Bepotastine Sensitizes Ovarian Cancer to PARP Inhibitors through Suppressing NF-κB–Triggered SASP in Cancer-Associated Fibroblasts

This article is featured in Highlights of This Issue, p. 419

Treatment of advanced high-grade serous ovarian cancer is challenging due to the lack of durable therapeutic options following debulking surgery and chemotherapy (1). PARP inhibitors (PARPis) can trigger single-strand breaks, leading to enhanced double-strand breaks that require other repair mechanisms such as homologous recombination (HR) repair (2). Therefore, PARPis have been approved as maintenance therapy for ovarian cancer that harbors HR deficiencies, such as BRCA1/2 mutations, in a synthetic lethal manner (3–6). However, acquired resistance to long-term PARPi maintenance therapy is ubiquitous. Although resistance mechanisms in tumor cells, such as reversion mutations and epigenetic modifications, can explain some occurrences, other resistance mechanisms involving the tumor microenvironment (TME) remain unclear (7). In our previous study, we found that PARPi-challenged fibroblasts displayed a tumor-promoting activation phenotype through CCL5/NF-κB–dependent autocrine signaling, and targeting CCL5 partially reversed the activation phenotype and sensitized the tumors to PARPis (8). However, in addition to CCL5, PARPis also triggered other tumor-promoting cytokines in cancer-associated fibroblasts (CAFs). The nature of the PARPi-induced secretion profile, its direct effect on tumor resistance to PARPis, and the feasibility of targeting the secretion profile for improved killing require further assessment.

Cellular senescence is traditionally described as a state of generally irreversible cell growth arrest that involves multiple internal and external causes, including oncogenic activation, oxidative and genotoxic stress, and mitochondrial dysfunction (9). Senescent cells share some characteristics, including enhanced activity of senescence-associated β-galactosidase (SA-β-gal), increased p16 or p21 expression, persistent telomeric DNA damage, enlarged morphology, and secretion of inflammatory cytokines, growth factors, chemokines, and matrix metalloproteinases, collectively termed senescence-associated secretory phenotype (SASP; refs. 10, 11). Therapy-induced senescence (TIS) is common in tumors and is heterogeneous in the ability and mechanisms of senescent cells to be antiapoptotic, escape growth arrest, and affect therapeutic efficacy determined by the tissue of origin, nature of treatment, and damage duration (12). TIS can shape the tumor response to therapy in many ways, among which SASP is crucial and involved in both tumor-promoting and tumor-suppressing effects (12, 13). For example, CAFs exposed to radiation express SASP factors and promote resistance to neoadjuvant therapy in rectal cancer (14). In contrast, TIS can trigger SASP-dependent vascularization and endothelial activation, which facilitates drug uptake and T-cell infiltration, leading to therapeutic vulnerabilities in pancreatic cancer (15). Because TIS as well as SASP may play opposite roles in antitumor therapy depending on the context, more intensive research is needed to explore the mechanisms underlying the effect and to develop novel TIS- or SASP-targeted strategies for improved efficacy.

Exploring the mechanism and targeted therapies of PARPi-related TIS has the potential for better efficacy. Recent studies have identified reversible TIS in ovarian and breast cancer cells treated with PARPis, and the combination of PARPis and ABT-263 (a senolytic that selectively eliminates senescent cells) can curb tumor growth in ovarian and breast cell line–derived xenograft (CDX) models (16). In another study, olaparib-resistant prostate cancer cells overcame TIS, while CDK1 inhibition reversed resistance (17). These studies suggest a reversible and escapable TIS in PARPi-treated tumor cells that can be targeted for enhanced efficacy. However, whether PARPi-related TIS occurs in other cell populations and whether targeting PARPi-induced SASP works in combination remain unclear. As an important source of inflammatory factors in the TME, SASP in stromal cells can be a promising target for improving PARPi killing, but remains to be explored (18).

