Claudin-4 Stabilizes the Genome via Nuclear and Cell-Cycle Remodeling to Support Ovarian Cancer Cell Survival

Abstract

Abstract: Alterations in the interplay between the nucleus and the cell cycle during cancer development lead to a state of genomic instability, often accompanied by observable morphologic aberrations. Tumor cells can regulate these aberrations to evade cell death, either by preventing or eliminating genomic instability. In epithelial ovarian cancer, overexpression of claudin-4 significantly contributes to therapy resistance through mechanisms associated with genomic instability regulation. However, the molecular mechanisms underlying claudin-4 overexpression in epithelial ovarian cancer remain poorly understood. In this study, we modified claudin-4 expression and employed a unique claudin mimic peptide to investigate claudin-4’s function. Our findings show that claudin-4 supports ovarian cancer cell survival by stabilizing the genome through nuclear and cell-cycle remodeling. Specifically, claudin-4 induced nuclear constriction by excluding lamin B1 and promoting perinuclear F-actin accumulation, thereby altering nuclear structure and dynamics. Similarly, cell-cycle modifications due to claudin-4 overexpression resulted in fewer cells entering the S-phase and reduced genomic instability in tumors. Importantly, disrupting claudin-4’s biological effects using claudin mimic peptide and forskolin increased the efficacy of PARP inhibitor treatment, correlating with alterations in the oxidative stress response. Our data indicate that claudin-4 protects tumor genome integrity by modulating the crosstalk between the nucleus and the cell cycle, leading to resistance to genomic instability formation and the effects of genomic instability–inducing agents.

Significance: High-grade serous ovarian carcinoma is marked by chromosomal instability, which can serve to promote disease progression and allow cancer to evade therapeutic insults. The report highlights the role of claudin-4 in regulating genomic instability and proposes a novel therapeutic approach to exploit claudin-4-mediated regulation.

Figure 1

Cell-cycle progression during claudin-4 modulation. Ovarian tumor cells were cultured and stained for PI at 24, 48 (OVCAR8 and OVCA429 cells), and 96 hours (OVCAR3 cells). Subsequently, cell-cycle progression was evaluated via flow cytometry. A, Representative histograms of cell-cycle phases during claudin-4 overexpression; right, percentages of cells at each cell-cycle stage. Similarly, the effect of claudin-4 downregulation in OVCA429 (B) and OVCAR3 (C). D, Model illustrating that claudin-4 overexpression reduces the proportion of tumor cells in the S-phase of the cell cycle, whereas its downregulation results in an accumulation of cells in the G2–M phase and a decrease in the G0–G1 phase. (four independent experiments, three independent experiments for OVCAR3 at 96 hours; two-tailed unpaired t test; significance P < 0.05). Graphs show min to max (+ indicates mean). KD, knockdown; PI, propidium iodide.

Figure 2

Claudin-4’s association with various forms of genome instability. Ovarian tumor cells were cultured for 24, 48, and 96 hours. Afterward, cells were PI-stained to quantify aneuploidy, a type of genomic instability in vitro. Likewise, another type of genomic instability (chromosomic amplifications) was quantified in human tumor samples [TCGA; accession number PHS000178; 2.32% (95% CI, 2.28–2.36) vs. 5.00% (95% CI, 4.93–5.08); P < 0.0001]. A, Percentages of hypertetraploid (% of variation relative to hypertetraploid observed in WT cells), a form of aneuploidy, during claudin-4 overexpression in OVCAR8 or its downregulation in OVCA429 (B) and OVCAR3 cells (C), respectively. D, Correlation of genomic instability (indicated as % of altered chromosomic copy numbers) in human ovarian tumors associated with levels of claudin-4 expression (four independent experiments, and three independent experiments for OVCAR3 at 96 hours; two-tailed unpaired t test; significance P < 0.05). Graphs show the mean and SEM. KD, knockdown; PI, propidium iodide.

Figure 3

Remodeling of nuclear morphology and the nuclear lamina during claudin-4 disruption. Ovarian tumor cells were treated with CMP (400 µmol/L) for 48 hours, and then cells were stained to mark the nuclear lamina (using antibodies against lamin B1 and lamin A/C) and the nuclei (using DAPI). Subsequently, we performed a morphometric characterization. A, Top, confocal images showing (maximum projections) nuclei size of OVCAR8 as well as OVCA429 (B) and OVCAR3 cells (C). Bottom, corresponding quantification for (A–C; n = OVCAR8, 1711 cells; OVCA429, 2,630 cells; OVCAR3, 2,365 cells; two-tailed Mann–Whitney test, Kruskal–Wallis test with Dunn’s multiple comparisons; three independent experiments). D, Top, selected confocal images (maximum projections) showing nuclear lamina components corresponding to OVCAR8 cells overexpressing claudin-4 and claudin-4 downregulation in OVCA429 (E) and OVCAR3 cells (F). Bottom, corresponding quantification of nuclear accumulation of lamin B1 and lamin A/C (yellow arrowheads highlight comparison of lamin B1 during claudin-4 overexpression and downregulation) under different conditions (three independent experiments; Kruskal–Wallis test with Dunn’s multiple comparisons. (Significance, P < 0.05). Graphs show mean and SEM, scale bar, 5 µm. KD, knockdown; MFI, mean fluorescence intensity; Veh, vehicle.

