The significance of the placental genome and methylome in fetal and maternal health

Abstract

The placenta is a crucial organ for supporting a healthy pregnancy, and defective development or function of the placenta is implicated in a number of complications of pregnancy that affect both maternal and fetal health, including maternal preeclampsia, fetal growth restriction, and spontaneous pre-term birth. In this review, we highlight the role of the placental genome in mediating fetal and maternal health by discussing the impact of a variety of genetic alterations, from large whole-chromosome aneuploidies to single-nucleotide variants, on placental development and function. We also discuss the placental methylome in relation to its potential applications for refining diagnosis, predicting pathology, and identifying genetic variants with potential functional significance. We conclude that understanding the influence of the placental genome on common placental-mediated pathologies is critical to improving perinatal health outcomes.

Introduction

The placenta is the key mediator of maternal and fetal health during pregnancy. It is a highly adaptive organ that responds to maternal and fetal signals, shows remarkable variability in size and shape, and can tolerate some degree of localized pathology while still able to support fetal growth to term. Nevertheless, abnormal development or function of the placenta underlies many complications of pregnancy including miscarriage, maternal preeclampsia (PE), fetal growth restriction (FGR), preterm birth (PTB), and fetal malformation (Burton and Jauniaux 2018; Morgan 2016). Chromosome imbalance, genetic variation, and epigenetic changes have been implicated and/or associated with these conditions, although our knowledge of normal and abnormal genomic variation in the placenta is still maturing.

The placental genome is normally identical to that of the fetus, as both are developmentally derived from the conceptus. However this can differ due to mosaicism (or more rarely, chimerism) discussed further below. The placenta also interfaces directly with the maternal decidua, which derives from a thickening and modification of the uterine wall during pregnancy and is shed along with the placenta at parturition. Within the decidua, there is an intermingling of placental-derived cells and maternal cells, as a subset of placental trophoblast cells migrate into the maternal tissue both to anchor the placenta into the uterus and to remodel the maternal uterine spiral arteries—opening up blood flow to the placenta (Figure 1). The maternal blood comes into direct contact with the syncytiotrophoblast, a multinucleated syncytium that forms the outer layer of the placental chorionic villi. This syncytium not only provides a protective sheath, but can selectively uptake nutrients and oxygen from maternal blood and transport these into the fetal circulation.

Figure 1.

The placenta’s role in mediating fetal and maternal health. Schematic representation of a placental villous tree, the functional unit of the placenta, and its interaction with the maternal uterus. 1) Interactions between placental extravillous trophoblast cells (EVT) and maternal uterine natural killer cells (uNK) in the maternal decidua promote immune tolerance of the mother to the developing fetus, mediate EVT invasion into the maternal uterus, and aid in resource allocation to the fetus. 2) Deficiencies in remodeling of the uterine spiral arteries by EVTs are associated with health complications of maternal preeclampsia and fetal growth restriction. 3) Placental sFLT1 in maternal blood binds with placental growth factor and vascular endothelial growth factor (VEGF). Altered levels of secretion of sFLT1 by the syncitiotrophoblast (STB) has been implicated in defective endothelial dysfunction and alterations in angiogenesis associated with PE and FGR. 4) The STB is in direct contact with maternal blood in the intervillous space, thus debris shed from this trophoblast cell layer composes the majority of “fetal” material in maternal blood, including cell-free placental DNA used in non-invasive prenatal testing. 5) Defects in differentiation of cytotrophoblast (CTB) cells to form the STB is seen in placentas associated with trisomy 21. 6) Microbial infection results in altered number of placental macrophages, Hofbauer cells (HC), and triggers an innate immune response in the trophoblast cells. uNK, uterine natural killer cell; EVT, extravillous trophoblast; STB, syncitiotrophoblast; CTB, cytotrophoblast; cfp-DNA cell-free placental DNA; HC, Hofbauer cell.

In contras, to the genome, the epigenome undergoes substantia, remodeling throughout development (Hanna et al. 2018), such that the placental and embryonic epigenomes are very distinct. Much of our knowledge of epigenomic reprogramming in human development comes from studies of DNAme, which reaches its lowest level in the blastocyst around the time of implantation. However, the early passive loss of DNAme on the maternal genome is less extensive in human as compared to mouse embryos, which may explain the large number of imprinted (parent-of-origin specific) gene expression and DNAme in the human placenta (Hamada et al. 2016; Hanna et al. 2018). After implantation, de novo DNAme continues as cells differentiate and, while the placenta retains a lower level of DNAme than the rest of the embryo, it also undergoes continued changes in DNAme throughout gestation (Novakovic et al. 2011). Factors such as embryo culture, maternal nutrition, and chemical exposures, may impact placental and fetal health in part via effects on both imprinted and non-imprinted epigenetic marks. However, the evidence for this in humans is still under development and discussed below.

Early events in placentation set the stage for fetal growth and pregnancy health

During the initial stages of implantation, the trophoblast cells produce substances that allow for adhesion and invasion into the uterine wall, alter maternal immune cell phenotype to prevent embryo rejection, and prevent menstruation. Defects in these processes due to chromosomal abnormality in the conceptus are the most common causes of early pregnancy loss, with 70% of early miscarriages having an abnormal karyotype, of which ~80% are whole chromosome aneuploidies (Soler et al. 2017). In a recent study in mouse, it was demonstrated that the vast majority of gene knockouts showing an embryonic lethal effect caused defects in placentation, suggesting that an abnormal placenta is often an unappreciated cause of early lethality (Perez-Garcia et al. 2018).

Trophoblast invasion and maternal spiral artery remodeling is also dependent on maintaining a delicate balance between maternal and placental immune signaling, which can be affected by both maternal and placental/fetal genetic make-up (Moffett et al. 2015). Insufficiency of this remodeling process underlies most cases of severe PE and/or FGR. Furthermore, the placenta also acts to some degree as a barrier that can protect the developing fetus from pathogens and other adverse environmental exposures. Inflammation of the placenta due to maternal anti-fetal rejection or infection can lead to a breakdown of this barrier and increase risk for infection which can lead to inflammation of fetal membranes commonly linked to PTB (Kim et al. 2015). Both PTB and presence of placental inflammatory lesions have significant recurrence risks and this is at least partly genetically influenced, though environmental factors such as smoking, maternal obesity, and maternal infection can also be important.

Placental genetics in maternal and fetal medicine

Large-scale chromosome abnormalities in the placenta have significant impact on fetal and maternal health outcomes

Spontaneously-arising de novo chromosome imbalances have clear and well-established roles in poor pregnancy outcomes. One of the more extreme examples of this is the imbalance of whole haploid complements of chromosomes. Most commonly, this presents as triploidy, the presence of an extra set of chromosomes that could be maternal (digyny) or paternal (diandry) in origin. Triploidy is present in ~8% of miscarriages, but contributes to ~22% of miscarriages occurring after 17 weeks of gestational age (Hardy et al. 2016). Phenotypes differ depending on the parental origin of the extra chromosome complement, highlighting the importance of genomic imprinting. Digynic triploids have small placentas that are generally insufficient to support fetal growth to term, and are associated with severe growth restriction and congenital malformations in the fetus (McFadden and Robinson 2006). Diandric triploid fetuses can be normally-sized or growth restricted, with congenital malformations, and have placentas often presenting as a partial hydatidiform mole (PHM) (McFadden and Pantzar 1996; McFadden and Robinson 2006). This placental phenotype of a PHM overlaps that of a complete hydatidiform mole (CHM), both being characterized by villous trophoblast hyperplasia and hydropic degeneration of the chorionic villi. CHMs are typically due to an androgenetic conceptus (diploid, but chromosomes are entirely paternally-derived), though rarely may be biparental with loss of maternal methylation at imprinted loci (Sanchez-Delgado et al. 2015). Distinguishing these with genetic testing is relevant, as recurrence risk in the latter case is high, since biparental CHMs can be due to rare maternal mutations that appear to affect imprint setting in the oocyte (discussed below). In CHM and PHM, not only is placental development severely abnormal, but embryonic development is abnormal (PHM) or lacking (CHM) and there is an increased risk for PE and trophoblastic tumors, though risk is much lower in cases of partial compared to complete moles (Jauniaux. 1999; Petts et al. 2014).

Chromosome aneuploidy (monosomy or trisomy) is the most common cause of miscarriage, and is detectable in approximately 50–70% of early embryos (Baart et al. 2006). Early miscarriage involving aneuploidy is likely caused by defects in trophoblast functions that are crucial for implantation and establishment of exchange of materials. Although trisomy 13, 18, 21 and sex chromosome aneuploidies can be present in a viable fetus, the role of the placenta in intrauterine survival and phenotypes associated with these aneuploidies is under-appreciated. The vast majority of embryos with trisomy 13 and 18 are spontaneously aborted, but the presence of normal diploid cells in the placental trophoblast cells may allow survival of these fetuses to term (Kalousek et al. 1989). Similar placental mosaicism is not common for viable trisomy 21, suggesting that the additional chromosome 21 has less of a detrimental impact on placental function, and thus embryo survival, as for chromosomes 13 and 18. Despite this, placental development is impacted in the case of trisomy 21, as there is defective formation of the syncytiotrophoblast (Frendo et al. 2000). This affects placental secretion and transfer of nutrients, and is the mechanism underlying the increased levels of human chorionic gonadotropin (hCG) in maternal blood used for prenatal screening for trisomy 21 (Pidoux et al. 2007).

