Endogenous DNA 3′ blocks are vulnerabilities for BRCA1 and BRCA2 deficiency and are reversed by the APE2 nuclease

Summary

The APEX2 gene encodes APE2, a nuclease related to APE1, the apurinic/apyrimidinic endonuclease acting in base excision repair. Loss of APE2 is lethal in cells with mutated BRCA1 or BRCA2, making APE2 a prime target for homologous recombination-defective cancers. However, since the function of APE2 in DNA repair is poorly understood, it is unclear why BRCA-deficient cells require APE2 for viability. Here, we present the genetic interaction profiles of APE2-, APE1- and TDP1-deficiency coupled to a biochemical and structural dissection of APE2. We conclude that the main role of APE2 is to reverse blocked 3′ DNA ends, problematic lesions that preclude DNA synthesis. Our work also suggests that TOP1 processing of genomic ribonucleotides is the main source of 3′-blocking lesions relevant to the APEX2-BRCA1/2 synthetic lethality. The exquisite sensitivity of BRCA-deficient cells to 3′ blocks indicates that they represent a tractable vulnerability in homologous recombination-deficient tumor cells.

Graphical Abstract

eTOC blurb

Álvarez-Quilón and Wojtaszek et al propose that APE2 is the main human enzyme reverting endogenous DNA 3′ blocks, problematic lesions that preclude DNA synthesis. TOPI conversion of genomic ribonucleotides into 3′ blocked lesions underlies the APEX2-BRCA1/2 synthetic lethality. 3′-adducted DNA damage represents a tractable vulnerability in HR-deficient tumor cells.”

Introduction

Synthetic lethality is a type of genetic interaction in which loss of viability is observed only when two mutations co-occur (O’Neil et al., 2017). There is interest in oncology to develop drugs that recapitulate synthetic lethality, as they are predicted to have both an exquisite therapeutic window and an ability to target tumor suppressors, a class of cancer alterations that are largely refractory to precision oncology approaches (Hartwell et al., 1997; Kaelin, 2005; O’Neil et al., 2017). Despite this promise, there are only very few examples of drugs based on the concept of synthetic lethality and only a handful targets are currently disclosed. The best-known example is the lethality caused by loss of PARP1 in BRCA1 or BRCA2-mutated cells (Bryant et al., 2005; Farmer et al., 2005; Lord and Ashworth, 2017). This synthetic lethal interaction is not only recapitulated with poly (ADP-ribose) polymerase (PARP) inhibitors, it is in fact enhanced through the trapping activity of these agents (Lord and Ashworth, 2017).

The advent of CRISPR-based screens will likely lead to additional synthetic lethal interactions with therapeutic potential (Mair et al., 2019; O’Neil et al., 2017). One such example is the lethality caused by the loss of the APEX2 gene in cells with mutations in BRCA1 and BRCA2 (Mengwasser et al., 2019; Roy et al., 2011). While BRCA1/2 are well-established as key mediators of DNA repair by homologous recombination, the function of APEX2 is much less understood (Mengwasser et al., 2019; Roy et al., 2011). APEX2 codes for the APE2 protein, a paralog of the apurinic/apyrimidinic (AP) endonuclease APE1. APE2 is a poor AP endonuclease that displays robust 3′ phosphodiesterase and 3′–5′ exonuclease activities that may be involved in the repair of oxidative DNA lesions (Burkovics et al., 2009; Wallace et al., 2017) or the activation of ATR (Lin et al., 2018; Wallace et al., 2017; Willis et al., 2013). APE2 also differs from APE1 by the presence of a ssDNA-binding Zf-GRF domain and a PCNA-interacting motif (PIP) in its C-terminus, presumably involved in promoting the 3′–5′ exonuclease activity (Wallace et al., 2017). However, it remains unclear which activity of APE2 is necessary for its cellular function.

One noteworthy feature of the BRCA1/2-APEX2 synthetic lethality is that it occurs in the absence of exogenous DNA damage, indicating that endogenous DNA lesions of unknown origins are the root cause of the BRCA/APEX2 synthetic lethality. Candidate sources of DNA lesions include genome-embedded ribonucleotides, AP sites and other types of oxidative DNA base damage. Genomic ribonucleotides are especially intriguing for two reasons: first, they outnumber all other forms of aberrant nucleotides in DNA during an otherwise unperturbed cell cycle (Wallace and Williams, 2014). Secondly, loss of RNase H2, the enzyme that removes embedded ribonucleotides from DNA (Williams et al., 2016), is synthetic lethal with BRCA1/2-deficiency, indicating that genomic ribonucleotides can be a source of endogenous lesions necessitating HR for viability (Zimmermann et al., 2018).

One means by which embedded ribonucleotides cause DNA damage is following their incidental cleavage by topoisomerase I (TOP1), which can lead to the formation of a DNA-protein crosslink along with the formation of 3′ blocked single-stranded DNA breaks (SSBs) with terminal 2′,3′ cyclic ribonucleotides (Kim et al., 2011; Sekiguchi and Shuman, 1997; Williams et al., 2013). 3′ DNA blocks are hazardous given that the presence of a 3′ hydroxyl group is a requirement for DNA polymerization and ligation. Ribonucleotide processing, however, is not the only means by which 3′ blocks can occur in DNA. Other sources include the trapping of TOP1 near AP sites (Pourquier et al., 1997) or following the cleavage of AP sites by DNA lyases that result in unsaturated ribose moieties at the 3′ termini through an α, β-elimination reaction (Kim and Wilson, 2012). Cells therefore guard against the accumulation of 3′ blocks by relying on a dedicated set of enzymes that suppress formation of 3′ blocked lesions. These enzymes include RNase H2, APE1 and TDP1, which process embedded ribonucleotides, AP sites and TOP1 cleavage complexes (TOP1cc), respectively.

In this manuscript, we investigated the physiological role of APE2 in DNA repair by combining genome-scale genetic interaction studies with a biochemical and structural investigation of APE2. This work leads us to propose a mechanism for the basis of the synthetic lethality between APEX2 and BRCA1/2. We found that APE2 acts as a major 3′ end unblocking nuclease that overcomes a variety of endogenously arising 3′ blocking lesions, consistent with activities recently reported for the yeast homolog Apn2 (Li et al., 2019). We propose that it is this activity of APE2 that promotes the viability of HR-deficient cells. More generally, our results indicate that the 3′ blocked DNA lesions are a major vulnerability in BRCA1/2-deficient cells, and that modulation of their accumulation or repair could represent an attractive route for the development of new therapeutics against HR-deficient tumors.

RESULTS

Two converging lines of investigation led us to examine the function of APE2 in DNA repair. Firstly, single guide (sg) RNAs targeting APEX2 cause ATR inhibitor sensitivity in multiple cell lines, suggesting that APE2 prevents the accumulation of DNA lesions that challenge DNA replication (Hustedt et al., 2019b; Wang et al., 2018). Secondly, we observed that loss of APEX2 is synthetic lethal with BRCA1 or BRCA2 deficiency in genome-scale CRISPR screens (manuscript in preparation), a genetic interaction also reported by Mengwasser et al. (2019). Together, this work hinted that APE2 inhibition could be a potential therapeutic strategy for homologous recombination (HR)-deficient tumors and that APE2 plays a much more prominent role in promoting genome integrity than was suggested in previous studies.

We validated these findings in clonogenic survival and competitive growth assays, and observed that sgRNAs targeting APEX2 resulted in a loss of fitness in both engineered RPE1-hTERT p53^−/− BRCA1^−/− cells (these cells also expressed Cas9) and DLD-1 BRCA2^−/− cells, but not their BRCA⁺ counterparts (Figures 1AB and S1A). Furthermore, the prior introduction of an sgRNA-resistant wild type APEX2 transgene, but not a transgene expressing catalytically-inactive APE2-E48Q/D197N, rescued the APEX2-BRCA1 (Figures 1A) and APEX2-BRCA2 (Figure 1B) synthetic lethality, indicating that the catalytic activity of APE2 is essential in the BRCA1/2-deficient setting.

Figure 1. Characterization of the APEX2-BRCA1/2 synthetic lethality.

Clonogenic survival of RPE1 (A) and DLD1 cells (B) of the indicated genotypes expressing wild-type APE2 or the APE2 E48Q/D197N mutant (sgRNA-resistant) upon transduction with a sgRNA targeting APEX2. Shown are results of a two-way ANOVA with Dunnet post-test that compared the mock and APE2 E48Q/D197N mutant to the WT APE2 condition in the BRCA1^−/− (A) or BRCA2^−/− (B) cells (*** P ≤ 0.001). (C) Clonogenic survival of the indicated cell lines upon transduction with vectors expressing Cas9 and an sgRNA targeting LacZ (control), PARP1, or APEX2. APEX2 editing efficiency. % indel by TIDE is indicated. Also shown are results of a t-test that compared sgRNA vs LacZ sgRNA control in BRCA1 mutated cell lines (* FDR ≤ 0.05; ** FDR ≤ 0.01). (D) Clonogenic survival of RPE1-hTERT p53^−/− Cas9 cells of the indicated genotypes upon transduction with virus expressing an sgRNA targeting LacZ or APEX2. (E) Affinity purifications from 293T cell lysates transiently expressing C-terminal SBP/HA-tagged APE2 wild-type, Y396A/F397A or ΔPIP mutants. Bound proteins and whole-cell extracts (WCE) were immunoblotted with the indicated antibodies. (F) Quantitation of clonogenic survival of RPE1 wild-type or BRCA1^−/− cells expressing APE2 or the indicated mutants transduced with an sgRNA targeting endogenous APEX2. Also shown are the results of a two-way ANOVA with Dunnet post-test (APE2 mutants vs WT APE2) (* P ≤ 0.05). Data in all panels are shown as mean ± standard deviation (n ≥ 3 biologically independent experiments), circles. WT, wild-type. See Figure S1 for controls and additional data. See Table S6 for details on statistical tests.

