Epigenomics

ObjectiveExposure to cigarette smoking during pregnancy has been robustly associated with cord blood DNA methylation. However, little is known about such effects on the placenta; in particular, whether cigarette smoking before pregnancy could also induce epigenetic alterations in the placenta of former smokers is unknown. Design and resultsPlacental DNA methylation levels were measured in 568 women and compared among non-smokers and women either smoking during their pregnancy or who had ceased smoking before pregnancy. An Epigenome Wide Association Study identified 344 Differentially Methylated Regions (DMRs) significantly associated with maternal smoking status. Among these 344 DMRs, 262 showed "reversible" alterations of DNA methylation, only present in the placenta of current smokers, whereas 44 were also found altered in former smokers, whose placenta had not been exposed directly to cigarette smoking. This observation was further supported by a significant demethylation of LINE-1 sequences in the placentas of both current (-0.43 (-0.83 to -0.02)) and former smokers (-0.55 (-1.02 to -0.08)) compared to nonsmokers. A comparative analysis of the epigenome landscape based on the ENCODE placenta data demonstrated an enrichment of all 344 DMRs in enhancers histone marks. Additionally, smoking-associated DMRs were found near and/or overlapping with 13 imprinting gene clusters encompassing 18 imprinted genes. ConclusionsDNA methylation patterns alterations were found in 344 genomic regions in the placenta of women smoking during their pregnancy, including 44 DMRs and LINE-1 elements, where methylation changes persisted in former smokers, supporting the hypothesis of an "epigenetic memory" of exposure to cigarette smoking before pregnancy. Enhancers regions, including imprinting control regions were also particularly affected by placenta methylation changes associated to smoking, suggesting a biological basis for the sensitivity of these regions to tobacco exposure and mechanisms by which fetal development could be impacted.

BackgroundPlacental-originated cell-free DNA (cfDNA) provides unique opportunities to study (epi)genetic placental programming remotely, but studies investigating the cfDNA methylome are scarce and usually technologically challenging. Methylated DNA sequencing (MeD-seq) is well-compatible with low cfDNA concentrations and has a high genome-wide coverage. We therefore aim to investigate the feasibility of genome-wide methylation profiling of first trimester maternal cfDNA using MeD-seq, by identifying placental-specific methylation marks in cfDNA. MethodsWe collected cfDNA from non-pregnant controls (female n=6, male n=12) and pregnant women (n=10), first trimester placentas (n=10), and paired preconceptional and first trimester buffy coats (total n=20). Differentially methylated regions (DMRs) were identified between pregnant and non-pregnant women. We investigated placental-specific markers in maternal cfDNA, including RASSF1 promoter and Y-chromosomal methylation, and studied overlap with placental and buffy coat DNA methylation. ResultsWe identified 436 DMRs between cfDNA from pregnant and non-pregnant women which were validated using male cfDNA. RASSF1 promoter methylation was higher in maternal cfDNA (fold change 2.87, unpaired t-test p<0.0001). Differential methylation of Y-chromosomal sequences could determine fetal sex. DMRs in maternal cfDNA showed large overlap with DNA methylation of these regions in placentas and buffy coats, indicating a placental and immune-cell contribution to the pregnancy-specific cfDNA methylation signature. Sixteen DMRs in maternal cfDNA were specifically found only in placentas. These novel potential placental-specific DMRs were more prominent than RASSF1. ConclusionsMeD-seq can detect (novel) genome-wide placental DNA methylation marks and determine fetal sex in maternal cfDNA. This study supports future research into maternal cfDNA methylation using MeD-seq. Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=134 SRC="FIGDIR/small/610227v1_ufig1.gif" ALT="Figure 1"> View larger version (51K): org.highwire.dtl.DTLVardef@b8fd55org.highwire.dtl.DTLVardef@ffe942org.highwire.dtl.DTLVardef@12afbf4org.highwire.dtl.DTLVardef@1036a6a_HPS_FORMAT_FIGEXP M_FIG C_FIG Studies investigating the maternal cell-free DNA (cfDNA) methylome are scarce and generally technologically challenging. We identified 436 autosomal differentially methylated regions (DMRs) between cfDNA from pregnant and non-pregnant women, using the innovative methylated DNA sequencing (MeD-seq) technique. Y-chromosomal methylation could determine fetal sex, we show hypermethylation of the placental-marker RASSF1, and identify 16 novel placental-specific markers in maternal cfDNA including DMRs related to TMEM240, DHRS3, and PCMTD2. This pilot study supports future research into the maternal cfDNA methylome using MeD-seq.

