ӳ����ý

Welcome to the October 29, 2021 installment of Research Roundup, a recurring snapshot of recent studies published by scientists at the ӳ����ý and their collaborators.

Immunotherapy’s impact on a dangerous cancer complication

A new study finds immunotherapy may benefit people with leptomeningeal disease, a dangerous cancer complication in which tumor cells infiltrate tissue covering the brain and spinal cord. Researchers led by Sanjay Prakadan, Christopher Alvarez-Breckenridge, Samuel Markson, associate member Priscilla Brastianos of the Cancer Program and of Massachusetts General Hospital, and institute member Alex Shalek used single-cell RNA and cell-free DNA profiling to examine cerebrospinal fluid from patients over time in two clinical trials of immune checkpoint inhibitors. The numbers of certain immune cells and expression of particular genes were higher following treatment, suggesting promising benefits to patients and potential biomarkers to guide patient selection. Read more in and an .

An EVEolving view into variants' effects

Predicting human genetic variants' implications remains a major challenge; more than 98 percent of variants discovered so far have unknown consequences. Associate member Debora Marks in the Klarman Cell Observatory, Jonathan Frazer and Mafalda Dias (HMS), and Pascal Notin and Yarin Gal (Oxford) and colleagues have developed EVE (Evolutionary model of Variant Effect), an artificial intelligence model that uses nonhuman species' variation patterns to predict the human variants' impacts and performs on par with high-throughput lab experiments. After processing 36 million protein sequences and 3,219 disease-associated genes from multiple species, EVE proposed new classifications for 256,000 human variants of unknown significance. Learn more in , an HMS , and a by Jonathan.

Getting an eToehold on eukaryotic translation

Synthetic approaches for detecting and responding to RNAs could enable biotechnology applications ranging from cell-type targeting to genetic circuits. Previously, researchers developed “toehold switches”: RNA-based prokaryotic switches that can detect specific RNAs. Now, Evan Zhao, Angelo Mao, Emma Chory, institute member James Collins of the Infectious Disease and Microbiome Program, and colleagues have developed “eToeholds”, which function in eukaryotic cells. eToeholds enable controlled translation of a desired protein only in the presence of specific trigger RNAs. In , the researchers show eToeholds can induce a 16-fold change in translation of reporter genes, function in human and yeast cells, and can distinguish between cell types and states. Read more in an .

Prioritizing AML vulnerabilities

In vitro CRISPR screens have identified hundreds of genetic vulnerabilities in acute myeloid leukemia (AML). To help prioritize those with translational potential, Shan Lin, institute member Kimberly Stegmaier in the Cancer Program, and team developed Cas-9 competent patient-derived xenograft mouse models of AML and an in vivo CRISPR screening method to evaluate the physiological relevance of known genetic vulnerabilities in AML. They identified two promising targets, SLC5A3 and MARCH5. Inhibition of MARCH5, which inhibits apoptosis, improved response to venetoclax treatment, while inhibition of SLC5A3 impaired uptake of myo-inositol, a metabolite for AML. This screening approach could help identify more drug targets for AML. Read more in and a Boston Children’s Hospital .

ClusterMap: pinpointing RNA in cells and tissues

In situ transcriptomic methods generate maps of RNA and their locations inside intact tissues. Yichun He, Xin Tang, Jia Liu (Harvard), core institute member Xiao Wang, and colleagues have developed a computational tool called ClusterMap for integrative analysis of in situ transcriptomic data. ClusterMap is an unsupervised, annotation-free framework that clusters RNA into cell bodies, subcellular structures, and tissue regions in 2D and 3D space, by incorporating RNAs' physical location and gene identity and identifying biological structures by density peak clustering. ClusterMap works on various mouse tissue types and human cardiac organoids, and revealed gene expression patterns and other key features from images with high-dimensional transcriptomic profiles. Read more in .

A Tangram of single cell puzzle pieces

A biological atlas should link high resolution molecular and histological information with spatial data to provide insight into biological function. However, single-nucleus RNA sequencing generates transcriptome-wide data without spatial information, and spatial technologies are often limited in resolution, sensitivity, or throughput. Now, Tommaso Biancalani, Gabriele Scalia, Chuck Vanderburg, institute member Evan Macosko of the Stanley Center for Psychiatric Research, core institute member (on leave) Aviv Regev, and colleagues describe Tangram, a deep learning framework that connects single-nucleus RNA-seq data with various kinds of spatial and anatomical information. The team used Tangram to assemble an atlas of the somatomotor area of a mouse brain with approximately 30,000 genes at single-cell resolution. Read more in .

Analyzing protein modifications

As cells build proteins, certain molecules may be added to modify the proteins in different ways. To study the role of these changes, researchers need the ability to find and collect such altered proteins. In , a team led by Sam Myers (La Jolla Institute for Immunology), Rajan Burt, and institute scientist and senior director of Proteomics Steve Carr evaluated an approach using antibodies to capture proteins that are modified by the addition of N-acetylglucosamine to serine and threonine residues (a modification called O-GlcNAc). The method, tested in mouse brain tissue, simplifies the analysis of O-GlcNAc signaling and will help elucidate the role of this protein modification in health and disease.