Epigenetics represents the key to unlocking the door of ‘personalized‘ medicine, and we are excited about the potential. The concept of personalized medicine is particularly important to the field of pediatrics and the ‘developmental origins of disease‘ hypothesis. Multiple environmental exposures within the perinatal period predispose infants towards adult diseases such as diabetes, obesity, hypertension and cardiovascular disease. These exposures span the continuum of the human experience in both developing and developed countries. Relevant exposures include toxins, maternal smoking, placental dysfunction and malnutrition of either the mother or infant.
As a pediatrician, a struggle exists because exposure does not guarantee development of adult disease. The urgency of this struggle lies within the biological truism that all interventions contain inherent risks themselves; the more extreme the intervention, the higher the price. As biomedical and pediatric communities increase their understanding of the ‘meaning‘ of perinatal epigenetic responses to environmental exposures, we hope to identify those individuals most predisposed towards adult diseases. This predisposition makes subsequent interventions probabilistically worthwhile.
The field is now focused on stratifying infants at risk and developing targeted interventions as a result. An important concept is that under normal circumstances, the epigenetic profile (or pattern of epigenetic modifications) of genes is dynamic and evolves with development. Put another way, the epigenetic profile present at a given time point, for example birth, forms the platform upon which future epigenetic changes take place. When this platform is altered by a perinatal environmental exposure, the new profile deflects the expected developmentally programmed changes away from the norm. Environmentally induced epigenetic modifications to DNA subsequently represent a personalized history of environmental exposures. The fidelity of these environmentally induced epigenetic changes, when considered in the context of the developmental origins of disease, present unique opportunities as we pursue the vision of personalized medicine.
The field of environmental epigenetics and developmental origins of disease contains a plethora of data showing associations between environmental exposures during development and changes in DNA methylation. Human studies generally involve examining nucleated blood cells for changes in DNA methylation because of the accessibility of samples and the stability of DNA methylation following collection. Studies can generally by classified into those looking at global methylation changes and those looking at gene-specific methylation changes. Global methylation can be nonspecifically assessed by looking at overall DNA methylation of the genome, or by looking at specific CpG methylation within repeat elements. Perhaps the most sensitive quantification of DNA methylation in developmental origins is that of specific CpG sites associated with genes known to have a functional relationship to disease.
In an elegant study published in the Journal of Pediatrics earlier this year, Perkins et al. quantified methylation of IGF-2 in cord-blood-leukocyte DNA from 204 infants Citation[1]. Levels of cord blood methylation were correlated with anthropometric and feeding data for those infants at 1 year of age. Infants were classified into weight for age (WFA) ≤85th percentile, or ≥85th percentile. When all infants were considered, the study demonstrated that the average methylation fraction at one of the insulin growth factor 2 differentially methylated regions, the H19 locus, was higher in the infants that had WFA ≥85th percentile compared with those WFA ≤85th percentile (p = 0.05). However, perhaps one of the most exciting outcomes from this study was that when infants were stratified based upon having ever breastfed, H19 methylation increases were only observed in the infants that were never breastfed (p = 0.01), while infants that were breastfed did not have a significant increase in H19 methylation (p = 0.82) Citation[1]. While somewhat difficult to interpret, this finding raises the potential that a postnatal behavior may interact with the prenatal methylation status with the potential to influence outcome.
Changes in DNA methylation are also associated with maternal tobacco smoke and toxin exposure. Methylation changes at the IGF-2 locus as well as within repetitive elements (LINE-1 and Alu) have been observed in cord blood of infants whose mothers smoked during pregnancy, in some cases with the most pronounced effect in males infants Citation[2,3]. Prenatal exposure to polycyclic aromatic hydrocarbons, carcinogenic environmental pollutants commonly found in tobacco smoke, air pollution and charbroiled foods, is associated with decreased global and gene-specific DNA methylation in cord blood Citation[4,5]. Prenatal arsenic exposure is associated with changes in global DNA and LINE-1 methylation Furthermore, these associations are sex dependent Citation[6,7].
The above studies are key first steps, but an important caveat is that the above relationships between early exposures and DNA methylation changes are purely associative. Casual relationships have not been established. Furthermore, many studies fail to take into account potential confounders such as gender, cell type and paternal or transgenerational contributions. This limitation decreases the sensitivity of such studies from detecting further associations. Similarly, practicality limits most studies from taking into account other environmental factors and molecular–biochemical interactions that potentially confound associations. The reality of funding and funding agencies turns the vast majority of scientists into ‘naive reductionists‘, which unfortunately leaves little opportunity through which to anticipate complex interactions, which are intrinsic to mammalian biology. Furthermore, a temptation exists to interpret changes in DNA methylation as a ‘yes‘ or ‘no‘ answer, with increased DNA methylation presumed to be ‘no‘ (transcription). However, DNA methylation does not always mean ‘no‘, and the answer often depends upon where within a gene‘s organization DNA methylation changes occur. As a result, these limitations and confounders resultantly limit our ability to dogmatically interpret results, and we must be appropriately thoughtful in our conclusions. Overstating our results will only lead to unmet expectations of the epigenetics field by the greater scientific community and society.
