LIG1 Is a Synthetic Lethal Target in BRCA1 Mutant Cancers

Abstract

Synthetic lethality approaches in BRCA1/2-mutated cancers have focused on PARP inhibitors, which are subject to high rates of innate or acquired resistance in patients. In this study, we used CRISPR/Cas9-based screening to identify DNA ligase I (LIG1) as a novel target for synthetic lethality in BRCA1-mutated cancers. Publicly available data supported LIG1 hyperdependence of BRCA1 mutant cells across a variety of breast and ovarian cancer cell lines. We used CRISPRn, CRISPRi, RNAi, and protein degradation to confirm the lethal effect of LIG1 inactivation at the DNA, RNA, and protein level in BRCA1 mutant cells in vitro. LIG1 inactivation resulted in viability loss across multiple BRCA1-mutated cell lines, whereas no effect was observed in BRCA1/2 wild-type cell lines, demonstrating target selectivity for the BRCA1 mutant context. On-target nature of the phenotype was demonstrated through rescue of viability with exogenous wild-type LIG1 cDNA. Next, we demonstrated a concentration-dependent relationship of LIG1 protein expression and BRCA1 mutant cell viability using a titratable, degradable LIG1 fusion protein. BRCA1 mutant viability required LIG1 catalytic activity, as catalytically dead mutant LIG1^K568A failed to rescue viability loss caused by endogenous LIG1 depletion. LIG1 perturbation produced proportional increases in PAR staining in BRCA1 mutant cells, indicating a mechanism consistent with the function of LIG1 in sealing ssDNA nicks. Finally, we confirmed LIG1 hyperdependence in vivo using a xenograft model in which LIG1 loss resulted in tumor stasis in all mice. Our cumulative findings demonstrate that LIG1 is a promising synthetic lethal target for development in patients with BRCA1-mutant cancers.

Introduction

Inactivation of the BRCA1 gene is responsible for a large percentage of inherited predispositions to breast and ovarian cancers in women, with associated lifetime risk estimates of 80% and 40% to 65%, respectively (1). Loss or mutation of BRCA1 has also been associated with colon and prostate cancers in men (2). In December 2014, the FDA approved the first PARP inhibitor for patients with BRCA1/2-mutated ovarian cancer, providing clinical proof-of-concept for synthetic lethality as an anticancer strategy. Synthetic lethality describes a gene interaction in which mutation or disruption of one of the interacting genes is compatible with cell viability, but perturbation of both genes produces lethality (3, 4). This concept was first reported in Drosophila melanogaster (5) and later articulated in an oncological context by Hartwell and colleagues (6) in the late 1990s. PARP enzymes are involved in base-excision repair of ssDNA breaks as well as nucleotide excision repair and the regulation of end-joining repair of double-strand DNA (dsDNA) breaks (7). In patients with homologous recombination deficiency (HRD), including those with germline BRCA (BRCA1 and/or BRCA2) mutations or with nongermline HRD-positive tumors, PARP inhibition is associated with the accumulation of dsDNA breaks and subsequently cell death due to intolerable replication stress at multiple difficult-to-replicate loci (8). Discovery of the synthetic lethal interaction between PARP and BRCA spurred the clinical development of PARP inhibitors as a novel class of anticancer agents and opened new gateways for oncological drug development (9, 10).

To date, several PARP inhibitors have been used successfully as synthetic lethal anticancer agents in the treatment of patients with HRD tumors, specifically, BRCA1- and BRCA2-mutated ovarian, breast, and pancreatic cancers (11), with ongoing clinical trials in chemoresistant germline or somatic BRCA1/2-mutated breast, ovarian, lung, and pancreatic cancers. However, use of these agents, especially for long-term maintenance therapy, is considerably limited by inherent and acquired resistance: approximately half of patients with BRCA1/2 mutations exhibit pre-existing intrinsic resistance to PARP inhibitors, and more than two-thirds of patients receiving PARP inhibitor therapy eventually develop acquired resistance (12–14). Moreover, the available spectrum of PARP inhibitors is associated with variable tolerability, necessitating dose reduction in some patients (15). Alongside strategies seeking to resensitize patients to treatment, there is a vested interest in identifying other targets with efficacy in BRCA1/2 mutant cancers that can provide good efficacy and tolerability as stand-alone monotherapy targets, act in concert with PARP inhibitors to improve efficacy, or provide an alternative for patients with intrinsic or acquired resistance to PARP inhibitors.

Molecular target identification for BRCA1/2-mutated cancers has been greatly aided by recent advances in CRISPR/Cas9 technology (16). Beyond PARP1/2, studies have characterized synthetic lethal interactions of BRCA1/2 with the gene product of SIRT1/6, wherein loss of sirtuin activity leads to PARP1/HPF1-mediated serine ADP-ribosylation and loss of viability (17), as well as the gene product of POLQ, wherein loss of Polθ activity is selectively lethal in cells deficient for homologous recombination (HR; 18–21). We previously used a CRISPR/Cas9 screen to identify USP1 as a target for synthetic lethality with BRCA1/2 and demonstrated that perturbation of USP1 in BRCA1/2 mutant cells is associated with decreased DNA synthesis and accumulation of S phase–specific DNA damage, likely mediated through accumulation and aberrant processing of mono- and poly-ubiquitinated proliferating cell nuclear antigen (PCNA; ref. 22).

