Targeted genome editing is a broadly applicable approach for efficiently modifying essentially any sequence of interest in living cells or organisms. This technology relies on the use of engineered nucleases, that is, artificial proteins composed of a customizable sequence-specific DNA-binding domain fused to a nuclease that cleaves DNA in a non-sequence-specific manner. These targetable nucleases are used to induce double-strand breaks (DSBs) into specific DNA sites, which are then repaired by mechanisms that can be exploited to create sequence alterations at the cleavage site. Nuclease-mediated genome editing enables genetic studies that were previously difficult or impossible to perform. This technology might also provide therapeutic avenues for genetic disorders including monogenic diseases such as sickle cell anemia or cystic fibrosis. Reflecting its broad importance, genome editing with engineered nucleases was named “Method of the Year” for 2011 by Nature Methods.¹

The vast majority of targeted genome editing described in the literature (including initial foundational studies) has been performed using zinc finger nucleases (ZFNs) (Box 1). ZFNs have been used to modify endogenous genes in a wide range of organisms and cell types². Several types of genomic alterations can be introduced with ZFNs including point mutations, deletions, insertions, inversions, duplications, and translocations, thus providing researchers with unprecedented tools to perform genetic manipulations. Furthermore, ZFNs can potentially be used for therapeutic purposes; for example, ZFNs designed to disrupt the expression of the HIV co-receptor CCR5 (chemokine receptor 5) gene³ have entered Phase 2 clinical trials for treatment of HIV/AIDS.

Recently, transcription activator-like effector nucleases (TALENs) have rapidly emerged as an alternative to ZFNs for genome editing and introducing targeted DSBs. TALENs are similar to ZFNs and comprise a non-specific FokI nuclease domain fused to a customizable DNA-binding domain. This DNA-binding domain is composed of highly conserved repeats derived from transcription activator-like effectors (TALEs), which are proteins secreted by Xanthomonas bacteria to alter transcription of genes in host plant cells⁴(Figure 1a and 1b).

TALENs have generated much interest and excitement because they can be very easily and rapidly designed by researchers using a simple ‘protein-DNA code’ that relates modular DNA-binding TALE repeat domains to individual bases in a target-binding site. Over the last two years, leveraging technologies and methodologies previously developed for the use of ZFNs, several groups have used TALENs to modify endogenous genes in yeast⁵, fruit fly⁶, roundworm⁷, crickets⁸, zebrafish^9–11, frog¹², rat¹³, pig¹⁴, cow¹⁴, thale cress¹⁵, rice¹⁶, silkworm¹⁷, and human somatic^{15, 18, 19} and pluripotent stem cells²⁰ (Supplementary information S1) and presumably the technique will continue to extend to additional organisms. Furthermore, a recent large-scale test demonstrated that TALENs have a very high success rate and can be used to target essentially any DNA sequence of interest in human cells¹⁸ (Supplementary information S2). Although ZFNs and TALENs have not been directly compared, many studies have shown that TALENs and ZFNs have comparable efficiencies when targeted to the same gene^{9, 13, 18, 20}. Thus, the ease of design, high rates of cleavage activity, and the essentially limitless targeting range of TALENs make them suitable for the use by non-specialist researchers.

In this Innovation article, we will first briefly describe the simple modular strategy used to design customized TALE repeat DNA-binding domains and then review recent progress on the use of TALENs to introduce different types of genome alterations in a range of organisms and cell types. In addition, we provide a comparison of the various publicly available methods for constructing TALENs. Finally, we outline important goals for future research that will further enhance the utility of these tools for research and therapeutic applications.

The fundamental building block used to engineer the DNA-binding domain of TALENs is a highly conserved repeat derived from naturally occurring TALEs encoded by Xanthomonas proteobacteria. These TALEs are injected into host plant cells via a Type III secretion system and bind to genomic DNA to alter transcription in these cells, thereby facilitating pathogenic bacterial colonization.⁴ DNA binding by these TALEs is mediated by arrays of highly conserved 33–35 amino acid repeats flanked by additional TALE-derived domains at the amino- and carboxy-terminal ends of the array (Figure 1c).

Individual TALE repeats in an array each specify a single base of DNA determined by the identities of two hypervariable residues typically found at positions 12 and 13 of the domain (Figure 1c and d). Experimental evidence for this simple recognition code was first provided by Bonas and colleagues in 2009.²¹ The researchers observed that the number of repeats in an array corresponded to the length of its target site and this insight enabled them to deduce a simple correlation between the hypervariable residues and the base bound by each repeat. Moreover, they found that a thymine is conserved at the position just 5’ to the base bound by the first repeat in the array (Figure 1d). This group provided experimental evidence for the TALE repeat code by constructing the first examples of engineered TALE repeat arrays with novel specificities.²¹ The TALE repeat code was also confirmed by another group by performing a computational analysis of the binding specificities of naturally occurring TALEs²². Subsequent reports provided additional evidence that engineered TALE repeats with desired specificities can be created using the code.^{19, 23–25}

