DNA methylation-based estimator of telomere length.
Lu AT, Seeboth A, Tsai PC, Sun D, Quach A, Reiner AP, Kooperberg C, Ferrucci L, Hou L, Baccarelli AA, Li Y, Harris SE, Corley J, Taylor A, Deary IJ, Stewart JD, Whitsel EA, Assimes TL, Chen W, Li S, Mangino M, Bell JT, Wilson JG, Aviv A, Marioni RE, Raj K, Horvath S. · 2019
Read DNA methylation-based estimator of telomere length. by Lu AT, Seeboth A, Tsai PC, Sun D, Quach A, Reiner AP, Kooperberg C, Ferrucci L, Hou L, Baccarelli AA, Li Y, Harris SE, Corley J, Taylor A, Deary IJ, Stewart JD, Whitsel EA, Assimes TL, Chen W, Li S, Mangino M, Bell JT, Wilson JG, Aviv A, Marioni RE, Raj K, Horvath S. (2019) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc6738410.txt.
Introduction
Lu and co-authors built a DNA-based estimator of telomere length. Telomeres are the caps on chromosome ends, and their length tracks with aging diseases. The new estimator reads 140 chemical mark sites on DNA in white blood cells. It works across the full human age span.
The estimator tracked age far more strongly than directly measured telomere length. It also beat direct measures in predicting time to death, heart disease, and heart failure. Links to smoking history were stronger too. Validation used about 10,000 samples from seven large studies. Cell experiments suggest the marks partly record how often cells have divided, not length alone.
Catalog record
| CREATOR | Lu AT, Seeboth A, Tsai PC, Sun D, Quach A, Reiner AP, Kooperberg C, Ferrucci L, Hou L, Baccarelli AA, Li Y, Harris SE, Corley J, Taylor A, Deary IJ, Stewart JD, Whitsel EA, Assimes TL, Chen W, Li S, Mangino M, Bell JT, Wilson JG, Aviv A, Marioni RE, Raj K, Horvath S. |
|---|---|
| YEAR | 2019 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 88 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
DNA methylation-based estimator of telomere length By Ake T Lu, Anne Seeboth, Pei-Chien Tsai, Dianjianyi Sun, Austin Quach, Alex P Reiner, Charles Kooperberg, Luigi Ferrucci, Lifang Hou, Andrea A Baccarelli, Yun Li, Sarah E Harris Published in: Aging (Albany NY) DOI: 10.18632/aging.102173 ABSTRACT Abstract Telomere length (TL) is associated with several aging-related diseases. Here, we present a DNA methylation estimator of TL (DNAmTL) based on 140 CpGs. Leukocyte DNAmTL is applicable across the entire age spectrum and is more strongly associated with age than measured leukocyte TL (LTL) (r ~-0.75 for DNAmTL versus r ~ -0.35 for LTL). Leukocyte DNAmTL outperforms LTL in predicting: i) time-to-death (p=2.5E-20), ii) time-to-coronary heart disease (p=6.6E-5), iii) time-to-congestive heart failure (p=3.5E-6), and iv) association with smoking history (p=1.21E-17). These associations are further validated in large scale methylation data (n=10k samples) from the Framingham Heart Study, Women's Health Initiative, Jackson Heart Study, InChianti, Lothian Birth Cohorts, Twins UK, and Bogalusa Heart Study. Leukocyte DNAmTL is also associated with measures of physical fitness/functioning (p=0.029), age-at-menopause (p=0.039), dietary variables (omega 3, fish, vegetable intake), educational attainment (p=3.3E-8) and income (p=3.1E-5). Experiments in cultured somatic cells show that DNAmTL dynamics reflect in part cell replication rather than TL per se. DNAmTL is not only an epigenetic biomarker of replicative history of cells, but a useful marker of age-related pathologies that are associated with it. Keywords: telomere length, DNA methylation, molecular biomarker, aging INTRODUCTION Telomeres are repetitive nucleotide sequences at the end of chromosomes that shorten with replication of somatic cells. Since the number of cell replication in vivo increases with age, telomere length (TL) is negatively correlated with age of proliferating somatic cells. Meta-analysis of 124 cross-sectional studies and 5 longitudinal studies showed that the correlation between leukocyte telomere length (LTL) and age ranges between r=-0.295 and r=-0.338 across adults [1]. TL variation within somatic tissues of the individual is much smaller than that between individuals. Within the individual, TL variation across somatic tissues such as blood, skin, muscle and fat largely reflects their replicative history prior to adulthood, given that the rates of TL shortening in these tissues are similar during adulthood [2]. Shorter LTL is associated with cardiovascular disease, psychological stress, and lifespan [3–10]. Another DNA-based biomarker that changes with age is methylation of cytosine residues of cytosine-phosphate-guanine dinucleotides (CpGs). Machine learning-based analyses of these changes generated algorithms, known as epigenetic clocks that use specific CpG methylation levels to estimate age (i.e., DNAm age) [11–14] and/or physiological age [15–17]. Although both DNAm age and LTL are associated with chronological age, they exhibit only weak correlations with each other after adjusting for age [18–20], suggesting the distinct nature of their underlying mechanisms. DNA methylation assays are already highly robust and ready for biomarker development [21]. By contrast, despite two decades of population-based telomere research, the measurement of TL remains challenging and can be subject to technical confounding factors including but not limited to different methods of DNA extraction [22–24]. Furthermore, the terminal restriction fragments (TRFs), measured by Southern blotting, the accepted ‘gold standard’ of TL measurements, include not only the canonical region of telomeres but also the potentially variable sub-telomeric region [22,25]. It would be ideal if the robustness inherent in DNA methylation analyses can be extended to TL measurement. Although there are reports of TL-related DNA methylation changes [26], it was unknown whether these reflect actual TL or associated biological features, including health outcomes. We present here a novel DNAm TL estimator (DNAmTL) based on methylation profiles of 140 CpGs. This epigenetic biomarker was developed by regressing measured LTL on blood methylation data from n=2,256 individuals (training set). We show that DNAmTL correlates negatively with age in different tissues and cell types and outperforms TRF-based LTL in predicting mortality and time-to-heart disease, as well as being associated with smoking history and other age-related conditions. We also validated the applicability of DNAmTL on a large-scale data set (N=9,345) and uncovered associations between age-adjusted DNAmTL with diet and clinical biomarkers. Monitoring cultured cells with or without telomerase revealed that DNAmTL records cell replication independently of telomere attrition. RESULTS Training and validation data from 3 cohorts In stage 1 of our project, we evaluated data from n=3,334 individuals for whom both LTL and Illumina methylation array data were available. These were from three different studies: Framingham Heart Study offspring cohort (FHS, N=878), Women’s Health Initiative (WHI, N=818) and Jackson Heart Study cohort (JHS, N=1638, Table 1 and Supplementary Note 1). The same laboratory measured LTL by Southern blotting of the terminal restriction fragments [25]. DNA methylation levels were measured in different labs using the Illumina Infinium methylation array platform. [Table: Overview of training and test data.] An overview of the data sets is found in Table 1. These US cohorts were comprised of two ethnic groups: 41% of European ancestry and 59% of African Ancestry. The age of the individuals ranged from 22 to 93 years. The training set used for constructing DNAmTL was comprised of N=2,256 individuals from the WHI and JHS cohorts for whom LTL and DNAm data were assessed from the same blood sample (collected at the same time). Although fewer than 20% of individuals in the training set were of European ancestry, our test data demonstrated that the resulting DNAmTL estimator applied equally well to individuals of European ancestry. We used two test data sets. The first test data set involved N=1,078 individuals comprised of N=100 from the WHI, N=100 from JHS, and N=878 from the FHS cohorts. The second test data set was collected in stage 2 of our analysis: it involved N=9,815 DNA methylation samples from additional cohorts (Bogalusa, Twins UK, Lothian Birth cohorts, InCHIANTI) to evaluate correlations between LTL and numerous age-related conditions and lifestyle factors. We also evaluated DNAmTL in publicly-available data from adipose tissue (N=648 from the Twins UK study [27,28]), liver (N=85) [28,29], and monocytes (n=1264 from the Multi-Ethnic Study of Atherosclerosis) [30]. Finally, we tested DNAmTL in in vitro studies to ascertain its applicability to cultured cells and to probe the nature of DNAmTL’s association with TL. Additional details of these studies can be found in Supplementary Notes 1 and 2. DNAmTL versus measured TL in blood and adipose tissue We restricted the analysis to CpGs that are present on both the Illumina Infinium 450K array and the Illumina EPIC methylation array (Methods). Using the training data (n=2,256), we regressed measured LTL (mean TRFs) on blood CpG methylations using an elastic net regression model [31]. This resulted in the automatic selection of 140 CpGs whose methylation levels best-predicted LTL (Supplementary data 1). The linear regression model allows a direct prediction of TL based on DNA methylation levels. The predicted TL value, also referred to as DNAmTL, possesses the same units (kilobase) as that of mean TRF. The correlation coefficient between DNAmTL and LTL in the training data was r=0.63, which was overly optimistic, as subsequent independent validation with test data sets produced lower correlations of r>0.40 (Figure 1). Further, using 12 large validation data sets, we found that the correlation between DNAmTL and LTL ranged from r=0.38 to r=0.5 (last column of Table 1) with the exception of the Lothian Birth cohorts (where it was close to zero).