Epigenetics in health and disease: heralding the EWAS era

Abstract

Success in the identification of genetic variants that affect complex human phenotypes, such as height, weight, and common diseases, is one of the major achievements in contemporary biomedical research. Insight into the functional complexity of the genome also draws attention to the probable role of non-sequence-based genomic variation in health and disease. Notably, substantial attention is focused on the role of epigenetic processes that might regulate gene expression via modifications to DNA, histone proteins, and chromatin in medical traits. Although the role of epigenetic mechanisms in some rare developmental syndromes and in cancer is well established, systematic examination of their contribution to common non-malignant disease phenotypes is only just beginning.New microarray-based and sequencing-based technologies allow economical, high-throughput profiling of epigenetic marks, with a primary focus on DNA methylation; the era of the epigenome-wide association study (EWAS) of large numbers of samples has begun. In The Lancet, Katherine Dick and colleagues describe the first systematic analysis of the association between variation in DNA methylation and body-mass index (BMI).1Dick KJ Nelson CP Tsaprouni L et al.DNA methylation and body-mass index: a genome-wide analysis.Lancet. 2014; (published online March 13.)http://dx.doi.org/10.1016/S0140-6736(13)62674-4PubMed Google Scholar They report significant associations between methylation at three probes targeting specific CpG sites within intron 1 of HIF3A and BMI in a discovery cohort, and subsequently confirm them in two independent cohorts. For every 10% increase in methylation of the most significant probe—cg22891070—BMI increased by 3·6% (95% CI 2·4–4·9), equating to about 0·98 kg/m2 for a person in the discovery cohort with a BMI of 27 kg/m2 on average.1Dick KJ Nelson CP Tsaprouni L et al.DNA methylation and body-mass index: a genome-wide analysis.Lancet. 2014; (published online March 13.)http://dx.doi.org/10.1016/S0140-6736(13)62674-4PubMed Google Scholar The increase in BMI was higher in individuals who had had a myocardial infarction (4·6%, 2·9–6·3) than in blood donors (2·3%, 0·4–4·1).1Dick KJ Nelson CP Tsaprouni L et al.DNA methylation and body-mass index: a genome-wide analysis.Lancet. 2014; (published online March 13.)http://dx.doi.org/10.1016/S0140-6736(13)62674-4PubMed Google Scholar To put the size of this epigenetic association into perspective, the minor allele of FTO—robustly associated with obesity-related traits—accounts for a more modest 0·39 kg/m2 increase in BMI.2Loos RJ Yeo GS The bigger picture of FTO—the first GWAS-identified obesity gene.Nat Rev Endocrinol. 2014; 10: 51-61Crossref PubMed Scopus (341) Google Scholar HIF3A encodes a component of the hypoxia inducible transcription factor that mediates the cellular response to hypoxia by regulating expression of many downstream genes.3Greer SN Metcalf JL Wang Y Ohh M The updated biology of hypoxia-inducible factor.EMBO J. 2012; 31: 2448-2460Crossref PubMed Scopus (412) Google Scholar This transcription factor has been previously implicated in metabolism4Shin MK Drager LF Yao Q et al.Metabolic consequences of high-fat diet are attenuated by suppression of HIF-1alpha.PLoS One. 2012; 7: e46562Crossref PubMed Scopus (48) Google Scholar and obesity,5Jiang C Qu A Matsubara T et al.Disruption of hypoxia-inducible factor 1 in adipocytes improves insulin sensitivity and decreases adiposity in high-fat diet-fed mice.Diabetes. 2011; 60: 2484-2495Crossref PubMed Scopus (209) Google Scholar providing a biologically plausible mechanism behind the reported association with BMI.Epigenetic epidemiology is an area of great research interest; in the past year, EWAS have been reported for several other human health phenotypes, such as multiple sclerosis,6Huynh JL Garg P Thin TH et al.Epigenome-wide differences in pathology-free regions of multiple sclerosis-affected brains.Nat Neurosci. 