EXO1 as a therapeutic target for Fanconi Anaemia, ZRSR2 and BRCA1-A complex deficient cancers

Abstract

Exonuclease EXO1 performs multiple roles in DNA replication and DNA damage repair (DDR). However, EXO1 loss is well-tolerated, suggesting the existence of compensatory mechanisms that could be exploited in DDR-deficient cancers. Using CRISPR screening, we find EXO1 loss as synthetic lethal with many DDR genes somatically inactivated in cancers, including Fanconi Anaemia (FA) pathway and BRCA1-A complex genes. We also identify the spliceosome factor and tumour suppressor ZRSR2 as synthetic lethal with loss of EXO1 and show that ZRSR2-deficient cells are attenuated for FA pathway activation, exhibiting cisplatin sensitivity and radial chromosome formation. Furthermore, FA or ZRSR2 deficiencies depend on EXO1 nuclease activity and can be potentiated in combination with PARP inhibitors or ionizing radiation. Finally, we uncover dysregulated replication-coupled repair as the driver of synthetic lethality between EXO1 and FA pathway attributable to defective fork reversal, elevated replication fork speeds, post-replicative single stranded DNA exposure and DNA damage. These findings implicate EXO1 as a synthetic lethal vulnerability and promising drug target in a broad spectrum of DDR-deficient cancers unaddressed by current therapies.

Fig. 1. Identification of cancer vulnerabilities synthetic lethal with loss of EXO1.

a Immunoblotting for EXO1 protein levels in eHAP iCas9 EXO1 KO clones (alpha-tubulin as control for equal loading; n = 2; MW molecular weight). b Viability of eHAP iCas9 EXO1 KO clones based on CellTitre-Glo assay (n = 4), mean with standard deviation (SD). c Sensitivity of eHAP iCas9 EXO1 KO clones to olaparib (n = 3, mean and error with SD). d Sensitivity of eHAP iCas9 EXO1 KO clones to cisplatin (n = 3, mean and error with SD). e Sensitivity of eHAP iCas9 EXO1 KO clones to ionising radiation (IR) (n = 3, mean and error with SD). f Scheme of genome-wide CRISPR dropout screen in eHAP iCas9 wild type and EXO1 KO isogenic pair of cell lines. g GO Biological Process (2023) term analysis for top synthetic lethal hits from the screen. Based on log fold change and significance parameters, 547 genes were selected from both time points. Significance of the enrichment for top GO terms is represented in a graph for −log of P value, which was calculated with Fisher’s exact test. h Overview of relevant categories of synthetic lethal hits. Three categories of genes of interest are described in separate boxes, with their identity listed in the respective box and denoted with the same colour as is used for the gene on the volcano plots in panels (i) and (j). i Volcano plot of ‘day 6’ comparison of EXO1 KO vs. WT cells. Data was plotted based on parameters from the CRISPR screen MAGeCK analysis (log2 fold change on the X-axis and −log10 MAGeCK score on the Y-axis), with genes of interest highlighted as in (h). j Volcano plot of ‘day 16’ comparison of EXO1 KO vs. WT cells. Equivalent to the comparison described in (i). Source data for a–e are provided as a Source Data file.

Fig. 2. EXO1 is synthetic lethal with FA pathway genes and BRCA1-A complex genes, which are somatically mutated in cancers.

a Validation of EXO1 and FANCG synthetic lethal interaction in eHAP iCas9 cells by clonogenic assay (representative images, n = 3, well diameter 15.5 mm). b Quantification of clonogenic assay (n = 3, mean with SD) in (a) with ordinary one-way ANOVA statistical analysis (F = 158.8). c Validation of EXO1 and FANCG synthetic lethal interaction in eHAP iCas9 cells by CellTitre-Glo viability assay (n = 3, mean with SD) with ordinary one-way ANOVA statistical analysis (F = 132.5). d Apoptosis of EXO1 and FANCG double KO cells (n = 3). e Quantification of cell cycle phases (n = 4, based on gating shown in Supplementary Fig. 5g, mean with SD). f Micronuclei increase in double KO cells. Paired t-test analyses were performed for each of the double KO to compare to ‘EXO1 KO + sgNT’ (n = 3, mean with SD, with each percentage calculated on the basis of >2181 nuclei; two-tailed P values). g Increase in phosphorylated RPA at serine 33 (RPA-pS33) foci in double KO cells. Dots in orange, blue and green represent the mean of a biological repeat, with the red line as their mean. Paired t-test analyses were performed on the means of biological repeat for each of the double KO cell lines in comparison to the ‘EXO1 KO + sgNT’ cell line (n = 3, mean with SD, with each percentage calculated on the basis of >754 nuclei; RPA-pS33 signal was measured within nuclei; two-tailed P values). h Immunoblotting of total cell extracts and chromatin for ‘WT + sgNT’, ‘WT + sgFANCG’, ‘EXO1 KO + sgNT’ and ‘EXO1 KO + sgFANCG’ eHAP iCas9 cells at day 5 post Cas9 expression induction. Samples were probed for phosphorylated levels of KAP1, p53, CHK2 and H2AX with controls, and EXO1, alpha-tubulin and histone H3 to control for sample identity and equal loading (representative experiment, n = 3). Source data for b–h are provided as a Source Data file.

