Uncovering genetic interactions in the DNA repair network in response to endogenous damage and ionizing radiation

Benjamin Nebenfuehr; Lynn Sanford; Erin R Taylor; Kerri Ball; Catherine E Woods-Killam; Hannah I Ghasemi; Bruce Proctor, III; Raquel Ortega; Colin Sempeck; Robin D Dowell; Nausica Arnoult

doi:10.1016/j.celrep.2025.116850

. Author manuscript; available in PMC: 2026 Mar 10.

Published in final edited form as: Cell Rep. 2026 Jan 16;45(1):116850. doi: 10.1016/j.celrep.2025.116850

Uncovering genetic interactions in the DNA repair network in response to endogenous damage and ionizing radiation

Benjamin Nebenfuehr ¹, Lynn Sanford ^1,², Erin R Taylor ¹, Kerri Ball ¹, Catherine E Woods-Killam ¹, Hannah I Ghasemi ¹, Bruce Proctor III ¹, Raquel Ortega ¹, Colin Sempeck ¹, Robin D Dowell ^1,², Nausica Arnoult ^1,^3,^*

PMCID: PMC12970561 NIHMSID: NIHMS2142944 PMID: 41546867

SUMMARY

Genomic integrity relies on a complex network of DNA damage response (DDR) pathways that repair endogenous and exogenous lesions, yet how individual factors operate within this broader landscape remains unclear. We performed a large-scale combinatorial CRISPR-Cas9 knockout screen targeting 461 DNA repair genes, disrupting over 100,000 gene combinations under basal conditions and after ionizing radiation (IR). This approach uncovered thousands of genetic interactions spanning pathways that respond to endogenous damage and those specific to double-strand break repair. From this dataset, we validated both positive and negative interactions under basal and irradiated conditions, including a synthetic lethal relationship between MRE11A and the E3 ligase UBR5, a role for Ku70/80 in preventing unscheduled nuclease activity at telomeres, an IR-specific vulnerability upon co-disruption of CYREN and PARG, and a link between CYREN-mediated radioresistance and innate immunity. This resource enables mechanistic insight and reveals therapeutic vulnerabilities in DNA-repair-deficient cancers.

In brief

Nebenfuehr et al. perform a combinatorial CRISPR-Cas9 screen targeting 461 DNA damage response genes under basal and irradiated conditions. The study maps thousands of genetic interactions, generating a rich resource to uncover mechanisms of genome maintenance and inspire future studies.

Graphical Abstract

graphic file with name nihms-2142944-f0006.jpg

INTRODUCTION

DNA is the blueprint of life, and as such, organisms have evolved robust mechanisms to preserve genomic integrity. These mechanisms detect damage, orchestrate the recruitment of repair factors, mediate the repair process, and are subsequently disassembled once repair is complete. Traditionally, these repair pathways have been studied in isolation, assuming each employs specific factors to repair distinct types of DNA damage. However, expanding insights into the DNA damage response (DDR) network revealed a far more intricate landscape. Many repair factors function across multiple lesions and pathways, and significant redundancy exists within the system. While such redundancy enhances protection against genomic instability, it adds complexity to studying the DDR.

The identification of genetic interactions is a powerful strategy for revealing alternative or unexpected functions of known DDR proteins, as well as the regulatory mechanisms governing them.^1–3 For instance, screening for synthetically lethal genetic interactions with homologous recombination (HR) genes BRCA1, BRCA2, and PALB2 led our group to identify a function for the APE2 nuclease in microhomology-mediated end joining (MMEJ) and the Durocher group to demonstrate its role at 3′ blocking lesions.^4–6 Motivated by these observations, we aimed to take a broad, systems-level view of the genetic interactions among DNA repair genes by assessing the effects of combinatorial gene knockouts (KOs) across the DDR network. This strategy allows us to better understand the network contexts in which individual genes operate, thereby providing deeper insight into their functional roles. When selecting genetic interactions to evaluate, we noted that previous studies investigating the genetic interactions of individual DNA repair genes have predominantly identified other known or predicted DDR-related genes,^1–7 suggesting a primarily self-contained genetic network. Based on this, we exclusively focused on genes already annotated as part of the DDR.

Identifying genetic interactions yields two key types of information. First, such interactions can uncover previously unrecognized roles of certain proteins in DNA repair pathways or reveal the regulatory mechanisms that govern them. Unchallenged genetic screens—those performed under baseline conditions—can expose mechanisms involved in responding to endogenous DNA damage. In contrast, combinatorial screens conducted under specific genotoxic stresses can illuminate pathways that respond to particular types of damage, such as DNA double-strand breaks (DSBs) or replication stress. Second, and more translationally relevant, genetic interactions can highlight therapeutically exploitable vulnerabilities. A well-known example is the synthetic lethality between HR deficiency and inhibition of MMEJ-associated DNA polymerase theta (Polθ), which has led to the development of Polθ inhibitors that are now in clinical trials for HR-deficient cancers, such as breast and ovarian cancers.⁸

Given these promising possibilities, we set out to explore genetic interactions between DDR genes. This strategy has been employed by other groups using slightly different strategies.⁹ We chose to especially focus on the combinatorial response to ionizing radiation (IR), an approach motivated by two factors. First, approximately half of all patients with cancer receive radiation therapy as part of their treatment regimen.^10,11 Thus, mapping genetic interactions in this context could identify targets for radio sensitization. Second, since IR primarily induces DNA DSBs,¹² this screening strategy offers an opportunity to dissect combinatorial mechanisms within the end-joining repair pathways.

We leveraged CRISPR-Cas9 and oligo synthesizing technologies to build a combinatorial KO library of 907,236 guide pairs, roughly twelve times the size of a genome-wide CRISPR-KO library such as the Brunello¹³ (76,441 guide RNAs [gRNAs]) or Toronto Knockout v.3 (TKOv3) Library^14,15 (70,948 gRNAs). This combinatorial library allowed us to measure pairwise genetic interactions among 461 DDR genes and their contextual contributions to cell viability under basal conditions or upon IR.

We uncovered thousands of previously unidentified genetic interactions, from which we validated interactions that shed light on molecular mechanisms involved in telomere maintenance, resection regulation, and DSB repair. Yet, these validated interactions represent only a minuscule fraction of those observed across our two screens. Thousands more remain to be explored and merit further investigation. As such, this database offers a tremendous resource to inform future studies of the DDR in the contexts of both endogenous damage and IR.

RESULTS

Construction of a combinatorial gRNA library to target the DNA repair network

We set out to analyze the genetic interactions between genes within the DDR network. We selected 461 genes matching the ‘‘DNA repair’’ Gene Ontology (GO: 0006281) and pulled four guides per gene from the Brunello Genome Wide Library and the TKOv3 Library, along with 61 non-targeting (NT) guide sequences.^13–15 These genes comprise diverse aspects of the DDR and associated cellular processes (Figure 1A). To construct our double-KO DNA damage (DDR-DKO) library, we adapted the CombiGEM strategy,¹⁶ which is summarized in Figures S1A and S1B. In brief, we designed and synthesized two types of pooled oligos: pool A, consisting of guides 1 and 2 for each gene, and pools B1–B7, smaller sub-pools containing guides 3 and 4 against each gene. Both oligo pools contained a 20 nt gRNA sequence, tracR, restriction sites for cloning, and barcodes unique to each guide sequence (Figures 1B, S1A, and S1B). Pool B oligos contained a modified tracR sequence (tracR-2) to prevent recombination during library cloning.¹⁷ These pools were combined over two cloning steps to yield seven pools of dual-guide plasmids, containing an hU6 promoter, gRNA-1, tracR-1, an hH1 promoter, gRNA-2, tracR-2, and finally, barcode-2 directly followed by barcode-1 (Figures 1B, S1A, and S1B). We deep sequenced these seven pools to confirm that 90% of guide pairs fell within a 10-fold distribution, indicating even representation across all guide pairs (Figure S1C). We then combined these seven dual-guide pools equimolarly into a single library consisting of 907,236 unique guide pairs targeting 106,260 gene combinations (Figure 1B).

Figure 1. — (A) Enrichment analysis of genes represented in our combinatorial library (ShinyGO).

(B) Schematic representation of the dual-guide construct used to target two DDR genes simultaneously. 461 genes were each targeted by four sgRNA sequences (two per locus) along with 26 non-targeting sequences in the first locus (red) and 35 non-targeting sequences in the second locus (blue). Each locus is expressed by distinct RNA polymerase III (RNA Pol III) promoters (hH1 or hU6) and supported by unique tracR scaffolds (tracR-1 and tracR-2). Unique barcodes match each guide sequence. This results in 907,236 guide pairs representing 106,260 gene combinations.

(C) Timeline for basal and ionizing radiation (IR) screens. For each of two replicates, 2.3 billion eHAP-iCas9-cl#6 cells were transduced with a lentiviral (LV) library containing our DDR-DKO library at a multiplicity of infection (MOI) of <0.2. 48 h after puromycin (puro) selection, 500 million cells were pelleted from each replicate (day 0 time point) and 2 × 500 million cells were seeded in doxycycline (Dox; 1 mg/mL) for the basal and IR screens. Basal screen cells were split every 48 h and pelleted on day 14 (basal final) for library amplification. Cells in the IR screen were split every 3 days with 0.6 Gy IR 24 h after every split (red arrows) and collected at day 15 (IR final). −dox indicates when Dox was withdrawn from the media.

(D) Cumulative population doublings for each replicate of the basal and IR screens. The regular IR treatments correspond to an LD30 relative to the basal replicates.

(E) Graphical representation of screen analysis. Guide-pair abundance was used to build gene-pair observed phenotypes, where significance considers the similarity of each of the eight guide pairs against a pair of genes. These observed phenotypes for basal and IR screens can also be visualized relative to one another.

See also Table S1 and Figures S1 and S2.

Basal and irradiation screens

For our combinatorial screen, we chose to work with engineered haploid (eHAP) cells (Horizon Discovery C669), which are derived from HAP1 human leukemia cancer cells. We reasoned that our dual-KO library would be more efficient at simultaneously knocking out two genes in a haploid genome, but we found that these cells tend to revert to diploidy after several passages in culture.¹⁸ However, their small size remained a key advantage in performing such a large screen.

We also wanted to perform the screen using an inducible Cas9 (iCas9) to have temporal control over KO initiation. Indeed, when using a constitutively active Cas9, one can only compare the final time point to a time 4–5 days post-library infection, when some genetic interactions may have already affected cell growth. At the level of the parental population, eHAP cells with doxycycline-iCas9 presented unreliable Cas9 activity with inconsistent induction (Figure S1D). We therefore sought to work with a clonal population with stronger control of Cas9’s activity. We isolated over 100 clones of eHAP cells transduced with iCas9 and screened these clones using a plasmid expressing GFP, BFP, and a gRNA targeting GFP.¹⁹ Using this reporter, we measured GFP loss as a proxy for Cas9 activity. We selected clone #6 (eHAP-iCas9-#6), which displayed less than 10% GFP loss in DMSO (low Cas9 leakage) and over 95% GFP loss upon doxycycline treatment (robust Cas9 induction) (Figures S1D and S1E).

For each replicate of the screen, we prepared a separate batch of lentivirus from our pooled DDR-DKO library. We infected 2.3 billion eHAP-iCas9-#6 cells at a multiplicity of infection (MOI) of ~0.2 to ensure single integration at 500× coverage (Figure 1C). After puromycin selection, cells were pelleted for DNA extraction as our pre-induction sample (day 0) or re-seeded with doxycycline for our basal and IR screens. Basal screen cells were grown for 14 days, representing 13 population doublings (PDs) (Figures 1C and 1D) with doxycycline withdrawal after 7 days. For each passage, 500 million cells were seeded to maintain library coverage. Inherently, cells exposed to regular doses of IR grew more slowly, and these cells were passaged until 10 PDs had been reached, on day 15 of the screen (Figures 1C and 1D), representing a lethal dose (LD) of 30 over the time course of the screen. Upon completion, we froze cell pellets and subsequently extracted genomic DNA from 500 million cells per sample to prepare for next-generation sequencing (see STAR Methods).

Sequencing and analysis of the combinatorial screens

While we initially planned to sequence only the barcode regions, results from other studies identifying recombination bias in combinatorial libraries^20,21 prompted us to sequence the entire dual-guide cassette instead, spanning both guide and barcode sequences using 250 bp paired-end sequencing. On average, the sequencing yielded ~300 million reads per sample (Figure S2A).

We found that the recovery and mapping of guide-1 was excellent, but guide-2 was successfully mapped in less than half the reads, most likely due to progressively poor read quality across the length of read-2 (Figures S2B and S2C). Since guide-1 and barcode-1 had much better quality and mapping rates, we checked how often guide-1 correctly matched its barcode-1. Strikingly, we found that in reads where guide-1 and barcode-1 were successfully identified and mapped, only 29%–30% of sequences showed proper coupling between the two (Figures S2A and S2C). This highlights the very high rate of recombination that occurs during library cloning, sequencing preparation, or both. Analysis of recombined libraries is likely to significantly affect the quality of the produced data. We reasoned that if guide-1 and barcode-1—which are located at opposite edges of our construct—were coupled, it was unlikely that recombination occurred between them to affect guide-2 and barcode-2. Therefore, to circumvent both the problem of library recombination and the problem of poor guide-2 read quality, we moved forward by analyzing barcode-1 and barcode-2 from sequences that showed proper coupling between guide-1 and barcode-1. This yielded an average of 73.5 million observed guide pairs per sample, corresponding to over an 80× coverage of the library. Though this final coverage is slightly lower than recommended, we prioritized accuracy by focusing on high-confidence reads, minimizing the impact of recombination on our analysis. In total, we recovered 104,563 gene combinations, or 98.4% of the input. These gene combinations showed similar distributions in both day 0 replicates (Figure S2D), suggesting that our replicates had similar initial conditions.

For each gene, we defined an ‘‘observed phenotype’’ based on fold changes in day 14 or 15 samples compared to the corresponding day 0 samples, calculated as the log₂(fold change) (Figure 1E). We then identified significant gene combinations in relation to the bulk of gene-gene phenotypes observed with those genes and filtered for gene combinations with a Fisher-combined p < 0.05 and an observed phenotype score greater than one standard deviation from the mean of all phenotypes in that condition (Figures 2A–2C).

Screen results overview

Overall, we recovered 98.4% of the input gene combinations (Figure 2A). Among them, using the analysis described above, thousands of candidate interactions were significant (17,084 or 16.34% of all gene combinations in the basal screen and 17,404 or 16.51% of all gene combinations in the IR screens; Figures 2A–2D). Moreover, 16.3% (4,917 gene combinations) of the significant interactions in basal conditions were also found in the IR screen (Figures 2A and 2D, red), indicating genetic interactions that exist independently of external DNA damage sources. In contrast, only 0.32% of total interactions (335 gene combinations or 1.9% of significant basal interactions) showed opposing phenotypes between the basal and IR screens (Figures 2A and 2D, purple). While some of these may represent true interactions with markedly different effects on cell growth in response to IR, they are more likely to be false positives. Therefore, we excluded these ‘‘opposite’’ interactions from further analysis.

As a first step toward validation, we looked for genetic interactions that we would expect to find in our basal results. We identified numerous known genetic interactions, such as BRCA1 x PARP10, BRCA1 x RAD18,²² and the well-established synthetic lethality between PARP1 and PARP2,²³ in our basal screen (Figure 2B). Interestingly, other known interactions, such as synthetic lethality between HR and MMEJ genes, did not appear depleted in our screen. It is unclear whether these are false negatives or are non-significant due to the temporal kinetics of genetic interactions: in response to gene loss, cells frequently rely on alternative genes or pathways, resulting in synthetic lethality. Yet, those same two genes might not present as synthetically lethal when concurrently depleted, as no adaptation occurred (see discussion).

Since combinatorial screens like ours are likely to produce a higher false positive and false negative rate than standard CRISPR screens, we took a more holistic approach when identifying compelling genetic interaction candidates to validate. First, we looked at the significant interactions for each gene, as defined above, and prioritized reciprocally significant interactions for both genes. We then selected some of these interactions for further validation in separate cell lines from the one screened: HT1080-6TG fibrosarcoma cells and U2-OS osteosarcoma cells.

