Research Roundup: December 3, 2021

Imaging ferroptosis sensitivity, using AI to analyze aorta genetics, double dependencies in cancer, and more

Susanna M. Hamilton
Credit: Susanna M. Hamilton

Welcome to the December 3, 2021 installment of Research Roundup, a recurring snapshot of recent studies published by scientists at the Broad Institute and their collaborators.

A PALP-able way to measure ferroptosis potential

Inducing ferroptosis, a type of cell death, is an emerging therapeutic strategy for cancer. Emily Graham, core institute member Stuart Schreiber in the Chemical Biology and Therapeutics Science Program, and colleagues have developed a method that can detect ferroptosis sensitivity in live cancer cells and tissues in vitro and in vivo. The method, called photochemical activation of membrane lipid peroxidation (PALP), uses targeted lasers to induce localized peroxidation of polyunsaturated fatty acyl (PUFA)-phospholipids, which triggers ferroptosis. The PALP signal depends on PUFA-phospholipid levels, providing a measure of ferroptosis susceptibility. With further development, the technology could be used to stratify patients according to their ferroptosis sensitivity. Read more in Cell Chemical Biology.

Repurposing an alcoholism drug for COVID-19

An inexpensive drug commonly used to treat alcoholism may be worth studying as a potential COVID-19 treatment. Ciyue Shen, Cancer Program associate member Chris Sander, and collaborators conducted a retrospective epidemiological study using clinical records from the US Veterans Affairs healthcare system. They found that use of the drug disulfiram was associated with decreased risk of both infection by SARS-CoV-2 and death from COVID-19. The team hypothesized that disulfiram inhibits viral replication and the inflammation that occurs in severe COVID-19. Read more in PLOS ONE and in a Harvard Medical School news story.

Optimizing power with scPOST

Detecting clusters, or cell states, correlated with disease in single-cell studies can offer insight into disease mechanisms. Large-scale power analyses are needed in order to detect differences in cluster abundance between conditions, but current methods are computationally expensive and don’t model intersample variation. In Cell Reports Methods, Nghia Millard, institute member Soumya Raychaudhuri of the Program in Medical and Population Genetics, and colleagues describe single-cell POwer Simulation Tool (scPOST) — a flexible framework for simulating multi-sample single-cell datasets to optimize study design. They show that scPOST simulates datasets that mimic input data structure, and find that increasing sample number and multiplexing samples generally improves statistical power more than increasing cells per sample or sequencing depth.

Deep learning enables genetic analysis of the aorta

Aortic aneurysm is a dangerous, often undiagnosed expansion of the aorta that can lead to sudden cardiac death. To identify the genetic basis of aortic size, a team led by James Pirruccello, Mark Lindsay, Patrick Ellinor, institute member and director of the Cardiovascular Disease Initiative and the Precision Cardiology Laboratory, and colleagues trained a deep learning model to evaluate 4.6 million cardiac MRIs from the UK Biobank, informing a genome-wide association study of nearly 40,000 individuals. The researchers reported novel loci associated with ascending and descending aorta size and explored their relationships to other traits, including aortic aneurysm, aiming to support diagnostic and therapeutics development. Read the full story in Nature Genetics.

Perturbation of paralog pairs

Screening studies have identified many single-gene dependencies in cancer, but new targets for many common drivers haven’t been readily found, in part due to functionally redundant paralogous genes. A team led by Takahiro Ito, core institute member and Cancer Program director Bill Sellers, and colleagues in the Genomics Platform and Genetic Perturbation Platform developed an approach to perturb pairs of genes and profile dependencies of 815 distinct paralog families. Dual deletion of DUSP4 and DUSP6 selectively impaired growth in NRAS and BRAF mutant cells through hyperactivation of MAPK signaling. This approach can unveil novel digenic vulnerabilities, which could represent new therapeutic targets. Read more in Nature Genetics.

BMI mediates diabetes risk from cholesterol drugs

Drugs that lower LDL cholesterol modestly increase the risk of type 2 diabetes, and reduced LDL is also associated with weight gain. Whether the effect on diabetes risk is mediated through increased body mass index (BMI) is unclear. In Diabetes Care, postdoctoral scholar Jordi Merino in the Metabolism Program and collaborators used multivariable Mendelian randomization analysis to investigate the interplay between LDL cholesterol, weight gain, and diabetes risk in more than 900,000 people. They found that the diabetogenic effect of lowering LDL cholesterol was partially mediated by increased BMI. This finding could advance our understanding of diabetes pathophysiology and help people using these medications take measures to prevent diabetes.

Cellular view of kidney disease

Many kidney diseases originate in the glomerulus, the filtration unit of the kidney. Disruption of cell-cell interactions between podocytes, endothelial cells, and mesangial cells of the glomerulus contribute to chronic kidney disease (CKD). Abbe Clark, Kidney Disease Initiative director and institute member Anna Greka, and colleagues used single-cell RNA sequencing to identify, in a mouse model, the earliest changes in intraglomerular cell-cell interactions driven by podocyte injury that ultimately lead to CKD. The resulting detailed single-cell map of the post-injury mouse glomerulus may help identify actionable targets for much-needed CKD therapies. Read more in the American Journal of Pathology.

To learn more about research conducted at the Broad, visit, and keep an eye on