Efficient RNA/Cas9-mediated genome editing in Xenopus tropicalis

Bacterial and archaeal clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated (Cas) adaptive immune systems rely on small RNAs in complex with Cas proteins to silence foreign nucleic acids, including viruses and plasmids. There are three major types of CRISPR/Cas systems (Makarova et al., 2011; Wiedenheft et al., 2012). In the type II CRISPR system, the Cas9 protein forms a complex with two short non-coding RNAs, namely the spacer-containing RNA (crRNA) and the trans-activating CRISPR RNA (tracrRNA), to selectively cleave the invading DNA. With the recapitulation of this DNA cleavage activity in vitro with purified Cas9 and an engineered single guide RNA (gRNA) molecule containing the minimal features of both spacer and tracr RNAs (Jinek et al., 2012), it was anticipated that the system could potentially be used in place of zinc-finger nucleases (ZFNs) or transcription activator-like effector nucleases (TALENs) for targeted genomic cleavage in higher organisms (Carroll, 2012; Jinek et al., 2012). Indeed, it has been shown that, via generation of site-specific DNA double-strand breaks in the target loci, RNA-guided Cas9 nuclease facilitates genome editing in yeast, nematode, fly, zebrafish and mice, in mouse and human cell lines, as well as in plants (Bassett et al., 2013; Chang et al., 2013; Cho et al., 2013; Cong et al., 2013; DiCarlo et al., 2013; Dickinson et al., 2013; Ding et al., 2013; Fujii et al., 2013; Gratz et al., 2013; Hwang et al., 2013; Jiang et al., 2013; Mali et al., 2013; Shen et al., 2013; Wang et al., 2013; Yang et al., 2013).

In the past decade, the diploid frog Xenopus tropicalis has emerged as an excellent amphibian genetic model (Harland and Grainger, 2011). We and others have established ZFN- or TALEN-mediated gene targeting protocols in this species (Ishibashi et al., 2012; Lei et al., 2012; Young et al., 2011). In comparison to ZFNs, TALENs are more effective in frogs. Despite the ease of designing TALE modules, it is by no means trivial to generate TALENs in the laboratory for use in large-scale reverse genetics. Thus, efficient genome engineering tools that can be easily and cost-effectively generated are still highly desirable. Here, we report that gRNA/Cas9 can serve as an easy, economic, efficient and reliable tool for targeted gene disruption in X. tropicalis.

First, we injected Cas9 mRNA at a dose of 500 pg per embryo together with gRNAs (50 pg/embryo) targeting ptf1a/p48, hhex or pat into one-cell stage X. tropicalis embryos. All three injection groups showed high levels of dead and deformed embryos (Fig. 1A,B), indicating non-specific toxicity. We then chose hhex and pat gRNAs to optimize the doses of Cas9 mRNA and gRNA for X. tropicalis embryos based purely on the morphological phenotype. The data obtained indicate that the optimal Cas9 mRNA dose is 300 pg/embryo (Fig. 1C) and the quantity of gRNA should not exceed 500 pg/embryo (Fig. 1D,E). In all subsequent experiments, we set the Cas9 mRNA dose at 300 pg/embryo; for a given locus, the gRNA dose was further optimized in the range 1-500 pg per embryo.

We initially designed gRNAs targeting 12 loci in ten different genes (Table 1; supplementary material Table S1). Those targeting elastase-T1, ets2, tm4sf4-T2, grp78, elastase-T2 and ptf1a/p48 readily exhibited targeting efficiencies above 72% at a dose of 50 pg/embryo and the first three even achieved 100% efficiency (Fig. 2A; supplementary material Fig. S1). The mutagenesis rates induced by gRNAs targeting hhex, tm4sf4-T1 and tyrosinase were raised from 31.3%, 60% and 60% to 100%, 86.7% and 82.4% when the gRNA doses were increased from an initial 50 pg/embryo to 500, 200 and 400 pg/embryo, respectively (Fig. 2A-D; supplementary material Fig. S1). The highest efficiency obtained for ets1 gRNA was 33.3%, and gRNAs targeting pat and pdx1 showed either very low efficiency or no effect at the various gRNA doses tested (Fig. 2A,E-G; supplementary material Fig. S1). We then designed two additional gRNAs for each of ets1, pat and pdx1. The data obtained indicate that all caused mutations with high efficiencies (45-79%) at 50 pg/embryo (Fig. 2H; supplementary material Fig. S2). Thus, all of the genes tested were readily targeted by gRNA/Cas9 with efficiencies above 45%. Finally, we chose one of the most effective gRNAs to scale down the gRNA dose, and found that 15 pg/embryo of elastase-T1 gRNA still exhibited 83.3% efficiency (Fig. 2I). Together, our results strongly suggest that gRNA/Cas9 can efficiently target most loci in the X. tropicalis genome.

To investigate the specificity of the gRNA/Cas9 system for genome editing in whole organisms, we chose two loci (ets2 and tm4sf4-T2) that displayed 100% targeting efficiency and systematically analyzed the consequence of single-nucleotide mismatches between the spacer and the protospacer sequences for targeting efficiency in X. tropicalis embryos. For both genes, point mutations up to the eleventh base pair upstream of the protospacer adjacent motif (PAM) completely abolished the targeting activity of gRNA/Cas9. By contrast, gRNA/Cas9-mediated target cleavage is partially tolerant to point mutations 12, 13, 15, 17, 18, 19 and 20 bp 5′ of the PAM (Fig. 3; supplementary material Figs S3 and S4).

To further assess whether gRNA/Cas9 creates any off-target mutations in frog embryos, we first computationally identified all the potential off-target sites with up to five mismatches to all the loci targeted in this study (supplementary material Table S2). Since no sites with one mismatch were identified, we selected 119 sites in total, including all four sites with two mismatches, seven sites with three mismatches distal to the PAM sequence, and all sites with up to four mismatches for the ets2, ptf1a/p48 and tyrosinase target loci, and performed a T7EI assay to identify any off-target disruptions (supplementary material Table S3). In contrast to the on-target sites, no potential off-target sites analyzed showed reliable gRNA/Cas9-dependent T7EI assay positive signals (data not shown). Our study suggests that the frequency of cleavage within potential off-target sites with two to four mismatches is too low to be detected by our T7EI assay.

To test whether this approach is suitable for multiplexed editing of genomic loci in Xenopus embryos, we co-injected Cas9 mRNA together with two gRNAs targeting grp78 and elastase-T1. The data indicate that the targeting efficiencies for each gene from the co-injection are almost identical to those obtained from the individual injections (Fig. 4A; supplementary material Fig. S5). Single-cell analysis for stage 9 embryos (blastulae) indicates that both alleles of the two loci targeted were mutated in the same cell (Fig. 4B,C). Given the high targeting efficiency in the founder embryos and high germ line transmission rates observed in this study with other genes, these data suggest that double or triple knockout lines of genes of interest in X. tropicalis could be established from a single injection of Cas9/gRNAs, which also appears to be achievable in mice (Wang et al., 2013).

In principle, the high efficiency of gene disruption induced by Cas9 nuclease could allow for direct phenotype assessment in gRNA/Cas9-injected Xenopus embryos. Our data indicated that the expression of the pancreas-specific marker gene pdip is indeed completely inhibited in a portion of ptf1a/p48 gRNA-injected embryos (Fig. 5A,E). The rest of the targeted embryos showed severe inhibition of pdip expression (Fig. 5B,F), with hardly any showing the strong signals seen in wild-type or elastase-targeted embryos (Fig. 5I-L). Co-injection of a dexamethasone-inducible variant of Ptf1a/p48 (Ptf1a/p48GR), which was activated after gastrulation, resulted in 100% rescue of the ptf1a/p48 gRNA-induced phenotype (Fig. 5D,H). Together, these findings are reminiscent of those obtained upon application of ptf1a/p48 morpholinos to Xenopus laevis embryos (Afelik et al., 2006; Jarikji et al., 2007). We also dissected six ptf1a/p48-targeted froglets that all showed severe pancreatic hypoplasia (Fig. 5N), consistent with our previous findings with ptf1a/p48 TALENs (Lei et al., 2012).

As a second example, we chose to phenotype the disruption of tyrosinase, which causes the ablation of pigmentation (Beermann et al., 2004; Damé et al., 2012; Ishibashi et al., 2012; Koga et al., 1995; Oetting et al., 2003). Upon tyrosinase gRNA/Cas9 injection, the majority of tadpoles (100/165, ∼61%) showed severe perturbation of pigmentation, with two showing almost full albinism, whereas the remainder displayed partial albinism with none showing the pigmentation pattern seen in wild-type siblings (Fig. 5O-V). The various levels of albinism were maintained to adulthood (Fig. 5X-Z).

