Epigenetic age prediction and rejuvenation

Abstract

Ageing is a complex, multi-faceted process that afflicts all humans. It invariably increases susceptibility to a range of diseases such as cancer and neurological disorders. Drugs that mimic calorie restriction show promise in slowing down ageing, but very few treatments appear to be able to actively reverse ageing. Partially reprogrammed stem cells have shown potential as an anti-ageing therapy when used to safely rejuvenate mice without tumour incidence. The question remained as to what exactly occurred at a cellular level. Were a subpopulation of cells dedifferentiating, or partially dedifferentiating and causing a rejuvenative effect by being more stem-like? Or, were the cells epigenetically rejuvenated, where cells became more youthful without loss of somatic cell identity? To test either of these hypotheses, two biomarkers were required to track (i) biological ageing and (ii) dedifferentiation state.

By analysing a previously published dataset of fibroblasts dedifferentiating to induced pluripotent stem cells (iPSCs) over a 49-day time-course, I helped assess the dynamics of cellular ageing. Epigenetic age was used a proxy for biological age, while RNA microarray data was used to assess the state of dedifferentiation (ie. by comparing fibroblast specific gene expression with pluripotency gene markers). Partially reprogrammed cells (between days 7 and 15 of dedifferentiation) declined in predicted age (also known as epigenetic age, eAge), while somatic cell identity was maintained. This shows that loss of somatic gene expression and epigenetic age follow different kinetics, suggesting that they can be uncoupled and a possible “safe period” exists where rejuvenation can be achieved with a minimized risk of cancer.

While epigenetic clocks appear to confer biological age in many respects, their true underlying function remains a mystery, and the precise aspects of ageing they capture is unclear. For example, differences between epigenetic age and chronological age that are associated with ageing disease states, could be caused by biological and technical biases. Biological biases can arise from mutations affecting the DNA methylation machinery, resulting in global sweeps in methylation. Technical biases may arise from errors in bisulphite (BS) conversion, which could cause a slight overestimation in percentage methylation and therefore alter eAge estimates.

To explore how robust the epigenetic clocks are to sweeps of global methylation, incremental increases and decreases of global methylation were simulated in a large cohort. I showed that epigenetic clocks are not impervious to gradual, global changes in methylation. I also showed how discrete alterations in methylation state can cause a significant difference in eAge compared to a control group, which conceivably could occur in experiments testing rare genetic diseases.

I also present an epigenetic clock based on average methylation over genomic regions, rather than individual CpGs. This clock provides a more robust method of predicting age, which may pave the way for more accurate age predictors using mouse RRBS data.

This thesis has demonstrated that epigenetic clocks are invaluable tools for exploring health-span extending therapies. However, caution must be taken when analysing epigenetic data, as mutations and technical issues may confound analysis. Nonetheless, epigenetic clocks have shown great potential in the molecular ageing field. By understanding the precise nature of eAge, avenues to achieve therapeutic anti-ageing therapies may also be achieved.