We next assembled an 8-cell line panel consisting of four BRCA1-mutated cell lines (SUM149PT, JHOS-2, COV362 and UWB1.28) as well as four BRCA1⁺ cell lines (SK-OV3, OV-90, COV644 and RMUG-S). The cell lines were infected with a virus expressing either a non-targeting sgRNA designed against LacZ, one of two independent APEX2 sgRNAs, or an sgRNA against PARP1, a positive control. We observed that both APEX2-targeting sgRNAs led to a reduction in clonogenic survival selectively in the BRCA1-deficient cell lines (Figure 1C) and the extent of reduction in clonogenic survival matched the proportion of cells with indel mutations calculated by TIDE analysis (Brinkman et al., 2014). These results suggest that the loss of APE2 impairs the viability of BRCA1- and BRCA2-mutated cancer cells, irrespective of cell type.

The observation that APEX2 loss is synthetic lethal with both BRCA1 and BRCA2 deficiency suggest that cells with impaired HR are vulnerable to APE2 loss. Indeed, we observed that BRCA1^−/− 53BP1^−/− or BRCA1^−/− SHLDZ^−/− double-knockout cell lines, in which HR is partially restored due to loss of DNA end-protection (Noordermeer et al., 2018), were largely insensitive to APEX2 sgRNAs (Figure 1D). These results indicate that in the absence of APE2, cells rely on HR for viability.

APE2 consists of an endonuclease-exonuclease-phosphatase (EEP) nuclease core, a PCNA-interacting peptide (PIP) motif and a C-terminal ssDNA-binding Zf-GRF domain (Figure 1E). Both the PCNA-interaction and Zf-GRF domains play a role in stimulating processive 3ʼ–5ʼ exonuclease activity by APE2 (Burkovics et al., 2009; Wallace et al., 2017). We prepared versions of untagged APE2 that are unable to interact with PCNA (Figure 1E) either via the Y396A/F397A mutation or by complete deletion of the PIP motif (residues 390–397, ΔPIP). A version of APE2 that delete the Zf-GRF domain was also prepared (residues 442–516, ΔZf-GRF). When these APE2 variants were expressed in BRCA1^−/− cells also expressing APEX2-targeting sgRNAs, the PCNA-interaction mutants rescued the BRCA1-APEX2 synthetic lethality to wild type APE2 levels (Figures 1F and S1B) whereas the Zf-GRF deletion partially rescued synthetic lethality. In a similar set of experiments, RPE1-hTERT p53^−/− APEX2^−/− cells generated by gene editing (Hustedt et al., 2019a) (Figures S1C) were first transduced with lentiviruses expressing wild type or mutant APE2 and then subjected to two-color growth competition assays with sgRNAs targeting AAVS1 and BRCA2. As expected, catalytically inactive APE2 was unable to rescue the APEX2-BRCA2 synthetic lethality but the APE2 Y396A/F397A, ΔPIP and ΔZf-GRF variants rescued lethality as much as wild type APE2 (Figure S1D). We conclude that the APE2-PIP box and, by extension, processive 3ʼ–5ʼ exonucleolytic activity, are not essential for the viability of BRCA1/BRCA2-deficient cells. This conclusion differs from that of Mengwasser et al. (2019), but under closer examination the putative PIP-box mutation made in that study, R339A, resides outside the PIP-box motif (Figure 1E) perhaps explaining the discrepancy.

A genetic interaction map centered on APEX2

The repertoire of the biochemical activities of APE2 overlap with those of APE1 and TDP1. As a means to uncover unique roles for APE2, we determined and compared the genetic interaction profiles of an isogenic set of cell lines with mutations in APEX2, APEX1 and TDP1 (Figures S1C and S2A). The screens were performed in technical duplicates, and a score reflecting the likelihood of “essentiality” for each gene was computed using the BAGEL and JACKS algorithms (Figure S2B and Table S1) (Allen et al., 2019; Hart and Moffat, 2016). To identify genes that are essential in the knockout, but not parental cell lines, we applied a set of score thresholds described in the Methods section. This analysis identified 90, 44 and 7 genes whose inactivation caused loss of fitness in the APE2-, APE1 or TDP1-deficient background, respectively (Figure 2A and Table S2). These genes were enriched in DNA repair factor-coding genes, as determined by Gene Ontology term enrichment, suggesting that the screens probed their DNA repair function, as expected (Figure S2C). The observation that APEX2 deficiency engendered vulnerabilities to many more genes than APEX1 or TDP1 loss suggests that APE2 is important for protecting cells against endogenous DNA lesions. In support of this possibility, we observed that loss of APEX2 was lethal in p53-proficient RPE1-hTERT cells but viable in their p53-mutated counterparts (Figure 2B). This observation indicates that APE2 loss in RPE1 cells causes accumulation of endogenous DNA lesions that triggers p53-dependent cell death.

Figure 2. APEX2, APEX1 and TDP1 genetic interaction landscape.

(A) Genetic interaction with with APEX2, APEX1 or TDP1 (source nodes) determined by CRISPR screens (see Table S2). Thickness of the edges are scaled on ΔBF values. Source nodes are coloured in blue and DNA repair-coding genes in red. See Figures S1C and S2 for cell line validation and additional analysis of the screens. (B) Representative image of RPE1-hTERT Cas9 cells clonogenic survival assays of the indicated genotypes transduced with sgRNA targeting AAVS1 or APEX2. Data presented as mean ± standard deviation. The results of a t-test that compares sgAPEX2 to the sgAAVS1 condition (*** FDR ≤ 0.001) are shown. (C) Competitive growth assays in wild-type or BRCA1^−/− RPE1-hTERT p53^−/− Cas9 cells transduced with virus expressing the indicated sgRNAs. Data presented as mean ± standard deviation (n=3 biologically independent experiments). The results of a one-way ANOVA test that compared sgRNA targeting each sgRNA condition in BRCA1^−/− to their WT counterpart (* P ≤ 0.05, *** P ≤ 0.001) are shown. (D) Quantitation of sister chromatid exchanges (SCEs) on metaphase spreads of RPE1-hTERT p53^−/− Cas9 wild-type or two independent APEXZ^−/− cell lines. Black lines indicate the means (n=2 biologically independent experiments). The results of a t-test comparing APEX2 clones to WT is shown (FDR *** ≤ 0.001). (E) Quantitation of chromatid breaks (left) and radial chromosomes (right) in metaphase spreads from RPE1-hTERT p53^−/− Cas9 cells with indicated genotypes upon transduction with sgRNA targeting APEX2 (10 metaphases scored from at least 2 biologically independent experiments). The results of a pair-wise Wilcox (Mann Whitney) test comparing sgAPEX2 to sgAAVS1 conditions are shown (* FDR < 0.05; ** FDR < 0.01). (F) Representative image of a metaphase from a BRCA1^−/− cell sgAPEX2 showing multiple broken or aberrant chromosomes (arrowheads). See Table S6 for details on statistical tests.

The screens also showed that genes coding for APE2, APE1 and TDP1 buffer the loss of each other, consistent with their overlapping biochemical activities (Figure 2A; see also below). Furthermore, while BRCA1 scored as synthetic lethal in all three screens, APEX2 knockout cells were vulnerable to mutations in many more genes coding for factors involved in HR, interstrand crosslink (ICL) repair and the regulation of fork progression or stability, than APE1 or TDP1-deficient cells (Figure 2A). For example, the gene encoding PALB2 scored as essential solely in APEX2^−/− cells. In support of a key role for HR in supporting viability of APE2-deficient cells, growth competition assays (Figure 2C) showed that an sgRNA targeting APEX2 caused a much more profound loss of fitness in BRCAT^−/− cells than those targeting TDP1 or APEX1, consistent with a model whereby the loss of APE2 leads to accumulation of spontaneous DNA lesions that necessitates HR or the recombinational repair of distressed replication forks. In support of this possibility, we find that loss of APE2 causes elevated spontaneous sister chromatid exchanges (SCEs; Figure 2D) and that the removal of APE2 in BRCAT^−/− cells results in marked increases in chromatid breaks and radial chromosomes, the latter likely reflecting illegitimate end-joining reactions at broken chromosomes, a phenotype observed also when HR-deficient cells are treated with PARP inhibitors (Bunting et al., 2010) (Figure 2EF).

APE2 opposes DNA damage caused by genomic ribonucleotides

In view of these results, we searched for potential sources of endogenous DNA lesions that require APE2 action to safeguard the genome. The synthetic lethality between the loss of APEX2 and that of the three genes coding for the RNase H2 complex (RNASEH2A/B/C) (Figure 2A) was particularly striking largely because these interactions were specific to APEX2^−/−. Indeed, the lack of TDP1-RNASEH2A synthetic lethality was initially surprising given that in the absence of RNase H2, genomic ribonucleotides accumulate and are processed by TOP1, which could conceivably necessitate TDP1 action (Sparks and Burgers, 2015). We confirmed the synthetic lethal interaction in competitive growth assays where an effective sgRNA targeting RNASEH2A led to a striking loss of fitness in APEX2^−/− cells but not in the APEX1^−/− or TDP1^−/− cell lines (Figure 3A).

Figure 3. APE2 reverses 3′ blocked DNA lesions.

(A) Competitive growth assays in RPE1-hTERT p53^−/− Cas9 cells of the indicated genotypes transduced with virus expressing an sgRNA targeting RNASEH2A. Results of a two-way ANOVA with Dunnet post-test comparing APEX2^−/− to WT is shown (P*** < 0.001). (B) As in A, but using an sgRNA targeting APEX2. Statistical test compares each knockout cell lines to WT (* P ≤ 0.05; ** P≤ 0.01; *** P< 0.001). (C) Competitive growth assays in cells transduced with an sgRNA targeting BRCA2 in RPE1 hTERT p53^−/− Cas9 wild-type or RNASEH2A^−/− cells overexpressing either RNASEH2A, RNASEH2A-P40D/Y210A (ribonucleotide excision repair deficient; RED), or RNASEH2A-D34A/D169A (nuclease-dead; ND). (D) Competitive growth assays in wild-type or BRCA1^−/− RPE1-hTERT p53^−/− Cas9 cells expressing APE2 or the E48Q/D197N mutant transduced with the indicated sgRNAs. (E) Clonogenic survival of RPE1 hTERT p53^−/− Cas9 wild-type or APEX2^−/− cells transfected with the indicated siRNA and exposed to camptothecin (CPT). Immunoblot of the depletion of TDP1 by siRNA is also shown. NT, non-targeting control siRNA. Results of a Fit of exponential decay (quadratic survival) with extra-sum-of-squares F test comparing APEX2^−/− siTDP1 condition to WT siNT, WT siTDP1 and APEX2^−/− siNT is shown (*** P ≤ 0.001) (F) Clonogenic survival of RPE1-hTERT p53^−/− Cas9 cells of the indicated genotypes exposed to KBrO₃. Results of a Fit of exponential decay (quadratic survival) with extra-sum-of-squares F test comparing APEX1^−/− to WT, APEX2^−/− or TDPT^−/− are shown (*** P ≤ 0.001). Data in panels A-F are represented as the mean ± standard deviation (n =3 biologically independent experiments). See Figure S3 for controls and additional data. See Table S6 for details on statistical tests.