The E3 ubiquitin ligase UBR5 is synthetic lethal with MRE11

We first examined our basal screen for depleted interactions and identified a shared synthetic lethal relationship between the E3 ubiquitin ligase UBR5 and the nuclease component of the damage-sensing MRN (MRE11-RAD50-NBN) complex, MRE11A. Although this pair was not among the top interactions overall (Figure 3A), each gene ranked as one of the most depleted hits in the other’s KO background (Figure 3B). Moreover, the same phenotype was observed in both the basal and IR screens, supporting the robustness of this interaction (Figure S3A). Chemical inhibition using mirin, a potent inhibitor of MRE11 exonuclease activity, provided us with an avenue for easily investigating this interaction.^24,25 First, we used an inducible, dual-gRNA system to better mimic the screen conditions. We transduced HT1080-6TGs expressing a nuclear EGFP²⁶ with an iCas9 and UBR5 or NT gRNA. A control experiment assessing GFP²⁶ loss upon sgEGFP-NT expression was used to assess this inducible dual-gRNA system (Figure S3B). We mixed these cells 1:1 with HT1080-6TGs expressing a nuclear RFP and followed the growth of the GFP/RFP population upon doxycycline induction and mirin treatment (Figures 3C and S3C). Cells treated with doxycycline showed increased sensitivity to mirin compared to uninduced UBR5-guide-harboring cells and NT doxycycline-induced controls (Figures 3D, 3E, and S3D), reproducing the growth defect observed in the screens. As an orthogonal approach, we electroporated HT1080-6TGs with a ribonucleoprotein (RNP) targeting UBR5 and performed a mirin dose curve. sgUBR5-a cells were more sensitive to mirin than cells electroporated with an NT guide (Figures S3E–S3G). We next isolated UBR5-KO clonal HT1080-6TG lines and performed colony formation assays with increasing doses of mirin. Again, UBR5-KO cells were significantly more sensitive to MRE11 inhibition than wild-type (WT) cells (Figures 3C, 3D, S3H, and S3I). Finally, we confirmed this sensitivity in another cell line, U2-OS (Figures 3H and 3I). These results suggest that UBR5 may regulate a mechanism that compensates for the loss of MRE11. While UBR5 is known to regulate p53 and ATM signaling,^27,28 the complete picture of its function in the DDR is yet to be characterized. As such, this interaction represents an unexpected finding and paves the way for further study of the regulatory functions of UBR5 in the DDR.

Figure 3. — (A) Distribution of the observed phenotypes depleted in the basal screen results, indicating where the MRE11A-UBR5 gene pair falls relative to other depleted interactions.

(B) Ranked observed phenotypes from the basal screen for both *UBR5* and *MRE11A*. Blue/red regions highlight significantly enriched/depleted *UBR5*-gene and *MRE11A*-gene interactions, respectively. Blue and red regions highlight gene combinations that are enriched above (blue) or depleted below (red) one standard deviation (+0.6981 and −0.7791) of the mean of all basal interactions and possess a Fisher-combined p < 0.05.

(C) Time-lapse images of growth from Dox-induced dual-guide vectors with NT-1 x *UBR5*-g2 or NT-1 x NT-2 in nuclear-EGFP-expressing cells co-cultured with nuclear-RFP-expressing cells grown in DMSO or 6.25 μM mirin (+mirin). Images are from one well per condition over the first 72 h of growth. Scale bar: 100 μM.

(D) Measurements are calculated as the log₂ fold change (log₂FC) of Dox induced over uninduced growth for the labeled condition, relative to the growth of co-cultured RFP+ cells, as seen in (C). Data represent mean ± SEM of 3 biological repliactes. The gray bar indicates the 72 h of growth shown in (C), though imaging continued.

(E) Comparative growth differences at 96 h of two separate, inducible UBR5 guides from experiments performed similarly to those shown in (C) and (D). Bars are only shown for Dox-treated conditions but are normalized to uninduced, DMSO-treated control wells for each genotype and replicate. Mirin: 5 μM. Blue prediction bars are calculated from an additive model assuming no interaction, with the average effect of mirin on NT cells plus the average effect of *UBR5* depletion for each guide and replicate. Data represent mean ± SEM of 2 biological replicates, and statistics represent a two-way ANOVA with Dunnett multiple comparisons test; p = 0.0277.

(F) Western blot depicting robust *UBR5* KO in two selected HT1080-6TG clones.

(G) Dose curve of HT1080-6TG clones shown in (F) grown in increasing doses of mirin in a colony formation assay. IC50 was calculated from a sigmoidal interpolation of all datasets. Error bars: ± SEM from three biological replicates. Statistics match in color the clone they reference, relative to NT cells. Statistics: a two-way ANOVA with Dunnett’s multiple comparisons test between NT-treated cells and *UBR5*-KO clones at each concentration; *p < 0.01 and ****p < 0.0001.

(H) Western blot showing UBR5 depletion in U2-OS cells electroporated with either an sgNT-RNP or an sgUBR5-RNP.

(I) Log₂ of the relative confluence over time (relative to DMSO) for sgNT-RNP- or sgUBR5-RNP-treated U2-OS cells from (H), grown with 25 μM mirin over 120 h. The shaded area around each solid line indicates the SEM from three biological replicates. Statistics: a two-way ANOVA at 120 h with Šídák’s multiple comparisons test; p = 0.0317.

See also Figure S3.

Suppression of Apollo and XPF rescues cellular fitness upon partial Ku70/80 loss

It has previously been established that human cells depend on the Ku70/Ku80 heterodimer (Ku), encoded by the genes XRCC6 and XRCC5, respectively. One explanation for this lethality is a role of Ku in protecting telomeres,^29,30 although other mechanisms have been proposed recently.³¹ We reasoned that combinatorial KOs capable of rescuing this lethality would reveal the mechanistic basis of Ku’s essentiality. We therefore turned to our basal screen results for gene KOs that are enriched in the Ku-depleted cells.

Among the top enriched interactions with XRCC6, we identified the E3 ubiquitin ligase RNF8, as well as three genes known to act at telomeres: the nucleases ERCC4 and DCLRE1B, which encode XPF and Apollo, respectively, and STN1, a member of the CST (CTC1-STN1-TEN1) complex (Figures 4A and 4B).

Figure 4. — (A) Distribution of the observed phenotypes (log₂FC of day 14 over day 0) for chosen genes paired with *XRCC6* among the enriched basal screen results.

(B) Ranked observed phenotypes from the basal screen for all *XRCC6* interactions.

(C–F) Quantified growth from colony-forming assays using inducible dual-guide vectors targeting *XRCC5* and *XRCC6* guides in combination with a non-targeting (NT) guide or with guides against (C) *DCLRE1B*, (D) *RNF8*, (E) *ERCC4*, and (F) *STN1*. Predicted interactions (blue boxes) are from an additive model of guide interactions, as described before. In this case, predictions were calculated for each gene pair of each replicate and plotted in the same manner as the measured phenotypes. Experiments were performed for a minimum of three biological replicates with three technical replicates each. Hollow circles indicate guide-1 or guide pair-1, while filled circles indicate guide-2 or guide pair-2. Horizontal bars across the boxes indicate the median of the log₂FC across replicates and whiskers mark the range of the data.

(G) Representative images of telomere FISH with intact telomeres or with telomere free ends (white arrows). PNA probes TelC-FAM (green) and CENPB-Cy3 (red) label telomeres and centromeres, respectively. False-colored DAPI labels DNA in gray. Scale bar: 5 μM.

(H) Quantification of free-end rates per chromosome for listed guide combos and their controls after 4 days of doxycycline induction across at least two biological replicates.

(I) Relative confluence over 4 days in HCT116-*XRCC5*^F/− post-tamoxifen addition in cells electroporated with the indicated sgRNA-RNPs. Cells were treated with either DMSO (−) or 1 μg/mL Tamoxifen (+), and growth was normalized to sgNT-RNP, DMSO-treated control. The shaded areas above and below the solid line averages indicate ± SEM from three biological replicates.

Statistics for (C)–(F) and (H): ordinary one-way ANOVA with Šídák’s multiple comparisons test. Statistics for (I): repeated-measure one-way ANOVA with Tukey multiple comparisons test between sgNT-3-treated cells and target-sgRNA-treated cells, for both DMSO- and tamoxifen-treated conditions. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. ns, non-significant. See also Figure S4.

We used our CRISPR-Cas9 inducible system with dual gRNAs to validate these interactions’ impact on cell growth. We built a series of plasmids to KO XRCC5 or XRCC6, as well as each candidate gene (DCLRE1B, RNF8, ERCC4, and STN1), in combination with each other or with an NT gRNA. Importantly, the gRNAs used in the validation were distinct from those used in the screen library (see STAR Methods). We then transduced these plasmids into HT1080-6TGs, selected cells, seeded them for colony formation assays, and treated half of the seeded wells with doxycycline to induce gene KO. Colony formation was assessed 9 days later (Figures 4C–4F, S4A, and S4B). As we had seen before, this system only partially depleted Ku (Figure S5C). However, this approach enabled us to directly compare induced vs. uninduced wells, thereby avoiding seeding biases in our colony formation assays. As expected, XRCC5/6 KO, combined with NT guides, led to decreased growth in doxycycline-treated wells (Figure S4A, top right, black box), yet we did not see population-wide death, again due to incomplete KOs. Strikingly, all our combinatorial KOs grew better than their XRCC5/6-NT counterparts (Figures 4C–4F) and significantly better than the additive prediction of gene-gene combinations (Figures 4C–4F, blue boxes). These results were consistent across two guides per target (Figures 4C–4F, solid vs. filled dots). Importantly, Ku was depleted to similar levels in the DKOs and single KOs (Figure S4C).

Since losing the Ku heterodimer was reported to result in telomere free ends,³⁰ we reasoned that Ku could act by blocking deleterious telomere trimming events by these nucleases. We performed metaphase spreads and telomere fluorescence in situ hybridization (FISH) and quantified the percentage of telomeres with free ends in NT-1-sgXRCC6-g2 compared to DCLRE1B/ERCC4-XRCC6. Though the CST complex is not expected to resect telomeres, we decided to also include STN1 in this analysis. Remarkably, while we reproduced the telomere-free-end phenotype in NT-1-sgXRCC6-g2 cells, combinatorial depletion of XRCC6 with ERCC4 or DCLRE1B suppressed this telomere loss phenotype and displayed baseline numbers of free ends (Figures 4G and 4H). These results suggest that both Apollo and XPF nuclease activities mediate Ku-depletion-associated telomere loss, which could explain why dual gRNAs targeting XRCC6 with these nucleases show improved growth. On the contrary, we observed no change in free ends when depleting XRCC6 in combination with STN1, indicating that Stn1 plays a distinct role in the context of Ku loss.

We next sought to determine whether suppression of Apollo, XPF, and Stn1 could fully rescue the lethality associated with the complete loss of Ku. To test this, we used RNP electroporation to KO these genes in HCT116 XRCC5^F/− cells,²⁹ seeded them for a live-imaging growth assay, and induced XRCC5 KO with tamoxifen. Under these conditions, XRCC5 loss led to complete cell death in a few days (Figure S4D), but none of our gene KOs conferred rescue (Figures 4I and S4E–S4G). To ensure that this was not due to cell line differences, we isolated clones of ERCC4- and DCLRE1B-KO HT1080-6TGs (Figures S4H and S4I) and electroporated them with an RNP targeting XRCC5, leading to full depletion of Ku (Figure S4J). Again, depletion of XPF and Apollo failed to rescue the full lethality of Ku loss (Figure S4K). Therefore, while suppression of XPF and Apollo alleviates the growth defects of cells with partial Ku depletion, it is insufficient to support the survival of Ku-KO cells. One possible explanation is a rescue of Ku haploinsufficiency. Indeed, despite Ku’s abundance, loss of one allele leads to proliferation defects and reduced NHEJ activity, indicating that a reduction in Ku levels can affect its capacity to act at discrete genomic sites.³² Our data therefore suggest a potential role for Ku at human telomeres in preventing unscheduled activity of Apollo and XPF nucleases and that reduced Ku levels compromise this function, leading to fitness defects. However, the essential, telomere-independent activity of Ku complicates the precise characterization of its telomeric functions. Importantly, beyond highlighting a telomeric role for Ku, these findings also demonstrate that our screen can capture the effects of hypomorphic gene suppression.

Regulatory interplay between CYREN and PARG

Finally, we turned to interactions that were especially strong upon irradiation. Comparing our basal and irradiated screen results allowed us to identify an apparent synthetic lethal interaction between the poly(ADP-ribose) glycosylase PARG^33,34 and the cell cycle regulator of NHEJ, CYREN. In our basal screen, CYREN-PARG guide pairs were mildly depleted (−0.7 log₁₀ fold change [LFC]), but this phenotype was notably exacerbated in the IR screen (−1.7 LFC) (Figures 5A, 5B, and S5A). This suggests a dependence on CYREN and PARG after IR, as each gene is among the most depleted in the other’s gene list.

Figure 5. — (A) Volcano plots from basal and IR screens highlighting the change in gene-pair phenotypes for *CYREN-PARG* (red) and *cGAS-CYREN* (purple) between conditions.

(B) Observed phenotypes of gene-gene interactions from basal vs. IR screens for *CYREN*, *PARG*, and *cGAS*. The gene-gene *CYREN-PARG* phenotypes are marked in red, and that of cGAS-CYREN is marked in purple.

(C) Quantified log2FC confluence of colony formation assay performed with WT or *CYREN*-KO HT1080 cells with PARG inhibitor PDD00017273 (PDD) with or without 2 Gy ionizing radiation. Fold changes were normalized to unirradiated, DMSO-treated conditions for each cell line. Blue bars: predicted additive interactions. Data represent mean ± SEM from three biological replicates, and significance was determined by a two-way ANOVA with Dunnett multiple comparisons test; *p = 0.0116.

(D) The incidence of micronuclei (MNs) quantified as a percentage of total nuclei counted for WT or *CYREN*-KO HT1080 cells with either DMSO or 2 μM PDD, with or without 2 Gy ionizing radiation. Data represent mean ± SEM from three biological replicates, with >500 nuclei counted for each replicate. Statistics: ordinary one-way ANOVA with Šídák’s multiple comparisons test.

(E) Western blot from WT HT1080 cells showing doxycycline-inducible (+Dox) *CYREN* overexpression vectors (Cy1, full-length *CYREN*; Cy2, truncated *CYREN*) compared to an empty vector (EV) and uninduced populations.

(F) Change (Δ) in MN incidence ratios from PDD-/DMSO-treated coverslips from irradiated vs. unirradiated WT or CYREN-KO cells grown in Dox and harboring either EV or *CYREN* overexpression vectors shown in (E). Each circle represents a separate biological replicate with >500 nuclei counted per condition ± SEM. Statistics: a mixed-model two-way ANOVA with Šídák’s multiple comparisons.

(G) Western blot showing abrogated cGAS expression in an sgcGAS-RNP electroporated and clonally expanded HT1080-6TG clone compared to WT HT1080-6TG cells.

(H) Log₂ of the change in growth from shRNA-expressing WT or *cGAS*-KO cells from (G) with or without 6 Gy ionizing radiation. Data represent mean ± SEM from three biological replicates. Statistics: a one-way ANOVA. *p < 0.05 and **p < 0.01. ns, not significant.

See also Figure S5.

To validate this interaction, we treated WT HT1080 or previously generated CYREN-KO HT1080 cells³⁵ with increasing doses of an available PARG inhibitor, PDD00017273 (PDD; MedChemExpress), exposed them to 2 Gy IR, and performed colony formation assays. Although CYREN-KO cells were generally more sensitive to irradiation, we found that they displayed even greater sensitivity when also treated with PDD, well beyond what was seen with their WT counterparts (Figures 5C and S5B). Notably, the CYREN-KO cells’ sensitivity was greater than the additive effect of IR and PDD, suggesting a bona fide synthetic lethal interaction between CYREN and PARG. Similar results were obtained upon irradiation of U2-OS cells expressing short hairpin RNAs (shRNAs) targeting CYREN (Figures S5C and S5D).

A marker of global genomic instability is the occurrence of micronuclei (MNs).^36,37 To begin dissecting the mechanistic interplay between PARG inhibition and CYREN loss, we looked for MN formation upon radiation treatment. While previous studies reported a moderate increase in MN abundance in WT cells treated with PDD,³⁸ we observed no significant difference between DMSO- and PDD-treated WT cells in our system (Figures 5D, S5E, and S5F). Suggestive of inherent genomic instability in the absence of CYREN, CYREN-KO cells exhibited a higher baseline level of MNs than the WT, which was further elevated upon irradiation. Strikingly, although PDD treatment did not increase MN levels in irradiated WT cells or unirradiated CYREN-KO cells, it significantly elevated MN levels in irradiated CYREN-KO cells (Figures 5D and S5F).

