A machine learning algorithm that can predict the biological age of brain cells has helped scientists identify hundreds of potential anti-aging treatments to prevent cognitive decline and neurodegeneration as we get older.
“Aging is the primary risk factor for several neurodegenerative disorders that most older adults eventually face,” stated the researchers led by Antonio del Sol, professor of computational biology at the Luxembourg Centre for Systems Biomedicine (LCSB) and research professor at CIC bioGUNE in Spain. “The global population is aging rapidly, with over two billion people projected to be above the age of 60 by 2050. Therefore, discovering effective strategies to protect the aging population from neurodegeneration is critical.”
To train the machine learning model, the team collected data from brain samples of 778 healthy individuals with ages ranging from 20 to 97 years. Rather than looking at the genetic code, the model focuses on the transcriptome — the collection of RNA molecules transcribed from DNA — in order to gauge the level of activity of each gene in each brain sample.
The algorithm identified 365 gene transcripts that, together, could accurately predict the age of a person from a brain sample within a five-year range. Only 25% of these genes were directly involved in brain processes; Instead, most of them were linked to DNA repair and regulation, which are known to be closely connected to aging across all tissues.
In samples of patients diagnosed with neurodegenerative conditions, such as Alzheimer’s or traumatic brain injury, this “aging clock” model predicted their brains to have a significantly higher biological age.
“This was particularly evident in samples coming from donors aged 60 to 70, with the neurodegenerative samples having a transcriptional age 15 years higher than the healthy individuals,” reported del Sol. “These findings show that transcriptional age is negatively correlated with brain function, supporting the view of neurodegeneration as a form of accelerated aging.”
Next, the machine learning model analyzed data from thousands of samples of neurons and neural progenitor cells, looking for gene expression changes that reduced the predicted age of the sample. This allowed the computer algorithm to find 478 drugs with a rejuvenating effect on brain cells.
“Although several compounds predicted by our model have been shown to extend lifespan, the vast majority have not been studied in the context of health or lifespan extension,” added del Sol. “Moreover, many predicted compounds are still experimental, and their mechanism of action remains unknown.”
The team selected three compounds identified by the algorithm and tested their effects on old mice over the course of four weeks. Treatment with the three compounds significantly reduced anxiety and improved memory in the mice, while shifting the genetic expression of their brain cells towards a younger transcriptional profile.
While these preliminary results show promise, more research will be needed to validate the effects of these and other compounds identified by the machine learning model. The goal is to one day develop drugs with potent anti-aging and neuroprotective effects.
According to del Sol and colleagues, the anti-aging field currently lacks systematic methods for drug discovery, making their computer algorithm a valuable resource for identifying promising therapeutic compounds.
“Our computational platform represents a valuable resource for identifying interventions that may counteract age-related brain decline in brain function,” concluded del Sol. “The hundreds of compounds predicted by our platform require validation across diverse multiple biological systems to assess their efficacy, offering extensive opportunities for future research and therapeutic development.”
Featured image credit: Micheile Henderson via Unsplash