BackgroundEpigenetic alterations during fetal development have been proposed as key factors explaining associations between maternal lifestyle during pregnancy and later health outcomes in the offspring, pertaining to the developmental origin of health and disease (DOHAD) hypothesis. ObjectivesTo assess the association of maternal lifestyle with offsprings birth weight and underlying epigenetic mediatory mechanisms in the NorthPop prospective birth cohort. MethodsA three-step analytic pipeline was applied. In 722 mother-child pairs, overall associations between 10 maternal lifestyle factors and the offsprings standardized birth weight were first evaluated by multiple linear regression. Three high dimensional mediation methods (HDMA, HIMA, and HIMA2) were then applied on the beta methylation matrix to identify candidate CpG mediators in cord blood driving the significant overall associations. Finally, robust- and ordinary least squares-regression-based classical mediation, including single- and multiple-(parallel and serial) mediator models were assessed. ResultsGestational weight gain (GWG) ({beta}-adj = 0.03; p = 2x10-5) and maternal BMI at the beginning of pregnancy ({beta}-adj = 0.036; p = 1x10-4) were significantly associated with the offsprings standardized birth weight. High dimensional mediation analyses identified pooled sets of four (cg19242268; cg08461903; cg14798382; cg21516291) and five (cg17040807; cg19242268; cg26552621; cg04457572; cg06457011) candidate CpG mediators related to GWG and BMI at the beginning of pregnancy, respectively. For both exposures, classical mediation analyses revealed a range of significant single- and multiple (both serial and parallel) mediator models via both robust- and OLS-regression based approaches. These indicated the likely presence of individual-, causally linked multiple-, and causally independent multiple mediatory pathways underlying the two significant overall associations. ConclusionsOur findings support the hypothesis that neonatal health effects related to maternal lifestyle may be partly mediated by epigenetic alterations. Findings also suggest the possible involvement of multiple DNA methylation sites via various mediatory pathways.

BackgroundMicroRNAs (miRNAs) are a class of small non-coding RNAs regulating gene expression. They are involved in many biological processes, including adaptation to pregnancy. The identification of genetic variants associated with gene expression, known as expression quantitative trait loci (eQTL), helps to understand the underlying molecular mechanisms and determinants of complex diseases. Using data from the prospective pre-birth Gen3G cohort, we investigated associations between maternal genotypes and plasmatic miRNA levels measured during the first trimester of pregnancy of 369 women. ResultsAssessing the associations between about 2 million SNPs and miRNA proximal pairs using best practices from the GTEx consortium, a total of 22,140 significant eQTLs involving 147 unique miRNAs were identified. Elastic-net regressions were applied to select the most relevant SNPs to build genetic risk scores (GRS) for each of these 147 miRNAs. For about half of the circulating miRNAs, the GRS captured >10% of the variance abundance. As a demonstration of the usefulness of the identified eQTLs and derived GRS, we used the GRSs as instrumental variables to test for association between the circulating levels of miRNAs quantified before the 16th week of pregnancy and the development of pregnancy complications (gestational diabetes [GDM] or pre-eclampsia [PE]) developing more than three months later on average. Using predicted miRNA levels derived from instrumental variables, we found 18 significant associations of miRNAs with potential support of causal inference for GDM or PE. ConclusionsOur results represent a valuable resource to understand miRNA regulation and highlight the potential of genetic instruments in predicting circulating miRNA levels and their possible contribution in disease development.