In this study, we characterize a second gene hit identified in our CRISPR/Cas9 screen, DNA ligase 1 (LIG1), as a target for synthetic lethality in BRCA1-mutant cancers. The ATP-dependent DNA ligase LIG1 is encoded by the LIG1 gene and functions in DNA replication and recombination as well as base excision repair (23), and specifically the processing of Okazaki fragments (24). Our in vitro and in vivo genetic studies validate LIG1 as a target for oncological development and suggest that synthetic lethality with BRCA1 is dependent on LIG1 catalytic activity, wherein PAR accumulation leads to genomic instability and subsequently cell death.

Materials and Methods

Information for compounds, antibodies, siRNAs, short hairpin RNA (shRNA), and single-guide RNA (sgRNA) sequences, cDNA sequences, and cell lines used in this study are listed in Supplementary Tables S1–S5.

Cloning of sgRNA and cDNA constructs

A lentiviral system was used for both CRISPR-Cas9 and doxycycline inducible CRISPR-dCas9-KRAB cloning. All guide and cDNA sequences are reported in Supplementary Table S4. Wild-type (WT) and mutant cDNA constructs were cloned using gBlocks purchased from Integrated DNA Technologies. dTAG constructs were cloned by using a GS linker to fuse FKBP12 WT or mutant sequences to the C-terminus of the WT LIG1 cDNA sequence. All constructs were verified using Sanger sequencing.

Cell culture and cell line engineering

All BRCA mutant and WT cell lines were cultured and maintained at 37°C under 5% CO₂. Cell lines were tested for Mycoplasma weekly using the Lonza MycoAlert Detection Kit according to manufacturer’s instructions (Lonza Bioscience, LT07-318). Cell line sources and culture media formulations are listed in Supplementary Table S2. Cell lines were authenticated by short tandem repeat performed at Labcorp. All cells were used at or below passage 15 and within 8 weeks of thawing.

For generation of stable cell lines, constructs and lentiviral packaging mix (Cellecta, CPCP-K2A) were transfected into Lenti-X cells using Lipofectamine 3000 (Thermo Fisher Scientific, L3000015). After incubation with transfection reagents overnight, media was replaced with DMEM supplemented with 30% FBS. Media containing viruses was collected 48 hours after transfection and filtered with a 0.45 mm membrane. For infection, cells were incubated overnight in media containing 8 mg/mL polybrene (Sigma-Aldrich) and virus. Infected cells were recovered for 24 hours before selection with puromycin (Gibco, A1113802), blasticidin (Gibco, A1113903), or geneticin (Gibco, 10131035) corresponding to the antibiotic resistance cassette in each plasmid. Where applicable, LIG1 cDNA expression was titrated to correspond with endogenous levels of LIG1 protein.

For generation of single-cell clones, cells were plated at 2,000 cells/15 cm dish and allowed to grow until single colonies visible. Media was aspirated, and colonies were covered with sterile cloning discs (Thermo Fisher Scientific, 0790710B) soaked in 0.25% trypsin-EDTA and subsequently transferred to a sterile plate containing culture media, allowed to adhere overnight, and cultured as needed for experiments. Successful gene editing was confirmed by Western blotting and Sanger sequencing of genomic DNA (gDNA).

CRISPR-Cas9 screening and manual curation of BRCA mutations

The methodology for CRISPR-Cas9 screening and complete findings have been reported previously (25). For corroboration of LIG1 candidacy in the CRISPR-Cas9 screens, we performed manual curation of BRCA mutations in multiple cell lines using publicly available and in-house analyses to evaluate LIG1 dependence. BRCA mutation status was assigned to breast and ovarian cancer cell lines in the Project Achilles database by first aggregating mutation calls from different sources and sequencing technologies [Broad whole-exome sequencing (WES), Sanger WES, CGA WES, and Hybrid capture]. Next, the Ensembl Variant Effect Predictor was used to re-annotate the mutation calls to derive their relative impact scores (high, moderate, etc.). Sample allele counts were aggregated and used to derive the genotype status (zygosity) of each mutation. Loss of function samples had a high impact mutation or a moderate impact, homozygous mutation in BRCA1.

Colony formation assays

Engineered cells were seeded onto tissue culture plates at densities such that cells would reach 80% to 90% confluency at endpoint. For doxycycline (DOX)-inducible experiments, cells were seeded with or without 0.5 μg/mL doxycycline; experiments using dTAG-based protein degradation were seeded with the indicated concentration of dTAGv1. Media and treatment were refreshed every 3 to 4 days, and cells were allowed to grow for 14 days. For siRNA experiments, cells were transfected with SMARTpool siRNAs (Horizon Discovery) targeting either a nontargeting control sequence, LIG1 or PLK1 using Lipofectamine 3000 RNAiMAX (Thermo Fisher Scientific 13778100), according to manufacturer’s protocol. Cells were incubated with siRNAs overnight and received fresh media the following morning. Cells were re-transfected every 3 to 4 days over a 14-day period. At endpoint, cells were stained with crystal violet, dried, and imaged. For quantification, acetic acid was used to solubilize the dye, and absorbance was measured.

PAR staining and immunofluorescence

Engineered cells were seeded onto 8-well chamber slides (30,000 cells/well) with or without 0.5 μg/mL doxycycline. Cell media and treatment were refreshed on treatment day 2. On day 4, cells were treated with PARGi, fixed in methanol, and subsequently permeabilized and incubated with the antibodies listed in Supplementary Table S2. For each chamber, five fields/image were taken at 20× magnification.

In-cell Western blotting

Engineered cells were seeded onto black-sided, clear-bottom, 96-well tissue culture plates (15,000 cells/well) with or without 0.5 μg/mL doxycycline. The next day, cells were treated with PARGi and methyl methanesulfonate (MMS), fixed in methanol, and subsequently permeabilized and incubated with the antibodies listed in Supplementary Table S2. Fluorescence was visualized using a LI-COR Odyssey imaging system.

