Introduction
Ovarian cancer remains the most lethal gynecologic malignancy, in part because many patients present with advanced disease and subsequently develop recurrence or treatment resistance [1, 2, 3, 4, 5]. Cytoreductive surgery and platinum‐based chemotherapy remain central to treatment. Poly (ADP‐ribose) polymerase (PARP) inhibitors have improved progression‐free survival (PFS), particularly in tumors with homologous recombination deficiency, but acquired resistance and an inconsistent overall survival (OS) benefit remain important limitations [6, 7, 8, 9, 10, 11].
Defining molecular regulators of ovarian cancer growth and DNA‐damage responses may therefore identify biomarkers and therapeutic vulnerabilities.
UHRF1 is a multidomain epigenetic regulator containing ubiquitin‐like (UBL), tandem Tudor (TTD), plant homeodomain (PHD), SET‐ and RING‐associated (SRA), and RING domains. Through recognition of histone marks and hemimethylated DNA, UHRF1 coordinates DNA methylation, chromatin regulation, and protein ubiquitination. Its depletion can also increase cellular sensitivity to genotoxic stress [12, 13, 14].
UHRF1 is overexpressed in several cancers, including colorectal, lung, and pancreatic cancer, where it supports oncogenic phenotypes through epigenetic and metabolic mechanisms [15, 16, 17, 18, 19, 20]. PARP1 has also been reported to associate with UHRF1 in heterochromatin regulation and methylation‐dependent homologous recombination [21, 22]. Whether UHRF1 reciprocally controls PARP1 protein turnover in ovarian cancer has not been established.
Ubiquitination is a sequential post‐translational modification catalyzed by E1 activating enzymes, E2 conjugating enzymes, and E3 ligases that regulates protein homeostasis and cancer biology [23, 24]. Different ubiquitin‐chain linkages can determine distinct protein fates: K48‐ and K11‐linked chains commonly promote proteasomal degradation, whereas K63‐linked chains frequently regulate protein interactions and signaling [25, 26]. RNF4 is a SUMO‐targeted RING E3 ubiquitin ligase that can ubiquitinate PARP1 under stress or PARP‐trapping conditions, promoting PARP1 processing or turnover [27, 28].
Mounting evidence implies that E3 ubiquitin ligases and deubiquitinating enzymes represent attractive therapeutic targets for boosting anti‐tumor immunity in cancer immunotherapy [26, 29]. Whether UHRF1 influences this pathway in ovarian cancer has not been established.
We investigated whether UHRF1 contributes to ovarian cancer progression and regulates PARP1 protein stability. By combining clinical analyses, UHRF1‐depletion models, interaction proteomics, assays of protein turnover, and PARP1 re‐expression, we identify UHRF1 as a PARP1‐associated protein and test a model in which UHRF1 limits PARP1 polyubiquitination and proteasomal turnover.
Materials and Methods
Study Design
The study integrated public‐dataset analyses, validation in human ovarian tissues, functional studies in ovarian cancer cell lines, two mouse models, and biochemical analyses of UHRF1‐associated proteins. Experiments compared stable UHRF1‐knockdown cells with a non‐targeting shRNA control. PARP1 re‐expression was used to test whether reduced PARP1 abundance contributed to the cellular phenotypes of UHRF1 depletion.
Bioinformatic Analyses
Pan‐cancer UHRF1 transcript abundance was examined using GEPIA. UHRF1 and PARP1 protein abundance were evaluated through UALCAN using Clinical Proteomic Tumor Analysis Consortium datasets. The association between UHRF1 transcript abundance and survival was assessed using Kaplan–Meier Plotter, including analyses of OS, PFS, and the optimal‐debulking subgroup.
Patients were divided using the platform‐defined optimal cutoff. PARP1 expression by stage and grade was examined using CSIOVDB [30, 31].
Patients and Tissue Specimens
Fresh tissue samples, including 10 ovarian cancer specimens and 10 benign ovarian controls, were collected from the First Affiliated Hospital of Sun Yat‐sen University from October 2020 to December 2022. Tissue microarray samples were obtained from surgical patients treated at the same institution from January 2014 to December 2022. All subjects provided written informed consent, and this study was approved by the Ethics Committee of the hospital (approval no. [2021]726).
