The transcriptome imbalance associated with ageing is a result of a length-driven systemic imbalance
The aging process manifests as a decline in homeostasis, cellular functions and physiological processes. The efforts to identify a common cause of vertebrate aging are fraught with challenges. For example, although there have been documented differences in the expression levels of hundreds of mRNAs across tissues and species, the results have been inconsistent. We analyzed age resolved transcriptomic data of 17 mouse organs, and 51 human organs with unsupervised machine-learning3 -5 in order to identify the architectural and regulation characteristics that are most informative about the differential expression genes as they age. We describe a previously unknown phenomenon: a systemic, age-dependent, length-driven imbalance in transcriptomes that disrupts homeostasis between short and longer transcript molecules. This occurs for humans, mice, rats and killifishes. In a mouse model for healthy aging we also show that a length-driven imbalance in transcriptome correlates with the expression of splicing factors proline and glutamine (Sfpq), regulating transcriptional elongation based on gene length. We also show that environmental hazards and pathogens can trigger length-driven transcriptionome imbalance. Our findings confirm the idea that aging is a breakdown of systemic homeostasis and provide a plausible explanation for how diverse insults can affect age-dependent phenotypes similarly.
The transcriptome is able to respond rapidly, selectively and strongly to many different molecular and physiologic insults that an organism experiences. The transcripts of thousands genes are reported to alter with age. However, the changes in magnitude of most transcripts is very small compared with classic examples of gene regulation2,8. There is also little agreement among the different studies. Our hypothesis is that aging may be associated with an uncharacterized process which affects the transcriptome systemically. We predict that this process would integrate a variety of environmental insults, each with a molecularly distinct signature, to promote the phenotypic manifestations associated with aging.
To identify age-dependent transcriptome changes, we use an unsupervised machine-learning approach3 -5. In order to achieve this, we measured and surveyed the transcriptomes of 17 mouse organs in 6 biological replicas raised under standard conditions at 5 different age ranges between 4-24 months (Fig. 1A). We examine information about the structure of genes and transcripts and the knowledge of the binding of regulatory molecules, such as transcription factors (miRNAs), and knowledge of the binding of regulatory molecules (Fig. 1B). Age-dependent fold-changes are defined as the log2-transformed relative ratio of the transcripts at a certain age to those of the same gene in the tissues of four-month-old mouse organs. The predicted fold-changes are in line with what is expected from models that capture the most measurable changes of transcript abundance (Fig. As expected for models that capture most measurable changes in transcript abundance, the predicted fold-changes (Fig. S2 and S3.
Source:
https://www.biorxiv.org/content/10.1101/691154v1.full
