A linear regression model predicts human brain ageing and reveals differential neuronal biological ageing relevant for Parkinson's disease susceptibility

Abstract

Biological brain ageing is a major risk factor for neurodegenerative diseases, which are characterized by selective degeneration of particular neuron types. We analyzed the impact of ageing on the transcriptome of neurons in the ventral tegmental area (VTA), substantia nigra pars compacta (SNc) and locus coeruleus (LC), that show differential vulnerabilities to Parkinsons disease. Neurons were isolated from human post mortem brain tissues originating from 48 individuals ranging from 17 to 102 years of age and subjected to Smart-seq2 RNA sequencing. We identified 2,764 genes that were correlated with chronological ageing. This gene expression data was used to develop a feature selection-based Time Traversal algorithm, utilizing functionally grouped gene sets, GO terms, with high predictive accuracy of biological brain ageing. We identified 59 GO terms that can predict biological age using a linear regression model, where leave-one-out cross validation demonstrated a strong correlation between chronological age and predicted biological age (Pearson correlation coefficient = 0.946; adjusted R{superscript 2} = 0.771). The algorithm was validated on five independent datasets with high predictive performance, demonstrating shared ageing features across the human brain. Nonetheless, our analysis also highlights brain region and neuron type specificity in particular ageing features. Resilient neurons showed a weaker association with age-related transcriptional changes, indicating that they age slower than their vulnerable counterparts, thus revealing targets that may be used to slow down ageing and prevent disease development.