Engineering and Application of Zinc Finger Proteins and TALEs for Biomedical Research

Abstract

Engineered DNA-binding domains provide a powerful technology for numerous biomedical studies due to their ability to recognize specific DNA sequences. Zinc fingers (ZF) are one of the most common DNA-binding domains and have been extensively studied for a variety of applications, such as gene regulation, genome engineering and diagnostics. Another novel DNA-binding domain known as a transcriptional activator-like effector (TALE) has been more recently discovered, which has a previously undescribed DNA-binding mode. Due to their modular architecture and flexibility, TALEs have been rapidly developed into artificial gene targeting reagents. Here, we describe the methods used to design these DNA-binding proteins and their key applications in biomedical research.

INTRODUCTION

The Cys2-His2 (C2H2) domain is the most common type of DNA-binding motif found in eukaryotes. The C2H2 ZF domain contains multiple cysteine and histidine residues, which are the most common ligands for the zinc ion in proteins since they use zinc coordination to stabilize their folds (Segal and Meckler, 2013). The DNA-binding activity of ZF domains has been extensively studied, and a number of studies have been conducted to create ZF proteins (ZFPs) that recognize any desired DNA sequence to provide useful new tools for numerous biomedical research applications such as gene regulation, genome engineering and diagnostics.

A new class of DNA binding domain has been recently discovered, which is called a transcriptional activator-like effector (TALE). The recent discovery of TALEs has enabled many scientists to exploit an alternative platform for engineering DNA-binding proteins. TALEs are naturally secreted proteins from plant pathogenic bacteria of the genus Xanthomonas. They contain modular DNA-binding domains composed of 33–35 amino acid repeat arrays, where each repeat domain specifies a single DNA base. TALEs have been deployed in DNA targeting for many applications, such as genome engineering and gene regulation (Miller et al., 2011). Numerous studies have shown considerable progress in understanding DNA recognition of ZFPs and TALEs and in developing design methods that could open up more widespread applications for a number of biomedical research opportunities.

ZINC FINGER PROTEINS – STRUCTURE AND DNA RECOGNITION

The C2H2 ZF domain is the most common type of ZF and is one of the most abundantly expressed proteins in eukaryotic cells. ZFs are small, functional and independently folded domains coordinated with zinc molecules in their structure. The C2H2 ZF folds into a compact ββα structure, which is stabilized by zinc coordination and by the conserved hydrophobic core. This ββα framework provides an insight into how ZFPs interact with DNA. Twenty-five of the thirty amino acids in the repeat folds around the zinc to form a ‘finger’ and the rest of the five amino acids (TGEK(R)P) provide a short consensus linker between consecutive fingers (Moore et al., 2001). The zinc ion is tetrahedrally coordinated between two cysteine and two histidine residues, which stabilizes the fingers.

Amino acids in each ZF have affinity towards specific nucleotides, causing each finger to selectively recognize 3–4 nucleotides of DNA. Multiple ZFs can be arranged into a tandem array and recognize a set of nucleotides on the DNA. It is possible to create modules of six ZFs that can potentially recognize 18 bp of DNA, which would be sufficient enough to recognize a unique DNA sequence in the human genome. The α-helix of each finger fits into the major groove of the DNA, causing the protein to wrap around the DNA, as shown in the crystal structure of Aart (Fig. 1) (Segal et al., 2006). Three amino acid residues at positions −1, 3, and 6 on the α-helix make contacts with the 3′, middle, and 5′ nucleotides, respectively. In addition, amino acids at positions −2, 1, and 5 can make direct or water-mediated contacts to the phosphate backbone of the DNA (Segal et al., 2003). The amino acid at position 2 is also involved in the contact with other helix residues. Different ZFPs of various lengths can be generated, which allow for recognition of almost any desired DNA sequence out of the possible 64 triplet subsites (Dreier et al., 2001).

Fig. 1.

DNA bound structure of a zinc finger protein (ZFP).

ZFP Aart (PDB ID: 2I13) is flanking the major groove of DNA, making contact with the edge of the nucleotide bases. Aart is a designed six finger protein with pentapeptide linkers, which recognizes an A-rich 18 bp sequence (Segal et al., 2006). The DNA is slightly unwound in order to allow 11.3 bp/turn.

ENGINEERING OF ZINC FINGER PROTEINS (ZFPs)

Modular assembly

Multiple approaches have been taken to identify optimized individual ZF modules to recognize one of the 64 possible 3 bp DNA subsites using a combination of rational design and selection (Beerli and Barbas, 2002; Dreier et al., 2000; 2005; Segal et al., 1999; Wu et al., 1995). Phage display libraries were constructed and selected, which contained all amino acid residues randomized in the α-helix of the ZF (Beerli and Barbas, 2002; Wu et al., 1995). By selecting for phage using oligonucleotides that contain a specific 3 bp subsite, ZF recognition modules that bind to specific 3 bp subsites were isolated. In principle, multi-finger proteins can be constructed by assembling predefined ZF modules in any order to recognize any desired DNA sequence in a modular fashion, which is referred to as modular assembly (Bhakta and Segal, 2010). A set of modular assembly fingers described here was developed by Barbas and his colleagues. Another set of modular assembly fingers was developed by ToolGen, which was based on selections of ZFs occurring naturally in the human genome as opposed to synthetic variants of Zif268 (Bae et al., 2003). The Barbas and ToolGen domains are the two most commonly used sets of modular assembly fingers. Both domains cover all 3 bp GNN, most ANN, many CNN and some TNN triplets (where N can be any of the four nucleotides). Both have a different set of fingers, which allows for searching and coding different ZF modules as needed (Bhakta and Segal, 2010; Gersbach et al., 2014). The main advantage of this approach is that ZFs can be assembled in any order and no selection step is required. Many of the engineered ZF domains that were constructed via modular assembly were shown to have higher specificity as compared to naturally occurring ZF domains by greater than 100-fold in the three finger context (Gersbach et al., 2014). Since modular assembly is one of the most popular methods for constructing ZFPs, it has been widely used in numerous applications, such as nucleases, transposases, recombinases, integrases, and gene regulators (Camenisch et al., 2008; Gordley et al., 2009; Kolb et al., 2005).

Oligomerized pool engineering (OPEN)

To minimize context-dependent effects of modular assembly involving the position of a finger in the protein and the sequence of neighboring fingers, a combinatorial selection-based oligomerized pool engineering (OPEN) strategy was developed by J. Keith Joung for constructing multi-finger arrays (Maeder et al., 2009). Before the OPEN approach, the Pabo group developed the bacterial two-hybrid (B2H) system, which involves a two-step process (Joung et al., 2000). The first step is performed to enrich a finger with out-of-context 3 bp subsites as in modular assembly. Then, a second round of selection enriches ZFs in the context of the full, intended DNA target sequence. Joung and colleagues developed OPEN as an optimized version of this method. An archive of pre-selected ZF pools is used in OPEN, each consisting of a maximum of 95 different fingers targeted to a specific 3 bp subsite at a defined position (Maeder et al., 2008). Appropriate finger pools from the archive are recombined to create a small library of multi-finger arrays for a target 9 bp site of interest. Members of this library are then screened using the B2H selection system, where ZF binding to its designed site activates the expression of selectable marker genes. The efficiency and success rate of this method is approximately 70–80% for obtaining ZF arrays capable of activating transcription in B2H strains (Maeder et al., 2009). OPEN ZF arrays and zinc finger nucleases (ZFNs) are publicly available from the Zinc Finger Consortium Database (Maeder et al., 2009). Although OPEN has sequence constraints in their ZFNs, it has been used to develop successful ZFNs targeting sites in human cells and plants (Curtin et al., 2011; Sebastiano et al., 2011).

APPLICATION OF ZINC FINGER PROTEINS IN BIOMEDICAL RESEARCH

Zinc finger nucleases (ZFNs)

Upon linking ZFs to a nuclease domain, ZFNs are constructed to recognize and cleave DNA at a desired location. The cleavage domain from the type IIs restriction enzyme FokI is fused to ZFs, thereby creating a DNA double-strand break (DSB) at targeted sites. FokI domains must dimerize to cleave DNA. Hence, two ZFs are fused with two FokI cleavage domains to assemble functional ZFNs. Two ZF-FokI monomers bind independently in an inverted tail-to-tail orientation and with a 5–7 bp spacer sequence recognized by the cleavage domain between the binding sites (Gaj et al., 2013a) (Fig. 2).

Fig. 2.

Schematic diagram of a zinc finger nuclease (ZFN) dimer bound to target DNA.

