Using Drug Screens And CRISPR To Make Better Cancer Drugs

• New research improves understanding of the biological mechanisms underpinning drug response in cancer
• The findings could facilitate more efficient drug development for cancer
• The success rate in drug development is very low, but it could be improved by using this type of analysis

The success rate in drug discovery is very low with fewer than 10% of prospective compounds proceeding to clinical trials. For drugs that treat cancer, this is in part because researchers do not fully understand the mechanism by which a drug kills cancer cells, also called its mechanism-of -action.

To gain new insights into the process, researchers from the Wellcome Sanger Institute, EMBL’s European Bioinformatics Institute and AstraZenca have combined drug response data with CRISPR genetic screens across hundreds of cell lines. This allowed them to better understand how drugs target cancer cells.

Two methods reveal new insights
The study, published in Molecular Systems Biology, identified the mechanism-of-action in 50% of the 397 drugs tested. This improved understanding of the biological mechanisms underpinning drug response will facilitate faster, more efficient development of new cancer drugs and brings us closer to precision medicine for cancer patients.

Researchers used two methods in the study. Firstly, they used pharmacological screens, which test the activity of anti-cancer drugs to identify how sensitive particular cancers are to particular compounds. Secondly, they used CRISPR-Cas9 technology to edit the genes in cancer cell lines, turning them off one by one, to measure how critical they are for the cancer to survive.

For the first time, researchers combined CRISPR-Cas9 screens with pharmacological screens for 397 unique anti-cancer compounds across 484 cancer cell lines. The data was sourced from EMBL-EBI’s ChEMBL database of bioactive molecules with drug-like properties, and included FDA-approved cancer drugs, drugs in clinical development, and compounds in early development.

The team investigated the extent to which drug sensitivity corresponded to CRISPR knock-out of drug targets by searching for associations between the two datasets across the 484 cell lines. They identified 865 significant associations between drug response and gene dependency.

Advancing drug development
“The effect of knocking out a gene and the effect of inhibiting a protein that the gene produces aren’t necessarily the same thing,” explains Emanuel Gonçalves, Postdoctoral Researcher at the Wellcome Sanger Institute. “But when a molecular pathway or function is associated with both drug response data and CRISPR screen data, it gives us a much clearer idea of how a drug is working at the molecular level, and the ability to detect when a drug is not working as we expect.”

“Fully understanding the molecular pathways involved is the key to understanding why a drug may work on one patient’s cancer but not on another’s, says Aldo Segura-Cabrera, Chemogenomics Analyst at EMBL-EBI. “The association between the MARCH5 and MCL1 genes in breast cancer, for example, suggests an important molecular relationship that we were unaware of. This in turn helps us to understand the mechanism-of-action of MCL1 protein inhibitors and in which cancer cases these drugs will be effective. This study also illustrates how the integration of pharmacological and functional genomics data can help us to provide a detailed and expanded view of how the anti-cancer drugs might work at the molecular level.”

The study illustrates an innovative application of the extensive bioactivity data available in ChEMBL. The study also suggests some future areas of research, for example for those drugs that act by the polypharmacological modulation of multiple proteins.