The high efficiency of somatic targeting in gRNA/Cas9-injected embryos would suggest a similarly high targeting efficiency in germ cells of G0 frogs. Just as expected, all five founder male frogs from the targeting of either elastase-T1, elastase-T2 or tyrosinase transmitted their targeted mutations through the germ line with high efficiencies ranging from 40-100% (Fig. 6). The data indicate that gRNA/Cas9-induced mutagenesis in X. tropicalis is highly heritable.

We have shown that gRNA/Cas9 is an efficient, simple and robust tool for X. tropicalis genome editing with high precision and specificity. Specificity in genome editing is crucial to both basic research and therapeutic application. Our data from the systematic analysis of the effects of single-nucleotide mismatches between the spacer and the protospacer sequences on Cas9-mediated gene targeting efficiencies in X. tropicalis embryos are consistent with findings obtained in vitro and in bacteria and mammalian cell lines (Cong et al., 2013; Fu et al., 2013; Hsu et al., 2013; Jinek et al., 2012; Sapranauskas et al., 2011), further highlighting the importance of the 3′ protospacer sequence close to the PAM in designing gRNAs to eliminate off-target effects. In contrast to the high level of off-target cleavage reported in human cell lines using the CRISPR/Cas system (Cradick et al., 2013; Fu et al., 2013; Hsu et al., 2013), our data suggest that the gRNA/Cas9-induced off-target mutation rate is very low in X. tropicalis embryos, consistent with data obtained with mouse embryos (Yang et al., 2013). Future studies using whole-genome sequencing would generate more comprehensive information. Meanwhile, the use of paired gRNA/Cas9 nickases significantly improves the specificity (Ran et al., 2013).

Sequencing data indicate that both Cas9 and TALEN induce diverse indels (supplementary material Fig. S1) (Lei et al., 2012), suggesting that different embryonic cells are likely to harbor different genotypes of the targeted gene. If some of the mutations do not result in a loss of function, the injected embryos would display a mosaic phenotype. It is also possible that a few cells carry a disruption in only one allele, or even in neither allele. Thus, the mosaicism and monoallelic mutagenesis presumably explain why the sequence disruption efficiency of ptf1a/p48 and tyrosinase was high (72% and 82%, respectively), but the complete inhibition of pdip expression or ablation of pigmentation was low (14% and 1%, respectively); indeed, the rest of the targeted embryos showed intermediate mosaic phenotypes. tyrosinase TALENs also cause mosaic phenotypes in X. tropicalis embryos (Ishibashi et al., 2012). Obviously, the higher the targeting efficiency of the gRNA/Cas9 (especially for those with 100% gene targeting efficiency), the better the opportunity for phenotype assessment in G0 embryos.

Among the 12 gRNAs initially designed, two (∼17%) showed almost no activity. Inactive gRNAs have also been reported for zebrafish embryos (Hwang et al., 2013). We found that when faced with these rare inactive gRNAs, one could easily find active substitutes in the neighboring loci.

In conclusion, our results demonstrate that gRNA/Cas9 is a superb tool for multiplex genome editing in X. tropicalis. Given the simplicity and low cost of gRNA construction, the high targeting efficiency of the gRNA/Cas9 system, the high efficiency of germ line transmission, the relatively short generation time of frogs (4-6 months), and the availability of the frog genome sequence (Hellsten et al., 2010), there is no doubt that the diploid frog X. tropicalis is jumping into the future of developmental genetics.

The recently reported (Cong et al., 2013) codon-optimized Streptococcus pyogenes Cas9 cDNA together with the two attached nuclear localization signals (3xFLAG-NLS-SpCas9-NLS) was synthesized by GenScript and cloned into the pCS2+ vector (Rupp et al., 1994) (supplementary material Fig. S6). The construct was linearized with NotI and transcribed with the mMessage mMachine SP6 Kit (Ambion) to produce capped Cas9 RNA, which was purified with the RNeasy Mini Kit (Qiagen) according to the RNA clean protocol.

