The discovery of RNA interference (RNAi) has profoundly transformed the field of gene silencing. Initially identified in Caenorhabditis elegans, RNAi has emerged as a powerful tool for combating viral and parasitic infections and for exploring gene functions (Fire et al., 1998; Kim and Rossi, 2008; Obbard et al., 2009; Rosso et al., 2009; Huvenne and Smagghe, 2010; Escobedo-Bonilla, 2013; Mohr and Perrimon, 2012). Despite its extensive applications, it is important to recognize that RNAi may result in hypomorphic phenotypes that do not fully recapitulate the effects of genetic mutations (Boettcher and McManus, 2015). Concerns regarding gene flow, horizontal gene transfer, and unintended effects on non-target organisms have motivated ongoing research to refine RNAi-based tools for safer and more effective gene manipulation.

RNAi is a highly conserved and potent mechanism for gene silencing, observed across various eukaryotic species, including plants, protozoa, nematodes, fungi, insects, and vertebrates. However, this process is absent in prokaryotes, highlighting its specialized role in eukaryotic cellular regulation (Das et al., 2011). RNAi acts as a defense mechanism against transposons and viruses, offering genome protection, particularly in plants (Obbard et al., 2009; Kemp et al., 2013; Nicolas et al., 2013).

The RNAi pathway is initiated by the introduction of double-stranded RNA (dsRNA) molecules that are perfectly complementary to the target gene. These dsRNAs are recognized and processed by the RNaseIII-like enzyme Dicer, generating small interfering RNAs (siRNAs) or microRNAs (miRNAs) (Gil-Humanes et al., 2010). These short RNAs, typically 20-24 base pairs long, possess 3′ hydroxyl termini, 2-nucleotide 3′ overhangs, and 5′ phosphorylated termini. They are incorporated into the RNA-induced silencing complex (RISC), where they guide sequence-specific degradation or translational repression of target messenger RNAs (mRNAs) (Das et al., 2011).

This process predominantly occurs in the cytoplasm. Additionally, signal amplification and intercellular propagation of RNAi have been observed, particularly in plants and C. elegans. The RNA-dependent RNA polymerase (RdRp) enzyme plays a pivotal role in amplifying siRNA signals and generating secondary dsRNAs, thereby enhancing the silencing effect ( Figure 1 ) (Cohen and Xiong, 2011; Das et al., 2011).

Recent studies have demonstrated that RNAi can be triggered by various dsRNA molecules, including transposon transcripts, viral satellites, and transgenes (Barba and Hadidi, 2009). In plants, RNAi is more efficiently induced compared to mammals, nematodes, or flies. Instead of chemically synthesized short hairpin RNAs (shRNAs) or siRNAs, plants rely on expression cassettes that produce self-complementary hairpin RNAs (Hirai and Kodama, 2008; Eamens and Waterhouse, 2011). These cassettes are commonly incorporated into vectors such as pH/KANNIBAL and GATEWAY, which generate intron-containing hairpin RNAs (ihpRNAs) (Eamens and Waterhouse, 2011; Yan et al., 2012).

The transcription of RNAi-inducing cassettes in these vectors yields dsRNA molecules with two distinct regions: a single-stranded loop and a double-stranded stem (Eamens and Waterhouse, 2011). Following the development of RNAi vectors, advanced cloning strategies, including ligation-independent cloning (LIC), the Golden Gate cloning strategy, and one-step PCR cloning, have enabled efficient production of hairpin RNAs for gene silencing in plants, animals, and insects (Xu et al., 2010; Yan et al., 2012; Baghban-Kohnehrouz and Nayeri, 2016).

RNAi technology has been extensively employed to develop transgenic plants for improved traits, enhanced nutritional content, and better resistance to pests and diseases. For instance, RNAi has facilitated the production of seedless fruits, extended shelf life, modified flower colors, and improved secondary metabolite pathway. Moreover, RNAi-mediated genetic transformations have been instrumental in controlling plant pathogens and promoting sustainable agricultural practices (Das and Sherif, 2020). Table 1 summarizes the applications of RNAi in various plant species for natural product biosynthesis.

Recognizing the significance of DSBs in genome engineering, the initial artificial system to generate site-specific DSBs involved the use of meganucleases (homing endonucleases). This approach triggered DNA repair pathways, resulting in specific modifications to the DNA, such as indels or single nucleotide polymorphisms (SNPs) in eukaryotic genomes. One of the first meganucleases to be discovered and characterized was the I-SceI meganuclease, which was identified in the 1970s and 1980s from the mitochondria of S. cerevisiae (yeast) (Jacquier and Dujon, 1985; Bos et al., 1978; Faye et al., 1979).

