I’d like to knock a synthetic construct (something without any sort of endogenous locus) into the worm genome. Are there any dpy’s or unc’s that are healthy and have a nice visible phenotype when completely null? Also if they are highly expressed it would be nice to use their promoter…
Yeah, there are a whole bunch of Uncs and Dpys that are strongly phenotypic (ie Unc or Dpy) and viable as nulls. If you just go through the first dozen dpy mutants and your favorite dozen or two unc mutants on WormBase, you can read up on their phenotypes and check the Genetics tab for available, phenotypic molecular nulls (not to request them, just to make sure they exist). I had a list of this sort someplace (for making initial CRISPR plans), but can’t rapidly find it. In any case, you can reconstruct it in fifteen minutes. Be aware that even though a molecular null is viable and phenotypic, that doesn’t mean it’s quite so nice as the canonical allele you might be used to; a true null can have a stronger phenotype, a smaller brood size, etcetera.
RE the promoters, I have no idea offhand - but I’m not entirely clear what your motivation is. There are various established, tested promoter constructs available that would seem likely to be more predictable.
In general, your approach seems odd. Why would you want your knock-in to disrupt an endogenous locus and cause a possibly confounding visible phenotype (and most visible phenotypes would seem likely to get in your way - though if you insist on this approach I’d nominate lon-2, which has an instantly recognizable visible phenotype, as a null, and you’re likely to find its phenotype far less inconvenient than most)? Wouldn’t MosTIC seem a more obvious choice, integrating into one of the standard intergenic sites commonly used for this purpose by a bunch of labs?
I’ll look into lon-2 nulls…as far as using CRIPRs rather than MosTIC, and targeting a selectable marker…the gene I’m inserting isn’t selectable, by disrupting a selectable gene I can select for my insert…this has worked rather well for yeast, which is haploid, but aside from having to wait a generation to homozygous the integrant I don’t see why MosTIC is better…if you have a reason I’d appreciate hearing
I guess I’m kind of with Hillel on this one: I’m a bit reticent to delete genes just to produce a visible phenotype to select for your knock-in. You never know how messing with TGF-beta signaling (lon-2), cuticle morphology (many dpys), or nervous system/muscle function (uncs) could affect planned or future experiments. It could always lead to an unanticipated or unwanted genetic interaction, depending what this gene you are inserting is expected to do. Additionally, your planned approach will likely require a fair amount of screening anyways: if you plan to drop your gene of interest into lon-2, deleting the entire lon-2 gene (using CRISPR/Cas9, I presume), then any NHEJ-mediated indel that causes a loss of function will also produce a Lon animal. And this seems like a more probable outcome than a knock-in. I feel why make a mutant worm unless you really want to do so, particularly with emerging editing methods.
Seems like another approach is to insert the cassette into the “safe harbor” locus where the chromosome II ttTi5605 transposon is inserted for Mos approaches) and use one of the recent co-CRISPR (unc-22) or co-conversion (dpy-10 or pha-1) selections to enrich for your knock-in. These approaches all involve either outcrossing the selectable knock-in or mutation, or restore a WT animal. Junction-specific PCR should let you test whether you got an insert rapidly.
If you do want to do a phenotypic selection, you might consider finding a healthy visible phenotypic mutant with a mutation close to the 3’ end of the gene, target that mutation site with a gRNA (or three), and provide a recombinational repair rescue construct that will restore wild-type sequence at the mutation site (maybe with a few more silent mutations introduced to discourage cleavage targeted by the sgRNAs you used), followed by your transgene (including some transcriptional control elements). There could be some issues of proximity (we don’t know about insulator sequences in elegans, which would come in handy in such a construct) and some difficulties of getting exactly the transcriptional regulation you like, and avoiding recombination within the 3’ UTR that repairs but excludes your transgene is an interesting issue (in a pinch, you could try substituting the briggsae 3’ UTR, or better yet you could repair a deletion that extends from the 3’ end of the gene into an intergenic region). Still, with this approach you’ll be restoring your phenotypic marker to wild-type function instead of causing problems for yourself; also, you should only get rescue when the recombination does occur, and the rescue will be dominant (rather than recessive as in the case of your insertion causing a phenotypic defect).
As to CRISPR versus MosTIC: I’m a CRISPR booster, albeit generally agnostic about which method to use (whichever works easiest!), so the issue isn’t some preference for Mos: it’s an appreciation of the benefit of integrating into an intergenic region, and into one generally accepted and previously used as being innocuous.
Thanks for the advice guys!
I hadn’t seen that coconversion paper from Andy Fire’s lab, it’s exactly what I was looking for…I’ll probably try the experiment both ways…just knocking it straight into dpy-10 and the dpy-10 mutation/coconversion elsewhere to produce a marker free strain.
On the assumption Jordan is too polite to mention it again, you should also look at his recent paper; also the Paix et al paper, and there may well be others that offer useful methodological tips for doing knock-ins and co-conversion.
Thanks again Hillel…I’d seen the paper from Geraldine’s lab but not Jordan’s…the TS pha-1 strategy sounds very nice as well!
Thanks Hillel BTW, in case people try my pha-1(ts) method, I’ve put the method with details and considerations up on my website; I found Dan Dickinson’s CRISPR methods page very helpful and wanted to do something similar: