PARP1 trapping activates ATM-mediated NF-κB signaling independent of replication in response to TOP1 blockade Authors: Elodie Bournique, Ambrocio Sanchez, Kim Ha, Katrina M Yan, Lavanya Manjunath, Gisselle Santiago, Pedro Ortega, Rémi Buisson Published: 2026 ABSTRACT AbstractThe NF-κB signaling pathway is a key driver of inflammation and can be activated by many genotoxic stresses. Yet, the mechanisms by which different types of DNA damage trigger NF-κB remain poorly understood. In this study, we find that NF-κB activation by topoisomerase 1 (TOP1) inhibition is strongly increased when cells are treated in combination with PARP inhibitors. Mechanistically, we demonstrate that TOP1 inhibition activates ATM-mediated NF-κB signaling through two distinct pathways: a replication-dependent pathway triggered by replication fork collapse and a replication-independent pathway revealed upon PARP1 inhibition. We further show that PARP1 enzymatic activity is not required to suppress NF-κB signaling after TOP1 inhibition during the replication-independent pathway. Instead, PARP inhibitors trap PARP1 at DNA lesions, thereby blocking repair and promoting ATM-dependent NF-κB signaling, mimicking the cell response to TDP1 depletion. These findings reveal an unexpected role for PARP1 as an NF-κB activator when trapped at DNA lesions induced by TOP1 inhibition, providing a therapeutic opportunity to use PARP inhibitors to enhance inflammatory responses and potentially improve the efficacy of TOP1-targeted cancer therapies. Introduction The mammalian NF-κB (nuclear factor kappa-light-chain-enhancer of activated B cells) regulates various components of both innate and adaptive immune responses, acting as a central mediator of inflammation [1–3]. The NF-κB transcription factor family includes five members: NF-κB1 (p105/p50), NF-κB2 (p100/p52), RELA (p65), RELB (RelB), and REL (c-REL). These proteins share a conserved Rel homology domain, a sequence of 300 amino acids that is responsible for DNA binding, dimerization, and interaction with inhibitory IκB proteins. NF-κB subunits can form a variety of homo- and heterodimers, enabling functional diversity in gene regulation in response to different types of stimuli and stresses [1–3]. The p65/p50 heterodimer is the most abundant and transcriptionally active form in many cell types. In unstressed cells, p65/p50 is sequestered in the cytoplasm through interactions with IκBα. Upon activation by various stimuli such as cytokines (e.g. TNF-α, IL-1β), microbial components, or DNA damage, IκBα is phosphorylated and degraded via the ubiquitin-proteasome pathway. This process releases NF-κB dimers, allowing p65/p50 to translocate into the nucleus and control the expression of many genes involved in inflammation, immune responses, cell survival, and proliferation. [1–3]. DNA damage caused by various genotoxic stresses is one of several intrinsic stimuli that can activate the NF-κB pathway independently of transmembrane receptor signaling [3, 4]. Depending on the type of DNA lesions, cell lines, and the timing after damage, genotoxic stress triggers acute NF-κB activation through distinct mechanisms. In the canonical pathway, DNA double-strand breaks (DSBs) activate the ATM kinase, which in turn stimulates the ubiquitin ligase TRAF6. TRAF6 activates the IKK complex [IKKα, IKKβ, and IKKγ (NEMO)], leading to proteasomal degradation of IκBα and the translocation of p65/p50 into the nucleus, promoting the expression of NF-κB target genes [5–9]. In contrast, DNA lesions that block transcription, such as ultraviolet (UV)-induced damage, activate NF-κB through a distinct mechanism that is independent of ATM [7]. Transcriptional stress causes the secretion of interleukin-1α (IL-1α) from damaged cells, which engages the IL-1R1–Myd88 receptor complex of neighboring cells [7, 10]. This signaling cascade activates IRAK1, which, like ATM, stimulates TRAF6 to promote p65-mediated transcription of NF-κB target genes [7]. In addition, persistent DNA damage responses can lead to the accumulation of cytosolic double-stranded DNA, either from micronuclei formed during defective mitosis or from DNA fragments released during nuclear repair [11, 12]. cGAS–STING senses cytosolic DNAs, which activate IRFs and NF-κB to trigger an interferon (IFN) response [11, 12]. Alternatively, noncanonical activation of STING by ATM and IFI16, but independently of cGAS, has been shown to induce NF-κB signaling in keratinocyte cells [13]. However, in many cancer cells, activation of NF-κB through the IL-1α–IL-1R1, cGAS–STING, and IFI16 pathways is suppressed, as cancer cells decrease their expression to avoid inflammatory signaling [7, 14–16]. These findings highlight that NF-κB activation is not a generic response to DNA damage, but rather a lesion-specific process governed by distinct signaling mechanisms. Therefore, a deeper understanding of how various types of DNA lesions activate NF-κB signaling over time and across the cell cycle is necessary. Poly(ADP-ribose) polymerase 1 (PARP1) is a nuclear enzyme that plays a crucial role in DNA repair and the maintenance of genome integrity by sensing various types of DNA lesions [17, 18]. Once activated, PARP1 catalyzes the attachment of poly(ADP-ribose) (PAR) to itself or to surrounding proteins, thereby promoting repair by recruiting repair factors [17, 18]. In addition to its function in DNA repair, PARP1 stimulates NF-κB following certain types of DNA damage [5, 13, 19, 20]; however, the mechanism by which PARP1 regulates NF-κB activation remains unclear and is a subject of debate, necessitating further investigation. In this study, we aimed to determine the role of PARP1 in regulating the NF-κB pathway depending on the type of DNA lesions encountered by cells. We conducted an unbiased targeted screen by generating distinct types of DNA lesions in cells, in the presence or absence of a PARP inhibitor, and monitored p65 nuclear localization as a marker of NF-κB activation. Surprisingly, we found that cell treatment with PARP inhibitors significantly increases p65 nuclear localization, specifically after DNA damage caused by topoisomerase 1 (TOP1) inhibitors, resulting in the expression of many inflammatory