BackgroundColorectal cancer (CRC) is associated with patient demographics, lifestyle exposures and molecular alterations. However, it is not possible to determine which adenomas will progress to CRC, as ethically it is unacceptable to leave and follow adenomas. We hypothesised that certain lifestyle exposures at high levels could precondition exposed bowel tissue by changing and aging it, increasing the risks of deleterious DNA methylation and genetic alterations. We used a novel study design comparing adenomas with concurrent CRC (thus more likely exposed to deleterious lifestyle effects) to single adenomas in bowels with no history of CRC; we called these high (HR) and low-risk (LR) adenomas respectively. MethodsWe carried out a discovery and replication epigenome-wide association study (EWAS) on 106 HR and 111 LR adenomas, profiled with MethylationEPIC BeadChips. In order, to identify differentially methylated positions (DMP), regions (DMR), and DNAm (DNAmethylation) lifestyle exposures and risks, with adjustment for confounders, and gene ontology (GO) and pathway enrichment. Then, two open-source gene expression omnibus (GEO) validation datasets (52, 57 and 49, 48 HR and LR normal bowel tissues respectively) were analysed for these DNAm lifestyle exposures and risks, with adjustment for confounders. ResultsOur EWAS found 5 Bonferroni significant DMPs with absolute delta betas [&ge;] 5%, and 14 significant DMRs with absolute mean DMR delta betas [&ge;] 5%, replicated in the GPX7, RGS3 and TMEM135 cancer-associated genes. DNAm high alcohol exposures were strongly associated with increased risk of HR adenomas (odds ratio (OR) per standard deviation (SD) = 2.16 (95% confidence interval (CI) 1.55 - 3.09, p-value = 9.7 x 10-6)). In the validation datasets, DNAm high alcohol (ORperSD = 2.12 (95% CI 1.35 - 3.55, p-value = 2.0 x 10-3) and ORperSD = 1.79 (95% CI 1.14 - 2.96, p-value = 1.7 x 10-2)), and high body mass index (BMI) exposures (ORperSD = 1.72 (95% CI 1.13 - 2.73, p-value = 1.5 x 10-2)) were associated with increased risk of HR normal bowel tissues. ConclusionsHigh alcohol and BMI exposures may precondition normal bowel tissues and adenomas for increased risk of DNA methylation alterations associated with CRC progression. The DNAm exposure signatures and our newly identified genes may be useful epigenetic biomarkers for CRC prevention.

The Illumina Infinium MethylationEPIC v2.0 BeadChip (EPICv2 array) is a microarray for assessment of the human epigenome. Sites on the EPICv2 array are annotated with an open-source file provided by Illumina, the EPICv2 manifest. Of the 923,452 unique genomic sites targeted by the EPICv2 array, the Illumina manifest identifies just 214,808 as mapping to a gene, excluding many sites located within a gene body. Based on the genomic coordinates of probes, we have mapped each site assayed on the Illumina EPICv2 array using publicly available data, comprehensively annotating affiliated genes and regulatory elements. We have found that a total of 700,392 EPICv2 array sites are located within a gene body (exon, intron, or UTR) according to the GENCODE Human release 47 (GENCODEv47) database. 509,940 of these sites were not annotated as being within a gene in the Illumina EPICv2 manifest, primarily because the Illumina manifest does not annotate introns - 498,407 of the excluded sites, or 97.74%, are located within the intron of at least one transcript. The Illumina EPICv2 manifest annotates 358,539 sites as being within 1500bp of a transcription start site (TSS). Using a distance-based approach, we have labelled 267,183 sites as being within promoter distance of a gene (<1500bp upstream or <500bp downstream of the TSS), and 140,123 sites as being within enhancer distance (1501-5000bp upstream of the TSS, excluding sites located within a gene body). We re-annotated the EPICv2 manifest using GENCODEv47 data to label intragenic features, and a distance-based approach to label the regulatory genome. We also include a column indicating whether a site is located in any promoter or enhancer, according to the GeneHancer database. The re-annotated manifest additionally labels which sites are required for the Horvath DNA Methylation Age Calculator and MethylDetectR epigenetic clocks, to facilitate data preparation for these tools. In conclusion, we have re-annotated the EPICv2 manifest, allowing more complete assessment of EPICv2 sites associated with gene bodies and regulatory regions during the interpretation of epigenetic studies. The re-annotated manifest is publicly available - see the Data Availability section of this article.