Immunoblotting

Protein lysate preparation and immunoblotting were performed, as described previously with minor modifications (25). Antibodies are summarized in Supplementary Table S2. Briefly, cells were rinsed in cold PBS and lysed in EBC buffer supplemented with protease and phosphatase inhibitors and universal nuclease. Lysates were cleared of insoluble material by centrifuging at 20,000 × g for 10 minutes at 4°C, and protein concentration was determined with the bicinchoninic acid protein assay.

For traditional Western blotting, 20 to 40 μg of protein in equal volumes was heated in LDS-sample buffer containing dithiothreitol (DTT) for 5 minutes at 95°C. Samples were centrifuged at 20,000 × g, separated by SDS-PAGE electrophoresis in 4% to 12% Bis-Tris gels and transferred to nitrocellulose membranes.

For simple Western blotting, 1.8 μg of protein in equal volumes was heated in 0.1× sample buffer containing DTT and fluorescent master mix for Jess (Bio-Techne), loaded onto Jess plates, and processed according to the manufacturer’s specifications. For detailed information on all immunoblotting procedures, see the Supplementary Methods.

NAD/NADH assay

NAD+ levels were measured using a NAD/NADH Assay kit (Abcam #ab65348). Assay procedure was performed according to the manufacturer’s protocol. MDA-MB-436 cells expressing a DOX-inducible CRISPR-resistant LIG1 cDNA and endogenous LIG1 knockout (KO) were treated with or without 0.5 μg/mL doxycycline for 6 days. Approximately 2 million cells per condition were harvested for NADH/NAD extraction. After removing NAD-consuming enzymes with a 10 kDa spin column (Abcam #ab93349), half of each sample was transferred to new tubes and incubated at 60°C for 30 minutes to decompose NAD+. NADtotal and NAD+ decomposed samples were diluted 1:3 in Extraction Buffer, then incubated with reaction mixture at room temperature for 5 minutes. NADH Developer was added to each well, and the reaction cycled at room temperature for 4 hours. Absorbance was measured at OD 450 nm on a SpectraMax microplate reader (Molecular Devices).

Immunofluorescence staining and assessment of micronuclei

Cells were seeded on #1.5 22 × 22 mm coverslips (Epredia, 152222) in 6-well plates (Corning). After being allowed to adhere for at least 16 hours, cells were washed with PBS and then fixed with 100% methanol (Fisher Chemical, A411-4) for 15 minutes at −20°C. After a PBS wash, 0.5% Triton X-100 (Thermo Fisher Scientific, 85112) in PBS was used to permeabilize cells for 15 minutes at room temperature. After being washed twice with PBS, coverslips were incubated in PBS + 10 μmol/L Hoechst 33342 (Thermo Fisher Scientific, 62249) for 5 minutes at room temperature to stain for DAPI, and then washed with PBS. After drying at room temperature, coverslips were mounted on slides (Thermo Fisher Scientific, 22-178-277) with ProLong Gold Antifade Mountant (Invitrogen, P36930) and allowed to dry for at least 16 hours at room temperature. Cells were imaged via an inverted Zeiss Axio Observer microscope using Zeiss ZEN pro imaging software. Analysis was completed using the cell counter functions of ImageJ software.

Xenograft model in NOG mice

Animal studies were conducted at Pharmaron. Female NOG mice (Vital River) 6 to 8 weeks of age were inoculated subcutaneously on the right flank with p5 MDA-MB-436-pTG59-CloneA or MDA-MB-436-pTG53-CloneD cells (1 × 10⁷ cells per animal) and 50% Matrigel (1:1) in RPMI1640. Doxycycline (25 mg/kg/day, orally) treatment was initiated immediately following implantation. When xenografts had reached a mean volume of ∼200 mm³, animals were randomized into DOX+ (25 mg/kg, every day orally) and DOX− (saline 10 μL/g, every day orally) groups.

Tumor measurements were conducted twice weekly using a caliper and estimated using the formula: tumor volume = a × b × b/2, where “a” and “b” are the long and short diameters of the tumor, respectively. Animals were euthanized on day 39 for tissue harvesting and protein quantification. All procedures related to animal handling, care, and treatment were approved by the local Institutional Animal Care and Use Committee and performed in accordance with Association for Assessment and Accreditation of Laboratory Animal Care International guidance.

Statistical analysis

Image quantification was conducted using ImageJ (NIH, Bethesda, MD; RRID: SCR_003070). All statistical analyses were performed using GraphPad Prism (GraphPad, Software Inc.; RRID: SCR_002798).

Data availability

The data generated in this study are available within the article and its Supplementary Materials. Additional information is available on request.