genes. Mechanistically, we demonstrate that PARP1 acts as an NF-κB activator, not through its enzymatic activity, but by being trapped on DNA, which results in an increased level of DNA damage and subsequently activates ATM-mediated NF-κB and the expression of inflammatory genes. This suggests that, in addition to increasing DNA damage levels in tumor cells, PARP inhibitor treatment in combination with TOP1 blockade may promote tumor-immune interactions by influencing the recruitment of immune cells. Materials and methods Cell lines RKO cells were maintained in DMEM/F12 GlutaMAX supplemented with 10% fetal bovine serum (FBS) and 1% penicillin–streptomycin. MCF10A cells were cultured in DMEM/F12 supplemented with 5% horse serum, 2 ng/ml epidermal growth factor, 0.5 μg/ml hydrocortisone, 100 ng/ml cholera toxin, 10 μg/ml insulin, and 1% penicillin–streptomycin. The HCC1806 cell line was cultured in RPMI GlutaMAX supplemented with 10% FBS and 1% penicillin–streptomycin. The U2OS lines were cultured in DMEM supplemented with 4 mM L-glutamine, 10% FBS, and 1% penicillin–streptomycin. Reagents Chemicals, oligonucleotides (siRNA, gRNA, and DNA primers), and antibodies used in this study are listed in Supplementary Tables S1–S5.] Immunofluorescence Cells grown on glass coverslips were fixed with paraformaldehyde [3% paraformaldehyde and 2% sucrose in 1 × phosphate buffered saline (PBS)] for 20 min, washed twice with 1 × PBS and permeabilized with a permeabilization buffer (1 × PBS and 0.2% Triton X-100) for 5 min. Subsequently, cells were washed twice with 1 × PBS and blocked in PBS-T (1 × PBS and 0.05% Tween-20) containing 2% bovine serum albumin (BSA) and 10% milk for 45 min. Cells were then incubated with the primary antibody diluted in 1 × PBS containing 2% BSA and 10% milk at room temperature for 2 h. Coverslips were washed three times for 5 min with PBS-T before incubation (1 h) with the appropriate secondary antibodies conjugated to fluorophores (Alexa Fluor 488 or Cy3). After three washes of 5 min with PBS-T, cells were stained with 4',6-diamidino-2-phénylindole (DAPI; 0.5 µg/ml, MilliporeSigma, #D9542), and the coverslips were mounted using slow-fade mounting medium (Thermo Fisher Scientific, #S36936). When indicated, cells pulse-labeled with EdU (10 μM for 30 min) were labeled using the Click-iT EdU Alexa Fluor 594 Imaging Kit (Thermo Fisher Scientific, #C10639) following the fixation and permeabilization steps, according to the manufacturer’s protocol. The immunofluorescence protocol was then performed as described above. Images were captured using a Leica DMi8 THUNDER microscope. Quantitative image-based cytometry Using a Leica DMi8 THUNDER microscope with a Leica HC PL APO × 20/0.80 objective, 25 to 40 individual images were taken for each sample. The intensity signals of p65, γH2AX, geminin, KAP1-pS824, and DAPI in individual cells were quantified using the CellProfiler software. The intensity levels (arb. units) of γH2AX, KAP1-pS824, and geminin were quantified in DAPI-stained nuclei, and the intensity levels were normalized between the different images by subtracting the background signal from the nuclear intensity signal. Relative nuclear levels of p65 were calculated by subtraction of p65 cytoplasmic intensity (mean intensity measured from the extension of the nucleus) from its nuclear intensity (mean intensity measured from the DAPI-stained area). The relative nuclear levels of p65 or the nuclear intensity of KAP1-pS824 were color-coded into three groups: ‘no or low’, ‘medium’, and ‘high’ intensity. The threshold between the ‘no or low’ and ‘medium’ groups was set to include 97%–98% of the untreated cells in the ‘no or low’ group. The threshold between the ‘medium’ and ‘high’ groups was defined to label between 20% and 40% of positive cells in the ‘high’ group. The threshold between the ‘medium’ and ‘high’ groups was adapted in different experiments to best visually reflect the differences in p65 or KAP1-pS824 intensity levels between samples. Dot blot graphs were generated using GraphPad Prism software by randomly selecting 500 cells (for single-target dot plot graphs) or 5000 cells (for dot plot graphs quantifying p65, KAP1-pS824, geminin, γH2AX, and DAPI or geminin intensity levels). RNA interference siRNA transfections were performed through reverse transfection with Lipofectamine RNAiMAX (Thermo Fisher Scientific, #13 778 150). siRNAs were purchased from Thermo Fisher Scientific (Silencer Select siRNA). Cells were transfected with siRNA for 48 h before the indicated treatments (20 nM). The sequences of the siRNAs used in this study are listed in Supplementary Table S2. CRISPR–Cas9 knockout cells TDP1 and PARP1 CRISPR-Cas9 knockout RKO cell lines were performed through transfection with Lipofectamine CRISPRMAX (Thermo Fisher Scientific, #CMAX00008) of TrueGuide Synthetic CRISPR gRNA and TrueCut Cas9 Protein v2 (Thermo Fisher Scientific, #A36497), according to the manufacturer’s instructions. PARP1 KO in RKO cells was generated using two independent gRNAs. PARP1/TDP1 double KO cells were generated by knocking out TDP1 in PARP1 KO cells. The gRNAs used in this study are listed in Supplementary Table S3. Quantitative reverse-transcription polymerase chain reaction Total RNA was extracted from cells using Quick-RNA MiniPrep Kit (Zymo Research, #R1055), according to the manufacturer’s instructions. Following extraction, total RNA was reverse transcribed using the High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific, #4368814). The resulting cDNA was analyzed by real-time quantitative polymerase chain reaction (qPCR) using SYBR Green (PowerUp SYBR Green Master Mix, Thermo Fisher Scientific, #A25742) in the QuantStudio 3 Real-Time polymerase chain reaction (PCR) detection system (Thermo Fisher Scientific). For each sample tested, the levels of indicated mRNA were normalized to the levels of ACTB messenger RNA (mRNA). To plot multiple genes with different expression levels on the same graph, the mRNA levels of each gene were normalized to its highest expressed condition. The sequences of the PCR primers used in this study are listed in Supplementary Table S4. Neutral comet assay Neutral comet assay was performed to assess DNA DSBs using the