Targeting BARD1 suppresses a Myc-dependent transcriptional program and tumor growth in pancreatic ductal adenocarcinoma

Abstract

Pancreatic ductal adenocarcinoma (PDAC) remains one of the deadliest cancers demanding better and more effective therapies. BARD1 or BRCA1-Associated -Ring Domain-1 plays a pivotal role in homologous recombination repair (HRR). However, its function and the underlying molecular mechanisms in PDAC are still not fully elucidated. Here, we demonstrate that BARD1 is overexpressed in PDAC and its genetic inhibition suppresses c-Myc and disrupts c-Myc dependent transcriptional program. Mechanistically, BARD1 stabilizes c-Myc through ubiquitin-proteasome system by regulating FBXW7. Importantly, targeting BARD1 using either siRNAs or CRISPR/Cas9 deletion blocks PDAC growth in vitro and in vivo, without any signs of toxicity to mice. Using a focused drug library of 477 DNA damage response compounds, we also found that BARD1 inhibition enhances therapeutic efficacy of several clinically relevant agents (fold changes ≥4), including PARPi, in HRR proficient PDAC cells. These data uncover BARD1 as an attractive therapeutic target for HRR proficient PDAC.

Keywords: BARD1; DNA damage; PARP inhibitor; Pancreatic ductal adenocarcinoma; c-Myc.

Fig. 1

BARD1 is overexpressed in PDAC. A) Immunohistochemistry staining of PDAC Tissue Microarrays with BARD1 antibody (left panel). The inset indicates BARD1 expression in the ductal cells. A representative graph of BARD1 IHC score is shown (right panel). *p < 0.05. B) mRNA expression and C) protein expression of BARD1 in PDAC cell lines compared to normal epithelial cell line, HPNE. Protein expression was normalized to α-tubulin and mRNA expression was normalized to 18s. ****p < 0.0001. D) BARD1 gene expression levels in PDAC and normal tissue from GEPIA using TCGA/GTEx data (left panel). |Log2FC| Cutoff: 0.5, p-value Cutoff: 0.01; GEO dataset GSE16515 (middle panel); GEO dataset GSE28735 (right panel) ****p < 0.05.

Fig. 2

RNA-seq analysis of Panc-1 PDAC cells treated with either SCR or BARD1 siRNA. A) Principal component analysis of all samples. B) Volcano plot showing gene-expression changes between SCR control and BARD1 siRNA (siB#1) cells. Downregulated transcripts, 186; upregulated transcripts, 138 (padj< 0.05; log2FoldChange > 1 and <−1). The x axis represents the fold change in gene expression and the y axis, the -log10 (adjusted p value). C) Hierarchical clustering map (heatmap) for differential expression genes in all samples (n = 4) based on Z-score (FPKM). Upregulated and downregulated genes in siB condition are marked. D) List of top 10 Hallmark genesets affected by targeting BARD1 in Panc-1 cells. P < 0.05. E) Gene Set Enrichment Analysis of siBARD1 (siB#1) vs. siSCR (SCR) (n = 4) samples in Panc-1 cells, showing negative enrichment of the Hallmark_MYC_targets and Schuhmacher_MYC_targets genesets. P Value, negative enrichment score (NES) and FDR q-value are mentioned. E) RT-qPCR analyses of selected genes from the MYC targets in MiaPaCa-2 and Panc-1 cell lines transfected with siB#2 and SCR. 6 common genes out of the top significant genes in the Hallmark_myc_v1 dataset (n = 35) were chosen ****p < 0.001.

