Research Article

Epigenetic signature of human immune aging in the GESTALT study

Laboratory of Molecular Biology and Immunology, National Institute on Aging, United States
Translational Gerontology Branch, National Institute on Aging, United States
Laboratory of Neurogenetics, National Institute on Aging, United States
Clinical Research Core, National Institute on Aging, United States
Flow Cytometry Unit, National Institute on Aging, United States
Laboratory of Genetics and Genomics, National Institute on Aging, United States
Laboratory of Clinical Investigation, National Institute on Aging, United States

Aug 17, 2023

https://doi.org/10.7554/eLife.86136

Open access
Copyright information

Version of Record

Accepted for publication after peer review and revision.

Download
Cite
Share
CommentOpen annotations (there are currently 0 annotations on this page).

Version of Record published: September 18, 2023 (This version)
Accepted Manuscript published: August 17, 2023 (Go to version)
Accepted: August 16, 2023
Preprint posted: January 23, 2023 (Go to version)
Received: January 12, 2023

1. Of interest
Copy number variation and population-specific immune genes in the model vertebrate zebrafish

Yannick Schäfer, Katja Palitzsch ... Jaanus Suurväli

Research Article Jun 4, 2024
Further reading

Abstract
Editor's evaluation
Introduction
Results
Discussion
Materials and methods
Data availability
References
Article and author information
Metrics

Abstract

Age-associated DNA methylation in blood cells convey information on health status. However, the mechanisms that drive these changes in circulating cells and their relationships to gene regulation are unknown. We identified age-associated DNA methylation sites in six purified blood-borne immune cell types (naive B, naive CD4⁺ and CD8⁺ T cells, granulocytes, monocytes, and NK cells) collected from healthy individuals interspersed over a wide age range. Of the thousands of age-associated sites, only 350 sites were differentially methylated in the same direction in all cell types and validated in an independent longitudinal cohort. Genes close to age-associated hypomethylated sites were enriched for collagen biosynthesis and complement cascade pathways, while genes close to hypermethylated sites mapped to neuronal pathways. In silico analyses showed that in most cell types, the age-associated hypo- and hypermethylated sites were enriched for ARNT (HIF1β) and REST transcription factor (TF) motifs, respectively, which are both master regulators of hypoxia response. To conclude, despite spatial heterogeneity, there is a commonality in the putative regulatory role with respect to TF motifs and histone modifications at and around these sites. These features suggest that DNA methylation changes in healthy aging may be adaptive responses to fluctuations of oxygen availability.

Editor's evaluation

This fundamental work advances our understanding of chromatin changes that may be associated with aging across six distinct immune cell types. It highlights a non-uniform process of expression of aging signatures while a core signature is preserved across different cell types. The research employs solid validated and robust analysis methodologies. The findings would be of interest to researchers studying DNA methylation clock and aging biology.

https://doi.org/10.7554/eLife.86136.sa0

Introduction

Human aging is associated with site-specific changes of DNA methylation. Summary measures of DNA methylation called ‘epigenetic clocks’ are extensively used in aging research to estimate biological age (Horvath and Raj, 2018; Hannum et al., 2013; Bocklandt et al., 2011). Epigenetic clocks closely approximate chronological age and beyond age, predict adverse health conditions, including frailty (Gale et al., 2018), Alzheimer’s disease (McCartney et al., 2018), and mortality (Marioni et al., 2015; Chen et al., 2016).

Research suggest that changes in DNA methylation with aging are regulated by specific mechanisms rather than by a stochastic drift (Teschendorff et al., 2013). For example, a loss-of-function mutation in the H3K36 histone methyltransferase has been associated with epigenetic aging in mice (Martin-Herranz et al., 2019). In humans, polymorphisms in the telomerase gene (TERT) (Lu et al., 2018) and age-dependent gain of methylation in the Polycomb repressive complex 2 have been related to accelerated aging (Teschendorff et al., 2010). However, so far, no sound hypothesis exists that explains the association of DNA methylation with aging and pathology.

A main obstacle in understanding mechanisms driving age-associated changes of DNA methylation is that most human studies were performed in mixed blood cell types. The few studies that investigated select immune circulating cells failed to propose a unifying biological hypothesis explaining predictable changes of DNA methylation with aging (Dozmorov et al., 2017; Reynolds et al., 2014; Tserel et al., 2015; Bell et al., 2012; Kananen et al., 2016; Acevedo et al., 2015; Marttila et al., 2015).

We analyzed age-associated methylation in six purified blood-borne cell types sorted from peripheral blood mononuclear cells (PBMCs) from 55 donors of ages ranging from 22 to 83 years. To minimize the confounding of age-associated pre-clinical and clinical diseases, participants were ascertained to be healthy by trained health professionals according to strict clinical criteria. We looked for CpGs differentially methylated with aging in the same direction in multiple cell types. Next, in each cell type, we conducted enrichment analyses of genes close to age-associated CpGs. Finally, we looked for chromatin accessibility markers and transcription factor (TF)-binding sites close to the same age-associated CpGs. Our findings suggest that changes in methylation with aging are related to fluctuation of energetic metabolism during the life course.

Results

Age-associated methylation in individual cell types

A principal component analysis (PCA) was performed on normalized DNA methylation data for all cell types from all the 55 donors (Figure 1A and Supplementary file 1). The PCA showed that clustering by cell types was stronger than by age (Figure 1—figure supplement 1A). The genes associated with the top 500 probes corresponding to PC1, PC2, and PC3 were enriched pathways linked to innate and adaptive lineage development (Supplementary file 2).

Figure 1 with 1 supplement see all

Download asset Open asset

Study design and identification of age-associated methylation probes.

(A) Study design. (B) Age-associated CpG methylation (False Discovery Rate or FDR p < 0.05) in six cell types. (**C, D**) SuperExactTest circular plots to show the number of age-associated hypo- and hypermethylated probes shared among different combinations of cell types (indicated by green boxes), respectively. The outermost bars show the number of probes shared among each cell-type combination (regardless of other cell types). For example, probes hypomethylated with age in B + CD4 + CD8 + gran + mono (n = 222) includes probes also hypomethylated in NK cells (n = 181) and probes not hypomethylated with age in NK cells (n = 41). Based on the exact probability distributions of multi-set intersections, all the overlaps shown are highly statistically significant (p < 10⁻¹⁰⁰). (E) Graphical representation of age-associated hypomethylation in promoter region of RCAN1 in all six cell types. (F) Graphical representation of age-associated hypermethylation in promoter region of KLF14. The methylation status in peripheral blood mononuclear cell (PBMC) and buffy coat are also shown. Missing methylation data are represented in white.

Age-associated CpGs were identified through sex-adjusted beta-regression models (FDR corrected p-value <0.05). The number of hypo- or hypermethylated sites varied considerably between cell types (Figure 1B) with highest numbers in CD4⁺ T cells (Figure 1—figure supplement 1B and Supplementary files 3 and 4). Using a different approach of comparing between young (≤35 years, 25th percentile) and old (≥70 years, 75th percentile) individuals, we observed >90% overlap with beta-regression-derived hypomethylated sites and 70–95% overlap with hypermethylated sites in all cell types except CD8⁺ T cells (9–14% overlap) (Figure 1—figure supplement 1C). Having fewer old donors with CD8⁺ T cells may have contributed to differences (Supplementary file 1).

Like other studies, we found that a significant proportion of age-hypomethylated CpGs were in the intergenic and open sea (>4 kb from CpG island) regions while age-hypermethylated CpGs were in promoters and CpG islands (Chi sq test p < 0.001) (Figure 1—figure supplement 1D, E). Additionally, age-associated differentially methylated sites in PBMC poorly recapitulate age-dependent changes that take place in specific primary immune cells (Figure 1—figure supplement 1E, F). These findings point to a wide heterogeneity of age-differential CpG methylation across immune blood cells and suggest that studies in PBMC poorly represents the changes that take place in specific cell types with aging.

Shared age-associated methylation across cell types

Only 181 age-associated hypomethylated sites and 169 hypermethylated sites were shared between all 6 cell types. These numbers increased to 776 (age-hypomethylated) and 404 (age-hypermethylated) sites in 5 or more cell types (Figure 1C, D). Thus, most age-related methylation changes are cell specific. Of note, only 10 of the sites overlap with the 359 CpGs in Horvath’s pan-tissue epigenetic clock (Horvath, 2013). Several reasons can be attributed to this poor overlap including (1) use of methylation array with about 21,369 CpGs for development of the clock in contrast to the analyses in this study based on ~850,000 CpG sites; (2) use of data from peripheral or whole blood for development these clocks in contrast to data from flow-sorted circulating immune cells in this study. While the number of shared age-hypo- or hypermethylated CpGs across cells was relatively small, it was significantly much higher than that expected based on chance alone, suggesting that common underlying epigenetic mechanisms exist across the considered cell types (Figure 1C, D). For example, CpG sites adjacent to RCAN1 (calcineurin 1) and KLF14 (Krϋppel-Like Factor 14) show similar age-associated patterns in all cell types (Figure 1E, F).

