CRISPR for genetically modified mice? Not so fast…

The use of genetically modified mice as model organisms for the study of human disease remains as relevant as ever¹ and CRISPR-based genome editing is rapidly changing the field. However, whereas the potential of CRISPR technology is very exciting, there are some limitations to its usefulness and drawbacks inherent to the technology can easily be overlooked.

CRISPR-based genome editing typically relies on the induction of a double strand break within the target genomic sequence by a nuclease, such as Cas9, as directed by a guide RNA^2,3. The nuclease is delivered to a zygote or early stage embryo and may either be delivered as a protein in a complex with sgRNA, as mRNA that is subsequently translated or as DNA. The nuclease may remain present in cells for some time, particularly when expressed from plasmid DNA, and is not 100% efficient. Consequently, the occurrence of nuclease-mediated double strand breaks is stochastic and can occur across several early cell divisions⁴. Repair of these double strand DNA breaks occurs via a number of competing pathways that have evolved to maintain integrity of the genome, such as non-homologous end-joining (NHEJ), microhomology-mediated end joining (MMEJ) and homology directed repair (HDR)⁵. With the exception of HDR, repair pathways are mutagenic and repair of CRISPR-mediated double stand breaks by NHEJ typically results in small insertions and deletions (INDELS) of less than 20 bp. CRISPR-mediated double strand breaks can induce large deletions (kilobases), complex genomic rearrangements and chromosomal truncations that are often overlooked by the genotyping methods employed^6,7.

Consequently, CRISPR founder animals are almost always mosaic in that they harbour multiple, and potentially complex, alleles within a single individual^8,9. It is therefore necessary to have a genotyping strategy that allows for the detection of the desired allele amongst the mixture of different alleles¹⁰. Founder animals can be screened for the presence of the desired allele and its presence verified in the F1 offspring, assuming the allele contributes to the germline. Depending on the nature of the study, this can add considerable additional time and cost as well as increased animal numbers. For example, the authors of one study generating large knockouts concluded that while CRISPR was faster at generating the desired genomic deletions, understanding the nature of the alleles it generated may ultimately involve more time and effort than using classical Cre-Lox systems¹¹.

In many instances, CRISPR is used to generate non-conditional knockout lines where the generation of multiple alleles may not be of primary concern, provided that they are predicted to function as null alleles. However, CRISPR is also commonly used for the generation of knock-in (KI) lines via the repair of double strand breaks utilising HDR with a repair template. As HDR typically occurs at a lower rate than competing repair pathways¹², the efficiency of generating knock-in lines by CRISPR can vary. For example, the generation of conditional knockout (conKO) lines by CRISPR requires the simultaneous insertion of two loxP sites. An initial approach utilised two guide RNAs and two single-stranded oligonucleotides each encoding a loxP site¹³. However, a consortium of core facilities has reported the generation of only 15 conKO mice from a total of 17,887 engineered zygotes, indicating that the technique is very inefficient and difficult to reproduce¹⁴. Subsequently, an improved technique utilising a single long single-stranded oligonucleotide to introduce the two loxP sites in a single HDR event has been described^15,16. However, the technique still requires electroporation or pro-nuclear injection of a large numbers of zygotes in order to generate correctly targeted conKO (floxed) mice, and the use of lssODNs is limited by size and sequence constraints. Analysis of founder mice generated by this technique shows that they are typically mosaic with a range of alleles, again highlighting the need for extensive genotyping so that mice harbouring additional or incorrect alleles are not used in experimentation¹⁷.

A recent study systematically compared the efficiency of CRISPR to classic gene targeting by homologous recombination for the generation of large KIs (>1.5kb)¹⁸. The authors analysed 221 experiments that targeted 128 loci in murine embryonic stem (ES) cells. While CRISPR often produced more targeted ES cells than classic gene targeting when screen by overlapping PCR, further investigation by Southern hybridisation revealed that many of these clones had large genomic deletions and rearrangements. When this result was taken into consideration, classic gene targeting was found to produce clones with the correct integration of the KI sequence at a higher frequency than CRISPR. The authors stated that the observation of genomic deletions and rearrangements occurring with CRISPR ‘ultimately rendered CRISPR/Cas9 less efficient than classical homologous recombination for the production of fully validated clones.’

In addition to these limitations to CRISPR-mediated genome engineering, it is also possible that the nuclease may induce a double strand break at a region within the genome other than the target site. There has been concern that such off-target effects could confound experimental results in instances where off-target mutations result in a phenotype. Recent modifications to the CRISPR technique, such as improved gRNA design and the development of improved nucleases with higher specificity have alleviated some of this concern and recent studies have shown that off-target mutations occur less frequently in engineered animals as opposed to cell lines¹⁹. However, off-target mutations do still occur. Whole-genome sequencing of ten genome-engineered mouse embryos found 43 off-target mutations, of which only 30 were predicted by an in silico prediction tool²⁰. Where they do occur, off-target mutations will segregate from the targeted allele, if not genetically linked, with subsequent breeding. The possible confounding effect of off-target mutations remains a consideration for lines generated in a pre-clinical setting and remains a significant concern for the application of genome engineering as a human therapeutic²¹.

Successful generation of an animal model by CRISPR depends on many variables²². These factors include the locus being targeted, the developmental stage of the embryo being targeted, the experimental design and sequence of the guide RNA, the nuclease being used, the template used (if any) for HDR, and the method of delivery of CRISPR reagents. Mosaicism of the founder animals should be expected and a means of genotyping either the founders and/or their offspring is required to select animals that harbour the desired alleles. In instances where the desired allele is present at low frequency, a large number of founder or F1 mice may need to be generated. The possibility of off-target effects should also be considered and investigated where appropriate. When these factors are considered, the generation of genetically modified animal lines by CRISPR may not always be as technically simple or cost effective as commonly expected.

