Multiplexed CRISPR/Cas9 genome editing increases the efficacy of homologous-dependent repair of donor sequences in mammalian cells

AFFILIATIONS: 1Medical Oncology Research Unit, Department of Internal Medicine, School of Clinical Medicine, University of the Witwatersrand, Johannesburg, South Africa 2Gene Expression and Biophysics Group, Synthetic Biology–Emerging Research Area, Council for Scientific and Industrial Research, Pretoria, South Africa 3Unit of Biophysics and Gene Expression, Institute of Molecular Medicine, Faculty of Medicine, University of Lisbon, Lisbon, Portugal 4Antiviral Gene Therapy Research Unit, Department of Molecular Medicine and Haematology, School of Pathology, University of the Witwatersrand, Johannesburg, South Africa 5HIV Pathogenesis Research Unit, Department of Molecular Medicine and Haematology, School of Pathology, University of the Witwatersrand, Johannesburg, South Africa 6Department of Molecular and Experimental Medicine, The Scripps Research Institute, La Jolla, California, USA


Introduction
Precise genome editing represents a powerful new paradigm for both forward and reverse genetics studies in model systems.2][3] The error-prone non-homologous end joining (NHEJ) repair pathway involves the re-ligation of the broken DNA ends, and in the process introduces small mutagenic insertions and deletions (indels).5][6] These tools have enabled functional studies based on the systematic knockout and knockin of genes 7 , the modelling of human diseases in cell-or animal-based systems [8][9][10] , and the generation of isogenic cell lines for stable transgene expression 11,12 .
Conventional genome editing strategies make use of a single cleavage event to stimulate gene conversion with either ssDNA or dsDNA homologous repair templates. 368][39] This situation can be partially resolved by the simultaneous introduction of two separate DSBs at a targeted locus to induce deletion and replacement events over large genomic regions. 26,27,40However, the efficiency of this approach has not been determined and, to date, the HDR frequencies among approaches that apply single or multiplexed DSBs remain unknown.Using the CRISPR/Cas9 adopted from Streptococcus pyogenes, we exploited the capabilities of this system to induce DSBs to introduce an oncogenic KRAS (c.35G>T) mutant variant in a selectable human cell-based model system.We demonstrate that the facile introduction of two simultaneous DSBs improves the frequency, relative to a single DSB, of reconstituting dsDNA homologous donor sequences, but not ssDNA ODNs (oligodeoxynucleotides), at the desired locus.These results point to an improved approach for efficient HDR-mediated repair using the facile CRISPR/Cas9 editing system.
Each pair of oligonucleotides was composed of a 20-nucleotide guide sequence for their target and appropriate 5ʹ overhangs that allowed for cohesive-end ligation into the BsmBI (Thermo Scientific, USA) digested pcDNA.H1.sgRNA expression plasmid.Sanger sequencing was performed to validate the insertion of the guide sequence into the sgRNA expression plasmid.

T7 Endonuclease I assay
At 24 h prior to transfection, HEK 293 cells were seeded at a density of 250 000 cells/well in a six-well plate.The cells were transfected with 2 μg of the sgRNA expression plasmid and 2 μg of the Cas9 expression plasmid (pX330) (Addgene plasmid #42230) 16 using 12 μL of branched polyethylenimine (Sigma Aldrich, USA).At 48 h posttransfection, the cells were harvested and the gDNA was extracted from these cells using the QIAamp DNA Mini Kit (Qiagen, Germany).Genomic regions surrounding the CRISPR/Cas9 target sites were amplified from this gDNA with primers for the 5ʹ cleavage region (5ʹ KRAS forward: 5ʹ -CGCAGAACAGCAGTCTGGC -3ʹ and 5ʹ KRAS reverse: 5ʹ -CTACGCCACCAGCTCCAAC -3ʹ), and the 3ʹ cleavage region (3ʹ KRAS forward: 5ʹ -AAGGCCTGCTGAAAATGACTG -3ʹ and 3ʹ KRAS reverse: 5ʹ -GCACAGAGAGTGAACATCATGG -3ʹ).The T7 Endonuclease I (T7E1) assay was performed as previously described. 36,41Briefly, 600 ng of DNA amplified from the genomic target region was denatured and annealed to form heteroduplexes.Amplicons (150 ng) were digested with 3 U of T7E1 (New England Biolabs, Germany) and resolved on an 8% polyacrylamide gel stained with SYBR Green I Nucleic Acid Gel Stain (Invitrogen, USA).

