Efficient TALEN construction and evaluation methods for human cell and animal applications


  • Communicated by: Masayuki Miura

Correspondence: tybig@hiroshima-u.ac.jp


Transcription activator–like effector nucleases (TALENs) have recently arisen as effective tools for targeted genome engineering. Here, we report streamlined methods for the construction and evaluation of TALENs based on the ‘Golden Gate TALEN and TAL Effector Kit’ (Addgene). We diminished array vector requirements and increased assembly rates using six-module concatemerization. We altered the architecture of the native TALEN protein to increase nuclease activity and replaced the final destination vector with a mammalian expression/in vitro transcription vector bearing both CMV and T7 promoters. Using our methods, the whole process, from initiating construction to completing evaluation directly in mammalian cells, requires only 1 week. Furthermore, TALENs constructed in this manner may be directly applied to transfection of cultured cells or mRNA synthesis for use in animals and embryos. In this article, we show genomic modification of HEK293T cells, human induced pluripotent stem cells, Drosophila melanogaster, Danio rerio and Xenopus laevis, using custom-made TALENs constructed and evaluated with our protocol. Our methods are more time efficient compared with conventional yeast-based evaluation methods and provide a more accessible and effective protocol for the application of TALENs in various model organisms.


Genome editing with artificial nucleases is hailed as a next-generation technology, enhancing gene targeting in previously nonpermissive model organisms (McMahon et al. 2012). Artificial nuclease pairs comprise specific, juxtaposed DNA-binding domains and nonspecific nuclease domains, which dimerize to induce DNA double-strand breaks (DSBs) at targeted loci, triggering endogenous repair pathways and stimulating homologous recombination (de Souza 2012). So far, zinc-fingers (ZFs) and transcription activator-like effectors (TALEs) have been developed to provide custom-engineered DNA-binding proteins. Fused to the FokI nuclease domain, both zinc-finger nucleases (ZFNs) and TALE nucleases (TALENs) have been shown to be effective for genome editing in cultured cells and various organisms (Urnov et al. 2010; Carroll 2011; Mussolino & Cathomen 2012).

Between the two nuclease technologies, there are clear advantages and disadvantages. ZFNs generally comprised three to six ZF domains that recognize 9–18-bp DNA, whose protein size is much more compact than that of TALEs. However, tandem repeats of ZF domains are known to interfere with each other, making effective construction of de novo ZF arrays difficult (Cathomen & Joung 2008). TALEs, however, show no context dependency and thus may be designed relatively easily (Bogdanove & Voytas 2011). The DNA recognition code of TALEs comprises only two amino acid variations, known as a repeat-variable di-residue (RVD), in a highly conserved 34 amino acid repeat unit that binds one nucleotide (Boch et al. 2009). In accordance with this advantage of TALENs, many methods to construct designer TALEs have been reported (Cermak et al. 2011; Li et al. 2011, 2012; Morbitzer et al. 2011; Sander et al. 2011; Weber et al. 2011; Zhang et al. 2011; Briggs et al. 2012; Reyon et al. 2012). One of the most popular open-source TALEN construction kits, the ‘Golden Gate TALEN and TAL Effector Kit’, is available from Addgene (Cambridge, MA). This kit contains 72 vectors, including TALE DNA-binding module plasmids, vector plasmids for the intermediary arrays and the final destination vectors containing the Xanthomonas oryzae pv. Oryzae-derived PthXo1-based TALE scaffold is well characterized, along with AvrBs3 from X. campestris pv. vesicatoria (Christian et al. 2010). Serial assembly of modules is based on a cycling type IIs restriction endonuclease reaction, reportedly taking only 5 days for TALEN construction (Cermak et al. 2011). Yet, the rate of success in independent laboratories is variable.

Importantly, many aspects of this new TALEN technology remain uncertain; in particular, how to optimize fusion of the DNA-binding and nuclease domains. Recent studies have shown that various truncations of the N- and C-terminal regions of the native TALE proteins affect DSB-forming activity (Miller et al. 2011; Mussolino et al. 2011; Sun et al. 2012).