In addition to impacts on the fetus, defective placental development and function caused by aneuploidy can impact maternal health. For example, trisomy 13 is associated with increased incidence of maternal PE (Dotters-Katz et al. 2018; Tuohy and James 1992). Of the many genes on chromosome 13 with increased dosage, FLT1 is of particular interest, as levels of soluble FLT1 (sFLT1) in maternal serum have been associated with PE in multiple independent studies (Maynard et al. 2003; Robinson et al. 2006). FLT1 is expressed mainly in the syncytiotrophoblast and released into maternal circulation where it binds to placental growth factor (PlGF) and vascular endothelial growth factor (VEGF), blocking them from binding to receptors on endothelial cells (Jim and Karumanchi 2017). High levels of sFLT1 lead to endothelial damage in the mother (Maynard et al. 2003), thus directly implicating it in the pathogenesis of PE.

While we tend to focus on the impacts of aneuploidy in the fetus, aneuploidy confined to extraembryonic lineages, i.e. confined placental mosaicism (CPM), can also impact fetal development. Placental mosaicism is identified in approximately 1–2% prenatal chorionic villus samples (CVS), however it is only confirmed upon amniocentesis approximately 10% of the time (Ledbetter et al. 1992; Phillips et al. 1996; Wang et al. 1993). Furthermore, even non-mosaic trisomy detected on CVS is commonly confined to the placenta when involving a non-viable trisomy diagnosed in ongoing pregnancies. Although the fetus may not carry detectable levels of abnormal cells, CPM detected prenatally has been associated with numerous complications, including stillbirth, FGR, PTB, and congenital malformations (Johnson et al. 1990; Kalousek et al. 1991; Toutain et al. 2018; Yong et al. 2003). In particular, CPM may account for a significant proportion of cases of idiopathic FGR, being present in the placentas of approximately 10% of growth restricted newborns but at much lower rates in controls (Robinson et al. 2010; Stipoljev et al. 2001; Wilkins-Haug et al. 1995). Outcomes of prenatally-detected CPM depend on the chromosome involved and levels of placental trisomy, with more severe outcomes and higher levels of trisomy associated from trisomy arising from a meiotic rather than post-zygotic error (Robinson et al. 1997; Wolstenholme. 1996). For example, trisomy 16 is almost always due to maternal meiotic nondisjunction with the diploid cell population arising from a post-zygotic rescue, thus tends to be associated with high levels of trisomy in the placenta, and newborn birth weights are nearly always below the population mean. Placental trisomy 16 is also associated with an increased risk of maternal PE (Yong et al. 2006), and higher levels of trisomy 16 on direct CVS (trophoblast) are associated with increased risk of congenital malformations (Yong et al. 2003). Follow-up of infants with CPM of trisomy 16 show the majority of infants with FGR have catch-up growth, and though there are few reports of global developmental delay, these are only in the subset of cases where mosaicism was also detected in amniotic fluid (Langlois et al. 2006).

Placental mosaicism in the age of NIPT

In addition to the risks for fetal and maternal health, it is increasingly important to consider the possibility of abnormalities being confined to the placenta as we move toward widespread use of noninvasive prenatal testing (NIPT) to diagnose aneuploidy and other genetic abnormalities of the fetus. In NIPT, the “fetal fraction” of circulating cell-free DNA in maternal blood is evaluated, but this “fetal” DNA is largely of placental trophoblast origin (Alberry et al. 2007). It may derive from cells (extravillous trophoblast) and trophoblast-derived microvessicles/exosomes that are actively released into the maternal space, however a substantial portion derives from syncytiotrophoblast debris resulting from apoptosis/necrosis and release of multicellular fragments. There is growing evidence of CPM contributing to false positive and false negative NIPT results; in a recent meta-analysis, 33% of false-positives due to a biological or technical reason were attributed to CPM (Hartwig et al. 2017). Currently, confirmation of any positive screening result from NIPT using invasive diagnostic methods, such as amniocentesis, is recommended by the Society of Obstetricians & Gynecologists of Canada, the Canadian College of Medical Genetics, and the American College of Medical Genetics (Audibert et al. 2017; Gregg et al. 2016). As NIPT potentially moves into testing for chromosome abnormalities beyond common trisomies and sex chromosome aneuploidies, CPM becomes more relevant to consider. Currently, NIPT has very low positive predictive value for rare autosomal trisomies (Pescia et al. 2017), potentially in part due to the higher likelihood of the aneuploid cell population being confined to the placenta. Furthermore, some trisomies, such as trisomy 8, may persist only when trisomy is absent from the trophoblast (Wolstenholme 1996), thus leading to increased false negatives. Should testing for rare autosomal trisomies using NIPT move forward, CPM may contribute to significantly more false-positives in these cases. Recently, methods to predict placental mosaicism from NIPT results have been developed (Brison et al. 2018), which have the potential to both improve predictive value of NIPT and better our understanding of the prevalence and impact of placental mosaicism in ongoing pregnancies.