In this study, we revealed that PARPis induce tumor stromal senescence in a p21-dependent manner. Moreover, SASP within the CAFs impaired PARPi cytotoxicity. Through database screening, we identified that bepotastine reversed PARPi-related SASP in CAFs and improved cell killing by PARPis, which was mediated by NF-κB pathway inhibition. In the patient-derived xenograft (PDX) models, bepotastine significantly augmented the efficacy of PARPis without evident toxicity. Our results highlighted the previously unappreciated role of TIS within CAFs to fuel PARPi resistance and identified bepotastine, an approved H1-antihistamine, as a candidate agent to eliminate SASP in CAFs and potentiate PARPis.

The epithelial ovarian cancer cell lines OVCAR8 and OVCAR3 were purchased from ATCC (Rockville, MD) and cultured in RPMI1640 medium (Gibco). Primary ovarian cancer cells and CAFs were isolated and purified from the tumor tissues of patients with ovarian cancer enrolled in a clinical trial (NCT04507841; ref. 19) as previously described and cultured in DMEM/F-12 (1:1; Gibco). MRC5 was cultured with TGFβ (50 ng/mL) or condition media from ovarian cancer cells to obtain the transformed fibroblasts named MRC5-CAFs (20). All patients provided written informed consent at recruitment, and the study was conducted in accordance with the current version of the Declaration of Helsinki and was approved by the Ethics Committee of Tongji Hospital and Huazhong University of Science and Technology (Hubei, China). A list of clinical information of the patients and primary fibroblasts is provided in Supplementary Table S1. All growth media were supplemented with 1% penicillin/streptomycin (Thermo Scientific) and 10% FBS (Gibco). All the above-mentioned cell lines and primary cells were cultured in 5% CO₂ at 37°C. All cell lines were authenticated by STR DNA profiling analysis, and the cell lines were tested for Mycoplasma contamination every 3 months by Mycoplasma DNA fluorescent staining. Olaparib (S1060, CAS No. 763113-22-0) and bepotastine (S5940, CAS No. 125602-71-3) were purchased from Selleck (https://www.selleck.cn/). Niraparib (HY-10619, CAS No. 1038915-60-4) and ABT-263 (HY-10087, CAS No. 923564-51-6) were purchased from MCE (https://www.medchemexpress.cn/), and the chemical structure was shown as previously described (21).

SA-β-gal activity was determined using an SA-β-gal staining kit, according to the manufacturer's protocol (Beyotime, C0602). Briefly, sections of PARPi-treated tumor tissues or PARPi-treated cells cultured in 6-well plates were fixed and incubated with the staining working solution at 37°C for 12 to 24 hours. SA-β-gal–positive cells were identified as bluish-green stained cells and counted using an Olympus microscope. The experiments were performed in triplicates and repeated three times.

For intracellular protein detection, primary fibroblasts were lysed in RIPA lysis buffer (Beyotime) supplemented with a protease inhibitor mixture (Roche). For supernatant protein detection, a 100-kDa ultrafiltration centrifuge tube and a 30-kDa ultrafiltration centrifugal tube were successively used to concentrate the supernatant. The concentrated supernatants were lysed in RIPA lysis buffer. For each sample, 40 μg of protein was separated using SDS-PAGE, transferred onto a nitrocellulose membrane, and incubated with the following primary antibodies: p21 (ABclonal, A19094, 1:1,000), IL6 (Abcam, ab9324, 1:1,000), CCL5 (CST, No. 36467,1:1,000), p65 (CST, No. 8242,1:1,000), p-p65 (ser536; CST, No. 3031,1:1,000), HRH1 (ABclonal, A1422, 1:1,000), and β-actin (ABclonal, AC038, 1:10,000). Finally, horseradish peroxidase–linked secondary antibody (anti-gene) was added and detected using an enhanced ECL system (Pierce).