To apply TALENs as a standardized technology in various model organisms, evaluation methods and criteria for constructed TALENs and derivatives stand as additional bottlenecks. Presently, in vitro cleavage assays and yeast-based assay systems are mainly used for the evaluation of custom-made TALENs (Cermak et al. 2011; Li et al. 2011, 2012; Mahfouz et al. 2011). These systems are robust but time-consuming, and it is unclear whether the activities in vitro or in yeast parallel that of higher eukaryotes including animal embryos and cultured cells. Cel-I or T7 endonuclease I cleavage assays directly screen DSB-forming activity by detecting imperfect nonhomologous end-joining repair induced by TALENs (Hockemeyer et al. 2011; Reyon et al. 2012; Sanjana et al. 2012). Such cleavage assays are suitable for direct evaluation of TALEN activity against their intended endogenous genes, but, unlike in vitro assays, are cell line or animal dependent and not necessarily applicable as universal TALEN validation assays. Furthermore, application of TALENs in essential model animals has gradually been realized by various independent groups using different TALEN architectures (Huang et al. 2011; Sander et al. 2011; Tesson et al. 2011; Wood et al. 2011; Cade et al. 2012; Liu et al. 2012; Moore et al. 2012), yet still no systematic methods exist which cover the construction and evaluation of TALENs that can be commonly used for all such cells and animals.

Here, we report unified methods for the construction and evaluation of TALENs, and we furthermore show the effectiveness of our TALENs. We modified vector constructs and protocols based on the ‘Golden Gate TALEN and TAL Effector Kit’, not only clarifying assembly but also TALEN evaluation and application. We show that re-engineering the TALEN architecture enhances activity in assays and practical applications. Furthermore, we validated our tools and TALEN construction methods by successful endogenous gene modification of cultured cells and multiple animal species. Our method ensures a convenient and reliable system of custom TALEN development for use in various cells and animal applications.


Adaptations permitting rapid construction of custom TALENs

We first modified the intermediary array vectors of the ‘Golden Gate TALEN and TAL Effector Kit’ to realize more robust first-step assembly. On a trial basis, we found that less than six-module ligation showed reproducible results in various independent laboratories. On average, success rate of ten-module assembly was approximately 10%, meanwhile success rate of six-module assembly was almost 100% in our groups. Additionally, like the original kit, six modules permit the construction of TALENs with up to thirty-one repeats without increasing the number of cloning steps (Fig. 1A). To incorporate our six-module ligation system to the Voytas kit, we have only to replace the pFUS_A, A30A and A30B capture vectors with our pFUS_A1A, A2A, A2B, A3A, A3B, A4A and A4B capture vectors. Note that pFUS_A3C and A4D are identical to pFUS_A2B, and pFUS_A4C is identical to pFUS_A3B. In using these specialized 6-module capture vectors, we decrease the number of RVD module vectors by 40% (from 40 to 24), minimizing efforts for RVD plasmid library preparation and the complexity of the first Golden Gate reaction. As TALEN technology is in a highly dynamic phase of development, this adaptation ensures that incorporation of new variant RVD vectors, such as NH and NK, recently shown to provide more context-relevant ‘G’ nucleotide recognition (Streubel et al. 2012), will require reduced effort. By simply incorporating our vectors into the original kit, we diminish the efficiency barrier of the first Golden Gate cloning step.

Figure 1.

Simplified method for the rapid construction and evaluation of custom TALENs. (A) Schematic overview of the six-module assembly method. Six or less than six modules are ligated into array plasmids in a first step. Subsequently, constructed arrays are joined into a mammalian expression vector in a second step. Bases in white and pink boxes represent overhangs left by BsaI and Esp3I, respectively. The vectors typed in red are modified vectors from the original kit. Spec, spectinomycin; Amp, ampicillin; CMV, cytomegalovirus promoter. (B) Scheme of the SSA assay. HEK293T cells were transfected with three types of plasmids, comprising TALEN-expressing plasmids, a reporter plasmid and a reference plasmid for dual-luciferase assay. The reporter plasmid encodes two split inactive parts of the luciferase gene with overlapping repeated sequences. Following a DSB caused by the TALENs, a functional luciferase gene is generated by an SSA reaction. (C) Timeline of TALEN construction and evaluation. It takes only 5 days for the construction and the evaluation of the activity completes in day six.

