Martin Jinek, Krzysztof Chylinski, Ines Fonfara, Michael Hauer, Jennifer A. Doudna, Emmanuelle Charpentier
8 min
Genome engineering in eukaryotic cells has historically relied on the design and production of complex proteins—such as Zinc-Finger Nucleases (ZFNs) and Transcription Activator-Like Effector Nucleases (TALENs)—that must be specifically engineered for every new DNA target site. Researchers sought to determine if the bacterial CRISPR/Cas9 system, which naturally uses simple RNA molecules to guide DNA cleavage, could be repurposed as a more flexible and efficient tool for genome editing in human cells.
The researchers tested the Type II CRISPR/Cas9 system in human HEK293T cells. They expressed a human-codon-optimized version of the Cas9 protein and a custom single-guide RNA (sgRNA) designed to target the CLTA (clathrin light chain A) gene. To assess whether the Cas9-sgRNA complex could successfully induce site-specific DNA breaks, the team utilized the Surveyor nuclease assay—a method that detects DNA mismatches created by the cell's imperfect repair of double-strand breaks via the non-homologous end joining (NHEJ) pathway. They also performed Western blotting, Northern blotting, and in vitro cleavage assays to analyze the expression, stability, and functional limitations of the system.
The study demonstrated that Cas9 and sgRNA can be successfully co-expressed and localized to the nucleus in human cells to generate targeted double-strand breaks. These breaks successfully triggered endogenous DNA repair mechanisms, resulting in detectable genomic mutations at the targeted CLTA locus. The researchers found that the expression and assembly of the sgRNA into the Cas9 protein act as the primary limiting factors for efficient cleavage. Furthermore, they discovered that extending the 3′ end of the sgRNA sequence significantly improved DNA targeting activity in vivo. This work established that the Cas9 system is a viable, easier-to-program alternative to current protein-based genome engineering methods.
This paper provides the foundational proof-of-concept for CRISPR-based genome editing in humans. Because the system relies on simple Watson-Crick base pairing between the guide RNA and the target DNA, rather than requiring the labor-intensive engineering of new proteins for every target site, it enables a much faster and more versatile approach to genetic modification. This technology has the potential to transform research in gene therapy, disease modeling, and biotechnology.
Type II CRISPR immune systems in bacteria use a dual RNA-guided DNA endonuclease, Cas9, to cleave foreign DNA at specific sites. We show here that Cas9 assembles with hybrid guide RNAs in human cells and can induce the formation of double-strand DNA breaks (DSBs) at a site complementary to the guide RNA sequence in genomic DNA. This cleavage activity requires both Cas9 and the complementary binding of the guide RNA. Experiments using extracts from transfected cells show that RNA expression and/or assembly into Cas9 is the limiting factor for Cas9-mediated DNA cleavage. In addition, we find that extension of the RNA sequence at the 3' end enhances DNA targeting activity in vivo. These results show that RNA-programmed genome editing is a facile strategy for introducing site-specific genetic changes in human cells.DOI:http://dx.doi.org/10.7554/eLife.00471.001.
Alex: Is matching the RNA enough, or does the DNA need another feature before Cas9 can cut?
Sam: It also needs a nearby recognition sequence called a protospacer-adjacent motif, usually shortened to PAM. In this paper, that requirement includes adjacent G bases. When the requirements are met, the guide pairs with the target DNA and Cas9 cuts both strands. So RNA matching makes targeting programmable, but the neighboring sequence still constrains target choice.
Alex: And the genetic change comes from repairing that cut. They aren’t demonstrating that the RNA itself writes a replacement sequence into the chromosome.
Sam: Correct. They test non-homologous end joining, a repair pathway that reconnects broken DNA ends. That repair can leave small insertions or deletions. Those changes are the editing outcome demonstrated here, rather than installation of a specified replacement sequence.
Alex: How did they distinguish those repair changes from simply having Cas9 and RNA present in the same cells?
Sam: They used the Surveyor assay, a test that detects sequence differences through mismatch-sensitive cutting. First they copied the targeted region from the cells’ DNA. Then they separated and rejoined the strands, allowing altered sequences to pair with unchanged ones. An enzyme cut the resulting mismatches, producing detectable DNA fragments.
Alex: So the assay detects the aftermath of repair, rather than watching the chromosome break directly. What makes that evidence persuasive here?
Sam: The signal depended on supplying both Cas9 and the targeting RNA. A zinc-finger nuclease pair aimed at a nearby site served as a positive control. They also sequenced cloned copies of the target region and identified four types of insertions or deletions. Those came from examining sixty-eight cloned DNA copies, not that many independent cell experiments.
Alex: What was the editing frequency, and how did it compare with the established tool?
Sam: The authors report mutation frequencies consistently between six and eight percent at this target. That was lower than the nearby zinc-finger nuclease comparison, though within the broader range reported for that approach and transcription activator-like effector nucleases. It establishes feasibility here, not superior efficiency.
Alex: A modest frequency could reflect several bottlenecks. How did they separate a protein problem from an RNA problem?
Sam: This is the most informative troubleshooting experiment. They broke open the treated cells and tested whether the extracts could cut a separate DNA molecule carrying the target. Extract from cells given Cas9 alone did not cut it. But adding laboratory-made guide RNA produced robust cutting.
Alex: That shows the protein in the extract can work when supplied with a guide. What happened with extracts from cells that had already received both components?
Sam: Those extracts could cut the target DNA. Adding more guide RNA substantially increased cutting further. The authors interpret that as evidence that guide availability, or loading the guide into Cas9, limits activity. It does not distinguish those possibilities, and adding RNA to an extract is not itself a demonstration of improved editing inside intact cells.
Alex: Did that diagnosis lead to a change they could test inside cells?
Sam: They extended the guide’s protein-binding region and its tail, drawing on the natural bacterial RNA sequence. The extended guides supported stronger targeting signals in the cellular assay. But the authors explicitly say more quantitative comparisons are needed to confirm improved efficiency. The direction looks promising; the size and reliability of the improvement remain unresolved.
Alex: The other concern is specificity. Cutting the chosen site doesn’t tell us whether other sites were also cut. How far does this study go on that?
Sam: It demonstrates editing at the selected CLTA site, but does not report a genome-wide test of unintended cutting. The cellular editing experiments also focus on that target in that cell line. My reading is that the evidence is convincing for feasibility, but cannot establish broad reliability across targets, cell types, or possible unintended sites.
Alex: The discussion also proposes targeting several sites with different guides in the same cell. Is that part of the demonstrated result?
Sam: No. The authors explicitly describe that as untested here. They propose it for targeting several genes or improving targeting at one site. Likewise, therapeutic and agricultural uses are potential applications, not outcomes measured in these experiments.
Alex: Given that scope, who should read the full document, and where should they start?
Sam: Researchers building genome-editing methods should read the Results, especially the cell-extract experiments with added guide RNA. They show how the authors identified a limiting step rather than just reporting mutations. Then read the Discussion’s efficiency comparison and qualifications. For everyone else, remember this: the targeting instructions can sit in an RNA, rather than a newly engineered protein.
Alex: Thanks for listening.