2013; 17: 121-130Crossref PubMed Scopus (181) Google Scholar rheumatoid arthritis,7Liu Y Aryee MJ Padyukov L et al.Epigenome-wide association data implicate DNA methylation as an intermediary of genetic risk in rheumatoid arthritis.Nat Biotechnol. 2013; 31: 142-147Crossref PubMed Scopus (673) Google Scholar pain sensitivity,8Bell JT Loomis AK Butcher LM et al.Differential methylation of the TRPA1 promoter in pain sensitivity.Nat Commun. 2014; 5: 2978Crossref PubMed Scopus (103) Google Scholar and metabolic traits.9Petersen AK Zeilinger S Kastenmuller G et al.Epigenetics meets metabolomics: an epigenome-wide association study with blood serum metabolic traits.Hum Mol Genet. 2013; 23: 534-545Crossref PubMed Scopus (140) Google Scholar Dick and colleagues1Dick KJ Nelson CP Tsaprouni L et al.DNA methylation and body-mass index: a genome-wide analysis.Lancet. 2014; (published online March 13.)http://dx.doi.org/10.1016/S0140-6736(13)62674-4PubMed Google Scholar used a powerful sequential-replication design to do one of the most systematic epigenetic studies of a human physiological phenotype yet reported. However, the results of any EWAS need to be interpreted carefully, with clear caveats when compared with genetic studies of complex traits. Unlike genetics, a range of potentially important confounding factors need to be considered, such as tissue or cell type, age, sex, drug exposure, and reverse causation.10Mill J Heijmans BT From promises to practical strategies in epigenetic epidemiology.Nat Rev Genet. 2013; 14: 585-594Crossref PubMed Scopus (248) Google ScholarA primary concern in epigenetic epidemiology is the tissue-specific nature of the epigenome. In large well-phenotyped sample cohorts, such as the discovery cohort used in Dick and colleagues' study,1Dick KJ Nelson CP Tsaprouni L et al.DNA methylation and body-mass index: a genome-wide analysis.Lancet. 2014; (published online March 13.)http://dx.doi.org/10.1016/S0140-6736(13)62674-4PubMed Google Scholar DNA from peripheral tissues (normally whole blood) is often the only source of biological material available. To circumvent this issue, Dick and colleagues subsequently examined the relation between DNA methylation at their top-ranked loci and BMI in adipose and skin tissue from an independent sample cohort, recording strong associations in adipose tissue but not skin.1Dick KJ Nelson CP Tsaprouni L et al.DNA methylation and body-mass index: a genome-wide analysis.Lancet. 2014; (published online March 13.)http://dx.doi.org/10.1016/S0140-6736(13)62674-4PubMed Google Scholar Furthermore, the investigators examined the correlation between DNA methylation and HIF3A expression in adipose tissue, reporting a significant inverse correlation and drawing attention to the potential functional relevance of epigenetic variation at the identified locus. This result is important, because it suggests that assessment of DNA methylation in whole blood can identify robust and biologically relevant epigenetic variation related to BMI.Of course, blood itself is a heterogeneous mix of epigenetically distinct cell types, with variation in cell counts between individuals a potentially huge confounder in EWAS analyses. Dick and colleagues included basic cell-count data that allowed some control for cellular heterogeneity;1Dick KJ Nelson CP Tsaprouni L et al.DNA methylation and body-mass index: a genome-wide analysis.Lancet. 2014; (published online March 13.)http://dx.doi.org/10.1016/S0140-6736(13)62674-4PubMed Google Scholar another approach is the use of robust algorithms to infer cellular composition from epigenomic data.11Houseman EA Molitor J Marsit CJ Reference-free cell mixture adjustments in analysis of DNA methylation data.Bioinformatics. 2014; (published online Feb 14.)https://doi.org/10.1093/bioinformatics/btu029Crossref PubMed Scopus (298) Google Scholar However, if the cellular content of a tissue is strongly associated with the trait being studied—as is likely for inflammatory or neurodegenerative disorders, for example—any apparent trait-associated epigenetic differences could partially reflect differences in cellular composition, even after statistical correction.Another important issue concerns the ability to distinguish between cause and effect. For example, the epigenetic variation reported by Dick and colleagues1Dick KJ Nelson CP Tsaprouni L et al.DNA methylation and body-mass index: a genome-wide analysis.Lancet. 