crystal fish imaging

Note the original request for this protocol asked how to generate the F0 crystal mutants. Some notes:

– The crystal mutant is a triple knockout of mitfa, mpv17, slc45a2. We targeted each gene at three loci, therefore injecting a mix of nine CRISPR-Cas9 RNPs in the egg (yolk) at the single-cell stage.

– The sequences of the crRNAs to order from IDT can be found in Supplementary File 1.

– Please follow our protocols.io for instructions on how to prepare the CRISPR-Cas9 RNPs. The protocol assumes you are preparing three RNPs, but the procedure is exactly the same for the nine RNPs used to generate the F0 crystal mutants. At step 5 "Pool RNPs", you will simply pool the nine RNPs together (~ 2 µL each, so ~ 18 µL total).

– How much should you inject? There are two logics you could follow here. When mutating one gene, we inject ~ 1 nL of the injection mix produced at the end of the protocol, which corresponds to 5029 pg Cas9 and 1070 pg gRNA (for three targets, so 357 pg gRNA per target). You could either assume this is the maximum that can be safely injected with an acceptable rate of unviability (logic #1), or assume that 357 pg gRNA per target is the minimum necessary to obtain high mutagenesis at every target (logic #2). Following logic #1, you would inject 1 nL of the nine-RNP mix, this would keep the amount you inject to 5029 pg Cas9 and 1070 pg gRNA, but it would mean you only inject 1070 pg / 9 = 119 pg per target. Following logic #2, you would inject 3 nL to keep 357 pg gRNA per target when injecting nine RNPs. On the one hand, we have some evidence that 357 pg gRNA is in excess of what is necessary for high mutagenesis (supports logic #1), but the main cause of unviability is likely to be the double-strand breaks, so injecting more or less while targeting the same number of loci may not have a large effect on unviability (supports logic #2 if there is no cost to injecting a large volume). We would recommend an in-between solution, say 1.5–2 nL.

– Note we obtained high unviability when generating the F0 crystal mutants, likely because of the high number of double-strand breaks generated early in development. We do not think the exact same protocol could be readily applied to routinely generating crystal fish for experiments, but it would likely be straightforward to improve it by only injecting the gRNAs which we found are highly effective (high frameshift mutation rate) in the goal of reducing the number of loci that are targeted simultaneously, for example from nine (three per gene) to six (two per gene). You can find some more notes about this in Author Response 11).

– Finally, because of the unviability and the cost of reagents (nine crRNAs are ~ 450£, 9 µL Cas9 are ~ 90£), we would not recommend generating F0 crystal mutants as a first pilot experiment to learn the method and test injections. For this, targeting slc24a5 is a much better choice. You can use gRNAs slc24a5-AA + slc24a5-AB + slc24a5-AD (sequences). Do not use slc24a5-AC, which does not work for mysterious reasons (we used it as a negative control in some of the headloop PCR experiments).