Acute myeloid leukemia (AML) is a devastating disease characterized by poor patient outcome and suboptimal chemotherapeutics. Here, a high-throughput screen identified diosmetin, a citrus flavonoid, with anti-AML activity. Diosmetin imparted selective toxicity against leukemia and leukemia stem cells in vitro and in vivo with no effect on normal hematopoietic stem cells. Mechanistically, we demonstrated that diosmetin targets estrogen receptor (ER) β. ERβ expression conferred cell sensitivity, as patient-derived AML cells with high levels of ERβ were sensitive, whereas cells with low ERβ were insensitive to diosmetin. Knockdown of ERβ confirmed resistance, whereas overexpression enhanced sensitivity to diosmetin, which was demonstrated to be mediated by reactive oxygen species signaling. In summary, these studies highlight targeting of ERβ with diosmetin as a potential novel therapeutic strategy for the treatment of AML. Mol Cancer Ther; 16(11); 2618–26. ©2017 AACR.

Acute myeloid leukemia (AML) is an aggressive hematologic malignancy with poor patient outcome characterized clinically by an accumulation of immature myeloid cells in the bone marrow (1). Current induction therapy fails to achieve long-term remission, and the 5-year-survival rates for adult patients (i.e., >60 years old) is less than 10% (2). New therapeutics to improve AML patient outcomes are clearly needed.

To identify novel anti-AML drugs, we created a unique library of food-derived bioactive compounds and screened this library against TEX leukemia cells, an AML and surrogate leukemia stem cell (LSC) line (3) used successfully in high-throughput screens (4–6). Through this approach, we identified diosmetin as a novel drug with anti-AML activity. Mechanistically, we show that diosmetin binds to estrogen receptor (ER) β and demonstrate that ERβ is functionally important to diosmetin's activity.

Cell culture

Cell lines from the ATCC were cultured in 5% CO2 at 37°C. They were obtained in 2012, and no authentication was performed thereafter. Additional details can be found in the Supplementary Materials. U2OS human osteosarcoma cells with epitope (FLAG)-tagged ERβ under the doxycycline promoter (hereafter denoted U2OS-ERβ) were grown in RPMI and were generated by Monroe and colleagues (7). Primary human AML samples were provided by Drs. Mark Minden (Princess Margaret Cancer Center), Joanna Graczyk, Janet MacEachern, and Robert Stevens (Grand River Cancer Centre) and were cultured in IMDM, and supplemented with 20% FCS and antibiotics. These cells were isolated from the peripheral blood of consenting AML patients who had at least 80% malignant cells among the mononuclear cells. Normal hematopoietic stem cells (HSC) were purchased from Stem Cell Technologies or provided by Dr. David Hess (Western University). The collection and use of human tissue for this study was approved by Institutional Ethics Review boards (University Health Network, Toronto, ON, Canada, Western University, and University of Waterloo, Waterloo, Ontario).

Cell growth and viability

Cell growth and viability was measured using the 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium inner salt (MTS) reduction assay (Promega) or Annexin V and propidium iodide (ANN/PI) staining (Biovision), according to the manufacturer's protocol and as previously described (6, 8). Colony formation assays were performed, as previously described (9) and as outlined in the Supplementary Methods.

Chemical screen and reactive oxygen species measurements

Food-derived bioactive molecules were obtained from Chengdu Biopurify Phytochemicals Ltd. (n = 288) and screened as outlined in the Supplementary Methods. Reactive oxygen species (ROS) measurements were conducted as detailed in the Supplementary Methods (10).

Target prediction using structural bioinformatics and molecular docking

Starting with diosmetin, we searched the Protein Data Bank (PDB) for protein structures capable of binding diosmetin-like ligands. We identified an initial set of 9 structures cobound with four diosmetin-like ligands: biochanin A (QSO), genistein (GEN), kaempherol (KMP), and luteolin (LU2). This set was expanded to include proteins with similar binding site compositions using the PoSSuM (Pocket Similarity Search using Multi-Sketches) tool with default parameters (11, 12). This search yielded a final set of 92 proteins with predicted diosmetin-binding potential. The Database for Annotation, Visualization and Integrated Discovery (DAVID) tool (13) was then used to summarize and identify overrepresented functions common to this group of potential target proteins.

The molecular docking experiments were conducted using the computational software Discovery Studio (DS), Structure-Based-Design program from BIOVIA/Accelrys Inc., as previously described (14, 15). Briefly, diosmetin was built using the small molecules module in DS. The crystal structures of ERα (pdb id: 1 × 7R) and ERβ (pdb id: 1 × 7J) with genistein bound was obtained from PDB. The protein was prepared using the macromolecules module in DS. Ligand-binding site was defined by selecting a 12 Å radius sphere using genistein crystal structure after which it was deleted. Then molecular docking was performed using the receptor–ligand interactions module in DS. The CDOCKER algorithm was used to find the most appropriate binding modes of diosmetin using CHARMm force field. The docked poses obtained were ranked based on the CDOCKER energies and CDOCKER interactions energies in kcal/mol.

mRNA and protein detection

qPCR was performed as previously described (9) in triplicate using an ABI 7900 Sequence Detection System (Applied Biosystems) with 5 ng of RNA equivalent cDNA, SYBR Green PCR Master mix (Applied Biosystems), and 300 nmol/L of ER-specific primers (forward ERβ: 5′-TGCTCAATTCCAGTATGTACC-3′, reverse ERβ: 5′-ATGAGGTGAGTGTTTGAGAG-3′, forward ERα: 5′-CATTATGGAGTCTGGTCCTG-3′, reverse ERα: 5′-TTCGTATCCCACCTTTCATC-3′). Relative mRNA expression was determined using the ΔΔCT method as previously described (18), and efficiency was calculated using LinRegPCR analysis software (20, 21). Western blotting was performed as previously described (6) and as outlined in the Supplementary Methods.

ER reporter assay

A Human Estrogen Receptors Reporter Assay Panel (Indigo Biosciences) was performed according to the manufacturer's protocol. Briefly, to investigate agonist activity, reporter HeLa cells were seeded in white 96-well plates and immediately dosed with increasing concentrations of diosmetin. To investigate antagonist activity, reporter cells were plated as above, but17-β-estradiol (E2), a known ERα and ERβ agonist, was added at the constant submaximal (EC75) concentrations. Following a 24-hour incubation period, a luciferase detection reagent was added to the wells and light emission was quantified using a luminometer.