Total RNA was extracted from primary CAFs using the RNeasy kit (Qiagen) and subjected to reverse transcription using the TSINGKE TSK302S Goldenstar RT6 cDNA Synthesis Kit Ver.2 (TSK302S) according to the manufacturer's protocols. The cDNAs were further analyzed using the 2 × TSINGKE Master qPCR Mix in triplicate using the Bio-Rad CFX96 system with SYBR Green, and the relative mRNA expression levels were calculated. The primer sequences used are listed in Supplementary Table S2. The experiments were performed in triplicates and repeated three times.

Calcein was used to stain viable cells according to the protocol of the Calcein/PI Cell Viability/Cytotoxicity Assay Kit (Beyotime, C2015S) and images were acquired using an inverted fluorescence microscope (Olympus). Immunofluorescence of cells and formalin-fixed, paraffin-embedded (FFPE) slides was performed as previously described (8). Briefly, cells were washed with PBS, fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton-X100, and blocked with 1% BSA in PBS. The cells were then incubated with the primary antibody (γH2AX, Abcam, ab22551, 1:200) overnight at 4°C and with Alexa Fluor 488-conjugated secondary antibody for 1 hour, followed by DAPI staining for 5 minutes. For F-actin visualization, rhodamine phalloidin (R415, 1:100; Invitrogen) was stained for 30 minutes at room temperature without prior blocking or subsequent secondary antibody incubation. For double labelling immunofluorescence, the tumor sections were firstly incubated with anti-pan-CK (Abcam, ab7753), anti-γH2ax (CST, 9718), anti-vimentin (Abcam, ab230171), anti-p21 (Abcam, ab109520), anti–p-p65 (Abcam, ab76302, 1:100), anti-IL6 (Abcam, ab233706, 1:50), and anti-CCL5 (Novus, AF-278-NA, 1:100), and were subsequently incubated with Goat Anti-Rabbit IgG (Alexa Fluor 594) and Goat Anti-mouse IgG (Alexa Fluor 488).

Multiplex IHC was performed on FFPE sections of human ovarian cancer tissues. After deparaffinization in xylene, antigen retrieval, the tumor sections were incubated with the following primary antibodies: p21 (ABclonal, A19094, 1:200), pan-CK (Abcam, ab7753, 1:200), and αSMA (Abcam, ab7817, 1:200). Subsequently, Opal IHC Multiplex Assay kit (NEL821001KT, Akoya Biosciences) were used for the chromogenic detection of the above targets in human tissues. Images were acquired using the Vectra3 automated quantitative pathology imaging system (Akoya Biosciences). The experiments were performed in triplicates and repeated three times.

Five thousand tumor cells or 3,000 CAFs were seeded in 96-well plates per well, followed by treatment with serial concentrations of olaparib, niraparib, and bepotastine (200 nmol/L) or ABT-263 (1 nmol/L) in the presence or absence of conditioned media for 48 hours. Cell viability and growth were examined using a Cell Counting Kit-8 assay (Dojindo Laboratories; catalog No. CK04) according to the manufacturer's protocol. The experiments were performed in triplicates and repeated three times.

Cells seeded in 12-well plates were treated for 48 hours and harvested for further analysis. For cell-cycle analysis, live cells were fixed overnight with 70% ethanol and incubated with RNase A and propidium iodide for 30 minutes at room temperature. For cell death analysis, all cells, including dead cells floating in the culture media, were collected and stained using an apoptosis detection kit (BD Pharmingen, catalog No. 556547), according to the manufacturer's protocol. A minimum of 10,000 events were counted per sample using a CytoFLEX Flow Cytometer (Beckman Coulter) and analyzed using the FlowJo software. The experiments were performed in triplicates and repeated three times.

Animal experiments were approved by the Committee on Ethics of Animal Experiments at Tongji Hospital. Six-week-old female NOD/SCID mice were subcutaneously inoculated into the right flank with homologous recombination–deficient (HRD)-positive PDX tumors (3 × 3 mm). After a week, the mice were treated with placebo, PARPis (olaparib 50 mg/kg/day, niraparib 50 mg/kg/day by oral gavage) alone, bepotastine (2 mg/kg/day by oral gavage) alone, or a combination followed by tumor volume measurement. Thirty days later, all mice were sacrificed under anesthesia, and the tumors were dissected, weighed, and paraffin-embedded for further analysis.