CYREN is expressed as two protein isoforms: CYREN-1 is the full-length and main isoform, while CYREN-2, which is shorter, lacks the C-terminal domain that interacts with ATM and the MRN complex.^35,39 To determine whether both forms of CYREN are involved in the synthetic lethal interaction with PARG upon irradiation, we performed a complementation experiment and analyzed the formation of MNs as a proxy for genome instability. Notably, exogenous expression of CYREN-1 reduced the occurrence of MNs, while expression of the short form, CYREN-2, had no effect (Figures 5E, 5F, and S5G).

Since the precise roles of both CYREN and PARG at DSBs remain unclear, the mechanism underlying their genetic interaction in response to irradiation requires further investigation. Nonetheless, these results suggest that CYREN and PARG may cooperatively regulate DSB repair, leading to increased genome instability upon loss of CYREN and the accumulation of PAR chains.

Suppression of cGAS rescues CYREN-associated sensitivity to irradiation

The high level of MNs observed in CYREN-KO cells following irradiation suggested elevated genomic instability and accumulation of cytoplasmic DNA, which can trigger a cGAS-mediated innate immune response. Interestingly, CYREN and cGAS emerged as the top enriched gene KOs in each other’s interactions in the irradiated, but not the basal, screen (Figures 5A and 5B). This observation prompted us to test whether cGAS suppression could mitigate the irradiation-induced lethality associated with CYREN loss. We generated cGAS-KO HT1080-6TG clonal cells (Figure 5H), transduced them with shRNAs targeting CYREN,³⁵ and monitored cell growth following irradiation. Irradiation caused only minor growth defects in both WT and cGAS-KO cells expressing a control shRNA. As expected, shCYREN expression markedly sensitized WT cells to irradiation. In contrast, cGAS-KO cells expressing shCYREN showed no reduction in viability upon irradiation (Figures 5I and S5H). These results suggest that cGAS loss rescues the irradiation sensitivity of CYREN-deficient cells, possibly by preventing activation of the innate immune response triggered by the accumulation of cytoplasmic DNA after irradiation.

DISCUSSION

We performed a combinatorial genetic interaction screen to assess the genetic dependencies and contributions of over 100,000 gene combinations associated with the DDR to endogenous damage (basal) and upon recurring doses of IR. This study granted us insight into the complex network of DNA repair pathways. In all, the screen identified 29,571 total significant genetic interactions, of which we validated the observed fitness rescue of lethal genes and synthetic lethal interactions in both basal conditions and in response to irradiation.

Our analysis showed that combinatorial screens present unique technical difficulties that can complicate the interpretation of the raw sequencing data (Figures S2A–S2C). To address this issue, we excluded all sequences from our analysis that exhibited recombination between guide-1 and barcode-1, which are positioned at the edges of the dual-guide cassette. While this approach effectively eliminated recombination bias and allowed us to work with cleaner data, it also removed over 70% of our sequencing reads—likely diminishing the strength of certain interactions and possibly yielding false positive and false negative interactions. One potential solution would be to construct combinatorial libraries with shorter distances between the two gRNAs, as is possible when using Cas12 rather than Cas9. Additionally, other groups performing combinatorial screens have opted for inhibition rather than KO,^9,40 which may yield important and intriguing differences in the genetic interactions identified by each approach.

When examining known genetic interactions, we found several instances where our screen did not recover previously described relationships. This variation likely reflects the intrinsic noise of combinatorial CRISPR screens, as dual perturbations inherently increase both biological and technical variability. Comparing outcomes across independent screens will therefore be highly useful to distinguish bona fide genetic interactions from stochastic effects. However, some of these discrepancies may stem from differences between concurrent and sequential gene loss: while KO cells can adapt over time and develop compensatory dependencies, simultaneous loss may not elicit the same synthetic lethal effect. In this regard, concurrent gene depletion may not fully recapitulate how cancer cells acquire dependencies through the gradual accumulation of mutations in DDR and the following cellular adaptation to their loss. Nonetheless, the identified interactions provide critical groundwork to uncovering underlying basic mechanisms, reveal context-specific vulnerabilities, and inform the development of new therapeutic strategies based on drug combinations.

Among the depleted interactions in our basal screen, we selected the potential synthetic lethality between UBR5 and MRE11 for validation—an example of an interaction unlikely to have been identified via traditional single-gene screens. We confirmed that cells expressing sgUBR5 are highly sensitive to mirin, a potent inhibitor of MRE11’s exonuclease activity (Figures 3 and S3). Interestingly, other components of the MRN complex, NBS1 and RAD50, did not appear as depleted in combination with UBR5 guides in our screen (Figure S3A). While this may simply reflect false negative interactions, it could also suggest that MRE11’s exonuclease role, independent of the MRN complex,^41–43 is UBR5 mediated. It would be interesting to test whether UBR5 KO also sensitizes cells to inhibition of MRE11’s endonuclease activity. However, as the ubiquitin targets and broader roles of UBR5 in the DDR remain poorly understood, it is difficult to speculate on an exact mechanism for this interaction. Nonetheless, this finding exemplifies the strength of combinatorial screening in uncovering unexpected genetic interactions and opening new avenues for investigation.

Our basal screen also revealed enriched genetic interactions between the XRCC5-XRCC6 genes and telomere-associated factors, including the Apollo and XPF nucleases, as well as the CST complex component Stn1 (Figure 4). Given that suppression of the Ku heterodimer is lethal in human cells, these interactions suggested that the loss of these genes may rescue Ku’s associated lethality. We validated these interactions with an inducible dual-gRNA system, which yielded partial Ku depletion, leading to a mixed population of cells with zero, one, or two alleles knocked out. Conversely, suppression of XPF, Apollo, or Stn1 was unable to prevent cell death upon full KO of Ku, suggesting that suppression of these genes could alleviate Ku haploinsufficiency but not Ku loss. Ku is a highly abundant protein, and one could argue that partial suppression should leave cells with enough protein to sustain activity. However, previous studies found that loss of one allele of Ku results in a significant growth defect and a ~50% reduction in NHEJ.³² Thus, even a single-allele loss is sufficient to compromise Ku function, generating sublethal phenotypes.

When investigating the potential underlying mechanism of this fitness rescue, we found that partial suppression of Ku led to telomere free ends that were Apollo and XPF dependent (Figure 4H), suggesting that Ku restrains unscheduled resection by these nucleases to maintain telomere integrity.

The Apollo rescue is surprising considering that recent findings show that DNA-PKcs is required to recruit Apollo to telomeres.⁴⁴ Apollo then initiates resection of the leading-strand telomere end,^45,46 thus preventing leading-leading telomere fusions.⁴⁷ Notably, while these fusions are mediated by MMEJ,⁴⁴ they are also abolished by Ku suppression.⁴⁷ Therefore, DNA-PKcs and the Ku heterodimer might play antagonizing roles in regulating Apollo at telomeres to maintain the balance between end resection and end protection.

Meanwhile, the nuclease XPF is particularly efficient at cleaving 3′ single-stranded DNA overhangs^48,49 and has been shown to act on telomere overhangs in the absence of TRF2, enabling the NHEJ machinery to act at deprotected telomeres.⁵⁰ Together, these data and our observations suggest that Ku may play a protective role at TRF2-bound telomeres by shielding overhangs from aberrant XPF-mediated excision. This model could explain the telomere fusions that other groups observed upon Ku depletion,^51,52 as overhang loss would render telomeres vulnerable to inappropriate repair. Conversely, in the absence of both Ku and 3′ overhangs, telomeres might become substrates for other nucleases, leading to rapid telomere erosion and the accumulation of free ends. Alternatively, Ku could prevent XPF from acting on other types of telomere structures, such as internal loops (i-loops) or RNA-DNA hybrids (R-loops).^53,54

Nevertheless, the telomeric defect observed upon Ku loss is unlikely to explain Ku’s essentiality, as neither XPF nor Apollo depletion rescues Ku full KO. Instead, Ku was recently shown to prevent innate immune activation by RNAs generated from ALU repeats.³¹ We therefore propose that Ku’s protective role at telomeres coexists with fundamental, non-telomeric activities that are required for cell survival. While Ku’s essentiality complicates the direct dissection of its telomeric functions, partial-loss or haploinsufficient contexts may provide a useful framework to uncover how Ku promotes telomere end protection.

We also uncovered a synthetic lethal relationship between the cell-cycle-dependent regulator of NHEJ, CYREN, and the poly(ADP-ribose) glycohydrolase PARG in irradiated cells (Figures 5A, 5B, and S5A). Both gene products have distinct repair roles that could influence the response to DNA damage,^35,39,55 yet they have not previously been linked. Notably, the combined loss of CYREN and PARG proved significantly more detrimental following IR than would be predicted by an additive model (Figure 5C). Moreover, CYREN-depleted cells displayed increased MN formation—a hallmark of genomic instability—upon irradiation, further exacerbated by PARG inhibition. While PARG’s function at DNA DSBs remains incompletely understood, it removes PARylation chains to facilitate the disassembly of repair complexes.^55,56 We have previously shown that CYREN inhibits NHEJ during S/G2 at specific types of breaks to maintain genome stability following irradiation.³⁵ Conversely, during G0/G1, it was found to promote NHEJ, particularly in the absence of XLF.³⁹ Notably, although CYREN exists in two isoforms, only CYREN-1 was able to rescue the synthetic lethal interaction with PARG. Both CYREN-1 and CYREN-2 share the N-terminal DNA-PK-interacting domain, but CYREN-2 lacks the C-terminal region required for binding the MRN complex.³⁹ In addition, both isoforms can inhibit NHEJ during S/G2.³⁵ Thus, the function of CYREN that mediates its synthetic lethal interaction with PARG likely depends on its MRN-binding activity, but is independent of its S/G2-specific NHEJ inhibition. Further investigation into the mechanistic basis of this interaction is needed to clarify the context-specific regulation of DSB repair by CYREN and PARG.

Nonetheless, these findings underscore how genetic interactions uncovered in our screen can reveal previously unrecognized relationships and open entirely new avenues for mechanistic and therapeutic exploration.

Conclusion

Despite some limitations, our combinatorial screens offer a powerful resource for uncovering unanticipated gene-gene interactions and investigating their underlying mechanisms. While we have validated both enriched and depleted interactions in the basal screen, as well as genetic interactions that become lethal specifically in response to irradiation, these findings represent only a minuscule fraction (0.04% of the 29,571 significant interactions) of the total interaction landscape uncovered by our screens. We hope the resource provided by these data will enable the field to explore novel research avenues and deepen our understanding of the DDR to both endogenous stress and IR.

Limitations of the study

This study faced several limitations inherent to combinatorial CRISPR screening. The largest of these was the high rate of template switching and recombination between gRNA sequences and their corresponding barcodes, which required the removal of a significant amount of sequencing reads and may have weakened true interaction signals while also introducing false positives and false negatives. Dual-gene perturbations also increased biological and technical noise, which can result in discrepancies with previously reported genetic interactions and can complicate efforts to distinguish genuine dependencies from stochastic gene effects. Finally, gene essentiality, as with Ku loss, restricted the ability to interrogate complete gene loss. In these instances, the inhibition-based approaches used by other groups may reveal different interaction signatures that better capture the functional molecular contexts of essential genes.

RESOURCE AVAILABILITY

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Nausica Arnoult (nausica.arnoult@colorado.edu).

Materials availability

Materials used in this study are available upon reasonable request.

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Cell lines and cell culture

eHAP cells are human engineered haploid cells derived from HAP1, a near-haploid human cell line derived from the KBM-7 chronic myelogenous leukemia (CML) cell line. eHAP cells were purchased from Horizon Discovery (C669) and were not further authenticated. A clonal population expressing an inducible Cas9 was generated via viral transduction of an inducible Cas9 construct (pLenti-Blast-iCas9, modified from Addgene #87360 which was a gift from Adam Karpf) and subsequent clonal selection. pLentiBlast-iCas9 was made by removing the U6, tracR, and GFP elements of TLCV2 an dCas9 activity was scored using a GFP/BFP/sgGFP reporter¹⁹ (Addgene #67980) to assess leaky Cas9 expression and total activation of Cas9 upon doxycycline induction (Figures S1C, S1D, and S1E). Clone #6 had the lowest Cas9 activity without doxycycline and had robust Cas9 activity upon treatment, providing strong temporal control of screen initiation.

HT1080s are patient-derived fibrosarcoma human cells. HT1080 6TG cells are a derivative of HT1080 with a mutation in TP53. HT1080 were purchased from ATCC (#CCL-121), HT1080-6TG were obtained from the Karlseder lab. Both cell lines were authenticated by STR profiling and TP53 status was verified by sequencing. HT1080-6TG with nuclear fluorescence were generated by transducing WT HT1080-6TG cells with plasmids pLentiBlast_eGFP_3xNLS and pLentiPuro_RFP_3xNLS (a generous gift from Jim Lambert).²⁶ We then isolated clones and selected ones that displayed bright, uniform expression using an Echo Revolve (Cytena).

U2-OS are human cells derived from a differentiated bone sarcoma. U2-OS were commercially obtained (ATCC #HTB-96) and not further authenticated.

HCT116 are human cells derived from colon cancer. HCT116-XRCC5^F/− cells are a generous gift from the Hendrickson Lab²⁹ and were not further authenticated, except for the verification that tamoxifen addition induces rapid cell death.

Lenti-X 293T were derived from a transformed human embryonic kidney cell line, purchased from Takara (#632180), and used for less than 5 passages. They were not further authenticated.

All cell lines and their derivates were cultured in IMDM (eHAP) or DMEM (HT1080, HT1080-6TG, HCT116, U2-OS, and Lenti-X 293T) supplemented with 10% cosmic calf serum (Cytiva #SH30087.04), 1x Penicilin/Streptomycin/L-glutamine (Corning #30-009-Cl) and 1X MEM Non-essential Amino Acids (Corning #25-025CI) and maintained in a low oxygen condition (3% O2 and 7.5% CO2), and tested for mycoplasma.

METHOD DETAILS

DDR-DKO library cloning

Cloning the combinatorial DNA Damage Response Double-Knock Out (DDR-DKO) Library was inspired by methods described in Wong et al. 2016.¹⁶ Oligos containing guide RNA sequences, scaffolds, and barcodes were designed to target 461 DNA Repair genes (Listed in Table S1) matching GO:0006281. Guide sequences (Listed in Table S1) were pulled predominantly from the Brunello Genome Wide Library,¹³ unless they would introduce an undesired restriction site. In these cases, guide sequences were drawn from the Toronto Knock-Out v3 library (TKOv3).^14,15 Each guide was paired with a unique barcode as well. Furthermore, oligos destined for each site in the lentiviral vector (Figures 1B, S1A, and S1B) were associated with a unique tracR sequence¹⁷ (tracR-1 or tracR-2) to prevent recombination during bacterial cloning steps.

The cloning strategy is depicted in Figure S1B. The 1,905 oligos were synthesized as 8 separate pooled oligo libraries (oPools) by Integrated DNA Technologies (IDT). The first pool, oPool A, consisted of 948 oligos containing guide sequences 1 and 2 against each gene along with 25 non-targeting (NT) sequences. This cassette would end up in the first position on the vector. The remaining seven pools were broken into smaller B oPools, containing ~135 oligos apiece. These were named oPools B1-B7 and contained guide sequences 3 and 4 against each gene along with 4–8 NT sequences per pool, for 36 total NT guides spread across the seven B pools. These A and B pools were assembled separately before the guide and barcode containing cassettes from each B pool were cloned into the intervening site within the A pool locus using traditional restriction digest and ligation cloning. These seven unique A + B pools were mixed equimolarly according to the slightly different number of unique constructs they each contained. This resulted one large, final plasmid library with 907,236 guide pairs from 948 unique guide sequences in the first location and 957 unique sequences in the second location. Our cloning methods are described in detail below.