To further explore the genetic relationship between the genes coding these enzymes, we carried out competitive growth assays with an effective sgRNA targeting APEX2 in APEX1^−/−, TDPV^−/−, RNASEH2A^−/− or RNASEH2B^−/− cells (Figure 3B). We observed rapid and marked cell death in RNASEH2-deficient cells upon loss of APE2 whereas APEX2 mutations caused a milder decrease in fitness in the APEX1^−/− and TDP1^−/− backgrounds (Figure 3B). Similarly, targeting RNASEH2A, RNASEH2B, APEX1 and TDP1 with sgRNAs in APEX2-deficient cells led to identical conclusions (Figure S3AB). These results suggest that APE2 has a unique activity in counteracting DNA damage caused by embedded ribonucleotides.

The loss of RNase H2 is also lethal in the absence of BRCA1/2 (Zimmermann et al., 2018) but whether this is due to the loss of ribonucleotide excision repair or the persistence of RNA-DNA hybrids is unclear, especially in light of a role for BRCA2 in RNA-DNA hybrid processing (Bhatia et al., 2014; D’Alessandro et al., 2018). We therefore assessed whether cells solely expressing catalytically inactive RNASEH2A-D34A/D169A (ND, for nuclease-dead) or the ribonucleotide-excision deficient (RED) RNASEH2A-P40D/Y210A variants were sensitive to the loss of BRCA2 (Cerritelli and Crouch, 2019; Zimmermann et al., 2018). We found that the ribonucleotide excision activity of RNase H2 is critical to promote the viability of BRCA2-depleted cells (Figures 3C and S3CD). These results suggest that DNA damage arising from the accumulation of genome-embedded ribonucleotides is toxic in HR-deficient cells.

Given the role of APE2 in mitigating DNA damage caused by genomic ribonucleotides, we next considered the possibility that APE2 may be limiting in RNase H2-deficient cells. We overexpressed APE2 or the catalytically inactive APE2-E48Q/D197N protein in parental or BRCA1^−/− cells and subjected the resulting cell lines to a competitive growth assays with sgRNAs targeting RNASHE2A/B. APE2 overexpression completely suppressed the lethality induced by RNase H2 loss in BRCA1-deficient cells in a manner that required its catalytic activity (Figures 3D and S3E). These results indicate that APE2 is a critical determinant in suppressing the ribonucleotide-ssociated DNA damage causing lethality in BRCA1-deficient cells. These data also hint that 3ʼ blocks deriving from ribonucleotide processing represent the main source of DNA lesions causing the APEX2-BRCA1/2 synthetic lethality.

APE2 reverses 3ʼ blocked DNA lesions

The loss of fitness in TDP1-APEX2 double-mutants suggests that suppressing DNA damage that arises from abortive TOP1 reactions may be another activity of APE2. In support of this idea, loss of APE2 decreased clonogenic potential following exposure to camptothecin (CPT), a TOP1-trapping poison, and this sensitivity was enhanced upon partial TDP1 depletion by short-interfering (si) RNAs (Figure 3E). Together, these results suggest that APE2 and TDP1 may carry out partly redundant activities in the repair of TOP1-induced 3′ blocking lesions.

Genomic ribonucleotides and TOP1 trapping ultimately causes 3′ blocking lesions. The abasic sites accumulating in APE1-deficient cells can also be processed by lyases to produce 3′-blocking lesions. Therefore, the APEX2-APEX1 genetic interaction might reflect a role of APE2 in the repair of 3′ blocks instead of its proposed role as a backup pathway for base excision repair (BER). APEX2^−/− cells are much less sensitive to KBrO3, an oxidizing agent that causes lesions repaired by BER, than APEX1^−/− cells (Figure 3F). We conclude that APE2 has a major role in reversing 3′ end blocks rather than acting principally as an AP endonuclease.

Biochemical analysis of APE2

We compared the activity of recombinant APE2, in this case Xenopus laevis APE2, on model oligonucleotide substrates bearing 3′-terminal blocks mimicking a nominal TOP1cc (3′-phosphotyrosine, 3′-pTyr), or the products of TOP1 cleavage at genome-incorporated ribonucleotides (2′,3′-cyclic-phosphate, 2′3′-P) (Figure 4A). APE2 had robust 3′-nucleolytic activity on unmodified 3′-hydroxyl (3′-OH) (Figure 4A, lanes 4–6), as well as on blocked 3′-pTyr (Figure 4A, lanes 14–16) and 2′3′-P (Figure 4A, lanes 23–25) substrates. APE2 activity required an intact catalytic site, as mutation of a metal binding ligand (X. laevis APE2 E34Q) or active site nucleophile (X. laevis APE2 D183N) completely abolished activity on a 3′-FAM-labeled substrate (Figure S4A). Despite not being essential for 3′ unblocking, PCNA did stimulate APE2 exonuclease activity on ends blocked with 3′-pTyr (Figure S4B). In contrast, APE1 failed to process both 3′-Tyr and 2′3′-P substrates (Figure 4A, lanes 18–20 and 27–29), but removed 1–2 nucleotides from the 3′ end of an undamaged (3′-OH) substrate (Figure 4A, lanes 8–10), and had robust AP-endonuclease activity on a substrate with a model internal abasic site, as anticipated (Figure 4A, lanes 38–40) (Whitaker et al., 2018). While APE2 displays some AP-endonuclease activity (Figure 4A lanes 34–36), our findings are consistent with APE2 also being endowed with broad DNA 3′-end-processing functions that extend beyond the previously assumed role as a backup AP-endonuclease.

Figure 4. APE2 nucleolytic activity facilitates resolution of blocked 3′ ends.

(A) Denaturing gel electrophoresis of dsDNA oligonucleotide substrates carrying a 3′ hydroxyl (lanes 1–10), 3′ phosphotyrosine (lanes 11–20), terminal 2′,3′-cyclic phosphate (lanes 21–30) or internal model abasic site (THF; tetrahydrofuran, lanes 31–40) incubated with recombinant X. laevis APE2, human APE1, or with MBP as controls. Asterisk, position of FAM label. Numbers, oligonucleotide lengths. (B) Scheme of TOP1 suicide substrate containing either a deoxythymidine (dT) or ribouridine (rU) at the TOP1 incision hot spot, two nucleotides from the 3′ end of a 5′ Cy5-labeled scissile strand. (c) Scheme of TOP1cc processing assay performed in D and F. (D) Denaturing gel electrophoresis of products obtained from dT-containing TOP1 suicide substrate incubated with the indicated recombinant proteins. Low and high molecular weight products are shown cropped from the same gel. Blue numbers in circles annotate the reaction products summarized in panel (E). (F) Same as in (D), but rU-containing TOP1 suicide substrate was used. Gray numbers in circles annotate the reaction products corresponding to the schematic in panel (G). Asterisk annotates a background band. See Figures S4 and S5 for additional data and controls.

Since our genetic analyses suggests that APE2 can repair TOP 1-generated DNA damage, we compared the activity of human APE2 or TDP1 on TOP1cc and 2′,3′-P substrates generated through in vitro reconstitution of TOP1 DNA damage (Figures 4B–G). TOP1cc were generated via the reaction of recombinant TOP1 with a suicide substrate (Interthal et al., 2001), that contains an embedded ribonucleotide or a deoxyribonucleotide counterpart (Figure 4B). The assays are schematically depicted in Figure 4C. This experimental design promotes TOP1 incision proximal to the 3′ end of the substrate and generates TOP1cc with TOP1 covalently linked to 9–12 nucleotide fragments (Figures 4D, lane 4; schematized in Figure 4E, product 1).

Neither APE2 nor TDP1 efficiently reversed the intact TOP1cc (Figures S4CDFGI and 4DF, lanes 5–10, product 1). However, APE2 and TDP1 processed minimal oligo-peptides (TOP1cc-tryp) that result from the tryptic digestion of TOP1ccs (Interthal and Champoux, 2011) (Figures 4D, lanes 15–20, product 2). These results are consistent with the proteolytic degradation of TOP1cc facilitating 3′ DNA end healing during TOP1cc resolution (Interthal and Champoux, 2011; Lopez-Mosqueda et al., 2016; Stingele et al., 2016). TDP1 reaction products migrate coincident with 3′-phosphorylated species (Figures S5A, 4D and 4F, lanes 19–20) (Flett et al., 2018) whereas APE2 degradation of TOP1cc-tryp yielded products coincident with 3′-hydroxyl terminated products (Figures S5A, 4D and 4F, lanes 17–18). Since TDP1 and APE2 both process degraded TOP1cc but leading to different molecular outcomes (Figure 4E), we conclude that these enzymes can process TOP1cc in distinct cellular contexts.

With the substrate bearing an embedded ribonucleotide, reaction with TOP1 yields both TOP1cc (Figures S5B and 4F, products 1 and 4), as well as a TOP1-dependent endonucleolytic product with a cyclized 2′3′-P end (product 2 in Figures S5B, lane 3–4 and 4F, lane 4), as predicted from previous studies (Huang et al., 2015; Kim et al., 2011; Sekiguchi and Shuman, 1997; Sparks and Burgers, 2015). Processing of TOP1cc generated by the ribonucleotide substrate also requires tryptic proteolysis prior to action by TDP1 and APE2 (Figures S4CEFHI and 4F, lanes 5–10 and lanes 15–20, product 1 and 5). We observed striking differences in the ability of the enzymes to resolve the 2′3′-P ribonucleotide-blocked species. TDP1 fails to degrade the 12-nucleotide cyclic phosphate substrate generated by TOP1 incision at a ribonucleotide (Figure 4F, lanes 9–10 and 19–20). However, APE2 degrades the 2′3′-P ends to 9–10 nucleotide products, similar to the all deoxynucleotide substrate (Figures S5A and 4F, lanes 5–8 and 15–18, products 3 and 6). APE2 exonucleolytic processing is stimulated by PCNA (Figure 4F, lanes 5–8 and 15–18), but PCNA is not strictly required for unblocking 2′3′-P termini (Figure 4A, lanes 22–25).

