Increased expression of BRD4 isoforms long (BRD4-L) and short (BRD4-S) promotes chemotherapy resistance in high-grade serous ovarian carcinoma

Abstract

Chemoresistance in ovarian carcinoma is a puzzling issue that urges understanding of strategies used by cancer cells to survive DNA damage and to escape cell death. Expanding efforts to understand mechanisms driving chemoresistance and to develop alternative therapies targeting chemoresistant tumors are critical. Amplification of BRD4 is frequently associated with chemoresistant ovarian carcinoma, but little is known about the biological effects of the overexpression of BRD4 isoforms in this malignancy. Here, we described the consequences of BRD4-L and BRD4-S overexpression in ovarian carcinoma shedding a light on a complex regulation of BRD4 isoforms. We demonstrated that the BRD4-L transcript expression is required to generate both isoforms, BRD4-L and BRD4-S. We showed that the BRD4-S mRNA expression positively correlated with BRD4-S protein levels, while BRD4-L isoform showed negative correlation between mRNA and protein levels. Moreover, we demonstrated that an overexpression of BRD4 isoforms is associated with chemoresistance in ovarian cancer.

Keywords: BRD4 amplification; BRD4 long; BRD4 short; chemoresistance; high-grade serous ovarian carcinoma.

Figure 1. Overexpression or knockdown of BRD4 isoforms induce their distinct transcription patterns.

(A) Wes (ProteinSimple) showing protein expression of BRD4 long (BRD4-L) and BRD4 short (BRD4-S) in ovarian cancer cell lines representing high-grade serous ovarian carcinoma. (B) Protein expression assessed by Wes of BRD4 isoforms in Ovcar4 cell lines following BRD4 isoforms overexpression or knockdown (KD). The expression of housekeeping proteins GAPDH and vinculin was used as loading control. Whole Wes images are included in Supplementary Figures 1, 2. (C–E) Relative expression of transcripts encoding BRD4 isoforms in Ovcar4 cell lines. (C) Total (endogenous and exogenous) levels of BRD4 mRNA. (D) Endogenous levels of BRD4-L mRNA. (E) Endogenous levels of BRD4-S mRNA. Fold change of BRD4 isoform expression in genetically engineered Ovcar4 cell lines is presented as relative to the expression of respective BRD4 isoform in control (parental Ovcar4 cells). Results are presented as average ± SE, ^*p < 0.05, NS indicate not statistically significant.

Figure 2. Overexpression of distinct BRD4 isoforms has opposing roles in promoting ovarian cancer cell proliferation in vitro.

(A) Cumulative cell population doubling quantified by 3T5 cell proliferation assay of Ovcar4 cells overexpressing BRD4-L or BRD4-S compared to control cells. (B) Cell proliferation rate of Tyk-nu cells following BRD4-L or BRD4-S knockdown compared to parental cells. Results are represented as average ± SD; ^*p < 0.05, and ^**p < 0.01 (one-way ANOVA followed by Tukey’s post-hoc). (C–E) Anchorage-independent growth of Ovcar4 cells overexpressing individual BRD4 isoforms was assessed by colony formation assay (soft agar colony formation assay). Number of colonies was quantified in (C) and their size in was measured in (D). (E) Representative images of cell colonies grown in soft agar captured at day 11. (F–H) Anchorage-independent growth of Tyk-nu cells following BRD4 isoform knockdown. Number of colonies was quantified in (F) and their size in was measured in (G). (H) Representative images of cell colonies grown in soft agar captured at day 11. Results are presented as average ± SE, ^*p < 0.05, NS not significant when compared to control (unpaired t-test). More detailed colony formation assay data are provided in Supplementary Figure 4.

Figure 3. Ovarian carcinoma cells overexpressing BRD4-S isoform become arrested in G2/M phase of the cell cycle.

(A) Mitotic Index (MI) defined as the percentage of cells undergoing mitosis in a given population of cells was quantified for Ovcar4 cell lines overexpressing BRD4 isoforms vs. Ovcar4 control cells. Results are presented as average ± SE, NS - not significant when compared to control (unpaired t-test). (B) Images of mitotic cells detected in Ovcar4 and Ovcar4-BRD4-S cell lines. Immunofluorescence staining for α-tubulin (green) shows mitotic spindles, and DAPI staining (blue) shows chromatin conformation in different phases of mitotic cell division. (C) Percentage of cells in mitosis among Tyk-nu cell lines depleted of respective BRD4 isoforms vs. parental control cells. Results are presented as average ± SE, ^*p < 0.05, NS - not significant when compared to control (unpaired t-test). (D) Images of mitotic cells detected in Tyk-nu and Tyk-nu-BRD4-S-KD cell lines. (E, F) Flow cytometry analysis of cell cycle distribution in Ovcar4 cells (E) and Ovcar4-BRD4-S cells (F). (G) Quantification of cells in different cell cycle phases. Results are presented as average ± SE, ^*p < 0.05. Statistical significance was evaluated by comparing the % of cells in a given cell cycle phase between BRD4 isoform overexpressing cell line vs. control (parental) cell line (one-way ANOVA followed by Tukey’s post-hoc). (H) Percentage of polyploid giant cells in Ovcar4 cell lines. Data are presented as average ± SE, ^*p < 0.05, ^**p < 0.01, NS - not significant when compared to control (unpaired t-test).

