Corrigendum to GFP to BFP Conversion: A Versatile Assay for the Quantification of CRISPR/Cas9-mediated Genome Editing

To the Editor: Genome editing via programmable endonucleases enables us to generate site-specific double-strand breaks at virtually any position in a target genome.1,2,3 Exploiting cellular repair mechanisms, this can be used for targeted gene disruption via nonhomologous end joining (NHEJ) or for the precise manipulation of a target sequence through homology-directed repair (HDR) in the presence of a suitable DNA template. The latter carries great promise for the field of gene therapy as it can be utilized for the correction of disease-causing mutations. Earlier this year, De Ravin et al. reported HDR rates >50% in human hematopoietic stem and progenitor cells using zinc finger nucleases, demonstrating that therapeutic levels of gene correction can be achieved in clinically relevant cell types.4 However, the efficiency of HDR remains considerably lower than that of NHEJ in many experimental settings and a background of mutagenic NHEJ is currently limiting the usefulness of genome editing for gene therapy approaches. This limitation signifies a need to identify conditions that bias genome editing toward HDR. Strategies have been developed to encourage HDR over NHEJ, including stimulation with small molecules and inhibition or disruption of DNA ligase 4 activity, but optimal conditions still need to be established.5,6,7 Reliable quantification of HDR and NHEJ is essential to the identification of conditions that favor HDR over NHEJ. This was first achieved through the generation of single-cell clones,2 which is impractical for the determination of overall NHEJ and HDR frequencies. The Traffic Light Reporter system provided the first fluorescence-based assay for the simultaneous quantification of HDR and NHEJ.8 However, this system requires the generation of reporter cell lines and therefore can not be applied easily in primary cells or animal models. Sophisticated methods such as single molecule real time sequencing or sib-selection/droplet digital polymerase chain reaction allow for the quantification of HDR and NHEJ at endogenous loci without the necessity of generating individual clones.9,10 However, downstream sample processing requirements limit the use of these techniques in a high-throughput format. As an alternative, we propose a simple strategy for the simultaneous quantification of HDR and NHEJ by targeting the ubiquitous enhanced green fluorescent protein (EGFP) fluorescent reporter (Figure 1a, b). In 1994, Heim et al. discovered that a single base substitution (196T > C) in the chromophore of wild-type (wt) GFP could shift its fluorescence absorption and emission toward the blue spectrum, thus creating blue fluorescent protein (BFP).11 Here, we demonstrate that EGFP can be converted into BFP in EGFP-expressing cell lines using the clustered regularly interspaced short palindromic repeats (CRISPR) and CRISPR-associated protein 9 (Cas9) system. HDR and NHEJ can subsequently be quantified as blue fluorescence and loss of fluorescence, respectively. K562 cells carrying an EGFP-modified human β-globin locus in the AAVS-1 site in chromosome 19,12 and HEK293T cells that were stably transduced with an integration competent lentiviral EGFP expression construct (K562-50 and HEK293T-EGFP, Figure 2a) were used in this study. Two guide RNA (gRNA) vectors based on px330-IRES mCherry were designed to target Cas9 into close proximity to the target site (Figure 1b). A double-stranded BFP PCR reaction product amplified from the vector pLMP (primers 5′-CCTGAAGTTCATCTGCACCACC-3′ and 5′-GACGTAGCCTTCGGGCATGG-3′) was compared with two single-stranded repair templates (ssODN) (Figure 1c). The gRNA/Cas9 plasmids (5 µg) and HDR templates (100 pmol) were coelectroporated into the target cells using a BioRad Gene Pulser II electroporator. GFP and BFP fluorescence were assessed 10 days later using flow cytometry. HDR and NHEJ were quantified as the percentage of BFP+ cells and nonfluorescent cells, respectively. HDR/total editing ratios (R) were determined using the formula: R = (HDR)/(NHEJ + HDR) * 100. Our data shows that a single 196T > C substitution using ssODN1 is sufficient to convert GFP to BFP. However, low fluorescence intensity and a low HDR frequency were observed in comparison with the other templates in K562-50 cells (Figure 1d,e). An additional 194C > G substitution in ssODN2, corresponding to a reversion of the EGFP amino acid sequence back to that of wild-type GFP, was sufficient to restore BFP fluorescence intensity to that observed with the PCR template (Figure 1e). The low HDR frequency observed with ssODN1 was theorized to result from recutting of the repaired sequence by Cas9, as the sequence resulting from HDR retains the complete target sequence for gRNA1 and contains only one mismatch in the gRNA2 target site. The 194C > G substitution introduces an additional mismatch in the gRNA2 target site and eliminates the gRNA1 protospacer adjacent motif sequence. To further reduce the target sequence similarity with gRNA1 after HDR, ssODN2 was designed with an additional silent mutation (201C > G). In accordance with our expectations, the highest HDR frequency was achieved with ssODN2 in both K562-50 and HEK293T-EGFP cells (5.8% and 23.3%, respectively, Figure 2a). No significant difference was observed between sense and antisense configuration of ssODN2 (Figure 1d). The assay was validated through sequencing of clones grown from the GFP+, BFP+, and nonfluorescent populations after editing with gRNA1 and ssODN2 in K562-50 cells (Figure 2b). In summary, we demonstrate that GFP to BFP conversion is a reliable and simple method for the quantification of HDR and NHEJ. The high sensitivity of the GFP chromophore region to single amino acid deletions demonstrated by Arpino et al. supports our hypothesis that even +3 and −3 insertions/deletions can be detected as loss of fluorescence.13 We have applied this to the optimization of a HDR template for GFP to BFP conversion and verified the strategy through sequencing. While we used EGFP+ cells as targets, wt GFP may also be a target in place of EGFP. This strategy could be used in a high-throughput screen to identify conditions that enhance HDR frequency (Figure 2c). In order to address mechanistic differences that uniquely affect genome-editing rates dependent on the donor template used, the screen can easily be adapted to use a different template type (e.g., dsDNA, adeno-associated virus). The abundance of EGFP-expressing cell lines and animal models permit the application of this strategy for optimization of HDR in a wide range of primary and transformed cells for the establishment of in vivo gene repair strategies. The authors acknowledge the financial support from the National Health and Medical Research Council, the Murdoch Childrens Research Institute, the Victorian Government’s Operational Infrastructure Support Program and Thalassaemia Australia. Kwaku Dad Abu-Bonsrah (Murdoch Childrens Research Institute, Parkville, Australia) kindly provided the px330-IRES-mCherry plasmid encoding Cas9 and the gRNA scaffold. The authors declare that they have no disclosures.