Results

Identification of LIG1 as a synthetic lethal target in BRCA1 mutant cell lines

To identify genes, whose inactivation conferred lethality in BRCA1 mutant, but not WT cells, we conducted CRISPR-Cas9 screens in 13 BRCA1/2 WT (MDA-MB-231, KP4, HCC1954, HCC70, AU565, A549, HCC38, HGC-27, FADU, 23132/87, SNU638, NUGC-4, and SNU-5) and two BRCA1 mutant (MDA-MB-436 and SUM149PT) cancer cell lines using a druggable genome library. In addition to the identification of known synthetic lethal targets PARP1 (26), USP1 (22, 27), and Polθ (18), we identified LIG1 as a candidate synthetic lethal target in BRCA1 mutant cell lines (Fig. 1A). Next, we conducted an internal analysis of BRCA mutation status for individual breast and ovarian cancer cell lines in the Project Achilles database. LIG1 lethality scores were tabulated for all included cell lines and confirmed selective hyperdependence of BRCA1 mutant cell lines on LIG1 (Fig. 1B). To ask whether the dependence on LIG1 would be extrapolatable to the BRCA2 mutant context, we analyzed publicly available CRISPR screening data in DepMap (Supplementary Fig. S1A; ref. 28). Cell lines harboring damaging mutations in BRCA2 were significantly more dependent on LIG1 as compared with BRCA2 intact cell lines as demonstrated by more negative Chronos gene effect scores, indicating that mutation in BRCA1 or BRCA2 could confer sensitivity to LIG1 inactivation. We also assessed whether a DLD1 cell line pair that is isogenic for BRCA2 would be differentially sensitive to LIG1 knockdown based on BRCA2 status (Supplementary Fig. S1B–S1E). DLD1 cells that were deficient for BRCA2 were dependent on LIG1 for survival, whereas DLD1 cells with BRCA2 intact were not impacted by loss of LIG1.

Figure 1.

Hyperdependence of multiple BRCA1-mutated cancer cell lines on LIG1. A, Volcano plot of a MAGeCK analysis comparing CRISPR-Cas9 screens performed in two BRCA1 mutant and 13 BRCA1/2 WT breast cancer cell lines (BRCA1/2 WT: MDA-MB-231, KP4, HCC1954, HCC70, AU565, A549, HCC38, HGC-27, FADU, 23132/87, SNU638, NUGC-4, and SNU-5. BRCA1 mutant: MDA-MB-436 and SUM149PT.) Genes on the left side of the plot are preferentially depleted in the BRCA1 mutant cells (potential drug targets), whereas those on the right side are preferentially depleted in BRCA1/2 WT cells (potential resistance markers) or resulted in enhanced growth of BRCA1 mutant cells. Known BRCA dependencies (PARP1, USP1, and POLQ) are highlighted. B, Waterfall plot of LIG1 lethality scores for breast and ovarian cancer cell lines. BRCA1 mutations were manually curated for individual cell lines from the Project Achilles database (see “Materials and Methods”).

Perturbation of LIG1 inhibits growth of BRCA1 mutant cells

To assess the robustness of the synthetic lethal relationship between LIG1 and BRCA1, we used multiple strategies for single-gene perturbation using the MDA-MB-436 cell line, which has a BRCA1 mutation that results in loss of BRCA1 protein expression (29). To assess whether inactivation of LIG1 at the mRNA level affects growth of BRCA1 mutant cells, cells were first engineered to express a DOX-inducible CRISPRi guide RNA targeting the promoter of either an intron-targeting control sequence or LIG1. Following validation of LIG1 knockdown by Western blot, a 14-day colony formation assay was performed to assess the impact of LIG1 knockdown on colony growth (Fig. 2A). LIG1 knockdown reduced colony formation of BRCA1 mutant cells by 89%. Colony formation of cells expressing the intron-targeting control sequence was unaffected by DOX treatment (Supplementary Fig. S2).

Figure 2.

Perturbation of LIG1 is lethal in BRCA1-mutated cancer cells. Colony formation assays (treatment day 14) in MDA-MB-436 cells engineered to express (A and B) a DOX-inducible CRISPRi guide RNA targeting either an intron-targeting control sequence (Supplementary Fig. S2) or the promoter of LIG1, or a CRISPR-resistant LIG1 cDNA fused to either the FKBP12mut (dTAG) degron (degradable; C–E) or the FKBP12wt degron (nondegradable; F–H), with KO of endogenous LIG1. B, E, and H, Model validation with Western blotting on treatment day 4 (quantified in Supplementary Fig. S3). E and G, Corresponding quantifications of relative viability on treatment day 14 (N = 2 biological replicates).

To determine whether perturbation of LIG1 at the protein level affects viability of BRCA1 mutant cells, we engineered MDA-MB-436 cells to express a CRISPR-resistant LIG1 cDNA (crLIG1) fused to either the FKBP12wt degron (nondegradable) or FKBP12mut (dTAG) degron (degradable) to produce a titratable degradable LIG1 protein [for dTAG model details, see (30)]. Viral titer and promoter strength were selected to approximate levels of endogenous LIG1 protein expression. These cells were then edited to knock out endogenous LIG1. Treatment with increasing concentrations of the degron-mediated PROTAC dTAGv1 produced concentration-dependent loss of LIG1-dTAG protein as well as concentration-dependent decreases in colony formation (Fig. 2B; Supplementary Fig. S3), whereas no effect was observed in cells expressing the nondegradable LIG1 cDNA (crLIG1-FKBP12wt). These findings support the hyperdependence of the BRCA1 mutant cell line MDA-MB-436 on LIG1 expression.

LIG1 synthetic lethality is on-target and selective for BRCA1 mutant lines

To ask whether the LIG1 dependence discovered in our screen was on-target and specific for the BRCA1 mutant context, we engineered the BRCA1 mutant cell line MDA-MB-436 and the BRCA1/2 WT HCC1954 cell line to express a DOX-inducible CRISPR-resistant LIG1 cDNA and edited the endogenous LIG1 gene to produce complete KO of endogenous LIG1. In this model, withdrawal of DOX results in complete depletion of all LIG1 (and thus loss of LIG1 protein). DOX withdrawal reduced colony formation in BRCA1 mutant cells by 87% but had no effect on colony formation of BRCA1/2 WT cells (Fig. 3A and B). These data demonstrate that the loss of viability in BRCA1 mutant cells is on-target through LIG1 as the inducible LIG1 cDNA fully rescued the effect of endogenous LIG1 KO, and that the dependence on LIG1 is specific to the BRCA1 mutant context. These results were recapitulated using MDA-MB-436 cells engineered to be isogenic through re-expression of BRCA1. Using this isogenic pair, knockdown of LIG1 eliminates colony formation in BRCA1 mutant MDA-MB-436 cells expressing an empty vector control plasmid, whereas re-expression of BRCA1 rescues this viability defect despite similar level of LIG1 depletion (Supplementary Fig. S4).

