Search

Identification of therapeutic targets from genome-wide association studies (GWAS) requires insights into downstream functional consequences. We harmonized 8,613 RNA-sequencing samples from 14 brain datasets to create the MetaBrain resource and performed cis- and trans-expression quantitative trait locus (eQTL) meta-analyses in multiple brain region- and ancestry-specific datasets (n ≤ 2,759). Many of the 16,169 cortex cis-eQTLs were tissue-dependent when compared with blood cis-eQTLs. We inferred brain cell types for 3,549 cis-eQTLs by interaction analysis. We prioritized 186 cis-eQTLs for 31 brain-related traits using Mendelian randomization and co-localization including 40 cis-eQTLs with an inferred cell type, such as a neuron-specific cis-eQTL (CYP24A1) for multiple sclerosis. We further describe 737 trans-eQTLs for 526 unique variants and 108 unique genes. We used brain-specific gene-co-regulation networks to link GWAS loci and prioritize additional genes for five central nervous system diseases. This study represents a valuable resource for post-GWAS research on central nervous system diseases.

Genetic testing in patients with suspected hereditary kidney disease may not reveal the genetic cause for the disorder as potentially pathogenic variants can reside in genes that are not yet known to be involved in kidney disease. To help identify these genes, we have developed KidneyNetwork, that utilizes tissue-specific expression to predict kidney-specific gene functions.KidneyNetwork is a novel method that we used to enrich a kidney RNA-sequencing co-expression network of 878 samples with a multi-tissue network of 31,499 samples. It then uses expression patterns to predict which genes have a kidney-related function and which (disease) phenotypes might result from variants in these genes, based on established gene-phenotype associations. We applied KidneyNetwork to prioritize rare variants in exome sequencing data from 13 kidney disease patients without a genetic diagnosis.KidneyNetwork can accurately predict kidney-specific gene functions and (kidney disease) phenotypes for disease-associated genes. Applying it to exome sequencing data of kidney disease patients allowed us to highlight a convincing candidate gene for kidney and liver cysts: ALG6.We present KidneyNetwork, a kidney-specific co-expression network that accurately predicts which genes have kidney-specific functions and can result in kidney disease. We show the added value of KidneyNetwork by applying it to kidney disease patients without a molecular diagnosis and consequently, we propose ALG6 as candidate gene in one of these patients. We designed an easy-to-use online interface that allows clinicians and researchers to use gene expression and co-regulation data and gene-phenotype connections to accelerate advances in hereditary kidney disease diagnosis and research.

Allele specific expression (ASE) concerns divergent expression quantity of alternative alleles and is measured by RNA sequencing. Multiple studies show that ASE plays a role in hereditary diseases by modulating penetrance or phenotype severity. However, genome diagnostics is based on DNA sequencing and therefore neglects gene expression regulation such as ASE. To take advantage of ASE in absence of RNA sequencing, it must be predicted using only DNA variation. We have constructed ASE models from BIOS (n = 3432) and GTEx (n = 369) that predict ASE using DNA features. These models are highly reproducible and comprise many different feature types, highlighting the complex regulation that underlies ASE. We applied the BIOS-trained model to population variants in three genes in which ASE plays a clinically relevant role: BRCA2, RET and NF1. This resulted in predicted ASE effects for 27 variants, of which 10 were known pathogenic variants. We demonstrated that ASE can be predicted from DNA features using machine learning. Future efforts may improve sensitivity and translate these models into a new type of genome diagnostic tool that prioritizes candidate pathogenic variants or regulators thereof for follow-up validation by RNA sequencing. All used code and machine learning models are available at GitHub and Zenodo.

Coffee and tea are extensively consumed beverages worldwide which have received considerable attention regarding health. Intake of these beverages is consistently linked to, among others, reduced risk of diabetes and liver diseases; however, the mechanisms of action remain elusive. Epigenetics is suggested as a mechanism mediating the effects of dietary and lifestyle factors on disease onset. Here we report the results from epigenome-wide association studies (EWAS) on coffee and tea consumption in 15,789 participants of European and African-American ancestries from 15 cohorts. EWAS meta-analysis of coffee consumption reveals 11 CpGs surpassing the epigenome-wide significance threshold (P-value, While coffee and tea consumption has been associated with risk of diseases, their mechanisms of action remain elusive. Here the authors present a large EWAS on coffee and tea consumption in cohorts of European and African-American ancestries, finding that coffee consumption is associated with differential DNA methylation levels at multiple CpGs.