Micro-deletion detection
HEK 293 cells were transfected using polyethylenimine with 2 μg of the 5ʹ cleavage sgRNA expression plasmid, 2 μg of the 3ʹ sgRNA expression plasmid and 2 μg of pX330.gDNA was extracted using the QIAamp DNA Mini Kit (Qiagen, Germany) 48 h post-transfection.Polymerase chain reaction (PCR) amplification of the deletion junction was performed using Phusion Flash High-Fidelity PCR Mix (Thermo Scientific, USA) with an initial denaturation at 98 °C for 10 s, followed by 35 cycles of denaturation at 98 °C for 1 s, annealing at 65.4 °C for 5 s, extension at 72 °C for 10 s and a final extension at 72 °C for 1 min using the 5ʹ KRAS forward and 3ʹ KRAS reverse primers described in the T7E1 assay.The PCR products were then visualised on an ethidium bromide (0.5 μg/mL) stained TAE agarose gel (2% (w/v)).

Restriction fragment length polymorphism analysis for detection of HDR
HEK 293 cells were transfected with 2 μg of sgRNA expression plasmid, 2 μg of pX330 and 2 μL of the ssODN repair template (100 μM) for the single cut strategy.For the dual cut strategy, 2 μg of each sgRNA expression plasmid, 2 μg of pX330 and 2 μL of the ssODN repair template (100 μM) were used.After 72 h, the gDNA was extracted from these transfected cells and the repaired region was amplified using PCR with the 5ʹ KRAS forward and 3ʹ KRAS reverse primers (described for the T7E1 assay).The PCR products were purified and 300 ng of the DNA was digested with EcoRI (Thermo Scientific, USA).The products were resolved and visualised on a 8% polyacrylamide gel which was stained with SYBR Green I Nucleic Acid Gel Stain (Invitrogen, USA).

dsDNA donor sequences
The ssDNA ODN sequences used were chemically synthesised (IDT, USA) (EcoRI site underlined): The dsDNA donor construct was assembled by fusing five PCR products in an overlap extension PCR: The plasmid pCI-NeoR-EGFP was used as a template for SV40 driven NeoR.Amplicons comprising the 5ʹ and 3ʹ homology arms, KRAS sequence and SV40-Neo-2A-eGFP sequences were assembled by overlap-extension PCR.The dsDNA donor was ligated into the pJet1.2/bluntCloning Vector from the CloneJet PCR Cloning Kit (Thermo Scientific, USA) and subsequently sequenced.The dsDNA donor was linearised by PCR amplification using Phusion Flash High-Fidelity PCR Mix (Thermo Scientific, USA) and the 5ʹ HA F2 and 3ʹ HA R2 primers, with an initial denaturation step at 98 °C for 10 s, followed by 35

CRISPR/Cas9 can facilitate KRAS mutagenesis
3][44] In order to model an oncogenic c.35G>T variation in codon 12 of KRAS, we identified suitable sgRNA target sites within a 250-bp region around the point mutation. 45he sgRNA guide sequences were chosen on the basis of the best returned score of each target site to minimise off-target DSBs.Two of the target sites were upstream of the c.35G>T mutation (5ʹ cleavage sites) and two were downstream of the mutation (3ʹ cleavage sites) (Figure 1a).A fifth sgRNA that made use of a degenerate NAG protospacer adjacent motif (PAM) was also included (sgRNA F2 225).The H1 Pol-III promoter drove the expression of the inserted guide sequence and the RNA scaffold, as a single chimeric sgRNA molecule comprising an optimised sgRNA architecture 22 (Figure 1b).The distribution of sgRNAs allowed for potential gene deletions to span the mutation site.To determine the functionality of each sgRNA, we used a T7E1 assay to detect the indel frequency at the targeted KRAS locus.All the sgRNAs introduced indels at the DSB repair sites at varying efficiencies (1.28-6.01%)(Figure 1c).Guide RNAs sgRNA F1 224 and sgRNA R2 228 were the most effective, and ensured that effective targeting occurred at positions 5ʹ and 3ʹ of the KRAS mutation site.These sgRNAs were applied simultaneously, resulting in the deletion of 199 bp from the KRAS locus as determined by PCR amplification with primers that flank the deletion junction (Figure 1a  and 1d).

Micro-deletions can be repaired precisely by HDR
In order to precisely modify genes, the CRISPR/Cas9 system needs to be able to stimulate the HDR pathway for the repair of DSBs according to specific exogenous repair templates.ssODNs have been shown to be able to seamlessly repair and modify targeted single DSB sites via HDR. 46hese are convenient repair templates that allow for small, targeted gene modifications to be made without the need for constructing large donor DNA constructs.Based on an optimised design from the Church laboratory 38 , ssODNs harbouring the EcoRI restriction site were designed to be homologous to the sense strand of the KRAS target region and serve as repair templates.Two 90-mer ssODNs were generated to directly flank and repair individual cleavage events generated by sgRNA F1 224 and R2 228 (Figure 2a).To demonstrate the HDR in the presence of two DSBs, a ssODN was designed to have distal homology arms, which served to bridge the resultant gap of the micro-deletion when sgRNA F1 224 and R2 228 were used concurrently (Figure 2a).The ability of the CRISPR/Cas9 to stimulate the HDR of this targeted DNA damage was assessed by restriction fragment length polymorphism analysis (Figure 2b).Guide sgRNA R2 228 facilitated HDR with an efficiency of 9.27%.The micro-deletion introduced by sgRNA F1 224 and R2 228 was efficiently repaired as was determined by the detection of a 451-bp deletion allele by PCR (Figure 2c).Interestingly, an increased abundance of the deletion allele was observed in the presence of the F1/R2 ssODN, suggesting that an oligo bridging the deletion gap facilitated NHEJ repair.
The micro-deletion repair predominantly favoured the NHEJ and not the HDR pathway.

Dual cleavage generates selectable c.35G>T KRAS mutant cells
The application of ssODNs in tandem with targeted endonucleases to HDR-mediated genome editing is limited, even though complex and long-distance editing functions have been described. 39However, to generate selectable recombinant cells, larger repair constructs were needed to insert expression cassettes that will allow for the selection of the modified cells.Here, we employed the CRISPR/Cas9 system in a dual cleavage strategy to introduce a de novo oncogenic c.35G>T point mutation in the human KRAS gene for positive selection.In order to do this without disrupting the gene function, we generated a knockin selection cassette for insertion into an intronic site 94 bp away from the mutation site in exon 2. Conventional strategies make use of a single DSB to stimulate HDR with dsDNA donor construct that is homologous to the sequences flanking a single DSB (Figure 3a).To improve on this approach, a dual cut repair strategy with a large dsDNA donor, using the CRISPR/Cas9 system, was predicted to efficiently facilitate the simultaneous c.35G>T transversion in exon 2 and the insertion of a selectable expression cassette into intron 1 of the KRAS gene.We used sgRNA F1 224 and R2 228 to introduce two concurrent DSBs to delete the intron 1-exon 2 junction, which was then reconstituted with the dsDNA donor (Figure 3a and 3b).To test the performance of the dual cleavage strategy (Figure 3b), a qPCR strategy was devised to quantify the modified KRAS DNA.Genomic DNA was extracted from the entire cell population without marker selection to minimise the enrichment of illegitimate recombinants.Various DNA species were quantified in the cell population 3 and 14 days post-transfection, using primer sets A-D.This quantification allowed for stable genomic integrants to be detected and monitored over time, while the residual episomal DNA from the transient transfection was diluted through continued cell proliferation.Approximately 1.9-2.5-foldmore of the target DNA was detected with the dual cleavage strategy using primer set A, which spanned the gene-donor junction (Figure 3b and 3c) and ensured detection of the integrated dsDNA donor in the KRAS gene locus.Transfections that were sgRNA deficient showed no detectable dsDNA donor integration, confirming that donor DNA integration did not occur spontaneously.
A mock donor failed to be detected, pointing to the specificity of the  dsDNA donor for HDR.The dilution of the episomal DNA was apparent after 14 days, with a decrease of approximately 99% in each sample (primer set B; Figure 3b and 3c).Despite this dilution, readily detectable amounts of episomal DNA were still present in the transfected cells after 14 days of sub-culture (600-1000-fold relative to β-actin).The amount of detected dsDNA donor integrated into the KRAS gene (data from primer set A), taken as a percentage of the total KRAS DNA detected (data from primer set C; Figure 3b and 3c), was used as a metric for site-specific donor integration efficiency (Figure 3d).The simultaneous use of the two sgRNAs proved to be the most efficient at facilitating the integration of the heterologous donor DNA sequences, and occurred at a frequency of 0.028% and 0.123% in transfected cells, after 3 and 14 days of selection-free culture, respectively.The c.35G>T KRAS mutation confers a slight growth advantage in the modified cells. 43e observed an enrichment of modified KRAS mutants as the amount of integrated donor DNA increased after 14 days of selection-free culture of the transfected cell population (Figure 3c and 3d).

