Predicting Novel Therapies and Targets: Regulation of Notch3 by the Bromodomain Protein BRD4

Abstract

Systematic approaches for accurate repurposing of targeted therapies are needed. We developed and aimed to biologically validate our therapy predicting tool (TPT) for the repurposing of targeted therapies for specific tumor types by testing the role of Bromodomain and Extra-Terminal motif inhibitors (BETi) in inhibiting BRD4 function and downregulating Notch3 signaling in ovarian cancer.Utilizing established ovarian cancer preclinical models, we carried out in vitro and in vivo studies with clinically relevant BETis to determine their therapeutic effect and impact on Notch3 signaling.Treatment with BETis or siRNA-mediated BRD4 knockdown resulted in decreased cell viability, reduced cell proliferation, and increased cell apoptosis in vitro. In vivo studies with orthotopic mouse models demonstrated that treatment with BETi decreased tumor growth. In addition, knockdown of BRD4 with doxycycline-inducible shRNA increased survival up to 50% (P < 0.001). Treatment with either BETis or BRD4 siRNA decreased Notch3 expression both in vitro and in vivo BRD4 inhibition also decreased the expression of NOTCH3 targets, including HES1 Chromatin immunoprecipitation revealed that BRD4 was present at the NOTCH3 promoter.Our findings provide biological validation for the TPT by demonstrating that BETis can be an effective therapeutic agent for ovarian cancer by downregulating Notch3 expression.The TPT could rapidly identify candidate drugs for ovarian or other cancers along with novel companion biomarkers.

Figure 1.. Determining BRD4 as a therapeutic avenue for targeting Notch3 in ovarian cancer with the Therapy Predicting Tool.

(A) Screenshot of the TPT demonstrating where the user chooses tumor type, gene of interest (BRD4) and correlate expression to potential downstream target (Notch3) based on reverse phase protein array (RPPA) data available for ovarian cancer. TPT can also demonstrate the importance of the downstream target, Notch3, in ovarian cancer by comparing mRNA expression levels with protein levels provided by RPPA as well as correlate protein level (RPPA) with copy number alterations. Blue box pop up summarizes the query results. The scale bar shown in this figure panel only pertains to the highlighted row in the heat map and not to the entire heat map displayed in the screen shot. (B) GTEx analysis comparing expression of BRD4 in normal ovarian tissue (blue box) with that in ovarian cancer (red box). (C) Survival curves for patients with ovarian cancer from The Cancer Genome Atlas data sets comparing low BRD4 expression (green curve) with high BRD4 expression (red curve) in ovarian cancer tumors (p < 0.02). Statistical analysis was done by applying the log-rank test in the “survival” R package. (D) RPPA data heatmap expanded from Figure 1A correlating BRD4 mRNA levels with protein expression levels of indicated proteins.

Figure 2.. BRD4 inhibition using BETis has a beneficial therapeutic effect in ovarian cancer cells.

(A) Standard curve qRT-PCR analysis of relative BRD4 mRNA relative fold change levels in ovarian cancer cell lines compared with human isolated primary fallopian tube epithelial cells (FTE 4). Experiment done in 2 independent biological duplicates, **** p < 0.001. One-way ANOVA was used to test for significance. (B) BRD4 protein levels in ovarian cancer cells relative to those of FTE 4. Experiment done in 2 independent biological duplicates. (C and D) MTT viability assays of ovarian cancer cells treated for 72 hours with BETi CPI203 or CN210, respectively. Experiment done in 3 independent biological duplicates. (E and F) Colony formation assay of ovarian cancer cells treated for 10 days with BETi CPI203 or CN210, respectively. Experiment done in 3 independent biological duplicates. (G and H) EdU flow cytometry analysis of ovarian cancer cells treated with BETi CPI203 or CN210, respectively, for 72 hours. * p < 0.05, ** p < 0.01, *** p < 0.001, ns (non-significant). Experiment done in 3 independent biological duplicates. Statistical significance was determined by conducting unpaired Student t-test comparing the mean of EdU incorporation of vehicle control versus BETi treated cells. Experiment done in 3 independent biological duplicates. (I and J) Annexin V flow cytometry analysis of ovarian cancer cells treated with BETi CPI203 or CN210, respectively, for 72 hours. * p < 0.05, ** p < 0.01, *** p < 0.001, ns (non-significant). Experiment done in 3 independent biological duplicates. Statistical significance was determined by conducting unpaired Student t-test comparing the mean Annexin V staining of vehicle control versus BETi treatments.