RNAi knockdown of ERβ

Lentiviral infections were performed as described (9). Briefly, OCI-AML2 cells (5 × 106) were resuspended and incubated overnight at 37°C in media (6.5 mL) containing protamine sulfate (5 μg/mL) and 1 mL of virus cocktail (which contains a puromycin antibiotic resistance gene and the shRNA sequence). Next, the virus was removed by centrifugation, and the cells were washed and resuspended in fresh media containing puromycin (1 μg/mL). Following puromycin selection, live cells were plated for viability assays. The shRNA coding sequences (Sigma Chemical) were: ERβ (TRCN0000003328) shRNA clone 16 5′-CCGGGATGCTTTGGTTTGGGTGATTCTCGAGAATCACCCAAACCAAAGCATCTTTTT-3′, (TRCN0000003327) ERβ shRNA clone 17 5′-CCGGCCTTAATTCTCCTTCCTCCTACTCGAGTAGGAGGAAGGAGAATTAAGGTTTTT-3′; and a TRC scramble control.

In vivo models

NOD/SCID gamma mice (NSG; Jackson Laboratory or University of Western Ontario) were used for xenograft and engraftment assays as previously described (5, 6, 16) and detailed in the Supplementary Methods. For engraftment experiments, human myeloid cells (hCD45+/CD33+) were detected in femoral bone marrow by flow cytometry following a 6-week engraftment period. All animal studies were carried out according to the regulations of the Canadian Council on Animal Care and with the approval of the University of Waterloo, Animal Care Committee.

Statistical analysis

Unless otherwise stated, in vitro results are presented as mean ± SD, whereas in vivo results are presented as mean ± SEM. Data were analyzed with GraphPad Prism 4.0 (GraphPad Software) using one-way ANOVA with Tukey post hoc analysis for between-group comparisons or standard Student t tests where appropriate and Mann–Whitney t tests were used for animal experiments. P ≤ 0.05 was accepted as being statistically significant.

A screen for novel anti-AML compounds identifies diosmetin

To identify novel anti-AML compounds, we screened our unique in-house library against TEX cells using the MTS assay. Compound screens against TEX cells have identified novel anti-AML targets (4–6) and drugs that have been evaluated in human clinical trials (17). In our screen, the flavonoid diosmetin was one of two compounds that imparted the greatest reduction in TEX cell viability (Fig. 1A; Supplementary Fig. S1A; arrow indicates diosmetin; inset: chemical structure). Diosmetin's ability to reduce growth and proliferation was also shown in a panel of AML cell lines using the MTS assay following a 96-hour incubation period (Fig. 1B; EC50 values: 3–15 μmol/L). We next assessed diosmetin's effects on functionally defined subsets of primitive human AML and normal hematopoietic cell populations. Treatment of TEX cells with diosmetin at sublethal doses reduced their ability to engraft in the marrow of immune-deficient mice (Fig. 1C; U11 = 1; P < 0.01). Adding diosmetin into the culture medium had no effect on the clonogenic growth of normal hematopoietic cells (Fig. 1D; n = 3) but reduced clonogenic growth in a subset of AML patient cells (Fig. 1E, n = 8, t11 = 6.5; P < 0.001; Fig. 1F, n = 4; Supplementary Tables S1–S2 for patient characteristics). Taken together, diosmetin selectively targets primitive leukemia cells.

Figure 1.

Diosmetin is identified as a novel anti-AML agent. A, A screen of an in-house library identified diosmetin (arrow) as a potent compound capable of reducing TEX cell viability, an AML and surrogate LSC cell line (inset: diosmetin structure). Viability was assessed using the MTS assay. B, Diosmetin's ability to reduce growth and proliferation was tested in a panel of AML cell lines using the MTS assay. Data presented as an average of three independent experiments. C, TEX cells were treated with suboptimal doses of diosmetin (i.e., 10 μmol/L for 24 hours) or a vehicle control, and then live cells were injected via tail vein into NSG mice (n = 6/7 group). After 6 weeks, human AML cells (hCD45+/CD33+) in mouse bone marrow were detected by flow cytometry. Data presented as % human cells in individual mouse bone marrow relative to control. D–E, Normal hematopoietic cells (n = 3) or patient-derived AML cells (D: n = 8; E: n = 4) were cultured with diosmetin for 7 to 14 days, and clonogenic growth was assessed by enumerating colonies as described in the Materials and Methods section. Data are presented as % clonogenic growth compared with control ± SEM, similar to previously described (5). **, P < 0.01 and ***, P < 0.001.

Figure 1.

Diosmetin is identified as a novel anti-AML agent. A, A screen of an in-house library identified diosmetin (arrow) as a potent compound capable of reducing TEX cell viability, an AML and surrogate LSC cell line (inset: diosmetin structure). Viability was assessed using the MTS assay. B, Diosmetin's ability to reduce growth and proliferation was tested in a panel of AML cell lines using the MTS assay. Data presented as an average of three independent experiments. C, TEX cells were treated with suboptimal doses of diosmetin (i.e., 10 μmol/L for 24 hours) or a vehicle control, and then live cells were injected via tail vein into NSG mice (n = 6/7 group). After 6 weeks, human AML cells (hCD45+/CD33+) in mouse bone marrow were detected by flow cytometry. Data presented as % human cells in individual mouse bone marrow relative to control. D–E, Normal hematopoietic cells (n = 3) or patient-derived AML cells (D: n = 8; E: n = 4) were cultured with diosmetin for 7 to 14 days, and clonogenic growth was assessed by enumerating colonies as described in the Materials and Methods section. Data are presented as % clonogenic growth compared with control ± SEM, similar to previously described (5). **, P < 0.01 and ***, P < 0.001.

Close modal

ERs are the predicted molecular target

In primary and validation screens, 5 of the 12 most potent compounds were structurally similar (Fig. 2A). So to investigate the cellular target, a unique structural bioinformatics approach was used to mine existing protein structural data for known interactions involving these compounds (Fig. 2A). We first screened the PDB for crystallographic or NMR-derived protein structures cobound with diosmetin and these structurally similar compounds. This resulted in an initial set of nine structures, which were subsequently used as templates to find additional proteins of similar ligand-binding site composition using the Pocket Similarity Search using Multi-Sketches (PoSSuM) tool. This analysis yielded a final list of 92 predicted diosmetin-binding proteins, 52 of which were human. To summarize this list of potential targets, the DAVID tool was used, which reports statistically overrepresented functional or protein family categories in gene lists. The analysis predicted nuclear hormone receptors, specifically ERs, as a dominant class of diosmetin-binding proteins (Fig. 2B).

