Distinct neural stem cell populations give rise to disparate brain tumors in response to N-MYC

Abstract

The proto-oncogene MYCN is mis-expressed in various types of human brain tumors. To clarify how developmental and regional differences influence transformation, we transduced wild-type or mutationally stabilized murine N-myc(T58A) into neural stem cells (NSCs) from perinatal murine cerebellum, brain stem, and forebrain. Transplantation of N-myc(WT) NSCs was insufficient for tumor formation. N-myc(T58A) cerebellar and brain stem NSCs generated medulloblastoma/primitive neuroectodermal tumors, whereas forebrain NSCs developed diffuse glioma. Expression analyses distinguished tumors generated from these different regions, with tumors from embryonic versus postnatal cerebellar NSCs demonstrating Sonic Hedgehog (SHH) dependence and SHH independence, respectively. These differences were regulated in part by the transcription factor SOX9, activated in the SHH subclass of human medulloblastoma. Our results demonstrate context-dependent transformation of NSCs in response to a common oncogenic signal.

Figure 1. GLT1 drives MYCN-dependent, predominantly SHH-negative medulloblastoma and is a fate determinant for NSCs

(A) Immunoblot of MYCN in GTML MB spheres (passages 6–10). (B–C) Proliferation of MYCN-low (GTML1) and MYCN-high (GTML2) spheres as indicated. (D–E) All GTML sphere lines except GTML1 were insensitive to the SMO inhibitor cyclopamine (5μM). Tomatidine, 5μM, was used as control. (F–I) Relative expression of neuronal markers Ngn1 and Syp; and Olig2 and Sox9 that are expressed in glial cells in embryonic (E16 WT), and postnatal (P0 WT) normal neurospheres and tumor spheres GTML1-4. (J–K) GLT1-positive origin shown by X-gal fate mapping in NSCs isolated from postnatal (P0) cerebellum and postnatal (P0) forebrain from a Glt1-tTA; TRE-Cre-Rosa26-LSL-LacZ mouse. Error bars: +/− Standard Deviation (SD). Scale: 100μm. See also Figure S1.

Figure 2. NSC-selective infection with N-myc^T58A alters proliferation and differentiation

(A) NSCs isolated from E16 and P0 forebrain (ventricular zone), E16 and P0 hindbrain (total cerebellum) and E14, E16 and P0 brain stem (ventricular zone/lower rhombic lip) from mice transgenic for Gtv-a were cultured for 7d in NB media on low-affinity plates with one or two passages. RCAS viruses containing GFP, N-myc^WT (N-MYC) or N-myc^T58A (T58A) came from supernatants of DF-1 virus producing cells (see Figure S2). NSCs were transduced with RCAS viruses mixed 1:1 with fresh NB media. Cells were infected (72h), dissociated and cultured in fresh NB media. Cells were orthotopically transplanted in nude mice 7 days after RCAS infection. (B) Selective infection of GFAP-positive P0 cerebellar NSCs, evidenced by co-expression of RCAS-GFP (green) with GFAP (blue) 72h after growth factor depletion. TUJ1 (red). Scale: 25 μm. (C) Immunoblot of FLAG-tagged N-myc^WT or N-myc^T58A in P0 cerebellar NSCs. NGN1 and β-actin control are also shown. (D–F) Proliferation at 3, 5 and 7 days after transduction with GFP, N-myc^WT or N-myc^T58A of P0 NSCs isolated from cerebellum (D), forebrain (E) and dorsal brain stem/lower rhombic lip (F) respectively. (G–J) Growth factor independence assay showing sphere diameter (as indicated) and number of secondary spheres in P0 cerebellar NSCs transduced with GFP, N-myc^WT or N-myc^T58A and in GTML2 spheres. Although not indicated, there are significantly more spheres irrespective of size in “GF conditions” (FGF, EGF or FGF+EGF) as compared to “no GF conditions” when analyzing GFP-, N-MYC- and T58A-transduced NSCs, respectively, when using Student’s t-test (p < 0.05). Error bars: +/− SD. See also Figure S2.

Figure 3. The N-myc^T58A-driven neuronal program depends on regional origin and NSC age

(A) Levels of N-myc and Gfap from Affymetrix exon arrays (log₂ values) and immunoblots of SOX9 and β-actin in E16 and P0 cerebellar and P0 forebrain NSCs transduced with GFP or N-myc^T58A, respectively. (B–E) Proliferation (7 days) of cerebellar and forebrain NSCs transduced with GFP (B–C) or N-myc^T58A (D–E) and treated with cyclopamine (cyc) or tomatidine (tom) in 5μM final concentrations for E16 and P0 ages, respectively. Error bars: +/− SD. * indicates p < 0.05 from Student’s t-test. See also Figure S3.

Figure 4. N-myc^T58A induces brain tumors from both cerebellum and forebrain

(A–C) Survival of mice orthotopically transplanted with 100,000 NSCs isolated from forebrain, cerebellum and brain stem/LRL, transduced with N-myc^WT or N-myc^T58A viruses. n=number of mice per group. Mice injected with N-myc^WT NSCs (green colors) or N-myc^T58A E16 and P0 brain stem/LRL (red and green, respectively) generated no tumors. (D–I) Histopathology of representative P0 cerebellar (D and E), P0 forebrain (F and G), and E14 brain stem/LRL (H–I) N-myc^T58A tumors (H&E staining). Cerebellar tumors displayed MB features, including cell wrapping in a tumor with prominent LC/A features (D, arrowhead) and Homer Wright rosettes (E, arrowhead). Forebrain tumors displayed diffuse invasion of tumor cells along white matter tracts (arrowhead) (G). LRL tumors displayed marked cellular and nuclear pleomorphism (I, arrowhead). Box in H (200x) denotes region shown at higher magnification in I (400x). (J–U) Immunostaining, of representative N-myc^T58A-induced cerebellar, forebrain, and LRL tumors demonstrating greater neuronal differentiation in cerebellar tumors, greater glial differentiation in forebrain tumors, and intermediate differentiation in LRL tumors. * indicate tumor region in pictures where an adjacent normal brain is present. Scale: 100μm. See also Figure S4 and Table S1.

Figure 5. SOX9 marks a SHH-dependent brain tumor profile

(A) Expression analysis of 103 primary MB samples clustered in four MB subgroups, WNT, SHH, Group 3 and Group 4 with levels normalized to normal adult cerebella (n=5) as described previously (Northcott et al., 2011). Dark blue represents the lowest and dark red the highest expression. (B–E) SOX9 immunostaining in representative human MB (B and C) and glioma (D and E) sections, without and with MYCN amplification, as determined by fluorescent in situ hybridization (Perry et al., 2009). Scale: 100μm. See also Figure S5 and Table S2.

Figure 6. N-myc^T58A tumors, GTML tumors, and normal murine tissues show discrete expression signatures

Dendrogram showing unsupervised hierarchical clustering from Affymetrix exon array data obtained from: (1) Primary and Tumor-Derived Cells (yellow): Four individual GFP-transduced E16 and P0 cerebellar (sorted or unsorted for GFP-infection using cell sorting) NSCs, P0 forebrain NSCs, two N-myc^T58A-transduced E16 and P0 cerebellar NSCs (in low passages) and three cell lines (in low passages) derived from orthotopic E16 and P0 cerebellar and P0 forebrain tumors. (2) Orthotopic Tumors (red): Nine freshly isolated orthotopic E16 and P0 cerebellar and P0 forebrain N-myc^T58A-tumors. (3) GTML Tumors (purple): 32 isolated GTML tumors and one cultured GTML tumor cell line (GTML2). (4) Normal Cerebellum (blue): Four samples from P7 GTML double heterozygous total cerebella. The distance in the clustering determines how well samples are related to each other. See also Figure S6.

Figure 7. N-MYC drives both SHH-dependent and SHH-independent MB

(A–C) Relative expression of Sox9, N-myc and Math1 in five disparate N-MYC-driven murine tumors, E16 MB (n=4), P0 MB (n=6), E16 glioma (n=4), P0 glioma (n=6) and E14 LRL tumor (n=3). n=number of individual tumors. Boxes are the first to third quartile with the median indicated. No boxes are shown for LRL tumors because of few samples (n=3). Whiskers go from min to max values. (D) Proliferation of cultures of E16 MB, P0 MB and P0 glioma isolated from individual tumors and treated with cyclopamine (cyc) or tomatidine (tom) used at final concentrations of 5μM. * indicates statistical significance (p < 0.05) from Student’s t-test. Error bars: +/− SD. See also Figure S7 and Table S3.

Figure 8. SOX9 promotes self-renewal and N-MYC-driven brain tumor incidence

(A) Immunoblot of NSCs generated from single cells of GFP- and SOX9-transduced P0 cerebellar NSCs from normal (six individual clones) and N-myc^T58A transduced NSCs (three individual clones). (B) Limiting dilution assay showing secondary (2°) spheres cultured starting from one single cell per well (with or without (wo) EGF and FGF). Numbers indicates mean values (from two experiments) of spheres per well from a total of 60 wells counted (shown as %). Clone numbers used (from A) are indicated. Assay results from single clones from a P0 MB tumor (P0C-T3 (Cl.1)) and from a GTML tumor (GTML2 (Cl.1)) are also included. * indicates statistical significance (p < 0.05) from Student’s t-test. (C–D) Relative expression of SOX9 and Gli2 in N-myc^T58A-transduced P0 cerebellar NSCs and N-myc^T58A-transduced P0 cerebellar NSCs further transduced with SOX9 (P0C-T58A-SOX9 Cl.3), respectively. For SOX9 expression and clone numbers, see A. * indicates statistical significance (p < 0.05) from Student’s t-test. (E) Immunoblot showing GLI2 protein expression in N-myc^T58A-transduced P0 cerebellar normal (P0C-T58A Cl.1) or SOX9-transduced (P0C-T58A-SOX9 Cl.3) clones. (F) Survival after orthotopic transplantation of 100,000 cells from N-myc^T58A P0 cerebellar NSCs (P0C-T58A, as used in Figure 4A; n=30) and N-myc^T58A P0 cerebellar NSCs further transduced with SOX9 (P0C-T58A-SOX9, n=10), respectively. Same clone (Cl.3) was used as in C–E. (G) Re-expression of SOX9 in representative GTML spheres (GTML3) treated with dox (1μg/ml) after 72h as shown in an immunoblot, with GTML1 cells as controls. * indicates statistical significance (p < 0.05) from Student’s t-test. Error bars: +/− SD. See also Figure S8.