Genomic and phenomic insights from an atlas of genetic effects on DNA methylation

Josine L. Min, Gibran Hemani, Eilís Hannon, Koen F. Dekkers, Juan Castillo‐Fernandez, René Luijk, Elena Carnero‐Montoro, Daniel J. Lawson, Kimberley Burrows, Matthew Suderman, Andrew D. Bretherick, Tom G. Richardson, Johanna Klughammer, Valentina Iotchkova, Gemma C. Sharp, Ahmad Al Khleifat, Aleksey Shatunov, Alfredo Iacoangeli, Wendy L. McArdle, Karen Ho, Ashish Kumar, Cilla Söderhäll, Carolina Soriano‐Tárraga, Eva Giralt‐Steinhauer, Nabila Kazmi, Dan Mason, Allan F. McRae, David L. Corcoran, Karen Sugden, Silva Kasela, Alexia Cardona, Felix R. Day, Giovanni Cugliari, Clara Viberti, Simonetta Guarrera, Michael C. Lerro, Richa Gupta, Sailalitha Bollepalli, Pooja R. Mandaviya, Yanni Zeng, Toni‐Kim Clarke, Rosie M. Walker, Vanessa Schmoll, Darina Czamara, Carlos Ruiz-Arenas, Faisal I. Rezwan, Riccardo E. Marioni, Tian Lin, Yvonne Awaloff, Marine Germain, Dylan Aïssi, Ramona A. J. Zwamborn, Kristel van Eijk, Annelot M. Dekker, Jenny van Dongen, Jouke‐Jan Hottenga, Gonneke Willemsen, Cheng‐Jian Xu, Guillermo Barturen, Francesc Català‐Moll, Martin Kerick, Carol A. Wang, Phillip E. Melton, Hannah R. Elliott, Jean Shin, Manon Bernard, İdil Yet, Melissa Smart, T.j. Gorrie-Stone, Chris Shaw, Ammar Al‐Chalabi, Susan M. Ring, Göran Pershagen, Erik Melén, Jordi Jiménez-Conde, Jaume Roquer, Debbie A. Lawlor, John Wright, Nicholas G. Martin, Grant W. Montgomery, Terrie E. Moffitt, Richie Poulton, Tōnu Esko, Lili Milani, Andres Metspalu, John R. B. Perry, Ken K. Ong, Nicholas J Wareham, Giuseppe Matullo, Carlotta Sacerdote, Avshalom Caspi, Louise Arseneault, France Gagnon, Miina Ollikainen, Jaakko Kaprio, Janine F. Felix, Fernando Rivadeneira, Henning Tiemeier, Marinus H. van IJzendoorn, André G. UitterlindenmedRxiv 2020 · DOI

TL;DR

This study presents the largest-ever atlas of genetic influences on DNA methylation (DNAm), identifying over 270,000 genetic associations across more than 420,000 DNAm sites. This comprehensive resource reveals that DNAm is highly polygenic, shaped by natural selection, and intricately linked to complex traits and diseases, offering a new lens to understand gene regulation and its impact on health.

What Problem Does This Paper Address?

DNA methylation (DNAm) is a crucial epigenetic mechanism that regulates gene expression without altering the underlying DNA sequence. While it's known to play a role in various biological processes and diseases, understanding the genetic underpinnings of DNAm variation has been challenging. Previous studies have primarily focused on "cis" effects (genetic variants near the DNAm site), but the vast majority of genetic influences are thought to be "trans" (acting from a distance) and have small effect sizes. Mapping these trans effects requires massive sample sizes to uncover the complex genetic architecture of DNAm and its functional consequences for human health. This paper aims to address this gap by building a comprehensive map of these genetic influences.

Methods

The researchers employed a two-phase discovery study design, leveraging data from the Genetics of DNA Methylation Consortium (GoDMC), an international collaboration.

