Genetically engineered mice are valuable subjects for developmental and pathomechanism studies. However, the traditional gene targeting method through embryonic stem cells (ESCs) has been time-consuming and costly. In 2013, the Jaenisch group introduced conducting gene modified mice in a one-step generation using clustered regularly interspaced short palindromic repeats (CRISPR)-CRISPR associated protein 9 (Cas9) genome engineering technology (1, 2). Since the CRISPR/Cas9-mediated system originated from the prokaryotic immune system (3–6), it enabled rapid and efficient genome editing in mammalian cells (7–11).

This system opened a new era in genome biology fields including animal, plants, and human genetic disease (12–15). Programmable endonuclease Cas9 with guide RNA (gRNA) induce DNA double-strand breaks (DSBs) on the target DNA sequences, and DSBs are repaired by non-homologous end-joining (NHEJ) or homology-directed repair (HDR) pathway, mainly (16–18). Among them, NHEJ is a predominant repair mechanism in higher eukaryotic cells or organisms. Therefore, after DSBs, NHEJ works dominantly and generates small insertions or deletions (indels), resulting in frame shifts at target genes eventually (19–21). Taking advantage of these characteristics, the efficient knock-out study through NHEJ pathway has been developed extensively in the genome editing field. However, since the NHEJ repair mechanism induces uncontrollable random mutations on target loci, NHEJ conjugated technologies showed limitations for precise genome editing, such as designated insertions and single-nucleotide substitutions (2, 22).

To overcome these limitations, many scientists made an effort to develop methods to insert donor template DNA using the HDR pathway, to perform precise gene editing. However, it was difficult to use HDR mechanism in gene editing unrestricted because of its extremely low efficiency. In mammalian cells, NHEJ is the major source of the DNA repair mechanism competing with the HDR pathway. Therefore, for more efficient HDR-mediated precise genome editing, numerous researchers have attempted to enhance HDR pathway or/and suppress NHEJ pathway by targeting key factors (23–25).

Recently, a new technology called base editors (BEs) has been introduced to overcome low accuracy of NHEJ and low efficiency of HDR. These powerful editing tools can change single nucleotide without DNA DSBs in cells (26, 27). BEs are composed of catalytically impaired Cas9 variant with deaminase classified as cytosine base editors (CBEs) and adenine base editors (ABEs), allowing direct conversion from C to T or A to G (28–30). Recent reports showed that various applications using base editors enable single nucleotide substitutions in mammalian genome successfully (31–35). Although it is clear that base-editing technique is an innovative development, limitations remain in the case of single base substitution, as well as insufficient accuracy/efficacy in vivo.

In this review, we will report recently developed methods for precise gene editing as enhanced HDR-mediated gene engineering and direct base editing in mammal species. Diverse strategies to increase HDR efficiency are introduced. One is optimization of the HDR pathway by controlling the length of homology arms of template donor DNA. Another is the inhibition of NHEJ pathway which competes with HDR. Additionally, we also introduce BEs, a method for tailored single nucleotide substitution.