2014; (published online March 13.)http://dx.doi.org/10.1016/S0140-6736(13)62674-4PubMed Google Scholar could have arisen before any alteration in BMI, contributing directly to obesity-related phenotypes. Alternatively, it could represent a secondary, downstream effect of variation in BMI itself or another exposure associated with variation in BMI. Therefore, the most robust design for epigenetic epidemiology involves the longitudinal assessment of epigenetic changes within the context of a prospective cohort study, so that epigenetic variation can be related to temporal changes in exposures and phenotype.10Mill J Heijmans BT From promises to practical strategies in epigenetic epidemiology.Nat Rev Genet. 2013; 14: 585-594Crossref PubMed Scopus (248) Google ScholarDick and colleagues attempt to address the issue of causality by applying a mendelian randomisation approach12Relton CL Davey Smith G Two-step epigenetic Mendelian randomization: a strategy for establishing the causal role of epigenetic processes in pathways to disease.Int J Epidemiol. 2012; 41: 161-176Crossref PubMed Scopus (260) Google Scholar to interrogate the causal relation between HIF3A methylation and BMI. This approach uses a genetic proxy for DNA methylation (namely, methylation quantitative trait loci) to identify a causal relation between an exposure or trait and epigenetic variation, assuming that genetic associations are largely immune to residual confounding and reverse causation. Dick and colleagues identified two upstream single nucleotide polymorphisms that were independently associated with DNA methylation at a HIF3A locus in both the discovery and replication cohorts. However, these single nucleotide polymorphisms were not associated with BMI in the study cohorts or the high-powered GIANT consortium dataset,13Speliotes EK Willer CJ Berndt SI et al.Association analyses of 249,796 individuals reveal 18 new loci associated with body mass index.Nat Genet. 2010; 42: 937-948Crossref PubMed Scopus (2206) Google Scholar suggesting that hypermethylation at the HIF3A locus is likely to be a result of increased BMI rather than a causal association between increased methylation and BMI. A non-causal association between methylation and a phenotype could still be informative as a diagnostic or prognostic biomarker—eg, HIF3A methylation might predict disease phenotypes associated with BMI, such as cancer and cardiovascular disease.Dick and colleagues' study1Dick KJ Nelson CP Tsaprouni L et al.DNA methylation and body-mass index: a genome-wide analysis.Lancet. 2014; (published online March 13.)http://dx.doi.org/10.1016/S0140-6736(13)62674-4PubMed Google Scholar represents an important advance for both obesity-related research and the specialty of epigenetic epidemiology. BMI is a good phenotype for population-based epigenomic studies: it is an accurate measure that is routinely collected in most cohort studies. The widespread uptake of instruments such as the Illumina 450K HumanMethylation array means that large collaborative EWAS meta-analyses can be done, building on the success of similar approaches in genetics.13Speliotes EK Willer CJ Berndt SI et al.Association analyses of 249,796 individuals reveal 18 new loci associated with body mass index.Nat Genet. 