Figure 2.

ERβ is diosmetin's predicted molecular target. A, A flow chart outlining the structural bioinformatics approach for prediction and analysis of diosmetin-binding proteins from the PDB. B, Initial list of human target proteins identified in the PDB as cobound with diosmetin-like compounds. C, Molecular docking experiments using the receptor-ligand interactions module (Discovery Studio software) were performed to explore the binding interactions between diosmetin and human ERα and ERβ receptors.

Figure 2.

ERβ is diosmetin's predicted molecular target. A, A flow chart outlining the structural bioinformatics approach for prediction and analysis of diosmetin-binding proteins from the PDB. B, Initial list of human target proteins identified in the PDB as cobound with diosmetin-like compounds. C, Molecular docking experiments using the receptor-ligand interactions module (Discovery Studio software) were performed to explore the binding interactions between diosmetin and human ERα and ERβ receptors.

Close modal

Molecular docking was next utilized to explore binding interactions of diosmetin in the ligand-binding domain of human ERα and ERβ receptors (14). The Structure-Based-Design program predicted diosmetin to bind with greater affinity and stability to ERβ than ERα (ERα and ERβ stability: –22.73 kcal/mol vs. –24.03 kcal/mol, respectively; affinity: –37.94 kcal/mol vs. –39.05 kcal/mol, respectively; Fig. 2C). As a final bioinformatics approach, we interrogated publically available AML databases (18, 19) to determine whether there is a population of AML patients who may respond to an ERβ-specific small molecule. Interestingly, within these databases we identified a subpopulation of AML patient cells that overexpress ERβ while underexpressing ERα, further suggesting a potential clinical utility for targeting ERβ (Supplementary Figs. S2 and S3).

Diosmetin is an ERβ agonist

To determine if ERβ is diosmetin's molecular target, we first performed ER agonist and antagonist luciferase reporter assays. In the agonist assay, ER reporter cells were treated with diosmetin, and luminescence increased in a dose-dependent manner for ERβ (Fig. 3A; F5,12 = 162; P < 0.0001) and ERα (Fig. 3B; F5,12 = 245.9; P < 0.0001). Given the similarity of the ERα and ERβ ligand-binding domain, dual receptor binding was not surprising. Nonetheless, diosmetin clearly demonstrated preferential binding toward ERβ (Fig. 3C; F5,30 = 53.75; P < 0.0001). In the antagonist assay, reporter cells were coincubated with the ERα and ERβ ligand, 17β-estradiol (E2), and diosmetin for 24 hours. E2-induced luminescence was not affected demonstrating that diosmetin does not act as an ERα or β antagonist (Fig. 3D and E). ER agonist activity at 24 hours had no effect on the viability of these luciferase promoter cells (Supplementary Fig. S4).

Figure 3.

Diosmetin is an ER agonist. ERβ agonist activity was tested by adding increasing concentrations of diosmetin to HeLa cells containing (A) ERβ or (B) ERα under the control of a luciferase reporter. C, Comparison of luminescence at individual diosmetin concentrations highlights diosmetin's preferentially ERβ binding. D and E, Diosmetin was tested as an ERβ antagonist by adding increasing concentrations of diosmetin in the presence of estradiol (E2) to HeLa cells containing ERα or ERβ under the control of a luciferase reporter. Data are presented as relative luciferase intensity (RLU) compared with vehicle control–treated cells. Data presented as an average of three independent experiments. *, P < 0.05; **, P < 0.01; and ***, P < 0.001.

Figure 3.

Diosmetin is an ER agonist. ERβ agonist activity was tested by adding increasing concentrations of diosmetin to HeLa cells containing (A) ERβ or (B) ERα under the control of a luciferase reporter. C, Comparison of luminescence at individual diosmetin concentrations highlights diosmetin's preferentially ERβ binding. D and E, Diosmetin was tested as an ERβ antagonist by adding increasing concentrations of diosmetin in the presence of estradiol (E2) to HeLa cells containing ERα or ERβ under the control of a luciferase reporter. Data are presented as relative luciferase intensity (RLU) compared with vehicle control–treated cells. Data presented as an average of three independent experiments. *, P < 0.05; **, P < 0.01; and ***, P < 0.001.

Close modal

Patient-derived AML cell sensitivity is linked to ERβ expression

Because diosmetin preferentially bound to ERβ, we next determined whether ERβ expression predicted the differential response (i.e., diosmetin sensitive or insensitive) observed in the clonogenic growth assays (Fig. 1D and E). Interestingly, ERβ levels were elevated in diosmetin-sensitive primary cells (Fig. 4A). In contrast, primary cells (normal and patient-derived) that were insensitive to diosmetin had no-to-low ERβ but elevated ERα mRNA levels (Fig. 4B), suggesting that ERβ/ERα ratios confer patient-derived AML cell sensitivity to diosmetin. In addition, AML cell lines sensitive to diosmetin (Fig. 1B) express ERβ protein, as measured by immunoblotting (Supplementary Fig. S5).

Figure 4.

ERβ expression confers patient-derived AML cell sensitivity to diosmetin. ERα and ERβ gene expressions were measured by qPCR data in primary AML samples sensitive (A; n = 5) and insensitive (B; n = 6) to diosmetin. Data presented as relative mRNA expression normalized to GAPDH. (Normal hematopoietic cells denoted as N; primary insensitive AML samples denoted as IN.)

Figure 4.

ERβ expression confers patient-derived AML cell sensitivity to diosmetin. ERα and ERβ gene expressions were measured by qPCR data in primary AML samples sensitive (A; n = 5) and insensitive (B; n = 6) to diosmetin. Data presented as relative mRNA expression normalized to GAPDH. (Normal hematopoietic cells denoted as N; primary insensitive AML samples denoted as IN.)