Cell Lines
Human ovarian cancer cell lines A2780, CAOV3, ES2, OVCAR8, and SKOV3, the immortalized ovarian epithelial cell line IOSE80, and HEK293T cells were maintained at 37°C in a humidified atmosphere containing 5% CO2. Cells were cultured in DMEM or RPMI 1640 supplemented with 10% fetal bovine serum and 1% penicillin–streptomycin, as appropriate for each line.
Gene Knockdown and Plasmid Transfection
Stable UHRF1‐knockdown A2780 and SKOV3 cells were generated using two independent UHRF1‐targeting shRNAs (sh10 and sh11); a non‐targeting shRNA (shNC) served as the control. Knockdown was confirmed by immunoblotting. Tagged full‐length and deletion constructs of UHRF1 and PARP1, HA‐tagged ubiquitin, and UHRF1‐targeting siRNA were introduced into the indicated cells for interaction, ubiquitination, and rescue experiments.
RNA Extraction and Quantitative Real‐Time PCR
Total RNA was extracted using TRIzol reagent (Thermo Fisher Scientific) and reversely transcribed into cDNA following the kit instructions (Takara). qRT‐PCR was performed using a SYBR Green PCR Kit (EZBioscience). Gene expression levels were normalized to GAPDH, and relative quantification was determined via the 2−ΔΔCt algorithm.
Immunoblotting and Co‐Immunoprecipitation
Cells were lysed in RIPA buffer containing protease inhibitors. Proteins were separated by SDS–PAGE, transferred to PVDF membranes, blocked with 5% non‐fat milk, incubated with primary antibodies overnight at 4°C, and detected with horseradish‐peroxidase‐conjugated secondary antibodies and enhanced chemiluminescence. Primary antibodies included UHRF1 (Abcam, ab213223), PARP1 (Cell Signaling Technology, 9532), RNF4 (Proteintech, 17810‐1‐AP), Myc tag (Cell Signaling Technology, 2276), Flag tag (Cell Signaling Technology, 14793), HA tag (Cell Signaling Technology, 3724), and GAPDH (Proteintech, 60004‐1‐Ig).
For co‐immunoprecipitation, clarified lysates were incubated with the indicated antibodies and protein A/G beads; washed complexes were analyzed by immunoblotting or silver staining.
Immunohistochemistry
Formalin‐fixed paraffin‐embedded sections were deparaffinized, rehydrated, subjected to antigen retrieval, blocked, incubated with primary antibodies, and visualized using a polymer detection system (ZSGB‐Bio, PV‐6000) and 3,3′‐diaminobenzidine. Cleaved caspase‐3 was detected with Proteintech antibody 25128‐1‐AP. UHRF1 staining was evaluated in the tissue microarray and mouse tumors.
Cell Proliferation, Clonogenic Growth, and Migration
Cell proliferation was measured using Cell Counting Kit‐8. Clonogenic growth was assessed by plating cells at low density, allowing colonies to form, and staining with crystal violet. Migration was assessed in 8‐μm‐pore Transwell inserts.
Results were normalized to the corresponding shNC group where indicated.
Apoptosis and γH2AX Immunofluorescence
Apoptosis was quantified by flow cytometry after Annexin V‐Alexa Fluor 647/propidium iodide staining. γH2AX immunofluorescence was examined in A2780 and SKOV3 cells, with nuclei counterstained using DAPI, under basal conditions and after cisplatin exposure (10.41 μM overnight). The relative fluorescence intensity of γH2AX foci was quantified.
Interaction Proteomics and RNA Sequencing
Flag‐UHRF1‐associated proteins were isolated from transfected HEK293T cells and visualized by silver staining. Candidate interactors were identified by mass spectrometry and analyzed for pathway enrichment. RNA sequencing was performed in stable UHRF1‐knockdown and control cells, followed by differential‐expression and Gene Ontology enrichment analyses.
