Genome editing using artificial site-specific nucleases in zebrafish



Zebrafish is a model vertebrate suitable for genetic analysis. Forward genetic analysis via chemical mutagenesis screening has established a variety of zebrafish mutants that are defective in various types of organogenesis, and the genes responsible for the individual mutants have been identified from genome mapping. On the other hand, reverse genetic analysis via targeted gene disruption using embryonic stem (ES) cells (e.g., knockout mouse) can uncover gene functions by investigating the phenotypic effects. However, this approach is mostly limited to mice among the vertebrate models because of the difficulty in establishing ES cells. Recently, new gene targeting technologies, such as the transcription activator-like effector nucleases (TALEN) and clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9 systems, have been developed: that can directly introduce genome modifications at the targeted genomic locus. Here, we summarize these new and powerful genome editing techniques for the study of zebrafish.


Genetic analysis, which includes forward genetics and reverse genetics, is commonly used to study the molecular mechanisms regulating biological processes in animal models. Zebrafish (Danio rerio) is one of the simplest model vertebrates amenable to genetic analysis and has a high-quality genomic sequence assembly available (Howe et al. 2013). A zebrafish embryo rapidly develops external to the mother and is transparent during early embryogenesis (Dawid 2004). Indeed, most organs start to function within several days after fertilization. Another important reason that the zebrafish makes a good model is that its gene structure and function in organogenesis are highly conserved with vertebrates (Thisse & Zon 2002). Using forward genetics, zebrafish mutants defective in organogenesis were established from random mutagenesis screenings, and the genes responsible for particular phenotypes have been identified. Thus, unbiased zebrafish forward genetics elucidated new insights into gene function during vertebrate development (Haffter et al. 1996; Fukui et al. 2009).

Recently, using forward genetics, both Stainier's group and our group identified a novel physiological function of lipid mediator sphingosine-1-phosphate (S1P), which regulates cardiac development in zebrafish (Kupperman et al. 2000; Hisano et al. 2012b). Both spns2 (an S1P transporter) and s1pr2 (an S1P receptor/S1PR2) zebrafish mutants displayed the two hearts phenotype known as cardia bifida (Kupperman et al. 2000; Osborne et al. 2008; Kawahara et al. 2009), indicating that Spns2-S1PR2 signaling regulates the migration of cardiac progenitor cells. We found that zebrafish Spns2 and mammalian SPNS2 are expressed in the plasma membrane and transport intracellular S1P out of the cell (Nishi et al. 2013). In SPNS2-knockout mice the number of lymphocytes in the peripheral blood was significantly decreased compared to that of wild-type mice, because S1PR1-mediated lymphocyte egress from the thymus to the peripheral blood was suppressed (Hisano et al. 2012a). S1P specifically binds to G-protein coupled S1P receptors (S1PR1-S1PR5) on the target cells, leading to various types of cellular responses such as cell proliferation, differentiation and cell migration (Hla et al. 2008; Spiegel & Milstien 2011). However, comprehensive understanding of the physiological role of S1P signaling remains obscure. For this reason, we are using reverse genetics to establish S1PR knockout zebrafish to investigate their in vivo phenotypes.

Reverse genetics is a powerful approach for investigating the physiological functions of individual genes by genetically disrupting a specific gene. The technique has already been used in mice by gene targeted embryonic stem (ES) cells by homologous recombination. However, targeted gene disruption through homologous recombination is not available in many model vertebrates including zebrafish, because the requisite ES cells have not been established. As an alternative, TILLING (Targeting Induced Local Lesions in Genomes) was developed (Wienholds et al. 2002), but it requires large-scale sequencing to isolate the knockout fish. Another option is temporary gene knockdown using antisense morpholinos, which easily enables the examination of the gene function of interest (Nasevicius & Ekker 2000). However, such knockdown analysis is incomplete and it may result in unpredictable off-target effects. In fact, it is well known that morpholinos often induce cell death mediated by the nonspecific activation of p53 (Robu et al. 2007).

