Transcription activator-like effector nuclease (TALEN)-mediated genome editing is highly specific due to the tightly stringent protein-nucleotide recognition principle and the obligatory heterodimers of the FokI domain, which makes it a powerful tool for targeted genome editing in diverse cell types and organisms. TALENs comprise a TALE DNA-binding region and a FokI restriction endonuclease region. DNA binding to TALE proteins is mediated by a central array of TALE repeat units, and each unit comprises a 33 to 35 amino acid motif specifying one DNA base pair. The repeats are highly identical except for the repeat variable di-residues (RVDs) at positions 12 and 13, which recognize the targeted DNA sequence. The RVDs, NI, NN, NG, and HD were used to specifically recognize nucleotides A, G, T, and C, respectively (Boch et al., 2009; Moscou and Bogdanove, 2009). However, the highly identical TALE repeat sequences make it challenging to assemble TALEs using conventional cloning approaches, and multiple repeats in one plasmid are easily catalyzed for homologous recombination in bacteria. The one-step TALEN assembly system is based on the principle of Golden Gate cloning (Engler et al., 2008; 2009). Type IIS enzymes, such as BsaI, cleave at sites beyond the recognition site. We designed BsaI recognition sites at the 5′ and 3′ ends of every TALE module DNA sequence in the inverse orientation, and these sequences will be removed during the enzyme digestion reaction, leaving sticky overhangs at both sites. Adjacent TALE modules sharing compatible sticky overhangs, termed fusion sites, can be ligated. Thus, this system simplifies the assembly of TALENs.

TALEN design and backbone vector selection

Online tools, such as the TAL Effector Nucleotide Targeter 2.0 (https://talent.cac.cornell.edu/), can be used for TALEN target design, followed by NCBI BLAST to eliminate off-target risks on the specific genome. Binding sites for either the left or right arm range from 12 to 19 bp, with a spacer 12 to 21 bp sequence in the middle, which facilitates FokI dimerization and cleavage. The nucleotide at 5-terminal ‘0’ position of the targeting sequence should be a T, and the last nucleotide is flexible. For knock-outs based on frame-shift induced non-homologous end-joining (NHEJ), a targeting site designed in the first coding exon is recommended. Target sequence verification to eliminate polymorphisms is highly recommended. In order to obtain one TALEN pair with high efficiency, we recommend designing TALEN pair combination 2 × 3 (or 2 × 2) at one site for one gene.

Once binding sites are selected, appropriate TALEN backbone vectors can be selected. An array of TALEN backbone vectors (left arm and right arm) are designed for the TALEN assembly for gene modification in various cell lines and species, containing EF1α, CMV, T7/SP6 or Ubiquitin/35S promoter, respectively. All vectors contain basic structures for TALEN expression, including 3× Flag, NLS, N-terminal, C-terminal, FokI coding sequence and Kanamycin resistant gene. FokI coding sequence on left and right arms are heterologous, and the protein products will undergo dimerization to function as an active endonuclease. A nucleotide specific half monomer is predesigned onto each backbone vector, which is 1/2A, 1/2T, 1/2C, and 1/2G at the C-terminal end, respectively, covering all potential TALEN-binding sites. Most left arm vectors (except plant vectors) contain IRES and puromycin resistance gene, however puromycin-free left arm vectors are more suitable for TALEN vectors in vivo transcription. Some right arm TALEN vector contains EGFP which allows the observation of successful transfection.

Cell culture and transfection

HEK293T cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Invitrogen) supplemented with 10% fetal bovine serum (FBS, Gibco). TALEN vector pairs and the EGFP-IRES-PURO vector were transfected into 293T cells using FuGENE HD transfection reagent (Promega). Fluorescent signals were observed, and 1 μg/ml puromycin was applied for successful transfection selection.

TALEN efficiency detection

TALENs were assembled using the FastTALE TALEN Assembly kit, please refer to Supplemental Experimental Procedures for more detailed description of these procedures. In brief, divide the TALEN recognizing sequence into 9 pieces, excluding the last 3′ nucleotide. Each of the 9 pieces was assigned with a number starting from 1 to 9 with a 5′ to 3′directionality. A master mix of 9 modules, Left/Right arm TALEN backbone vector, BsaI with T4 DNA ligase was prepared, followed with a ligation program in a thermal cycler. Upon completion of the ligation, Plasmid-Safe™ ATP-Dependent DNase was added to selectively remove residual linear DNA fragments. The vectors were transformed into competent cells, colonies were screened and confirmed using enzyme digestion and DNA sequence analysis.

At three days post transfection, the cells were harvested for genome isolation. Raw efficiency was read from multiple peaks of the PCR sequencing of mixed cells; accurate efficiency was calculated after sequencing single clonal-competent cells, followed by ligation of the PCR products into the T-vector and transformation.

Encouraged by the successful reprogramming of fibroblast to induced pluripotent stem cells (iPS cells) by the simultaneous introduction of 24 factors (Takahashi and Yamanaka, 2006), we attempted to put all the modules and one TALEN backbone vector together, to simplify TALEN assembly into a one-step process. We assembled 18 TALE repeats by ligating 9 dimers using Golden Gate cloning, which has never been reported previously.

This one-step assembly method employs a library of a total 180 purified DNA fragments, or TALE modules, in either monomer or dimer forms (Fig. 1A). The 180 TALE modules are divided into 9 positions including 16 dimers and 4 monomers at each position. Each position is determined based on its specific sticky ends, ensuring that all modules are properly ligated (Fig. 1A). Divide the TALEN recognizing sequence into 9 pieces, excluding the last 3′ nucleotide, which will be encoded by the expression vector. Each of the 9 pieces was assigned with a number starting from 1 to 9 with a 5′ to 3′ directionality. For example, to target sequence ATTCTGCTAACTCATAT, after removing the last ‘T’ the rest are numbered as A1, T2, TC3, TG4, CT5, AA6, CT7, CA8, and TA9. The CT5 module comprises an HD-containing TALE repeat and an NG-containing TALE repeat that recognizes a CT with unique 5′ and 3′ ends located at the 5^th position. An array of TALE monomers/dimers can be assembled by selecting 9 modules from the 1^st to the 9^th position, followed by ligation into the circular TALEN backbone vectors to target a specific genomic locus. Each circular TALEN backbone vector contains a heterodimeric FokI domain and can be selected to express a TALEN protein in specific cell lines and species, based on differences between the promoter and selection marker (Fig. 1A). To investigate the efficiency of this TALEN assembly, five TALENs forming six TALEN pairs targeting one locus of the human gene PDX1 were designed and constructed, followed by analysis of the assembly rates (Fig. 1B).

After assembly using this system, the ligation products were transformed into E. coli, and five colonies were picked for each TALEN. The assembly rate was assessed by enzyme digestion (Fig. 1C) and confirmed using DNA sequencing. The rate of TALEN assembly ranged from 40% to 80%, with an average of 56.0% (Fig. 1D). Every TALEN pair was transfected into HEK293T cells, and the cells were harvested for genomic DNA preparation at 5 days post-transfection. The amplified PCR products of the targeted loci were sequenced to detect mutations. A multiple-peak sequencing pattern indicates a targeted mutation (Fig. 1E). PCR products containing mutations were ligated into T vectors, and thirty single colonies were sequenced. Various types of mutations in the target genomic region were observed. One selected pair (PDX1-1L1+PDX1-1R1) targeting PDX1 led to 17/30 mutation types (Fig. 1E). The same analysis was performed for other loci in a second locus of PDX1 (Fig. 2A), human NESTIN (NES) (Fig. 2B), and ALBUMIN (ALB) (Fig. 2C). The overall TALEN cleavage efficiency of all four validated pairs ranged from 10.0% to 56.7% (Fig. 1F).

This system largely simplifies the assembly of TALENs by shortening the manual operation time of TALEN assembly into less than half an hour, followed by verification within three days. This strategy has three major advantages: (A) Ease. This one-step assembly system requires only 9 modules and one circular TALEN backbone vector to complete assembly, avoiding the traditional steps of digestion, gel extraction and ligation. (B) Speed. One-step assembly facilitates the assembly of functional TALEN vectors within one day, followed by verification steps, such as enzyme digestion or DNA sequencing within three days. (C) Flexibility. High flexibility in selecting a TALE target sequence is endowed by the combinatorial use of dimers and monomers from the library. Considering a half monomer module, which recognizes a single nucleotide A, T, C or G on the TALEN backbone vector, this system facilitates the assembly of TALE repeats from a minimum of 9.5 to a maximum of 18.5 nucleotides in one-step cloning.

Because TALE was combined with FokI to function as genome scissors, the methods for TALEN assembly have been continuously improved based on the Golden Gate cloning method (Cermak et al., 2011; Ding et al., 2013; Kim et al., 2013; Li et al., 2011; 2012; Morbitzer et al., 2011; Weber et al., 2011; Zhang et al., 2011; 2013). Several one-step assembly methods based on trimer or tetramer module libraries have been reported (Ding et al., 2013; Kim et al., 2013). To confirm the potential and flexibility of the TALEN design, the sizes of the module libraries are inevitably large, containing 424 and 832 modules, respectively. This results in operating complexity in selecting modules and maintaining libraries of such sizes. The two methods based on trimer or tetramer module libraries can be used only to build TALENs with a fixed number of recognition repeats, markedly diminishing the flexibility of selecting recognition sequences.

Other methods based on monomer or dimer recognition modules have required tedious serial operation steps and/or troublesome multiple enzyme utilization, increasing operational inconvenience. Other methods can be used only to build TALENs with a fixed number of recognition repeats, e.g., for 13 bp or 17 bp nucleotides (Li et al., 2011; 2012).

To our knowledge, this study is the first to ligate 9 modules and one circular TALEN backbone vector in one step, generating 9.5 to 18.5 repeat sequences with an overall assembly rate higher than 50%. Compared with the above methods, this system based on a monomer/dimer library at 9 positions ensures a smaller library size, renders higher designing flexibility and maintains high targeting specificity.

Although we focused on TALENs assembly in the present study, this TALE assembly system can also be used to fuse the engineered TALE repeat arrays with other functional proteins, including catalytic hydrolase, to generate target chimeric proteins.

The CRISPR/Cas9 system was developed as a genome editing tool in 2013. As the construction of CRISPR/Cas9 guiding RNA is simple and former TALEN assembly is complicated, the CRISPR/Cas9 genome editing tool has been widely used despite its severe off-target effect due to a 10?12 bp core recognition sequence. The off-target issues are still a concern in clinical application. Recently, many efforts are made to improve CRISPR/Cas9 target specificity and to decrease off-target effect. Here, we developed an assembly system that enables easier TALEN assembly. This assembly system makes TALEN-mediated genome editing a more convenient and promising tool which is comparable to CRISPR/Cas9 genome editing.