Rapid Assembly of Customized TALENs into Multiple Delivery Systems

Transcriptional activator-like effector nucleases (TALENs) have become a powerful tool for genome editing. Here we present an efficient TALEN assembly approach in which TALENs are assembled by direct Golden Gate ligation into Gateway® Entry vectors from a repeat variable di-residue (RVD) plasmid array. We constructed TALEN pairs targeted to mouse Ddx3 subfamily genes, and demonstrated that our modified TALEN assembly approach efficiently generates accurate TALEN moieties that effectively introduce mutations into target genes. We generated “user friendly” TALEN Entry vectors containing TALEN expression cassettes with fluorescent reporter genes that can be efficiently transferred via Gateway (LR) recombination into different delivery systems. We demonstrated that the TALEN Entry vectors can be easily transferred to an adenoviral delivery system to expand application to cells that are difficult to transfect. Since TALENs work in pairs, we also generated a TALEN Entry vector set that combines a TALEN pair into one PiggyBac transposon-based destination vector. The approach described here can also be modified for construction of TALE transcriptional activators, repressors or other functional domains.


Introduction
Generation of genetically modified animal models, as well as creation of mutations and correction of mutants in human cells, have provided many important tools for investigating gene function, genetic disease and drug development [1,2]. Modification of a specific gene/locus in mammalian cells has been a time-consuming task, because it usually involves homologous recombination, which happens at a low rate in nature. However, the efficiency of genetic modification at the target locus could be increased hundreds to thousands fold by creation of a site-specific double-stand break (DSB) through designing "customized" nucleases. Repair of a DSB can occur by two potential pathways: non-homologous end-joining (NHEJ) or homologous recombination (HR). Currently, there are at least four different customized nuclease systems have been developed (see below), and assembly of customized nucleases systems is becoming more efficient [3][4][5][6].
Zinc-finger nucleases (ZFNs), which emerged a decade ago [7], are artificial nucleases consisting of engineered sequence-specific Cys2His2 zinc-finger DNA-binding domains (3 or more zinc fingers) and a FokI endnucleases cleavage domain. Although tremendous progress has been made to improve the specificity and affinity of zinc finger domains that recognize desired DNA sequences, the target site overlap and crosstalk between individual fingers make it complicated to design and select sequence-specific ZFNs [7], which has impeded researcher access. Homing endonucleases, such as LAGLIDADG family members, have also been engineered for genome editing. Homing endonucleases directly involve DNA sequence recognition and cutting process as well. The drawback is that the production of homing endonucleases targeted to specific sequences appeared to be more complicated [8][9][10][11].
The CRISPR/Cas system is a defense system evolved in bacteria and archaea that use short RNA to directly degrade foreign nucleic acids. Recently, the S. pyogenes type II CRISP/Cas 9 based system has been developed for genome editing in mammalian cells. In this system, the mature crRNA base-paired to trans-activating trans-crRNA that directs the Cas protein to introduce DSBs in target DNA [44][45][46][47]. Designing CRISP/Cas 9 system to target a specific sequence is simpler than the TALEN system. However, a relative largescale analysis demonstrated that it shows significant off-target activity which may limit its application, especially for cell therapy [48].
Several different scaffolds and protocols for assembly of TALE RVDs have been reported [15][16][17]19,20,32]. In this study, based on the TALEN scaffold of Zhang's group [15], we constructed a set of RVDs and TALEN entry vectors which allow us to efficiently and accurately assemble customized TALENs. Furthermore, through Gateway ® LR recombination, the assembled TALEN can be easily transferred into an adenoviral vector system, which facilitates delivery of TALEN into cells. We also demonstrated that a pair of TALENs can be merged into a single plasmid, which is convenient for transfection or injection. Using these modified tools, we generated three pairs of TALENs which target the Ddx3 subfamily of DEAD-box RNA helicase genes in mouse, Ddx3x, Ddx3y and D1Pas1(Pl10) and validated their function by Surveyor mutation analysis and sequencing of target genes. The TALEN assembly system described here has the flexibility for application to other TALE related vectors.

