As we age, genes are expressed differently throughout our body. The most obvious examples are the hormonal changes seen in adolescence and in the menopause. In many models of how genes are expressed during older age, one of the prevalent models - the programmed ageing model - claims that ageing is caused by genetically programmed cell death. In other words, ageing is programmed into our cells - little blame is normally to be put on random or environmental factors.
Opposed to this model we have the normally less favoured stochastic model, which claims that random biological events play a significant role in ageing. In a recent report (PDF) (see also more extensive material here (PDF)) in Current Biology a team of researchers from the Max Planck Institute for Evolutionary Anthropology, headed by Mehmet Somel, present evidence supporting this model. Using an in-house method (called "age-correlated heterogeneity of expression", or ACHE) to assess heterogeneity of gene expression, they found that the expression becomes more heterogeneous as we age. And this goes not only for humans. Rats show the same pattern, too. They find that their results "are compatible with ACHE being an outcome of the accumulation of stochastic effects at the cellular level". In other words, something that is not programmed but due to other biological factors.
You can see this in the following figure from the article:
Explanation:"An example from the human brain data set B, the log-expression versus age plot for a probe set detecting the gene PIM-1, for which the ACHE test p-value was calculated as0.0002." Looks like heterogeneity to me … although I'd like to see some more data, maybe on 100-200 subjects.
Note also from that article that the heterogeneity is different depending on where in the body we take the sample. What I find interesting is that the brain seems to be one of the organs where genes become most heterogeneous during ageing (see bar 2-5). This also goes for the rat hippocampus. Click the image to see a larger version:
Explanation: The heights of the bars indicate observed to expected ratios of the number of probe sets at different cutoffs within the ACHE test p-value distribution
But what is ACHE really a measure of? Here's the answer from Somel et al.:
Our results indicate that ACHE is a general — but weak — effect in the transcriptome.This is compatible with ACHE being the outcome of accumulating stochastic effects in the soma, such as cellular damage and mutations. These effects will influence each cell in a unique way so that expression variation among aging cells will be equalized at the tissue level. If somatic mutations mostly cause decreases in expression level, the overlap between ACHE and age- related decrease in expression levels can also be explained within this framework. ACHE supports the stochastic nature attributed to the aging process. It implies a weakening of expression regulation with age, contrary to previous observations and hypotheses based on measurements on a small number of gene.
While browsing around for links to this post, I found a great forthcoming publication (PDF) in Trends in Ecology and Evolution by Partridge and Gems. See also this nice yet brief post at Wikipedia about the ageing brain.