New genome editing techniques such as ZFN (zinc finger nuclease), TALEN (transcription activator-like effector nuclease) and CRISPR (clustered regularly interspaced short palindromic repeats)/Cas9 can introduce genome modifications in a variety of cell types (e.g., induced pluripotent stem/iPS cells) and in various model organisms including animals and plants (Carroll 2011; Cong et al. 2013; Joung & Sander 2013; Mali et al. 2013). In this review, we focus on the TALEN and CRISPR/Cas9 systems and their applicability for genome editing in zebrafish.

Genome editing using TALEN and CRISPR/Cas9 systems

Transcription activator-like effector nuclease consists of a chimeric molecule (Cermak et al. 2011; Sakuma et al. 2013) that includes the TALE (transcription activator-like effector) DNA binding domain and the FokI nuclease catalytic domain (Fig. 1A). TALE, which was originally discovered in the plant pathogenic bacteria Xanthomonas, contains a 34 amino acid repeat, with the exception of two hypervariable residues (RVD: repeat variable diresidues). Individual TALE repeats containing RVDs recognize a single nucleotide using simple rules (TALE code: NG=T, HD=C, NI=A, NN=G or A; Fig. 1A). Because the FokI catalytic domain functions as a dimer, targeted DNA double-strand breaks (DSBs) occur in the spacer region between the forward- and reverse-TALENs.

Figure 1.

Genome editing techniques of transcription activator-like effector nucleases (TALEN) and clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9. (A) The TALEN system. TALEN consists of the TALE DNA binding domain and the FokI nuclease catalytic domain. Individual TALE repeats containing RVDs (repeated variable diresidues) recognize a single nucleotide with simple rules (TALE code; NG=T, HD=C, NI=A, NN=G or A). DNA double strand breaks (DSBs) are induced in the spacer region (green line) between the forward- and reverse-TALEN target sites (blue lines). (B) The CRISPR/Cas9 system. CRISPR/Cas9 consists of a guide RNA and the Cas9 nuclease. The guide RNA recognizes sequences (20 bases) to the genomic target adjacent to the protospacer-adjacent motif (PAM) site of NGG (N: any nucleotide). CRISPR/Cas9 induces DSBs in the targeted genomic locus. (C) DSBs can be repaired by non-homologous end joining (NHEJ) or homologous recombination (HR). In NHEJ, both broken ends are rejoined with a high frequency of insertions and/or deletions (indels), resulting in frameshift-mediated gene disruption. HR is an error-free repair pathway that uses homologous templates.

In clear contrast to TALEN, the CRISPR/Cas9 system has two components (Cong et al. 2013; Mali et al. 2013): guide RNA (gRNA) and the Cas9 nuclease (Fig. 1B). In bacteria, the CRISPR system functions as acquired immunity against invading foreign DNA (viruses and plasmids). Guide RNA recognizes sequences to the target genome site (20 bases), which is followed by the protospacer-adjacent motif (PAM) sequence NGG (N: any nucleotide); DSBs are induced at this adjacent site (Fig. 1B). Thus, when combined with the Cas9 nuclease, the CRISPR/Cas9 system enables targeted genome modifications in various model organisms.

Figure 2.

Sphingosine-1-phosphate (S1P) signaling in zebrafish. (A) S1P is produced within the cells and is released through the S1P transporter Spns2. Secreted S1P binds to S1P receptors (S1PR1-S1PR5) on the target cell surface. Because both spns2 and s1pr2 mutants exhibit the two hearts phenotype (cardia bifida), as reported previously (Kawahara et al. 2009; Hisano et al. 2013a), it has been concluded that S1P signaling via Spns2-S1PR2 regulates the migration of cardiac progenitor cells. The position of the heart is indicated by the arrowhead. (B) Cardiac morphology (ventral view) was visualized using monomeric red fluorescent protein (mRFP) or enhanced green fluorescent protein (EGFP) expression driven by the cmlc2 (cardiac myosin light chain 2) cardiac-specific promoter. Using transcription activator-like effector nucleases (TALEN) technology, we established s1pr2 mutant zebrafish that presented cardia bifida at 28 h postfertilization (hpf). This phenotype was often observed in embryos injected with S1PR2-TALEN (400 pg) or S1PR2-gRNA (5 pg) with Cas9 mRNA (100 pg) at 28 hpf. WT, wild-type embryo at 28 hpf.