Figure 3.

LIG1 is a selective synthetic lethal target in BRCA1-mutated cell lines. Representative image (A) and quantification (B) of colony formation assay (treatment day 14) in MDA-MB-436 (BRCA1 mutant) and HCC1954 (BRCA WT) cells engineered to express a DOX-inducible CRISPR-resistant LIG1 cDNA with KO of endogenous LIG1. Right, model validation with Western blotting on treatment day 4. Representative colony formation assay (treatment day 14; C) and quantification (E) in five BRCA1 mutant cell lines (MDA-MB-436, COV362, SUM149PT, HCC1395, and UWB1.289) and three BRCA1/2 WT cell lines (HCC1954, HCC1419, and AU565) using DOX-inducible shRNA against LIG1, PLK1 (pan-lethal control), or a control sequence. D, Model validation with quantification of Western blots on treatment day 4.

To investigate whether our findings in MDA-MB-436 and HCC1954 cell lines were representative of responses in a wider range of BRCA1 mutant and BRCA WT cell lines, we engineered DOX-inducible shRNAs targeting LIG1, PLK1 (pan-lethal control), or a control sequence in five BRCA1 mutant cell lines (MDA-MB-436, COV362, SUM149PT, HCC1395, and UWB1.289) and three BRCA1/2 WT cell lines (HCC1954, HCC1419, and AU565). LIG1 knockdown selectively reduced viability in BRCA1 mutant cell lines but not BRCA1/2 WT cell lines (Fig. 3C and D). These data demonstrate that the synthetic lethality between LIG1 and BRCA1 is true across a variety of cell lines, indicating the robustness of the relationship. Furthermore, we asked whether the homologous recombination deficient (HRD+) but BRCA1/2 WT cell line HCC1806 (31) was dependent on LIG1 for survival. LIG1 knockdown inhibited colony formation in this cell line (Supplementary Fig. S5), indicating that LIG1 dependence may be a feature of HRD beyond BRCA1 inactivation.

Catalytic activity of LIG1 is required for BRCA1 mutant cell viability

We next asked whether the catalytic activity of LIG1 is necessary for BRCA1 mutant cell viability. To perform its enzymatic function of sealing single-stranded nicks in the DNA backbone, the LIG1 protein transfers an AMP molecule onto the DNA nick. Lys568 is critical for the catalytic activity of LIG1 due to its role in hydrolyzing ATP and coordinating the AMP moiety, and the K568A mutant produces a catalytically inactive form of the enzyme (Fig. 4A; refs. 32, 33). BRCA1 mutant MDA-MB-436 cells were engineered to express a shRNA-resistant WT or LIG1^K568A mutant cDNA alongside a DOX-inducible shRNA targeting LIG1. Consistent with previous observations, LIG1 knockdown reduced growth of MDA-MB-436 cells (no cDNA) in 14-day colony formation assays. This loss of viability was rescued by expression of an shRNA-resistant WT LIG1 cDNA. In contrast, expression of the catalytically dead LIG1^K568A mutant was not sufficient to rescue the loss of viability resultant from endogenous LIG1 depletion (Fig. 4B and C). These findings indicate that the catalytic activity of LIG1 is required for survival of BRCA1 mutant cells.

Figure 4.

Catalytic activity of LIG1 is required for viability in BRCA1-mutated cells. A, Illustrated crystal structure (PDB 6P0A) of LIG1 with nicked dsDNA. Lys568 is shown in cyan blue, AMP is shown in purple, and DNA is shown in coral. The K568A mutation produces a catalytically inactive LIG1 protein. B, Model validation with Western blotting on treatment day 4. C, Colony formation assay (treatment day 14) in MDA-MB-436 cells engineered to express an shRNA-resistant WT or LIG1^K568A mutant cDNA alongside a DOX-inducible shRNA targeting LIG1.

LIG1 inactivation leads to increased PARylation and chromosomal instability

LIG1 suppresses ssDNA gaps and DNA nicks through its activity ligating Okazaki fragments and through its role in DNA damage repair. In the absence of ligation, unligated DNA nicks are subject to PARylation to signal for the recruitment of DNA repair machinery (34, 35). We examined the effect of functional LIG1 expression on PARylation to further elucidate the mechanism of LIG1 synthetic lethality with BRCA1. Cells expressing an shRNA-resistant WT or LIG1^K568A mutant cDNA alongside a DOX-inducible shRNA targeting LIG1 were subject to immunofluorescence (IF)-based staining for PAR. Knockdown of LIG1 led to a ∼19× increase in PAR staining, and this effect could be completely rescued by expression of the shRNA-resistant LIG1. In contrast, expression of the catalytically inactive LIG1^K568A mutant was not sufficient to rescue induction of PAR caused by endogenous LIG1 knockdown (Fig. 5A and B). To corroborate the PAR signal observed with LIG1 mRNA knockdown, we engineered to express DOX-inducible LIG1 cDNA with endogenous LIG1 gene KO to ask whether the same phenomenon could be observed when inactivating LIG1 at the DNA level. Consistent with our shRNA knockdown results, DOX withdrawal (and thus loss of LIG1 expression) led to an increase in PARylation as measured by in-cell Western blotting (Supplementary Fig. S6). These data demonstrate that loss of functional LIG1 expression is associated with increased PARylation in BRCA1 mutant cells, consistent with its role in sealing DNA nicks.