Altogether, these biochemical analyses reveal that while there is a division of labor between APE2 and TDP1 for the processing of TOP1cc, and APE2 and APE1 for abasic site cleavage, APE2 is uniquely endowed with the capacity to process the cyclic phosphate substrates (2′3′-P) generated by TOP1 incision at ribonucleotides (Figures 4E and 4G). Furthermore, we could not detect endonucleolytic activity of APE2 on an embedded ribonucleotide substrate (Figure S5C). Thus, our data suggest that the APEX2-RNASEH2 synthetic lethality is rooted in a deficiency in the removal of DNA 3ʼ blocks arising from TOP1 metabolism of ribonucleotides (Williams et al., 2013), rather than the direct processing of genome-embedded ribonucleotides by APE2.

APE2 catalytic domain architecture

To better define APE2 structure/activity relationships we mapped crystallizable domains of vertebrate APE2 homologs and determined the structure of the N-terminal EEP (endonuclease/exonuclease/phosphatase) domain of X laevis APE2 (APE2^cat, amino acids 1–355) to 1.65 Å resolution. Overall, the APE2^cat structure is typified by an anti-parallel β-sheet (β1–13) sandwiched between twelve helices (α1–12; Figure 5A–D). The APE2 active center bears a single octahedrally coordinated magnesium ion metal which is liganded by N9, E34 and D299 (Figure 5B and S6). The presumed catalytic base, D183, binds a water nucleophile that marks the base of the DNA binding cleft (Figure 5B).

Figure 5. APE2 X-ray structure.

(A) Domain architecture of APE2. (B) Magnesium ion coordination site with electron density maps overlaid (2F_o-F_c, 1.5 o at 1.54Å). (C) Crystal structure of the N-terminal EEP domain of X. laevis APE2. A single active site-bound magnesium ion is indicated in green. (D) An orthogonal view of the catalytic domain structure. (E) Structural overlay of APE2 (blue) with the APE1-THF-DNA complex (grey). AP site recognition residues (in grey surfaces) are also divergent in APE2. (F) Closeup view of the APE1-APE2 comparison. Critical APE1 AP site recognition residues R177, M270 and W280 interrogate the DNA base stack and the DNA lesion binding pocket, which is capped by a closed L5 loop. In APE2, these residues are not conserved, and are replaced by smaller side chain P148, T270 and A260. By comparison, L5 in APE1 is closed and secures the AP recognition pocket. (G) A surface representation of the APE1-DNA complex compared to APE2 in (H). See Figure S6 for additional data and Table S4 for crystallographic data collection and refinement statistics.

When compared to the EEP-family proteins APE1 (Mol et al., 2000) and TDP2 (Schellenberg et al., 2012), structural and sequence alignments reveal marked divergence in six surface-exposed loops flanking the APE2 active site trench (loop 1-loop 6, Figures S6A–S6C). Relative to APE1, key deviations in the APE2 substrate binding pocket are found in loop 5 (L5), and the AP (apurinic site) recognition pocket assembled by APE1 residues R177 and W280 (Figure 5E–H). The APE2 DNA binding cleft is comparatively highly accessible, particularly in the region expected to engage the substrate 3′ to the cleavage site (compare Figures 5G and 5H). Here, the L5 loop adopts an exposed platform for substrate binding (Figures 5F and 5H), consistent with an active site similarly accessible on the peptide egress side for 5′ phosphotyrosine-linked TOP2cc processing described for TDP2 (Schellenberg et al., 2012). Altogether, the structure of the APE2 catalytic domain is consistent with the broad substrate specificity of APE2 for accommodating 3′ blocking lesions.

SLX4-XPF-ERCC1 may process 3′ ends in APE2-deficient cells

Given that recombination-associated DNA synthesis requires ends with a 3′-hydroxyl group to proceed, the ability of BRCA1/2 to promote viability of APEX2-mutated cells suggests that additional activities remove 3′ blocks as part of the recombination process. A candidate is the SLX4-XPF-ERCC1 complex since the genes encoding these proteins, SLX4, ERCC4 (XPF) and ERCC1, were synthetic-lethal with APEX2 (Figure 2A), a finding that we validated in competitive growth assays (Figures 6AB, S7A and S7E–G). XPF and ERCC1 form a nuclease that exists either as a dimer or as part of a larger complex with the SLX4 scaffold protein (Dehe and Gaillard, 2017). SLX4 also interacts with SLX1 and the MUS81-EME1 nucleases to form the SMX trinuclease complex (Dehe and Gaillard, 2017; Wyatt et al., 2017) but loss of SLX1 or MUS81 is viable in APEX2^−/− cells, suggesting that APE2-deficiency necessitates specifically the SLX4-XPF-ERCC1 complex (Figures 2A and S7A). To test this possibility further, we introduced SLX4 or an SLX4 variant deficient in XPF interaction (L530A/W531A) in SLX4^−/− cells (Figure 6C and S7B–D) and challenged those cells with an sgRNA targeting APEX2. We found that the SLX4/XPF-ERCC1 interaction is essential to maintain the viability of APE2-deficient cells, suggesting that SLX4-XPF-ERCC1 may act as a backup to APE2. Loss of SLX4 in APEXZ^−/− cells results in accumulation of chromatid breaks and radials, reminiscent of the situation seen in cells co-depleted of APE2 and BRCA1 (Figure 6D).

Figure 6. SLX4-XPF-ERCC1 may process 3′ ends in APE2-deficient cells.

(A) Competitive growth assays in RPEl-hTERT p53^−/− Cas9 cells of the indicated genotypes transduced with virus expressing sgRNAs targeting SLX4 or AA VS1. Data presented as the mean ± standard deviation (n=3 biologically independent experiments). Results of a two-way ANOVA with Dunnet post-test comparing APEX2^−/− to WT are shown (*** P ≤ 0.001). (B) As in A but cells were transduced with virus expressing an sgRNA targeting APEX2. Results of a two-way ANOVA with Dunnet post-test comparing knock-out cell lines to WT are shown (*** P≤ 0.001). (C) Clonogenic survival of RPE1 hTERT p53^−/− Cas9 wild-type or SLX4^−/− cells with transgenes expressing either wild-type SLX4 or its L530A/W531A mutant. These cells were transduced with an sgRNA targeting either AAVS1 or APEX2. Data are presented as the mean ± standard deviation (n=4 biologically independent experiments). Results of a t-test comparing GFP-SLX4 to GFP-only and GFP-SLX4-L530A/W531A are shown (*** FDR ≤ 0.001). (D) Quantitation of chromatid breaks (left) and radial chromosomes (right) in metaphase spreads from RPE1-hTERT p53^−/− Cas9 cells with indicated genotypes upon transduction with an sgRNA targeting SLX4 (10 metaphases scored from at least 2 biologically independent experiments). The results of a Pair-wise Wilcox (Mann Whitney) comparing sgSLX4 to sgAAVSl conditions are shown (* FDR ≤ 0.05). See Figure S7 for controls and additional data. See Table S6 for details on statistical tests.

Discussion

The presence of a free 3′ hydroxyl end on DNA is a prerequisite to the DNA synthesis reactions associated with DNA replication, repair and recombination. Therefore, 3′ blocks are a class of highly deleterious DNA lesions. Our work suggests that 3′ block avoidance and removal can be conceptually organized in a two-tier system (Figure 7) where cells first attempt to prevent the accumulation of 3′ blocks via the action of lesion-specific enzymes such as TDP1, RNase H2 and, under some circumstances, APE1. In cases where these enzymes fail to act to prevent formation of 3′ blocks, we propose that APE2 acts to restore an extendible 3′ end (Figure 7). This conclusion is based on both the genetic interaction profile of APEX2-deficient cells, the ability of APE2 to remove a variety of 3′ blocks in vitro and the structure of APE2 that shows how it can accommodate multiple types of 3′ blocking DNA lesions.

Figure 7. DNA 3′ block metabolism.

DNA 3′ blocks are managed by a two-tier system. In Tier 1, RNase H2, TDP1 and APE1 promote 3′ block avoidance. Tier 2 consists of APE2. BRCA1/2-dependent recombination preserves cell viability and genomic integrity when cells undergo replication in the presence of 3ʼblocks.

The function of APE2 in excising 3′ blocks in human cells is largely consistent with the recently described role of its yeast homolog, Apn2 (Li et al., 2019). We thus conclude that the removal of 3′ blocks is the primary and evolutionarily conserved function of APE2. Our work suggests that it is this activity of APE2, and not a potential role in BER that is critical for the viability of HR-deficient cells. This is based on two sets of observation. First, increasing the formation of 3′ blocks via the depletion of RNase H2, results in a synthetic lethality with HR deficiency that can be suppressed by increasing the dosage of APEX2. Secondly, our data indicates that the PCNA-interacting PIP-box on APE2 is not critical for the viability of BRCA-deficient cells. Similarly, mutation of the PIP box in yeast Apn2 does not cause lethality in conditions where 3′ blocks accumulate (Li et al., 2019). This argues against a model where APE2 promotes viability of HR-deficient cells by engaging in a PCNA-dependent BER subpathway.

