Non-homologous end joining (NHEJ) is the dominant double-strand break repair pathway in mammalian cells, operating throughout the cell cycle without requiring a template sequence. When CRISPR-Cas9 creates a double-strand break, NHEJ rapidly re-ligates the broken ends, but this process is error-prone and frequently introduces small insertions or deletions (indels) at the junction. If these indels occur within a protein-coding region and shift the reading frame, they effectively knock out gene function, making NHEJ the basis for the majority of CRISPR gene disruption experiments.
The error-prone nature of NHEJ, while useful for gene knockout, presents challenges when precise editing is desired. The indel spectrum at a given target site is not entirely random; certain repair outcomes are favored based on local sequence context and the microhomologies flanking the break. Researchers have used machine learning to predict indel outcomes, enabling the selection of guide RNAs that preferentially produce specific frameshift-causing deletions. Tools like inDelphi, developed at the Broad Institute, provide probabilistic predictions of repair outcomes that help guide experimental design.
In therapeutic genome editing, NHEJ-mediated gene disruption has advanced furthest in clinical development. The approach is simpler than HDR-based correction because it requires only the Cas nuclease and guide RNA without a donor template, simplifying delivery. Clinical programs from Intellia Therapeutics targeting transthyretin amyloidosis and from CRISPR Therapeutics targeting beta-globin regulators in sickle cell disease rely on NHEJ-mediated disruption of specific gene targets. The high efficiency of NHEJ in most cell types and its cell-cycle independence make it particularly suitable for in vivo therapeutic applications where controlling the target cell's proliferative state is not feasible.