High BET gene expression correlates with poor survival in basal breast cancers
BET proteins are involved in the transcriptional regulation of oncogenes and CSC-associated genes. To determine the prognostic relevance of BET proteins in TNBC patients, we performed Kaplan–Meier survival (KM) analysis on TCGA data of basal breast cancer patients (n = 309). High levels of BET gene expression were correlated with worse overall survival compared to low levels of expression (Fig. 1a–d). This correlation occurred for individual BET proteins BRD2, BRD3, and BRD4, with the greatest significance for combined BRD2, BRD3, and BRD4 expression. Therefore, we hypothesized that BET proteins are an important clinical target, and that targeted inhibition or degradation of BET proteins might improve patient outcomes. We identified the pan-BET degrader ZBC260 as a promising therapeutic agent. Consistent with previous reports , ZBC260 reduced breast cancer cell levels of BRD2, BRD3, and BRD4 proteins in a concentration-dependent manner. ZBC260 treatment also decreased the protein level of Myc, a gene known to be transcriptionally regulated by BRD4 (Fig. 1e and Additional file 1: Fig. S1a-b).
ZBC260 decreases cellular viability and tumor growth
The effect of ZBC260 treatment on the cellular viability of four TNBC cell lines, SUM149, SUM159, MDA-MB-453 and MDA-MB-468, was tested in vitro. Our results demonstrated that ZBC260 treatment at nanomolar concentrations significantly decreased cellular viability in all tested cell lines (Fig. 2a), ranging from 1.56 nM (SUM149) to 12.5 nM (SUM159). To further evaluate the potency of ZBC260 compared to the well-known BET inhibitor JQ1, a comparative cell viability analysis was conducted. The results showed that ZBC260 was approximately tenfold more potent than JQ1 in inhibiting the viability of SUM149 and SUM159 cells (Fig. 2b). Additionally, the effect of ZBC260 on the ability of tumor cells to survive and form colonies, indicative of their replicative efficiency, was evaluated by colony formation assay. Treatment with ZBC260 reduced the survival fraction (SF) of cells. These findings suggest that ZBC260 not only inhibits cellular viability but also impairs the replicative efficiency of tumor cells to form colonies (Fig. 2c–d).
These in vitro results demonstrated that treatment with ZBC260 efficiently degraded BET proteins and reduced the viability of TNBC cells, thus the effect of ZBC260 on tumor growth and tumor initiation cell (TIC) frequency in vivo was determined. Mice (n = 10) bearing SUM149 tumor cells were treated with ZBC260 or vehicle. Compared to the control group, tumors treated with ZBC260 showed diminished tumor growth over 6 weeks of treatment (Fig. 2e, and Additional file 1: Fig. 2). Extreme limiting dilution analysis (ELDA) was performed to analyze the effect of ZBC260 treatment on TIC frequency, . Tumors treated with ZBC260 had a non-significantly lower frequency of TICs compared to vehicle treated cells (Fig. 2f); coupled with the significantly decreased overall tumor size, this finding is consistent with potent drug effect against both CSCs and bulk cells.
ZBC260 treatment decreases stemness markers, tumorsphere formation and enhances differentiation
Given the strong anti-tumor effect of ZBC260 on bulk cells and CSCs, the effect of ZBC260 on breast cancer stemness was further explored. Many CSC-targeted agents exert effects at concentrations near or even below the IC50 for the drug, and therefore, SUM149 and SUM159 cell lines were treated with ZBC260 across a range of low concentrations to assess the effect of ZBC260 on ALDH activity and CD44/CD24 expression. These two cell lines were chosen for their well-defined CSC populations: SUM149 CSCs exist as two populations defined by ALDH+ or CD44+/CD24−, while SUM159 CSCs are best defined by ALDH expression as most cells are CD44+/CD24− in this very mesenchymal cell line .
The percentage and absolute number of ALDH+ cells in SUM149 cells transiently increased at very low concentration of ZBC260, followed by a decrease at higher concentrations (Fig. 3a–b and Additional file 1: Fig. 3a–b), and the percentage and absolute number of CD44+/CD24− cells in this cell line decreased in a concentration-dependent manner (Fig. 3a and c and Additional file 1: Fig. 3c). Correspondingly, the percentage of CD44+/CD24+ cells increased, inversely mirroring the observed decrease in CD44+/CD24− cells (Fig. 3a and d).
In SUM159 cells the percentage and absolute cell number of ALDH+ and CD44+/CD24− cells decreased with ZBC260 treatment (Fig. 3e–g), albeit at slightly higher concentrations compared to SUM149 cells, consistent with the higher measured IC50 in this cell line (Additional file 1: Fig. 3a, d and e). Similarly, the percentage of CD44+/CD24+ cells increased at higher concentrations (Fig. 3e and h).
The effect of ZBC260 on the absolute numbers of ALDH+ and CD44+ /CD24− cells compared to the effect on relative percentage of these cell populations indicated that both bulk cells and CSCs are sensitive to ZBC260, with the effect being more pronounced in the CSC population (Additional file 1: 4a–d).
Tumorsphere formation assays were performed to assess the effect of ZBC260 on cellular self-renewal capacity . Consistent with the changes in stemness markers, we observed a concentration-dependent decrease in primary tumorsphere numbers in both cell lines. Secondary tumorsphere numbers were also decreased at higher concentrations (Fig. 3i and j). In addition to changes in tumorsphere number, the size of the tumorspheres also decreased for both primary and secondary spheres at all tested concentrations (Fig. 3k and Additional file 1: 4e–f). These results demonstrate an inhibitory effect of ZBC260 on cellular self-renewal, a key criterion for stemness.
ZBC260 Treatment decreases stemness index and impairs the expression of stemness-related genes
To further characterize the observed changes in stemness using a molecular approach, RNA-seq was performed on SUM159 ALDH+ and ALDH− cells treated with ZBC260 or vehicle. Principal component analysis (PCA) plot demonstrated that ALDH+ and ALDH− populations were minimally overlapping at baseline yet become significantly distinct after ZBC260 treatment, consistent with our hypothesis that ZBC260 exerts differential effects on the ALDH+ and ALDH− cell populations (Fig. 4a). To determine the global effect on stemness, a previously developed stemness index (SI)  was applied. To demonstrate the validity of applying this calculation to cultured cells, SI for treated samples was first compared to the TCGA samples originally used in the generation of the index, and the SI for our test samples fell within the range of SI values determined for TCGA samples (Additional file 1: Fig. 5a). Next, the relative SI index was determined for our 4 treatment conditions (Fig. 4b). As expected, ALDH+ cells had a higher SI score at baseline compared to ALDH− cells. After ZBC260 treatment the SI score for ALDH+ population was significantly reduced (−37.11% P = 0.05), while ALDH− cells exhibited a non-significant change in SI score (−17.26% P = 0.36). These findings further suggest that ZBC260 primarily affects stemness in CSCs.
Differential gene expression (DGE) analysis showed that ZBC260 treatment significantly (FDR < 0.05) modulated the expression of 4111 genes in ALDH+ cells and 2813 genes in ALDH− cells. Of these genes, 1831 genes were shared between both cell populations, 2280 genes were unique to ALDH+ cells, and 982 were unique to ALDH− cells (Fig. 4c). The top 20 significantly modulated genes in both cell populations were identified using unbiased gene analysis (Fig. 4d and Additional file 1: Fig. 5b). The most significantly downregulated and upregulated genes in ALDH+ cells were IL18R1, ARHGEF2, and CBS; and NQO1, LOXL2, and H2AFX, respectively. The corresponding genes in ALDH− cells were IL18R1, ARHGEF2, and CASC19; and H2AFX, NQ01, and TUBB4B.
The expression of genes involved in three processes relevant to our investigation was analyzed: known BET-regulated genes, stemness-associated genes, and EMT/MET plasticity genes. Regarding known BET-protein-regulated genes, the BRD4 target genes MYC and CCND2 were significantly downregulated, while P21, a gene known to negatively regulate MYC expression , was significantly upregulated. These changes occurred in both the ALDH+ and ALDH− populations to a similar degree (Fig. 4e). Changes in some stemness-associated genes were similar in ALDH+ and ALDH− cells, namely downregulation of KLF4 and upregulation of CD44; however, other genes were specifically affected in only one cell population, namely CD24 which was upregulated in ALDH+ cells and ALDH1A1 which was downregulated in ALDH− cells (Fig. 4f). We did not detect the expression of pluripotency genes SOX-2 and NANOG consistent with previous studies in this cell line . Finally, the expression of genes involved in EMT-MET states was determined. Some mesenchymal genes were significantly upregulated, namely VIM, CDH2 (N-cadherin), BMP1, and CTNNB1 (b-catenin), while another mesenchymal gene FOXC2 was downregulated. Similarly, the epithelial gene CLDN4 was significantly upregulated, while MUC1 was downregulated. The EMT transcription factor ZEB1 was significantly downregulated, but no significant changes in other transcription factors were found. We observed no expression of CDH1 (E-cadherin) gene and EPCAM gene, consistent with existing characterization of SUM159 cells . These results suggest minimal effects on EMT and that the effects of ZBC260 are not via modulation of the CSC EMT/MET phenotype (Fig. 4g).