More recently, co-crystal structures of TALE DNA-binding domains bound to their cognate sites have shown that individual repeats comprise two-helix v-shaped bundles that stack to form a superhelix around the DNA and the hypervariable residues at positions 12 and 13 are positioned in the DNA major-groove. The residues at position 8 and position 12 within the same repeat make a contact with each other that may stabilize the structure of the domain while the residues at position 13 can make base-specific contacts with the DNA.^{26, 27}

Nearly all engineered TALE repeat arrays published to date use four domains with the hypervariable residues NN, NI, HD and NG, for the recognition of G, A, C, and T, respectively. Another repeat with hypervariable residues NK has been reported to be more specific for G than the NN repeat (which can also recognize A)^{19, 22} but arrays using the NK repeats show less activity than those using NN repeats.^{10, 28} More recently, a repeat with hypervariable residues NH has been reported to be more specific than the NN repeat, but with only slightly lower activity.^{28, 29} Additional studies with a greater number of repeat arrays are needed to determine the optimal repeat domain for G recognition. It will also be interesting to explore whether repeats bearing other hypervariable residue combinations will have higher or different specificities for one or more DNA nucleotides.

A substantial body of literature demonstrates that normal cellular repair of ZFN-induced DSBs by non-homologous end-joining (NHEJ) or homology-directed repair (HDR) can be exploited to introduce targeted genome alterations in a wide range of organisms and cell types^{2, 30}. NHEJ-mediated repair of a nuclease-induced DSB leads to the efficient introduction of variable length insertion/deletion (indel) mutations that originate at the site of the break (Figure 2a). Thus, NHEJ-mediated repair of DSBs introduced into gene coding sequences will often yield frameshift mutations that can lead to knockout of gene function.

Alterantively, if a double-stranded DNA “donor template” is supplied, HDR of a nuclease-induced DSB can be used to introduce precise nucleotide substitutions or insertions of up to 7.6 kb at or near the site of the break³¹ (Figure 2a). Recent work has also shown that oligonucleotides can be used with ZFNs to introduce precise alterations, small insertions, and large deletions³². ZFNs have been used to introduce NHEJ- or HDR-mediated gene alterations in fruit fly^{33, 34}, roundworm^{7, 35}, zebrafish^{36, 37}, rainbow trout³⁸, catfish³⁹, sea urchin⁴⁰, frog⁴¹, pig⁴², cattle⁴³, cricket⁸, rabbit⁴⁴, silkworm⁴⁵, butterfly⁴⁶, mouse^{47, 48}, rat^{49, 50}, soybean^{51, 52}, thale cress^{53, 54}, corn⁵⁵, tobacco⁵⁶, petunia⁵⁷, hamster cells⁵⁸, and human somatic^{59, 60} and pluripotent stem cells^61–63. In most of these organisms and cell types, the high absolute rates of mutagenesis that can be achieved with ZFNs have enabled researchers to screen for mutations without the need for selection.

As summarized in Figure 2b, ZFNs and the I-SceI homing endonuclease have also been used to induce other more complex types of genome alterations in mammalian cells. These include large deletions induced by introduction of two DSBs with subsequent deletion of intervening sequence of up to 15 Mbps in length⁶⁴, translocations induced by two DSBs introduced on different chromosomes^{65, 66} and inversions of chromosomal sequence between two DSBs induced on the same chromosome⁶⁷. Given the requirement to introduce two DSBs, it is not surprising that these more complex alterations are obtained with lower efficiencies compared to alterations that are dependent on a single DSB.

Although TALENs were first described only two years ago, these nucleases have already been utilized in a large number of applications. TALENs have been used to generate NHEJ-mediated mutations in a wide variety of organisms with generally high efficiencies (summarized in Supplementary information S1). TALENs have also been used to introduce specific insertions in human somatic and pluripotent stem cells using double-stranded donor templates.^{19, 20}

As noted above, the rapid development of customized ZFNs has already substantially expanded the scope of genetic research that can be performed in a broad range of different organisms and cell types. The high efficiencies of alterations observed have already inspired efforts to use ZFNs as a potential therapeutic approach for genetic-based diseases. The relative simplicity with which TALENs can be engineered will further spur efforts to explore the research and therapeutic applications of customized nuclease technology. In this section, we review recent progress on the use of TALENs in various organismic and cellular contexts and briefly discuss prospects for their future applications.

