Age-related somatic evolution of clonal haematopoiesis

Abstract

Over a lifetime, human cells continually acquire mutations, some of which may alter cell division's complex homeostasis and lead to the subsequent expansion of somatic clones. Such expansions are frequent in the haematopoietic system and become detectable as we age.

Haematopoiesis is a complex and hierarchical system that generates millions of functionally diverse cells daily. This multi-tiered system allows for the rapid regeneration of our blood system in response to stress whilst protecting the pool of long-lived haematopoietic stem and progenitor cells (HSPCs) from excessive replicative stress. Haematopoiesis can function with high fidelity for many decades but is inevitably challenged by ageing and the time-dependent accumulation of somatic variation.

Clonal Haematopoiesis of Indeterminate Potential (CHIP) is defined as the expansion of HSPCs in healthy-aged individuals that results from genetic alterations. Although mostly inconsequential, the constant rate of the acquisition of mutations in HSPCs (17 mutations/year) leads to an increasing probability, with respect to age, of a variant occurring in a gene that dysregulates the tightly maintained mechanism of haematopoiesis. In healthy individuals, the differentiated cells that comprise our blood are the progeny of equally contributing stem cells that produce a genetically diverse, polyclonal population. CHIP, however, is marked by the population of blood cells becoming increasingly dominated by single (or multiple) genetic clone(s) that are genotypically identical.

In 2014, several independent studies confirmed that CHIP is a condition that increases with age: more than 10% of the population over 65 years are affected, with a prevalence that increases dramatically over subsequent decades. It has been associated with all-cause mortality, cardiovascular disease and haematological malignancies – a risk that scales with clone size.

Observing the relationship between CHIP and age-related pathologies, we sought to test the relationship between clonal haematopoiesis and ageing using a range of published epigenetic clocks (which use DNA methylation states to predict biological age) to assess any association with biological ageing. In Robertson et al., we characterised the landscape of somatic mutations in a range of core haematopoietic marker genes in the Lothian Birth Cohorts (LBCs). The LBCs are two parallel studies of ageing that consist of individuals over 70 and 79 years, in LBC36 and LBC21, respectively. We observed a significant association with biological ageing in several published cell-intrinsic clocks in participants that harboured a mutation in one of the six most prevalent CHIP genes versus our control group. CHIP status was associated with a significant increase in Horvath age acceleration: with an increase of 4.5 (SE 0.9) years in LBC1936 and 3.7 (SE 1.2) years in LBC1921 (p = 2.3 x 10-6 and 2.5 x 10-3, respectively). In addition, we note significant epigenetic age accelerations in both DNMT3A and TET2 in isolation – the most commonly affected genes in clonal haematopoiesis. This result might indicate that CHIP is either driven or a driver of systemic ageing, explaining its links to non-haematological age-dependent pathologies.

A triptych of fundamental forces shape evolution: mutation, drift and selection. Whilst the first two are essentially stochastic processes, the third is the driving force: aiming to maximize fitness within an environment. Currently, it’s not understood whether mutations in differing CHIP genes lead to distinct fitness advantages that would lead to patient stratification. Since mutations in HSPCs often instigate leukaemia, we hypothesize that HSC fitness substantially contributes to the transformation to disease states. We again leverage the LBCs, using a particularly unique aspect of their curation - the collection of peripheral blood over 12 years of later life - to develop a longitudinal assay for HSPC fitness using error-corrected sequencing. We then create a novel method we call the likelihood-based filter for time series data (LiFT) to determine fitness effects across our longitudinal data, quantifying the growth potential of somatic mutations within each participant. This approach discriminates naturally drifting populations of cells that typically harbour synonymous or non-functional variants and those that harbour driver mutations that give rise to rapidly growing clones. We characterise the fitness effects of mutations in many known CHIP driver genes and observe that differences in gene-specific fitness effects outweigh inter-individual variation, which could constitute a new method of personalised clinical management.

This work has shown that CHIP confers a strong association with biological ageing through the prism of epigenetic clocks. Furthermore, we have begun characterising the fitness effects of genes that characterise CHIP whilst beginning to understand the molecular mechanisms behind their distinct fitness effects. We hope this should aid in a greater understanding of the pathogenesis of CHIP and assist in the improved stratification of patients.