How VNTRs affect human traits, investigating interneuron diversity, deep learning dives into prostate cancer, and more
Research Roundup: September 24, 2021
Welcome to the September 24, 2021 installment of Research Roundup, a recurring snapshot of recent studies published by scientists at the Broad Institute and their collaborators.
A view into VNTRs
Variable number tandem repeats (VNTRs) are repeated sections of the genome that stretch from seven up to thousands of base pairs in length, and their contribution to human traits are understudied. Ronen Mukamel, Bob Handsaker, institute member Steve McCarroll in the Stanley Center for Psychiatric Research, associate member Po-Ru Loh in the Medical and Population Genetics Program, and colleagues have identified five VNTRs in protein-coding regions that are strongly associated with nearly two dozen traits, including height, hair curl, and kidney disease risk. The team developed new methods to analyze human exome sequencing data and examine potential VNTRs, and also used data from 415,000 UK Biobank participants to assess whether length variation within VNTRs was associated with traits. Read more in Science and a Broad news story.
A tale of two interneurons
Cellular diversity underlies complex human thought, but scientists don’t fully understand how this diversity arises. In Nature, Kathryn Allaway, Mariano Gabitto (Flatiron Institute), Orly Wapinski, Richard Bonneau (Flatiron Institute), institute member Gord Fishell of the Stanley Center, and colleagues use cell-type specific gene regulatory networks to show that changes in transcription and chromatin structure control the development of two types of interneurons in the cerebral cortex — PV and SST cells — from a single progenitor. The findings could provide a model for studying cellular diversity and shed light on the etiology of neurodevelopmental and psychiatric disorders, which don’t affect all cell types equally. Read more in a Broad news story.
Catching prostate cancer with a P-NET
Molecular profiling technologies can yield an abundance of information about tumors, but researchers often struggle to synthesize that data into prognoses for patients. Using machine learning, affiliate researcher Haitham Elmarakeby, associate member Eliezer Van Allen of the Cancer Program, and colleagues developed P-NET, a biologically-informed and interpretable deep learning model that can differentiate between genomic profiles of prostate cancers that are lethal and those unlikely to cause symptoms or death. The interpretability of P-NET could help researchers identify candidate genes and pathways related to lethal prostate cancer for experimental and clinical validation. Read more in Nature and a Broad news story.
Ribosomes pick up the PACE
A team including Fan Liu, Siniša Bratulić, Alan Costello, and former Broad Fellow Ahmed Badran (now at Scripps Research Institute) developed a versatile platform for rapidly evolving ribosomal RNAs towards unnatural bioactivities, known as orthogonal ribosome-dependent phage-assisted continuous evolution (oRibo-PACE). They used the system to generate evolved rRNA mutants from three bacterial species, including mutants that enhanced the rate of protein synthesis in natural and engineered contexts, such as incorporation of noncanonical amino acids. The findings demonstrate that ribosomes can be evolved for improved protein yield, enhanced genetic code expansion, and faster translation rates in living cells. Read more in Nature Communications.
A better antibody builder
Antibodies can be powerful research tools, but despite the growth of a number of in vitro technologies, by and large, scientists still rely on animal-based production methods. Xun Chen, Matteo Gentili, institute member Nir Hacohen, and core institute member (on leave) Aviv Regev, all in the Cell Circuits Program, have developed CeVICA, a cell-free platform for engineering nanobodies (single-chain antibody fragments also called VHH antibodies) against desired targets. CeVICA combines synthetic nanobody library production, ribosome display, and computational prediction to rapidly generate and select nanobodies that bind targets efficiently and strongly. In Nature Communications, they discuss how they used CeVICA to engineer neutralizing nanobodies against SARS-CoV-2.