Perturb-ME combines genome-wide CRISPR knockout, cell sorting, and single-cell analysis to map cellular mechanisms
Perturb-ME combines genome-wide CRISPR knockout, cell sorting, and single-cell analysis to map cellular mechanisms
Researchers at Genentech and the Broad Institute introduced Perturb-ME, a method for mapping causal networks, in an August 18 preprint. The approach combines genome-wide gene knockout with cell selection based on target protein levels, followed by simultaneous measurement of the knocked-out gene, RNA, and surface markers in each cell. In an experiment using melanoma cells, the authors reconstructed a regulatory network comprising 221 regulatory genes and 1998 response genes. They made the code and computational pipelines available in an open repository.
A central obstacle in the study of complex cellular processes is the gap between large sets of genetic correlations and testable causal relationships. When biologists want to determine how each gene controls a phenotype, they must choose between two limited approaches. A pooled genetic screen knocks out thousands of genes but shows only the overall change in cell survival or protein abundance across a cell culture, without revealing the molecular mechanism. Perturb-seq measures the activity of thousands of genes in each individual cell after targeted disruption of a DNA region. Applying this analysis across the entire genome, however, requires sequencing millions of cells and remains too expensive.
The new Perturb-ME approach (Perturb-seq with Marker Enrichment) addresses this limitation through a two-stage design. First, the CRISPR system is used to knock out genes across the genome in a pooled cell population. The population is then passed through a flow cytometer, which selects only the lowest and highest five percent of cells according to the level of the protein being studied. This step removes cells without a pronounced response and enriches the sample for mutations that produce a measurable effect.
Only after sorting do the researchers subject the cells to multimodal single-cell profiling. Within each cell, they simultaneously measure the guide RNA, the messenger RNA profile, and surface protein abundance using DNA-tagged antibodies.
Perturb-seq enables pooled genetic screens with rich single-cell profiling, but genome-wide analysis remains expensive and may not be linked to functional phenotypes. Perturb-ME, combined with agent-based interpretation, provides a scalable framework for functional discovery.
The authors tested the method using the major histocompatibility complex class I (MHC-I) in human melanoma cells. MHC-I presents fragments of intracellular proteins on the cell membrane, allowing cytotoxic T cells to determine whether a cell is healthy, infected with a virus, or transformed into a tumor cell. Reduced MHC-I abundance is a common mechanism by which tumor cells and senescent cells evade immune surveillance.
After sequencing 353 thousand selected cells, the authors constructed a mathematical model of regulation that identified 221 key regulators and 1998 response genes. These interactions formed seven functional modules. As a positive control, the model identified the canonical interferon-gamma pathway. It also revealed six other mechanisms that regulate antigen presentation, including the retromer system, which controls endosomal sorting and protein recycling, quality-control systems in the endoplasmic reticulum, vesicular transport through the Golgi apparatus, and maintenance of protein homeostasis.
The development team included Genentech research director and Human Cell Atlas cofounder Aviv Regev, who helped develop the original Perturb-seq method in 2016.
Combining functional sorting with deep single-cell measurements makes it possible to construct detailed maps of intracellular mechanisms without sequencing the entire cell population. This approach shortens the path from lists of candidate genes to specific therapeutic targets.