How CRISPR-Cas9 Cuts DNA—and Why Cutting Is Not Correction
Explain SpCas9 guide RNA, PAM recognition and its two nuclease domains, then read DNA repair outcomes separately from gene-function evidence.
Publication & corrections: infostoa · Editorial standards
Individual author and specialist reviewer information is not provided for this article.
1. A guide and a cutting enzyme work together
CRISPR-Cas9 editing is adapted from bacterial defenses against foreign DNA. This explanation concerns Streptococcus pyogenes Cas9, or SpCas9. CRISPR systems do not all share the same enzyme, PAM or cutting pattern.
Natural crRNA and tracrRNA work together with Cas9. An engineered single-guide RNA (sgRNA) combines their roles in one molecule. Changing the target-specific guide region can redirect the system, but targeting a desired sequence is different from obtaining only the desired editing outcome.
2. PAM and cutting the two strands
SpCas9 needs a PAM adjacent to the DNA sequence corresponding to the guide. A representative PAM is 5′-NGG-3′, where N means any base. The PAM is a requirement in the target DNA, not a three-base segment to append to the guide.
In Jinek and colleagues' 2012 biochemical experiments, the HNH domain cut the DNA strand complementary to the guide, while the RuvC-like domain cut the opposite strand. Representative plasmid cleavage occurred three base pairs upstream of the PAM; broader positions were observed for the noncomplementary strand of short DNA substrates. A perfectly identical clean cut should therefore not be assumed for every substrate. The work also demonstrated redirection using a guide that joined the two RNAs.
3. Cellular repair determines the outcome after cutting
Cong and colleagues' 2013 study demonstrated editing at targeted loci in human and mouse cells. Repair through nonhomologous end joining (NHEJ) can produce insertions and deletions, or indels. An indel alone does not establish loss of protein function: the actual sequence change and the relevant gene function need checking.
Homology-directed repair (HDR) using homologous donor DNA can introduce desired sequence changes. At EMX1, the same study introduced restriction sites using a repair template and confirmed the outcome by sequencing. Providing a donor does not mean every cell acquires the desired edit.
For reading results, distinguish evidence of cutting → repaired DNA sequence → RNA, protein or functional change. The original paper used SURVEYOR assays and Sanger sequencing and showed different insertion/deletion outcomes. This reading sequence is explanatory, not an experiment performed here. Some guide–target mismatches retained activity, so successful on-target editing does not establish the absence of off-target cutting. Clinical treatment and agricultural effects or safety require evidence beyond these foundational experiments.