Discussion
We used the CRISPR/Cas9 system for the improved efficacy of donor-led HDR using a multiple editing approach which allows for extensive and wide-ranging gene modifications to be achieved while maintaining gene function.As an example, the introduction of a putative oncogenic point mutation in the KRAS gene in a cell-based model was adopted in this study.We initially demonstrated the efficiency of introducing single cleavage events in the KRAS gene sequence using RNA-guided CRISPR/Cas9.These single cleavage events were repaired either by the NHEJ pathway, resulting in the introduction of mutagenic indels, or by the HDR pathway, which seamlessly repaired the DSB by using homologous ssODN templates.The NHEJ and HDR activity induced by However, DSBs mediated by sgRNA R2 228 clearly favoured the HDR pathway for DSB resolution.Furthermore, HDR of the micro-deletion, with a ssODN, occurred at relatively low levels compared to the deletion events detected, indicating that NHEJ repair was more favourable.These repair pathway preferences of DNA damage could be related to cellular mechanisms that prioritise the repair of certain genetic elements. 47Nevertheless, such repair preferences could potentially alter the efficiency of the desired gene editing outcome.Further observation of gene editing at different genomic sites, on a more high-throughput scale, may provide further insight.
We demonstrated that the CRISPR/Cas9 system can be used in multiplex to improve the efficiency of HDR for precise modifications in the KRAS gene.We predicted that by applying two cleavage events, the gene region of interest could be deleted and efficiently reconstituted with our mutant-containing dsDNA donor, which would allow for a more expansive genomic region to be precisely modified, without being limited by the short gene conversion tracts that branch out from a single break site.Furthermore, the design of the dsDNA donor for this strategy removes the point mutation from the homology arms and relocates it into the reconstitution sequence, allowing for complete homology with the target sequence to be maintained.This strategy should improve the efficiency of gene conversion for single, as well as multiple, point mutations, as small levels of sequence divergence in the homology arms have been shown to greatly decrease the levels of recombination. 37o illustrate this, we targeted two distal sites within the KRAS gene locus for concurrent modification, using a single donor.This approach allowed for the concurrent insertion of a selection cassette into a nearby intron of the oncogenic point mutation site in the KRAS protein-coding sequence, preserving the integrity and function of the gene and eliminating the need for any post-modification gene restoration to rescue function.

Conclusions
A dual cut gene editing strategy allows for distally located sites to be targeted simultaneously, enabling for extensive and expansive gene modifications to be achieved efficiently.We have shown that this method is particularly useful for maintaining the native function of the gene, in which selection cassettes used to identify cells with the modifications of interest can be inserted in neighbouring introns.We significantly improved on current HDR-based gene editing strategies, which can only modify regions within close proximity to the DSB site.Furthermore, the CRISPR/Cas9 system is well suited for this dual cleavage approach, as it is simple and cost effective to employ in a multiplex fashion.This method should improve the success rate of producing recombinant cell lines, and thus expedite efforts to gain a better understanding of causal genetic elements in biological systems.

