Prediction of biological age and evaluation of genome-wide dynamic methylomic changes throughout human aging

Abstract The use of DNA methylation signatures to predict chronological age and aging rate is of interest in many fields, including disease prevention and treatment, forensics, and anti-aging medicine. Although a large number of methylation markers are significantly associated with age, most age-prediction methods use a few markers selected based on either previously published studies or datasets containing methylation information. Here, we implemented reproducing kernel Hilbert spaces (RKHS) regression and a ridge regression model in a Bayesian framework that utilized phenotypic and methylation profiles simultaneously to predict chronological age. We used over 450,000 CpG sites from the whole blood of a large cohort of 4,409 human individuals with a range of 10-101 years of age. Models were fitted using adjusted and un-adjusted methylation measurements for cell heterogeneity. Un-adjusted methylation scores delivered a significantly higher prediction accuracy than adjusted methylation data, with a correlation between age and predicted age of 0.98 and a root-mean-square error (RMSE) of 3.54 years in un-adjusted data, and 0.90 (correlation) and 7.16 (RMSE) years in adjusted data. Reducing the number of predictors (CpG sites) through subset selection improved predictive power with a correlation of 0.98 and an RMSE of 2.98 years in the RKHS model. We found distinct global methylation patterns, with a significant increase in the proportion of methylated cytosines in CpG islands and a decreased proportion in other CpG types, including CpG shore, shelf, and open sea (p < 5e-06). Epigenetic drift seemed to be a widespread phenomenon as more than 97% of the age-associated methylation sites had heteroscedasticity. Apparent methylomic aging rate (AMAR) had a sex-specific pattern, with an increase in AMAR in females with age related to males.

Download Full-text

Genome-wide methylation prediction of biological age using reproducing kernel Hilbert spaces and Bayesian ridge regressions

10.1101/2020.08.25.266924 ◽

2020 ◽

Cited By ~ 2

Author(s):

Mahmoud Amiri Roudbar ◽

Mehdi Momen ◽

Seyedeh Fatemeh Mousavi ◽

Siavash Salek Ardestani ◽

Fernando Brito Lopes ◽

...

Keyword(s):

Hilbert Spaces ◽

Reproducing Kernel ◽

Cpg Islands ◽

Reproducing Kernel Hilbert Spaces ◽

Specific Pattern ◽

Chronological Age ◽

Global Methylation ◽

Aging Rate ◽

Kernel Hilbert Spaces ◽

Methylation Patterns

ABSTRACTThe use of DNA methylation signatures to predict chronological age and the aging rate is of interest in many fields, including disease prevention and treatment, forensics, and anti-aging medicine. Although a large number of methylation markers have been found to be significantly associated with age, most age-prediction methods use a small number of markers selected based on either previously published studies or datasets containing methylation information. Here, we implemented reproducing kernel Hilbert spaces (RKHS) regression and ridge regression model in a Bayesian framework that utilized phenotypic and methylation profiles simultaneously to predict chronological age. We used over 450,000 CpG sites from the whole blood of a large cohort of 4,409 human individuals with a range of 10-101 years of age. Models were fitted using adjusted and un-adjusted methylation measurements for cell heterogeneity. Non-adjusted methylation scores delivered a significantly higher prediction accuracy than adjusted methylation data, with a correlation between age and predicted age of 0.98 and a root-mean-square error (RMSE) of 3.54 years in non-adjusted data, 0.90 (correlation) and 7.16 (RMSE) years in adjusted data. Reducing the number of predictors through subset selection improved predictive power with a correlation of 0.98 and an RMSE of 2.98 years in the RKHS model. We found distinct global methylation patterns, with significant hypermethylation in CpG islands and hypomethylation in other CpG types including CpG shore, shelf, and open sea (p < 5e-06). Epigenetic drift seemed to be a widespread phenomenon as more than 97% of the age-associated methylation sites had heteroscedasticity. Apparent methylomic aging rate (AMAR) had a sex-specific pattern, with an increase in AMAR in females with age compared to males.