Single–Cell Gel Electrophoresis. RKO cells were treated with camptothecin (CPT; 5 µM) in the presence or absence of PARP inhibitor (talazoparib, 2.5 µM). Cells were collected, embedded in 1% low-melting-point agarose, and placed on microscope slides. The slides were immersed in cold lysis solution (2.5 M NaCl, 100 mM ethylenediaminetetraacetic acid, 10 mM Tris, 10% diméthylsulfoxyde (DMSO), 1% Triton X-100, pH 10) for 1 h at 4°C. Electrophoresis was performed in 1× Tris-borate-EDTA (TBE) buffer at 24 V (0.8 V/cm) for 14 min at 4°C. Following electrophoresis, DNA was stained with SYBR Safe. Images were acquired using a Leica DMi8 THUNDER microscope. A total of 200 cells per condition were analyzed, and DNA damage was quantified as Olive tail moment using CometScore software, version 2.0.0.38. RNA-sequencing and data analysis RKO cells were treated with CPT (2.5 µM) with or without PARPi (talazoparib, 2.5 µM) and ATM inhibitor (ATMi ;10 µM) for 10 h, and total RNA was extracted using the Quick-RNA MiniPrep Kit (Zymo #11-328). Poly(A) RNAs from five micrograms of total RNA for each sample were then selected using NEBNext Poly(A) mRNA magnetic isolation module (#E7490L) and were then used to construct the sequencing library using Invitrogen Collibri Stranded RNA Library Prep Kit (Thermo Fisher Scientific, #A39114096), according to the manufacturer’s protocols. Libraries were amplified using 12 PCR cycles. Library size distributions were measured using a BioAnalyzer and quantified through qPCR. Libraries were sequenced on a NovaSeq X Plus platform. For RNA-sequencing data analysis, the fastq files were mapped on the human genome (GRCh38.114) using the STAR algorithm [21], and PCR duplicates were removed using Picard tools. Differential gene expressions were analyzed using R studio. For visual representation, only genes with more than 40 reads (mean between two biological replicates) in at least one treatment condition were included in the analysis. Statistical analysis Statistical analyses were performed with GraphPad Prism software version 11. The statistical test used for each panel was indicated in the figure legends. For bar plots from three independent experiments, the mean was highlighted as a bar, with error bars showing the standard deviation (SD). In scatter plots, the median was marked with a red line. Results PARP inhibitors specifically enhance NF-κB signaling triggered by TOP1 blockade Monitoring the activation of the NF-κB pathway in response to stress can be achieved using several complementary techniques: western blotting allows assessment of the activation state of key signaling components, such as p65 phosphorylation, IKK complex activation, or IκBα degradation; electrophoretic mobility shift assays can detect NF-κB DNA-binding activity; qPCR is commonly used to measure the expression of NF-κB target genes; or a luciferase reporter assay which involves introducing a plasmid with a luciferase gene under the control of NF-κB responsive element located upstream of a minimal promoter. However, these methods provide bulk measurements across entire cell populations and fail to capture cell-to-cell variability or differences related to cell cycle phase. As a result, changes in NF-κB activity within individual cells may be masked if they are offset by opposing changes in other cells within the same population. This highlights the need to monitor NF-κB activation at the single-cell level to fully understand its dynamic and heterogeneous regulation. Our laboratory has recently developed a quantitative image-based cytometry (QIBC) approach using high-content microscopy to assess NF-κB pathway activation at the single-cell level [7]. This method quantifies p65 subcellular localization by measuring nuclear-to-cytoplasmic distribution in individual cells, enabling precise detection of NF-κB activation dynamics across a cell population at single-cell resolution. To investigate the role of PARP1 in regulating NF-κB signaling in response to DNA damage, we performed an unbiased targeted screen using QIBC to monitor p65 nuclear localization after treatment with DNA-damaging agents that generate distinct types of DNA lesions, in combination with the PARP inhibitor talazoparib (PARPi). To pinpoint the DNA lesions that directly trigger p65 activation, we reasoned that assessing its nuclear recruitment at early time points (1–4 h) would capture the immediate regulatory mechanisms while minimizing confounding effects from broader DNA damage responses, such as IFN signaling, which can indirectly activate NF-κB at later time points (24–72 h) [22]. From this screen, we found a significant increase in p65 nuclear localization after PARPi treatment, specifically following exposure to topoisomerase 1 (TOP1) inhibitor CPT (Fig. 1A and B) [23]. Notably, this activation occurred across multiple CPT concentrations (low or high) (Supplementary Fig. S1A) and was rapid, within 2–4 h of CPT and PARPi treatment (Fig. 1A and B), consistent with acute NF-κB stimulation. In contrast, treatment with etoposide (ETP), ionizing radiation (IR), actinomycin D (ActD), or UV irradiation also led to nuclear accumulation of p65, but PARP inhibition had no significant effect (Fig. 1B and Supplementary Fig. S1B). However, treatment with 5-fluorouracil (5-FU), mitomycin C (MMC), or aydroxyurea (HU) failed to induce p65 activation within this time frame, regardless of PARPi treatment (Fig. 1B), even though all three treatments induced significant levels of DNA damage, as measured by γH2AX levels (Supplementary Fig. S1C). PARP inhibition only enhanced γH2AX levels after CPT treatment and not with other DNA-damaging agents (Supplementary Fig. S1C), indicating a specific role for PARP1 in protecting cells from DNA break formation in response to TOP1 blockade. Using the neutral comet assay, we found that the increase in γH2AX levels following CPT treatment, alone or in combination with PARP inhibition, correlated with an increase in comet tail moment (Supplementary Fig. S1D), demonstrating the formation of DSBs induced by these treatments. We then validated the CPT-induced increase in p65 nuclear localization using a second PARP inhibitor, olaparib (PARPi#2), as well as in the nontumorigenic