DNA methylation in pre-eclampsia: methodologies and analytics

Abstract

The placenta is a heterogeneous and multifunctional organ that grows and adapts according to increasing and fluctuating fetal demands in the intrauterine environment. As expected, the placenta plays a major role in pregnancy outcomes. Pre-eclampsia (PET) is one of the most common adverse pregnancy complications, that occurs in approximately 10% of pregnancies worldwide, yet the underlying cause is unknown. Unfortunately, the only known cure is delivery of the fetus and placenta. Little progress has been made on treatment and management of pre-eclampsia because the condition usually remains undiagnosed until after the 24th week of gestation, by which time the condition is well established. The use of biomarkers to accurately identify women with an increased risk of developing pre-eclampsia would be a major step forward in antenatal care.

Over the last decade, a growing number of studies have sought to define a correlation between epigenetic modifications and the development of pre-eclampsia. However, reported results have been inconsistent. This may reflect small sample numbers and a lack of standardisation of analytical methods. The lack of independent replication of key findings constrains our understanding of the biological role of DNA methylation in pre-eclampsia. Over the past few years, the field has become focused on using the epigenetic features of cell-free fetal DNA (cff-DNA) to create a universal ‘fetal’ marker in maternal plasma. The molecular analysis of cff-DNA may therefore lead to new non-invasive approaches for assessing placental (fetal) health and detecting pregnancy complications.

This PhD project seeks to identify a panel of candidate epigenetic biomarkers for pre-eclampsia that can ultimately be detected in cff-DNA in maternal plasma. The work presented in this thesis has three main objectives. The first objective is to discover and validate epigenetic (DNA methylation) differences in pre-eclamptic placentas by performing and analysing reduced representation bisulfite sequencing (RRBS) data and publicly available 450K data. The second objective was to establish a targeted bisulfite sequencing protocol for genomic DNA and cf-DNA. To do this, I used the Illumina MiSeq platform to validate differentially methylated candidates that were generated from the discovery phase (objective 1). I also used MiSeq sequencing to quantify placental methylation of cf-DNA in pregnant maternal plasma. Targeted bisulfite sequencing generates a large amount of high quality and high coverage sequence data. Thus, the third objective of this PhD project was to develop a bioinformatic pipeline for the analysis of high-throughput targeted bisulfite sequencing (MiSeq) data.

In the discovery phase of this project, methylation on a genome-wide scale at single-nucleotide resolution was conducted to screen for methylation biomarkers, with 15 selected placental samples(9 PET and 6 CTL). We identified several genomic locations where variation in methylation between pre-eclamptic tissues and unaffected controls could be associated with pre-eclampsia. Interestingly, among the differentially methylated genes, there was enrichment for a set of genes that have been found in cancer studies to function as tumour suppressor genes. We propose that these loci are epigenetically regulated and may be functionally relevant in the pathophysiology of pre-eclampsia. Complementary to the RRBS analysis, in-silico analysis of published 450K DNA methylation data identified several differentially methylated positions (DMPs). Among those loci identified by discovery analysis, 28 were selected for independent validation. Targeted high-throughput bisulfite sequencing was chosen as it provides a more focused analysis of the selected regions. Our data demonstrated correlation between the discovery tools (RRBS and 450K) and the validation technology (targeted bisulfite sequencing). Our validation analysis confirmed three differentially methylated fragments from RRBS analysis (PCDHB17, GPR123, XRN2), two differentially methylated cytosines (AL353597.1, TBX15), and four differentially methylated positions, identified by in-silico analysis, in an independent cohort. We demonstrated the targeted bisulfite sequencing approach to be a powerful validation method on predefined regions and surrounding CpG sites to confirm aberrant methylation in the pre-eclamptic methylome. Since all targets were amplified and sequences as a multiplexed pool, this method proved efficient in reducing the cost and time while enhancing the coverage. Our results also demonstrate the powerful of bioinformatics pipeline developed for downstream analysis.

In the cff-DNA part of this study, I selected five epigenetic marker genes from the literature that displayed opposite and extreme methylation differences between the placenta and somatic tissues. I then established a targeted bisulfite sequencing (Illumina MiSeq) protocol to quantify the methylation of the epigenetic marker genes (RASSF1A, SERPINB5, AIRE58, AIREA1 and PSMB5) in plasma samples taken from healthy pregnant and non-pregnant individuals. By quantifying the methylation of these marker genes, I was able to determine the proportion of placental DNA in maternal blood. Isolation and quantification of cff-DNA is a critical step in developing non-invasive screening strategies for the identification of epigenetic biomarkers in maternal plasma. By knowing the exact contribution of placental DNA in maternal plasma, a future project will be able to continue the second phase and ultimate aim of this project which is the quantification of methylation biomarkers in maternal plasma as a predictor for pregnancy complications, particularly, pre-eclampsia. The study also demonstrated a clear correlation between increasing cff-DNA and increasing gestational age.

To the best of this author’s knowledge, this study is the first to report the validation of RRBS and 450K-derived candidate biomarkers using Illumina MiSeq. Moreover, I believe it is the first to establish cff-DNA quantification by targeted bisulfite sequencing. Further optimisation is warranted before confirming the utility as a clinical biomarker. Genomic data sets and the rapidly expanding bioinformatic tools to analyse them provide a unique opportunity to quickly evaluate biomarker concepts with minimal cost and identify promising markers for prospective validation on precious clinical specimens.