CRISPR has been grabbing headlines because of its potential as a form of gene therapy. But CRISPR also deserves attention because of its contributions to genome-wide screening. Although these contributions usually escape public notice, they are nothing less than revolutionary.
Soon after CRISPR gene editing was introduced to the world in 2012, researchers led by Feng Zhang, PhD, core institute member at the Broad Institute of MIT and Harvard, set to work building a genome-scale loss-of-function screen that relied on the CRISPR-Cas9 nuclease instead of small interfering RNA (siRNA).1 Then CRISPR screening quickly began acquiring various refinements. (For details, see the below sidebar, “Establishing the Basics of CRISPR Screening.”)
By now, CRISPR screening has become a familiar technique. But it continues to become more powerful and sophisticated. In this article, we will see how CRISPR screening is enabling advances in drug development, functional human genomics, and basic science. Also, we will look into blue-sky developments that will allow CRISPR screening to do even more.
Characterizing coding and noncoding elements
Neville Sanjana, PhD, who worked with the Zhang laboratory on early CRISPR screens, says his group at the New York Genome Center (NYGC) and New York University (NYU) is now developing gene editing technologies for functional genomics applications, with the ambitious goal of understanding the function of all the elements of the human genome—both coding and noncoding.
The Sanjana laboratory has been developing screens that do more than just target genes. “[We have been] looking at noncoding regions like transcription factor binding sites [or] noncoding RNAs, or [at] regions deep in intergenic space that are associated with common or rare diseases,” Sanjana says. “[Some of this work] has led to amazing clinical translation.”
The principal example of that clinical translation was a noncoding CRISPR screen that identified a region of the genome involved in repressing fetal hemoglobin. The screen showed that when that noncoding region was perturbed, fetal hemoglobin could be derepressed. This finding led to a new therapeutic target for sickle-cell disease (SCD) and β-thalassemia, two diseases affecting adult hemoglobin.2
“The exact guide RNA from our CRISPR screen is what is actually used for the first-in-human CRISPR therapy,” notes Sanjana. That breakthrough therapy uses CRISPR-Cas9 gene editing to repress the BCL11A gene specifically in blood stem cells, causing them to reactivate the fetal hemoglobin gene. The engineered cells were then successfully used to treat two patients—one with SCD (Victoria Gray), and one with β-thalassemia. (The work was led by investigators from CRISPR Therapeutics and Vertex Pharmaceuticals.)
A major challenge in gene editing therapies is getting edits to happen in the right cell types. In this case, cell-type specificity is a product of editing the noncoding genome, because this particular noncoding region is functional only in blood cells, according to Sanjana.3
Identifying host factors for SARS-CoV-2 infection
More recently, Sanjana and colleagues have tackled COVID-19 by deploying a CRISPR screen to knock out every gene in the human genome in lung cells to find genes and pathways required for SARS-CoV-2 infection. Sanjana explains that the goal was to find multiple points of attack for blocking the SARS-CoV-2 virus.
The approach was inspired by the successful development of antiretroviral cocktails for HIV that prevent viral escape mutations. Those mutations otherwise occur when the virus is challenged with a single antiviral drug. For some of the top candidate genes involved in viral entry, a search for existing drugs that inhibit the protein products of these genes came up empty.
“We thought, let’s see if there might be some other common mechanism or convergent pathway that we can target,” Sanjana recalls. To conduct this work, Sanjana and colleagues picked up another new CRISPR screening tool. The tool is called “enhanced CRISPR-compatible indexing of transcriptomes and epitopes with sequencing,” or ECCITE-seq. It takes CRISPR screening to the single-cell level while detecting proteins and transcripts in parallel.
That analysis pointed to the cholesterol biosynthesis pathway and increasing intracellular cholesterol as a potential block on viral infection. This key discovery opened a wealth of heart disease drugs for potential repurposing. With the help of other team members, Sanjana zeroed in on the calcium channel blocker amlopidine, an FDA-approved drug, and found that it had a powerful inhibitory effect on the virus.
Commercializing CRISPR screening
Companies specializing in CRISPR screening for drug discovery and development are taking advantage of single-cell analysis capabilities. For example, single-cell CRISPR analysis is a key service of Cellecta, which offers pooled lentiviral libraries with barcoded guide RNA (gRNA) expression cassettes to enable the analyses known as Perturb-seq and CROP-seq (CRISPR droplet sequencing).
