Effective gene targeting in rabbits using RNA-guided Cas9 nucleases

Dear Editor, Recently, zinc finger nuclease, transcription activator-like effector nuclease, and RNA-guided Cas9 endonuclease (Cas9) have emerged as powerful means for genome editing (Conklin, 2013; Gaj et al., 2013). These nucleases are efficient in generating double-strand breaks in the genome that can be repaired by error-prone nonhomologous end joining leading to a functional knockout (KO) of the targeted gene or used to integrate a DNA sequence at a specific locus through homologous recombination. Although the Cas9 system has been shown highly efficient in generating genetically engineered mice and rats (Li et al., 2013a, b; Wang et al., 2013), its feasibility in the rabbits still needs be determined. Here we report the use of Cas9 system to effectively generate targeted mutations in rabbit embryos and the production of KO rabbits. The rabbit is a classic model animal species. It is useful for the study of many human diseases such as atherosclerosis, cystic fibrosis, and acquired immunodeficiency syndrome. However, production of gene targeted transgenic (GTT) rabbits has been an extreme challenge. This is mainly due to the lack of germline transmitting embryonic stem cells and the very low efficiency of somatic cell nuclear transfer in rabbits (Chesne et al., 2002). In the present work, we used the Cas9 system to target the rabbit genome. Because embryo transfer work in large animal species is costly, we first established an in vitro system to test the efficacy of the RNA-guided Cas9 nucleases. Individual single guide RNAs (sgRNAs) were designed (Figure 1A and Supplementary Table S1) to target nine rabbit genes: apolipoprotein E (APOE), cluster of differentiation 36 (CD36), cystic fibrosis transmembrane conductance regulator, low-density lipoprotein receptor (LDLR), apolipoprotein CIII, scavenger receptor class B, member 1 (SCARB1), leptin, leptin receptor, and ryanodine receptor 2 (RyR2). For each gene, RNA mixture of Cas9 constructs (150 ng/μl Cas9 mRNA plus 6 ng/μl sgRNA) was microinjected into cytoplasm of pronuclear stage rabbit embryos (n = 290). We chose to use the concentration at 6 ng/μl for sgRNA because higher concentrations (12, 18, or 24 ng/μl) did not significantly improve the mutation rates (data not shown), and we speculate that higher quantity of sgRNA may increase the frequency of off-target events. Embryos were cultured in vitro, collected at blastocyst (BL) stage, and subjected to single embryo PCR and sequencing (n = 116) to identify mutations in the corresponding target locus. All nine sgRNAs generated mutations on their corresponding targeting loci with efficiencies ranging from 10% to nearly 100% (Figure 1D). High percentage (four out of nine) of the sgRNAs resulted in mutation rates higher than 50%; interestingly, bi-allelic mutations were also identified in these four, but not in those where mutation rates were 50% or lower (Figure 1D). Figure 1 Generation of KO rabbits with RNA-guided Cas9 nucleases. (A) Constructs of Cas9 RNA system used in this study. NLS, nuclear localization signal; bGH-pA, bovine growth hormone poly-A; PAM, protospacer adaptor motif. (B) Representative T7 endonuclease 1 ... After validating the in vitro gene targeting capacity of these Cas9 constructs, we continued to use four of them (APOE, CD36, LDLR, and RYR2) to produce KO rabbits. CD36, LDLR, and APOE KO rabbits are useful to study lipid metabolisms and atherosclerosis. The fourth line (RyR2 KO) will be used as a model to study heart arrhythmia. This is in recognition of the increasing demand for non-murine models for cardiovascular diseases in the research community. Cardiac physiology of rabbit better mimics that of human in a number of key areas than mouse does (Fan and Watanabe, 2003). For example, cholesteryl ester transfer protein, which plays a central role in the atherosclerotic process, is abundant in both human and rabbit plasma but absent in the mouse. Like humans, rabbits are very susceptible to diet-induced atherosclerosis, whereas wild-type (WT) mice do not develop atherosclerosis naturally. A total of 301 embryos were injected with one of the four Cas9 constructs and transferred to 10 pseudo-pregnant recipient rabbits (20–35 embryos per recipient). After 1 month gestation, nine (90%) recipients gave birth to 68 live kits (7.6/L), out of which 38 were identified as positive KO after initial T7 endonuclease assay and final confirmation by PCR sequencing (Figure 1B, C, and E). The term rate calculated as total term kits/total embryos is 22.6% (68/301). The KO rate calculated as total KO kits/total term is 55.9% (38/68). Consistent with the prediction based on the in vitro results, three (i.e. APOE, CD36, and RyR2) out of the four Cas9 constructs resulted in higher than 50% mutation rates and