Next, we wanted to investigate whether the top age-associated genes are the ones which are shared across cell types. For this we arranged the age-associated probes with decreasing order of adjusted p-value and looked at the annotated genes to identify the top 15 genes in each cell type (Figure 2A, B, Figure 2—figure supplement 1A–E, and Supplementary file 5). THSD4 and CCDC102B were the most significant age-associated hypomethylated genes shared by five or more cell types, while ELOVL2, KLF14, LHFP14, and GPR158 were among the most significant age-hypermethylated genes in five or more cell types. This count increased to 5 and 13 genes, respectively, when the list was expanded to 50 top genes (Supplementary file 5). It is noteworthy that only 13–15% of these ‘top’ age-associated probes overlapped with the list of age-associated probes shared across cell types (181 hypomethylated and 169 hypermethylated probes). These findings suggest that most CpGs with age-associated methylation consistent across cell types undergo moderate (although significant) methylation changes with aging.

Figure 2 with 2 supplements see all

Download asset Open asset

Characteristics of age-associated probes.

(**A, B**) Manhattan plot of age-associated hypo- and hypermethylated CpG sites in B cells, respectively. Most significant genic probes (−log p_adj >10) are labeled. (C) Correlation between beta-regression coefficients of age-differentially methylated CPGs in GESTALT and longitudinal InCHIANTI study. X-axis – InCHIANTI, Y-axis – B cell (C) and CD4⁺ T cell coefficients (D). Blue dots – age-hypomethylated CpGs, yellow triangles – age-hypermethylated CpGs. (**E, F**) Scatter plot of age-associated CpGs showing opposite trends in different immune cells. (E) cg27123256 (in BCL11B promoter) is hypomethylated with older age in B, monocytes, and NK while is hypermethylated with older age in CD4⁺ T cells. (F) cg03530364 (in FAM19A1 promoter) is hypermethylated with older age in B, granulocytes, monocytes, and NK cells while it is hypomethylated with older age in CD4⁺ T cells.

Longitudinal validation of age-associated CpG sites

We hypothesized that the age-associated CpGs identified across the six immune cells in this cross-sectional study would also show longitudinal changes of the size and direction predicted. We used DNA methylation data (Illumina 450K microarray on DNA from buffy coats) assessed at baseline and 9- and 13-year follow-up in 699 participants of the InCHIANTI study (Ferrucci et al., 2000). Of the 181 hypomethylated and 169 hypermethylated CpGs with age in all cell types in GESTALT, 72 and 135, respectively, were represented in the 450K microarray (Moore et al., 2016). The beta-coefficients for age of the 207 CpG probes (72 + 135) estimated from the GESTALT study and their corresponding values estimated longitudinally from the InCHIANTI study were highly and significantly correlated (hypomethylated with age CpGs: r = 0.49, p = 1.2e−09 and hypermethylated with age CpGs: r = 0.5, p = 6.9e−06 for average beta coefficients across six cell types, Figure 2C, D and Figure 2—figure supplement 2). Thus, CpGs identified as differentially methylated with aging across cell types in GESTALT also change longitudinally with aging.

Age-associated probes with opposite trends in different immune cells

Several CpGs showed significant but opposing age trends in different cell types, especially in B, CD4⁺ T cells, and monocytes (Figure 2—figure supplement 1F, G). For example, cg27123256 in the gene body of BCL11B was age hypomethylated in non-T cells and significantly age hypermethylated in naive CD4⁺ T cells (Figure 2E). Our observations implicate BCL11B in aging-related changes in naive CD4⁺ T cell function, distinct from its proposed role in effector cells (Tserel et al., 2015; Gray et al., 2014; Yui and Rothenberg, 2014). Conversely, cg03530364 in the body of FAM19A1 gene was hypermethylated in non-T cells but age-hypomethylated in CD4⁺ T cells (Figure 2F). Of note, none of these CpGs were differentially age-methylated in PBMC. Thus, opposite age-methylation trends in specific cell types may cancel each other and obscure their relevance for aging when mixed cell-type samples are assessed.

Pathway analysis of age-associated genes

Gene set enrichment analyses were performed on genes associated with at least one CpG significantly age-hypo- or hypermethylated in five or more cell types. We identified 30 pathways (q-value <0.05) (Figure 3 and Supplementary file 6). Probes commonly age-hypomethylated in five or more cell types (n = 776) pointed to genes enriched in collagen biosynthesis, complement cascade, and GTPase pathways (left-most column in bottom panel of Figure 3) that highlighted inflammatory and metabolic pathway in aging. Genes associated with shared age-hypermethylated probes (n = 404) were enriched for neural pathways previously implicated to brain aging along with G-protein-coupled receptors pathways (de Oliveira et al., 2019; Ewing et al., 2019) (left-most column in top panel of Figure 3). A recent study by Karagiannis et al. also identified neuronal genes in their PBMC aging data emphasizing a possible interlink between immune-aging and neuronal pathways (Karagiannis et al., 2023). Other key pathways are highlighted, with associated genes displayed in boxes on the right-hand side.

Figure 3

Download asset Open asset

Pathway analysis of methylated probes.

Enrichment analysis of genes annotated to age-associated hypo- and hypermethylated CpGs in ≥5 cell types (left-most column) and in individual cell types. Red/green shades indicate enrichment scores in hyper- (red) and hypo- (green) methylated genes. Yellow indicates ambiguous pathways associated with both hypo- and hypermethylated genes in individual cell types. Not significant pathways are shown in gray. Full results in Supplementary file 6.

Functional annotation of age-associated probes

To further interrogate the relationships between DNA methylation and other epigenetic states, we mapped the methylation age-associated sites to cell-specific chromHMM-derived chromatin profiles (Ernst and Kellis, 2012). As controls, we annotated all sites in the EPIC array to the 18-state chromHMM model of respective primary cell type. Granulocytes were excluded from this analysis because reference data were not available.

Age-associated hypomethylated CpGs were significantly enriched for weak/active enhancers (yellow bar, Figure 4A) whereas, confirming previous reports, age-hypermethylated CpGs, were enriched in bivalent/polycomb regions compared to control set (brown and dark gray bars, respectively, in Figure 4A). Results for cell-type-specific analyses are shown in Figure 4B.

Figure 4 with 1 supplement see all

Download asset Open asset

Functional annotation of age-associated probes along with their grouping based on sharedness.

(A) ChromHMM annotation of age-associated CpGs. (B) Proportion of CpGs mapping to weak/active enhancers (left, orange box), bivalent enhancers/TSS (inset, brown box) and polycomb repressor regions (right, gray box) in age-associated hypo- (blue line), hypermethylated (red line) CpGs as compared to all MethylationEPIC CpGs (gray line). (C) DeepTools plots showing the distribution of accessible chromatin (DNase hypersensitive sites) and H3K4me1 histone mark in and around ±3 kb region of age-differentially methylated CpGs. The age-associated sites were divided into shared (blue) (common between five or more immune cells) and selective sites (green). The top row shows the pattern for age-associated hypomethylated CpGs while the bottom row is for the age-associated hypermethylated CpGs in B and CD4⁺ T cells.

We further mapped the profile of four epigenetic markers from the ENCODE project in and around (±3 kb) age-associated methylation sites. For B and CD4⁺ T cells, we observed a V-shaped peak-valley-peak pattern of DNase hypersensitivity at sites of age-associated hypomethylation, which is characteristic of promoter sites (Figure 4C; Pundhir et al., 2016). Both age-associated hypo- and hypermethylated sites showed evident H3K4me1 peaks, a marker commonly associated with active and primed enhancers (Figure 4C; Bae and Lesch, 2020). No specific trend was observed for H3K4me3 and H3K27ac (Figure 4—figure supplement 1). These patterns were highly consistent across cell types (Figure 4—figure supplement 1) and strongly suggest a functional connection between methylation and chromatin status. However, as the DHS and histone data in the ENCODE database were only available for either one of two donors (a 21-year-old male and 37-year-old female), we could not verify whether the patterns observed are stable with change in age.

Pattern of TF-binding motifs around age-associated CpGs

Specific TFs binding may induce loss of DNA methylation or bind DNA that is methylated (Medvedeva et al., 2014; Moore et al., 2013). Through our de novo HOMER analysis, we looked for TF-binding motifs in a 200-bp window around the age-associated methylated sites in each cell type. We observed that the binding motif for aryl hydrocarbon receptor nuclear translocator (ARNT, also named HIF1β) was associated with age-hypomethylated CpGs across most cell types (Figure 5A). The only exception was naive CD8⁺ T cells where the top enriched motif was B-cell lymphoma gene 6 (BCL6). BCL6 code for a zinc finger TF that plays a critical role in the generation of memory and effector cells in acute infection (Kim et al., 2020). Another motif associated with age-hypomethylated CpGs across most cell types was chromatin architectural protein CTCF and its closely related gene BORIS. Methylation changes at CTCF sites have been reported to reflect large-scale genome reorganization in immune cells in older individuals (van Ruiten and Rowland, 2021; Bhat et al., 2021).

Figure 5 with 1 supplement see all

Download asset Open asset

Association of transcription factor (TF)-binding motifs with age-differentially methylated CpGs.

