Indels frameshifts and protein loss

A one-base insertion, a shifted reading frame, and an empty lane on a blot are three different observations. Research write-ups collapse them when a guide "worked". This page keeps them apart for anyone about to claim that a protein is gone. How the cut arises is in how CRISPR-Cas9 editing works in research. How the DNA is read without fooling yourself is in checking whether a genome edit worked.

Antibodies, gels and PCR reagents are classes in the molecular biology catalogue. The claim you need to support should be explicit on the quote request.

What an indel is, before anyone mentions frame

An indel is an insertion or a deletion in the DNA. At a blunt SpCas9 cut, a few bases upstream of the PAM, the common products of end joining are small. One-base insertions are frequent at many sites. Short deletions are frequent at others. Larger deletions happen and are easy to miss if your PCR primers sit too close and the deleted allele simply fails to amplify. The indel is a sequence fact. It does not, by itself, name a protein.

Record the change against the parental sequence, not only against a reference download. A two-base difference that was already in the line is not your edit. Ensembl and GenBank identify the reference. The parent identifies the background. If you did not sequence the parent, you do not know which.

Count the length in coding sequence if the protein is the point. Add the insertion length, subtract the deletion length, and ask whether the net change is divisible by three. That arithmetic is the frameshift test. It applies only when the lesion sits in the coding frame you think is used. An indel in an intron, in a UTR, or in an exon the isoform skips, can pass or fail that test and still not do what the cartoon shows.

Frameshift, premature stop, and nonsense-mediated decay

If the net length change is not a multiple of three, the ribosome reads a different frame downstream until it meets a stop. Those stops are often premature relative to the normal stop. The mRNA may then be unstable. In mammalian cells, a widely used rule of thumb is that a premature stop more than about 50 to 55 nucleotides upstream of the final exon-exon junction is a substrate for nonsense-mediated decay. A stop in the last exon, or within that short distance of the last junction, frequently escapes. Decay is not uniform across genes, and it is not complete. Residual mRNA can remain.

Nonsense-mediated decay is an RNA observation. You can measure it by comparing transcript abundance, with the controls set out in from cells to a gene expression result and the assay limits in RT-qPCR for relative expression. A lower mRNA makes protein loss more plausible. It does not demonstrate an empty proteome. Truncated polypeptides can still be translated from the fraction of mRNA that escaped, or from a start codon downstream of the lesion.

If the net change is a multiple of three, do not draw a frameshift. You have removed or added amino acids. The protein may fold, may be degraded, or may lose one activity and keep another. UniProt is a place to see whether those residues sit in a described domain, before you predict the whole protein is irrelevant. Domain maps are clues, not functional data.

Exon skipping, restart, and extra copies

Cells have several ways to keep a polypeptide after a coding lesion. A guide that damages a splice site, or an exon that contains an exonic splicing enhancer, can cause the exon to be skipped. If the skipped exon has a length divisible by three, the downstream frame survives and the protein lacks that segment. The DNA trace from a genomic PCR still shows the indel. The mRNA does not use it. If you only sequenced genomic DNA, you have not seen this.

A downstream AUG can reinitiate translation and produce an N-terminally truncated protein. An antibody raised against the missing N-terminus will report a knockout. An antibody against a downstream domain will report a smaller band. Both can be honest. Knowing the epitope is part of choosing and checking a primary antibody. Loading and transfer controls for the blot itself are in western blot from gel to membrane.

Aneuploidy and copy-number variation do the simplest version of the same trick. Many cultured lines are not diploid. Frameshifts on two copies leave a third copy making protein. Cancer lines are famous for this, and some widely used embryonic kidney derivatives are not tidy diploids either. Without a copy-number statement, a pair of sequenced alleles is an incomplete inventory. Say what you counted.

Alternative promoters and alternative last exons mean one guide can ruin one transcript and spare another. Name the transcript you sequenced and the transcript your assay detects. They are sometimes not the same isoform.

What a protein method may add

A western blot with a working parental lysate, a loading control, and an antibody whose epitope you can point to, can show that the epitope is strongly reduced in the clone. That supports loss of the region the antibody sees, under those extraction conditions. It does not prove that no fragment exists, and it does not prove the function is gone if the function lived in a different part of the protein. A second antibody, or a mass-spectrometric search for peptides from the protein, is the orthogonal step when the claim is absolute. The culture of that kind of protein evidence is what HUPO exists to discuss. It is not a kit you run beside the blot by default.

A phenotype is a third layer. Loss of a band plus loss of a known activity is stronger than either alone. Restoration of the activity by a cDNA that lacks the guide site is stronger again, because it ties the phenotype to that protein rather than to a passenger mutation. If the cDNA does not rescue, the clone may carry more than the indel you like.

When the DNA looks edited and the band remains

Run through the escapes in order. Is the length change a multiple of three. Is there an unedited copy left. Does the mRNA skip the exon. Does the antibody see a fragment from a downstream start. Is the band a cross-reacting protein that was always going to survive, which you would know if the parent and a second antibody had been on the membrane. Is the clone mixed with unedited cells that supply the protein. A pool is especially prone to the last of these. The decision to leave the pool is in editing pools versus single-cell clones.

If the band is gone and the activity remains, the activity was not this protein, or a redundant paralogue covers it. That is a biological result. Editing a paralogue by accident because the guide matched both is an off-target problem of a special kind: the second site is a gene you might have called a control. Check paralogues when you choose the spacer, not after the phenotype refuses to match the blot.

Do not rescue a weak DNA claim with a strong adjective on the blot. A smeared lane and an overexposed parent is not loss. Repeat the extraction.

Research claims, not a clinical genotype

Nothing in a frameshift diagram diagnoses a person or approves a therapy. Edited research lines remain under institutional biosafety review. This page does not describe embryo or germline editing. A truncated protein can have a new activity, including a dominant fragment. If that possibility matters for safety or for interpretation, look for the fragment instead of assuming decay removed every peptide. Handling acrylamide or other gel hazards follows the safety data for those chemicals and the practice in your protein laboratory. The biology of nonsense-mediated decay is not a handling rule.

Lysates in a humid room

Protein work fails in ordinary ways that then get blamed on exon skipping. A lysate that warmed during a power cut, an antibody aliquot that collected condensation in a humid freezer, and a transfer that ran too hot will all empty a lane. The parental lysate on the same gel is the control that saves you. Store lysates and antibodies as their own notes require, in the spirit of storing biological samples from fridge to freezer. Label the clone and the passage on the tube. A box of "knockout lysates" without genotypes is how an in-frame clone becomes a figure.

What to put on the enquiry

State whether you need reagents to genotype an indel, to measure a transcript, or to detect a protein, because those are different shopping lists. Name the epitope or the domain if you are asking about an antibody class, and name the copy-number problem if the line has one. Sequencing of the locus can be discussed via the CRISPR validation sequencing enquiry reference as an enquiry reference only. It does not mean a validation service is operated. Protein detection sits against the molecular biology pathway. Ask whether a quotation is possible. Do not write knockout as the specification for an antibody that has not been shown to see the fragment you might still produce.