bi-allelic mutations. It remains to be tested whether mutations in these founder animals will faithfully transmit to the next generation. In agreement with previous reports, Δ15 mutant alleles were repeatedly discovered in six out of nine LDLR KO founder rabbits (Figure 1C, left lower panel), likely caused by microhomology-mediated end joining (Wang et al., 2013). One main concern with the Cas9 system for gene targeting is the off-target effects (Fu et al., 2013; Hsu et al., 2013). Recently, Hsu et al. (2013) examined Cas9-induced off-target mutation events in human 293T and 293FT cells. They found that sgRNA can tolerate as much as 4 nt changes in the seed sequence, and that the change of the protospacer adaptor motif (PAM) sequence from NGG to NAG does not totally abolish the targeting capacity. Their work suggested that there may be up to hundreds of potential off-target loci in a mammalian genome for one particular Cas9 seed sequence. In the present work, we examined off-target effects in all the KO founders (LDLR, RYR2, CD36, and APOE), following similar strategies described by Wang et al. (2013) in their mouse work. We used BLASTn to identify exact match to the 15 nt sequence (12 nt seed region and 3 nt NGG). A total of 160 potential off-target loci were identified, of which 9 were within an exon region (Supplementary Table S2). Considering the fact that mutations in an exon region are more likely to cause gene mutation and phenotype changes in the animals, we looked at these nine loci (Supplementary Table S3). None of the 38 founders contained mutations in these exon regions. This finding substantially alleviated the concerns of using Cas9-based gene targeting in rabbits. Notably, we used the most stringent way (i.e. exact match of 12 nt seed sequence plus the 3 nt NGG in the PAM), yet failed to identify any off-target events. Several factors may have contributed to the low frequency of off-target events. First, we used very low concentration of the sgRNA (6 ng/μl). In contrast, 12.5–50 ng/μl were used in similar mouse and rat studies (Li et al., 2013a; Wang et al., 2013). Secondly, we used RNA, whereas Hsu et al. (2013) used plasmid DNA. The much shorter half-life of RNAs may also reduce the off-target frequencies. Thirdly, we worked in an in vivo system of a different species, quite different from the in vitro system using immortalized human cell lines (e.g. 293T and 293FT) (Hsu et al., 2013). Consistent with our findings, recent reports of GTT mice and rats generated via Cas9 system also had very few detectable off-target mutations (Li et al., 2013b; Wang et al., 2013). Nevertheless, we admit that additional work may be necessary to examine all potential off-target loci. It is also noteworthy that the current assembly of rabbit genome is still incomplete. On the other hand, in the context of transgenic animal production, we argue that the initial focus could be narrowed to those that fall in an exon region, therefore making off-target examination feasible and affordable. Researchers who use these novel rabbit models in their future studies should however bear in mind that there may be mutations, although remotely, in other regions of the animal genome. To our knowledge, our work represents the first report of successful generation of GTT rabbits using RNA-guided Cas9 nucleases. We used 10 donor rabbits to harvest embryos, transferred 301 embryos to 10 recipient rabbits, and generated 38 KO founders. On average, for a given targeted gene, five rabbits (donors and recipients) were used and nine founder KO kits were produced in as few as two months. Such high success rates and time efficiency make production of GTT rabbits technically and economically feasible for biomedical research. In the near future, Cas9 system will likely contribute to the generation of knock-in, multiplex GTT, conditional GTT rabbits too, as has been demonstrated in mice (Wang et al., 2013; Yang et al., 2013). Taken together, we have established a highly effective Cas9-based system to produce KO rabbits. We achieved 100% success rates in all nine genes and generated four novel lines of KO rabbits (CD36, LDLR, RyR2, and APOE). Analysis on potential off-targets showed no effects in the Exon regions of KO rabbits, supporting the application of the Cas9 technology on generating GTT rabbits. Importantly, this success positions rabbits as a brand new animal model amendable to gene targeting in translational biomedical research arena. Supplementary material is available at Journal of Molecular Cell Biology online. This work was funded by grants from the National Institutes of Health (HL117491, HL088391, HL114038, and HL068878 to Y.E.C). This work utilized Core Services supported by Center for Advanced Models for Translational Sciences and Therapeutics (CAMTraST) at the University of Michigan Medical Center.