Organoids, three-dimensional miniature organs grown from stem cells, are powerful tools for studying development and disease, for testing new drugs, and potentially even for transplantation. But so far, growing and imaging them at scale has proven difficult. Now, a team at the University of Bordeaux, in collaboration with scientists at the National University of Singapore, has designed an automated technique that takes mere seconds to image an organoid in 3D (Nat Meth, 19:881–92, 2022).
Organoids grown under the same conditions can still develop differently, so “you need to have a lot of organoids in the same condition to be able to understand what’s going on,” explains National University of Singapore cell biologist Anne Beghin, who helped develop the technique. But capturing a 3D image of the entire structure is tricky, she adds, because light is toxic to cells, and most 3D imaging methods are light-intensive, not to mention slow. Previous methods could image just “ten or twenty” organoids at a time, Beghin says.
In 2015, Bordeaux physicist Jean-Baptiste Sibarita and his colleagues developed a method to capture super-resolution 3D images of live single cells. The technique, Single-Objective Selective-Plane Illumination Microscopy (soSPIM), differs from traditional cell-imaging methods because it only illuminates a single plane of the sample at a time and uses only one objective, minimizing cells’ light exposure (Nat Meth, 12:641–44, 2015).
To apply soSPIM to organoids, Sibarita, Beghin, and their colleagues designed JeWell chips: high-density arrays that contain cavities composed of four mirrors arranged in a pyramid. The shape of each JeWell cavity keeps the organoid grown within from spilling out and facilitates soSPIM imaging. A laser positioned below the JeWell chip bounces off a mirror to illuminate thin slices of either fixed or live organoids tagged with fluorescent markers. A camera then captures the reflected light and assembles a 3D image, layer by layer.
Beghin successfully grew neural, liver, and cancer cell–derived organoids, among others, inside the JeWells and applied the imaging approach. The researchers also adapted machine learning–based tools to pick out cells undergoing mitosis or apoptosis and taking on properties of organoids.
Using this approach, the researchers imaged a single organoid in seven seconds—and roughly 300 organoids in an hour—employing a single color of fluorescence. Using three colors to tag three separate proteins, they could image about 96 organoids per hour. “There is a huge demand in terms of getting organoids close to the pipeline to drug discovery,” says Beghin, adding that this approach should help meet it by helping pharmaceutical companies integrate organoids into high-throughput drug screening protocols.
Cardiff University cell biologist Trevor Dale, who researches organoids but was not involved in the study, says he worries the JeWell’s unique shape may prevent important structure-giving molecules from reaching the organoids, rendering the technique unsuitable for growing certain types of organoids. However, he adds that the imaging definitely “appears to increase the rate at which you can acquire 3D data,” whereas similar techniques he’s tried have “taken ages.”