Cell Reports Article

An Aberrant Network Essential for Wnt Signaling and Stem Cell Maintenance in Glioblastoma

Esther Rheinbay,1,2,5,6,9 Mario L. Suva` ,1,2,5,9 Shawn M. Gillespie,1,2,5 Hiroaki Wakimoto,3 Anoop P. Patel,3 Mohammad Shahid,8 Ozgur Oksuz,2 Samuel D. Rabkin,3 Robert L. Martuza,3 Miguel N. Rivera,2,5 David N. Louis,2 Simon Kasif,6,7 Andrew S. Chi,1,2,4,5,* and Bradley E. Bernstein1,2,5,* 1Howard Hughes Medical Institute, Chevy Chase, MD 20815, USA 2Department of Pathology and Center for Cancer Research 3Department of Neurosurgery 4Divisions of Neuro-Oncology and Hematology/Oncology and Department of Neurology Massachusetts General Hospital and Harvard Medical School, Boston, MA 02114, USA 5Broad Institute of Harvard and MIT, Cambridge, MA 02142, USA 6Bioinformatics Program 7Department of Biomedical Engineering Boston University, Boston, MA 02215, USA 8Department of Surgery, Massachusetts General Hospital, Boston, MA 02114, USA 9These authors contributed equally to this work *Correspondence: [email protected] (A.S.C.), [email protected] (B.E.B.) http://dx.doi.org/10.1016/j.celrep.2013.04.021

SUMMARY cer cells maintain distinctive transcriptional programs that reflect their lineage, differentiation stage, and cellular environment. Glioblastoma (GBM) is thought to be driven by a sub- These transcriptional programs are driven by transcription population of cancer stem cells (CSCs) that self- factors (TFs) that interact with regulatory sequences and by renew and recapitulate tumor heterogeneity yet that modulate the chromatin state of specific loci. Tran- remain poorly understood. Here, we present a scriptional and epigenetic programs can exhibit striking hetero- comparative analysis of chromatin state in GBM geneity within a tumor and may distinguish cancer stem cells CSCs that reveals widespread activation of (CSCs) or other subpopulations of clinical significance. However, these malignant programs and their variability remain poorly normally held in check by Polycomb . understood. These activated targets include a large set of de- Chromatin state maps provide a general and systematic velopmental transcription factors (TFs) whose means for gauging the activity and epigenetic state of pro- coordinated activation is unique to the CSCs. We moters, genes, and other regulatory elements within a particular demonstrate that a critical factor in the set, ASCL1, cell type (Bell et al., 2011; Zhou et al., 2011). These maps are ac- activates Wnt signaling by repressing the negative quired by coupling chromatin immunoprecipitation with deep regulator DKK1. We show that ASCL1 is essential sequencing (ChIP-seq) and are typically applied to histone mod- for the maintenance and in vivo tumorigenicity of ifications that mark different types of functional sequence ele- GBM CSCs. Genome-wide binding profiles for ments. Chromatin data can be integrated with TF recognition ASCL1 and the Wnt effector LEF-1 provide mecha- motifs or binding profiles to gain more specific insight into the nistic insight and suggest widespread interactions regulators and pathways that activate these sequence elements in specific cellular contexts (Dunham et al., 2012; Ernst et al., between the TF module and the signaling pathway. 2011). Although these approaches have been applied to cancer Our findings demonstrate regulatory connections models to a limited extent (Baylin and Jones, 2011), their poten- among ASCL1, Wnt signaling, and collaborating tial has yet to be explored systematically. TFs that are essential for the maintenance and Glioblastoma (GBM) is the most common malignant brain tu- tumorigenicity of GBM CSCs. mor in adults and is associated with poor prognosis despite aggressive treatment. Transcriptional profiling studies have re- INTRODUCTION vealed biologically relevant GBM subtypes associated with sur- vival and response to therapy as well as specific dysregulated The importance of epigenetic regulation in cancer initiation and cellular pathways (Huse and Holland, 2010). Furthermore, epige- progression is now well recognized (Baylin and Jones, 2011; Pu- netic regulators, including the Polycomb repressors EZH2 and jadas and Feinberg, 2012). Aberrant DNA methylation patterns BMI1, and specific DNA methylation changes have been linked and recurrent mutations in genes encoding chromatin-modifying to disease pathology, prognosis, and therapeutic responses enzymes have been documented in a wide range of tumors. Can- (Suva` et al., 2009; Bruggeman et al., 2007; Lee et al., 2008).

Cell Reports 3, 1567–1579, May 30, 2013 ª2013 The Authors 1567 AB D 5 5 10 10 4 4 10 10 3 3 10 PE-A 10 GBM CSCs Astrocytes (NHA) FITC-A 2 2 010 010

0102 103 104 105 0102 103 104 105

SSEA1-FITC CD133-PE Histone modification ChIP-Seq

C DNA methylation array Epigenomic profiles

2 mm 100 µm 100 µm

E F

H3K4me3 H3K4me3 H3K27me3 Neither mark +H3K27me3

0.5 MGG4 NHA 12284 MGG6 promoter state promoters MGG8 1784 1244 5110 promoters promoters promoters

3% 0.2 0.3 0.4 94% 27% 8% 1% 32% 12% 9% CSC majority 5% 11% 3% promoter state 59% 48% 88% probes in indicated tumor Fraction of hypermethylated 0.0 0.1 H3K4me3 H3K4me3+ H3K27me3 Neither H3K27me3 mark H3K4me3 H3K4me3+ H3K27me3 Neither mark H3K27me3 NHA promoter state

G

10 5 kb 5 kb 10 10 kb 10 kb 10 10 kb 0 0 0 3 3 3 0 0 0 2 2 2 0 0 0 2 2 2

MGG8 CSC 0 MGG6 CSC 0 MGG4 CSC 0

10 10 10 0 0 0 3 3 3 0 0 0 2 2 2 0 0 0 2 2 2 0 0 0

10 10 10 0 0 0 2 2 2 0 0 0

NSC2 MGG8 serum NSC2 MGG6 serum NSC2 MGG4 serum 0 0 0 10 10 10 0 0 0 3 3 3 0 0 0 2 2 2

