Genome editing using engineered nucleases such as transcription activator-like effector nucleases (TALENs) has become a powerful technology for reverse genetics. In this study, we have described efficient detection methods for TALEN-induced mutations at endogenous loci and presented guidelines of TALEN design for efficient targeted mutagenesis in medaka, Oryzias latipes. We performed a heteroduplex mobility assay (HMA) using an automated microchip electrophoresis system, which is a simple and high-throughput method for evaluation of in vivo activity of TALENs and for genotyping mutant fish of F1 or later generations. We found that a specific pattern of mutations is dominant for TALENs harboring several base pairs of homologous sequences in target sequence. Furthermore, we found that a 5′ T, upstream of each TALEN-binding sequence, is not essential for genomic DNA cleavage. Our findings provide information that expands the potential of TALENs and other engineered nucleases as tools for targeted genome editing in a wide range of organisms, including medaka.
Medaka, Oryzias latipes, is a small freshwater teleost that serves as an excellent vertebrate animal for genetic experiments. Although its biological features are mostly similar to those of the zebrafish, medaka has several additional advantages, such as smaller genome size, availability of a large number of highly polymorphic inbred strains, which differ in various aspects, and the ability to survive in a wider range of temperatures (Takeda & Shimada 2010). Medaka has been used in scientific studies on a wide range of topics, including early embryogenesis, evolutionary biology, functional genomics, and toxicology.
Recently, genome editing with engineered nucleases, such as zinc-finger nucleases (ZFNs) and transcription activator-like (TAL) effector nucleases (TALENs), has become a powerful technology for reverse genetics approaches (Carroll 2011; Joung & Sander 2013). These nucleases can induce site-specific DNA double-strand breaks (DSBs) that can be repaired by non-homologous end joining (NHEJ) or homology-directed repair (HDR), resulting in targeted mutagenesis by insertion and deletions (indels) or in targeted gene integrations via homologous recombination, respectively (Urnov et al. 2010). Our group and other researchers have previously demonstrated successful targeted mutagenesis in medaka using ZFNs (Ansai et al. 2012; Chen et al. 2012) and TALENs (Ansai et al. 2013). Although it is expensive and labor-intensive to generate zinc finger domain binding specifically to a certain genomic sequence, the modularity of the DNA-recognition domain in TALENs allows researchers to customize TALENs easily using modular assembly methods (Cermak et al. 2011; Reyon et al. 2012; Sanjana et al. 2012; Sakuma et al. 2013). The TALEN technology has become an effective and robust tool for inducing genomic modifications in a wide range of organisms, including medaka.
Here, we have described efficient detection methods for TALEN-induced mutations at endogenous loci and guidelines for TALEN design for targeted mutagenesis in medaka. We performed a heteroduplex mobility assay (HMA) using an automated microchip electrophoresis system, which is a simpler and high-throughput method for the detection of TALEN-induced mutations than previously described methods such as restriction fragment length polymorphism (RFLP) analysis (Huang et al. 2011; Ansai et al. 2013); DNA-cleaving assay with mismatch sensitive nucleases, for example, CEL1 endonuclease and T7 endonuclease I (Miller et al. 2011; Mussolino et al. 2011); high-resolution melting analysis (HRMA) (Dahlem et al. 2012); the LacZ disruption/recovery assay (Hisano et al. 2013); and HMA using poly-acrylamide gel electrophoresis (PAGE) (Chen et al. 2012; Ota et al. 2013). We found that some pairs of TALENs can dominantly induce specific patterns of mutations, which are thought to be due to short homologous sequences in the target genomic sequences. Furthermore, we also found that a 5′ T adjacent to the half site of each TALEN, which is the only limitation in design of the TALENs (Reyon et al. 2012), is not necessary for genomic DNA cleavage.
Materials and methods
Three strains of medaka were used in this study. The d-rR strain (Yamamoto 1953) was used for the DJ-1 experiment. The Cab strain (Furutani-Seiki et al. 2004) was used for the snca, mc4r, notch1b, and tbx6 experiments. The Sakyo population, a dark-colored wild-type (B/B) fish collected at Sakyo-ku, Kyoto, Japan, was used for the slc45a2 experiment.
The fish were handled in accordance with The Regulation on Animal Experimentation at Kyoto University. The fish were maintained in an aquarium with recirculating water under a 14/10-h day/night cycle at 26°C.
Design and construction of TALENs
We searched for potential TALEN target sites using TALE-NT 2.0 (https://tale-nt.cac.cornell.edu/) (Doyle et al. 2012) with the following parameters: (i) spacer length of 14–17; (ii) repeat array length of 15–18; and (iii) the upstream base of T only. TAL effector repeats were assembled as described previously (Ansai et al. 2013). Briefly, up to six TAL effector modules (NI for A, HD for C, NN for G, and NG for T) were cloned into the array plasmid pFUS (Cermak et al. 2011; Sakuma et al. 2013), and then the resulting repeat arrays were cloned into the expression vectors pCS2TAL3DD or pCS2TAL3RR (Dahlem et al. 2012). All the assembled RVD modules for each target site have been described in Table S1.
RNA preparation and microinjection
The TALEN expression vectors were linearized using digestion with NotI. Capped RNA sequences were synthesized using the mMessage mMachine SP6 kit (Life Technologies). The transcribed RNAs were purified using the RNeasy Mini kit (Qiagen) according to the RNA clean-up protocol. The RNAs were diluted with Yamamoto's Ringer's Solution (0.75% NaCl, 0.02% KCl, 0.02% CaCl2, and 0.002% NaHCO3, pH 7.3) (Kinoshita et al. 2000) and then microinjected into fertilized eggs essentially as described (Kinoshita et al. 2000).
Genomic DNA extraction
Embryos (after breaking the egg envelope [chorion] with fine forceps), larvae, and caudal fin clips from adult fish were lysed individually in 25 μL of alkaline lysis buffer (25 mmol/L NaOH and 0.2 mmol/L ethylenediaminetetraacetic acid [EDTA], pH 8.0) at 95°C for 15 min. After neutralization with 25 μL of 40 mmol/L Tris–HCl (pH 8.0), they were used as genomic DNA samples.
Heteroduplex mobility assay
To detect TALEN-induced mutations, we performed HMA (Chen et al. 2012; Ota et al. 2013) with the following modifications: a short fragment (80–200 bp) containing the target site of TALENs was PCR-amplified from DNA samples (described above) as templates using either KOD FX (Toyobo) or HybriPol DNA polymerase (Bioline). The primers used are listed in Table S2. The resulting amplicons were analyzed using a microchip electrophoresis system (MCE-202 MultiNA; Shimazu) with the DNA-500 reagent kit or 15% polyacrylamide gel (SuperSep DNA, Wako).
To detect homozygous mutants, 5 μL of a PCR product amplified from each sample was mixed with 5 μL of the PCR product from a wild-type fish. The mixtures were denatured and reannealed at 95°C for 5 min, followed by cooling to room temperature. The resulting samples were analyzed using the MultiNA system.
The genomic region that contained the target site of snca-TALENs was amplified using the primers snca-FW1 and snca-RV1 (Table S2) with HybriPol DNA polymerase. The resulting PCR product was precipitated with ethanol and was digested at 37°C for overnight in 10 μL of the MseI restriction digestion solution that consisted of NEBuffer 4, 100 µg/mL of bovine serum albumin (BSA), and 2.5 units of MseI (New England Biolabs). The digestion products were analyzed using agarose gel electrophoresis.
We examined the fish under a fluorescence stereomicroscope MZFLIII (Leica Microsystems), and captured the images with a digital color cooled charge-coupled camera device (VB-7010, Keyence).
For somatic mutation analysis of the fish injected with slc45a2-TALENs, those genomic regions that contained the target site of TALENs were amplified from the genomic DNA pools extracted from 12 TALEN-injected embryos using the primers listed in Table S2. The amplicons were subcloned into the EcoRI/XhoI site of the pBluescript KS II (+) vector. The fragment containing the cloned genomic sequence was amplified from each colony by using the M13 forward and reverse primers (Table S2), and each fragment was sequenced using a T7 promoter primer (Table S2).
For direct sequencing analysis of mutant fish generated using TALENs, the genomic region, which includes the target site of the TALENs was PCR-amplified and sequenced using the primers listed in Table S2.
Evaluation of the TALEN activities in the medaka embryos by HMA
As an initial step in creating genome-edited fish by using TALENs, we needed to evaluate each TALEN's activity on an endogenous targeting sequence. In our previous study (Ansai et al. 2013), we had evaluated TALEN activity by using an RFLP method, which involves laborious steps such as ethanol precipitation for purification and restriction enzyme treatment. To accelerate the production of gene knockout strains by using TALENs, we aimed to develop a more simple and high-throughput evaluation method. Thus, we analyzed the HMA results using the MultiNA system, an automated microchip electrophoresis system with high resolution, in order to detect mutations in medaka target genes of TALEN-injected fish. In this experiment, we designed and constructed two pairs of TALENs (site #1 and site #2) that target the medaka slc45a2 gene, known as the gene responsible for the oculocutaneous albinism type 4 (OCA4), whose mutation is carried by the b mutant strain of medaka (also called himedaka in Japanese) (Fukamachi et al. 2001, 2008). A total of 250 ng/μL of each pair of TALENs was injected into fertilized eggs of the Sakyo population. Although wild-type fish (B/B) show melanin pigmentation in both retinal pigment epithelium and skin melanophores at 5 days postfertilization (dpf) (Fig. 1a), some TALEN-injected embryos at 5 dpf displayed areas devoid of melanin pigmentation. The coverage of melanin-less areas was higher in embryos injected with the TALENs for site #2 (Fig. 1c) than for site #1 (Fig. 1b). To confirm that this lack of melanin pigmentation was due to lesions in the slc45a2 gene, 211 or 148 bp genomic fragments including the targeting sequence of the TALENs were PCR-amplified from each 5-dpf embryo injected with slc45a2-TALENs for site #1 or #2, respectively. MultiNA analysis of the PCR amplicons showed that multiple bands with lower mobility were observed in all analyzed TALEN-injected embryos while one dominant amplicon was observed in control embryos that were not injected with TALENs (Fig. 1d,e). The number of observed multiple bands is larger in the embryos injected with the TALENs for site #2 than #1. To confirm that these multiple bands are the result of the formation of heteroduplexes, a larger target genomic region (468 and 325 bp for site #1 and #2, respectively) was amplified from genomic DNA pools of 12 TALEN-injected embryos. These PCR products were subcloned into pBluescript KS II (+), and then each clone was sequenced. Seven of 23 (30%) and nine of 17 (53%) sequenced clones had altered sequences induced by the TALENs for site #1 (Fig. 1f) and site #2 (Fig. 1g), respectively. These indicated that HMA with the MultiNA system (HMA–MultiNA) can efficiently detect TALEN-induced mutations. The site #2 TALEN showed more mutant PCR products and a higher variety of mutation types than did the site #1 TALEN (Fig. 1d–g). These data suggest that the multiplicity of HMA profiles correlates with the TALEN activity as reported in a previous study using PAGE (Ota et al. 2013). As mentioned above, we found that combining HMA with the MultiNA system can evaluate the activity of different TALENs in vivo in injected medaka embryos.
Genotyping of both heterozygous and homozygous mutants by using HMA
Because the TALEN-injected G0 founders usually generate more than one type of mutant F1 fish, we need to characterize the genotype of each F1 fish. If TALENs induce mutations at low frequency, then many F1 fish would have to be screened to find mutants. Therefore, identification of a successful mutant strain requires a simple and high-throughput method. To confirm that combining HMA with the MultiNA system meets this purpose, we tried to determine genotypes of F1 fish derived from two G0 founders (founder #1F and #8M) injected with DJ1–TALENs (Ansai et al. 2013). We mated each G0 founder with a wild-type fish, and then the F1 progeny were raised to adulthood. Genomic DNA was extracted from fin clips of each F1 fish, and PCR amplicons containing the target site of the TALENs were subjected to HMA–MultiNA. Two unique HMA profiles were found in 11 of the 20 F1 fish (55%) analyzed from founder #1F (Fig. 2a), and three unique HMA profiles were present in all of the 17 fish analyzed (100%) from founder #8M (Fig. 2b). Direct sequencing of the PCR amplicons including the target genomic region revealed that the F1 fish from founder #1F had either a 7 bp deletion (Fig. 2a, ∆7) or an 8 bp deletion (Fig. 2a, ∆8) mutant allele together with the wild-type allele, whereas the F1 fish from the founder #8M had either a 4 bp deletion (Fig. 2b, ∆4), or a 7 bp deletion (Fig. 2b, ∆7), or a 15 bp deletion (Fig. 2b, ∆15) mutant allele together with the wild-type allele. These results indicate that HMA–MultiNA is an effective and rapid method for genotyping of F1 fish harboring TALEN-induced mutations.
Another issue to establish mutant strains is how efficiently homozygotes for a mutation can be distinguished from heterozygotes or wild-type fish. We investigated whether homozygotes can be identified by the HMA–MultiNA method. Heterozygous F1 fish harboring an 8-bp deletion induced by the DJ1–TALENs were mated. Each F2 fish was subjected to HMA–MultiNA, and we identified eight homozygotes (Fig. 2c, b, e, i, j, k, l, m, and p), five heterozygotes (Fig. 2c, a, c, f, h, and o), and three wild-type fish (Fig. 2c, d, g, n) in the 16 analyzed F2 fish. We also showed that comparable results were obtained with HMA using PAGE (Fig. S1). However, little difference between the sizes of PCR amplicon from homozygotes and wild-type fish is likely to make it difficult to distinguish between them. We therefore added a slight modification to the HMA–MultiNA method to reliably identify homozygous mutants. The PCR product amplified from a wild-type fish was added to each of the 11 PCR products derived either from the homozygotes for the mutation or wild-type fish. The mixtures were denatured and reannealed and were then analyzed using the MultiNA system again. We found that eight samples (Fig. 2d, b′, e′, i′, j′, k′, l′, m′, and p′) showed heteroduplex bands, and we identified them as homozygous mutants carrying an 8-bp deletion in both alleles. These results indicate that HMA is an easy and simple method for the detection of homozygous mutations by using the tandem approach with the PCR products amplified from wild-type fish samples.
Microhomology induces a specific pattern of mutations
The targeting genomic sequences of slc45a2-TALENs for site #1 and #2 includes two stretches of a 6-bp homologous sequence that are located 8 and 16 bp away from each other, respectively (shown in Fig. 1f,g as green underlined letters). Sequencing analysis of the embryos injected with slc45a2-TALENs for site #1 showed that six of the seven mutant clones (86%) identified had the same pattern of 14-bp deletions between the 6-bp regions of homologous sequences (Fig. 1f). In contrast, no specific pattern of mutations was found in the embryos injected with slc45a2-TALENs for site #2 (Fig. 1g). The induction of a specific pattern of mutations by TALENs was also observed in germline mutations of fish injected with snca-TALENs or mc4r-TALENs. Six G0 founder fish injected with snca-TALENs were mated with wild-type fish, and each F1 larva was genotyped using PCR-RFLP analysis with the MseI restriction enzyme and subsequent direct sequencing. All six G0 founders produced F1 progeny harboring an 11-bp deletion between the 3-bp of homologous sequence located 8 bp away from each other (shown as green underlined letters in Fig. 3a), with high efficiency (14–59% of the mutated fish; Fig. 3a). Similarly, each F1 embryo from seven G0 founders injected with mc4r-TALENs genotyped using HMA and subsequent direct sequencing, showed that all five mutagenized G0 founders produced F1 progeny harboring the same type of an 11-bp deletion between the 6-bp of homologous sequences located 5 bp away from each other (shown as green underlined letters in Fig. 3b), with high efficiency (66–100% of the mutated fish; Fig. 3b). These results suggest that each pair of TALENs harboring several base pairs of homologous sequences within their targeting genomic sequence has the potential to induce a specific pattern of mutations.
5′ T adjacent to a half site of each TALEN is not required for genomic DNA cleavage
Almost all the engineered TALEN monomers had binding sequences preceded by a 5′ T according to the design guidelines proposed in previous studies (Cermak et al. 2011; Reyon et al. 2012). To test whether the 5′ T adjacent to the half sites of each TALEN is necessary for genomic DNA cleavage, we constructed two pairs of TALENs at each targeting genomic locus, one having the TALEN binding sequence with a 5′ T and the other without a 5′ T. Two genomic loci, notch1b and tbx6, were selected as the target sites, and then TALENs with and without the 5′ T specificity (indicated with light blue and orange lines in Fig. 4, respectively) were constructed. A total of 160 ng/μL of each pair of TALENs was injected into the fertilized eggs of the Cab strain, and genomic DNA was extracted from each 5–7 dpf TALEN-injected embryo. Multiple heteroduplex bands were observed in HMA for all embryos injected with TALENs, indicating that all pairs of TALENs induced mutations in their target sequence (Fig. 4). However, embryos injected with tbx6-TALENs without a 5′ T displayed less multiple bands of slow migration (Fig. 4b). These results indicate that a 5′ T flanking the half site of each TALEN is not necessary for genomic DNA cleavage.
In this study, we describe an efficient method for the detection of TALEN-induced mutations by HMA with the MultiNA system. This approach has some advantages over previously described methods for the detection of TALEN-induced mutations. First, the induction of mutations can be detected at any desired loci without the requirement of a restriction enzyme site for RFLP analysis. Second, this approach consists of a simple procedure that includes PCR amplification of a genomic fragment containing the TALEN targeting sequence and subsequent electrophoretic analysis using the MultiNA system. Third, this approach is more widely applicable to detect mutant alleles in F1 or later generations induced not only by TALENs but also by other engineered nucleases, such as ZFNs and the CRISPR/Cas system. In addition to these characteristics that were similar to those of HMA performed using PAGE (Chen et al. 2012; Ota et al. 2013), our method has an additional advantage because it allows for high-throughput analysis (simultaneous analysis of a maximum of 96 samples) with automated treatment by MultiNA. Thus, HMA–MultiNA is a simple and high-throughput system to detect TALEN-induced genome modifications, but this system needs special equipment, such as MCE-202 MultiNA.
We found that some pairs of TALENs can frequently induce a specific pattern of mutation, that is, a deletion between two short (comprising several base pairs) regions of homologous sequence near the spacer region of these TALENs. The DSB repair–mediated deletions are mainly caused by three DSB repair pathways: classical non-homologous end joining (cNHEJ), microhomology-mediated end joining (MMEJ), and single-strand annealing (SSA) (McVey & Lee 2008). From these pathways, MMEJ uses microhomologous sequences to align the broken ends before joining, thereby causing deletions flanking the original break (McVey & Lee 2008). This mechanism suggests that the patterns we identified (∆14 in slc45a2-TALEN site #1, ∆11 in snca-TALENs, and ∆11 in mc4r-TALENs; Fig. 1f and 3) are likely to be produced by MMEJ. Although the targeting site of slc45a2-TALENs site #2 has two 6-bp homologous sequences located 16 bp away from each other, an expected specific pattern of 22-bp deletions was not observed in the embryos injected with those TALENs (Fig. 1g). This observation indicates that when a longer deletion is expected (probably >20 bp), the frequency of inducing the corresponding pattern of deletion is likely to decrease. Additionally, we found that the embryos injected with slc45a2-TALENs site #1 showed fewer bands than those injected with slc45a2-TALENs site #2 in their MultiNA gel images of HMA. This indicates that each HMA profile in the TALEN-injected G0 embryos can be influenced by each specific pattern of mutation.
From a practical standpoint, our findings suggest that microhomologous sequences in the targeting site of TALENs allow us to predict frequently induced patterns of deletions. In other words, consideration of microhomologies in the targeting site helps induce desired patterns and/or avoid unwanted patterns of mutations. Therefore, a search for homologous sequences can become one of the necessary steps in the protocol for the selection of a targeting site of TALENs. The relevant events can be induced by DSBs and therefore by their repair pathway, suggesting that this mechanism can be applicable to other engineered nucleases for targeted genome editing.
Previous studies found that nearly all TAL effector binding sites observed in nature are preceded by a 5′ T (Boch et al. 2009; Moscou & Bogdanove 2009). Furthermore, the crystal structure of a TAL effector bound to its DNA target suggests that two degenerate repeat folds N-terminal to the canonical repeats associate with a 5′ thymine (Mak et al. 2012). In fact, it was reported that a mutation of this thymine reduces TAL effector activity at the target site (Boch et al. 2009; Römer et al. 2010). However, we find in medaka that TALENs recognizing a targeting sequence without a 5′ T can induce targeted mutations as effectively as those with a 5′ T (Fig. 4). Similarly, mc4r-TALENs induced mutations in a targeting sequence preceded by a G or a C (Fig. 3b). This is confirmed by a report that TALENs can induce mutations at a targeting sequence preceded by a C in human cells (Miller et al. 2011). In summary, a T at the 5′ end of TALEN-targeting sites is not required for efficient induction of DNA cleavage. This observation expands the range of sequences targetable by custom-designed TALENs.
This work was supported by a Grant-in-Aid for the Japan Society for the Promotion of Science Fellows (to SA). The authors thank Tetsushi Sakuma and Takashi Yamamoto for the pFUS_A2A and pFUS_A2B vectors, Daniel F. Voytas for Golden Gate TALEN and TAL Effector Kit (Addgene; #1000000016), and Kazuyuki Hoshijima and David J. Grunwald for the pCS2TAL3DD and pCS2TAL3RR vectors.