Skip to main content
Premium Trial:

Request an Annual Quote

Broad, UCSF Researchers Combine CRISPR, Single-Cell RNA-Seq for Complex Phenotype Screening


NEW YORK (GenomeWeb) – In two papers published today in Cell, researchers from the Broad Institute and the University of California, San Francisco described research carried out on a new experimental platform combining CRISPR-based pooled screens and single-cell RNA sequencing.

Coupling a random oligonucleotide barcode to each DNA-targeting single guide RNA (gRNA), the system, dubbed "Perturb-seq," provides researchers information on cells that have received one or more genetic perturbations, alongside the rich readout of single-cell transcriptional profiling.

Meanwhile, another study led by Ido Amit of the Weizmann Institute and also published today in Cell, describes a separate platform combining single-cell RNA-seq and pooled CRISPR screening. The three papers further establish a strong trend toward merging single-cell RNA-seq with CRISPR. Earlier this year Christoph Bock of the CeMM Research Center for Molecular Medicine of the Austrian Academy of Sciences posted a paper to the preprint sever bioRxiv on a similar type of experimental platform.

"It used to be that you had to choose: You could get the scale of a pooled screen but the readout was simple, or you could get a sophisticated readout from microscopy, or RNA-seq, or many kinds of assays, but then you only could do it for a small number of things," Aviv Regev, core Broad member and a senior author on one of the papers, told GenomeWeb.

Perturb-seq helps provide the best of both worlds. Combining the technologies "will greatly speed our ability to understand how different genes that encode for the components of cells are normally wired together," Jonathan Weissman, a professor of cellular and molecular pharmacology at UCSF, said in a statement.

Furthermore, the platform can incorporate a number of different types of perturbations. Weissman and Regev collaborated while developing the new technology, but each led a study in their biological application of interest. Regev's lab used Perturb-seq featuring CRISPR/Cas9-induced frame shift mutations to study genetic interactions in dendritic cells, while Weissman's lab incorporated CRISPR-based transcriptional interference (CRISPRi) to study the unfolded-protein response in K562 cells.

"We wanted to show that we can put it in lots of settings and it gives us valuable results," Regev said.

Achieving the scale of genome-wide screens and rich readouts at the same time was a longstanding goal for Regev. She collaborated with Harvard University's Steven McCaroll and David Weitz on Drop-seq, a massively parallel RNA-sequencing technology, publishing a study on it last year.

"[Drop-seq] serves many purposes, but for us, one of the motivations was to be able to do it along with CRISPR screening," she said.

Regev visited Weissman's lab in early 2015 and the collaboration on Perturb-seq began in earnest around that time. With droplet-based RNA-seq working, she wanted to combine it with pooled screening. The two labs agreed they would benefit from collaboration. They designed the Perturb-seq vector together and the first tests were performed at the Broad.

"When things were working promisingly, we thought there was value to test it in different modalities," Regev said.

Regev's lab chose to use CRISPR/Cas9-based editing in dendritic cells, looking at transcription factors regulating the cell's response to lipopolysaccharide. It's a setup her lab is extremely familiar with: Of the 24 transcription factors interrogated, Regev said her lab already had ChIP-seq data on 22 of them, providing an important positive control.

"Our idea was that we'd be able to compare this to the ChIP-seq data and see if they also bind those genes' promoters and enhancers," she said.

An important aspect of the study looked at the number of cells needed to provide results on each perturbation, and how many sequencing reads per cell would be necessary.

"We definitely went beyond the level we thought would be necessary," she said, assaying hundreds of cells per perturbation and sequencing the library "well beyond saturation." There's a point at which extracting more data no longer provides information, but that can only be done computationally. "You need to go beyond what you actually need," Regev said. "We thought we'd be well positioned to make that investment and then give that to others so they wouldn't have to do the same in their system."

Regev and her co-authors provided some information about where that point of diminishing returns is in the study. "It matters what the answer you want to get at the end is," she said. "If you want an answer about gene signatures, such as inflammatory response, you can go for pretty low numbers of cells — in the dozens — and very low numbers of reads per cell. But if you want answers on the effect of every individual perturbation and each gene target, then you need a higher number of cells and more reads per cell — maybe hundreds of cells."

"That's valuable because most people want to know a lot about signatures and they won't need to know every individual gene and how it's affected," she said. "They can save a lot of money this way, or assess more perturbations for the same amount of money."

Regev added that the preliminary results suggest generally that genetic interactions are rare. "If this were not true, then we'd likely be in trouble as organisms," she said. "If everything interacted with everything, every time you changed something, it would all go awry."

But that's not to say there aren't interactions. And the results suggest that the rules governing them can be inferred, she said. When things do interact, they form modules that affect things in similar ways. "Instead of having to assay not only every pairwise interaction, but three- and four- way interaction, we can see if we can randomly assess the system and still understand a lot about how it's made," Regev said.

Meanwhile, Weissman's lab used CRISPRi to simultaneously repress up to three genes in a K562 cell line. Their analysis allowed them to investigate the unfolded protein response, a quality control pathway that senses stress and detects errors in protein production machinery.

"These studies provide insight into how the three separate sensors of [endoplasmic reticulum] homeostasis allow the cell to monitor distinct types of ER stress," Weissman and his co-authors wrote, adding that their study highlights the ability of Perturb-seq to generally dissect complex cellular processes.

"With the unbiased readout we get from single-cell RNA sequencing, we can potentially discover things about biological pathways without a prior hypothesis," Weissman lab postdoc Tom Norman, co-first author of the UCSF-led study, said. "It opens up new possibilities that might not be evident from more targeted studies."

"The fact that Perturb-seq is not limited to one type of CRISPR-mediated perturbation should be incredibly enabling and allow other research groups to take advantage of the new method," said Britt Adamson, co-first author of the UCSF-led study and a postdoc in the Weissman lab.

To that end, plasmids will be made available for research uses on Addgene, though the Broad and UCSF said in a statement that they have applied for patents related to Perturb-seq.

The Scan

Shape of Them All

According to BBC News, researchers have developed a protein structure database that includes much of the human proteome.

For Flu and More

The Wall Street Journal reports that several vaccine developers are working on mRNA-based vaccines for influenza.

To Boost Women

China's Ministry of Science and Technology aims to boost the number of female researchers through a new policy, reports the South China Morning Post.

Science Papers Describe Approach to Predict Chemotherapeutic Response, Role of Transcriptional Noise

In Science this week: neural network to predict chemotherapeutic response in cancer patients, and more.