Background and Aims Genetic testing in patients with suspected hereditary kidney disease does not always reveal the genetic cause for the patient's disorder. Potentially pathogenic variants can reside in genes that are not known to be involved in kidney disease, which makes it difficult to prioritize and interpret the relevance of these variants. As such, there is a clear need for methods that predict the phenotypic consequences of gene expression in a way that is as unbiased as possible. To help identify candidate genes we have developed KidneyNetwork, in which tissue-specific expression is utilized to predict kidney-specific gene functions. Method We combined gene co-expression in 878 publicly available kidney RNA-sequencing samples with the co-expression of a multi-tissue RNA-sequencing dataset of 31,499 samples to build KidneyNetwork. The expression patterns were used to predict which genes have a kidney-related function, and which (disease) phenotypes might be caused when these genes are mutated. By integrating the information from the HPO database, in which known phenotypic consequences of disease genes are annotated, with the gene co-expression network we obtained prediction scores for each gene per HPO term. As proof of principle, we applied KidneyNetwork to prioritize variants in exome-sequencing data from 13 kidney disease patients without a genetic diagnosis. Results We assessed the prediction performance of KidneyNetwork by comparing it to GeneNetwork, a multi-tissue co-expression network we previously developed. In KidneyNetwork, we observe a significantly improved prediction accuracy of kidney-related HPO-terms, as well as an increase in the total number of significantly predicted kidney-related HPO-terms (figure 1). To examine its clinical utility, we applied KidneyNetwork to 13 patients with a suspected hereditary kidney disease without a genetic diagnosis. Based on the HPO terms “Renal cyst” and “Hepatic cysts”, combined with a list of potentially damaging variants in one of the undiagnosed patients with mild ADPKD/PCLD, we identified ALG6 as a new candidate gene. ALG6 bears a high resemblance to other genes implicated in this phenotype in recent years. Through the 100,000 Genomes Project and collaborators we identified three additional patients with kidney and/or liver cysts carrying a suspected deleterious variant in ALG6. Conclusion We present KidneyNetwork, a kidney specific co-expression network that accurately predicts what genes have kidney-specific functions and may result in kidney disease. Gene-phenotype associations of genes unknown for kidney-related phenotypes can be predicted by KidneyNetwork. We show the added value of KidneyNetwork by applying it to exome sequencing data of kidney disease patients without a molecular diagnosis and consequently we propose ALG6 as a promising candidate gene. KidneyNetwork can be applied to clinically unsolved kidney disease cases, but it can also be used by researchers to gain insight into individual genes to better understand kidney physiology and pathophysiology. Acknowledgments This research was made possible through access to the data and findings generated by the 100,000 Genomes Project; http://www.genomicsengland.co.uk.

Celiac disease is an auto-immune disease in which an immune response to dietary gluten leads to inflammation and subsequent atrophy of small intestinal villi, causing severe bowel discomfort and malabsorption of nutrients. The major instigating factor for the immune response in celiac disease is the activation of gluten-specific CD4+ T cells expressing T cell receptors that recognize gluten peptides presented in the context of HLA-DQ2 and DQ8. Here we provide an in-depth characterization of 28 gluten-specific T cell clones. We assess their transcriptional and epigenetic response to T cell receptor stimulation and link this to genetic factors associated with celiac disease. Gluten-specific T cells have a distinct transcriptional profile that mostly resembles that of Th1 cells but also express cytokines characteristic of other types of T-helper cells. This transcriptional response appears not to be regulated by changes in chromatin state, but rather by early upregulation of transcription factors and non-coding RNAs that likely orchestrate the subsequent activation of genes that play a role in immune pathways. Finally, integration of chromatin and transcription factor binding profiles suggest that genes activated by T cell receptor stimulation of gluten‑specific T cells may be impacted by genetic variation at several genetic loci associated with celiac disease.