Figure 1 :
Figure 1: CRISPR/Cas9-mediated targeting of the human KRAS gene.(a) A schematic of the KRAS gene locus chosen for CRISPR/Cas9 targeting.The sgRNA target sites were designed around an oncogenic c.35G>T mutation in the KRAS gene, with the cleavage sites depicted as arrowheads.Two primer sets were used to amplify cleavage regions that were 5ʹ and 3ʹ of the point mutation.(b) The sgRNA construct was driven by an H1 Pol-III promoter for the expression of the guide sequence and the RNA scaffold as a single chimeric molecule.The sgRNA architecture used was previously optimised to improve sgRNA expression and Cas9 loading. 22(c) A T7E1 assay showed the performance of each sgRNA at generating indels (n=3; mean±SD, *p<0.05,**p<0.01,unpaired Student's t-test).(d) The simultaneous application of sgRNA F1 224 and sgRNA R2 228 led to the deletion of 199 bp in the KRAS gene, which was detected with PCR.

Figure 2 :
Figure 2: HDR of the individual DSBs and the micro-deletion with ssODNs.(a) 90-mer ssODNs were designed to repair DSBs from a single cut (sgRNA F1 224 or sgRNA R2 228) (left) as well as a micro-deletion from the use of a dual cut (both sgRNA F1 224 and sgRNA 228) (right).The homology arms of the ssODNs for the repair of single cuts flanked sequences directly adjacent to their respective DSBs.The micro-deletion repair ssODN was designed to flank distal sequences that were outside of the F1 224 and R2 228 DSBs.Each ssODN carried the EcoRI recognition site that was targeted for integration into the repair site.(b) Restriction fragment length polymorphism analysis of the PCR products demonstrated the HDR of single DSBs and the micro-deletion as the genomic integration of the EcoRI restriction site (n=3; mean±SD, *p<0.05,unpaired Student's t-test).(c) PCR products of the repaired region amplified from gDNA, showing the full length (650 bp) and the deletion amplicons (451 bp).

Figure 3 :
Figure 3: The envisioned dual cut strategy versus the traditional single cut strategy to modify the KRAS gene.(a) The extent of gene conversion is limited with the traditional single cut strategy, which uses a dsDNA donor design that has two homology arms (HAs) that directly flank the DSB site.The successful co-transfer of both the gene mutation and the selection marker cassette (SM) into sites that are 94 bp apart will be very inefficient with a single cut, as a result of short gene conversion tracts.The dual cut strategy should overcome this inefficiency by deleting the entire region between the two sites and then reconstituting this region with a modified gene sequence that contains the mutation and the selection marker.(b) The dsDNA donor was designed to functionally separate the intended gene modification elements (the selection cassette and the point mutation) from the homology arms.This separation ensured that the homology arms maintained complete homology with the target sequence and functioned solely to direct the construct to the target region.The selection cassette was designed to be expressed in the opposite direction of KRAS gene transcription to produce a chimeric eGFP-NeoR protein that is cleaved by an embedded 2A self-cleaving peptide, allowing for a fluorescence or antibiotic-based approach for selection.Fluorescence micrographs showed the expression of eGFP in HEK293 cells that were transfected with this construct (right).A qPCR strategy was used to determine the integration efficiency of the dsDNA donor into the KRAS gene by measuring the amount of various DNA components (left).Quantification was done on extracted gDNA 3 and 14 days post-transfection.(c) Relative amounts of detected target DNA of each primer set, normalised to β-actin.(d) The percentage of KRAS alleles that successfully underwent integration of the selection cassette at the KRAS gene locus, as calculated by taking the data from primer set A (integration events) as a percentage of primer set C (total KRAS DNA) (mean±SD, ND denotes no detected amplification).
cycles of denaturation at 98 °C for 1 s, annealing at 62.7 °C for 5 s, extension at 72 °C for 45 s and a final extension at 72 °C for 1 min.This linear dsDNA http://www.sajs.co.za