(A) Top 5 TF motifs at and around (±200 bp) of CpG sites that are hypomethylated with age. All the age-hypomethylated sites were considered for the analysis in each cell type. Recurring motifs like ARNT and CTCF/BORIS are highlighted. (B) Top 5 TF motifs at and around (±200 bp) CpG sites that are hypermethylated with age. All the age-hypermethylated sites were considered for the analysis in each cell type. Recurring motifs like REST and Sp100 are highlighted. (C) Hypoxia-centric model of age-associated sites with ARNT and REST motifs. CpG sites hypomethylated with aging across six different cell types are significantly more likely to host-binding motifs for ARNT, the core hub for the hypoxia response. On the contrary, CpG sites hypermethylated with aging are significantly more likely to host-binding motifs for REST, a hypoxia-response transcriptional repressor. On the right are selected age-associated genes that carry the motifs for ARNT or REST TFs.

Repressor Element 1-Silencing Transcription Factor (REST) was the TF motifs most frequently associated with age-hypermethylated CpGs in five of six cell types (Figure 5B). Age-hypermethylated sites in PBMCs have been previously shown to be enriched for REST, which is known to repress stress response genes and is lost in cognitive impairment and Alzheimer’s disease pathology (Yuan et al., 2015; Lu et al., 2014). The top enriched TF motif associated with age-hypermethylated sites in monocytes was Arid5A (p < 10⁻²⁷) that binds to selective inflammation-related genes, such as IL6 and STAT3 and stabilize their expression (Nyati et al., 2020; Wilsker et al., 2002). We further repeated the analysis with a smaller 50 bp window size for TF motif search. Motifs for ARNT, CTCF, and REST remained the top hits in most cell types (Supplementary file 7). However, BCL6 and ARID5A were no longer the top motifs in the search indicating that motifs for these TFs appear to be farther from the age-associated CpG sites.

The recurring enrichment of ARNT and REST with age-associated CpGs observed across multiple cell types, despite relatively few shared genomic region locations, suggests a common mechanism of gene regulation. We found that only 17 and 44 age-associated hypo- and hypermethylated probes, respectively, shared ARNT or REST motifs across all cells (Figure 5—figure supplement 1A, B), suggesting these overlaps are not random and have a specific function (Figure 5—figure supplement 1A, B).

Remarkably, ARNT mRNA was significantly overexpressed in older age in three of the six cell types and REST mRNA showed a significant decrease of expression with age in most cell types (Supplementary files 8 and 9). These findings suggest that age-associated changes in expression levels of REST and ARNT can affect the epigenetic status of their target genes.

Age-related differential methylation and oxygen sensing

ARNT, REST, and BCL6, three TFs most associated with differentially methylated regions, are implicated in hypoxia response (Figure 5C). ARNT is the beta subunit of Hypoxia Factor 1 (HIF-1), which is stabilized during hypoxia and shuttled to the nucleus where it binds to DNA hypoxia-response elements and triggers a complex response that include upregulation of angiogenesis and erythropoiesis and reprogramming of energetic metabolism from oxidative phosphorylation to anaerobic glycolysis (Semenza, 2000). Hypoxia also upregulates the transcription of REST which is the master regulator of the transcriptional repression arm of the response to hypoxia. Released REST is shuttled to the nucleus where it binds to DNA and regulates approximately 20% of the hypoxia-repressed genes, including genes involved in proliferation, translation, and cell cycle progression. We identified 35 genes that were hypomethylated with aging and had close by an ARNT motif in all six cell types (Supplementary file 10). Ten of these genes (right side of Figure 5C, genes under orange headings) have been linked to hypoxia response (Craps et al., 2021; Stegmann, 1998; Chakraborty and Ain, 2017; Gusdon et al., 2012; Wang et al., 2009; Hsu et al., 2010; Pamenter et al., 2020; Pangou et al., 2016; Lazarou et al., 2013; Cai et al., 2020). Similarly, we found 26 genes with probes hypermethylated with age and with REST motif in the vicinity in all six cell types (Supplementary file 10). Four of these (right side of Figure 5C, genes under green heading) are known to be downregulated in hypoxia (Tan et al., 2012; Liu et al., 2020; Dasgupta, 2004; Carmeliet and Jain, 2011). These results strongly suggest a link between age-associated DNA methylation and oxygen sensing through putative regulation by TFs like ARNT and REST in the various immune cells.

Association with inflammatory cytokines

Low-grade inflammation has been reported to be part of healthy aging. In order to investigate whether age-related pro-inflammatory state may explain the age-related changes in methylation observed in this study, we analyzed the SomaScan protein data of seven pro-inflammatory cytokines (IL6, IL1RN, IL1A, IL1B, TNF, TNFRSF1A, and TNFRSF1B) for the same cohort of donors from Tanaka et al., 2018. For each cell type, all CpGs reported as significantly hypo- or hypermethylated with age (in beta-regression analyses adjusted for sex) were reanalyzed by incorporating data on seven pro-inflammatory cytokines (see Materials and methods for details). Briefly, by comparing a model with a cytokine as explanatory variable (CpG ~ age + sex + cytokine) with another model without it (CpG ~ age + sex), we explored the robustness of age as an explanatory variable of methylation change, as well as possible mediating effects arising from pro-inflammatory cytokines. Detailed results are provided in OSF and summary statistics for each cell type, hypo-/hypermethylation association and pro-inflammatory cytokine are provided in Supplementary file 11. By comparing the results from the two abovementioned regression models, we observed in case of hypermethylated sites in CD4 cells, the number of CpGs dropped by 10% on adding TNFRSF1A to the model, a cytokine that appears significantly associated with 7507 of those ge-associated CpGs (Column I). In addition, TNFRSF1B appears significantly associated with 9058 of the age-hypermethylated CpG sites in CD4 cells. For other cell types like B naive and monocytes, TNF-alpha was associated with 65–124 age-associated CpG sites, respectively. Fewer associations are observed for the remaining analytes. These results suggest a possible link between TNF-alpha signaling pathway, aging, and DNA methylation change in circulating immune cells.

Discussion

Novel and important conclusions arise from our observations. First, only few CpG sites are hypo- and hypermethylated with aging across all circulating cells while majority of the significant age-associated methylation changes are cell selective. Indeed, several CpGs show differential age methylation in opposite directions in different cell types and are unchanged in PBMC, suggesting that they may be missed when studying mixed cell samples. Noteworthy, age-related methylation differences in this cross-sectional study were strongly and significantly correlated with longitudinal age-associated methylation changes in an independent population.

Second, age-associated hypomethylated sites were significantly enriched for active enhancers whereas age-hypermethylated sites were enriched for bivalent/polycomb regions, confirming previous findings in whole blood (Yuan et al., 2015). Age-differential methylation coincided with specific chromatin status and histone markers patterns, suggesting that their position in proximity of promoter and active enhancer regions is connected with chromatic accessibility and potentially modulation of gene expression. Since the ENCODE data were only from two donors, it will be worthwhile to see how the histone or chromatin accessibility patterns change with age at and around these age-associated CpG sites.

Third, distinct TF-binding motifs co-localize with CpGs differentially methylated with aging despite wide variation in the distribution of such sites across cell types, suggesting a specific regulatory function. Noteworthy, the top age-associated TF identified, ARNT and REST act in coordination in hypoxia response (Cavadas et al., 2016). BCL6, another top TF-binding motif associated with age-differentially methylated CpG has also been shown to protects cardiomyocyte from damage during hypoxia (Gu et al., 2019). These findings support the hypothesis that systematic methylation changes with aging may be induced by fluctuations in oxygen availability and energy metabolism. Interestingly, the mRNA encoding ARNT significantly increases with age in all cell types except monocytes, while mRNA coding for REST declines with aging in four cell types and shows no significant change in naive CD8⁺ T cells and NK cells. mRNAs coding for CTCF showed strong age association across numerous cell types (Supplementary file 8). The hypothesis that oxygen sensing regulates directly or indirectly DNA methylations is consistent with studies showing that in replicating fibroblasts, biological age estimated by DNA methylation slows down under hypoxia compared to normoxia (Matsuyama et al., 2019). Further, many genes close by to ‘shared’ age-differentially methylated CpG identified in our analyses play important roles in hypoxia response (Figure 5C).

The specific mechanisms connecting age-related changes in DNA methylation in genes which also contain binding motifs the master hypoxia-response mediators remain unknown. Shahrzad et al. reported an inverse correlation between the severity of hypoxia and the degree of DNA methylation (Shahrzad et al., 2007). There is evidence that hypoxia-induced hypermethylation may be due to reduced TETs activity (Thienpont et al., 2016). Our findings add to this literature by suggesting that a direct interaction between hypoxia-related TFs and DNA methylation at specific DNA sites occur with aging, perhaps as an adaptive response triggered by fluctuations in oxygen levels that occur in many age-related conditions. This hypothesis is consistent with oxygen availability been the most important environmental factor that requires physiological adaptation during pregnancy and development and extends this concept in a life course perspective.

A limitation of this study is that we have focused on circulating cells and, therefore, our findings may not apply to age methylation in other tissues. In addition, our findings were not replicated in an independent cross-sectional study population. Despite these limitations, this study has unique features: a cohort of exceptionally healthy donors and percent methylation was assessed in specific cell types obtained by cytapheresis and sorted by using state-of-the art methods.