Gene-lifestyle interaction analyses have identified genetic variants whose effect on cardiovascular risk-raising traits is modified by alcohol consumption and smoking behavior. The biological mechanisms of these interactions remain largely unknown, but may involve epigenetic modification linked to perturbation of gene expression. Diverse, individual-level datasets including genotypes, methylation and gene expression conditional on lifestyle factors, are ideally suited to study this hypothesis, yet are often unavailable for large numbers of individuals. Summary-level data, such as effect sizes of genetic variants on a phenotype, present an opportunity for multi-omic study of the biological mechanisms underlying gene-lifestyle interactions. We propose a method that unifies disparate, publicly available summary datasets to build mechanistic hypotheses in models of smoking behavior and alcohol consumption with blood lipid levels and blood pressure measures. Of 897 observed genetic interactions, discovered through genome-wide analysis in diverse multi-ethnic cohorts, 48 were identified with lifestyle-related differentially methylated sites within close proximity and linked to target genes. Smoking behavior and blood lipids account for 37 and 28 of these signals respectively. Five genes also showed differential expression conditional on lifestyle factors within these loci with mechanisms supported in the literature. Our analysis demonstrates the utility of summary data in characterizing observed gene-lifestyle interactions and prioritizes genetic loci for experimental follow up related to blood lipids, blood pressure, and cigarette smoking. We show concordance between multiple trait-or exposure-related associations from diverse assays, driving hypothesis generation for better understanding gene-lifestyle interactions.

In the United States, cardiovascular disease is the leading cause of death, and the rate of maternal mortality remains among the highest of any industrialized nation. Maternal cardiometabolic health throughout gestation and postpartum is representative of placental health and physiology. Both proper placental functionality and placental microRNA expression are essential to successful pregnancy outcomes, and both are highly sensitive to genetic and environmental sources of variation. While placental pathologies, such as preeclampsia, are associated with maternal cardiovascular health and may contribute to the developmental programming of cardiovascular disease, the role of more subtle alterations to placental function and microRNA expression in this relationship remains poorly understood. To develop a more comprehensive understanding of how cardiometabolic health influences placental microRNA expression, and how this shapes placental functionality, we performed small RNA sequencing to investigate microRNA in the placentae from the Rhode Island Child Health Study (n=230). We modeled microRNA counts on maternal family history of cardiovascular disease using negative binomial generalized linear models, and identified microRNAs that were differential expressed (DEmiRs) at a false discovery rate (FDR) less than 0.10. Utilizing parallel mRNA sequencing data and bioinformatic target prediction software, we identified potential mRNA targets of these DEmiRs. We identified 9 DEmiRs, with predicted targets of those miRNA enriched overwhelmingly in the TGF{beta} signaling pathway but also in pathways involving cellular metabolism and immunomodulation. Overall, we identified a robust association existing between familial cardiovascular disease and placental microRNA expression which may be implicated in both placental insufficiencies and the developmental programming of cardiovascular disease.

Circulating cell-free DNA (cfDNA) carries fragmentation patterns that serve as biomarkers of cancer, but standard sequencing approaches miss large portions of the fragment length spectrum. In addition to altered fragmentation patterns, cancer patients often have elevated levels of cfDNA, but the underlying mechanisms are not well understood. To address both questions, we analyzed cancer cases with elevated cfDNA levels using Oxford Nanopore (ONT) sequencing. Long-read ONT sequencing captures the full spectrum of cfDNA fragment lengths and enables cell type inference based on DNA methylation markers. One cohort included cases from several cancer types with elevated cfDNA levels, and a second consisted of patients from a single neuroendocrine cancer study. In each cohort, cases with the highest cfDNA levels showed either hypofragmentation (excess fragments of 1-4 kb) or hyperfragmentation (excess fragments <145 bp). Hypofragmentation reflected blood cell DNA released during delayed sample processing, bearing DNASE1L3-associated hallmarks, while in one cohort we also observed ultra-long fragments (>7.5 kb) lacking these hallmarks and consistent with plasma lysis. By contrast, hyperfragmented samples often had elevated levels of both cancer- and blood-derived DNA, indicating an inflammatory or other system process rather than cancer-specific origin. These findings clarify the distinction between biological and artifactual fragmentation, expand our understanding of cfDNA biology, and highlight long-read sequencing as a powerful tool for biomarker discovery.