Close modal

ERβ is functionally important to diosmetin's activity

To further define the role of ERβ in diosmetin-induced cell death, we utilized doxycycline-inducible flag-tagged ERβ U2OS cells (U2OS-ERβ cells; ref. 7). Treatment with 100 ng/mL doxycycline for 24 hours increased ERβ 4-fold compared with vehicle control–treated cells (Fig. 5A). In the absence of doxycycline (i.e., no ERβ), these cells were insensitive to diosmetin (Fig. 5B); however, when treated with 100 ng/mL doxycycline (i.e., increased ERβ expression), diosmetin imparted toxicity (Fig. 5B; F1,8 = 24.44; P < 0.01). Because diosmetin is a flavonoid and flavonoids generate ROS (20–22), we confirmed that diosmetin's activity is characterized by increases in intracellular ROS, as measured by DCF-DA and flow cytometry (Supplementary Fig. S6). Consistent with this observation, ROS increased in cells treated with doxycycline and diosmetin but not with diosmetin alone (Fig. 5C; t4 = 5.254; P < 0.01).

Figure 5.

ERβ is diosmetin's molecular target. A, U2OS-ERβ human osteosarcoma cells, with ERβ under a doxycycline-inducible promoter, were incubated in the presence or absence of doxycycline (100 ng/mL) for 72 hours. Expression of ERβ was measured in the presence of increasing doxycycline concentrations (100–500 ng/mL) using Western blotting. Experiments were performed three times, and representative blots are shown. B, Viability of U2OS-ERβ cells was measured in the presence or absence of diosmetin with or without doxycycline (100 ng/mL) for 72 hours using the ANN/PI assay. Data presented as an average of three independent experiments and as percent viable (ANN/PI) relative to zero controls (C) ROS was measured in U2OS-ERβ cells treated with diosmetin (24 hours at 10 μmol/L) or a vehicle control in the presence of doxycycline (100 ng/mL), using DCF-DA staining as outlined in the Materials and Methods section. D, ERβ (gene: ESR2) gene silencing was achieved by lentiviral-mediated transduction and confirmed by immunoblotting. Densitometry calculated as outlined in the Materials and Methods section. E, Viability of ERβ knockdown cells was tested with increasing concentrations of diosmetin by the ANN/PI assay. Data presented as an average of three independent experiments. F, ROS was measured in knockdown cells using DCF-DA staining, as outlined in the Materials and Methods section. Data presented as an average of three independent experiments. *, P < 0.05; **, P < 0.01; and ***, P < 0.001.

Figure 5.

ERβ is diosmetin's molecular target. A, U2OS-ERβ human osteosarcoma cells, with ERβ under a doxycycline-inducible promoter, were incubated in the presence or absence of doxycycline (100 ng/mL) for 72 hours. Expression of ERβ was measured in the presence of increasing doxycycline concentrations (100–500 ng/mL) using Western blotting. Experiments were performed three times, and representative blots are shown. B, Viability of U2OS-ERβ cells was measured in the presence or absence of diosmetin with or without doxycycline (100 ng/mL) for 72 hours using the ANN/PI assay. Data presented as an average of three independent experiments and as percent viable (ANN/PI) relative to zero controls (C) ROS was measured in U2OS-ERβ cells treated with diosmetin (24 hours at 10 μmol/L) or a vehicle control in the presence of doxycycline (100 ng/mL), using DCF-DA staining as outlined in the Materials and Methods section. D, ERβ (gene: ESR2) gene silencing was achieved by lentiviral-mediated transduction and confirmed by immunoblotting. Densitometry calculated as outlined in the Materials and Methods section. E, Viability of ERβ knockdown cells was tested with increasing concentrations of diosmetin by the ANN/PI assay. Data presented as an average of three independent experiments. F, ROS was measured in knockdown cells using DCF-DA staining, as outlined in the Materials and Methods section. Data presented as an average of three independent experiments. *, P < 0.05; **, P < 0.01; and ***, P < 0.001.

Close modal

To further confirm its functional importance, we generated ERβ knockdown OCI-AML2 cells using two independent shRNA vectors (denoted 16 and 17); knockdown was confirmed by immunoblotting (Fig. 5D). ERβ knockdown cells are resistant to diosmetin-induced death compared with scramble controls (Fig. 5E; F4,18 = 8.604, P < 0.001). Moreover, consistent with our observation that ERβ activation is characterized by increased ROS, diosmetin increased ROS in a dose-dependent manner in transduced controls but had little effect in the resistant knockdown cells (Fig. 5F; F4,18 = 5.152; P < 0.01). Together, these data demonstrate that ERβ is functionally important to diosmetin's activity.

ERβ targeting imparts anti-AML activity in vivo

Given the cytotoxicity of diosmetin in AML cells, we performed additional experiments in vivo in a functionally defined subset of primitive human AML and normal hematopoietic cell populations. First, normal HSCs were injected into the tail vein of NSG mice and after 1 week, intraperitoneal injections of diosmetin (50 mg/kg/every other day) were administered for 6 weeks. Treatment with the ERβ agonist had no effect on the presence of normal human myeloid cells (hCD45+/CD33+) in mouse bone marrow compared with vehicle control, as measured by flow cytometry (Fig. 6A).

Figure 6.

Diosmetin targets leukemia stem cells in vivo. A, Normal CD34+ cells from cord blood were injected via tail vein into NSG mice (n = 5/group). After 1 week, mice were treated intraperitoneally with 50 mg/kg/every other day with diosmetin. After 6 weeks, normal human myeloid cells (hCD45+/CD33+) in mouse bone marrow were detected by flow cytometry. B, Patient-derived AML cells were injected via tail vein into NSG mice (n = 10/group). After 1 week, mice were treated intraperitoneally with 50 mg/kg/every other day with diosmetin. After 6 weeks, human AML cells (hCD45+/CD33+) in mouse bone marrow were detected by flow cytometry. C, Secondary engraftments were performed, as detailed in the Materials and Methods section, Human AML cells (hCD45+/CD33+) in mouse bone marrow were detected by flow cytometry after 6 weeks. For all experiments, data were normalized to vehicle-treated controls and presented as percent engraftment. D–I, Compared with control-treated animals, diosmetin-treated mice had no changes in body weights, white blood cells, red blood cells, bilirubin levels, or serum levels of alkaline phosphatase (marker or kidney function) or creatine kinase (marker of liver damage). *, P < 0.05.

Figure 6.

