You are here

Validating Morpholino phenotypes with CRISPRs

I've been enjoying a conversation with Martin Blum about Morpholinos, CRISPR mutants, funding and publication. He wrote:

Good morning Jon,

My study section at the German Research Council DFG now accepts MO-proposals, no problem. Still, many people particularly working in zebrafish are too careless with controls and rescues. That might backfire at one point in the future.

My take is that we will use CRISPR to validate MOs and then use MOs most of the time.

Best wishes,

My response

Hi Martin,

You want a CRISPR mutant to validate the Morphlino, but we know compensation in the mutant can conceal the effect of losing the protein, which can be revealed by a knockdown. What do you do if the Morpholino shows a phenotype and the CRISPR doesn't? Is this validation useful? More below...

I heartily agree that specificity controls need to be tight. Five-mispair oligos are worthless. Rescues are very good if they work, but ectopic expression can mess them up. The two-nonoverlapping-oligo phenocopy is good but there have been cases where both oligos have off-target interactions, so that can take more oligos to nail down. Checking for dose synergy with two non-overlapping oligos is stronger when combined with the one-oligo-at-a-time phenocopy previously mentioned. But my favorite specificity control today is:

You have a Morpholino and it produces a phenotype when injected into a wild-type creature. You also have the CRISPR mutant for the same target gene, it has no phenotype. You inject the Morpholino into the mutant, you see no phenotype. That tells us that compensation in the mutant is concealing the phenotype that the Morpholino reveals, and further tells us that the Morpholino phenotype is due to interaction with the targeted transcript and not an unexpected RNA. If that is the validation of a Morpholino with a CRISPR mutant that you are proposing, then I am enthusiastically on-board.

Alex reviewed this message prior to sending and pointed out that if the Morpholino has changed physical state, e.g. through aggregation in the solution state, then relying on a negative result (no Morpholino phenotype in the mutant) could be dangerous. To avoid that, the oligo should be injected to wild-type embryos and mutant embryos in the same session, so if there is an issue with the oligo's physical state then you won't see the Morpholino phenotype in the wild-type embryo.

There is a confounding problem that could show up for early phenotypes. If a heterozygous mother is bred to produce a homozygous mutant embryo, there could be some wild-type maternal transcript that a translation-blocking Morpholino could shut down. This might lead to a more extreme early phenotype in the Morpholino-injected embryos than in the homozygous mutants from heterozygous mothers, which have some wild-type maternal transcript. In this case, the Morpholino could be specific but the phenotype would persist in the mutant background.

I'll post this discussion in my blog; if you like I would be happy to credit you with stimulating this response (I won't do so unless you tell me to -- I could simply add to the blog your message in order to set up the response).

Best wishes to you too,

- Jon


Hi Jon,

Thank you so much for this reasoning on controls, and please: post it on you blog. I am aware of all you are saying, and we do struggle with specificity controls in some cases quite a bit. So far we have always managed to prove it and to convince referees of manuscripts.

With best wishes,


Note (Jon): I first became aware of a Morpholino-in-a-mutant strategy from Didier Stainier's group's paper on genetic compensation in morphants:
Rossi A, Kontarakis Z, Gerri C, Nolte H, Hölper S, Krüger M, Stainier DYR. Genetic compensation induced by deleterious mutations but not gene knockdowns. Nature. 2015;[Epub ahead of print] doi:10.1038/nature14580

Add new comment