Amplification and purification of oligo pools

50ng of each oligo pool was individually amplified using New England Biolabs (NEB) Q5 NEXT II Ultra polymerase for 7 cycles with the ‘Oligo-FWD’ and ‘Oligo-REV’ primers (Key Resources Table) and purified using a DNA Clean and Concentrator Kit (Zymo Research, D4031). Yields were measured by High-Sensitivity (HS) dsDNA Qubit (Invitrogen, Q32851). To remove primer regions, the purified PCR products were then digested with Esp3I (NEB #R0734) in rCutSmart Buffer (NEB #B6004) at 37°C for one hour, and again purified with DNA Clean and Concentrator.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Ku70 (V540) Rabbit Antibody	Cell Signaling Technology	4104; RRID: AB_1904185
Ku80 (C48E7) Rabbit mAB	Cell Signaling Technology	2180; RRID: AB_2218736
spCas9 (7A9-3A3) Mouse mAb	Cell Signaling Technology	14697; RRID: AB_2750916
β-Actin (D6A8) Rabbit mAb	Cell Signaling Technology	8457; RRID: AB_10950489
UBR5 (D6O8Z) Rabbit mAb	Cell Signaling Technology	65344; RRID: AB_2799679
XPF (D3G8C) Rabbit mAb	Cell Signaling Technology	13465; RRID: AB_2798227
cGAS (D1D3G) Rabbit mAB	Cell Signaling Technology	15102; RRID: AB_2732795
Flag = = Dykddddk tag (93A) Mouse mAb	Cell Signaling Technology	8146; RRID: AB_10950495
anti-Mouse-HRP	Cell Signaling Technology	7076S; RRID: AB_330924
anti-Rabbit-HRP	Cell Signaling Technology	7074; RRID: AB_2099233
Bacterial and virus strains
NEB Stable	New England Biolabs	C3040
Lucigen Endura Electrocompetent E.coli	Biosearch Technologies Inc	602422
Chemicals, peptides, and recombinant proteins
Blasticidin	RPI	B12200–0.05
Polybrene	EMD millipore Coprporation	TR-1003-G
Brilliant Blue G	Sigma-Aldrich	B0770-SG
Carbenecillin	IBI Scientific 3P	IB02025
Colcemid	Gibco KaryoMAX	15212–012
Doxycycline	TCI	D4116
4-hydroxy Tamoxifen	Cayman Chemical Company	17308
Fluoroshield with DAPI	Sigma-Aldrich	F6057
HEPES	GIBCO	15630080
MEM Nonessential Amino Acids	Corning	25-025CI
Mirin	Selleck Chemical LLC	S809610MG
PDD00017273 (PARGi)	MedChemExpress	HY-108-360
Penicilin/Streptomycin	Lonza	09-757F
Puromycin	Alfa Aesar	61278-MC
NEBNext Ultra II Q5 Master Mix	NEB	M0544X
Apex Taq RED 2x Master Mix	Genesee Scientific	42–138
TelC-FAM	PNA Bio Inc.	F1001
CENPB-Cy3	PNA Bio Inc.	F3002
Triton X-	MP Biomedicals	9002-93-1
PEI Max	Polysciences Inc.	24765–1
HyClone Cosmic Calf Serum	Cytiva	SH30087.04
BbsI-HF	NEB	R3539
BamHI-HF	NEB	R3136
FastDigest BgIII	ThermoFischer	FD0083
EcoRI-HF	NEB	R3101
Esp3I	NEB	R0734
MfeI-HF	NEB	R3589
FastDigest MunI	ThermoFischer	FD0754
T4 Ligase	NEB	M0202
Critical commercial assays
BCA Protein Assay	ThermoFischer	23227
MacheryNagel NucleoBond Xtra	Machery-Nagel	740420.5
Midi EF Kit
Qubit High Sensitivity dsDNA Kit	Invitrogen	Q32851
QuiAMP Blood Maxi Kit	Quiagen	51194
High Sensitivity D1000 Reagents	Agilent	5067–5585
High Sensitivity D1000 ScreenTape	Agilent	5067–5584
DNA Clean and Concentrator 5-G	Zymo Research	D4013
DNA Clean and Concentrator 25-G	Zymo Research	D4033
DNA Clean and Concentrator 500-G	Zymo Research	D4031
Zymoclean Gel DNA Recovery Kit	Zymo Research	D4001
Super Signal West Dura	Genesee Scientific	20–301
Super Signal West Femto	Genesee Scientific	20–302
PEI Max	Polysciences Inc.	24765–1
Amersham Hybond P 0.2μm PVDF	Cytiva	10600021
Phosphatase inhibitor cocktail	Roche	5892791001
Bolt 4%–12% Bis-Tris Plus gels	Invitrogen	NW04120
Cas9 protein	IDT	1081060
Nucleofection Buffer	MaxCyte	EPB-1
Electropuration cuvette	MaxCyte	SOC-25x3
Gene Pulser Electroporation	BIO-RAD	1652089
Cuvettes 0.1cm
Deposited data
Raw sequencing files	This study	GEO GSE290153
Experimental models: Cell lines
eHAP	Horizon Discovery	C669
HT1080	ATCC	CCL-121
HT1080-6TG	gift from the Karlseder Lab
U2-OS	ATCC	HTB-96
HCT116-XRCC5 F/−	gift from the Hendrickson Lab²⁹
Lenti-X 293-T	Takara	632180
Oligonucleotides
Combinatorial Library Guides	IDT DNA	see Table S1
NT-1	IDT DNA	AAAAAGCTTCCGCCTGATGG
NT-2	IDT DNA	TTGCAGTTATTACCCTGTTA
sgNT-3	Synthego	GCACUACCAGAGCUAACUCA
sgEGFP	IDT DNA	GAAGTTCGAGGGCGACACCC
sgDCLRE1B-g1	IDT DNA	CCGAGGACTCAATACCACCC
sgDCLRE1B-g2	IDT DNA	GCCATTCCAACATGCTGCGT
sgDCLRE1B-a	Synthego	CCGAGGACUCAAUACCACCC
sgERCC4-g1	IDT DNA	CCATAACCCATCGCTTGAAG
sgERCC4-g2	IDT DNA	TGAAGTGGAGCCAAGATACG
sgERCC4-a	Synthego	CAUAACCCAUCGCUUGAAG
sgRNF8-g1	IDT DNA	GGCTCAAGGGCAGTCAACTT
sgRNF8-g2	IDT DNA	GCATCCAACTTGGAGTGCCT
sgSTN1-g1	IDT DNA	ATAAAGTGGACGACCCAGTG
sgSTN1-g2	IDT DNA	TCTTCACACCGGCTGGATCC
sgUBR5-g1	IDT DNA	CGGGTGAACCACGAAATGGA
sgUBR5-g2	IDT DNA	GTAGGACAAGCGACTCTCCA
sgUBR5-a	Synthego	CGGGUGAACCACGAAAUGGA
sgXRCC5-g1	IDT DNA	GGAGCCAATGGTCAGTCGGC
sgXRCC5-g2	IDT DNA	GAACTATATCACTTCCATAG
sgXRCC5-a	Synthego	AUCCAAAGCAUGAAUCAGGG
sgXRCC6-g1	IDT DNA	GCTAGAAGACCTGTTGCGGA
sgXRCC6-g2	IDT DNA	TGAGACCTCTTGGTATCGCT
sgcGAS-a	Synthego	GAACUUUCCCGCCUUAGGCA
Transfer PCR FWD	IDT DNA	CCCAAATCCAAAGACATTTCACG
Transfer PCR REV	IDT DNA	CCCTAAAGCTTGCCCTAC
Storage PCR FWD	IDT DNA	GAGGGCCTATTTCCCATGATT
Storage PCR REV	IDT DNA	TAACTCAATGCGGATCCAAAAAAGCACC
Oligo-FWD	IDT DNA	GAGGGTCCTGCCTTCATATCTG
Oligo-REV	IDT DNA	CAAGTATGTCAGCCTCATCGGG
PCR1 FWD	IDT DNA	CCTACACGACGCTCTTCCGATCTATGACGTCAGCGAAACAC
PCR1 REV	IDT DNA	TTCAGACGTGTGCTCTTCCGATCTCCCTAAAGCTTGCCCTAC
Illumina 502	IDT DNA	AATGATACGGCGACCACCGAGATCTACACGTTACAGCA CACTCTTTCCCTACACGACGCTCTTCCGATCT
Illumina 503	IDT DNA	AATGATACGGCGACCACCGAGATCTACACCGGTTCAAA CACTCTTTCCCTACACGACGCTCTTCCGATCT
Illumina 504	IDT DNA	AATGATACGGCGACCACCGAGATCTACACGCTGGATTA CACTCTTTCCCTACACGACGCTCTTCCGATCT
Illumina 505	IDT DNA	AATGATACGGCGACCACCGAGATCTACACAGGCGAAGA CACTCTTTCCCTACACGACGCTCTTCCGATCT
Illumina 506	IDT DNA	AATGATACGGCGACCACCGAGATCTACACTAATCTTAA CACTCTTTCCCTACACGACGCTCTTCCGATCT
Illumina 507	IDT DNA	AATGATACGGCGACCACCGAGATCTACACCAGGACGT ACACTCTTTCCCTACACGACGCTCTTCCGATCT
Illumina 508	IDT DNA	AATGATACGGCGACCACCGAGATCTACACGTACTGAC ACACTCTTTCCCTACACGACGCTCTTCCGATCT
Illumina 702	IDT DNA	CAAGCAGAAGACGGCATACGAGATCGATATCGGTGA CTGGAGTTCAGACGTGTGCTCTTCCGATCT
Illumina 703	IDT DNA	CAAGCAGAAGACGGCATACGAGATTAACTCGGGTGA CTGGAGTTCAGACGTGTGCTCTTCCGATCT
Illumina 704	IDT DNA	CAAGCAGAAGACGGCATACGAGATTAACAGTTGTGA CTGGAGTTCAGACGTGTGCTCTTCCGATCT
Illumina 705	IDT DNA	CAAGCAGAAGACGGCATACGAGATTTCTGAATGTGA CTGGAGTTCAGACGTGTGCTCTTCCGATCT
Illumina 706	IDT DNA	CAAGCAGAAGACGGCATACGAGATACGAATTCGTG ACTGGAGTTCAGACGTGTGCTCTTCCGATCT
Illumina 707	IDT DNA	CAAGCAGAAGACGGCATACGAGATAGCTTCAGGTGA CTGGAGTTCAGACGTGTGCTCTTCCGATCT
Illumina 708	IDT DNA	CAAGCAGAAGACGGCATACGAGATGCGCATTAGTGA CTGGAGTTCAGACGTGTGCTCTTCCGATCT
Recombinant DNA
pLentiPuro-Transfer	Guan et al.⁵⁷	Addgene #39481
pNB-Storage	Wong et al.¹⁶	Addgene #73850
pLentiBlast-iCas9	Barger et al.⁵⁸	Addgene #87360
pLentiPuro-DualGuide	This paper	This paper
pKLV2-U6gRNA5(gGFP)-PGKBFP2AGFP-W	Tzelepis et al.¹⁹	Addgene #67980
pLentiBlast_eGFP_3xNLS	Persenaire et al.²⁶	gift from the Lambert Lab
pLentiBlast_RFP_3xNLS	Persenaire et al.²⁶	gift from the Lambert Lab
pRSITUH_shLuciferase	Arnoult et al.³⁵	Cellecta
pRSITUH_shCyren-A	Arnoult et al.³⁵	Cellecta
pRSITUH_shCyren-A	Arnoult et al.³⁵	Cellecta
pLPT_EmptyVector	Arnoult et al.³⁵	–
pLPT_CYREN1-3xFLAG	Arnoult et al.³⁵	–
pLPT_CYREN2-3xFLAG	Arnoult et al.³⁵	–
Software and algorithms
GraphPad Prism9	GraphPad	9.0.1
ImageJ	NIH	–
Rstudio	open access	2022.07.0 + 548
SnapGene	Domatics	v7.0.2
FastQC	Babraham Bioinformatics	0.11.8
bbmap	open access	v38.05
HISAT2	open access	v2.1.0
Samtools	open access	v1.10
Code used in this study	This study	https://zenodo.org/records/17704799
Other
Agilent TapeStation 4150	Agilent	G2992A
BIO-RAD Gene Pulser Xcell with PC and CE Modules	BIORAD	1652660
Cytena CellCyteX	Discover Echo GmbH	–
Cytena Echo Revolve	Discover Echo GmbH	–
Cabinet X-ray System Model RX-650	Faxitron	RX-650
Nikon Widefield Eclipse Ti-E	Nikon	RRID: SCR_018302
Nunc Square BioAssay Dishes	Thermo Scientific	240835
NovaSeq X	Illumina	RRID: SCR_021984
G:Box Chemi XX6	Syngene	–
MaxCyte ATx system	MaxCyte	E-ATx
Biotec MACSQuant VYB Analyzer	Miltenyi	–

Open in a new tab

Cloning vectors

pLentiPuro-Transfer, which served for cloning of Pool A, was cloned by modifying the pLenti-puro (Addgene #39481, a gift from Ie-Ming Shih) by intropducing a human H1 (hH1) promoter and Esp3I and MfeI coning sites using synthesized gBlocks (IDT DNA) and InFusion (Takara Bio USA Inc. 63948). pNB-Storage, which was used for cloning the seven B Pools, was generated by modifying pAWp28 (Addgene #73850, a gift from Timothy Lu) to include a longer distance between BbsI and MfeI sites using synthesized gBlocks and InFusion as well.

Cloning of oligo pools into plasmids

16μg of pLentiPuro-Transfer plasmid was digested (1 h at 37° C) with Esp3I (NEB #R0734) and MfeI-HF (NEB #R3589) to provide compatible 5′-CACC-3′ and 5′-AATT-3′ ends, respectively, for orienting oligos in the pool appropriately. For each B pool, 8μg of pNB_Storage was digested 1 h at 37°C with BbsI-HF (NEB #R3539) and MfeI-HF (NEB #R3589), resulting in the same overhangs as above. Digested plasmids were then run in an agarose gel, gel extracted and purified using the Zymo Gel DNA Recovery Kit (D4001). The digested oligo pools and corresponding digested vector were ligated in a 3:1 oligo:vector molar ratio (2.17ng oligo +50ng vector) using T4 Ligase (NewEnglandBiolabs, M0202) in an overnight reaction at 16°C. Four ligation reactions were performed for the A pool. These ligation reactions were cleaned up using DNA Clean and Concentrator Kit and yields were again measured by HS dsDNA Quibit.

Transformation

Ligated pools were electroporated using Lucigen Endura Electrocompetent E. Coli (Biosearch Technologies Inc, #602422) as described in Sanjana et al. 2014.⁵⁹ In brief, 1mm cuvettes (Bio-Rad, 1652089) were loaded with 25μL electrocompetent cells and 2μL of cleaned ligation product from above, before electroporating using a BIO-RAD Gene Pulser Xcell (BIO-RAD, 1652660) with settings for 10μF, 600 Ohms, and 1800 Volts. To increase yields and maintain coverage, 4 electroporation reactions were performed for the A pool. Each electroporation reaction was then plated on a Nunc Square BioAssay (Thermo Scientific, 240835) LB + Carbenicillin culture plates and cultured overnight at 37°C. In tandem, a 1:20,000 dilution was grown to determine the final colony number for each electroporation and control that we obtained at least 80x colonies per unique oligo for each pool. After >16hr of growth, colonies from the large culture plates were collected and the libraries extracted using a Machery-Nagel NucleoBond Xtra Midi EF Kit (Machery-Nagel, cat#740420.50) per manufacturer’s instructions, and yields again quantified using High-Sensitivity dsDNA Qubit.

Quality control

To ensure a high abundance of appropriately cloned plasmids, we performed colony PCR on at least 50 colonies per pool from the dilution plate, using the Apex Taq RED polymerase (Genesee Scientific, 42–138) and primers ‘Transfer PCR FWD’/‘Transfer PCR REV’ for the A pool and ‘Storage PCR FWD’/‘Storage PCR REV’ for the B pools (Key Resources Table). PCR products were run on an agarose gel for visualization. Vectors that did not take up an oligo would show a ~1400bp band while those that did showed a ~500bp band. We ensured that at least 95% of colonies showed proper cloning.

Cloning of combinatorial library

We then cloned our ‘B’ cassettes from the storage vector to our transfer vector, thereby creating the combinatorial library. For each B pool, 8ug of cloned storage vector was digested with BglII and MunI (an isoschizomer of MfeI, ThermoFischer, FD0754) for one hour at 37°C. The reaction was run in an agarose gel and the hU6-B-oligo cassette (396bp) was gel extract and purified using the Zymo Gel DNA Recovery Kit (Zymo Research, D4001). 16μg of the cloned pool A-containing transfer vector was digested with BamHI-HF and EcoRI-HF (one hour at 37°C) and gel purified (7134bp, Zymo Gel DNA Recovery Kit). Purified cassette and vector concentrations were quantified by HS dsDNA Qubit. Each B-pool was then ligated with Pool A-containing vector (A + B1, A + B2, etc). Ligation was performed as described above with a 3:1 molar ratio of cassette to vector (8.8ng cassette, 50ng vector per 50μL reaction, eight reactions per B pool). Electro-transformation for the combined library was performed as above, except that eight transformation reactions were performed per combined pool, and plated on eight Nunc Square BioAssay (Thermo Scientific, 240835) LB + Carbenicillin culture plates and cultured overnight at 37°C. Using a dilution plate, we controlled that at least 76,000,000 colonies were grown per combined pool to maintain 80x coverage. The final library pools were collected as above and stored at −80°C until use.