TALEN activity in HEK293T-based SSA assay parallels genome modification frequency

Next, we converted the evaluation system to a mammalian cell-based SSA assay (Fig. 1B), which has been previously characterized for the evaluation of ZFN activity in cultured cells and several animal embryo applications (Ochiai et al. 2010, 2012; Ansai et al. 2012; Kawai et al. 2012; Song et al. 2012; Watanabe et al. 2012). To conveniently transition constructed TALENs to a human cell–based assay, we replaced the capture vector of the second Golden Gate assembly step with a mammalian expression vector bearing cytomegalovirus (CMV) and T7 promoters (Fig. 1A). As the result, assembled TALEN plasmids can be directly used in the SSA assay, transfection into target cells or mRNA synthesis for embryo injection without additional cloning. These improvements allow us to construct, evaluate and apply custom TALENs easily and quickly (Fig. 1C).

To validate the mammalian cell–based SSA assay (Fig. 2A), we constructed HPRT1_B TALENs similar to those originally described in the Cermak paper (Cermak et al. 2011). Compared with cells transfected with TALENs and a reporter vector bearing unrelated target sequences, cells transfected with TALENs and the associated SSA reporter showed marked activation in a dose-dependent manner (Fig. 2A). Additional previously reported TALENs HPRT1_A, CFTR_A, CFTR_B, GFP and eGFP_A (Cermak et al. 2011) were similarly constructed using our 6-module construction system. As eGFP-A TALENs showed low activity in the previous article (Cermak et al. 2011), we designed and constructed another TALEN pair for the eGFP gene in addition to them (eGFP_B). Our SSA assay showed that constructed TALENs had a variety of activities (Fig. 2B). Interestingly, despite the full accordance of amino acid sequences, the previously reported yeast-based assay and our HEK293T-based assay showed disparate TALEN activities, especially for CFTR_A and GFP (Cermak et al. 2011) (Fig. 2B). Therefore, we used the Cel-I assay (Guschin et al. 2010) for the HPRT1_A, HPRT1_B, CFTR_A and CFTR_B TALENs (Fig. 2C) as readout for direct genome modification. By using Cel-I nuclease, amount of mutated DNAs can be easily measured. After the TALEN introduction, PCR amplification around target sequence, followed by denaturation and reannealing, is carried out so that they can generate heteroduplexes when TALENs induce mutations. As Cel-I nuclease digests DNAs with base-mismatches, we can analyze approximate mutation rate by simple electrophoresis of Cel-I-digested product. As a result, cleaved products appeared in the HPRT1_A or HPRT1_B TALEN-treated genomes, but no cleaved products were observed in the CFTR_A or CFTR_B TALEN-treated genomes, showing some correlation with our SSA assay data (Fig. 2B,C). In contrast, the previous yeast-based assay showed that CFTR_A TALEN had higher activity than HPRT1_A (Cermak et al. 2011), which is strikingly different from our Cel-I data. These results suggest that our HEK293T cell-based SSA assay system represents a relatively accurate activity score compared with the yeast-based assay, at least in human cells.

Figure 2.

Functional evaluation of engineered TALENs by SSA and Cel-I assay. (A) Evaluation of SSA activity of a TALEN pair targeted to human gene, HPRT1. Data are expressed as means ± SEM (n = 3). (B) Comparison of the activity of several custom TALENs by SSA assay. Relative activity is defined as the ratio of measured activity to the activity score of pSTL-ZFA36. Data are expressed as means ± SEM (n = 3). (C) Cel-I assay using custom TALENs for human genes. Arrowheads indicate the expected positions of the digested products. % NHEJ (nonhomologous end joining) was estimated using ImageJ software as previously described (Hansen et al. 2012).

