Researchers used machine learning models to identify three compounds that could potentially fight aging. They say their approach could be an effective way to identify new drugs, especially for complex diseases.
Cell division is necessary for our bodies to grow and renew tissues. Cellular senescence is a phenomenon in which cells remain in the body without dividing permanently, causing tissue damage and aging throughout the body’s organs and systems.
Senescent cells are normally cleared from the body by the immune system. However, as we age, our immune system becomes less effective at removing these cells, and their numbers increase. An increase in senescent cells is associated with diseases such as cancer, Alzheimer’s disease, and features of aging such as poor vision and reduced mobility. Given the possible negative effects on the body, there is a need to develop effective senolytics, compounds that eliminate senescent cells.
Previous studies have identified several promising senolytic agents, but they are often toxic to healthy cells. Now, a study led by researchers at the University of Edinburgh in Scotland used a pioneering method to search for chemicals that can safely and effectively eliminate these defective cells.
They developed a machine learning model and trained it to recognize key features of chemicals with senolytic properties. Training data for the model came from multiple sources, including academic publications and commercial patents, and was integrated with compounds from two existing chemical libraries, including a wide range of FDA-approved and clinical-stage compounds.
The full dataset contained 2,523 compounds and included compounds with both senolytic and non-senolytic properties to ensure unbiased machine learning algorithms. Using this algorithm he then screened over 4,000 chemicals, of which 21 potential candidates were identified.
Researchers testing these candidates found that three chemicals, gingethin, periplocin, and oleandrin, eliminated senescent cells without harming healthy cells. Of the three, oleandrin proved to be the most effective. All three are natural products found in traditional Chinese medicine.
Oleandrin is an oleander plant (oleander), which has similar properties to the drug digoxin, which is used to treat heart failure and certain abnormal heart rhythms (arrhythmias). Studies have shown that oleandrin has anti-cancer, anti-inflammatory, anti-HIV, antibacterial and antioxidant properties. Oleandrin has a high supratherapeutic level of toxicity and its levels are very narrow in humans, which hinders its clinical application. As such, it is not approved as a prescription drug or dietary supplement by any regulatory agency.
Like oleandrin, gingethin has been shown to exhibit anticancer, anti-inflammatory, antibacterial, antioxidant, and neuroprotective effects. Ginkgothin is extracted from ginkgo biloba (ginkgo) s tree. Its leaves and seeds are the oldest surviving tree species that have been used in herbal medicine for thousands of years.highly concentrated ginkgo Extracts made from the dried leaves of the tree are available over the counter. This is his one of the best selling herbal supplements in the US and Europe.
Periprosin is isolated from the root bark of the Chinese silk tree (periplocasepium). Studies have shown that in addition to improving heart function, it can block cell proliferation and cause cell death in cancer cells.
The researchers say their results show that these compounds are as or more potent than senolytic drugs reported in previous studies. More importantly, they say, their machine-learning-based method was so efficient that it reduced the number of compounds that needed to be screened by more than 200-fold.
The researchers say their AI-based approach marks a milestone in identifying new drugs, especially for complex diseases.
“This study shows that AI can be highly effective in identifying new drug candidates, especially in the early stages of drug discovery and in diseases with complex biology or few known molecular targets,” said the researcher. Corresponding author Diego Oyarzun said.
They also say that this approach is more cost-effective than standard drug screening methods such as preclinical and clinical trials.
“This research is the result of an intensive collaboration between data scientists, chemists and biologists,” said Vanessa Semell Barreto, lead author of the study. “By leveraging the strength of this interdisciplinary combination and using only publicly available data for model training, we were able to build robust models and save on screening costs. We hope to open up new opportunities to accelerate the application of this exciting technology.”
The study was published in a journal Nature Communications.
Source: University of Edinburgh
