EXO1 protects BRCA1-deficient cells against toxic DNA lesions

Abstract

Inactivating mutations in the BRCA1 and BRCA2 genes impair DNA double-strand break (DSB) repair by homologous recombination (HR), leading to chromosomal instability and cancer. Importantly, BRCA1/2 deficiency also causes therapeutically targetable vulnerabilities. Here, we identify the dependency on the end resection factor EXO1 as a key vulnerability of BRCA1-deficient cells. EXO1 deficiency generates poly(ADP-ribose)-decorated DNA lesions during S phase that associate with unresolved DSBs and genomic instability in BRCA1-deficient but not in wild-type or BRCA2-deficient cells. Our data indicate that BRCA1/EXO1 double-deficient cells accumulate DSBs due to impaired repair by single-strand annealing (SSA) on top of their HR defect. In contrast, BRCA2-deficient cells retain SSA activity in the absence of EXO1 and hence tolerate EXO1 loss. Consistent with a dependency on EXO1-mediated SSA, we find that BRCA1-mutated tumors show elevated EXO1 expression and increased SSA-associated genomic scars compared with BRCA1-proficient tumors. Overall, our findings uncover EXO1 as a promising therapeutic target for BRCA1-deficient tumors.

Keywords: BRCA1; DNA double-strand break repair; EXO1; cancer; homologous recombination; single-strand annealing; synthetic lethality.

Figure 1.. EXO1 loss is synthetically lethal with BRCA1 deficiency

(A) Selected results of a gene essentiality screen in BRCA1-proficient and -deficient RPE1 cells. Plotted is the CCA score: a higher score indicates a unique essentiality in BRCA1-deficient cells compared with -proficient cells. Dashed line indicates the cut-off for a significant CCA score (based on Adam et al.). (B) RPE1 hTERT cells expressing Cas9, either TP53^−/− (black lines) or TP53^−/− BRCA1^−/− (blue lines), were infected with indicated sgRNA together with GFP, or with an empty vector together with mCherry. GFP- and mCherry-positive cells were mixed 1:1, and the frequency of GFP-positive cells in the population was determined at multiple time points (n = 4, mean ± SEM). Western blot of lysates shown in Figure S1A. (C) RPE1 hTERT PAC^−/− TP53^−/− cells were genetically modified to generate a BRCA1-mAID-GFP fusion gene at the endogenous BRCA1 locus. In this genetic background, two clonal EXO1^−/− cell lines were generated. Cells were treated with auxin (500 μM) for 48 h or left untreated, and lysates were analyzed by western blotting. (D) The cell lines described in (C) were treated with 500 μM auxin, or left untreated, and clonogenic survival was determined. Right panel shows a representative experiment, left panel shows the quantification (n = 4, mean + SEM, ns = p > 0.05, *p < 0.05, **p < 0.01, paired t test). (E) BRCA1-mutated MDA-MB-436 cells, either WT or reconstituted with BRCA1 cDNA, were infected with empty vector (CTRL) or EXO1-targeting sgRNA and viability was measured using CellTiter-Glo (n = 3, mean + SD, **p < 0.01, paired t test). (F) Lysates of the MDA-MB-436 cell lines described in (E) were analyzed by western blotting. (G) RPE1 hTERT TP53^−/− BRCA1-mAID-GFP cell lines were virally transduced to express Cas9 cDNA and the indicated gRNAs, followed by a clonogenic survival assay in presence or absence of 500 μM auxin (n = 3, mean + SD, ns = p > 0.05, **p < 0.01 paired t test). Western blot of lysates shown in Figure S2K.

Figure 2.. Elevated EXO1 expression in BRCA1-deficient tumors

(A) Gene expression of the indicated genes in BRCA1 WT or mutant tumors of a breast cancer cohort (red line indicates median, ns = p > 0.05, **p < 0.01, ***p < 0.001, Kolmogorov-Smirnov). (B) For breast cancer tumors of the TCGA cohort (n = 1,048), the correlation between the HRD score and expression level of an individual gene was determined. Plotted are the Pearson correlation coefficient for each of the 60,000 measured transcripts. (C) HRD score and expression of the indicated genes were plotted for each tumor sample in the TCGA breast cancer dataset (n = 1,048). Red line is the linear regression curve, PC = Pearson correlation coefficient.

