In this study, we used electroporation to deliver two CRISPR RNPs and two ssODNs into single cell embryos for creating floxed alleles with a success rate of 85% (59/69) from a single round of four mouse sessions and 100% (63/63), if supplemented with a second round targeting (Additional file 2: Fig. S1). Floxed alleles require the insertion of two loxP sites into the same copy of the target gene after two double strand breaks are introduced. Various editing events can happen and compete with the formation of a floxed allele. Mainly, indels mediated by nonhomologous end joining can occur at each target site, and deletions can form between the two cut sites. An inverse relationship has been reported between deletion frequency at two targets and their distance from each other, where gRNA targets distanced < 10 kb from each other can result in up to 20–30% deletion products spanning both loci [24]. An allele can also have an indel at one site and loxP inserted in the other. The different editing outcomes are listed (see Additional file 2: Fig. S2). So floxing is a relatively low efficiency editing event and can be difficult to obtain under nonoptimal conditions. Here, we report a reliable floxing strategy routinely used by our center with success: electroporating embryos with validated reagents, assessing loxP phasing using an in vitro Cre assay, and, when needed, retargeting a male founder with a single loxP site via IVF.
Prior to embryo electroporation, validating synthetic gRNAs and ssODNs by transfecting Neuro-2a cells and analyzing the target sites using NGS is an important step to avoid unnecessary waste of time and money by faulty reagents or human errors. Even though the majority of reagents pass validation, the simple protocol is a worthy effort, given it takes at least 6 weeks to find out about a failed mouse session.
Among the 69 targets, we obtained various numbers of animals with both loxP sites, up to 12% of live births, from single round targeting. We observed about 50% of founders with both loxP sites carried a floxed allele, so when there is only one or two potential founders, floxing may not be successful. Confirmation by germline transmission takes 8–10 weeks from the time founders are genotyped. Instead, we used the in vitro Cre assay, where recombinant Cre is incubated with low concentrations of genomic DNA to encourage intramolecular reaction, so that Cre-mediated excision only occurs between the loxP sites in cis, forming a deletion product and a circular product. The circular product is detected by an excision-specific PCR with high sensitivity, even in founders with relatively low percentages of loxP-containing reads (Fig. 1). All animals positive in the in vitro Cre assay transmitted a floxed allele, when bred. On one hand, only founders positive for the Cre assay need to be bred to the F1 generation. More importantly, if no animal has a floxed allele, the second round targeting can be planned right away, saving up to 10 weeks of time. Whereas the in vitro Cre assay is sensitive enough to detect even a low percentage floxed allele in a founder, about 5% of targets turned out to be difficult to detect, including a very large floxed region (> 400 kb, greater than the average size of gDNA fragments, a target not reported here) and targets with multiple high homology sequences elsewhere in the genome. A negative Cre assay on all founders with both loxP sites is not conclusive, and multiple founders should be bred for confirmation if available, given about 50% of animals with both loxP sites carry a floxed allele.
Among the 69 targets we tried to flox, 10 did not result in a floxed allele with the first four electroporation sessions. Recognizing that many alleles with a single loxP insertion have an indel in cis, we used a third gRNA to target the indel specifically in embryos obtained by IVF with sperm from a male founder and wild type oocytes. Half of the fertilized eggs resulting from IVF have one wild type allele and an allele with a loxP at one target site and an indel at the other. Specific insertion of loxP into only the indel ensures that F1 animals with both loxP sites carry a floxed allele (Fig. 2). A higher percentage of floxed F1 animals were obtained when retargeting was against an indel in phase with a loxP at the second site (22/124, 3/32) than against the wild type sequence (1/40). The latter resembles a sequential targeting strategy, where one set of RNP/ssODN is electroporated to obtain male founders with one loxP insertion and sperm will be used in IVF for single cell embryos, which will then be electroporated with the second set of RNP/ssODN. The sequential strategy circumvents the competition from deletion alleles but is obligated to have a second round of targeting. Essentially, two-round targeting breaks down the relatively low efficiency floxing event into two relatively high efficiency events: insertion of a single loxP mediated by a CRISPR RNP and an ssODN. To date, all four attempted retargetings were successful.