Download Full-text

The effect of age on DNA methylation in whole blood among Bangladeshi men and women

BMC Genomics ◽

10.1186/s12864-019-6039-9 ◽

2019 ◽

Vol 20 (1) ◽

Cited By ~ 1

Author(s):

Rick J. Jansen ◽

Lin Tong ◽

Maria Argos ◽

Farzana Jasmine ◽

Muhammad Rakibuz-Zaman ◽

...

Keyword(s):

Dna Methylation ◽

Whole Blood ◽

Cpg Islands ◽

Chronological Age ◽

Cpg Sites ◽

Age Related ◽

Genome Wide ◽

A Genome ◽

Effect Of Age ◽

Age Related Changes

Abstract Background It is well-known that methylation changes occur as humans age, however, understanding how age-related changes in DNA methylation vary by sex is lacking. In this study, we characterize the effect of age on DNA methylation in a sex-specific manner and determine if these effects vary by genomic context. We used the Illumina HumanMethylation 450 K array and DNA derived from whole blood for 400 adult participants (189 males and 211 females) from Bangladesh to identify age-associated CpG sites and regions and characterize the location of these age-associated sites with respect to CpG islands (vs. shore, shelf, or open sea) and gene regions (vs. intergenic). We conducted a genome-wide search for age-associated CpG sites (among 423,604 sites) using a reference-free approach to adjust for cell type composition (the R package RefFreeEWAS) and performed an independent replication analysis of age-associated CpGs. Results The number of age-associated CpGs (p < 5 x 10− 8) were 986 among men and 3479 among women of which 2027(63.8%) and 572 (64.1%) replicated (using Bonferroni adjusted p < 1.2 × 10− 5). For both sexes, age-associated CpG sites were more likely to be hyper-methylated with increasing age (compared to hypo-methylated) and were enriched in CpG islands and promoter regions compared with other locations and all CpGs on the array. Although we observed strong correlation between chronological age and previously-developed epigenetic age models (r ≈ 0.8), among our top (based on lowest p-value) age-associated CpG sites only 12 for males and 44 for females are included in these prediction models, and the median chronological age compared to predicted age was 44 vs. 51.7 in males and 45 vs. 52.1 in females. Conclusions Our results describe genome-wide features of age-related changes in DNA methylation. The observed associations between age and methylation were generally consistent for both sexes, although the associations tended to be stronger among women. Our population may have unique age-related methylation changes that are not captured in the established methylation-based age prediction model we used, which was developed to be non-tissue-specific.

Download Full-text

Trends in DNA methylation with age replicate across diverse human populations

10.1101/073171 ◽

2016 ◽

Author(s):

Shyamalika Gopalan ◽

Oana Carja ◽

Maud Fagny ◽

Etienne Patin ◽

Justin W. Myrick ◽

...

Keyword(s):

Dna Methylation ◽

Meta Analysis ◽

Chronological Age ◽

Human Populations ◽

Strongly Correlated ◽

Hunter Gatherers ◽

Cpg Sites ◽

Age Related ◽

Genome Wide ◽

Methylation Patterns

AbstractAging is associated with widespread changes in genome-wide patterns of DNA methylation. Thousands of CpG sites whose tissue-specific methylation levels are strongly correlated with chronological age have been previously identified. However, the majority of these studies have focused primarily on cosmopolitan populations living in the developed world; it is not known if age-related patterns of DNA methylation at these loci are similar across a broad range of human genetic and ecological diversity. We investigated genome-wide methylation patterns using saliva and whole blood derived DNA from two traditionally hunting and gathering African populations: the Baka of the western Central African rainforest and the ≠Khomani San of the South African Kalahari Desert. We identify hundreds of CpG sites whose methylation levels are significantly associated with age, thousands that are significant in a meta-analysis, and replicate trends previously reported in populations of non-African descent. We confirm that an age-associated site in the gene ELOVL2 shows a remarkably congruent relationship with aging in humans, despite extensive genetic and environmental variation across populations. We also demonstrate that genotype state at methylation quantitative trait loci (meQTLs) can affect methylation trends at some known age-associated CpG sites. Our study explores the relationship between CpG methylation and chronological age in populations of African hunter-gatherers, who rely on different diets across diverse ecologies. While many age-related CpG sites replicate across populations, we show that considering common genetic variation at meQTLs further improves our ability to detect previously identified age associations.