Conclusion

Age-associated DNA methylation profiles of the six purified primary immune cell populations in the blood show more cell specificity than sharedness. However, we observe common regulatory features with respect to TF-binding motifs and histone modifications. Based on the consistent association of these methylated sites with ARNT and REST, which are master hypoxia regulators, we hypothesize that oxygen sensing and hypoxia drive mechanisms for changes in methylation. This hypothesis should be further explored in animal models with manipulation of oxygen levels and serial measures of DNA methylation in circulating immune cells.

Materials and methods

Cohort details

Request a detailed protocol

Buffy coat, PBMCs, and granulocytes were collected from Genetic and Epigenetic Signatures of Translational Aging Laboratory Testing study (GESTALT) study participants (N = 55; 34 men and 21 women; age 22–83 years) who were free of diseases (except controlled hypertension or history of cancer silent for >10 years), not on medications (except one antihypertensive drug), had no physical or cognitive impairments, non-smokers, weighed >110 lbs, had body mass index <30 kg/m² (Roy et al., 2021; Ubaida-Mohien et al., 2019). GESTALT was approved by the institutional review board of the National Institutes of Health and participants explicitly consented to participate.

Isolation of PBMC and immune cell populations

Request a detailed protocol

PBMCs were isolated from cytapheresis packs by density gradient centrifugation using Ficoll-Paque Plus. Total B, CD4⁺, and CD8⁺ T cells were enriched by negative selection using EasySep Negative Human kits specific for each cell type; monocytes were negatively enriched using ‘EasySep Human Monocyte Enrichment Kit w/o CD16 depletion’. Natural killer cells were negatively enriched by depleting PBMCs with antibodies against CD3, CD4, CD14, CD19, and Glycophorin-A in HBSS (Hanks' Balanced Salt Solution) buffer. Enriched cell populations were FACS (fluorescence-activated cell sorting) sorted by flow cytometry as per Human Immunophenotyping Consortium (HIPC) phenotyping panels (Maecker et al., 2012). Gating strategies and post-sort purity were analyzed by FlowJo software (LLC, Ashland, OR) (Roy et al., 2021). Granulocytes were positively selected from whole blood using EasySep Human Whole Blood CD66b Positive Selection Kit. Purified cells and PBMC were washed with phosphate-buffered saline, snap frozen and stored at −80°C. All sorted cells were >95% pure by flow cytometry (Roy et al., 2021).

Assessment of DNA methylation

Request a detailed protocol

DNA was isolated from 1 to 2 million cells using DNAQuik DNA Extraction protocol and the Qiagen DNeasy Kit. 300 ng of DNA was treated with sodium bisulfite using Zymo EZ-96 DNA Methylation Kit. The methylation of ~850,000 CpG sites was determined using Illumina Human MethylationEPIC BeadChip, and data preanalyzed by GenomeStudio 2011.1.

Data processing and functional annotation of CpG sites

Request a detailed protocol

Analyses were performed by the R minfi package (Aryee et al., 2014; Fortin et al., 2017). Probes with low detection p-values (cutoff 0.01) were filtered out (Moran et al., 2016). Data were normalized using noob and BMIQ (Liu and Siegmund, 2016), batch corrected by ComBat function (sva package), and β values were used for differential methylation analyses. Following the MethylationEPIC probe annotation (IlluminaHumanMethylationEPICanno-.ilm10b2.hg19) to the UCSC RefSeq genes (hg19), we grouped the locations into three categories: (1) promoter group – TSS 1500 (from 201 to 1500 bp upstream of TSS), TSS 200 (≤200 bp upstream of TSS), 5′UTR, first exon; (2) genebody – exons (all exons except exon1), exon intron boundary, intron and 3′UTR; and (3) intergenic probes. The first gene in the annotation package was considered. Probes were divided into three groups – within CpG islands (CGI), within CpG shore (0–2 kb from CGI), CpG shelf (2–4 kb from CGI), and open sea (>4 kb from CGI).

Definition of age-associated probes

Request a detailed protocol

Age- and sex-adjusted CpG-specific beta-regressions were performed on normalized β values using the R betareg function. p-values were adjusted for multiple testing (Benjamini–Hochberg [BH] adjusted p < 0.05). Probes with FDR p < 0.05 for age and FDR p > 0.05 for sex were considered age-differentially methylated CpGs. Beta-regression estimate value was used to group the age-associated probes as hypo- (Estimate_age <0) or hypermethylated (Estimate_age >0). The overlap of probes across multiple combinations of the six cell types was assessed using R package SuperExactTest (v.1.1.0) (Wang et al., 2015).

Gene set enrichment analysis

Request a detailed protocol

Based on the EPICarray annotation, genes were classified as differentially hypo- or hypermethylated with age. Genes with both age hypo- and hypermethylated CpGs were removed from the analysis. Enrichment analysis was performed by the tmodHGtest method in the tmod v.0.46.2 R package, comparing a foreground list of genes found in ≥5 cell types against reference gene set collections ‘Hallmarks’ and ‘Canonical Pathways’ (which includes Reactome, KEGG, WikiPathways, PID, and Biocarta gene sets) from the Molecular Signature Database MSigDB (v.7.4) (Subramanian et al., 2005).

For the gene enrichment analysis of the principal components, the top 500 CpG probes corresponding to the positive and negative directions along PC1, PC2, and PC3 were extracted and annotated to nearest gene as per the manufacturer’s annotation file. The ambiguous genes with probes associated with both positive and negative PC directions were removed from the analysis. The remaining genes were run through the abovementioned enrichment analysis pipeline. A filter based on q-value <0.05 was imposed to find the most significant pathways.

Visualization of histone peaks and DHS peaks

Request a detailed protocol

Primary cell DHS and chromatin ChIP-Seq bigwig files were downloaded from ENCODE (Roy et al., 2021). DeepTools was used to visualize DHS and histone peaks in +3 kb region surrounding age-associated shared and non-shared methylated sites. For plotting purposes, the order of methylated probes was determined based on descending score of DHS peaks and followed for all histone marks (H3K4me1, H3K4me3, and H3K27ac).

Annotation of age-associated methylated probes using chromHMM

Request a detailed protocol

The 18-state chromHMM models (based on 6 chromatin marks H3K4me3, H3K4me1, H3K36me3, H3K27me3, H3K9me3, and H3K27ac) for various immune cells (E032 – primary B cell, E038 – primary naive CD4⁺ T cells, E047 – primary naive CD8⁺ T cells, E029 – monocyte, E046 – NK cell) were downloaded from Roadmap epigenomics project. Bedops tool was used to map the age-associated methylated sites to the respective chromHMM profiles. All Infinium MethylationEPIC array probes were also partitioned using each of the immune cell chromHMM profiles as controls.

Prediction of de novo TF-binding motifs by HOMER

Request a detailed protocol

All the age-associated methylation sites were considered for HOMER analysis. A region of ±200 bp around each age-associated methylated site was provided as input for analysis in HOMER using de novo setting (Heinz et al., 2010). As a background, we used the default background list that HOMER creates by matching the GC% in the input list. The output from the stringent de novo analysis was considered for downstream data interpretation.

InCHIANTI longitudinal study cohort

Request a detailed protocol

InCHIANTI (Invecchiare in Chianti) is a population-based cohort of individuals ≥20 years old from the Chianti region of Tuscany, Italy (PMID: 11129752). The Italian National Institute of Research and Care on Aging Institutional Review Board approved the study protocol and all participants explicitly consented to participate. DNA methylation from 699 participants (1841 observations) was used for the analysis. CpG methylation of 485,577 CpGs was determined by the Illumina Infinium HumanMethylation450 BeadChip (Illumina Inc, San Diego, CA) and data processed by the R package ‘sesame’. Mean rates of change were estimated from 2 to 3 longitudinal timepoints.

RNA-Seq sample extraction, processing, and data analysis

Request a detailed protocol

Total RNA was extracted from 2 × 10⁶ cells, depleted from ribosomal RNA and 50 ng was used for cDNA synthesis and library preparation. Libraries were sequenced for 138 cycles on Illumina HiSeq 2500. After adapter removal and end trimming of raw FASTQ files, transcript abundances were quantified with reference to hg19 transcriptome using kallisto 0.44 (with options --single -l 250 -s 25). Transcripts were aggregated to genes with tximport and filtered out if less than 10 TPM were detected in more than 33% of the samples. Linear regression models (~phase + age*sex) were used on TPM normalized expression values to study expression changes of selected TFs with age. Only the regression coefficient and p-value for the three TF genes – ARNT, CTCF, and REST were used in this study.

Inflammatory cytokine analysis

Request a detailed protocol

Published SomaScan protein data from the same cohort of donors used in the present study were extracted to look for age-associated changes in seven cytokines (IL6, IL1RN, IL1A, IL1B, TNF, TNFRSF1A, and TNFRSF1B) (Tanaka et al., 2018). Briefly, plasma proteomics was measured using the 1.3k SomaScan assay (SomaLogic, Boulder, CO) followed by standard quality control and normalization procedures as described in previous publications (Candia et al., 2017; Candia et al., 2022). Normalized data for seven cytokines were extracted (detailed annotation provided in Supplementary file 10). To complement the age association analysis of CpGs adjusted by sex (CpG ~ age + sex), we performed additional beta-regression analyses separately including each target pro-inflammatory cytokine as an explanatory variable in the form: CpG ~ age + sex + cytokine. Details about the cytokines are provided in OSF.

