A glossary of genome-editing terms

A methods sentence can stack PAM, sgRNA, indel, and knockout as if they were one event. They are not. A glossary of genome-editing terms is a way to choose the word that matches the molecule or the result you actually have. The mechanism those words point at is in how CRISPR-Cas9 editing works in research. How far a sequenced allele can be pushed is in checking whether a genome edit worked. Use the entries below when a notebook, a figure legend, or an order form is about to blur two of them.

PAM

A protospacer adjacent motif is a short DNA sequence that the nuclease itself recognises beside the target. For the widely taught Streptococcus pyogenes Cas9, that motif is often written NGG. The guide RNA does not have to base-pair with those bases. The protein will generally ignore a perfect spacer match that lacks a motif it can accept.

The mistake this prevents is pasting the PAM into the spacer oligo, or designing a spacer for one enzyme and ordering a different Cas protein. Motifs are compared, including what a SNP can do to them, in PAM sequences and why the genome context matters.

sgRNA

A single guide RNA is one RNA molecule that joins a targeting spacer to the scaffold the Cas protein binds. Older experiments used a separate CRISPR RNA and tracrRNA. Both formats can position the same class of protein. The spacer is written 5-prime to 3-prime and is the part you change. The scaffold is the part the protein expects, and it is not a free choice of sequence.

The mistake this prevents is ordering only the spacer DNA and assuming a protein will use it, or calling every RNA in the tube a guide when the tube also holds a tracrRNA that must be annealed. Record which format you used, as set out in recording guides so an experiment can be repeated.

Indel

An indel is an insertion or a deletion of bases at a repaired break, relative to the sequence you chose as the reference. End joining often produces a mixture of small indels, not one designed allele. The size can be one base or many. A gel that cannot resolve one base will still show a band, so an indel can be present and invisible until the amplicon is sequenced.

The mistake this prevents is treating "edited" as a synonym for a specific insertion you drew. Unless a donor was used and the junctions were read, the honest noun is often a mixture of indels. Name the alleles you sequenced.

NHEJ

Non-homologous end joining is a repair route that rejoins a broken DNA end without using a long homologous donor as the template. In many mammalian cells it is the common fate of a Cas9 break. The products are the indels above. The route is not a single enzyme you pipette. It is a cellular activity, and its balance with other repair routes depends on the cell and the cell cycle.

The mistake this prevents is promising a precise base change from a cut alone. A cut plus NHEJ is a scar, and the scar is empirical. If the sentence you need is a defined new sequence, you are talking about a different repair path and you must show that path was used.

HDR

Homology-directed repair, in this literature, means the break was repaired using a homologous donor: a single-stranded oligo, a double-stranded DNA fragment, or a longer template with homology arms. The cell has to use that donor. Many cells do not. A donor in the transfection mix is evidence of intent, not evidence of outcome.

The mistake this prevents is calling every drug-resistant colony a knock-in. Random integration, or an indel that happens to destroy a restriction site, can mimic a crude screen. Read the junctions. The same caution applies when you compare bacterial homology repair with mammalian HDR. The cellular machinery is not interchangeable even when the word homology appears in both protocols.

RNP

A ribonucleoprotein is the Cas protein already bound to its guide RNA, delivered as a complex rather than as a plasmid that must be transcribed and translated inside the cell. Exposure of the genome to the nuclease is often shorter than with a stable expression plasmid. That is a kinetic difference, not a guarantee of purity or of on-target behaviour.

The mistake this prevents is writing "RNP" on an order when the tube you need is a plasmid, or assuming a fluorescent plasmid marker will be present when the delivery was protein plus RNA. The complex does not encode a marker unless you added one separately. Delivery class belongs in the enquiry next to the enzyme name. Reagent classes live in the molecular biology catalogue.

dCas9

Dead Cas9 carries amino-acid changes that damage the nuclease sites, so the protein can still be guided to DNA and still bind, but it does not cut both strands the way active Cas9 does. Fusions and tethers turn that binding into repression, activation, or a label. Those experiments are not knockouts. A nickase, which cuts one strand, is a third design and should not be filed under the same heading as dCas9.