Download Full-text

Genome-Wide DNA Methylation Analysis Shows Enrichment of Differential Methylation in “Open Seas” and Enhancers and Reveals Hypomethylation in DNMT3A Mutated Cytogenetically Normal AML (CN-AML)

Blood ◽

10.1182/blood.v120.21.653.653 ◽

2012 ◽

Vol 120 (21) ◽

pp. 653-653 ◽

Cited By ~ 2

Author(s):

Ying Qu ◽

Andreas Lennartsson ◽

Verena I. Gaidzik ◽

Stefan Deneberg ◽

Sofia Bengtzén ◽

...

Keyword(s):

Gene Expression ◽

Dna Methylation ◽

Cpg Island ◽

Cpg Islands ◽

Differential Methylation ◽

Methylation Analysis ◽

Cpg Sites ◽

Genome Wide ◽

Genomic Regions ◽

Methylation Patterns

Abstract Abstract 653 DNA methylation is involved in multiple biologic processes including normal cell differentiation and tumorigenesis. In AML, methylation patterns have been shown to differ significantly from normal hematopoietic cells. Most studies of DNA methylation in AML have previously focused on CpG islands within the promoter of genes, representing only a very small proportion of the DNA methylome. In this study, we performed genome-wide methylation analysis of 62 AML patients with CN-AML and CD34 positive cells from healthy controls by Illumina HumanMethylation450K Array covering 450.000 CpG sites in CpG islands as well as genomic regions far from CpG islands. Differentially methylated CpG sites (DMS) between CN-AML and normal hematopoietic cells were calculated and the most significant enrichment of DMS was found in regions more than 4kb from CpG Islands, in the so called open sea where hypomethylation was the dominant form of aberrant methylation. In contrast, CpG islands were not enriched for DMS and DMS in CpG islands were dominated by hypermethylation. DMS successively further away from CpG islands in CpG island shores (up to 2kb from CpG Island) and shelves (from 2kb to 4kb from Island) showed increasing degree of hypomethylation in AML cells. Among regions defined by their relation to gene structures, CpG dinucleotide located in theoretic enhancers were found to be the most enriched for DMS (Chi χ2<0.0001) with the majority of DMS showing decreased methylation compared to CD34 normal controls. To address the relation to gene expression, GEP (gene expression profiling) by microarray was carried out on 32 of the CN-AML patients. Totally, 339723 CpG sites covering 18879 genes were addressed on both platforms. CpG methylation in CpG islands showed the most pronounced anti-correlation (spearman ρ =-0.4145) with gene expression level, followed by CpG island shores (mean spearman rho for both sides' shore ρ=-0.2350). As transcription factors (TFs) have shown to be crucial for AML development, we especially studied differential methylation of an unbiased selection of 1638 TFs. The most enriched differential methylation between CN-AML and normal CD34 positive cells were found in TFs known to be involved in hematopoiesis and with Wilms tumor protein-1 (WT1), activator protein 1 (AP-1) and runt-related transcription factor 1 (RUNX1) being the most differentially methylated TFs. The differential methylation in WT 1 and RUNX1 was located in intragenic regions which were confirmed by pyro-sequencing. AML cases were characterized with respect to mutations in FLT3, NPM1, IDH1, IDH2 and DNMT3A. Correlation analysis between genome wide methylation patterns and mutational status showed statistically significant hypomethylation of CpG Island (p<0.0001) and to a lesser extent CpG island shores (p<0.001) and the presence of DNMT3A mutations. This links DNMT3A mutations for the first time to a hypomethylated phenotype. Further analyses correlating methylation patterns to other clinical data such as clinical outcome are ongoing. In conclusion, our study revealed that non-CpG island regions and in particular enhancers are the most aberrantly methylated genomic regions in AML and that WT 1 and RUNX1 are the most differentially methylated TFs. Furthermore, our data suggests a hypomethylated phenotype in DNMT3A mutated AML. Disclosures: No relevant conflicts of interest to declare.

