Abstract Although platinum salts or PARP inhibitors are effective in delivering anti-tumor responses in people with BRCA1 or BRCA2 mutated cancers, drug resistance is common and is often caused by secondary BRCA1/2 reversion mutations that restore function. By collating and analyzing 848 BRCA1/2 reversion mutations in 384 cancer patients with drug resistance, we confirm that pathogenic BRCA1/2 mutation type influences the acquisition of reversions, and that large BRCA1/2 deletions are an underappreciated form of reversion. Integrating reversion data with systematic CRISPR-Cas9 screens that delete BRCA1/2 exons, we also show that both proteins contain privileged domains whose structure is essential for drug resistance, including the PALB2 interacting domains of both BRCA1 and BRCA2. Reversions in PALB2 also conserve both BRCA1 and BRCA2 binding domains. Surprisingly, exon 11 of BRCA2, which encodes BRC repeats 1-8, is not essential for resistance. Using this patient and functional information, we estimate the likelihood of pathogenic BRCA2 mutations to revert. We show that risk of reversion correlates with both the presence of clinical reversions and the response to treatment, suggesting that the propensity to revert could be a useful clinical parameter.

Although Rho kinase (ROCK) has been studied in tumor progression, the reliance of some cancer cells on ROCK-Myosin II for survival remains poorly understood. Using systematic analysis of ROCK inhibitor sensitivity in hundreds of cancer cell lines, we find that ROCK inhibition reduces survival of highly de-differentiated, invasive cancer cells from solid tumors. Transcriptomic analysis reveals enrichment in epithelial-to-mesenchymal transition, migration, proliferation, and inflammation genes, with reduced expression of differentiation and cell-cell junction genes like E-cadherin (CDH1). Acute myeloid leukemia (AML) shows high ROCK inhibitor response among hematological malignancies. Using in vitro and in vivo approaches, we validate biomarkers of ROCK inhibitor sensitivity in breast cancer, melanoma, and AML, demonstrating their unique addiction to Rho-ROCK-myosin II signaling for survival. Our work has important pre-clinical implications while cautions against wider use of ROCK inhibitors in patient-derived organoid cultures, where they may deplete important cancer cell populations.

Supplementary File containing methods, figures and uncropped membranes.

SUMMARY Homology-directed repair deficiency (HRd) encompasses mutations in multiple genes yet is treated clinically as a single entity. Here, through parallel analyses of isogenic knockouts of multiple HR pathway genes, integrating multi-omic analyses with genome-wide CRISPR-Cas9-dependency and resistance screens, we show that HRd is not a single entity but exists along a molecular and functional continuum. BRCA1, BRCA2, PALB2, RAD51C, and RAD51D mutants shared many HRd-associated mutational signatures, while RAD51B, BRIP1 , CDK12 exhibited distinct genomic patterns. Functional heterogeneity was equally apparent: synthetic lethal interactions including CIP2A and a novel dependency on PRDX1 were penetrant across most HRd genotypes, whereas FANCM dependency was linked to HRd subtypes characterized by tandem duplications. PARPi resistance screens in distinct HRd contexts uncovered BRIP1 and RECQL5 as new BRCA2 -specific resistance genes. HRd is thus a complex continuum, underscoring why modernizing the molecular taxonomy utilizing all genomic features available per patient is crucial to informing precision interventions.

Invasive lobular carcinoma (ILC) accounts for 15% of breast cancers yet lacks specific therapy because ILCs are underrepresented in clinical trials and preclinical models are lacking. In this study, we established intraductal xenograft models to test whether the clinical pan-lysyl-oxidase inhibitor PXS-5505, now in phase I/IIa trials for myelofibrosis, can exploit the collagen-rich matrix dependency in ILC created by CDH1 loss. PXS-5505 remodeled fibrillar collagen and halted tumor expansion and metastatic seeding across estrogen receptor-positive and triple-negative models without systemic toxicity. Genome-wide CRISPR screens revealed ITGAV and ITGB5 as synthetic lethal partners of CDH1, and LOX inhibition downregulated their expression, together with MYC, NF-κB, and AP-1 transcriptional programs. Collagen fiber density/alignment and MYC/AP-1 gene signatures served as pharmacodynamic readouts of drug activity. These data uncover a tractable extracellular matrix-integrin-MYC axis in ILC and nominate PXS-5505, alone or with endocrine therapy, for window-of-opportunity trials in this understudied breast cancer subtype. SIGNIFICANCE:A clinical stage LOX inhibitor slows tumor progression and alters multiple molecular endpoints in invasive lobular carcinoma, providing a translatable therapeutic strategy for this frequent breast cancer subtype currently lacking specific therapies.