Figure 4. The BRD4-S overexpression or knockdown increases DNA damage in ovarian carcinoma cell lines.

(A) Representative images of Ovcar4 and Ovcar4-BRD4-S cell lines assessed for presence of DNA damage by quantification of a number of cells with pH2AX foci. Immunofluorescence staining identified phosphorylated histone H2AX (pH2AX) foci (red) indicating cells with DNA damage. The cells were counterstained with DAPI (blue) as a chromatin marker for nucleus visualization. (B) Representative images of Tyk-nu-BRD4-S-KD cell line showing DNA damage (pH2AX, red) in a cell undergoing mitosis (white star). Mitotic spindles were visualized by α-tubulin staining (green), and cell nuclei were visualized by DAPI staining (blue). (C) Quantification of the percentage of cells with more than 20 pH2AX foci in cell lines with overexpression or knockdown of BRD4 isoforms vs. parental cells. Data expressed as average ± SE; ^*p < 0.05, NS - not significant when compared to control (unpaired t-test). (D) Quantification of mitotic cells with pH2AX foci (as those shown in B) in Tyk-nu cell lines depleted of BRD4 isoforms vs. control cells. Data expressed as average ± SE; ^*p < 0.05, NS - not significant when compared to control (unpaired t-test). (E) DNA repair array showing the fold change in the expression of genes involved in the DNA-damage repair in Ovcar4-BRD4-S cell line relative to the control (Ovcar4 parental cells). Data expressed as average ± SE.

Figure 5. Overexpression of BRD4 isoforms in ovarian carcinoma promotes chemotherapy resistance in vitro and in vivo.

(A–I) Drug dose-response assay performed in cell lines with overexpression or knockdown of respective BRD4 isoforms. Data are presented as EC50 values of selected compounds (cisplatin (A–C), paclitaxel (D), dBET6 (E), JQ1 (F), cisplatin/paclitaxel (G), JQ1/paclitaxel (H), and adavosertib/paclitaxel in I) assessed for each cell line. The lower the EC50 value, the more potent drug in killing cancer cells. Data expressed as average ± SE; ^*p < 0.05, NS - not significant when compared to control (unpaired t-test). (J–M) Assessment of tumor growth rate in vivo of subcutaneously implanted Ovcar4 cell lines with and without overexpression of BRD4 isoforms in NOD/SCID mice. Animals received 3 weekly cycles of cisplatin/paclitaxel chemotherapy or vehicle control. Data are expressed as average ± SE; ^*p < 0.05 (one-way ANOVA followed by Tukey’s post-hoc).

Figure 6. Patient-derived xenografts with high expression of BRD4-L and BRD4-S are resistant to cisplatin/paclitaxel chemotherapy.

(A) RNA-Seq data were used to calculate the percentage of spliced-in (PSI) index of BRD4-L mRNA and BRD4-S mRNA. PSI represents the percentage of specific BRD4 isoform transcripts out of all BRD4 isoforms transcripts. (B) RPPA analysis (presented as a heatmap) illustrates quantification of BRD4-L protein expression in ovarian PDXs. PDXs with the highest BRD4-L expression are denoted in red, while those with the lowest BRD4-L expression are denoted in green. (C) The correlation between BRD4-L isoform transcript and protein expression was assessed by the Pearson’s correlation analysis, which revealed a negative correlation between mRNA and protein levels. (D) Wes image of selected PDX models showed a similar BRD4-L protein expression profile (quantified in E) as observed in RPPA results. Vinculin was used as a loading control. (F) The Pearson’s correlation analysis revealed a positive correlation between BRD4-S mRNA and protein levels. (G) Assessment of tumor growth rate in vivo and response to cisplatin/paclitaxel chemotherapy of subcutaneously implanted BRD4-high (PDX-0083) and BRD4-low (PDX-0003) tumor models. Animals received 4 weekly cycles of cisplatin/paclitaxel chemotherapy or vehicle control. (H) Waterfall plots show the response of individual PDXs to cisplatin/paclitaxel treatment during therapy (week 1–4) and during follow up period (week 6–18). Percentage change of tumor volume values below 0 indicate tumor regression, while the values above 0 indicate increased tumor volume when compared to initial tumor volume (prior treatment). Data expressed as average ± SE; ^**p < 0.01, NS - not significant when compared to control (unpaired t-test).