Landscape of target:guide homology effects on Cas9-mediated cleavage

Abstract

To study target sequence specificity, selectivity, and reaction kinetics of Streptococcus pyogenes Cas9 activity, we challenged libraries of random variant targets with purified Cas9::guide RNA complexes in vitro. Cleavage kinetics were nonlinear, with a burst of initial activity followed by slower sustained cleavage. Consistent with other recent analyses of Cas9 sequence specificity, we observe considerable (albeit incomplete) impairment of cleavage for targets mutated in the PAM sequence or in 'seed' sequences matching the proximal 8 bp of the guide. A second target region requiring close homology was located at the other end of the guide::target duplex (positions 13-18 relative to the PAM). Sequences flanking the guide+PAM region had measurable (albeit modest) effects on cleavage. In addition, the first-base Guanine constraint commonly imposed by gRNA expression systems has little effect on overall cleavage efficiency. Taken together, these studies provide an in vitro understanding of the complexities of Cas9-gRNA interaction and cleavage beyond the general paradigm of site determination based on the 'seed' sequence and PAM.

Figure 1.

Assaying a diverse population of potential targets for Cas9 cleavage. (a) Schematic of target recognition by Cas9 (redrawn from Pattanayak et al. (8)). Cas9 uses 20 bp of gRNA homology and PAM to recognize and target DNA and cleave both strands of substrate. (b) Target sequences assayed using random variant libraries. From the 5′ end the bases are indexed starting at −5 and ending with 29 (left to right). Positions 21–23 correspond to the PAM (NGG). Positions 11–20 correspond to the seed region. For both guides, the 0th position contained a guanine added on by the gRNA expression vector. c. Cas9 in vitro assay. Each random variant library was incubated with Cas9:gRNA complexes. PCR was used to amplify uncut Cas9 target regions. The resulting amplicons were subjected to high-throughput DNA sequencing and computational analysis described in Materials and Methods.

Figure 2.

Cleavage results for two independent unc-22A random variant libraries. (A) Each variant is indicated by a dot noting differences from the canonical sequence. Retention at various time points following Cas9 addition is depicted as a dot colored according to time point. For the target region and constrained PAM nucleotides (positions 1–20 and 22–23), each dot represents median retention amongst matching library sequences carrying the variant of interest (either A,C,G or T) and distinct flanking sequences. For the flanking sequence and the N in the PAM, each dot represents median retention for library species with fully matched target/PAM regions and with the indicated flanking position constrained to the indicated base. Only species with > = 50 reads in the control (uncut) library were considered in calculating this median. Two negative (labeled PS4) and one positive (labeled WT) controls are shown at the left. Sequences with 4–7 mismatches in the target region, and sequences from a ps4 ‘spike in’ behave in aggregate as similar ‘uncut’ pools, while a median from all sequences that match the guide and PAM sequences perfectly provides a positive control. The retention scores shown are from experiments done on unc-22A random variant library-2 (AF_SOL_516, using 0.2 ug of enzyme per reaction). Negative retention scores indicate increasing cleavage, while a retention score of zero corresponds to samples whose cleavage is comparable to the pool of highly mutated sites (with 4–7 mutations) and to trace amounts of unrelated (protospacer 4) DNA. Some samples showed a slightly positive retention score, indicating a lack of cleavage with either noise or consistent PCR effects leading to a slight over-representation in the post-cleavage sample. (B) Detailed time course experiments for unc-22A random variant library-1. Methods of analysis were as in Figure 2A. The retention scores shown are from experiment AF_SOL_515, using 0.2 ug of enzyme per reaction. (C) Heat map summarizing cleavage for single variants for 180 min time point experiments shown in part b. The heat map shows (median variant retention)/(median WT retention). Boxes with asterisks indicate the WT base. Values with the score of ∼1 (white) are variants with retentions that do not greatly deviate from the median for all perfectly matched (guide+PAM) targets. Variants with values that are above 1 (blue) correspond to variants where read counts indicated a more efficient Cas9-based depletion than WT. Variants with values around ∼0 (red) correspond to variants that either do not cleave or cleave very inefficiently. Retention scores shown are from experiment AF_SOL_515, using 0.2 ug of enzyme per reaction.

Figure 3.

Cleavage results for the protospacer 4 (ps4 (7)) library analysis. Methods of analysis from Figure 2 were used to analyze the ps4 variant library. A reference population used for normalization with ps4 libraries consists of sequences with 4–7 mismatches. The log retention scores shown are from experiment AF_SOL_513, using 0.2 ug of enzyme per reaction. (A) Cleavage results for ps4. (B) Heat map in vitro results for ps4. Black boxes represent variants that were not assayed due to insufficient representation in the library.

Figure 4.

Kinetics of Cas9 cleavage for exemplary variant categories. Graph depicts the median retention for each exemplary variant as a function of time for the unc-22A random variant library (AF_SOL_515). The error bars represent the standard deviation of the median retentions for each exemplary variant at the specific time point.

Figure 5.

Double variant analysis for unc-22A random variant library (AF_SOL_515). Each position of the target sequence is a box with a 4×4 grid labeled ‘ATCG’. The 4×4 grid denotes the variant and the positions of the double variants (shown in the figure legend). Because transitions and transversions have diverse effects on cleavage efficiency, the data are represented in a detailed heat map (92×92). Only the 23 bp target sequence of the gRNA is shown in this heat map (positions 1–23). Only species with > = 50 reads in the control (uncut) library were considered. (A) Table showing bases of gRNA and corresponding position (for reference to heat map positions). (B) Detailed double variant heat map of log retentions. Each box represents the median log retention of the double variant of interest. For reference, the diagonal provides median log retention of the single variants. Red colors correspond to variants that do not cleave efficiently. Blue colors correspond to variants with highly efficient cleavage. (C) Detailed double variant synergy heat map. Each box represents the value of the double variant divided by the corresponding the multiplication of the single variants: (LR_M12/LR_WT) / [(LR_M1/LR_WT) * (LR_M2/LR_WT)]. Single mutants with no detectable cleavage (no significant difference in representation between cleaved and uncleaved) were not considered in this analysis, as a synergy value cannot generally be calculated. A synergy value of 1 corresponds to double variants whose kinetic influence on the reaction rate is the product of the influences of the two single variants. Synergy values below 1 indicate negative synergistic effects of variants, while values above 1 indicate a higher residual cleavage activity on the double variant that would be expected from the individual effects of the single variants.