(Yin et al. 2015).
Recombineering is rapidly becoming the method of choice to manipulate biosynthetic gene clusters but it is also increasingly used to evolve the bacterial genome as pioneered by the work of Church´s group. Hang HH, Isaacs FJ et al. in their landmark paper of 2019 described the use of recombineering to accelerate bacterial genome evolution by an automated multiplex recombineering strategy that they named MAGE (Wang et al. 2009). MAGE is using a pool of ssDNA oligonucleotide coupled with the expression of a ssDNA‐binding protein to install thousands of functionalized genome variants in E. coli genome. MAGE has been recently applied to lower Eukaryotes such as Saccharomyces cerevisiae (Si et al. 2017).
One of the problems associated with high‐throughput Recombineering is the need to select for the functional variant because there is usually no selective advantage for the recombined bacterial cell. To overcome this limitation, several strategies have been developed to combine the precision and efficiency of Recombineering with the strong selection pressure that occurs in bacteria after the generation of DSB caused by endonucleases like Cas9/CRISPR (Jiang et al. 2013; Jiang et al. 2015; Baker et al. 2016). Despite the large success of Recombineering applied to E. coli genetic engineering, the engineering of large biosynthetic pathways is still inefficient in endogenous hosts where a Recombineering system is not efficient or where Homologous Recombination is not efficient. Therefore, alternative strategies need to be developed to overcome this limitation. One possibility would be to exploit alternative pathways (Su et al. 2016) of DNA recombination/repair such as non‐homologous end joining (NHEJ) as previously done in mammalian cells (Maresca et al. 2013). A more promising approach relies on the recently described integration system by Transposon‐encoded CRISPR‐Cas (Klompe et al. 2019; Strecker et al. 2020). These strategies can facilitate genome engineering of bacteria and can possibly be implemented for the genetic manipulation of Eukaryotic genomes that are less prone to homologous recombination as described in the next section.
2.5 Genetic Engineering in Higher Eukaryotes
The genomes of Higher Eukaryotes and in particular mammalian genomes are quite resilient to genetic manipulation. It was only in the 1986 that the first gene targeting experiment to generate a functional Knock‐Out of the HPRT gene was published by scientists in Capecchi´s group (Thomas et al. 1986). The same targeting strategy was further optimized by Janish´s group leading to the establishment of a gene targeting pipeline in mESC where potentially any gene can be modified to generate genetically engineered mice models. Despite several attempts to further optimize the method to reduce homology arms length, it became evident that this gene targeting strategy in mESC requires long stretches of homology to the target sites to obtain an optimal efficiency of recombination. High frequency of recombined cells allows the isolation of engineered clones. An increase in the homology arm length by using targeting engineered BACs could further boost this efficiency (Yang and Seed 2003). One characteristic of this canonical gene targeting strategy is that most of the targeting events result in mono‐allelic insertion. Consequentially, additional gene targeting steps or animal breeding strategies are required to obtain bi‐allelic targeting. As discussed above, Recombineering has recently become the method of choice to generate the complex targeting cassettes with long homology arms needed for mammalian genome manipulation.
This classical gene targeting approach works relatively well in mESC and in DT40, a chicken bursal lymphoblast cell line (Buerstedde and Takeda 1991). It is still not clear to this day the reason why this approach is difficult to implement in other cell types. However, in 2002, David Russel lab showed that it is possible to use Adeno Associated Virus (AAV) to promote gene targeting in mammalian cell lines (Hirata et al. 2002). Despite the importance of this finding, this system is still laborious and pretty inefficient in inducing bi‐allelic gene targeting. A promising approach to introduce small indels/mutations by single‐stranded oligonucleotide in mammalian cells has been described by Kmiec´s lab using chimeric oligonucleotides, but it has not been extensively applied for the generation of cellular model due to its low efficiency and due to lack of reproducibility when targeting different loci (Cole‐Strauss et al. 1996).
Experiments from Jasin´s group clearly demonstrated that a targeted DSB induced by I‐SceI enzyme can overcome the anti‐recombinogenic feature of mammalian genome by promoting homologous recombination. This experiment was based on previous observations in budding yeast on mating‐type switching (Strathern et al. 1982) and on the model of recombination dependent on DSB generation (Szostak et al. 1983). There are two important findings as result of the work from Jasin´s group. The first one is that DSBs are preferentially repaired by NHEJ in mammalian cells and although it is difficult to evaluate the amount of perfect repair by precise NHEJ, imprecise NHEJ, using additional enzymes to process the broken ends of the DSBs, can result in template‐independent gene disruption. The second important finding is that DSBs can promote bi‐allelic DNA recombination if the break occurs in both alleles.
2.6 Targeted Endonucleases
The use of DSBs to promote homologous recombination in mammalian cells was limited by the lack of a modular site‐specific endonuclease technology that could be directed to any gene of interest. The solution to this problem came from the work of Chandrasegaran´s group where for the first time the engineering of Zinc Finger‐based endonucleases (ZFN) was described (Kim et al. 1996). This system combines the modular design of Zinc Finger Domain to the powerful nuclease activity of the restriction enzyme FokI that requires dimerization to induce DSBs. The requirement of an obligate dimerization of FokI for an efficient cleavage guarantees high specificity to the system. The Zinc Finger DNA‐binding domain was selected based on the previous work from the labs of Klug and Pabo directed to the discovery of ZF domain (Miller et al. 1985) and the understanding of its structure (Pavletich and Pabo 1991).
Following the publication of the ZFN architecture, Carroll´s group showed for the first time efficient gene targeting using ZFNs in the Drosophila genome (Bibikova et al. 2002). This experiment not only resulted in efficient targeting by installation of random indels via NHEJ but it also demonstrated that ZFNs were precise enough to induce a number of DSBs in the genome compatible with embryonic development. In follow‐up works, gene insertion by Homology Directed Repair after ZFN cleavage was demonstrated by Carroll and Baltimore/Porteus groups, respectively, in Drosophila (Bibikova et al. 2001) and human cells (Porteus and Baltimore 2003).
The scene was set for preclinical genome editing, that is the demonstration of gene editing at disease‐relevant targets. The work from Fyodor Urnov and colleagues at Sangamo therapeutics to target the X‐linked (SCID) mutation in the IL2Rγ led to a new era in genome engineering (Urnov et al. 2005). The other important message coming from this work was the demonstration of bi‐allelic genome editing using targeted ZFNs, something that is very infrequent with classical gene targeting strategies as discussed above.
After the demonstration of preclinical genome editing, an important advancement in the field was the development of systems to engineer targeted endonucleases. These protein engineering pipelines resulted in the generation of precise endonucleases that could safely target the human genome without inducing strong toxicity as shown in the work of Joung, Kim, Cathomen & Voytas´ groups (Maeder et al. 2008; Kim et al. 2009).
It resulted evident from these early experiments that NHEJ is the main pathway of repair exploited by mammalian cells to repair DSBs. Based on these observations, most of the initial preclinical genome editing approaches were directed to exploit NHEJ for the generation of loss of function alleles. Scientists at Sangamo Therapeutics in collaboration with June´s group suggested that CCR5 could be an ideal target for the NHEJ‐dependent loss of function approach. This strategy was based on genomics data associating CCR5 loss of function to protection from HIV infection. Their work led to the first use of genome editing in primary human cells with efficient and safe genome editing (Perez et al. 2008; Maier et al. 2013).
The application of the NHEJ‐based loss of function strategy has also been important in biopharmaceuticals