ZF arrays are fused to the FokI nuclease domain to make a custom nuclease that can recognize unique left and right half-sites. The two ZFNs must bind in an inverted tail-to-tail orientation with their C-termini facing each other. The optimal spacing between the half-sties is 5–7 bp.

A ZFN-induced DSB will stimulate cellular DNA repair mechanisms either by error-prone non-homologous end joining (NHEJ) or precise homologous recombination (HR). NHEJ-mediated repair often results in small insertion or deletion (indel) errors at the targeted site. HR involves a precise addition of an exogenous nucleotide sequence that is complementary to the sequence on the broken double-stranded DNA, making it easy to incorporate any nucleotide sequence of choice into the DNA.

Gene editing mediated by ZFNs has been applied to correct disease-causing mutations associated with sickle cell disease (Sebastiano et al., 2011; Zou et al., 2011), α₁-antitrypsin deficiency (Yusa et al., 2011), hemophilia B (Li et al., 2011a) and Parkinson’s disease (Soldner et al., 2011). For the correction of Parkinson’s disease-associated mutations, ZFN-mediated genome editing was combined with induced pluripotent stem cells (iPSCs) technology (Soldner et al., 2011). This approach enables the genetic correction of point mutations in the α-synuclein gene in patient-derived human iPSCs. Another example of combination of ZFNs with iPSCs technology was also found in the correction of the E6V mutation in the β-globulin gene for sickle cell disease (Sebastiano et al., 2011).

ZFN-mediated gene disruption has been taken to clinical trials (NCT00842634 and NCT01044654) for treating HIV (Holt et al., 2010; Perez et al., 2008). ZFNs have been used to confer HIV-1 resistance in CD4⁺ T cells by disrupting the co-receptor chemokine (C-C motif) receptor type 5 (CCR5). This ZFN approach potentially results in a heritable gene knockout of CCR5 and consequently HIV resistance (Urnov et al., 2010). This could allow ZFN-modified CD4⁺ T cells to potentially reconstitute immune function in patients with HIV/AIDs by maintaining an HIV-resistant CD4⁺ T cell population (Urnov et al., 2010). ZFNs were also used to engineer hematopoietic stem and progenitor cells (HSPCs) so that the stem cells could directly mutate and become resistant to HIV (Li et al., 2013).

Zinc finger recombinases (ZFRs)

Site-specific recombinases (SSRs) are capable of recognizing 30–40 bp sequences and catalyzing excision, inversion, or integration between defined segments of DNA (Grindley et al., 2006). Due to their strict target specificity it poses a challenge to use SSRs in cells and organisms that have artificially induced recombination sites or pre-existing recombination sites (Gaj et al., 2013b). To address this challenge, ZFRs were introduced as an effective alternative to the conventional site-specific recognition systems (Gordley et al., 2007). ZFRs catalyze recombination between specific ZF target sites. Like ZFNs, they consist of two inverted ZFs on double-stranded DNA with the recombinase domain exercising its catalytic activity in the 20 bp central flanking region (Smith and Thorpe, 2002). Successful re-engineering of serine recombinases was studied to explore the specificity and effectiveness of ZFRs. Using this approach, Gosh. et al generated enhanced hybrid recombinases based on activated catalytic domains derived from the resolvase/invertase family of serine recombinases (Gin, Hin, Tn3 and γδ) (Gaj et al., 2014b). Through rational design and directed evolution, they reengineered the serine recombinase dimerization interface. The re-engineered hybrid recombinases showed higher specificity with low toxicity, indicating the potential of these enzymes in a wide range of applications for genome engineering and gene therapy (Gaj et al., 2013b).

Zinc finger transposases

The Sleeping Beauty (SB) transposon is an integrating vector system capable of inserting expression cassettes with high stability. However, the SB insertion profile is close-to-random in the genome, and random genomic insertion can cause unwanted mutagenesis of endogenous genes (Voigt et al., 2012). To address this problem, an attempt for targeted transposon insertion has been made such that the transposase or the transposon vector DNA is physically linked to a DNA-binding domain (DBD). In this way, the transposase/transposon complex is tethered to defined sites in the genome and is able to facilitate integration of the transposon into the adjacent intended DNA (Voigt et al., 2012). Fusion of a ZF with SB transposase resulted in a fourfold enrichment of the transposon insertion as compared to native SB transposase (Voigt et al., 2012). In another study, Zif268 was fused to the C-terminus of ISY100 transposase, resulting in highly specific integration into TA dinucleotides positioned 6–17 bp to one side of a binding site for Zif268 (Feng et al., 2010). The classical Gal4 DBD was also fused to a piggyback (PB) transposase to bias genomic insertion to specific sites, upstream activating sequence (UAS) Gal4 recognition sites (Owens et al., 2012). Gal4-PB fusion proteins were able to target transposition near to UAS sites, which were randomly integrated throughout the genome, as compared to native PB transposase.

