MELK-dependent FOXM1 phosphorylation is essential for proliferation of glioma stem cells

Abstract

Glioblastoma multiforme (GBM) is a life-threatening brain tumor. Accumulating evidence suggests that eradication of glioma stem-like cells (GSCs) in GBM is essential to achieve cure. The transcription factor FOXM1 has recently gained attention as a master regulator of mitotic progression of cancer cells in various organs. Here, we demonstrate that FOXM1 forms a protein complex with the mitotic kinase MELK in GSCs, leading to phosphorylation and activation of FOXM1 in a MELK kinase-dependent manner. This MELK-dependent activation of FOXM1 results in a subsequent increase in mitotic regulatory genes in GSCs. MELK-driven FOXM1 activation is regulated by the binding and subsequent trans-phosphorylation of FOXM1 by another kinase PLK1. Using mouse neural progenitor cells (NPCs), we found that transgenic expression of FOXM1 enhances, while siRNA-mediated gene silencing diminishes neurosphere formation, suggesting that FOXM1 is required for NPC growth. During tumorigenesis, FOXM1 expression sequentially increases as cells progress from NPCs, to pretumorigenic progenitors and GSCs. The antibiotic Siomycin A disrupts MELK-mediated FOXM1 signaling with a greater sensitivity in GSC compared to neural stem cell. Treatment with the first-line chemotherapy agent for GBM, Temozolomide, paradoxically enriches for both FOXM1 (+) and MELK (+) cells in GBM cells, and addition of Siomycin A to Temozolomide treatment in mice harboring GSC-derived intracranial tumors enhances the effects of the latter. Collectively, our data indicate that FOXM1 signaling through its direct interaction with MELK regulates key mitotic genes in GSCs in a PLK1-dependent manner and thus, this protein complex is a potential therapeutic target for GBM.

Figure 1

FoxM1 expression is restricted in neural progenitor cells in the mouse brain. (A): Reverse transcription polymerase chain reaction (RT-PCR) analysis for FoxM1 and Melk expression during brain development. (B): RT-PCR analysis for FoxM1 and Melk in neural progenitor cells with or without differentiation. Expression is compared with the differentiation markers for neurons (neurofilament heavy chain [Nfh]), astrocytes (Gfap), and oligodendrocytes (proteolipid protein [Plp]). For (A) and (B), Gapdh is used as internal control. (C): Left panels, top two images indicate immunohistochemistry for FoxM1 (green) and Sox2 (red) in mouse brains of embryonic day 17 (E17). Lower four pictures, immunofluorescence for FoxM1 (green) and Tuj1 (red) in mouse brains of E17. Original magnification is ×10 and ×40. Scale bars = 100 and 20 μm. (a) CZ, (b) GZ, (c) ST represents magnified pictures of FOXM1 (green) and Tuj-1 (red). Original magnification is ×10 and ×40. Scale bars = 100 and 20 μm. Right panels represent immunohistochemistry for FoxM1 (green), Gfap (red), BrdU (red), and Dcx in brains of the postnatal mouse day 30. Original magnification is ×20–×40. Scale bars = 50–20 μm, respectively. (D): Immunohistochemistry of FOXM1 (brown) in adult mouse brain at the SVZ (left panel) and at the subgranular zone (right panel). Original magnification is ×40. Scale bars = 20 μm. (E): Bar graph representing the relative number of NS formed from EGFP- or FoxM1-expressing NS derived from mouse E17 cerebral cortices. Asterisks (*) indicate statistical significance by t test. (F): Left panel exhibiting diagram of two different isoforms of FoxM1 protein. Right panel indicating the relative number of NS formed from FoxM1a or FoxM1a/b siRNA transfected NS derived from mouse E17 cerebral cortices. NS forming assay was performed in triplicate in 96-well plate and repeated three times independently. Abbreviations: BrdU, Bromodeoxyuridine; CC, corpus callosum, CZ, cortical zone; DCX, doublecortin; DC, differentiated cells; EGFP, enhanced green fluorescent protein; ES, embryonic stem cells; E, embryonic brain; GFAP, glial fibrillary acidic protein; GZ, germinal zone; NS, neurospheres; P, postnatal brain; SVZ, subventricular zone; ST, striatum; V, ventricle; VZ, ventricular zone.

