We have generated several "knockout" cell lines using CRISPR-Cas9. Of the five bi-allelic "knockout" clones we characterized by WB, all five have regained the ability to express the protein (in some cases the protein was a similar/same size as the WT protein and in others it was truncated), which was detectable by an antibody against the WT protein. In the case where truncated proteins are generated, we predict based on sequence data that the cell may be utilizing cryptic or alternative ORFs. Can anyone confirm this with their first-hand observations or publications?

Not necessary to read further, but want to share an exchange I had with a colleague regarding this phenomenon. The exchange may provide some clarity and context for those wanting more (e-mail exchange revised for brevity):

I write:

The observation was made after introducing indels into an early exon of our gene of interest (both alleles were edited, but each with a different indel). We were able to successfully generate the KO cell line (several, in fact), which we confirmed by Sanger sequencing and Western blotting. However, after performing an experiment to test the effect of the knockout, we performed a second Western blot. To our surprise, what appeared to be a truncated protein was observed on this blot, where it was not detected in an earlier passage of the KO line (the WT was still absent).

Turns out (by in silico translation of the mutant sequence) that the indel introduced into one of the alleles results in the availability of new AUGs, which are predicted to result in the encoding of the remaining WT protein (i.e., protein is truncated on the N-terminus end as opposed to the C-terminus as would be expected with an early frameshift). These in silico data agree with the WB results.

We made this observation in at least two of our KO cell lines, with a different set of indels, but in the same location.

My preliminary interpretation of the results is that by the introduction of an indel, the cell has attempted to rescue the loss of the expression of this gene by utilizing a new AUG. While an early passage of the cells did not result in the utilization of this alternative AUG, they acquired this ability later under an unknown selective pressure.

Colleague responds with:

Thank you for your email. I think your interpretation sounds correct. We have a similar activation of a cryptic AUG that results from a Cre/lox deletion of an early exon. I have heard of others but will not be able to remember references. The mechanisms determining what initiation codons are active are not well understood. There are quite a few people working on this kind of thing in relation to upstream ORFs and non-canonical initiation codons such as CUG. You might look for papers by [name redacted]. One thing that might be worth exploring is whether your gene produces this truncated form even without the mutation as an alternative isoform.

My response:

Had a quick look at the NCBI RefSeq database…in addition to the the NM transcripts, there are a few XM transcripts that appear to coincide with our experimental observation. While the XM designation refers to a computational prediction, it would appear that we have activated this downstream AUG in our cells by knocking out the ability of the gene to produce the preferred isoform. However, this appears to be dependent on the sequence/length of the indel that is introduced as not all of our KO lines appear to result in the activation of this downstream AUG. Interestingly, in a different KO line we observe that a different downstream AUG is preferred than the one I described to you below and we detect a larger truncated protein in this line consistent with the location of this AUG.

[End of correspondence]