NHA 0 NHA 0 NHA 0 2 2 2 0 0 0 ASCL1 EN2 OLIG2 OLIG1 LHX2 HEY2

H3K4me3 H3K4me1 H3K27me3 H3K36me3

Figure 1. Characterization of the GBM CSC Chromatin Landscape (A) GBM CSCs used for this study grow as gliomaspheres in serum-free neurobasal media. (B) FACS analysis of MGG8 GBM CSCs shows positivity for the GBM stem cell markers SSEA-1 and CD133. (C) Mouse brain cross-section after orthotopic xenotransplantation of MGG8 GBM CSCs (left). Higher magnification of tumor tissue depicts cytonuclear pleo- morphism, mitotic and apoptotic figures (center), and infiltration along white matter tracks (right). (D) Schematic overview of the study strategy. (E) Breakdown of TSS chromatin state in NHA (H3K4me3 only, H3K4me3 + H3K27me3, H3K27me3 only, neither mark) and their consensus chromatin state in the GBM CSCs (see Experimental Procedures). A large fraction (59%) of genes bivalent in NHA become activated (H3K4me3 only, green) in GBM CSCs. (legend continued on next page)

1568 Cell Reports 3, 1567–1579, May 30, 2013 ª2013 The Authors Recent work has also shown that epigenetic resetting by brillary acidic (GFAP) expression, respectively (Figures induced pluripotent stem cell reprogramming followed by line- S1B). Orthotopic xenotransplantation of a limited number of age differentiation can override the malignant properties of GBM CSCs leads to formation of tumors that recapitulate GBM (Stricker et al., 2013). However, little is known about the GBM morphology with diffuse infiltration of the brain paren- control mechanisms that drive these transcriptional programs chyma (Figure 1C). and their contribution to the malignant phenotype in GBM. We systematically examined the transcriptional and epige- Recent studies have documented subpopulations of GBM netic landscapes of GBM CSCs by profiling histone H3 lysine 4 cells with tumor-propagating capacity that are believed to trimethylation (H3K4me3; a marker of transcriptional initiation), constitute the tumor’s driving force and to play a major role in tu- lysine 36 trimethylation (H3K36me3; transcriptional elongation), mor recurrence and resistance to therapy (Bao et al., 2006; Chen and H3 lysine 4 monomethylation (H3K4me1; candidate en- et al., 2012; Singh et al., 2004). This subpopulation can be pro- hancers) (Figure 1D; Table S1)(Zhou et al., 2011). We also map- spectively isolated ex vivo with specific cell-surface markers ped H3 lysine 27 trimethylation (H3K27me3), a repressive mark (Singh et al., 2003; Son et al., 2009) or defined culture conditions catalyzed by the Polycomb protein EZH2. For comparison, we (Lee et al., 2006; Lottaz et al., 2010; Wakimoto et al., 2009) and also profiled normal human astrocytes (NHAs) isolated from fetal expanded in vitro as gliomaspheres. Upon serial xenotransplan- brain, embryonic stem cell (ESC)-derived neural stem cells tation, these cells initiate tumors that closely phenocopy the pa- (NSCs; Figures S1C–S1F) (Conti et al., 2005), and previously tient’s parental tumor both morphologically and genetically (Galli characterized serum-cultured GBM cell lines derived from the et al., 2004; Singh et al., 2004; Wakimoto et al., 2009). Because same patient tumors as our GBM CSCs (Wakimoto et al., of their unlimited self-renewal capacity and their ability to seed 2009). These traditional GBM lines grow as adherent mono- malignant tumors in vivo, this subpopulation satisfies major layers, do not express GBM CSC-surface markers, and fail to criteria for cancer stem cells (CSCs) (Valent et al., 2012). initiate tumors upon orthotopic xenotransplantation in limiting Here, we combined chromatin profiling, computational anal- dilution assays (Figures S1G–S1J). ysis, and directed cellular perturbations to characterize tran- We initially focused our attention on the four GBM CSC lines scriptional and epigenetic regulatory programs in GBM CSCs. using the NHAs as normal comparators. For each cell type, we Comparative analysis of chromatin maps for GBM CSCs, differ- classified over 20,000 promoters as ‘‘active’’ (only entiated GBM cells, and nonmalignant neural cells reveals a H3K4me3 detected) or ‘‘repressed’’ (H3K27me3 detected, with module of developmental TFs that is coordinately activated in or without H3K4me3, or neither mark) (Figure S2A; Table S2). the CSCs. Functional analysis suggests that these TFs play H3K27me3-marked promoters are maintained in an inactive essential roles in GBM CSCs. In particular, we show that state by Polycomb complexes, while promoters with neither ASCL1 directly activates Wnt signaling and is essential for mark are frequently repressed through DNA hypermethylation GBM CSC maintenance and in vivo tumorigenicity. Genome- (Meissner et al., 2008). We compared assignments between wide maps for ASCL1 and the Wnt effector lymphoid enhancer cell types in order to identify genes that are differentially regu- binding factor (LEF-1) suggest specific mechanisms and wide- lated in a majority of the GBM CSC lines relative to NHAs. Pro- spread interplay between the TF module and the signaling moters that are ‘‘active’’ in NHAs retain this state in GBM pathway. Our findings thus provide global and mechanistic CSCs in a vast majority of cases (n = 11,586; 94% of NHA insight into the regulatory programs that drive a CSC-like popu- H3K4me3 genes) (Figure 1E), consistent with many of these be- lation critical for GBM pathogenesis. ing housekeeping genes (Mikkelsen et al., 2007). In contrast, promoters that are ‘‘repressed’’ in NHAs frequently change their RESULTS chromatin state in GBM CSCs in the following manner. Repressed promoters with H3K27me3 and H3K4me3 (‘‘biva- Comparative Epigenomic Analysis of GBM CSCs and lent’’) in NHAs predominantly lose H3K27me3 and become acti- Human Astrocytes vated in GBM CSCs (n = 1,057; 59% of NHA bivalent promoters) We focused our study on four GBM CSC lines derived from (Figure 1E). Repressed promoters with only H3K27me3 in NHAs different human tumors that were defined functionally through often lose H3K27me3 and switch to the unmarked state in GBM their ability to initiate tumors in a xenotransplantation model (Wa- CSCs (n = 591; 48% of NHA H3K27me3 genes). Repressed pro- kimoto et al., 2009, 2012). These lines grow as gliomaspheres moters that are unmarked in NHAs tend to retain this state in (Figure 1A; see Experimental Procedures) and express the GBM CSCs (n = 5,110; 88%) (Figure 1E). Notably, an analogous CSC cell surface markers CD133 and stage-specific embryonic comparison of NSCs to GBM CSCs also revealed frequent antigen 1 (SSEA-1) (Figure 1B) and the neural progenitor interme- switching of loci from bivalent (or H3K27me3 only) in the nonma- diate filament Nestin (Figure S1A). GBM CSCs show differentia- lignant cells to active/H3K4me3 in the CSCs (61%; Figure S2B). tion potential toward the neuronal and astrocytic lineages, as In contrast, only 33% of genes with bivalent chromatin state in shown by microtubule-associated protein 2 (MAP-2) and glial fi- NSCs switched to active/H3K4me3 in the nonmalignant NHAs