2010; 42: 937-948Crossref PubMed Scopus (2206) Google Scholar Whether EWAS will be as successful for other clinical phenotypes—especially those manifest in more inaccessible tissues such as brain, or more directly affected by confounding factors such as cellular heterogeneity, environmental exposures, and drugs—remains to be seen.We declare that we have no competing interests. Success in the identification of genetic variants that affect complex human phenotypes, such as height, weight, and common diseases, is one of the major achievements in contemporary biomedical research. Insight into the functional complexity of the genome also draws attention to the probable role of non-sequence-based genomic variation in health and disease. Notably, substantial attention is focused on the role of epigenetic processes that might regulate gene expression via modifications to DNA, histone proteins, and chromatin in medical traits. Although the role of epigenetic mechanisms in some rare developmental syndromes and in cancer is well established, systematic examination of their contribution to common non-malignant disease phenotypes is only just beginning. New microarray-based and sequencing-based technologies allow economical, high-throughput profiling of epigenetic marks, with a primary focus on DNA methylation; the era of the epigenome-wide association study (EWAS) of large numbers of samples has begun. In The Lancet, Katherine Dick and colleagues describe the first systematic analysis of the association between variation in DNA methylation and body-mass index (BMI).1Dick KJ Nelson CP Tsaprouni L et al.DNA methylation and body-mass index: a genome-wide analysis.Lancet. 2014; (published online March 13.)http://dx.doi.org/10.1016/S0140-6736(13)62674-4PubMed Google Scholar They report significant associations between methylation at three probes targeting specific CpG sites within intron 1 of HIF3A and BMI in a discovery cohort, and subsequently confirm them in two independent cohorts. For every 10% increase in methylation of the most significant probe—cg22891070—BMI increased by 3·6% (95% CI 2·4–4·9), equating to about 0·98 kg/m2 for a person in the discovery cohort with a BMI of 27 kg/m2 on average.1Dick KJ Nelson CP Tsaprouni L et al.DNA methylation and body-mass index: a genome-wide analysis.Lancet. 2014; (published online March 13.)http://dx.doi.org/10.1016/S0140-6736(13)62674-4PubMed Google Scholar The increase in BMI was higher in individuals who had had a myocardial infarction (4·6%, 2·9–6·3) than in blood donors (2·3%, 0·4–4·1).1Dick KJ Nelson CP Tsaprouni L et al.DNA methylation and body-mass index: a genome-wide analysis.Lancet. 2014; (published online March 13.)http://dx.doi.org/10.1016/S0140-6736(13)62674-4PubMed Google Scholar To put the size of this epigenetic association into perspective, the minor allele of FTO—robustly associated with obesity-related traits—accounts for a more modest 0·39 kg/m2 increase in BMI.2Loos RJ Yeo GS The bigger picture of FTO—the first GWAS-identified obesity gene.Nat Rev Endocrinol. 2014; 10: 51-61Crossref PubMed Scopus (341) Google Scholar HIF3A encodes a component of the hypoxia inducible transcription factor that mediates the cellular response to hypoxia by regulating expression of many downstream genes.3Greer SN Metcalf JL Wang Y Ohh M The updated biology of hypoxia-inducible factor.EMBO J. 2012; 31: 2448-2460Crossref PubMed Scopus (412) Google Scholar This transcription factor has been previously implicated in metabolism4Shin MK Drager LF Yao Q et al.Metabolic consequences of high-fat diet are attenuated by suppression of HIF-1alpha.PLoS One. 2012; 7: e46562Crossref PubMed Scopus (48) Google Scholar and obesity,5Jiang C Qu A Matsubara T et al.Disruption of hypoxia-inducible factor 1 in adipocytes improves insulin sensitivity and decreases adiposity in high-fat diet-fed mice.Diabetes. 2011; 60: 2484-2495Crossref PubMed Scopus (209) Google Scholar providing a biologically plausible mechanism behind the reported association with BMI. Epigenetic epidemiology is an area of great research interest; in the past year, EWAS have been reported for several other human health phenotypes, such as multiple sclerosis,6Huynh JL Garg P Thin TH et al.Epigenome-wide differences in pathology-free regions of multiple sclerosis-affected brains.Nat Neurosci. 2013; 17: 121-130Crossref PubMed Scopus (181) Google Scholar rheumatoid arthritis,7Liu Y Aryee MJ Padyukov L et al.Epigenome-wide association data implicate DNA methylation as an intermediary of genetic risk in rheumatoid arthritis.Nat Biotechnol. 