Zinc finger-artificial transcription factor (ATF)

DNA-binding domains including ZFs can be engineered to regulate expression of specific genes by fusing them to transcriptional or epigenetic effector domains, thus generating artificial transcription factors (ATFs). In principle, ATFs are comprised of a DNA-binding domain, a transcriptional activator (VP16 and p65 domains) or repressor (KRAB domain), and a nuclear localization signal (NLS) to ensure the efficient transport of ATFs into the nucleus. Engineered ZFPs were fused to VP64 and KRAB domains to create synthetic activators and repressors, respectively (Beerli et al., 1998). These ATPs were demonstrated to up- and down-regulate the endogenous ERBB2 and ERBB3 genes in human cells (Beerli et al., 2000). Later, engineered ZF repressors were shown to down-regulate the HIV promotor and reduce HIV replication in primary cells up to 100-fold (Segal et al., 2004). As a therapeutic approach to sickle cell disease and β-thalassemia, ZF ATFs were designed to target the promotor region of the γ-globin gene, resulting in up-regulation of γ-globin expression in human cell lines (Graslund et al., 2005) and activation of the silent γ-globin gene in primary human hematopoietic stem cells in an in vivo mouse model (Wilber et al., 2010). ZF ATFs have been designed to reactivate the paternal UBE3A gene that is silenced by imprinting in an attempt to develop a molecular therapy for Angleman syndrome (Bailus and Segal, 2014).

Protein delivery

ZFNs are intrinsically cell permeable (Gaj et al., 2012), which is attributed to the net positive charge of ZF domains (Gaj et al., 2014a). Cell penetrating ZF domains were successfully proven to be good protein transduction reagents (Gaj et al., 2014a). Gaj et al. demonstrated that when the N-terminus of firefly luciferase was genetically fused to two or three fingers, it resulted in cell penetrating properties as effective as Lipofectamine-mediated plasmid transfection. These protein-fused ZFs are capable of delivering functional proteins into primary and transformed mammalian cells (Gaj et al., 2014a). This study also showed that ZFPs enter the cells mainly through macropinocytosis and at low frequencies through caveolin-dependent endocytosis.

DNA diagnostics

A system called Sequence Enabled Reassembly of β-lactamase (SEER-LAC) consists of two split enzymatic domains of β-lactamase that would reassemble into a full-length enzyme upon ZFPs binding to their target DNA (Ooi et al., 2006). Kim et al. (2011) developed a ZFP array combined with the SEER-LAC system for DNA diagnostic applications. The ZFP array with the SEER-LAC system generated DNA-dose dependent signals with a visual readout and allowed for a quantitative assay. Their result suggested the potential use of this system to develop a point-of-care (POC) diagnostic for pathogen detection.

Cytosine on CpG is frequently methylated in certain genes, causing epigenetic silencing. Detection of DNA methylation is an excellent diagnostic tool for early detection of different carcinomas and adenoma. Ghosh et al. (2006) used the SEER-LAC system fused with an engineered ZFP and a methyl binding domain for direct detection of methylated dsDNA.

TRANSCRIPTIONAL ACTIVATOR-LIKE EFFECTORS (TALEs) – STRUCTURE AND DNA RECOGNITION

TALE represents the largest effector family and functions in transcriptional activation of plant genes. Their unique structure encompasses a DNA-binding region that enables TALEs to bind specifically to the promoter region on DNA. Their binding specificity can be predictable since two hypervariable amino acids within the repeat domain known as repeat variable di-residues (RVDs) determine the nucleotide to which the particular repeat binds (Boch et al., 2009).