BackgroundGenetic testing in patients with suspected hereditary kidney disease may not reveal the genetic cause for the disorder as potentially pathogenic variants can reside in genes that are not yet known to be involved in kidney disease. To help identify these genes, we have developed KidneyNetwork, that utilizes tissue-specific expression to predict kidney-specific gene functions.MethodsKidneyNetwork is a co-expression network built upon a combination of 878 kidney RNA-sequencing samples and a multi-tissue dataset of 31,499 samples. It uses expression patterns to predict which genes have a kidney-related function and which (disease) phenotypes might result from variants in these genes. We applied KidneyNetwork to prioritize rare variants in exome sequencing data from 13 kidney disease patients without a genetic diagnosis.ResultsKidneyNetwork can accurately predict kidney-specific gene functions and (kidney disease) phenotypes for disease-associated genes. Applying it to exome sequencing data of kidney disease patients allowed us to identify a promising candidate gene for kidney and liver cysts: ALG6.ConclusionWe present KidneyNetwork, a kidney-specific co-expression network that accurately predicts which genes have kidney-specific functions and can result in kidney disease. We show the added value of KidneyNetwork by applying it to kidney disease patients without a molecular diagnosis and consequently, we propose ALG6 as candidate gene in one of these patients. KidneyNetwork can be applied to clinically unsolved kidney disease cases, but it can also be used by researchers to gain insight into individual genes in order to better understand kidney physiology and pathophysiology.Significance statementGenetic testing in patients with suspected hereditary kidney disease may not reveal the genetic cause for the patient’s disorder. Potentially pathogenic variants can reside in genes not yet known to be involved in kidney disease, making it difficult to interpret the relevance of these variants. This reveals a clear need for methods to predict the phenotypic consequences of genetic variation in an unbiased manner. Here we describe KidneyNetwork, a tool that utilizes tissue-specific expression to predict kidney-specific gene functions. Applying KidneyNetwork to a group of undiagnosed cases identified ALG6 as a candidate gene in cystic kidney and liver disease. In summary, KidneyNetwork can aid the interpretation of genetic variants and can therefore be of value in translational nephrogenetics and help improve the diagnostic yield in kidney disease patients.

Gaining insight into the downstream consequences of non-coding variants is an essential step towards the identification of therapeutic targets from genome-wide association study (GWAS) findings. Here we have harmonized and integrated 8,727 RNA-seq samples with accompanying genotype data from multiple brain-regions from 14 datasets. This sample size enabled us to perform both cis- and trans-expression quantitative locus (eQTL) mapping. Upon comparing the brain cortex cis-eQTLs (for 12,307 unique genes at FDRcis-eQTL analysis (n=31,684 samples), we observed that brain eQTLs are more tissue specific than previously assumed.We inferred the brain cell type for 1,515 cis-eQTLs by using cell type proportion information. We conducted Mendelian Randomization on 31 brain-related traits using cis-eQTLs as instruments and found 159 significant findings that also passed colocalization. Furthermore, two multiple sclerosis (MS) findings had cell type specific signals, a neuron-specific cis-eQTL for CYP24A1 and a macrophage specific cis-eQTL for CLECL1.To further interpret GWAS hits, we performed trans-eQTL analysis. We identified 2,589 trans-eQTLs (at FDRWe also generated a brain-specific gene-coregulation network that we used to predict which genes have brain-specific functions, and to perform a novel network analysis of Alzheimer’s disease (AD), amyotrophic lateral sclerosis (ALS), multiple sclerosis (MS) and Parkinson’s disease (PD) GWAS data. This resulted in the identification of distinct sets of genes that show significantly enriched co-regulation with genes inside the associated GWAS loci, and which might reflect drivers of these diseases.

Our genetic code is stored in our DNA, which consists of four bases: adenine (A), cytosine (C) guanine (G) and thymine (T). More than three billion of these bases are strung together in 23 chromosomes, forming our genome. Each of our cells has two copies of our genome, containing instructions that guide cellular processes such as growth, development, signalling, and many more, but can occasionally also be the basis of disease. A large part of the genetic instructions are contained in our genes, which are small regions in our genome which contain information to make ribonucleic acid molecules (RNA), which can be translated to proteins. Variation in our genomes can change the regulation of genes or the functionality of the protein products. The work in this thesis shows the effects of genetic variation on the activity (the amount of RNA that is produced) of genes. Often, the effect of genetic variation differs between tissues and cell types. We developed a method to study the difference in genetic regulation between different cell types and applied this to samples taken from brain and blood. We show that there are large differences in genetic regulation between brain and blood, and we identify putatively causal disease genes for several neuro-psychiatric disease which could not be identified using only data from blood. This thesis expanded our knowledge of the difference in genetic regulation of gene expression between brain and blood, which can help us in further understanding genetic diseases and in designing drug targets.