Placental copy number variants: another avenue to investigate placental genetics in fetal and maternal health?

Though the study of placental aneuploidies has revealed significant associations with pregnancy complications, there have been few studies investigating the role of submicroscopic genetic imbalances, copy number variants (CNVs), in the placenta in association with perinatal health outcomes. A number of rare pathogenic CNVs have been identified in fetal samples from miscarriage and stillbirths (Ernst et al. 2015; Rajcan-Separovic et al. 2010), however, to date, results concerning the role of CNVs in the placenta are conflicting. It has been suggested that there is an increased load of CNVs in the placenta, and that this load is highest in healthy term pregnancies compared to numerous pregnancy complications, including PE, large- and small-for-gestational age babies, gestational diabetes, and recurrent pregnancy loss (Kasak et al. 2015; Kasak et al. 2017). However, another study found the opposite, that more CNVs were present in placentas associated with FGR and PE compared to healthy controls, and that load correlated with the severity of the pathology (Biron-Shental et al. 2016). Such discrepancies may be due to small sample sizes, different analysis methods, or factors such as DNA quality and cell composition that vary depending on source or gestational age, and can influence detection of CNVs. Additionally, it is yet unknown if CPM of CNVs is a common occurrence, and whether it may contribute to pregnancy complications. At least one case of a submicroscopic CNV of clinical relevance detected by CVS and confined to placental tissue in a healthy live born male has been reported (Karampetsou et al. 2014). While exciting, the investigation of placental CNVs is still in its infancy, and will require studies with larger sample sizes and careful attention to data quality and rigorous analysis to establish the associations with pregnancy complications in both the fetus and the mother.

Rare genetic variants affecting placental function can contribute to pregnancy complications

Common pregnancy complications such as PE, FGR, and PTB are heterogeneous conditions, and while it seems likely that rare genetic mutations may lead to these disorders, their total contribution is yet undetermined. From genetic linkage studies of familial PE, rare maternal mutations in genes relevant to placental function have been reported, including CORIN, which facilitates proper trophoblast invasion and spiral artery remodeling (Cui et al. 2012), and STOX1, hypothesized to be important in recruitment of uterine natural killer cells (uNK) and monocytes by placental EVTs for successful interaction at the maternal-placental interface (Dunk et al. 2016). Though promising, such findings have failed to replicate in some cohorts (Iglesias-Platas et al. 2007; Kivinen et al. 2007). Rare variants in genes known to be involved in growth have been associated with FGR. For example, a rare report of a paternally-inherited IGF2 nonsense mutation in a family associated with severe fetal and postnatal growth restriction with a Silver-Russell-like phenotype was documented (Begemann et al. 2015). While loss of function of IGF2 has known implications on fetal growth, it may also impact function in placental cells, as IGF2 promotes cytotrophoblast proliferation in vitro (Forbes et al. 2008) and trophoblast-specific loss of Igf2 in mice generates placental and fetal growth restriction (Fowden et al. 2002). Maternal mutations in NLRP7 and KHDC3L have been reported to cause recurrent biparental CHM due to abnormal setting or maintenance of maternal imprints in the oocyte, leading to widespread impacts on imprinting in the placenta (Djuric et al. 2006; Parry et al. 2011; Sanchez-Delgado et al. 2015). Mutations in these genes are not commonly associated with androgenetic CHM, triploidy, recurrent pregnancy loss or infertility (Aghajanova et al. 2015; Mahadevan et al. 2013; Manokhina et al. 2013), thus alternate genetic variants may be involved (Nguyen et al. 2018).