Cancer cells lacking BRCA1 or BRCA2 must adapt to survive and proliferate despite defective DNA repair and high genomic instability. Restoration of homologous recombination (HR) through loss of DNA end protection rescues growth defects and promotes PARP inhibitor (PARPi) resistance in BRCA1-deficient cells; however, the genetic basis of adaptation to BRCA2 loss remains largely unexplored. To delineate BRCA1- and BRCA2-specific adaptation trajectories, we established a fully isogenic screening platform in mouse embryonic stem cells engineered for acute depletion of either protein. This approach uncovered hundreds of genes with shared, distinct, or even opposing effects between the two BRCA-deficiency states. We identified FANCM and its interacting partners RMI1/2, CENPS/MHF1, CENPX/MHF2, and FAAP24 as essential in BRCA1-deficient cells but toxic in BRCA2- deficient contexts. Loss of FANCM in BRCA2-deficient cells alleviated genomic instability and proliferation defects by preventing RMI1/2-dependent replication fork degradation, without rescuing HR. Moreover, we show that the FANCM-RMI1/2 complex drives PARPi sensitivity in both mouse and human BRCA2-deficient cells, in contrast to the 53BP1-SHLD-CST axis in BRCA1-deficient cells. These findings reveal distinct adaptation routes to BRCA1 and BRCA2 loss and establish FANCM as a determinant of BRCA2-specific vulnerability and therapeutic response, with direct implications for tailoring prevention and therapy according to BRCA mutation status. ### Competing Interest Statement C.J.L. makes the following disclosures: receives and/or has received research funding from: AstraZeneca, Merck KGaA, Artios, Neophore, FoRx. Received consultancy, SAB membership or honoraria payments from: FoRx, Syncona, Sun Pharma, Gerson Lehrman Group, Merck KGaA, Vertex, AstraZeneca, Tango Therapeutics, 3rd Rock, Ono Pharma, Artios, Abingworth, Tesselate, Dark Blue Therapeutics, Pontifax, Astex, Neophore, Glaxo Smith Kline, Dawn Bioventures, Blacksmith Medicines, ForEx, Ariceum. Has stock in: Tango, Ovibio, Hysplex, Tesselate, Ariceum. C.J.L. is also a named inventor on patents describing the use of DNA repair inhibitors and stands to gain from their development and use as part of the ICR Rewards to Inventors scheme and also reports benefits from this scheme associated with patents for PARP inhibitors paid into CJL's personal account and research accounts at the Institute of Cancer Research. 020

Proteogenomic analyses have identified an association between LIG1 (DNA Ligase I) loss and chemotherapy resistance in a subset of triple negative breast cancer (TNBC) enriched for TP53 mutations. Here, we demonstrate that co-occurrence of TP53 mutations and LIG1 loss is associated with upregulated DDR activity, including homologous recombination, likely contributing to reduced platinum sensitivity. Unbiased genetic and monotherapy drug screens identified PARP inhibitors (PARPi) as a potential treatment for LIG1-depleted tumors; however, the increase in sensitivity was modest and lower than that observed in TNBC models with homologous recombination deficiency. Subsequently, a screen of PARP inhibition in combination with each of 120 clinically relevant DDR inhibitors revealed that PARPi sensitivity in LIG1-loss cells was significantly enhanced by the addition of an ATR inhibitor (ATRi). Olaparib and ceralasertib demonstrated synergistic cytotoxicity in LIG1-loss cell line models; the combination significantly reduced tumor volume in a LIG1-low PDX model compared to either monotherapy, and showed greater ex vivo cytotoxicity in a LIG1-low PDXO model versus a LIG1-high control. Hence, this study highlights LIG1 status as a stratification factor for ongoing and future clinical trials of DDR-targeted combinations in TNBCs.