Even though all four projects reported here had clearcut retargetable alleles in males, there are many more male founders with one loxP site and one or more indels at the second site at frequencies < 90% for each indel. In these cases, it can be difficult to decide which male founder to retarget. Additionally, larger indels at gRNA cleavage sites have been reported [25], which may not be detectable by NGS and can mislead the interpretation of genotyping. One solution is to breed the founder to the F1 generation and retarget F1 males with a loxP site and an indel. However, an additional 3 months would be added to the timeline. Alternatively, sperm can be frozen from all males with a loxP site and one or more indels at the second site. A straw of sperm from each male could be used for in vitro fertilization of wild type oocytes to obtain a small number of blastocysts to genotype and determine phasing of the loxP site and the indel. This way, retargeting will only be done with sperm of confirmed F0s, that have the desired genotype in phase without adding to the timeline. This genotype confirmation step also ensures sperm samples were collected from the correct animals.
A usable PAM site is needed for retargeting at an indel. Most of CRISPR-mediated indels are small. The original PAM site was maintained in all retargeted projects reported here, primarily because we observe small indels most commonly by CRISPR editing. However, it is possible that a deletion removes the PAM site, and there is no convenient one nearby. Testing individual blastocysts from multiple founder sperm samples via IVF increases the chance of identifying a retargetable animal with a PAM site. Yet, it would be prudent at the design stage to pick out sites with nearby PAMs when possible in the event a larger indel compromises retargeting potential. The original ssODN can usually be used for retargeting, unless the indel significantly reduced one or both of the homology arms in the ssODN.
For decades microinjection has been used to create transgenenic animals, and then for nuclease-mediated embryo manipulation [26]. It remains the go-to method for delivery of large molecules, such as DNA plasmid, long single-stranded DNA, and mRNAs. However, it is usually the bottleneck for throughput, taking microinjectionist hours under the scope to inject a few hundred embryos. For protein molecules, small DNA or RNA molecules, electroporation of single cell embryos has been very effective using various apparatuses [27,28,29,30]. Electroporation is much less labor intensive and time consuming than microinjection, and the conditions are more reproducible from operator to operator. The limit of the number of embryos to be electroporated in a given day is determined by the availability of embryos and recipients rather than available time under the scope and skilled microinjectionists. Higher embryo survival and thus birth rates are consistently observed after electroporation compared to microinjection owed in large part to less physical damage to the embryo [28, 29, 31]. If combined with using HyperOVA [32] and IVF to produce fertilized eggs, it is possible to electroporate large numbers of eggs with RNPs plus ssODNs in one day to obtain over 100 live births. Additionally, in our hands and those of others, electroporation of RNPs consistently results in higher editing efficiency [28, 33]. The electroporation protocol transfer to a second mouse core on campus was straightforward and produced similar success in floxing using validated reagents, NGS genotyping and in vitro Cre assay for seven targets (not shown). Five of the targets reached germline transmission with one round, one with Cre assay-positive founders and one being retargeted.
The timeline to identify F1s with a floxed allele is around five months with either one-round or two-round targeting (Fig. 3). Starting from electroporation, it takes about 8 weeks to identify floxed founders by using the in vitro Cre assay for one-step floxing and 20 weeks to reach F1 generation with confirmed germline transmission or to validate multiple floxed animals resulted from a second-round of targeting.
A common alternative method is to use long single-stranded DNA (lssDNA) donors [17], which have been reported to be efficient to flox a relatively small region. lssDNAs can be difficult to synthesize, require a double-stranded template, and are limited by size. There are times floxing more than a few kb is necessary, such as genes with multiple splicing isoforms needing more than one exon to be floxed. When using two gRNAs and lssDNA to flox a gene, deletion and individual indels still occur as competing events. One attractive solution to overcome these competing events is to use a nuclease-dead Cas9 to knock-in the loxP sites, such as prime editors [34], once targeting efficiencies for the system improve.
The two RNPs/two ssODNs method, delivered in one round or two, is highly flexible with the size of the region to be floxed. The largest region floxed in this study was 160 kb. If necessary, much larger sequences can in theory be floxed using the method, given the two loxP insertions are generally independent. Synthetic gRNAs and ssODNs can be obtained commercially and validated in cultured cells within a month at relatively low cost. The small size of ssODNs is compatible with efficient embryo electroporation and high-throughput NGS genotyping. Combined with the in vitro Cre assay and optional second round targeting, by electroporating two CRIPSR RNPs and two ssODNs, one can reliably obtain several floxed F1 animals in roughly five months either by one round or two round targeting.