Figure 3.. EXO1 loss is not lethal in BRCA2-deficient cells

(A) A published gene essentiality screen in BRCA2-proficient and -deficient DLD1 cells was mined to extract the CCA scores for the indicated genes. A higher CCA score indicates a unique essentiality in BRCA2-deficient cells compared with proficient cells. Dashed line indicates the cut-off for a significant CCA score (based on Adam et al.). (B) DLD1 WT and BRCA2^−/− cells were infected with empty vector (CTRL) or EXO1-targeting sgRNA and viability was measured using CellTiter-Glo (n = 3, mean + SD, ns = p > 0.05, paired t test). (C) Lysates of the indicated DLD1 cells indicated in (B) analyzed by western blotting. (D) H1299 cells carrying a doxycycline-inducible BRCA2 shRNA were transduced with an AAVS1-targeting (CTRL) or EXO1-targeting sgRNA, followed by a clonogenic survival assay in the absence or presence of 10 μg/mL doxycycline (n = 3, mean + SD, ns = p > 0.05, paired t test). (E) Western blot analysis of the lysates of the H1299 TetOn shBRCA2 cells studied in (D). (F) CHORD analysis was applied to determine the HR-status of tumors of a pan cancer dataset (n = 1,823) and EXO1 expression was plotted (red line indicates median, ns = p > 0.05, **p < 0.01, Kruskal-Wallis with Dunn’s multiple comparisons test).

Figure 4.. EXO1 depletion causes increased S phase PARylation

(A) RPE1 hTERT TP53^−/− BRCA1^−/− cells expressing Cas9, either TP53BP1^+/+ or TP53BP1^−/−, were transduced with an AAVS1-targeting (CTRL) or EXO1-targeting sgRNA, followed by a clonogenic survival assay (n = 3, mean + SD, **p < 0.01, paired t test). Western blot of lysates shown in Figure S4A. (B) Analysis of replication fiber length after incubation with and without S1 nuclease. Left panel shows setup of experiment (top) and images of representative fibers (bottom), right panel shows quantification (n = 3, at least 100 fibers were analyzed per condition, per replicate; black or red dots indicate median per experiment and the line indicates their average, ns = p > 0.05, ****p < 0.0001, Kruskal-Wallis followed by Dunn’s multiple comparison test). (C) Indicated RPE1 cell lines were incubated with EdU and PARGi, followed by IF microscopy to quantify PAR levels. A representative of two independent experiments is shown, black line indicates median (****p < 0.0001, Mann Whitney).

Figure 5.. Genomic instability in BRCA1-deficient cells is exacerbated by EXO1 depletion

(A) Nuclear γH2AX intensity in S phase (EdU+) cells was analyzed by IF microscopy in BRCA1-mutated MDA-MB-436 cells (+BRCA1-reconstituted with BRCA1 cDNA). A representative of two independent experiments is shown, black line indicates median (****p < 0.0001, Mann Whitney). (B) RPE1 hTERT TP53^−/− BRCA1-mAID-GFP cells, either EXO1^+/+ or EXO1^−/−, were treated with 500 μM auxin for 48 h to deplete BRCA1 or left untreated. Nuclear γH2AX intensity in S phase (EdU+) cells was analyzed by automated IF microscopy imaging. A representative of two independent experiments is shown, black line indicates median (****p < 0.0001, Mann Whitney). Figure S5A shows representative microscopy images. (C) BRCA1-mutated MDA-MB-436 cells were infected with empty vector (CTRL) or EXO1-targeting sgRNA. This was followed by DAPI staining and microscopic quantification of the number of micronuclei (n = 3, mean + SD, *p < 0.05, ratio paired t test). (D) RPE1 hTERT TP53^−/− BRCA1-mAID-GFP cells, either EXO1^+/+ or EXO1^−/−, were treated with 500 μM auxin for 48 h to deplete BRCA1 or left untreated. This was followed by Hoechst staining and microscopic quantification of the number of micronuclei (n = 5, mean + SD, ns = p > 0.05, *p < 0.05, ratio paired t test). (E) MEFs with a homozygous exon 11 deletion in the BRCA1 gene (Δ11) were infected with control or EXO1-targeting shRNA, followed by metaphase spread analysis (n = 3, >40 metaphases per replicate, mean + SD, *p < 0.05, ratio paired t test).