  • Meta-analysis: They performed a standardized meta-analysis of DNA methylation quantitative trait loci (mQTL) across 36 population-based and disease datasets, integrating genotype and DNAm data from nearly 28,000 participants.
  • Linkage Disequilibrium (LD) Clumping: This technique was used to identify independent genetic associations (SNPs) with DNAm sites.
  • Conditional Analysis: This was used to identify multiple independent SNPs influencing a single DNAm site.
  • Replication: Findings were replicated in the Generation Scotland cohort.
  • Network Construction: They built networks to identify "genomic communities" of DNAm sites that share genetic control.
  • Enrichment Analyses: Various enrichment analyses were conducted to understand the functional and biological properties of mQTLs, including TF binding, chromatin states, and associations with complex traits and diseases.
  • Bi-directional Mendelian Randomization (MR): This was used to investigate potential causal relationships between DNAm levels and complex traits/diseases in both directions.
  • Selection Analyses: They used various metrics (Fst, SDS, iHS, XPEHH) to assess evidence of natural selection acting on mQTL SNPs.

Datasets

  • Discovery Cohorts: Data from 36 population-based and disease datasets, comprising approximately 27,750 European participants.
  • Replication Cohort: The Generation Scotland cohort (n = 5,101).
  • GWAS Summary Statistics: Publicly available GWAS summary statistics for 37 phenotypes across 11 disease/trait categories and later expanded to 116 complex traits and diseases.
  • Reference Panels: The 1000 Genomes reference panel was used for genotype imputation.
  • Epigenomic Data: Data from 127 cell types (including 27 blood cell types) for combinatorial chromatin states.
  • TF Binding Site (TFBS) Data: Data from 171 TFs across 27 cell types.

Key Insights and Ideas

  • DNAm is Highly Polygenic: The study reveals that DNAm levels are influenced by a vast number of genetic variants, not just a few major ones, underscoring the complexity of epigenetic regulation.
  • Trans Effects are Crucial: A significant portion of genetic influences on DNAm are long-range (trans), highlighting the need for large-scale studies to capture these effects.
  • Genomic Communities: The construction of networks revealed "genomic communities" where multiple DNAm sites are co-regulated by shared genetic factors, suggesting coordinated regulatory mechanisms.
  • Natural Selection Shapes DNAm: Evidence of both positive and negative natural selection acting on mQTL SNPs suggests that DNAm patterns have been shaped by evolutionary pressures, potentially preserving ancestral states or adapting to new environments.
  • Complex Genotype-Phenotype Map: While shared genetic factors link DNAm to complex traits and diseases, the relationship is not always a simple causal one, indicating a more intricate interplay between genes, epigenetics, and phenotype.

Key Results

  • Massive mQTL Atlas: Identified over 270,000 independent mQTL associations, linking genetic variants to 420,509 DNAm sites.
  • Trans Associations Dominate: 8.5% of identified mQTLs were long-range (trans) associations.
  • Variance Explained: Identified mQTL associations explain 15-17% of the additive genetic variance of DNAm.
  • Replication Success: High replication rates (76%) in the Generation Scotland cohort demonstrated the robustness of the findings.
  • Functional Enrichments: mQTL SNPs and sites showed significant enrichments for regulatory elements, TF binding sites, and gene annotations, providing functional context.
  • Trait Associations: Found enrichments for mQTLs with 35% of complex traits and diseases, suggesting a role for DNAm in various health conditions.
  • Bi-directional MR: While some evidence for causal relationships was found, the overall picture suggests that DNAm is not always on the direct causal pathway from genotype to phenotype, pointing to more complex interactions.
  • Selection Signatures: Strong evidence of positive selection was found for cis-acting variants, while negative selection appeared to shape the ancestral DNAm structure.

Limitations and Future Work

  • Microarray Coverage: The study was limited by the microarray technology used, which only covered about 2% of the genome and was biased towards promoters, underrepresenting regulatory elements. Future work using EPIC arrays or sequencing technologies can expand coverage.
  • Computational Constraints: The multi-stage study design limited the detection of mQTLs with smaller effect sizes (r² < 1%).
  • Causal Inference Complexity: While MR analyses were performed, establishing definitive causal relationships between DNAm and complex traits remains challenging due to potential pleiotropy and complex interactions.
  • Tissue Specificity: While some tissue specificity was observed, the study primarily focused on blood. Further investigation into tissue-specific mQTLs is warranted.
  • Future Directions: The authors anticipate that this atlas will be invaluable for future studies of genome regulation, second-generation EWAS, and understanding the complex genotype-phenotype map. They highlight the need for large-scale, multi-omic datasets to dissect intricate mechanisms.