Diosmetin targets leukemia stem cells in vivo. A, Normal CD34+ cells from cord blood were injected via tail vein into NSG mice (n = 5/group). After 1 week, mice were treated intraperitoneally with 50 mg/kg/every other day with diosmetin. After 6 weeks, normal human myeloid cells (hCD45+/CD33+) in mouse bone marrow were detected by flow cytometry. B, Patient-derived AML cells were injected via tail vein into NSG mice (n = 10/group). After 1 week, mice were treated intraperitoneally with 50 mg/kg/every other day with diosmetin. After 6 weeks, human AML cells (hCD45+/CD33+) in mouse bone marrow were detected by flow cytometry. C, Secondary engraftments were performed, as detailed in the Materials and Methods section, Human AML cells (hCD45+/CD33+) in mouse bone marrow were detected by flow cytometry after 6 weeks. For all experiments, data were normalized to vehicle-treated controls and presented as percent engraftment. D–I, Compared with control-treated animals, diosmetin-treated mice had no changes in body weights, white blood cells, red blood cells, bilirubin levels, or serum levels of alkaline phosphatase (marker or kidney function) or creatine kinase (marker of liver damage). *, P < 0.05.

Close modal

Primary engraftment was next assessed where patient-derived AML cells were injected into the tail vein of NSG mice and, after 1 week, intraperitoneal injections of diosmetin (50 mg/kg/every other day) or vehicle control were administered for 6 weeks. Human myeloid cells were then measured in mouse bone marrow. Treatment with diosmetin significantly reduced LSC primary engraftment compared with vehicle control–treated mice (Fig. 6B; U = 21; P < 0.05). In secondary transplant assays, an equal number of viable leukemia cells from control or diosmetin-treated mice were injected into the tail vein of NSG mice (n = 5/group) and, after 6 weeks, engraftment was assessed in mouse bone marrow. Human cells isolated from mice treated with the diosmetin had significantly reduced LSC secondary engraftment compared with vehicle control–treated mice (Fig. 6C; U = 2; P < 0.05). Analysis of complete blood counts, bilirubin, serological markers (i.e., alkaline phosphatase or creatine kinase, which are markers of kidney function or liver function, respectively), and mouse body weights showed no difference between control and diosmetin treatment groups (Fig. 6C–H). Together, this demonstrates the selective activity of the ERβ agonist diosmetin against primitive leukemia cells.

A high-throughput screen identified diosmetin as a novel anti-AML compound, and validation studies determined that this flavonoid was capable of inducing selective toxicity in leukemia and LSCs with no effect on normal cells in vitro and in vivo. We subsequently demonstrated that selective AML toxicity was related to ERβ targeting, highlighting this ER subtype as a novel anti-AML target.

ER isoforms, ERα and ERβ, are encoded by genes at different chromosomal locations and perform different cellular roles (23). While ERα activation results in cell proliferation, ERβ targeting results in cell senescence or death (24, 25). ERβ activation induced death in prostate (25), colon (26), and breast (26, 27) cancer tissue, and loss of ERβ in mice results in prostate hyperplasia and myeloproliferative disease (e.g., increased myeloid cell numbers in bone marrow, blood, spleen, and lymph nodes) resembling chronic myeloid leukemia (CML; ref. 28, 29). In this study, ERβ targeting with diosmetin-induced selective toxicity that was dependent on AML cell ERβ expression. Patient-derived AML cells with elevated ERβ mRNA were diosmetin sensitive, whereas insensitive cells had low-to-nondetectable ERβ levels. Moreover, ERβ knockdown conferred resistance, whereas overexpression, using doxycycline-inducible cells, resulted in enhanced cell death following ERβ targeting with diosmetin. Indeed, ERβ ligand activation or transfection of ER-negative cell lines with ERβ results in reduced proliferation, induction of apoptosis, and/or inhibition of tumor formation in prostate (30), colon (31), lymphoma (32, 33), and breast (26, 27) cancer models. Thus, ERβ targeting exerts antiproliferative effects and is demonstrated here, for the first time, as a novel anti-AML/LSC target.

High ratios of ERβ/ERα conferred AML cell sensitivity. Interrogation of publically available AML patient genetic databases identified an AML patient subpopulation with this phenotype confirming the potential prognostic significance of ERβ and clinical utility of ERβ drug targeting. Similarly, elevated ERβ levels were observed in CLL patients compared with normal samples (34). Low ERβ/ERα ratios were observed in breast tumors compared with matched normal adjacent tissue (35), and the presence of ERβ (ratio, 1.0–1.5) resulted in favorable patient response to chemotherapy (36). Downregulation and degradation of ERβ by Pescadillo (PES1) resulted in lower ERβ/ERα ratios and breast cancer progression in mice; elevated ERα and PES1 coupled with downregulated ERβ was also found in breast cancer patients (37). Moreover, ER hypermethylation in AML is a common clinical characteristic associated with improved patient outcome. In one study, ERα was specifically tested; however, no differentiation was made between ERα and ERβ in the other two studies (38–40). ERβ methylation (i.e., decreased ERβ activity) is observed in colon or breast tumors (41, 42). It is important to note; however, that while receptor expression is critical, other factors influence the cellular response to ER interactions with estrogen hormones or selective estrogen receptor modulators (SERMs; e.g., tamoxifen, raloxifene; ref. 43). Indeed, tamoxifen is an ER antagonist in breast, but a weak ER agonist in bone tissue, whereas raloxifene is a strong agonist in bone but a weak agonist in breast tissue (44). Although future studies are needed to further define the role of ERs in myeloid cells and leukemia pathogenesis, our study points to the direct antitumorigenic and prognostic potential of ERβ in AML.

Although a wide spectrum of ERβ agonists are under clinical development (AUS-131, fosfestrol, KB-9520, sulfestrol; refs. 42 and 45), none are FDA approved or have been tested clinically in AML. Diosmetin is widely available and when provided orally as diosmin at 10 mg/kg (the aglycone prodrug form), diosmetin reached a Cmax of 126 mmol/L within 2 hours with a half-life of 26–48 hours (46). In addition, administration of 500 mg of daflon, a common product used for chronic venous insufficiency that contains 450 mg of diosmin, results in 0.17–1.32 mmol/L diosmetin plasma concentrations (47, 48) with minimal reported toxicities (49, 50). Our study provides a clear rationale for preclinical testing and/or potential clinical evaluation of ER-based therapies for AML, specifically focusing on potent ERβ agonists such as diosmetin. Similar to our anti-AML in vivo observations, treatment with ERβ agonists has shown antitumor effects in lymphoma (33), prostate (30), and breast (51, 52) cancer models. Tamoxifen also enhanced cytarabine's antitumor activity in AML xenografts (53). However, tamoxifen decreased bone marrow cell numbers and increased myeloid progenitor apoptosis in healthy mice (53), which was convincingly shown to be ERα mediated (i.e., effects not shown in ERβ−/− mice). This emphasizes the need for caution when applying SERM therapies, as these agents exert tissue- and context-dependent activities and both ERα and ERβ are found on hematopoietic progenitors (54–56). Hence, clinical evaluation, specifically within the hematopoietic compartment, of potent and selective ERβ agonists is required to translate these observations into patient benefits. Nonetheless, our work demonstrates a potential clinical benefit of ERβ agonists in AML therapy, particularly in a subpopulation of patients who present with favorably high ERβ/ERα ratios.

