Researchers have used a machine learning model to identify three compounds that could combat aging. They say their approach could be an effective way of identifying new drugs, especially for complex diseases.
Cell division is necessary for our body to grow and for tissues to renew themselves. Cellular senescence describes the phenomenon where cells permanently stop dividing but remain in the body, causing tissue damage and aging across body organs and systems.
Ordinarily, senescent cells are cleared from the body by our immune system. But, as we age, our immune system is less effective at clearing out these cells and their number increases. An increase in senescent cells has been associated with diseases such as cancer, Alzheimer’s disease and the hallmarks of aging such as worsening eyesight and reduced mobility. Given the potentially deleterious effects on the body, there has been a push to develop effective senolytics, compounds that clear out senescent cells.
Previous studies have identified some promising senolytics, but they’re often toxic to healthy cells. Now, a study led by researchers from the University of Edinburgh in Scotland has used a pioneering method to seek out chemicals that can safely and effectively eliminate these defective cells.
They developed a machine learning model and trained it to recognize the key features of chemicals with senolytic properties. The model training data came from multiple sources, including academic papers and commercial patents, and was integrated with compounds from two existing chemical libraries that contain a wide range of FDA-approved or clinical-stage compounds.
The full dataset contained 2,523 compounds and included compounds with both senolytic and non-senolytic properties so as not to bias the machine-learning algorithm. The algorithm was then used to screen more than 4,000 chemicals from which 21 potential candidates were identified.
Testing these candidates, the researchers found that three chemicals – ginkgetin, periplocin, and oleandrin – removed senescent cells without harming healthy cells. Of the three, oleandrin was found to be the most effective. All three are natural products found in traditional herbal medicines.
Oleandrin is extracted from the oleander plant (Nerium oleander) and has properties similar to the drug digoxin used to treat heart failure and certain abnormal heart rhythms (arrhythmias). Studies have shown that oleandrin possesses anticancer, anti-inflammatory, anti-HIV, antimicrobial and antioxidant properties. Oleandrin is highly toxic beyond therapeutic levels, which in humans is a very narrow window, hindering its clinical application. As such, it hasn’t been approved by regulatory agencies as a prescription drug or dietary supplement.
Like oleandrin, ginkgetin has been shown to exhibit anticancer, anti-inflammatory, antimicrobial, antioxidant and neuroprotective properties. Ginkgetin is extracted from the Gingko (Ginkgo biloba) tree, the oldest living tree species whose leaves and seeds have been used in Chinese herbal medicine for thousands of years. A highly concentrated Ginkgo biloba extract made from the tree’s dried leaves is available over the counter. It’s one of the best-selling herbal supplements throughout the US and Europe.
Periplocin is isolated from the root bark of the Chinese silk vine (Periploca sepium). Studies have shown that it can improve heart function as well as block cell growth and cause cell death in cancer cells.
The researchers say their findings demonstrate that these compounds have a potency comparable to or higher than senolytics described in previous studies. More importantly, they say, their machine-learning-based method was extremely efficient, reducing the number of compounds that needed to be screened by more than 200-fold.
The researchers say their AI-based approach represents a milestone in identifying new drugs, particularly for complex diseases.
“This study demonstrates that AI can be incredibly effective in helping us identify new drug candidates, particularly at early stages of drug discovery and for diseases with complex biology or few known molecular targets,” said Diego Oyarzún, corresponding author of the study.
They also say the approach is more cost-effective than standard drug screening methods, such as preclinical and clinical trials.
“This work was borne out of intensive collaboration between data scientists, chemists and biologists,” said Vanessa Smer-Barreto, lead author of the study. “Harnessing the strengths of this interdisciplinary mix, we were able to build robust models and save screening costs by using only published data for model training. I hope this work will open new opportunities to accelerate the application of this exciting technology.”
The study was published in the journal Nature Communications.
Source: University of Edinburgh