Protein Turnover, Proteasome Inhibition, and Ubiquitination Assays
For experiments involving RNF4, the association between RNF4 and PARP1 was examined in the indicated UHRF1‐depletion and reconstitution groups. In separate experiments, RNF4‐targeting or non‐targeting siRNA was introduced into the indicated UHRF1 backgrounds to assess PARP1 turnover and ubiquitination. Full‐length UHRF1 or the TTD‐deletion construct (D2) was re‐expressed in the indicated knockdown cells for the domain‐rescue experiments.
For the domain‐rescue experiments, stable shNC and sh10 A2780 and SKOV3 cells were transfected with the indicated control plasmid, full‐length UHRF1‐Flag, or UHRF1‐D2‐Flag. D2 lacks amino acids 133–284, encompassing the TTD‐containing region. For co‐immunoprecipitation, lysates were prepared under non‐denaturing conditions and subjected to anti‐Flag immunoprecipitation; co‐precipitated PARP1 and immunoprecipitated Flag‐tagged protein were detected by immunoblotting.
CHX‐chase experiments used the same treatment conditions, with samples collected at 0, 6, 12, 18, and 24 h. PARP1 signals were normalized to GAPDH and then to the 0‐h value within each group.
For the D2‐rescue ubiquitination assay, the indicated UHRF1 construct was co‐transfected with Myc‐PARP1 and HA‐ubiquitin in a separate set of samples. MG132 was applied using the same concentration and exposure duration. Myc‐PARP1 was immunoprecipitated with anti‐Myc, and PARP1‐associated HA‐ubiquitin was detected by immunoblotting with anti‐HA.
For the RNF4–PARP1 association assay, PARP1 was immunoprecipitated from the indicated control, UHRF1‐depleted, and UHRF1‐reconstituted groups; RNF4 and PARP1 were detected in the input and immunoprecipitated fractions. In a separate experiment, stable shNC and sh10 (shUHRF1) A2780 and SKOV3 cells were transfected with siNC or siRNF4 in a 2 × 2 design: shNC/siNC, shUHRF1/siNC, shNC/siRNF4, and shUHRF1/siRNF4. RNF4 knockdown was verified by immunoblotting in both cell lines (Figure 8B).
CHX‐chase treatment and sample collection followed the conditions described above.
For ubiquitination assays in the RNF4‐targeting siRNA experiment, Myc‐PARP1 and HA‐ubiquitin were co‐transfected into the indicated siNC‐ or siRNF4‐treated groups in a separate set of samples. MG132 treatment used the same concentration and exposure duration described above. Myc‐PARP1 immunoprecipitates were analyzed by anti‐HA immunoblotting to assess PARP1‐associated HA‐ubiquitin.
Mouse Models
Female nude mice were used to establish subcutaneous xenograft and peritoneal metastasis models. OC cells with stable UHRF1 knockdown were injected subcutaneously or intraperitoneally. Subcutaneous tumor weight and volume were measured at terminal collection.
Peritoneal lesions were documented by macroscopic photographs and in vivo bioluminescence imaging. Tumor tissues were collected for immunohistochemical analysis. Animal procedures were approved by the institutional committee (approval no. [2022]020).
Statistical Analysis
Analyses were performed using SPSS version 24.0 and GraphPad Prism. Associations between UHRF1 expression and clinicopathological characteristics were assessed using the χ2 test. OS and PFS were analyzed using Kaplan–Meier curves and log‐rank tests.
Prognostic variables were evaluated by univariable and multivariable Cox proportional‐hazards models and are reported as hazard ratios (HRs) with 95% confidence intervals (CIs). Two‐sided p < 0.05 was considered statistically significant.
Results
UHRF1 Is Elevated in Ovarian Cancer and Is Associated With Adverse Outcomes
Public datasets showed elevated UHRF1 transcript abundance across multiple tumor types and higher UHRF1 protein abundance in primary ovarian tumors than in normal samples (Figure 1A,B). In Kaplan–Meier Plotter, high UHRF1 expression was not significantly associated with OS in the overall cohort (HR 1.17, 95% CI 0.96–1.44; log‐rank p = 0.13), but was associated with shorter PFS (HR 1.30, 95% CI 1.04–1.63; p = 0.022). In the optimal‐debulking subgroup, high UHRF1 was associated with shorter OS (HR 1.75, 95% CI 1.04–2.93; p = 0.031) and PFS (HR 1.66, 95% CI 1.10–2.51; p = 0.015; Figure 1C).