ZBC260 Suppresses inflammatory and stemness pathways
These data demonstrated that ZBC260 causes significant gene expression changes in ALDH+ cells, with changes seen in stemness-related genes but not related to EMT/MET states. To further understand the specific effects of ZBC260 on CSCs, gene set enrichment analysis (GSEA) was performed on differentially expressed genes after ZBC260 treatment of both ALDH+ and ALDH− cells. GSEA of KEGG pathways identified significantly (FDR < 0.05) modulated pathways in both cell types. In total 16 significant pathways were identified in ALDH+ cells and 12 pathways in ALDH− cells; 7 pathways were shared between both cell populations, (Fig. 5a). Of the 16 significantly modulated KEGG pathways identified in ALDH+ cells, 3 pathways were positively enriched and 13 were negatively enriched (Fig. 5b). The cytokine–cytokine receptor interaction, NOD-like receptor signaling pathway, and JAK-STAT signaling pathway were the most significantly negatively enriched. In ALDH− cells, 12 significantly regulated pathways were identified: 6 positively enriched and 6 negatively enriched (Additional file 1: Fig. 6).
To understand the drivers of the ZBC260 inhibitory effect on CSCs, the top 20 differentially expressed genes in the significantly downregulated pathways in ALDH+ cells were determined (Fig. 5c). Most of the identified genes were involved in inflammation, with many encoding cytokines and cytokine receptors. Top genes included IL1A, IL1B, IL18R1, PDGFRA, CSF3, INHBE, MEFV, and chemokines CCL5, CSF3, and CXCL10, all of which were reduced more than twofold after treatment in ALDH+ cells. Additional important genes involved in JAK-STAT signaling, including STAT4, STAT5A, and LIF, were also found to be differentially downregulated (Additional file 1: Fig. 7a, top 20 genes in the 3 most downregulated pathways). The top 20 differentially expressed genes in significantly upregulated pathways were also identified. (Additional file 1: Fig. 7b).
ZBC260 decreased the expression of inflammatory genes and Stat proteins
RNA-seq data identified that the effect of ZBC260 in ALDH+ cells is mainly via downregulation of the expression of genes known to be involved in inflammation, potentially via JAK-STAT pathway signaling. The effect of ZBC260 on the expression of select inflammatory genes in ALDH+ and ALDH− cells was analyzed by qPCR. These genes included the most differentially downregulated ones (CSF3, CCL5, CXCL10, IL18R1, PDGFRA), a known target gene of BRD4 (MYC), and additional genes involved in inflammatory signaling (IL-6, and LIF). ZBC260 treatment led to a significant downregulation of all the tested genes compared to control. CCL5, CSF3 and CXCL10 uniquely were more highly expressed in ALDH+ cells at baseline and these genes also decreased to a greater extent after ZBC260 treatment as compared to ALDH− cells (Fig. 6a–i). These findings were substantiated by ELISA, showing a significant decrease in CCL5 and CSF3 secretion exclusively in ALDH+ cells following ZBC260 treatment, while IL6 levels decreased in both cell populations (Fig. 6j).
The expression of genes that encode STAT1, STAT3, and STAT5A, was significantly decreased in ALDH+ cells, but not ALDH− cells, after ZBC260 treatment (Fig. 7a–c). Western blot analysis showed that this decrease in gene expression led to a corresponding decrease in Stat protein levels after treatment (ZBC260) in both ALDH+ and ALDH− populations, with a greater decrease observed in the ALDH+ cells compared to the ALDH− cells (Fig. 7d–g). In addition to the observed decrease in total protein levels, Stat protein activation was also decreased after ZBC260 treatment as measured by phosphorylated protein levels. Levels of phosphorylated STAT3 and STAT5A were more significantly changed in the ALDH+ population compared to the ALDH− population. These findings support a dual effect on STAT signaling in ALDH+ CSCs, via inhibition of both gene expression and protein activation, consistent with our RNA-seq findings which identified changes in gene expression patterns related to STAT and its signaling activators.
Overall, our data suggest that ZBC260 treatment not only inhibited the growth of tumor cells but also reduced the expression of CSC markers and overall stemness. ZBC260 treatment resulted in a targeted downregulation of inflammatory signaling pathways in CSCs. This concept is visually depicted in Fig. 8.