The repetitive nature of TALE DNA-binding domains led to the binding code being deciphered in 2009 (Boch et al., 2009). As shown in Fig. 3, amino acids at positions 12 and 13 within a 34-amino acid repeat are called RVDs, which direct nucleotide specificity on the target DNA. The tandem polymorphic amino acid repeats of TALEs are located in the central DNA-binding region. Each RVD recognizes a single DNA base and different RVDs have variable affinity for different nucleotides. The four most common RVDs are HD, NG, NI, and NN, specifying C, T, A, and G, respectively (Boch et al., 2009). There are about 24 known unique RVDs with seven of the most common being HD, NG, HG, NN, NS, NI and N* (N* corresponds to a 33 amino acid repeat with a missing residue within the RVD loop) (Mak et al., 2013). Thus, the number of repeats including the last truncated repeat and the series of RVDs determine the length and the nucleotide composition of the target that they would recognize. TALEs flank the major groove of the DNA helix (Deng et al., 2012; Mak et al., 2012) with the RVDs making contact with the DNA target as shown in Fig. 4. One of these structures is the PthXho1 bound to its target DNA, which shows the presence of two α-helices connected by a loop of RVDs that makes contact with the DNA. The target sequence of all naturally occurring TALEs begins with a thymine (T) nucleotide at the 5′ end, which is important for the functionality of the TALE’s activity (Boch et al., 2009). By deciphering the TALE RVD code, it has been possible to program TALEs with high target specificity and selectivity (Moscou and Bogdanove, 2009). However, a very remarkable study (Rogers et al., 2015) was carried out recently using 21 TALE proteins of different lengths containing all possible consecutive pairs of repeats to identify the influence of these repeats on TALE-DNA binding specificity. Their results infer that not only the affinity of the RVD governs DNA binding, but binding also depends strongly on the base disfavored by the RVD. For example, HD had the highest affinity for C and strongly disfavored G.

Fig. 3.

DNA binding recognition of TALEs.

A central DNA-binding region of TALEs contains an array of multiple repeats that are almost identical except for two amino acids at positions 12 and 13 termed repeat variable diresidues (RVDs). Each RVD specifies one DNA base.

Fig. 4.

Crystal structure of PthXo1 bound to DNA.

PthXo1 contains 23.5 repeats. The figure shows PthXo1 making contact with a 36 bp dsDNA and the HD RVDs at the 12^th and 13^th position within the repeat that recognize a single DNA base (PDB ID: 3UGM) (Mak et al., 2012).

CONSTRUCTION OF TALEs

Golden Gate cloning

The major obstacle in the routine usage of TALEs is that the assembly of repeat TALE arrays can be challenging because of extensive identical repeat sequences. To address this issue, methods for achieving rapid assembly of TALE arrays have been studied. One of these is Golden Gate cloning, which allows several DNA fragments to be assembled in a single cloning step (Engler et al., 2008). The assembly may be manipulated to generate sequences of choice without depending on site-specific restriction enzymes since Golden Gate cloning utilizes Type IIs restriction enzymes that cleave outside the recognition sites, creating 4 bp overhangs (sticky ends). The 4 bp overhang can be any four nucleotide sequence of choice, enabling multiple compatible DNA fragments to be ligated together linearly in a single cloning restriction-ligation step.

Morbitzer et al. (2011) developed a rapid, efficient, and low-cost approach for Type IIs enzyme-mediated assembly of repeat modules, which involves fusion of individual TALE repeat modules into a tandem array. This approach allows them to fuse two repeat sub-arrays containing seven and ten repeat-modules into a functional designer TALE (dTALE). dTALE assembly was carried out by two consecutive BsaI cut-ligation steps, followed by BpiI cut-ligation. Their approach resulted in generating a full-length dTALE gene with a high level of sequence fidelity based on sequence-validated plasmids, and not involving PCR.

Another efficient method for assembly of TALE constructs was reported by Cermak et al. (2011) using Golden Gate cloning. The approach allows for assembly of novel repeat arrays for TALE nucleases (TALENs), TALEs, and TALE fusion proteins in just two cloning steps using a set of sequence-verified modules. Golden Gate reaction 1 was performed to build arrays of 1–10 repeats, followed by a Golden Gate reaction 2 to join arrays in a backbone vector to create the final construct TALEN monomer with a 16 RVD array. The software used to design TALENs in their study is available for use as an online tool (http://boglabx.plp.iastate.edu/TALENT/) to identify sequences of recognition sites for the left and right TALEN monomers and the spacer sequence. The identified binding sites are converted into appropriate RVD sequences using the four most commonly used RVDs (NI, HD, NN and NG, specific to A, C, G and T, respectively). Their TALENs were demonstrated to be active in a yeast DNA cleavage assay and effective in gene targeting in human cells and Arabidopsis. Similarly, Zhang and his group developed a toolbox to construct different TALE-TFs (TALE transcription factor) and TALENs (TALE nucleases) that can bind to different lengths of DNA (Sanjana et al., 2012).