Background: Allele-specific expression (ASE) refers to divergent abundances of allelic copies of a transcript and is quantified by RNA sequencing. Multiple studies have shown that ASE likely plays a role in many diseases by modulating the pathogenesis or phenotypes. However, routine genome diagnostics is based on DNA sequencing and therefore neglects the regulation of gene expression such as ASE. To take advantage of ASE information in absence of RNA sequencing, ASE must be predicted using only DNA variation. Results: We have constructed ASE models from BIOS and GTEx that predict local ASE using DNA features. These models are highly stable as shown by performance metrics, cross-validation, and feature importance. Many different types of features are involved in these predictions, highlighting the complex regulation that underlies ASE. We applied the BIOS-trained model to population variants in three genes in which ASE plays a role in disease penetrance or modulation: BRCA2, RET, and NF1. This resulted in predicted ASE effects for 27 out of 8,957 variants. Ten of these were known pathogenic variants, while the rest are interesting candidates for further elucidation of disease etiology. Conclusions: We demonstrated that ASE can be stably and reliably predicted from DNA features using machine-learning models. Future efforts may further improve sensitivity and translate these models into a new class of genome diagnostic in silico prediction tools to prioritize candidate pathogenic variants or regulators thereof for follow-up validation by RNA sequencing.

Background Expression quantitative trait loci (eQTL) studies are used to interpret the function of disease-associated genetic risk factors. To date, most eQTL analyses have been conducted in bulk tissues, such as whole blood and tissue biopsies, which are likely to mask the cell type-context of the eQTL regulatory effects. Although this context can be investigated by generating transcriptional profiles from purified cell subpopulations, current methods to do this are labor-intensive and expensive. We introduce a new method, Decon2, as a framework for estimating cell proportions using expression profiles from bulk blood samples (Decon-cell) followed by deconvolution of cell type eQTLs (Decon-eQTL). Results The estimated cell proportions from Decon-cell agree with experimental measurements across cohorts (R ≥ 0.77). Using Decon-cell, we could predict the proportions of 34 circulating cell types for 3194 samples from a population-based cohort. Next, we identified 16,362 whole-blood eQTLs and deconvoluted cell type interaction (CTi) eQTLs using the predicted cell proportions from Decon-cell. CTi eQTLs show excellent allelic directional concordance with eQTL (≥ 96–100%) and chromatin mark QTL (≥87–92%) studies that used either purified cell subpopulations or single-cell RNA-seq, outperforming the conventional interaction effect. Conclusions Decon2 provides a method to detect cell type interaction effects from bulk blood eQTLs that is useful for pinpointing the most relevant cell type for a given complex disease. Decon2 is available as an R package and Java application (https://github.com/molgenis/systemsgenetics/tree/master/Decon2) and as a web tool (www.molgenis.org/deconvolution).

Most diseases have both an environmental and genetic component. Although many diseases are strongly heritable, individual genetic variants typically confer only a small effect on disease, and thus these diseases are strongly polygenic. Paradoxically, molecular traits, such as gene expression, methylation, protein or metabolite levels, typically have a lower heritability, but sometimes individual genetic variants show much higher effect sizes on these traits. In this review, we discuss the genetic architecture of these molecular traits, and contrast this to the genetic architecture of complex diseases, and provide explanations why strong effects of individual genetic variants on molecular traits do not necessarily need to translate into increased risk of disease.

Summary - Significant insights into disease pathogenesis have been gleaned from population-level genetic studies; however, many loci associated with complex genetic disease contain numerous genes, and phenotypic associations cannot be assigned unequivocally. In particular, a gene-dense locus on chromosome 11 (61.5–61.65 Mb) has been associated with inflammatory bowel disease, rheumatoid arthritis, and coronary artery disease. Here, we identify TMEM258 within this locus as a central regulator of intestinal inflammation. Strikingly, Tmem258 haploinsufficient mice exhibit severe intestinal inflammation in a model of colitis. At the mechanistic level, we demonstrate that TMEM258 is a required component of the oligosaccharyltransferase complex and is essential for N-linked protein glycosylation. Consequently, homozygous deficiency of Tmem258 in colonic organoids results in unresolved endoplasmic reticulum (ER) stress culminating in apoptosis. Collectively, our results demonstrate that TMEM258 is a central mediator of ER quality control and intestinal homeostasis., Leona M. and Harry B. Helmsley Charitable Trust (2014PG-IBD016), Crohn's and Colitis Foundation of America, National Institutes of Health (U.S.) (grant DK043351), National Institutes of Health (U.S.) (grant DK097485)

Background: DNA methylation at GFI1-locus has been repeatedly associated with exposure to smoking from foetal period onwards. We explored whether DNA methylation maybe a mechanism that links exposure to maternal prenatal smoking with offspring's adult cardio-metabolic health. Methods: We meta-analysed the association between methylation at GFI1-locus with maternal prenatal smoking, adult own smoking, and cardio-metabolic phenotypes in 22 population-based studies from Europe, Australia, and USA (n=18212). DNA methylation at GFI1-locus was measured in whole-blood. Multivariable regression analyses were fitted to examine its association with exposure to prenatal and own adult smoking. DNA methylation levels were analysed in relation to body mass index (BMI), waist circumference (WC), fasting glucose (FG), high-density lipoprotein cholesterol (HDL-C), triglycerides (TG), diastolic, and systolic blood pressure (BP). Findings: Hypomethylation at four out of eight GFI1-CpGs was associated with exposure to maternal prenatal smoking, whereas, all eight CpGs were associated with adult own smoking. Cg14179389 hypomethylation, the strongest maternal prenatal smoking locus, was associated with increased WC and BP when adjusted for sex, age, and adult smoking with Bonferroni-corrected P < 0·012. In contrast, hypomethylation at cg09935388, the strongest adult own smoking locus, was associated with decreased BMI, WC, and BP (adjusted 1 x 10 - 7 < P < 0·01). Similarly, hypo-methylation at cg12876356, cg18316974, cg09662411, and cg18146737 was associated with decreased BMI and WC (5 x 10 - 8 < P < 0·001). Hypomethylation at all the CpGs was consistently associated with higher TG levels. Interpretation: Epigenetic changes at GFI1 were linked to smoking exposure in-utero/in-adulthood and robustly associated with cardio-metabolic risk factors. This introduces a molecular mechanism and biomarker of smoking exposure. Funding: Not applicable Declaration of interests: The authors declare no competing interests. Ethical Approval Statement: Ethical approval for the study was obtained from the ALSPAC Ethics and Law Committee and the Local Research Ethics Committees.

Genetic risk factors often localize to noncoding regions of the genome with unknown effects on disease etiology. Expression quantitative trait loci (eQTLs) help to explain the regulatory mechanisms underlying these genetic associations. Knowledge of the context that determines the nature and strength of eQTLs may help identify cell types relevant to pathophysiology and the regulatory networks underlying disease. Here we generated peripheral blood RNA-seq data from 2,116 unrelated individuals and systematically identified context-dependent eQTLs using a hypothesis-free strategy that does not require previous knowledge of the identity of the modifiers. Of the 23,060 significant cis-regulated genes (false discovery rate (FDR) ≤ 0.05), 2,743 (12%) showed context-dependent eQTL effects. The majority of these effects were influenced by cell type composition. A set of 145 cis-eQTLs depended on type I interferon signaling. Others were modulated by specific transcription factors binding to the eQTL SNPs.

Transcription factors often form protein complexes and give rise to intricate transcriptional networks. The regulation of transcription factor multimerization plays a key role in the fine-tuning of the underlying transcriptional pathways and can be exploited to modulate synthetic transcriptional modules. A novel regulation of protein complex formation is emerging: microProteins-truncated transcription factors-engage in protein-protein interactions with transcriptional complexes and modulate their transcriptional activity. Here, we outline a strategy for the discovery, design, and test of putative miPs to fine-tune the activity of transcription factors regulating synthetic or natural transcriptional circuits.

Transcription factors often form protein complexes and give rise to intricate transcriptional networks. The regulation of transcription factor multimerization plays a key role in the fine-tuning of the underlying transcriptional pathways and can be exploited to modulate synthetic transcriptional modules. A novel regulation of protein complex formation is emerging: microProteins-truncated transcription factors-engage in protein-protein interactions with transcriptional complexes and modulate their transcriptional activity. Here, we outline a strategy for the discovery, design, and test of putative miPs to fine-tune the activity of transcription factors regulating synthetic or natural transcriptional circuits.