human mammary epithelial cell line MCF10A, which also triggers NF-κB signaling in response to DNA damage (Fig. 1C and D) [7]. In addition, combining PARPi with topotecan (TPT), another TOP1 inhibitor [23], further enhanced p65 activation compared to TPT alone (Fig. 1E), confirming the results obtained after CPT. Importantly, depletion of TOP1 prior to treatment with CPT and PARPi impeded the activation of p65 (Fig. 1F and Supplementary Fig. S2A), further demonstrating that PARPi promotes the nuclear localization of p65 in response to TOP1 blockade by CPT, rather than as a consequence of an off-target effect of CPT treatment. Taken together, these results show that PARP inhibition increased NF-κB response to TOP1 inhibition specifically by raising the levels of DSBs. PARP1 inhibitors enhanced p65 activation following TOP1 blockade. (A) Representative immunofluorescence images of p65 localization in RKO cells treated with DMSO (control, CTL), CPT (5 µM), or CPT + PARPi (talazoparib, 2.5 µM) for 2 h. (B) Levels of nuclear p65 intensity relative to cytoplasmic p65 intensity were measured by immunofluorescence in RKO treated with CPT (5 µM), IR (10 Gy), ActD (5 µM), UV (10 mJ/cm2), ETP (50 µM), MMC (3 µM), HU (2 mM), or 5-FU (10 µM) and PARPi (2.5 µM) for the indicated time. Red lines indicate the median. n = 500 cells. (C, D) Quantification of p65 nuclear relative intensity in RKO cells treated with CPT (5 µM) (C) or MCF10A cells treated with CPT (2.5 µM) (D) and/or PARPi#2 (olaparib, 10 µM) for 2 h. Red lines indicate the median. n = 500 cells. (E) p65 nuclear relative intensity was quantified by immunofluorescence in RKO cells treated with Topotecan (TPT, 5 µM) in the presence or the absence of PARPi (2.5 µM). Red lines indicate the median. n = 500 cells. (F) RKO cells were transfected with the indicated siRNA for 48 h, then were treated with CPT (5 µM) and PARPi (2.5 µM) for 2 h. Relative levels of p65 in the nucleus were measured by immunofluorescence. Red lines indicate the median. n = 500 cells. All P-values were calculated with a two-tailed Mann–Whitney U test. To further investigate the impact of PARP inhibition on CPT-induced NF-κB signaling, we monitored IκBα, which sequesters p65 in the cytoplasm [2, 24]. CPT treatment decreased IκBα levels, and this reduction was further enhanced by co-treatment with PARPi (Supplementary Fig. S2B), while PARPi treatment upon ETP treatment did not impact the degradation of IκBα (Supplementary Fig. S2C). These results are consistent with the increased nuclear localization of p65 observed under these conditions (Fig. 1B). IκBα degradation is mediated by the IKK complex (IKKα, IKKβ, and IKKγ [NEMO], which is activated downstream of TRAF6 in response to DNA damage [5, 13, 25]. Accordingly, CPT-induced phosphorylation of IKKα/β was further increased in the presence of PARPi (Supplementary Fig. S2B). Moreover, depletion of TRAF6 or NEMO abolished p65 activation following CPT treatment, either alone or in combination with PARPi (Supplementary Fig. S2D and E). Collectively, these findings demonstrate that PARP1 regulates NF-κB activation in response to TOP1-induced DNA damage through the TRAF6-IKK signaling axis. PARP inhibitor triggers a replication-independent ATM-mediated NF-κB activation after TOP1 blockade We next investigated the spatial regulation of the NF-κB pathway by PARP1 in response to TOP1 inhibition throughout the cell cycle. Along with monitoring p65 nuclear localization, we first quantified DNA damage levels using γH2AX as a marker of DSBs and replication stress [26], and measured DNA content by DAPI staining. Following CPT treatment, p65 nuclear localization was detected exclusively in S-phase cells with high levels of DNA damage (γH2AX-positive) (Fig. 2A and B). In contrast, PARPi treatment led to p65 activation across all cell-cycle phases, no longer restricted to S phase (Fig. 2B). We further validated this result by quantifying p65 activation in relation to geminin and EdU levels. Geminin was used as a marker of cell-cycle progression, as its expression gradually increases from early S phase through G2 phase [27]. In contrast, EdU incorporation occurs exclusively during DNA replication [28]. Following CPT treatment, p65 nuclear localization was primarily detected in EdU-positive cells and cells with intermediate geminin levels, corresponding to S-phase cells (Fig. 2C and Supplementary Fig. S3A). By contrast, PARP inhibition induced p65 nuclear localization across the entire cell population, irrespective of geminin or EdU intensity (Fig. 2C and Supplementary Fig. S3A). Notably, the increased p65 nuclear localization occurring in G1 and G2 cells correlated with elevated γH2AX levels (Fig. 2A and B, and Supplementary Fig. S3B), indicating that PARP inhibition enhances CPT-induced DNA damage and thereby promotes NF-κB activation. Ongoing transcription has previously been shown to promote DSB formation in response to TOP1 inhibition in nonreplicating cells [29–32]. To determine whether p65 activation in nonreplicating cells is transcription-dependent, we treated cells with the transcription inhibitor DRB (5,6-dichlorobenzimidazole 1-β-D-ribofuranoside) in combination with CPT or CPT + PARPi. DRB treatment suppressed p65 activation in G1 and G2 phase cells but had no effect in S-phase cells, where p65 activation occurs through a replication-dependent mechanism (Fig. 2D). These results indicate that, in nonreplicating cells, p65 activation is driven by transcription occurring in close proximity to unrepaired single-strand breaks (SSBs) caused by TOP1 inhibition, consistent with previous studies [29–32]. ATM-dependent p65 activation following PARP1 and TOP1 inhibition. (A) Representative Immunofluorescence of p65 localization and γH2AX activation in RKO cells treated with CTL or CPT (5 µM), in the presence or the absence of PARPi (2.5 µM) for 2 h. (B) Quantification of p65 nuclear relative levels, γH2AX intensity, and DNA content (arb. units) in 5000 RKO cells treated as in panel (A), using QIBC analysis. Each cell was color-coded according to the intensity of p65 staining in the nucleus compared with its cytoplasmic staining. The green area indicates the population of γH2AX-negative