Both systems induce DSBs at a specific genomic locus. These DSBs can be repaired by two systems: non-homologous end joining (NHEJ) and homologous recombination (HR; Fig. 1C). NHEJ connects the end of the broken strands and efficiently introduces insertion/deletion (indel) mutations at the target site, which lead to frameshift-mediated gene disruptions (knockout). In the presence of a homologous fragment, HR is also operative, leading to gene replacement (knockin). HR-based knockin has been reported in zebrafish (Bedell et al. 2012; Zu et al. 2013), but its efficiency is very low.

Using the TALEN technology, we have established s1pr2 mutant zebrafish that exhibits the two hearts phenotype, which is identical to the previously reported s1pr2 or spns2 (an S1P transporter) mutant phenotypes (Fig. 2; Kupperman et al. 2000; Osborne et al. 2008; Kawahara et al. 2009). The injection of S1PR2-TALENs or S1PR2-gRNA with Cas9 mRNA into zebrafish embryos often causes the two hearts phenotype (Fig. 2B), indicating that both systems efficiently disrupt both alleles of the s1pr2 gene and work well in zebrafish.

TALEN-mediated genome modifications in zebrafish

We recently established a method for efficient generation of TALEN-mediated knockout zebrafish (Fig. 3). We first evaluated the in vivo TALEN activity by examining genomic DNAs from forward- and reverse-TALEN mRNA-injected embryos. If TALENs possessing enough activity were obtained, potential F0 founders injected with these TALENs were grown to adulthood. Potential F0 founders were mated with wild-type fish, and the germline transmission of the TALEN-mediated mutations was confirmed by examining whether the mutated alleles were genetically transmitted into the F1 embryos. Finally, F1 fish containing the specific mutant alleles were identified by genotyping using genomic DNA prepared from their fin clips.

Figure 3.

Strategy of generating knockout zebrafish using transcription activator-like effector nucleases (TALEN). TALEN mRNAs were injected into zebrafish embryos at the 1 cell stage. To examine the in vivo TALEN activity, genomic DNA was isolated from uninjected or TALEN-injected embryos. In vivo TALEN activity for an endogenous target locus was measured using the lacZα disruption assay. Potential F0 founders were mated with wild-type fish (WT), and the F0 founders that produced useful mutant alleles were determined using the heteroduplex mobility assay (HMA) with genomic DNA from individual F1 embryos. Genotyping of the growing F1 fish containing the useful mutant alleles were identified by HMA using genomic DNA from individual fin clips.

To establish knockout zebrafish, TALEN-induced genome modifications at the target site have to be detected at three different steps: the evaluation of the in vivo TALEN activity in the F0 embryos, the identification of potential F0 founders and the genotyping of growing the F1 fish containing the mutant alleles (Fig. 3). To evaluate the first step, the in vivo TALEN activity, we developed a simple lacZα disruption assay (Hisano et al. 2013a). For the second and third steps, the identification of the potential F0 founders and F1 fish containing heterogeneous mutant alleles, we found the heteroduplex mobility assay (HMA) to be very useful (Ota et al. 2013). These two methods have proven themselves to be powerful tools for the detection of genome modifications induced by artificial site-specific nucleases.

Evaluation of in vivo TALEN activity by the lacZα disruption assay

Transcription activator-like effector nuclease predominantly induces small indel mutations that efficiently lead to frameshifts at the target site. To measure the incidence of these TALEN-induced frameshifts, we developed a lacZα assay that uses α-complementation of the β-galactosidase gene, which contains a multi-cloning site (MCS; Hisano et al. 2013a) (Fig. 4). The principle of our lacZα disruption assay is very simple. TALEN-induced frameshifts are assessed as the β-galactosidase activity that is rescued by the lacZα gene, which contains a short genomic fragment derived from the TALEN-target site. The entire MCS is replaced by this fragment, which was polymerase chain reaction (PCR) amplified from genomic DNA derived from wild-type or TALEN-injected embryos. If locus-specific primers are used to fuse the wild-type fragment in-frame with the lacZα gene, all colonies containing the wild-type fragment become blue on the X-gal-treated plates. However, colonies containing the fragment amplified from the TALEN-injected embryos stochastically become white on the plate because the lacZα gene is disrupted by the TALEN-mediated frameshifts. Thus, the in vivo TALEN activity can be estimated by the number of blue and white colonies. Furthermore, we can easily determine the possible TALEN-induced indel mutations by sequencing the plasmid DNA derived from the white colonies, as these colonies contain the TALEN-mediated frameshifts. Thus, the lacZα assay is a versatile method for the quantitative evaluation of the genome editing activity of artificial site-specific nucleases.

