PI-3K Inhibitors Preferentially Target CD15+ Cancer Stem Cell Population in SHH Driven Medulloblastoma

Abstract

Sonic hedgehog (SHH) medulloblastoma (MB) subtype is driven by a proliferative CD15+ tumor propagating cell (TPC), also considered in the literature as a putative cancer stem cell (CSC). Despite considerable research, much of the biology of this TPC remains unknown. We report evidence that phosphatase and tensin homolog (PTEN) and phosphoinositide 3-kinase (PI-3K) play a crucial role in the propagation, survival and potential response to therapy in this CD15+ CSC/TPC-driven malignant disease. Using the ND2-SmoA1 transgenic mouse model for MB, mouse genetics and patient-derived xenografts (PDXs), we demonstrate that the CD15+TPCs are 1) obligately required for SmoA1Tg-driven tumorigenicity 2) regulated by PTEN and PI-3K signaling 3) selectively sensitive to the cytotoxic effects of pan PI-3K inhibitors in vitro and in vivo but resistant to chemotherapy 4) in the SmoA1Tg mouse model are genomically similar to the SHH human MB subgroup. The results provide the first evidence that PTEN plays a role in MB TPC signaling and biology and that PI-3K inhibitors target and suppress the survival and proliferation of cells within the mouse and human CD15+ cancer stem cell compartment. In contrast, CD15+ TPCs are resistant to cisplatinum, temozolomide and the SHH inhibitor, NVP-LDE-225, agents currently used in treatment of medulloblastoma. These studies validate the therapeutic efficacy of pan PI-3K inhibitors in the treatment of CD15+ TPC dependent medulloblastoma and suggest a sequential combination of PI-3K inhibitors and chemotherapy will have augmented efficacy in the treatment of this disease.

Fig 1

CD15+ TPCs isolated from SmoA1 Tg medulloblastoma mouse model are highly proliferative and shows elevated PI-3K signaling (A) Left panel shows proliferation of CD15+ and CD15- cells isolated from Smo A1 tumors. FACS sorted CD15+ and CD15− tumor cells (n = 4) were cultured for 48 hours in serum-free medium, followed by addition of AlamarBlue® and incubation of plates at 37°C in 5% CO₂ for 6 hours. Fluorescence signals were read as emission at 590 nm after excitation at 560 nm. Right panel shows BrDU incorporation in CD15+ and CD15- cells isolated from the Smo A1 Tg. CD15+ and CD15- from Smo tumors were pulsed with BrDU as described in Materials and Methods. (B) FACS sorted CD15+ and CD15- cells were analyzed for expression of PTEN and phosphorylation of Akt, 4EBP1, P70S6K and Erk. β-actin was used as a loading control. (C) Quantitative PCR analysis of mRNA for expression of PTEN in FACS sorted CD15+ and CD15- cells (n = 3). Relative expression levels were normalized to GAPDH expression. Experiment was repeated 3 times with similar results.

Fig 2

Preferential targeting of TPCs by PI-3K inhibitors in vitro (A) Effect of PI-3K inhibitors on proliferation of CD15+ and CD15- cells. CD15+ and CD15- cells were cultured in serum-free media containing no additive, DMSO (vehicle), BKM-120, BEZ-235, PF-04691502, and SF1126 at different conc. After 48 hr, AlamarBlue® was added and plates were incubated at 37°C in 5% CO₂ for 6 hours. Fluorescence signals were read as emission at 590 nm after excitation at 560 nm. (B) CD15+ and CD15- cells were treated with 100nM conc. of cisplatinum, TMZ, NVP-LDE-225 either alone or in combination with BKM 120 and analyzed for cell viability using Alamar Blue. (C) CD15+ TPCs were treated with BKM120 for 30 minutes followed by stimulation with IGF (50 ng/ml). Cell lysates were analyzed by Western blot for phosphorylation of substrates of PI-3K signalling. (D) Relative expression of SHH pathway genes in BKM120 treated (0.2, 2.0 μM) and untreated CD15+ TPCs. Relative expression levels were normalized to GAPDH. Graphs present mean ± SEM of 3–4 mice in each group for B and D. Statistical significance is assessed by two sample t-test where *denotes P<0.05, ** denotes P<0.01 and *** denotes P<0.001. Experiment was repeated 4–5 times with similar results.

Fig 3. PI-3K inhibitors induce cell cycle arrest & apoptosis in CD15+ TPCs.

(A) Fig shows cell cycle analysis data on BKM120 (0.2 μM, 2.0 μM and 10 μM) treated and untreated CD15+ and CD15- cells. Cells were harvested and analyzed for cell cycle profile (p < 0.05 for all phases compared to untreated group). (B) Western blot data for expression of p27^Kip1 and p21^Waf1/Cip1 in BKM120 (2.0 μM) treated CD15+ cells (30 min treatment). (C) Relative gene expression of cell cycle genes in BKM120 treated (0.2, 2.0 μM) CD15+ TPCs. Relative expression levels were normalized to GAPDH expression. (D) CD15+ and CD15- cells were treated with BKM120 (0.2 μM, 2.0 μM and 10 μM) for 24 hours, assayed for apoptosis by using the Annexin V ^FITC assay. (E) Increased caspase 3 activation with increasing conc. of BKM120. CD15+ and CD15- TPCs were treated with different conc. of BKM120 and used for fluorometric caspase 3 activity according to manufacturer’s instructions (Roche Diagnostic GmbH, Germany). (F) Western blotting of pMdm2 (S166), p53, Bax, Bad, cleaved caspase3 and cleaved PARP in CD15+ TPCs treated with different concentrations of BKM120. Results are mean ± SEM (n = 3–4 mice) for 3 independent experiments performed in triplicate (A, C, D & E). *P <0.05, **P <0.01 and ***P <0.001 vs. untreated, t test.