As above, the final library pools were subjected to a colony PCR to qualitatively assess construction success. The same colony PCR was performed as for the A pool above with the expected band, representing both oligos appropriately inserted, now falling at 754 bp.

Library sequencing

Oligo abundances for each final pool (P1-7) of the DDR-DKO plasmid library was assessed by commercially sequencing the barcode region with Novogene. 90% of barcode-pairs were determined to fall within a 10-fold distribution for each pool (Figure S1C, vertical demarcations) which is on par with single-guide libraries.¹³ Each pool of the DDR-DKO library was finally combined equimolarly with respect to the number of sequences present in each.

Virus production

Viral production of the combinatorial library

For the large volume of virus required to infect one billion cells at an appropriate multiplicity of infection MOI, 5 million Lenti-X 293T cells were seeded in 25 × 15cm Ø plates with media supplemented with 10mM HEPES (GIBCO, #15630080) for buffering. The following day, each plate was transfected with packaging plasmids (5μg pCMV-VSV-G, 11μg pD8.9), 850μg of the DDR-DKO library constructed above, and with 73.5μg PEI MAX (Polysciences Inc., cat#24765-1). Media was refreshed 24 h later. The supernatant was collected a further 48 h later, vacuum filter sterilized (0.45μm), and flash frozen in LN2. This was repeated for two separate viral preparations to serve as biological replicates. An aliquot of each prep was thawed to titer transduction efficiency for each viral prep.

Other viral preps

500,000 Lenti-X 293T cells were seeded in one well of a 6-well plate. The next day, cells were transfected with packaging plasmids (0.5μg pCMV-VSV-G, 1.1μg pD8.9), 0.85μg transfer plasmid, 7.35μg of PEI MAX (Polysciences Inc. - Cat #: 24765-1) and 10mM HEPES. Media was refreshed 24h later with 3mL of complete DMEM. Virus-containing supernatant was collected 48h and 72h post transfection, cleared through a 0.45μm PES Filter membrane (Whatman Uniflo #9914-2504) and directly used to infect recipient cells in the presence of 5 μg.mL polybrene (EMD Millipore #TR-1003-G). Forty-eight hours after infection, cells were washed and selected with Puromycin (Alfa Aesar #J61278-MC, 1μg/ml for HT1080-6TG, U2-OS, and HCT116 or 0.25μg/ml for eHAPs) or Blasticidin (RPI #B12200–0.05, 10μg/ml for HT1080-6TG, U2-OS, and HCT116 or 2.5μg/ml for eHAPs). Selection was performed until complete death of control uninfected cells.

Ionizing radiation

When stated, cells were dosed with ionizing radiation using a Faxitron Cabinet X-ray System Model RX-650 (Lincolnshire, IL) at 115 kV and 5.33 rad/s. In all cases, cells were irradiated within 24 h of seeding or media changes.

Basal and ionizing radiation DDR-DKO screens

For each replicate, 2.3 billion eHAP-iCas9-cl6 cells were transduced with the DDR-DKO lentiviral library at an MOI of ~0.2. Infections were supplemented with 8μg/mL polybrene (EMD Millipore Corporation, cat#TR-1003-G) and seeded across 76 15cm tissue culture plates. Media was changed the following morning, and cells were allowed 24 h to recover. Cells were then treated with Puromycin (0.25μg/mL, Alfa Aesar, cat#61278-MC) selection for 48 h (Figure 1C). Post selection, at least 500M cells were collected for our day 0 (d0) sample to ensure coverage of our library and another 500M were re-seeded with Doxycycline (1μg/mL, TCI) and Blasticidin (10μg/mL, RPI cat# B12200–0.05) to activate and maintain Cas9 expression for each the Basal and Ionizing Radiation (IR) screens. For the basal screen, cells were split every 48 h. Cells were collected on day 14 (d14) and pelleted for genomic extraction. For the IR screen, cells were grown with Doxycycline and Blasticidin as above, but splits occurred every three days as plates were exposed to 0.6Gy ionizing radiation (Faxitron Cabinet X-ray System) 24 h after each split. These cells were kept in culture for 15 days to account for the slowed growth in response to radiation (Figure 1D), at which point cells reached at least 10PD and pellets were harvested for genomic DNA extraction. At each split and for all replicates of the screens cell counts were kept above 500M to maintain >500x coverage of the library.

Library amplification and sequencing

Genomic DNA (gDNA) was harvested from frozen 500M cell pellets for each sample using a QIAMP Blood Maxi Kit (Quiagen, #51194) per manufacturer’s instructions – 10 columns were used par sample.

The library was amplified in a way to minimize polymerase chain reaction (PCR) cycle number to avoid undesirable amplification biases. To introduce seed sequences for subsequent Illumina-primer amplification, each sample was amplified in a first PCR (PCR1) as follow: 1.8mg of DNA, 18mL of 2x NEBNext Ultra II Q5 polymerase (NEB, M0544), and 72μL of each PCR forward and reverse primers (primer table) at 100mM, in a total volume of 36mL. PCR reactions were then loaded as 100μL aliquots in qPCR plates and amplified for 12 cycles (2 min at 98°C followed by 12 cycles of 98°C 10 s, 65°C 75 s, followed by a final 2 min extension at 65°C). Completed reactions were re-pooled and 12mL of the product was then purified and concentrated using DNA Clean and Concentrator 500-G columns (Zymo Research, D4031). The concentrated PCR1 product was then run on an 0.6% agarose gel at 128V for 30 min. A ‘blind’ gel extraction was performed on each lane between ~400 and 1000bp to isolate the PCR1 fragment and remove most of the genomic DNA. These were then purified using four DNA Clean and Concentrator 25-G columns (Zymo Research, D4033) per sample. Each column was double eluted with 25μL nuclease free H₂O by centrifugation. Yields from the columns corresponding to each sample were pooled together and concentrations were again measured by HS dsDNA Qubit.

The second sequence PCR (PCR2) added Illumina adapter sequences. The entire purified PCR1 yield was used as a template for PCR2 and scaled such that each 100μL PCR2 reaction would not contain more than 1μg of PCR1 DNA. Each sample was also assigned a unique pair of Illumina Adapters (Key Resources Table). Thermocycler settings were the same as for PCR1 though PCR2 only underwent six amplification cycles to minimize adapter bias in sequencing.

PCR2 product was directly quantified by TapeStation (Agilent, TapeStation 4150) to confirm clean amplification of the entire guide-barcode region. Samples that only showed primer contaminants were purified using DNA Clean and Concentrator 5-G columns (and double-eluted with 15μL nuclease-free H2O. Some samples showed residual genomic DNA contamination; these were gel purified to isolate PCR2 product from genomic DNA and re-assessed by TapeStation for purity. Final sample concentrations were confirmed by HS dsDNA Qubit. Samples were then pooled at 10mM per sample and submitted to the University of Colorado, Anschutz Cancer Center Genomics Core (Cancer Center Support Grant P30CA046934) for sequencing.

Pairwise reads for 250 cycles covered guide-1 with read 1 and guide-2, barcode-2, and barcode-1 with read 2. Sequencing was performed on a NovaSeq X (RRID: SCR_021984) with our samples distributed across seven SP flow cells. On average, each sample produced 300M paired reads (Figure S2A).

Sequencing analysis

Overall quality of the sequenced library was assessed with FastQC (0.11.8). Raw reads were then trimmed to 20 base guide- and barcode-sequences using bbmap (v38.05) to remove flanking sequences. Specifically, guide-1 was trimmed from the forward read (read 1) in three steps with the following bbduk parameters.

Removing any read including the reverse sequencing primer (literal=GGGCAAGCTTTA, k=12, hdist=1)
Trimming sequence 5′ of Guide 1 (literal=ACGTCAGCGAAACACCG, ktrim=l, k=17, mink=8, hdist=2)
Trimming sequence 3′ of Guide 1 (literal=GTTTCAGAGCTA, ktrim=r, k=12, hdist=1, maq=10, minlen=20, maxlen=20).

Guide-2 was filtered similarly from the reverse read (read 2) except the initial step filtered out reads containing the forward sequencing primer. The three steps used the following parameters.

literal = TTTCGCTGACGT k = 12 hdist = 1 rcomp = f
literal = ATTTCTAGCTCTAAAAC ktrim = l k = 17 mink = 8 hdist = 2 rcomp = f
literal = CGGTGTTTCGTC ktrim = r k = 12 hdist = 1 maq = 10 rcomp = f minlen = 20 maxlen = 20

Barcodes were also isolated from read 2 with four trimming steps with the following bbduk parameters.

Trimming sequence 5′ of the barcodes (literal=GCCCTACTATCCAATTG, ktrim=l, k=17, mink=8, hdist=2)
Trimming sequence 3′ of the barcodes (literal=GAATTCCGTTGC, ktrim=r, k=12, hdist=1, maq=10, minlen=46, maxlen=46)
Isolating barcode-1 on the 5′ end of the spacer (literal=GAATTG, ktrim=l, k=6, hdist=0, maq=10, minlen=20, maxlen=20)
Isolating barcode-2 on the 3′ end of the spacer (literal=GAATTG, ktrim=r, k=6, hdist=0, maq=10, minlen=20, maxlen=20).

Each of the four sequences was then separately mapped to the expected guide-barcode sequences using HISAT2 (v2.1.0) with the following parameters: --ignore-quals --no-softclip --mp2,1 --score-min L,0,-0.4 --rdg 20,20 --no-unal --no-spliced-alignment. Guide-1 was mapped using the --norc parameter and guide-2 and barcode sequences using the --nofw parameter. Mapped data was pared from these SAM files to the most relevant information with Samtools (v1.10) using the command cut -f 1,3,12,13 ‘‘$samfile’’ | grep -v @. These were the files used for read matching.

Due to low read quality and low mapping rates at the Guide 2 position compared to the other sequences, we omitted Guide 2 from further analysis. We implemented custom Python scripts (Python v3.6.3) to match read IDs for mapped Guide 1/Barcode sequences. We discarded read pairs when any of the three sequences remained unmapped; otherwise we sorted sequences as ‘properly coupled’ or ‘decoupled’ based on whether Guide 1 correctly matched Barcode 1.

We combined all properly coupled sequences from all technical replicates and then counted guide pairs within each biological replicate. Biological replicate 3 had extremely low read depth at Day 0. Furthermore, at Day 14 the data for that replicate was largely unmapped, and the remainder of the data was strangely distributed with large counts for some guide pairs and very few for most guide pairs. We therefore dropped Replicate 3 from downstream analysis.

We calculated fold changes and statistics on guide pair counts with a custom Python script. All raw counts below 10 were discarded, then a pseudocount of 1 was added to all values. We then normalized counts to the totals of these filtered/pseudocount observations for each sample. We calculated fold changes as Day 14 counts/Day 0 counts, and we calculated observed guide-guide phenotypes as log2(fold change).

In order to determine single-guide and single-gene phenotypes, we took the following steps: We separated out all guides paired with nontargeting (nt) sequences by gene (4 guides per gene). Not all possible guide-nt pairs were represented due to some failing to appear or being filtered out in the initial processing step. We then compared the fold change for each guide-nt pair to the set of all guide-nt pair fold changes for each gene to determine outliers, and we omitted any pairs yielding p < 0.1 from the phenotype calculation. We averaged the fold changes for all remaining guide-nt pairs for each guide and for each gene to yield single-guide and single-gene phenotypes. We then calculated predicted interactions as products of fold changes for each guide in a guide-guide pair or gene in a gene-gene combinations and we reported these as log2(fold change product).

We calculated observed gene-gene fold changes by averaging the fold changes for all guide-guide pairs in a gene-gene combinations (typically eight, though initial filtering of lowly-observed guide-guide pairs or outlier guides based on single-guide phenotypes led to fewer in many cases). We then reported observed gene-gene phenotypes as log2(observed gene-gene fold change).

Before generating models to assess combinatorial gene effects, we correlated observed vs. predicted guide-guide and gene-gene phenotypes, expecting decent correlations on the assumption that most genes are likely not interacting with each other. However, almost all guides/genes showed Pearson’s or Spearman’s correlation coefficients of <0.2, likely due to wide variance of guide-nt phenotypes. We therefore calculated combinatorial effects only with observed gene-gene phenotypes rather than using a model incorporating single-gene phenotypes.

To determine potential combinatorial effects, we calculated p-values for each gene1-gene2 combination in relation to the distribution of all gene1-geneX pairs. Thus, while the observed phenotype was the same for any given gene-gene combination, the observed phenotypes and p-values depended on which gene was designated gene1. To reduce false positives, we combined these calculated p-values using Fisher’s method.

Dual-guide validation

Combinatorial KO vector cloning was achieved using the Golden Gate SAM technique⁶⁰ using pLentiPuro-DualGuide as a vector (Key Resources Table). pLentiPuro-DualGuide was constructed by modifying TLCV2 (Addgene #87360, a gift from Adam Karpf) to include a human H1 PolIII promoter (hH1), our tracR-2, and goldenGate cloning sites with a synthesized gBlock (IDT DNA) and the EGFP and other Esp3I, BamHI, and EcoRI restriction sites in the backbone were removed using site directed mutagenesis PCR. The plasmid features two guide RNA cloning sites, one with an H1 promoter, Esp3I cloning site, and tracR-1. The second contains a human U6 promoter, BbsI cloning site, and tracR-2.¹⁷ Two guides per gene were chosen via IDT’s CRISPR design tool to have strong on- and off-target scores. All combinatorial vectors and their intermediates were verified by Sanger sequencing (Quintara Biosciences). To verify the efficiency of KO from both loci, we cloned an sg-eGFP guide into either locus (sgEGFP-1 or sgEGFP-2) and transduced the vectors into HT1080-6TG nuclear-eGFP cells (Figure S3B). We then quantified the number of GFP+ cells over time and observed a reduction in GFP+ cells.

Other interactions were tested using commercially available drugs as listed and appropriate KO cell lines made previously or for this study (see below) (Key Resources Table).

Electroporation of Cas9-synthetic gRNAs to generate KO cells

UBR5 and cGAS were knocked out in HT1080-6TG and UBR5 was also knocked out in U2-OS cells using ribonucleoprotein (RNP) electroporation. Cas9 protein (IDT; 1081060) and synthetic sgRNA targeting respective genes (Synthego) were used to assemble the RNP complex. To prepare the complex, 2 μL of Cas9 (20 μM) was mixed with 5 μL of appropriate sgRNA (10 μM) and 6.4 μL nuclease-free water. The mixture was gently pipetted to ensure thorough mixing and incubated at room temperature for 20 min. Meanwhile, 1.5 × 10⁶ cells were collected and spun down at 500g for 3 min, then washed twice with 1X PBS. After the final wash, PBS was carefully aspirated from the cell pellet, which was then resuspended in 25 μL of nucleofection buffer (MaxCyte; EPB-1) and combined with the pre-assembled RNP complex, loaded into an OC-25x3 electroporation cuvette (MaxCyte; SOC-25x3) and transfected using the “Optimization 7” setting for HT1080-6TG and “U2OS” setting for U2-OS on the MaxCyte ATx system (MaxCyte; E-ATx). Following transfection, cells were transferred to a 10 cm² plate pre-warmed with 10 mL of supplemented DMEM and cultured under standard conditions. For HT1080-6TG lines, clones were then selected from these population-level knockouts using traditional clonal expansion procedures.

HCT116 XRCC5^F/− cells were electroporated with RNPs targeting the genes DCLRE1B, ERCC4, and STN1 (Figures S4E–S4G). RNP mixtures and HCT116 cells were prepared and treated as above and the MaxCyte setting “HCT116” was used for transfection.

Colony forming assay

To assess our dual-guide-validation vectors’ genetic contributions to growth, 200 post-selection cells were seeded per well in a 6-well plate and grown for 9 days either without (3 wells - uninduced) or with (3 wells – induced) doxycycline (Dox, 2μg/mL, TCI D4116) to induce Cas9 expression. The three wells served as technical replicates for induced and uninduced cells. At least three biological replicates were performed for each experiment. Predicted scores were calculated assuming no interaction between genetic combinations. To do this, we added the average Log₂FC for a one condition to the average Log₂FC of the other condition to predict the Log₂FC effect on cell growth without a genetic interaction. For example, the prediction boxes in Figure 3E take the average effect Mirin treatment had on NT-1 x NT-2 cells plus the average effect of each UBR5-guide individually in DMSO to predict the resulting phenotype. In Figure 4, these predictions were calculated per guide-pair and per replicate and plotted in the same manner as the measured Log₂FC’s.

