Stem cells lead to improved neurotoxicity tests

MADISON, Wis.—Testing the toxicity of new drugs is often a costly and time-consuming process. Moreover, current methods of testing, which typically involve animal studies, leave much to be desired in terms of accuracy. But a new procedure developed by scientists at the University of Wisconsin–Madison (UW-Madison) and Madison-based Morgridge Institute for Research suggest that a better method of testing is on the horizon—at least when it comes to testing toxicity to the developing brain.

The researchers devised a method to test for neurotoxicity that relies on using stem cells to create a model of the developing brain. The new procedure has the potential to streamline the drug development process by allowing researchers to predict neurotoxicity much earlier in the preclinical stages. The research team also believes their new procedure, which could be used to test chemicals and pesticides as well as drug candidates, could support efforts to address the potential role of environmental chemicals in the rising incidence of neurodevelopmental disorders such as autism.

“Animal testing is expensive and often unreliable as a tool for predicting human neurotoxicity,” Michael Schwartz, a co-lead author of the study and an assistant scientist in biomedical engineering at UW-Madison, tells DDNews. “Human cells in culture provide a potential alternative to animal testing, but cellular models for predicting developmental neurotoxicity still need to be validated.”

One of the main challenges involved in using cellular models to predict developmental neurotoxicity is the limited understanding surrounding how the human brain forms and how it is disrupted by chemical exposure. The researchers addressed this challenge by devising a method to culture stem cells into differentiated cells that have a role in the composition and function of the brain, including neural-progenitor cells, vascular cells and microglia on polyethylene glycol hydrogels. These constructs could then be folded into three-dimensional structures called organoids, which in this case were able to mimic the human brain in early development.

“Our human pluripotent stem cell-derived neural tissue model is the first to incorporate an organized vasculature and microglia, which are components of the human brain that are crucial to normal development and which are potential targets for toxic chemicals,” Schwartz tells DDNews. “We also achieved high sample uniformity that allowed us to test 280 individual neural tissues in a single experiment, which is unprecedented for ‘organoid’ culture models.”

Once the culture models were created, the researchers were able to test their potential to predict toxicity. RNA-sequencing data was collected from neural tissue constructs that were individually exposed to 60 different “training” chemicals—both safe compounds and known toxins—and machine learning was used to build a predictive model from these results. After training with known chemicals using duplicate samples and two time points, the model correctly classified nine out of 10 additional chemicals in a blinded trial.

Schwartz says that the research team is extremely satisfied with that level of accuracy, which exceeds standard methods of testing for neurotoxicity. He adds that the research team believes it “will be able to further improve the accuracy of our model by expanding the size and diversity of our training chemicals.”

The new method of testing neurotoxicity was developed as part of a joint initiative of the National Institutes of Health, the Defense Advanced Research Projects Agency and the U.S. Food and Drug Administration. The goal of the initiative is to create a human “tissue chip” that incorporates 10 model tissues on a single platform to better predict the effects of drug and other toxins by modeling interactions between the individual systems.

Schwartz says that the screening method developed by his team offers a valuable bridge between testing a single layer of cells in a dish and testing on animals.

“These model neural tissues capture a lot more of the complexity than you would find in a monolayer of cells,” he says. “They also mimic human physiology, and should be more relevant for predicting toxicity than animal models. The fact that we could apply a machine learning model to achieve 90-percent accuracy this early in the process is fantastic.”

It may be some time until the new screening method becomes a viable alternative to animal testing in the drug development process, but Schwartz says the procedure could have immediate applications. “Our protocol should be useful as a discovery tool right away by enabling systematic and quantitative assessment of cellular interactions within the human brain that can be targeted for therapeutic intervention,” he tells DDNews. “However, the value of our model for making predictions is dependent on a biologically relevant cellular model of human brain development and training sets that represent all potential toxicity mechanisms.”