Nutr Soc 60, 187194.
34. Nedergaard, J, Bengtsson, T & Cannon, B (2007) Unexpected evidence for active brown adipose tissue in adult humans. Am J Physiol Endocrinol Metab 293, E444E452.
35. Christensen, CR, Clark, PB & Morton, KA (2006) Reversal of hypermetabolic brown adipose tissue in F-18 FDG PET imaging. Clin Nucl Med 31, 193196.
36. Goetze, S, Lavely, WC, Ziessman, HA et al. (2008) Visualization of brown adipose tissue with 99mTc-methoxyisobutylisonitrile on SPECT/CT. J Nucl Med 49, 752756.
37. Langley-Evans, SC (1997) Hypertension induced by foetal exposure to a maternal low-protein diet, in the rat, is prevented by pharmacological blockade of maternal glucocorticoid synthesis. J Hypertens 15, 537544.
38. Kwong, WY, Wild, AE, Roberts, P et al. (2000) Maternal undernutrition during the preimplantation period of rat development causes blastocyst abnormalities and programming of postnatal hypertension. Development 127, 41954202.
39. D'Angelo, G, Elmarakby, AA, Pollock, DM et al. (2005) Fructose feeding increases insulin resistance but not blood pressure in Sprague-Dawley rats. Hypertension 46, 806811.
40. Bunag, RD (1973) Validation in awake rats of a tail-cuff method for measuring systolic pressure. J Appl Physiol 34, 279282.
41. Fernandez-Twinn, DS, Ekizoglou, S, Wayman, A et al. (2006) Maternal low-protein diet programs cardiac beta-adrenergic response and signaling in 3-mo-old male offspring. Am J Physiol Regul Integr Comp Physiol 291, R429R436.
42. Hoppe, CC, Evans, RG, Moritz, KM et al. (2007) Combined prenatal and postnatal protein restriction influences adult kidney structure, function, and arterial pressure. Am J Physiol Regul Integr Comp Physiol 292, R462R469.
43. Brennan, KA, Kaufman, S, Reynolds, SW et al. (2008) Differential effects of maternal nutrient restriction through pregnancy on kidney development and later blood pressure control in the resulting offspring. Am J Physiol Regul Integr Comp Physiol 295, R197R205.
44. Festing, MFW (2006) Design and statistical methods in studies using animal models of development. ILAR J 47, 514.
45. Symonds, ME & Gardner, DS (2006) Experimental evidence for early nutritional programming of adult health in animals. Curr Opin Clin Nutr Metab Care 9, 278283.
46. Dandrea, J, Wilson, V, Gopalakrishnan, G et al. (2001) Maternal nutritional manipulation of placental growth and glucose transporter-1 abundance in sheep. Reproduction 122, 793800.
47. Edwards, LJ, Symonds, ME, Warnes, K et al. (2001) Responses of the fetal pituitary-adrenal axis to acute and chronic hypoglycaemia during late gestation in the sheep. Endocrinology 142, 17781785.
48. Symonds, ME, Budge, H, Mostyn, A et al. (2007) Maternal diet through pregnancy – the key to future good health of the next generation? In Nutrition Research Advances, pp. 223240 [Watkins, SVeditor]. New York: Nova Science Publishers, Inc.
49. Gardner, DS, Pearce, S, Dandrea, J et al. (2004) Peri-implantation undernutrition programs blunted angiotensin II evoked baroreflex responses in young adult sheep. Hypertension 43, 17.
50. Gardner, DS, Tingey, K, van Bon, BWM et al. (2005) Programming of glucose-insulin metabolism in adult sheep after maternal undernutrition. Am J Physiol Regul Integr Comp Physiol 289, R947R954.
51. Hayes, EJ, Paul, D, Ness, A et al. (2007) Very-low-birthweight neonates: do outcomes differ in multiple compared with singleton gestations? Am J Perinatol 24, 373376.
52. Gopalakrishnan, G, Gardner, DS, Dandrea, J et al. (2005) Influence of maternal pre-pregnancy body composition and diet during early-mid pregnancy on cardiovascular function and nephron number in juvenile sheep. Br J Nutr 94, 938947.
53. Symonds, ME, Stephenson, T & Budge, H (2009) Early determinants of cardiovascular disease – The role of early diet in later blood pressure control. Am J Clin Nutr (In the Press).
54. Gilbert, JS, Lang, AL, Grant, AR et al. (2005) Maternal nutrient restriction in sheep: hypertension, decreased nephron number in offspring at 9 months of age. J Physiol 565, 137148.
55. Ford, SP, Hess, BW, Schwope, MM et al. (2007) Maternal undernutrition during early to mid-gestation in the ewe results in altered growth, adiposity, and glucose tolerance in male offspring. J Anim Sci 85, 12851294.
56. Mühlhäusler, BS, Adam, CL, Findlay, PA et al. (2006) Increased maternal nutrition alters development of the appetite-regulating network in the brain. FASEB J 20, 12571259.
57. Cleal, JK, Poore, KR, Boullin, JP et al. (2007) Mismatched pre- and postnatal nutrition leads to cardiovascular dysfunction and altered renal function in adulthood. Proc Natl Acad Sci USA 104, 95299533.