UBR5-KO cells or control cells were seeded at 200 cells per well in 6-wells plates and treated with either dimethyl sulfoxide (DMSO) or an Mre11 inhibitor mirin (Selleck Chemical LLC, S809610MG) at doses listed.

For the CYREN and PARG genetic interaction validation, HT1080 CYREN-KO cells had been previously generated.³⁵ These or WT HT1080 cells were seeded at 200 cells per well in a 6-well plate and treated with either DMSO or a listed dose of PARG inhibitor (PDD00017273, MedChemExpress, HY-108-360) and/or a 2 Gray (Gy) of ionizing radiation.

For all clonogenic assays, nine days after seeding, wells were rinsed with 1x Phosphate Buffered Saline (PBS) and incubated for five minutes in −20°C methanol to fix colonies. Methanol was then removed and each well incubated with ~2mL of a mix of 50% v/v methanol and 0.5% m/v brilliant Blue G (SigmaAldrich, #B0770-SG) for another five minutes. Plates were carefully rinsed with water until residual dye was gone and dried overnight.

Individual wells were imaged using an iPhone 12 and analyzed using ImageJ (v1.53). In brief, images were color-thresholded for the blue dye, converted to a binary image, and subjected to hole filling and water-shedding algorithms to better identify individual colonies. The ‘Analyze Particles’ function was then used to count the number, size, and percent area the colonies inhabited (confluence) within each well. These were then further analyzed using Microsoft Excel (v.16.66.1) and visualized with GraphPad Prism (v9.0.1).

CellCyteX live-cell growth quantification

For live cell experiments to quantify cell growth, listed cells were seeded in 12-, or 48-well formats at either 30k or 3k per well, respectively, to ensure sufficient time for growth before the wells reach confluence. When appropriate, cells were seeded directly in Doxycycline (1μg/mL) or Tamoxifen (1μg/mL) containing media. When small molecules were added or samples needed to be, these occurred 24 h after seeding to allow cells time to adhere. Image quantification always began immediately after drug addition or IR treatment. Images were acquired using Enhanced Contour and CELLCYTE Studio was used to mask cells in images to quantify growth. When cells were treated with an shRNA, this was induced for 48 h prior to treatment and image acquisition.

Telomere FISH

Telomere FISH was performed as described previously.⁴ In brief, 6TGs were induced with doxycycline for four days before being synchronized in metaphase for three hours with 0.1μg/mL colcemid (Gibco KaryoMAX, #15212-012). Media and cells were collected, pelleted, and exposed to a hypotonic shock in 75mM KCl for 20 min at 37°C. Cells were fixed with a 3:1 methanol:glacial acetic acid (v/v) solution before metaphases were dropped only superfrost microscope slides and dried overnight. FISH was performed identically to before⁴ with the only modification being the inclusion of a centromere probe (CENPB-Cy3, PNA Bio Inc.) on top of the telomeric probe. Slides were mounted using Fluoroshield with DAPI (SigmaAldrich, #F6057) and imaged at 100× magnification on a Nikon Widefield Ti-E epi-fluorescence inverted microscope at the BioFrontiers Institute’s Advanced Light Microscopy Core (RRID: SCR_018302). The Nikon Ti-E Widefield is supported by NIH grant R01CA107098S1. Telomere free ends were counted manually using ImageJ software (v1.53).

Immunoblotting

Whole cell lysates were collected by trypsinizing cells, pelleting, and washing with PBS. Protein was extracted using RIPA buffer and protease and phosphatase inhibitor cocktail (Roche, 05892791001). Protein concentrations were measured using a bicinchoninic acid (BCA) assay (ThermoFisher, #23227). Equivalent amounts (5-30mg) of total protein for each sample were separated on Bolt 4–12% Bis-Tris Plus gels (Invotrogen, NW04120) before being transferred to PVDF membranes (Amersham Hybond P 0.2μm PVDF, Cytiva, 10600021). Membranes were blocked with 5% milk in TBS with 0.5% tween (TBS-T) for 1 h at room temperature and subsequently incubated overnight at 4°C with primary antibodies diluted in blocking medium. Membranes were then washed three times with TBS-T, incubated for one hour at room temperature with HRP-conjugated secondary antibodies diluted in 5% milk in TBS-T, and rinsed three more times in TBS-T before being detected using enhanced chemiluminescence (Super Signal West Dura- or Femto- Plus, Genesee Scientific, 20–301, 20–302). Chemiluminescence was imaged on a G:Box Chemi XX6 (Syngene).

Antibodies used: Cas9 (Cell Signaling, #14697S, diluted 1:1000), UBR5 (Cell Signaling #D608Z, diluted 1:1000), Ku70 (Cell Signaling, #4104S, diluted 1:1000), Ku80 (Cell Signaling, #2180T, diluted 1:1000), XPF (Cell Signaling, #13465S, diluted 1:1000), cGAS (Cell Signaling, #15102, diluted 1:1000), FLAG (Cell Signaling, #8146S, diluted 1:1000), Actin (Cell Signaling, #8457S, diluted 1:5000), secondary anti-Mouse-HRP (Cell Signaling, #70765, diluted 1:3000), and secondary anti-Rabbit-HRP (Cell Signaling, #7074P2, diluted 1:3000).

Micronuclei (MN) quantification

WT or CYREN-KO HT1080 cells were seeded on coverslips and allowed to settle. Sixteen hours before IR treatment, either DMSO or 2μM PDD00017273 (MedChemExpress, HY-108-360) was added. For Cyren rescue experiments, cells harboring inducible overexpression vectors were seeded with doxycycline for an additional 24hours before DMSO or PDD00017273 addition. Cells were then treated with 2Gy ionizing radiation and returned to normal growth conditions for another 24 h. Coverslips were then fixed using 3.7% formaldehyde and permeabilized with 0.5% Triton X-(MP Biomedicals, 9002-93-1) in PBS. They were then mounted to slides using Fluoroshield with DAPI (SigmaAldrich, F6057) and imaged using an Echo Revolve microscope (Cytena) at 20× magnification. At least 500 nuclei per condition were imaged per replicate. MN were quantified relative to nuclei abundance using ImageJ.

QUANTIFICATION AND STATISTICAL ANALYSIS

For each experiment, three biological replicates were performed unless otherwise denoted. Statistical analyses for each experiment are described in the corresponding figure legends or corresponding methods sections. Graphs plot the mean of the data, error bars represent the SEM, and p-values are defined in the corresponding figure legends or in the figures themselves.

Supplementary Material

NIHMS2142944-supplement-1.pdf^{(6.6MB, pdf)}

NIHMS2142944-supplement-2.xlsx^{(40.3KB, xlsx)}

NIHMS2142944-supplement-3.xlsx^{(4.4MB, xlsx)}

NIHMS2142944-supplement-4.xlsx^{(4.4MB, xlsx)}

Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2025.116850.

Highlights.

A combinatorial screen identifies DDR genetic interactions upon ionizing radiation
Combined loss of UBR5 and Mre11 is lethal
Loss of XPF, Apollo, or Stn1 rescues the growth defect associated with Ku haploinsufficiency
CYREN-KO hypersensitivity to IR is amplified by PARG inhibition but rescued by cGAS loss

ACKNOWLEDGMENTS

We thank the University of Colorado, Anschutz Medical Campus Cancer Center, for their Genomics Shared Resource (RRID: SCR_021984), supported by Cancer Center Support Grant P30CA046934. We also thank the MCDB Flow Cytometry Facility, as well as Joe Dravagon and the University of Colorado, Boulder BioFrontiers Institute’s Advanced Light Microscopy Core (RRID: SCR_018302), whose Nikon Ti-E Widefield microscope is supported by NIH grant R01CA107098S1. This work was funded by the NIH (R01CA266100 and R35GM143108 to N.A., AI156739 to R.D.D., T32GM142607 to B.N. and E.R.T., and T32GM008759 to B.N.), the NSF (GRFP DGE2040434 to E.R.T.), the Boettcher Foundation (AWD-21-02-0209 to N.A.), and the V Foundation for Cancer Research (V2021-031 to N.A.).

Footnotes

DECLARATION OF INTERESTS

The authors declare no conflicts of interest.

DECLARATION OF GENERATIVE AI AND AI-ASSISTED TECHNOLOGIES IN THE WRITING PROCESS

During the preparation of this work, the authors used ChatGPT4.0 in order to assist with enhancing the clarity and readability of certain sentences in the manuscript. After using this tool/service, the author reviewed and edited the content as needed and take full responsibility for the content of the publication.

Data and code availability

The list of gRNAs used in the screens, results of the basal screen, and results of the IR screen are available in Tables S1, S2, and S3. Raw sequencing files are available on GEO (GEO: GSE290153). The code is available on GitHub (https://zenodo.org/records/17704799).