Data availability

Researchers interested in using the data from the previously published InCHIANTI study are invited to submit a proposal for consideration, for full details please see https://www.nia.nih.gov/inchianti-study. Code and data processing scripts (including a de-identified version of the GESTALT dataset) are available on OSF. DNA methylation EPIC 850k data are available at GEO under accession number GSE184269.

The following data sets were generated

1. Roy et al
(2023) Open Science Framework
ID rxw6h. Epigenetic signature of human immune aging: the GESTALT study.

https://osf.io/rxw6h/

The following previously published data sets were used

1. Kaileh M
2. Roy R
3. Ramamoorthy S
4. Boller S
5. Grosschedl R
6. De S
(2021) NCBI Gene Expression Omnibus
ID GSE184269. Specification of human immune cell epigenetic identity by combinations of transcription factors (MethylationEPIC).

https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE184269

References

1. Acevedo N
2. Reinius LE
3. Vitezic M
4. Fortino V
5. Söderhäll C
6. Honkanen H
7. Veijola R
8. Simell O
9. Toppari J
10. Ilonen J
11. Knip M
12. Scheynius A
13. Hyöty H
14. Greco D
15. Kere J
(2015) Age-associated DNA methylation changes in immune genes, histone modifiers and chromatin remodeling factors within 5 years after birth in human blood leukocytes
Clinical Epigenetics 7:34.
https://doi.org/10.1186/s13148-015-0064-6
- PubMed
- Google Scholar
(2014) Minfi: a flexible and comprehensive Bioconductor package for the analysis of Infinium DNA methylation microarrays
Bioinformatics 30:1363–1369.
https://doi.org/10.1093/bioinformatics/btu049
- PubMed
- Google Scholar
1. Bae S
2. Lesch BJ
(2020) H3K4me1 distribution predicts transcription state and poising at promoters
Frontiers in Cell and Developmental Biology 8:289.
https://doi.org/10.3389/fcell.2020.00289
- PubMed
- Google Scholar
1. Bell JT
2. Tsai P-C
3. Yang T-P
4. Pidsley R
5. Nisbet J
6. Glass D
7. Mangino M
8. Zhai G
9. Zhang F
10. Valdes A
11. Shin S-Y
12. Dempster EL
13. Murray RM
14. Grundberg E
15. Hedman AK
16. Nica A
17. Small KS
18. MuTHER Consortium
19. Dermitzakis ET
20. McCarthy MI
21. Mill J
22. Spector TD
23. Deloukas P
(2012) Epigenome-wide scans identify differentially methylated regions for age and age-related phenotypes in a healthy ageing population
PLOS Genetics 8:e1002629.
https://doi.org/10.1371/journal.pgen.1002629
- PubMed
- Google Scholar
(2021) Nuclear compartmentalization as a mechanism of quantitative control of gene expression
Nature Reviews. Molecular Cell Biology 22:653–670.
https://doi.org/10.1038/s41580-021-00387-1
- PubMed
- Google Scholar
1. Bocklandt S
2. Lin W
3. Sehl ME
4. Sánchez FJ
5. Sinsheimer JS
6. Horvath S
7. Vilain E
(2011) Epigenetic predictor of age
PLOS ONE 6:e14821.
https://doi.org/10.1371/journal.pone.0014821
- PubMed
- Google Scholar
1. Cai Z-L
2. Liu C
3. Yao Q
4. Xie Q-W
5. Hu T-T
6. Wu Q-Q
7. Tang Q-Z
(2020) The pro-migration and anti-apoptosis effects of HMGA2 in HUVECs stimulated by hypoxia
Cell Cycle 19:3534–3545.
https://doi.org/10.1080/15384101.2020.1850970
- PubMed
- Google Scholar
1. Candia J
2. Cheung F
3. Kotliarov Y
4. Fantoni G
5. Sellers B
6. Griesman T
7. Huang J
8. Stuccio S
9. Zingone A
10. Ryan BM
11. Tsang JS
12. Biancotto A
(2017) Assessment of variability in the SOMAscan assay
Scientific Reports 7:14248.
https://doi.org/10.1038/s41598-017-14755-5
- PubMed
- Google Scholar
1. Candia J
2. Daya GN
3. Tanaka T
4. Ferrucci L
5. Walker KA
(2022) Assessment of variability in the plasma 7k SomaScan proteomics assay
Scientific Reports 12:17147.
https://doi.org/10.1038/s41598-022-22116-0
- PubMed
- Google Scholar
1. Carmeliet P
2. Jain RK
(2011) Molecular mechanisms and clinical applications of angiogenesis
Nature 473:298–307.
https://doi.org/10.1038/nature10144
- PubMed
- Google Scholar
1. Cavadas MAS
2. Mesnieres M
3. Crifo B
4. Manresa MC
5. Selfridge AC
6. Keogh CE
7. Fabian Z
8. Scholz CC
9. Nolan KA
10. Rocha LMA
11. Tambuwala MM
12. Brown S
13. Wdowicz A
14. Corbett D
15. Murphy KJ
16. Godson C
17. Cummins EP
18. Taylor CT
19. Cheong A
(2016) REST is a hypoxia-responsive transcriptional repressor
Scientific Reports 6:31355.
https://doi.org/10.1038/srep31355
- PubMed
- Google Scholar
1. Chakraborty S
2. Ain R
(2017) Nitric-oxide synthase trafficking inducer is a pleiotropic regulator of endothelial cell function and signaling
The Journal of Biological Chemistry 292:6600–6620.
https://doi.org/10.1074/jbc.M116.742627
- PubMed
- Google Scholar
1. Chen BH
2. Marioni RE
3. Colicino E
4. Peters MJ
5. Ward-Caviness CK
6. Tsai P-C
7. Roetker NS
8. Just AC
9. Demerath EW
10. Guan W
11. Bressler J
12. Fornage M
13. Studenski S
14. Vandiver AR
15. Moore AZ
16. Tanaka T
17. Kiel DP
18. Liang L
19. Vokonas P
20. Schwartz J
21. Lunetta KL
22. Murabito JM
23. Bandinelli S
24. Hernandez DG
25. Melzer D
26. Nalls M
27. Pilling LC
28. Price TR
29. Singleton AB
30. Gieger C
31. Holle R
32. Kretschmer A
33. Kronenberg F
34. Kunze S
35. Linseisen J
36. Meisinger C
37. Rathmann W
38. Waldenberger M
39. Visscher PM
40. Shah S
41. Wray NR
42. McRae AF
43. Franco OH
44. Hofman A
45. Uitterlinden AG
46. Absher D
47. Assimes T
48. Levine ME
49. Lu AT
50. Tsao PS
51. Hou L
52. Manson JE
53. Carty CL
54. LaCroix AZ
55. Reiner AP
56. Spector TD
57. Feinberg AP
58. Levy D
59. Baccarelli A
60. van Meurs J
61. Bell JT
62. Peters A
63. Deary IJ
64. Pankow JS
65. Ferrucci L
66. Horvath S
(2016) DNA methylation-based measures of biological age: meta-analysis predicting time to death
Aging 8:1844–1865.
https://doi.org/10.18632/aging.101020
- Google Scholar
1. Craps S
2. Van Wauwe J
3. De Moudt S
4. De Munck D
5. Leloup AJA
6. Boeckx B
7. Vervliet T
8. Dheedene W
9. Criem N
10. Geeroms C
11. Jones EAV
12. Zwijsen A
13. Lambrechts D
14. Fransen P
15. Beerens M
16. Luttun A
(2021) Prdm16 supports arterial flow recovery by maintaining endothelial function
Circulation Research 129:63–77.
https://doi.org/10.1161/CIRCRESAHA.120.318501
- PubMed
- Google Scholar
1. Dasgupta P
(2004) Somatostatin analogues: multiple roles in cellular proliferation, neoplasia, and angiogenesis
Pharmacology & Therapeutics 102:61–85.
https://doi.org/10.1016/j.pharmthera.2004.02.002
- PubMed
- Google Scholar
(2019) Gi/o-protein coupled receptors in the aging brain
Frontiers in Aging Neuroscience 11:89.
https://doi.org/10.3389/fnagi.2019.00089
- Google Scholar
(2017) Age-associated DNA methylation changes in naive CD4⁺ T cells suggest an evolving autoimmune epigenotype in aging T cells
Epigenomics 9:429–445.
https://doi.org/10.2217/epi-2016-0143
- PubMed
- Google Scholar
1. Ernst J
2. Kellis M
(2012) ChromHMM: automating chromatin-state discovery and characterization
Nature Methods 9:215–216.
https://doi.org/10.1038/nmeth.1906
- PubMed
- Google Scholar
1. Ewing E
2. Kular L
3. Fernandes SJ
4. Karathanasis N
5. Lagani V
6. Ruhrmann S
7. Tsamardinos I
8. Tegner J
9. Piehl F
10. Gomez-Cabrero D
11. Jagodic M