Figure 4.

Evaluation of the activity of engineered nucleases in vivo. (A) Schematic representation of the lacZα disruption assay. Genomic DNA was prepared from uninjected and S1PR2-transcription activator-like effector nucleases (TALEN)-injected embryos, as reported previously (Hisano et al. 2013a). In the lacZα disruption assay, primer pairs for the TALEN-target site were designed to generate an in-frame fusion to the lacZα gene, resulting in blue colonies in the X-gal-treated selection plate. S1PR2-TALEN induces indel mutations at the target site, preventing functional lacZα from being generated as the amplified fragment contained TALEN-mediated frameshifts (white colonies). Thus, white colonies contain sequences with frameshift mutations at the target site. (B) Blue/white colonies of the lacZα disruption assay. Most colonies derived from the uninjected embryos were blue, whereas the number of white colonies (red circles) increased with the injection of S1PR2-TALEN. (C) Sequences of wild-type (WT) and S1PR2-TALEN-induced indel mutations. Deletions are indicated by the red dashes, and insertions are indicated by the red letters. WT shows the endogenous TALEN-target site sequence. The number of nucleotides deleted (−) or inserted (+) is indicated to the left. Blue: S1PR2-TALEN target sequences; green: spacer sequences.

The lacZα assay is also useful in evaluating the in vivo off-target effects mediated by the engineered nucleases and the RNA-guided nuclease. For example, using this assay, we confirmed that the cleavage activity of our TALENs for 3- to 6-base mismatched off-target sites (25–29 nucleotides binding target) was undetectable (Hisano et al. 2013a). It was reported that the TALENs failed to cleave 6-base mismatched off-target sites (36 nucleotides binding target), but partially cleaved 2-base mismatched off-target sites (Dahlem et al. 2012). Although TALEN has a highly specific recognition activity for the targeted genomic locus, CRISPR/Cas9 may have more off-target effects because of its short recognition motif (20 bases). It has been demonstrated that the CRISPR/Cas9 system has substantial off-target cleavage activity (Cradick et al. 2013). Because the lacZα assay is a very simple method, it should be very useful in detecting the in vivo off-target effects mediated by the engineered nucleases.

Detection of TALEN-induced mutations using the heteroduplex mobility assay

Examining the variation of TALEN-induced mutations and the rate of germline transmission is very important for the identification of potential zebrafish F0 founders. We have established a novel application of the heteroduplex mobility assay (HMA) in genome editing. Because TALEN induces various types of indel mutations at the target site, both homoduplexes and heteroduplexes are formed when the TALEN-targeted genomic locus is amplified by PCR using locus-specific primers from the genomic DNA of F1 embryos (Fig. 5A). HMA is based on the principle that heteroduplexes migrate more slowly than homoduplexes during polyacrylamide gel electrophoresis (PAGE) because of an opened single-strand configuration surrounding the mismatched region.

Figure 5.

Identification of mutant alleles by heteroduplex mobility assay (HMA). (A) The heteroduplex mobility assay. Heteroduplexes (two bands) migrate more slowly than wild-type (WT) homoduplexes and deletion mutant allele homoduplexes (deletion). Four base pairs (red and blue letters) were deleted in the mutated allele. (B) The potential F0 founders (S1PR1-transcription activator-like effector nucleases (TALEN)-injected F0 fish) were mated with wild-type fish as reported previously (Ota et al. 2013). The polymerase chain reaction (PCR) amplicons from the individual F1 genomic DNA were electrophoresed on a 15% polyacrylamide gel. PCR amplicons showing the HMA profiles (red numbers: 2, 4, 6, 11, 12 and 13) contained various mutant alleles, while the PCR amplicons with no heteroduplexes contained only the wild-type sequences. (C) Sequences of the wild-type (WT) and S1PR1-TALEN-induced indel mutations. The deleted and inserted nucleotides in the DNA sequences are indicated by red dashes and red letters, respectively. The wild-type sequence is shown at the top. Blue: TALEN target sequences; green: spacer sequences.