In institutional specimens, UHRF1 mRNA and protein abundance were higher in ovarian cancer than in benign ovarian controls (Figure 1E,F). UHRF1 staining was evaluated across the 180‐case tissue microarray, and case‐level high/low categories were used in the statistical and survival analyses (Figure 1D). Figure 1G shows representative cores illustrating the high‐ and low‐staining categories.
Higher‐resolution images are provided in Supplementary Figures S2 and S3. High UHRF1 was associated with shorter OS (HR 1.838; log‐rank p = 0.0110) and PFS (HR 1.961; p = 0.0036; Figure 1D). In multivariable Cox models, high UHRF1 remained associated with OS (HR 2.73, 95% CI 1.40–5.34; p = 0.003) and PFS (HR 2.84, 95% CI 1.46–5.50; p = 0.002; Table S1).
UHRF1 Depletion Suppresses Ovarian Cancer Cell Growth and Migration and Increases Apoptosis
UHRF1 protein was detected in five ovarian cancer cell lines and was low in IOSE80 cells (Figure 2A). Two independent shRNAs reduced UHRF1 protein in A2780 and SKOV3 cells (Figure 2B). Both shRNAs reduced colony formation, Transwell migration, and CCK‐8 signal relative to shNC controls (Figure 2C–G), supporting a role for UHRF1 in clonogenic growth, migration, and proliferation.
UHRF1 depletion increased the Annexin V‐positive population in A2780 and SKOV3 cells (Figure 3A,B). γH2AX staining increased after UHRF1 depletion and was prominent after cisplatin exposure (Figure 3C) in both cell lines. Relative γH2AX foci fluorescence intensity was quantified (Figure 3D). These observations are consistent with increased apoptosis and DNA‐damage signaling after UHRF1 depletion.
UHRF1 Depletion Reduces Tumor Growth and Intraperitoneal Tumor Burden In Vivo
Subcutaneous tumors formed by UHRF1‐depleted A2780 cells were smaller and had lower weight and volume than shNC tumors at terminal collection (Figure 4A–C). Knockdown tumors showed lower UHRF1 staining and higher cleaved‐caspase‐3 staining (Figure 4D). In the intraperitoneal SKOV3 model, UHRF1 depletion was accompanied by lower bioluminescent tumor signal and fewer or smaller visible abdominal lesions (Figure 4E,F; higher‐resolution photographs are provided in Supplementary Figures S4 and S5), together with lower UHRF1 and higher cleaved‐caspase‐3 staining (Figure 4G).
UHRF1 Associates With PARP1 Through Defined Domain‐Containing Regions
Immunoprecipitation–mass spectrometry of Flag‐UHRF1 complexes identified proteins enriched in pathways related to DNA replication and repair (Figure 5A–C). RNA‐sequencing analysis of UHRF1‐depleted cells showed enrichment of apoptosis‐related Gene Ontology terms (Figure 5D). PARP1 was among the UHRF1‐associated proteins, with 58% sequence coverage and 59 unique peptides (Figure 5C).
Reciprocal co‐immunoprecipitation in HEK293T cells supported an association between tagged UHRF1 and PARP1 (Figure 5E).
To determine the domains responsible for the interaction between UHRF1 and PARP1, we employed a series of truncated UHRF1 [32] (Figure 5F) and PARP1 constructs [33] (Figure 5G). The results revealed that the TTD domain (aa133–284) of UHRF1 interacts with PARP1 (Figure 5H). Additionally, the ZF3 domain (aa227–374) of PARP1 was found to interact with UHRF1 (Figure 5I).
These findings identify PARP1 as a critical binding partner of UHRF1, supporting that UHRF1 regulates DNA repair and apoptosis in ovarian cancer via associating with PARP1.
UHRF1 Maintains PARP1 Protein Stability and Limits Its Polyubiquitination
PARP1 protein abundance was higher in primary ovarian tumors than in normal controls in CPTAC data and increased across the stage and grade groups shown in CSIOVDB [30, 31] (Figure S1). In A2780 and SKOV3 cells, UHRF1 depletion did not significantly change PARP1 mRNA but reduced PARP1 protein abundance (Figure 6A,B). Cycloheximide‐chase assays showed faster loss of PARP1 after UHRF1 depletion (Figure 6C,D).