(F) Fraction of DNA hypermethylated (b R 0.75) probes in MGG4, MGG6, and MGG8 GBM CSCs contingent on NHA chromatin state of the probe. Probes marked with H3K27me3 only are twice as likely to become DNA methylated than those marked with both H3K4me3 and H3K27me3. (G) Chromatin state of six CSC-TFs in one representative GBM CSC line, matched serum-grown GBM cell line, NSC, and NHA. All TFs are active (H3K4me3 at promoter, H3K36me3 over the transcript) in GBM CSCs, but not in NHA or serum-grown GBM cells, as indicated by large domains of H3K27me3. See also Figures S1 and S2.

Cell Reports 3, 1567–1579, May 30, 2013 ª2013 The Authors 1569 A Figure 2. Aberrant Activation and Repres- differentiation sion of TF Polycomb Targets in GBM CSCs sequence-specific DNA binding (A) Representative top-scoring functional terms cell morphogenesis enriched in genes active (H3K4me3 only) in GBM cell projection organization CSCs but repressed in NHA (see also Table S3). transcription factor activity cell morphogenesis involved in differentiation Scores are calculated based on Benjamini-Hoch- 0 5 10 15 20 25 berg corrected p values (see Experimental Pro- -log p-value 10 cedures). B C (B) H3K4me3 (green) and H3K27me3 signal (red) at aberrantly activated TF loci (À2.5kb to +2.5 kb from TSS) for indicated cell types. Orange in- NHA NSC MGG23 CSCMGG4 CSCMGG6 CSCMGG8 CSCMGG4 serumMGG6 serumMGG8 serum NHA NSC MGG23MGG4 CSCMGG6 CSCMGG8 CSC CSC dicates overlap of H3K4me3 and H3K27me3

OSR2 OSR2 signal (bivalent). Genes were clustered based on HOXA13 HOXA13 SIM2 SIM2 ONECUT2 ONECUT2 H3K27me3 signal. HMX3 HMX3 DMRTA2 DMRTA2 (C) Microarray data for acti- OSR1 OSR1 DLX1 DLX1 SIX2 SIX2 vated TFs confirms chromatin state data HOXC10 HOXC10 HOXD1 HOXD1 ESRRG ESRRG (GSE46016). Red indicates high and blue FOXD2 FOXD2 TP73 TP73 indicates low expression normalized by row. The

Cancer TFs HOXC4 HOXC4 HOXC9 HOXC9 PITX1 PITX1 expression changes are consistent with the ISL2 ISL2 SHOX2 SHOX2 TFAP2A TFAP2A chromatin changes, although the magnitude of OTX1 OTX1 SIX1 SIX1 expression change across samples is more HOXC5 HOXC5 HOXC6 HOXC6 EN2 EN2 variable. HEY2 HEY2 CSC TFs LHX2 LHX2 (D) Chromatin state and gene expression data for OLIG2 OLIG2 ASCL1 ASCL1 OLIG1 OLIG1 NHA-active, GBM-CSC-repressed TF loci. Color SALL2 SALL2 RFX4 RFX4 CITED1 CITED1 scheme as in (B) and (C). POU3F2 POU3F2 NR5A2 NR5A2 See also Figure S3. FOXC1 FOXC1 PROX1 VAX2 PROX1 MEIS3 VAX2 SOX21 MEIS3 BACH2 SOX21 EBF4 BACH2 FOXJ1 EBF4 HES6 FOXJ1 HES6

Stem-cell TFs KLF15 SOX2 MYCL1 KLF15 POU3F3 MYCL1 SOX5 POU3F3 SOX8 SOX5 or no change in their expression, suggest- TBX2 SOX8 MSX1 TBX2 MSX1 ing that transcriptional changes are NR4A2 KLF4 NFATC1 NR4A2 MITF NFATC1 unlikely to underlie this epigenetic NPAS2 MITF NKX2-2 NPAS2 DLX2 NKX2−2 switch (Figure S2E). We therefore con- NR6A1 DLX2 GRHL1 NR6A1 EGR3 GRHL1 sidered the possibility that these loci MYB EGR3 HEY1 MYB SALL3 HEY1 may become DNA methylated in the ZNF467 SALL3 ZNF704 ZNF467 SOX1 ZNF704 CSCs, as would be consistent with prior EGR2 SOX1 ZIC4 EGR2 ZIC4_2 ZIC4 ZNF219 ZNF219 reports of Polycomb targets becoming ZEB2 ZEB2 SCML2 SCML2 LASS4 LASS4 hypermethylated in cancer (Ohm et al.,