Three-dimensional (3D) culture was performed as previously described, with some modifications (22). 1 × 10⁶ OVCAR8 and OVCAR3 were cultured in Matrigel, followed by treatment with PARPis and bepotastine in the presence or absence of PARPi-primed CAF CM for 72 hours. The organotypic models were harvested, fixed, and embedded in paraffin for hematoxylin and eosin staining and analysis.

One milliliter of blood was collected from each mouse and incubated for 30 minutes at room temperature. The serum was further centrifuged at 8,000 g for 8 minutes. The clear supernatant was collected and measured using a Murine ALT ELISA kit (mlbio, ml063179) and Murine AST ELISA kit (mlbio, ml058577) according to the manufacturer's protocol.

Control siRNA, siRNA-HRH1-1#,2#,3# and siRNA-p65-1# were purchased from RiboBio (China) and transfected into primary CAFs according to the manufacturer's protocol. Briefly, Lipofectamine 3000 reagent (Invitrogen) and siRNA were respectively incubated in Opti-MEM (Gibco) for 5 minutes, then the two mixtures were combined and incubated at room temperature for 30 minutes and added to primary CAFs. Western blotting was performed to determine the efficiency of siRNA knockdown.

All data are presented as the mean values ± SEM, based on at least three independent experiments. Student t test was used to assess the statistical significance between the two groups. One-way ANOVA followed by Tukey post hoc test was conducted to assess statistical significance among multiple groups. Statistical significance was set at P < 0.05.

GSE40595 and GSE164088 profiling data were downloaded as raw data from the Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/), and process data were obtained after interpretation, normalization, and log2 scaling using the R software. Gene set enrichment analysis (GSEA) was performed using the publicly available desktop application from the Broad Institute (http://www.gsea-msigdb.org/gsea/index.jsp).

SASP and NF-κB signatures were obtained from the Molecular Signatures Database (http://www.gsea-msigdb.org/gsea/msigdb/index.jsp) and previous studies (18, 23, 24), and are listed in Supplementary Table S3. Differential gene expression analysis between PARPi-treated and untreated CAFs was conducted for each specimen, and Venn diagrams were used to identify overlapping genes. These 24 genes are listed in Supplementary Table S4.

Single-sample gene set enrichment analysis (ssGSEA) was used to calculate the signature scores of NF-κB and SASP in GSE40595 using the ssGSEA module in GenePattern (https://www.genepattern.org/). The signature scores are listed in Supplementary Table S5.

The WEB-based GEne SeT AnaLysis Toolkit (http://www.webgestalt.org/option.php) was used to screen potential drugs targeting PARPi-induced SASP factors. Overrepresentation analysis was selected for enrichment analysis. The gene list of PARPi-triggered SASP factors was uploaded, and the GLAD4U database was selected for further drug screening. The rank scores of the screened drugs are presented in Supplementary Table S6.

To explore whether PARPi lead to TIS in ovarian cancer, paired human ovarian cancer sections before and after niraparib monotherapy were obtained from the NANT trial (NCT04507841), the first single-armed, phase II prospective study to evaluate the effectiveness and safety of niraparib as neoadjuvant therapy in newly diagnosed advanced HRD ovarian cancer (19). SA-β-gal staining showed that niraparib significantly increased senescence in ovarian cancer sections (Fig. 1A). Further multiplex IHC analysis showed that PARPi preferentially triggered p21 expression in αSMA+ stromal areas, suggesting p21-involved cell-cycle arrest in CAFs (Fig. 1B). Moreover, SA-β-gal staining in primary cells isolated from BRCA1-mutant tumors was more significantly elevated in the tumor stroma than in the tumor epithelia at 48 hours (Fig. 1C), implying a senescence-prone phenotype in CAFs. These results are consistent with previous reports that at least 3 to 6 days of PARPi exposure is required to induce senescence in tumor cells (16). To directly investigate CAFs in response to PARPis, purified primary CAFs and MRC5-CAFs were obtained as described in the Methods section and previous study (20) and were treated with PARPis. Significantly increased γH2AX foci, spread cell morphologic alternations, and G₂–M block were observed in primary CAFs (Fig. 1D and E; Supplementary Fig. S1A). Meanwhile, enhanced SA-β-gal staining was detected in both primary CAFs and MRC5-CAFs (Fig. 1F; Supplementary Fig. S1B). Moreover, GSEA analysis and heatmap of our previous RNA sequencing (RNA-seq) data of CAFs treated with PARPis (GSE164088) revealed enrichment of genes related to the p21-associated senescent phenotype (Fig. 1G and H; ref. 8). Consistently, p21 protein expression and γH2AX levels were elevated in PARPi-treated CAFs (Fig. 1I; Supplementary Fig. S1C). In addition, p21 knockdown in both primary CAFs and MRC5-CAFs using siRNA significantly attenuated PARPi-induced senescence (Fig. 1J; Supplementary Fig. S1D–S1F). Last but not least, ABT-263, known as a senolytic agent, downregulated PARPi-induced SA-β-gal staining in CAFs, accompanied by a decrease in cell density, indicating a specific clearance of senescent cells (Fig. 1K). Overall, these results indicate that PARPis induces senescence in CAFs in a p21-dependent manner.

SASP is the most commonly reported phenotype involved in TIS-induced resistance (18). GSEA of RNA-seq data of PARPi-treated CAFs revealed significant enrichment of the SASP signature in olaparib-treated CAFs (Fig. 2A). Furthermore, combined with RNA-seq and cytokine microarray data analysis, we determined that PARPis induced SASP in CAFs and identified candidate genes that harbor tumor-promoting potential (Fig. 2B; ref. 8). Previous studies have reported that NF-κB signaling regulates SASP gene expression (25). The NF-κB signature score showed a significant positive correlation with the SASP signature score in the microdissected ovarian cancer stromal profile (n = 31) of the GSE40595 dataset (Fig. 2C; ref. 26). Immunoblotting assays revealed enhanced p-p65 expression in PARPi-primed CAFs, whereas additional p65 knockdown (siRNA) reversed the PARPi-induced p-p65 expression and SASP expression profiles (Fig. 2D and E), suggesting that PARPi regulates the SASP in CAFs in an NF-κB–dependent manner. To further explore the effect of stromal SASP on PARPi resistance, supernatants were collected from si-NC and si-p65 CAFs with or without PARPi treatment for 72 hours and added to ovarian cancer cell lines to assess drug sensitivity (Fig. 2F). After 48 hours of exposure to PARPis, OVCAR8 and OVCAR3 cells cultured with CM from PARPi-primed CAFs exhibited evident less apoptosis and more resistance compared to those cultured with CM from DMSO-primed CAFs. Moreover, CM-induced resistance in PARPi-primed CAFs was reversed when p65 was knocked down (Fig. 2G–J; Supplementary Fig. S2), implying that PARPi-induced CAFs’ SASP plays a crucial role in tumor resistance. Interestingly, ABT-263 reversed PARPi-induced SASP and CM-mediated tumor resistance (Fig. 2K–N). Taken together, PARPis can induce SASP in CAFs via NF-κB signaling, which is sufficient to confer PARPi resistance in ovarian tumor cells.

Although PARPi-induced SASP is regulated by NF-κB, most drug and intervention strategies targeting the NF-κB pathway have been limited to preclinical studies. Therefore, reusing approved drugs could be an attractive strategy to counteract the SASP. Using the GLAD4U database, we screened a group of potential antagonists targeting PARPi-induced SASP genes, among which bepotastine, an approved H1-antihistamine (27), was predicted to be the most promising drug candidate (Fig. 3A). Considering the potential translational significance of the approved drugs, 200 nmol/L bepotastine was used for further experimental verification based on clinically relevant plasma concentration ranges (28). qRT-PCR analysis showed that the expression of SASP in CAFs augmented by PARPis was reversed by bepotastine (Fig. 3B and C). Moreover, while bepotastine decreased the baseline expression levels of IL6 or CCL5 in CAFs with inflammatory phenotype to some extent, it abrogated PARPi-triggered SASP factor expression more significantly (Fig. 3D).

Because bepotastine effectively inhibited PARPi-induced SASP in CAFs, we investigated whether bepotastine could sensitize tumors to PARPis. Bepotastine did not directly impair the viability of OVCAR8, either alone or in combination with PARPis (Fig. 4A and B). Similar results were observed for OVCAR3 (Supplementary Fig. S3). In contrast, bepotastine attenuated CAF CM-mediated tumor cells resistance to PARPis (Fig. 4C–F), indicating an indirect antitumor effect by targeting PARPi-triggered SASP. The same effect was observed in the 3D culture system of OVCAR8 and OVCAR3, where bepotastine significantly diminished the survival benefits provided by PARPi-primed CAF CM (Fig. 4G and H). These results suggest that bepotastine may be a promising drug for sensitizing tumor cells to PARPis by targeting SASP in CAFs.

To investigate the mechanism by which bepotastine inhibits the SASP of PARPi-primed CAFs, we first investigated whether bepotastine attenuates PARPi-triggered senescence or eliminates senescent CAFs. Intriguingly, bepotastine did not affect the G₂–M block or regulate cellular senescence in CAFs caused by PARPis (Fig. 5A and B). Similarly, bepotastine, alone or in combination with PARPis, did not alter p21 expression in CAFs (Fig. 5C). Meanwhile, bepotastine did not display a senescence-specific cytotoxic effect on CAFs (Fig. 5D). These results suggest that, unlike the mechanisms of action for other senolytic drugs, TIS modulation and senescent cell elimination are dispensable for bepotastine-driven downregulation of PARPi-induced SASP. Therefore, we investigated whether bepotastine affected NF-κB signaling, the driver of SASP in PARPi-primed CAFs. NF-κB activation is accompanied by the enhanced phosphorylation and nuclear translocation of p65. Importantly, immunoblotting also showed that bepotastine downregulated p-p65 expression activated by PARPis (Fig. 5E). Moreover, bepotastine inhibited the PARPi-triggered nuclear translocation of p65 in CAFs (Fig. 5F). These results confirmed that bepotastine significantly inhibited NF-κB signaling in PARPi-primed CAFs, but had no effect on the NF-κB pathway in untreated CAFs.

Because histamine content and its receptors are elevated in a variety of tumors (including ovarian cancer; ref. 29), to further explore whether the H1-histamine receptor participates in bepotastine-driven NF-κB signaling suppression, HRH1 in CAFs was knocked down using siRNA (Fig. 5G). p-p65 expression was not significantly different between HRH1 knockdown and control PARPi-primed CAFs treated with bepotastine, suggesting HRH1-independent regulation of the NF-κB signaling pathway (Fig. 5H). Taken together, we demonstrated that PARPis trigger the SASP of CAFs in an NF-κB–dependent manner and that bepotastine inhibits NF-κB activation and SASP induced by PARPis, providing rationale to leverage bepotastine as a promising adjuvant drug to potentiate PARPis (Fig. 5I).

In vivo studies were conducted in two HRD-positive ovarian PDX models to evaluate the antitumor effects of the combined bepotastine and PARPi therapy. For single agent, while bepotastine exhibited negligible antitumor efficacy, olaparib and niraparib could curb tumor growth compared with vehicles, but the tumor load was still high (Fig. 6A–F). The synergistic therapy using PARPis and bepotastine effectively controlled tumor growth and low tumor volume maintained until vehicle-treated tumors reached the endpoint (Fig. 6A–F). To assess the potential toxicity of the combination therapy, body weight, serum ALT, and AST levels were measured, and no significant weight loss or liver injury was observed (Supplementary Fig. S4). To better understand the improved antitumor efficacy of the combination therapy in PDX models, double labelling immunofluorescence of reserved humanized CAFs was performed in PDX tumor tissues as previously described (30) to validate the senescence-like and SASP phenotypes and NF-κB signaling in vivo. Enhanced γH2AX in tumor cells, increased p21, IL6, CCL5, and p-p65 expression in CAFs were observed in the PARPi monotherapy groups compared with the control group, suggesting that PARPi attenuated tumor proliferation and simultaneously triggered senescence, SASP, and NF-κB signaling activation in the tumor stroma. Meanwhile, the addition of bepotastine reversed SASP and NF-κB signaling activation and further suppressed tumor proliferation in the combination groups without affecting PARPi-induced senescence (Fig. 6G–I). Collectively, these results suggest that bepotastine sensitizes tumors to PARPis in HRD PDX models.

Although TIS is considered as an attractive target against PARPi resistance, most studies that explore the impact of TIS in PARPi treatment have mainly focused on tumor cells (16, 17, 31). Herein, we extended the investigation of PARPi-related TIS from tumor cells to stromal fibroblasts and discovered that PARPi-treated stromal fibroblasts exhibited a more pronounced and rapid senescent phenotype than tumor cells. PARPi-primed fibroblasts expressed increased levels of SASP factors and promoted tumor resistance. Bepotastine, an approved antihistamine, has been shown to reverse PARPi-induced SASP and sensitize PARPi in vitro and in HRD-positive PDX models by inhibiting NF-κB signaling. Taken together, our work highlights the importance of stromal TIS in PARPi resistance, and targeting SASP with an approved agent may be a promising strategy to overcome PARPi resistance.

Senescence in tumor cells shares the same definition as that in normal cells and is traditionally described as a state of irreversible cell growth arrest and suppressed proliferation, which supports senescence as a favorable response to antitumor therapies (32, 33). Therefore, TIS have been considered for cancer treatment. However, growing evidence shows that senescence in tumor cells can also be a crucial cell fate transition, in which tumor cells escape from cell-cycle blockade, maintain senescence-associated stemness (SAS), fuel more aggressive growth, and acquire senescence-related resistance (34, 35). For example, lung cancer cells overcome TIS and develop senescence-related chemoresistance via autophagy (36). Breast and lung cancer cells escape from TIS via SLC1A5-dependent glutamine intake (37). Ovarian cancer cells acquire cisplatin-induced SAS through NAMPT-mediated NAD biosynthesis (38). In addition to the SAS of tumor cells, the SASP of stromal cells within the TME also plays a crucial role in TIS-related resistance (39). Endothelial cells express SASP factors mediated by PI3K/Akt signaling and p38, leading to chemoresistance in breast cancer (40, 41). CAF-derived SASP factors promote chemoresistance in prostate cancer that involves WNT16B, β-catenin, and NF-κB signaling (42). Because TIS is heterogeneous and context-dependent in its mechanisms of tumor-promoting effects (43), SASP is also mediated by various mechanisms (25, 44–47). Thus, targeting TIS or SASP combined with PARPis in ovarian cancer may not simulate the mechanism and effect conditioned by a combination with traditional genotoxic agents and remains to be explored.

Reversible TIS in PARPi therapy provides a treatment window for further elimination of senescent tumor cells (16). However, it is unclear whether PARPi-induced senescence in nonmalignant cells antagonizes the therapeutic response to PARPi. Our previous study has revealed that CAFs are relatively insensitive to PARPis. Instead, PARPi triggers CCL5 autocrine signaling-dependent stromal fibroblast activation, which contributes to PARPi resistance (8). In this study, we found that PARPis caused more pronounced and rapid senescence and SASP in CAFs than in tumor epithelium. It seems contradictory that CAFs are less vulnerable to PARPi-induced apoptosis, but more susceptible to PARPi-induced senescence than tumor cells. First, senescence phenotype enables CAFs to evade apoptosis (48). Second, CAFs inherently exhibit a senescence-prone phenotype primed by stimulants such as TGFβ (49, 50). Third, genomic and epigenetic changes in malignant transformation endow tumor cells with limitless replicative potential and the ability to evade senescence (51).

Because the heterogeneity of CAFs has been widely recognized, both the stromal effect on tumor and response to therapy need to be revisited (52). The previously observed multiple response of CAFs could be attributed to cumulative effects from overlapping or even mutually exclusive subtypes of CAFs at various ratios. In our study, PARPi-induced senescent CAFs shared similar activated signal (NF-κB) and characters of secretion (IL6, CCL5 etc.) with inflammatory CAFs (iCAFs). In another study, senescent iCAFs were demonstrated to induce tumor resistance (14). These studies indicated an overlap between senescent CAFs and iCAFs. However, further evidence is still needed to prove this hypothesis.

Antihistamines are widely used as chemotherapy premedication to relieve and prevent chemotherapy-induced symptoms, such as nausea and vomiting (29). Retrospective studies have shown that antihistamine use is associated with improved prognosis in some cancer types, and the repurposing of these drugs as antitumor therapy has garnered considerable attention (53, 54). The antitumor mechanisms of antihistamines can be summarized in two ways. First, antihistamines interact with histamine receptors in certain cell types or inhibit the release of histamine by mast cells, subsequently inhibiting histamine-driven aggressive tumor phenotypes (29, 54–56). Second, antihistamines with cationic amphiphilic drug characteristics can directly cause lysosomal-associated cell death in tumor cells (53, 57). Recently, H1-antihistamine treatment was reported to enhance immunotherapy response by blocking histamine receptor H1 on macrophages (54). In this study, bepotastine, a second-generation H1-selective antihistamine, antagonized PARP-induced SASP in CAFs by inhibiting NF-κB signaling independent of the histamine H1 receptor. Moreover, bepotastine alone or in combination with PARPis did not directly affect tumor cell viability, whereas combined administration in vivo potentiated PARPis. These results suggest that bepotastine sensitizes tumors to PARPi therapy in a TME-dependent manner. Because SASP can modulate the immune microenvironment, the antitumor efficacy of PARPis plus bepotastine in immunocompetent mice remains unclear. In addition, although bepotastine seemed to suppress baseline SASP or inflammatory factor expression in proinflammatory CAF, it remains unclear if PARPi-specific mechanisms are involved in the inhibitory effect, and there is no evidence for the antitumor potential of bepotastine as monotherapy.

In summary, this study showed that PARPis can induce senescence and SASP in CAFs, which is distinct but highly complementary to its established role in promoting CAF activation. Targeting SASP by bepotastine can effectively increase tumor sensitivity to PARPis. Taken together, our results highlight that PARPi-induced senescence and SASP in CAFs play crucial roles in PARPi resistance. Bepotastine, an approved H1-antihistamine, may be used as a novel SASP inhibitor to sensitize PARPi therapy.

No disclosures were reported.

P. Jin: Conceptualization, resources, data curation, formal analysis, validation, visualization, methodology, writing–original draft, writing–review and editing. X. Li: Data curation, software, formal analysis, investigation, visualization, methodology. Y. Xia: Supervision, project administration. H. Li: Writing–review and editing. X. Li: Software, formal analysis, methodology. Z.-Y. Yang: Supervision, methodology. Z. Wang: Formal analysis, methodology. C. Xu: Methodology. T. Fang: Methodology. D. Zhou: Methodology. X. Xiong: Visualization, methodology. S.-Y. Wang: Methodology. S. Xu: Conceptualization, resources, methodology, writing–original draft. Q. Gao: Resources, supervision, funding acquisition, project administration.

This work was supported by grants from the National Key Research and Development Program (No. 2022YFC2704200 and No. 2022YFC2704205) and the National Science Foundation of China (grant No. 82072889 and No. 81802611).

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.