REFERENCES

1.Dhoonmoon A, Schleicher EM, Clements KE, Nicolae CM, and Moldovan GL (2020). Genome-wide CRISPR synthetic lethality screen identifies a role for the ADP-ribosyltransferase PARP14 in DNA replication dynamics controlled by ATR. Nucleic Acids Res. 48, 7252–7264. 10.1093/nar/gkaa508. [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Srinivas US, Tay NSC, Jaynes P, Anbuselvan A, Ramachandran GK, Wardyn JD, Hoppe MM, Hoang PM, Peng Y, Lim S, et al. (2022). PLK1 inhibition selectively induces apoptosis in ARID1A deficient cells through uncoupling of oxygen consumption from ATP production. Oncogene 41, 1986–2002. 10.1038/s41388-022-02219-8. [DOI] [PubMed] [Google Scholar]
3.Adam S, Rossi SE, Moatti N, De Marco Zompit M, Xue Y, Ng TF, Álvarez-Quilón A, Desjardins J, Bhaskaran V, Martino G, et al. (2021). The CIP2A-TOPBP1 axis safeguards chromosome stability and is a synthetic lethal target for BRCA-mutated cancer. Nat. Cancer 2, 1357–1371. 10.1038/s43018-021-00266-w. [DOI] [PubMed] [Google Scholar]
4.Fleury H, MacEachern MK, Stiefel CM, Anand R, Sempeck C, Nebenfuehr B, Maurer-Alcalá K, Ball K Proctor B 3rd, Belan O, et al. (2023). The APE2 nuclease is essential for DNA double-strand break repair by microhomology-mediated end joining. Mol. Cell 83, 1429–1445.e8. 10.1016/j.molcel.2023.03.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Mengwasser KE, Adeyemi RO, Leng Y, Choi MY, Clairmont C, D’Andrea AD, and Elledge SJ (2019). Genetic Screens Reveal FEN1 and APEX2 as BRCA2 Synthetic Lethal Targets. Mol. Cell 73, 885–899.e6. 10.1016/j.molcel.2018.12.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Alvarez-Quilon A, Wojtaszek JL, Mathieu MC, Patel T, Appel CD, Hustedt N, Rossi SE, Wallace BD, Setiaputra D, Adam S, et al. (2020). Endogenous DNA 3’ Blocks Are Vulnerabilities for BRCA1 and BRCA2 Deficiency and Are Reversed by the APE2 Nuclease. Mol Cell 78, 1152–1165.e1158. 10.1016/j.molcel.2020.05.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Feng W, Simpson DA, Carvajal-Garcia J, Price BA, Kumar RJ, Mose LE, Wood RD, Rashid N, Purvis JE, Parker JS, et al. (2019). Genetic determinants of cellular addiction to DNA polymerase theta. Nat. Commun 10, 4286. 10.1038/s41467-019-12234-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Bazan Russo TD, Mujacic C, Di Giovanni E, Vitale MC, Ferrante Bannera C, Randazzo U, Contino S, Bono M, Gristina V, Galvano A, et al. (2024). Poltheta: emerging synthetic lethal partner in homologous recombination-deficient tumors. Cancer Gene Ther. 31, 1619–1631. 10.1038/s41417-024-00815-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Fielden J, Siegner SM, Gallagher DN, Schröder MS, Dello Stritto MR, Lam S, Kobel L, Schlapansky MF, Jackson SP, Cejka P, et al. (2025). Comprehensive interrogation of synthetic lethality in the DNA damage response. Nature 640, 1093–1102. 10.1038/s41586-025-08815-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Abdel-Wahab M, Gondhowiardjo SS, Rosa AA, Lievens Y, El-Haj N, Polo Rubio JA, Prajogi GB, Helgadottir H, Zubizarreta E, Meghzifene A, et al. (2021). Global Radiotherapy: Current Status and Future Directions-White Paper. JCO Glob. Oncol 7, 827–842. 10.1200/GO.21.00029. [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Baskar R, Yap SP, Chua KLM, and Itahana K (2012). The diverse and complex roles of radiation on cancer treatment: therapeutic target and genome maintenance. Am. J. Cancer Res 2, 372–382. [PMC free article] [PubMed] [Google Scholar]
12.Cannan WJ, and Pederson DS (2016). Mechanisms and Consequences of Double-Strand DNA Break Formation in Chromatin. J. Cell. Physiol 231, 3–14. 10.1002/jcp.25048. [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Doench JG, Fusi N, Sullender M, Hegde M, Vaimberg EW, Donovan KF, Smith I, Tothova Z, Wilen C, Orchard R, et al. (2016). Optimized sgRNA design to maximize activity and minimize off-target effects of CRISPR-Cas9. Nat. Biotechnol 34, 184–191. 10.1038/nbt.3437. [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Mair B, Tomic J, Masud SN, Tonge P, Weiss A, Usaj M, Tong AHY, Kwan JJ, Brown KR, Titus E, et al. (2019). Essential Gene Profiles for Human Pluripotent Stem Cells Identify Uncharacterized Genes and Substrate Dependencies. Cell Rep. 27, 599–615.e12. 10.1016/j.celrep.2019.02.041. [DOI] [PubMed] [Google Scholar]
15.Hart T, Tong AHY, Chan K, Van Leeuwen J, Seetharaman A, Aregger M, Chandrashekhar M, Hustedt N, Seth S, Noonan A, et al. (2017). Evaluation and Design of Genome-Wide CRISPR/SpCas9 Knockout Screens. G3 (Bethesda) 7, 2719–2727. 10.1534/g3.117.041277. [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Wong ASL, Choi GCG, Cui CH, Pregernig G, Milani P, Adam M, Perli SD, Kazer SW, Gaillard A, Hermann M, et al. (2016). Multiplexed barcoded CRISPR-Cas9 screening enabled by CombiGEM. Proc. Natl. Acad. Sci. USA 113, 2544–2549. 10.1073/pnas.1517883113. [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Dang Y, Jia G, Choi J, Ma H, Anaya E, Ye C, Shankar P, and Wu H (2015). Optimizing sgRNA structure to improve CRISPR-Cas9 knockout efficiency. Genome Biol. 16, 280. 10.1186/s13059-015-0846-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Hewitt G, Borel V, Segura-Bayona S, Takaki T, Ruis P, Bellelli R, Lehmann LC, Sommerova L, Vancevska A, Tomas-Loba A, et al. (2021). Defective ALC1 nucleosome remodeling confers PARPi sensitization and synthetic lethality with HRD. Mol. Cell 81, 767–783.e11. 10.1016/j.molcel.2020.12.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Tzelepis K, Koike-Yusa H, De Braekeleer E, Li Y, Metzakopian E, Dovey OM, Mupo A, Grinkevich V, Li M, Mazan M, et al. (2016). A CRISPR Dropout Screen Identifies Genetic Vulnerabilities and Therapeutic Targets in Acute Myeloid Leukemia. Cell Rep. 17, 1193–1205. 10.1016/j.celrep.2016.09.079. [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Sack LM, Davoli T, Xu Q, Li MZ, and Elledge SJ (2016). Sources of Error in Mammalian Genetic Screens. G3 (Bethesda) 6, 2781–2790. 10.1534/g3.116.030973. [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Hegde M, Strand C, Hanna RE, and Doench JG (2018). Uncoupling of sgRNAs from their associated barcodes during PCR amplification of combinatorial CRISPR screens. PLoS One 13, e0197547. 10.1371/journal.pone.0197547. [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Khatib JB, Dhoonmoon A, Moldovan GL, and Nicolae CM (2024). PARP10 promotes the repair of nascent strand DNA gaps through RAD18 mediated translesion synthesis. Nat. Commun 15, 6197. 10.1038/s41467-024-50429-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Menissier de Murcia J, Ricoul M, Tartier L, Niedergang C, Huber A, Dantzer F, Schreiber V, Ame JC, Dierich A, LeMeur M, et al. (2003). Functional interaction between PARP-1 and PARP-2 in chromosome stability and embryonic development in mouse. EMBO J. 22, 2255–2263. 10.1093/emboj/cdg206. [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Shibata A, Moiani D, Arvai AS, Perry J, Harding SM, Genois MM, Maity R, van Rossum-Fikkert S, Kertokalio A, Romoli F, et al. (2014). DNA double-strand break repair pathway choice is directed by distinct MRE11 nuclease activities. Mol. Cell 53, 7–18. 10.1016/j.molcel.2013.11.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Dupre A, Boyer-Chatenet L, Sattler RM, Modi AP, Lee JH, Nicolette ML, Kopelovich L, Jasin M, Baer R, Paull TT, and Gautier J (2008). A forward chemical genetic screen reveals an inhibitor of the Mre11-Rad50-Nbs1 complex. Nat. Chem. Biol 4, 119–125. 10.1038/nchembio.63. [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Persenaire C, Babbs B, Yamamoto TM, Nebbia M, Jordan KR, Adams S, Lambert JR, and Bitler BG (2024). VDX-111, a novel small molecule, induces necroptosis to inhibit ovarian cancer progression. Mol. Carcinog 63, 1248–1259. 10.1002/mc.23721. [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Ling S, and Lin WC (2011). EDD inhibits ATM-mediated phosphorylation of p53. J. Biol. Chem 286, 14972–14982. 10.1074/jbc.M110.182527. [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Zhang T, Cronshaw J, Kanu N, Snijders AP, and Behrens A (2014). UBR5-mediated ubiquitination of ATMIN is required for ionizing radiation-induced ATM signaling and function. Proc. Natl. Acad. Sci. USA 111, 12091–12096. 10.1073/pnas.1400230111. [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Li G, Nelsen C, and Hendrickson EA (2002). Ku86 is essential in human somatic cells. Proc. Natl. Acad. Sci. USA 99, 832–837. [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Wang Y, Ghosh G, and Hendrickson EA (2009). Ku86 represses lethal telomere deletion events in human somatic cells. Proc. Natl. Acad. Sci. USA 106, 12430–12435. 10.1073/pnas.0903362106. [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Zhu Y, Li A, Maji S, Lee BJ, Korn SM, Gertie JA, Dorrity TJ, Wang J, Wang KJ, Pelletier A, et al. (2025). Ku limits RNA-induced innate immunity to allow Alu expansion in primates. Nature 643, 562–571. 10.1038/s41586-025-09104-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Li J, Williams BL, Haire LF, Goldberg M, Wilker E, Durocher D, Yaffe MB, Jackson SP, and Smerdon SJ (2002). Structural and functional versatility of the FHA domain in DNA-damage signaling by the tumor suppressor kinase Chk2. Mol. Cell 9, 1045–1054. [DOI] [PubMed] [Google Scholar]
33.Slade D, Dunstan MS, Barkauskaite E, Weston R, Lafite P, Dixon N, Ahel M, Leys D, and Ahel I (2011). The structure and catalytic mechanism of a poly(ADP-ribose) glycohydrolase. Nature 477, 616–620. 10.1038/nature10404. [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Gupte R, Liu Z, and Kraus WL (2017). PARPs and ADP-ribosylation: recent advances linking molecular functions to biological outcomes. Genes Dev. 31, 101–126. 10.1101/gad.291518.116. [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Arnoult N, Correia A, Ma J, Merlo A, Garcia-Gomez S, Maric M, Tognetti M, Benner CW, Boulton SJ, Saghatelian A, and Karlseder J (2017). Regulation of DNA repair pathway choice in S and G2 phases by the NHEJ inhibitor CYREN. Nature 549, 548–552. 10.1038/nature24023. [DOI] [PMC free article] [PubMed] [Google Scholar]
36.Umbreit NT, Zhang CZ, Lynch LD, Blaine LJ, Cheng AM, Tourdot R, Sun L, Almubarak HF, Judge K, Mitchell TJ, et al. (2020). Mechanisms generating cancer genome complexity from a single cell division error. Science 368, eaba0712. 10.1126/science.aba0712. [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Fenech M, Kirsch-Volders M, Natarajan AT, Surralles J, Crott JW, Parry J, Norppa H, Eastmond DA, Tucker JD, and Thomas P (2011). Molecular mechanisms of micronucleus, nucleoplasmic bridge and nuclear bud formation in mammalian and human cells. Mutagenesis 26, 125–132. 10.1093/mutage/geq052. [DOI] [PubMed] [Google Scholar]
38.Gravells P, Neale J, Grant E, Nathubhai A, Smith KM, James DI, and Bryant HE (2018). Radiosensitization with an inhibitor of poly(ADP-ribose) glycohydrolase: A comparison with the PARP1/2/3 inhibitor olaparib. DNA Repair 61, 25–36. 10.1016/j.dnarep.2017.11.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Hung PJ, Johnson B, Chen BR, Byrum AK, Bredemeyer AL, Yewdell WT, Johnson TE, Lee BJ, Deivasigamani S, Hindi I, et al. (2018). MRI Is a DNA Damage Response Adaptor during Classical Non-ho-mologous End Joining. Mol. Cell 71, 332–342.e8. 10.1016/j.molcel.2018.06.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Simpson D, Ling J, Jing Y, and Adamson B (2023). Mapping the Genetic Interaction Network of PARP inhibitor Response. Preprint at bioRxiv. 10.1101/2023.08.19.553986. [DOI] [Google Scholar]
41.Paull TT, and Gellert M (1998). The 3’ to 5’ exonuclease activity of Mre 11 facilitates repair of DNA double-strand breaks. Mol. Cell 1, 969–979. [DOI] [PubMed] [Google Scholar]
42.Ceccaldi R, and Cejka P (2025). Mechanisms and regulation of DNA end resection in the maintenance of genome stability. Nat. Rev. Mol. Cell Biol 26, 586–599. 10.1038/s41580-025-00841-4. [DOI] [PubMed] [Google Scholar]
43.Trujillo KM, and Sung P (2001). DNA structure-specific nuclease activities in the Saccharomyces cerevisiae Rad50*Mre11 complex. J. Biol. Chem 276, 35458–35464. [DOI] [PubMed] [Google Scholar]
44.Sonmez C, Toia B, Eickhoff P, Matei AM, El Beyrouthy M, Wallner B, Douglas ME, de Lange T, and Lottersberger F (2024). DNA-PK controls Apollo’s access to leading-end telomeres. Nucleic Acids Res. 52, 4313–4327. 10.1093/nar/gkae105. [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Wu P, van Overbeek M, Rooney S, and de Lange T (2010). Apollo contributes to G overhang maintenance and protects leading-end telomeres. Mol. Cell 39, 606–617. 10.1016/j.molcel.2010.06.031. [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Wu P, Takai H, and de Lange T (2012). Telomeric 3’ Overhangs Derive from Resection by Exo1 and Apollo and Fill-In by POT1b-Associated CST. Cell 150, 39–52. 10.1016/j.cell.2012.05.026. [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Lam YC, Akhter S, Gu P, Ye J, Poulet A, Giraud-Panis MJ, Bailey SM, Gilson E, Legerski RJ, and Chang S (2010). SNMIB/Apollo protects leading-strand telomeres against NHEJ-mediated repair. EMBO J. 29, 2230–2241. 10.1038/emboj.2010.58. [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Mu D, Park CH, Matsunaga T, Hsu DS, Reardon JT, and Sancar A (1995). Reconstitution of human DNA repair excision nuclease in a highly defined system. J. Biol. Chem 270, 2415–2418. 10.1074/jbc.270.6.2415. [DOI] [PubMed] [Google Scholar]
49.de Laat WL, Appeldoorn E, Jaspers NG, and Hoeijmakers JH (1998). DNA structural elements required for ERCC1-XPF endonuclease activity. J. Biol. Chem 273, 7835–7842. 10.1074/jbc.273.14.7835. [DOI] [PubMed] [Google Scholar]
50.Zhu XD, Niedernhofer L, Kuster B, Mann M, Hoeijmakers JHJ, and de Lange T (2003). ERCC1/XPF removes the 3’ overhang from uncapped telomeres and represses formation of telomeric DNA-containing double minute chromosomes. Mol. Cell 12, 1489–1498. [DOI] [PubMed] [Google Scholar]
51.Samper E, Goytisolo FA, Slijepcevic P, van Buul PP, and Blasco MA (2000). Mammalian Ku86 protein prevents telomeric fusions independently of the length of TTAGGG repeats and the G-strand overhang. EMBO Rep. 1, 244–252. [DOI] [PMC free article] [PubMed] [Google Scholar]
52.Hsu HL, Gilley D, Galande SA, Hande MP, Allen B, Kim SH, Li GC, Campisi J, Kohwi-Shigematsu T, and Chen DJ (2000). Ku acts in a unique way at the mammalian telomere to prevent end joining. Genes Dev. 14, 2807–2812. [DOI] [PMC free article] [PubMed] [Google Scholar]
53.Mazzucco G, Huda A, Galli M, Piccini D, Giannattasio M, Pessina F, and Doksani Y (2020). Telomere damage induces internal loops that generate telomeric circles. Nat. Commun 11, 5297. 10.1038/s41467-020-19139-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
54.Sollier J, Stork CT, García-Rubio ML, Paulsen RD, Aguilera A, and Cimprich KA (2014). Transcription-coupled nucleotide excision repair factors promote R-loop-induced genome instability. Mol. Cell 56, 777–785. 10.1016/j.molcel.2014.10.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
55.Feng X, and Koh DW (2013). Roles of poly(ADP-ribose) glycohydrolase in DNA damage and apoptosis. Int. Rev. Cell Mol. Biol 304, 227–281. 10.1016/B978-0-12-407696-9.00005-1. [DOI] [PubMed] [Google Scholar]
56.Miwa M, and Sugimura T (1971). Splitting of the ribose-ribose linkage of poly(adenosine diphosphate-robose) by a calf thymus extract. J. Biol. Chem 246, 6362–6364. [PubMed] [Google Scholar]
57.Guan B, Wang TL, and Shih IM (2011). ARID1A, a factor that promotes formation of SWI/SNF-mediated chromatin remodeling, is a tumor suppressor in gynecologic cancers. Cancer Res. 71, 6718–6727. 10.1158/0008-5472.CAN-11-1562. [DOI] [PMC free article] [PubMed] [Google Scholar]
58.Barger CJ, Branick C, Chee L, and Karpf AR (2019). Pan-Cancer Analyses Reveal Genomic Features of FOXM1 Overexpression in Cancer. Cancers (Basel) 11, 251. 10.3390/cancers11020251. [DOI] [PMC free article] [PubMed] [Google Scholar]
59.Sanjana NE, Shalem O, and Zhang F (2014). Improved vectors and genome-wide libraries for CRISPR screening. Nat. Methods 11, 783–784. 10.1038/nmeth.3047. [DOI] [PMC free article] [PubMed] [Google Scholar]
60.Konermann S, Brigham MD, Trevino AE, Joung J, Abudayyeh OO, Barcena C, Hsu PD, Habib N, Gootenberg JS, Nishimasu H, et al. (2015). Genome-scale transcriptional activation by an engineered CRISPR-Cas9 complex. Nature 517, 583–588. 10.1038/nature14136. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

NIHMS2142944-supplement-1.pdf^{(6.6MB, pdf)}

NIHMS2142944-supplement-2.xlsx^{(40.3KB, xlsx)}

NIHMS2142944-supplement-3.xlsx^{(4.4MB, xlsx)}

NIHMS2142944-supplement-4.xlsx^{(4.4MB, xlsx)}

Data Availability Statement

[R1] 1.Dhoonmoon A, Schleicher EM, Clements KE, Nicolae CM, and Moldovan GL (2020). Genome-wide CRISPR synthetic lethality screen identifies a role for the ADP-ribosyltransferase PARP14 in DNA replication dynamics controlled by ATR. Nucleic Acids Res. 48, 7252–7264. 10.1093/nar/gkaa508. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R2] 2.Srinivas US, Tay NSC, Jaynes P, Anbuselvan A, Ramachandran GK, Wardyn JD, Hoppe MM, Hoang PM, Peng Y, Lim S, et al. (2022). PLK1 inhibition selectively induces apoptosis in ARID1A deficient cells through uncoupling of oxygen consumption from ATP production. Oncogene 41, 1986–2002. 10.1038/s41388-022-02219-8. [DOI] [PubMed] [Google Scholar]

[R3] 3.Adam S, Rossi SE, Moatti N, De Marco Zompit M, Xue Y, Ng TF, Álvarez-Quilón A, Desjardins J, Bhaskaran V, Martino G, et al. (2021). The CIP2A-TOPBP1 axis safeguards chromosome stability and is a synthetic lethal target for BRCA-mutated cancer. Nat. Cancer 2, 1357–1371. 10.1038/s43018-021-00266-w. [DOI] [PubMed] [Google Scholar]

[R4] 4.Fleury H, MacEachern MK, Stiefel CM, Anand R, Sempeck C, Nebenfuehr B, Maurer-Alcalá K, Ball K Proctor B 3rd, Belan O, et al. (2023). The APE2 nuclease is essential for DNA double-strand break repair by microhomology-mediated end joining. Mol. Cell 83, 1429–1445.e8. 10.1016/j.molcel.2023.03.017. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R5] 5.Mengwasser KE, Adeyemi RO, Leng Y, Choi MY, Clairmont C, D’Andrea AD, and Elledge SJ (2019). Genetic Screens Reveal FEN1 and APEX2 as BRCA2 Synthetic Lethal Targets. Mol. Cell 73, 885–899.e6. 10.1016/j.molcel.2018.12.008. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R6] 6.Alvarez-Quilon A, Wojtaszek JL, Mathieu MC, Patel T, Appel CD, Hustedt N, Rossi SE, Wallace BD, Setiaputra D, Adam S, et al. (2020). Endogenous DNA 3’ Blocks Are Vulnerabilities for BRCA1 and BRCA2 Deficiency and Are Reversed by the APE2 Nuclease. Mol Cell 78, 1152–1165.e1158. 10.1016/j.molcel.2020.05.021. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R7] 7.Feng W, Simpson DA, Carvajal-Garcia J, Price BA, Kumar RJ, Mose LE, Wood RD, Rashid N, Purvis JE, Parker JS, et al. (2019). Genetic determinants of cellular addiction to DNA polymerase theta. Nat. Commun 10, 4286. 10.1038/s41467-019-12234-1. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R8] 8.Bazan Russo TD, Mujacic C, Di Giovanni E, Vitale MC, Ferrante Bannera C, Randazzo U, Contino S, Bono M, Gristina V, Galvano A, et al. (2024). Poltheta: emerging synthetic lethal partner in homologous recombination-deficient tumors. Cancer Gene Ther. 31, 1619–1631. 10.1038/s41417-024-00815-2. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R9] 9.Fielden J, Siegner SM, Gallagher DN, Schröder MS, Dello Stritto MR, Lam S, Kobel L, Schlapansky MF, Jackson SP, Cejka P, et al. (2025). Comprehensive interrogation of synthetic lethality in the DNA damage response. Nature 640, 1093–1102. 10.1038/s41586-025-08815-4. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R10] 10.Abdel-Wahab M, Gondhowiardjo SS, Rosa AA, Lievens Y, El-Haj N, Polo Rubio JA, Prajogi GB, Helgadottir H, Zubizarreta E, Meghzifene A, et al. (2021). Global Radiotherapy: Current Status and Future Directions-White Paper. JCO Glob. Oncol 7, 827–842. 10.1200/GO.21.00029. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R11] 11.Baskar R, Yap SP, Chua KLM, and Itahana K (2012). The diverse and complex roles of radiation on cancer treatment: therapeutic target and genome maintenance. Am. J. Cancer Res 2, 372–382. [PMC free article] [PubMed] [Google Scholar]

[R12] 12.Cannan WJ, and Pederson DS (2016). Mechanisms and Consequences of Double-Strand DNA Break Formation in Chromatin. J. Cell. Physiol 231, 3–14. 10.1002/jcp.25048. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R13] 13.Doench JG, Fusi N, Sullender M, Hegde M, Vaimberg EW, Donovan KF, Smith I, Tothova Z, Wilen C, Orchard R, et al. (2016). Optimized sgRNA design to maximize activity and minimize off-target effects of CRISPR-Cas9. Nat. Biotechnol 34, 184–191. 10.1038/nbt.3437. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R14] 14.Mair B, Tomic J, Masud SN, Tonge P, Weiss A, Usaj M, Tong AHY, Kwan JJ, Brown KR, Titus E, et al. (2019). Essential Gene Profiles for Human Pluripotent Stem Cells Identify Uncharacterized Genes and Substrate Dependencies. Cell Rep. 27, 599–615.e12. 10.1016/j.celrep.2019.02.041. [DOI] [PubMed] [Google Scholar]