cells. (C) Quantification by QIBC of p65 nuclear levels, EdU intensity, and DNA content in 5000 RKO cells treated with CPT (5 µM) for 2h, in the presence or the absence of PARPi (2.5 µM) for 2 h. (D) Levels of nuclear p65 intensity relative to cytoplasmic p65 intensity were measured by immunofluorescence in RKO cells treated with CPT (5 µM) with or without PARPi (2.5 µM) for 2 h. When indicated, cells were pretreated with DRB (100 µM) for 1 h prior to CPT treatment. n = 5000 cells. (E, F) QIBC of RKO cells stained for γH2AX, DNA content, and p65 (E) or KAP1-pS824 (F). Cells were treated with 5 µM of CPT and/or 2.5 µM of PARPi for 2 h in the presence or absence of ATMi (10 µM). n = 5000 cells. Each cell was color-coded according to the intensity of p65 staining in the nucleus compared with its cytoplasmic staining (E) or according to the nuclear intensity of KAP1-pS824 (F). The green area indicates the population of γH2AX-negative cells. We then examined whether p65 nuclear translocation was mediated by ATM. Treatment with an ATM inhibitor (ATMi) completely abrogated p65 activation induced by CPT alone or by CPT in combination with PARPi (Fig. 2E). Similar results were obtained following ATM depletion by siRNA (Supplementary Fig. S4A and B). In contrast, inhibition of ATR or DNA-PK had no effect on p65 nuclear localization (Supplementary Fig. S4C), confirming that ATM activity is specifically required for NF-κB activation in response to TOP1 inhibition. We then monitored phosphorylation of the ATM target KAP1 at Serine 824 (KAP1-pS824) as a readout of ATM kinase activity [33]. PARPi treatment significantly increased KAP1 phosphorylation levels after CPT treatment (Supplementary Fig. S4D), and this effect was suppressed by ATMi (Supplementary Fig. S4E). Notably, PARPi induced KAP1 phosphorylation specifically in G1 and G2, correlating with the increased levels of γH2AX, whereas after CPT alone, KAP1 phosphorylation occurred exclusively in the S phase (Fig. 2F). Consistently, ATMi partially decreased γH2AX levels in G1 and G2 (Fig. 2E and Supplementary Fig. S4F), which is also a target of ATM after TOP1 inhibition [34–36]. Collectively, these results indicate that PARP inhibition promotes the formation of TOP1-induced DNA breaks in G1/G2 cells, leading to ATM activation and subsequent NF-κB pathway activation. Because DNA damage can also activate p65 through IFN signaling independently of the canonical NF-κB pathway mediated by ATM-TRAF6, we examined IRF3 activation. Similar to p65, IRF3 translocates from the cytoplasm to the nucleus upon activation downstream of the cGAS-STING pathway [11, 12]. However, we did not observe nuclear recruitment of IRF3 following CPT or CPT + PARP inhibitor treatment, in contrast to poly(I:C) transfection, which is known to induce an IFN response and cause IRF3 nuclear localization (Supplementary Fig. 5A and B) [37, 38]. Consistent with this observation, depletion of either cGAS did not impair p65 activation following CPT treatment, with p65 levels comparable to those observed in control siRNA-treated cells (Supplementary Fig. S5C and D). Together, these results support the model that early p65 activation in response to DNA damage occurs through the canonical ATM–TRAF6–NF-κB signaling pathway rather than through IFN-mediated signaling. Two distinct mechanisms drive ATM-dependent NF-κB activation following TOP1 inhibition We then sought to understand the shift in NF-κB activation from S-phase-restricted in response to TOP1 inhibition to G1- and G2-phase activation when PARP1 is inhibited. CPT does not block TOP1’s ability to cleave DNA but instead traps TOP1 on DNA [39, 40]. During cleavage, the active-site tyrosine of TOP1 becomes covalently linked to the DNA backbone, forming the TOP1 cleavage complex (TOP1cc) [39, 40]. CPT binding prevents the religation step, leading to the accumulation of DNA SSBs [39, 40]. If not rapidly repaired by the base excision repair (BER) pathway, these SSBs are converted into DSBs during S phase when replication forks encounter them and collapse [39, 40]. Replication fork collapse following CPT treatment can be prevented by blocking DNA synthesis with HU or aphidicolin (APH) [41, 42]. Consistently, HU and APH treatment strongly reduced p65 nuclear translocation (Fig. 3A), while γH2AX levels remained unchanged, as both stalled and collapsed forks induce γH2AX [43, 44]. In striking contrast, the high levels of nuclear p65 localization monitored after combined treatment with CPT + PARPi were unaffected by either HU or APH (Fig. 3A), indicating that after PARP1 inhibition, CPT-induced NF-κB activation occurs independently of ongoing DNA replication across all cell cycle phases, including S-phase. These findings suggest that PARP inhibition uncouples TOP1 inhibitor-induced NF-κB activation from S-phase-specific replication stress, enabling DNA damage signaling and ATM-dependent NF-κB activation across all cell cycle phases. PARP inhibitor and TDP1 loss induce replication-independent activation of p65 after TOP1 blockade. (A) Top: Model highlighting how HU and APH block DNA replication fork collapse caused by TOP1 inhibition. Bottom: Quantification of p65 nuclear levels, γH2AX intensity, and DNA content (arb. units) of 5000 RKO cells treated with CTL, CPT (5 µM), or CPT + PARPi (2.5 µM) for 2 h. When indicated, cells were pretreated with HU (2 mM) or APH (1 µg/ml) for 5 min before treatment. (B) Top: Model showing TOP1cc degradation and removal by TDP1. Bottom: QIBC analysis of relative p65 nuclear levels, γH2AX intensity, and DNA content (arb. units) of RKO WT or TDP1 KO cells treated with or without 5 µM of CPT for 2 h. n = 5000 cells. (C–E) Quantification of p65 nuclear relative intensity in RKO WT or TDP1 KO cells treated with CPT (5 µM) and ATMi (10 µM) (C), HU (2 mM, pretreatment 5 min) (D), or PARPi (2.5 µM) (E) for 2 h. Red lines indicate the median. n = 500 cells. All P-values were calculated with a two-tailed Mann–Whitney U test. We next investigated how TOP1 inhibition triggers p65 activation, independent of replication fork collapse, in the absence of PARP1 activity. The initial step in repairing TOP1-induced damage is the removal of trapped TOP1cc from DNA by tyrosyl–DNA phosphodiesterase 