As we begin to unravel the cause-and-effect relationship between environmental epigenetics and the developmental origins of disease, two things become apparent moving forward. The first is that epigenetic changes may serve purely as a biomarker, for example, as an ‘echo‘ of an early-life event that predicts risk for an adult disease, but has nothing to do with the etiology of the disease. Second, epigenetic changes may allow us to target molecular responses to specific environmental perinatal events, for example, as a ‘fingerprint‘ that identifies specific pioneering mechanisms that led to relevant adult diseases. In this case, the prenatal exposure produced an epigenetic change that causes subsequent disease. Both aspects potentially improve clinical care, but only if we learn to focus on understanding what epigenetics ‘means‘ and ‘how it happens‘.
The clinical utility of an epigenetic biomarker in the era of personalized medicine is clear. It is important that a biomarker be specific as a predictor and be available at a time point when relevant interventions exist. In addition, a biomarker needs to be sensitive enough to be detectable, consistent/robust and noninvasive. In the case of an epigenetic biomarker in human samples, the noninvasive nature of nucleated blood cells, either sampled directly from the individual or from cord blood, is an attractive option. When samples of convenience are to be used, it will be essential to perform validation with epidemiologic and mechanistic studies Citation[8]. Ideally, we will also learn to integrate our epigenetic biomarkers with other potential personal predictors of disease such as phenotype, metabolomics and genome-wide sequences.
Unfortunately, we have failed as a translational community to integrate an understanding of the histone code from nucleated blood cells as a personalized biomarker of adult disease. This is particularly true within the context of a specific site or gene. Considering that each cell has approximately 4 × 1030 possible histone modification permutations, the amount of potential environmental exposure data stored in a ‘personalized‘ histone code appears to be extensive, and it has yet to be unlocked. However, we continue to struggle as a community in interpreting the histone code. Accurate interpretation likely requires further understanding of how different histone covalent modifications interact; acknowledgment that location within a gene‘s organization affects the impact of a specific histone covalent modification; and recognition of direct and indirect interactions from sites quite distal to the site or gene(s) of interest.
Alternatively, if an epigenetic change is the footprint of a pioneering event, an understanding of how that change occurs will lead to the development of potential interventions. At present, in vivo epigenetic studies in animal models provide the best means of establishing causation and the hierarchy of events. In vivo animal models also provide the opportunity to examine the global effects of potential interventions, including dietary manipulations Citation[9,10]. However, as the goal is to apply these findings to interventions, epigenetic manipulations must be considered with caution because the potential for off-target negative effects is high.
Our growing understanding of the fidelity of environmental epigenetic responses represents a new and exciting opportunity in terms of personalized medicine. The complexity intrinsic to this fidelity provides a depth and breadth of information that may eventually allow environmental epigenetic databases to reveal meaningful information about individual health and subsequently guide a more personalized approach to medicine.
Financial & competing interests disclosure
LA Joss-Moore is funded by the NIH (grant number: 5K01DK084036-03), and RH Lane is supported by the NIH (grant number: 1R01HL110002-01). The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.
No writing assistance was utilized in the production of this manuscript.
Additional information
Funding
References
- Perkins E , MurphySK, MurthaAPet al. Insulin-like growth factor 2/H19 methylation at birth and risk of overweight and obesity in children. J. Pediatr. 161(1) , 31–39 (2012).
- Murphy SK , AdigunA, HuangZet al. Gender-specific methylation differences in relation to prenatal exposure to cigarette smoke. Gene 494(1) , 36–43 (2012).
- Breton CV , ByunHM, WentenM, PanF, YangA, GillilandFD. Prenatal tobacco smoke exposure affects global and gene-specific DNA methylation. Am. J. Respir. Crit. Care Med.180(5) , 462–467 (2009).
- Herbstman JB , TangD, ZhuDet al. Prenatal exposure to polycyclic aromatic hydrocarbons, benzo[a]pyrene-DNA adducts, and genomic DNA methylation in cord blood. Environ. Health Perspect. 120(5) , 733–738 (2012).
- Tang WY , LevinL, TalaskaGet al. Maternal exposure to polycyclic aromatic hydrocarbons and 5´-CpG methylation of interferon-gamma in cord white blood cells. Environ. Health Perspect. 120(8) , 1195–1200 (2012).
- Kile ML , BaccarelliA, HoffmanEet al. Prenatal arsenic exposure and DNA methylation in maternal and umbilical cord blood leukocytes. Environ. Health Perspect. 120(7) , 1061–1066 (2012).
- Pilsner JR , HallMN, LiuXet al. Influence of prenatal arsenic exposure and newborn sex on global methylation of cord blood DNA. PLoS ONE 7(5) , e37147 (2012).
- Talens RP , BoomsmaDI, TobiEWet al. Variation, patterns, and temporal stability of DNA methylation: considerations for epigenetic epidemiology. FASEB J. 24(9) , 3135–3144 (2010).
- Joss-Moore LA , WangY, BaackMLet al. IUGR decreases PPARgamma and SETD8 expression in neonatal rat lung and these effects are ameliorated by maternal DHA supplementation. Early Hum. Dev. 86(12) , 785–791 (2010).
- Burdge GC , LillycropKA, PhillipsES, Slater-JefferiesJL, JacksonAA, HansonMA. Folic acid supplementation during the juvenile-pubertal period in rats modifies the phenotype and epigenotype induced by prenatal nutrition. J. Nutr.139(6) , 1054–1060 (2009).