2013; 31: 142-147Crossref PubMed Scopus (673) Google Scholar pain sensitivity,8Bell JT Loomis AK Butcher LM et al.Differential methylation of the TRPA1 promoter in pain sensitivity.Nat Commun. 2014; 5: 2978Crossref PubMed Scopus (103) Google Scholar and metabolic traits.9Petersen AK Zeilinger S Kastenmuller G et al.Epigenetics meets metabolomics: an epigenome-wide association study with blood serum metabolic traits.Hum Mol Genet. 2013; 23: 534-545Crossref PubMed Scopus (140) Google Scholar Dick and colleagues1Dick KJ Nelson CP Tsaprouni L et al.DNA methylation and body-mass index: a genome-wide analysis.Lancet. 2014; (published online March 13.)http://dx.doi.org/10.1016/S0140-6736(13)62674-4PubMed Google Scholar used a powerful sequential-replication design to do one of the most systematic epigenetic studies of a human physiological phenotype yet reported. However, the results of any EWAS need to be interpreted carefully, with clear caveats when compared with genetic studies of complex traits. Unlike genetics, a range of potentially important confounding factors need to be considered, such as tissue or cell type, age, sex, drug exposure, and reverse causation.10Mill J Heijmans BT From promises to practical strategies in epigenetic epidemiology.Nat Rev Genet. 2013; 14: 585-594Crossref PubMed Scopus (248) Google Scholar A primary concern in epigenetic epidemiology is the tissue-specific nature of the epigenome. In large well-phenotyped sample cohorts, such as the discovery cohort used in Dick and colleagues' study,1Dick KJ Nelson CP Tsaprouni L et al.DNA methylation and body-mass index: a genome-wide analysis.Lancet. 2014; (published online March 13.)http://dx.doi.org/10.1016/S0140-6736(13)62674-4PubMed Google Scholar DNA from peripheral tissues (normally whole blood) is often the only source of biological material available. To circumvent this issue, Dick and colleagues subsequently examined the relation between DNA methylation at their top-ranked loci and BMI in adipose and skin tissue from an independent sample cohort, recording strong associations in adipose tissue but not skin.1Dick KJ Nelson CP Tsaprouni L et al.DNA methylation and body-mass index: a genome-wide analysis.Lancet. 2014; (published online March 13.)http://dx.doi.org/10.1016/S0140-6736(13)62674-4PubMed Google Scholar Furthermore, the investigators examined the correlation between DNA methylation and HIF3A expression in adipose tissue, reporting a significant inverse correlation and drawing attention to the potential functional relevance of epigenetic variation at the identified locus. This result is important, because it suggests that assessment of DNA methylation in whole blood can identify robust and biologically relevant epigenetic variation related to BMI. Of course, blood itself is a heterogeneous mix of epigenetically distinct cell types, with variation in cell counts between individuals a potentially huge confounder in EWAS analyses. Dick and colleagues included basic cell-count data that allowed some control for cellular heterogeneity;1Dick KJ Nelson CP Tsaprouni L et al.DNA methylation and body-mass index: a genome-wide analysis.Lancet. 2014; (published online March 13.)http://dx.doi.org/10.1016/S0140-6736(13)62674-4PubMed Google Scholar another approach is the use of robust algorithms to infer cellular composition from epigenomic data.11Houseman EA Molitor J Marsit CJ Reference-free cell mixture adjustments in analysis of DNA methylation data.Bioinformatics. 2014; (published online Feb 14.)https://doi.org/10.1093/bioinformatics/btu029Crossref PubMed Scopus (298) Google Scholar However, if the cellular content of a tissue is strongly associated with the trait being studied—as is likely for inflammatory or neurodegenerative disorders, for example—any apparent trait-associated epigenetic differences could partially reflect differences in cellular composition, even after statistical correction. Another important issue concerns the ability to distinguish between cause and effect. For example, the epigenetic variation reported by Dick and colleagues1Dick KJ Nelson CP Tsaprouni L et al.DNA methylation and body-mass index: a genome-wide analysis.Lancet. 