NHA-repressed TFs NHA-repressed TUB TUB ZNF395 ZNF395 SETBP1 SETBP1 2007; Schlesinger et al., 2007; Widsch- ARNT2 ARNT2 E2F2 DBP DBP wendter et al., 2007). To test this, PBX4 PBX4 BATF2 BATF2 MEIS1 MEIS1 we profiled DNA methylation in three IRX5 IRX5 NFIB NFIB HES4 HES4 GBM CSC lines (see Experimental Proce- MXI1 MXI1 D NR4A3 NR4A3 dures). We found that loci marked by POU2F2 POU2F2 ZBED2 ZBED2 H3K27me3 only in the NHAs become MEIS3P1 MEIS3P1 SOX30 SOX30 ZNF540 hypermethylated in the GBM CSCs at a ZNF154 ZNF540 RARA ZNF154 SMAD3 RARA rate that is 2-fold higher than bivalent ELF1 SMAD3 ZNF438 ELF1 MAFG ZNF438 ZNF668 MAFG loci (Figure 1F). Taken together, these BNC2 ZNF668 GLIS1 BNC2 BNC1 GLIS1 findings suggest that repressive Poly- FOXL1 FOXL1 CSC-repressed TFs FOXC2 FOXC2 comb complexes are lost from a subset -2.5 kb TSS 2.5kb of their target loci in GBM CSCs, with initially bivalent genes undergoing tran- (Figure S2C). These statistics suggest that genes marked by scriptional activation and H3K27me3-only genes acquiring H3K27me3 in nonmalignant contexts frequently lose the Poly- DNA hypermethylation. comb-associated mark and switch their chromatin state in GBM CSCs. Widespread TF Activation in GBM CSCs We next examined how the chromatin state transitions To gain insight into the regulatory consequences of the chro- correspond to changes in gene expression. As expected, genes matin state transitions, we examined the identities of genes that switch from bivalent in NHA to H3K4me3-only in GBM CSC with active chromatin in GBM CSCs but repressed chromatin are expressed at significantly higher levels in the malignant cell in NHAs. Unbiased functional enrichment analysis (Dennis lines (Figure S2D). In contrast, genes that switch from et al., 2003) revealed a significant overrepresentation of terms H3K27me3-only in NHAs to unmarked in GBM CSCs show little related to development and transcriptional regulation (Figure 2A;