To distinguish DNA methylation (DNAm) from cell proportion changes in whole placental tissue research, we developed a robust cell type-specific DNAm reference to estimate cell composition. We collated newly collected and existing cell type DNAm profiles quantified via Illumina EPIC or 450k microarrays. To estimate cell composition, we deconvoluted whole placental samples (n=36) with robust partial correlation based on the top 50 hyper- and hypomethylated sites per cell type. To test deconvolution performance, we evaluated RMSE in predicting principal component one of DNAm variation in 204 external placental samples. We analyzed DNAm profiles (n=368,435 sites) from 12 cell types: cytotrophoblasts (n=18), endothelial cells (n=19), Hofbauer cells (n=26), stromal cells (n=21), syncytiotrophoblasts (n=4), six lymphocyte types (n=36), and nucleated red blood cells (n=11). Median cell composition was consistent with placental biology: 60.4% syncytiotrophoblast, 17.1% stromal, 8.8% endothelial, 4.5% cytotrophoblast, 3.9% Hofbauer, 1.7% nucleated red blood cells, and 1.2% neutrophils. Our expanded reference outperformed an existing reference in predicting DNAm variation (15.4% variance explained, IQR=21.61) with cell composition estimates (RMSE:10.51 vs. 11.43, p-value<0.001). This cell type reference can robustly estimate cell composition from whole placental DNAm data to detect important cell types, reveal biological mechanisms, and improve casual inference.

BackgroundDNA methylation (DNAm) may play an important role in birth outcomes. Material and MethodsGenome wide DNAm was analysed in peripheral blood DNA of women at multiple time points during gestation. A novel empirical method was used to identify CpG sites with high temporal variance in methylation associating with preterm birth. ResultsHigh variability at 1296 CpG sites from the promoter regions of 1197 genes significantly associated with PTB. These genes belonged to pathways involved in signalling by platelet derived growth factor, platelet homeostasis, collagen degradation, extracellular matrix and circadian clock. ConclusionsThe findings provide novel information which might help in development of predictive biomarkers of preterm birth outcome.

Preterm birth is a global public health crisis which results in significant neonatal and maternal mortality. Yet little is known regarding the molecular mechanisms of idiopathic spontaneous preterm birth, and we have few diagnostic markers for adequate assessment of placental development and function. Previous studies of placental pathology and our transcriptomics studies suggest a role for placental maturity in idiopathic spontaneous preterm birth. It is known that placental DNA methylation changes over gestation. We hypothesized that if placental hypermaturity is present in our samples, we would observe a unique idiopathic spontaneous preterm birth DNA methylation profile potentially driving the gene expression differences we previously identified in our placental samples. Our results indicate the idiopathic spontaneous preterm birth DNA methylation pattern mimics the term birth methylation pattern suggesting hypermaturity. Only seven significant differentially methylated regions fitting the idiopathic spontaneous preterm birth specific (relative to the controls) profile were identified, indicating unusually high similarity in DNA methylation between idiopathic spontaneous preterm birth and term birth samples. We identified an additional 1,718 significantly methylated regions in our gestational age matched controls were the idiopathic spontaneous preterm birth DNA methylation pattern mimics the term birth methylation pattern, again indicating a striking level of similarity between the idiopathic spontaneous preterm birth and term birth samples. Pathway analysis of these regions revealed differences in genes within the WNT and Cadherin signaling pathways, both of which are essential in placental development and maturation. Taken together, these data demonstrate that the idiopathic spontaneous preterm birth samples are molecularly more mature than expected given their respective gestational age which likely impacts birth timing.

Preeclampsia (PE) is a severe pregnancy complication that threatens maternal and neonatal health. Previous epigenome-wide association studies (EWAS) on PE have produced inconsistent results, possibly due to inadequate adjustment for confounders. Here, we analyzed DNA methylation changes in cord blood from newborns affected by PE, using a multi-ethnic cohort from Hawaii. We comprehensively adjusted for clinical variables (maternal age, BMI, parity) and estimated cell proportions. Additionally, we re-analyzed two public datasets with similar adjustments and conducted a meta-analysis combining all three datasets to increase statistical power. To further address confounding by gestational age, we also included idiopathic preterm samples as controls. After adjusting for cell type proportions and clinical characteristics, all previously reported significant CpG methylation changes associated with severe PE disappeared across our data, the two public datasets, and the meta-analysis. This result remained even after including idiopathic preterm samples. Instead, severe PE was associated with shifts in CD8T and natural killer (NK) cell proportions. We validated this lack of CpG changes using multiple published cord blood methylation datasets. Moreover, we observed that gestational progression itself is accompanied by significant changes in granulocyte, nRBC, CD8T, and B cell proportions. In summary, our study demonstrates that many previously reported DNA methylation changes in severe PE are artifacts caused by confounding factors such as cell type heterogeneity and gestational age. Severe PE is associated with changes in cell proportions rather than direct methylation alterations. These findings emphasize the importance of rigorous confounder adjustment in EWAS.