The most precise genome editing method is utilizing HDR mechanism to insert artificial DNA sequences to target locus or to induce single-nucleotide substitutions. However, the efficiency of HDR pathway in nature is extremely low (2, 36–38). Recently, several studies reported new methods to overcome low efficiency by optimizing template donor DNA. Researchers modulated the length of homology arms and types of donor DNA, such as single strand DNA (ssDNA) or double strand DNA (dsDNA) (Table 1). Renaud et al. explained that using single-stranded oligo DNA nucleotides (ssODNs) as template donors with chemical modifications such as phosphorothioate or LNA could improve precise knock-in efficiency, rather than using double-stranded oligo DNA nucleotides (dsODNs) (39). Paquet et al. delivered ssODN donor templates which comprise silent mutations. These mutations prevented re-cleavage of inserted sequences by CRISPR/Cas9 and increased precise knock-in efficiency (40). Easi-CRISPR was reported as a new method to generate mutant mice efficiently with insertion of exogenous artificial DNA sequences. DNA donors were prepared as ssODNs approximately 1 kb long. They delivered directly components such as ssODN donor templates, gRNAs, and Cas9 mRNA, into mouse zygotes using microinjection. They also successfully generated knock-in mice using CRISPR ribonucleoproteins (RNPs) (41). Some research groups attempted to modify Cas9 protein and gRNAs to increase HDR efficiency. Most recently, the Rossant group has shown that combining two-cell homologous recombination (2C-HR)-CRISPR with a modified biotin-streptavidin approach in mice, can increase knock-in efficiency over standard methods by more than 10-fold (up to 95 %) (42). The Gordon group demonstrated that HDR efficiency could be increased up to 30-fold using Cas9 and Porcine Circovirus 2 (PCV) Rep fusion protein delivered with ssODNs containing 13 bp PCV recognition sequences at 5'-end (43). Other groups attempted NHEJ or microhomology-mediated end-joining (MMEJ) mediated knock-in, to insert exogenous DNA sequences more efficiently to the target loci, instead of HDR pathway requiring shorter homology arms compared with HDR-mediated. A new knock-in method using MMEJ pathway, termed the precise integration into target chromosome (PITCh), was reported. They generated vectors exquisitely, which contain short micro-homology sequences approximately 5–25 bp, and enabled insertion of large DNA fragments to the target sites of various cell lines and organisms (44, 45). Also, Yao et al. successfully knocked in tagging sequences in-vivo and ex-vivo by MMEJ-mediated manner. Donor DNA sequences contain short homology arms including microhomology sequences (46). Also, they reported a new method, called homology-mediated end-joining (HMEJ) strategy. The vector for HMEJ based knock-in contains CRISPR-Cas9 cleavage sites, identical to target sequences on the genome, and approximately 800 bp-long homology arms. These methods were tested in mouse and monkey embryos and showed greater results than HDR, NHEJ, and MMEJ mediated knock-in efficiency (47). Most recently, Yao et al. demonstrated Tild-CRISPR (targeted integration with linearized dsDNA-CRISPR). They provided donor DNA with 800 bp-long homology arms by PCR-amplification. This method is based on HMEJ strategy and has advantages in preparing template donor DNA by PCR, efficiently. They claimed that it shows high integration efficiency for both, in vitro (mouse/human embryo cell) and in vivo (mouse brain) scale (48). Representative studies are summarized at Table 1.

NHEJ mediated genome editing induces random mutations such as small indels on target sites. Therefore, these kinds of mutations led the frame shift on targeted genes and is proper for knock-out studies but not for inducing precise mutations, such as point mutations or knock-in studies. Conversely, HDR repair system is good in generating precise point mutations and for inserting external artificial DNA sequences. However, low efficiency has always been a major obstacle to broad use. A number of studies have attempted to increase HDR efficiency by regulating DSBs repair mechanisms (Fig. 1A). It is well known that NHEJ and HDR pathways are in competition (49–51). Several studies have shown that suppression of key molecules involved in the NHEJ pathway could increase efficiency of HDR. Many proteins are known to be relevant with NHEJ pathway including Ku heterodimers (Ku70/80), DNA-dependent protein kinase catalytic subunits (DNA-PKcs), DNA ligase IV, the X-ray repair cross-complementing protein 4 (XRCC4), and the XRCC4-like factor (XLF) as core complexes (52–54). Among these related proteins Chu et al. suppressed DNA ligase IV by Scr7, a DNA ligase IV inhibitor, and adenovirus 4 E1B55K and E4orf6 proteins, inducing proteasomal degradation of DNA ligase IV. HDR efficiency increased 4–5-fold or 8-fold, respectively (55). Also, Maruyama et al. showed that treatment of Scr7 in a mammalian cell line and mouse zygotes increase HDR efficiency approximately 19-fold (56). Yu et al. identified small molecules, L755505 and Brefeldin A. The function of these molecules in NHEJ pathway has not been clarified. However, both small molecules enhanced HDR efficiency approximately 2–3-fold for large fragment knock-in and 9-fold increase for inducing point mutation, respectively (57). Risenberg et al. identified effective small molecules to increase HDR efficiency in human induced pluripotent stem cells (hiPSCs) by screening of small molecules related with DNA repair mechanisms. The combination of small molecules termed CRISPY mix containing NU7026, Trichostatin A, MLN4924, and NSC 15520 showed the most effective HDR efficiency. Also, the related small molecules affected key molecules of major DNA repair mechanisms (58).