1570 Cell Reports 3, 1567–1579, May 30, 2013 ª2013 The Authors Table S3). In contrast, we failed to detect significant enrichment together, these results establish ASCL1 as a regulator of Wnt for genes that were repressed in GBM CSCs but active in NHAs signaling in GBM CSCs. (Table S3). We therefore focused our attention on the set of To assess the functional role of ASCL1 in CSC maintenance, developmental TFs (n = 90) that are differentially activated in we performed single-cell sphere-formation assays of GBM GBM CSCs (Table S4). These include factors previously associ- CSCs after shRNA-mediated ASCL1 knockdown (Clement ated with GBM such as SOX2 (Gangemi et al., 2009), OLIG2 et al., 2007; Galli et al., 2004). Only 2% of ASCL1-depleted (Ligon et al., 2004), HEY1 (Hulleman et al., 2009), and several CSCs retained the capacity to reform gliomaspheres after HOX genes (Murat et al., 2008). However, most of the TFs we de- 14 days compared to nearly 20% of control cells (Figures 3F tected have not yet been described in this context. Although the and 3G). In addition, spheres generated by ASCL1 knockdown CSC lines used in this study were derived from different patients cells were significantly smaller (p < 10À14; Figure 3H). Mice ortho- and harbor different genetic aberrations (Wakimoto et al., 2012), topically injected with ASCL1-depleted GBM CSCs showed a majority of the implicated TFs is active in all four CSC lines (Fig- prolonged survival compared to those xenotransplanted with ure 2B). Conventional messenger RNA (mRNA) microarray pro- control CSCs (p < 0.01; Figure 3I). Thus, ASCL1 is essential for files confirm that these TFs are expressed at significantly higher maintenance and propagation of GBM CSCs. levels in the four CSCs than NHAs (Figure 2C). Consistent with We next examined ASCL1 expression in primary GBMs using their role as developmental regulators, most of these TFs are published gene expression profiles (Verhaak et al., 2010; Sun marked with bivalent chromatin or H3K27me3 in ESCs and et al., 2006). ASCL1 is expressed at markedly higher levels in tu- NSCs, as well as in primary nonmalignant brain sections (Fig- mors relative to normal brain. However, its expression varies ure S3)(Zhu et al., 2013). In contrast, we identified few TFs significantly between molecular subtypes (Kruskal-Wallis p < (n = 16) that are selectively repressed in the CSCs (Figure 2D). 10À16), with significantly higher levels in proneural tumors (Fig- This suggests that a common set of TFs loses Polycomb repres- ure 3K). ASCL1 also correlates with tumor grade, with increasing sion and is induced in GBM CSCs. expression in grade II and III astrocytomas and grade IV GBMs of To better understand their specificities, we also studied the proneural subtype (Figure 3L). The functional link between chromatin states of the activated TF genes in representative ASCL1 and AXIN2 is further supported by significantly correlated cell models, including NHA, ESC-derived NSCs, the CSCs, and expression across the GBM tumors (Pearson’s r = 0.52; p < serum-grown GBM cell lines. Clustering the TFs on the basis 10À14; Figure 3J). Furthermore, flow cytometric staining in of their promoter H3K27me3 signals distinguished four subsets: acutely dissociated human GBM shows that ASCL1 is ex- (1) ‘‘cancer TFs’’ active in GBM CSCs and traditional GBM cell pressed in a subpopulation of cells with high levels of SOX2, a lines, (2) ‘‘CSC TFs’’ exclusively active in GBM CSCs, (3) marker for GBM subpopulations with CSCs properties (Fig- ‘‘stem-cell TFs’’ active in CSCs and NSCs, and (4) NHA- ure 3M; Figure S4). This result is supported by in situ hybridiza- repressed TFs active in all the other cell types (Figure 2B). tion (ISH) in human GBM samples where ASCL1 expression is restricted to a subset of SOX2-positive cells (Figure 3N). ASCL1 Induces Wnt Signaling in GBM CSCs The distinctive chromatin patterns of the CSC TFs prompted us ASCL1 Promotes Wnt Signaling Directly by Repressing to further explore their functional significance (Figures 1G and DKK1 2B). We specifically examined whether any of these TFs affect To identify the direct regulatory targets of ASCL1, we used ChIP- signaling pathways that are essential for GBM CSC prolifera- seq to map ASCL1 in MGG8 CSCs (see Experimental Proce- tion, such as Notch, Hedgehog, and Wnt (Clement et al., dures; Figure S5). The mapped ASCL1 binding sites are highly 2007; Shih and Holland, 2006; Zheng et al., 2010). We ectopi- enriched for its cognate sequence motif and include the known cally expressed each CSC TF in NHAs and measured the target DLL1 (Figure S5). We found that ASCL1 binds to candidate expression of canonical target genes for each of these path- enhancers (marked by H3K4me1) in the vicinity of several genes ways (Figure 3A). involved in Wnt regulation, including FZD5, DKK1, TCF7, and We found that ASCL1 (also known as MASH1) induces AXIN2, TCF7L1 (Figure 4A). To test whether these candidates represent a canonical Wnt target (Figure 3A). ASCL1 mRNA is highly ex- direct functional targets, we ectopically expressed ASCL1 in pressed in all four CSC cell lines but not in NHAs or NSCs (Fig- NHAs and measured changes in their expression. The effect of ure 3B). To confirm the functional connection between ASCL1 ASCL1 on Wnt pathway genes was moderate with the exception and AXIN2, we depleted ASCL1 in GBM CSCs by small hairpin of 10-fold downregulation of DKK1 (Figure 4B). This is consistent RNA (shRNA)-mediated knockdown (Figure 3C), whereupon with a previous report of ASCL1 being a negative regulator of we observed significantly reduced AXIN2 expression (Figure 3D). DKK1 in small cell cell lines (Osada et al., 2008). In We also used a Wnt-responsive luciferase reporter system to GBM CSCs, ASCL1 binds to an H3K4me1-marked regulatory test whether ASCL1 control of AXIN2 is associated with a general element located 5.7 kb upstream of the DKK1 transcription start induction of Wnt signaling and TCF/LEF transcriptional regula- site (TSS) (Figure 4C), which adopts a chromatin configuration tion (Firestein et al., 2008; Veeman et al., 2003). Indeed, ASCL1 characteristic of an inactive or ‘‘poised’’ enhancer. In NHA, which increased reporter gene expression >10-fold compared to a do not express ASCL1, this element is also enriched for control vector with mutated TCF/LEF sites (Figure 3E). ASCL1- H3K27ac and thus assumes an ‘‘active enhancer’’ state, consis- mediated induction of the reporter was dramatically enhanced tent with the high expression of DKK1 in these cells (Figure 4D). by costimulation with Wnt3a, suggesting a synergistic mode of To test whether repression of DKK1 is the primary mechanism action with autocrine or paracrine Wnt stimulation. Taken by which ASCL1 modulates Wnt signaling, we simultaneously

Cell Reports 3, 1567–1579, May 30, 2013 ª2013 The Authors 1571 A B 12 ASCL1 EN2 6 10 HEY2 5 8 LHX2 OLIG1 4 6 OLIG2 3

4 2 ASCL1 expression

2 10 1 Fold expression difference Log after expression of indicated TF 0 * **** 0 AXIN2 LEF1 TCF7 CCND1 HEY1 HEY2 HES1 GLI1 GLI2 GLI3 NSC NHA MGG4 MGG6 MGG8 Wnt targets Notch targets Shh targets MGG23

C DE I 1.2 1.2 70 ASCL1- expression * 1.0 1 1 Wnt3a 60 shASCL1 (n=4) vector Control (n=4) 0.8 0.8 50 40 0.6 0.6 30 0.4 0.4 ** 20 Fraction alive TOP/FOP ratio TOP/FOP 0.2 0.2 Measure 10

Relative AXIN2 expression Relative reporter signal p<0.01 0.0 Relative ASCL1 expression Relative 0 0 0 0 20406080100120140 Days Control Control Control Wnt3a ASCL1 ASCL1 shASCL1 shASCL1 +Wnt3a

FGH J 25 p<10-14 7.5 Control shASCL1 20

15

10

5 Sphere diameter [ µ m] % Sphere formation

20 40 60 80 100 r=-0.52 300x 300x TCGA AXIN2 expression p<10-14 0 5 4 TCGA ASCL1 expression 11

Control Control shASCL1 shASCL1 K L M N p<4 × 10-17 p<8 × 10-6 p<2.5 × 10-7 TCGA AXIN2 expression TCGA ASCL1 expression 5.5 6.0 6.5 7.0 05 Sun et al. ASCL1 expression ASCL1 5678910 Non- IIIII IV (GBM) Proneural SOX2 600x neoplastic GBM only Neural Neural brain Astrocytoma ClassicalProneural ClassicalProneural tumor grade Mesenchymal Mesenchymal

Figure 3. ASCL1 Is an Upstream Regulator of the Wnt Pathway (A) Relative expression of Wnt, Notch, and Shh targets after ectopic lentiviral expression of indicated TF in NHA measured by quantitative RT-PCR (qRT-PCR). *Not detectable. (B) mRNA levels of ASCL1 in NSCs, NHAs, and GBM CSCs measured on Affymetrix microarray (GSE46016). (C) Relative levels of ASCL1 after shRNA-mediated knockdown in MGG4 CSCs by qRT-PCR. (D) Repression of AXIN2 upon knockdown of ASCL1 in MGG4 CSCs (t test, p < 0.005). (E) Schematic of Wnt activation experiment and relative luciferase expression for a TCF/LEF-responsive promoter (TOPFLASH-Firefly) relative to a scrambled response element (FOPFLASH-Renilla) in 293T cells after lentiviral transfection with ASCL1 and addition of Wnt3a. *p < 0.05; **p < 0.01 (one-tailed t test). (F and G) Individual examples (F) and quantification (G) of MGG4 CSC sphere-forming capacity in control and ASCL1-depleted cells. (H) Quantification of sphere diameter in control and ASCL1-depleted cells. (I) Kaplan-Meier survival curve for mice injected with 5,000 control (blue line) or ASCL1-depleted MGG4 CSCs (log-rank p value < 0.01). (legend continued on next page)

1572 Cell Reports 3, 1567–1579, May 30, 2013 ª2013 The Authors transfected DKK1 and ASCL1 into NHA and measured Wnt acti- likely to occur in normal physiologic contexts. Deregulation of vation using the Wnt reporter. As shown in Figure 4E, exogenous this network is associated with loss of H3K27me3 at TF pro- DKK1 expression completely abrogates the inducing effect of moters and may reflect ineffective Polycomb repression in the ASCL1. These data support a model in which ASCL1 activates CSCs. We speculate that diminished epigenetic silencing and Wnt signaling by repressing a regulatory element upstream of promiscuous TF activation might confer a competitive advantage the negative Wnt regulator DKK1. The association is also sup- by allowing GBM CSCs to respond to the varied requirements of ported in clinical contexts by a significant anticorrelation in their malignant state, altered genetic makeup, and environment. ASCL1 and DKK1 expression across primary GBM tumors Among the CSC TFs, ASCL1 emerged as a potent upstream (Pearson’s r = À0.6; p < 10À20; Figure 4F). Given the depen- regulator of the Wnt signaling pathway, which was recently dency of GBM CSCs on Wnt signaling (Zheng et al., 2010; Zhang shown to be critical for GBM CSC maintenance (Zheng et al., et al., 2011), this regulatory function may explain the critical role 2010; Zhang et al., 2011). ASCL1 overexpression induces Wnt of ASCL1 in GBM CSC maintenance. signaling in normal astrocytes, while its knockdown in GBM CSCs markedly reduces activity of the pathway. We show that LEF-1 Mediates Reciprocal Interactions between Wnt Wnt activation is mediated through DKK1, a secreted Wnt inhib- Signaling and CSC TFs itor that is directly repressed by ASCL1. We also find evidence To gain further insight into the Wnt signaling pathway in GBM that Wnt signaling feeds back upon the CSC TF genes via mul- CSCs, we mapped LEF-1, a high-mobility group TF that regu- tiple LEF-1 target sites in their vicinity. The clinical relevance of lates Wnt targets (see Experimental Procedures; Figure S6). these interactions is supported by increased ASCL1 expression We identified over 3,000 LEF-1 binding sites, which are highly in primary astrocytomas and GBMs and a correlation between enriched for the LEF/TCF motif and include the known Wnt target ASCL1 and the Wnt target gene AXIN2 across tumor samples. CCND1 (Clevers, 2006). The vast majority of these sites (82%) In conclusion, we describe an aberrant epigenetic landscape reside outside of promoter regions, consistent with the original in GBM CSCs and the induction of a nonphysiologic TF module identification of LEF-1 at an enhancer (Travis et al., 1991). that is linked to Wnt signaling and essential for CSC maintenance Notably, nearly half of the CSC and stem cell TF genes are prox- and tumorigenicity. Our findings thus shed light on the regulatory imal to H3K4me1-marked candidate regulatory elements bound circuitry of this CSC model and propose specific factors and in- by LEF-1 (Figure 5A; Table S5). To test whether these TFs are teracting pathways as candidates for translational study. downstream of the Wnt pathway, we measured their expression in NHAs stimulated with ASCL1 complementary DNA (cDNA) EXPERIMENTAL PROCEDURES and Wnt3a protein. We detected significant induction of six TFs, indicating their responsiveness to Wnt signaling (Figure 5B). Cell Culture Taken together, these results suggest that crosstalk between a Surgical specimens of GBM tumors were collected at Massachusetts General Hospital with approval by the institutional review board (IRB). Mechanically network of activated TFs and Wnt signaling is critical for main- minced tissue was triturated and then cells were grown as gliomaspheres in taining GBM CSC regulatory programs (Figure 5C). serum-free neural stem cell medium (Neurobasal medium [Invitrogen] sup- plemented with 3 mmol/l L-glutamine [Mediatech], 13 B27 supplement DISCUSSION [Invitrogen], 0.53 N2 supplement [Invitrogen], 2 mg/ml heparin [Sigma-Aldrich], 20 ng/ml recombinant human EGF [R&D Systems], 20 ng/ml recombinant hu- 3 Cancer genome sequencing and complementary mechanistic man FGF2 [R&D Systems], and 0.5 penicillin G/streptomycin sulfate), as pre- viously described (Wakimoto et al., 2009). Genomic copy number alterations studies have accelerated our understanding of cancer genetics. and tumor xenograft histopathology for these CSCs have been described else- However, technical issues and the heterogeneity typical of where (Wakimoto et al., 2012). From the same tumors, traditional GBM cells many tumors have limited comprehension of the epigenetic aber- lines, grown as adherent monolayer in Dulbecco’s modified Eagle’s medium rations that contribute to tumor pathology. Here, we combined (DMEM) and 10% fetal calf serum (FCS) were derived as previously described recently established epigenomic technologies with an in vitro (Wakimoto et al., 2009). GBM CSC differentiation was induced using 5% FCS CSC model with tumor-initiating capacity in order to characterize and withdrawal of growth factors for 7 days on poly-L-ornithine- and laminin- coated plates (see below for details). Staining was performed for nestin (Santa the epigenetic and transcriptional programs that drive malignant Cruz, 1: 400), MAP-2 (Chemicon, 1: 150), and GFAP (Sigma, 1: 400). brain tumors. By comparing these GBM CSCs to serum-grown Human-ESC-derived NSCs generated from H9 ES cells were obtained GBM lines and nonmalignant neural cell models, we identify a from Millipore and grown and passaged in NSC medium consisting of a 1:1 large network of TFs activated in the CSCs in combinations un- mix of DMEM/F12:Neurobasal (Invitrogen), 0.53 N2 (Invitrogen), 0.53 B27