Figure 2

FoxM1 expression is markedly elevated during gliomagenesis. (A): Representative images of immunohistochemistry for FoxM1 (brown) in the SVZ of wild-type mice, SVZ of pretumorigenic Mut6 mice, and GBM-like tumors in cerebral cortex or normal areas in brain stem in Mut6 mice. Original magnification is ×10. Scale bars = 100 μm. Magnified images are shown below. Original magnification is ×40. Scale bars = 20 μm. (B): Quantitative reverse transcription polymerase chain reaction (qRT-PCR) analysis for FoxM1 expression in GBM tissues in Mut6 mice compared to normal side brain tissues derived from the same mice. Asterisks (*) indicate statistical significance by t test. Experiment repeated in three Mut-6 mouse-derived GBM tissues. (C): Phase bright representative images of NS derived from either normal SVZ or GBM-like tumors developed in Mut6 mice. Original magnification: ×10. Scale bars = 100 μm. Lower panel indicates qRT-PCR data for FoxM1 expression in NS derived from GBM-like tumors (GBM NS) in Mut6 mice or normal SVZ (normal NS). Asterisk (*) indicates statistical significance by t test. Experiments were done in triplicate and repeated three times independently. (D): Microarray analysis of relative FOX genes family expression in GBM (n = 30) compared to normal astrocytes cultures (n = 4). Expression of astrocytes was normalized as one. Expression of FOXM1 is shown in red. (E): Relative expression of FOXM1 of 10 GSCs samples compared to normal astrocytes cultures. FOXM1 expression of normal astrocytes was normalized as one. (F): Relative mRNA expression of FOXM1 and MELK in GBM157 and GBM528 CD133 (+) and CD133 (−) cells collected after cell sorting (n = 3). (G): Representative immunocytochemistry of GBM NS and GBM cells propagated in serum-containing medium (SPGCs) derived from GBM30. Cells were double-stained for FOXM1 (red) in combination with one of the following neural progenitor cell markers Nestin, SOX2 and differentiation markers TuJ1, GFAP (green). Hoechst dye is used for nuclear staining (blue). Original magnification: ×40. Scale bars = 20 μm. Right panel indicates of the proportions of FOXM1 positive cells coexpressing SOX2, Nestin, GFAP, and TuJ1 positive cells. For quantification, GBM30 sample were used and the experiment was repeated four times. (H): Flow cytometry analysis for the expression of FOXM1, SOX-2, and GFAP using GBM NSs and SPGCs, both of which are derived from GBM30. Abbreviations: FITC, fluorescein isothiocyanate; FSC, forward scatter; GFAP, glial fibrillary acidic protein; GBM, glioblastoma multiforme; NS, neurosphere; SPGC, serum-propagated GBM cells; SVZ, subventricular zone.

Figure 3

FOXM1 is a substrate for MELK. (A): Comparison of MELK and FOXM1 expression profile (Affymetrix Human Genome U133A Array) indicates statistically significant correlation of the expression of these two genes in grade III glioma (left) (n = 24) and in GBM (middle) (n = 56). The Cancer Genome Atlas (TCGA) analysis of MELK and FOXM1 expression profile in 218 GBM patient samples indicates statistical significance (right panel). (B): Representative images of immunocytochemistry with GBM30 neurospheres for FOXM1 (red), MELK (green). Hoechst dye for nuclear staining (blue). Original magnification: ×40. Scale bar = 20 μm. For quantification, GBM30 samples were used and the experiment was repeated four times. (C): Reverse transcription polymerase chain reaction (RT-PCR) analysis for MELK and FOXM1 expression in GBM neurospheres treated with siRNA targeting three different sequences for FOXM1 (left). RT-PCR analysis for MELK and FOXM1 expression in GBM neurospheres treated with different doses of siRNA targeting MELK (right) n = 5. (D): Upper panel: Overexpression of Empty-EGFP (control) + MELK-Flag (Flag-MELK), EGFP-FOXM1 + MELK-Flag, or EGFP-FOXM1 + MELK D150A-Flag plasmids in HEK293 cells are processed to GFP-trap followed by immunoblotting with anti-Flag antibody. Middle panel: Autoradiogram displaying in vitro phosphorylation of FOXM1 by the kinase domain of MELK (1–340). Lane 1: FOXM1 + ATP-Mg (no kinase); lanes 2 and 6: FOXM1 + ATP-Mg + MELK (1–340); lane 3: FOXM1 + MELK (1–340) (no ATP-Mg); lane 4: FOXM1 + ATP-Mg + CDK2/CyclA; lane 5: FOXM1 + ATP-Mg + CDK2/CyclA + roscovitine; lane 7: FOXM1+ ATP-Mg + MELK1–340(D150A). Lower panels: Coomassie staining of the samples subjected to autoradiography. Abbreviations: GBM, glioblastoma multiforme; IP, immunoprecipitation.

Figure 4

MELK phosphorylates FOXM1 and regulates FOXM1 activity, leading to upregulation of mitotic gene expression. (A): Graph indicating the FOXM1 promoter activity in 293T cells transfected with the 6×FOXM1 TATA-luciferase plasmid together with expression vectors encoding wild-type FOXM1 and increasing amounts of plasmids encoding either WT MELK or kinase-dead mutant form of MELK (D150A). The experiment was performed in triplicate in 96 well plates and repeated three times independently. (B): Relative mRNA expression levels of Survivin, CyclinB1, CDC25B, and Aurora B by quantitative reverse transcription polymerase chain reaction (qRT-PCR) in GBM30 cells transfected with GFP, FOXM1, FOXM1 + WT MELK, and FOXM1 + MELK D150A mutant. qRT-PCR was performed in triplicate and repeated three times independently. Abbreviations: EGFP, enhanced green fluorescent protein; WT, wild type.

Figure 5

MELK-driven FOXM1 phosphorylation is dependent on PLK-1. (A): Graph indicating the FOXM1 promoter activity in 293T cells with transfection of the FOXM1 reporter plasmid together with the plasmids encoding WT FOXM1 (FOXM1WT) and MELK WT, PLK-1 WT, or combination, in the presence of different doses (100, 250, and 500 nM) of the PLK-1 inhibitor ON-01910 for 48 hours. (B): Graph indicating the FOXM1 promoter activity in 293T cells with transfection of the FOXM1 reporter plasmid together with the plasmids encoding either FOXM1WT and plasmids encoding for WT or kinase-dead mutant (D150A) of MELK or PLK-1 WT and DN PLK-1 mutant. (C): Graph indicating the FOXM1 promoter activity in indicated cells with transfection of the FOXM1 reporter plasmid together with the plasmids encoding constitutive active mutant FOXM1 (FOXM1EE) and plasmids encoding for WT or kinase-dead mutant (D150A) of MELK. (D): Graph indicating the FOXM1 promoter activity in indicated cells with transfection of the FOXM1 reporter plasmid together with the plasmids encoding either WT FOXM1 (FOXM1WT) or FOXM1 mutants (715A, 724A, or 715/724A) and plasmids encoding for WT or kinase-dead mutant (D150A) of MELK. (E): Graph indicating the FOXM1 promoter activity in indicated cells with transfection of the FOXM1 reporter plasmid together with the plasmids encoding either WT FoxM1 (FOXM1WT) or FOXM1 mutants (S596A, S678A, or TSAA) and plasmids encoding for WT or kinase-dead mutant (D150A) of MELK. All luciferase experiments were performed using 96 well plate and repeated three times independently. Abbreviations: DN, dominant negative; GBM, glioblastoma multiforme; WT, wild type.

Figure 6

SM abrogates MELK-driven FOXM1 activity and inhibits growth of glioma stem-like cells (GSCs) but not normal progenitors in mouse and human. (A): Flow cytometry analysis on GBM30 spheres treated with either DMSO or SM. The upper panels indicate the proportions of FOXM1 (+) cells in GBM30 spheres with indicated dose of SM treatment for 72 hours. The lower panels display the proportions of MELK (+) cells in each condition. Experiment was repeated three times and similar expression pattern was observed. (B): Graph indicating the FOXM1 promoter activity in indicated cells transfected with the plasmids encoding for FOXM1 wild-type (WT) with MELK WT or kinase dead D150A. The cells were then treated with indicated doses of SM after transfection for 48 hours, followed by detection of FOXM1 activity by luciferase assay. (C): Graph showing cell cycle analysis with flow cytometry of GBM30 GSCs. The cells were treated with either control (DMSO 1%), SM (500 nM), FOXM1 wild-type overexpression together with SM (500 nM), MELK wild-type overexpression together with SM (500 nM) or MELK, and FOXM1 wild-type overexpression together with SM (500 nM) for 48 hours. (D): Left panel indicates the relative neurosphere numbers formed from mouse subventricular zone or GBM-like tumors with SM (500 nM) treatment or DMSO. The right panel shows the relative neurosphere numbers formed from GBM30 GSCs and normal spheres (16wf).The different numbers of cells were seeded in each well, as shown on x-axis. DMSO concentration is 1% and SM was 500 nM. The experiments were done in triplicates using 96 well plates and repeated three times. Abbreviations: DMSO, dimethyl sulfoxide; FITC, fluorescein isothiocyanate; FSC, forward scatter; GBM, glioblastoma multiforme; SM, Siomycin A.

Figure 7

Combined treatment of TMZ with SM on glioma stem-like cell (GSC)-derived mouse tumors yields better survival than monotherapy with TMZ. (A): Flow cytometry analysis for FOXM1 (upper panels) and MELK (lower panels) with GBM30 SPGCs (serum-propagated GBM cells) treated with varying doses of TMZ for 72 hours. Experiment was repeated three times for confirmation of results. (B): Representative images of mouse brains with intracranial xenograft tumors derived from GBM30 neurospheres (left top panel). Middle and lower pictures indicate H&E staining. N indicates necrotic area in the tumor. Original magnifications: ×2 (middle panel) and ×10 (lower panel). Scale bar = 500 μm (middle panel) and 100 μm (lower panel). (C): Ki-67 immunohistochemistry of mouse tumors treated with either DMSO or SM. Mice were sacrificed at 2 days post-SM treatment. Original magnifications: ×20. Scale bars = 50 μm. Graph (right) indicating the proportion of Ki-67(+) cells in DMSO- and SM-injected tumors analyzed by Image J software. (n = 4 for each group) Asterisk (*) indicates statistical significance by t test. (D): Graph indicating the relative neurosphere numbers derived from mouse tumor tissues following DMSO and SM treatment for 2 days. (n = 3) Asterisk (*) indicates statistical significance by t test. (E): Kaplan Meier survival curve of mice harboring GBM30 neurosphere-derived tumors treated with DMSO (control), TMZ (10 mg/kg), or TMZ (10 mg/kg) combined with SM injection (2.5 nM). Table (right) indicates the mean and medial survival periods of the three groups. Abbreviations: DMSO, dimethyl sulfoxide; FITC, fluorescein isothiocyanate; FSC, forward scatter; SM, Siomycin A; TMZ, Temozolomide.