Single-cell profiling of multiple cell lines can reveal cancer vulnerabilities and drug mechanisms

A new method uses single-cell RNA sequencing to measure the effect of drugs or genetic perturbations on dozens of cancer cell lines simultaneously. This allows scientists to rapidly observe gene activity changes in a wide range of different cancer cell types in response to drugs or genetic tweaks. These insights could reveal new details about how specific drugs work in cells and support the development of new cancer therapies.

Known as MIX-Seq, the method builds upon the PRISM technology pioneered by the Broad Institute’s Cancer Program, in which researchers tag hundreds of cancer cell lines with unique DNA barcodes, pool the cells together, expose them to treatment, and then measure their abundance to determine their sensitivity to drugs. MIX-Seq instead uses the cells’ own genetic fingerprints to identify and track the cells. And instead of simply measuring cell viability, MIX-Seq uses single-cell RNA sequencing to efficiently generate information-rich readouts from dozens of cancer cell lines at once.

Scientists with the Broad’s Cancer Dependency Map Project and the lab of core institute member Aviv Regev (on leave) describe MIX-Seq in a paper in Nature Communications.

“With MIX-Seq, we were able to combine the efficiency of pooling cells together with the granular, single-cell level information provided by RNA sequencing,” said James McFarland, associate director of data science in the Broad’s Cancer Program and co-first author on the study. “The single-cell resolution of MIX-Seq, combined with an existing trove of data on these cells, lets us ask questions that we couldn’t otherwise answer.”

A cell’s signature in the sequence

In developing the method, the research team demonstrated that each cell type’s unique fingerprint of single-letter mutations, known as SNPs, could serve as a ready-made system to track cells.

In MIX-Seq, researchers grow dozens of cancer cell types together, expose the cells to a drug or genetic perturbation, and then read the RNA being transcribed by individual cells. They use the unique patterns of SNPs in the RNA to track which cell type each RNA transcript originated from. The RNA sequences, and the genes from which they arose, also provide clues to what’s happening inside different cancer cell types when they are altered by drugs or when a specific gene is turned on or off. Patterns of gene expression changes can indicate which cellular pathways have been affected by the treatment, revealing how drugs exert their effects on the cell.

The data generated as part of this work include single-cell RNA sequencing measurements for more than 200 cancer cell lines across 30 different treatment conditions. The team has made all data publicly available, as described on the Cancer Data Science Blog.

To demonstrate the method’s utility, the researchers showed that the single-cell resolution of MIX-Seq can efficiently measure changes in the cell cycle, which can support efforts to develop cancer therapies that inhibit cellular division. The method also revealed unappreciated cellular heterogeneity within the cancer cell lines that can be used to identify subpopulations of cells that each respond differently to treatment.

Further, the team used antibody tags to label cells at different time points after treatment. This allowed them to pool the cells together into one RNA sequencing run and efficiently observe changes in gene expression over time.

Single-cell RNA sequencing could also be used to assess the drug sensitivity of tumor cells from patients. Standard viability assays are done several days after exposure to a drug, but tumor cells don’t often survive that long in a dish. The team showed that early gene expression changes, even measured in small populations of individual cells, can predict which cells will later succumb to the treatment, suggesting that MIX-Seq could help pave the way towards tests that can directly and rapidly measure the drug sensitivities of a patient’s tumor.

The work was also led by co-first author Brenton Paolella, a research scientist for the Cancer Dependency Map Project; co-senior author and associate director of the Cancer Dependency Map Project Francisca Vazquez; co-senior author Andrew Aguirre, an associate member in the Cancer Program and assistant professor of medicine at Harvard Medical School and Dana-Farber Cancer Institute; and co-senior author and Cancer Program scientific advisor Aviad Tsherniak.