(J) Scatter plot shows correlation between AXIN2 and ASCL1 expression across 200 primary GBM samples (Verhaak et al., 2010). Each point denotes a single tumor sample. (K) TCGA ASCL1 (left) and AXIN2 (right) expression (Verhaak et al., 2010) correlate across molecular subtypes. Distributions for subtypes are significantly different from each other (Kruskal-Wallis p < 4 3 10À17 [ASCL1] and p < 8 3 10À6 [AXIN2]). (L) ASCL1 expression (Sun et al., 2006) is increased in GBM, most strongly in the proneural subtype and lower-grade astrocytomas, relative to nonneoplastic brain, suggesting that ASCL1 induction may be an early event in gliomagenesis. (M) Intracellular staining and flow cytometric detection of the nuclear stem cell marker SOX2 and ASCL1 in primary GBM. The population of ASCL1+ cells (23.8%) is entirely contained within the SOX2+ compartment. Notably, the ASCL1+ subpopulation also displays highest levels of SOX2 expression. (N) RNA-ISH for SOX2 (blue dots) and ASCL1 (red dots) in primary GBM shows expression of ASCL1 in a restricted subset of SOX2+ cells. Error bars in qRT-PCR experiments indicate SEM. See also Figure S4.

Cell Reports 3, 1567–1579, May 30, 2013 ª2013 The Authors 1573 A 5 kb 5 kb MGG8 CSC NR_038277 DKK1 LRP5

5 kb 10 kb

MGG8 CSC FZD5 TCF7L1 H3K4me3 H3K4me1 ASCL1

BC10 2.5 kb

1 MGG8 CSC

0.01 Expression fold change

0.001 NHA H3K4me3 H3K4me1 H3K27ac TLE1 H3K27me3 H3K36me3 ASCL1 DKK1 FZD5 LGR5 LPR5 TCF7 TCF7L1 AXIN2 DKK1 NR_038277

D E F 12 600 20

15

10

expression 5 TOP/FOP ratio TOP/FOP DKK1 microarrary

0 TCGA DKK1 expression ρ=-0.6 0 p<10-20 NSC NHA Wnt3a Wnt3a Wnt3a 2 MGG4MGG6 MGG8 4.5 10.5 MGG23 +ASCL1 +ASCL1 TCGA ASCL1 expression +DKK1

Figure 4. ASCL1 Regulates Wnt Signaling through DKK1 (A) ChIP-seq of ASCL1 in MGG8 CSCs (pink track) reveals enrichment at H3K4me1-marked distal elements in several Wnt pathway gene loci (gray shading). (B) Relative mRNA level change for ASCL1-bound Wnt pathway genes in NHA after ectopic expression of ASCL1. (C) ChIP-seq maps depict the chromatin environment of the DKK1 gene locus and the ASCL1-bound enhancer (gray shading) in MGG8 CSCs and NHA. In the absence of ASCL1, the element is activated (as indicated by H3K27 acetylation in NHA) and DKK1 is expressed (increase in active marks H3K4me3, H3K4me1, H3K27ac, and H3K36me3). (D) Expression levels of DKK1 in NSCs, NHAs, and GBM CSCs measured by microarray. (E) Expression changes of a TCF/LEF-responsive reporter (TOPFLASH-Firefly) relative to scrambled response elements (FOPFLASH-Renilla) in 293T cells after stimulation with the indicated combinations of Wnt3a protein and lentivirally transfected ASCL1 and DKK1. DKK1 overexpression abrogates ASCL1-mediated Wnt induction. (F) DKK1 and ASCL1 expression patterns are inversely correlated in 200 GBM samples (Verhaak et al., 2010). Each point denotes a tumor sample. Error bars in qRT-PCR experiments indicate SEM. See also Figure S5.

(Invitrogen), 13 Glutamax, and 0.1 mM b-mercaptoethanol, which was supple- of laminin (Sigma) in PBS was added to plates and plates were incubated at mented with 20 ng/ml of both epidermal growth factor (EGF) and fibroblast 37C for at least 3 hr. Cells were passaged using manual dissociation. growth factor-2 (FGF-2) (R&D). NSCs were grown on poly-L-ornithine- and lam- For differentiation into astrocytes, when cells were 80%–90% confluent, inin-coated plates. Poly-L-ornithine/laminin plates were generated as follows: the media was changed to NSC medium with 3% FCS and without EGF A20mg/ml solution of poly-L-ornithine (Sigma) in water was added to plates or FGF-2. After 4 days, cells were fixed for immunofluorescence. For differ- and plates were incubated at 37C for 1 hr. The poly-L-ornithine solution was entiation into , NSCs were grown to 90% confluency. Then, medium then removed, plates were washed three times with PBS, and 5 ml/ml solution was changed to either NSC medium but without N2 and supplemented

