Efficient Homology-Directed Repair with Circular Single-Stranded DNA Donors

Abstract

While genome editing has been revolutionized by the advent of CRISPR-based nucleases, difficulties in achieving efficient, nuclease-mediated, homology-directed repair (HDR) still limit many applications. Commonly used DNA donors such as plasmids suffer from low HDR efficiencies in many cell types, as well as integration at unintended sites. In contrast, single-stranded DNA (ssDNA) donors can produce efficient HDR with minimal off-target integration. In this study, we describe the use of ssDNA phage to efficiently and inexpensively produce long circular ssDNA (cssDNA) donors. These cssDNA donors serve as efficient HDR templates when used with Cas9 or Cas12a, with integration frequencies superior to linear ssDNA (lssDNA) donors. To evaluate the relative efficiencies of imprecise and precise repair for a suite of different Cas9 or Cas12a nucleases, we have developed a modified traffic light reporter (TLR) system (TLR-multi-Cas variant 1 [MCV1]) that permits side-by-side comparisons of different nuclease systems. We used this system to assess editing and HDR efficiencies of different nuclease platforms with distinct DNA donor types. We then extended the analysis of DNA donor types to evaluate efficiencies of fluorescent tag knockins at endogenous sites in HEK293T and K562 cells. Our results show that cssDNA templates produce efficient and robust insertion of reporter tags. Targeting efficiency is high, allowing production of biallelic integrants using cssDNA donors. cssDNA donors also outcompete lssDNA donors in template-driven repair at the target site. These data demonstrate that circular donors provide an efficient, cost-effective method to achieve knockins in mammalian cell lines.

FIG. 1.

Comparisons of the integration efficiencies of different donor topologies on HDR using the TLR-MCV1 cassette in human cells. (A) The schematic depicts the TLR-MCV1 system showing the SFFV promoter driving the expression of GFP and mCherry, separated by a ribosome-skipping T2A signal. The yellow arrow depicts the SFFV promoter driving the expression of the GFP-T2A-mCherry cassette. The orange box indicates the insertion disrupting GFP containing target sequences for different Cas effectors. The sequence of insertion is shown below the schematic of TLR-MCV1. Sequences and arrows in blue indicate overlapping PAMs and a common cut site associated with SpyCas9, Nme1Cas9, CjeCas9, and SauCas9. The bolded black sequence and black arrow depict the Nme2Cas9 PAM and cut site, respectively. The magenta text shows PAMs associated with Cas12a effectors, and their approximate cut sites are shown by magenta lines. The PAMs associated with Geo1Cas9 and Geo2Cas9 are highlighted in green italicized text and brown-underlined italicized text, respectively. The cut sites for these two Cas9s are shown by green and brown arrows, respectively. DSBs at any of the sites may be imprecisely repaired through the NHEJ pathway resulting in mCherry expression (shown on the left) if repair results in productive translation due to a +1 frameshift. In the presence of donor DNA, HDR-mediated correction of the “broken” GFP region results in restoration of GFP expression (shown on the right). (B) Efficacy of distinct DNA templates in driving HDR. The graph depicts the percentage of GFP-positive cells obtained after codelivery of SpyCas9 or AspCas12a RNP with cssDNA, T-lssDNA, B-lssDNA, or plasmid DNA repair templates into TLR-MCV1 K562 cells (upper gray box) and TLR-MCV1 HEK293T cells (lower blue box). Bars represent the mean from three independent biological replicates and error bars represent the s.e.m. (C) Schematic of the approach used to generate circularized B-lssDNA. A short oligonucleotide (red) is hybridized to the B-lssDNA containing a 5′-phosphorylated end such that the oligo spans the 5′ and 3′ ends of the lssDNA. The sample is treated with Escherichia coli DNA ligase to ligate the ends. The lssDNA sample is then treated with exonucleases (I and III) to eliminate residual uncircularized lssDNA. The native agarose gel shows linear and ligated lssDNA before and after treatment with exonucleases, which digest unprotected, linear DNA species. (D) The graphs depict the percentage of GFP-positive cells obtained after codelivery of SpyCas9 with B-lssDNA and circularized B-lssDNA DNA repair templates into TLR-MCV1 K562 cells (upper gray box) and TLR-MCV1 HEK293T cells (lower blue box). Bars represent the mean from three independent biological replicates and error bars represent (s.e.m.). n.s., p value not significant; ***p < 0.001. B-lssDNA, biotin-based affinity purified linear ssDNA; cssDNA, circular ssDNA; DSB, double-strand break; GFP, green fluorescent protein; HDR, homology-directed repair; lssDNA, linear ssDNA; NHEJ, nonhomologous end joining; PAM, protospacer adjacent motif; RNP, ribonucleoprotein; s.e.m., standard error of the mean; ssDNA, single-stranded DNA; TLR-MCV1, traffic light reporter multi-Cas variant 1; T-lssDNA, reverse-transcription generated linear ssDNA.

FIG. 2.

Characterization of the HDR efficiencies of cssDNA and T-lssDNA. (A) Comparison of cssDNA- and T-lssDNA-mediated HDR efficiency upon treatment of TLR-MCV1 cells with distinct Cas effectors. The graphs depict the percentage of GFP-positive cells obtained after codelivery of SpyCas9, AspCas12a, LbaCas12a, or FnoCas12a with cssDNA and T-lssDNA DNA repair templates into TLR-MCV1 K562 cells (upper gray box) and TLR-MCV1 HEK293T cells (lower blue box). Bars represent the mean from three independent biological replicates and error bars represent the s.e.m. (B) Effect of cssDNA and T-lssDNA donor orientation on HDR efficiency. The graphs depict the percentage of GFP-positive cells obtained after codelivery of SpyCas9 or AspCas12a (targeting the same strand) with sense (S) and antisense (AS) strand cssDNA and T-lssDNA, DNA repair templates into TLR-MCV1 K562 cells (upper gray box) and TLR-MCV1 HEK293T cells (lower blue box). Bars represent the mean from three independent biological replicates for K562 cells and six independent replicates for HEK293T cells. Error bars represent s.e.m. (C) Dose dependence of cssDNA and T-lssDNA donor template-mediated HDR efficiency. The graphs depict the percentage of GFP-positive cells as a function of increasing cssDNA and T-lssDNA donor DNA in the presence of SpyCas9 and AspCas12a proteins in TLR-MCV1 K562 cells (left) and HEK293T cells (right). Points represent the mean from three independent biological replicates, and error bars represent s.e.m. n.s., p value not significant; *p < 0.05; ***p < 0.001.

FIG. 3.

Comparison of the type of DNA donor on the efficiency of HDR at endogenous genomic loci in human cells. (A) Schematic of fluorescent protein tagging. The left panel shows a schematic of a genomic region containing the SpyCas9 target site and the design of a donor template containing the fluorescent protein of interest flanked by HA. The right panel shows a schematic of each target genomic locus and the arrangement of the fluorescent tag (EGFP, dTomato, or iTag-RFP) following integration. Three of the donors (targeting ACTB, TOMM20, and SEC61B) produce direct fusions of the tag to the endogenous protein. The donor designed to fluorescently tag the GAPDH locus contains an IRES and a bGH polyadenylation sequence. (B–E) Bar graphs displaying the percentages of fluorescent cells obtained upon codelivery of 20 pmoles of SpyCas9 complexed with 25 pmoles of guide-RNA targeting the (B) ACTB, (C) TOMM20, (D) SEC61B, or (E) GAPDH locus with or without cssDNA or T-lssDNA as a donor template. Bars represent the mean from three independent biological replicates and error bars represent s.e.m. n.s.: p value not significant; ***p < 0.001. bGH, bovine growth hormone; EGPF, enhanced green fluorescence protein; GAPDH, glyceraldehyde 3-phophate dehydrogenase; HA, homology arms; IRES, internal ribosome entry site; RFP, red fluorescent protein.

FIG. 4.

Biallelic tagging of endogenous proteins using two different cssDNA donor templates. (A) The graphs show the percentage of fluorescent cells tagged with GFP (shown in cyan), dTomato (shown in red), or both (shown in yellow) at each locus (TOMM20, SEC61B, or GAPDH) in K562 cells (top panel) and HEK293T cells (bottom panel). Twenty picomole of SpyCas9 RNPs were codelivered with 0.5 pmol of each cssDNA template. Bars represent the mean from three independent biological replicates and error bars represent s.e.m. (B) Competition between cssDNA and lssDNA templates as donors for HDR. The graph shows the percentage of cells tagged with GFP (shown in cyan), iTAG-RFP (shown in red), or both GFP and iTAG-RFP (shown in yellow) at the ACTB locus. Bars represent the mean from three independent biological replicates and error bars represent s.e.m.