In these analyses, genetic perturbations (gene knockouts or knockdowns) are applied, and the resulting phenotypes are studied at the level of the transcriptome to infer genetic function. Barcoding of the gRNAs allows the pooled perturbations to be deconvoluted and associated with specific phenotypes.
“A typical use is an experiment in a heterogeneous cell population,” says Donato Tedesco, PhD, Cellecta’s director of research and development. “If you want to identify a subset of cells within the general population that have different responses to a knockout, the only way to do that is single-cell analysis.”
He relates that Cellecta’s customers often use CRISPR screens looking for knockouts that either antagonize or synergize with an experimental drug, in experiments aiming at elucidating the drug’s mechanism of action.
Another company offering CRISPR screening services is Synthego, which works with pharmaceutical and biotechnology firms. For Robert Deans, PhD, chief scientific officer of Synthego, riding the single-cell analysis trend in CRISPR screening comes down to automation and robotics—and lots of it.
Synthego has been applying machine learning and automation to handle the massive data output from these types of experiments. That powerful analysis feeds back into the screening. The company can boost efficiency by tweaking aspects of guide design and optimizing the ribonuclear protein complex.
“The name of the game, both in target identification and in editing for GMP, is precision,” Deans says. “You’re reducing the complexity of the screen. With greater precision and accuracy, your downstream interrogation simplifies.”
He reports that last year, the company used powerful data crunching and an efficient solid-phase synthesis platform to help a research collaboration led by the University of California, San Francisco, accelerate the study of potential treatment targets for coronaviruses. Synthego synthesized more than 1,000 gRNAs in just two days, and it used these gRNAs to engineer cells that incorporated more than 300 knockouts to screen host–coronavirus protein interactions. Just one person was needed to scribe in the automation and workflow.
The collaboration assisted by Synthego expanded on an earlier study, one in which 332 high-confidence SARS-CoV-2 protein–human protein interactions were mapped to identify drug repurposing opportunities.4 In the earlier study, several drugs were mentioned, including plitidepsin, a cancer drug that is marketed by PharmaMar.
To expand on the earlier study, the collaboration in which Synthego participated sought to determine which coronavirus protein–human protein interactions (not just those involving SARS-CoV-2, but other coronaviruses) might be good targets for drugs meant to block viral replication.5 The effort involved a combination of in vitro virus infectivity assays and in silico modeling.
In a press release about the pan-coronaviral study, Synthego included a quote attributed to Nevan J. Krogan, PhD, a UCSF researcher and one of the study’s leaders: “The precision and reproducibility of CRISPR were key to helping us study how SARS-CoV-2 affects cellular pathways and ultimately causes disease, enhancing our validation of promising therapeutic targets that may offer broad protection against infection from coronaviruses.”
Both the SARS-CoV-2 and pan-coronaviral interactome studies informed a subsequent study that focused on plitidepsin. This study, which included Krogan among its corresponding authors, indicated that the drug is effective against SARS-CoV-2 because it targets the host protein eEF1A.6 Plitidepsin is now the subject of a Phase III study that is set to enroll patients requiring hospitalization for the management of moderate COVID-19 infection.
Improving the analysis of multimodal perturbation screens
A new computational approach to analyze data from CRISPR screening has been developed by NYU and NYGC researchers. The approach, which is called mixscape, can improve the signal-to-noise ratio in studies that combine the use of a pooled CRISPR screen to perturb genes, and the use of multimodal single-cell sequencing technologies to amass mRNA and surface protein profiles. Essentially, mixscape identifies and removes confounding sources of variation.
According to a recent study, the researchers developed mixscape to help them understand how cancer cells alter the regulation of key genes, such as the gene that encodes the immune checkpoint molecule PD-L1, to avoid detection and evade the body’s immune system.7 Before mixscape, the researchers had difficulty interpreting data from a multimodal ECCITE-seq screen, mainly because a subset of cells “escaped” perturbation by the screen’s gRNAs, creating a lot of noise in the data.