Figure 5.

Perturbation of LIG1 induces PAR accumulation and micronuclei formation in BRCA1-mutated cells. A and B, Representative image (A) and quantification (B, top) of PAR IF on treatment day 4 in MDA-MB-436 cells expressing an shRNA-resistant WT or LIG1^K568A mutant cDNA alongside a DOX-inducible shRNA targeting LIG1. B, Bottom, model validation with Western blotting on treatment day 3. C, Top, quantification of PAR/DRAQ5 (800 nm/700 nm) IF ratio on treatment day 4 in a model of titratable degradation of LIG1, using MDA-MB-436 cells engineered to express LIG1-dTAG (or LIG1-FKBP12wt control) fusion proteins with endogenous LIG1 KO. Bottom, model validation with Western blotting. D, Quantification of micronuclei relative to total cell count measured by IF staining (one-way ANOVA P < 0.0001, Tukey multiple comparisons test sgLIG1 +DOX vs. −DOX P < 0.0001) in MDA-MB-436 cells expressing Dox-inducible LIG1 cDNA and either endogenous LIG1 KO (sgLIG1) or an intron-targeting control sgRNA (sgITC). ****, P < 0.0001 Endog., endogenous; Exog., exogenous.

We next asked whether there was a concentration-dependent relationship between the level of LIG1 protein and accumulation of PARylation using titratable degradation of LIG1 fusion proteins (LIG1-dTAG or LIG1-FKBP12wt control) in cells with endogenous LIG1 KO. Increasing concentrations of dTAGv1 resulted in concentration-dependent loss of LIG1-dTAG protein as well as concentration-dependent increases in the ratio of PAR/DRAQ5 IF, whereas no effect was observed in cells expressing the nondegradable LIG1 fusion protein (LIG1-FKBP12wt; Fig. 5C). Findings support the hypothesis that loss of LIG1 activity is directly responsible for associated increases in PARylation observed in BRCA1 mutant cells, and that this effect is proportional to the amount of LIG1 protein available to seal DNA nicks. PARP1 uses the cofactor NAD+ to add PAR chains to unligated DNA nicks. Consistent with this requirement, we observed a significant decrease in NAD+ level in cells lacking LIG1 in a model where endogenous LIG1 has been knocked out using CRISPR (sgLIG1) and an exogenous Dox-inducible LIG1 cDNA had been stably expressed (Supplementary Fig. S5C). No significant change in NAD+ level was observed in cells expressing an sgRNA targeting an intron and expressing exogenous Dox-inducible LIG1 cDNA.

The increase in PARylation observed when LIG1 is inactivated is consistent with our hypothesized mechanism in which loss of LIG1 activity leads to accumulation of unligated DNA and genomic instability. To assess this, we measured the number of micronuclei in cells with or without LIG1 knockdown using IF and normalized to total cell number (Fig. 5D). We observed a significant increase in the number of micronuclei in cells with LIG1 knockdown as compared with cells harboring shRNA targeting a control sequence. The formation of micronuclei is consistent with genomic instability as a mechanism leading to cell death.

Loss of LIG1 inhibits tumor growth in vivo

A murine xenograft model was used to ask whether the results of our in vitro studies could translate to an in vivo setting. MDA-MB-436 cells engineered to express a DOX-inducible CRISPR-resistant LIG1 cDNA in the presence (sgLIG1) or absence (sgITC) of endogenous LIG1 KO were xenografted into NOG mice, and tumors were allowed to form under DOX treatment (25 mg/kg/day, orally). Once tumors were established, mice were randomized into two groups, and DOX treatment was maintained (+DOX) or withdrawn (−DOX) to sustain or remove LIG1 expression, respectively. Whereas DOX treatment had no influence on tumor volume in sgITC clone tumors expressing endogenous LIG1 [Fig. 6A; 2,602 mm³ vs. 2,326 mm³, T/C value = 88%, P = not significant (N.S.)], withdrawal of DOX (and subsequent loss of exogenous LIG1 expression) was associated with significant tumor growth inhibition in tumors lacking endogenous LIG1 (Fig. 6B; 2019 mm³ vs. 351 mm³, T/C value = 8%, P < 0.0001). Loss of LIG1 expression was confirmed by Western blotting in sgLIG1-DOX mice at an early time point (day 3; Fig. 6C) to confirm the model and at endpoint (day 29; Supplementary Fig. S7). Consistent with our observations, in vitro, loss of LIG1 suppressed BRCA1 mutant tumor growth in vivo.

Figure 6.

LIG1 inactivation inhibits BRCA1 mutant tumor growth in a murine xenograft model. A and B, Immunodeficient mice were grafted with MDA-MB-436 cells engineered to express a DOX-inducible CRISPR-resistant LIG1 cDNA in the presence (sgLIG1, A) or absence (sgITC, B) of endogenous LIG1 KO. Following tumor establishment under DOX treatment, DOX treatment was maintained (+DOX) or withdrawn (−DOX) to sustain or remove LIG1 expression, respectively. Mice grafted with sgLIG1 and sgITC clones were euthanized on treatment days 37 and 32, respectively. Representative image (C) and quantification (D) of model validation with Western blotting on treatment day 3.