Summary updated

Abstract

Abstract Characterizing genetic influences on DNA methylation (DNAm) provides an opportunity to understand mechanisms underpinning gene regulation and disease. Here we describe results of DNA methylation-quantitative trait loci (mQTL) analyses on 32,851 participants, identifying genetic variants associated with DNAm at 420,509 DNAm sites in blood. We present a database of &gt;270,000 independent mQTL of which 8.5% comprise long-range ( trans ) associations. Identified mQTL associations explain 15-17% of the additive genetic variance of DNAm. We reveal that the genetic architecture of DNAm levels is highly polygenic and DNAm exhibits signatures of negative and positive natural selection. Using shared genetic control between distal DNAm sites we construct networks, identifying 405 discrete genomic communities enriched for genomic annotations and complex traits. Shared genetic factors are associated with both blood DNAm levels and complex diseases but in most cases these associations do not reflect causal relationships from DNAm to trait or vice versa indicating a more complex genotype-phenotype map than has previously been hypothesised.
BibTeX
@misc{min2020genomic,
  author = {Josine L. Min and Gibran Hemani and Eilís Hannon and Koen F. Dekkers and Juan Castillo‐Fernandez and René Luijk and Elena Carnero‐Montoro and Daniel J. Lawson and Kimberley Burrows and Matthew Suderman and Andrew D. Bretherick and Tom G. Richardson and Johanna Klughammer and Valentina Iotchkova and Gemma C. Sharp and Ahmad Al Khleifat and Aleksey Shatunov and Alfredo Iacoangeli and Wendy L. McArdle and Karen Ho and Ashish Kumar and Cilla Söderhäll and Carolina Soriano‐Tárraga and Eva Giralt‐Steinhauer and Nabila Kazmi and Dan Mason and Allan F. McRae and David L. Corcoran and Karen Sugden and Silva Kasela and Alexia Cardona and Felix R. Day and Giovanni Cugliari and Clara Viberti and Simonetta Guarrera and Michael C. Lerro and Richa Gupta and Sailalitha Bollepalli and Pooja R. Mandaviya and Yanni Zeng and Toni‐Kim Clarke and Rosie M. Walker and Vanessa Schmoll and Darina Czamara and Carlos Ruiz-Arenas and Faisal I. Rezwan and Riccardo E. Marioni and Tian Lin and Yvonne Awaloff and Marine Germain and Dylan Aïssi and Ramona A. J. Zwamborn and Kristel van Eijk and Annelot M. Dekker and Jenny van Dongen and Jouke‐Jan Hottenga and Gonneke Willemsen and Cheng‐Jian Xu and Guillermo Barturen and Francesc Català‐Moll and Martin Kerick and Carol A. Wang and Phillip E. Melton and Hannah R. Elliott and Jean Shin and Manon Bernard and İdil Yet and Melissa Smart and T.J. Gorrie-Stone and Chris Shaw and Ammar Al‐Chalabi and Susan M. Ring and Göran Pershagen and Erik Melén and Jordi Jiménez-Conde and Jaume Roquer and Debbie A. Lawlor and John Wright and Nicholas G. Martin and Grant W. Montgomery and Terrie E. Moffitt and Richie Poulton and Tōnu Esko and Lili Milani and Andres Metspalu and John R. B. Perry and Ken K. Ong and Nicholas J Wareham and Giuseppe Matullo and Carlotta Sacerdote and Avshalom Caspi and Louise Arseneault and France Gagnon and Miina Ollikainen and Jaakko Kaprio and Janine F. Felix and Fernando Rivadeneira and Henning Tiemeier and Marinus H. van IJzendoorn and André G. Uitterlinden},
  title = {{Genomic and phenomic insights from an atlas of genetic effects on DNA methylation}},
  year = {2020},
  doi = {10.1101/2020.09.01.20180406},
  howpublished = {medRxiv},
}
Figure