N- and C-terminal deletions of PthXo1 TALE scaffold increase TALEN activity

Based on the previous report that describes deletion scaffolds of AvrBs-based TALENs with high activity and low toxicity (Mussolino et al. 2011), we similarly designed N- and C-terminal truncated PthXo1-based TALENs without disrupting the native N-terminal nuclear localization signal (NLS; Fig. 3A). In the previous article, TALENs with a +153 N-terminal domain and a +47 C-terminal domain, named NC scaffold, showed the highest cleavage activity when separated by 12 to 15 bp spacer lengths (Mussolino et al. 2011). Therefore, we constructed truncation variants of HPRT1_B TALENs, named HPRT1_B TALEN-C and HPRT1_B TALEN-NC, and evaluated them both by SSA and Cel-I assays (Fig. 3A–C). TALEN-C retains a full N-terminal domain and +47 C-terminal domain, whereas TALEN-NC has a +153 N-terminal domain and +47 C-terminal domain (Fig. 3A). RVD arrays may be assembled directly in these truncation variants by the usual Golden Gate cloning method with pcDNA-TAL-C or pcDNA-TAL-NC capture vectors instead of pcDNA-TAL. SSA assay results using reporter constructs containing 7, 11 and 15 spacer sequences showed that our truncation variants had higher activity than the original TALENs when the spacer length was 11 or 15 bp (2- to 2.5-fold), consistent with the previous report using analogous AvrBs-based truncations (Mussolino et al. 2011) (Fig. 3C). Furthermore, we observed the efficiency of genome modification by these TALEN variants using the Cel-I nuclease assay (Fig. 3B). The HPRT1_B TALEN-NC pair showed higher activity compared with the original, thus confirming the usefulness of our deletion scaffolds.

Figure 3.

N- and C-terminal deletions of the TALE scaffold enhanced the activity of the HPRT1_B TALENs. (A) Schematic of TALENs harboring original scaffold and truncated scaffold. (B) Cel-I assay. Arrowheads indicate the expected positions of the digested products. (C) Relative TALEN activity in relation to spacer length measured by SSA assay. Relative activity is defined as the ratio of measured activity to the activity score of pSTL-ZFA36. Data are expressed as means ± SEM (n = 3).

Truncated PthXo1-based TALEN architecture achieves homozygous gene disruption in hiPSC

To assess the functionality of our TALEN-NC architecture, we made use of an HPRT disruption assay in human induced pluripotent stem cells (hiPSCs). The HPRT1 gene is X-linked and a component of the purine salvage pathway, permitting drug counter-selection (Caskey & Kruh 1979). Whereas HPRT1+/+ cells are 6-thioguanine sensitive (6-TGS) and resistant to hypoxanthine, aminopterin and thymidine (HATR), HPRT1−/− cells are 6-thioguanine resistant (6-TGR) and hypoxanthine, aminopterin and thymidine sensitive (HATS). The hiPSC line 201B7 is derived from female fibroblasts (Takahashi et al. 2007) and has been shown to have two active X-chromosomes (Tomoda et al. 2012); thus, TALEN-induced resistance to 6-TG would require disruption of both copies of the HPRT gene. Spontaneous inactivation of HPRT is extremely rare (Thomas & Capecchi 1987).

We electroporated the HPRT1_B TALEN or TALEN-NC expression vectors into female 201B7 hiPSCs. After 6-TG selection, cells electroporated with the original TALEN architecture were unable to form colonies, suggesting a homozygous disruption rate below the sensitivity level of this assay. Conversely, cells electroporated with the TALEN-NC pair consistently formed colonies (Fig. 4A,B). Under these conditions, electroporation of only the left or right TALEN gave rise only to rare false positives devoid of mutation in the PCR-screened target region (clone 201L2, Fig. 4C). All clones transfected with the HPRT1_B TALEN-NC pair were HATS, indicating true functional knockout events (Fig. 4C). Furthermore, sequence analysis of the TALEN target site showed that TALEN-induced 6-TG resistance is the result of compound heterozygous InDels spanning the spacer region (Fig. S1 in Supporting Information). These results confirm the utility of the engineered TALEN-NC scaffold in hiPSCs.

Figure 4.

HPRT1 homozygous knockout in female human iPSCs. (A) Number of 6-TGR colonies. Data are expressed as means. (B) Normal morphology of a 6-TGR HPRT−/− hiPSC colony. (C) Crystal violet staining of cultures after 4 days of 6-TG or HAT selection. Clones 201B7, 201L2 and 201LR3 display representative growth properties of parental, nontargeted and targeted cells, respectively.

