Using CRISPR Single-Cell Screening to Reveal Gene Function at High Resolution

Genome-wide functional genomics has been transformed by CRISPR-based screening approaches, enabling systematic assessment of gene function at a large scale. In practice, these screens rely on the efficient delivery of CRISPR components into the target cells. Cells used for a pooled lentiviral CRISPR screen must express both Cas9 and guide RNA (gRNA). This can be achieved either using an All-in-one vector expressing Cas9 and gRNA or by introducing gRNAs into a cell line that expresses Cas9 or dCas9 fusion systems (for example, CRISPR interference (CRISPRi) or CRISPR activation (CRISPRa)). The CRISPR library contains a mixture of gRNAs targeting hundreds to thousands of gene targets that are introduced into a large cell population where ultimately each cell hosts a single gRNA integrant. In a pooled CRISPR screen, next-generation sequencing (NGS) is used to identify the spacer of each gRNA, allowing identification of gene perturbations that impact viability or proliferation, using the gRNA abundance as the readout.

CRISPR single-cell (CRISPRsc) screening builds upon the traditional pooled CRISPR screen by adding molecular resolution through the integration of advanced single-cell technologies with the precision of CRISPR tools. By combining targeted gene perturbation using CRISPR with high-resolution profiling of single-cell transcriptomics (including gRNA capture), each gene perturbation can be directly linked to transcriptional consequences.

In this article, two Revvity scientists, Josien Levenga, Ph.D., R&D Manager on the Dharmacon Reagents R&D team, and Clarence Mills, Sr. Scientist on the Dharmacon Reagents R&D team, address what CRISPRsc screening is, outlining its core methods, as well as its benefits and limitations.

Biocompare: What are the benefits of CRISPR single-cell (CRISPRsc) screening?

Josien Levenga: “While a traditional pooled CRISPR screen reveals whether a gene affects cellular fitness and can be scaled rapidly, a CRISPRsc screen provides mechanistic resolution to understand gene function. Using CRISPRsc screening technology, scientists can systematically dissect cellular heterogeneity, allowing them to map gene regulatory networks and uncover how each gene perturbation influences cellular state at high resolution. Through computational analysis, this technology has led to discoveries of new or rare cell states, and it can tease apart direct versus downstream effects. It allows for comprehensive profiling of the transcriptome within complex cellular ecosystems. Integrating pooled CRISPR screens with single-cell technology enables scientists to gain more profound insights into cellular behavior.”

Biocompare: How do CRISPRsc screening platforms differ from traditional pooled screens?

Clarence Mills: “As Josien said, single-cell CRISPR screening offers much higher resolution than traditional pooled screens, but the workflows needed to achieve this are consequently more complex and come with the additional cost considerations.

“Single cells must be isolated, and the gRNA expressed in each must be identified to link each perturbation with the transcriptomic and/or proteomic readout. A key differentiator between scCRISPR platforms is how gRNAs are captured. Some methods, such as optical pooled screening, rely on RNA polymerase II expression of the guide spacer from a CROP-seq style vector. With other platforms, like 10x Genomic’s 5’ CRISPR chemistry, S. pyogenes Cas9 gRNAs transcribed from most library vectors can be captured directly. Earlier generation Perturb-seq methods utilize a Pol II expressed barcode as a proxy for spacer identity or capture the gRNAs through a synthetic sequence encoded in the guide scaffold. Far more sequencing depth is needed to analyze these screens and with that comes more complex analyses. Transcriptomic and/or proteomic data from cells with matched perturbations are compared to cells expressing non-targeting controls and/or those subjected to different treatments. In a large screen this can be hundreds or thousands of differential expression analyses.

“As a result, Simon Scrace from the Preclinical Service team at Revvity Discovery says for now these screening technologies are better suited to understanding mechanism of action and target validation rather than genome-wide target ID.”

Biocompare: What are the most important factors to consider when designing single-cell CRISPR libraries?

Josien Levenga: “CRISPRsc screening requires significantly more sequencing compared to a pooled CRISPR screen. For every target screened, increasing the number of guides per target and controls scales up the required sequencing depth to maintain adequate cell numbers per guide and, as a result, the overall cost of the experiment. The use of uniform pooled libraries, those with low 90/10 skew ratios, ensures that most targeted genes are perturbed in statistically meaningful numbers of cells. Combinatorial libraries can multiplicatively increase this requirement for sequencing depth as well as push the limits of cell numbers that can be perturbed in practice. As Clarence mentioned, several CRISPR single-cell platforms are available, and therefore it’s important to decide which technology to use for gRNA capture (e.g., direct capture or barcoding), particularly if you already have a screening library in hand. As discussed earlier, the ability to process more complex data sets will be a consideration as well.

“From a technical standpoint, the Revvity Preclinical Services team also mentions that it is essential to consider the cell model in combination with the given method; it’s important that the cells survive the manipulation, as sensitive cells might not survive the procedure, which will lead to a decreased recovery rate and reduced coverage. Furthermore, it is important to determine at what time point the samples need to be collected: are you only interested in gene expression or does the experiment require change in protein expression as well? And do you take long-lived proteins into consideration as well?”