1 (TDP1) [35, 39, 45]. TDP1 cleaves the covalent bond between the TOP1 active-site tyrosine and the 3′ DNA end, enabling subsequent repair of the SSBs. We found that removal of TDP1 increased p65 nuclear translocation across all cell cycle stages following CPT treatment, mimicking the effects of PARP1 inhibition (Fig. 3B and Supplementary Fig. S6A and B). Additionally, CPT-treated TDP1 KO cells showed increased γH2AX levels in G1 and G2 phases, along with elevated ATM activation, as indicated by KAP1 phosphorylation (Fig. 3B, and Supplementary Fig. S6C and D), again recapitulating the effects observed with PARPi after CPT. Moreover, ATM inhibition fully suppressed p65 activation in CPT-treated TDP1 KO cells (Fig. 3C), indicating that enhanced ATM signaling in TDP1-deficient cells also drives NF-κB pathway activation, similar to PARPi. Furthermore, as observed after PARPi treatment, blocking ongoing replication with HU or APH after CPT did not affect the nuclear localization of p65 in TDP1 KO cells (Fig. 3D and Supplementary Fig. S6E), suggesting that fork collapse is not required for NF-κB activation in TDP1 KO cells. Importantly, treating TDP1 KO cells with PARPi did not further increase p65 activation (Fig. 3E), demonstrating that PARPi and TDP1 depletion are epistatic in promoting CPT-induced NF-κB activation. Taken together, these results reveal two distinct modes of NF-κB activation in response to TOP1 inhibition: a replication-dependent pathway, where unrepaired SSBs caused by TOP1 inhibition lead to fork collapse and ATM activation, resulting in NF-κB activation specifically in S-phase cells; and a replication-independent pathway, which occurs when cells are treated with a PARP inhibitor or when TDP1 is depleted, while still relying on ATM for NF-κB activation in all cell cycle phases. PARP1 trapping, not enzymatic inhibition, causes NF-κB activation after TOP1 blockade To investigate how PARP inhibitors enhance p65 activation after TOP1 inhibition, we asked whether PARP1-mediated PARylation is required to block p65 nuclear localization. First, cell treatment with a PARG inhibitor (PARGi), which prevents PAR chain degradation [46, 47], did not alter p65 nuclear localization after CPT treatment (Supplementary Fig. S7A), indicating that elevated PAR levels do not influence NF-κB activation upon TOP1 inhibition. Second, PARP1 KO cells did not phenocopy the effect of PARP inhibition (Fig. 4A and B). Indeed, PARP1 depletion did not increase p65 nuclear localization following CPT treatment. Importantly, we found that the depletion of PARP1 also prevented p65 nuclear localization when treated with CPT in combination with PARPi (Fig. 4A and B and Supplementary Fig. S7B), as well as the increase in DNA damage levels observed in the G1 and G2 phases (Fig. 4A and B), suggesting that PARP1 enzymatic activity is not required to protect nonreplicative cells from TOP1 inhibitors. Together, these results demonstrate two key points. First, since PARP1 KO cells suppressed p65 activity after CPT and PARPi treatment, it implies that PARPi enhances p65 activation after TOP1 inhibition specifically through PARP1, rather than via other PARPs, such as PARP2, which PARPi also targets [48]. Second, because PARP1 depletion does not mimic the effect of PARPi, PARP1-mediated PARylation at SSBs is not required to prevent p65 activation after DNA damage. Instead, these findings suggest that PARP1 trapping at DNA breaks caused by PARPi, rather than inhibition, is necessary to promote p65 activation in response to TOP1 blockade. PARP inhibitors cause noncovalent trapping of PARP1 on DNA by stabilizing the interaction between PARP1 and DNA, preventing its release from the DNA and the repair of the SSB [49–51]. Consistent with this model, treatment with PJ34, a PARP inhibitor with minimal PARP1-trapping activity [52], [53], did not significantly enhance CPT-induced p65 nuclear localization, in contrast to talazoparib, which has strong PARP1-trapping activity (Fig. 4C and D) [49, 54], despite both inhibitors showing similar inhibition of PARP1 catalytic activity (Supplementary Fig. S7C). Therefore, these findings indicate that the removal of PARP1 from SSBs, similar to the removal of TOP1cc by TDP1, is a critical step in promoting repair and preventing the activation of the NF-κB pathway. PARP1 trapping promotes p65 activation after TOP1 inhibition. (A) RKO WT or PARP1 KO cells were treated with CPT (5 µM) and/or PARPi (2.5 µM) for 2 h. Relative levels of p65 in the nucleus were measured by immunofluorescence. Red lines indicate the median. n = 500 cells. (B) Relative intensity of p65 nuclear levels, γH2AX intensity, and DNA content were then analyzed by QIBC in WT or PARP1 KO cells treated as indicated. n = 5000 cells. (C) Quantification of p65 nuclear relative intensity in RKO WT cells treated with CPT (5 µM) in combination or not with either talazoparib (PARPi, 2.5 µM) or PJ34 (PARPi#3, 10 µM) for 2 h. Red lines indicate the median. n = 500 cells. (D) QIBC of RKO WT cells treated as indicated, and stained for p65, γH2AX, and DNA content. n = 5000 cells. (E) Indicated cell lines were treated with CPT (5 µM) for 2 h, and p65 nuclear relative intensity was quantified by immunofluorescence. Red lines indicate the median. n = 500 cells. (F) RKO WT, PARP1 KO, TDP1 KO, or PARP1/TDP1 dKO cells were treated as indicated. p65 nuclear levels, γH2AX intensity, and DNA content (arb. units) were analyzed by QIBC. n = 5000 cells. All P-values were calculated with a two-tailed Mann–Whitney U test. Because both TDP1 KO cells and PARP1 trapping by PARPi exhibit similar phenotypes after CPT treatment, we next investigated whether PARP1 trapping affects TDP1 activity, thereby preventing DNA damage and p65 activation. We generated double-knockout (dKO) cells for PARP1 and TDP1 (Supplementary Fig. S7D) and monitored both p65 activation and DNA damage following TOP1 inhibition. While PARP1 knockout suppressed PARPi-induced p65 nuclear localization after CPT treatment, PARP1/TDP1 dKO cells still exhibited strong p65 activation, similar to WT cells treated with PARPi or TDP1 KO cells after CPT (Fig. 4E and F), suggesting that PARP1 is not required for TDP1’s ability to remove TOP1cc and instead