TALEN-mediated disruption of exogenous eGFP transgenes in various animals

To elucidate the utility of our TALEN architecture for animal applications, we examined whether our uniquely constructed eGFP_B TALENs could disrupt integrated eGFP transgenes in flies, frogs and zebrafish. First, we tested eGFP_B TALEN-mediated gene knockout in the fly, Drosophila melanogaster. We used the protein trap line, P{PTT-GA}Jupiter[G00147] (Morin et al. 2001), bearing an actively expressing eGFP insertion at the Jupiter locus and a second eGFP insertion at a silent locus for TALEN injection. Among 79 embryos injected, 35 grew up to fertile adults (44%). Among 35 vials of fertile cross, seven eGFP progeny, with at least one allele disrupted, were obtained (20%, Table 1). Fourteen eGFP-negative progenies were balanced, and their eGFP coding regions sequenced (Fig. 5). At least one of the two eGFP insertions was disrupted, and double mutations were found in three cases. Most of eGFP mutations caused frame shifts and reduced viability of homozygous flies. Because the original protein trap line is homozygous viable, our eGFP knockout lines are thought to be new Jupiter mutants. Further phenotypic characterization is underway.

Table 1. Summary of TALEN-induced mutagenesis after microinjection of mRNAs into Drosophila melanogaster
eGFP+ onlyeGFP− progenyeGFP+eGFP−
Figure 5.

Sequences observed in eGFP knockout D. melanogaster lines. The wild-type sequence of eGFP is shown at the top with the TALEN-targeting sequences (capital letters). Deletions are indicated by dashes and insertions by capital letters with an underline. Asterisks indicate the same mutation types from independent lines.

Based on the success of invertebrate gene knockouts, we applied the same eGFP_B TALEN set to the vertebrates, Danio rerio (zebrafish) and Xenopus laevis (frog). In the case of zebrafish, we crossed Tg(flk1:mRFP) lines and Tg(fli1a:eGFP) lines, which express mRFP and eGFP in vascular endothelial cells, respectively. Subsequently, TALENs were injected at one- to four-cell stages of the double transgenic embryos. Small populations of endothelial cells judged by mRFP expression were completely suppressed in eGFP expression, suggesting the mosaic disruption of the eGFP transgene in the F0 progenitors (Fig. 6A–C). Consistent with this observation, eGFP gene modification was confirmed by Cel-I assay (Fig. 6D).

Figure 6.

Transcription activator–like effector nuclease-mediated eGFP disruption in vertebrates. (A–C) Fluorescence microscopy images of eGFP_B TALEN-injected zebrafish embryos. eGFP_B TALEN mRNAs (100 pg each) were injected at 1–4 cell stage zebrafish embryos obtained by reciprocal crossing of Tg(fli1a:eGFP) and Tg(flk1:mRFP). A merged image is shown in panel a, eGFP fluorescence is shown in panel b, and mRFP fluorescence is shown in panel (C). The eGFP expression in a small population of endothelial cells indicated by arrows was specifically suppressed, presumably due to the disruption of eGFP locus by the TALEN. Similar mosaic eGFP expression was observed in eight independent embryos. (D) Cel-I assay of Mock or TALEN-injected zebrafish embryos. Arrowheads indicate Cel-I-digested fragments. (E) Bright field and fluorescence microscopy images of eGFP_B TALEN-injected frog embryos. (F) Cel-I assay of TALEN-injected frog embryos. Arrowheads indicate Cel-I digested fragments. Right: A tadpole injected 600 pg right eGFP_B TALEN mRNA. Left and Right: A tadpole injected each 300 pg of left and right eGFP_B TALENs mRNA.

Meanwhile, much more dramatic TALEN effects were observed in frogs. We used in vitro fertilized eggs of ubiquitously expressing eGFP Tg(CMV:eGFP) lines for injection. Surprisingly, maternally transmitted eGFP expression of the TALEN-injected F0 progenies nearly vanished in most embryos (Fig. 6E). We obtained the similar results in Tg(CMV:eGFP) transgene disruption for paternal transmission (Fig. S2 in Supporting Information). Furthermore, the Cel-I assay showed that mutation rates of eGFP genes were extremely high compared with the results obtained using zebrafish (Fig. 6D,F). High-frequency eGFP gene mutation was also confirmed by DNA sequence analysis (Fig. S3 in Supporting Information).