The mistake this prevents is interpreting a loss of transcript from CRISPRi as a broken gene, or ordering a dead mutant when the assay needs a break. The evidence difference is set out in comparing knockout and knockdown evidence.

Off-target

An off-target cut is cleavage at a site that is not the site you named, usually a related sequence that still sits next to a motif the enzyme accepts. Prediction tools rank candidates. They do not close the list. Sequencing the intended locus says nothing about the others.

The mistake this prevents is the adjective "specific" used as data. Report the sites you looked at, or say you looked only on target. A phenotype that appears with one guide and not with a second guide is a reason to suspect an off-target or a passenger event, not a reason to ignore it.

Mosaicism

Mosaicism means more than one genotype is present in a body, an embryo, or a tissue that began as a smaller group of cells. If a nuclease is still cutting while cells divide, daughter cells can inherit different repairs. A bulk culture of transfected cells is also a mixture, but that mixture is usually called a pool. Reserve mosaic for the biological situation in which lineages inside one organism differ.

The mistake this prevents is writing a single genotype for an embryo from one biopsy. The concept, and the limit of that biopsy, is in mosaicism in edited embryos as a concept. This glossary does not add a procedure.

Frameshift

A frameshift is an indel whose length is not a multiple of three, placed in a coding sequence so that codons downstream of the lesion no longer match the original frame. A premature stop often follows. Cells can still splice around the lesion, start translation at another AUG, or keep a fragment. A multiple of three can delete amino acids and leave the frame intact, which is a different allele and sometimes a different protein.

The mistake this prevents is equating any indel with loss of protein. Say frameshift when the arithmetic of the reading frame is what you measured. Say knockout only with the extra evidence that word demands.

Choosing the word when you write the result

Start from the object in the tube, then the event in the cell, then the evidence on the gel or in the file. Protein plus RNA is an RNP. A plasmid that encodes them is not an RNP until the cell has made the complex, and even then you usually still call the delivery a plasmid. A break followed by end joining is an indel problem until sequencing names the alleles. A frameshift is an interpretation of a coding indel. A knockout is a claim about the product. An off-target is a claim about somewhere else, and only if you looked.

If two people can read your sentence and order different molecules, the sentence has not chosen a term yet. Spell the enzyme species, the guide format, and the repair you hope for. Link the assembly name from a browser such as Ensembl so the coordinates have a home.

Where loose wording becomes the wrong experiment

Ordering dCas9 for a knockout wastes the well and then invites a false-negative sequencing result, because there was nothing to cut. Ordering an active nuclease for a repression experiment cuts the locus you only meant to bind. Calling a pool mosaic, or a mosaic a clone, sends the analysis down the wrong sampling plan. Calling HDR complete because a donor was added will publish a scar as a knock-in.

Public design tools such as CRISPOR ask for the enzyme and the motif before they score a spacer. That prompt is a useful habit even when you are only writing a legend. The Addgene CRISPR materials use the same family of words. Read them as definitions with limits, then match your own figure to the definition you can defend.

Safety and the words that overclaim

None of these terms is a clinical indication or a biosafety category. An RNP can still be a recombinant experiment under your local rules. A dCas9 experiment can still use a viral vector that changes containment. Institutional review sits outside the vocabulary and outside a reagent quotation. Do not use a precise term as decoration on a therapeutic claim this page does not support.

What to put on the order

On the quote request, write the enzyme class, nuclease versus dead protein, guide format, and whether a donor is in scope. Mention the organism. If a later read of alleles is part of the scientific question, point at the CRISPR validation sequencing enquiry reference as a discussion prompt, not as an approval. The molecular biology pathway is the surrounding research context. Ask whether a quotation is possible, and keep the nouns in this glossary stable between the email and the tube label.