[R15] 15.Hart T, Tong AHY, Chan K, Van Leeuwen J, Seetharaman A, Aregger M, Chandrashekhar M, Hustedt N, Seth S, Noonan A, et al. (2017). Evaluation and Design of Genome-Wide CRISPR/SpCas9 Knockout Screens. G3 (Bethesda) 7, 2719–2727. 10.1534/g3.117.041277. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R16] 16.Wong ASL, Choi GCG, Cui CH, Pregernig G, Milani P, Adam M, Perli SD, Kazer SW, Gaillard A, Hermann M, et al. (2016). Multiplexed barcoded CRISPR-Cas9 screening enabled by CombiGEM. Proc. Natl. Acad. Sci. USA 113, 2544–2549. 10.1073/pnas.1517883113. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R17] 17.Dang Y, Jia G, Choi J, Ma H, Anaya E, Ye C, Shankar P, and Wu H (2015). Optimizing sgRNA structure to improve CRISPR-Cas9 knockout efficiency. Genome Biol. 16, 280. 10.1186/s13059-015-0846-3. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R18] 18.Hewitt G, Borel V, Segura-Bayona S, Takaki T, Ruis P, Bellelli R, Lehmann LC, Sommerova L, Vancevska A, Tomas-Loba A, et al. (2021). Defective ALC1 nucleosome remodeling confers PARPi sensitization and synthetic lethality with HRD. Mol. Cell 81, 767–783.e11. 10.1016/j.molcel.2020.12.006. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R19] 19.Tzelepis K, Koike-Yusa H, De Braekeleer E, Li Y, Metzakopian E, Dovey OM, Mupo A, Grinkevich V, Li M, Mazan M, et al. (2016). A CRISPR Dropout Screen Identifies Genetic Vulnerabilities and Therapeutic Targets in Acute Myeloid Leukemia. Cell Rep. 17, 1193–1205. 10.1016/j.celrep.2016.09.079. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R20] 20.Sack LM, Davoli T, Xu Q, Li MZ, and Elledge SJ (2016). Sources of Error in Mammalian Genetic Screens. G3 (Bethesda) 6, 2781–2790. 10.1534/g3.116.030973. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R21] 21.Hegde M, Strand C, Hanna RE, and Doench JG (2018). Uncoupling of sgRNAs from their associated barcodes during PCR amplification of combinatorial CRISPR screens. PLoS One 13, e0197547. 10.1371/journal.pone.0197547. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R22] 22.Khatib JB, Dhoonmoon A, Moldovan GL, and Nicolae CM (2024). PARP10 promotes the repair of nascent strand DNA gaps through RAD18 mediated translesion synthesis. Nat. Commun 15, 6197. 10.1038/s41467-024-50429-3. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R23] 23.Menissier de Murcia J, Ricoul M, Tartier L, Niedergang C, Huber A, Dantzer F, Schreiber V, Ame JC, Dierich A, LeMeur M, et al. (2003). Functional interaction between PARP-1 and PARP-2 in chromosome stability and embryonic development in mouse. EMBO J. 22, 2255–2263. 10.1093/emboj/cdg206. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R24] 24.Shibata A, Moiani D, Arvai AS, Perry J, Harding SM, Genois MM, Maity R, van Rossum-Fikkert S, Kertokalio A, Romoli F, et al. (2014). DNA double-strand break repair pathway choice is directed by distinct MRE11 nuclease activities. Mol. Cell 53, 7–18. 10.1016/j.molcel.2013.11.003. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R25] 25.Dupre A, Boyer-Chatenet L, Sattler RM, Modi AP, Lee JH, Nicolette ML, Kopelovich L, Jasin M, Baer R, Paull TT, and Gautier J (2008). A forward chemical genetic screen reveals an inhibitor of the Mre11-Rad50-Nbs1 complex. Nat. Chem. Biol 4, 119–125. 10.1038/nchembio.63. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R26] 26.Persenaire C, Babbs B, Yamamoto TM, Nebbia M, Jordan KR, Adams S, Lambert JR, and Bitler BG (2024). VDX-111, a novel small molecule, induces necroptosis to inhibit ovarian cancer progression. Mol. Carcinog 63, 1248–1259. 10.1002/mc.23721. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R27] 27.Ling S, and Lin WC (2011). EDD inhibits ATM-mediated phosphorylation of p53. J. Biol. Chem 286, 14972–14982. 10.1074/jbc.M110.182527. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R28] 28.Zhang T, Cronshaw J, Kanu N, Snijders AP, and Behrens A (2014). UBR5-mediated ubiquitination of ATMIN is required for ionizing radiation-induced ATM signaling and function. Proc. Natl. Acad. Sci. USA 111, 12091–12096. 10.1073/pnas.1400230111. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R29] 29.Li G, Nelsen C, and Hendrickson EA (2002). Ku86 is essential in human somatic cells. Proc. Natl. Acad. Sci. USA 99, 832–837. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R30] 30.Wang Y, Ghosh G, and Hendrickson EA (2009). Ku86 represses lethal telomere deletion events in human somatic cells. Proc. Natl. Acad. Sci. USA 106, 12430–12435. 10.1073/pnas.0903362106. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R31] 31.Zhu Y, Li A, Maji S, Lee BJ, Korn SM, Gertie JA, Dorrity TJ, Wang J, Wang KJ, Pelletier A, et al. (2025). Ku limits RNA-induced innate immunity to allow Alu expansion in primates. Nature 643, 562–571. 10.1038/s41586-025-09104-w. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R32] 32.Li J, Williams BL, Haire LF, Goldberg M, Wilker E, Durocher D, Yaffe MB, Jackson SP, and Smerdon SJ (2002). Structural and functional versatility of the FHA domain in DNA-damage signaling by the tumor suppressor kinase Chk2. Mol. Cell 9, 1045–1054. [DOI] [PubMed] [Google Scholar]

[R33] 33.Slade D, Dunstan MS, Barkauskaite E, Weston R, Lafite P, Dixon N, Ahel M, Leys D, and Ahel I (2011). The structure and catalytic mechanism of a poly(ADP-ribose) glycohydrolase. Nature 477, 616–620. 10.1038/nature10404. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R34] 34.Gupte R, Liu Z, and Kraus WL (2017). PARPs and ADP-ribosylation: recent advances linking molecular functions to biological outcomes. Genes Dev. 31, 101–126. 10.1101/gad.291518.116. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R35] 35.Arnoult N, Correia A, Ma J, Merlo A, Garcia-Gomez S, Maric M, Tognetti M, Benner CW, Boulton SJ, Saghatelian A, and Karlseder J (2017). Regulation of DNA repair pathway choice in S and G2 phases by the NHEJ inhibitor CYREN. Nature 549, 548–552. 10.1038/nature24023. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R36] 36.Umbreit NT, Zhang CZ, Lynch LD, Blaine LJ, Cheng AM, Tourdot R, Sun L, Almubarak HF, Judge K, Mitchell TJ, et al. (2020). Mechanisms generating cancer genome complexity from a single cell division error. Science 368, eaba0712. 10.1126/science.aba0712. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R37] 37.Fenech M, Kirsch-Volders M, Natarajan AT, Surralles J, Crott JW, Parry J, Norppa H, Eastmond DA, Tucker JD, and Thomas P (2011). Molecular mechanisms of micronucleus, nucleoplasmic bridge and nuclear bud formation in mammalian and human cells. Mutagenesis 26, 125–132. 10.1093/mutage/geq052. [DOI] [PubMed] [Google Scholar]

[R38] 38.Gravells P, Neale J, Grant E, Nathubhai A, Smith KM, James DI, and Bryant HE (2018). Radiosensitization with an inhibitor of poly(ADP-ribose) glycohydrolase: A comparison with the PARP1/2/3 inhibitor olaparib. DNA Repair 61, 25–36. 10.1016/j.dnarep.2017.11.004. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R39] 39.Hung PJ, Johnson B, Chen BR, Byrum AK, Bredemeyer AL, Yewdell WT, Johnson TE, Lee BJ, Deivasigamani S, Hindi I, et al. (2018). MRI Is a DNA Damage Response Adaptor during Classical Non-ho-mologous End Joining. Mol. Cell 71, 332–342.e8. 10.1016/j.molcel.2018.06.018. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R40] 40.Simpson D, Ling J, Jing Y, and Adamson B (2023). Mapping the Genetic Interaction Network of PARP inhibitor Response. Preprint at bioRxiv. 10.1101/2023.08.19.553986. [DOI] [Google Scholar]

[R41] 41.Paull TT, and Gellert M (1998). The 3’ to 5’ exonuclease activity of Mre 11 facilitates repair of DNA double-strand breaks. Mol. Cell 1, 969–979. [DOI] [PubMed] [Google Scholar]

[R42] 42.Ceccaldi R, and Cejka P (2025). Mechanisms and regulation of DNA end resection in the maintenance of genome stability. Nat. Rev. Mol. Cell Biol 26, 586–599. 10.1038/s41580-025-00841-4. [DOI] [PubMed] [Google Scholar]

[R43] 43.Trujillo KM, and Sung P (2001). DNA structure-specific nuclease activities in the Saccharomyces cerevisiae Rad50*Mre11 complex. J. Biol. Chem 276, 35458–35464. [DOI] [PubMed] [Google Scholar]

[R44] 44.Sonmez C, Toia B, Eickhoff P, Matei AM, El Beyrouthy M, Wallner B, Douglas ME, de Lange T, and Lottersberger F (2024). DNA-PK controls Apollo’s access to leading-end telomeres. Nucleic Acids Res. 52, 4313–4327. 10.1093/nar/gkae105. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R45] 45.Wu P, van Overbeek M, Rooney S, and de Lange T (2010). Apollo contributes to G overhang maintenance and protects leading-end telomeres. Mol. Cell 39, 606–617. 10.1016/j.molcel.2010.06.031. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R46] 46.Wu P, Takai H, and de Lange T (2012). Telomeric 3’ Overhangs Derive from Resection by Exo1 and Apollo and Fill-In by POT1b-Associated CST. Cell 150, 39–52. 10.1016/j.cell.2012.05.026. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R47] 47.Lam YC, Akhter S, Gu P, Ye J, Poulet A, Giraud-Panis MJ, Bailey SM, Gilson E, Legerski RJ, and Chang S (2010). SNMIB/Apollo protects leading-strand telomeres against NHEJ-mediated repair. EMBO J. 29, 2230–2241. 10.1038/emboj.2010.58. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R48] 48.Mu D, Park CH, Matsunaga T, Hsu DS, Reardon JT, and Sancar A (1995). Reconstitution of human DNA repair excision nuclease in a highly defined system. J. Biol. Chem 270, 2415–2418. 10.1074/jbc.270.6.2415. [DOI] [PubMed] [Google Scholar]

[R49] 49.de Laat WL, Appeldoorn E, Jaspers NG, and Hoeijmakers JH (1998). DNA structural elements required for ERCC1-XPF endonuclease activity. J. Biol. Chem 273, 7835–7842. 10.1074/jbc.273.14.7835. [DOI] [PubMed] [Google Scholar]

[R50] 50.Zhu XD, Niedernhofer L, Kuster B, Mann M, Hoeijmakers JHJ, and de Lange T (2003). ERCC1/XPF removes the 3’ overhang from uncapped telomeres and represses formation of telomeric DNA-containing double minute chromosomes. Mol. Cell 12, 1489–1498. [DOI] [PubMed] [Google Scholar]

[R51] 51.Samper E, Goytisolo FA, Slijepcevic P, van Buul PP, and Blasco MA (2000). Mammalian Ku86 protein prevents telomeric fusions independently of the length of TTAGGG repeats and the G-strand overhang. EMBO Rep. 1, 244–252. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R52] 52.Hsu HL, Gilley D, Galande SA, Hande MP, Allen B, Kim SH, Li GC, Campisi J, Kohwi-Shigematsu T, and Chen DJ (2000). Ku acts in a unique way at the mammalian telomere to prevent end joining. Genes Dev. 14, 2807–2812. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R53] 53.Mazzucco G, Huda A, Galli M, Piccini D, Giannattasio M, Pessina F, and Doksani Y (2020). Telomere damage induces internal loops that generate telomeric circles. Nat. Commun 11, 5297. 10.1038/s41467-020-19139-4. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R54] 54.Sollier J, Stork CT, García-Rubio ML, Paulsen RD, Aguilera A, and Cimprich KA (2014). Transcription-coupled nucleotide excision repair factors promote R-loop-induced genome instability. Mol. Cell 56, 777–785. 10.1016/j.molcel.2014.10.020. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R55] 55.Feng X, and Koh DW (2013). Roles of poly(ADP-ribose) glycohydrolase in DNA damage and apoptosis. Int. Rev. Cell Mol. Biol 304, 227–281. 10.1016/B978-0-12-407696-9.00005-1. [DOI] [PubMed] [Google Scholar]

[R56] 56.Miwa M, and Sugimura T (1971). Splitting of the ribose-ribose linkage of poly(adenosine diphosphate-robose) by a calf thymus extract. J. Biol. Chem 246, 6362–6364. [PubMed] [Google Scholar]

[R57] 57.Guan B, Wang TL, and Shih IM (2011). ARID1A, a factor that promotes formation of SWI/SNF-mediated chromatin remodeling, is a tumor suppressor in gynecologic cancers. Cancer Res. 71, 6718–6727. 10.1158/0008-5472.CAN-11-1562. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R58] 58.Barger CJ, Branick C, Chee L, and Karpf AR (2019). Pan-Cancer Analyses Reveal Genomic Features of FOXM1 Overexpression in Cancer. Cancers (Basel) 11, 251. 10.3390/cancers11020251. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R59] 59.Sanjana NE, Shalem O, and Zhang F (2014). Improved vectors and genome-wide libraries for CRISPR screening. Nat. Methods 11, 783–784. 10.1038/nmeth.3047. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R60] 60.Konermann S, Brigham MD, Trevino AE, Joung J, Abudayyeh OO, Barcena C, Hsu PD, Habib N, Gootenberg JS, Nishimasu H, et al. (2015). Genome-scale transcriptional activation by an engineered CRISPR-Cas9 complex. Nature 517, 583–588. 10.1038/nature14136. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Uncovering genetic interactions in the DNA repair network in response to endogenous damage and ionizing radiation

Benjamin Nebenfuehr

Lynn Sanford

Erin R Taylor

Kerri Ball

Catherine E Woods-Killam

Hannah I Ghasemi

Bruce Proctor III

Raquel Ortega

Colin Sempeck

Robin D Dowell

Nausica Arnoult

SUMMARY

In brief

Graphical Abstract

INTRODUCTION

RESULTS

Construction of a combinatorial gRNA library to target the DNA repair network

Figure 1. A combinatorial CRISPR-KO screen of DDR genes under basal conditions and with ionizing radiation.

Basal and irradiation screens

Sequencing and analysis of the combinatorial screens

Figure 2. Combinatorial genetic interactions differentially contribute to fitness under basal or IR conditions.

Screen results overview

The E3 ubiquitin ligase UBR5 is synthetic lethal with MRE11

Figure 3. MRE11A and UBR5 display a synthetic lethal genetic interaction.

Suppression of Apollo and XPF rescues cellular fitness upon partial Ku70/80 loss

Figure 4. Suppression of Apollo and XPF rescues cellular fitness upon partial Ku70/80 loss.

Regulatory interplay between CYREN and PARG

Figure 5. Loss of CYREN and PARG is synthetic lethal upon ionizing radiation.

Suppression of cGAS rescues CYREN-associated sensitivity to irradiation

DISCUSSION

Conclusion

Limitations of the study

RESOURCE AVAILABILITY

Lead contact

Materials availability

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Cell lines and cell culture

METHOD DETAILS

DDR-DKO library cloning

Amplification and purification of oligo pools

Cloning vectors

Cloning of oligo pools into plasmids

Transformation

Quality control

Cloning of combinatorial library

Library sequencing

Virus production

Viral production of the combinatorial library

Other viral preps

Ionizing radiation

Basal and ionizing radiation DDR-DKO screens

Library amplification and sequencing

Sequencing analysis

Dual-guide validation

Electroporation of Cas9-synthetic gRNAs to generate KO cells

Colony forming assay

CellCyteX live-cell growth quantification

Telomere FISH

Immunoblotting

Micronuclei (MN) quantification

QUANTIFICATION AND STATISTICAL ANALYSIS

Supplementary Material

Highlights.

ACKNOWLEDGMENTS

Footnotes

Data and code availability

REFERENCES

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases