Continuous evolution of SpCas9 variants compatible with non-G PAMs

Abstract

The targeting scope of Streptococcus pyogenes Cas9 (SpCas9) and its engineered variants is largely restricted to protospacer-adjacent motif (PAM) sequences containing G bases. Here we report the evolution of three new SpCas9 variants that collectively recognize NRNH PAMs (where R is A or G and H is A, C or T) using phage-assisted non-continuous evolution, three new phage-assisted continuous evolution strategies for DNA binding and a secondary selection for DNA cleavage. The targeting capabilities of these evolved variants and SpCas9-NG were characterized in HEK293T cells using a library of 11,776 genomically integrated protospacer-sgRNA pairs containing all possible NNNN PAMs. The evolved variants mediated indel formation and base editing in human cells and enabled A•T-to-G•C base editing of a sickle cell anemia mutation using a previously inaccessible CACC PAM. These new evolved SpCas9 variants, together with previously reported variants, in principle enable targeting of most NR PAM sequences and substantially reduce the fraction of genomic sites that are inaccessible by Cas9-based methods.

Figure 1.. Phage-assisted non-continuous evolution (PANCE) of SpCas9 binding activity on non-G PAMs.

(a) Original selection scheme for Cas9 DNA binding in which ω-dSpCas9 expressed by ΔgIII selection phage (SP) binds to a designated protospacer and PAM sequence upstream of gIII on an accessory plasmid (AP) in host E. coli cells. Host cells and infecting SP are continuously mutagenized by a mutagenesis plasmid (MP). (b) Binding activity of SpCas9 or xCas9 on all 64 possible NNN PAMs, determined by fold propagation of SP expressing ω-dSpCas9 (top) or ω-dxCas9 (bottom) on host cells containing APs bearing these PAM sequences upstream of gIII. (c) Schematic overview of PANCE workflow. Host cells containing an AP and MP are grown to log phase in a deep-well plate or tube before being infected with SP. Mutagenesis is induced and SP are allowed to propagate for 6–18 hours before cells are pelleted and the SP-containing supernatant is collected. The SP pool is then used to infect host cells in the next iteration of PANCE. (d) Consensus mutations arising from evolution of ω-dSpCas9 (N1) or ω-dxCas9 (N2) on NAA (red), NAT (blue), or NAC (green) PAM sequences.

Figure 2.. Selection improvement enables evolution of SpCas9 variants with robust activity on non-G PAMs.

(a) Four selections for Cas9:DNA binding in PACE. 1. Original selection scheme. 2. Dual-AP selection where ω-dSpCas9 binds two distinct protospacer-PAM to drive either half of a split-intein pIII. 3. Use of split-intein Cas9 limits total Cas9 concentration in host cells, thus avoiding saturation of protospacer-PAM binding sites. Residues 574–1368 of SpCas9 fused to NpuC (dSpCas9_C) is expressed by the SP and ω–dSpCas9(1–573) fused to NpuN (ω–dSpCas9_N) is encoded by a low-copy complimentary plasmid (CP) in host cells. 4. Combination of the selection principles from (2) and (3) through use of gVI as an additional PACE-compatible selection marker for phage propagation and ΔgIIIΔgVI SP. (b) Mutations from further PACE of original PANCE mutants evolved to bind NAA PAMs. Mutations in darker red were acquired in later PACE generations. (c) Comparison of human cell base editing efficiency of evolved clones shown in (b) in HEK293T cells on NRA PAM sites. Bars represent mean and standard error (SEM) of n=3 independent biological replicates, with individual values shown as dots. (d) Mutated residues in the PID of TAA-P4s-4 mapped onto the SpCas9 crystal structure (4UN3). (e) Mutations from further PACE of original PANCE mutants evolved to bind NAT PAMs. Mutations in darker blue were acquired in later PACE generations. We note that the Y1131C mutation was found to be inactivating (Supplementary Fig. 2a). (f) Mutations from further PACE of original PANCE mutants evolved to bind NAC PAMs. Mutations in darker green were acquired in later PACE generations. (g) Mutated residues in the PID of TAT-P5–1 mapped onto the SpCas9 crystal structure (4UN3). (h) Mutated residues in the PID of TAC-P9s-3 mapped onto the SpCas9 crystal structure (4UN3).

Figure 3.. Comprehensive characterization of PAM preferences using a genomically integrated human cell target sequence library.

(a) Schematic overview of human cell base editing library experiments. A library of 11,776 matched sgRNA and protospacer target sites spanning all NNNN PAMs was genomically integrated in HEK293T cells. Library cells were transfected with and selected for genomic integration of plasmids encoding CBE variants. After antibiotic selection, the integrated sgRNA/protospacer site was amplified by PCR for HTS analysis. (b) Violin plots of base editing activity on the 11,776-member NNNN PAM library in HEK293T cells, with positions 2 and 3 of all NRN PAMs defined. For each construct, the editing across all sites containing the designated PAM over two independent biological replicates is shown, with solid lines indicating median and dotted lines indicating first and third quartile for n=666 to 763 target sites per PAM (see Supplementary Table 6 for exact values) (c) Relative editing activities on the subset of NANN PAMs in the library for SpCas9-CBE, CBE-NRRH, CBE-NRTH, and CBE-NRCH, and CBE-NG by nucleotide at the first, third, or fourth position of the PAM. Full data on the genomically integrated 11,776-member library can be found in Supplementary Table 5.

Figure 4.. Mammalian cell indel formation and DNA specificity of evolved Cas9 variants.

(a) Summary of indel formation efficiencies in HEK293T cells across 48 endogenous human sites containing NANH (H=non-G) PAMs for SpCas9-NRRH, -NRTH, -NRCH, and SpCas9-NG. Mean and standard deviation (SD) of all individual values of n=3 independent biological replicates are plotted. (b) Indel formation in primary human fibroblasts across give endogenous human sites containing NR PAMs for SpCas9-NRRH, -NRTH, -NRCH, and SpCas9-NG. Bars represent mean and standard deviation (SD) of n=3 independent biological replicates, with individual values shown as dots. (c) GUIDE-seq on-target reads (indicated by an asterisk) and off-target reads for SpCas9, xCas9, and evolved variants SpCas9-NRRH, -NRTH-, and NRCH at HEK site 4. Off-target reads that represent less than 1% of total reads are not shown but are available in Supplementary Table 4. (d) DNA targeting specificity of SpCas9, xCas9, and evolved variants SpCas9-NRRH, -NRTH-, and NRCH resulting from GUIDE-seq analysis on HEK site 4, VEGFA site 2, HEK site 1, and EMX1 in U2OS cells as determined by number of off-target sites detected. See Supplementary Fig. 6 for additional GUIDE-seq results.

Figure 5.. Mammalian cytosine and adenine base editing activity and scope of evolved variants and SpCas9-NG.

(a) Summary of cytosine base editing in HEK293T cells across 48 endogenous human sites containing NANH (H=non-G) PAMs for CBE-NRRH, CBE-NRTH, CBE-NRCH, and CBE-NG. Mean and SEM of three independent biological replicates are shown. (b) Adenine base editing in HEK293T cells across 27 endogenous human sites containing NANN PAMs for ABE-NRRH, ABE-NRTH, ABE-NRCH, and ABE-NG. Bars represent mean and SEM of three independent biological replicates, shown as dots. (c) Fraction of pathogenic SNPs in the ClinVar database that in principle can be corrected by a C•G to T•A (left) or A•T to G•C (right) base conversion using NR PAMs. (d) Mean number of possible sgRNAs capable of targeting each pathogenic SNP in the ClinVar database using NR, NG, or NGG PAMs. Mean and SEM of the number of targeting sgRNA are shown for n=3,919, n=12,095, n=1,154, n=9,740, n=1,132, or n=3,841 individual ClinVar entries for CBE/NR, ABE/NR, CBE/NG, ABE/NG, CBE/NGG, and ABE/NGG, respectively.

Figure 6.. Evolved SpCas9 variants enable correction of pathogenic SNPs with non-G PAMs.

(a) Overview of adenine base editing strategy for correcting the sickle hemoglobin (HbS) SNP. In HbS, the Glu (GAG codon) at position 6 of normal β-globin (HBB) is mutated to a Val (GTG codon). Targeting this SNP with adenine base editing on the reverse strand enables a Val to Ala (GTG to GCG) base conversion, leading to the rare Makassar β-globin variant (HbG) that is thought to be non-pathogenic. (b) Adenine base editing in HEK293T cells engineered with the HbS mutation using a CATG PAM by ABE-NRRH, ABE-NRTH, ABE-NRCH, and ABE-NG. This PAM places the target A at position 4, and an off-target A at position 6 that leads to a silent Pro (CCT) to Pro (CCC) mutation when converted to a G. (c) Adenine base editing in HEK293T cells engineered with the HbS mutation using a CACC PAM by ABE-NRRH, ABE-NRTH, ABE-NRCH, and ABE-NG. This PAM places the target A at position 7, and an off-target A, which leads to a silent pro (CCT) to pro (CCC) mutation, at position 9. For (b) and (c), bars represent mean and SEM of n=3 independent biological replicates, with individual values shown as dots. (d) Recommended Cas9 variants for accessing all possible PAMs within NRNN PAM space. Only Cas9s that require recognition of three or fewer defined nucleotides in their PAMs are listed. The variants evolved and characterized in this study are highlighted in blue.