Reference methylomes, used in deconvolution algorithms to determine cell-free DNA tissue sources, were based on driver CpGs from either microarray or sequencing platforms. Cross-validation of these algorithms is important to allow interpretation of data across studies, select optimal sequencing depth, and thus reduce costs of cf-DNA deconvolution assays. Towards this end, we assessed the performance of two reference-based deconvolution algorithms: cfDNAme, sequencing-based methylome signatures, and Meth-Atlas, a microarray-based methylome signatures using a cfDNA bisulfite sequencing. While both algorithms use NNLS model, cfDNAme uses CpG windows, while Meth-Atlas uses individual CpGs as cell or tissue signatures. We determined the optimal the number of informative CpGs signatures, and the best sequencing depths for precise deconvolution. We found that above 5-fold coverage, much lower coverage than what is frequently used, there is little difference between our two chosen algorithms, both identifying the correct tissue make-up with a high accuracy, suggesting that whole genome bisulfite sequencing for tissue of origin identification can be completed in a much more cost-effective manner than previously thought.

Fetal sex-specific differences in placental morphology and physiology have been associated with sexually dimorphic health outcomes. However, the molecular mechanisms underlying these sex differences are not well understood. We performed whole genome bisulfite sequencing in 133 placenta samples and discovered a significant difference in DNA methylation (DNAm) at the ZNF300 gene locus between male and female offspring and replicated this result in 6 independent datasets. Additionally, the sex-specific pattern appears to be placenta-specific, is robust to a wide range of gestational ages and adverse health outcomes and is present in sorted placenta villous cytotrophoblast cells. Integration of DNAm, genetic, and placental morphology data from the same individuals revealed ZNF300 methylation is also associated with placenta area, perimeter, and max diameter, genetic variants on chromosomes 5 and X, and may mediate the effects of genetic variation on placental area.

Antenatal corticosteroids (ACS) are provided to improve perinatal survival when there is risk of preterm birth. Though evidence suggests increased risk of developing neurobehavioural disorders in exposed offspring, the mechanisms that mediate this relationship remain largely unknown. Here, we investigated the DNA methylation patterns in the prefrontal cortex (PFC) of exposed offspring. We hypothesized that differential methylation will be evident at both newborn and juvenile ages. Pregnant guinea pigs were administered saline or betamethasone (1mg/kg) on gestational days 50/51 to mimic a single course of ACS. gDNA was isolated from the PFC of term-born offspring on postnatal day 1 (PND1) and PND14 to identify differentially methylated CpG sites (DMCs) using reduced representative bisulfite sequencing. In the PND1 PFC, 1521 DMCs, annotating to 145 genes were identified following ACS. Identified genes were involved in pathways regulating developmental cellular process. In the PND14 PFC, 776 DMCs representing 46 genes were identified, and were enriched in synaptic signalling pathways. Though no individual DMCs were identified at both PND1 and PND14, differential methylation was consistently observed at the binding sites of transcription factors PLAGL1, TFAP2C, ZNF263, and SP1 at both ages. In this study, we identified an altered DNA methylome in the PFC of ACS-exposed guinea pig offspring at both newborn and juvenile ages. Notably, a unique methylation signature was consistently observed at four key transcription factor binding sites at multiple post-natal time points, indicating a persistent change which may predispose the development of altered neurobehavioural phenotypes that have been described in exposed offspring.

Examining DNA in a liquid biopsy for non-invasive cancer detection relies on identifying dilute signal in a high background. This study aims to identify DNA methylation biomarkers for multi-cancer detection. Utilizing large tissue datasets, we apply novel search algorithms to discover confined biomarker panels capable of distinguishing tumor from normal and determining the tissue of origin. We explore the applicability to blood-based testing using targeted methylation sequencing followed by machine learning classification. We present an 8-marker panel, which successfully predicts tumors across 14 types with a 91% average sensitivity, maintaining a low false positive rate (< 0.04%). Additionally, a panel of 39 CpG sites exhibits accuracies ranging from 69% to 98% for identifying tissue of origin. When tested on 114 patient plasma samples (colon, liver, pancreatic, prostate, and stomach cancer), the 8-marker panel obtains an AUC of 0.78 with a 78% sensitivity among 32 early-stage patients (stage I-II), and 60% overall. Using the 39-marker panel in a multi-class classification model selecting only the best match, 54% of tumor samples were on average correctly assigned to the tissue of origin, and up to 80% when allowing more inclusive criteria. Using a limited set of biomarkers, our work contributes to advancing non-invasive cancer diagnostics.