Download Full-text

Age-related clonal haemopoiesis is associated with increased epigenetic age

10.1101/600759 ◽

2019 ◽

Author(s):

Neil A. Robertson ◽

Ian J. Deary ◽

Kristina Kirschner ◽

Riccardo E. Marioni ◽

Tamir Chandra

Keyword(s):

Somatic Mutations ◽

Chronological Age ◽

Accelerated Ageing ◽

Healthy Individuals ◽

Main Risk Factor ◽

Cpg Sites ◽

Age Related ◽

Epigenetic Age ◽

Increased Risk ◽

Chronic Ischemic Heart Failure

Age-related clonal haemopoiesis (ARCH) in healthy individuals was initially observed through an increased skewing in X chromosome inactivation. More recently, several groups reported that ARCH is driven by somatic mutations. The most prevalent ARCH mutations are in the DNMT3A and TET2 genes, previously described as drivers of myeloid malignancies. ARCH is associated with an increased risk for haematological cancers. ARCH also confers an increased risk for non-haematological diseases such as cardiovascular disease, atherosclerosis, and chronic ischemic heart failure, for which age is a main risk factor. Whether ARCH is linked to accelerated ageing has remained unexplored. The most accurate and commonly-used tools to measure age acceleration are epigenetic clocks. They are based on age-related methylation differences at specific CpG sites correlating chronological age accurately with epigenetic age. Deviations from chronological age towards an increased epigenetic age have been associated with increased risk of earlier mortality and age-related morbidities. Here we present evidence of accelerated epigenetic age in individuals with ARCH.

Download Full-text

Do age-related changes in DNA methylation play a role in the development of age-related diseases?

Biochemical Society Transactions ◽

10.1042/bst20120358 ◽

2013 ◽

Vol 41 (3) ◽

pp. 803-807 ◽

Cited By ~ 26

Author(s):

Sanne D. van Otterdijk ◽

John C. Mathers ◽

Gordon Strathdee

Keyword(s):

Gene Expression ◽

Dna Methylation ◽

Cpg Islands ◽

Large Body ◽

Epigenetic Mechanism ◽

Aberrant Methylation ◽

Gene Promoters ◽

Cpg Sites ◽

Pathological Conditions ◽

Age Related

DNA methylation is an important epigenetic mechanism in mammalian cells. It occurs almost exclusively at CpG sites and has a key role in a number of biological processes. It plays an important part in regulating chromatin structure and has been best studied for its role in controlling gene expression. In particular, hypermethylation of gene promoters which have high levels of CpG sites, known as CpG islands, leads to gene inactivation. In healthy cells, however, it appears that only a small number of genes are controlled through promoter hypermethylation, such as genes on the inactivated X-chromosome or at imprinted loci, and most promoter-associated CpG islands remain methylation-free regardless of gene expression status. However, a large body of evidence has now shown that this protection from methylation not only breaks down in a number of pathological conditions (e.g. cancer), but also already occurs during the normal process of aging. The present review focuses on the methylation changes that occur during healthy aging and during disease development, and the potential links between them. We focus especially on the extent to which the acquisition of aberrant methylation changes during aging could underlie the development of a number of important age-related pathological conditions.

Download Full-text

Differences in DNA Methylation Patterns Between Vegans and Non-vegetarians in the AHS-2 Cohort (FS11-06-19)

Current Developments in Nutrition ◽

10.1093/cdn/nzz037.fs11-06-19 ◽

2019 ◽

Vol 3 (Supplement_1) ◽

Author(s):

Fayth Miles ◽

Andrew Mashchak ◽

Gary Fraser

Keyword(s):

Dna Methylation ◽

Cpg Islands ◽

White Blood Cells ◽

Small Sample ◽

Cpg Methylation ◽

Vegan Diet ◽

Gene Region ◽

Base Pairs ◽

Cpg Sites ◽

Methylation Patterns

Abstract Objectives It is unclear how diets differing in content of animal products affect DNA methylation patterns. We sought to determine if DNA methylation patterns differed between vegans and non-vegetarians in the Adventist Health Study-2 (AHS-2) cohort. Methods DNA methylation was profiled using the Infinium Human Methylation 450 K BeadChip in white blood cells of 137 participants in the AHS-2 cohort classified as vegan (57) or non-vegetarian (80) based on validated diet history data. Linear regression models were generated to test associations between diet pattern and methylation, where the response variable represented methylation intensity for individual CpG sites. This was repeated for sites separated into gene regions or in relation to CpG islands. CpG methylation was also averaged across each gene. The permutation-identified null distribution, false discovery rate of Storey et al was used to adjust for multiple testing. Results A total of 53,809 individual CpG sites were estimated to be differentially methylated (non-null). Of these, with this small sample, we could individually identify only up to 5.5% (differing by gene region) at FDR <0.05. A vegan diet was associated almost exclusively with hypomethylation of individual CpG sites. Significant CpG sites numbered 2504 within CpG islands, 749 and 13 CpG sites 0–200 base pairs or 201–1500 base pairs upstream of the transcription start site (TSS200 and TSS1500), respectively, and 51 and 458 sites falling within the 5’ UTR and first exon, respectively. The greatest difference in methylation was observed for sites mapping to TSS200 and CpG islands, where methylation was 9.2% lower in vegans. No differences were found when CpG methylation was averaged to obtain a cumulative value for all sites (no FDR criterion) within each gene region. A total of 328 genes (averaging all methylation regions) were significantly hypomethylated in vegans relative to non-vegetarians at FDR <0.05. The greatest number of genes were hypomethylated when considering methylation of only CpG islands, followed by the TSS200, first exon, and TSS1500 regions. Conclusions Our findings suggest substantial differences in methylation of CpG sites and genes, particularly in regulatory regions, between vegans and non-vegetarians, with a preponderance of hypomethylation among vegans. Funding Sources National Institute of Health, National Cancer Institute, World Cancer Research Fund (UK).

Download Full-text

Characterising sex differences of autosomal DNA methylation in whole blood using the Illumina EPIC array

10.1101/2021.09.02.458717 ◽

2021 ◽

Author(s):

Olivia A Grant ◽

Yucheng Wang ◽

Meena Kumari ◽

Nicolae Radu Zabet ◽

Leonard C Schalkwyk

Keyword(s):

Dna Methylation ◽

Sex Differences ◽

Whole Blood ◽

Cpg Islands ◽

Disease Etiology ◽

Cpg Sites ◽

Males And Females ◽

Methylation Patterns ◽

First Time ◽

Epic Array

Sex differences are known to play a role in disease etiology, progression and outcome. Previous studies have revealed autosomal epigenetic differences between males and females in some tissues, including differences in DNA methylation patterns. Here, we report for the first time an analysis of autosomal sex differences in DNAme using the Illumina EPIC array in human whole blood (n=1171). We identified 554 sex-associated differentially methylated CpG sites (saDMPs) with the majority found to be hypermethylated in females (70%). These saDMP's are enriched in CpG islands and CpG shores and located preferentially at 5'UTRs, 3'UTRs and enhancers. Additionally, we identified 311 significant sex associated differentially methylated regions (saDMRs). Transcription factor binding site enrichment revealed enrichment of transcription factors related to critical developmental processes and sex determination such as SRY and SOX9. Our study reports a reliable catalogue of sex associated CpG sites and elucidates several characteristics of these sites.

Download Full-text

Development of Epigenetic Clocks for Key Ruminant Species

Genes ◽

10.3390/genes13010096 ◽

2021 ◽

Vol 13 (1) ◽

pp. 96

Author(s):

Alex Caulton ◽

Ken G. Dodds ◽

Kathryn M. McRae ◽

Christine Couldrey ◽

Steve Horvath ◽

...