Figure 6.. Survival of BRCA1-deficient cells is dependent on the function of EXO1 in SSA

(A) Indicated RPE1 cells expressing eGFP-RAD52 were transfected with a control (siCTRL) or EXO1-targeting siRNA, exposed to IR (10 Gy) and analyzed by IF microscopy for RAD51 and RAD52 foci formation 3 h post-IR. Left panel shows foci quantification (n = 3, mean + SD, *p < 0.05, paired t test), middle panel shows the mean eGPP-RAD52 foci intensity (a representative of three independent experiments is shown, black lines indicate mean) and right panel shows representative microscopy images of eGFP-RAD52 foci. Western blot of lysates shown in Figure S6A. (B) HEK293T cells carrying the DSB-Spectrum_V3 reporter, either WT or EXO1^−/−, were transfected with a control (siCTRL), BRCA1- or BRCA2-targeting siRNA, followed by a second round of transfection with a Cas9 cDNA and sgRNA targeting the reporter locus. Next, cells were analyzed by flow cytometry to quantify repair by the indicated pathways (n = 4, mean ± SEM, ns = p > 0.05, *p < 0.05, ratio paired t test). Western blot of lysates shown in Figure S6G. (C) HEK293T DSB-Spectrum_V3 cells, either WT control (sgAAVS1) or depleted for EXO1 (sgEXO1) were transfected with indicated siRNAs. Next, cells were transfected with Cas9 cDNA and an HBB-targeting sgRNA. The SSA repair product was PCR-amplified and analyzed by agarose gel electrophoresis. Representative agarose gel of three replicates. (D) Clonogenic survival assay of Cas9-expressing RPE1 hTERT TP53^−/− BRCA1^−/− cells that were transduced to express the indicated sgRNAs (n = 6, mean + SD, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, one-way ANOVA, post hoc Dunnett’s, compared with CTRL). Western blot of lysates shown in Figure S7A. (E and F) As in (B), but now for DSB-Spectrum_V3 reporter cells depleted for RAD52, XPF, or ERCC1 (HR: E, SSA: F) (n = 5, mean ± SEM, ns = p > 0.05, *p < 0.05, **p < 0.01, ratio paired t test). Western blot of lysates shown in Figure S7D. (G and H) RPA foci in S phase (EdU+) cells were analyzed by IF microscopy in BRCA1-mutated MDA-MB-436 cells (G) or DLD1 BRCA2^−/− cells (H) infected with empty vector (CTRL) or EXO1-targeting sgRNA 3 h post-5 Gy IR (n = 3, mean, ns = p > 0.05, ****p < 0.0001, Mann Whitney). RPA foci formation in BRCA1-complemented MDA-MB-436 control samples is shown in Figure S7F.

Figure 7.. BRCA1-deficient tumors have more SSA scars than BRCA1-proficient tumors

(A) Model of the mechanism causing synthetic lethality between BRCA1-deficiency and EXO1 loss. Levels of transparency indicate levels of pathway activity. (B) Whole genome sequencing data of pan cancer tumor samples65 was analyzed to quantify the number of genetic scars indicative of DSB repair by SSA, here defined as deletions flanked by homologous sequences of >10 bp. CHORD analysis was used to determine HR-status3 (****p < 0.0001, Kruskal-Wallis with Dunn’s multiple comparisons test). (C) Tumor samples from a pan-cancer cohort were binned based on SSA scar count, and the EXO1 expression was plotted for each tumor sample.