Two look-alike cancer drugs diverge in 3D lab cell clusters

Two cancer drugs look the same on a flat dish. Grow the same cells into small 3D balls, and they come apart. That is what a team at Uppsala University, in Sweden, reports after testing 52 drugs used in or developed for cancer treatment. The chemotherapy 5-fluorouracil and the targeted drug olaparib were nearly indistinguishable in ordinary flat cultures. In three-dimensional clusters, they behaved differently.

The tool is Cell Painting, a widely used imaging method in drug discovery. It colours the main parts of a cell, such as the nucleus, the mitochondria and the internal scaffolding, so each can be photographed in detail under a microscope. Until now it had mostly been used on cells grown in flat layers. The Uppsala team grew bowel cancer cells as minispheres in standard laboratory plates, then made them transparent with a simple solution of sugar and glycerol. The microscope could then see individual cells deep inside each sphere, and the researchers measured thousands of details of each cell's shape and structure. The difference appeared only when they looked at single cells, not when they looked at each minisphere as a whole.

Follow-up experiments produced the surprise. 5-fluorouracil caused less DNA damage than olaparib, yet it was better at stopping cells from dividing. Gene activity pointed to the p53 pathway, a central stress response in cells, which 5-fluorouracil activated more strongly. The explanation is a hypothesis the team traced through its data, not a mechanism confirmed in patients.

The method also left the lab where it was built. The team ran it together with colleagues at AstraZeneca, a test that the researchers, led by Ola Spjuth and first authors Christa Ringers and David Holmberg, describe as a check on whether others can use it. For drug developers, the appeal is a screen that keeps single-cell detail while testing many compounds at once. The limits are clear: the work used two cancer cell lines in the laboratory and does not show how the drugs work in patients. The next step, according to the researchers, is more complex 3D models, such as organoids grown from patient tissue. No date is given for that step. All images, data and code are openly available.