Fig. 6: Beyond p53: experimental evidence identifying KRAS as another major cancer driver whose mutation can be potentially selected for by CRISPR-Cas9.

a CRISPR-Cas9 and CRISPRi screens of the top KRAS CDE gene knockouts were performed in isogenic MOLM13 and MOLM13-KRAS-G12D cell lines. The box plot shows the trend that the sgRNAs of the KRAS CDE+ genes are more depleted in KRAS WT cells vs KRAS mutant cells and vice versa for KRAS CDE- genes in CRISPR-Cas9 screens, but there is no such trend in the CRISPRi screens. The P values shown are of one-tailed Wilcoxon signed-rank tests. Top 200 candidates of both CDE+ and CDE- genes were used to derive this statistic with N = 10 unique sgRNA per gene. b Analysis of published genome-wide CRISPR-Cas9 and shRNA screens in KRAS-isogenic DLD1 cell line. The box plot shows the trend that the CRISPR-KO of KRAS CDE+ genes reduces cell viability more in KRAS WT cells than KRAS mutant cells, while there is no such trend in the shRNA screen. The P values of one-tailed Wilcoxon signed-rank tests are shown. 861 CDE+ and 185 CDE-genes were used to derive this statistic with N = 3 unique sgRNA per gene. c The mean difference in the percentage of KRAS mutant cells between Day 15 and Day 0 in co-culture (y-axis), under the CRISPR-KO of different KRAS CDE+ genes (x-axis). The y-axis values were obtained by fitting a linear model for each gene, with the percentage of KRAS mutant cells as dependent variable and time (day, as a continuous variable) and sgRNA as independent variables. The linear model coefficients associated with the variable "day" multiplied by 15 are plotted. Error bars represent standard errors of the coefficients as estimated from the linear models. NTC: non-targeting control sgRNA. N = 4 unique sgRNAs were used per gene. NTC: non-targeting control sgRNA. d A box plot showing the comparison of stable Cas9 activity measured via GFP reporter assay10 in N = 1375 WT vs N = 226 KRAS (distribution of total data points) mutant cancer cell lines, two-sided Wilcoxon rank-sum test P-value 2.4E–04. e A scatter plot showing the difference in Cas9 activity between cell lines with a driver WT vs mutant for each driver (effect size, x-axis) and a corresponding Wilcoxon two-sided significance for this difference (negative log10-P-value, y-axis). f The change in mutant allele frequency (x-axis) of the KRAS mutations detected in different cell lines (cell line-mutation pair on the y-axis) after induced Cas9 expression, compared to the corresponding parental cell lines, based on data from Enache et al. [13]. The starts and ends of arrows correspond to the mutant allele frequencies in the parental and the Cas9-expressed cell lines, respectively. Cases of increased allele frequency are colored in red, and those with decreased frequency are colored in blue. g A scatter plot showing the median change in mutant allele frequency after induced Cas9 expression across all cell lines in [13] (x-axis) and the corresponding Wilcoxon signed-rank test significance (negative log10-P value, y-axis) for the 61 major cancer driver genes from Vogelstein et al. [27]. The p-values are calculated using two-sided Wilcoxon Rank Sum tests unless not specified otherwise. In the boxplots of panels a, b, and d, the center line, box edges, and whiskers denote the median, interquartile range, and the rest of the distribution in respective order, except for points that were determined to be outliers using a method that is a function of the interquartile range, as in standard box plots.