The GEIS workflow generates RELA/p65 S276C HEK293T cells within 1 month.
p65 is a REL-associated protein involved in NF-κB heterodimer formation, nuclear translocation, and downstream gene transactivation 13. We applied GEIS to generate the S276C mutation in RELA/p65. A lentiCRISPR-v2 plasmid carrying sgRNA targeting intron 8 of the RELA gene was used to generate DSBs. To avoid disrupting RNA splicing, we did not target the splicing element. The donor DNA template contained a cytomegalovirus (CMV) promoter-driven DsRed-expressing cassette between the left and right homology arms (HAs), while the desired S276C mutation was located on the left arm (Fig. 1A). The lentiCRISPR-v2 plasmid and donor DNA were cotransfected into HEK293T cells for 24 hours, and then puromycin selection was conducted for 72 hours to kill nontransfected cells. The surviving cells were subjected to fluorescence-activated cell sorting (FACS) to isolate the DsRed-positive cells. To increase the selection efficiency, a second round of FACS was performed. The sorted cells were seeded into 96-well plates for single-cell clone growth. We obtained positive cell clones with the S276C mutation within 1 month with this workflow (Fig. 1B, C). Reverse transcription PCR (RT-PCR) and quantitative PCR (qPCR) showed that the inclusion of the CMV-DsRed cassette in the intron neither disturbed the splicing of the two adjacent exons nor affected mRNA transcription (Fig. 1D, E).
HDR with an ssDNA template reduces production of false-positive cell clones.
In this strategy, the use of dsDNA as donor DNA produces false-positive cell clones via direct transcription and translation or via random integration into the genome through canonical NHEJ (c-NHEJ) 14. Recent studies have demonstrated that ssDNA donors show superior performance compared to dsDNA donors in mammalian systems by reducing the probability of NHEJ 15. To effectively obtain ssDNA sequences as large as 5000 nt, we denatured the dsDNA from PCR at 95°C and with 100 mM NaCl for 10 minutes (Fig. 2A). Our data demonstrated that a single-stranded CMV-DsRed donor led to significantly lower fluorescence intensity than a double-stranded donor (Fig. 2B, S1A). We speculated that the use of an ssDNA donor would increase the true-positive rate of FACS-enriched DsRed-expressing cells. As shown in Fig. 2C, with ssDNA, the recombination rate for RELA S276C reached 66.7% (8 out of 12), while with dsDNA, it was only 16.7% (2 out of 12) after two rounds of sorting. Elevated recombination rates were also observed at the NABP2 and EGFR loci, and no abnormal splicing or mRNA changes were detected (Fig. 2C, D, E).
A conversion tract longer than 490 bp is necessary for NABP2 GEIS.
Despite the efficient selection of successfully recombined cells, GEIS still exhibits a low editing efficiency when the conversion tract is too long. Because the CMV-DsRed cassette must be located in an intron to avoid disrupting endogenous gene splicing and expression, the sgRNA target site should usually be intronic, but the expected conversion site is usually exonic. The distance from the DSB to the conversion site (conversion tract) affects the gene editing efficiency 16.
To estimate the influence of conversion tract length on editing efficiency, we first evaluated the HA length required for efficient insertion of the selection cassette into the intron. Using the EGFR locus as an example, we designed a series of donors with 250, 500, 800 and 1000 bp HAs. HAs longer than 500 bp were determined to be necessary for recombination at this locus (Fig. S2 A, B). Next, we designed donor DNA with a left HA (800 bp) containing nucleotide variations 45, 90, 171, 281, 490, 596 and 696 bp away from the DSB site for GEIS of NABP2 (Fig. 3A). The genomic DNA of the GEIS-processed cell group was PCR-amplified with the forward primer located outside the left HA on the genome and the reverse primer at the DsRed cassette. The PCR product was cloned into pLV-MCS-puro-Green for Sanger sequencing. A total of 624 clones were sequenced, and the conversion efficiency was calculated. The results showed that the conversion efficiency decreased as the tract became further away from the DSB. In contrast to previous research based on 80 cell clones, which reported an efficiency of only 20% when the tract was 200 bp long 16, our data showed that nearly 60% of recombinants were converted at the 490 bp site (Fig. 3B).
Because introns adjacent to the target exon on both the left and right sides are available for GEIS DSB generation, a nearer intron can always be found for the exonic editing site for GEIS, which needs less than half of the exon length as the conversion tract. The conversion tract of 490 bp indicated that GEIS has an approximately 60% probability of generating mutations for exons as large as 980 bp in the locus (Fig. S2C).
To assess the applicability of GEIS in the human and mouse genomes, we analyzed the distributions of exon length in these two species from the Consensus CDS (CCDS) Project (Fig. S2D) 17–19. Most exons longer than 1000 bp were the first or last exons, which contain long 5’ or 3’ untranslated regions (UTRs); however, DSBs can still be introduced by sgRNA in the first or last intron. When we excluded the UTRs and reanalyzed the distribution of exon lengths, only approximately 1% of exons had lengths greater than 1000 bp (Fig. 3C). Based on the conversion tract analysis from the NABP2 locus, we speculate that GEIS might be able to edit 99% of gene targets with relatively high efficiency.
GEIS has the potential to introduce multiple DNA variations.
To evaluate the possibility of introducing multiple genome alterations in one GEIS reaction, we analyzed the mutation distributions in each of the 624 clones. A heatmap was created to show the percentage of alterations that occurred at the remaining sites (horizontal axis) when an alteration occurred at the indicated site (vertical axis) (Fig. 3D). According to the map, mutations at a further site largely indicated successful editing of the nearer site, and genome editing showed a high extent of linkage rather than independence, indicating that multiple genome alterations can be introduced in one GEIS reaction.