The study’s lead author, Efthymia Papalexi, says the idea for mixscape was borrowed from the field of image recognition. Mixscape models each perturbation as a mixture of different cell responses. By doing so, it can identify and remove cells that have escaped CRISPR perturbation, allowing researchers to focus on the true biological signals.
“With this approach, we can knock out a gene, discover the molecular pathways it controls, and then associate these pathways to cellular behaviors,” Papalexi explains. For example, when applying this method in cancer cell lines, the researchers discovered that two genes frequently mutated in lung cancers—the gene for the kelch-like protein KEAP1, and the gene for the transcriptional activator NRF2—regulate the expression levels of PD-L1. This finding suggests that the genes are important in tumor development and progression.
Looking into the future of CRISPR screening
The horizon beyond these foundational CRISPR-Cas9 screening components and conditions is already being explored. The Broad Institute’s John Doench, PhD, is blazing a trail in the area of alternate Cas enzymes. The standard CRISPR-Cas9 system originates from Streptococcus pyogenes. But CRISPR systems are abundant in nature, and there are plenty of other Cas nucleases that can be exploited, says Doench, who is focusing on a protein called Cas12a from Acidaminococcus.
“It has a particularly useful property: if you want to express multiple gRNAs, you can do so from just one small transcript,” Doench points out. “Because of that, the Cas12a enzyme is particularly useful for studying combinations.”
Combinations are key to exploring synthetic lethality, where a combination of genes contribute to cell death, rather than one alone. In cancer drug discovery, synthetic lethality can link two or more genes that, individually, might not be effective targets, but when combined are lethal to cancer cells. BRCA1 and PARP are the most famous synthetic lethal pair. Doench asserts that with Cas12a, it’s now possible to screen for synthetic lethal pairs to potentially identify new cancer targets.
In addition, blue-sky variations on CRISPR screening are emerging to build upon or even replace current technologies. Those include techniques like CRISPR activation, where an enzymatically inactive Cas9 is tailored to activate gene expression, and base editor screening, where individual mutations or variants are targeted instead of genes. With that kind of fine-grained targeting, researchers could pin down the effect of a single amino acid change on a drug’s activity.
Base editor screens could finally crack the so-called V-to-F (variants to function) challenge for the human genome, enabling a map of every possible variant of every human gene with a function. A database containing all variants with their functions could unlock the vast potential for personalized medicine.
Realizing all of these possibilities with CRISPR screening will require upgrades in other technologies. “We still need lots of good systems, and lots of good assays to read out,” Doench states. “[CRISPR screening] is one of the tools in the toolbox, but it’s by no means the only one.”
References
1. Shalem O, Sanjana NE, Hartenian E, et al. Genome-scale CRISPR-Cas9 knockout screening in human cells. Science 2014; 343(6166): 84–87. DOI: 10.1126/science.1247005.
2. Canver MC, Smith EC, Sher F, et al. BCL11A enhancer dissection by Cas9-mediated in situ saturating mutagenesis.Nature 2015; 527(7577): 192–197. DOI: 10.1038/nature15521.
3. Frangoul H, Altshuler D, Cappellini MD, et al. CRISPR-Cas9 Gene Editing for Sickle Cell Disease and β-Thalassemia.N. Engl. J. Med. 2021; 384(3): 252–260. DOI: 10.1056/NEJMoa2031054.
4. Gordon DE, Jang GM, Bouhaddou M, et al. A SARS-CoV-2 protein interaction map reveals targets for drug repurposing.Nature 2020; 583: 459–468. DOI: 10.1038/s41586-020-2286-9.
5. Gordon DE, Hiatt J, Bouhaddou M, et al. Comparative host-coronavirus protein interaction networks reveal pan-viral disease mechanisms.Science 2020; 370(6521): eabe9403. DOI: 10.1126/science.abe9403.
6. White KM, Rosales R, Yildiz S, et al. Plitidepsin has potent preclinical efficacy against SARS-CoV-2 by targeting the host protein eEF1A.Science 2021; 371: 926–931. DOI: 10.1126/science.abf4058.
7. Papalexi E, Mimitou EP, Butler AW, et al. Characterizing the molecular regulation of inhibitory immune checkpoints with multimodal single-cell screens.Nat. Genet. 2021; 53(3): 322–331. DOI: 10.1038/s41588-021-00778-2.
