Human liver cells treated with a cancer drug are imaged under a microscope using a method called Cell Painting.
Image Credit: Axiom Bio
Scientific Frontline: Extended "At a Glance" Summary: Cell Painting for Drug Safety Testing
The Core Concept: Cell Painting is a scalable, image-based cellular profiling method that utilizes fluorescent dyes and artificial intelligence to measure thousands of molecular and structural changes in human cells following chemical exposure.
Key Distinction/Mechanism: Unlike conventional cell-based toxicity tests that typically measure single endpoints, Cell Painting labels eight different cellular components and leverages a trained AI model to simultaneously analyze thousands of morphological changes. This high-content approach allows researchers to detect cellular harm—including specifically which proteins and biochemical pathways are affected—at much lower chemical concentrations and with significantly greater detail than traditional assays.
Major Frameworks/Components:
- Fluorescent Multiplexing: The application of specific dyes to visualize eight distinct cellular compartments and structures simultaneously under a microscope.
- High-Throughput Image Analysis: Culturing human primary cells (such as liver cells), exposing them to over 1,000 different chemicals at varying concentrations, and capturing the resulting morphological data.
- Artificial Intelligence and Machine Learning: The deployment of predictive computational models trained on extensive cell-based toxicity datasets to recognize biological signatures indicative of adverse effects.
Branch of Science: Toxicology, Pharmacology, Cellular Biology, and Computational Biology.
Future Application: This method is actively being deployed to rapidly evaluate the toxicity of thousands of previously uncharacterized chemicals used in manufacturing and everyday consumer products. In pharmaceutical development, it will enable researchers to design better drugs by screening out toxic candidate compounds earlier in the pipeline, thereby minimizing off-target effects.
Why It Matters: Many drug candidates fail in early-stage clinical trials due to human-specific toxicities that animal models fail to predict. By accurately identifying human cellular toxicity in vitro, Cell Painting provides a faster, more cost-effective, and highly detailed alternative to early-stage animal testing, ultimately streamlining drug development and improving consumer safety.
Many drug candidates fail in early-stage clinical trials because of toxic effects in humans that weren’t seen in animal tests. Testing these chemicals in human cells in the lab could help detect these adverse effects sooner and at lower cost, while reducing the need for animal testing.
In a new study published in Cell Systems, Broad Institute researchers, in collaboration with scientists with the Omics for Assessing Signatures for Integrated Safety (OASIS) consortium, have shown how a scalable, image-based cell profiling method called Cell Painting promises to be a much more efficient way to detect toxicity than current methods.
“I’m really excited about this work because there’s a huge problem that most of the chemicals that we use, for example in manufacturing and consumer products, haven’t been well tested or monitored for toxicity,” said Jess Ewald, first author on the paper and a group leader at the European Molecular Biology Laboratory in Cambridge UK who did the study as a postdoc in the laboratory of study co-authors Anne Carpenter and Shantanu Singh at the Broad.
“I think we could improve human health quite a lot if we had better, faster, cheaper tools like this to get a better handle on the effects of drugs and other chemicals on our cells,” she said.
What did the researchers do?
Cell Painting uses fluorescent dyes to label eight different cell components and measure thousands of molecular changes in the cells in response to chemical exposure, which are then captured by a microscope. In the new study, the team trained a Cell Painting-based AI model on large amounts of data from hundreds of cell-based chemical toxicity tests.
They exposed human liver cells in a dish to more than 1,000 different chemicals at different concentrations. They took Cell Painting images of the cells, and then used their AI model to analyze the images and look for signs of biological changes that can lead to toxic effects. The team ran conventional cell-based toxicity tests on those same cells and compared how their Cell Painting method performed.
What did they find?
The scientists discovered that compared to existing cell toxicity tests, Cell Painting could rapidly provide a more detailed view on how chemicals are harming cells — such as which proteins or biochemical pathways were affected. The method captured hundreds of changes in the cells, generating insights from a single measurement that would normally come from hundreds of individual tests.
Cell Painting could also accurately detect changes in the cells exposed to chemicals known to be harmful to cells. And it could do so at much lower concentrations of the chemicals, confirming that Cell Painting can pick up on many more molecular changes in the cell than conventional tests.
Why is this important?
Knowing the details on how chemicals affect cells could lead to a better understanding of not just which chemicals are potentially toxic, but also why. In drug development, this insight could help researchers design better drugs with fewer side effects.
The scientists are already using the tools they developed for this study to assess the effects of hundreds to thousands of chemicals that haven’t been well studied at all.
Funding: Support for this research was provided by Banting and SOT Syngenta postdoctoral fellowships, the National Institutes of Health, the Novo Nordisk Foundation, and the Omics for Assessing Signatures for Integrated Safety (OASIS) Consortium, which is supported by a grant from the Massachusetts Life Sciences Center.
Published in journal: Cell Systems
Title: Cell Painting for cytotoxicity and mode-of-action analysis in primary human hepatocytes
Authors: Jessica D. Ewald, Katherine L. Titterton, Alex Bäuerle, Alex Beatson, Daniil A. Boiko, Ángel A. Cabrera, Jaime Cheah, Beth A. Cimini, Bram L. Gorissen, Joshua Harrill, Thouis R. Jones, Konrad J. Karczewski, Christine E. Crute, David Rouquie, Srijit Seal, Erin Weisbart, Brandon White , Anne E. Carpenter, and Shantanu Singh
Source/Credit: Broad Institute | Corie Lok
Reference Number: phar033126_01