The ‘Golden Gate TALEN and TAL Effector Kit’ from the Voytas laboratory implements a two-step assembly method to make custom TAL effector–based constructs, including TALEs and TALENs in yeast expression vectors (Cermak et al. 2011). This sophisticated assembly system is particularly appealing compared with other open-source TALEN construction kits available (www.addgene.org/TALEN). The ‘TALE Toolbox kit’, supplied from the Zhang laboratory, is based on a PCR-mediated rapid assembly system to make the constructs with a standard of 18.5 TALE repeats. Limitations with this kit are the potential for PCR-generated errors and a lower flexibility in repeat lengths, which in turn confounds sequence confirmation. ‘TAL Effector Engineering Reagents’, supplied from the Joung laboratory, are based on the REAL (Restriction Enzyme And Ligation) assembly method. This method can precisely assemble TALE repeats, but completing the construction is time-consuming. Recently, the Joung laboratory developed an automated assembly method, FLASH (fast ligation-based automatable solid-phase high-throughput) (Reyon et al. 2012). It is currently the fastest assembly method, but requires significant start-up costs sophisticated equipment and a huge amount of effort to maintain. The investment of time and resources required is suitable only for laboratories intending large-scale construction of TALENs.

Taking overall comparison into account, the Golden Gate kit is thought to be the best choice to make custom-made TALENs rapidly and precisely in a casual laboratory scale. However, there still remain some issues in the kit to be resolved, especially for animal applications. The first-step assembly method of the original kit requires ten-module ligation using Golden Gate cloning, which the robustness of the reaction is typically not high enough to be widely applied. In addition, the original evaluation system is yeast-based, which takes too much time and possesses lower reliability in the measured activity for the animal cells and embryos.

In this study, we have successfully developed and used simplified methods for TALEN construction and evaluation and showed the availability of our TALEN architecture. Based on the Golden Gate kit, we have minimized the required components for construction, while increasing efficacy. Using our protocol, the whole process, from construction to evaluation in mammalian cells, can be completed within 1 week. With this system, we succeeded in high-efficiency genome modification in various cells including HEK293T cells and hiPSCs. In addition, we showed the disruption of exogenous eGFP transgenes in flies, zebrafish and frogs using TALENs. We expect that our system will reduce barriers faced by entry-level TALEN application in cells and animals. Toward this end, the vectors described herein and additionally constructed CMV early enhancer/chicken beta actin (CAG) promoter-driven destination vector will be deposited in Addgene for public access, contributing globally to genome editing research.

Experimental procedures

Backbone vector construction

pFUS_A1A, A2A, A2B, A3A, A3B, A3C, A4A, A4B, A4C, A4D plasmids were constructed by inverse PCR using the primers listed in Table S1 in Supporting Information. For the construction of the final capture vector pcDNA-TAL, a SacI/BglII fragment from pTAL4 (Cermak et al. 2011) was inserted into pcDNA3.1 (Invitrogen, Carlsbad, CA, USA). Deletion constructs of pcDNA-TAL (pcDNA-TAL-C or pcDNA-TAL-NC) were constructed by inverse PCR using the primers listed in Table S1 in Supporting Information.

Construction of TALEN expression plasmids

The protocol for TALEN assembly follows the original report (Cermak et al. 2011) with some key modifications. The new 6-module backbone vectors for intermediary arrays are depicted in Fig. 1A. Repeat assembly was conducted using a Golden Gate reaction, transformed to XL1-Blue competent cells and screened precisely assembled clones by colony PCR using the pCR8_F1 and pCR8_R1 primers (Cermak et al. 2011). Constructed array plasmids and the appropriate last repeat are captured during a second Golden Gate reaction in the mammalian expression vectors, pcDNA-TAL, pcDNA-TAL-C or pcDNA-TAL-NC, described above. Final assembly, transformation and colony PCR screening for the second Golden Gate reaction followed the previous report (Cermak et al. 2011).

For consistency in analysis, TALEN plasmids used in this article were mainly designed to recapitulate those of the previous report (Cermak et al. 2011) with the exception of eGFP_B TALENs. Target sequences of eGFP_B TALENs are as follows: left 5′-CTTCAAGGACGACGGCAACT-3′ and right 5′-CGCCCTCGAACTTCACCT-3′.