Aims and methodsOur aim was to characterise the methylation level of a polymorphically imprinted gene, VTRNA2-1/nc886, in human populations and somatic tissues. We utilised 48 datasets, consisting of >30 different tissues and >30 000 individuals. ResultsWe show that the nc886 methylation status is associated with twin status and ethnic background, but the variation between populations is limited. Monozygotic twin pairs present concordant methylation, while [~]30% of dizygotic twin pairs present discordant methylation in the nc886 locus. The methylation levels of nc886 are uniform across somatic tissues, except in cerebellum and skeletal muscle. ConclusionWe hypothesize that the nc886 imprint is established in the oocyte and that after implantation, the methylation status is stable, excluding a few specific tissues.

The phenotype of human placental extravillous trophoblast (EVT) at the end of pregnancy reflects both first trimester differentiation from villous cytotrophoblast (CTB) and later gestational changes, including loss of proliferative and invasive capacity. Invasion abnormalities are central to two major placental pathologies, preeclampsia and placenta accreta spectrum, so characterization of the corresponding normal processes is crucial. In this report, our gene expression analysis, using purified human CTB and EVT cells, highlights an epithelial- mesenchymal transition (EMT) mechanism underlying CTB-EVT differentiation and provides a trophoblast-specific EMT signature. In parallel, DNA methylation profiling shows that CTB cells, already hypomethylated relative to non-trophoblast cell lineages, show further genome- wide hypomethylation in the transition to EVT. However, a small subgroup of genes undergoes gains of methylation (GOM) in their regulatory regions or gene bodies, associated with differential mRNA expression (DE). Prominent in this GOM-DE group are genes involved in the EMT, including multiple canonical EMT markers and the EMT-linked transcription factor RUNX1, for which we demonstrate a functional role in modulating the migratory and invasive capacities of JEG3 trophoblast cells. This analysis of DE associated with locus-specific GOM, together with functional studies of an important GOM-DE gene, highlights epigenetically regulated genes and pathways acting in human EVT differentiation and invasion, with implications for obstetric disorders in which these processes are dysregulated.

Despite considerable advances in identifying risk factors for obesity development, there remains substantial gaps in our knowledge about its etiology. Variation in obesity (defined by BMI) is thought to be due in part to heritable factors; however, obesity-associated genetic variants only account for a small portion of heritability. Epigenetic regulation defined by genetic and/or environmental factors with changes in gene expression, may account for some of this "missing heritability". Epigenetic studies of obesity have largely been conducted in populations of European ancestry, despite the disproportionate burden of obesity in African Americans (AAs). To address race/ethnic (RE)-differences in obesity, we conducted a BMI epigenome-wide association study (EWAS) meta-analysis using AA participants from the Jackson Heart Study (JHS, n=1604) and the Multi-Ethnic Study of Atherosclerosis (MESA, n=179). Analyses using a linear regression model with methylation as the outcome and continuous BMI as the predictor were stratified by study and sex, then meta-analyzed. There were 208 methylation sites (CpGs) that reached epigenome-wide significance (p< 8.72x10-8); 151 of these were novel. Of the novel CpGs, 29 CpGs were available for replication testing in a separate sample of AA and 20 replicated. Differentially methylated region (DMR) analysis resulted in 54 DMRs significantly associated with BMI. Several regions are proximal to, or include, genes previously associated with obesity traits (e.g., SOCS3, ABCG1, and TGFB1) in GWAS. Gene and trait enrichment and pathway analysis showed enrichment for genes in immune system and inflammation related pathways (e.g., the IL-6/JAK/STAT pathway). In conclusion, EWAS of BMI in AAs replicated previously known associations identified in European and multi-ethnic EWAS and identified novel obesity-associated CpGs.