Figure 3.. BRD4 inhibition reduces tumor growth and prolongs survival.

(A) Schematic for CPI203 treatment in OVCAR 5 tumor-bearing mice. (B) Tumor nodules (white arrows) in CPI203- or PBS-treated mice. (C) Mean tumor weights in PBS (n = 8) and CPI203-treated mice (n = 6), * p < 0.05. Statistical significance was determined using Student t-test for mean difference in tumor weights. (D) IHC staining of OVCAR 5 tumors for Ki67. Scale bar, 100 μm. N = 4 per group. (E) Quantification of Ki67-positive cells from PBS- or CPI203-treated mice, n = 4 per group. Quantification was done using pictures of five random fields from OVCAR 5 tumors. Statistical significance was determined using Student t-test for mean difference in Ki67 positive cells. (F) Timeline for doxycycline-inducible shBRD4 survival experiment. (G) IVIS imaging of OVCAR 5 tumor-bearing mice 21 days after shRNA induction. (H) IHC staining for BRD4 in OVCAR 5 tumor-bearing mice to confirm BRD4 knockdown in vivo. Scale bar, 50 μm. (I) Quantification of BRD4-positive cells from tumors in (H), ** p < 0.01, *** p < 0.001. Quantification was done using pictures of five random fields from OVCAR 5 tumors and statistical significance was determined using Student t-test for mean difference in BRD4 positive cells (n = 3 per indicated group). (J) Survival curve of each indicated group after shBRD4 induction, *** p < 0.001. Comparison was made to test for any significant difference between curves using Log-rank (Mantel-Cox) test.

Figure 4.. BRD4 inhibition decreases Notch3 expression in ovarian cancer.

(A, C, and E) Protein expression analysis at 72 hours of full-length (FL) and cleaved (C) Notch3 after BRD4 inhibition using a BETi (A and C) or siRNA-mediated knockdown (E). (B, D, and F) qRT-PCR analysis of Notch3 mRNA levels after BRD4 inhibition using a BETi (B and D) or siRNA-mediated knockdown (F). * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001, ns (non-significant). Statistical analysis was done by applying unpaired Student t-test of mean fold change for vehicle control cells compared to BETi treated cells. Experiments (A) through (F) have been repeated at least in 3 independent biological replicates. (G) Notch3 IHC staining of OVCAR 5 tumor-bearing mice treated with either vehicle control (n = 3) or CN210 (n = 3). (H) Notch3 IHC staining of OVCAR 5 tumors after Control (n = 5) or BRD4 (n = 5) siRNA delivery by DOPC nanoliposomes.

Figure 5.. BRD4 is present on the NOTCH3 promoter and impacts Notch3 downstream signaling.

(A) Schematic of the NOTCH3 promoter and transcription start site. (B) Chromatin immunoprecipitation (ChIP) PCR reactions after BRD4 pulldown at the indicated gene promoter regions. Experiments repeated in 3 independent biological duplicates. (C) ChIP qPCR of NOTCH3 transcription start site after BRD4 pulldown, * p < 0.05. Experiments repeated in 3 independent biological duplicates. Statistical analysis was done by applying unpaired Student t-test for mean fold enrichment of BRD4 over IgG control. (D and E) NOTCH3 gene signature generated from RPPA data using NetWalker software. Fold change calculated based on NormLog2 expression difference of DMSO treated cells versus CPI203 treated cells. Samples were submitted as 2 independent biological duplicates. (F and G) Western blot of cells treated with BETis. Experiments repeated in 2 independent biological duplicates (H) qRT-PCR analysis of HES1 after BETi treatment. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Experiments repeated in 3 independent biological duplicates. Statistical analysis was done by applying unpaired Student t-test of mean fold change for vehicle control cells compared to BETi treated cells.

Figure 6.. Summary model figure for use of BETis to for targeting Notch3 in ovarian cancer.

(A) BRD4 promoting NOTCH3 transcription and driving ovarian cancer tumor growth. (B) BETi inhibiting BRD4 mediated NOTCH3 transcription and inhibiting ovarian cancer tumor growth.