Since PARP1 is mainly degraded via the proteasome [28], we hypothesized that UHRF1 knockdown promotes PARP1 proteasomal degradation. Treatment with the proteasome inhibitor MG132 restored PARP1 protein levels in UHRF1‐deficient cells (Figure 6E), indicating that UHRF1 supports PARP1 stability through a proteasome‐sensitive process. Furthermore, co‐transfection of Myc‐PARP1 and HA‐ubiquitin followed by immunoprecipitation showed increased PARP1‐associated polyubiquitin after UHRF1 depletion in both A2780 and SKOV3 (Figure 6F).
The ubiquitin‐linkage type was not determined. Ubiquitination is a post‐translational modification wherein ubiquitin is covalently attached to target proteins via an E1–E2–E3 enzymatic cascade. E3 ubiquitin ligases determine substrate specificity by recognizing specific target proteins.
As RNF4 is the primary E3 ligase mediating PARP1 proteasomal degradation [28, 34], we next examined whether UHRF1 abundance was associated with RNF4–PARP1 co‐immunoprecipitation. In HEK293T cells, UHRF1 overexpression reduced RNF4 recovered with Myc‐PARP1, whereas UHRF1 knockdown increased it (Figure 6G).
PARP1 Partially Rescues Effects of UHRF1 Depletion
To assess whether reduced PARP1 contributes to the phenotypes caused by UHRF1 depletion, we re‐expressed PARP1 in UHRF1‐knockdown A2780 and SKOV3 cells. PARP1 re‐expression partially restored colony formation (Figure 7A,B), cell migration (Figure 7C,D), and proliferation (Figure 7E) and attenuated the increase in apoptosis under low‐dose cisplatin exposure (Figure 7G,H). These results indicate that reduced PARP1 abundance contributes to the phenotypes caused by UHRF1 depletion.
The proposed working model is presented in Figure 7F.
UHRF1 TTD Contributes to PARP1 Stability
In A2780 and SKOV3 cells, anti‐Flag immunoprecipitation recovered more PARP1 with re‐expressed full‐length UHRF1 than with the D2 deletion construct, which lacks the TTD‐containing region (Figure 8A). In CHX‐chase experiments, full‐length UHRF1 restored PARP1 stability more effectively than D2 in UHRF1‐depleted cells (Figure 8C,D). The ubiquitination assay showed a corresponding reduction in PARP1‐associated HA‐ubiquitin with full‐length UHRF1, whereas D2 did not restore the signal to the same extent (Figure 8E).
These findings identify the TTD‐containing region as a functional contributor to UHRF1–PARP1 association and PARP1 stability.
RNF4 Contributes to PARP1 Turnover
Immunoblotting confirmed lower RNF4 protein in siRNF4‐treated than in matched siNC groups in both A2780 and SKOV3 cells (Figure 8B). To examine RNF4–PARP1 association in the context of UHRF1 depletion and rescue, we analyzed PARP1 immunoprecipitates from shNC+NC‐Flag, shUHRF1+NC‐Flag, shUHRF1+UHRF1‐WT, and shUHRF1+UHRF1‐D2 cells. RNF4 input levels were comparable across these groups, whereas RNF4 recovery with PARP1 increased after UHRF1 depletion and decreased after re‐expression of UHRF1‐WT; the D2 construct was less effective in reducing this recovery (Figure 8F).
In separate experiments, siRNF4‐treated cells showed slower PARP1 loss during CHX chase than their matched siNC groups in both shNC and shUHRF1 (sh10) backgrounds (Figure 8G,H), together with a lower PARP1‐associated ubiquitin signal (Figure 8I). Together, these results support a contribution of RNF4 to PARP1 turnover and indicate that UHRF1 status affects PARP1–RNF4 co‐recovery.
Discussion
This study links UHRF1 abundance to PARP1 protein stability in ovarian cancer models. UHRF1 was elevated in clinical specimens and associated with adverse outcomes in the institutional tissue‐microarray cohort. In cell models, UHRF1 depletion reduced growth‐associated phenotypes, increased apoptosis, shortened PARP1 persistence during CHX chase, and increased PARP1‐associated polyubiquitin.