Download technical note as PDF

1. Nadeau JH, Auwerx J. The virtuous cycle of human genetics and mouse models in drug discovery. Nat Rev Drug Discov. 2019;18(4):255-272. doi:10.1038/s41573-018-0009-9
2. Jinek M, Chylinski K, Fonfara I, Hauer M, Doudna JA, Charpentier E. A Programmable Dual-RNA–Guided DNA Endonuclease in Adaptive Bacterial Immunity. Science. 2012;337(6096):816. doi:10.1126/science.1225829
3. Cong L, Ran FA, Cox D, et al. Multiplex Genome Engineering Using CRISPR/Cas Systems. Science. 2013;339(6121):819. doi:10.1126/science.1231143
4. Mehravar M, Shirazi A, Nazari M, Banan M. Mosaicism in CRISPR/Cas9-mediated genome editing. Dev Biol. 2019;445(2):156-162. doi:10.1016/j.ydbio.2018.10.008
5. Ranjha L, Howard SM, Cejka P. Main steps in DNA double-strand break repair: an introduction to homologous recombination and related processes. Chromosoma. 2018;127(2):187-214. doi:10.1007/s00412-017-0658-1
6. Kosicki M, Tomberg K, Bradley A. Repair of double-strand breaks induced by CRISPR–Cas9 leads to large deletions and complex rearrangements. Nat Biotechnol. 2018;36(8):765-771. doi:10.1038/nbt.4192
7. Cullot G, Boutin J, Toutain J, et al. CRISPR-Cas9 genome editing induces megabase-scale chromosomal truncations. Nat Commun. 2019;10(1):1136. doi:10.1038/s41467-019-09006-2
8. Yen S-T, Zhang M, Deng JM, et al. Somatic mosaicism and allele complexity induced by CRISPR/Cas9 RNA injections in mouse zygotes. Dev Biol. 2014;393(1):3-9. doi:10.1016/j.ydbio.2014.06.017
9. Oliver D, Yuan S, McSwiggin H, Yan W. Pervasive Genotypic Mosaicism in Founder Mice Derived from Genome Editing through Pronuclear Injection. Fujii H, ed. PLOS ONE. 2015;10(6):e0129457. doi:10.1371/journal.pone.0129457
10. Mianné J, Codner GF, Caulder A, et al. Analysing the outcome of CRISPR-aided genome editing in embryos: Screening, genotyping and quality control. Methods. 2017;121-122:68-76. doi:10.1016/j.ymeth.2017.03.016
11. Boroviak K, Fu B, Yang F, Doe B, Bradley A. Revealing hidden complexities of genomic rearrangements generated with Cas9. Sci Rep. 2017;7(1):12867. doi:10.1038/s41598-017-12740-6
12. Chu VT, Weber T, Wefers B, et al. Increasing the efficiency of homology-directed repair for CRISPR-Cas9-induced precise gene editing in mammalian cells. Nat Biotechnol. 2015;33:543.
13. Yang H, Wang H, Shivalila CS, Cheng AW, Shi L, Jaenisch R. One-Step Generation of Mice Carrying Reporter and Conditional Alleles by CRISPR/Cas-Mediated Genome Engineering. Cell. 2013;154(6):1370-1379. doi:10.1016/j.cell.2013.08.022
14. Gurumurthy C, Quadros R, Adams J, et al. Re-Evaluating One-Step Generation of Mice Carrying Conditional Alleles by CRISPR-Cas9-Mediated Genome Editing Technology. Genetics; 2018. doi:10.1101/393231
15. Quadros RM, Miura H, Harms DW, et al. Easi-CRISPR: a robust method for one-step generation of mice carrying conditional and insertion alleles using long ssDNA donors and CRISPR ribonucleoproteins. Genome Biol. 2017;18(1):92. doi:10.1186/s13059-017-1220-4
16. Miyasaka Y, Uno Y, Yoshimi K, et al. CLICK: one-step generation of conditional knockout mice. BMC Genomics. 2018;19(1):318. doi:10.1186/s12864-018-4713-y
17. Codner GF, Mianné J, Caulder A, et al. Application of long single-stranded DNA donors in genome editing: generation and validation of mouse mutants. BMC Biol. 2018;16(1):70. doi:10.1186/s12915-018-0530-7
18. Rezza A, Jacquet C, Le Pillouer A, et al. Unexpected genomic rearrangements at targeted loci associated with CRISPR/Cas9-mediated knock-in. Sci Rep. 2019;9(1):3486. doi:10.1038/s41598-019-40181-w
19. Kimberland ML, Hou W, Alfonso-Pecchio A, et al. Strategies for controlling CRISPR/Cas9 off-target effects and biological variations in mammalian genome editing experiments. J Biotechnol. 2018;284:91-101. doi:10.1016/j.jbiotec.2018.08.007
20. Anderson KR, Haeussler M, Watanabe C, et al. CRISPR off-target analysis in genetically engineered rats and mice. Nat Methods. 2018;15(7):512-514. doi:10.1038/s41592-018-0011-5
21. Herai RH. Avoiding the off-target effects of CRISPR/cas9 system is still a challenging accomplishment for genetic transformation. Gene. 2019;700:176-178.
22. Peng R, Lin G, Li J. Potential pitfalls of CRISPR/Cas9-mediated genome editing. FEBS J. 2016;283(7):1218-1231. doi:10.1111/febs.13586