CreERT2 knock-in mice
As one can see in the collection here, several different creERT2 knock-in mice were created over the years as work done for my doctoral and post-doctoral research training. The creERT2 was expressed from the endogenous promoter (Sox2-creERT2), from an EMCV IRES (Id1-IRES-creERT2), or from a T2A ribosome skip peptide (Lrig1-T2A-iCreERT2, Csrp2-T2A-sfGFPiCreERT2, and Id1-T2A-iCreERT2).
The constructs used to create these mouse lines were sequence verified by Sanger sequencing. More recently generated constructs were also sequence verified by NGS sequencing.
Through the projects, I've followed the same techniques that I was taught during my bachelor's degree education. Recently, 165 different constructs that I created were NGS sequenced at Addgene during the deposit QC (Capecchi lab at Addgene, Benezra lab at Addgene). There were a few constructs that had oligonucleotide primer-derived errors, one or two recombination-prone plasmids, and one exceptional case of counter-selection for mutations in the Id1-LSL-TagRFP-T-drePBD* construct. However, other than those, the NGS sequences clearly indicated I had handled the plasmid DNA correctly all along because the plasmids weren't mutated during culture or anything like that.
With the alleles generated from those DNA, we can compare their performances to our and others' mouse lines. For example, the Id1-IRES-creERT2 and Id1-T2A-iCreERT2 lines; the Sox2-creERT2 line; the Lrig1-T2A-iCreERT2 line.
The comparisons were similar. For example, in the case of the Id1-IRES-creERT2 allele and the Id1-T2A-iCreERT2 allele, both alleles "worked" similarly, although the recombination efficiency could have been somewhat different.
So, this means that, in the cases of previous targeting vectors that were not NGS sequenced, the targeting vector homology arm sequences must have been correct, and they must have directed the knock-in into the intended locus.
Next, when the creERT2 recombination efficiencies in adult mice were compared, Sox2 and Csrp2 recombined well, whereas Id1 and Lrig1 did not recombine as well. One might speculate that the knock-in's damaged the loci in some way to perhaps reduce the expression level of the endogenous gene and the reporter gene, but that's probably a very rare occurrence, if it even occurs at all.
Thus, an insight gleaned from all these mice is the locus dependence of the creERT2 recombination efficiency, rather than one type of human error. Perhaps an example of the expression characteristics driving a difference in the recombination efficiency may be derived from the Lrig1-T2A-iCreERT2 mouse. A recent study indicated the LRIG1 expression level is highest in early development during embryogenesis and neonatal days (Jeong et al., 2020). Perhaps consistent with this, using the same driver and reporter combination, i.e., the Lrig1-T2A-iCreERT2/+; Rosa26-Ai14/+ mice, the recombination efficiency from the Lrig1 allele was much higher at the same tamoxifen dose adjusted for mouse weight (i.e., same mg per kg) when induced at postnatal day 0 or postnatal day 21 compared to 12 weeks (Nam and Capecchi, 2023).
Having done all of this work, my impression remains that the expression characteristics of the locus make most of the difference, rather than the specifics of the allele design. That said, I think the knock-in knock-out design (i.e., where the reporter knock-in knocks out the endogenous gene) is no longer a good option in 2021 especially for transcription factors, with the progress made on IRES or 2A sequence. Next, whether an IRES or a T2A is used probably won't make too much of a difference, at least for creERT2 alleles. In contrast to expressing fluorescent proteins, I think the cre protein (expressed from the bacteriophage cre sequence or the codon-optimized iCre sequence) is highly active in mouse cells such that both IRES and T2A constructs would "work." As an aside, the sequence of the specific IRES deserves special attention, as there are variants from different labs due to intentional or unintentional cloning decisions (for example, compare plasmid #1 and plasmid #2, see Bochkov and Palmenberg, 2006 for a definitive sequence). To reiterate, I think marker gene selection is more important than the choice to use an IRES or a T2A. If the selected gene doesn't show the characteristics of a robust genetic marker gene, and thus unsuitable to be used as such, neither the IRES-creERT2 nor the T2A-creERT2 allele will be very good regardless of that choice. Whereas, if the marker gene is robustly expressed, either the IRES or the T2A allele should recombine sufficiently well. That said, because (1) T2A is smaller in size and thus easier to clone than IRES and (2) IRES might depend on additional cellular proteins that may be cell type-dependent, the T2A may be favored over the IRES.
To wrap up, there is also the concern of "leaky" creERT2 if its expression level is too high. As aforementioned, I think the cre protein is actually very efficient, so this "too high" of the expression level might not be very high in absolute terms. In that case, use of two ERT2 fragments, i.e., ERT2-iCre-ERT2, or less efficient recombinases, e.g., FLPo-ERT2 or dre-ERT2, might be appropriate.
Finally, it is also worth noting that how the tamoxifen is formulated seems to make a difference in the efficiency of in vivo inductions. Nam and Capecchi, 2020 describes briefly what I do. Also see "TMX induction tips.pdf" (link - clicking will download a file).