In conclusion, ERβ expression conferred AML cell sensitivity to the ERβ agonist, diosmetin, highlighting diosmetin as a novel drug and ERβ as a potential novel anti-AML target.

No potential conflicts of interest were disclosed.

Conception and design: S.-G. Rota, P.A. Spagnuolo

Development of methodology: S.-G. Rota, A. Roma, D.A. Hess, A.C. Doxey, P.A. Spagnuolo

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): I. Dude, C. Ma, R. Stevens, J. MacEachern, P.N. Rao, P.A. Spagnuolo

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S.-G. Rota, A. Roma, I. Dude, C. Ma, S.M.G. Espiritu, P.N. Rao, P.A. Spagnuolo

Writing, review, and/or revision of the manuscript: S.-G. Rota, A. Roma, D.A. Hess, A.C. Doxey, P.A. Spagnuolo

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S.-G. Rota, A. Roma, C. Ma, M.D. Minden, E. Kreinin, D.A. Hess, P.A. Spagnuolo

Study supervision: P.A. Spagnuolo

We are grateful to Drs. John E. Dick and David G. Monroe for their generous gift of TEX cells and doxycycline-inducible ERβ cells, respectively; Rose Hurren, Jean Flanagan, Martin Ryan, and Nancy Gibson for assisting with the engraftment assays; and the practitioners at the Cancer Centres of Grand River and Princess Margaret for their assistance in procuring primary patient samples.

