History of Epigenetic Clocks

Written by: Sabrina Sables

In 1984, Dr. Barry Marshall drank a broth containing Heliobacter pylori bacteria to prove that they cause stomach ulcers (14). While self-experimentation is usually frowned upon in the scientific process, this particular case led to a major breakthrough in the medical field. Self-experimentation is also how the epigenetic clocks started, albeit not as extreme as drinking ulcer-causing bacteria. Dr. Steve Horvath, the scientist who is responsible for the development of the epigenetic clocks, was a trial volunteer alongside his twin brother for the first clock developed at UCLA in 2011 (1). Self-experimentation isn’t exactly a surprise from Dr. Horvath, since he was interested in aging from a young age. He and his twin brother, along with a friend, made a pact to dedicate their careers to pursuing longevity (13). 

Before diving further into the epigenetic clocks, it is helpful to know two terms: DNA methylation and biological age. DNA methylation (DNA-m) is a biological process in which a methyl group is added to parts of a DNA molecule. This process does not change the sequence of the DNA, but it can change the activity surrounding it, such as gene expression. In recent years, DNA-m has been used as a marker for biological aging. Biological age can show how old the body is biologically as opposed to just chronological age. The measure of biological age using DNA-m as a marker is known as epigenetic age. Identical twins in particular are an interesting model for aging studies because they start with identical methylation patterns which can end up varying due to external factors such as environment (1). The first epigenetic predictor of age developed by Dr. Horvath had the accuracy of predicting an individual’s age within 5.2 years, on average. DNA methylation in 88 sites of the genome was correlated with biological age using saliva samples, with results shown in the figure below: 

Figure 1 Predicted versus observed age using DNA-m values from saliva samples (1). 

Figure 1 Predicted versus observed age using DNA-m values from saliva samples (1). 

Those 88 sites are parts of DNA known as CpGs, which are cytosines linked to guanine by a phosphate group. These are the specific sites where DNA methylation occurs and can be measured as a biomarker. Saliva samples were a good starting point, however, Dr. Horvath applied the original concept across most tissue and cell types. In 2013, he developed a multi-tissue predictor of age that allows the estimation of DNA-m age of most tissue and cell types, better known as the Horvath Pan-Tissue Clock (2). This clock uses 353 CpGs conserved across the mammalian genome, which give rise to a measure of epigenetic age based on levels of methylation, and allows the estimation of the age of an individual. DNA methylation was found by Dr. Horvath to have the following properties (2): it is close to zero for embryonic and induced pluripotent stem cells (iPSCs, cells turned into stem cells from adult cells), it correlates with cell passage number (number of times cells have been re-grown from a single culture), gives rise to a highly heritable measure of age acceleration, and is applicable to chimpanzee tissues. This clock had an accuracy of estimation within 3.6 years. It is also important to note that in 2015, the initial clock was corrected to reflect that cancerous tissue is not epigenetically older than normal tissue (3). These discoveries paved the way to be able to apply the epigenetic clock across most tissues, mammals, and for drug-screening studies.  

In the quest to get an even more accurate estimate of DNAm age, in 2018 the Skin & Blood Age clock was developed (4). Dr. Horvath and his team developed an accurate clock to apply to DNA-m age of cultured human skin cells, buccal cells, skin, blood and saliva samples. When used on fibroblasts (connective tissue cells) from Hutchinson Gilford Progeria Syndrome patients, for example, the Skin & Blood age clock accurately revealed an epigenetic age acceleration that was below the sensitivity levels of other existing clocks that use DNA-m as a biomarker. The Skin & Blood age clock is most accurate out of any epigenetic age predictors, as well as the most accessible since buccal swabs can be used with greater accuracy than before. The accuracy on cultured cells also opens the door for in-vitro studies involving drugs that could prolong lifespan by reducing DNA-m age and thus leading to a healthier aging process. A great example of this is the screening of the drug rapamycin, which affects the cellular mTOR pathway (5). The mTOR complex governs many pathways related to cellular metabolism. Rapamycin, a drug originally developed as an immunosuppressant, has been known to inhibit the mTOR kinase. The mTOR kinase is part of the complex which promotes cell growth, proliferation, and survival. Rapamycin has also been shown to extend the lifespan of different species, such as yeast, flies and mice (5). The Skin & Blood age clock was used to confirm the apparent anti-aging effect on cultured primary human keratinocytes. DNA methylation profiles from four passages of keratinocytes, with known numbers of population doubling were obtained and their ages were estimated by the Skin & Blood Age clock. Rapamycin appeared to slow the epigenetic aging of these cells regardless of cell senescence, proliferation, differentiation and telomere elongation (5). 

Cell senescence, proliferation, differentiation and telomere elongation are known markers of aging (6). Telomeres have especially been a topic of interest when studying aging. Telomeres are repetitive sequences of non-coding DNA at the end of chromosomes which protect from DNA damage. Every time a cell divides, telomeres shorten until a cell can no longer divide. When a cell can no longer grow and divide, it enters a period called senescence. It is thought that elongation of telomeres would prevent aging, however, this hypothesis assumes that senescence is the sole cause of aging. Telomeres and senescence, along with the other biomarkers mentioned above are but few pieces of the puzzle of biological aging. DNA methylation is another important piece of that puzzle. Interestingly, epigenetic aging was shown to continue in cells which were manipulated to bypass senescence (6). While bypassing senescence and thus having longer telomeres may create a younger phenotype, epigenetic aging still goes on unprevented by increased telomere expression. It is possible that to halt the aging process, treatments would need to target epigenetic aging and reduce it.  The availability of an accurate clock in vitro (Skin & Blood Age) can pave the way for the large-scale testing of compounds, conditions and existing “anti aging” treatments and how they affect epigenetic aging. This kind of research, along with what is known about other biomarkers like telomeres and senescence, can advance the understanding of the biological process of aging and perhaps uncover the molecular pathways involved. For example, the effects of telomere maintenance, cellular senescence and cellular proliferation have been found to be unrelated to epigenetic ageing. On the other hand, inhibition of mTOR by rapamycin, caloric restriction, hypoxia, and growth hormone appear to be related to epigenetic aging (8). 

One of the most exciting features of DNAm biomarkers is that epigenetic aging is reversible (7), raising the prospect that DNAm age estimates might be very useful for identifying or validating anti-aging interventions. The epigenetic clock theory of aging “views biological ageing as an unintended consequence of both developmental and maintenance programmes for which the molecular footprints give rise to DNAm age estimators.” (7) This means that biological aging happens as a side effect of normal developmental processes across the lifespan, and DNA methylation measures provide a “footprint” for tracking this effect. 

The ability to reverse epigenetic age would in theory increase an individual’s healthspan, or lifespan spent healthy. Epigenetic aging predictors would be a valuable tool for achieving this across the human population.  To enable this, newer DNAm based measurements were also developed (7) that were trained on large populations to predict human mortality.  These predictors benefit from samples that have been banked for decades and where long-outcomes data is available for trial volunteers.  

For instance, the phenotypic age estimator or DNAm PhenoAge, stands out because of its predictive accuracy for time to death, its association with smoking status and its association with various markers of senescence. Hannum’s clock is accurate for blood tissues and uses the least CpGs, at 71. The issue these measurements have is that they lead to bias in non-blood tissues and in children, which the Horvath pan-tissue clock can overcome. The figure below from a review article (7) highlighting these technologies provides a comparison of each by showing what they correlate with the most, and where they may fall short: 

Figure 2: Comparison of different DNA-m based biomarkers of aging. The Horvath pan-tissue clock stands out because it is most correlated with chronological age across tissue types, is accurate in children, correlates with gestational age, among other attributes shown in the figure. “AA” denotes age acceleration. (7)

A large-scale meta analysis has shown that epigenetic age can indeed predict lifespan (9). The study “strengthens the evidence that epigenetic age predicts all cause mortality above and beyond chronological age and traditional risk factors…individuals whose epigenetic age was greater than their chronological age (i.e., individuals exhibiting epigenetic “age acceleration”) were at an increased risk for death from all causes, even after accounting for known risk factors.” (9). A second generation clock called GrimAge is currently the best epigenetic clock for predicting mortality risk (10). This clock was constructed using seven DNAm based estimators of plasma proteins including those of plasminogen activator 1 and growth differentiation factor 15. Aside from these proteins, GrimAge is also based on smoking pack years, since smoking is a known contributor to accelerating mortality. GrimAge stands out because it can predict time-to-death, time-to coronary heart disease, time-to-cancer, among other factors that accelerate epigenetic age and thus time to mortality. GrimAge could prove to be useful when conducting epidemiological studies to predict health outcomes and intervene to decelerate epigenetic aging. The overall goal of reversing epigenetic age is to enable the population to age with dignity and have an increased healthspan. 

Aside from exciting in-vitro studies and population studies about aging, the future of epigenetic clocks will likely be applied to all species. Having a reliable clock available for different species can allow for applications such as pet health monitoring, wildlife conservation, and more. Dr. Horvath has over the years developed successful clocks for different species, and has realized that it is difficult to ignore that aging may have a defined and shared mechanism in conserved areas of the genome across mammalian species. In an article currently undergoing review, Dr. Horvath and his team highlight their most recent effort to achieve this as a universal mammalian epigenetic clock. They generated 10,000 methylation arrays, each profiling up to 37,000 CpGs in highly-conserved stretches of DNA from over 59 tissue types from 128 mammalian species (11). This is one of the most ambitious epigenetic clock projects to date. Specific CpGs have been identified, with methylation levels that change with age across mammalian species. These findings may support the hypothesis that aging is evolutionarily conserved and coupled to developmental pathways across all mammalian species. 

To further prove that the future of epigenetic clocks is sound, independent studies (12) have shown that epigenetic age and the clocks are the most promising predictors of biological age. DNAm and other markers of biological age can be used to monitor an increasingly aging global population, with the goal of increasing healthspan and decreasing healthcare burdens. 

Works Cited

1. Bocklandt S, Lin W, Sehl ME, Sánchez FJ, Sinsheimer JS, Horvath S, Vilain E. Epigenetic predictor of age. PLoS One. 2011;6(6):e14821. doi: 10.1371/journal.pone.0014821. Epub 2011 Jun 22. PMID: 21731603; PMCID: PMC3120753.

2. Horvath, S. DNA methylation age of human tissues and cell types. Genome Biol 14, 3156 (2013). https://doi.org/10.1186/gb-2013-14-10-r115

3. Horvath, S. Erratum to: DNA methylation age of human tissues and cell types. Genome Biol 16, 96 (2015). https://doi.org/10.1186/s13059-015-0649-6

4. Horvath S, Oshima J, Martin GM, Lu AT, Quach A, Cohen H, Felton S, Matsuyama M, Lowe D, Kabacik S, Wilson JG, Reiner AP, Maierhofer A, Flunkert J, Aviv A, Hou L, Baccarelli AA, Li Y, Stewart JD, Whitsel EA, Ferrucci L, Matsuyama S, Raj K. Epigenetic clock for skin and blood cells applied to Hutchinson Gilford Progeria Syndrome and ex vivo studies. Aging (Albany NY). 2018 Jul 26;10(7):1758-1775. doi: 10.18632/aging.101508. PMID: 30048243; PMCID: PMC6075434.

5. Horvath S, Lu AT, Cohen H, Raj K. Rapamycin retards epigenetic ageing of keratinocytes independently of its effects on replicative senescence, proliferation and differentiation. Aging (Albany NY). 2019;11(10):3238-3249. doi:10.18632/aging.101976

6. Kabacik S, Horvath S, Cohen H, Raj K. Epigenetic ageing is distinct from senescence-mediated ageing and is not prevented by telomerase expression. Aging (Albany NY). 2018;10(10):2800-2815. doi:10.18632/aging.101588

7. Horvath, S., Raj, K. DNA methylation-based biomarkers and the epigenetic clock theory of ageing. Nat Rev Genet 19, 371–384 (2018). https://doi.org/10.1038/s41576-018-0004-3

8. Raj K, Horvath S. Current perspectives on the cellular and molecular features of epigenetic ageing. Exp Biol Med (Maywood). 2020 Nov;245(17):1532-1542. doi: 10.1177/1535370220918329. Epub 2020 Apr 10. PMID: 32276545; PMCID: PMC7787550.

9. Chen BH, Marioni RE, Colicino E, Peters MJ, Ward-Caviness CK, Tsai PC, Roetker NS, Just AC, Demerath EW, Guan W, Bressler J, Fornage M, Studenski S, Vandiver AR, Moore AZ, Tanaka T, Kiel DP, Liang L, Vokonas P, Schwartz J, Lunetta KL, Murabito JM, Bandinelli S, Hernandez DG, Melzer D, Nalls M, Pilling LC, Price TR, Singleton AB, Gieger C, Holle R, Kretschmer A, Kronenberg F, Kunze S, Linseisen J, Meisinger C, Rathmann W, Waldenberger M, Visscher PM, Shah S, Wray NR, McRae AF, Franco OH, Hofman A, Uitterlinden AG, Absher D, Assimes T, Levine ME, Lu AT, Tsao PS, Hou L, Manson JE, Carty CL, LaCroix AZ, Reiner AP, Spector TD, Feinberg AP, Levy D, Baccarelli A, van Meurs J, Bell JT, Peters A, Deary IJ, Pankow JS, Ferrucci L, Horvath S. DNA methylation-based measures of biological age: meta-analysis predicting time to death. Aging (Albany NY). 2016 Sep 28;8(9):1844-1865. doi: 10.18632/aging.101020. PMID: 27690265; PMCID: PMC5076441.

10. Lu AT, Quach A, Wilson JG, Reiner AP, Aviv A, Raj K, Hou L, Baccarelli AA, Li Y, Stewart JD, Whitsel EA, Assimes TL, Ferrucci L, Horvath S. DNA methylation GrimAge strongly predicts lifespan and healthspan. Aging (Albany NY). 2019 Jan 21;11(2):303-327. doi: 10.18632/aging.101684. PMID: 30669119; PMCID: PMC6366976.

11. Universal DNA methylation age across mammalian tissues. MAMMALIAN METHYLATION CONSORTIUM, Ake T. Lu, Zhe Fei, Amin Haghani, Todd R. Robeck, Joseph A. Zoller, Caesar Z. Li, Joshua Zhang, Julia Ablaeva, Danielle M. Adams, Javier Almunia, Reza Ardehali, Adriana Arneson, C. Scott Baker, Katherine Belov, Pete Black, Daniel T. Blumstein, Eleanor K. Bors, Charles E. Breeze, Robert T. Brooke, Janine L. Brown, Alex Caulton, Julie M. Cavin, Ioulia Chatzistamou, Hao Chen, Priscila Chiavellini, Oi-Wa Choi, Shannon Clarke, Joseph DeYoung, Christopher Dold, Candice K. Emmons, Stephan Emmrich, Chris G. Faulkes, Steven H. Ferguson, Carrie J. Finno, Jean-Michel Gaillard, Eva Garde, Vadim N. Gladyshev, Vera Gorbunova, Rodolfo G. Goya, Matthew J Grant, Erin N. Hales, M. Bradley Hanson, Martin Haulena, Andrew N. Hogan, Carolyn J. Hogg, Timothy A. Hore, Anna J. Jasinska, Gareth Jones, Eve Jourdain, Olga Kashpur, Harold Katcher, Etsuko Katsumata, Vimala Kaza, Hippokratis Kiaris, Michael S. Kobor, Pawel Kordowitzki, William R. Koski, Brenda Larison, Sang-Goo Lee, Ye C. Lee, Marianne Lehmann, Jean-Francois Lemaitre, Andrew J. Levine, Cun Li, Xinmin Li, David TS Lin, Nicholas Macoretta, Dewey Maddox, Craig O. Matkin, Julie A. Mattison, June Mergl, Jennifer J. Meudt, Khyobeni Mozhui, Asieh Naderi, Martina Nagy, Pritika Narayan, Peter W. Nathanielsz, Ngoc B. Nguyen, Christof Niehrs, Alexander G. Ophir, Elaine A. Ostrander, Perrie O’Tierney Ginn, Kim M. Parsons, Kimberly C. Paul, Matteo Pellegrini, Gabriela M. Pinho, Jocelyn Plassais, Natalia A. Prado, Benjamin Rey, Beate R. Ritz, Jooke Robbins, Magdalena Rodriguez, Jennifer Russell, Elena Rydkina, Lindsay L. Sailer, Adam B. Salmon, Akshay Sanghavi, Kyle M. Schachtschneider, Dennis Schmitt, Todd Schmitt, Lars Schomacher, Lawrence B. Schook, Karen E. Sears, Andrei Seluanov, Dhanansayan Shanmuganayagam, Anastasia Shindyapina, Kavita Singh, Ishani Sinha, Russel G. Snell, Elham Soltanmaohammadi, Matthew L. Spangler, Lydia Staggs, Karen J. Steinman, Victoria J. Sugrue, Balazs Szladovits, Masaki Takasugi, Emma C. Teeling, Michael J. Thompson, Bill Van Bonn, Sonja C. Vernes, Diego Villar, Harry V. Vinters, Mary C. Wallingford, Nan Wang, Robert K. Wayne, Gerald S. Wilkinson, Christopher K. Williams, Robert W. Williams, X. William Yang, Brent G. Young, Bohan Zhang, Zhihui Zhang, Peng Zhao, Yang Zhao, Joerg Zimmermann, Wanding Zhou, Jason Ernst, Ken Raj, Steve Horvath. bioRxiv 2021.01.18.426733; doi: https://doi.org/10.1101/2021.01.18.426733

12. Jylhävä J, Pedersen NL, Hägg S. Biological Age Predictors. EBioMedicine. 2017 Jul;21:29-36. doi: 10.1016/j.ebiom.2017.03.046. Epub 2017 Apr 1. PMID: 28396265; PMCID: PMC5514388.

13. Gibbs, W. Biomarkers and ageing: The clock-watcher. Nature 508, 168–170 (2014). https://doi.org/10.1038/508168a

14. Ahmed N. 23 years of the discovery of Helicobacter pylori: is the debate over?. Ann Clin Microbiol Antimicrob. 2005;4:17. Published 2005 Oct 31. doi:10.1186/1476-0711-4-17