acts downstream of TOP1cc removal by TDP1. Collectively, our results demonstrate that TDP1 and PARP1 function independently but converge to repair TOP1-induced DNA lesions, thereby preventing the accumulation of DSBs and subsequent ATM-mediated NF-κB signaling. PARP1 trapping promotes TOP1 inhibitor-induced inflammatory gene expression To assess how PARPi affects the cellular response to TOP1 inhibition by regulating inflammatory gene expression, we performed RNA-seq after CPT or CPT + PARPi treatment. Transcriptomic analysis of CPT-treated cells revealed a significant increase in the expression of various inflammatory genes, including CCL20, CXCL1, and CXCL8, which were further upregulated by PARPi treatment (Fig. 5A and Supplementary Data 1). The expression of these genes was strictly dependent on ATM, as ATM inhibition completely suppressed their expression (Fig. 5A–C). Importantly, we confirmed these results by measuring gene expression with qPCR after treating cells with TPT rather than CPT (Supplementary Fig. S8A) and across multiple cell lines (Supplementary Figs S8B–D). Note that not all similar inflammatory genes were expressed across all cell types, and we only illustrated those we found to be induced by PARPi after CPT treatment in an ATM-dependent manner. Then, to verify that the NF-κB pathway mediated the expression of these genes, we knocked down p65. Depletion of p65 completely abolished the expression of CXCL1, CXCL8, and CCL20 after both CPT treatment and CPT combined with PARPi (Fig. 5D and Supplementary Fig. S8E), demonstrating that NF-κB drives the expression of these genes. Finally, we investigated the effect of PARPi after ETP treatment, as previous studies have shown that PARP1 activation promotes p65-dependent gene expression following TOP2 inhibition [5, 13, 20]. Consistently, we observed a decrease in the expression of many inflammatory genes after treatment with ETP in combination with PARPi compared to ETP alone (Supplementary Fig. S8F), further highlighting that PARP1 inhibition-induced p65-targeted genes are specific to TOP1 blockade. Together, these findings establish that PARPi enhances inflammatory gene expression specifically following TOP1 inhibition via an ATM-dependent NF-κB-mediated pathway. PARPi enhances TOP1 inhibitor-induced inflammatory gene expression. (A) Volcano plot of differentially expressed genes in RKO cells treated with CPT (2.5 µM) for 10 h in the presence or the absence of PARPi (2.5 µM). Only genes induced after CPT treatment (≥2-fold change) were shown in the volcano plot. Genes induced more than 1.3-fold by PARPi following CPT treatment are highlighted in purple. Among this gene population, those whose expression decreased by 50% or more upon ATMi (10 µM) treatment are highlighted in pink. Data were derived from n = 2 biological replicates. (B) Heatmap illustrating gene expression fold change for two independent biological replicates in RKO cells treated with CPT (5 µM) for 10 h, with or without PARPi (2.5 µM) or ATMi (10 µM). Gene reads were normalized to the mean expression after CPT and PARPi treatment of both replicates. (C) Quantification of CCL20, CXCL1, and CXCL8 mRNA levels by reverse transcription quantitative PCR (RT-qPCR) in RKO cells treated with CPT (2.5 µM) for 10 h in the presence or the absence of PARPi (2.5 µM) and/or ATMi (10 µM). Mean values ± SD (n = 3). (D) RKO cells were transfected with a siCTL or a siRNA targeting p65, then were treated with CPT (2.5 µM) and PARPi (2.5 µM) for 10 h. CCL20, CXCL1, and CXCL8 mRNA levels were monitored by RT-qPCR. Mean values ± SD (n = 3). (E, G) CCL20, CXCL1, and CXCL8 mRNA levels were measured by RT-qPCR in RKO WT and/or TDP1 KO cells treated with CPT (2.5 µM), with or without ATMi (10 µM) (E) or PARPi (2.5 µM) (G) for 10 h. When indicated, cells were transfected with CTL or p65 siRNA 48 h before treatment. Mean values ± SD (n = 3). (H) RKO cells were treated with CPT (2.5 µM), PARPi (2.5 µM), and/or HU (2 mM) for 10 h. CCL20 and CXCL1 mRNA levels were monitored by RT-qPCR. Mean values ± SD (n = 3). All P-values were calculated with a two-tailed paired t-test. Next, we treated TDP1 KO cells with CPT and monitored gene expression via qPCR. Notably, these genes were further induced in the absence of TDP1 in an ATM- and p65-dependent manner (Fig. 5E and F), while PARPi treatment of TDP1 KO cells treated with CPT had no significant effect on their expression (Fig. 5G). These findings further demonstrate that PARP1 trapping and TDP1 depletion act epistatically to promote NF-κB activation induced by TOP1 inhibition. Conversely, treatment with HU, which blocks ongoing DNA replication, decreased CPT-induced expression of inflammatory genes (Fig. 5H). However, HU treatment did not alter their expression when combined with CPT and PARPi (Fig. 5H). Therefore, TOP1 inhibition triggers inflammatory gene expression through ATM and NF-κB via both replication-dependent and replication-independent pathways, with PARP inhibition specifically enhancing the replication-independent NF-κB response. Collectively, these findings demonstrate a role for PARP1 in promoting NF-κB-mediated inflammation when trapped on DNA lesions in response to TOP1 inhibition. Discussion In this study, we uncover a previously unrecognized role for PARP1 in regulating the NF-κB pathway in response to TOP1 blockade. We show that TOP1 inhibition activates NF-κB specifically in S-phase cells by promoting replication fork collapse and ATM activation. In contrast, TOP1 inhibition in combination with PARP inhibitors induces NF-κB activation across the entire cell population, independently of DNA replication. Mechanistically, we demonstrate that both replication-dependent and replication-independent NF-κB activation pathways rely on ATM. In nonreplicating cells, PARP inhibition enhances DNA damage by trapping PARP1 on DNA lesions, blocking repair, and causing DSBs. This, in turn, activates ATM-mediated NF-κB signaling and induces the expression of inflammatory genes. Together, our findings reveal a mechanism in which PARP1 activates the NF-κB pathway independently of its enzymatic activity or replication but through PARP1 trapping in response to TOP1 inhibitors, highlighting a unique therapeutic opportunity to induce inflammation in tumor cells. PARP1 is involved in NF-κB activation following DNA damage caused by IR or ETP treatment [5, 13, 19, 20]. However, the mechanism by which PARP1, a nuclear protein, regulates the p65/p50-IκBα complex present in the cytoplasm remains debated, with conflicting studies proposing different models of regulation, underscoring the need for further investigation. Indeed, previous reports have proposed that PARylation, mediated by PARP1 upon IR or ETP treatment in mouse cells, acts as a scaffold for the assembly of NEMO, PIASy, and ATM into a complex, leading to the degradation of IκBα and the activation of NF-κB [9, 19]. Another study proposed that PARP1 forms a complex with ATM and IFI16 to induce a noncanonical activation of STING, independent of cGAS, leading to NF-κB activation [13]. Additionally, a study using human cells has shown that neither PARP1’s enzymatic nor its DNA-binding activity is required for its NF-κB coactivator function [20]. Our data failed to detect an effect of PARP inhibitors on p65 nuclear localization following either ETP or IR treatment (Fig. 1B). Our QIBC method offers the advantage of directly monitoring NF-κB signaling without perturbing the cells, unlike extract-based assays that can disrupt nuclear-cytoplasmic compartmentalization. Previous studies have shown that PARP1 can act as a promoter-specific coactivator of NF-κB by directly interacting with p65 and p50 on the promoter of NF-κB target genes [55–59]. Although we did not observe any major change in p65 nuclear levels after ETP treatment in combination with PARPi, we found that cell treatment with a PARP inhibitor decreases inflammatory gene expression following ETP (Supplementary Fig. S8F). These findings support a model in which PARP1 activates NF-κB after this treatment not by controlling p65 localization, but by promoting transcription of p65 target genes, consistent with previous reports [55–59]. Therefore, PARP1 plays multiple roles in the regulation of NF-κB, and our data further stress the importance of dissociating the role of PARP1 in regulating NF-κB according to the type of DNA lesion, which may vary across cell lines, cell cycle stage, or other specific contexts. PARP1 has been implicated at multiple levels in the response to TOP1 inhibition [60]. First, PARP1-mediated PARylation of TOP1 facilitates TDP1 recruitment [45, 61], yet PARP1 activity is not directly required for TOP1cc cleavage [62]. Indeed, PARylation inhibits TDP1’s cleavage activity, as PAR polymers prevent the proteasomal degradation step of TOP1 that is required for making the phosphotyrosyl bond accessible to cleavage by TDP1 [62]. Consistently, our data showed that PARP1 depletion did not increase DNA damage after TOP1 inhibition in G1/G2 (Fig. 4B), contrary to TDP1 depletion (Fig. 3B), suggesting that PARP1 enzymatic activity may provide a fine-tuned regulatory mechanism but is not essential for TDP1-mediated removal of TOP1cc. Second, PARylation by PARP1 also plays an important role in recruiting BER factors, such as XRCC1, which serves as a scaffold that coordinates the recruitment of many repair proteins, including polymerase β, Ligase 3, and APLF [63]. However, we found that PARP1 activity in promoting BER was not necessary to prevent NF-κB activation, as only PARPi, but not PARP1 depletion, increased DNA damage levels. The absence of phenotypes in G1/G2 cells lacking PARP1 during TOP1 inhibition might be due to PARP2, which, though less active than PARP1 at DNA damage sites, could help maintain minimal PARylation at TOP1-induced lesions, sufficient to regulate TDP1 and BER in G1/G2 [64]. These findings contrast with PARP inhibitor treatment, which caused significant DNA damage only in wild-type cells when TOP1 was blocked, but not in PARP1 knockout cells. This indicates that replication-independent DNA breaks are specifically caused by PARP1 trapping at DNA lesions, whereas trapping of PARP2 by PARPi has little effect, as PARP1 depletion alone can prevent p65 activation in G1/G2 in response to TOP1 inhibition. A similar mechanism occurs with TDP1 depletion, where increased DNA damage results not from deficient repair but from the trapping of TOP1cc on DNA lesions [29, 65], analogous to the effect of PARP inhibitors. Therefore, PARP1 trapping might impact TDP1’s ability to remove TOP1cc, or it is the trapping of proteins at lesions caused by TOP1 inhibition that drives replication-independent DSB formation and activates ATM-mediated NF-κB signaling. It has been proposed that DSB formation in the absence of TDP1 or after PARPi treatment involves a transcription-related process triggered by TOP1cc, which blocks transcription and leads to R-loop structures targeted by nucleases that cause DSBs [29–32]. Consistently, we showed that cell pretreatment with a transcription inhibitor suppressed NF-κB activation in nonreplicative cells exposed to TOP1 inhibition in combination with a PARP inhibitor (Fig. 2D). PARP1 trapping at TOP1 lesions might act similarly by impeding ongoing transcription to increase R-loop-mediated DSBs, explaining the lack of DSB increase and p65 activation in PARP1 KO cells. This raises the question of how PARP1 or TDP1 trapping leads to DSB formation. Several enzymes, including MUS81, XPF, XPG, FEN1, and MRE11, are linked to DSB formation in TDP1 KO cells following TOP1 inhibitor treatment [29, 35, 65]. This suggests that several enzymes or pathways might redundantly process unrepaired DNA damage caused by TOP1. However, further studies are needed to elucidate the detailed step-by-step mechanism of DSB formation caused by unrepaired lesions from TDP1 loss or PARP inhibition. In summary, we identify an additional mechanism by which PARP1 regulates NF-κB signaling through ATM in response to DNA damage induced by TOP1 inhibition. This mechanism is independent of PARP1 enzymatic activity and instead occurs when PARP1 is trapped at DNA lesions, independently of DNA replication. Our data indicate that combining TOP1 and PARP inhibitors may not only potentiate cancer cell killing through increased DNA damage but also enhance inflammatory signaling to promote tumoral innate immunity. Supplementary Material