Common genetic variants associated with pregnancy complications converge on a few major pathways of placental function

While the importance of maternal genetic effects in pregnancy complications is well-established, the significance of fetal/placental genetic factors cannot be ignored. In fact, fetal (placental) genetic effects explain 20% of the variation in PE (Cnattingius et al. 2004). Despite the compelling evidence from epidemiological studies supporting fetal contribution to the heritability of PE, genome-wide association studies (GWAS) investigating fetal (placental) single nucleotide polymorphisms (SNPs) in relation to PE are limited. A SNP near the FLT1 locus (rs4769613) is the only PE-associated risk variant that convincingly replicated in an independent European cohort in a GWAS (Gray et al. 2018; McGinnis et al. 2017). This result is convincing both due to its reproducibility and its biological relevance, as sFLT1 levels have been extensively associated with PE. Interestingly, these SNPs were also found to be associated with red blood cell count (McGinnis et al. 2017) and may influence erythropoiesis in the placenta.

A few additional studies have investigated the association between SNPs in candidate genes and PE, though rarely is the placental genotype investigated. Placental genotype at a common SNP in MTHFR (rs1801133) has been associated with PE in one study (Cheaui et al. 2015), and though this failed to replicate in a separate study, there was a trend of increased “TT” genotypes in the combined PE and/or FGR cohort (Del Gobbo et al. 2018). Differences between studies may depend on geographical location, as MTHFR variants may play a less significant role in populations such as North America where there is folate supplementation. Other candidate studies have identified an association between placental variants in immune system genes such as HLA-G and IL10 and an increased susceptibility to PE (Makris et al. 2006; Moreau et al. 2008), but these findings have not yet been replicated in larger cohorts. While rarely are the same variants identified to be associated with PE, alterations in gene pathways involved in angiogenesis, hypoxia and immune responses are often demonstrated, suggesting dysregulation of common biological processes in PE.

Similarly, although few reproducible SNP associations with spontaneous PTB have been reported, common biological pathways underlying spontaneous PTB have been implicated based on genetic association studies. The infection-inflammation response network is the most consistent pathway associated with spontaneous PTB (Uzun et al. 2013). Histologic evidence of inflammation in the placenta, fetal membranes, or umbilical cord is commonly observed in association with spontaneous PTB (Andrews et al. 2006; Salafia et al. 1991). Placental inflammatory lesions are also frequently observed, and a high recurrence risk for these lesions has been well-demonstrated (Ghidini and Salafia 2005), supporting the role of the placenta in genetic susceptibility for spontaneous PTB. Additionally, genetic variation in inflammatory genes in the placenta such as TLR and TNF is associated with preterm rupture of membranes (Rey et al. 2008), which leads to PTB.

A few studies have examined the interaction between maternal and fetal/placental genotypes, identifying compounded risk when the mother and the infant carry specific combinations of genetic variants. For example, maternal IL1B “GG” genotype (rs16944) and fetal IL6 “GC” genotype (rs1800795) in a South American population was associated with increased risk of PTB (Pereyra et al. 2016). No significant association was observed when the variants were analyzed independently, likely because both alleles were predisposing in individuals with European ancestry. Further, it has been shown that the maternal KIR activating “AA” genotype in combination with a fetal HLA-C “C2” allele is associated with an increased risk of placental insufficiency leading to PE and/or low birth weight (Hiby et al. 2014). The risk is greater if the placental C2 is expressed from the paternal chromosome. The KIR receptors on uNK cells bind directly to the HLA-C molecules expressed on trophoblast cells and these interactions likely play an important role in regulating the degree of invasion and allocating resources between the fetus and the mother.

Placental epigenetics in maternal and fetal medicine

While the study of genetic variation in the placenta is important to understanding how some pregnancies may be predisposed to adverse reproductive outcome, there is much information to be leveraged from the placental epigenome as well. Epigenetic modifications such as DNA methylation (DNAme) and histone modifications (e.g. methylation, phosphorylation, acetylation, and ubiquitinylation) are stable marks that are involved in the regulation of gene expression that is key to cell differentiation. Histone modifications and DNAme tend to occur in tandem, working to recruit the other modification to stably activate or repress genes (Cedar and Bergman 2009; Kondo. 2009). Several studies in model organisms and stem cells in vitro, have established the crucial role for histone modifications in early cell fate of the blastocyst extraembryonic lineage and in the development and differentiation of placental cell types (Rugg-Gunn et al. 2010; Saha et al. 2013; Torres-Padilla et al. 2007). For example, the trophectoderm initially has lower levels of H3K27me3, a repressive mark at gene promoters and enhancers, as compared to the inner cells mass (Saha et al. 2013); although this changes dynamically in development and does not correlate with typical patterns of chromatin condensation in trophoblast populations, suggesting unusual properties of histone methylation in these cells (Fogarty et al. 2015). Histone modifications and the enzymes catalyzing them are necessary for the regulation of expression of key placental proteins, including syncytin (Chuang et al. 2006), maspin (Dokras et al. 2006), pregnancy-specific glycoproteins (Camolotto et al. 2013), and the human growth hormone protein family (Ganguly et al. 2015). Altered expression of these genes or in resulting histone modifications have been associated with pregnancy complications including PE, FGR, and gestational diabetes (Alahari et al. 2018; Paauw et al. 2018; Xie et al. 2019).

