Homology-directed repair versus end joining

Cas9 opens the duplex. The allele you write down afterwards is whichever repair path won, in that cell, at that moment in the cell cycle. Homology-directed repair is the path knock-in drawings depend on. End joining is the path that usually dominates. This page is a troubleshooting map for that competition. The cut itself is described in how CRISPR-Cas9 editing works in research. What a sequenced scar is allowed to mean for a protein is in indels, frameshifts and protein loss.

Guide RNA, donor nucleic acid and genotyping reagents sit in the molecular biology catalogue. Specify the repair outcome you actually need when you use the quote request.

Three ways a break can close

Classical non-homologous end joining recognises the broken ends, often through Ku and DNA-dependent protein kinase, and ligates them. It is available in G1 as well as later in the cycle. At the blunt cut SpCas9 usually leaves, a few bases upstream of the PAM, the products are typically small insertions and deletions. A one-base insertion at the cut is common at many SpCas9 sites. Re-ligation without a scar also happens, which means a cut can occur and then vanish from a bulk assay.

Microhomology-mediated end joining, sometimes discussed together with polymerase theta-mediated end joining, is a different end-joining route. Short stretches of identical sequence, a few bases up to around 20, anneal and the intervening DNA is lost. If you placed the cut between two such repeats on purpose, the deletion can be more predictable than a random indel. That predictability is still end joining. It does not use the donor you added, and it does not restore an arbitrary new sequence.

Homology-directed repair copies information from a homologous template. In an unperturbed cell the template is usually the sister chromatid, which exists in S and G2. In a knock-in experiment the template is the donor you supplied. The machinery that invades that donor is a minority activity in many mammalian cell types, especially when classical end joining is intact. Quiescent and post-mitotic cells are poor hosts. A protocol copied from a cycling cancer line will not create an S phase where the biology does not have one.

The practical consequence is numerical and local. In many research lines, indel alleles outnumber precise donor insertions, sometimes widely. The ratio is a property of the locus, the donor, the cut-to-edit distance and the cell. It is not a constant to copy from a figure in a different species.

Cell cycle is an input, not a footnote

If the goal is a disruptive indel, end joining is the pathway you are already counting on, and forcing cells into S phase is optional rather than logical. If the goal is a donor allele, the cells need a chance to run homology-directed repair. That can mean using a cell type that cycles, timing delivery toward S or G2, or accepting a low rate and screening more clones. Chemical synchronisation exists and is stressful. It changes checkpoints and can change the phenotype you wanted the edit to explain. Use it only with a reason, and with the recognition that a synchronised culture is not the same biological object as the unsynchronised one.

Measure where you are before you blame the donor. A culture that is mostly G1 will under-represent homology-directed repair even when the donor sequence is perfect. Flow cytometry of DNA content is one way laboratories see that state. It is described as a method class in how a flow cytometer reads a cell. A viability dye is not a cell-cycle measurement.

Editing early after plating, while many cells are still recovering from passaging, is a different state again. Record the growth condition. Two "same" transfections done on a confluent plate and on a sparse cycling plate are not the same repair experiment.

A branch point when the donor sequence is missing

Sequence the locus. If you see only indels, the cut happened and the donor path lost. Check, in order, whether the donor matched this genome. Polymorphisms in the homology arms are a quiet killer of recombination. A donor built on a reference the parent does not share will behave as a poor template. Sequence the arms against the parent, or design the arms from parental sequence in GenBank or a local assembly.

Then check distance. For a single-stranded oligo, precise installation is usually planned with the cut within about 10 to 20 base pairs of the edited bases, and the yield often drops as that gap grows. A cut 80 bases away is a weak plan for an oligo. Move the guide, or move to a donor class that can carry more homology, as discussed in donor templates for a knock-in.

Then check whether the repaired allele can be recut. If the donor leaves the PAM and the seed intact, Cas9 cuts the successful allele and end joining destroys it. A blocking change in the PAM or the seed, silent if it falls in a codon, is part of the design whenever the intended edit does not itself destroy recognition. If you omitted it, a rare homology-directed event can be erased before you genotype.

If even indels are absent, you are not yet looking at a repair-pathway problem. You are looking at delivery, guide match, or chromatin. Fix that before you redesign the donor. An unedited control that shows the donor sequence is contamination, not repair. Stop and separate the genotyping from the donor stocks.

Patterns that send the troubleshooting the wrong way

A bright PCR band the size of the unedited locus does not mean repair failed. Small indels hide inside that band. A band that shifts does not mean the donor inserted. Microhomology can delete more DNA than a one-base scar and still have nothing to do with the oligo. Junction PCR with one primer outside the homology is the minimum for a knock-in claim. Internal primers amplify the donor whether or not it recombined.

Selecting for a marker on the donor selects for the marker, including random insertion. Those colonies can be devoid of the intended junction. Genotype before you expand the "positive" clone. The traps are laid out in selecting edited clones without fooling yourself.

Inhibitors of DNA-dependent protein kinase and related tricks are sometimes used to tilt the balance away from classical end joining. They are stressful, locus-dependent and easy to over-interpret. If you use one, the control is the same transfection without the inhibitor, sequenced the same way. A rise in donor alleles with a rise in cell death is a trade, not a better pathway in the abstract.

Off-target cuts are repaired by the same menu. A phenotype that appears without the on-target donor allele may be an off-target scar, a response to the break, or a passenger change. Do not repair the on-target story by adding adjectives. Assay the site.

Damage, approval and scope

A double-strand break is a lesion. Planning for it in a research cell line still requires the institutional biosafety and ethics decision for that line and that nuclease. This article does not authorise clinical use and does not give a procedure for human embryo or germline editing. Pathway diagrams are not a quality system. The WHO laboratory biosafety manual is background for laboratory practice, not a waiver.

If you suppress end joining with drugs or with genetic lesions in repair genes, you have changed the cell as well as the edit rate. Report that background. A clone selected under a repair inhibitor is not a neutral wild-type genome plus one allele.

Incubators, timing and an interrupted run

Homology-directed repair depends on cells that are actually progressing through the cycle. An incubator that overshoots because the room is hot, or a power cut that cools a room and stalls a culture, changes the fraction of cells in S phase before it changes anything about your donor sequence. After such an interruption, do not genotype a distressed plate and call the repair ratio biological. Recover the culture, or repeat the delivery. Write the passage and the confluence in the same note as the guide. Those facts explain more failed knock-ins than a missing comma in the oligo.

What to ask when repair is the dispute

Say whether you need indels or a donor allele, which donor class you intend, how far the cut sits from the edit, and which cell type must do the repair. Include the parental assembly or a statement that the arms were checked against the parent. Sequencing of junctions can be raised through the CRISPR validation sequencing enquiry reference as a discussion of method. That reference is not an operated genotyping service. The molecular biology pathway frames the rest of the experiment. Ask whether a quotation is possible for the reagents the plan actually uses. A request for knockout reagents will not arrive as a donor.