Construction of SSA reporter plasmids

Transcription activator–like effector nuclease plasmids constructed using this modified protocol may be immediately evaluated using a single-strand annealing (SSA) assay. The pGL4-SSA reporter vector, containing inactive fragments of the luciferase gene that bear 700-bp regions of homologous overlap and are driven by a cytomegalovirus (CMV) immediate-early enhancer/promoter, was generated as described previously (Ochiai et al. 2010). For the addition of TALEN target sequences, sense and antisense oligonucleotides (Table S2 in Supporting Information) were annealed and inserted into BsaI sites between the bisected luciferase elements of pGL4-SSA.

pGL4-SSA-HPRT1_B-7 and pGL4-SSA-HPRT1_B-15 were constructed using restriction enzymes and T4 DNA polymerase (Takara Bio, Otsu, Japan). Plasmid pGL4-SSA-HPRT1_B-11 was digested with SacI or XhoI, and protruding ends of these DNA fragments were resected or filled in using T4 DNA polymerase. Polished fragments were then self-ligated to obtain pGL4-SSA-HPRT1_B-7 and pGL4-SSA-HPRT1_B-15, respectively.

SSA assay using HEK293T cells

The SSA assay was carried out as previously described (Ochiai et al. 2010) with slight modifications. HEK293T cells were grown in DMEM supplemented with 10% FCS. Fifty thousand cells were cotransfected with 200 ng of each of the TALEN expression plasmids, 100 ng of the pGL4-SSA reporter plasmid and 20 ng of the pRL-CMV reference vector (Promega, Madison, WI, USA) using Lipofectamine LTX (Invitrogen) in a 96-well plate. After 24 h, dual-luciferase assays were conducted using the Dual-Glo luciferase assay system (Promega) in a TriStar LB 941 plate reader (Berthold Technologies, Bad Wildbad, Germany) following the manufacturer's instructions. For the ZFN-positive control, a ZFN expression vector, pSTL-ZFA36, was constructed as previously described (Ochiai et al. 2010) and co-transfected with reporter plasmids as described herein.

Transfection and Cel-I assay

Transfection for the Cel-I assay was carried out as follows: the day before transfection, 150 000 HEK293T cells were plated to 35-mm dishes. The day of transfection, 1.5 μg of TALEN plasmids was transfected using Lipofectamine LTX (Invitrogen) according to manufacturer's instructions. The day after transfection, cells were moved to 60-mm dishes. At 72 h post-transfection, cells were collected and their genomic DNA isolated with the DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany). Genomic PCR was carried out with primers listed in Table S3 in Supporting Information. Amplified products were purified (PCR Purification Kit, Qiagen), and 400 ng of purified DNA was used for the SURVEYOR Mutation Detection Kit Cel-I assay (Transgenomic, Omaha, NE, USA) according to manufacturer's instructions. Products were analyzed by electrophoresis in 3% agarose gels and ethidium bromide staining.

6-Thioguanine (6-TG) screening of HPRT−/− human induced pluripotent stem cells (hiPSCs)

The female human iPS cell line 201B7 was derived from fibroblasts using retroviral reprogramming vectors (Takahashi et al. 2007). Maintenance of 201B7 and TALEN-targeted derivatives was carried out on SNL feeder cells in ReproCELL media as described.

HPRT1_B TALENs were transferred by standard cloning into a human iPS cell transient expression vector driven by the CAG promoter (Niwa et al. 1991). TALEN-mediated gene disruption was carried out using transient transfection by electroporation with the Neon system (Invitrogen). 1 × 106 viable 201B7 cells prepared by feeder depletion and single-cell dissociation with Accutase (Stemcell Technologies, Vancouver, Canada) were electroporated with 5 μg of each TALEN pair (1400 V, 20 ms, 1 pulse, 100 μL Neon tip). Cells were plated at a density of 15 000 cells/cm2 in hiPSC media containing Rho-Kinase inhibitor (Y-27632; Wako, Osaka, Japan). Two days after electroporation, the media were changed to 15 μm 6-TG (A-4660; Sigma, St. Louis, MO, USA)-containing media, and reduced to 5 μm between days 8 and 10. Compact colonies were counted and picked for expansion and genotyping in the absence of drug. Counter-selection with 1x HAT (H0262; Sigma) was carried out during standard cell passage.