Domain‐rescue experiments show that the TTD‐containing region contributes to UHRF1–PARP1 association, while the RNF4‐targeting siRNA experiments support a contribution of RNF4 to PARP1 destabilization after UHRF1 depletion. Together, these findings support a role for UHRF1 in maintaining PARP1 stability in ovarian cancer models.
The clinical analyses support the biological observations while also defining their limits. Public transcript data associated UHRF1 with PFS but not OS in the overall cohort, whereas the institutional tissue microarray associated high UHRF1 protein with both outcomes. These differences may reflect assay modality, cohort composition, cutoff selection, treatment, and follow‐up.
Although multivariable models supported an association independent of the included clinicopathological variables, external validation, and prespecified cutoffs are needed before UHRF1 can be considered a prognostic biomarker.
The unchanged PARP1 mRNA, accelerated PARP1 loss during CHX treatment, MG132 rescue, and increased PARP1‐associated polyubiquitin are consistent with post‐translational regulation. Domain mapping identified regions that contribute to recovery of UHRF1 and PARP1 in the same complexes. The D2 deletion construct showed weaker association and less effective rescue than full‐length UHRF1, supporting a functional contribution of the TTD‐containing region.
Further biochemical studies can define the molecular basis of this association.
The RNF4‐related experiments address two distinct questions. In Figure 8F, RNF4 input was comparable across the UHRF1 control, depletion, and reconstitution groups, whereas RNF4 recovery with PARP1 increased after UHRF1 depletion and was lower after full‐length UHRF1 re‐expression; D2 was less effective. In the separate siRNA experiments, immunoblotting confirmed RNF4 knockdown in both cell lines (Figure 8B). siRNF4‐treated cells showed slower PARP1 loss during CHX chase and a lower PARP1‐associated ubiquitin signal, including in the UHRF1‐depleted background (Figure 8G–I).
Together, the co‐immunoprecipitation and siRNA results support a contribution of RNF4 to PARP1 turnover and indicate that UHRF1 status affects RNF4–PARP1 association.
The study used two established ovarian cancer cell lines, and additional tumor models will help define the generalizability of these findings. γH2AX foci‐associated fluorescence intensity was quantified in both cell lines. Time‐course measurements could further characterize DNA‐damage resolution. Subcutaneous tumor weight and volume were assessed at the experimental endpoint, and representative bioluminescence images illustrate the peritoneal model.
Future studies incorporating longitudinal tumor measurements and ROI‐based imaging quantification would extend these in vivo findings. The study also did not test PARP‐inhibitor response or platinum‐resistant models, important directions for future work.
In conclusion, UHRF1 supports ovarian cancer cell growth‐associated phenotypes and maintains PARP1 protein stability. The available data indicate that the TTD‐containing region contributes to UHRF1–PARP1 association, while the RNF4‐targeting siRNA results support a contribution of RNF4 to PARP1 destabilization after UHRF1 depletion. Direct competition and ubiquitin‐linkage specificity remain unresolved.
Author Contributions
Yuying Chen: writing – original draft, visualization, formal analysis. Linxiang Wu: validation, writing – original draft, formal analysis. Jihui Kang: writing – original draft, resources.
Huishan Xu: investigation, visualization. Qin Ling: investigation, methodology. Siqi Yang: investigation, methodology.
Tingting Sun: funding acquisition, data curation. Huiling Lai: funding acquisition, data curation. Hao Tan: investigation.
Guofen Yang: conceptualization, writing – review and editing. Weipeng He: conceptualization, supervision, writing – review and editing.
Funding
This study was supported by the National Natural Science Foundation of China (grant 82303484 to Tingting Sun and grant 82303279 to Huiling Lai).
Ethics Statement
Human tissue procedures were approved by the Ethics Committee of The First Affiliated Hospital of Sun Yat‐sen University (approval no. [2021]726), and all participants provided written informed consent. Animal experiments were approved by the Institutional Ethics Committee for Clinical Research and Animal Experiments of The First Affiliated Hospital of Sun Yat‐sen University (approval no. [2022]020).
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information