ӳ����ý

Visualizing cells after editing specific genes can help scientists learn new details about the function of those genes. But using microscopy to do this at scale can be challenging, particularly when studying thousands of genes at a time.

Now, researchers at the ӳ����ý of MIT and Harvard, along with collaborators at Calico Life Sciences, have developed an approach that brings the power of microscopy imaging to genome-scale CRISPR screens in a scalable way. 

PERISCOPE — which stands for perturbation effect readout in situ via single-cell optical phenotyping — combines two technologies developed by ӳ����ý scientists: Cell Painting, which can capture images and key measures of subcellular compartments at scale, and Optical Pooled Screening, which “barcodes” cells and uses CRISPR to systematically turn off individual genes to study their function in those cells. 

The new technique lets scientists study the effects of perturbing over 20,000 genes on hundreds of image-based cellular features. Generating data with the method is more than 10 times less expensive than comparable high-dimensional approaches such as high-throughput single-cell RNA sequencing and can be adapted to study a wide variety of cell types. In , the researchers applied PERISCOPE to execute three whole-genome CRISPR screens to create an open-source atlas of cell morphology.

The study is the result of a collaboration between the labs of ӳ����ý co-senior authors JT Neal, institute scientist and co-director of type 2 diabetes systems genomics; Anne Carpenter, institute scientist and senior director of the Imaging Platform; and Paul Blainey, core institute member; as well as Calico, led by co-senior author . ӳ����ý staff scientists Meraj Ramezani and Erin Weisbart, research associate Julia Bauman, and postdoctoral researcher Avtar Singh were all co-first authors on the study.

“This atlas itself is a first-in-class genome-scale resource for linking cell morphology to gene function, and I think it has a lot of discovery potential,” said Neal. “But maybe more importantly, all of the data is open access, and we’ve built really robust industrial-grade analysis pipelines that can be repurposed by anybody to analyze optical pooled screening data at scale.”

Morphological insights

First developed by Carpenter in , Cell Painting is an assay designed to image multiple cell organelles at the same time and at scale, “painting” them different colors such as nuclei blue and mitochondria red. Machine-learning models then detect hundreds of subtle changes in the images, such as the texture of the cytoplasm or area of the nucleus, allowing researchers to test the effects of a gene perturbation, for example, on the cell.

When Neal joined the ӳ����ý in 2017 as a researcher in the Cancer Program, he was intrigued by the possibility of applying Cell Painting to cancer cells to learn the function of key genes and variants, which could help in the development of targeted therapies. He struck up a conversation with Carpenter, and also with Blainey, whose lab had developed . The three started discussing how they might combine the two techniques and, in 2019, started building PERISCOPE, including an associated computational pipeline that integrates with Cell Profiler, in collaboration with Calico.