Research Article: Leukocyte telomere length, T cell composition and DNA methylation age

Date Published: September 20, 2017

Publisher: Impact Journals LLC

Author(s): Brian H. Chen, Cara L. Carty, Masayuki Kimura, Jeremy D. Kark, Wei Chen, Shengxu Li, Tao Zhang, Charles Kooperberg, Daniel Levy, Themistocles Assimes, Devin Absher, Steve Horvath, Alexander P. Reiner, Abraham Aviv.


Both leukocyte telomere length (LTL) and DNA methylation age are strongly associated with chronological age. One measure of DNA methylation age-the extrinsic epigenetic age acceleration (EEAA)-is highly predictive of all-cause mortality. We examined the relation between LTL and EEAA. LTL was measured by Southern blots and leukocyte DNA methylation was determined using Illumina Infinium HumanMethylation450 BeadChip in participants in the Women’s Health Initiative (WHI; n=804), the Framingham Heart Study (FHS; n=909) and the Bogalusa Heart study (BHS; n=826). EEAA was computed using 71 DNA methylation sites, further weighted by proportions of naïve CD8+ T cells, memory CD8+ T cells, and plasmablasts. Shorter LTL was associated with increased EEAA in participants from the WHI (r=-0.16, p=3.1×10−6). This finding was replicated in the FHS (r=-0.09, p=6.5×10−3) and the BHS (r=−0.07, p=3.8x 10−2). LTL was also inversely related to proportions of memory CD8+ T cells (p=4.04×10−16) and positively related to proportions of naive CD8+ T cells (p=3.57×10−14). These findings suggest that for a given age, an individual whose blood contains comparatively more memory CD8+ T cells and less naive CD8+ T cells would display a relatively shorter LTL and an older DNA methylation age, which jointly explain the striking ability of EEAA to predict mortality.

Partial Text

Aging eludes precise definition at the systemic level and denotes a multitude of processes at the cellular level. Two of these processes-age-dependent telomere shortening [1] and DNA methylation (DNAm) profiles of cytosine phosphate guanines (CpGs) [2-4] have been used as indices of biological age. The age estimates resulting from multivariable regression models of DNAm profiles are referred to as ‘DNAm age’ or ‘epigenetic age’.

Major characteristics of participants from the WHI (the discovery cohort), the FHS and the BHS are displayed in Table S1 and Figure S1 (available as Supplementary data on line).

The two key observations of this study are: (a) LTL is inversely correlated with EEAA; and (b) the LTL-EEAA correlation largely reflects the proportions of imputed naïve and memory CD8+ T cell populations in the leukocytes from which DNA was extracted. These correlations were independently replicated in two well-characterized cohorts, providing confidence in their validity. To our knowledge, this is the first study showing association between LTL and a specific formulation of the epigenetic age, but only when it was weighted by the proportions of T naïve cells, T memory cells and plasmoblats (i.e., the EEAA). A previous study, using the Hannum formulation [3], showed no significant association between LTL and epigenetic age [14]. Overall, these findings might explain the ability of EEAA to predict all-cause mortality, given that EEAA captures not only leukocyte DNAm age but also a key aspect of immune senescence (principally naïve and memory T cells), which increases risks of a host of age-related diseases and of death [15].

Participants originated from the Women’s Health Initiative (WHI), the Framingham Heart Study (FHS) and the Bogalusa Heart Study (BHS); all signed informed consents approved by respective institutional review boards. All participants consented for the use of their DNA in genetic research. Analytic codes can be obtained from authors (BHC, CLC) upon request.




Leave a Reply

Your email address will not be published.