This work was supported by grants to P.A. Spagnuolo by the Stem Cell Network, Leukemia and Lymphoma Society of Canada, University of Waterloo, Ontario Research Fund, and the Canadian Foundation of Innovation; to Rao PPN by the Natural Sciences and Engineering Research Council of Canada (NSERC) and the Early Researcher Award; and to A.C. Doxey by NSERC.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1
Shipley
JL
,
Butera
JN
. 
Acute myelogenous leukemia
.
Exp Hematol
2009
;
37
:
649
58
.
2
Lowenberg
B
,
Downing
JR
,
Burnett
A
. 
Acute myeloid leukemia
.
N Engl J Med
1999
;
341
:
1051
62
.
3
Warner
JK
,
Wang
JC
,
Takenaka
K
,
Doulatov
S
,
McKenzie
JL
,
Harrington
L
, et al
Direct evidence for cooperating genetic events in the leukemic transformation of normal human hematopoietic cells
.
Leukemia
2005
;
19
:
1794
805
.
4
McDermott
SP
,
Eppert
K
,
Notta
F
,
Isaac
M
,
Datti
A
,
Al-Awar
R
, et al
A small molecule screening strategy with validation on human leukemia stem cells uncovers the therapeutic efficacy of kinetin riboside
.
Blood
2012
;
119
:
1200
7
.
5
Skrtic
M
,
Sriskanthadevan
S
,
Jhas
B
,
Gebbia
M
,
Wang
X
,
Wang
Z
, et al
Inhibition of mitochondrial translation as a therapeutic strategy for human acute myeloid leukemia
.
Cancer Cell
2011
;
20
:
674
88
.
6
Lee
EA
,
Angka
L
,
Rota
SG
,
Hanlon
T
,
Mitchell
A
,
Hurren
R
, et al
Targeting mitochondria with avocatin B induces selective leukemia cell death
.
Cancer Res
2015
;
75
:
2478
88
.
7
Monroe
DG
,
Getz
BJ
,
Johnsen
SA
,
Riggs
BL
,
Khosla
S
,
Spelsberg
TC
. 
Estrogen receptor isoform-specific regulation of endogenous gene expression in human osteoblastic cell lines expressing either ERalpha or ERbeta
.
J Cell Biochem
2003
;
90
:
315
26
.
8
Tcheng
M
,
Samudio
I
,
Lee
EA
,
Minden
MD
,
Spagnuolo
PA
. 
The mitochondria target drug avocatin B synergizes with induction chemotherapeutics to induce leukemia cell death
.
Leuk Lymphoma
2017
;
58
:
986
8
.
9
Spagnuolo
PA
,
Hurren
R
,
Gronda
M
,
Maclean
N
,
Datti
A
,
Basheer
A
, et al
Inhibition of intracellular dipeptidyl peptidases 8 and 9 enhances parthenolide's anti-leukemic activity
.
Leukemia
2013
;
27
:
1236
44
.
10
Angka
L
,
Lee
EA
,
Rota
SG
,
Hanlon
T
,
Sukhai
M
,
Minden
M
, et al
Glucopsychosine increases cytosolic calcium to induce calpain-mediated apoptosis of acute myeloid leukemia cells
.
Cancer Lett
2014
;
348
:
29
37
.
11
Ito
J
,
Tabei
Y
,
Shimizu
K
,
Tsuda
K
,
Tomii
K
. 
PoSSuM: a database of similar protein-ligand binding and putative pockets
.
Nucleic Acids Res
2012
;
40
:
D541
8
.
12
Ito
J
,
Ikeda
K
,
Yamada
K
,
Mizuguchi
K
,
Tomii
K
. 
PoSSuM v.2.0: data update and a new function for investigating ligand analogs and target proteins of small-molecule drugs
.
Nucleic Acids Res
2015
;
43
:
D392
8
.
13
Huang da
W
,
Sherman
BT
,
Lempicki
RA
. 
Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources
.
Nat Protoc
2009
;
4
:
44
57
.
14
Manas
ES
,
Xu
ZB
,
Unwalla
RJ
,
Somers
WS
. 
Understanding the selectivity of genistein for human estrogen receptor-beta using X-ray crystallography and computational methods
.
Structure
2004
;
12
:
2197
207
.
15
Rao
PP
,
Mohamed
T
,
Teckwani
K
,
Tin
G
. 
Curcumin binding to beta amyloid: a computational study
.
Chem Biol Drug Des
2015
;
86
:
813
20
.
16
Sachlos
E
,
Risueno
RM
,
Laronde
S
,
Shapovalova
Z
,
Lee
JH
,
Russell
J
, et al
Identification of drugs including a dopamine receptor antagonist that selectively target cancer stem cells
.
Cell
2012
;
149
:
1284
97
.
17
Reed
GA
,
Schiller
GJ
,
Kambhampati
S
,
Tallman
MS
,
Douer
D
,
Minden
MD
, et al
A phase 1 study of intravenous infusions of tigecycline in patients with acute myeloid leukemia
.
Cancer Med
2016
;
5
:
3031
40
.
18
Valk
PJ
,
Verhaak
RG
,
Beijen
MA
,
Erpelinck
CA
,
Barjesteh
van Waalwijk van Doorn-Khosrovani S
,
Boer
JM
, et al
Prognostically useful gene-expression profiles in acute myeloid leukemia
.
N Engl J Med
2004
;
350
:
1617
28
.
19
Haferlach
T
,
Kohlmann
A
,
Wieczorek
L
,
Basso
G
,
Kronnie
GT
,
Bene
MC
, et al
Clinical utility of microarray-based gene expression profiling in the diagnosis and subclassification of leukemia: report from the International Microarray Innovations in Leukemia Study Group
.
J Clin Oncol
2010
;
28
:
2529
37
.
20
Wang
IK
,
Lin-Shiau
SY
,
Lin
JK
. 
Induction of apoptosis by apigenin and related flavonoids through cytochrome c release and activation of caspase-9 and caspase-3 in leukaemia HL-60 cells
.
Eur J Cancer
1999
;
35
:
1517
25
.
21
Bishayee
K
,
Ghosh
S
,
Mukherjee
A
,
Sadhukhan
R
,
Mondal
J
,
Khuda-Bukhsh
AR
. 
Quercetin induces cytochrome-c release and ROS accumulation to promote apoptosis and arrest the cell cycle in G2/M, in cervical carcinoma: signal cascade and drug-DNA interaction
.
Cell Prolif
2013
;
46
:
153
63
.
22
Andueza
A
,
Garcia-Garzon
A
,
Ruiz de Galarreta
M
,
Ansorena
E
,
Iraburu
MJ
,
Lopez-Zabalza
MJ
, et al
Oxidation pathways underlying the pro-oxidant effects of apigenin
.
Free Radic Biol Med
2015
;
87
:
169
80
.
23
Liang
J
,
Xie
Q
,
Li
P
,
Zhong
X
,
Chen
Y
. 
Mitochondrial estrogen receptor beta inhibits cell apoptosis via interaction with Bad in a ligand-independent manner
.
Mol Cell Biochem
2015
;
401
:
71
86
.
24
Heldring
N
,
Pike
A
,
Andersson
S
,
Matthews
J
,
Cheng
G
,
Hartman
J
, et al
Estrogen receptors: how do they signal and what are their targets
.
Physiol Rev
2007
;
87
:
905
31
.
25
McPherson
SJ
,
Hussain
S
,
Balanathan
P
,
Hedwards
SL
,
Niranjan
B
,
Grant
M
, et al
Estrogen receptor-beta activated apoptosis in benign hyperplasia and cancer of the prostate is androgen independent and TNFalpha mediated
.
Proc Natl Acad Sci U S A
2010
;
107
:
3123
8
.
26
Hartman
J
,
Lindberg
K
,
Morani
A
,
Inzunza
J
,
Strom
A
,
Gustafsson
JA
. 
Estrogen receptor beta inhibits angiogenesis and growth of T47D breast cancer xenografts
.
Cancer Research
2006
;
66
:
11207
13
.
27
Strom
A
,
Hartman
J
,
Foster
JS
,
Kietz
S
,
Wimalasena
J
,
Gustafsson
JA
. 
Estrogen receptor beta inhibits 17beta-estradiol-stimulated proliferation of the breast cancer cell line T47D
.
Proc Natl Acad Sci U S A
2004
;
101
:
1566
71
.
28
Shim
GJ
,
Wang
L
,
Andersson
S
,
Nagy
N
,
Kis
LL
,
Zhang
Q
, et al
Disruption of the estrogen receptor beta gene in mice causes myeloproliferative disease resembling chronic myeloid leukemia with lymphoid blast crisis
.
Proc Natl Acad Sci U S A
2003
;
100
:
6694
9
.
29
Imamov
O
,
Morani
A
,
Shim
GJ
,
Omoto
Y
,
Thulin-Andersson
C
,
Warner
M
, et al
Estrogen receptor beta regulates epithelial cellular differentiation in the mouse ventral prostate
.
Proc Natl Acad Sci U S A
2004
;
101
:
9375
80
.
30
Cheng
J
,
Lee
EJ
,
Madison
LD
,
Lazennec
G
. 
Expression of estrogen receptor beta in prostate carcinoma cells inhibits invasion and proliferation and triggers apoptosis
.
FEBS Lett
2004
;
566
:
169
72
.
31
Hartman
J
,
Edvardsson
K
,
Lindberg
K
,
Zhao
C
,
Williams
C
,
Strom
A
, et al
Tumor repressive functions of estrogen receptor beta in SW480 colon cancer cells
.
Cancer Res
2009
;
69
:
6100
6
.
32
Yakimchuk
K
,
Hasni
MS
,
Guan
J
,
Chao
MP
,
Sander
B
,
Okret
S
. 
Inhibition of lymphoma vascularization and dissemination by estrogen receptor beta agonists
.
Blood
2014
;
123
:
2054
61
.
33
Yakimchuk
K
,
Iravani
M
,
Hasni
MS
,
Rhonnstad
P
,
Nilsson
S
,
Jondal
M
, et al
Effect of ligand-activated estrogen receptor beta on lymphoma growth in vitro and in vivo
.
Leukemia
2011
;
25
:
1103
10
.
34
Yakimchuk
K
,
Norin
S
,
Kimby
E
,
Hagglund
H
,
Warner
M
,
Gustafsson
JA
. 
Up-regulated estrogen receptor beta2 in chronic lymphocytic leukemia
.
Leuk Lymphoma
2012
;
53
:
139
44
.
35
Leygue
E
,
Dotzlaw
H
,
Watson
PH
,
Murphy
LC
. 
Altered estrogen receptor alpha and beta messenger RNA expression during human breast tumorigenesis
.
Cancer Res
1998
;
58
:
3197
201
.
36
Madeira
M
,
Mattar
A
,
Logullo
AF
,
Soares
FA
,
Gebrim
LH
. 
Estrogen receptor alpha/beta ratio and estrogen receptor beta as predictors of endocrine therapy responsiveness-a randomized neoadjuvant trial comparison between anastrozole and tamoxifen for the treatment of postmenopausal breast cancer
.
BMC Cancer
2013
;
13
:
425
.
37
Cheng
L
,
Li
J
,
Han
Y
,
Lin
J
,
Niu
C
,
Zhou
Z
, et al
PES1 promotes breast cancer by differentially regulating ERalpha and ERbeta
.
J Clin Invest
2012
;
122
:
2857
70
.
38
Toyota
M
,
Kopecky
KJ
,
Toyota
MO
,
Jair
KW
,
Willman
CL
,
Issa
JP
. 
Methylation profiling in acute myeloid leukemia
.
Blood
2001
;
97
:
2823
9
.
39
Niijima
T
,
Matsumura
Y
,
Katayama
Y
,
Morinaga
O
,
Ike
N
. 
A study on tumor of the bladder: clinical and prognostic studies on 426 cases (author's transl)
.
Nihon Hinyokika Gakkai Zasshi
1976
;
67
:
1057
63
.
40
Issa
JP
,
Zehnbauer
BA
,
Civin
CI
,
Collector
MI
,
Sharkis
SJ
,
Davidson
NE
, et al
The estrogen receptor CpG island is methylated in most hematopoietic neoplasms
.
Cancer Res
1996
;
56
:
973
77
.
41
Zhao
C
,
Lam
EW
,
Sunters
A
,
Enmark
E
,
De Bella
MT
,
Coombes
RC
, et al
Expression of estrogen receptor beta isoforms in normal breast epithelial cells and breast cancer: regulation by methylation
.
Oncogene
2003
;
22
:
7600
6
.
42
Rody
A
,
Holtrich
U
,
Solbach
C
,
Kourtis
K
,
von Minckwitz
G
,
Engels
K
, et al
Methylation of estrogen receptor beta promoter correlates with loss of ER-beta expression in mammary carcinoma and is an early indication marker in premalignant lesions
.
Endocrine-related Cancer
2005
;
12
:
903
16
.
43
Deroo
BJ
,
Korach
KS
. 
Estrogen receptors and human disease
.
J Clin Invest
2006
;
116
:
561
70
.
44
Riggs
BL
,
Hartmann
LC
. 
Selective estrogen-receptor modulators – mechanisms of action and application to clinical practice
.
N Engl J Med
2003
;
348
:
618
29
.
45
NIH
. 
2015
.
Available from
: https://clinicaltrials.gov/ct2/home.
46
Cova
D
,
De Angelis
L
,
Giavarini
F
,
Palladini
G
,
Perego
R
. 
Pharmacokinetics and metabolism of oral diosmin in healthy volunteers
.
Int J Clin Pharmacol Ther Toxicol
1992
;
30
:
29
3
.
47
Campanero
MA
,
Escolar
M
,
Perez
G
,
Garcia-Quetglas
E
,
Sadaba
B
,
Azanza
JR
. 
Simultaneous determination of diosmin and diosmetin in human plasma by ion trap liquid chromatography-atmospheric pressure chemical ionization tandem mass spectrometry: application to a clinical pharmacokinetic study
.
J Pharm Biom Anal
2010
;
51
:
875
81
.
48
Spanakis
M
,
Kasmas
S
,
Niopas
I
. 
Simultaneous determination of the flavonoid aglycones diosmetin and hesperetin in human plasma and urine by a validated GC/MS method: in vivo metabolic reduction of diosmetin to hesperetin
.
Biomed Chromatogr
2009
;
23
:
124
31
.
49
Thanapongsathorn
W
,
Vajrabukka
T
. 
Clinical trial of oral diosmin (Daflon) in the treatment of hemorrhoids
.
Dis Colon Rectum
1992
;
35
:
1085
8
.
50
Pecking
AP
,
Fevrier
B
,
Wargon
C
,
Pillion
G
. 
Efficacy of Daflon 500 mg in the treatment of lymphedema (secondary to conventional therapy of breast cancer)
.
Angiology
1997
;
48
:
93
8
.
51
Lazennec
G
,
Bresson
D
,
Lucas
A
,
Chauveau
C
,
Vignon
F
. 
ER beta inhibits proliferation and invasion of breast cancer cells
.
Endocrinology
2001
;
142
:
4120
30
.
52
Treeck
O
,
Pfeiler
G
,
Horn
F
,
Federhofer
B
,
Houlihan
H
,
Vollmer
A
, et al
Novel estrogen receptor beta transcript variants identified in human breast cancer cells affect cell growth and apoptosis of COS-1 cells
.
Mol Cell Endocrinol
2007
;
264
:
50
60
.
53
Sanchez-Aguilera
A
,
Arranz
L
,
Martin-Perez
D
,
Garcia-Garcia
A
,
Stavropoulou
V
,
Kubovcakova
L
, et al
Estrogen signaling selectively induces apoptosis of hematopoietic progenitors and myeloid neoplasms without harming steady-state hematopoiesis
.
Cell Stem Cell
2014
;
15
:
791
804
.
54
Thurmond
TS
,
Murante
FG
,
Staples
JE
,
Silverstone
AE
,
Korach
KS
,
Gasiewicz
TA
. 
Role of estrogen receptor alpha in hematopoietic stem cell development and B lymphocyte maturation in the male mouse
.
Endocrinology
2000
;
141
:
2309
18
.
55
Igarashi
H
,
Kouro
T
,
Yokota
T
,
Comp
PC
,
Kincade
PW
. 
Age and stage dependency of estrogen receptor expression by lymphocyte precursors
.
Proc Natl Acad Sci U S A
2001
;
98
:
15131
6
.
56
Nakada
D
,
Oguro
H
,
Levi
BP
,
Ryan
N
,
Kitano
A
,
Saitoh
Y
, et al
Oestrogen increases haematopoietic stem-cell self-renewal in females and during pregnancy
.
Nature
2014
;
505
:
555
8
.