Genome editing efficiency based on native DNA length
In order to easily evaluate the possibility of genome editing using ssODNs, human cells (HEK293T) with an inactivated EGFP cassette (Figure S1) were generated by introducing a cassette gene (Fig. 1A) containing eight foreign bases not required for EGFP. Genome editing was then attempted by lipofecting cells with 20–100 nt ssODNs (Table 1 and Fig. 1A) designed to encode the native GFP sequence, which does not contain the eight bases introduced into the inactivated EGFP. Following transfection, genomic DNA was extracted from the transfected cell population and deep sequenced by NGS to determine the percentage of deletion of the added 8-base sequence. Genome editing efficiencies for 8-base deletions were less than 0.0001%, 0.0002%, 0.0019%, 0.0091% and 0.0154% for 20, 40, 60, 80 and 100 nt, respectively. Thus, although the efficiency was not exceptionally high, site-specific editing could be performed by introducing ssODNs into cells, with the efficiency increasing as the length of the transfected ssODNs increased (Fig. 1B). Flow cytometry analysis of a subset of the transfected cell population also revealed, as expected, an increased number of cells expressing EGFP fluorescence following genome editing as a function of the length of the transfected ssODNs (Fig. 1C). However, the efficiency of this assay was consistently lower than that of NGS. Unlike NGS analysis, which directly examines the sequence of a normally edited genome, flow cytometry evaluates a phenotypic expression system that counts fluorescent proteins normally expressed after genome editing. Therefore, it may not detect cells in an insufficiently expressed state. Nevertheless, flow cytometry analysis is more convenient for this study, in which reproducibility needs to be confirmed in a large number of cells, due to its fast turnaround time and much simpler analysis process compared to NGS. Therefore, in future studies, we used this method to investigate the conditions of the ssODN method before incorporating NGS analysis for the more detailed studies required (Figure S2).
These studies confirmed the benefits of increasing ssODN length to improve the efficiency of genome editing using native DNA; however, there is a limit to the length of ssODNs that can be stably synthesised and delivered. In addition, increasing the chain length reduces the efficiency of ssODN incorporation into the nucleus. It is thought that there is a practical limit to the length of the chain that can be increased to significantly improve the efficiency of genome editing. We therefore decided to investigate whether the use of LNA, which is expected to improve binding affinity with genomic DNA, would facilitate genome editing compared to the introduction of unmodified ssODNs, even when the ssODNs have the same strand length.
Genome editing efficiency depending on the number of LNAs introduced
In general, it has been reported that instead of placing LNAs in a contiguous arrangement, inserting an appropriate number of DNAs between LNAs enhances their editing efficiency in genome editing using ssODNs (19, 26). Based on this fact, LNAs were introduced at both ends (3′ and 5′) of 80 nt ssODNs, a length suitable for detecting genome editing efficiency. Three variations were created: ssODNs containing 10 (80-10L), 12 (80-12L) and 14 (80-14L) LNAs to investigate the effect of LNA introduction (Fig. 2, Group 2).
In Fig. 2, the efficiency of unmodified 80nt native ssODN is denoted as 1 to facilitate comparison of length and genome editing efficiency. The LNA results are shown in Fig. 2, Group 2, and the editing efficiency with native ssODNs is also shown in the same graph (Fig. 2, Group 1).
As a result, 80-10L showed 1.23 times higher efficiency than native DNA of the same length, and 80-12L with 12 LNAs showed 1.5 times higher efficiency, equivalent to that of 100 nt native DNA. In particular, 80-14L with 14 LNAs produced more than twice the efficiency of 80-12L and a 3.77-fold increase in editing efficiency compared to 80-nt native DNA. These results suggest that the introduction of 10 or more LNAs at both ends of an 80-nt ssODN significantly increases the efficiency of genome editing as the number of LNAs introduced increases.
Efficiency of genome editing according to the position of LNAs
LNA introduction was also effective for relatively long ssODNs, with the effect being particularly pronounced in 80-14L, where an additional pair of LNAs (2 in total) was introduced within 80-12L (Fig. 2, Group 2). Indeed, the binding affinity and stability of ssODNs vary depending on the number and position of LNAs [14, 21, 22, 25]. However, no data have been reported on the optimal position of 60-nt or longer ssODNs for genome editing in human cells. Therefore, we investigated the effect of LNA modification position on genome editing.
Differences in genome editing efficiency according to LNA position were compared using multiple ssODNs with the innermost LNA changed while maintaining the length and number of LNAs at 80-14L, which had the highest editing efficiency in previous studies (Fig. 2, Group 3). Interestingly, the results showed significant differences in editing efficiency depending on the location of the additional LNAs, even when the same number of LNAs were introduced. These characteristics could be classified into three patterns, as described below: LNA positions that increase editing efficiency (80-12L-50c2L and 80-12L-40c2L), LNA positions that have minimal effect on editing efficiency (80-12L-10c2L and 80-12L-c2L), and regions that decrease editing efficiency (80-12L-30c2L and 80-12L-20c2L).
In particular, when LNAs were introduced 25 nt from the centre of the ssODN (80-12L-50c2L) and 20 nt from the centre (80-12L-40c2L), their efficiency was more than double that of 80-12L and almost equal to that of 80-14L, which introduced LNAs at 27 nt from the centre of the ssODN. Genome editing efficiency was significantly increased when LNA was introduced into the region 20–27 nt from the centre of the ssODN (Fig. 2, Group 3).
Subsequently, 80-12L-10c2L and 80-12L-c2L, in which LNAs were introduced closer to the centre, showed almost identical editing efficiencies to 80-12L without any effect of LNA integration. In contrast, 80-12L-30c2L and 80-12L-20c2L, in which LNAs were introduced 15 nt and 10 nt from the centre, respectively, showed lower efficiency than native DNA of the same length, despite the 14 LNAs introduced in ssODNs. Thus, the introduction of additional LNAs in inappropriate locations conversely hinders genome editing, suggesting that it is very important to consider the optimal location of LNA modifications (Fig. 2, Group 3).
This observation suggested that the position of LNAs in the ssODN, rather than the absolute number of LNAs introduced, may play an important role in improving genome editing efficiency. Therefore, we further investigated genome editing efficiency using ssODNs of different lengths with 10 LNAs at the ends (Fig. 2, Group 4).
Interestingly, 90 nt (90-10L) and 100 nt (100-10L) sequences showed lower editing efficiencies than native DNA of the same length. Conversely, 60 nt (60-10L), 70 nt (70-10L) and 80 nt (80-10L) showed higher editing efficiencies than their corresponding native DNA sequences of the same length (Fig. 2, Group 4). Among these, a significant increase in editing efficiency was observed for 70-10L, which was 8.5 times higher than that of native DNA of the same length (DNA-70) and much more effective than conditions involving longer lengths with more LNA introduced (e.g. 80-14L, 80-12L-50c2L and 80-12L-40c2L). In 70-10L, LNA was introduced 25–35 nt away from the ssODN centre, which partially overlapped with the 20–30 nt region where editing efficiency had previously increased with the introduction of additional LNA. In addition, 60-10L introduced LNAs at the same position (20–30 nt from the centre of ssODN) where an increase in editing efficiency was observed that was higher than native DNA of the same length (Fig. 2, Group 1, DNA-60). However, in the present study, it was concluded that the 60 nt ssODN was not long enough for genome editing; hence the effect of LNA in this length ssODNs was not pronounced. Therefore, it is clear that in order to increase the efficiency of genome editing methods using ssODNs, it is necessary to use ssODNs of appropriate length, taking into account the position of LNA introduction.
Optimal placement of LNA in ssODNs to improve editing efficiency
In order to investigate the ssODNs that show the most effective genome editing efficiency, we used 70-10L, which has the highest editing efficiency based on the results of the experiments confirmed so far, as a standard and performed the following studies: 1) addition of LNA within ssODN, 2) deletion of LNA at the end sides of ssODN, 3) adjustment of the length of the nucleic acid (Fig. 2, Group 5).
