Epigenetics is the study of mechanisms that change the pace at which proteins are produced from the blueprints encoded in DNA, a process known as gene expression. The first step in gene expression is the generation of RNA, which is what is actually usually measured. Rates of production - and thus numbers of specific RNA and protein molecules in circulation - are the switches and dials that control cellular processes. These rates of production change constantly, as the various processes of production and activities of molecular machinery interact with one another in response to internal and external circumstances. Researchers have in recent years discovered that they can identify more stable patterns of changes in the noise, patterns that correspond to age. This is promising on many levels. On the practical front it is a path to biomarkers of aging that accurately reflect biological rather than chronological age, and thus can be used to greatly speed up the assessment and develoment of rejuvenation therapies. When it comes to fundamental research, patterns of epigenetic alterations to the production of biomolecules are another tool by which the course of aging and its mechanisms can be mapped.
It is fairly well established that most natural genetic variations are only important to health and mortality in later life, and even then only collectively. Individual genetic variants have tiny effects and very few are consistent across study populations. Young people see no measurable impact, but when old, with a high level of cell and tissue damage, many gene variants grant small, differing levels of protection or accommodation of damage. Obviously these are not large effects on the whole: the best of combinations raises the odds of living an extremely long life from minuscule to merely tiny, and beneficiaries are still frail and dependent at the end, crushed by high levels of damage in the biological machinery that sustains life.
If genetic variants are only really important in later life, then should we also expect the pace of change in epigenetic patterns and rates of gene expression to be much higher in later life? At the very high level, and even from simple metrics like activity, skin elasticity, and grip strength, we know that aging isn't a linear process. It is a downward spiral that proceeds ever faster as damage feeds on damage, and the final collapse into terminal ill health at the end of life is often a rapid thing after years of much slower decline. It would be surprising if measures that truly reflect the progression of mechanisms of aging turned out to show something different. In this paper, the researchers look at the RNA levels resulting from gene expression, and use statistical methods to mark the ages at which more significant changes start, going gene by gene to build the beginning of a picture. They find that most of the identifiable and robust changes occur in old age.
Genome-wide alterations in RNA expression profiles are age-associated. Yet the rate and temporal pattern of those alterations are poorly understood. Most often, age-associated physiological and molecular alterations are extracted using linear regression models. Linear regression assumes a constant change over time and therefore might be appropriate for organisms that aged over a short period. In humans, however, adulthood spans from 50 to 80 years. It is very unlikely that the rate of age-associated changes progresses at a constant rate. The fitness of different regression models to describe age-associated physiological features demonstrated that a quadratic or a parabolic regression model are most suitable to describe age-associated changes. Quadratic models have fewer assumptions compared with a linear model. Moreover, a quadratic model could be employed to identify the age when major changes occur (named here age-position). Using cross-sectional transcriptome studies, it was suggested that most transcriptional alterations in brain frontal cortex occur around the age of 42, and in Vastus lateralis muscle major changes occur already in the fourth decade. In both studies, two linear regression models were applied to identify the age-positions. Applying a quadratic regression model we indeed confirmed that major expression profiles are changed first in the fourth decade in both brain frontal cortex and Vastus lateralis muscle.
Ideally, the pattern of aging-associated molecular changes could be extracted from population-based datasets. These datasets are cross-sectional, covering a broad age-range, and all subjects are included. Most population-based datasets are skewed in the old age, making a linear regression model unfit. Here we investigated age-associated molecular changes in whole blood from two population datasets. The Rotterdam Study (RS) cohort III and the SHIP-TREND cohort were independently generated using RNA microarrays. After correcting for the skewed sample distribution across age, we demonstrate that an age-associated pattern of molecular changes is highly similar between the two datasets. We show that in whole blood major molecular changes occur only at the seventh decade, predominantly affecting the translation and immune cellular machineries.
The RS dataset was split into two subsets with the age of 65 years being selected as a cut-off point. At 65 years the age-position was found in the two age-matched datasets. The younger group (less than 65 years) comprised 606 individuals, whilst the older group (greater than 65 years) included 156 individuals. In the younger group, only 128 probes were found to be significantly age-associated and those were not enriched in any functional group. Those probes were not found among the overlapping genes between RS and SHIP-TREND datasets. This suggests that molecular changes in blood prior to 65 years are neither robust nor consistent. In contrast, in the older age group 1319 probes were age-associated and those were mapped to the immune system, translation and the defense response functional groups. 65% of those genes overlapped with the significant probes from the SHIP-TREND. This indicates that major expression profile alterations in blood occur from the seventh decade onwards. This age-position is in agreement with a recent study showing that the number of immune cells and T-cell receptor reduces from the seventh decade onwards.
Whether or not the rate of molecular aging is similar between tissues is poorly understood. In whole blood, we identified only a single age-position during the seventh decade. A single age-position was found in kidney cortex, also during the seventh decade. However, in brain frontal cortex and in Vastus lateralis muscle two age-positions were identified, the first during the fifth decade and a second one during early eighth decade. This suggests that in humans the age at which major molecular changes occur differs between tissues. This conclusion in agreement with physiological studies suggesting that the rate of age-associated tissue deterioration differs between tissues. Moreover, the prominent aging-associated gene networks also differ between tissues: translation and the immune system gene networks from blood were not identified in brain cortex or skeletal muscles tissues. An age-position could indicate an aging-associated disease risk for tissue-specific disorders and could be a consideration for treatments and interventions during aging.