Fig. 3. Loss of the spliceosome factor ZRSR2 confers replication stress, DNA damage and phenocopies FA pathway deficiency.

a Validation of EXO1 and ZRSR2 synthetic lethal interaction in eHAP iCas9 cells by clonogenic assay (representative images, n = 5, well diameter 15.5 mm). b Quantification of clonogenic assay (n = 5, mean with SD) in a with ordinary one-way ANOVA statistical analysis (F = 21.92). c Validation of EXO1 and ZRSR2 synthetic lethal interaction in eHAP iCas9 cells by CellTitre-Glo viability assay (n = 5, mean with SD) with ordinary one-way ANOVA statistical analysis (F = 17.84). d Volcano plot of the CRISPR screen ‘day 16’ comparison of EXO1 KO vs. WT cells with the indicated spliceosome factors. e Immunoblotting of total cell extracts and chromatin samples for eHAP iCas9 WT and ZRSR2 KO. Samples were probed for phosphorylated levels of RPA (pS33), CHK2 (pT68) with controls, as well as alpha-tubulin and histone H3 as a control for equal loading (representative experiment, n = 2). f Increase of RPA-pS33 foci in ZRSR2 KO cells in comparison to wild-type cells. Dots in orange, blue and green represent the mean of a biological repeat, with the red line as their mean (mean with SEM, n = 3, two-tailed paired t-test analysis on the mean of each biological repeat, each mean was calculated on the basis of >711 nuclei). g Increase of H2AX-pS139 signal in ZRSR2 KO cells in comparison to wild type cells. Dots in orange, blue and green represent the mean of a biological repeat, with the red line as their mean (mean with SEM, n = 3, two-tailed paired t-test analysis on the mean of each biological repeat, each mean was calculated on the basis of >2060 nuclei). h Immunoblotting analysis of total cell extracts and chromatin samples for eHAP iCas9 WT and ZRSR2 KO without and with cisplatin treatment. Samples were probed for FANCD2, phosphorylated levels of KAP1 (pS824) and CHK2 (pT68) with controls, as well as histone H3 as a control for equal loading (representative experiment, n = 3). i Representative images of metaphase spreads for eHAP iCas9 ‘WT + MMC’, ‘FANCG KO + MMC’ and ‘ZRSR2 KO + MMC’ (MMC – mitomycin C). Blue arrows point to chromosome breaks, and pink arrows point to radial chromosomes. Scale bars: 10 μm. j Quantification of phenotypes observed in metaphase spreads for eHAP iCas9 ‘WT + MMC’, ‘FANCG KO + MMC’ and ‘ZRSR2 KO + MMC’, quantification based on >100 metaphases per repeat and per condition (n = 3, mean with SD). k Sensitivity of eHAP iCas9 ZRSR2 KO clones to cisplatin (n = 3, mean with SD). l Sensitivity of eHAP iCas9 ZRSR2 KO clones to olaparib (n = 3, mean with SD). Source data for b, c, e–h, j–l are provided as a Source Data file.

Fig. 4. Single and combination therapy opportunities for EXO1 inhibitor development.

a Immunoblotting for EXO1 protein levels after complementation with either empty vector, vector constitutively expressing wild type EXO1 (EXO1-WT) or vector constitutively expressing catalytic dead mutant of EXO1 (EXO1-CD) (alpha-tubulin as control for equal loading). b Quantification of clonogenic assay for the complementation experiment for EXO1 and FANCG synthetic lethal interaction (n = 3, mean with SD) with ordinary one-way ANOVA statistical analysis (F = 41.71). c Quantification of clonogenic assay for the complementation experiment for EXO1 and ZRSR2 synthetic lethal interaction (n = 3, mean with SD) with ordinary one-way ANOVA statistical analysis (F = 17.95). d Heatmap of log fold change (FC) of IC50 values for survival of ‘EXO1 KO + sgNT’, ‘WT + sgFANCG’ and ‘EXO1 KO + sgFANCG’ in comparison to ‘WT + sgNT’ in eHAP iCas9 background after treatment with DNA-damaging agents and inhibitors of DDR factors. IC50 values were calculated from sigmoidal curve fits, and the change was calculated with ‘WT + sgNT’ value as reference (n = 3). e Additive sensitivity of eHAP iCas9 double KO cells for EXO1 and FANCG to olaparib (n = 3, mean with SD). Example of an experiment that was used to generate the heatmap in (d). f Heatmap of log fold change (FC) of IC50 values for survival of ‘EXO1 KO + sgNT’, ‘WT + sgZRSR2’ and ‘EXO1 KO + sgZRSR2’ in comparison to ‘WT + sgNT’ in eHAP iCas9 background after treatment with DNA-damaging agents and inhibitors of DDR factors; as in (d) (n = 3). g Additive sensitivity of eHAP iCas9 double KO cells for EXO1 and ZRSR2 to olaparib (n = 3, mean with SD). Example of an experiment that was used to generate the heatmap in (f). h Additive sensitivity of eHAP iCas9 double KO cells for EXO1 and FANCG to ionising radiation (n = 3, mean with SD). i Additive sensitivity of eHAP iCas9 double KO cells for EXO1 and ZRSR2 to ionising radiation (n = 3, mean with SD). j Sigmoidal plot for ‘EXO1 KO + sgFANCG vs EXO1 KO + sgNT’ comparison in CRISPR rescue screen. Data was plotted based on MAGeCK analysis, with highlighted genes of interest. k Sigmoidal plot for ‘EXO1 KO + sgZRSR2 vs. EXO1 KO + sgNT’ comparison in CRISPR rescue screen. Data was plotted based on MAGeCK analysis, with highlighted genes of interest. l Rescue of ‘EXO1 KO + sgFANCG’ and ‘EXO1 KO + sgZRSR2’ sensitivity to TYMS inhibitor Pemetrexed (clonogenic assay) in comparison to ‘WT + sgNT’, ‘EXO1 KO + sgNT’, ‘WT + sgFANCG’ and ‘WT + sgZRSR2’ (n = 3, mean with SD). Source data for a–i, l are provided as a Source Data file.

