Fig. 4: Prime editing scanning screen of patient-derived variants identify EGFR-activating mutations in MCF10A cells.
From: Multimodal scanning of genetic variants with base and prime editing
a, Prime editing efficiency in MCF10A∆MLH1 cells expressing the PEmax enzyme. EGFR-targeting pegRNAs introducing 14 pathogenic variants were designed using PRIDICT2.0 (ref. 68). Cells were infected with the indicated pegRNAs in three biological replicates and editing efficiencies were measured by deep sequencing of the respective target sites on days 7, 14 and 22 after transduction. b, Schematic of the prime editing screening approach. Variants impacting EGFR exons or UTRs and shorter than 11 nucleotides were obtained from ClinVar, COSMIC and the base editing screens. Whenever possible, additional synonymous mutations were added to the adjacent codon and/or the Cas9 PAM sequence was mutated to increase prime editing efficiency. For each combination, pegRNAs introducing only the intended edit or the synonymous mutation were added to the library as controls. Each pegRNA was designed with three different PBS lengths and a 24-nt RTT. Lentiviral pegRNA library and PEmax delivery vectors are shown. c, Violin plot showing individual barcode counts per million derived from the deep sequencing of the final pegRNA library (n = 1) according to the PBS length of the corresponding pegRNAs. Solid horizontal lines represent population medians and dashed lines represent 0.25 and 0.75 quantiles. Generalized linear model (Poisson regression) coefficients: PBS length, −3.26 × 10−1 (P < 2 × 10−16); intercept = 9.82 (P < 2 × 10−16) on 53,007 degrees of freedom. d, LFCs of individual pegRNAs between nontreated and EGF-deprived cells along the EGFR protein. Colors denote the variant classifications for pegRNAs with −log10(FDR) > 2. e, Deep sequencing data for individual pegRNA validation. PEmax-expressing MCF10A∆MLH1 cells were infected with lentivirus encoding the indicated pegRNAs in triplicate and deprived of EGF from day 7 to day 15. The percentages of reads that correspond to perfect edits, including additional synonymous mutations, are shown, as well as P values for two-sided t-tests between nontreated and EGF-deprived samples on day 15.