Invasive lobular carcinoma (ILC) accounts for 15% of breast cancers yet lacks specific therapy because ILCs are underrepresented in clinical trials and preclinical models are lacking. We established intraductal xenograft models to test whether the clinical pan-lysyl-oxidase PXS-5505, now in phase trials for myelofibrosis can exploit the collagen-rich matrix dependency created by CDH1 loss. PXS-5505 remodels fibrillar collagen, and halts tumor expansion and metastatic seeding across ER+ and triple negative models without systemic toxicity. Genome-wide CRISPR screens reveal ITGAV and ITGB5 as synthetic lethal partners of CDH1 and LOX inhibition downregulates their expression together with MYC, NF-κB, and AP-1 transcriptional programmes. Collagen fibre density/alignment, and MYC/AP-1 gene signatures serve as pharmacodynamic readouts of drug activity. These data uncover a tractable ECM-integrin-MYC axis in ILC and nominate PXS-5505, alone or with endocrine therapy, for window of opportunity trials in this understudied breast cancer subtype. One Sentence Summary Targeting matrix remodelling in ILC inhibits ILC progression and alters multiple molecular endpoints, providing a translatable therapeutic strategy for this understudied subtype that requires better treatments. ### Competing Interest Statement The authors have declared no competing interest.

PARPi and ATRi have synergistic cytotoxic effects in models of DSRCT with high PARP1 expression. A and B, PARP1 expression (A) and PARylation levels (B) as assessed by IHC in a cohort of 16 DSRCT samples, compared with those of the JN1 and R cell lines (PARP1 and PAR expression levels are shown as H-scores). Representative cases (PARP1-high vs. PARP1-low tumors; PAR-high vs. PAR-low tumors) are shown to the right, compared with JN1 and R cells. C and D, Surface plots of Bliss independence scores calculated for the talazoparib–M4344 combination in JN1 (C) and R (D) cell lines at 7 days. E, The GR_13-PDX-O model was established from the primary peritoneal tumor of a patient with DSRCT, with confirmation of EWSR1::WT1 fusion by FISH and WT1-Cter IHC (Supplementary Fig. S8). F, Surface plot of Bliss independence scores calculated for the talazoparib–M4344 combination in the GR_13 PDX-O at 7 days. Mean ± SD; n = 3. Surface plots: the x-axis and y-axis values indicate drug concentrations, and the z-axis values indicate the associated synergy score; score < −10, antagonistic interaction; score = 0, absence of interaction; score > 10, synergistic interaction. G, Schematic illustration of an in vivo therapeutic experiment performed to evaluate the antitumor effect of PARPi talazoparib and ATRi M1774 in NSG mice engrafted with JN1 xenografts. H, Therapeutic responses to drug treatment in mice harboring JN1 xenografts. Mean tumor volume ± SD; two-way ANOVA and post hoc Dunnett test. I, Tumor volume at the time of mice sacrifice. Mean ± SD; one-way ANOVA and post hoc Šídák test. *, P < 0.01; ns, not significant. Tala, talazoparib.

Silencing of SMG1 resensitizes SMG8 and SMG9 knockout cells to ATR inhibition. A, Left, SMG1 was overexpressed in the SMG8 and SMG9 mutant cells after 48 hours of DMSO or 150 nmol/L berzosertib exposure compared with the WT cells. Right, Smg1 signal intensity relative to β-actin expression representing relative protein expression using ImageJ. Error bars represent SEM, considering three biological replicates. B and C, SMG1 knockdown experiment (384-well plates) showing an increase in the normalized percent of inhibition in the SMG8 and SMG9 mutant cells (red) compared with the WT cells (blue) following exposure to a range of SMG1 siRNA concentrations (from 1 to 20 nmol/L; negative controls, siCON1, siCON2; positive controls, siATM, siPLK1). D, Western blot analysis showing siSMG1 silencing in the YCC6 WT and SMG8 and SMG9 mutant cells. E–G, Silencing of SMG1 in YCC6 WT, SMG8, and SMG9 mutant cells resensitizes them to ATRi (384-well plate, 5-day assay). siCON2 was used as a negative control. All panels of this figure are representative of three or more biological replicates. ns, nonsignificant.