Gene-editing nucleases offer the potential to directly assess the impacts of gene disruption and of specific sequence variants on gene function in somatic cell-based models of disease. To date, TALENs have primarily been used to disrupt human genes via introduction of NHEJ-induced indels into coding sequence^{15, 18, 19, 68–71}. In principle, such loss-of-function mutations could be used to create somatic cell-based models of disease. Alternatively, precise insertions have also been introduced into endogenous human genes using HDR with TALENs and a double-stranded homologous donor template plasmid^{19, 20}. Targeted insertions could be used to create endogenous gene fusions to fluorescent proteins or epitope tags to visualize protein expression, distribution, or interactions. Beyond generation of such fusions, HDR-based approaches might also be used to create isogenic human or other mammalian cell lines bearing specific single nucleotide polymorphisms (SNPs) identified by large-scale GWAS, ENCODE, or other sequencing projects, potentially enabling studies to determine the functional significance of these sequence variants.

The construction of DNA encoding engineered TALE repeat arrays can be challenging due to the requirement to assemble multiple, nearly identical repeat sequences. Different platforms have been designed to facilitate the assembly of plasmids encoding TALE repeat arrays. These methods are depicted in Supplementary information S3 and can be grouped into three broad categories: standard restriction enzyme and ligation-based cloning, “Golden Gate” cloning, and solid-phase assembly. Detailed descriptions of these methods are provided in Supplementary Information S2.

As summarized in Supplementary information S4, these platforms vary in their throughput, molecular cloning techniques used, numbers of plasmids required (and time required to prepare these DNAs), use of potentially mutagenic PCR, flexibility in the length of arrays that can be constructed, ease with which the required reagents and detailed protocols can be acquired, and availability of author-supported web-based software for practicing the method. Reagent kits for three of these methods are available to academics through the non-profit plasmid distribution service Addgene (http://www.addgene.org/TALEN/). We have established and maintain an active and open newsgroup (currently with over 660 members) for discussion of TALE-related projects (http://groups.google.com/group/talengineering) as well as a “one-stop” comprehensive website with links to protocols, reagents, software, and other information about engineered TALE technology (http://www.TALengineering.org).

The specific architecture of a TALEN is an important factor for users to consider as they choose an assembly method. Various TALEN architectures have been utilized to date and one difference among these is the length and sequence composition of the amino- and carboxy-terminal TALE-derived sequences that flank the TALE repeat array. In the earliest TALENs described in the literature, large segments of naturally occurring TALE sequence were used to join the FokI domain to the carboxy-terminal end of engineered TALE repeat arrays.²⁴ The TALEN framework was then refined by showing that nuclease activities could be greatly enhanced by truncating the length of this carboxy-terminal TALE-derived sequence¹⁹,⁶⁸. In addition, although early studies used wild-type homodimeric FokI nuclease domains, more recent reports^{10, 20, 75} have employed various obligate heterodimeric domains originally developed and used with ZFNs⁷⁶ (Box 1). Therefore, because not all architectures are the same, we suggest that users should carefully consider the reported activity levels and potential specificities of TALENs made on these different frameworks when choosing a method of assembly. We note that as of the writing of this review, the most extensively tested and validated TALEN framework remains that described by Rebar and colleagues^{7, 9, 13, 18–20, 75, 77} (Supplementary information S1).

Although the development of engineered TALE technology has proceeded at an extremely rapid pace over the past three years, many important questions remain to be addressed if these proteins are to be used routinely for research and therapeutic applications. First, although TALENs and ZFNs can induce specific HDR events, competing mutagenesis by NHEJ can still lead to unwanted mutation of the original and, in some cases, the HDR-altered allele. It will therefore be important to develop generalizable methods that tip the balance away from NHEJ- and toward HDR-mediated repair. For example, recent work^{7879, 80} has demonstrated that ZFN-derived nickases that cleave only one strand, instead of both strands, of DNA can shift this balance, although the absolute rates of HDR-mediated repair can be lower than those induced by the ZFNs from which they are derived. Second, developing methods that enable definition of the genome-wide specificities of TALENs will be critical to minimizing off-target NHEJ-mediated mutagenesis. Third, optimization of methods for efficiently delivering TALENs or nucleic acids encoding them into cells will also be an important area for future research. It will be interesting to investigate whether purified native or modified TALEN proteins, like ZFNs⁸¹, might be efficiently taken up directly by cells.

Another potential area for future exploration will be creation of fusion proteins harbouring domains others than nucleases. TALE-based activators and repressors have already been described (Box 2). However, one can also envision that engineered TALE repeat arrays might be used to direct functional domains that induce epigenetic alterations (such as covalent histone or DNA modifications) to specific genomic loci to induce stable, heritable alterations in gene expression. TALE repeat arrays fused to a recombinase domain have recently been described⁸², raising the exciting possibility of enabling targetable site-specific recombination events.

Engineered TALE technology promises to facilitate and enhance genetic manipulations in different organisms and cell types. The simplicity with which TALENs can be designed together with their robust success rates has already spurred much broader adoption of genome editing technology. Although many interesting and challenging questions remain, the accessibility and power of TALENs makes this technology an exciting and important subject for future research and development.