Biocompare: Do you have any suggestion on how to optimize gRNA delivery and perturbation efficiency in single cells?

Clarence Mills: “Perturbation efficiency is especially important in pooled single cell screens as they are typically limited in size. Most gRNAs in the library should produce the desired perturbation in most cells in which they are expressed to maximize statistical power. Use a high-performing, validated algorithm when designing your library to ensure gRNAs are efficient and specific. Perturbation efficiency tends to correlate to CRISPR effector expression, particularly with CRISPRa and CRISPRi. For optimal expression, select a vector with a promoter that has robust activity in your cell model and subject the cells to stringent selection following delivery.

“The Poisson model of lentiviral integration relies on the assumption that each integration event is independent and can over-represent the number of single integration events. In a pooled screen where many gRNAs are co-packaged together, this results in more cells expressing multiple guides and therefore multiple perturbations that can confound analysis. For single cell screens, we recommend transducing cells at lower functional multiplicity of infections (MOIs) of 0.1–0.2 compared to the conventionally targeted MOI of 0.3 in traditional pooled screening to minimize multiple integration events.”

Biocompare: Are there limits to scalability in single-cell CRISPR screening and, if yes, how can they be resolved?

Clarence Mills: “Single-cell CRISPR screens are increasingly being performed at the gene family and even genome scale! That said, the method of single-cell capture and sequencing depth required to analyze a statistically meaningful number of perturbed cells are still limiting factors. With a minimum coverage of 50–100 single cells per gRNA, a 4 gRNA/gene whole genome library would require isolation and sequencing of 4 to 8 million single cells, before accounting for doublets, cells without detectable gRNAs, and multiple integration events.

“Workflows with rapid capture steps or those that are amenable to fixed cells can be easier to scale. Furthermore, there are published methods that enrich a representative panel of transcripts to reduce the number of sequencing reads needed per cell. More compact libraries also enable larger screens: some researchers use dual gRNA libraries to reduce the library size while increasing the likelihood that each unique construct produces a detectable perturbation. And single-cell platforms are improving rapidly, making large-scale CRISPR screens progressively more practical.”

Biocompare: What are the most common pitfalls you see in customer-designed CRISPR screens and how would you recommend avoiding?

Josien Levenga: “Researchers should consider expectations and be aware of the costs and complexity of the experiments. CRISPRsc represents a completely different process downstream as it’s more complicated and requires specialized analysis. It is also limited by low throughput, as fewer cells can be analyzed, which can impact analytical quality. Controls are important, but be careful about selecting controls: having enough, but not so many that you add significant numbers of guides. Consider what number of cells you aim for in the final readout, which requires you to know how many cells you are likely to lose during the process.”

Biocompare: Do you have a recommendation of number of guides per gene?

Josien Levenga: “As Clarence mentions earlier, there are reasonably well-accepted rules of thumb for the number of guides per gene and the number of cells per perturbation. The ‘right’ number for a CRISPRi/CRISPRa or a CRISPR knockout depends on the effect sizes you expect and the single-cell RNA-sequencing depth. The typical recommendation is 3–6 guides per gene. This allows for variable guide efficiency and consensus calling at the gene level. This essentially means trusting a gene-level result only when multiple independent guides targeting that gene produce consistent effects. Multiple guides also reduce false positives from off-target effects or low-activity guides. Next, it’s important to decide how many cells per guide you need for analysis, as there will be cell loss during the experiments and filtering process. Most groups plan to start the experiment with 2–3 times more cells per guide than they want in the final analyzed dataset. So, if you aim for 100 cells per guide, then starting with 250–300 cells per guide is a good starting point.

“Taken together, CRISPR single-cell screening is an evolution of functional genomics, shifting the question from whether a gene matters to how it shapes cellular state. Because of the added costs, complexity, and data burden, these approaches are not (yet) a replacement for large-scale pooled CRISPR screens. But they uniquely enable high-resolution mechanistic insights, target validation, and dissection of cellular heterogeneity. Thoughtful experimental design by balancing the library size, guide efficiency, cell model suitability, sequencing depth, and analysis capacity, is essential to realize the full potential of single-cell CRISPR screening. Used strategically, this can complement traditional pooled approaches to deliver deeper biological understanding.”

Clarence Mills is a Sr. Scientist on the Dharmacon Reagents R&D team at Revvity where he develops next-generation CRISPR tools. He is the inventor of a patented CRISPR interference (CRISPRi) system that demonstrates enhanced gene knockdown efficiency compared with first generation of CRISPRi effectors. In addition, he has developed a tightly regulated inducible CRISPR system designed to minimize unintended basal activity and avoid leakiness, improving precision and experimental control.

Josien Levenga, Ph.D., is an R&D manager on the Dharmacon Reagents R&D team at Revvity. She leads efforts to advance the CRISPR toolbox using both synthetic systems and lentivirally expressed platforms. In addition, she ensures continued support for the importance of RNAi technologies in an increasingly complex transcriptomic landscape.

In addition, the authors got input from members of the Preclinical Service team at Revvity Discovery in Cambridge, U.K.