Ageing DNAme signatures. Public deliverable 8.2

The epigenome is at the interface between intrinsic (genetic) and external

(environmental) factors influencing cellular behaviour. DNA methylation, defined as

the addition of methyl groups on the C5 position of cytosines on the DNA in the

context of CpGs, is a key part of the epigenome which integrates these intrinsic and

external factors and records them as a signature on the DNA. DNA methylation is

also influenced by cellular ageing and forms the basis of several existing “epigenetic

clocks” which utilise a combination of age-associated differentially methylated

positions to predict chronological age. Many age-related DNA methylation changes

are common across most tissue types. Whether discordant tick rates between cell

type-specific and general epigenetic clocks play a role in health or disease has

remained largely unexplored. Recently we reported that cervical samples, consisting

of a mixture of hormone-sensitive epithelial and immune cells, are a suitable

surrogate tissue for prediction of breast cancer and may reflect systemic epigenetic

programming defects. We therefore aimed to investigate if cell type-specific ageing

signatures in cervical samples are associated with breast cancer risk.

Using DNA methylation analysis of cervical samples, we developed the WID

(Women’s cancer risk IDentification) general clock, shared by both epithelial and

immune cells and optimised for cervical samples. Surprisingly, the WID general clock

outperforms two commonly used epigenetic clocks for age prediction in several

tissues. Via computational inference of sample composition, we also identified

epithelial- and immune-specific epigenetic clocks. A discordance between

WID-general and epithelial age (WID-relative-epithelial-age, WID-REA) in cervical

samples is associated with breast cancer. The same effect is observed in breast

tissue from individuals at increased risk of breast cancer (BRCA1/2 mutation carriers)

but could be rescued via risk-reducing pharmacological intervention in a clinical

study. Interestingly, the WID-REA also differs between women who received hormone

replacement therapy (HRT) and did or did not develop breast cancer, indicating a

potential to monitor individual breast cancer risk on HRT using the WID-REA.

We therefore provide novel general and tissue-specific epigenetic clocks which are

accurate at predicting chronological age and may further be an informatic measure of

disease risk.