Orthogonal Cas9-Cas9 chimeras provide a versatile platform for genome editing

Abstract

The development of robust, versatile and accurate toolsets is critical to facilitate therapeutic genome editing applications. Here we establish RNA-programmable Cas9-Cas9 chimeras, in single- and dual-nuclease formats, as versatile genome engineering systems. In both of these formats, Cas9-Cas9 fusions display an expanded targeting repertoire and achieve highly specific genome editing. Dual-nuclease Cas9-Cas9 chimeras have distinct advantages over monomeric Cas9s including higher target site activity and the generation of predictable precise deletion products between their target sites. At a therapeutically relevant site within the BCL11A erythroid enhancer, Cas9-Cas9 nucleases produced precise deletions that comprised up to 97% of all sequence alterations. Thus Cas9-Cas9 chimeras represent an important tool that could be particularly valuable for therapeutic genome editing applications where a precise cleavage position and defined sequence end products are desirable.

Fig. 1

Development of a functional Cas9–Cas9 nuclease framework. a Schematic of SpCas9^MT–dNm/SaCas9 fusions. PAM-interaction attenuated SpCas9 (yellow star) is C-terminally fused to a nuclease-dead Cas9 from N. meningiditis or S. aureus. Each Cas9 is loaded with its cognate sgRNA. b Top, schematic of parameters tested for target site organization. Four composite target site configurations are tested (D1:D4). The red arrow and rectangle represent the SpCas9 protospacer (in 5′–3′ orientation) and PAM, respectively, whereas the blue arrow and rectangle represent the Nm/SaCas9 protospacer (in 5′–3′ orientation) and PAM, respectively. Two dashed lines indicate the edges of each orthogonal Cas9-binding site, and x represents the number of intervening nucleotides. Bottom, activity profiles of SpCas9 (blue), SpCas9^MT3 (R1335K; gray), and C-terminal fusions for SpCas9^MT3–dNmCas9 (pink), and SpCas9^MT3-dSaCas9 (purple) in the GFP reporter assay. GFP reporter assay data are from three independent biological replicates performed on different days in HEK293T cells. c Changes in the activity profile of SpCas9^MT3-dSaCas9 nuclease as a function of the distance between sgRNA-binding sites at the AAVS1 locus. Top, schematic of the orientation of the target sites, where the SpCas9 site is fixed and the SaCas9 site is shifted away various distances to examine its distance-dependent activity (Supplementary Fig. 2) Bottom, deep sequencing data are from three independent biological replicates performed on different days in HEK293T cells, where the orientation and spacing between the orthogonal Cas9 sites is indicated below the x-axis (Supplementary Data 1). Error bars indicate ± s.e.m

Fig. 2

SpCas9^MT–dSa/NmCas9 fusions improve the specificity of editing. a Sequences of dual Cas9 genomic target sites at the VEGFA locus. The SpCas9 protospacer is bold underlined with its PAM is in red, the SaCas9 protospacer is double underlined with its PAM is green, and the NmCas9 protospacer is wavy underlined with its PAM in blue. b Lesion rates of the nuclease platforms are determined by deep sequencing. c, d Genome-wide off-target analysis of the nuclease platforms determined via GUIDE-seq (Supplementary Data 2). c The number of off-target peaks detected for the given nuclease. d Fold improvement of the specificity ratio of the Cas9^MT–dCas9 framework relative to SpCas9^WT. e Deep sequencing determined lesion rates of the nucleases at a small set of off-target sites discovered within the GUIDE-seq data. GUIDE-seq result is from a single experiment, whereas amplicon deep sequencing data are from three independent biological replicates performed on different days in HEK293T cells (Supplementary Data 1). Error bars indicate ± s.e.m

Fig. 3

Cas9–Cas9 dual nucleases generate uniform deletion products. a Top, sequence of SpCas9-NmCas9 target site: SpCas9 protospacer is underlined and its PAM is red; NmCas9 protospacer is wavy underlined and its PAM is blue. Purple arrows indicate expected double-strand break positions. Bottom left, genomic region containing the target site is PCR amplified; the higher band is the wild-type sequence or sequences with small indels, and the lower band is the segmental deletion product generated by dual nucleases. Bottom right, chromatogram from Sanger sequencing of the gel-extracted lower band (black arrow). The main product is the perfect junction of two double-strand break sites yielding a precise deletion (black rectangle). b Lesion rates and types are determined by deep sequencing. Single nucleases generate small indels at their cleavage sites, whereas dual nucleases (independent or fusion) may generate six types of lesion products. The majority of the lesions produced by dual-nuclease fusions is precise deletion. SpCas9^MT–dNmCas9 fusions behave like a monomeric SpCas9. Values above each bar indicate the precise deletion rate divided by the total lesions. c–e Activity profiles of SpCas9^WT (blue), Nm/SaCas9^WT (pink), SpCas9^WT + Nm/SaCas9^WT (orange), and SpCas9^WT − Nm/SaCas9^WT (green) nucleases at 12 genomic sites (six D1 and 6 D2 configurations) determined by deep sequencing. c Total lesion rates of the Cas9–Cas9 dual nucleases are higher than the monomeric Cas9s used in combination. d Cas9–Cas9 fusions generate higher rates of precise deletions than two independent Cas9 monomers. e Cas9–Cas9 dual nucleases primarily generate precise deletion products whereas lesion types of the two independent monomeric Cas9s are site dependent. Each box plot is drawn by GraphPad Prism, where the box represents the 25th and 75th percentile and the middle line is the median. Whiskers and outliers are defined by the Tukey method. Statistical significance is determined by one-way analysis of variance (ANOVA), ** and **** denote P < 0.01 and P < 0.0001, respectively. Deep sequencing data are from three independent biological replicates performed on different days in HEK293T cells (Supplementary Data 1). Error bars indicate ± s.e.m

Fig. 4

Cas9–Cas9 dual nucleases display enhanced precise deletions out to ~200 bp site separation. Changes in the activity profile of SpCas9^WT–SaCas9^WT nuclease as a function of the distance between sgRNA-binding sites at the AAVS1 locus. Boxed inset: schematic of the orientation of the target sites, where the SpCas9 site is fixed and the SaCas9 site is shifted to examine the distance-dependent activity of the Cas9–Cas9 fusions (Supplementary Fig. 2). Bar graph: deep sequencing data, where the various lesion types and rates at AAVS1 sites are determined by UMI-corrected deep sequencing. Data are from three independent biological replicates performed on different days in HEK293T cells, where the orientation and spacing between the orthogonal Cas9 sites is indicated below the x-axis (Supplementary Data 1). Error bars indicate ± s.e.m

Fig. 5

Cas9–Cas9 fusions expand the targeting range of SpCas9 allowing deletion of the GATA1-binding element in the BCL11A enhancer+58 kb. a Lesion rates and types at tandem target sites, with suboptimal SpCas9 but canonical SaCas9 PAMs (Supplementary Fig. 8) are determined by deep sequencing with bulk analysis. SaCas9 generates robust editing whereas SpCas9 displays low or no activity. In Cas9–Cas9 fusion format, SpCas9 cuts effectively at these protospacers as observed from the SpCas9–dSaCas9 fusions or the fused wild-type nucleases. b Sequence information of the four target sites chosen for more detailed assessment of the application of Cas9–Cas9 fusions for the deletion of the GATA1-binding element in the functional core of the BCL11A enhancer+58 kb (highlighted in gray). c Lesion rates and types at four target sites spanning the GATA1 element are determined by deep sequencing after UMI-correction. Deep sequencing data are from three independent biological replicates performed on different days in HEK293T cells (Supplementary Data 1). Error bars indicate ± s.e.m

Fig. 6

Cas9–Cas9 fusions achieve robust and specific genome editing. a Summary of the GUIDE-seq genome-wide off-target analysis of SpCas9^WT, Sa/NmCas9^WT, and SpCas9^WT-Sa/NmCas9^WT at four GATA1 target sites (Supplementary Data 2). b Deep sequencing determined lesion rates for these nucleases at a subset of off-target sites discovered by the GUIDE-seq data or predicted by CasOFFinder (Supplementary Data 3). The names of SpCas9, NmCas9, and SaCas9 off-target sites are colored in dark red, blue, and green. The GUIDE-seq result is from single experiment, and amplicon deep sequencing data are from three independent biological replicates performed on different days in HEK293T cells (Supplementary Data 1). Error bars indicate ± s.e.m. Statistical significance is determined by one-way analysis of variance (ANOVA), *, **, ***, and NS denote BH-adjusted P-values of <0.05, <0.01, <0.001, and not significant, respectively