Non-homologous DNA increases gene disruption efficiency by altering DNA repair outcomes

Abstract

The Cas9 endonuclease can be targeted to genomic sequences by programming the sequence of an associated single guide RNA (sgRNA). For unknown reasons, the activity of these Cas9-sgRNA combinations varies widely at different genomic loci and in different cell types. Thus, disrupting genes in polyploid cell lines or when using poorly performing sgRNAs can require extensive downstream screening to identify homozygous clones. Here we find that non-homologous single-stranded DNA greatly stimulates Cas9-mediated gene disruption in the absence of homology-directed repair. This stimulation increases the frequency of clones with homozygous gene disruptions and rescues otherwise ineffective sgRNAs. The molecular outcome of enhanced gene disruption depends upon cellular context, stimulating deletion of genomic sequence or insertion of non-homologous DNA at the edited locus in a cell line specific manner. Non-homologous DNA appears to divert cells towards error-prone instead of error-free repair pathways, dramatically increasing the frequency of gene disruption.

Figure 1. Non-homologous DNA increases gene disruption in multiple cell types.

(a) Single- and double-stranded linear non-homologous DNA stimulates indel formation in HEK293T cells. Cas9 was targeted to the EMX1 locus with or without nucleic acid carrier agents (−, no nucleic acid). Indel formation was measured using a T7 endonuclease I assay and is presented as the mean±s.d. of at least two independent experiments (see Supplementary Data 1 for uncropped gels). * denotes significant difference between sample and no nucleic acid (−) control (P<0.05, Welch's t-test). (b) Non-homologous single-stranded DNA treatment (NOE) increases editing rates in multiple cell types. Editing was performed as described in panel A in multiple cell types either with (dark blue bars, NOE) or without (light blue bars, RNP) 4.5 μg of N-oligo. (c) NOE increases the frequency of homozygous gene disruption. HEK293T cells edited in panel (b) were clonally isolated and amplicons were sequenced to determine genotype. Each horizontal bar represents a single clone with green (wildtype sequence) or magenta (mutations disrupting EMX1) divisions sized according to the percentage of sequencing reads in each category. Zygosity is summarized in the lower table.

Figure 2. NOE promotes error-prone DNA repair events that differ between cell types.

(a) NOE stimulates insertions and deletions in HEK293T cells. The allele frequency of deletions (left of plot) and insertions (right of plot) are shown for nucleofections performed with (dark blue, NOE) or without (light blue, RNP) N-oligo. Editing is summarized for each clone in Supplementary Fig. 3. Raw data is available in Supplementary Data 3. (b) Inserted sequences are derived from single- and double-stranded heterologous DNA. Wildtype EMX1 sequence is presented at top with the protospacer (bold), PAM (bold underline), and cut site (triangle) diagrammed. Four example alleles are presented below with sgRNA template (green) and/or non-homologous oligonucleotide (grey) sequence inserted in both orientations. Complete sequencing alignments are available in Supplementary Data 3. (c) NOE stimulates deletions in U2OS cells. Multiple sequence reads from edited cell populations are presented as described in Fig. 2a. (d) Insertion of non-homologous DNA primarily occurs in HEK293T and K562 cells. DNA harvested from unedited (-), RNP treated (R), or NOE treated (N) cell populations was evaluated using a panel of PCR reactions (diagrammed at top). Primers N1 and N2 anneal to the N-oligo sequence; primers T1 and T2 anneal to residual sgRNA template. Open arrow—position of the 700 bp ladder band.

Figure 3. NOE is locus-independent.

(a) NOE causes similar fold changes of editing at FANCF on-target and three known off-target sites in HEK293T cells. Cas9 was targeted to the FANCF locus with (dark blue, NOE) or without (light blue, RNP) non-homologous oligonucleotide and editing rates at FANCF or previously reported off-target (OT) loci were measured by NGS analysis. Data presented are the mean±s.d. of two biological replicates. *—denotes significant (P<0.05, Welch's t-test) difference between RNP and NOE samples. Full editing rates are presented in Supplementary Fig. 7. (b,c) NOE increases editing rates ∼2.9-fold at the HEK293 site 1 and HEK293 site 3 known off-target sites, but cannot increase editing at the already saturated on-target site. Data presented as described in panel (a).