Fig. 3

BARD1 enhances c-Myc stability in PDAC. A) Western blot analysis showing endogenous c-Myc protein levels in BARD1 CRISPRKO PDAC cells and cells transfected with either BARD1 siRNA (siB#1) or SCR. B) Correlation plots of BARD1 and c-Myc mRNA (left) and protein (right) levels from TCGA and CCLE datasets. Spearman r and p values are mentioned. C) Effect of cycloheximide (CHX, 25 μg/ml) on c-Myc stability in BARD1 depleted Panc-1 cells. The protein expression of BARD1 and c-Myc was analyzed by western blotting. Graph represents % c-Myc remaining at each time point. D) Effect of MG-132 (15 μg/ml) on c-Myc expression in BARD1 depleted (siB) and SCR control Panc-1 and MiaPaCa-2 cells. The protein expression of BARD1 and c-Myc was analyzed by western blotting and normalized to α-Tubulin. Numbers represent relative expression to SCR. E) Immunoprecipitation (IP) was used for the detection of c-Myc degradation in MiaPaCa-2 cells transfected with BARD1 siRNA and SCR siRNA. After treating indicated cells with MG-132 (15 μM) for 6 h, extracts were subjected to IP with c-Myc antibody and the polyubiquitination of c-Myc was assessed by western blot using ubiquitin antibody. Corresponding input controls are also shown. F) RT-qPCR analyses of FBXW7 in MiaPaCa-2 and Panc-1 cell lines transfected with siB and SCR. ****p < 0.001, n = 3.

Fig. 4

BARD1 is required for PDAC proliferation and invasion. Colony formation assay of A) PDAC cell lines (MiaPaCa-2, Bxpc-3, Panc-1) transfected with either BARD1 siRNA (siB#1) or Scramble siRNA, and B) BARD1 overexpressed (BARD1OE) and empty vector (EV) MiaPaCa-2 and Panc-1 cells. Colonies were stained with crystal violet solution after 14 days. Representative pictures and graph of relative colony area from n = 3 are shown. Growth of PDAC cell lines transfected with C) either BARD1 siRNA (siB#1) or Scramble siRNA (SCR), and D) BARD1 overexpression plasmid or EV. Graphs represent growth relative to day 1 as analyzed by Pico Green assay. E) MiaPaCa-2 and Panc-1 cells transfected with BARD1 siRNA (siB) or Scramble siRNA (SCR) for 48 h, were utilized in Matrigel Boyden chamber invasion assays. Representative images are shown. Graphs are Mean ± SEM for three independent experiments; each experiment was normalized to Scrambled (SCR) control. n.s=not significant, *p < 0.05, **p < 0.01, ***p < 0.001.

Fig. 5

CRISPR-Cas9 deletion of BARD1 slows PDAC growth in vitro and in vivo. A) Chromatograms from Sanger sequencing of wild type BARD1 WT (+/+) and homozygous mutant BARD1 (-/-) clones, generated by CRISPR-Cas9 transfection in MiaPaCa-2 cells. Highlighted in red is the specific mutations in the BARD1 (-/-) clone. B) Relative protein expression of BARD1 in CRISPR clones by western blot analysis. α-Tubulin was used as a loading control. C) Colony formation (top) and Matrigel Boyden invasion (bottom) assay with WT and BARD1 (-/-) MiaPaCa-2 cells. Representative images and graphs of relative colony area and relative invasion from n = 3 are shown.****p < 0.0001. D) Growth curves of WT and BARD1 (-/-) MiaPaCa-2 cells relative to day 1. E) Graph showing percentage of mice that developed palpable tumors of 100 mm3. F) Table showing percentage of mice with 100 mm3 tumors post injection. G) Kaplan Meier survival curve in a xenograft tumor model after injection of WT and BARD1 (-/-) CRISPR mutant cells into flanks of nude mice (n= 8 for WT, n = 15 for BARD1 (-/-)). H) Spaghetti plot of individual tumor volumes. I) Mouse weights plotted over time. Note: Mouse weights averages until day 45 are shown because of staggering survival events that decrease the number of mice in each experimental group.

Fig. 6

Targeting BARD1 enhances efficacy of various DNA damage response agents. A) Flow chart of drug screening experiment. B) Graphs showing percentage inhibition in survival of BARD1 siRNA (siB#1 and siB#2) and SCR control MiaPaCa-2 PDAC cells after exposure to different drugs from the drug library. Red dots depict compounds that show >50 % inhibition with siBARD1 and <50 % inhibition with Scramble (SCR) treatments. C) Venn diagram showing common agents in the two siRNA groups (siB#1 and siB#2). D) Table of common agents with % inhibition. E) Graphs showing % relative survival in WT, BARD1 (-/-) and BARD1 (±) CRISPR mutant cells after indicated drug treatments for 5 days, as analyzed by Pico Green assay. Table on the right shows IC50 (μM) and fold change values for each condition. All experiments were performed atleast n = 3.