In a preprint published in bioRxiv, Morgan Levine and colleagues have identified and grouped 5,717 epigenetic CpG sites into 12 different modules, conducting an in-depth examination into how epigenetic clocks work.
What is a CpG site?
Methylation cannot occur everywhere in the genome. The researchers explain that for a mammalian gene to be silenced through methylation, a methyl group must bind to a specific section of DNA, which consists of cytosine followed by guanine (two of the DNA ‘letters’) and is known by the shorthand CpG . The genome then wraps around that methyl group, thus making it inaccessible .
The researchers selected these particular 5,717 sites because they are well researched and their methylation states are known to change with aging; they appear in 15 different epigenetic clocks along with five new tissue-specific clocks that the researchers developed by selecting CpGs that were common between mice and people..
Then, they grouped these CpG sites based on their behaviors during both aging and epigenetic reprogramming. Methylation does not affect all CpG sites in the same way; different sites age at different rates. Sites that acted similarly were given the same ‘color’ as a means of identification.
A rainbow of reprogramming analysis
The researchers found that epigenetic reprogramming through administration of the Yamanaka factors did not entirely reset the epigenetic clock back to a youthful state. The researchers determined.that this did occur in many regions, particularly the green-yellow module, which lost methylation with aging and gained methylation upon reprogramming. This module consisted mostly of sites that had a low density of CpG regions.
The light blue behaved in a reverse, but expected, manner; sites there gained methylation with age and lost methylation with reprogramming. However, the cyan module went completely against the pattern. Sites in this module that lost methylation with aging also lost methylation with reprogramming.
The researchers also contended with technical concerns. Locations that were almost always methylated, or almost never methylated, were difficult to analyze as moving in one direction or the other. The researchers claim that due to the imprecision of current analysis techniques, any signal got lost in the noise.
Analyzing clocks with this new information
With their module analysis in hand, the researchers then turned their eyes towards current clocks, both old and new. They found that, while all of these clocks contain multiple different modules, the more advanced clocks are more strongly weighted towards CpGs that change more dramatically with age. The researchers suggest that many older clocks are weaker because they contain CpG sites that are less strongly, or even inversely, correlated with aging and mortality.
The researchers also analyzed other information associated with these clocks and found unsurprising results. For example, according to the GrimAge clock, smoking strongly affected CpG sites that the researchers placed in the yellow and red modules, and GrimAge also strongly associated epigenetic alterations in the yellow module with all-cause mortality.
There were also differences between tumor and normal cells. While many clocks have had trouble distinguishing from normal cells and cells undergoing uncontrolled division, this study found that cancerous, rapidly dividing cells increase their epigenetic aging in the navy and light blue sites.
As with other preprints published on bioRxiv, this paper hasn’t been peer reviewed. However, this isn’t research into whether or not a potential therapeutic works in an animal or cellular model; this is a paper that describes how DNA methylation sites can be grouped in order to better understand epigenetic aging and create more descriptive, more accurate measurements of it. As such, it is best understood as a look into the inner workings of how epigenetic clocks are developed and what makes these clocks tick.
However, if one thing can be clearly taken from this study, it is that epigenetic changes with aging are much, much more complicated than “younger” or “older”. Going beyond similarity-based, color-coded modules in order to understand which of these thousands of changes are harmless and which changes are ultimately dangerous is obviously a tremendous challenge; however, it is one that might be necessary to undertake.
 Jabbari, K., & Bernardi, G. (2004). Cytosine methylation and CpG, TpG (CpA) and TpA frequencies. Gene, 333, 143-149.
 Derreumaux, S., Chaoui, M., Tevanian, G., & Fermandjian, S. (2001). Impact of CpG methylation on structure, dynamics and solvation of cAMP DNA responsive element. Nucleic acids research, 29(11), 2314-2326.