To create a gRNA expression vector, we placed a T7 promoter followed by two BbsI sites upstream of the recently described gRNA scaffold (Mali et al., 2013), which was synthesized by GenScript and cloned into the pUC57-Simple vector (GenScript) (supplementary material Fig. S6). The gRNAs were designed to target protospacer sequences in genes of interest with the form 5′-GG-(N)₁₈-NGG-3′ (Table 1). NGG is the PAM. The locus-specific 20 bp protospacer containing the cloning cohesive sites was obtained by annealing two synthesized partially complementary oligonucleotides (Table 1), and was then cloned into BbsI-digested gRNA expression vector. The resulting construct was digested with DraI and transcribed using the MAXIscript T7 Kit (Ambion). The gRNA was purified by miRNeasy Mini Kit (Qiagen).

Capped ptf1a/p48GR mRNA was generated as described (Afelik et al., 2006).

X. tropicalis frogs were purchased from Nasco. Ovulation and in vitro fertilization were carried out according to the protocol described previously (Khokha et al., 2002; Young et al., 2011). The desired amount of Cas9 mRNA and gRNA in 2 nl was co-injected into one-cell stage embryos. During subsequent development, dead and abnormal embryos (mainly due to gastrulation defects) were sorted out and counted for the purposes of morphological phenotyping.

Forty-eight hours after microinjection (about stage 40), we randomly pooled five healthy embryos from each injection, extracted genomic DNA, amplified the targeted region by PCR (for primers see supplementary material Table S4), and then cloned the purified PCR products into the pMD18-T vector (Takara) by TA cloning. Twenty single colonies were randomly picked for DNA sequencing analysis to detect any insertion or deletion (indel) mutations resulting from error-prone non-homologous end joining (NHEJ)-based repair of Cas9-created double-strand breaks. The targeting efficiency was determined by the ratio of mutant to total colonies.

For single-cell analysis, stage 9 embryos co-injected with Cas9 mRNA together with two gRNAs targeting grp78 and elastase-T1 were freed from the vitelline membrane and dissociated in calcium- and magnesium-free medium (88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO₃, 7.5 mM Tris pH 7.6). Single blastomeres from the dissociated embryos were separately cultured with 1× MBS solution (88 mM NaCl, 2.4 mM NaHCO₃, 1 mM KCl, 0.82 mM MgSO₄, 0.33 mM Ca(NO₃)₂, 0.41 mM CaCl₂, 10 mM HEPES pH 7.4) in 24-well plates lined with 0.8% agar overnight and then subjected to proteinase K digestion, PCR amplification and TA cloning. Ten single colonies from each blastomere progeny were sequenced to determine the genotype of individual blastomeres.

gRNA/Cas9-injected X. tropicalis embryos were raised to sexual maturity. Male founder frogs were crossed with wild-type females and individual F1 embryos were collected 48 hours postfertilization for genomic DNA extraction. Evaluation of mutations was carried out by PCR amplification, TA cloning and DNA sequencing of single colonies. Ten embryos from each founder frog were analyzed, and for each F1 embryo ten colonies were sequenced.

All genomic loci containing up to five mismatches compared with the coding sequence for a given 20 nt gRNA followed by the NGG PAM sequence were identified by mapping the targeted site to X. tropicalis genome V4.1 using a PERL script developed according to the SeqMap method (Jiang and Wong, 2008).

The T7EI assay was performed essentially as described (Guschin et al., 2010). For each injection, gRNA/Cas9-injected embryos or uninjected control embryos at stage 40 were pooled in groups of five for genomic DNA extraction. The regions of interest containing the off-target sites were amplified by PCR with gene-specific primers (supplementary material Table S3). PCR products were denatured and annealed under the following conditions: 95°C for 5 minutes, 95-85°C at -2°C/s, 85-25°C at -0.1°C/s, hold at 4°C. The annealed samples were digested with T7EI (NEB M0302L), separated and measured on an ethidium bromide-stained 10% polyacrylamide TBE gel and quantified using ImageJ software (NIH).

The digoxigenin-labeled antisense X. tropicalis pdip probe was transcribed with T7 RNA polymerase using an RT-PCR-amplified template containing the T7 promoter (forward, 5′-GAGGAGGAGACATCAGACGA-3′; reverse, 5′-CAGTAATACGACTCACTATAGGGAATACTCAAGGACCGAAGAAA-3′). Whole-mount in situ hybridization was performed as described (Harland, 1991).

We thank Yinying Long for technical assistance and frog husbandry.