An age old question: Which came first: the chicken or the egg? Determining the cause and effect relationship isn’t easy, and the same is true for the relationship between DNA methylation (DNAme) and gene expression. Who is strong-arming who? And even more important (at least to the pragmatists among us), how do both of these work in tandem to influence physiological traits and disease? In a large collaborative effort and massive display of bioinformatics muscle led by the labs of Francis Collins (National Human Genome Research Institute, USA), Laura Scott (University of Michigan Medical School, USA), and Ewan Birney (European Bioinformatics Institute, UK), this team tackles that very question.
To understand the scope of their undertaking by the numbers, the talented team analyzed skeletal muscle samples from 318 patients. They probed 20,952 genes using RNA-seq and 727,141 DNAme sites using the EPIC array to examine their influence on eight physiological traits: height, waist, weight, waist-hip ratio, BMI, fasting insulin, fasting glucose, and type 2 diabetes status. They continued by using QTL analysis to map both gene expression and DNAme quantitative trait loci at 7,128,878 autosomal single nucleotide variants. Finally, they integrated their DNAme and gene expression data with GWAS data spanning 534 disease/quantitative traits.
Using the statistical tool of Mendelian randomization to predict causal relationships, they generated thousands of data points. Here are the highlights:
- There are several hundred genes and DNAme sites associated with fasting insulin, waist, and BMI
- They identified 213 gene-DNAme site pairs (115 genes, 190 DNAme sites) that are predicted to have a causal relationship
- 137 pairs (64%) predict methylation to causally influence expression, and 76 pairs (36%) predict expression to influence methylation
- By overlapping gene expression and DNAme studies with GWAS data for disease/quantitative traits, they identified 86 gene-DNAme-trait relationships
As an example of this method’s potential, the authors identify a link between genetic regulators of the RXRA transcription factor gene and body composition traits, suggesting that increased RXRA expression results in lower lean body mass.
So was it the chicken or the egg? At least for DNAme and gene expression, the causality runs in both directions, and this study takes us one step closer towards fully identifying the physiological traits they influence as well.