In the experiment with 80 nt, LNA was added at the same position for 70-10L because the editing efficiency was improved by adding LNA at a distance of 20–25 nt from the centre of ssODN. As a result, compared to 70-10L, the editing efficiency of 70-10L-40c2L was approximately doubled just by adding a pair of LNAs inside each arm (Fig. 2, Group 5). This result shows the highest editing efficiency among the ssODNs to which the identified LNA has been added so far, and the addition of LNA to this region has always been shown to be effective in increasing genome editing efficiency. From these results, it was considered important for efficient genome editing to introduce a sufficient number of LNAs at appropriate intervals at positions 20–35 nt from the centre (ssODN), specifically five pairs at positions 25–35 nt from the centre and also one pair at positions 20–25 nt.
To check the validity of the position and number of LNAs introduced, an ssODN was created by removing one outermost and one innermost LNA from this ssODN. The editing efficiency of 70-(10 − 2)L-40c2L, in which the outermost LNAs were removed from 70-10L-40c2L, decreased dramatically. In addition, the editing efficiency of 70-(10 − 2)L, in which the innermost LNA was removed from 70-(10 − 2)L-40c2L, was lower (Fig. 2, Group 5). In other words, it was again shown that the LNAs in the region 25–35 nt from the centre, which is the position of the 70 nt end, play an important role and maintain high efficiency in cooperation with the LNAs 20 nt from the centre added with 70-10L-40c2L.
Next, to confirm whether the introduction of the terminal LNA is important for the efficiency of genome editing, as is commonly believed, we added 10 nt native DNA to both ends of 70-10L-40c2L to create 90 − 70 type-10L-40c2L, which has an LNA at the same position but is not terminally modified and measured its genome editing efficiency. The results showed that even if the LNA modification position was not changed and the strand length was increased so that there was no LNA at the end, no decrease in efficiency was observed and the genome editing efficiency was 1.5 times higher than that of 70-10L-40c2L, indicating that the introduction of the terminal LNA is not essential for editing efficiency in the using ssODN method (Fig. 2, Group 5). This result suggests that the presence of LNA at the above position is more important than the presence of LNA at the end of ssODNs.
Effect of LNA-introduced ssODN on base substitution
The introduction of LNAs has succeeded in deleting 8 bases from target genes as intended, with 17.95-fold efficiency (Fig. 2, Group 5, 70-10L-40c2L) compared to native DNA of the same length (Fig. 2, Group 1, DNA-70). In actual therapy, however, it is necessary to replace precisely the gene mutation that causes the genetic disease. Therefore, we decided to investigate the possibility of single nucleotide replacement by ssODNs, using the ssODNs that showed high genome editing efficiency in the previous experiment. The cell used in the experiment, '80stop EGFP cell' (Figure S3), is a cell in which mutant EGFP has been introduced into the genome by converting the 80th amino acid Lys (AAG) to a stationary codon (TAG). Using the same ssODN as in the previous experiment, it is possible to replace the T with an A instead of a deletion. The results of the efficiency comparison based on DNA-80 are shown in Fig. 2 (Group 4 and 5) and each specific base substitution frequency is shown in Fig. 3. The efficiency of base substitution was approximately 173.48-fold for 70-10L, 240.01-fold for 70-10L-40c2L and 280.92-fold for 90-70type-10L-40c2L when DNA-80 was set to 1. This was 33.95-fold, 22.28-fold and 16.61-fold higher than the efficiency of the corresponding 8-base deletions. As a result of confirming the efficiency of nucleotide sequence substitution using NGS (Fig. 3), which exceeded 0.42% for 70-10L, 0.52% for 70-10L-40c2L and 0.63% for 90 − 70 type-10L-40c2L, it was confirmed that nucleotide substitution was quite efficient. Thus, the efficiency of genome editing with LNA-introduced ssODNs was consistently higher for substitutions than for deletions, demonstrating that genome editing with ssODNs is an excellent genome editing method even for single nucleotide substitutions.