FLASH

The Fast Ligation-based Automated Solid-phase High-throughput (FLASH) assembly method was developed for rapid construction of large numbers of TALE repeat arrays (Reyon et al., 2012). FLASH allows for high-throughput construction of TALE repeat arrays and ligation of multiple individual TALE repeats in a unidirectional fashion. DNA fragments encoding TALE repeats are assembled on solid-phase magnetic beads, which enables serial restriction digestion reactions, purification and ligation, avoiding the need for column-based washing or purification. The interlaced washing steps between ligation events facilitate the desired order of ligations. The final full-length TALE repeat arrays are released from the beads after a restriction digest, which can be then cloned into a suitable expression vector of choice. One can construct DNA fragments encoding 24 or 96 different TALE repeat arrays in a day using manual or automated FLASH methods, respectively (Reyon et al., 2012). All of the 48 TALEN pairs assembled by FLASH were shown to possess significant EGFP gene disruption activities in a human cell-based assay (Reyon et al., 2012). In addition, FLASH-assembled TALENs were tested for modifying endogenous genes involved in human cancer and epigenetics in human cells. It was found that 84 of the 96 TALENs displayed efficient NHEJ-mediated mutagenesis at the intended target sites (Reyon et al., 2012).

To facilitate the high-throughput design of FLASH TALE repeat arrays, Joung and his group improved the Zinc Finger and TALE Targeter software (ZiFiT Targeter) (http://ZiFitBeta.partners.org) (Reyon et al., 2013). The upgraded software enables identification of potential target sites and provides all the information needed about the plasmids for the construction of TALE repeats with FLASH.

APPLICATION OF TALEs IN BIOMEDICAL RESEARCH

TALE nucleases (TALENs)

The non-specific FokI nuclease domain can be fused to TALEs to create TALE nucleases (TALENs), which can produce a DNA DSB. FokI cleavage domains as a dimer are attached to the C-terminal end of the two TALEs, with the two TALEs placed tail to tail (Fig. 5). TALENs are designed in pairs to make contact with the two opposing strands of the target DNA, separated by a spacer to provide the FokI nuclease domains with enough space to dimerize and create a DNA DSB, as described in the section on ZFNs. Repair of TALEN-mediated DSBs was shown to create efficient targeted alteration of endogenous genes in several model organisms, including plants (Cermak et al., 2011), yeast (Li et al., 2011b), zebrafish (Sander et al., 2011), human somatic cells (Cermak et al., 2011; Mussolino et al., 2011) and pluripotent stem cells (Hockemeyer et al., 2011).

Fig. 5.

Schematic diagram of a TALE nuclease (TALEN) dimer bound to target DNA.

TALE arrays are fused to the FokI nuclease domain to make a custom nuclease that can recognize unique left and right half-sites. The two TALE binding sites are separated by a spacer of 12–20 bp in length.

Sun et al. (2012) have constructed and optimized TALENs from a TALE AvrXa10 by manipulating the N-terminal and C-terminal extensions on either side of the repeat domain along with the spacer length of each effector binding element (EBE). Optimized TALENs showed efficient cleavage of target DNA in the human β-globin gene associated with sickle cell disease with little or no cytotoxicity. Ousterout et al. (2013) have successfully used TALENs to manipulate the nucleotide sequence of the protein dystrophin that is involved in Duchenne Muscular dystrophy disease. Exon 51 was deleted via NHEJ, which corrected the reading frame of the gene and caused successful expression of the protein. The TALEN system was used for HIV-1 gene therapy, resulting in approximately 45% disruption of the CCR5 gene (Mussolino et al., 2011). A similar level of gene disruption was also achieved using ZFNs. However, TALENs showed much lower cytotoxicity with significantly reduced off-target activity as compared to ZFNs.

