Developmental genes targeted for epigenetic variation between twin-twin transfusion syndrome children

Epidemiologic studies have clearly linked infant growth to antenatal environmental factors, including diet, xenobiotic exposures, stress and lifestyle factors, as a significant risk factor for long-term chronic disease, particularly cardiovascular disease and metabolic syndromes [1–3]. These studies, and a growing literature on the role of development on lifelong health, would suggest that a significant proportion of disease risk can be linked to the intrauterine environment.

On a molecular level, epigenetic mechanisms have been invoked to explain mechanistically how experiences during a narrow but critical and susceptible period of time can influence long-term processes leading to health and disease. DNA methylation, a key epigenetic mechanism, is a clear focus of studies on the developmental origins of health and disease. It represents a stable modification of DNA that can be propagated during cell division, yet is susceptible to environmental influence. This is particularly true during development when the cell-specific patterns of methylation that define cellular differentiation are set [4–6]. Studies of specific candidate genes as well as genome-wide examinations of DNA methylation have demonstrated clear relationships to infant birthweight, supporting the role of DNA methylation in mediating these risks [7, 8]. This is particularly relevant to correlations in genes involved in metabolism, growth and cardiovascular disease [9].

Studies of twins represent powerful approaches to understanding the importance of shared or disparate environment as well as the contribution of genetics on phenotype [10]. These studies also represent a powerful opportunity to examine the impact of the environment on epigenetic phenomena and, in turn, the contribution of epigenetics on various health outcomes [11, 12]. Twin studies are powerful in interrogating both the intrauterine and the postnatal environment, as generally monozygotic, and to a slightly lesser degree, dizygotic twins share the same intrauterine environment. Nonetheless, to date, most studies of twin epigenetics have focused on older and adult cohorts, ignoring the opportunity to consider the role of the intrauterine environment [13–17].

Twin-twin transfusion syndrome (TTTS) patients offer a unique opportunity to investigate how an adverse intrauterine environment compares to a rich environment, on DNA methylation and, potentially, on later health outcomes while controlling for genetic contribution. Approximately 70% of monozygotic twins are monochorionic and diamniotic (DiMo), following embryo division before the post-implantation blastocyst phase. These fetuses share a placenta, and virtually all develop placental vascular anastomoses, which allow blood to flow between the fetuses [18]. In cases of TTTS, which represents approximately 15% of DiMo cases, the blood supply sharing becomes unequal, due to as yet incompletely understood mechanisms that may include the number and/or type of anastomoses[19]. If severe and progressive, TTTS leads to almost 90% perinatal mortality [20]. The ‘donor’ twin becomes chronically hypovolemic and anuric, which leads to severe oligohydramnios. This twin is typically growth restricted as well – a consequence of the progressive cardiac failure, but also of the smaller placental share often associated with the donor. The ‘recipient’ twin becomes polyuric, resulting in severe polyhydramnios. This state can lead to cardiac failure, hydrops and fetal death as well, as chronic hypervolemia leads to (right) ventricular dilation, tricuspid regurgitation and global cardiac dyskinesia [21]. Thus, this syndrome results in a number of distinct pathophysiologic changes in each of the genetically identical twins. Although a number of treatment approaches have been proposed and utilized with varying success to treat this condition, in recent years selective fetoscopic laser photocoagulation of placental vessels has become more widely available and is now considered the only effective treatment in severe, progressive forms of the disease [22–24]. This treatment consists of obliterating all intertwin vascular connections at the placental level, effectively halting the twin-to-twin transfusion and restoring the balance of the twins’ blood supplies. While survival of at least one twin is seen in 80% of cases [25], not all fetuses fully recover. In some, the renal, cardiovascular or neurological effects of the syndrome are already irreversible [26, 27]; in addition, unequal placental share seen in some twin pregnancies (a problem that cannot be corrected by fetal surgery) causes persistent placental insufficiency and growth restriction, usually of the donor [28]. In part because of the inherent risks of twin gestations, the chronic morbidity of TTTS and the relative invasiveness of fetal intervention, TTTS pregnancies and infants may also suffer from miscarriage, preterm birth, neonatal intensive care unit admissions, respiratory distress syndrome and intraventricular hemorrhage [29]. Long term morbidity in survivors of TTTS can include renal, cardiovascular and neurodevelopmental deficits [30, 31], although a number of studies have suggested that morbidity rates in TTTS cases treated with fetoscopic laser photocoagulation are no different than those observed in similarly preterm populations [32–34]. More subtle long term effects, such as those on metabolic disease and neurobehavioral outcomes, have been less comprehensively evaluated. These effects could be of great interest as a model in defining underlying mechanisms of the developmental origins of these outcomes, including the epigenetic mechanisms potentially responsible. Moreover, the immediately stabilizing effect of fetoscopic laser photocoagulation on the uteroplacental environment may allow us to pinpoint critical time points during gestation when these epigenetic phenomena are critical. Finally, the co-occurrence of small placental share in some donor twins may help us differentiate between adverse uteroplacental factors that are correctible (TTTS) and those that are not (placental insufficiency).

In this study, we examined how DNA methylation profiles in cases of TTTS are impacted by donor-recipient status, a model of extreme adverse and rich intrauterine conditions. This condition occurred during a critical window of development during which DNA methylation patterns are being set. Our aim is to define broadly where variation in DNA methylation is occurring in this situation, as it may aid in defining not only key pathways potentially affected by the intrauterine growth disturbances that result from TTTS but also point to potential long-term consequences worthy of further study.

Using a unique population of TTTS survivors, we have identified subtle but potentially important variability in DNA methylation influenced by the physiologic environment during development. Epigenetic mechanisms, such as DNA methylation, are clearly important in mediating the developmental origins of lifelong health and disease. By taking advantage of this unique and valuable cohort, we have been able to demonstrate that regardless of tissue type, developmentally critical genes controlled by polycomb group transcription factors and modifiers are being targeted for altered DNA methylation. These findings provide additional evidence that through epigenetic alterations the intrauterine environment can affect cellular function beyond the developmental period. Additionally, our results may provide important insights into the mechanisms leading to the potential morbidities in survivors of TTTS, which are growing in number due to advances in successful surgical interventions of this previously mortal condition [25].

We identified subtle differences in global markers of DNA methylation, specifically the finding of increased methylation of the LINE1 element among recipients. These results are in line with recent findings that nutritional factors can influence global markers of DNA methylation. For example, fortification of the diet with folate has been correlated with increased peripheral blood LINE1 methylation in adults [40] although in a folate-replete population, little difference in LINE1 methylation was observed in infant cord blood based upon maternal intake of methyl donors [41]. LINE1 hypomethylation in cord blood has been associated with extremes in birth weight (both high and low) and with prematurity [42]. In adults, alterations in LINE1 methylation have been associated with various health outcomes including cancers and cardiovascular disease [43] but the mechanism through which non-target tissue alterations of methylation of these global markers leads to disease is not understood [43]. To date, there have been little data regarding any relationships with this marker in childhood or adolescent populations, making this study unique.

Overall, our examination revealed low levels of variability of DNA methylation between donors and recipients at the locus-specific, genome-wide level. However, when individual twin pairs were examined, these differences were clearer. The saliva samples showed greater variability between donors and recipients. Variability did not appear to occur consistently across all CpG sites and was lowest at the loci with beta values at the extremes (0 and 1) and at the midpoint (0.5). This would represent those loci with complete or nearly complete hypo- and hypermethylation. Loci with intermediate methylation potentially represent imprinted regions. Interestingly, in blood, comparing donor and recipient methylation in a five-month-old infant, there are a number of variable loci at the lowest levels of methylation. Importantly, since blood as well as saliva [44] represents a mixture of cell types, this may represent changes in the proportion of specific blood cell types characterized by methylation (or lack of methylation) at specific loci, such as the FOXP3 locus, which characterizes T-regulatory cells [45]. In fact, recent studies have shown that much of the variability identified in peripheral blood DNA methylation profiles represents variation in the underlying proportion of blood cell types [38, 39]. We examined if the loci with the greatest interclass correlation by donor-recipient status would be considered differentially methylated across blood cell subtypes, and saw little evidence. Yet, the data on blood cell differentially methylated regions were based on adult signatures and may not be completely comparable to the data examined here on infants and young children, thus making more specific examinations worthy of further explorations.

Our observation that there are a greater number of severely altered loci among twin pairs with later fetoscopic surgery is intriguing, although preliminary. From these findings, we could speculate that differences in methylation are more common, the longer the uneven placental environment exists, as infants undergoing surgery at the latest time points (18 to 20 weeks) had nearly twice as many loci with highly divergent methylation in their saliva samples as those with surgery at 15 weeks. Even with surgery at 15 weeks, there are a substantial number of divergent loci, suggesting that some of these effects may be occurring early in development, even before laser treatment or TTTS diagnosis, is possible. Coincidentally, the infants with early surgery were also the youngest pair, while those with later surgery were the oldest. It is, therefore, possible that these effects are derived postnatally. Yet, considering that these twins are being raised in similar environments postnatally, there remains strong reason to believe that early correction of the discordant uteroplacental environment may limit epigenetic divergence.

We observed the greatest divergence amongst loci considered polycomb group targets. Polycomb group (PcG) genes are a family of developmentally important genes, which play a role in chromatin post-translational modification and remodeling and which are responsible for silencing key developmentally regulated genes [46, 47]. The polycomb group family members are responsible for targeting developmentally regulated genes such as the HOX family and those involved in cellular differentiation, and evidence is accumulating that DNA hypermethylation is observed in many cancers at the sites of polycomb-mediated gene repression in embryonic cells [48]. Our finding that the genes exhibiting the greatest variation related to the donor recipient status of the child are over-represented by PcG targets highlights the potential importance of this variability. The role that such variability may play in downstream disease risk and morbidity deserves further study.

The strengths of this study include the unique study population and the opportunity to examine twin pairs in the context of TTTS. We employed state-of-the-art genome-wide methodologies for the assessment of DNA methylation. Further, we utilized appropriate and powerful statistical methodologies to quality control and analyze the resulting data. We recognize the limitation of sample size in our analyses, and have carefully constructed our analyses and our interpretations with this limitation in mind. Particularly, this is why we have not sought to identify genome-wide significant loci or genes with methylation associated with donor-recipient status or other features of the children involved. Instead, we devised strategies to broadly describe key features of those genes targeted for disruption by this condition. We also recognize that there is some potential confounding by genome in those samples from individuals whose twin did not survive to be included in this analysis. In addition, due to the small sample size, we could not fully address the modification of these effects by age or gender. Future studies are needed to identify more precisely those genes targeted for alteration, to understand better how various factors impact these alterations, and to link those alterations with downstream outcomes in the survivors of TTTS.

Overall, this study provides an intriguing, albeit preliminary, description of the epigenomic landscape of childhood survivors of TTTS. Studies of this condition and its consequences present a unique opportunity to consider more carefully mechanisms involved in the developmental origins of health and disease. Future evaluation of these cohorts holds great promise to identify the molecular basis of downstream health consequences faced by children following successful intervention for TTTS.