Dear Lena, I suppose one never knows if a morpholino knockdown is a true null. You can do a western as you say. We most often use splice donor targeting morpholinos and with them you can do RTPCR to demonstrate absence of normally spliced, wildtype mRNA. This is pretty efficient. Of course there is a possibility of in frame mis splicing so you have to pick your splice donors and confirm what you have by sequencing. You never know exactly what you are going to get (box of chocolates...).

The field has moved on however to TALENs and the CRISPR system. If you have only a few genes of interest, I would reccomend making genetic mutants. I think a good approach to making a likely null would be to target two sites in a gene with TALENs/CRISPR and generate a large deletion. There are a few papers describing this, one is:

Lim et al. 2013 A simple strategy for heritable chromosomal deletions in zebrafish via the combinatorial action of targeting nucleases

http://genomebiology.com/2013/14/7/R69/abstract

If you can create a null background for a gene orthologous to a human disease gene in zebrafish using either ENU mutants, morpholino knockdown, or TALEN targeted mutagenesis and you can observe a reliably scorable phenotype, then you can inject human mRNA in an attempt to rescue the phenotype to test the function different human disease alleles.

The first requirement is that an injected wildtype human mRNA rescues the zebrafish null phenotype. This is not always the case (for unknown reasons). In some cases, for instance when expressing mRNAs for morphogens or potent transcription factors, mRNA expression ubiquitously throughout the embryo produces an ectopic or overexpression phenotype on its own. So this won't work. Also, there is a limited time window for mRNA expression (1-2 days post fertilization roughly) so its best to focus on early phenotypes.

If you can achieve mRNA rescue and score a phenotype quantitatively, then you can make mRNA of different human gene alleles and test them for function in vivo. It is advisable to epitope tag your mRNAs to meausure protein expression in zebrafish (western). In some cases human mutations are destabilizing and no protein is made. In this case injecting more mRNA to try and see expression is one approach. Insuring that the mRNA can make protein in an in vitro translation assay might be another approach. Ideally you can detect mutant mRNA expression in injected zebrafish embryos but the phenotype persists, indicating a deleterious mutation or, see rescue with a mutant mRNA indicating a benign polymorphism.

A paper where we did this:

Panizzi, J.R., Becker-Heck, A., Castleman, V.H., Al-Mutairi, D.A., Liu, Y., Loges, N.T., Pathak, N., Austin-Tse, C., Sheridan, E., Schmidts, M., et al. (2012). CCDC103 mutations cause primary ciliary dyskinesia by disrupting assembly of ciliary dynein arms. Nat Genet 44, 714-719.

Dear Iain,

Thank you very much for this very clear answer. Obviously, this strategy works provided the KO is not lethal. I will read your article carefully.

Sébastien

A lethal KO would still work; the parent fish can be heterozygous carriers of a mutation and the screen done in homozygous offspring. schmalhans is a lethal mutation at 5 dpf.

It's quite powerful and convenient model indeed. It may be very useful for some of my current projects.

Due to a genome duplication that occurred in most fish (though not Gar), there are many instances where the zebrafish has two paralogues for a given human gene, some cases of which both genes need to be knocked out or knocked down to see a mutant phenotype. As Iain pointed out, you need a mutant phenotype to score first, and can assay by rescue.

In addition, in instances where mutant human mRNA rescue does not rescue, the zebrafish orthologous mRNA can be used to rescue. In the ~80% of instances where missense mutations causing phenotype in humans are conserved in zebrafish, the polymorphism in question can be created in the corresponding position in the zebrafish mRNA. The advantage of this is that one can then test ONE polymorphism at a time. This approach was demonstrated for three coding polymorphisms in two pigmentation genes, SLC24A5 and SLC45A2:

Tsetskhladze, Z.R., Canfield V.A., Ang K.C., Wentzel, S.M., Reid, K.P., Berg, A.S., Johnson, S.L., Kawakami, K., and Cheng, K.C. (2012) Functional assessment of human coding mutations affecting skin pigmentation using zebrafish. PLoS One published 10 Oct 2012, 10.1371/journal.pone.0047398.

Dear Keith, That's a very enlightening observation. I didn't know about the duplication and I will check whether the genes I'm interested in are in two copies.

Actually, if possible, I indeed intend to assess the effect of one polymorphism at a time, and if it works, I'd like to test combinations of polymorphisms that could lead to the expected phenotype.

I'm very happy to learn about such perspectives with the Zebrafish model, which, I thought, were restricted to mouse model.

Thanks for the note.

Sébastien

Dear Iain and Keith,

What if the gene is not expressed in Zebrafish? I have one gene of interest apparently expressed only in mammals and which does not blast Danio rerio.

Would you try this model anyway?

I feel pretty much like creating a monster, but I imagine that it could give clues for understanding the function of this precise gene which is barely known to date. On the other hand, if it's not naturally expressed in Zebrafish, it might have a "chimeric" function far from the one in man.

What is your opinion?

Sébastien

Maybe I too fast that my gene of interest was not present in the zebrafish. In fact, by querying the Ensembl database Danio_rerio PEP_ABINITIO taking into accountg Genscan predictions, I obtained a protein sequence identity of around 22-25%, together with homology between 40-50%. Probably this gene exists but is simply barely homologous between zebrafish and humans, as it diverged a lot from one species to another.

Dear Iain, I have a question on your first comment. Is it actually possible to create a null background just by the use of Morpholinos? Do you verify the efficiency of your Morpholinos by carrying out a Westernblot or how can you be sure that you actually got a null background?

Dear Lena, I suppose one never knows if a morpholino knockdown is a true null. You can do a western as you say. We most often use splice donor targeting morpholinos and with them you can do RTPCR to demonstrate absence of normally spliced, wildtype mRNA. This is pretty efficient. Of course there is a possibility of in frame mis splicing so you have to pick your splice donors and confirm what you have by sequencing. You never know exactly what you are going to get (box of chocolates...).

The field has moved on however to TALENs and the CRISPR system. If you have only a few genes of interest, I would reccomend making genetic mutants. I think a good approach to making a likely null would be to target two sites in a gene with TALENs/CRISPR and generate a large deletion. There are a few papers describing this, one is:

Lim et al. 2013 A simple strategy for heritable chromosomal deletions in zebrafish via the combinatorial action of targeting nucleases

http://genomebiology.com/2013/14/7/R69/abstract