Figure 4

Claudin-4’s effect on the actin cytoskeleton. Ovarian tumor cells were treated with CMP (400 µmol/L) for 48 hours and stained to mark the actin cytoskeleton using phalloidin. In addition, cells were engineered to express LifeAct to mark the actin cytoskeleton in living cells. Afterward, we performed a morphologic and kymograph analysis. A, Top, reconstructions of perinuclear F-actin and genomic DNA (from confocal z-stacks) for OVCAR8, OVCA429 (B), and OVCAR3 cells (C). Bottom, corresponding quantification of perinuclear F-actin for A–C, respectively. OVCAR8, 1711 cells; OVCA429, 2630 cells; OVCAR3, 2365 cells; two-tailed Mann–Whitney test, Kruskal–Wallis test with Dunn’s multiple comparisons). D, Remodeling effect of claudin-4 in the nuclear architecture, considering both nuclear lamina and perinuclear F-actin. E and F, left, Quantification of junctional F-actin from reconstructions (from confocal z-stacks). Right, confocal images (maximum projection) and zoom, followed by reconstruction of selected ROIs (at junctional F-actin) from OVCA429 cells (top) and OVCAR3 cells (bottom), respectively (n = OVCA429, 783 cells; OVCAR3, 825 cells; Kruskal–Wallis test with Dunn’s multiple comparisons). G and H, Kymographs illustrating the movement of junctional F-actin (vertical gray arrow) over time (horizontal blue arrow), generated from different ROIs during confocal live-cell imaging of transduced OVCA429 cells (top; n = 142) with LifeAct to mark F-actin (without any stimuli and cultured for 24 hours) and OVCAR3 cells (bottom; n = 116), respectively (two-tailed Mann–Whitney test; three independent experiments; significance, P < 0.05). Graphs show mean and SEM. KD, knockdown; MFI, mean fluorescence intensity; ROI, regions of interest; Veh, vehicle.

Figure 5

Effect of combining olaparib, FSK, and CMP on LAT1 expression. Ovarian tumor cells were treated with a tripartite combination of olaparib (600 nmol/L), FSK (5 µmol/L), and CMP (400 µmol/L) for different time points. Subsequently, cell lysates were obtained to carry out immunoblotting for LAT1. A–C LAT1 protein expression at different time points in OVCAR8, OVCA429, and OVCAR3 cells, respectively. On the right are graphs showing the corresponding quantification of LAT1 from A–C relative to loading control. KD, knockdown; Veh/veh, vehicle.

Figure 6

Impact of targeting claudin-4’s functional effects via CMP and FSK on ovarian cancer cell survival. Olaparib treatment was used as a reference (Supplementary Fig. S6D–S6F) to evaluate the effects of CMP and FSK on cell survival using the 7-day colony formation assay and crystal violet staining. Cells were treated as follows: two cycles of treatment (at days 0 and 3) for OVCAR8 and OVCAR3 cells and one treatment for OVCA429 (at day 0). A, Percentage of tumor cell survival during olaparib treatment vs. olaparib + FSK (5 µmol/L) and claudin-4 overexpression, and similar information during claudin-4 downregulation in OVCA429 cells (B) and OVCAR3 cells (C). D, Percentage of tumor cell survival during olaparib + FSK (5 µmol/L) vs. olaparib + FSK (5 µmol/L) + CMP (400 µmol/L) and claudin-4 overexpression, and similar information during claudin-4 downregulation in OVCA429 cells (E) and OVCAR3 cells (F), respectively (three independent experiments; two-way ANOVA; significance P < 0.05). Graphs show mean and SEM. KD, knockdown; Olap/olap, olaparib; Veh/veh, vehicle.

Figure 7

Cellular stress response evaluation in ovarian tumor cells treated with olaparib, FSK, and CMP. Ovarian tumor cells were treated with a tripartite combination of olaparib (600 nmol/L), FSK (5 µmol/L), and CMP (400 µmol/L) for 4 hours to analyze ROS as well as a reporter gene for HIF-1α via flow cytometry. The same cells were treated similarly for 96 hours and then stained using immunofluorescence to mark lamin B1. ROS generation is indicated as normalization relative median of OVCAR8 WT, OVCA429 WT, and OVCAR3 WT cells without treatment (A; two independent experiments; unpaired t test, red rectangle; one-way ANOVA and Tukey multiple comparison test, P < 0.05). Reported HIF-1α is indicated as normalization relative median of OVCAR8 WT, OVCA429 WT, and OVCAR3 WT cells without treatment (B; three independent experiments; unpaired t test; one-way ANOVA and Tukey multiple comparison test, P < 0.05). C and D, Confocal images showing HIF-1α during claudin-4 (CLDN4) overexpression in OVCAR8 cells treated or not as indicated above at 96 hours. E and F, Confocal images showing HIF-1α during claudin-4 downregulation in OVCA429 cells treated or not as indicated above at 96 hours. G and H, Confocal images showing HIF-1α during claudin-4 downregulation in OVCAR3 cells treated or not as indicated above at 96 hours. Graphs show the median with 95% CI. Scale bar, 10 µm. CI, confidence interval; KD, knockdown; Veh, vehicle.