When you think about it, fluorescent proteins (FPs) are pretty forgiving as tags. They tend not to strongly disrupt the protein they are stuck onto, though there are of course exceptions.
I bring this up because we CRISPR tagged one of our favorite genes, rheb-1, at the 5’ end with FP::AID::short linker (Gly-Ala-Ser). We know from a lot of experience that the rheb-1 null phenotype is an arrested L3, and we successfully tagged rheb-1 with mKate2::3xFlag (via the Dickinson SEC approach) without altering function.
To our surprise, the AID tag killed rheb-1 function, resulting in arrested L3s. Multiple alleles, checked by sequence and no mutations found. But then when I thought about it, I realized we had stuck a relatively unknown quantity, AID, in close proximity to a highly structured N-terminus of RHEb-1. RHEB-1 is a Ras family small GTPase. These proteins are little but a GTpase domain and a C-terminal membrane-targeting domain. So I suspect that we compromised RHEB-1 structure or function with the AID tag. (We did observe fluorescence, in this case mNG, in teh tagged animals. So stable protein is made.) We will circle back and try to put AID on the opposite side of the FP, giving AID::FP::RHEB-1. We have similarly tagged a close friend of RHEB-1, DAF-15, making DAF-15::FP::AID. Deletion of DAF-15 similarly causes L3 arrest, but this strain is just fine. Then again, the C-term of the mammalian ortholog of DAF-15, Raptor, is not known to be very structured.
Anyway, I opened this thread because AID is increasingly going to be used, people should be wary of this potential caveat, and it is worth having a public discussion so we can prevent many people from running into the same mistake.
Thanks for sharing, Dave. Really important to have these discussions. One idea might be to Micropublish the result. It’d be useful to create a citable list of CRISPR knock-in oddities, which could eventually be synthesized into some sort of perspective (I think Micropubs calls these Integrations). Getting more information on when a tag disables a protein, where it is relative to protein features (N/C termini, domains, etc.), linker lengths.
Looking at the AID tag, it looks like there are a few bulky aromatics and a lysine near the C-term of the degron tag. I wonder if a longer linker is helpful. Is there a FLAG or Myc tag also included? Would be curious if the defective RHEB-1 is unstable and/or non-functional. I’ve seen some weird results where a linker-GFP-BioTag-degron-3xFLAG caused a partial loss-of-function that was exacerbated at high temps, whereas another knock-in in the same site with a slightly different linker and no BioTag was totally fine.
Alas, our construct had the FP at the other end of the protein. Currently we are using strategies that have the AID on the other side of the FP from the protein of interest. This has worked in at least one case (DAF-15), and others are in the pipeline.
UPDATE: We will think about re-doing this with a longer linker. Right now, we need to submit the paper on this, then we can return to this question. I think the ideas of micropubs, and a later Integration, is a great idea to move us forward.
We have tried to tag around 10 different genes with an AID tag, of which approximately half worked and the other half didn’t. All our genes are essential, so any tags affecting gene function will not be viable.
In our first tries we used a 5 aa linker and this worked for some but not all tags. For the tags that did not work, we tried a linker of 15 aa, but this did not give us any insert either.
Looking further into this. it seems that whenever our tags worked, there was a predicted ‘ordered’ part of protein located at least 50-100 aa from the site of insertion, while the tags that did not work contain a relatively long stretch of ‘disordered’ protein domain. We are now trying to put the tag at the other side of the proteins, where the protein structure is predicted to be more ‘ordered’, hopefully that works. Alternatively, we are thinking of trying a rigid instead of a flexible linker (which apparently helps in some cases to keep two domains away from each other so they will not be able to disturb each others function) or a combination of a rigid and flexible linker.
In addition, one of our tags is not being degraded efficiently and we heard that tagging N-terminally might be beneficial in this case because that is where ubiquitination would normally take place.
Anyway, not a lot of conclusive information, but I thought I’d share our process and hopefully I will be able to share more of what we learned soon.
But these anecdotal observations are important until somebody does a systematic study. Certainly yours agree with ours. For now, I recommend that people put AID at the distal end of the FP from your protein of interest.
lrijnberk-One idea might be to Micropublish the unsuccessful AID alleles. It’s a nice way to get otherwise anecdotal observations into the community and into the record. One might be able to synthesize some potential rules from these failed strains and then empirically test them.