Due to the ease of its assessment, DNAme has been more extensively studied in human populations and has potential applications to maternal and fetal medicine. Characterization of DNAme in the placenta can improve our current understanding of disease pathogenesis and help identify biomarkers for various placental-mediated pregnancy outcomes. Although a subset of DNAme changes can reflect changes to gene expression in specific cell types, the information obtained from these molecular processes is not identical. For example, DNAme may reflect more stable long-term changes to gene regulation that might be present from earlier in development and therefore represent very early changes predictive of the disease. It might also reveal pathology-associated changes to cell ratios, which can be informative in sub-classifying placentas by phenotype. Additionally, genetic alterations may be reflected in DNAme and therefore help to identify functional genetic variation that may explain disease susceptibility (Figure 2).

Figure 2.

Examples of applications of placental DNAme in assessing and improving maternal and fetal health. These include: detecting altered DNAme linked to in utero exposures; detecting placental-specific DNAme signatures in maternal blood; quantifying maternal immune cells infiltrating the placenta; subclassifying placental phenotypes into more homogenous groups; diagnosing chromosomal imbalance or imprinting errors; and using methylation changes linked to disease to identify nearby predisposing SNPs (mQTLs).

Placental DNAme is unique

The DNAme profile of the placenta is distinct compared to somatic tissues, offering an avenue for development of biomarkers for assessing placental health, as this property can be used to distinguish placental from maternal DNA. Both the trophoblast and the mesenchyme core of the placental chorionic villi have a unique DNAme landscape compared with maternal decidua, fetal membranes (chorion and amnion), and embryonic tissues (brain, kidney, muscle, spinal cord) (Robinson and Price 2015). While on average the placenta is hypomethylated compared to somatic tissues (Ehrlich et al. 1982; Novakovic et al. 2010; Schroeder et al. 2013), this largely reflects that approximately 40% of the placental genome comprises large blocks (>100kb) of low to intermediate methylation, termed partially methylated domains (PMDs) (Schroeder et al. 2013). In contrast to the bimodal distribution of DNAme in somatic tissues, where most CpG sites are either highly (>90%) or lowly (<10%) methylated, the presence of PMDs and imprinted genes in the placenta results in a unique trimodal distribution of DNAme measures. Hypomethylation is also reported for several families of retrotransposable elements (REs) in the placenta compared to other somatic tissues, though the degree of this varies depending on the RE family and on the evolutionary age of the sequence (Chatterjee et al. 2016; Price et al. 2012; Reiss et al. 2007). Furthermore, the DNAme level of LINE-1 elements is highly correlated with location within versus outside of a PMD, rather than being an independent property of these repeat elements themselves (Schroeder et al. 2013). Hypomethylation of RE-derived promoters has been shown to contribute to placental-specific expression of genes or certain gene isoforms important for placental development and function, including: INSL4, involved in placental apoptosis (Macaulay et al. 2011); ILR2B, which contributes to immune communication at the maternal-fetal interface via proliferation and differentiation of uNK cells (Cohen et al. 2011); and expression of endogenous retroviral envelope proteins syntytin-1 and −2, which are involved in the fusion of cytotrophoblast cells to form the syncytiotrophoblast layer (Macaulay et al. 2011; Vargas et al. 2009).

Applications of placental DNAme in maternal & fetal health

Diagnosis and sub-classification of placental-mediated pregnancy complications

In genetic disorders linked to epigenetic dysregulation, specific patterns of DNAme in blood are diagnostic of the underlying epigenetic errors (Butcher et al. 2017). In the placenta, rapid DNAme-based assays at imprinted loci can be used for the diagnosis of parental genetic imbalances, such as those in CHM or triploidy, or to screen for abnormal imprint setting as observed in biparental CHMs (Bourque et al. 2011). Accurate diagnosis of the underlying mechanisms for placentas with mole-like features is important for assessing risk for choriocarcinoma, as altered DNAme patterns are associated with invasive choriocarcinomal cell lines (Novakovic et al. 2008). Further, DNAme at imprinted loci is sensitive enough to provide quantitative estimates of the level of normal and abnormal cells in cases of mosaicism or chimerism (Bourque et al. 2011). Loss of methylation at several imprinted genes has also been observed in a subgroup of placentas associated with infertility and/or assisted reproduction (Choufani et al. 2018).