Given how problematic 3′ blocks are, one might expect that APE2 loss should be deleterious to cell and organismal viability. Consistent with this idea, APEX2 is an essential gene in the p53-proficient RPE1-hTERT cell line (Figure 2). Furthermore, the probability that protein-truncating mutations in the APEX2 gene are not tolerated in humans is very high. Indeed, the pLI value associated with APEX2 in GnomAD database is 0.97 (Karczewski et al., 2019). pLI values > 0.9 indicate that the gene is likely to be highly intolerant to protein-truncating mutations. However, the phenotypes of the commonly used mouse APEX2^tm1Yun allele, which is likely a null allele (Ide et al., 2004), are relatively mild. This discrepancy in the impact of APE2 loss in human vs mice is reminiscent of mutations in other important DNA repair factor-coding genes, such as those coding for Fanconi anemia pathway and TC-NER factors, which cause only mild phenotypes in mice.

Cells without APE2 require HR but also factors involved in DNA replication (CLSPN, POLE3, HROB-MCM8/9 and genes encoding 9-1-1 complex components). APEXZ^−/− cells are also sensitive to ATR inhibitors (Hustedt et al., 2019a). Together, these genetic interactions suggest that the DNA lesions that accumulate in APE2-deficient cells require HR in the context of stalled or collapsed DNA replication forks. However, this also raises a conundrum as strand invasion reactions with 3′-blocked overhangs cannot be extended by DNA synthesis. Our work suggests a potential solution to this problem through the action of the SLX4-XPF-ERCC1 structure-specific nuclease. We speculate that this nuclease can recognize unextendible invasion reactions and cleave near ds/ssDNA junctions to remove the 3′ block. In other words, we propose that there might be a recombination-associated 3′ end unblocking pathway. However, alternative explanations exist for the APEX2-SLX4 synthetic lethality, e.g. SLX4-XPF-ERCC1 may act upstream of HR, and therefore this model will need to be thoroughly tested.

Finally, the highly penetrant synthetic lethal interactions between APEX2 and BRCA1/2, RNASEH2A and BRCA1/2 as well as APEX2 and RNASEH2A/B leads us to propose that 3′ blocks are highly deleterious to HR-deficient cells. This idea is further supported by the reduced fitness of BRCA-deficient cells upon APE1 or TDP1 depletion. Therefore, the development of APE2 inhibitors may provide a new tool for targeting HR-deficient cancers. The larger APE2 catalytic pocket makes it likely that potent and selective inhibitors can be identified and then elaborated to have desirable pharmacological properties. In addition to targeting HR-deficient tumors, the profound synthetic lethality between genes encoding RNase H2 and APEX2 suggests that tumors that frequently lose RNASEH2B, which is often co-deleted with the RB1 tumor suppressor (Zimmermann et al., 2018), could also be another target population for prospective APE2 inhibitors.

STAR methods

Resource Availability

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by Daniel Durocher (durocher@lunenfeld.ca).

Materials Availability

Materials included in this manuscript will be shared upon request

Data and Code Availability

The datasets generated and analyzed during this study are available at Mendeley Data (https://data.mendelev.com/datasets/cx297hzkik/draft9a=9817abdf-24c7-4a4c-8570-ecf8c2f29465). The coordinates for the X. laevis APE2 catalytic domain X-ray structure are available in the Protein Data Bank (PBD: 6WCD)

Experimental Model and Subject Details

Cell lines

Human cell lines used in this manuscript are summarized in the Key Resource Table.

KEY RESOURCE TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Monoclonal mouse anti-APE1	ThermoFisher	MA1–440
Monoclonal mouse anti-PCNA	Abcam	PC10 ab29
Polyclonal rabbit anti-ERCC4	Bethyl	A301–315A
Monoclonal mouse anti-ERCC1	Santa Cruz	D-10 sc-17809
Polyclonal rabbit anti-APE2	In house (Durocher D)	TO266
Monoclonal mouse anti-α-Tubulin	Millipore	DM1A cp06
Polyclonal rabbit anti-TDP1	Abcam	ab4166
Polyclonal rabbit anti-GFP	Abcam	ab290–50
Polyclonal sheep anti-GFP	In house (Rouse J)	S268B
Polyclonal sheep anti-SLX1	In house (Rouse J)	S701C
Monoclonal mouse anti-MUS81	Immuquest	MTA30 2G10/3
Monoclonal mouse anti-ERCC4	Thermo scientific	AB-1 clone 219
Polyclonal Sheep anti-EME1	In house (Rouse J)	S180D
Polyclonal Rabbit anti-PLK1	Cell signalling	4535
Monoclonal Mouse anti-GAPDH	Millipore Sigma	G9545
Monoclonal Mouse anti-α-Actinin	Millipore	AT6/172
Goat anti-mouse Irdye 800CW	LiCOR Bioscience	926–32210
Goat anti-rabbit Irdye 680RD	LiCOR Bioscience	926–68071
Bacterial and Virus Strains
BL21-AI E. coli	Invitrogen	C607003
BL21 Star™ (DE3) E.coli	Invitrogen	C601003
Chemicals, Peptides, and Recombinant Proteins
Hoechst 33342	Thermo Fisher	H1399
5-Bromo-2′-deoxyuridine	Sigma	B5002
KaryoMAX colcemid	Gibco	15212012
Puromycin	InvivoGen	Cat# ant-pr
Blasticidin	InvivoGen	Cat# ant-bl
Penicillin and streptomycin (Pen Strep)	GIBCO	Cat# 15140122
Fetal Bovine Serum (FBS)	Wisent Bioproduct	Cat# 080–150
Phosphate Buffer Saline (PBS)	GIBCO	Cat# 10010023
Dulbecco’s Modified Eagle’s Medium (DMEM)	GIBCO	Cat# C11965500BT
RPMI 1640 Media	GIBCO	Cat# 11875093
DMEM High Glucose	VWR	CA45000–304
MEGM bullet kit	VWR	CA101968–208
HyClone McCoy’s 5A Media	Fisher Scientific	SH30200FS
MCDB 105 Medium	Sigma-Aldrich	117–500
Medium 199	Sigma-Aldrich	M4530
DMEM/F12	VWR	CA45000–344
HAM’S F-12 W/GLN	VWR	CA45000–358
Sodium pyruvate	VWR	CA45000–710
Sodium bicarbonate 7.5% sln	VWR	CA45000–704
Hydrocortisone	Cedarlane Labs	07926
Insulin solution human	Sigma-Aldrich	I9278–5ML
Polybrene	Sigma-Aldrich	Cat# TR-1003
Potassium Bromate	Sigma Millipore	60085
Camptothecin (CPT)-1	Sigma Millipore	C9911
Recombinant human TDP1	NovusBio	NBP2–59661
Topo1 assay buffer	TopoGEN	TG4095
Trypsin	Sigma-Aldrich	T1426
FreeStyle™ 293 Expression medium	Gibco	12338018
Anti-GFP nanobody resin	(Schellenberg et al., 2018)	N/A
Recombinant E. coli RNase HII	New England Biolabs	M0288S
Recombinant human APE1	New England Biolabs	M0282S
Recombinant human APE2 full-length	This paper	N/A
Recombinant X. laevis APE2 full-length	This paper	N/A
Recombinant X. laevis APE2Cat (residues 1–355)	This paper	N/A
Recombinant human PCNA	This paper	N/A
Recombinant X. laevis PCNA	This paper	N/A
Recombinant human TOP1 (Top70)	This paper	N/A
Critical Commercial Assays
QIAamp Blood Maxi Kit	Qiagen	Cat# 51194
Q5 High-Fidelity 2X Master Mix	New England Biolabs	Cat# M5044L
NextSeq 500/550 High Output Kit v2.5	Illumina	Cat# 20024906
Deposited Data
Coordinate of X. laevis APE2 Catalytic Domain	This paper	PDB: 6WCD
sgRNA readcounts for the CRISPR screens	This paper	https://data.mendeley.com/datasets/cx297hzkjk/draft?a=9817abdf-24c7-4a4c-8570-ecf8c2f29465
Raw values for all graphs presented in this manuscript.	This paper	https://data.mendeley.com/datasets/cx297hzkjk/draft?a=9817abdf-24c7-4a4c-8570-ecf8c2f29465
Uncropped blot images	This paper	https://data.mendeley.com/datasets/cx297hzkjk/draft?a=9817abdf-24c7-4a4c-8570-ecf8c2f29465
Experimental Models: Cell Lines
Human: RPE1-hTERT Cas9 TP53-KO	(Zimmermann et al., 2018)	N/A
Human: RPE1-hTERT Cas9 TP53/APEX2 #1-KO	(Hustedt et al., 2019a)	Table S3
Human: RPE1-hTERT Cas9 TP53/APEX2 #2-KO	(Hustedt et al., 2019a)	Table S3
Human: RPE1-hTERT Cas9 TP53/APEX1-KO	This paper	Table S3
Human: RPE1-hTERT Cas9 TP53/TDP1-KO	This paper	Table S3
Human: RPE1-hTERT Cas9 TP53/SLX4 #1-KO	This paper	Table S3
Human: RPE1-hTERT Cas9 TP53/SLX4 #2-KO	This paper	Table S3
Human: RPE1-hTERT Cas9 TP53/ERCC4 #1-KO	This paper	Table S3
Human: RPE1-hTERT Cas9 TP53/ERCC4 #2-KO	This paper	Table S3
Human: RPE1-hTERT Cas9 TP53/ERCC1 #1-KO	This paper	Table S3
Human: RPE1-hTERT Cas9 TP53/RNASEH2A-KO	(Zimmermann et al., 2018)	N/A
Human: RPE1-hTERT Cas9 TP53/RNASEH2B-KO	(Zimmermann et al., 2018)	N/A
Human: RPE1-hTERT Cas9 TP53/BRCA1-KO	(Noordermeer et al., 2018)	N/A
Human: DLD1 (ATCC® CCL-221)	Horizon Discovery	N/A
Human: DLD1 BRCA2-KO	Horizon Discovery	HD 105–007
Human: SUM149PT	Asterand	N/A
Human: JHOS-2	Riken BRC	RCB1521
Human: COV362	Sigma Aldrich	07071910–1VL
Human: UWBA1.289	ATCC	CRL-2945
Human: SK-OV-3	ATCC	HTB-77
Human: OV-90	ATCC	CRL-11732
Human: COV644	Sigma Aldrich	07071908
Human: RMUG-S	JCRB	IFO50320
Human: Lenti-X 293T	Takara Bio USA, Inc.	#632180
Human: FreeStyle™ 293-F	Gibco	R79007
Oligonucleotides
TKOv3 CRISPR libraries	(Hart et al., 2017)	N/A
PCR#1 library primer forward V3_2_F: CTGCGTGCGCCAATTCTG	This paper	N/A
PCR#2 library primer reverse V3_1_R2: AGAACCGGTCCTGTGTTCTG	This paper	N/A
sgRNA sequences and TIDE PCR primers	This paper	Table S3
Substrate Oligonucleotides for in vitro reactions	This paper	Table S5
SMARTpool ON-TARGETplus TDP1 siRNA	Dharmacon	L-016112-00-0005
Recombinant DNA
psPAX2	Didier Trono	Addgene Cat#12260
pMD2.G	Didier Trono	Addgene Cat#12259
pMDLg/pRRE	(Dull et al., 1998)	Addgene Cat#12251
LentiCRISPRv2	(Sanjana et al., 2014)	Addgene Cat#52961
pDONR221	Thermo Fisher	12536017
pCDNA3.1–2xHA-C	Gift of Dr. R. McInnes	N/A
pGLUE	Randall Moon	Addgene Cat#15100
pBABE-puro	(Morgenstern and Land, 1990)	Addgene Cat#1764
LentiGuide-NLS-EGFP-Puro	(Noordermeer et al., 2018)	N/A
LentiGuide-NLS-mCherry-Puro	(Noordermeer et al., 2018)	N/A
LentiGuide-Puro	(Sanjana et al., 2014)	Addgene Cat#52963
pCW57.1	David Root	Addgene Cat#41393
pCW57.1-GFP	(Zimmermann et al., 2018)	N/A
pCW57.1-APE2	This paper	N/A
pCW57.1-APE2-E48Q/D197N	This paper	N/A
pCW57.1-APE2-Y395A/F396A	This paper	N/A
pCW57.1-APE2-ΔPIP	This paper	N/A
pCW57.1-APE2-ΔZf-GRF	This paper	N/A
pCDNA3.1-SBP-2xHA-C	This paper	N/A
pCDNA3.1-APE2-SBP-2xHA-C	This paper	N/A
pCDNA3.1-APE2-Y395A/F396A-SBP-2xHA-C	This paper	N/A
pCDNA3.1-APE2-ΔPIP-SBP-2xHA-C	This paper	N/A
pCW57.1-FLAG-RNASEH2A	(Zimmermann et al., 2018)	N/A
pCW57.1-FLAG-RNASEH2A-P40D/Y210A	(Zimmermann et al., 2018)	N/A
pCW57.1-FLAG-RNASEH2A-D34A/D169A	(Zimmermann et al., 2018)	N/A
pBabeD-puro-GFP-SLX4	This paper	DU33990
pBabeD-puro-GFP-SLX4-L530A/W531A	This paper	DU50145
2Cc-T XlAPE2Cat	(Wallace et al., 2017)	N/A
2Cc-T XlAPE2Cat E34Q/D183N	(Wallace et al., 2017)	N/A
pET28a-XlPCNA	(Lin et al., 2018)	N/A
pT7-HsPCNA	(Hishiki et al., 2008)	N/A
pcDNA6.2/N-YFP-DEST HsTopo70	This paper	N/A
Software and Algorithms
Acappella	PerkinElmer	http://bowtie-bio.sourceforge.net/bowtie2/index.shtml
g:Profiler	(Raudvere et al., 2019)	https://biit.cs.ut.ee/gprofiler/gost
Cytoscape version 3.7.1	Cytoscape	https://cytoscape.org
R version 3.5.2	R software	https://www.r-project.org
MAGeCK	(Li et al., 2014)	https://sourceforge.net/p/mageck/wiki/Home
BAGEL	(Hart et al., 2017)	https://github.com/hart-lab/bagel/blob/master/BAGEL.py
Adobe Illustrator CS6 (version 16.0.0)	Adobe Inc.	https://www.adobe.com/products/illustrator.html
Image Studio 3.1 and 5.0	LiCOR Bioscience	https://www.licor.com/bio/image-studio-lite
Prism 8	GraphPad	https://www.graphpad.com/scientific-software/prism
ggplot2 (version 3.2.1)		https://ggplot2.tidyverse.org
PHENIX	(Adams et al., 2010)	https://www.phenix-online.org
Coot	(Emsley and Cowtan, 2004)	https://www2.mrc-lmb.cam.ac.uk/personal/pemsley/coot
HKL2000	HKL Research Inc.	https://hkl-xray.com
PyMOL	Schrödinger	https://pymol.org/2
Fiji	ImageJ	https://imagej.net
Image Lab	Bio-Rad	https://www.bio-rad.com/en-us/product/image-lab-software
Ismeans R package		https://www.jstatsoft.org/article/view/v069i01
Rstatix R package 0.4.0.		https://CRAN.R-project.org/package=rstatix

