Enhanced Bacterial Immunity and Mammalian Genome Editing via RNA-Polymerase-Mediated Dislodging of Cas9 from Double-Strand DNA Breaks

Abstract

The ability to target the Cas9 nuclease to DNA sequences via Watson-Crick base pairing with a single guide RNA (sgRNA) has provided a dynamic tool for genome editing and an essential component of adaptive immune systems in bacteria. After generating a double-stranded break (DSB), Cas9 remains stably bound to DNA. Here, we show persistent Cas9 binding blocks access to the DSB by repair enzymes, reducing genome editing efficiency. Cas9 can be dislodged by translocating RNA polymerases, but only if the polymerase approaches from one direction toward the Cas9-DSB complex. By exploiting these RNA-polymerase/Cas9 interactions, Cas9 can be conditionally converted into a multi-turnover nuclease, mediating increased mutagenesis frequencies in mammalian cells and enhancing bacterial immunity to bacteriophages. These consequences of a stable Cas9-DSB complex provide insights into the evolution of protospacer adjacent motif (PAM) sequences and a simple method of improving selection of highly active sgRNAs for genome editing.

Keywords: CRISPR; Cas9; DNA repair; RNA polymerase; genome editing; phage biology; strand bias; transcription.

Figure 1. Transcription mediated displacement of Cas9 from the DSB increases genome editing frequencies and is strand dependent

A, Bimodal distribution of indel frequencies of 40 distinct mouse genes four days after transient transfection of Cas9 and sgRNA expression plasmids. Individual observations from biological duplicates for each sgRNA were binned according to their mutation frequencies (% indel), and the number of sgRNA that fell into each bin is displayed (count). See also Table S1. B, Indel frequencies associated with the 40 sgRNA in (A) were separated by whether the sgRNA annealed to the DNA strand used as the template for transcription by RNA polymerase II (RNAP), or the non-template DNA strand. There are 17 template strand and 23 non-template sgRNA. Each point represents a mutation frequency of independent transfections, n=2 for each sgRNA. **p < 0.01. C, Schematic illustrating orientation of Cas9, target DNA, and an approaching RNAP for the two possible RNAP and Cas9 collision orientations: with a template sgRNA and non-template sgRNA. D, Mutagenesis frequencies mediated by 20 different sgRNA targeting a genomic mCherry measured by T7E1 assays (see also Fig S2E). mCherry transcription is controlled by doxycycline (dox) (see Fig S2D). Plusses (+dox/mCherry expression) and cirlces (−dox/no mCherry expression) represent individual biological replicates testing the effect of transcription on mutagenesis levels mediated by each sgRNA. Genomic DNA was isolated 48h after transfection. *p<0.05. E, The strand bias was tested at a silent endogenous gene through synthetically activating transcription of the human TTN gene using Cas9-VPR construct. Nuclease active Cas9-VPR was targeted to activate transcription but not introduce DSBs through using a 14nt sgRNA. Simultaneously, a 20nt sgRNA targeted to either the template or non-template strand was provided to drive transcription mediated by 14nt-Cas9-VPR through Cas9 cleavage sites. Genomic DNA was harvest 48h after transfected and mutation frequencies were analyzed via T7E1 assays. Each point represents a biological replicate.

