Acute myeloid leukemia (AML) is a heterogeneous and aggressive malignant disease characterized by poor patient outcome and suboptimal front-line chemotherapy. To identify novel anti-AML compounds and potential molecular markers that confer drug-specificity, we performed a high-throughput screen of an in-house library consisting only of food-derived bioactive molecules (n=300). This screen was performed against the AML cell line (TEX), which has several properties of leukemia stem cells, the cells responsible for disease pathophysiology and patient relapse. This screen identified diosmetin as a novel anti-leukemia agent (IC50: 6.8 ± 1.7µm). Diosmetin (10µM) reduced clonogenic growth of primary AML patient cells with no effect on normal hematopoetic stem cells. In leukemia mouse xenografts, administration of diosmetin (50mg/kg) reduced tumor weight (>40%) without evidence of toxicity. Collectively, this shows that diosmetin is a novel anti-AML agent.
Diosmetin-induced apoptosis (e.g., increased Annexin V staining and DNA fragmentation) was abolished in the presence of caspase-inhibitors. Importantly, diosmetin, which is a flavonoid and a known ROS-inducer, induced death in the presence of anti-oxidants demonstrating that death was not related to ROS activity. Instead, we demonstrate that diosmetin-induced caspase 8-, not caspase 9, mediated apoptosis that was facilitated by increased TNF-α.
To identify diosmetin's molecular target we employed a novel method utilizing several free-online bioinformatics tools. Here, the protein data base (PBD), Pocket Similarity Search using Multi-Sketches (PoSSuM) and Database for Annotation, Visualization and Integrated Discovery (DAVID) tool, were used sequentially to identify potential diosmetin binding partners. This method identified the estrogen receptor (ER) as a potential molecular target for diosmetin's anti-leukemia activity.
There are two types of ERs. When activated, ERα increases cell proliferation whereas ERβ inhibits proliferation. These opposite and opposing roles have led to the hypothesis that ER expression patterns dictate the cell's response to estrogen, as hormones like estradiol can bind to both ERα and ERβ with different affinities. To first determine to which ER diosmetin bound, we mapped the binding sites of both ERα and ERβ using the Discovery Studio (DS), Structure-Based-Design program and determined the relative affinity of diosmetin to these ERs. Diosmetin bound with far greater affinity to ERβ than ERα. Next, we interrogated publically available AML patient datasets to determine the expression pattern of ERα versus ERβ. Interestingly, we noted a subset of AML patients that upregulate ERβ but not ERα expression (log4 fold increase; p<0.001) suggesting the clinical relevance of targeting ERβ in a subset of AML patients.
To assess the cell and molecular role of ERs, we measured ERα and ERβ levels in diosmetin sensitive and insensitive cell lines. Interestingly, diosmetin sensitive cell lines (TEX, LP1, AML2) display significantly elevated ER-β protein and mRNA levels (4 fold, as determined by Western blotting and qtPCR, respectively) compared to diosmetin insensitive cell lines (K562, DU145). This pattern was not observed for ER-α. Furthermore, this ER expression pattern was also observed in primary AML cells; as AML cells sensitive to diosmetin displayed an increase in ERβ, but not ERα, mRNA. Finally, ER reporter assays demonstrated that diosmetin binds and acts as a partial agonist in ERβ but not ERα reporter cells. Together, these results show that diosmetin binds to ERβ and that ERβ is functionally important to diosmetin's activity.
In summary, these studies highlight ERβ as a potential novel therapeutic target for the treatment of AML.
No relevant conflicts of interest to declare.