I have been trying to use CRISPR to generate a targeted mutation at a specific gene, and since I don’t know if the resulting mutation will necessarily have a phenotype (though I expect it will), I have been attempting to use Co-CRISPR to help make the screening process easier.
I have been following the methodology in Arribere et al. 2014: http://www.genetics.org/content/198/3/837.long
The basic idea is that by inducing a specific mutation (cn64) in dpy-10, I should be able to isolate heterozygous cn64/+ dpy-10 mutants for screening. Since the CRISPR mediated conversion worked for these homozygotes, presumably the odds of getting a mutation at my second locus should be pretty high, and I should be able to screen the offspring of the cn64/+ heterozygotes by PCR for my mutation of interest.
My understanding is that F1 cn64/+ mutants should be non dpy, but should have a left roller phenotype, whereas null dpy heterozygotes (o/+) worms should look totally wild-type.
I have been seeing very strange inheritance patterns in the F1s I’ve picked. First of all, the vast majority of abnormal looking F1 worms I’ve seen look dpy or partially dpy. Some F1 worms look partially dpy with a slight roller phenotype. Many of these slight roller worms have seemingly 100% wild-type looking offspring. When I singled some of these wild-type offspring, some of them led to broods with undeniably dumpy worms, suggesting that the original F1 progenitor was a o/+.
Yet, if the F1 parent I picked was o/+ for dpy-10, shouldn’t it have looked completely wild-type? I understand that a dpy-10 o/+ should look wild-type, and indeed many of my F2 heterozygotes do look wild-type.
Is dpy-10 o/+ incompletely penetrant? Should I expect to see some dpy heterozygous mutants in the F1?
Yeah, I sometimes get some of this dpy-10 weirdness too. I usually get a lot of Dpys and DpyRollers, with few pure rollers. Ideally you need clean rollers (cn64/+) for the technique to work. Are you getting any obvious rollers (even dpy rollers)? One variable could be your oligo. From some people who had problems with pha-1 co-conversion, the issue was with the pha-1(ts) repair oligo.
When you single the slight dpy rollers, do you get any rollers at all? One explanation could be that one allele got knocked out in the germline and the cn64 went into the other locus in the soma later (leading to some mosaicism). That could explain why the roller was weakly penetrant and non-heritable, how there could be F1 dpys, and why you saw dpys in the F2.
I’d be interested in hearing if others have seen similar results, ie whether the cn64 knock-in varies in efficiency from lab to lab.
One other thing to consider (once you start getting rollers), is to tweak your edit mixture similar to what Alex Paix and Geraldine Seydoux described in their recent Genetics paper on RNP editing. They got improved knock-in efficiency (even with plasmids) by reducing the dpy-10 sgRNA and repair oligo. I think that this could reduce the amount of dpy animals, but also bias the repair to the targeted locus, since you’ve reduced the efficiency of editing your selected locus (ie. if you’re seeing rollers, you’ve picked an animals that had very active Cas9-sgRNA complexes).
I, too, have seen cruddy Dpy-looking F1 worms in my CRISPR injections. I suspect that many of these are not dpy-10 edits, but instead are sickly worms that didn’t like the array. The reason that I suspect this is because I’ve done CRISPR injections that did not involve co-conversion (i.e., the injection mix was pRF4-based, and did not contain any dpy-10 sgRNA construct or repair oligo), and I saw a ton of the cruddy Dpys in the F1s. I would be interested to know if others have seen the “nonspecific” F1 Dpys in their non-coconversion CRISPR injections.
With regards to the dpy-10 co-conversion, my experience is that its efficiency varies from injection mix to injection mix, presumably because the other DNAs in the mix are having some effect. For a typical injected animal, I might get a handful of nice Rols. However, for a fraction (5%?) of these injected animals, I’ll get a bolus of Rols, >20.
In response to Jordan, I definitely am getting some bonafide rollers in my broods. Now that I’ve looked at these worms very closely for many hours, I can tell that cn64/+ heterozygotes have a really strong roller phenotype and stick their heads in the air. These bonafide rollers are definitely a minority, but they seem to be a good indicator to screen harder on the plates in which they appear for bonafide rollers.
In response to Sven, I don’t think these are merely sickly worms. After all, I did see F2 dumpy worms from the cruddy P0 dumpys suggesting that many if not all cruddy dumpys are o/+.
I am using the following ssDNA sequence ordered from IDT at a final injection concentration of 20 ng/ul as the dpy-10 repair template:
Sven is describing the “jackpot” of edited animals, observed by many labs. It would also be worth injecting more animals to try and get a few jackpot broods. It’s good that you’re getting rollers, but it seems like you’re missing the “insanely obvious packed with rollers” well. I seem to remember that 1/10 or so are supposed to be jackpots with plasmid-based editing approaches?
For the oligo, is is a 4nM Ultramer, and is it PAGE purified? If not purified, and you continue to have issues, it may be worth PAGE purifying. I typically avoid this cost, and try to inject/screen more, but if there seem to be issues a pure oligo could boost efficiency.
Yeah, your injection amount are definitely in the Paix ballpark, though it might be worth jacking up the “MYSITE” sgRNA amounts and repair template amounts. Depending on size of your desired insertion, you could go up to the max of 500 ng/µl of PCR-derived template described by Paix. I usually start with at least 50 ng/µl.
Yeah I have been getting some plates with a large amount of cruddy dumpies and some rollers. I’ll try boosting plasmid concentrations if my bonafide rollers turn out not to have the genotype I am looking for.
I am using the 4nM Ultramer without PAGE purification.
I also get many Dpys and partial-Dpys in F1 worms, and I suspect that conditional somatic knockouts of dpy-10 occur in many of these (like dpy-5 in Shen et al., 2014, Dev Cell, Fig. 1C: http://www.ncbi.nlm.nih.gov/pubmed/25155554).
Yeah, a 4nM IDT Ultramer should work ok, that’s what I’ve been using for both dpy and pha co-conversion. How do you prep your plasmids? I try to use either Qiagen midipreps or Invitrogen Purelink. Several people have reported lower efficiency with Qiagen miniprepped DNA. I agree that somatic edits happen, but hopefully the lower dpy-10 sgRNA concentration could offset this possibility.