Fig. 7: PAM scanning influences genome-wide transcriptional modulation in GBS.

RNA-seq was performed on CNCTC 10/84 Cas9 variants in triplicate biological replicates at two phases of growth (mid-log, O.D.₆₀₀ = 0.6; early stationary, O.D.₆₀₀ = 1.2). Upset plots of DESeq2 analysis after sequence data normalization and strain alignment showed that the majority of differentially regulated genes emerged from comparisons between PAM non-scanning strains (Δcas9 or scas9) and PAM scanning strains (WT or dcas9), which are highlighted in the red bars (A). Pearson correlation between scanner v. non-scanner comparisons was significantly higher than comparisons between scanner v. non-scanner and either scanner v. scanner or non-scanner v. non-scanner (B, two-tailed Student’s t test on R² values < 0.0001). Clustering of scanner v. non-scanner comparisons, at both growth time points, is illustrated in the principal component analysis of whole-genome RNA-seq data in (C). Gene set enrichment analysis of scanner v. non-scanner comparisons at both time points demonstrated decreased expression of genes in cluster of ortholog (COG) functional groups G (nucleotide metabolism/transport) and V (defense) in PAM scanning strains relative to non-scanners. COG group F (carbohydrate metabolism/transport) showed more expression in the scanners than non-scanners at both time points but was only significantly different (with a false discovery rate ≤ 0.10) at early stationary phase (D). Leading edge analyses of COG groups F, G, and V are illustrated as heat maps (E), where the values plotted are variance stabilizing transformation (VST) counts, which give a relative measure of gene expression levels that are normalized (for sequencing depth) and stabilized (for variance across sample replicates) allowing comparisons between different sample conditions⁷². There is no across-row or across-column scaling of these values, such that cross-comparison between rows, columns, and plots are valid.