PPP2R1A mutations cause ATR inhibitor sensitivity in ovarian clear cell carcinoma

Abstract

Identification of ARID1A/ATR synthetic lethality led to ATR inhibitor phase II trials in ovarian clear cell carcinoma (OCCC), a cancer of unmet need. Using multiple CRISPR-Cas9 mutagenesis and interference screens, we show that inactivation of protein phosphatase 2A (PP2A) subunits, including PPP2R1A, enhance ATRi sensitivity in ARID1A mutant OCCC. Analysis of a new OCCC cohort indicates that 52% possess oncogenic PPP2R1A p.R183 mutations and of these, one half possessed both ARID1A as well as PPP2R1A mutations. Using CRISPR-prime editing to generate new isogenic models of PPP2R1A mutant OCCC, we found that PPP2R1A p.R183W and p.R183P mutations cause ATRi-induced S phase stress, premature mitotic entry, genomic instability and ATRi sensitivity in OCCC tumour cells. p.R183 mutation also enhanced both in vitro and in vivo ATRi sensitivity in preclinical models of ARID1A mutant OCCC. These results argue for the assessment of PPP2R1A mutations as a biomarker of ATRi sensitivity.

Fig. 1. Genome wide CRISPR-Cas9 screens identify PP2A subunits as genetic determinants of ATR inhibitor sensitivity.

A, B Dose response curves illustrating the ATRi sensitivity of OCCC TOV21G cells. Cells were plated in 384 well plates and exposed to either AZD6738 (A) or VX970 (B) for five continuous days, at which point cell viability was assessed using CellTiter-Glo®. ARID1A^+/+ and ARID1A^–/– HCT116 cells are included as controls. Error bars represent standard error of the mean (SEM) from 5 replicates. C Screen schema for genome wide CRISPR-Cas9 mutagenesis screen performed in TOV21G cells. TOV21G cells expressing a doxycycline-inducible SpCas9 transgene were used. AZD6738 SF50 was 0.1 µM. D Waterfall plot showing the ranked gene-level NormZ scores from the genome wide CRISPR-Cas9 screen. PPP2R2A and PPP2CA were identified as sensitising ”hits” along with several other genes previously known to cause both sensitivity to ATRi such as POLE3, POLE4, CHEK1 and RAD9A; known resistance-causing effects such as CDK2 and CDC25B were also identified. E Guide-level Z scores from the genome wide screen for sgRNAs designed to target PPP2R2A or PPP2CA. F Experimental scheme for additional CRISPR mutagenesis (CRISPRn) (1, 2) or CRISPR interference (3,4) (CRISPRi) screens for genetic determinants of ATRi sensitivity. Non-tumour epithelial MCF10A p53^mutant cells, with a deleterious P53 mutations, expressing a doxycycline-inducible Cas9 transgene or constitutive dCas9-KRAB transgene were used as shown. AZD6738 SF80 in MCF10A p53mutant cells was 1 µM. VX970 SF80 was 0.1 µM. G Violin plots showing the quantile normalized NormZ scores for either 18,009 genes (CRISPRn screens) or 18,905 genes (CRISPRi screens) from the screens described in and labelled (1–4) as in F. NormZ scores for PPP2R1A (green), PPP2R2A (red) and PPP2CA (blue) and ATM (magenta; positive control) are highlighted. Dotted line represents a Z –2 threshold used to define “hits”. H Scatter plot showing the gene-level Norm Z score from the CRISPRi screens performed with AZD6738 and VX970 described in E and F. Data points for PPP2R1A (green), PPP2R2A (red) and PPP2CA (blue) and ATM (magenta; positive control) are highlighted.

Fig. 2. PPP2R1A mutations in OCCC cause ATRi synthetic lethality.

A Model of the PP2A holoenzyme (PDB 2NPP) with scaffolding subunit shown in blue, regulatory subunit shown in green and catalytic subunit shown in yellow. Somatic missense mutations in PPP2R1A (scaffolding subunit), highlighted in magenta, are conserved across multiple cancer types and are involved in direct interactions with regulatory B-subunits of the PP2A holoenzyme. Oncoprints illustrating PPP2R2A deletions in serous ovarian cancer (B) or PPP2R1A mutations (52% of cases) in ovarian clear cell carcinoma (OCCC – shown in C). Oncoprint in B generated from data in (ICGC/TCGA, 2020). Oncoprint in C generated by genotyping of 23 cases of OCCC from the Royal Marsden Hospital. D Heterozygous PPP2R1A missense mutations introduced to TOV21G cell line using CRISPR-prime gene editing. Cells were transfected with plasmids encoding prime editing guide (PEG), prime editor 2 (PE2) and nicking sgRNA before being single cell sorted, expanded and genotyped. Two resultant clones were generated with heterozygous PPP2R1A p.R183P or p.R183W missense mutations. DNA sequence traces from PPP2R1A mutant clones are shown. Each mutation is heterozygous, present in 5/10 TOPO-cloned sequences. E Western blot illustrating that PPP2R1A p.R183P mutation causes reduced levels of PPP2R2A and PPP2CA. Western blot performed from whole cell extracts from TOV21G PPP2R1A WT, p.R183P and p.R183W cells. F Volcano plot illustrating results from phosphoproteomic profiling of PPP2R1A mutant cells, indicating a significant enrichment (Log2 fold change >0, unpaired t-test <0.05) of known PP2A phosphorylation sites in SPRY1 (two sites) HDAC5 and IL6ST (highlighted). G Heterozygous PPP2R1A p.R183P or p.R183W mutation causes increased sensitivity to the ATR inhibitor AZD6738. Dose response curve for cells from D exposed to AZD6738 for two weeks. Error bars represent SEM from 4 replicates. Significance determined using two-way ANOVA.