Figure 2. The Cas9-DSB complex precludes DNA repair activities in vitro

A, Detection of phospo-H2AX levels 24hrs after transfecting mouse ES cells with pools of either template or non-template sgRNAs. sgRNA that mediated >30% indel were selected sgRNA (Fig 1A,B, Table S1). For each sgRNA, a new sgRNA annealing the opposite strand of the same gene was made. To compare strand among the same sets of genes, pools of 4 or 8 sgRNA consisted of the previously characterized and newly generated sgRNA. Western blot analysis was used to determine fold change of phospo-H2AX signal with denistometric measurement of bands and normalization to the loading control (β-actin) and the no sgRNA control. Four target genes: APC, FBXW7, PTPN11, and TSC1. Eight target genes: APC, FBXW7, PTPN11, TSC1, VPS16, VPS54, RAB7, and RANPBP3. B, Differences in Ku70/80 binding at template or non-template Cas9-generated DSBs was measured by chromatin immunoprecipitation of Ku70/80 bound DNA followed by quantitative PCR (ChIP-qPCR). ChIP DNA was isolated 24 hours after transfection of DNA to express the pool of four sgRNA from (A), DNA precipitated by Ku70/80 antibodies at each target site was measured through qPCR amplifying a sequence adjacent to each Cas9 cleavage site. For each transfected cell population, two biological replicates were harvested, and each was split into three technical replicates prior to immunoprecipitation. Data are expressed as enrichment of the target site compared to the negative control site (Gapdh). **p < 0.01, *p < 0.05. C, Agarose gel electrophoresis of an in vitro reaction where linear dsDNA was digested by Cas9 for 30 minutes, then treated with Proteinase K to release the cleaved DNA products. D, The ability of T4 DNA ligase to repair a Cas9-generated DSB in a circular plasmid DNA was measured through E. coli colony formation on ampicillin-containing plates (CFU) after transformation. Cas9 or restriction endonuclease (PmeI) digestion of plasmid DNA prevented CFU following transformation. T4 DNA ligase activity repaired the DSB and stimulated CFU if plasmid was cut with PmeI, or if Cas9 was denatured at 75C for 10m before addition of ligase. T4 DNA ligase did not stimulate CFU if Cas9 was not denatured. Values represent mean +/− s.d., n=3. E, Agarose gel analysis of a circular plasmid DNA incubated with T7 exonuclease and the conditions indicated above each lane. PmeI and heat denaturation of Cas9 were as described in (D). Cas9 prevented DNA ends from serving as a substrate for T7 exonuclease unless reactions were heat denatured prior to exonuclease addition. All reactions were treated with Proteinase K before gel loading. F, Schematic depicting the experiment in (G) to test if Ku70/80 can displace Cas9 from its DSB. The Cas9-DSB complex is formed on target DNA that is biotinylated on one end and fluorescein (FAM) conjugated on the other. If purified human Ku70/80 displaces Cas9 from the DSB, release of the fluorescent DNA end is measured as soluble fluorescence. G, Liberation of fluorescent DNA ends into the soluble fraction after challenging the target DNA with indicated conditions. NcoI is a restriction endonuclease that cuts the DNA substrate and functions as the control for maximum fluorescence, and maximum fluorescence of Cas9 digested DNA was assessed through Proteinase K treatment after Cas9-DSB formation. See also Fig S2D. Values represent mean +/− s.d., n=3.

Figure 3. The Cas9-DSB complex is disrupted by translocating RNA polymerases if the sgRNA anneals to the template strand

A, Schematic illustrating orientation of Cas9 RNP, target DNA, and T7 RNAP translocation colliding with the PAM-distal surface of Cas9 for a template sgRNA and disruption of the enzyme-product complex. B, DNA degradation by T5 exonuclease ability to access Cas9 generated DSB ends in the presence or absence of T7 RNAP transcription was visualized by agarose gel electrophoresis. Plasmid DNA harboring a T7 promoter was digested with Cas9 or PmeI restriction endonuclease for 30 min prior to incubation with T5 exonuclease and/or T7 RNAP. All reactions were treated with Proteinase K before gel loading. C, Schematic illustrating experiment in (D) to test whether T7 RNAP can evict Cas9 from the DSB and whether T7 RNAP-displaced Cas9 molecules retain activity. In the presence of inactive T7 RNAP, the Cas9-DSB complex was formed on a Target DNA 1, which contains a T7 RNAP promoter on either end of the DNA for either collision orientation. After 30 minute incubation, rNTPs and a second substrate (Target DNA 2) are simultaneously added. Target DNA 2 lacks a T7 promoter. Target DNA 1 and target DNA 2 each have the same DNA sequence targeted by Cas9. The addition of rNTPs and Target DNA 2 stimulates T7 RNAp transcription and provide a sensor of displaced Cas9 molecules. D, Agarose gel for the experiment described in (C). Template and Non-template refer to location of T7 promoter on Target DNA 1. Cleavage of target DNA 2 indicates displacement of active Cas9 from Target DNA 1 over time. E, The ability for T7 RNAP to displace Cas9 with various sgRNA was measured similar to (C), except target DNA 2 was biotinylated on one end and FAM conjugated on the other end, as illustrated in Fig S2D. The 20 mCherry sgRNA (from Fig 1D, S1C) were subjected to the assay in the presence or absence of rNTPs. The fold change in fluorescence levels as a result of T7 RNAP mediated displacement was measured through fluorescence in the soluble fraction. Values are mean +/− s.d.; n=3 for each sgRNA. F, Fold change Cas9 activity dislodged from mCherry DNA by mammalian RNA Pol II activity from nuclear extracts. Activity was measured by the soluble fraction fluorescent levels for a fluorescent, as above. RNA Pol II activity was controlled by addition of α-amanitin. See also Fig. S2D. Values are mean +/− s.d.; n=2 for each sgRNA.

Figure 4. Strand dependent ability of translocating T7 RNA polymerase to stimulate in vitro multi-turnover nuclease activity by Cas9

A, Multi-turnover nuclease capability of Cas9 was visualized by agarose gel analysis of hybrid reactions combining Cas9 nuclease and T7 RNAP transcription reactions. A T7 promoter was placed on either end of the target DNA to achieve template or non-template orientation. Cas9 was incubated with DNA for 30m as shown before initiating T7 RNAP with addition of rNTPs. B, The multi-turnover capacity of template strand Cas9 was measured through titration of in the presence or absence of T7 RNAP. Target DNA was held constant at 150nM. Values represent mean +/− s.d., n=3. (See also Fig. S3C). C, Titration of substrate in the presence or absence of T7 RNAP. Cas9 was held constant at 12.5nM. Values represent mean +/− s.d., n=2. (see also Fig S3D). D, Multi-turnover Cas9 activity on various sgRNA was examined through hybrid digestion and transcription reactions of target DNAs harboring T7 RNAP promoters in the template or non-template strand orientation. See also Fig. S4A for schematics and Fig S4B,C and Fig S6 for representative gels. Values represent mean +/− s.d. of fold changes in cleavage by indicated sgRNA in the presence or absence of T7 RNAP, n=3.

Figure 5. PAM sequences across Streptococci phage are more frequently oriented on the template strand

A, Of the 16 surveyed streptococcus phages, all harbor majority of PAM sequences on DNA strand corresponding to the transcription template strand. NNAGGAW = S. thermophilus PAM and NGG = S. pyogenes PAM. B, Distribution of all PAM sequences among genomes analyzed in A.

Figure 6. Template strand-targeted protospacers enhance phage interference

A, General organization of the phage ΦNM1 genome and targeting strategy of the lysogenic repressor gene in the ΦNM1h1 and ΦNM1γ6 mutant phages. The ΦNM1h1 mutant has defective lysogeny genes that transcriptionally active, while the ΦNM1γ6 mutant has transcriptionally silent lysogeny genes. 2 pairs of protospacers were designed to target the mutant phages so that each pair consists of crRNA annealing to either the template or non-template strand with PAM sites within 25bp of each other. S. aureus strains harboring S. pyogenes Cas9 and each of these spacers (RC1–4) were generated respectively for the experiment. B, Growth curves of S. aureus strains harboring spacers RC1–4 after infection with ΦNM1h1 or ΦNM1γ6. T: template strand orientation. NT: non-template strand orientation.