(2019) Combining evidence from four immune cell types identifies DNA methylation patterns that implicate functionally distinct pathways during Multiple Sclerosis progression
EBioMedicine 43:411–423.
https://doi.org/10.1016/j.ebiom.2019.04.042
- PubMed
- Google Scholar
(2000) Subsystems contributing to the decline in ability to walk: bridging the gap between epidemiology and geriatric practice in the InCHIANTI study
Journal of the American Geriatrics Society 48:1618–1625.
https://doi.org/10.1111/j.1532-5415.2000.tb03873.x
- PubMed
- Google Scholar
(2017) Preprocessing, normalization and integration of the Illumina HumanMethylationEPIC array with minfi
Bioinformatics 33:558–560.
https://doi.org/10.1093/bioinformatics/btw691
- PubMed
- Google Scholar
1. Gale CR
2. Marioni RE
3. Harris SE
4. Starr JM
5. Deary IJ
(2018) DNA methylation and the epigenetic clock in relation to physical frailty in older people: the Lothian Birth Cohort 1936
Clinical Epigenetics 10:101.
https://doi.org/10.1186/s13148-018-0538-4
- PubMed
- Google Scholar
(2014) T-cell differentiation
Immunological Reviews 261:157–168.
https://doi.org/10.1111/imr.12205
- PubMed
- Google Scholar
1. Gu Y
2. Luo M
3. Li Y
4. Su Z
5. Wang Y
6. Chen X
7. Zhang S
8. Sun W
9. Kong X
(2019) Bcl6 knockdown aggravates hypoxia injury in cardiomyocytes via the P38 pathway
Cell Biology International 43:108–116.
https://doi.org/10.1002/cbin.11028
- PubMed
- Google Scholar
1. Gusdon AM
2. Zhu J
3. Van Houten B
4. Chu CT
(2012) ATP13A2 regulates mitochondrial bioenergetics through macroautophagy
Neurobiology of Disease 45:962–972.
https://doi.org/10.1016/j.nbd.2011.12.015
- PubMed
- Google Scholar
1. Hannum G
2. Guinney J
3. Zhao L
4. Zhang L
5. Hughes G
6. Sadda S
7. Klotzle B
8. Bibikova M
9. Fan J-B
10. Gao Y
11. Deconde R
12. Chen M
13. Rajapakse I
14. Friend S
15. Ideker T
16. Zhang K
(2013) Genome-wide methylation profiles reveal quantitative views of human aging rates
Molecular Cell 49:359–367.
https://doi.org/10.1016/j.molcel.2012.10.016
- PubMed
- Google Scholar
1. Heinz S
2. Benner C
3. Spann N
4. Bertolino E
5. Lin YC
6. Laslo P
7. Cheng JX
8. Murre C
9. Singh H
10. Glass CK
(2010) Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities
Molecular Cell 38:576–589.
https://doi.org/10.1016/j.molcel.2010.05.004
- PubMed
- Google Scholar
1. Horvath S
(2013) DNA methylation age of human tissues and cell types
Genome Biology 14:10.
https://doi.org/10.1186/gb-2013-14-10-r115
- PubMed
- Google Scholar
1. Horvath S
2. Raj K
(2018) DNA methylation-based biomarkers and the epigenetic clock theory of ageing
Nature Reviews. Genetics 19:371–384.
https://doi.org/10.1038/s41576-018-0004-3
- PubMed
- Google Scholar
1. Hsu C
2. Morohashi Y
3. Yoshimura S-I
4. Manrique-Hoyos N
5. Jung S
6. Lauterbach MA
7. Bakhti M
8. Grønborg M
9. Möbius W
10. Rhee J
11. Barr FA
12. Simons M
(2010) Regulation of exosome secretion by Rab35 and its GTPase-activating proteins TBC1D10A-C
The Journal of Cell Biology 189:223–232.
https://doi.org/10.1083/jcb.200911018
- PubMed
- Google Scholar
(2016) Aging-associated DNA methylation changes in middle-aged individuals: the Young Finns study
BMC Genomics 17:103.
https://doi.org/10.1186/s12864-016-2421-z
- PubMed
- Google Scholar
(2023) Multi-modal profiling of peripheral blood cells across the human lifespan reveals distinct immune cell signatures of aging and longevity
EBioMedicine 90:104514.
https://doi.org/10.1016/j.ebiom.2023.104514
- PubMed
- Google Scholar
1. Kim C
2. Jin J
3. Weyand CM
4. Goronzy JJ
(2020) The transcription factor TCF1 in T cell differentiation and aging
International Journal of Molecular Sciences 21:6497.
https://doi.org/10.3390/ijms21186497
- Google Scholar
1. Lazarou M
2. Narendra DP
3. Jin SM
4. Tekle E
5. Banerjee S
6. Youle RJ
(2013) PINK1 drives Parkin self-association and HECT-like E3 activity upstream of mitochondrial binding
The Journal of Cell Biology 200:163–172.
https://doi.org/10.1083/jcb.201210111
- PubMed
- Google Scholar
1. Liu J
2. Siegmund KD
(2016) An evaluation of processing methods for HumanMethylation450 BeadChip data
BMC Genomics 17:469.
https://doi.org/10.1186/s12864-016-2819-7
- PubMed
- Google Scholar
1. Liu T-J
2. Yeh Y-C
3. Lee W-L
4. Wang L-C
5. Lee H-W
6. Shiu M-T
7. Su C-S
8. Lai H-C
(2020) Insulin ameliorates hypoxia-induced autophagy, endoplasmic reticular stress and apoptosis of myocardial cells: In vitro and ex vivo models
European Journal of Pharmacology 880:173125.
https://doi.org/10.1016/j.ejphar.2020.173125
- PubMed
- Google Scholar
1. Lu T
2. Aron L
3. Zullo J
4. Pan Y
5. Kim H
6. Chen Y
7. Yang T-H
8. Kim H-M
9. Drake D
10. Liu XS
11. Bennett DA
12. Colaiácovo MP
13. Yankner BA
(2014) REST and stress resistance in ageing and Alzheimer’s disease
Nature 507:448–454.
https://doi.org/10.1038/nature13163
- Google Scholar
1. Lu AT
2. Xue L
3. Salfati EL
4. Chen BH
5. Ferrucci L
6. Levy D
7. Joehanes R
8. Murabito JM
9. Kiel DP
10. Tsai P-C
11. Yet I
12. Bell JT
13. Mangino M
14. Tanaka T
15. McRae AF
16. Marioni RE
17. Visscher PM
18. Wray NR
19. Deary IJ
20. Levine ME
21. Quach A
22. Assimes T
23. Tsao PS
24. Absher D
25. Stewart JD
26. Li Y
27. Reiner AP
28. Hou L
29. Baccarelli AA
30. Whitsel EA
31. Aviv A
32. Cardona A
33. Day FR
34. Wareham NJ
35. Perry JRB
36. Ong KK
37. Raj K
38. Lunetta KL
39. Horvath S
(2018) GWAS of epigenetic aging rates in blood reveals a critical role for TERT
Nature Communications 9:387.
https://doi.org/10.1038/s41467-017-02697-5
- PubMed
- Google Scholar
(2012) Standardizing immunophenotyping for the human immunology project
Nature Reviews. Immunology 12:191–200.
https://doi.org/10.1038/nri3158
- PubMed
- Google Scholar
1. Marioni RE
2. Shah S
3. McRae AF
4. Chen BH
5. Colicino E
6. Harris SE
7. Gibson J
8. Henders AK
9. Redmond P
10. Cox SR
11. Pattie A
12. Corley J
13. Murphy L
14. Martin NG
15. Montgomery GW
16. Feinberg AP
17. Fallin MD
18. Multhaup ML
19. Jaffe AE
20. Joehanes R
21. Schwartz J
22. Just AC
23. Lunetta KL
24. Murabito JM
25. Starr JM
26. Horvath S
27. Baccarelli AA
28. Levy D
29. Visscher PM
30. Wray NR
31. Deary IJ
(2015) DNA methylation age of blood predicts all-cause mortality in later life
Genome Biology 16:25.
https://doi.org/10.1186/s13059-015-0584-6
- PubMed
- Google Scholar
(2019) Screening for genes that accelerate the epigenetic aging clock in humans reveals a role for the H3K36 methyltransferase NSD1
Genome Biology 20:146.
https://doi.org/10.1186/s13059-019-1753-9
- PubMed
- Google Scholar
1. Marttila S
2. Kananen L
3. Häyrynen S
4. Jylhävä J
5. Nevalainen T
6. Hervonen A
7. Jylhä M
8. Nykter M
9. Hurme M
(2015) Ageing-associated changes in the human DNA methylome: genomic locations and effects on gene expression
BMC Genomics 16:179.
https://doi.org/10.1186/s12864-015-1381-z
- PubMed
- Google Scholar
(2019) Epigenetic clock analysis of human fibroblasts in vitro: effects of hypoxia, donor age, and expression of hTERT and SV40 largeT
Aging 11:3012–3022.
https://doi.org/10.18632/aging.101955
- Google Scholar
1. McCartney DL
2. Stevenson AJ
3. Walker RM
4. Gibson J
5. Morris SW
6. Campbell A
7. Murray AD
8. Whalley HC
9. Porteous DJ
10. McIntosh AM
11. Evans KL
12. Deary IJ
13. Marioni RE
(2018) Investigating the relationship between DNA methylation age acceleration and risk factors for Alzheimer’s disease
Alzheimer’s & Dementia 10:429–437.
https://doi.org/10.1016/j.dadm.2018.05.006
- Google Scholar
(2014) Effects of cytosine methylation on transcription factor binding sites
BMC Genomics 15:119.
https://doi.org/10.1186/1471-2164-15-119
- Google Scholar
1. Moore LD
2. Le T
3. Fan G
(2013) DNA methylation and its basic function
Neuropsychopharmacology 38:23–38.
https://doi.org/10.1038/npp.2012.112
- PubMed
- Google Scholar
(2016) Change in Epigenome-Wide DNA Methylation Over 9 Years and Subsequent Mortality: Results From the InCHIANTI Study
The Journals of Gerontology. Series A, Biological Sciences and Medical Sciences 71:1029–1035.
https://doi.org/10.1093/gerona/glv118
- PubMed
- Google Scholar
(2016) Validation of a DNA methylation microarray for 850,000 CpG sites of the human genome enriched in enhancer sequences
Epigenomics 8:389–399.
https://doi.org/10.2217/epi.15.114
- PubMed
- Google Scholar
(2020) Arid5a, an RNA-Binding protein in immune regulation: RNA Stability, inflammation, and autoimmunity
Trends in Immunology 41:255–268.
https://doi.org/10.1016/j.it.2020.01.004
- PubMed
- Google Scholar
(2020) Cross-species insights into genomic adaptations to hypoxia
Frontiers in Genetics 11:743.
https://doi.org/10.3389/fgene.2020.00743
- PubMed
- Google Scholar
1. Pangou E
2. Befani C
3. Mylonis I
4. Samiotaki M
5. Panayotou G
6. Simos G
7. Liakos P
(2016) HIF-2α phosphorylation by CK1δ promotes erythropoietin secretion in liver cancer cells under hypoxia
Journal of Cell Science 129:4213–4226.
https://doi.org/10.1242/jcs.191395
- PubMed
- Google Scholar
1. Pundhir S
2. Bagger FO
3. Lauridsen FB
4. Rapin N
5. Porse BT
(2016) Peak-valley-peak pattern of histone modifications delineates active regulatory elements and their directionality
Nucleic Acids Research 44:4037–4051.
https://doi.org/10.1093/nar/gkw250
- Google Scholar
1. Reynolds LM
2. Taylor JR
3. Ding J
4. Lohman K
5. Johnson C
6. Siscovick D
7. Burke G
8. Post W
9. Shea S
10. Jacobs DR
11. Stunnenberg H
12. Kritchevsky SB
13. Hoeschele I
14. McCall CE
15. Herrington DM
16. Tracy RP
17. Liu Y
(2014) Age-related variations in the methylome associated with gene expression in human monocytes and T cells
Nature Communications 5:5366.
https://doi.org/10.1038/ncomms6366
- Google Scholar
1. Roy R
2. Ramamoorthy S
3. Shapiro BD
4. Kaileh M
5. Hernandez D
6. Sarantopoulou D
7. Arepalli S
8. Boller S
9. Singh A
10. Bektas A
11. Kim J
12. Moore AZ
13. Tanaka T
14. McKelvey J
15. Zukley L
16. Nguyen C
17. Wallace T
18. Dunn C
19. Wersto R
20. Wood W
21. Piao Y
22. Becker KG
23. Coletta C
24. De S
25. Sen JM
26. Battle A
27. Weng N-P
28. Grosschedl R
29. Ferrucci L
30. Sen R
(2021) DNA methylation signatures reveal that distinct combinations of transcription factors specify human immune cell epigenetic identity
Immunity 54:2465–2480.
https://doi.org/10.1016/j.immuni.2021.10.001
- PubMed
- Google Scholar
1. Semenza GL
(2000) HIF-1: mediator of physiological and pathophysiological responses to hypoxia
Journal of Applied Physiology 88:1474–1480.
https://doi.org/10.1152/jappl.2000.88.4.1474
- Google Scholar
(2007) Induction of DNA hypomethylation by tumor hypoxia
Epigenetics 2:119–125.
https://doi.org/10.4161/epi.2.2.4613
- PubMed
- Google Scholar
1. Stegmann TJ
(1998) FGF-1: a human growth factor in the induction of neoangiogenesis
Expert Opinion on Investigational Drugs 7:2011–2015.
https://doi.org/10.1517/13543784.7.12.2011
- PubMed
- Google Scholar
1. Subramanian A
2. Tamayo P
3. Mootha VK
4. Mukherjee S
5. Ebert BL
6. Gillette MA
7. Paulovich A
8. Pomeroy SL
9. Golub TR
10. Lander ES
11. Mesirov JP
(2005) Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles
PNAS 102:15545–15550.
https://doi.org/10.1073/pnas.0506580102
- Google Scholar
1. Tan WSD
2. Lee JJ
3. Satish RL
4. Ang E-T
(2012) Detectability of secretagogin in human erythrocytes
Neuroscience Letters 526:59–62.
https://doi.org/10.1016/j.neulet.2012.08.006
- PubMed
- Google Scholar
1. Tanaka T
2. Biancotto A
3. Moaddel R
4. Moore AZ
5. Gonzalez-Freire M
6. Aon MA
7. Candia J
8. Zhang P
9. Cheung F
10. Fantoni G
11. CHI consortium
12. Semba RD
13. Ferrucci L
(2018) Plasma proteomic signature of age in healthy humans
Aging Cell 17:e12799.
https://doi.org/10.1111/acel.12799
- PubMed
- Google Scholar
1. Teschendorff AE
2. Menon U
3. Gentry-Maharaj A
4. Ramus SJ
5. Weisenberger DJ
6. Shen H
7. Campan M
8. Noushmehr H
9. Bell CG
10. Maxwell AP
11. Savage DA
12. Mueller-Holzner E
13. Marth C
14. Kocjan G
15. Gayther SA
16. Jones A
17. Beck S
18. Wagner W
19. Laird PW
20. Jacobs IJ
21. Widschwendter M
(2010) Age-dependent DNA methylation of genes that are suppressed in stem cells is a hallmark of cancer
Genome Research 20:440–446.
https://doi.org/10.1101/gr.103606.109
- PubMed
- Google Scholar
(2013) Age-associated epigenetic drift: implications, and a case of epigenetic thrift?
Human Molecular Genetics 22:R7–R15.
https://doi.org/10.1093/hmg/ddt375
- PubMed
- Google Scholar
1. Thienpont B
2. Steinbacher J
3. Zhao H
4. D’Anna F
5. Kuchnio A
6. Ploumakis A
7. Ghesquière B
8. Van Dyck L
9. Boeckx B
10. Schoonjans L
11. Hermans E
12. Amant F
13. Kristensen VN
14. Koh KP
15. Mazzone M
16. Coleman ML
17. Carell T
18. Carmeliet P
19. Lambrechts D
(2016) Tumour hypoxia causes DNA hypermethylation by reducing TET activity
Nature 537:63–68.
https://doi.org/10.1038/nature19081
- Google Scholar
1. Tserel L
2. Kolde R
3. Limbach M
4. Tretyakov K
5. Kasela S
6. Kisand K
7. Saare M
8. Vilo J
9. Metspalu A
10. Milani L
11. Peterson P
(2015) Age-related profiling of DNA methylation in CD8+ T cells reveals changes in immune response and transcriptional regulator genes
Scientific Reports 5:13107.
https://doi.org/10.1038/srep13107
- PubMed
- Google Scholar
(2019) Physical activity associated proteomics of skeletal muscle: being physically active in daily life may protect skeletal muscle from aging
Frontiers in Physiology 10:312.
https://doi.org/10.3389/fphys.2019.00312
- PubMed
- Google Scholar
1. van Ruiten MS
2. Rowland BD
(2021) On the choreography of genome folding: A grand pas de deux of cohesin and CTCF
Current Opinion in Cell Biology 70:84–90.
https://doi.org/10.1016/j.ceb.2020.12.001
- PubMed
- Google Scholar
1. Wang J
2. Taba Y
3. Pang J
4. Yin G
5. Yan C
6. Berk BC
(2009) GIT1 mediates VEGF-induced podosome formation in endothelial cells: critical role for PLCgamma
Arteriosclerosis, Thrombosis, and Vascular Biology 29:202–208.
https://doi.org/10.1161/ATVBAHA.108.174391
- PubMed
- Google Scholar
1. Wang M
2. Zhao Y
3. Zhang B
(2015) Efficient test and visualization of multi-set intersections
Scientific Reports 5:16923.
https://doi.org/10.1038/srep16923
- Google Scholar
(2002)
ARID proteins: a diverse family of DNA binding proteins implicated in the control of cell growth, differentiation, and development

Cell Growth & Differentiation 13:95–106.
- PubMed
- Google Scholar
1. Yuan T
2. Jiao Y
3. de Jong S
4. Ophoff RA
5. Beck S
6. Teschendorff AE
(2015) An integrative multi-scale analysis of the dynamic DNA methylation landscape in aging
PLOS Genetics 11:e1004996.
https://doi.org/10.1371/journal.pgen.1004996
- PubMed
- Google Scholar
1. Yui MA
2. Rothenberg EV
(2014) Developmental gene networks: a triathlon on the course to T cell identity
Nature Reviews. Immunology 14:529–545.
https://doi.org/10.1038/nri3702
- PubMed
- Google Scholar

Article and author information

Author details

Roshni Roy

Laboratory of Molecular Biology and Immunology, National Institute on Aging, Baltimore, United States

Contribution
Conceptualization, Formal analysis, Supervision, Funding acquisition, Visualization, Methodology, Writing – original draft, Writing – review and editing

Competing interests
No competing interests declared
Pei-Lun Kuo

Translational Gerontology Branch, National Institute on Aging, Baltimore, United States

Contribution
Data curation, Formal analysis, Writing – review and editing

Competing interests
No competing interests declared
Julián Candia

Translational Gerontology Branch, National Institute on Aging, Baltimore, United States

Contribution
Formal analysis, Writing – original draft

Competing interests
No competing interests declared

"This ORCID iD identifies the author of this article:" 0000-0001-5793-8989
Dimitra Sarantopoulou

Laboratory of Molecular Biology and Immunology, National Institute on Aging, Baltimore, United States

Contribution
Formal analysis, Methodology, Writing – review and editing

Competing interests
No competing interests declared
Ceereena Ubaida-Mohien

Translational Gerontology Branch, National Institute on Aging, Baltimore, United States

Contribution
Software, Visualization

Competing interests
No competing interests declared

"This ORCID iD identifies the author of this article:" 0000-0002-4301-4758
Dena Hernandez

Laboratory of Neurogenetics, National Institute on Aging, Bethesda, United States

Contribution
Data curation, Supervision, Methodology, Writing – review and editing

Competing interests
No competing interests declared
Mary Kaileh

Laboratory of Molecular Biology and Immunology, National Institute on Aging, Baltimore, United States

Contribution
Conceptualization, Resources, Methodology, Project administration, Writing – review and editing

Competing interests
No competing interests declared

"This ORCID iD identifies the author of this article:" 0000-0003-2314-312X
Sampath Arepalli

Laboratory of Neurogenetics, National Institute on Aging, Bethesda, United States

Contribution
Data curation

Competing interests
No competing interests declared
Amit Singh

Laboratory of Molecular Biology and Immunology, National Institute on Aging, Baltimore, United States

Contribution
Visualization, Methodology, Writing – review and editing

Competing interests
No competing interests declared
Arsun Bektas

Translational Gerontology Branch, National Institute on Aging, Baltimore, United States

Contribution
Data curation, Writing – review and editing

Competing interests
No competing interests declared
Jaekwan Kim

Laboratory of Molecular Biology and Immunology, National Institute on Aging, Baltimore, United States

Contribution
Data curation

Competing interests
No competing interests declared
Ann Z Moore

Translational Gerontology Branch, National Institute on Aging, Baltimore, United States

Contribution
Data curation, Formal analysis

Competing interests
No competing interests declared
Toshiko Tanaka

Translational Gerontology Branch, National Institute on Aging, Baltimore, United States

Contribution
Data curation

Competing interests
No competing interests declared

"This ORCID iD identifies the author of this article:" 0000-0002-4161-3829
Julia McKelvey

Clinical Research Core, National Institute on Aging, Baltimore, United States

Contribution
Conceptualization, Data curation

Competing interests
No competing interests declared
Linda Zukley

Clinical Research Core, National Institute on Aging, Baltimore, United States

Contribution
Conceptualization, Data curation

Competing interests
No competing interests declared
Cuong Nguyen

Flow Cytometry Unit, National Institute on Aging, Baltimore, United States

Contribution
Data curation, Methodology

Competing interests
No competing interests declared
Tonya Wallace

Flow Cytometry Unit, National Institute on Aging, Baltimore, United States

Contribution
Data curation

Competing interests
No competing interests declared
Christopher Dunn

Flow Cytometry Unit, National Institute on Aging, Baltimore, United States

Contribution
Data curation

Competing interests
No competing interests declared

"This ORCID iD identifies the author of this article:" 0000-0001-7899-0110
William Wood

Laboratory of Genetics and Genomics, National Institute on Aging, Baltimore, United States

Contribution
Data curation

Competing interests
No competing interests declared
Yulan Piao

Laboratory of Genetics and Genomics, National Institute on Aging, Baltimore, United States

Contribution
Data curation, Methodology

Competing interests
No competing interests declared
Christopher Coletta

Laboratory of Genetics and Genomics, National Institute on Aging, Baltimore, United States

Contribution
Data curation

Competing interests
No competing interests declared
Supriyo De

Laboratory of Genetics and Genomics, National Institute on Aging, Baltimore, United States

Contribution
Software, Formal analysis, Methodology

Competing interests
No competing interests declared
Jyoti Sen

Laboratory of Clinical Investigation, National Institute on Aging, Baltimore, United States

Contribution
Supervision, Methodology, Writing – review and editing

Competing interests
No competing interests declared
Nan-ping Weng

Laboratory of Molecular Biology and Immunology, National Institute on Aging, Baltimore, United States

Contribution
Conceptualization, Methodology, Writing – review and editing

Competing interests
No competing interests declared
Ranjan Sen

Laboratory of Molecular Biology and Immunology, National Institute on Aging, Baltimore, United States

Contribution
Conceptualization, Supervision, Writing – review and editing

Competing interests
No competing interests declared
Luigi Ferrucci

Translational Gerontology Branch, National Institute on Aging, Baltimore, United States

Contribution
Conceptualization, Formal analysis, Supervision, Funding acquisition, Visualization, Methodology, Writing – original draft, Writing – review and editing

For correspondence
ferruccilu@grc.nia.nih.gov

Competing interests
No competing interests declared

"This ORCID iD identifies the author of this article:" 0000-0002-6273-1613

Funding

No external funding was received for this work.

Acknowledgements

This work was supported entirely by the Intramural Research Program of the National Institute on Aging. We are grateful to the GESTALT participants and the GESTALT Study Team at Harbor Hospital and NIA.

Ethics

GESTALT study was approved by the institutional review board of the National Institutes of Health. Informed consent as well as the consent to publish the data collected was obtained from every participant in the study. Since the study of gene expression and epigenetic regulation are essential aims of GESTALT, all participants were required to consent to DNA/RNA testing and storage at all visits in order to participate in the study. The GESTALT IRB approval number is 15-AG-0063.

Version history

Received: January 12, 2023
Preprint posted: January 23, 2023 (view preprint)
Accepted: August 16, 2023
Accepted Manuscript published: August 17, 2023 (version 1)
Version of Record published: September 18, 2023 (version 2)

Copyright

This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.

Metrics

1,032

views
158

downloads
1

citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Mendeley

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

Roshni Roy
Pei-Lun Kuo
Julián Candia
Dimitra Sarantopoulou
Ceereena Ubaida-Mohien
Dena Hernandez
Mary Kaileh
Sampath Arepalli
Amit Singh
Arsun Bektas
Jaekwan Kim
Ann Z Moore
Toshiko Tanaka
Julia McKelvey
Linda Zukley
Cuong Nguyen
Tonya Wallace
Christopher Dunn
William Wood
Yulan Piao
Christopher Coletta
Supriyo De
Jyoti Sen
Nan-ping Weng
Ranjan Sen
Luigi Ferrucci

(2023)

Epigenetic signature of human immune aging in the GESTALT study

eLife 12:e86136.

https://doi.org/10.7554/eLife.86136

Categories and tags

Research organism

Human

Share this article

Cite this article

Study design and identification of age-associated methylation probes.

Characteristics of age-associated probes.

Pathway analysis of methylated probes.

Functional annotation of age-associated probes along with their grouping based on sharedness.

Association of transcription factor (TF)-binding motifs with age-differentially methylated CpGs.

Author details

Roshni Roy

Contribution

Competing interests

Pei-Lun Kuo

Contribution

Competing interests

Julián Candia

Contribution

Competing interests

Dimitra Sarantopoulou

Contribution

Competing interests

Ceereena Ubaida-Mohien

Contribution

Competing interests

Dena Hernandez

Contribution

Competing interests

Mary Kaileh

Contribution

Competing interests

Sampath Arepalli

Contribution

Competing interests

Amit Singh

Contribution

Competing interests

Arsun Bektas

Contribution

Competing interests

Jaekwan Kim

Contribution

Competing interests

Ann Z Moore

Contribution

Competing interests

Toshiko Tanaka

Contribution

Competing interests

Julia McKelvey

Contribution

Competing interests

Linda Zukley

Contribution

Competing interests

Cuong Nguyen

Contribution

Competing interests

Tonya Wallace

Contribution

Competing interests

Christopher Dunn

Contribution

Competing interests

William Wood

Contribution

Competing interests

Yulan Piao

Contribution

Competing interests

Christopher Coletta

Contribution

Competing interests

Supriyo De

Contribution

Competing interests

Jyoti Sen

Contribution

Competing interests

Nan-ping Weng

Contribution

Competing interests