Discussion

In this study, we demonstrate synthetic lethality between LIG1 and BRCA1 using in vitro and in vivo genetic strategies. Perturbation of LIG1 resulted in on-target loss of viability in BRCA1 mutant cells across multiple cell lines and exhibited selectivity for BRCA1 mutant cells by lacking this effect in cell lines expressing functional BRCA1. This hyperdependence of BRCA1 mutant cells on LIG1 was corroborated by consistent results whether perturbing LIG1 at the DNA, mRNA, or protein level. Although the greatest depth of inquiry relies on MDA-MB-436 cells, our cell line panel in Fig. 3C indicates that the results are robust and likely generalizable for the BRCA mutant context. The growth inhibition induced by LIG1 inactivation could be completely rescued by reconstitution of exogenous LIG1, demonstrating the on-target nature of the effect but could not be rescued by supplementation of a catalytically inactive LIG1. Loss of LIG1 catalytic activity was associated with increased PARylation, formation of micronuclei, and inhibition of cell growth, consistent with the known role of LIG1 in DNA replication and damage repair. LIG1 was also necessary for BRCA1 mutant tumor growth in a murine xenograft model. Taken together, these data demonstrate that LIG1 is a validated target for synthetic lethality in BRCA1-associated cancers. The identification and development of novel stand-alone targets such as LIG1 for treating BRCA1 mutant cancers is important for potential future clinical practice in both initial and long-term maintenance treatment of patients harboring BRCA1 mutant tumors.

Clinically, PARP inhibitors are an important therapeutic option for patients with BRCA1-mutated cancers. However, the clinical utility of these agents is limited by inherent or acquired resistance (12–14), highlighting the need for alternative targets. Inhibitors of ideal targets should capably substitute for PARP inhibitors where necessary, facilitate resensitization, or act synergistically in combination therapy with PARP inhibitors or other anticancer agents. Recently, several new targets for synthetic lethality with BRCA1/2 have been proposed or identified in the literature and share a common mechanism in which loss of a compensatory DNA repair pathway leads to genomic instability (36). BRCA proteins play important roles in several steps of the DNA damage response, including cell-cycle checkpoint activation and repair of dsDNA breaks through HR (37, 38). Accordingly, synthetic lethality targets frequently interact with BRCA either directly or indirectly in processes that maintain genome integrity. Patel and colleagues (36) recently summarized such synthetic lethal interactions for BRCA1/2 and the gene products of PARP, POLQ, RAD52, FANCD2, FEN1, and APEX2. Common to many of these BRCA1/2 synthetic lethal targets is the generation of single-strand gaps or double-strand breaks (DSB) that result from their loss. Recent studies indicate that unresolved ssDNA gaps lead to synthetic lethality with BRCA1 (39–41), whereas an alternative hypothesis proposes that single-stranded gaps become DSBs, which then depend on HR for repair. Inactivation of LIG1 could contribute to both proposed mechanisms, through formation of ssDNA by failure to seal nicked DNA at Okazaki fragments due to the role of LIG1 in normal DNA replication or by virtue of loss of the backup DSB repair through the role of LIG1 in alternative end joining (42).

The present research identifies LIG1, the gene product of LIG1, as a stand-alone oncological target that is synthetic lethal with BRCA1 inactivation. LIG1 functions in DNA replication and recombination as well as base excision repair (26); specifically, LIG1 ligates Okazaki fragments and participates in nucleotide resection and DNA repair processes through interaction with PCNA (27) or through interaction with DNA polymerases. In addition to a clear avenue for synthetic lethality with BRCA1, these critical roles for LIG1 have also led to its implication as a target for oncological treatment development or as a prognostic biomarker in other cancer settings. Ali and colleagues (43) recently reported synthetic lethality of LIG1 in XRCC1-deficient cancers, and further demonstrated that LIG1 overexpression predicts platinum resistance in this subset. Overexpression of LIG1 has also been identified in several human breast cancers (44, 45). Loss of LIG1 activity has been proposed to mediate anticancer effects by blocking replication, by introducing mutations due to DNA repair pathway failures, and by sensitizing cells to DNA-damaging chemotherapeutic agents (43). To this end, LIG1 is a therapeutic target with wide potential applications in oncology, including its utility as a target for monotherapy in BRCA1-mutant cancers.

In previous research, we demonstrated synthetic lethality between USP1 and BRCA1/2 via a mechanism involving reduction in the level of PCNA, which acts as a sliding clamp for DNA replication machinery and furthermore illustrated synergy with PARP inhibition (22). Although our data demonstrate the potential for LIG1 as a stand-alone target in BRCA1-mutant cancers, it may be hypothesized that LIG1 inhibition has potential to exhibit synergy with PARP inhibitors in the same manner. Consistent with our observations in the present study, Hanzlikova and colleagues (34) demonstrated that perturbation of LIG1 increases S-phase PARylation due to an accumulation of unligated Okazaki fragments and subsequent PARP activity. Following LIG1 inactivation, PARylation led to the recruitment of XRCC1, indicating that PARP activity recruits the single-strand break repair pathway as “backup” to process unligated Okazaki fragments. Therefore, simultaneous inhibition of LIG1 and PARP in BRCA1 mutant cells may produce synergistic lethality, a possibility that merits future investigation in the interest of expanding the efficacy and utility of current PARP inhibitors. While many mechanisms for intrinsic and acquired resistance to PARP inhibitors have been proposed (46, 47), our data using BRCA1 re-expression to rescue LIG1 knockdown-induced growth inhibition in MDA-MB-436 cells (Supplementary Fig. S4) suggest that PARP inhibitor resistance caused by restoration of homologous repair capability would not likely present a context that would respond to LIG1 inactivation. Additional mechanisms with homologous recombination deficiencies (HRD+) may be dependent on LIG1, as demonstrated by the sensitivity to LIG1 knockdown observed in HRD+, BRCA1/2 WT HCC1806 cells (Supplementary Fig. S5).

In conclusion, the findings of this study identify and validate LIG1 as a promising target for synthetic lethality in BRCA1-mutant cancers. In cells with mutated BRCA1, loss of functional LIG1 expression reduces growth of BRCA1-mutant cancer cells and is associated with increased PARylation. LIG1 inhibition is thus an attractive option for complementing the current arsenal of anticancer agents for the treatment of patients with gynecological and other cancers.

Supplementary Material

Supplementary Figure S1

Cell lines with damaging mutations in BRCA2 are dependent on LIG1. A) Analysis of DepMap data (28) comparing LIG1 dependence of cell lines with damaging mutations in BRCA2 vs. BRCA2 wild-type cell lines. ***p=0.0004 unpaired t-test. B,C) Representative image (B) and quantification (C) of colony formation assay in DLD1 wild-type (WT) and BRCA2-/- cells transfected with siRNAs targeting the LIG1 gene or a control sequence. D,E) Representative image (D) and quantification (E) of Western blot assessing LIG1 knockdown at endpoint for the colony formation assay in (B).

Supplementary Figure S2

CRISPRi system does not reduce viability of BRCA mutant cells using an intron-targeting control guide RNA (sgITC). Representative image (A) and quantification (B) of colony formation assay (treatment day 14) in MDA-MB-436 (BRCA1 mutant) cells engineered to express a DOX-inducible CRISPRi guide RNA targeting an intron (sgITC). Knockdown was validated by Western blot (representative image in C, quantification in D). FC, fold-change.

Supplementary Figure S3

Quantification of western blots for dTAG-mediated LIG1 degradation. MDA-MB-436 cells were engineered to express a CRISPR-resistant LIG1 cDNA fused to either the FKBP12mut (dTAG) degron (degradable) (A) or the FKBP12wt degron (non-degradable) (B), with knockout of endogenous LIG1. Cells were then treated for 4 days with vehicle (DMSO) or the indicated doses of the VHL-targeting PROTAC dTAGv1 and then subject to immunoblotting for LIG1, HA tag, and a loading control. FC, fold-change.

Supplementary Figure S4

LIG1 inactivation is lethal in BRCA1-mutant MDA-MB-436 cells, and is rescued by exogenous expression of wild-type BRCA1. A) Representative colony formation assay (treatment day 14) using DOX-inducible shRNA against LIG1, PLK1 (pan-lethal control), or a control sequence in MDA-MB-436 cells stably infected with cDNA encoding wild-type BRCA1 or a control vector. B) Model validation with Western blotting. C) Western blots assessing LIG1 protein level at assay endpoint.

Supplementary Figure S5

LIG1 inactivation is lethal in HRD+ BRCA1/2 wild-type HCC1806 cells. A) Representative colony formation assay (treatment day 11) using DOX-inducible shRNA against LIG1, PLK1 (pan-lethal control), or a control sequence in HCC1806 cells. B) Western blots assessing LIG1 protein level in the models from (A) (treatment day 4).

Supplementary Figure S6

Perturbation of LIG1 induces proportional PAR accumulation in BRCA1-mutated cells in immunofluorescence and in-cell western assay formats. A) In-cell Western blotting of PAR on treatment day 4 in MDA-MB-436 cells engineered to express a DOX-inducible LIG1 cDNA with endogenous LIG1 knockout. B) Model validation with Western blotting. C) Quantification of relative NAD+ as deduced by total NAD (NADt)/NADH ratio normalized to +DOX condition (t-test sgLIG1 +DOX vs. -DOX p < 0.0001).

Supplementary Figure S7

Withdrawal of Doxycycline leads to LIG1 depletion that is sustained through the study. Image (A) and quantification (B) of western blotting data from tumors of immunodeficient mice grafted with MDA-MB-436 cells engineered to express a DOX-inducible CRISPR-resistant LIG1 cDNA with endogenous LIG1 knockout. Samples depicted are from endpoint (day 29 post-grouping).

Supplementary Data

Supplementary Materials

Authors’ Disclosures

L.C.M. Martires, L.G. Ahronian, C.B. Pratt, N.M. Das, X. Zhang, D.A. Whittington, H. Zhang, B. Shen, J. Come, P. McCarren, M.-S. Liu, C. Min, T. Feng, H. Jahic, J.A. Ali, D.R. Aird, F. Li, J.N. Andersen, A. Huang, W.D. Mallender, and H.E. Nicholson report other support from Tango Therapeutics outside the submitted work.

Authors’ Contributions

L.C.M. Martires: Data curation, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. L.G. Ahronian: Conceptualization, validation, investigation, visualization, methodology. C.B. Pratt: Validation, investigation, methodology. N.M. Das: Validation, investigation, methodology. X. Zhang: Validation, investigation, methodology. D.A. Whittington: Conceptualization, data curation, writing–review and editing. H. Zhang: Conceptualization, validation, investigation. B. Shen: Resources, investigation. J. Come: Supervision, writing–review and editing. P. McCarren: Supervision, visualization, writing–review and editing. M.-S. Liu: Investigation, writing–review and editing. C. Min: Conceptualization, investigation. T. Feng: Investigation. H. Jahic: Validation, investigation. J.A. Ali: Conceptualization, supervision. D.R. Aird: Investigation. F. Li: Data curation, software. J.N. Andersen: Conceptualization, supervision, writing–review and editing. A. Huang: Conceptualization, supervision. W.D. Mallender: Conceptualization, supervision, project administration, writing–review and editing. H.E. Nicholson: Conceptualization, data curation, formal analysis, supervision, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing.