PCBP2 Mediates Olaparib Resistance in Breast Cancer by Inhibiting m⁶A Methylation to Stabilize PARP1 mRNA

PCBP2-induced suppression of m⁶A methylation increases PARP1 to promote DNA damage repair and confer resistance to olaparib in BRCA-mutated breast cancer, making PCBP2 a potential therapeutic target to enhance PARP inhibitor sensitivity.

Abstract

Base excision repair (BER), a critical pathway for repairing DNA single-strand breaks, is mediated by PARP, which plays a pivotal role in maintaining genomic stability. Targeting PARP with PARP inhibitors (PARPi) has emerged as an effective strategy for treating BRCA-mutated breast cancers characterized by homologous recombination deficiency. However, PARPi resistance remains a major challenge in the treatment of BRCA-mutated breast cancer. Using bioinformatics analysis and cellular-level experiments, we discovered that the RNA-binding protein PCBP2 contributes to resistance to the PARPi olaparib in BRCA-mutated breast cancer by increasing PARP1 expression via interference with the m⁶A methylation machinery. PCBP2 was upregulated in olaparib-resistant cells, and PCBP2 overexpression in BRCA-mutated breast cancer cells increased resistance to olaparib and enhanced cell proliferation under treatment. Mechanistically, PCBP2 directly interacted with PARP1 mRNA, inhibiting m⁶A methylation and stabilizing the mRNA. PCBP2-mediated upregulation of PARP1 enhanced DNA repair activity, contributing to olaparib resistance. Together, these findings unveil a mechanism by which PCBP2 upregulates PARP1 to promote olaparib resistance in BRCA-mutated breast cancer, indicating that targeting this pathway could represent a therapeutic strategy to overcome PARPi resistance in breast cancer.

Significance:

PCBP2-induced suppression of m⁶A methylation increases PARP1 to promote DNA damage repair and confer resistance to olaparib in BRCA-mutated breast cancer, making PCBP2 a potential therapeutic target to enhance PARP inhibitor sensitivity.

Graphical Abstract

Introduction

Breast cancer is a global health concern and has become the most common malignancy worldwide, accounting for one-quarter of all cancer cases in women (1). Despite improvements in breast cancer survival rates, distant metastasis and treatment resistance remain closely associated with poor prognosis (2). Whereas advances in surgery, radiotherapy, chemotherapy, and targeted therapies have improved outcomes, the prognosis for advanced-stage breast cancer remains poor (3). In particular, patients with specific genetic mutations, such as breast cancer gene 1 (BRCA1) or breast cancer gene 2 (BRCA2) mutations, face a higher risk of recurrence and have limited treatment options. Therefore, further investigation into the underlying biological mechanisms of breast cancer drug resistance is crucial for identifying novel therapeutic targets and improving patient survival.

BRCA1 and BRCA2 germline mutations are present in the majority of hereditary breast cancer cases, significantly increasing the risk of developing both breast and ovarian cancers (4). Carriers of BRCA1/2 mutations have a 50% to 85% lifetime risk of developing breast cancer, with approximately 70% of these cases classified as triple-negative breast cancer (TNBC; ref. 5). Furthermore, the prevalence of BRCA mutations varies across different breast cancer molecular subtypes. TNBC has been reported to have the highest prevalence of BRCA mutations at 15.4% (6), compared with 5% in hormone receptor–positive (HR⁺) breast cancer (7, 8) and 4% in human epidermal growth factor receptor 2 (HER2)–positive breast cancer (9). Additionally, 60% to 80% of BRCA1-mutant breast cancers are classified as TNBC, whereas more than 75% of BRCA2-mutated breast cancers are of the luminal subtype (5, 10).

PARP inhibitors (PARPi), based on the concept of synthetic lethality, have been used to target homologous recombination (HR)–deficient cancers caused by BRCA1/2 mutations (11). To date, the FDA has approved four PARPis for cancer treatment, namely olaparib, niraparib, rucaparib, and talazoparib (12, 13). BRCA1 and BRCA2 genes play essential roles in maintaining genomic stability and DNA repair, primarily through HR to repair DNA double-strand breaks (14). PARP1 is a critical enzyme involved in the repair of DNA single-strand breaks (SSB) via the base excision repair (BER) pathway. In cells harboring BRCA1 or BRCA2 mutations, HR is impaired, leading to a reduced capacity for double-strand break repair. PARPis block the BER pathway, further preventing SSB repair, resulting in accumulation of DNA damage and ultimately inducing cell death (15). However, similar to traditional chemotherapeutics, PARPis also face the challenge of acquired resistance during clinical use. Despite their clinical efficacy, resistance to PARPis in HR-deficient (HRD) tumors can occur through several well-characterized mechanisms. These include reversal of the HRD phenotype, such as through secondary mutations restoring BRCA function or the loss of 53BP1, which allows HR to resume; stabilization of the DNA replication fork, preventing degradation by nucleases, such as MRE11; increased PARylation activity through alterations in poly(ADP-ribose) (PAR) metabolism, such as poly(ADP-ribose) glycohydrolase (PARG) loss; and drug efflux via upregulation of transporter proteins (12). Additionally, chromatin remodeling has been recognized as a contributor to PARPi resistance, facilitating DNA repair access and limiting drug efficacy. These mechanisms highlight the complexity of PARPi resistance and underscore the need to fully identify the molecular players involved in this process. Therefore, investigating novel mechanisms of resistance and identifying effective combination strategies remain of significant clinical value.

Recent research has found a link between drug resistance and aberrant expression of RNA-binding proteins (RBP), suggesting that RBPs could be potential targets for developing therapies to overcome resistance to targeted treatments (16). RBPs are key regulators in posttranscriptional gene regulation, responsible for controlling various stages of the mRNA life cycle (17). RBPs typically function by interacting with specific sequence motifs within target mRNAs, providing a regulatory basis for processes such as splicing, 3′ untranslated region (UTR) processing, mRNA stability, and translation (18). Poly(rC)-binding protein 2 (PCBP2) is a sequence-specific RBP that interacts with poly(C) motifs (19). Due to its poly(C)-binding ability, PCBP2 performs multiple functions, including transcriptional and posttranscriptional regulation, such as pre-mRNA splicing, mRNA stabilization, and translational control. PCBP2 contains three K homology (KH) domains (PCBP2 KH1, PCBP2 KH2, and PCBP2 KH3), separated by intervening sequences of variable length (20). The regulatory function of the KH domains at the posttranscriptional level was initially identified through studies of human α-globin mRNA. PCBP2’s KH domains bind with high affinity to the pyrimidine-rich 3′ UTR of α-globin mRNA, enhancing mRNA stability (21). We previously found that PCBP2 is highly expressed in breast cancer stem-like cells (22). Because the stem-like characteristics of breast cancer stem cells are closely associated with resistance to therapy, particularly to olaparib (23, 24), PCBP2 may play a critical role in the development of resistance to olaparib treatment in breast cancer.

N⁶-methyladenosine (m⁶A), the most prevalent internal modification in eukaryotic mRNAs, plays a crucial role in regulating various aspects of mRNA metabolism, including splicing, export, translation, and stability (25). m⁶A methylation is dynamically and reversibly regulated by a methyltransferase (writer) complex composed of METTL3, METTL14, and the cofactor WTAP, as well as by demethylases (erasers) such as fat mass and obesity-associated protein (FTO) and AlkB homolog 5 (ALKBH5), to regulate m⁶A-dependent functions mediated by readers, such as YT521-B homology domain proteins (YTHDF1/2; ref. 26). Increasing evidence suggests that m⁶A methylation contributes to cancer therapy resistance by modulating the stability of key gene transcripts, thus activating or inhibiting specific signaling pathways (25). For example, it has been reported that m⁶A methyltransferase METTL3 promotes oxaliplatin resistance in CD133⁺ gastric cancer stem cells by stabilizing PARP1 mRNA, leading to enhanced BER pathway activity (27). However, no reports have indicated that BRCA-mutated breast cancer cells can overcome olaparib resistance through alterations in m⁶A methylation.

This study aims to explore the role of PCBP2 in regulating olaparib resistance, specifically through its impact on PARP1 mRNA stability. We found that PCBP2 expression is upregulated in olaparib-resistant cell lines, and PCBP2 overexpression induces resistance of BRCA-mutated breast cancer cells to olaparib. Mechanistically, PCBP2 directly interacts with PARP1 mRNA and enhances its stability by inhibiting m⁶A modification. In olaparib-resistant cells, elevated PCBP2 levels decrease PARP1 mRNA decay, leading to upregulation of PARP1 expression and enhanced DNA repair, which in turn reduces cell sensitivity to olaparib, thereby driving resistance.

Materials and Methods

Cell culture

Human breast cancer SUM149PT (RRID: CVCL_3422), HCC1937 (RRID: CVCL_0290), and MCF7 (RRID: CVCL_0031) cell lines were acquired from the Cell Bank of the Chinese Academy of Sciences. SUM149PT and HCC1937 cells are BRCA1-mutant and serve as models of HRD breast cancer (28). MCF7 cells are HR⁺ breast cancer cells. All of the above cell lines were authenticated by DNA profiling (short tandem repeat), morphology, cell viability, isoenzymes, and Mycoplasma assays. HCC1937 and MCF7 cells were cultured in RPMI 1640 medium (Gibco, cat. #11875500BT), and SUM149PT cells were cultured in Ham’s F-12 (Pricella, cat. #PM150810) with 0.5 μg/mL hydrocortisone (MedChemExpress, cat. #HY-N0583) and 10 μg/mL insulin (Beyotime, cat. #P3376). All media had 10% FBS (ZETA, cat. #Z7181FBS-500) and 1% penicillin/streptomycin (Beyotime, cat. #C0222). All cells were cultured in a humidified incubator at 37°C with 5% CO₂. Olaparib (AZD2281, Selleck, cat. #S1060), veliparib (ABT-888, Selleck, cat. #S1004), talazoparib (BMN-673, Selleck, cat. #S7048), and niraparib (MK-4827, Selleck, cat. #S2741) were purchased from Selleck. Olaparib-resistant SUM149PT (SUM149PT^OlaR) and HCC1937 (HCC1937^OlaR) cells were generated by culturing cells in medium supplemented with increasing concentrations of olaparib for 6 months starting at 1 μmol/L and reaching a final dose of 10 μmol/L (SUM149PT^OlaR) or 20 μmol/L (HCC1937^OlaR), followed by continued culture in 10 μmol/L (SUM149PT^OlaR) or 20 μmol/L (HCC1937^OlaR) olaparib to maintain resistance. The IC₅₀ value for SUM149PT and HCC1937 cells with acquired resistance to olaparib was nearly three times that of the parental cells (see Fig. 2A).

Figure 2.

Stable knockdown of PCBP2 enhances olaparib sensitivity both in vitro and in vivo in olaparib-resistant breast cancer cells. A, BRCA-mutated cell lines SUM149PT and HCC1937 were subjected to a gradual increase in the concentration of olaparib to develop acquired resistance. The IC₅₀ values of olaparib-resistant SUM149PT^OlaR and HCC1937^OlaR, as well as the original parental (Par) cell lines, were determined by CCK-8 assay. B, Talazoparib IC₅₀ curves of parental SUM149PT and cells with acquired resistance to olaparib. C, Veliparib IC₅₀ curves of parental SUM149PT and cells with acquired resistance to olaparib. D, Niraparib IC₅₀ curves of parental SUM149PT and cells with acquired resistance to olaparib. E and F, Western blotting was used to assess protein levels of PCBP2 and PARP1, and qRT-PCR was performed to measure PCBP2 mRNA levels. G and H, SUM149PT^OlaR and HCC1937^OlaR cells were stably transduced with shNC and shPCBP2 lentiviruses. Western blotting was used to assess protein levels of PCBP2 and PARP1, and qRT-PCR was performed to measure PCBP2 mRNA levels. I, CCK-8 assay was used to measure cell viability in cells treated with olaparib for 72 hours. J, Left, EdU assays were used to characterize the proliferation of SUM149PT^OlaR and HCC1937^OlaR cells treated with olaparib. Right, percentage of EdU-positive cells was quantified. Cells were treated with 10 μmol/L olaparib for SUM149PT^OlaR and 20 μmol/L olaparib for HCC1937^OlaRfor 72 hours. Scale bar, 25 μm (high magnification). K, Left, clonogenic assays were conducted to assess the colony-forming ability of SUM149PT^OlaR and HCC1937^OlaR cells in the presence of olaparib. Cells were treated with 10 μmol/L olaparib for SUM149PT^OlaR and 20 μmol/L olaparib for HCC1937^OlaR for 7–14 days. Right, the number of colonies was quantified. L and M, Whole cell lysate (WCL) and chromatin-binding protein (CHR) were extracted and analyzed by Western blotting with the indicated antibodies. N, Schematic representation of the PCBP2 knockdown xenograft breast cancer mouse model using SUM149PT^OlaR cells and the treatment regimen. O, At the end of treatment, tumors were excised and photographed. P, Tumor size was measured at the indicated time intervals and calculated, and growth curves were plotted using average tumor volume within each experimental group at the designated time points. Q, Tumor weights. R, Representative IHC images of PCBP2, PARP1, and γH2AX in subcutaneous tumors. Scale bar, 50 μm (high magnification). Data are presented as the mean ± SEM (n = 3). *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Establishment of transfected cell lines

The human PCBP2 overexpression vector, short hairpin RNA vectors targeting PCBP2, and the empty lentiviral vector were purchased from GenePharma. Lentiviral plasmids were cotransfected with packaging plasmids psPAX2 and pMD2.G into HEK293T cells (RRID: CVCL_0063). Supernatants were collected at 48 and 72 hours after transfection and used to infect breast cancer cells. After 24 hours, stably transduced cell lines were selected using puromycin according to the manufacturer’s instructions. siRNAs targeting METTL3, METTL14, WTAP, FTO, ALKBH5, YTHDF1, and YTHDF2 were obtained from Genomeditech. PCBP2 and its three deletion mutants (T1, T2, and T3) were amplified by PCR and inserted into the pcDNA3.1-Flag vector to get pcDNA3.1-Flag-PCBP2 (full length), pcDNA3.1-Flag-PCBP2-KH1 domain (T1), pcDNA3.1-Flag-PCBP2-KH2 domain (T2), and pcDNA3.1-FlagPCBP2-KH3 domain (T3) constructs. The coding sequence, 3′ UTR, and 5′ UTR of PARP1 were amplified and then cloned into the pcDNA3.1 vector. All plasmids were acquired from GenePharma. Plasmid transfection was performed using Lipofectamine 3000 (Thermo Fisher Scientific, cat. #L3000015 following the manufacturer’s protocol for transient transfections. To construct the targeted RNA methylation system, standard molecular biology techniques were used, including restriction enzyme digestion, PCR, and subcloning. The sequences are provided in Supplementary Tables S1 and S2.

qPCR

Total RNA was extracted using TRIzol reagent (Invitrogen, cat. #15596026). cDNA was synthesized using Evo M-MLV RT Reaction Mix (Accurate Biology, cat. # AG11705). qRT-PCR was performed with SYBR Green Pro Taq HS Premix (Accurate Biology, cat. #AG11702). Gene expression levels were calculated using the 2^−ΔΔCT method. Each experiment was repeated at least three times. The sequences of all primers used in the study are provided in Supplementary Tables S3 and S4.

Western blotting

Cells were placed on ice and washed twice with physiologic saline. RIPA lysis buffer (Beyotime, cat. #P0013B) containing protease inhibitors was then added to the cells. The lysates were centrifuged at 12,000 rpm for 20 minutes, and the supernatant was collected. The supernatant was mixed with 4× loading buffer and heated at 100°C for 15 minutes. Proteins were separated using 8% to 15% gradient SDS-PAGE gels and transferred to polyvinylidene difluoride membranes. The membranes were blocked with 5% BSA at room temperature for 1 hour and incubated with primary antibodies at 4°C overnight. After washing, the membranes were incubated with horseradish peroxidase–conjugated secondary antibodies at room temperature for 1 hour. Protein expression was visualized using an enhanced chemiluminescence kit (Thermo Fisher Scientific, cat. #34580). GAPDH was used as an internal control.

IHC

Tissue slides were heated at 60°C for 15 minutes, deparaffinized in xylene, and rehydrated through a graded ethanol series. Antigen retrieval was performed using citrate buffer (pH 6.0). To block endogenous peroxidase activity, slides were immersed in 3% hydrogen peroxide (H₂O₂) for 15 minutes. After blocking with 5% BSA, the slides were incubated with primary antibodies overnight at 4°C. Afterward, secondary antibodies were applied at room temperature for 30 minutes. Visualization was performed using diaminobenzidine (Maixin Bio, cat. # DAB-0031), and counterstaining was carried out with hematoxylin. Staining intensity was scored on a scale of 0 to 3: 0 (negative), 1 (weak), 2 (moderate), and 3 (strong). The percentage of positive staining was scored as follows: 1 (0%–25%), 2 (26%–50%), 3 (51%–75%), and 4 (76%–100%). The final IHC score was determined by multiplying the intensity score by the proportion score, with the final score for each sample being the average of two independent measurements. Hematoxylin and eosin staining was used for morphologic examination, and the results were evaluated by two independent pathologists.

Cell Counting Kit-8 and determination of IC₅₀ value

Cells (2 × 10³) were seeded into each well of a 96-well microplate and cultured for 72 hours in medium containing a range of drug concentrations. After treatment, 10 μL of Cell Counting Kit-8 (CCK-8) reagent (Beyotime, cat. #C0038) was added to each well, and the cells were incubated for an additional 2 hours. Absorbance was measured at 450 nm using a microplate reader, with wells containing medium alone serving as the blank control. IC₅₀ values were calculated using GraphPad Prism (version 8.0, RRID: SCR_002798).

Colony formation and EdU cell proliferation assay

Cells (1,000 cells/well) were seeded as a single-cell suspension into 6-well plates and treated with the indicated drugs for 7 to 14 days. Colonies were then fixed with 4% formaldehyde for 15 minutes and stained with 0.05% crystal violet (Merck, cat. # 115940) for 15 minutes. Colony numbers were quantified using ImageJ software (version 1.52a, RRID: SCR_003070). Cells (5 × 10³ to 1 × 10⁴ per well) were seeded into 24-well plates. After a 2-hour incubation with 5-ethynyl-2′-deoxyuridine (EdU; Beyotime Biotechnology, cat. #C0071S), cells were promptly fixed and permeabilized, and DNA-incorporated EdU was labeled using Click-iT reaction cocktail. Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). Fluorescent images were acquired using a fluorescence microscope, and the percentage of EdU-positive cells was quantified using ImageJ software.

RNA immunoprecipitation and biotin RNA pulldown assay

RNA immunoprecipitation (RIP) was conducted using a RIP-Assay kit (MBL Life Science, RN1001) according to the manufacturer’s protocol. In brief, 10 μg of antibodies specific for PCBP2 (Abcam, cat. #ab236137, RRID: AB_2827629), METTL3 (Proteintech, cat. #15073-1-AP, RRID: AB_2142033), METTL14 (Proteintech, cat. #26158-1-AP, RRID: AB_2800447), WTAP (Proteintech, cat. #60188-1-Ig, RRID: AB_10859484), YTHDF1 (Abcam, cat. #ab220162, RRID: AB_2892231), YTHDF2 (Abcam, cat. #ab246514, RRID: AB_2891213), or control IgG (Thermo Fisher Scientific, cat. #02-6102, RRID: AB_2532938) were conjugated to protein A/G agarose beads and incubated at 4°C for 30 minutes. The antibody–bead complexes were then mixed with cell lysates and incubated at 4°C overnight. Following immunoprecipitation, RNA was extracted from the complexes and subjected to qRT-PCR for the identification and validation of target RNA molecules. The RNA pulldown assay was conducted using an RNA pulldown kit (BersinBio, cat. #Bes5102) according to the manufacturer’s protocol. In brief, 1 × 10⁷ cells were lysed and incubated overnight at 4°C with 3 μg of biotin-labeled oligonucleotide probes targeting the exogenous PARP1 3′ UTR, 5′ UTR, and coding sequence (Genomeditech). Streptavidin-coated magnetic beads were subsequently added, and the lysates were incubated for an additional 2 hours at room temperature. After thorough washing, pulled-down proteins were subjected to Western blot analysis to detect RNA-associated proteins.

m⁶A RIP analysis

m⁶A modification levels in target mRNA were assessed using Magna MeRIP m⁶A Kit (Millipore, cat. # 17-10499) following the manufacturer’s instructions. Briefly, total RNA was extracted and fragmented using RNA fragmentation buffer. One tenths of the fragmented RNA was saved for the input control. The remaining RNA was incubated with m⁶A antibody–conjugated magnetic beads (protein A/G). After immunoprecipitation, the RNA–antibody–bead complexes were thoroughly washed, eluted, and purified. Enriched mRNA was then analyzed by qRT-PCR.

Dual-luciferase reporter assay

Breast cancer cells were seeded into 24-well plates and transfected with pmirGLO-PARP1-3′ UTR reporter plasmids, containing either the wild-type 3′ UTR sequence or a mutant version with altered m⁶A methylation sites, using Lipofectamine 3000. The pmirGLO vector contains both firefly luciferase (as the reporter) and Renilla luciferase (as the internal control) within a single plasmid. After 48 hours, luciferase activities were measured using a Dual-Luciferase Reporter Assay System (Promega) following the manufacturer’s protocol. Relative luciferase activity (firefly/Renilla) was calculated to assess the effect of m⁶A site mutations on PARP1 mRNA stability and its regulation by PCBP2.

Xenograft mouse model

All animal experiments were approved by the Institutional Animal Care and Use Committee of Shantou University Medical College (approval number SUMC2023-461). Four-week-old female BALB/c nude mice (Beijing Vital River Laboratory Animal Technology Co., Ltd, RRID: IMSR_RJ: BALB-C-NUDE) were housed in a pathogen-free facility and acclimatized for 1 week prior to experimentation.

Xenograft PCBP2 knockdown, olaparib-resistant cell model: Stable PCBP2 knockdown (SUM149PT^OlaR-shPCBP2) or control (SUM149PT^OlaR-shNC) cells (1 × 10⁷) were suspended in Matrigel (Corning, cat. #356231) and subcutaneously injected into the axillary region of four-week-old female BALB/c nude mice (n = 4–6 per group). Once tumor volumes reached 200 mm³, mice were randomly divided into four groups and treated with either olaparib (50 mg/kg, intraperitoneally, five times per week) or DMSO as the vehicle control. Tumor size was measured every 2 days using the formula (length × width²)/2, and body weight was monitored. After 18 days of treatment, mice were euthanized, and tumors were harvested for subsequent analysis.

Xenograft model for PCBP2 overexpression combined with catalytically inactivated Cas13 (dCas13)-targeted m⁶A modification: Parental SUM149PT cells stably overexpressing PCBP2 and transduced with the dCas13-METTL3 system (ΔMETTL3, residues 273–580), combined with either nontargeting guide RNA (gRNA; gNC) or PARP1-targeting gRNA (gPARP1), were used for in vivo validation. Cells (1 × 10⁷) were mixed with Matrigel and subcutaneously injected into BALB/c nude mice as above (n = 4–6 per group). Once tumor volumes reached 200 mm³, mice were treated intraperitoneally with olaparib (50 mg/kg) or DMSO 5 days per week for 18 days. Tumor volumes and body weights were recorded every 2 days. At the endpoint, tumors were collected for histologic and molecular analyses.

Clinical specimen collection

This study included 10 patients with TNBC that were treated at the Affiliated Cancer Hospital of Shantou University Medical College, Shantou, Guangdong Province, China, between June 2023 and March 2025. The inclusion criteria were as follows: (i) female patients without menopause and between 20 and 55 years of age, (ii) patients diagnosed with TNBC, (iii) patients without a previous history of malignant tumors, (iv) patients without distant metastasis, and (v) patients that followed guidelines for the oral administration of olaparib as an adjuvant therapy after surgery. According to the guidelines by the American College of Clinical Oncology/College of American Pathologists for breast cancer, estrogen receptor and progesterone receptor positivity ≥1% is considered as positive. HER2 testing was performed according to the 2007 guidelines of the American Society for Clinical Oncology and Pathology. The olaparib therapy drug resistance standards for breast cancer were according to the ESO-ESMO Second Edition International Consensus Guidelines for Advanced Breast Cancer. This study was approved by the Ethics Committee of the Affiliated Cancer Hospital of Shantou University Medical College (approval number 2022141). Written informed consent was obtained from all patients prior to participation in the study, and their complete clinicopathologic and follow-up data were available for analysis. The clinical characteristics of the patients are detailed in Supplementary Table S5.

Statistical analysis

All experimental data were analyzed and visualized using GraphPad Prism 8. Differences between group means were evaluated using Student t test or one-way ANOVA, depending on the experimental design. Pearson correlation analysis was conducted to assess the relationships between variables. Kaplan–Meier analysis was used to estimate overall survival, with statistical significance determined using the log-rank test. Differences with a P value of less than 0.05 were considered statistically significant.

Results

PCBP2 is associated with olaparib sensitivity

We previously found that PCBP2 is highly expressed in breast cancer stem-like cells, which possess olaparib resistance (22–24). To identify a role for PCBP2 in olaparib resistance, we initially conducted RIP in BT549 TNBC cells using a PCBP2 antibody to enrich for RNA molecules bound to PCBP2. RNA sequencing analysis revealed PCBP2 to be highly enriched in PARP1 mRNA (Supplementary Fig. S1A). Given that stem cells with high expression of PCBP2 are resistant to olaparib, we examined the involvement of PCBP2 in olaparib resistance.

To determine whether PCBP2 expression is associated with sensitivity of BRCA-mutated breast cancer cells to PARPi, we established stable knockdown and overexpression of PCBP2 in BRCA1-mutant, HR-deficient breast cancer cell lines SUM149PT and HCC1937. The extent of knockdown (Fig. 1A and B) and overexpression (Fig. 1C and D) were determined by qRT-PCR and Western blot analysis. CCK-8 assays showed that PCBP2 knockdown enhanced the sensitivity of SUM149PT and HCC1937 cells to olaparib treatment (Fig. 1E and F), whereas PCBP2 overexpression induced resistance to olaparib (Fig. 1G and H). Additionally, PCBP2 knockdown led to a decrease in cell proliferation and colony formation upon olaparib treatment, but not in the absence of olaparib (Fig. 1I and J; Supplementary Fig. S1B and S1C), whereas PCBP2 overexpression had the opposite effect (Fig. 1K and L; Supplementary Fig. S1D and S1E).

Figure 1.

PCBP2 is associated with olaparib sensitivity. A–D, SUM149PT and HCC1937 cells were stably transduced with LV3 (shNC) or PCBP2 shRNA (shPCBP2) lentiviruses for PCBP2 knockdown experiments. SUM149PT and HCC1937 cells were transduced with LV5 (vector) and LV5-PCBP2 (oePCBP2) lentiviruses for control and PCBP2 overexpression experiments, respectively. Western blotting was used to assess protein levels of PCBP2 and PARP1, and qRT-PCR was performed to measure PCBP2 mRNA levels. E–H, CCK-8 assay was used to measure cell viability in cells treated with olaparib (Ola) for 72 hours. I, Left, clonogenic assays were performed to evaluate the colony-forming ability of SUM149PT cells with or without olaparib treatment (10 μmol/L) for 7–14 days. Right, the number of clones was quantified. J, Left, EdU assays were performed to assess the proliferation of SUM149PT cells following treatment with olaparib (10 μmol/L) or DMSO for 72 hours. Right, percentage of EdU-positive cells was quantified and plotted. Scale bar, 25 μm (high magnification). K, Clonogenic assays were performed as in I using SUM149PT cells overexpressing PCBP2. L, EdU assays were performed as in J using SUM149PT cells overexpressing PCBP2. M, Western blot analysis of PARP1 and PCBP2 protein levels in SUM149PT and HCC1937 cells. Cells were transfected with vector or oePCBP2, followed by siRNA targeting PARP1 (siPARP1) or negative control (siNC). GAPDH was used as the loading control. N and O, CCK-8 assay was used to measure cell viability in cells treated with olaparib for 72 hours. P–S, Whole cell lysate (WCL) and chromatin-binding protein (CHR) were extracted and analyzed by Western blotting with the indicated antibodies. Data are presented as the mean ± SEM (n = 3). ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001.

To further evaluate whether PCBP2 affects olaparib sensitivity in HR⁺ breast cancer cells, we performed PCBP2 knockdown in the MCF7 cell line. Western blot analysis confirmed effective knockdown of PCBP2, accompanied by a reduction in PARP1 protein levels, compared with the shNC control (Supplementary Fig. S1F). However, CCK-8 assays demonstrated that silencing PCBP2 did not significantly alter the sensitivity of MCF7 cells to olaparib across a concentration range of 0 to 30 μmol/L (Supplementary Fig. S1G). These findings suggest that the role of PCBP2 in modulating PARPi sensitivity may be context-dependent and more prominent in BRCA-mutated or HRD breast cancer cells.

As shown above, PCBP2 knockdown led to a clear reduction in PARP1 protein levels, whereas PCBP2 overexpression increased PARP1 expression in BRCA-mutated breast cancer cells (Fig. 1B and D), suggesting that PARP1 may mediate PCBP2-driven olaparib resistance. To determine whether this effect is specifically mediated by PARP1, we silenced PARP1 in cells overexpressing PCBP2. PARP1 knockdown did not affect PCBP2 levels (Fig. 1M) but fully reversed the resistance to olaparib induced by PCBP2 overexpression (Fig. 1N and O). We next evaluated whether other PARP isoforms were involved. PARP2 and PARP3 expression remained unchanged following PCBP2 knockdown or overexpression (Supplementary Fig. S1H and S1I). Moreover, silencing PARP2 or PARP3 did not affect olaparib sensitivity in PCBP2-overexpressing cells (Supplementary Fig. S1J–S1M). These data indicate that PCBP2 promotes PARPi resistance primarily through PARP1.

PARP1 is closely associated with DNA BER in response to DNA SSBs (29, 30). Upon detecting DNA damage, PARP1 undergoes auto-poly(ADP-ribosyl)ation and recruits XRCC1, which subsequently facilitates the assembly of DNA polymerase β (POLβ) and DNA ligase III (LIGIII) complexes at the damage site, coordinating and promoting DNA repair (12, 31). Overexpression of PCBP2 led to an increase in PARP1 expression along with elevated levels of chromatin-bound DNA SSB repair factors XRCC1, POLβ, and LIGIII, indicating enhanced DNA SSB repair activity (Fig. 1Q). Concurrently, a reduction in the DNA damage marker γH2AX suggests lower DNA damage levels, supporting the notion of improved repair efficiency (Fig. 1S). Conversely, knockdown of PCBP2 resulted in reduced levels of PARP1 and chromatin-bound BER repair factors (Fig. 1P), along with increased γH2AX expression (Fig. 1R), consistent with diminished repair efficiency and accumulation of DNA damage. Collectively, these findings indicate that PCBP2 plays a regulatory role in enhancing DNA BER activity, which may in turn reduce the sensitivity of SUM149PT and HCC1937 cells to olaparib. Furthermore, compared with the control group, following PCBP2 knockdown, olaparib increased the expression of proapoptotic proteins cleaved caspase-3 and Bax while decreasing the expression of the antiapoptotic protein Bcl2 (Fig. 1R). Conversely, PCBP2 overexpression reduced the expression of cleaved caspase-3 and Bax, while elevating Bcl2 levels, further supporting a role for PCBP2 in inducing apoptosis resistance in response to olaparib treatment (Fig. 1S). Collectively, our findings demonstrate that PCBP2 regulates olaparib sensitivity.

Stable knockdown of PCBP2 enhances olaparib sensitivity in olaparib-resistant breast cancer cells both in vitro and in vivo

To further explore the relationship between PCBP2 and olaparib resistance, we established acquired olaparib-resistant cell lines (SUM149PT^OlaR and HCC1937^OlaR) by gradually increasing the concentration of olaparib over 8 months. CCK-8 assays showed that for the SUM149PT cell line, the IC₅₀ value of the olaparib-resistant SUM149PT cells (OlaR) was 22.32 μmol/L, compared with 4.93 μmol/L in the parental cells (Fig. 2A). In the HCC1937 cell line, the calculated IC₅₀ value of the olaparib-resistant HCC1937 cells was 40.93 μmol/L, whereas that of the parental cells was 10.11 μmol/L (Fig. 2A). We also assessed the effects of three additional PARPis (niraparib, veliparib, and talazoparib) on both olaparib-resistant and parental cells (Fig. 2B–D; Supplementary Fig. S2A–S2C). The results indicated that olaparib-resistant cells exhibited broad resistance to multiple PARPis. Further RT-qPCR and Western blot analyses confirmed that PCBP2 mRNA and protein levels were elevated in the olaparib-resistant cells compared with the parental cells, suggesting that PCBP2 plays a critical role in olaparib resistance (Fig. 2E and F).

To further investigate the role of PCBP2 in olaparib resistance, we stably knocked down PCBP2 in the SUM149PT^OlaR and HCC1937^OlaR cells (Fig. 2G and H) and conducted a series of drug sensitivity assays. Cells were treated with varying concentrations of olaparib for 72 hours, and then CCK-8 assays were performed. Compared with the control group (shNC), PCBP2 knockdown restored the sensitivity of resistant cells to olaparib (Fig. 2I). Similarly, the results from EdU proliferation and colony formation assays demonstrated that PCBP2 knockdown in resistant cells led to a reduction in cell proliferation (Fig. 2J) and colony formation (Fig. 2K) upon olaparib treatment.

SUM149PT^OlaR and HCC1937^OlaR cells exhibited enhanced expression of chromatin-bound BER proteins—including XRCC1, POLβ, and LIGIII—upon drug treatment compared with their parental counterparts (Fig. 2L; Supplementary Fig. S2D). Under the same conditions, levels of the DNA damage marker γH2AX were reduced, indicating improved tolerance to DNA damage. These observations suggest that resistant cells maintain higher BER activity to cope with olaparib-induced lesions. In addition, under olaparib treatment, resistant cells displayed suppressed apoptosis, characterized by reduced expression of proapoptotic markers cleaved caspase-3 and Bax, and increased levels of the antiapoptotic protein Bcl2 (Fig. 2M; Supplementary Fig. S2E). Knockdown of PCBP2 in resistant cells led to a reduction in PARP1 expression (Fig. 2H) and attenuated the olaparib-induced upregulation of chromatin-bound BER components (Fig. 2L; Supplementary Fig. S2D). This was accompanied by reactivation of DNA damage signaling, as shown by increased γH2AX levels. Notably, PCBP2 knockdown also partially restored drug-induced apoptotic signaling, with increased cleaved caspase-3 and Bax and decreased Bcl2 levels under olaparib exposure (Fig. 2M; Supplementary Fig. S2E). These findings suggest that PCBP2 supports the drug-adaptive repair and survival responses of PARPi-resistant cells and that its depletion sensitizes these cells by disrupting these protective mechanisms.

To validate these in vitro findings, we further assessed whether abnormal PCBP2 expression affects tumor sensitivity to olaparib in a xenograft mouse model. SUM149PT^OlaR cells transduced with either shNC or shPCBP2 were subcutaneously injected into BALB/c nude mice. After approximately 10 days, mice were randomly divided into four groups and intraperitoneally administered 50 mg/kg olaparib for 5 days a week over a period of 18 days (Fig. 2N). Tumor volume and weight in the shPCBP2 group were lower than those in the shNC group (Fig. 2O–Q). We used IHC analysis to assess expression levels of PCBP2, PARP1, and γH2AX in tumor tissue sections from xenografted mice. IHC staining showed that, in the olaparib-treated shNC group, high levels of PCBP2 and PARP1 expression were observed, along with reduced γH2AX expression, indicating substantial DNA repair activity. In contrast, in the olaparib-treated shPCBP2 group, PCBP2 and PARP1 levels were significantly downregulated, accompanied by a notable increase in γH2AX expression, suggesting heightened DNA damage (Fig. 2R). These results provide in vivo confirmation that PCBP2 upregulates PARP1 expression and increases DNA repair efficiency, further supporting the potential of PCBP2 inhibition to enhance olaparib sensitivity through modulating DNA damage response pathways.

PCBP2 upregulates PARP1 by directly binding to the PARP1 mRNA 3′ UTR

Our above transcriptomic data (RIP sequencing) indicate that RNAs bound to PCBP2 are highly enriched in PARP1 mRNA (Fig. 1A). To further explore the regulatory mechanism of PCBP2 on PARP1, we utilized the bc-GenExMiner database (http://bcgenex.ico.unicancer.fr/BC-GEM/GEM-requete.php). A positive correlation between the expression of PCBP2 and PARP1 mRNA was observed in patients with TNBC (Fig. 3A). To validate this finding, we conducted qRT-PCR and Western blot experiments in both parental and resistant cell lines. In parental cells, knockdown of PCBP2 resulted in a decrease in PARP1 mRNA and protein levels (Figs. 1C and 3B), whereas PCBP2 overexpression led to an increase in both (Figs. 1E and 3C). Similar results were observed in olaparib-resistant cells, in which upregulation of PCBP2 was accompanied by increased levels of PARP1 mRNA and protein (Figs. 2F and 3D). Conversely, knockdown of PCBP2 in these resistant cells caused a reduction in PARP1 mRNA and protein levels (Figs. 2H and 3E). These results confirm that PCBP2 is positively correlated with PARP1 at both the mRNA and protein levels, suggesting that PCBP2 regulates PARP1 expression.

Figure 3.

PCBP2 upregulates PARP1 by directly binding to the PARP1 mRNA 3′ UTR. A, Pearson pairwise correlation plot of PCBP2 and PARP1 mRNA expression obtained from the bc-GenExMiner database using RNA sequencing analysis. B–E, qRT-PCR was performed to measure PARP1 mRNA levels. F, RIP assay for the interaction between PCBP2 and PARP1 mRNA in SUM149PT and HCC1937 cells, showing enrichment of PARP1 mRNA, as measured by qRT-PCR. RIP with nonspecific IgG was set as control. Western blot of PCBP2 showing equal amount of input PCBP2 protein in the two groups. G, Schematic structures of PCBP2 and deletion mappings of PCBP2. Vec, vector; FL, full length PCBP2; T, truncated fragment of PCBP2, with T1: 1–96; T2: 76–286; and T3: 163–365. H, RIP-qPCR, using full-length or truncated PCBP2 protein to identify the PARP1 mRNA–binding domains in PCBP2. I, RNA pulldown assay using a PARP1 mRNA 5′ UTR probe, coding sequence (CDS) probe, or 3′ UTR probe in SUM149PT and HCC1937 cells. PCBP2 protein was characterized by Western blotting in the indicated precipitates of the pulldown assay. J, Western blot analysis of HA-tagged PARP1 lacking the 3′ UTR [oePARP1(Δ3′ UTR)] in SUM149PT and HCC1937 cells transduced with shNC or shPCBP2. Cells were cotransfected with either vector or oePARP1(Δ3′ UTR) construct. K and L, Expression of PARP1 mRNA, measured by qRT-PCR, in cells transduced with the indicated lentiviruses and treated with α-amanitin for 0, 6, 12, 18, 24, and 30 hours. Data are presented as the mean ± SEM (n = 3). *, P < 0.05; **, P < 0.01; ***, P < 0.001.

RIP assay, with a PCBP2 antibody, further confirmed that PCBP2 is enriched in PARP1 mRNA in both SUM149PT and HCC1937 cells (Fig. 3F). PCBP2 is known to contain three KH domains (32), which are known to recognize and bind to the 5′ UTR or 3′ UTR of target mRNAs, thereby regulating their stability (33–35). To precisely identify the interaction domains between PCBP2 and PARP1 mRNA, we constructed tagged full-length PCBP2 and three overlapping truncated fragments covering the nonactive regions (Fig. 3G). RIP assays showed that PARP1 mRNA binds to the KH2 domain (T2) of PCBP2 (Fig. 3H). Furthermore, using RNA pulldown assays with in vitro–transcribed PARP1 mRNA probes, PCBP2 was pulled down by probes specifically containing the 3′ UTR of PARP1 mRNA but not by probes containing the 5′ UTR or coding sequence regions (Fig. 3I). To further verify that PCBP2 regulates PARP1 expression specifically through interaction with its 3′ UTR, we constructed an HA-tagged PARP1 overexpression plasmid lacking the 3′ UTR (oePARP1–Δ3′ UTR). SUM149PT and HCC1937 cells were transfected with this construct under PCBP2-silencing or control conditions. Western blot analysis revealed that knockdown of PCBP2 did not alter the expression of the exogenous PARP1 protein lacking the 3′ UTR, as shown by the unchanged HA signal (Fig. 3J).

Next, we treated both parental and resistant cells with α-amanitin to inhibit RNA polymerase II–mediated RNA synthesis and measured PARP1 mRNA levels at different time points. Overexpression of PCBP2 extended the half-life of PARP1 mRNA in parental cells (Fig. 3K), whereas knockdown of PCBP2 in resistant cells markedly shortened the half-life of PARP1 mRNA (Fig. 3L). In summary, these findings suggest that PCBP2 upregulates PARP1 protein expression by directly binding to the 3′ UTR of PARP1 mRNA, enhancing PARP1 mRNA stability.

Depletion of PCBP2 facilitates m⁶A methyltransferase complex recruitment and accelerates PARP1 mRNA degradation

Previous studies have shown that m⁶A methylation plays a crucial role in mRNA metabolism and stability (36). Given the link between PCBP2 and mRNA stabilization, PCBP2 might stabilize PARP1 mRNA by modulating m⁶A methylation. Using the catRAPID database (http://s.tartaglialab.com/page/catrapid_group), we analyzed the interaction between PCBP2 and PARP1 mRNA. Interestingly, database predictions from catRAPID identified a potential binding region between PCBP2 and the 3′ UTR of PARP1 mRNA, indicating a likely interaction site based on sequence and structural compatibility. Further analysis of PARP1 mRNA using the WHISTLE database (https://whistle-epitranscriptome.com/) identified an m⁶A modification site near the PCBP2-binding region on PARP1 mRNA (Fig. 4A). To explore the potential mechanism by which PCBP2 enhances PARP1 mRNA stability, we investigated the role of m⁶A modification in PARP1 mRNA stability. RIP assay, with an m⁶A antibody, showed enrichment of m⁶A at the predicted site (Fig. 4A). To identify which m⁶A-related proteins play a key role in the stability of PARP1 mRNA, we designed siRNAs targeting m⁶A methyltransferases (METTL3, METTL14, and WTAP) and demethylases (ALKBH5 and FTO) and transfected them into SUM149PT^OlaR and HCC1937^OlaR cells. Knockdown of METTL3, METTL14, and WTAP all increased PARP1 mRNA levels, whereas knockdown of ALKBH5 and FTO had little effect (Fig. 4B; Supplementary Fig. S3A). Furthermore, knockdown of METTL3, METTL14, and WTAP could partially reverse the PCBP2 knockdown–mediated decrease in PARP1 mRNA levels (Fig. 4C; Supplementary Fig. S3B). Interestingly, there were no changes in the levels of m⁶A-related proteins in PCBP2 knockdown cells (Supplementary Fig. S3C). Compared with the control group, PCBP2 knockdown increased the binding of METTL3, METTL14, and WTAP on the 3′ UTR of PARP1 mRNA (Fig. 4D–F; Supplementary Fig. S3D–S3F), whereas PCBP2 overexpression reduced the binding of these methyltransferases on the 3′ UTR (Fig. 4G–I; Supplementary Fig. S3G–S3I), which is consistent with the m⁶A enrichment results (Fig. 4J and K; Supplementary Fig. S3J and S3K). Additionally, dual-luciferase reporter assays revealed that PCBP2 knockdown led to a decrease in the luciferase activity of the luciferase-PARP1 3′ UTR reporter gene, but this change was not observed in constructs with a mutated PARP1 3′ UTR (Fig. 4L–N; Supplementary Fig. S3L and S3M). Similarly, the reduction in luciferase activity induced by PCBP2 knockdown could be abolished by the knockdown of METTL3, METTL14, or WTAP (Fig. 4O–Q; Supplementary Fig. S3N-S3P). Lastly, by knocking down METTL3, we found that the PCBP2 knockdown–induced decrease in the half-life of PARP1 mRNA was nearly completely restored (Fig. 4R and S). In summary, these findings suggest that loss of PCBP2 facilitates the recruitment of the m⁶A methyltransferase complex to the 3′ UTR of PARP1 mRNA, thereby reducing its stability.

Figure 4.

Depletion of PCBP2 facilitates m⁶A methyltransferase complex recruitment and accelerates PARP1 mRNA degradation. A, Top, predicted m⁶A site information was obtained from the WHISTLE database. Left, location of the PCBP2 binding site and a potential m⁶A site in the PARP1 mRNA 3′ UTR. Right, abundance of the PARP1 transcript in mRNA immunoprecipitated with an anti-m⁶A antibody was measured by qRT-PCR and normalized to IgG. B and C, The mRNA level of PARP1 was determined by qRT-PCR after transfection with the indicated siNC or siRNAs. D–I, METTL3, METTL14, and WTAP were immunoprecipitated, followed by qRT-PCR for assessing the association of PARP1 mRNA with the indicated m⁶A methyltransferase after overexpression or knockdown of PCBP2. RIP with nonspecific IgG was set as the control. J and K, Abundance of PARP1 mRNA, as measured by qRT-PCR, among mRNA immunoprecipitated with anti-m⁶A antibody from cells transduced with the indicated vectors. L, PARP1 mRNA 3′ UTR, either wild-type or with a mutated m⁶A consensus sequence (A–G), was cloned into a dual-luciferase reporter. M and N, Luciferase activity of the reporter gene with a PARP1 mRNA 3′ UTR was measured and normalized to Renilla luciferase activity in cells after transduction with shPCBP2, oePCBP2, or their corresponding empty vectors. O–Q, Luciferase activity was measured after cotransfection with shPCBP2 and siMETTL3/siMETTL14/siWTAP. R and S, Relative expression of PARP1 mRNA in cells cotransfected with shPCBP2 and either siNC or siMETTL3, as measured by qRT-PCR after α-amanitin treatment. T, Schematic illustration of the targeted RNA methylation system. U, Abundance of PARP1 among mRNA immunoprecipitated with anti-m⁶A antibody from cells transfected with the indicated vectors, as measured by qRT-PCR. V, Protein levels of PARP1 and PCBP2 were measured by Western blotting after transfection with the indicated vectors. W, Relative expression of PARP1 mRNA in cells transfected with the indicated vectors, as measured by qRT-PCR after α-amanitin treatment. X, At the end of treatment, tumors were excised and imaged. Data are presented as the mean ± SEM (n = 3). ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001.

A targeted RNA methylation system was applied to confirm that m⁶A modifications of PARP1 mRNA play a critical role in mediating the effects of PCBP2 on olaparib sensitivity in breast cancer cells. Tethering dCas13 to m⁶A writer/readers allows programmable installation of m⁶A at sites specified by a Cas13 gRNA (37). We utilized a programmable dCas13-based system fused to the catalytic core of METTL3 (ΔMETTL3/M3, residues 273–580) to achieve site-specific m⁶A methylation (Fig. 4T). MeRIP-qPCR showed that PCBP2 overexpression markedly reduced m⁶A levels on PARP1 mRNA in SUM149PT and HCC1937 cells. However, reintroduction of m⁶A at targeted sites via gRNA-guided ΔMETTL3 largely restored m⁶A methylation despite PCBP2 overexpression (Fig. 4U). Compared with the negative control (M3 + gNC), PARP1 protein levels were elevated in cells overexpressing PCBP2. Targeted m⁶A installation on PARP1 mRNA using dCas13-ΔMETTL3 and a PARP1-specific gRNA (M3 + gPARP1) partially reversed this increase, leading to reduced PARP1 expression (Fig. 4V). Correspondingly, targeted m⁶A installation on PARP1 mRNA partially reversed the PCBP2-induced mRNA stabilization, leading to enhanced PARP1 mRNA decay (Fig. 4W). To further investigate whether targeted m⁶A restoration on PARP1 mRNA can reverse the olaparib resistance induced by PCBP2, we performed CCK-8 and clonogenic assays in SUM149PT and HCC1937 cells. Compared with control cells (vector + M3 + gNC), PCBP2 overexpression (oePCBP2 + M3 + gNC) reduced olaparib sensitivity. Importantly, coexpression of dCas13–ΔMETTL3 (M3) and gPARP1 partially rescued olaparib sensitivity in PCBP2-overexpressing cells (Supplementary Fig. S4A). Consistently, clonogenic assays showed that the impaired colony-forming ability caused by olaparib was partially restored upon site-specific m⁶A reinstallation at the PARP1 3′ UTR (Supplementary Fig. S4B).

To further validate our in vitro findings, we assessed whether PCBP2 and m⁶A-targeted regulation of PARP1 affects olaparib sensitivity in vivo using a xenograft model derived from parental SUM149PT cells. Cells were transduced with PCBP2-overexpression constructs and dCas13-ΔMETTL3 (M3) combined with either a gNC or a PARP1-specific gRNA (gPARP1) and subcutaneously injected into BALB/c nude mice. Mice were treated with olaparib (50 mg/kg, i.p., 5 days/week) or DMSO for 18 days, starting 10 days after inoculation. PCBP2 overexpression markedly reduced the tumor-suppressive effect of olaparib, resulting in increased tumor volume and weight. Importantly, targeted reintroduction of m⁶A at the PARP1 3′ UTR, using M3 + gPARP1, partially reversed this effect, restoring olaparib sensitivity in the PCBP2-overexpressing tumors (Fig. 4X; Supplementary Fig. S4C and S4D). We performed IHC to evaluate the expression of PCBP2, PARP1, γH2AX, METTL3, and YTHDF2. In the olaparib-treated group, PCBP2 overexpression (oePCBP2 + M3 + gNC) led to increased PARP1 expression and reduced γH2AX staining, suggesting enhanced DNA repair. In contrast, targeted restoration of m⁶A on PARP1 mRNA (oePCBP2 + M3 + gPARP1) resulted in decreased PARP1 levels and increased γH2AX signal, indicating impaired DNA repair and elevated DNA damage (Supplementary Fig. S4E). In summary, these in vivo results corroborate our in vitro findings, further supporting the conclusion that PCBP2 promotes olaparib resistance by stabilizing PARP1 mRNA through m⁶A-dependent mechanisms and that site-specific restoration of m⁶A marks on PARP1 can partially overcome this resistance.

YTHDF2-dependent regulation of PARP1 mRNA stability modulates olaparib sensitivity

Readers can recognize and bind to m⁶A-modified transcripts, regulating gene expression by influencing various processes such as mRNA stability, structure, and translation efficiency (38). For example, YTHDF2 binds to and mediates DDX58 mRNA degradation in an m⁶A-dependent manner (39). Given our previous results showing that m⁶A modification, mediated by the methyltransferase complex, reduces the stability of PARP1 mRNA, we hypothesized that YTHDF2 might be the key m⁶A reader involved in regulating PARP1 mRNA stability. To test this possibility, we used siRNA to knock down YTHDF1 and YTHDF2 individually and then measured PARP1 mRNA levels. Knockdown of YTHDF2 led to a significant increase in PARP1 mRNA levels, whereas knockdown of YTHDF1 had no noticeable effect on PARP1 mRNA expression (Fig. 5A and B; Supplementary Fig. S5A). Further investigation showed that YTHDF2 knockdown could counteract the effect of PCBP2 knockdown on PARP1 mRNA levels (Fig. 5C and D), and RIP assays indicated increased YTHDF2 binding to PARP1 mRNA, especially in PCBP2 knockdown OlaR cells (Fig. 5E–H). In contrast, YTHDF1 did not show any notable binding to PARP1 mRNA (Fig. 5I and J; Supplementary Fig. S5B and S5C). This finding was further supported by dual-luciferase reporter assays, in which knockdown of PCBP2 led to a decrease in the luciferase activity of the PARP1 3′ UTR reporter, which could be partially restored by YTHDF2 knockdown (Fig. 5K and L), whereas YTHDF1 knockdown had no significant effect (Supplementary Fig. S5D and S5E). Additionally, YTHDF2 knockdown significantly increased luciferase activity in parental SUM149PT and HCC1937 cells transfected with the wild-type PARP1 3′ UTR reporter but not the m⁶A motif mutant construct (Supplementary Fig. S5F and S5G). Finally, mRNA half-life analysis showed that knockdown of YTHDF2 almost completely restored the shortened half-life of PARP1 mRNA induced by PCBP2 knockdown (Fig. 5M and N). In contrast, knockdown of YTHDF1 did not alter this effect. In summary, YTHDF2 regulates PARP1 mRNA stability by recognizing m⁶A modifications, thereby influencing olaparib sensitivity.

Figure 5.

YTHDF2-dependent regulation of PARP1 mRNA stability modulates olaparib sensitivity. A–D, The mRNA level of PARP1 was determined by qRT-PCR after transfection with siNC or siYTHDF1/2. E–J, YTHDF1/2 was immunoprecipitated, followed by qRT-PCR for assessing the association of PARP1 mRNA with YTHDF1/2 after overexpression or knockdown of PCBP2. RIP with nonspecific IgG was set as control. Western blot of YTHDF1/2 showed equal amount of input YTHDF1/2 protein in the two groups. K and L, Luciferase activity was measured after cotransfection with shPCBP2 and siYTHDF2. M and N, Relative expression of PARP1 mRNA in cells transduced with shPCBP2 and cotransfected with either siNC or siYTHDF2, as measured by qRT-PCR after α-amanitin treatment. O, Schematic illustration of the targeted RNA methylation system. P and Q, qRT-PCR was performed to measure PARP1 mRNA levels. R, Protein level of PARP1 was measured by Western blotting after transfection with the indicated vectors. S, Relative expression of PARP1 mRNA in cells transfected with the indicated vectors, as measured by qRT-PCR after α-amanitin treatment. Data are presented as the mean ± SEM (n = 3). ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001.

The targeted m⁶A read system was applied to confirm the above results (40). Guided by a gRNA targeting the 3′ UTR of PARP1, YTHDF2^1–400 (ΔYTHDF2/Y2) reduced both PARP1 mRNA and protein levels in PCBP2-overexpressing SUM149PT and HCC1937 cells (Fig. 5O–R). Messenger RNA stability assays further confirmed that Y2 + gPARP1 accelerated the decay of PARP1 transcripts compared with the nontargeting control (Y2 + gNC), partially counteracting the stabilizing effect induced by PCBP2 overexpression (Fig. 5S). To validate the functional consequence of this targeted degradation on PARPi response, we assessed cell viability and colony formation. Targeted recruitment of Y2 + gPARP1 reduced cell viability following olaparib treatment compared with Y2 + gNC in both SUM149PT and HCC1937 cells (Supplementary Fig. S5H and S5I). Consistent with this, clonogenic assays revealed that Y2 + gPARP1 restored olaparib sensitivity in PCBP2-overexpressing HCC1937 cells, as evidenced by decreased colony numbers (Supplementary Fig. S5J). Together, these results provide functional evidence that YTHDF2-mediated degradation of PARP1 is sufficient to reverse PCBP2-induced olaparib resistance.

Clinical relevance of the PCBP2–m⁶A–PARP1 drug resistance axis in patients with breast cancer

To discern the clinical relevance of the PCBP2–m⁶A–PARP1 axis in breast cancer, we assessed tumor samples from 10 patients with breast cancer who had undergone therapy with olaparib at Shantou University Medical College between 2023 and 2025. The patients were divided into two groups based on their clinical response to olaparib: the olaparib-sensitive group (n = 5) and the olaparib-resistant group (n = 5). IHC analysis revealed that PCBP2 and PARP1 protein levels were significantly elevated in the olaparib-resistant group, whereas γH2AX expression was reduced, indicating impaired DNA damage response. In contrast, the expression levels of METTL3 and YTHDF2 showed no significant differences between the two groups (Fig. 6A and B), consistent with our in vitro findings. These results suggest that PCBP2 promotes PARP1 mRNA stability and olaparib resistance not by altering the overall expression of m⁶A-related enzymes but likely through interfering with the recruitment of the m⁶A modification machinery to the PARP1 transcript. Collectively, our clinical observations corroborate the proposed mechanism that PCBP2 inhibits m⁶A-mediated decay of PARP1 mRNA, thereby contributing to PARPi resistance in BRCA1-mutant breast cancer.

Figure 6.

Clinical relevance of the PCBP2–m⁶A–PARP1 drug resistance axis in patients with breast cancer. A, Representative IHC staining of PCBP2, PARP1, γH2AX, METTL3, and YTHDF2 in breast cancer patient samples categorized by clinical response to PARPis (olaparib-sensitive vs. olaparib-resistant). Scale bar, 50 μm (high magnification). B, Quantitative IHC analysis of PCBP2, PARP1, γH2AX, METTL3, and YTHDF2 expression in samples of patients with breast cancer (n = 10, 5 olaparib-sensitive and 5 olaparib-resistant). Patients with olaparib resistance showed significantly higher PCBP2 and PARP1 expression and lower γH2AX levels compared with sensitive patients, whereas METTL3 and YTHDF2 expression showed no significant differences. Data are presented as the mean ± SEM. ns, not significant; *, P < 0.05; **, P < 0.01.

Discussion

RBPs play a crucial role in determining the fate of downstream RNA molecules, and increasing evidence links RBP dysregulation to cancer progression, altered tumor metabolism, drug resistance, cancer stem cell self-renewal, and immune evasion (41). We show a direct role for PCBP2 in regulating PARP1 mRNA stability, contributing to the development of resistance to olaparib therapy in BRCA-mutated breast cancer. Our findings indicate that targeting PCBP2 or its key downstream targets may be a promising approach to overcome resistance and improve the efficacy of olaparib therapy, offering new insights for molecular diagnosis and treatment of BRCA-mutated breast cancer.

The importance of PCBP2 as an RBP in transcriptional and translational regulation, as well as its impact on cancer progression, has been previously discussed (42). PCBP2 has been recognized as an oncogene in various cancer types, with growing evidence indicating its overexpression in a range of malignancies, including hepatocellular carcinoma (HCC; ref. 35), breast cancer (22), gastric cancer (43), esophageal squamous cell carcinoma (44), cervical cancer (45), and pancreatic ductal adenocarcinoma (46). Wang and colleagues (47) reported that PCBP2 is highly expressed in breast cancer, with its overexpression correlating with disease progression and poor prognosis. We show the induction of olaparib resistance is associated with elevated PCBP2 levels, a correlation further validated through functional assays. Importantly, specific knockdown of PCBP2 expression in olaparib-resistant breast cancer cells restored sensitivity to PARPi-targeted therapy. Using xenograft models, we further demonstrated that targeting PCBP2 could potentially reverse resistance to olaparib in vivo. These findings indicate that PCBP2 holds promise as a therapeutic target for BRCA-mutated breast cancers, particularly in cases with acquired resistance to PARPi therapy.

RBPs play essential roles in nearly every aspect of posttranscriptional regulation by influencing the function and turnover of their target transcripts. Consequently, they act as key regulators of oncogene and tumor-suppressor expression across various cancer types (41). The complex and precise regulatory functions of RBPs are primarily mediated by protein–RNA interactions. Current research shows that HuR, an RBP, enhances p21 protein expression by binding to the 3′ UTR of p21 mRNA, increasing its stability (48). In contrast, AUF1 also binds to the p21 mRNA 3′ UTR but promotes its degradation, resulting in decreased p21 expression (49). In this study, we show that PCBP2 directly interacts with the 3′ UTR of PARP1 mRNA, subsequently increasing PARP1 expression by stabilizing its mRNA. Additionally, we found that m⁶A modification of PARP1 mRNA is mediated by the m⁶A methyltransferase complex (METTL3, METTL14, and WTAP), with methylation-dependent PARP1 mRNA decay occurring in a YTHDF2-dependent manner.

As one of the most common chemical modifications in eukaryotic RNAs, m⁶A plays crucial roles in regulating RNA stability, localization, translation, splicing, and transport (26). m⁶A RNA modification is mediated by a methyltransferase complex composed of METTL3, METTL14, and WTAP, in which METTL3 functions as the catalytic subunit, METTL14 provides an RNA-binding platform, and WTAP acts as a regulatory component (50). The outcomes of m⁶A methylation on posttranscriptional regulation largely depend on m⁶A readers, with YTHDF1, YTHDF2, and YTHDF3 playing key roles (51). We observed that m⁶A labeling induces PARP1 mRNA degradation through YTHDF2. A significant finding of this study is the link between PCBP2-mediated mRNA stability and m⁶A methylation. Using RIP and pulldown assays, we demonstrate that the m⁶A modification site on PARP1 mRNA is located near the PCBP2-binding region. Depletion of PCBP2 increased recruitment of the METTL3–METTL14–WTAP m⁶A methyltransferase complex to PARP1 mRNA, suggesting that PCBP2 may shield PARP1 mRNA from m⁶A-mediated destabilization. Supporting this model, knockdown of the METTL3–METTL14–WTAP m⁶A methyltransferase complex partially counteracted the destabilizing effects induced by PCBP2 depletion, underscoring the functional interplay between PCBP2 and m⁶A methylation in regulating PARP1 mRNA stability.

In the context of BRCA-mutated cancers, PARPis, such as olaparib, leverage the synthetic lethality concept by targeting cells deficient in HR repair pathways. Cells with BRCA1/2 mutations are particularly sensitive to PARP inhibition, as they lack efficient HR repair mechanisms, relying instead on PARP1-mediated BER to manage DNA damage (52). However, our findings indicate that increased PARP1 expression may disrupt this synthetic lethality by enhancing DNA repair capabilities, including BER, thus leading to olaparib resistance. We show that PCBP2 directly binds to the 3′ UTR of PARP1 mRNA, stabilizing it and upregulating PARP1 expression. This stabilization was observed both in vitro and in vivo, in which olaparib-resistant models showed markedly higher levels of PARP1. This upregulation likely promotes DNA repair via the BER pathway, compensating for HR deficiency and undermining the intended effects of PARP inhibition. Consequently, the elevated PARP1 levels in resistant cells restore DNA repair capability, allowing them to survive despite PARPi therapy. Knockdown of PCBP2 in olaparib-resistant breast cancer cells restores PARPi sensitivity, suggesting that PCBP2-mediated stabilization of PARP1 mRNA is a key contributor to resistance. This functional link between PARP1 expression and olaparib resistance highlights PCBP2 as a potential therapeutic target. By targeting PCBP2 or disrupting its interaction with PARP1 mRNA, we could effectively reduce PARP1 expression, impair BER, and reestablish synthetic lethality in BRCA-mutated cancers. This strategy could offer a novel approach to overcoming PARPi resistance and improving the efficacy of olaparib therapy in patients with breast cancer with BRCA mutations.

m⁶A methylation is a dynamic and context-dependent RNA modification, and its targeting as an anticancer strategy requires careful evaluation. Whereas most studies have highlighted the therapeutic benefits of inhibiting m⁶A erasers such as FTO or ALKBH5 to restore m⁶A-mediated mRNA decay and sensitize tumors to therapy, there is growing recognition that loss of m⁶A modifications—whether through increased demethylase activity, reduced methyltransferase function, or regulatory RBPs such as PCBP2—may promote resistance to a variety of anticancer treatments, including PARPis. Our data suggest that PCBP2 suppresses m⁶A methylation on and thereby stabilizes PARP1 mRNA, leading to enhanced DNA repair capacity, which compromises the synthetic lethality induced by PARPi in BRCA-mutant cancer cells. This supports the broader hypothesis that m⁶A loss can be a mechanism of resistance, and targeting the m⁶A pathway without understanding the precise transcriptomic context may inadvertently stabilize oncogenic transcripts, such as PARP1. Therefore, the impact of manipulating m⁶A should be considered in a pathway-specific and cell type–specific manner. This concept is not limited to breast cancer, as similar m⁶A-dependent mechanisms of drug resistance have been observed in leukemia, glioblastoma, and lung cancer (53, 54).

Given the pivotal role of PCBP2 in stabilizing PARP1 mRNA and conferring resistance to PARP inhibition via m⁶A methylation suppression, our findings suggest that dual-targeting of PCBP2 and the m⁶A regulatory axis may offer a promising combinatorial therapeutic strategy to overcome PARPi resistance in BRCA-mutated breast cancers. Mechanistically, PCBP2 functions as a protective RBP that blocks the recruitment of the METTL3/METTL14/WTAP methyltransferase complex to the 3′ UTR of PARP1 mRNA, thereby preventing m⁶A deposition and subsequent YTHDF2-mediated degradation. Inhibition of PCBP2 not only restores m⁶A levels on PARP1 transcripts but also reactivates m⁶A-dependent decay pathways, resulting in downregulation of PARP1 protein expression and impaired DNA repair capacity (39). Combining PCBP2 silencing with pharmacologic activation of m⁶A pathways—for example, via inhibition of demethylases such as FTO or ALKBH5—may synergistically reduce PARP1 expression and reestablish synthetic lethality in HR-deficient cells. Clinically, this combinatorial approach could be especially valuable for patients with acquired or intrinsic resistance to PARPi monotherapy, offering a rational and targeted method to sensitize resistant tumors. Moreover, recent advances in programmable m⁶A editing systems using CRISPR/dCas13-based methyltransferase tethering have demonstrated the feasibility of locus-specific RNA methylation modulation in vivo (37). Such tools could potentially be used to selectively increase m⁶A modification at the PARP1 3′ UTR in resistant tumors, thereby degrading PARP1 transcripts and enhancing the efficacy of PARPi treatment. Furthermore, the development of small-molecule inhibitors of PCBP2, although still at a nascent stage, may complement these strategies and broaden the therapeutic window in resistant cancers.

In summary, our study not only identifies a novel PCBP2–m⁶A–PARP1 regulatory axis driving olaparib resistance but also proposes a combinatorial therapeutic concept that could be rapidly translatable to the clinic. Future work should explore the pharmacokinetics, safety, and efficacy of dual PCBP2/m⁶A-targeting regimens in preclinical BRCA-mutated models and ultimately in clinical trials.

Supplementary Material

Figure S1

Supplementary Figure S1 supports the Results section titled "PCBP2 is associated with olaparib sensitivity".

Figure S2

Supplementary Figure S2 supports the Results section titled "Stable knockdown of PCBP2 enhances olaparib sensitivity in olaparib-resistant breast cancer cells both in vitro and in vivo".

Figure S3

Supplementary Figure S3 supports the Results section titled "Depletion of PCBP2 facilitates m6A methyltransferase complex recruitment and accelerates PARP1 mRNA degradation".

Figure S4

Supplementary Figure S4 supports the Results section titled "Depletion of PCBP2 facilitates m6A methyltransferase complex recruitment and accelerates PARP1 mRNA degradation".

Figure S5

Supplementary Figure S5 supports the Results section titled "YTHDF2-dependent regulation of PARP1 mRNA stability modulates olaparib sensitivity".

Table S1

Supplementary Table S1: Target sequences of siRNAs and shRNAs

Table S2

Supplementary Table S2: Guide RNAs used for the targeted RNA methylation system.

Table S3

Supplementary Table S3: Primers used for quantitative RT-PCR

Table S4

Supplementary Table S4: Sequences of primers used for pcDNA3.1-PARP1 mRNA construction

Table S5

Supplementary Table S5: Clinical characteristics of 10 human BRCA-mutated breast cancer tissue samples

Data Availability

The data analyzed in this study were obtained from publicly available databases: Breast Cancer Gene-Expression Miner version 4.9 (bc-GenExMiner version 4.9; http://bcgenex.ico.unicancer.fr/BC-GEM/GEM-requete.php), catRAPID (http://s.tartaglialab.com/page/catrapid_group), and WHISTLE (https://whistle-epitranscriptome.com/). All data generated in this study are included in the article and its supplementary files. Further details or specific requests can be addressed by contacting the corresponding author.

Authors’ Disclosures

No disclosures were reported.

Authors’ Contributions

Z. Qi: Software, formal analysis, validation, methodology, writing–original draft. L. He: Resources, formal analysis. Z. Xu: Funding acquisition, validation. X. Luo: Methodology. L. Ji: Software. C. Lin: Validation, methodology. A.E. Giuliano: Writing–original draft. X. Cui: Writing–original draft. Z. Deng: Methodology. J. Wu: Conceptualization, resources. S.L. Lin: Conceptualization, formal analysis, methodology, writing–original draft. Y. Cui: Conceptualization, formal analysis, funding acquisition, methodology, writing–review and editing.