TALE recombinases

SSRs have emerged as genome engineering tools for manipulating DNA because of their high specificity. To alter the specificity of SSRs, the DBD of SSRs can be replaced by custom-designed DBDs such as ZFs and novel DNA-binding TALEs. In the TALE-recombinase system (TALER architecture), TALEs would bring specificity for inducing a DNA DSB while the recombinase would assure homology directed insertion of exogenous DNA. The first attempt to generate a chimeric TALER was carried out by Barbas’s group (Mercer et al., 2012). They created a library of truncated TALE variants to identify optimized TALER fusions with a catalytic domain from the DNA invertase from Gin. Their study showed that TALERs can be used to recombine any DNA sequence in bacteria and mammalian cells, which may overcome the limitation of the modular targeting capacity of ZFRs. They also demonstrated the reprogrammability of the recombinase’s catalytic specificity.

TALE transposases

The non-viral PB transposable element fused with the Gal4 DBD has been studied to address the problem of integrating viral vectors associated with insertions at unwanted sites (Owens et al., 2012). One year later, Owens et al. (2013) generated hyperactive PB transposases fused with custom-designed TALEs to target the first intron of the human CCR5 gene. They have demonstrated targeted transposition to the CCR5 genomic safe harbor, which allows for stable expression of a transgene across multiple cell types (Owens et al., 2013).

TALE-artificial transcription factors (TALE-ATFs)

Engineered TALEs can be fused to transcriptional activator and repressor domains to construct artificial transcription factors (ATFs). TALE-ATFs have been successfully used as gene-specific activators and repressors (Maeder et al., 2013; Mahfouz et al., 2012; Perez-Pinera et al., 2013; Zhang et al., 2011). Engineered TALEs fused with a VP64 domain were shown to target a wide spectrum of DNA sequences at a similar or greater level compared to ZF-ATF bearing a VP64 domain (Zhang et al., 2011). Maeder et al. (2013) constructed a large series of TALE activators with various numbers of repeats and tested their activity in stimulating expression of the endogenous human VEGF-A gene. In their study, most TALE activators bearing a VP64 domain significantly induced expression of the VEGF-A gene in a very wide range from 5.3- to 114-fold. They also found that both VP64 and p65 TALE activators synergistically induced endogenous gene expression at even higher levels compared to each activator expressed individually (Maeder et al., 2013).

CONCLUSIONS

Engineered ZFPs and TALEs when fused with nucleases, repressors or activators are useful for targeting and manipulating a DNA sequence of interest. ZFPs and ZFNs can be custom-designed depending on the target DNA sequence, but ZFPs have shown a certain sequence preference (5′-GNN-3′). TALEs on the other hand have a flexible and modular structure making it possible to target any desired DNA sequence with robust programmability. Both ZFNs and TALENs have been studied as therapeutic agents with numerous clinical and diagnostic applications as described here.

Distinct from ZFPs and TALEs, clustered regularly interspaced short palindromic repeats (CRISPRs) along with the CRISPR associated proteins (Cas) have recently emerged as an alternative DNA targeting platform. The CRISPR/cas systems depend upon a small database of CRISPR RNAs (crRNA) requiring only programming of a 20–22 bp single-guide RNA (sgRNA) (Jiang and Doudna, 2015). The type II CRISPR/cas system can be engineered as a chimeric’ single-guide RNA’ by simply connecting the 3′ end of the crRNA to the 5′ end of the transactivating crRNAs (tracrRNAs) with a linker sequence (Jiang and Doudna, 2015). This assembly can efficiently direct the Cas9 protein to a target DNA sequence matching the 20 bp RNA guide-sequence and induce a double-strand break in the genome of eukaryotic cells (Jinek et al., 2012). By changing the DNA target sequence within the guide RNA, Cas9 can be retargeted to cleave virtually any DNA sequence in the genome. The simplicity of guide RNA design is an advantage over ZFPs and TALEs since CRISPR/cas technology does not require protein engineering depending on target DNA sequence. Therefore, CRISPR/cas systems can serve as effective tools in genome engineering with greater efficiency and fewer off-target binding events. However, CRISPR/cas technology is still new and future studies are needed to address questions related to DNA-binding specificity in the context of complex genomes.

A wide range of applications have been developed using ZFPs and TALEs so far. ZFPs and TALEs clearly provide a powerful and versatile tool for gene targeting and genome engineering. This being said, there is still considerable potential for researchers to look for new applications for diverse biomedical studies. It will be interesting to see the full potential of CRISPR/cas9 technology for biomedical research and applications.