The zebrafish has the capacity to regenerate its heart after severe injury. While the function of a few genes during this process has been studied, we are far from fully understanding how genes interact to coordinate heart regeneration. To enable systematic insights into this phenomenon, we generated and integrated a dynamic co-expression network of heart regeneration in the zebrafish and linked systems-level properties to the underlying molecular events. Across multiple post-injury time points, the network displays topological attributes of biological relevance. We show that regeneration steps are mediated by modules of transcriptionally coordinated genes, and by genes acting as network hubs. We also established direct associations between hubs and validated drivers of heart regeneration with murine and human orthologs. The resulting models and interactive analysis tools are available at http://infused.vital-it.ch. Using a worked example, we demonstrate the usefulness of this unique open resource for hypothesis generation and in silico screening for genes involved in heart regeneration. We thank C. Hoffmann and J. Esposito for technical support, A. Feenstra (VUA) for co-supervision of N.K., and R. Schneider (LCSB) for co-supervision of G.A. during her work at LIH and current support at LCSB. This research was funded by Luxembourg's National Research Fund (FNR) and the Swiss National Research Foundation (SNF), INFUSED project (www.infused-project.eu). N.M. was supported by Comunidad de Madrid (Fibroteam P2010/BMD-2321) and the ERC (Starting Grant 2013 337703 zebraHeart). Zebrafish drawing in Figure 1 was adapted from Wikimedia Commons (http://bit.ly/1PL92vj). Sí

An emerging concept in transcriptional regulation is that a class of truncated transcription factors (TFs), called microProteins (miPs), engages in protein-protein interactions with TF complexes and provides feedback controls. A handful of miP examples have been described in the literature but the extent of their prevalence is unclear. Here we present an algorithm that predicts miPs and their target TFs from a sequenced genome. The algorithm is called miP prediction program (miP3), which is implemented in Python. The software will help shed light on the prevalence, biological roles, and evolution of miPs. Moreover, miP3 can be used to predict other types of miP-like proteins that may have evolved from other functional classes such as kinases and receptors. The program is freely available and can be applied to any sequenced genome.

Truncated transcription factor-like proteins called microProteins (miPs) can modulate transcription factor activities, thereby increasing transcriptional regulatory complexity. To understand their prevalence, evolution, and function, we predicted over 400 genes that encode putative miPs from Arabidopsis (Arabidopsis thaliana) using a bioinformatics pipeline and validated two novel miPs involved in flowering time and response to abiotic and biotic stress. We provide an evolutionary perspective for a class of miPs targeting homeodomain transcription factors in plants and metazoans. We identify domain loss as one mechanism of miP evolution and suggest the possible roles of miPs on the evolution of their target transcription factors. Overall, we reveal a prominent layer of transcriptional regulation by miPs, show pervasiveness of such proteins both within and across genomes, and provide a framework for studying their function and evolution.

Background: DNA methylation at the GFI1-locus has been repeatedly associated with exposure to smoking from the foetal period onwards. We explored whether DNA methylation may be a mechanism that links exposure to maternal prenatal smoking with offspring's adult cardio-metabolic health.Methods: We meta-analysed the association between DNA methylation at GFI1-locus with maternal prenatal smoking, adult own smoking, and cardio-metabolic phenotypes in 22 population-based studies from Europe, Australia, and USA (n= 18,212). DNA methylation at the GFI1-locus was measured in whole-blood. Multivariable regression models were fitted to examine its association with exposure to prenatal and own adult smoking. DNA methylation levels were analysed in relation to body mass index (BMI), waist circumference (WC), fasting glucose (FG), high-density lipoprotein cholesterol (HDL-C), triglycerides (TG), diastolic, and systolic blood pressure (BP).Findings: Lower DNA methylation at three out of eight GFI1-CpGs was associated with exposure to maternal prenatal smoking, whereas, all eight CpGs were associated with adult own smoking. Lower DNA methylation at cg14179389, the strongest maternal prenatal smoking locus, was associated with increased WC and BP when adjusted for sex, age, and adult smoking with Bonferroni-corrected P Interpretation: Epigenetic changes at the GFI1 were linked to smoking exposure in-utero/in-adulthood and robustly associated with cardio-metabolic risk factors. Fund: European Union's Horizon 2020 research and innovation programme under grant agreement no. 633595 DynaHEALTH. (c) 2018 The Authors. Published by Elsevier B.V.