Widespread DNAme changes are reported in placentas from pregnancies with severe PE, with a subset of these PE-associated DNAme sites being reproduced in multiple studies (Wilson et al. 2018; Yeung et al. 2016). Reproducible DNAme alterations can potentially be used to diagnose cases with characteristic pathology of placental insufficiency, i.e. advanced villous maturation and distal villous hypoplasia. Genes associated with these DNAme changes include those implicated in PE pathogenesis, such as FLT1, CXCL9, an inflammatory chemokine that inhibits angiogenesis and may contribute to inadequate vascular remodeling; JUNB, a transcription factor with altered placental expression in response to hypoxia in trophoblasts (Yuen et al. 2013); and INHBA, a subunit of Activin A, primarily produced by the placenta during pregnancy, that shows increased serum levels in PE-affected women. Clustering placentas from diverse outcomes using independently-validated PE-associated DNAme sites identified an additional cluster of placentas that largely included PTBs with a putative “inflammation” phenotype (Wilson et al. 2018), a key risk factor for many pregnancy complications. A similar clustering pattern with at least 5 distinct sub clusters was observed using gene expression profiling of healthy and affected placentas of mixed etiologies (Leavey et al. 2018). By integrating pathology information, histological lesions of the placenta have been linked to the gene expression-defined subclasses, providing improved characterization of PE subtypes (Benton et al. 2018). For example, one cluster was most highly associated with maternal vascular malperfusion, while another was strongly correlated with histological chorioamnionitis. Such analyses can be used to identify more homogeneous patient populations to improve biomarker development and understanding of the underlying factors contributing to pregnancy complications.

Quantify altered cell ratios in placental-mediated pregnancy complications

As DNAme is cell-specific, pathology-associated DNAme changes in the placenta may be in part indicative of changes to cell composition. For example, PE-associated DNAme changes may reflect an increase in syncytiotrophoblast mass, as well as its dysfunction. While placental cell-specific DNAme signatures are still being developed, testing for cell-specific changes may have applications in diagnosis. This may provide an approach to quantify low levels of maternal immune cell infiltration in various inflammation-mediated placental pathologies. An increase in infiltration of maternal neutrophils is common in acute chorioamnionitis, a commonly reported inflammatory lesion in placentas from spontaneous PTBs. Using neutrophil-specific DNAme sites, acute chorioamnionitis affected placentas could be largely seperated from non-affected placentas, suggesting some DNAme changes associated with the placental pathology may be attributed to an increase in neutrophils as a response to inflammation (Konwar et al. 2018). Similarly, an increased number of nucleated red blood cells in umbilical cord blood of PE-affected pregnancies is observed (Aali et al. 2007; Hebbar et al. 2014), suggesting enhanced erythropoiesis as a fetal response to a hypoxic environment. Cell-specific DNAme levels may potentially be useful to predict the severity of immune-specific responses both in the mother and fetus. While the utility of measuring this in the placenta is not yet explored, microchimeric maternal cells in cord blood have been shown to predict neonatal complications (Harrington et al. 2017).

Biomarker in maternal blood for placental-mediated pregnancy complications

In addition to its role in genetic diagnosis, placental DNA that is released into maternal blood during pregnancy may also be used to assess placental health based on DNAme profiles. Direct quantification of methylated placental DNA in maternal circulation is most feasible at sites where DNAme is exclusive to the placenta and absent in maternal blood, or vice versa. One study identified 958 CpGs that are consistently hypomethylated (β<0.25; β: methylation value ranging from 0–1) or hypermethylated (β>0.75) between maternal blood and placenta, thus providing a list of candidate CpGs with diagnostic potential (Hatt et al. 2015). For instance, the RASSF1A gene promoter is only methylated in placenta and has been used to directly measure placental DNA in maternal plasma, thus may aid in prediction of complications such as PE which is associated with increased trophoblast apoptosis and release of placental DNA into maternal circulation (Zhao et al. 2010). Alternatively, changes in placental DNAme at term may reflect altered protein levels earlier in gestation, providing another route of investigation for identifying biomarkers for placental insufficiency conditions (Wilson et al. 2015).