We previously mated potential F0 founders (e.g., S1PR1-TALEN-injected F0 fish) with wild-type fish and isolated genomic DNA from individual F1 embryos numbered 1–16 (Ota et al. 2013). When the mutant alleles were transmitted into the F1 embryos, we successfully observed unique heteroduplexes patterns (usually two bands) by PAGE (Fig. 5A). As shown in Fig. 5B, the F1 embryos 2, 4 and 6 showed similar HMA profiles and identical 4-base deletions based on the sequencing analysis. The F1 embryos 11 and 13 exhibited a different HMA profile and represented a 6-base deletion and 2-base insertion indel mutation. We did not observe any indel mutations in the F1 embryos that did not show heteroduplex bands, including F1 embryos 1, 3, 5, 7, 8, 9, 10, 14, 15 and 16. We therefore concluded that different HMA profiles in the assay correspond to distinct indel mutations. Furthermore, the efficiency of germline transmission by the F0 fish was estimated to be 38% (6 HMA profiles/16 embryos × 100), as judged by the incidence of HMA profiles.

We found that HMA is also very useful for genotyping F1 fish using genomic DNA from fin clips (Ota et al. 2013). Similarly, ZFN-mediated indel mutations have also been effectively detected using the HMA (Chen et al. 2012). Alternative methods, such as the T7 endonuclease I assay (Kim et al. 2009) and HRMA (high resolution melt analysis; Dahlem et al. 2012), can also be used to examine the genome modifications induced by the engineered nucleases, but these methods are not as effective at estimating the variations in the indel mutations. Therefore, HMA is a very useful method for detecting small genome alterations induced by engineered nucleases in various model organisms.

Conclusions and perspectives

In this review, we introduced two genome editing technologies, TALEN and CRISPR/Cas9, which directly induce genome modifications at targeted endogenous genomic loci in zebrafish. Reports have shown that specific gene disruption mediated by engineered nucleases is feasible in other animals, such as Drosophila, medaka, rat (Beumer et al. 2008; Ansai et al. 2013; Mashimo et al. 2013), and even plants, such as rice and Arabidopsis (Cermak et al. 2011; Li et al. 2012). For precise genome editing, it is very important to evaluate the engineered nuclease activity at the target site. The lacZα assay and the HMA are both very simple and useful tools for such evaluation. Moreover, the development of the TALEN and CRISPR/Cas9 systems has enabled the easy establishment of gene knockout animals and plants. Temporal knockdown of S1PR1 using antisense morpholinos causes vascular defects in zebrafish (Ben Shoham et al. 2012; Tobia et al. 2012; Mendelson et al. 2013). However, these vascular defects were not observed during early embryogenesis in s1pr1 zebrafish mutants established using the TALEN system (Hisano et al. 2013b), raising the possibility that the knockdown phenotype of this particular gene was not necessarily consistent with its knockout phenotype.

At present, there does not exist any efficient, targeted, knockin technology that uses either the TALEN or CRISPR/Cas9 systems. For example, although targeted integration of homologous fragments was performed in zebrafish (Bedell et al. 2012; Zu et al. 2013), its efficiency is quite low. It is apparent that gene targeting via homologous recombination can introduce more precise genome modifications, such as a particular point mutations, large deletions, gene replacement or conditional knockout, than other methods. Therefore, efficient targeted knockin technology could be instrumental in designing animal disease models for human genetic disorders in which the orthologue in the model animal is replaced with the gene responsible for the human genetic disease. Because of the genome information available, we foresee the ability to freely modify the genomes of a wide number of organisms using artificial site-specific nucleases and subsequently investigate the in vivo gene function.


We would like to thank Peter Karagiannis for valuable comments. This work was supported by the Program for Next Generation World-Leading Researchers (NEXT Program) and by the Japan Society for the Promotion of Science.