Major DNA repair pathways, NHEJ and HDR are not always activated during all cell cycle stages. NHEJ dominates over all M, G1, S, and G2 phases, while HDR can only compete with NHEJ, during S and G2 phases. HDR is down regulated during M phase and G1 phase (59–61). Various small molecules exert their effects by controlling such stages in part (Fig. 1B). Li et al. re-tested the function of Scr7 and L755505, in porcine fetal fibroblast. Additionally, resveratrol, a novel small molecule in this field was also tested. Scr7 and L755505 in porcine fetal fibroblast led a 2-fold increase similar as tested in other cell lines, and the resveratrol could raise approximately 3-fold in porcine fetal fibroblast. It is also reported that L755505 and resveratrol could arrest cells at S phage, wherein the HDR mechanism is activated. Treatment of three molecules such as Scr7, L755505, and resveratrol, up-regulated mRNA expression level of HDR key factors, such as BRCA1, BRCA2, RPA3, SPIDR, NBN, RAD50, RAD51, and RAD52, and down-regulated key molecules of NHEJ pathway such as LIG4, MRE11, DCLRE1C, and XRCC4 (62). Also, multiple researchers identified small molecules that affect cell cycle arrest to increase HDR. Nocodazole and Lovastatin synchronize the cell cycle in G2/M phase and early G1 phase, respectively. Lovastatin also inhibits at G2/M phase, partially. Mimosine, aphidicolin, thymidine, and hydroxyurea arrest cells at between G1 phase and S phase, before DNA replication (61, 63). Recently, Canny et al. regulated another key factor: 53BP1. It is significant at the beginning of the repair mechanism between NHEJ and HDR pathways on the DSBs loci. The 53BP1 blocks DNA end resection and recruitment of BRCA1 to DSBs. This study has shown that the 53BP1 inhibitor, i53, can increase HDR efficiency (64). Song et al. reported applying RS-1 could increase HDR efficiency by stimulating Rad51. Unlike previously reported studies, in which small molecules were used to inhibit the NHEJ pathway, this study used a small molecule, RS-1, to promote the HDR pathway (65). Most of the cases of treatment of small molecules are focused on suppression of NHEJ pathways since both repair mechanisms are in competition.

More than 50% of human pathogenic mutations are point mutations or single nucleotide polymorphisms (SNPs) (26). As the importance of precise medicine arises, accurate single nucleotide substitutions in the genome have been required for pathology or mechanistic studies. However, in the beginning of the CRISPR technology, specific nucleotide substitutions at desired target sites could only be induced by an HDR-based CRISPR system, despite its low efficiency. To overcome such limitation, new tools called Base Editors (BEs) were developed to induce single-nucleotide substitution, which do not need a template donor DNA (Fig. 2A and 2B) (28–30). Because these techniques do not introduce DSBs, they never use DNA repair mechanisms as NHEJ, MMEJ, or HDR pathways. BEs were composed of nuclease activity deficient Cas9, nickase Cas9 (nCas9) or dead Cas9 (dCas9), and cytidine deaminase or adenine deaminase. They enable conversion of C to T, or A to G, and vice versa. They are newly-developed methods not affected by HDR efficiency in case of inducing substitutions. These tools were verified through various research groups and applied to many other organisms, including mice and rabbits (31, 32, 66, 67). The substitution efficiency was higher than the HDR mechanism. However, the unique characteristic of BEs, such as base editing window which indicates the specific region occurring substitution, could be a limitation to inducing single-nucleotide substitution to the exact target base pair. So, some researchers attempted to change the base editing window. One study induced some mutations at cytidine deaminase domains to narrow the base editing window for more specific substitutions (68). Conversely, to extend coverage of BE systems, some researchers demonstrated that using the extended guided RNA could extend coverage of BEs and using Cas9 variants with different protospacer adjacent motif (PAM) sequences, such as xCas9 and VQR variants (32, 69, 70). There remain several improvements in the BE system. Accuracy and efficacy have not been satisfied for clinical demands and knock-in of external DNA sequences are impossible.

CRISPR/Cas9 mediated genome engineering applicable to a variety of organisms is crucial as a tool for research and clinical applications. In this review, we showed efforts to increase efficiency of HDR, one of the genetic manipulation strategies, for accurate and specific targeted knock-in. Recent efforts to improve HDR efficiency have focused on controlling the homology arm length, or suppressing the NHEJ pathway using small molecules. In particular, the Tild-CRISPR method, a method of controlling donor DNA homology arm length, is expected to greatly improve the efficiency of HDR. Based on these results, HDR efficiency is expected to be enhanced by combining NHEJ pathway inhibition with small molecules and the control of homology arm length. Additionally, the BEs (nucleotide substitution methods for specific target sites) are expected to be applied to studies of clinical pathology mechanism by allowing tailored point mutation. Recently, development of gene editing technology has suggested the possibility of clinical application as a genetic disease therapeutic agent. However, accuracy of gene correction fails to meet clinical demands and additionally, the stable in vivo delivery system is lacking. To overcome these problems and to apply clinical applications for therapeutic purposes, it is necessary to improve gene editing accuracy/efficiency and develop in vivo delivery systems delivery systems, simultaneously.

This study was supported by the Chung Yang, Cha Young Sun & Jang Hi Joo Memorial fund, Korea university grant (K1804351), and the Bio & Medical Technology Development Program of the National Research Foundation (NRF) of Korea (NRF-2018M3A9H3021707, NRF-2018R1D1A1B07048434, and NRF-2014M3A9D5A01075128).

The authors have no conflicting interests.