Methods Details

Gene-edited cell lines

Unless stated, cell lines were generating using RPEl-hTERT TP53^−/− Cas9 as parental cell lines. The APEX2^−/−, APEX1^−/−, TDP1^−/−, SLX4^−/−, ERCC1^−/−, and ERCC4^−/− knockout cell lines were generated by electroporation (Lonza Amaxa II Nucleofector) of LentiGuide plasmids expressing sgRNAs (Table S3). 24 h after transfection, cells were selected with 15 μg/ml puromycin for 24–48 h, followed by single clone isolation. Cell lines were tested by TIDE analysis and by immunoblot. The BRCAl^−/−, RNASEH2A^−/− and RNASEH2B^−/− knockout cell lines were described previously (Noordermeer et al., 2018)

Cell culture

RPE1-hTERT p53^−/− Cas9 (Zimmermann et al., 2018) and 293T cells were grown in DMEM supplemented with 10% FBS (Wisent #080150), GlutaMAX (life Technologies #35050–061) and 1% Pen/Strep (Wisent #450–201-EL). UWB1.28, SKOV3 and OV90 cells were purchased from ATCC. UWB1.28 cells were grown in RPMI: MEGM (1:1) supplemented with 3% FBS and 1% Pen/Strep. SKOV3 cells were grown in McCoys 5a supplemented with 10% FBS and 1% Pen/Strep. OV90 cells were grown in MCDB 105: Medium 99 (1:1) supplemented with 1% sodium bicarbonate (0.75 g/L), 15% FBS and 1% Pen/Step. JHOS2 were purchased from Riken BRC and grown in DMEM: F-12(1:1) supplemented with 10%FBS, 0.1mM NEAA. SUM149PT cells were purchased from Asterand BioScience and grown in a DMEM/F12 medium mixture supplemented with 5% FBS, 1% Pen/Strep, 1 μg/mL hydrocortisone and 5 μg/mL insulin. COV362 and COV644 were purchased from Sigma and grown in DMEM supplemented with 10% FBS and 1% Pen/Strep. RMUGS cells were purchased from JCRC cell bank and grown in Ham’s 12 supplemented with 10% FBS and 1% Pen/Strep. Wild-type and BRCA2^−/− DLD-1 cells were purchased from Horizon and maintained in RPMI medium supplemented with 10% FBS, 1% Pen/Strep and 2 mM sodium pyruvate. All cell lines were grown at 37 °C and 5% CO2.

Plasmids

For CRISPR-Cas9 genome editing, DNA corresponding to sgRNAs was cloned into LentiGuide-Puro, a modified form of LentiGuide-Puro (Addgene: 52963) in which Cas9 was replaced by NLS-tagged GFP or mCherry using Agel and BamHI (designated as LentiGuide-NLS-GFP or - mCherry). For APE2 overeexpression in cells, untagged APE2 coding sequence (CCDS14365.1) was cloned into pDONR221 (Thermo Fisher, 12536017), were mutagenesis was performed by PCR to generate an APE2 sgRNA resistant version carrying silent mutations (G180T, T183C, C186T, T189C, G192A and T195A), and the coding mutations E48Q/D197N, Y395A/F396A, deletion of aminoacids 390–397 (ΔPIP) and 442–518 (ΔZf-GRF). APE2 coding sequences were subcloned in the destination vector pCW57.1 (Addgene: 41393) used for transduction in cells. For APE2 streptavidin pulldown, APE2 was amplified by PCR from pDONR221-APE2 (excluding STOP codon) and flanking KpnI/Xhol restriction sites were added. PCR product was digested with KpnI/Xhol and cloned using those sites in a pCDNA3.1–2xHA-C (kind gift of Dr. R. Mclnnes). To generate an SBP tagged version of APE2, SBP was amplified by PCR from pGLUE (Addgene: 15100) adding flaking BamHI restriction sites. PCR was BamHI digested and cloned in frame using a BamHI site between APE2 and the 2xHA C-terminal tag in pCDNA3.1-APE2–2xHA. Y395A/F396A and ΔPIP mutants (described previously) were generated. A pCDNA3.1-SBP-HA empty version (control) was also generated by deletion of the APE2 coding sequence (with the exception of the first methionine codon nucleotide sequence). For SLX4 overexpression, cDNA coding for the SLX4 open reading frame (NM_032444.2) was originally generated by overlap PCR using IMAGE consortium EST clones 6527830 and 4340346 as template material. BamHI and NotI restriction enzyme sites were incorporated into the 5ʼ and 3ʼ oligonucleotides respectively. Full-length SLX4 as a BamHI-NotI fragment was cloned into pBabeD-puro-GFP-SLX4 (DU33990). Mutagenesis was performed to generate pBabeD-puro-GFP-SLX4-L530A/W531A (DU50145).

Lentiviral transduction

To produce lentivirus, 4 × 10⁶ 293T cells in a 10-cm dish were transfected with packaging plasmids (5 μg pVSVg, 3 μg pMDLg/pRRE and 2.5 μg pRSV-Rev, addgene #14888, 12251, 12253) along with 10 μg of transfer plasmid using PEI. Medium was refreshed 12–16 h later. Virus-containing supernatant was collected ~36–40 h post transfection, cleared through a 0.2-μm filter, supplemented with 4 μg/mL polybrene (Sigma) and used for infection of target cells.