Our experience with a few essential proteins are same. 2 failed (kind of sick worms) and 1 success (perfectly fine with great protein degratation). We have flexible linkers and tagged at the C terminus. We are trying the N terminus.
Hi Dave
On the topic of systematic studies… This recent paper in G&D (http://genesdev.cshlp.org/content/33/19-20/1441.full) suggests that the auxin-inducible degron generally results in only ~3-15% of native protein expression in the absence of auxin, at least in mammalian cell culture. So the problems may be unrelated to linkers etc. but may reflect dosage sensitivity? The same paper solves the problem by expressing a second protein (auxin response transcription factor (ARF)), which is how the system is natively regulated in plants. Might be worth a try?
Cheers
Christian
Great stuff Chris! We should try to directly compare some of our tags (with and without AID) by blotting and detecting the epitope tag. Meanwhile, we should look into this ARF improvement!
The ARF16 trick looks interesting, was thinking of trying it. However, I’m not super optimistic it will work for the bulk of exisiting AID-tagged strains. In that paper they use a larger AID tag that has additional domains. Buried in the methods: "Note that this plasmid contains the full-length AID, and the mini-AID tag is unlikely to function in the ARF-AID system due to mini-AID’s lack of domains III and IV. " So it could be worth testing out, but would be too bad that it likely wouldn’t fix auxin-independent degradation issues with existing strains.
Now we have AID tagging experiments for 5 genes, all of which are essential. All C terminal tags. 1 worked great. 2 gave hypomorphic phenotypes. The remaining 2 did not homozygose. These were all generated in TIR background, so we did not check if the problem was TIR expression, tough I am guessing it is not, because the strain that is successful is topo II, which is very dosage sensitive. I also dont think that C terminal tagging is the problem, because we tested tagging a few of these with GFP only and it worked fine. My conclusion from this limited experience is that the tag is doing something to some proteins. Yeah, I know. Such a useful conclusion
we also noted this in our recent G3 paper: https://doi.org/10.1534/g3.119.400781 that auxin-independent, TIR1 dependent degradation is definitely occurring in C. elegans like in other organisms (see Figure S4 from that paper where both ubiquitous GFP::AID (under eft-3) and endogenous nhr-25::GFP::AID both show depletion with TIR1 in the background w/o auxin). We’re hoping to have a fix or improvement for this soon!
Hello everyone! I have read all of your answers, and I would like to know your answers for this:
I am designing my CRISPR insertion of the degron sequence in a specific gene in order to do auxin-induced degradation in C. elegans.
We will insert Degron::emGFP sequence in our gene of interest by CRISPR. Should we add a linker? any particular size? I know that you guys have tested, and just wanted to know if you could give me maybe top 2 options to try…
How to decide where to place degron, whether 5’ or 3’ of the gene? Depending on the domains of the protein of targeted? Or do you suggest putting the degron::emGFP sequence 5’ of our gene of interest?
Any information is useful! Thank you!
Victoria Cerdeira
We try to put AID on the far side of the FP::epitope, just to minimize interference.
As for 5’ vs. 3’, that depends entirely on the protein. Check all the functional domains. For example, with small GTPases, which are nearly always modified by prenyl lipid at the C-term, the 5’ end is mandatory because 3’ would kill function. For one protein known to be heavily regulated, we actually consulted with a structural biologist who threaded the worm sequence onto a mammalian structure of a three protein complex. Good thing, too. An N-term mod would probably have been between the protein and the plasma membrane, and disrupted activity. But that’s the only one where we consulted. And in some cases we just make the best guess based on being distant from any structures or domains. And I know of at least one case where either end killed function and the group stuck the FP in mid-protein, between domains. But worth some homework to make a best guess.
For linkers we go at least 9 codons between the FP and yfg-1 (same variant of Gly, Ala, Ser). Be careful that PCR primers targeting a linker can sometimes cross-prime with other linker seqs in plasmids. Gly is GGN and GCN codons, so VERY GC-rich.
