Reply to “CRISPR screens are feasible in TP53 wild‐type cells”

Abstract

In their analyses, one out of three WT experiments (“UBC”, using TKOv3) outperforms their single p53 null experiment (done with TKOv2), likely because (i) it uses more partial replicates, (ii) the experimental design is more advanced, using a bottleneck after transduction to decrease clonal variance (Michlits et al, ; Schmierer et al, ), (iii) it uses better guide‐RNA library, and (iv) the measure used selectively exaggerates its performance due to overfitting (the guides for TKOv3 but not TKOv2 were designed based on the essential gene set they use to test performance). Under such synthetic tests, the true extent of the variance caused by p53 is normalized away. [...]consistent with a p53 response hampering screens in normal (untransformed) human cells, MAGeCK analysis of the raw data reveals that under both our and Brown et al's screen designs the performance of the p53 null RPE‐1 is higher than that of RPE‐1 p53 WT. [...]the very high Cas9 expression in their cells (Zimmermann et al, ) is a plausible explanation for why both screens detected the positive selection for loss of p53, but only our screens clearly distinguished between the drop‐out performance of p53 wild‐type and p53 null cells. [...]although our original work was focused on the implications of Cas9‐induced p53 activation for precision editing and for cell‐based therapies, we would like to take this opportunity to also strongly advise against ignoring p53—the guardian of the genome—in any study that makes use of reagents that introduce double‐strand DNA breaks.