Keyword(s):

Large Body ◽

Ovis Aries ◽

Farm Animal ◽

Tissue Type ◽

Capra Hircus ◽

Chronological Age ◽

Individual Species ◽

Epigenetic Clock ◽

Cpg Sites ◽

Age Related

Robust biomarkers of chronological age have been developed in humans and model mammalian species such as rats and mice using DNA methylation data. The concept of these so-called “epigenetic clocks” has emerged from a large body of literature describing the relationship between genome-wide methylation levels and age. Epigenetic clocks exploit this phenomenon and use small panels of differentially methylated cytosine (CpG) sites to make robust predictions of chronological age, independent of tissue type. Here, we present highly accurate livestock epigenetic clocks for which we have used the custom mammalian methylation array “HorvathMammalMethyl40” to construct the first epigenetic clock for domesticated goat (Capra hircus), cattle (Bos taurus), Red (Cervus elaphus) and Wapiti deer (Cervus canadensis) and composite-breed sheep (Ovis aries). Additionally, we have constructed a ‘farm animal clock’ for all species included in the study, which will allow for robust predictions to be extended to various breeds/strains. The farm animal clock shows similarly high accuracy to the individual species’ clocks (r > 0.97), utilizing only 217 CpG sites to estimate age (relative to the maximum lifespan of the species) with a single mathematical model. We hypothesise that the applications of this livestock clock could extend well beyond the scope of chronological age estimates. Many independent studies have demonstrated that a deviation between true age and clock derived molecular age is indicative of past and/or present health (including stress) status. There is, therefore, untapped potential to utilize livestock clocks in breeding programs as a predictor for age-related production traits.

Download Full-text

Examination of the dynamics of global DNA methylation pattern establishment during spermatogenesiss

Clinical & Investigative Medicine ◽

10.25011/cim.v30i4.2868 ◽

2007 ◽

Vol 30 (4) ◽

pp. 90

Author(s):

Kirsten Niles ◽

Sophie La Salle ◽

Christopher Oakes ◽

Jacquetta Trasler

Keyword(s):

Dna Methylation ◽

Germ Cell ◽

Germ Cells ◽

Epigenetic Modification ◽

Methylation Pattern ◽

Chromosome 9 ◽

Specific Gene ◽

Global Methylation ◽

Dna Methylation Pattern ◽

Methylation Patterns

Background: DNA methylation is an epigenetic modification involved in gene expression, genome stability, and genomic imprinting. In the male, methylation patterns are initially erased in primordial germ cells (PGCs) as they enter the gonadal ridge; methylation patterns are then acquired on CpG dinucleotides during gametogenesis. Correct pattern establishment is essential for normal spermatogenesis. To date, the characterization and timing of methylation pattern acquisition in PGCs has been described using a limited number of specific gene loci. This study aimed to describe DNA methylation pattern establishment dynamics during male gametogenesis through global methylation profiling techniques in a mouse model. Methods: Using a chromosome based approach, primers were designed for 24 regions spanning chromosome 9; intergenic, non-repeat, non-CpG island sequences were chosen for study based on previous evidence that these types of sequences are targets for testis-specific methylation events. The percent methylation was determined in each region by quantitative analysis of DNA methylation using real-time PCR (qAMP). The germ cell-specific pattern was determined by comparing methylation between spermatozoa and liver. To examine methylation in developing germ cells, spermatogonia from 2 day- and 6 day-old Oct4-GFP (green fluorescent protein) mice were isolated using fluorescence activated cell sorting. Results: As compared to liver, four loci were hypomethylated and five loci were hypermethylated in spermatozoa, supporting previous results indicating a unique methylation pattern in male germ cells. Only one region was hypomethylated and no regions were hypermethylated in day 6 spermatogonia as compared to mature spermatozoa, signifying that the bulk of DNA methylation is established prior to type A spermatogonia. The methylation in day 2 spermatogonia, germ cells that are just commencing mitosis, revealed differences of 15-20% compared to day 6 spermatogonia at five regions indicating that the most crucial phase of DNA methylation acquisition occurs prenatally. Conclusion: Together, these studies provide further evidence that germ cell methylation patterns differ from those in somatic tissues and suggest that much of methylation at intergenic sites is acquired during prenatal germ cell development. (Supported by CIHR)

Download Full-text