Plasmid construction
A plasmid array encoding TAL repeat variable diresidues (RVDs). By using the plasmids encoding four RVDs (NI, NN, HD or NG for recognition of A, G, C and T nucleotides, respectively), a total of forty TAL RVDs were amplified by using the primers as reported previously [15]. However, twelve RVDs  of the first position of hexamers were re-amplified through  primer  Ex-F:  5'-GACAGATCT  CGTCTC  ATGGCCAACCTTAAACCGGCCAACATAC-3' and reverse  primer  In-R1:  5'-TCTTATCGGTGCTTCGTTCTCGTCTCCCGTAAGTCCGTGC  GCTTGGACAC-3'. Thus, six RVDs can be ligated directly into  the intermediate plasmid pTemp-S (see below), rather than the original RVDs, which were ligated to form circular hexamers [15]. All of these RVDs fragments were then cloned into pGEM-T vector (Promega, A3600 ) by TA cloning and sequenced. The resulting RVD plasmid array will be made available for other researchers upon request.

Assembly of TAL effector RVDs into TALEN expression Entry vectors
We designed TALENs targeted to mouse Ddx3 gene family members, Ddx3x, Ddx3y, and D1Pas1. Ddx3x and Ddx3y are located on the X and Y chromosomes, respectively. D1Pas1 is an intronless retrotransposed copy of the Ddx3x gene and is located on chromosome 1. We designed TALEN pairs to target sequences such as the translation start site or intron/exon junctions that are most likely to disrupt gene expression. In addition, we targeted non-coding sequences, which are distinct among family members. For Ddx3x, the left TALEN target sequences are in the 5' untranslationed region (5'UTR) of exon 1, and the right TALEN targets 5'UTR close to the coding sequence. For Ddx3y, the left TALEN target sequences are in intron 4, the right TALEN target sequences are in exon 5, and the intervening sequence consists of part of intron 4 and exon 5. For the intronless D1Pas1, the TALEN targets were chosen from the sequences corresponding to exon 1 (left TALEN) and exon 2 (right TALEN) of Ddx3x. This combination is unlikely to mutate Ddx3x or Ddx3y due to the intervening intron sequence.
For a pair of TALENs, each TALEN targets 20 base pairs of sequences starting from "T" with a space of ~14-20 bases in between. The middle 18 bases of each target sequence (the RVD sequence) was divided into three hexamers that were ligated into pTemp-S [15] . The RVDs were assembled into TALEN Entry basic vectors through two continuous golden gate ligation [49]. The detailed protocol is in Supplementary Methods (Protocol S1 A). Half (~7 µl) of each final ligation products were used to transform chemical competent DH 5α cells. Six colonies were inoculated from each ligation. Plasmid DNAs were extracted and screened by SalI digestion, then  sequencing using primer TALE-Seq-F1, TALE-Seq-F2 and TALE-Seq-R1 [15].

Transfer of assembled TALENs to the Destination Vectors
To transfer the assembled TALENs into adenoviral destination vectors, 150 ng of TALEN Entry vector and 150 ng destination vector pAd/PL-DES), 2 µl Gateway ® LR Clonase ® II Enzyme Mix (Life Technologies, 11791-020) and TE buffer (pH 8.0) were added up to 10 µl. After incubation at room temperature for 2 hours, 4 µl of the product were directly added into 50 µl DH5α competent cells for transformation.
For the "2 in 1" system, two TALEN Entry vectors (150 ng each), the destination vector pPB-DEST or pPB-DEST-Puro (150 ng), 2 µl Gateway ® LR Clonase ® II Plus Enzyme Mix (Life Technologies, 12538-120) and TE buffer (pH 8.0) were added up to 10 µl . After incubation at room temperature for 4 hours to overnight, 4 µl of the product were directly added into 50 µl DH5α competent cells for transformation. The restriction digestion maps of all plasmids are presented in Protocol S1 B.

Production of Adenoviral vectors
Adenoviral vectors were produced in 293A cells according to the instructions of ViraPower TM Adenoviral Expression System (Life Technologies, K4940-00). To determine the titers, the viruses were serially diluted and added to 12-well plates preseeded with HEK 293 cells. Seventy hours after transduction, EGFP or mCherry expressing cells or colonies of cells were counted to calculate plaque-forming units or transducing units.