Gastric cancer genome-wide CRISPRn screen identifies genetic determinants of ATRi resistance. A, YCC6 and SNU5 gastric tumor cell lines show no ARID1A expression by Western blotting. Other ARID1A-proficient gastric tumor cell lines are represented in this Western blot analysis and HCT116 ARID1A isogenic cell lines were used as positive and negative controls. B and C, Dose–response survival curves (384-well plate, 5-day assay) show that YCC6 and SNU5 (blue) are sensitive to ATRi, compared with the ARID1A-proficient gastric cancer cell lines (black). HCT116 isogenic controls are highlighted in red. D, Schematic illustrating a genome-wide ATRi CRISPR/Cas9 screen using the YCC6 gastric cancer cell line. E, Western blot analysis showing that YCC6 gastric tumor cell line (clone 9) expresses Cas9 upon doxycycline induction after being transduced with an Edit-R Inducible Lentiviral hEF1a-Blast-Cas9 Nuclease vector and selected with blasticidin. F, YCC6 iCas9 cells have a catalytically active Cas9 as shown by flow cytometry. GFP/RFP/Empty represent iCas9 cells transduced with GFP- and RFP-expressing lentiviral constructs. GFP/RFP/gfp-sgRNA represents iCas9 cells that were additionally transduced with a sgRNA-targeting GFP that cleaves the GFP protein thereby decreasing green fluorescence emission. G, Scatter plot illustrating sgRNA z-score for ATRi-resistant cells retrieved at T1 compared with untreated cells retrieved at T0 plotted against the rank calculated from the rank product of z-score and MAGeCK analysis of sgRNA counts. The genes targeted by sgRNA that were most enriched (T1-T0) are highlighted in red at the top right corner of the graph comprising CDC25B, SMG8, SMG9, HUWE1, IRF9, HNRNPF, and CARD10. H, Diagram showing the validation screen workflow. I, Results of the deconvoluted CRISPRn validation screen in 96-well plate format. Each red dot represents an individual sgRNA. P values were calculated by conducting a t test comparing all sgRNAs per gene versus all sgRNAs per the control sgRNA. Dose–response curves, Western blot analysis, and the deconvoluted CRISPRn validation screen in 96-well plate format are representative of three or more biological replicates. ns, nonsignificant.

SMG8 and SMG9 deficiency causes resistance to ATR inhibition. A, YCC6 SMG8 mutant clones show lower levels of SMG8 protein expression compared with SMG8 WT YCC6 cells by Western blotting. B, YCC6 SMG9 mutant clones show no SMG9 protein expression compared with SMG9 WT YCC6 cells by immunoprecipitation. C–F, SMG8 and SMG9 mutant clones are resistant to ATRi (384-well plate, 5-day assay) compared with the WT cells. G, Western blot analysis showing doxycycline-inducible overexpression of HA-tagged SMG8 cDNA in SMG8 Mut 1 clone. Doxycycline-inducible GFP cDNA overexpression served as a negative control. H and I, ATRi dose–response survival curves (384-well plate, 5-day assay) illustrating a resensitization to berzosertib and AZD6738 in the SMG8 Mut 1 clone when SMG8 overexpression was induced by doxycycline exposure (dotted line, compared with the continuous line). J, SMG8 (blue) and SMG9 (red) mutant HAP1 cells are resistant to ATRi (384-well plate, 5-day assay). All panels of this figure are representative of three or more biological replicates.

To maintain cell fitness, deleterious genetic alterations are buffered by compensatory changes in additional genes. In cancer, buffering processes could be targeted by synthetic lethality. However, despite the large-scale identification of synthetic lethal effects in preclinical models, evidence that these operate clinically is limited. This impedes the application of synthetic lethal approaches. By integrating molecular profiling data from >9,000 cancers with synthetic lethal screens, we show that transcriptomic buffering of tumor suppressor gene (TSG) loss by hyperexpression of synthetic lethal partners is a common phenomenon, extending to multiple TSGs and histotypes. Transcriptomic buffering is also notable in cancers that phenocopy TSG loss, such as BRCAness cancers, where expression of BRCA1/2 synthetic lethal genes correlates with clinical outcome. Synthetic lethal genes that exhibit transcriptomic buffering also represent more robust synthetic lethal effects. These observations have implications for understanding how tumor cells tolerate TSG loss, in part explain transcriptomic architectures in cancer and provide insight into target selection.

Genomic defects caused by truncating mutations or deletions in the Retinoblastoma tumor suppressor gene (RB1) are frequently observed in many cancer types leading to dysregulation of the RB pathway. Here, we propose an integrative proteogenomic approach that predicts cancers with dysregulation in the RB pathway. A subset of these cancers, which we term as "RBness," lack RB1 genomic defects and yet phenocopy the transcriptional profile of RB1-defective cancers. We report RBness as a pan-cancer phenomenon, associated with patient outcome and chemotherapy response in multiple cancer types, and predictive of CDK4/6 inhibitor response in estrogen-positive breast cancer. Using RNA interference and a CRISPR-Cas9 screen in isogenic models, we find that RBness cancers also phenocopy synthetic lethal vulnerabilities of cells with RB1 genomic defects. In summary, our findings suggest that dysregulation of the RB pathway in cancers lacking RB1 genomic defects provides a molecular rationale for how these cancers could be treated.