Fig. 3. PPP2R1A p.R183W and p.R183P mutations cause ATRi-induced S phase stress, premature mitotic entry and genomic instability (previous page).

A FACS plots illustrating that PPP2R1A R183P mutation and ATRi exposure causes a significant increase in the non-replicating S phase fraction at the expense of the active (replicating) S phase fraction. Quantification of changes in fraction of cells in replicating S phase (B) or non-replicating S phase (C) from the experiment described in A. Error bars represent mean and SEM from three replicates. Pairwise significance determined by two-way ANOVA with Sidak correction for multiple comparisons. D FACS plots illustrating that PPP2R21A mutant cells more readily enter mitosis following AZD6738 exposure and are most likely to do so with sub-4n DNA content. Quantification of phospho-histone H3 positive cells (E) and sub 4n fraction of cells (F) from experiment shown in D. Error bars represent mean and SEM from three replicates. Pairwise significance determined by two-way ANOVA with Sidak correction for multiple comparisons. G Heterozygous PPP2R1A p.R183P mutation causes the formation of 53BP1 bodies. TOV21G PPP2R1A p.R183P isogenic cells were exposed to either DMSO or AZD6738 (500 nM) for 24 h before being fixed and immunostained for 53BP1, cyclin A and with DAPI. Representative confocal microscopy images are shown. Scale bar represents 10 µM. H Quantification of 53BP1 bodies from experiment described in G. Error bars represent mean and SEM from three replicates. Pairwise significance determined via Two-way ANOVA with Sidak’s correction for multiple comparisons. I Exposure to ATRi leads to a significant increase in the duration of metaphase in cells with a PPP2R1A mutation. Quantification of the duration mitotic cells spend in metaphase. Viable TOV21G PPP2R1A WT, p.R183P or pR183W were labelled with live cell stains for DNA, tubulin and actin before being exposed to AZD6738 (500 nM) or DMSO for 24 h with an image being captured every 5 min. Error bars represent SEM for 50 mitotic cells. Pairwise significance determined via Two-way ANOVA with Sidak’s correction for multiple comparisons (J) Exposure to ATRi leads to a significant increase in the proportion of cells with micronuclei in cells with a PPP2R1A mutation. TOV21G PPP2R1A isogenic cells were exposed to either DMSO or AZD6738 (500 nM) for 24 h before being fixed and stained with DAPI. Error bars represent SEM from three triplicate experiments. Pairwise significance determined via Two-way ANOVA with Sidak’s correction for multiple comparisons.

Fig. 4. PPP2R1A/ATRi synthetic lethality operates in vivo.

A In vivo study schema. Luciferase tagged PPP2R1A WT or p.R183W TOV21G cells were introduced into recipient mice by intraperitoneal injection. Treatment with either AZD6738 or vehicle commenced after tumours had established (after day six post IP injection) using on a five days on, two days off schedule. A low dose of ATRi (25 mg/kg) was used in these experiments to account for the pre-existing ARDI1A/ATR inhibitor synthetic lethality [15]. Total flux luminescence was measured twice per week. B Treatment with AZD6738 does not impair the growth of PPP2R1A WT tumours. Line plots showing fold change in luminescence compared to pre-treatment level for mice xenografted with TOV21G PPP2R1A WT cells. Error bars represent mean and SEM. Significance determined by two-way ANOVA. C Treatment with AZD6738 impairs the growth of PPP2R1A mutant tumours. Line plots showing fold change in luminescence compared to pre-treatment level for mice xenografted with TOV21G PPP2R1A p.R183W cells. Error bars represent mean and SEM from 15 mice. Significance determined by two-way ANOVA. D Representative images of vehicle and AZD6738 treated mice taken 14 days post initiation of treatment. E, F Low dose ATRi treatment had minimal effects on animal body weight. Weight of each animal expressed as a percentage relative to weight at time of tumour implantation.