Kidney cells generated from stem cells untreated (top) or treated with aristolochic acid (bottom), a toxic compound widely used in traditional Chinese medicine and other herbal remedies, which is banned in many countries.
Image credit: IBN Singapore
In the journal Scientific Reports, scientists from the Institute of Bioengineering and Nanotechnology (IBN) in Singapore describe how they developed the platform using human induced pluripotent stem cells (iPSCs) and machine learning methods.
As well as predicting toxicity, they note that the new platform also correctly identifies the injury mechanisms that a drug may induce, which can improve understanding of the compound and its effects.
The team anticipates the new tool will be very useful not only to pharmaceutical companies looking to improve and speed up drug development as well as reduce its costs, but also to the food and cosmetics industry.
In addition to raising ethical concerns, using animals to try and predict whether a new drug is safe to use in humans has many drawbacks. For instance, it is expensive, time-consuming and does not necessarily ensure reliable results because of inter-species differences.
The authors also note that often, a drug's toxicity is not discovered until the later stages of drug development, and sometimes only after it has been marketed. So there is a need for tools that can detect any toxic effects much earlier in the drug development phase.
Additionally, there are now many parts of the world, such as the EU, India and Israel, where it is illegal to sell cosmetic products if they have been tested on animals.
'Highly pure kidney cells suitable for compound screening'
Senior author Dr. Daniele Zink, IBN team leader and principal research scientist, says:
"We have developed the fastest and most efficient protocol for generating kidney cells from induced pluripotent stem cells. Within 8 days, it yielded highly pure kidney cells that were suitable for compound screening."
The new platform is the result of several years of work that evaluated earlier versions. These earlier versions used human primary renal proximal tubular cells or similar cells derived from human embryonic stem cells (hESCs). But these versions were not ideal because primary cells are not easy to source from the body, and there are ethical and legal concerns with using hESCs.
So the team turned to human iPSCs, which can can be generated from cells that are easily found in the body, for instance on skin. The other advantage of iPSCs is they can also be used to develop patient- and disease-specific models that can improve understanding of kidney disease and help the development of personalized therapies.
Machine learning increases accuracy of prediction
Another important feature of the screening platform is the use of machine learning methods. These greatly improve the accuracy of prediction. Machine learning uses computer methods that teach themselves to grow and change as they are fed new data. Dr. Zink notes:
"We were further able to identify injury mechanisms and drug-induced cellular pathways by using automated cellular imaging. We hope that our work will contribute to the development of safer products in future."
She and her colleagues are now looking to team up with industrial partners to further validate and apply their renal screening platforms.
Another growing area of concern surrounding the development of new drugs and compounds is the increasing pharmaceutical pollution of water systems worldwide. Medical News Today recently learned how scientists are working on a way to make drugs more biodegradable to protect water resources.