TAL effector RVD plasmid array
To generate TALEN constructs targeted to genes in mammalian cells, we adopted the TALEN toolbox from Zhang's group [15] with modifications described here. The assembly of TALENs by PCR in the original strategy is straightforward, but we frequently observed mutations or deletions that had been introduced into TALENs, probably because tandem repeats of the RVD sequences caused PCR errors. Therefore, we cloned the TAL effector RVD in into pGEM-T vector to generate a RVD array of forty plasmids for hexamer construction, and one intermediate plasmid, pTemp-S. For TALEN assembly, while position 2 to 5 of all hexamers share the same plasmids encoding RVDs, NI (A), HD (C), NG (T), or NN (G) respectively, different plasmids should be picked for the first and last RVDs for different hexamers ( Figure S1, Top ) to facilitate ligation of RVD sequences to generate each hexamer. The same intermediate plasmid pTemp-S is used for all hexamer assembly. Generally, it is not necessary to transform hexamer ligation products and screen the plasmids. However, this is an option if the TALEN assembly by one-step transformation fails ( Figure S1, Bottom).

TALEN Entry vectors
The Gateway ® technology (Life technologies), which is based on the bacteriophage lambda site-specific recombination system, is very efficient and specific for molecular cloning. In this system, the sequences in the first plasmid (Entry vectors) can be transferred to the second plasmid (Destination vector) through attL/attR (LR) recombination [50,51], which is suitable for choosing different TALEN delivery systems in the future.
We generated two sets of four TALEN Entry vectors containing either EGFP or mCherry reporters, which are driven by human EF1α promoter (Table 1). T7 RNA polymerase promoter following EF1α promoter in these vectors can be used for in vitro transcription to produce mRNAs.
We then used the new TALE RVD plasmid array and TALEN Entry vectors to construct TALEN pairs to target mouse Ddx3 gene family members (Ddx3x, Ddx3y and D1Pas1). Each TALEN binding sequence and RVDs are shown on Figure S2. Using a two continuous golden gate ligation approach ( Figure  1A, Protocol S1), we routinely obtained 4 to 6 out of 6 randomly picked colonies with the correct sequences. Sequence analyses did not reveal any mutations if the plasmid had the correct restriction enzyme patterns. Except of those RVDs which were synthesized chemically and ligated into TALEN expression vectors [17,19,34], the comparison of our strategy to others is provided in Table S1. We transfected TALEN Entry plasmids targeted to Ddx3x and D1Pas1 into mouse Neuro 2a cells. The transfection efficiency and TALEN expression were easily monitored via expression of fluorescent reporter genes, which were encoded from the same transcript as the TALENs with intervening 2A sequences (Figure 1B-D). We favored this approach over the approach of co-transfecting with a separate reporter plasmid [26,29]. Furthermore, if the donor DNAs containing either green or red fluorescent reporter are used for homologous recombination, EGFP or mCherry in these customized TALEN vectors can be simply removed by AscI digestion and re-ligation. Surveyor ® mutation analysis was performed to validate the activity of each TALEN pair. As shown in Figure 1E and F, any single TALEN (left or right) vector targeting Ddx3x or D1Pas1 did not introduced mutations around the target sequences. However, Co-transfection of TALEN pairs resulted in mutagenesis that was revealed by Surveyor mutation analysis.

Transfer of customized TALENs to adenoviral vector
Transfer of TALENs to non-integrating adenoviral vectors would theoretically significantly enhance the efficiency of TALEN delivery and broaden the potential applications of TALEN technology. Therefore, we recombined TALEN pairs, which targeted either Ddx3x or Ddx3y respectively, from their Entry vectors into adenoviral destination vectors pAD/PL-DEST (Figure 2A). The TALEN Entry vectors can be easily transferred to adenoviral vectors by Gateway ® LR recombination.
We then generated TALEN adenoviruses for Ddx3x and Ddx3y and tested them in either Neuro-2a cells or mouse fibroblasts. Based on reporter gene expression shown in Figure  2B-D, the efficiency of Ddx3x TALEN adenovirus transduction into Neuro-2a cells was qualitatively greater than that of sequential (2x) TALEN plasmid transfections (Compare Figure  1B-D and Figure 2B-D). Surveyor ® mutation analysis demonstrated that adenovirus transduction with single TALENs did not cause mutations at their target region, while adenovirus transduction with TALEN pairs did ( Figure 2H). The ratio of cut to non-cut DNA from adenovirus-TALEN transduced samples was approximately 50% ( Figure 2H). Thus, higher efficiency of adenoviral vector transduction resulted in higher genomic mutagenesis (compare Figure 2H and Figure 1E). Similar results were obtained for Ddx3y TALEN adenovirus pairs, which were tested in mouse male fibroblasts ( Figure 2F and I).
To show the tropism of virus and ubiquity of human EF1α promoter, the adenoviruses were also transduced into mouse G4 ES cells and rat C6 glioma cells ( Figure 2E and G). During the preparation of this manuscript, we note that Holkers and coworkers used a different plasmid system to introduce TALEN sequences in adenoviral vectors [52]. However, their cloning strategy required several extra steps (see Discussion).

Sequence analysis of Ddx3 genes after treatment with plasmid or adenoviral TALENs
We cloned the PCR products amplified at the target regions of Ddx3x and D1Pas1 from Neuro-2a cells which were transfected with TALEN plasmid pairs ( Figure 3A), and Ddx3y from male mouse fibroblasts transduced with adenoviral TALEN pairs ( Figure 3B). We confirmed that the cells treated with customized TALENs generated mutations around their target sequences (Figure 3). The largest deletion detected in Ddx3y was 193 bp (not shown).
Assembly of customized TALEN pair into one vector-"2 in 1". Because of the feature of FokI, TALENs must work as a pair. We believe it should be beneficial for plasmid transfection and DNA injection to engineer a customized TALEN pair into one plasmid. We thought it may be feasible to clone both left and right TALENs into one destination vector via the Multisite Gateway ® Pro system [53] (Life technologies). To enable this approach, different att sites must be chosen for recombination of two or more fragments into the Entry vectors.
We generated 12 TALEN Entry constructs. Four vectors, which have attL1 and attR5 flanking sequences and EGFP reporter for one side TALEN (e.g.,left TALEN) assembly, and the other eight vectors, which hold attL5 and attL2 flanking sequences and mCherry reporter for the other side TALEN (e.g., right TALEN) assembly ( Table 2). The assembly of RVDs into these Entry plasmids followed the same procedure described above. We also constructed two PiggyBac  transposon-based destination vectors, pPB-DEST and pPB-DEST-Puro ( Figure 4A). This system can be readily modified for other destination vector systems. As a proof of concept, we generated a pair of Entry TALEN vectors targeting exon 1 of the Ddx3x gene using same RVDs described above. We then recombined into the pPB-DEST and pPB-DEST-Puro destination vectors. The two fragment recombination was effective, despite the fact that thirty-six RVDs (eighteen RVDs for each side TALEN) were present in the sequences. It should be pointed out that it is impossible to sequence the RVDs in the final vectors. Nonetheless, we transfected each TALEN Entry plasmid ( Figure 4B and C) and the TALEN "2 in 1" plasmid ( Figure 4D) into Neuro-2a cells. Surveyor ® mutation analysis demonstrated that the TALEN pair assembled in one plasmid ("2 in 1") was capable of introducing mutations in genome ( Figure 4E).
Our TALEN assembly approach exploits the features of the methodology originally described by Sanjana and colleagues [15]. However, we assembled TALENs using a sequence validated array of RVD plasmid encoding repeat monomers, which eliminated possible errors introduced by PCR. We used these basic materials to assemble TALENs targeting three genes of Ddx3 gene family, which are located on chromosome 1 (D1Pas1), the X chromosome (Ddx3x) and the Y chromosome (Ddx3y) and did not observe any mutations in our assembled TALEN constructs. The system described here saves the time and labor of amplification and purification of PCR products, relative to PCR only [15,20,24,54] or PCRplasmid mixed assembly [16] approaches. In addition, our approach eliminated an extra step of transformation compared to other plasmid-based assembly approaches [18,29,38,62].
The TALEN Entry vectors generated in this study (Table 1) contain fluorescent reporters (EGFP or mCherry), which help to evaluate the transfection efficiency and can facilitate tracking and sorting of cells containing the TALEN moiety. Furthermore, the new TALEN Entry vector sets are in a Gateway ® Entry vector-based system, which can be easily transferred to other delivery system, such as viral or transposon vectors through in vitro LR recombination. Indeed, we successfully transferred TALEN-reporter expression units into adenoviral plasmids without losing any TALEN expression unit sequences. Adenoviral vectors efficiently transduce dividing or non-dividing mammalian cells and can be produced at high titers. We demonstrated in the current study that adenoviral TALENs can effectively modify target gene sequences. During preparation of this manuscript, Holkers and co-workers reported similar success using adenoviruses to introduce TALEN sequences into HeLa cells with a different strategy [52]. To generate adenoviral TALEN plasmid in that study, the authors first assembled the TALEN sequence into an expression vector, then isolated the TALEN sequences and inserted them into a shuttle vector, followed by HR in Escherichia coli BJ5183 pAdEasy-1 to get adenoviral TALEN vectors [52]. In the current study, we report a relatively simple strategy in which adenoviral TALEN vectors are assembled directly from TALEN Entry plasmids in the test tube via LR recombination. The fact that adenoviral vectors rarely integrate in the genome may be advantageous in some experimental circumstances. For some studies, the adenovirus-TALEN approach described here may provide a viable alternative to lentiviral vector-based small RNA knockdown systems that are currently widely used.
We also built a set of TALEN Entry plasmids with different att flanking sequences, which can be used to join TALEN pairs into PiggyBac transposon-based destination vectors (2 in 1 system). The targeting efficiency of one plasmid transfection should theoretically be higher than that of two plasmids. The individual Entry TALEN vectors for the "2 in 1" system can also be recombined to pAd/PL-DEST adenoviral vectors. However, because of adenovirus genome size limitations, the new generation of gutless adenoviral vectors [63] may be required to assemble the "2 in 1" TALEN system in adenoviruses. Cloning of TALEN expression units into PiggyBac vectors could also increase transfection efficiency if co-transfected with transposase. This system may not be limited for TALEN assembly. For example, by replacement of the FokI C-terminal domain in TALEN vectors with transcriptional activator or repressor domains, and generation of new multisite Gateway ® Entry plasmids, we could regulate 2 to 4 genes by using a single plasmid because the PiggyBac transposon system has been shown to deliver cargo sizes larger than 200 kb [64][65][66][67]. Furthermore, integrated TALE PiggyBac transposons can be removed from the genome by transient expression of transposase without leaving a "footprint" [68,69].
Future advances in the TALEN field will improve specificity, which is important for cell therapy. This could be done, for example, by adding new Asn-His (NH) or Asn-Lys (NK) RVDs, which have high biological activity and specificity for G [43,57] into the array or by use of mutated FokI C-terminal domain DD and RR in TALEN Entry vectors to reduce off-target nuclease activity [68]. In this work, we developed a robust and accurate TALEN assembly system and "user friendly" TALEN Entry vectors that can be modified by basic molecular biology methods to take advantage of new developments in the TALEN field. Figure S1. Repeat variable di-residue (RVD) plasmid library and hexamer assembly. Upper: RVDs library plasmids for TALEN assembly. Lower: example of assembly of 3 hexamers for the left TALEN of exon 1 of Ddx3x. (PDF) Figure S2. TALENs' target sequences of Ddx3x, Ddx3y and D1Pas1. The TALEN pair for Ddx3x targets the 5' untranslated region. The TALEN pair for Ddx3y targets intron 4 (left TALEN, red) and exon 5 (Right TALEN, blue). The TALEN pair of the D1Pas1 (retrotransposed autosomal copy of Ddx3x) targets sequences corresponding to exon 1 (for left TALEN, red) and exon 2 (for right TALEN, blue) of Ddx3x. The left and right TALEN target sequences for each gene are indicated with bold underlines. The assembled RVDs (with encoded nucleotides) for the left and right TALENs are below each target gene region. RVDs (NI, NG, NN and HD) are color coded in correspondence with the RVD library in Figure S2. Exon sequences are indicated with bold font. (PDF) Protocol S1. Protocol. (PDF )   Table S1.

Supporting Information
Comparison of the Strategy of TALEN Assembly.

Author Contributions
Conceived and designed the experiments: YS. Performed the experiments: ZZ SZ XH YS. Analyzed the data: ZZ YS KEO. Contributed reagents/materials/analysis tools: ZZ XH SZ YS KEO. Wrote the manuscript: YS KEO.