Fig 4. BKM120 inhibits tumor growth by depleting CD15+ TPCs in vivo.

(A) Tumor growth was suppressed in BKM120 treated CD15+TPC xenografts. Nu-nu mice subcutaneously implanted with 10 million CD15+ TPCs were treated for 21 days with vehicle (control) or BKM120 at 30 mg/kg dose. Data are expressed as mean ± SEM (n = 6) (***P <0.001, t test) (B) FACS analysis on CD15+ cells isolated from BKM120 treated tumors as compared to untreated control mice. (C) Passaging of CD15+ or CD15- cells from second generation subcutaneous tumors into NOD SCID gamma (NSG) mice brain (orthotopic transplantation) in vivo. Left panel shows MRI images indicating, only CD15+ cell implanted NSG mice and not CD15- cells implanted mice formed tumor (white circle) in the MRI. Right panel shows histopathologic analysis of the CD15+ and CD15- transplanted NSG mice showing cortical tumor growth in only CD15+ implanted NSG mice. (D) 2 million CD15+ TPCs were intracranially transplanted into nude mice. After 40 days, mice displayed neurologic symptoms and were treated with vehicle (15% ethanol &15% cremophore, or BKM120 (30 mg/kg/day), 6 days a week for 21 days by oral gavage (*P <0.05 compared to vehicle, student’s t-test). Experiments were repeated 3 times with 7–8 mice in each group.

Fig 5. Similarity of CD15+ population from SmoA1 PTEN +/+ tumors to the ‘c3/SHH’ subtype of human MB.

(A) The heatmap shows the degree of similarity as quantified by the SubMap method [47] between gene expression levels of murine SmoA1Tg tumors (n = 5 paired CD15+ and CD15 samples) and 199 human tumors previously classified into 6 MB subtypes and normal control samples [4]. Each block in the heatmap corresponds to the p-value for similarity between the row category (mouse model samples, CD15+ or CD15-) and the column category (human tumor samples, classed by subtype). Blue suggests no similarity and red suggests strong similarity in gene expression levels. Note that the CD15- murine samples show the strongest similarity with human tumors subtypes c4 (Group 4) and c7 (normal CBL), while the CD15+ samples show similarity with c1 (Group 3) and c3 (SHH) subtypes. (B) Upper panel shows two dimensional Principal Components Analysis (PCA) plot comparing expression levels of the 22 leading edge genes from the SubMap analysis of Fig 5A. Data are from n = 5 paired CD15+ and CD15- murine SmoA1 PTEN+/+ tumors (open diamonds) and n = 199 human tumor samples (solid dots) previously classified into one of 6 MB subtypes [4] and reference normal samples. The human tumors are labeled according to subytpes as in Cho et al [4]. The first Principal Component, PC1, is plotted as the x-axis, and shows the major mode of variation in the data. PC1 is observed to separate the human tumors into three major groups, horizontally from right to left: c3 (purple, SHH subtype), c6 (orange, WNT subtype), and the combined group c1/c2/c4/c5/c7. Importantly, the CD15+ mouse samples (purple diamonds) are observed to associate with the human samples of SHH subtype (group c3), while the CD15- mouse samples (green diamonds) are clustered with the leftmost combined group c1/c2/c4/c5/c7 group containing normal human samples. The second PC, PC2 (y-axis) shows the next largest mode of variation in the data, and is seen to further divide the human the samples in the c1/c2/c4/c5/c7 combined group. The human normal samples (c7, grey) are clustered at the extreme high values of PC2, are largely distinct from the human tumor samples. Human subtypes c2 (black) and c4 (green) are highly overlapped as expected, but some separation is observed between human subtypes c1 (brown) and c5 (yellow), suggesting that these subtypes have distinct molecular phenotypes. Lower panel shows the major classification of medulloblastoma in 4 major subgroups by Taylor et al.[55] (C) List of 22 leading genes obtained from submap analysis.

Fig 6. Augmented antitumor activity of BKM120 against human CD15+ cells isolated from SHH MB PDX l.

(A) FACS analysis of CD15 expression in a human SHH subgroup medulloblastoma tumor. (B), Total tumor cells, FACS sorted CD15+ and CD15− tumor cells were used for cell viability assay in presence of BKM120 using AlamarBlue®. (C) CD15+ and CD15- TPCs from human SHH MB patient tumor were cultured in serum-free media containing no additive, DMSO (vehicle), BKM 120 at different conc. followed by cell viability assay using AlamarBlue®. (D) Western blot analysis demonstrates suppression of PI-3K signaling in BKM120 treated CD15+ TPCs isolated from a human medulloblastoma tumor. (E) BKM120 treatment significantly prolonged survival of mice and reduce the tumor size. 2 million CD15+ TPCs were transplanted intracranially into nu/nu mice. After 90 days, MRIs taken in the coronal plane show the formation of a tumor (white circle) in nude mice followed by separation of mice into two random groups. Mice in group1 were treated with vehicle (15% ethanol &15% Cremophore) and those in group 2 were treated with BKM120 (30 mg/kg/day), 5 days / week for 14 days by oral gavage. (F) Sub-classification of human MB tumor and PDX by real time RT-PCR. SHH pathway genes (SFRP1, gli-1, gli-2 and Smo) are up-regulated in human MB (upper panel) and xenograft (lower panel) as compared to WNT pathway genes (Wnt-1, DKK1, DKK2) and non SHH/non Wnt (BMI-1), validated by real time RTPCR.