Sensor of environmental exposures

There has been widespread interest in using DNAme as a sensor of maternal environmental exposures during pregnancy. The most reproduced finding is an association between maternal smoking during pregnancy and altered placental DNAme at AHRR and CYP1A1 (Suter et al. 2010; van Otterdijk et al. 2017), genes reported to mediate toxic effects of nicotine metabolism. Interestingly, pesticide exposure has also been found to be associated with fewer PMDs with increased DNAme (Schmidt et al. 2016), with a greater impact than a number of other environmental exposures measured. In addition, altered placental DNAme at genes such as HTR2A and HSD11B2, which are involved in serotonin and glucocorticoid signaling in the placenta, in response to exposure to maternal stress has been associated with birth weight and neurobehavioural outcomes in infants (Green et al. 2015; Marsit et al. 2012). In another study, sites of altered placenta DNAme were linked to future development of autism; interestingly these overlapped genetic risk loci for autism, and their methylation was affected by prenatal vitamins (Zhu et al. 2018). Further, maternal health conditions such as gestational diabetes mellitus is also associated with altered DNAme in the placenta (Binder et al. 2015; Ruchat et al. 2013). As expected, these DNAme changes are enriched in metabolic and immune response pathways, supporting that a pro-inflammatory maternal environment likely alters the inflammatory profile in the placenta. It is important to note, however, that differences in placental DNAme related to inflammation-mediated diseases such as gestational diabetes may also reflect pathology-linked changes in immune cell populations in the placenta and the changes noted in such studies tend to be small.

Help identify genetic loci associated with poor maternal and fetal health outcomes

Emerging evidence shows that DNAme patterns are strongly influenced by underlying genetic variation. Based on studies in other tissues, approximately 20–80% of DNAme variance within a tissue can be attributed to genetic variants, referred to as methylation-quantitative trait loci (mQTL) (Do et al. 2016; Gertz et al. 2011). For example, a placental mQTL located in the enhancer region of IL6 is associated with both changes in IL6 expression and with acute chorioamnionitis in individuals of Asian ancestry (Konwar et al. 2019). In a comprehensive study of 300 human placentas comparing Illumina DNAme array to SNP array findings, 4,342 placental mQTLs were identified (Delahaye et al. 2018). Pathway analysis of placental mQTLs revealed an enrichment of inflammation-associated pathways primarily involving HLA- related genes that are known to regulate immunological responses during pregnancy. Similar to other tissues, placental mQTLs tended to be located in gene enhancer regions and were associated with transcription factor binding sites. This suggests a plausible impact of mQTLs on gene regulation, which in turn may affect susceptibility to various pregnancy complications. Interestingly, mQTLs identified in blood are frequently enriched among risk loci identified in association with complex diseases (McRae et al. 2018), aiding in prioritization of candidates implicated in disease pathogenesis. It is reasonable to speculate that placental-specific mQTLs may be used to refine GWAS signals for various placental- mediated complex outcomes such as PE, FGR, and spontaneous PTB.

Conclusion & Future Directions

Pregnancy health is currently routinely assessed by a combination of ultrasound, protein markers in serum, and biophysical measurements. As genetic diagnosis using placental cell free DNA circulating in maternal plasma has advanced, expanded opportunities for DNA-based assessment placental health can be considered. Although the role of chromosomal aneuploidy has been known for over half a century and has been the focus of much prenatal testing, the role of rare and common genetic variants, including copy number variants, in placental and fetal development is only beginning to be elucidated. Whether these types of errors may also commonly be mosaic and confined to the placenta, remains to be determined. In addition to diagnosing genetic errors, one can imagine identifying epigenetic biomarkers of risk. However, the application is limited largely by major gaps in our knowledge of placental health and the connections between epigenetic changes in the placenta at delivery and factors in maternal circulation. Understanding the underlying biology of the placenta and integrating pathological features with genetic and epigenetic changes is important in the development of new tools to predict maternal and fetal complications of pregnancy. An integrated assessment of the placenta at delivery also has potential for predicting future cognitive, immune, and metabolic development of the newborn.

A unique challenge to studies of the placenta is that we are not only trying to predict heterogeneous conditions, such as FGR and PTB, that clearly have both genetic and environmental contributions, but the interaction between maternal and placental/fetal genome is also critical. While obtaining large sample sizes is one approach to improve genetic association studies, there may be more power in focusing on improved phenotyping using pathology and molecular profiling, in combination with models that integrate genetic variation in mother and placenta along with common environmental exposures.