Clonogenic survival assays

Cells were seeded in 10-cm dishes (1000–4,000 cells per 10 cm plate, depending on cell line and genotype). For drug sensitivity assays cells were seeded into media containing a range of CPT and KBrO₃ concentrations. Plates were incubated at atmospheric O₂. Experiments performed with BRCAr^−/− cells and their controls were incubated at 3% O₂. Medium was refreshed after 7 days in all cases. At the end of the experiment medium was removed, cells were rinsed with PBS and stained with 0.4% (w/v) crystal violet in 20% (v/v) methanol for 30 min. The stain was aspirated and plates were rinsed twice in ddH₂O and air-dried. Colonies were manually counted and data were plotted as surviving fractions relative to untreated cells or sgAAVSl/LacZ-transduced controls. For clonogenic survival assays in Figure 1A–C the same procedure was used but 300–1200 cells/well (depending on cell lines) were seeded a into 6-well plate. Colonies were counted using a GelCount instrument (Oxford Optronix, GelCount).

Two-color competitive growth assays

The two-color competitive growth assays were performed as previously described (Hustedt et al., 2019a). Cells were transduced with virus particles expressing NLS-mCherry-sgAAVSl (control) or an NLS-GFP-sgRNA targeting a specific gene of interest (see Table S3). 24 h after transduction virally transduced cells were selected using 20 μg/mL puromycin (Life Technologies #A1113802) for 48 h. Four days post infection, mCherry- and GFP-expressing cells were mixed 1:1 (3,000 cells each) and seeded in a 12-well Eppendorf plate. Cells were imaged for GFP and mCherry signals 24 h after initial plating (t=0) and at the indicated timepoints using a 4X objective IN Cell Analyzer system (InCell 6000 Analyzer, GE Healthcare Life Sciences). During the course of the experiment, cells were subcultured when near-confluency was reached. Segmentation and counting of GFP- and mCherry-positive cells were performed using an Acapella script (PerkinElmer).

Affinity purification

For APE2 streptavidin pulldown, 5×10⁶ 293T cells were plated in 10 cm dishes. 48 h later were transfected with 2 μg of pCDNA3. 1-APE2-HA-SBP, -APE2(Y396A/F397A)-HA-SBP, - APE2(Δ390–397)-HA-SBP or empty vector using a standard PEI protocol. After 48 h, cells were washed with PBS, scraped, and lysed in 1 ml of lysis buffer (50 mM Tris-HCl pH 8.0, 100 mM NaCl, 2 mM EDTA, 0.5% NP-40 and 10 mM NaF, 10 mM MgCl₂ and 10 U/mL Benzonase) on ice for 30 min. Lysates were centrifuged at 15,000 ×g for 5 min at 4 °C, and supernatants were incubated with 100 μL of Streptavidin Sepharose High Performance beads (GE Healthcare) for 1 h at 4 °C. Beads were washed 5 times with lysis buffer and eluted with 10 mM D-biotin (Invitrogen) for 2 h at 4 °C. Whole cell extracts and eluted samples were boiled 5 min in SDS-PAGE sample buffer and analyzed by immunoblotting. For GFP fused SLX4 pulldown cells were transfected with plasmids encoding GFP-SLX4, GFP-SLX4 L530A W531A, or GFP only. Extracts were subjected to immunoprecipitation with GFP-Trap beads (Chromotek).

CRISPR screens

Two RPE1-hTERT TP53^−/− Cas9 (wild-type, WT) and RPE1-hTERT TP53^−/− Cas9 (knock-out, KO) APEX1^−/−, TDP1^−/− or APEX2^−/−−1 cells were transduced with the lentiviral TKOv3 library at a low MOI (~0.35) and puromycin-containing medium was added the next day to select for transductants. Selection was continued until 72 h post-transduction, which was considered the initial time point, t0. Cells were subcultured every 3 days. Cell pellets were frozen at day 18 for gDNA isolation. Screens were performed in technical duplicates and library coverage of ≥ 375 cells per sgRNA was maintained at every step. gDNA from cell pellets was isolated using the QIAamp Blood Maxi Kit (Qiagen) and genome-integrated sgRNA sequences were amplified by PCR using Q5 Mastermix (New England Biolabs Next UltraII, M5044S). i5 and i7 multiplexing barcodes were added in a second round of PCR and final gel-purified products were sequenced on Illumina NextSeq500 systems to determine sgRNA representation in t0 and t18 timepoints of each sample.

To identify the list of genes whose inactivation causes loss of fitness, we compared sgRNA depletion using JACKS and BAGEL algorithms (t0 vs t18) (Figure 2A and Table S1). Bona fide threshold values were chosen (Table S2) as to ensure we removed parental cell line essential genes, but captured the most expressive changes between wild type and knock-out conditions, being consistent and stringent enough to reduce false positives: JACKS average wild type (WT) > −1.2, delta average BAGEL > 12, knockout (KO) BAGEL > 15 and JACKS linear regression residuals > 0.3, where by “average” we mean the arithmetic mean of both WT scores and by “delta” we mean subtracting BAGEL scores in the KO sample from the average of two independent WT scores. The JACKS linear regression was performed in R treating the JACKS average WT as dependent variable and KO JACKS as independent variable. Both algorithms were implemented under their standard parameters, where the essential and nonessential gene lists (CEGv2.txt and NEGv1.txt) for BAGEL were obtained from Hart’s lab GitHub page (https://github.com/hart-lab/bagel, March 26^th 2019 update).

Antibodies

Primary antibodies: mouse anti-APEl (ThermoFisher MA1–440, 1:1000), mouse anti-PCNA (Abcam PC10 ab29, 1:1000), rabbit anti-ERCC4 (Bethyl A301–315A, 1:1000), mouse anti-ERCCl (Santa Cruz D-10 sc-17809, 1:1000), rabbit anti-APE2 (in house, 1:1000), mouse anti-α-Tubulin (Millipore DM1A cp06, 1:5000), rabbit anti-TDP1 (Abcam ab4166, 1:1000), rabbit anti-GFP (Abcam ab290–50, 1:1000), sheep anti-GFP (Figure S7X, In house S268B, 0.2 μg/ml), sheep anti-SLX1 (In house S701C, 0.5 μg/ml), mouse anti-MUS81 (Immuquest MTA30 2G10/3, 1:500), mouse anti-ERCC4 (Figure S7F, Thermo scientific AB-1 clone 219, 1:5000), Sheep anti-EME1 (In house S180D, 0.2 μg/ml), Rabbit anti-PLK1 (Cell signalling 4535, 1:500). Secondary antibodies: goat anti-mouse Irdye 800CW and anti-rabbit Irdye 680RD (926–32210 and 926–68071 LiCOR, 1:5000). Secondary antibody detection was achieved using an Odyssey Scanner (LiCOR).

Protein depletion

siRNA oligonucleotides were transfected in Opti-MEM reduced-serum medium using Lipofectamine RNAiMAX agent (Thermo-Fisher) following manufacturer recommended protocol. Fresh media was added to cells 16 h after transfection. Cells were used for clonogenic survival assay and immunobloting 48 h after transfection of the following siRNAs: SMARTpool ON-TARGETplus TDP1 siRNA L-016112-00-0005 and ON-TARGETplus non-targeting pool siCTRL-SP, D-001810-10-05 (Dharmacon/BD Technologies).

Cytogenetic analysis

5×10⁶ RPE1-hTERT cells were seeded in 10-cm dishes 4 days after transduction with virus particles expressing NLS-mGFP-sgAAVSl (control) or an NLS-GFP-sgRNA targeting a specific gene of interest (see Table S3). 24 h later BrdU (final concentration 10 μM) was added to the media and cells were grown for 48 h. 100 ng/mL KaryoMAX colcemid (Gibco/Thermo Fisher) was then added for 2 h, and cells were harvested as follows:

Growth medium was stored in a conical tube. Cells were gently washed and treated twice for 5 min with 1 mL of trypsin. The growth medium and the 2 ml of trypsinization incubations were centrifuged (1000 rpm, 5 min, 4°C). Cells were then washed with PBS and resuspended in 75 mM KCl for 15 min at 37°C. Cells were centrifuged again, the supernatant was removed, and cells were fixed by drop-wise addition of 1 mL fixative (ice-cold methanol: acetic acid, 3:1) while gently vortexing. An additional 9 mL of fixative was then added and cells were fixed at 4°C for at least 16 h. Once fixed, metaphases were dropped on glass slides and air-dried overnight, protected from Ugk.

To visualize chromosomal aberrations, slides were dehydrated in a 70%, 95% and 100% ethanol series (5 min each), air-dried and mounted in DAPI-containing ProLong Gold mounting medium (Molecular Probes/Thermo Fisher). To visualize sister chromatid exchanges (SCE) slides were rehydrated in PBS for 5 min and stained with 2 μg/mL Hoechst 33342 (Thermo Fisher) in 2xSSC (final 300 mM NaCl, 30 mM sodium citrate, pH 7.0) for 15 min. Stained slides were placed in a plastic tray, covered with a thin layer of 2xSSC and irradiated with 254 nM UV light (~5400 J/m2). Slides were subsequently dehydrated in a 70%, 95% and 100% ethanol series (5 min each), air-dried and mounted in DAPI-containing ProLong Gold mounting medium (Molecular Probes/Thermo Fisher). Images were captured on a Zeiss LSM780 laser-scanning confocal microscope.

Protein production

Full length MBP-fused APE2 from X. laevis (used in assays for Figures 4A, S4ABGH and S5C) along with the APE2^Cat protein (residues 1–355) from X. laevis (used for crystallography) were expressed and purified as described in (Wallace et al., 2017). X. laevis PCNA (used in Figure S4ABGH), was expressed and purified as described (Lin et al., 2018). Codon-optimized cDNAs encoding for human full-length APE2 protein with C-terminal His8-Flag tags (used for assays in Figures 4DF, S4DE and S5AC) was obtained from Genscript and cloned into the pTT5TM expression vector. His8-Flag-tagged APE2 was transfected in suspension HEK293–6E cells (Durocher et al., 2002) by the PEI method in FreeStyle F17 media (Invitrogen) as described in (L’Abbe et al., 2018). Transfected cells were lysed in TNG (50 mM Tris-Cl pH 7.5, 150 mM NaCl, 10% Glycerol, 0.1 % Nonidet P40 Substitute, 1 mM TCEP) supplemented with complete EDTA-free protease inhibitors (Roche). The lysate cleared by centrifugation was batch-adsorbed onto anti-FLAG M2 affinity gel resin (MilliporeSigma), was washed with TNG, followed by elution with 100 μg/mL FLAG peptide in TNG. Eluted APE2 was directly loaded onto a HiTrap nickel column (GE Healthcare), was washed with TNG with an intermediate step of TNG supplemented with 1 M NaCl, 3 mM MgCl2, 1 mM ATP to remove HSP70. After elution using a linear gradient of imidazole in TNG, fractions were pooled, buffer exchanged into TNG and stored at −80°C. Human PCNA (used for assays in Figures 4DF, S4DE and 6AC) was expressed and purified as previously described (Hishiki et al., 2008).

APE2 catalytic domain crystallization, data collection, structure solution and refinement APE2^Cat protein was buffer-exchanged into 10 mM Tris-Cl pH 8.0, 100 mM NaCl, 1 mM MgCl₂, 1 mM TCEP and concentrated to 10 mg/mL. Crystallization trials with commercially available screens (Qiagen) were utilized for initial hit discovery using the sitting drop vapor diffusion method. Initial crystals of the APE2^Cat domain were optimized to grow in 25% PEG 6000, 0.1 M MES pH 6.5 at 20 °C at a 2:1 and 1:1 protein: crystallant ratios. Crystals were cryoprotected with mother liquor supplemented with 26% ethylene glycol and flash-frozen in liquid nitrogen for X-ray data collection. X-ray data was collected at 105K on beamline 22-ID of the Advanced Photon Source. Data reduction and scaling was performed with the HKL2000 suite (Otwinowski and Minor, 1997). The APE2^Cat structure was solved by molecular replacement in PHENIX (Adams et al., 2010) with PDB 3G0R (Lakomek et al., 2010) as a search model. A combination of density modification with Autobuild and manual refinement in COOT (Emsley and Cowtan, 2004) produced a model that was refined to 1.54 Å in PHENIX. The final model (R=0.143, R_free=0.158) displays good geometry statistics (bonds=0.014, angles=1.34; Ramachandran statistics, 99% favored, 0% outliers). Data collection and refinement statistics are provided in detail in Table S4. The structure and associated experimental data was deposited in the RCSB PDB as 6WCD.

Preparation of Oligonucleotide Substrates

The sequences of DNA substrates used in the fluorescence-based nuclease and nuclease cleavage experiments are detailed in Table S5. All oligonucleotides were synthesized by IDT except 5′-FAM 3′-pTyr (Midland) and 5′-FAM 2′3’-P (Chemgenes). The DNA substrates used in the experiments with TOP1 are modified from (Interthal et al., 2001) to include a 5′-Cy5 label and bear a ribonucleotide U versus deoxyribonucleotide T at the 12 nt position of CL14N (Table S5).

Fluorescence-based Nuclease Assay

5 μL of XlAPE2 WT or E34Q/D183N mutant, as a 50 nM stock in storage buffer (5 mM Tris-Cl pH 8.0, 75 mM NaCl, 0.5 mM TCEP, 0.5 mM MgCl2, 50 % glycerol), was added to 1 μL 1 μM 3′-FAM substrate, 5 uL 10x reaction buffer (100 mM Tris-Cl pH 8.0, 10 mM TCEP, 10 mM MnCl2, 1 mg/mL BSA) and 39 μL ddH₂O at 37°C to initiate a reaction with final concentrations of 5 nM XlAPE2 and 20 nM substrate. For nuclease assays in the presence of XlPCNA, reactions were supplemented with 5 nM PCNA. The 50 μL reactions were monitored in black flat bottom 96 well plates (Costar) over 60 min at 37°C, and fluorescence measurements were collected with the POLARstar Omega microplate reader (BMG Labtech) using excitation and emission wavelengths of 485 and 520 nm, respectively. Results were plotted in GraphPad Prism and shown in Figure S4A.

Nuclease Cleavage Assays

2.5 μL of XlAPE2, as a 200 nM stock in storage buffer (5 mM Tris pH 8.0, 75 mM NaCl, 0.5 mM TCEP, 0.5 mM MgCl2, 50% glycerol), or 2.5 μL of 1 U/μL APE1 (NEB) was added to 2.5 μL 200 nM substrate, 1 uL 10x reaction buffer (100 mM Tris pH 8.0, 10 mM TCEP, 10 mM MnCl₂, 1 mg/mL BSA) and 4 μL ddH₂O at 37°C to initiate a reaction with final concentrations of 50 nM XlAPE2 and 50 nM substrate. For nuclease assays in the presence of PCNA, reactions were supplemented with 50 nM PCNA (Figure S4B). At noted timepoints, 10 μL reactions were stopped by the addition of 20 μL of formamide buffer (98% formamide, 10 mM EDTA) and immediate denaturation at 95°C for 5 minutes. Samples were resolved on 20% polyacrylamide TBE-Urea gels, imaged on a Typhoon 9500 imager (GE Healthcare), and viewed using ImageJ. For RNase H2 (NEB) reactions, 2 U/μL enzyme was used with NEB-supplied 10x ThermoPol reaction buffer (200 mM Tris-HCl pH 8.8, 100 mM (NH₄)₂SO₄, 100 mM KCl, 20 mM MgSO4, 1% Triton X-100).

TOP1 Protein Purification

The Topo70 protein, an active N-terminal truncated 70kDa form of human Topoisomerase I (Stewart et al., 1996), was purified from suspension HEK293F cells transfected by PEI in FreeStyle media (Invitrogen) with a Vivid Colors pcDNA6.2/N-YFP-DEST vector (Invitrogen) containing amino acids 175–765 of Topoisomerase I with an N-terminal YFP and TEV site. Transfected cells were lysed by sonication in 50 mM Tris-Cl pH 8, 600 mM NaCl, 1 % NP-40, 0.5 mM TCEP, 2 mM MgCl₂, EDTA-free cOmplete-mini protease inhibitors (Sigma-Aldrich), and the lysate cleared by centrifugation was loaded onto anti-GFP nanobody resin (Schellenberg et al., 2018). Resin was washed with 0.5% NP-40 modified lysis buffer, followed by 400 mM NaCl, 0.1% NP-40 modified lysis buffer, followed by TEV buffer (20mM Tris-Cl pH 8.0, 150mM NaCl, 0.1% Tween-20, 0.5 mM TCEP, 2 mM MgCl₂) then incubated with 50 μg/mL TEV overnight at 4°C to release the target protein. Cleaved and eluted Topo70 was loaded onto a Superdex 200 column in 20 mM Tris-Cl pH 8.0, 300 mM NaCl, 1 mM TCEP and fractions assessed by SDS-PAGE. Pure fractions were pooled, concentrated, and stored in 10 mM Tris-Cl pH 8.0, 150 mM NaCl, 0.5 mM TCEP, 50% glycerol at − 80°C.

Nuclease Assays with Topoisomerase I-generated suicide Substrates

200 nM Topo70 or Topo70 storage buffer was incubated with 100 nM DNA suicide substrate in 10 mM Tris-Cl pH 7.9, 150 mM NaCl, 0.1% BSA, 100 μM spermidine, 5% glycerol reaction buffer (TopoGEN) for 2 hours at 37°C. RNase-free trypsin (Sigma-Aldrich) or 20 mM MES pH 6, 150 mM NaCl buffer was then added to a final trypsin concentration of 50 ng/μL and incubated for 1 hour at 37°C followed by addition of PMSF to 10 mM for 10 minutes at room temperature. These reactions were then diluted four-fold and 75 or 300 nM hAPE2, hAPE2/hPCNA, TDP1 (NovusBio), or storage buffer (10 mM Tris-Cl pH8.0, 50 mM NaCl, 1 mM TCEP, 1 mM MgCl₂, 50% glycerol for APE2; PBS, 10% glycerol for TDP1) added to test nuclease activity on the TOP 1-generated substrates in either 10 mM Tris-Cl pH 8.0, 1 mM TCEP, 1 mM MnCl₂ for APE2 reactions or 10 mM Tris-Cl pH 7.5, 100 mM KCl, 1 mM EDTA, 1 mM DTT for TDP1 reactions. Following a 0.5 h incubation, the reactions were stopped by adding 2 volumes of formamide buffer to 1 volume of reaction followed by immediate denaturation at 95 °C for 5 minutes (See Figure 4C). For CIAP treated samples, 1 unit of CIAP (Invitrogen) was added to enzyme reactions and incubated for 30 min at 37°C before stopping as previously described. Samples were resolved on 20% polyacrylamide TBE-Urea gels, imaged on a Typhoon 9500 imager (GE Healthcare), and viewed using ImageJ. For high molecular weight resolution of Top1cc (Figure S4DEGH), samples were run on 4–12% BisTris-SDS precast mini gels, similarly imaged, and quantified by Image Lab (BioRad).

Quantification and Statistical Analysis

Statistical analysis summarized in Table S6 was performed using R (version 3.6.1) under the standard, “rstatix” and “lsmeans” packages, or using PRISM GraphPad for exponential decay analysis in clonogenic survival assays. Only statistically significant (p/fdr values ≤ 0.05) in key comparison supporting the conclusions of this manuscript are indicated in the figures.

Supplementary Material

3

Table S1. Scores reflecting the likelihood of “essentiality” for each gene using BAGEL and JACKS algorithms. Relates to Figure 2.

4

Table S2. List of gene hits from mutated isogenic cell lines. Relates to Figure 2.

5

Table S3. List of sgRNA and PCR primers used for CRISPR-Cas9 Editing and validation. Relates to Figures 1–3, 6, S1–S3 and S7

6

Table S6. Statistical analysis. Relates to Figures 1, 2, 3 and 6.

Highlights.

• Loss of APE2 is lethal in cells with mutations in BRCA1 or BRCA2

• The APE2 DNA repair nuclease removes endogenous DNA 3′ blocks

• 3′ blocks arising from ribonucleotides cause the APEX2-BRCA1/2 synthetic lethality

• DNA 3′ block-resolving pathways are vulnerabilities for HR-deficient tumor cells