2014; (published online March 13.)http://dx.doi.org/10.1016/S0140-6736(13)62674-4PubMed Google Scholar could have arisen before any alteration in BMI, contributing directly to obesity-related phenotypes. Alternatively, it could represent a secondary, downstream effect of variation in BMI itself or another exposure associated with variation in BMI. Therefore, the most robust design for epigenetic epidemiology involves the longitudinal assessment of epigenetic changes within the context of a prospective cohort study, so that epigenetic variation can be related to temporal changes in exposures and phenotype.10Mill J Heijmans BT From promises to practical strategies in epigenetic epidemiology.Nat Rev Genet. 2013; 14: 585-594Crossref PubMed Scopus (248) Google Scholar Dick and colleagues attempt to address the issue of causality by applying a mendelian randomisation approach12Relton CL Davey Smith G Two-step epigenetic Mendelian randomization: a strategy for establishing the causal role of epigenetic processes in pathways to disease.Int J Epidemiol. 2012; 41: 161-176Crossref PubMed Scopus (260) Google Scholar to interrogate the causal relation between HIF3A methylation and BMI. This approach uses a genetic proxy for DNA methylation (namely, methylation quantitative trait loci) to identify a causal relation between an exposure or trait and epigenetic variation, assuming that genetic associations are largely immune to residual confounding and reverse causation. Dick and colleagues identified two upstream single nucleotide polymorphisms that were independently associated with DNA methylation at a HIF3A locus in both the discovery and replication cohorts. However, these single nucleotide polymorphisms were not associated with BMI in the study cohorts or the high-powered GIANT consortium dataset,13Speliotes EK Willer CJ Berndt SI et al.Association analyses of 249,796 individuals reveal 18 new loci associated with body mass index.Nat Genet. 2010; 42: 937-948Crossref PubMed Scopus (2206) Google Scholar suggesting that hypermethylation at the HIF3A locus is likely to be a result of increased BMI rather than a causal association between increased methylation and BMI. A non-causal association between methylation and a phenotype could still be informative as a diagnostic or prognostic biomarker—eg, HIF3A methylation might predict disease phenotypes associated with BMI, such as cancer and cardiovascular disease. Dick and colleagues' study1Dick KJ Nelson CP Tsaprouni L et al.DNA methylation and body-mass index: a genome-wide analysis.Lancet. 2014; (published online March 13.)http://dx.doi.org/10.1016/S0140-6736(13)62674-4PubMed Google Scholar represents an important advance for both obesity-related research and the specialty of epigenetic epidemiology. BMI is a good phenotype for population-based epigenomic studies: it is an accurate measure that is routinely collected in most cohort studies. The widespread uptake of instruments such as the Illumina 450K HumanMethylation array means that large collaborative EWAS meta-analyses can be done, building on the success of similar approaches in genetics.13Speliotes EK Willer CJ Berndt SI et al.Association analyses of 249,796 individuals reveal 18 new loci associated with body mass index.Nat Genet. 2010; 42: 937-948Crossref PubMed Scopus (2206) Google Scholar Whether EWAS will be as successful for other clinical phenotypes—especially those manifest in more inaccessible tissues such as brain, or more directly affected by confounding factors such as cellular heterogeneity, environmental exposures, and drugs—remains to be seen. We declare that we have no competing interests. DNA methylation and body-mass index: a genome-wide analysisIncreased BMI in adults of European origin is associated with increased methylation at the HIF3A locus in blood cells and in adipose tissue. Our findings suggest that perturbation of hypoxia inducible transcription factor pathways could have an important role in the response to increased weight in people. Full-Text PDF Open Access