Genomic DNA was isolated from 6-TGR iPS cell clones using the DNeasy Blood and Tissue Kit (Qiagen). Genomic PCR was carried out with forward and reverse HPRT1_B primers (Table S3 in Supporting Information). Products were analyzed by gel electrophoresis, TA-cloned into pGEM-T (Promega), and the resulting plasmids sequenced by standard dye-terminator chemistry and analyzed on an ABI 3130x. Sequences alignments to determine insertions and deletions were carried out with Sequencer (Genecodes, Ann Arbor, MI, USA).

eGFP-transgenic fly and zebrafish strains

For gene disruption in Drosophila, a strain containing a homozygous third chromosome protein trap insertion in the Jupiter gene (encoding a microtubule-binding protein) P{PTT-GA}Jupiter[G00147] (Morin et al. 2001), and P{Ubi-tagRFP.CAAX}TK1 (Kondo and Hayashi, submitted) inserted into the attP site at the cytogenetic location 68A4 (Groth et al. 2004), was used. During the course of this study, an additional nonfluorescent insertion of P{PTT-GA} was discovered in this strain.

To establish Tg(flk1:mRFP) fish, both a Tol2 transposon plasmid containing a 7-kb flk1 promoter element driving mRFP and Tol2 transposase mRNA were co-injected into the blastomere of one-cell stage embryos. After these embryos were raised to adulthood Tg(flk1:mRFP)ko09, which stably expresses mRFP in vascular endothelial cells was isolated. Another endothelial reporter transgenic fish Tg(fli1a:EGFP)y1 was also used (Lawson & Weinstein 2002).

mRNA synthesis, embryo microinjection and Cel-I assay

Constructed and evaluated eGFP_B TALEN plasmids were used for mRNA synthesis directly from the pcDNA-TAL vector. TALEN plasmids were linearized by restriction enzyme digestion and were used as templates in mRNA synthesis using the mMessage mMachine T7 Ultra Kit (Ambion, Austin, TX, USA) following the manufacturer's instructions. Microinjection into animal embryos was carried out as follows:

Fly: Dechorionated Drosophila embryos were injected with 250 μg/mL each of synthetic eGFP_B TALEN mRNA at their posterior ends. All of eclosed adults expressed normal levels of eGFP and tagRFP, and among 35 fertile crosses, seven yielded eGFP-, tagRFP+ progenies that were analyzed by PCR amplification and DNA sequencing.

Zebrafish: eGFP_B TALEN mRNA pairs (100 pg each) were injected at 1- to 4-cell stage zebrafish embryos obtained by reciprocal crossing of Tg(fli1a:EGFP)y1 and Tg(flk1:mRFP)ko09. The embryos, anesthetized with tricaine, were observed with PlanApo20x/0.75 objective (Nikon, Tokyo, Japan) at 33 h postfertilization. Optically sectioned images of intersegmental veins were obtained with a laser scanning confocal microscope A1-VAAS (Nikon). Maximum projection image of the volume data is presented. Genomic DNA was recovered from each embryo for Cel-I assay after microscopic examination. eGFP was amplified from the genome with KOD-FX-Neo polymerase (TOYOBO, Osaka, Japan) using the primers described in Table S3 in Supporting Information. The amplified fragment was used for Cel-I assays, and digested products were analyzed by electrophoresis in 1% Synergel (Diversified Biotech, Boston, MA, USA) and ethidium bromide staining.

Frog: In vitro fertilized eggs were obtained by mating CMV:eGFP-transgenic lines with wild type of X. laevis adults. Six hundred picograms of eGFP TALEN mRNA was injected into one-cell stage embryos. Injected embryos were reared to the swimming tadpole stage and then were observed through a fluorescence stereomicroscope. Cel-I assay was carried out as described above.


We thank Sumihare Noji and Keishi Osakabe for discussions. We also thank Akihiko Kashiwagi and Nobuaki Furuno for assistance in frog experiments. We appreciate Kikumi Horiguchi for assistance in TALEN construction and Hidehiko Kawai for construction of CAG promoter-driven destination vector. The authors wish to express their thanks to Dr Daniel Voytas for supplying the Golden Gate TALEN and TAL Effector Kit (Addgene, TALEN kit #1000000016). The authors also thank the Cryogenic Center of Hiroshima University for supplying liquid nitrogen. This study was supported by a Grant-in-Aid for Scientific Research on Innovative Areas (No. 2020006) to T.Y. and a Grant-in-Aid for JSPS Fellows (No. 12J06938) to T.S.