Fig. 5. EXO1-FANCG double KO cells are deficient in replication fork reversal, which induces faster replication forks and results in synthetic lethality.

a Representative images of DNA fibres of ‘WT + sgNT’, ‘WT + sgFANCG’, ‘EXO1 KO + sgNT’, ‘EXO1 KO + sgFANCG’ after 5 days of Cas9 induction (n = 3), with an example of a staining for a DNA fibre (taken from image of ‘WT + sgNT’) that was measured for total fork speed. Scale bars: 10 μm. b Quantification of replication fork speed from the DNA fibres experiment shown in (a). Measurements of a minimum of 300 fibres per condition per biological repeat (n = 3). Black points represent the mean from each biological repeat, with red lines representing the mean with the standard deviation of those. Means are also noted below the X-axis in the numerical value. An ordinary one-way ANOVA statistical analysis was performed on the means of biological repeats (F = 6.513). c Representative electron micrographs of normal replication fork and reversed replication forks (n = 2). Scale bar for large panels: 250 nm, small panels: 50 nm. P: parental strand, D: daughter strand, and R: reversed strand. d Quantification of fork reversal in ‘WT + sgNT’, ‘WT + sgFANCG’, ‘EXO1 KO + sgNT’, ‘EXO1 KO + sgFANCG’ after 5 days of Cas9 induction, with and without hydroxyurea (HU) treatment following induction; percentage of fork reversal plotted. Error bars are representative of the mean with standard deviation from two independent experiments. e Representative electron micrographs of replication fork with extensive ssDNA behind the fork obtained from ‘EXO1 KO + sgFANCG’ cells after 5 days of Cas9 induction (n = 2). Scale bar for large panels: 250 nm, small panels: 50 nm. P: parental strand, D: daughter strand. f Quantification of ssDNA gaps in ‘WT + sgNT’, ‘WT + sgFANCG’, ‘EXO1 KO + sgNT’, ‘EXO1 KO + sgFANCG’ after 5 days of Cas9 induction, with and without HU treatment following induction. Gaps observed per fork are quantified as 0, 1, 2 and more than 2. Error bars are representative of the mean with standard deviation from two independent experiments. g Quantification of fork degradation rate (IdU to CldU ratio) via DNA fibres experiment following 4 mM HU treatment without and with mirin (representative experiment from three independent experiments, red lines represent median values). Measurements of a minimum of 100 fibres per condition per biological repeat. h Quantification of replication fork speed from DNA fibre experiment without and with TYMS inhibitor Pemetrexed. Measurements of a minimum of 300 fibres per condition per biological repeat (n = 3). Black points represent the mean from each biological repeat, with red lines representing the mean with the standard deviation of those. Means are also noted below the X-axis in the numerical value. Two-way ANOVA statistical analysis was performed on the means of biological repeats. Source data for b, d, f and h are provided as a Source Data file.

Fig. 6. Model for mechanistic causes of synthetic lethality with loss of EXO1 in FA-deficient cells.

In cells with a functional Fanconi Anaemia pathway, replication forks that encounter a DNA lesion can undergo fork reversal as a way of safeguarding genome integrity, ensuring sustainable replication fork speed and cell viability. However, in FA-deficient cells, in which genome stability is challenged with a higher incidence of DNA lesions, this mechanism of genome stability maintenance critically relies on EXO1. Loss of EXO1 in FA-deficient cells leads to a defect in replication fork reversal, elevated replication fork speed and an increase in post-replicative ssDNA gap formation, which ultimately contributes to DNA damage and results in cell death.