A rare-cutting enzyme, produced via an intron within the mitochondrial large 21 S ribosomal RNA subunit (LsrRNA), selectively identifies and cuts an 18 bp sequence in an intron-less version of the LsrRNA gene, causing a double-strand break (DSB) (Jacquier and Dujon, 1985; Stoddard, 2014). The homologous recombination (HR) pathway rectifies the resultant double-strand breaks (DSBs) by employing a variant of the 21 S rRNA gene that contains an intron, thereby facilitating the incorporation of the intron-embedded I-SceI open reading frame (ORF) region into the designated target locus (Jacquier and Dujon, 1985). The I-SceI endonuclease, characterized by its extensive DNA recognition sequence and pronounced specificity, can be expressed in a manner that does not compromise the integrity of host genomes or cellular structures. As a result, I-SceI has been effectively harnessed to induce DSBs at targeted loci, allowing for the investigation of DNA repair mechanisms across various eukaryotic genomes (Rouet et al., 1994a, 1994b; Choulika et al., 1995). In the year 1993, Puchta et al., demonstrated that the DSBs generated by I-SceI enhance HR not only in plant systems but also in Nicotiana tabacum (Puchta et al., 1993; Puchta, 1999).

Meganucleases require the ability to design nucleases with a high level of sequence specificity to achieve exact genome modification. However, one of the challenges in engineering these nucleases is the overlap between the DNA-binding domains and cleavage (Stoddard, 2011). This overlap can hinder the design process. On the other hand, altering the amino acid sequence to gain new DNA sequence particularity often compromises the catalytic action of the enzyme. To overcome this challenge, a semi-rational approach has been developed. This approach involves using the mathematical examination of the structural aspects of the protein-DNA interface to generate meganucleases with new specificities. This strategy has shown promise in the field (Ashworth et al., 2006).

The two-step combinatorial method has made it easier to create personalized meganucleases through a wider span of target sites. This has expanded the usefulness of designer meganucleases for different purposes. Among these is I-CreI, which serves as a molecular framework for generating innovative meganucleases with unique specificities. Nevertheless, there are instances where their binding and/or cleavage characteristics may not be optimal. But, by optimizing the framework, we can improve both the nuclease specificity and activity. Different approaches have been explored to achieve this goal (Arnould et al., 2007; Redondo et al., 2008; Grizot et al., 2009).

Meganucleases were assessed for their potential application in tomato and oilseed rape crops. Both I-SceI and a tailored meganuclease have been identified as capable of inducing homology-directed repair (HDR) and double-strand breaks mediated recombination in a reporter gene (Rahmanian and Seyeddokht, 2023). Despite being less effective than I-SceI, the customized meganuclease was successful in causing the removal of an exogenous transgene in tomato plants (Danilo et al., 2022). Engineered meganucleases based on I-Crel have been created to exhibit enhanced activity in recognizing sequences with specific central sequences. The development of the ARCUS platform represents a new generation of meganuclease technology that enables the production of nucleases with tailored activity and specificity, including the ability to differentiate between target sites that vary by just 1 base pair (Bartsevich et al., 2016).

Youssef et al. (2018) described for the first time the use of a customized meganuclease to cleave wheat DNA in vivo. They showed that double excisions removed previously inserted DNA cassettes containing the DsRed reporter gene and, in many cases, the meganuclease target site was correctly reconstructed, providing opportunities for subsequent insertion of accumulated genes to replace the selected gene. I-SceI nuclease catalyzed the precise integration of gene in maize at a pre-integrated target site, thereby inducing expression of the BAR gene (the BAR gene is known to confer resistance to phosphinothricin) (D’Halluin et al., 2008).

Studies have shown that the effectiveness of meganucleases can be impacted by various factors related to the chromosome context of the target sequences (Kim et al., 1996; Smith et al., 2000; Durai et al., 2005). However, the low frequency of genome editing in somatic cells was one of the major limitations of I-SceI and other meganucleases, recognition of natural target sites in the genome could not be easily available (Porteus, 2015). Therefore, to solve this problem, ZFNs and TALENs technologies with endonuclease activity domains for inducing targeted DSBs at specific genomic DNA loci have been developed.

Zinc finger nucleases (ZFNs) are engineered nucleases designed to create precise DSBs in DNA at specific genomic locations (Memon, 2021). These breaks can trigger DNA repair mechanisms, resulting in targeted mutagenesis or chromosomal segment removal via non-homologous end joining (NHEJ) or gene targeting through homologous recombination (HR) (Van Eck, 2020). The discovery of zinc finger protein domains in the early 1990s revolutionized genome engineering in model plants and crops. Using ZFNs to target specific sequences in genomic DNA enables various modifications through DNA repair pathways like NHEJ and HR (Weinthal et al., 2010; Tzfira et al., 2012; Voytas, 2013).

ZFNs are vital in activating nuclease dimers that cleave DNA away from the binding site. Each monomer contains a zinc finger DNA-binding domain-spanning 30 amino acids and featuring arrays of Cys2-His2 fingers with a Zn²⁺ binding ion—and a nonspecific FokI nuclease domain ( Figure 2 ). The zinc finger domain typically includes three or four patterns, with each pattern recognizing a specific group of three DNA bases. This enables precise recognition of an 18 or 24-base pair sequence using a ZFN pair (Zhang et al., 2015; Simmons and Douglas, 2016).

Advances in ZFN engineering have been achieved through two major platforms. The first, modular assembly, combines individual fingers with specific DNA-binding properties (Beerli and Barbas, 2002; Liu et al., 2002; Bae et al., 2003; Segal et al., 2003; Mandell and Barbas, 2006). However, modular assembly often has low efficiency (≤30%) and may exhibit limited activity or high toxicity (Cornu et al., 2008; Ramirez et al., 2008; Kim et al., 2011). The second approach, based on screening multi-finger databases, considers interactions between neighboring fingers. Sangamo Bioscience pioneered this method, branding it as CompoZr^® from Sigma Aldrich^® (Doyon et al., 2008). Academic methods include oligomerized pool engineering (OPEN) (Maeder et al., 2008) and context-dependent assembly (CoDA) (Sander et al., 2010a), both using improved two-finger databases (Gupta et al., 2012).

Online tools such as ZiFiT Targeter Version 4.2 (http://zifit.partners.org/ZiFiT/) (Sander et al., 2007, 2010b), Zinc Finger Database (ZiFDB 2.0) (https://zifdb.msi.umn.edu:8444/ZiFDB), and ZFNGenome (Reyon et al., 2011) facilitate ZFN design and selection through modular synthesis. Researchers continue exploring ZFN-mediated genome modifications like targeted mutagenesis, gene replacement, and stacking via HR repair pathways in various crops ( Table 1 ).

Despite its potential, ZFN technology faces challenges, including germinal transmission, chromatin accessibility, and off-target effects. ZFN-induced DSBs often result in small deletions (≤50 bp). Achieving large chromosomal deletions requires simultaneous creation of DSBs at both ends, with deletion frequencies remaining low (Qi et al., 2013). Tools have been developed to predict off-target sites in genomes outside plants (Cradick et al., 2011). In Arabidopsis, potential off-targets are identified using BLAST scans and verified for ZFN activity (Zhang et al., 2010). High-throughput methods have assessed off-target effects in maize. However, a major drawback of ZFNs remains the high cost and complexity of designing high-affinity DNA-binding domains essential for genome editing (Shukla et al., 2009).

ZFN-mediated genome modifications have been achieved in various plants, including Arabidopsis, petunia, tobacco, soybean, and maize (Zala et al., 2016). Although ZFN-induced modifications are often successful in somatic cells, germ cell modification rates remain low (Kamburova et al., 2017). Challenges such as low germ cell transmission and cellular toxicity must be addressed to maximize ZFN potential in plant genetic engineering (Qi, 2015).

The applications of ZFNs in targeting specific genes in plants include various examples. For instance, the ADH1 gene in tobacco, encoding the enzyme alcohol dehydrogenase, was targeted to investigate the feasibility of site-specific mutagenesis in plants and served as a model system for functional genomics (Townsend et al., 2009). Similarly, in Arabidopsis, targeted deletions and insertions in the ADH1 gene demonstrated the precision of ZFNs in genome editing (Qi and Zhang, 2014). In soybean, the DCL2 gene, involved in virus-induced gene silencing, was knocked out to enhance resistance to viral pathogens (Curtin et al., 2011). Additionally, in wheat, the GW2 gene, associated with grain weight, was edited to improve crop yield by enhancing grain size and weight (Wang et al., 2014).

Transcription Activator-Like Effector Nucleases (TALENs) are a highly effective gene-editing technology, capable of precisely modifying specific genomic sequences. TALENs consist of a DNA-binding domain that interacts with the target DNA and is connected to a DNA cleavage domain, enabling the induction DSBs at specific locations in the genome. These DSBs can be repaired through either homology-directed repair (HDR) or error-prone nonhomologous end joining (NHEJ), thereby significantly enhancing transformation efficiency (Harale et al., 2022). Initially discovered in Xanthomonas species, TALEs are a group of DNA-binding nuclease proteins that infect a variety of plants, including citrus, rice, pepper, cotton, and soybean (Mansfield et al., 2012). Research on TALEs has revealed their ability to bind to genomic DNA and activate gene expression by mimicking eukaryotic transcription factors (ETFs) (Nemudryi et al., 2014). The structure of TAL effector proteins includes an N-terminal type III secretion domain, a central DNA-binding region known as the repeat array, which consists of several repeats (ranging from 1.5 to 33.5 repeats of 34 amino acids each), and a C-terminal region that contains multiple nuclear localization signals (NLS) and an acidic activation domain (AAD), characteristic features of prokaryotic transcription factors (TF) ( Figure 3A ) (Boch and Bonas, 2010). Notably, there is significant variability in the sequence of TAL effector proteins, particularly at the C- and N-termini, as well as within the repeat array. The 12th and 13th residues within each repeat, known as repeat-variable di-residues (RVDs), display the greatest variability. While over 20 different RVDs have been identified in natural TAL effectors, four RVDs including HD (binding to C), NI (paired with A), NG (linked to T), and NN (recognizing both G and A) are the most prevalent, accounting for over 75% of RVDs ( Figure 3B ) (Moscou and Bogdanove, 2009). A conserved T at the -1 position is a common feature in all TAL effector sequences; any alteration to this T results in a decrease in gene upregulation (Boch et al., 2009).

TAL effector nucleases (TALENs) are engineered to function as dimers, with each monomer containing the catalytic domain of FokI nuclease. These TALENs have shown superior efficiency compared to zinc finger nucleases (ZFNs) in generating high-throughput DSBs in the target DNA (Christian et al., 2010). TALEN monomers are designed to bind to two half-sites of DNA with a spacer sequence between them. This unique design allows FokI monomers to dimerize and create a double-strand break in the spacer sequence separating the half-sites ( Figure 4 ) (Christian et al., 2010). Initially, designing new TAL effector arrays to recognize a specific sequence was time-consuming; however, researchers have developed efficient strategies to address this challenge. Numerous online tools are now available for designing and selecting TALEN pairs, including TAL Effector Nucleotide Targeter 2.0 (TALE-NT), Scoring Algorithm for Predicting TALEN Activity (SAPTA), TAL Effectors, E-TALEN, CHOPCHOP, TALEN Designer, ZiFit, and Mojo Hand (Gaj et al., 2013).

A widely adopted method for producing high-throughput engineered TALENs is the Golden Gate cloning system. This system facilitates the sequential assembly of multiple DNA fragments in a single reaction using Type IIS restriction endonucleases (Engler et al., 2008, 2009; Briggs et al., 2012; Reyon et al., 2012; Schmid-Burgk et al., 2013). The application of the Golden Gate method has greatly accelerated the development of novel TAL effector arrays and has further advanced genome engineering. TALENs have diverse applications, including enhancing plant traits through the introduction of novel genes and regulating gene expression (Behboudi et al., 2022). Additionally, TALENs have been employed to generate OVM-knockout chicken eggs, offering a potential source of safe, allergen-free food (Ezaki et al., 2023). In pigs, TALENs have been used to produce genetically inheritable knockout pigs with a mutation rate comparable to wild-type controls, making them a reliable resource for clinical applications (Barnett, 2018). The precision and safety of TALEN-based genome editing make them invaluable tools in genetic engineering (Choi et al., 2020).

Over the years, numerous studies have documented TALEN-mediated genome modifications in approximately 17 different model organisms, including a variety of plant species such as Arabidopsis protoplasts, tobacco, barley, rice, and Brachypodium (Cermak et al., 2011; Li et al., 2012; Mahfouz et al., 2012; Shan et al., 2013a; Wendt et al., 2013; Zhang et al., 2013). These studies highlight the potential of TALENs to revolutionize agricultural practices by enabling targeted genome editing in crops and livestock. For instance, TALENs have been used to knock out three TaMLO homoeologs in wheat, conferring resistance to powdery mildew (Wang et al., 2014). They have also been employed to generate rice resistant to Xanthomonas oryzae pv. oryzae (Li et al., 2012) and to knockout the vacuolar invertase (VInv) gene in potatoes, resulting in tubers with negligible levels of harmful reducing sugars during cold storage (Clasen et al., 2016). Furthermore, Zhang et al. (2013) described targeted genome modification in tobacco (Nicotiana tabacum) protoplasts, with TALENs directed at the ALS gene. Despite these advances, the construction of TALENs requires the development and synthesis of a new nuclease for each target DNA sequence, as well as the re-engineering of TALEN or ZFN. Therefore, assembling TALENs is a time-consuming and costly process that requires significant expertise in experimental design and molecular biology (Gaj et al., 2013).