1574 Cell Reports 3, 1567–1579, May 30, 2013 ª2013 The Authors A

5 kb 5 kb

MGG8 CSC OLIG1 SOX8

10 kb 25 kb MGG8 CSC NR5A2 SOX5 H3K4me3 H3K4me1 LEF1 SOX5

B C GBM serum cells and NHA 100

PRC Wnt signaling X ASCL1 DKK1

10 Cancer stem cells ? Expression fold change Wnt signaling PRC 1

AX2 X LEF/TCF HEY2 SOX2 HES6 V AXIN2 OLIG1 NR5A2 ASCL1 DKK1 TF

Figure 5. Crosstalk of the TF Module and Wnt Signaling in GBM CSCs (A) ChIP-seq of LEF-1 (purple track) in MGG8 CSCs reveals enrichment at H3K4me1-marked distal elements near several TF loci. (B) Relative mRNA level changes for Wnt-responsive TFs with distal elements bound by LEF-1 after transfection of NHA with ASCL1 and stimulation with Wnt3a measured by qRT-PCR. Error bars indicate SEM. (C) A model for crosstalk between aberrantly activated TFs and Wnt signaling in nonstem GBM cells and NHA (top) versus GBM CSCs (bottom). In nonstem GBM cells and NHA, Polycomb complexes repress TFs, including ASCL1. In the absence of ASCL1 protein, the DKK1 upstream regulatory elements is active, the locus is transcribed, and expressed DKK1 inhibits Wnt signaling. In GBM CSCs, Polycomb repression is lost at many TF loci, including ASCL1. ASCL1 binds to the DKK1 regulatory element, thereby repressing DKK1 expression and activating Wnt signaling. Active Wnt signaling feeds back upon loci encoding several other TFs that are aberrantly active in GBM CSCs. See also Figure S6. with 13 B27 or ENStem-A neuronal differentiation medium (Chemicon) FACS Analysis supplemented with L-glutamine (2 mM). After 2 weeks, cells were fixed for CD133 (Miltenyi Biotec CD133/1 PE cat # 130-080-801) and SSEA-1 (BD Bio- immunofluorescence. sciences cat # 560127) antibodies were used according to the manufacturer’s For immunofluorescence, cells were washed once with PBS and then fixed instructions. For TF staining in primary tumor, primary human glioblastomas with 4% paraformaldehyde for 30 min. Cells were washed three times with were obtained from patients operated at Massachusetts General Hospital in PBS and blocked for 2 hr (5% normal goat serum, 0.3% Triton X-100, PBS). accordance with an approved IRB protocol (2005-P-001609/16). Briefly, Then primary antibodies in blocking solution were added and cells were incu- tumors were dissociated to single-cell suspension using a brain tumor disso- bated overnight at 4C. The next day, cells were washed twice with PBS and ciation kit (Miltenyi Biotec) depleted for CD45-positive immune cells using then twice with blocking solution. Secondary antibody in blocking solution was microbeads and a MACS separator (Miltenyi Biotec). Cells were stained with added and cells were incubated for 2 hr at room temperature. Cells were then SOX2 (R&D Systems) and ASCL1 (BD Pharmingen) antibodies conjugated to washed three to five times with PBS and then counterstained with DAPI/ Alexa Fluor 647 or Alexa Fluor 488 using Alexa Fluor protein labeling kits (Invi- 1XPBS solution. Primary antibodies include anti-nestin 1:500 (Chemicon Cat trogen). Flow cytometric analysis was conducted with an LSR II flow cytometer SCR060), anti-Sox2 1:200 (Chemicon Cat SCR019), anti-BLBP 1:500 (Chem- (BD Biosciences) and analysis was performed with FlowJo software (Treestar). icon Cat AB9558), anti-GFAP (Chemicon Cat SCR019), anti-betaIII tubulin 1:500 (Chemicon Cat SCR060), and anti-MAP-2 1:200 (Chemicon SCR019). ChIP-Seq Assay Secondary antibodies were Alexa Fluor 488 (goat anti-rabbit immunoglobulin ChIP assays were carried out on cultures of approximately 1 3 106 cells per G [IgG], Invitrogen Cat A-11008) at 1:200 and Alexa Fluor 555 (goat anti-mouse sample and epitope, following the general procedures outlined previously IgG, Invitrogen Cat A-21422) at 1:200. (Ku et al., 2008; Mikkelsen et al., 2007). Immunoprecipitation was performed NHAs were obtained from Lonza and propagated according to the manufac- using antibodies against H3K4me3 (Millipore 07-473), H3K27me3 (Millipore turer’s specifications. 07449), H3K36me3 (Abcam 9050), H3K4me1 (Abcam 8895), ASCL1

Cell Reports 3, 1567–1579, May 30, 2013 ª2013 The Authors 1575 (Epitomics T091), or LEF-1 (Abcam 53293). ChIP DNA samples were used to Generation of Aberrantly Active TF Set prepare sequencing libraries, which were then sequenced on the Illumina TFs were identified as those aberrantly active genes that were contained in the Genome Analyzer II or HiSeq by standard procedures. We sequenced an GO:0003700 transcription factor activity set or in the set of human TFs defined input control for each cell type for use in normalization. Read alignment to by Vaquerizas et al. (2009). We manually removed TFs whose chromatin state the hg19 human reference genome was performed with Bowtie (Langmead was incorrectly of ambiguously called by our algorithm to generate the final list et al., 2009) and density maps were generated with read extension to of TFs. MYCN was also removed because of focal amplification in the MGG8 200 bp with IGVtools (Robinson et al., 2011; Thorvaldsdo´ ttir et al., 2013). cell line. For the TF chromatin state heatmap, we extracted H3K4me3 and When several reads with same start position and direction were present, H3K27me3 signal, respectively, in 40 bins covering a 5 kb region around the only one was kept for further analysis. Two replicates that were available for annotated TSS from density maps. For both H3K4me3 and H3K27me3, MGG8 GBM CSCs were merged into a single track. Visualization was per- several control genes with similar chromatin state in all samples were chosen formed with the Integrative Genomics Viewer. ChIP-seq data set statistics (Table S4) and served to scale signal for each sample. TFs were ordered based are summarized in Table S1 and data are available for viewing at http:// on their H3K27me3 signal using hierarchical clustering (R Development Core www.broadinstitute.org/cgi-bin/epigenomics/public/prod/cloneportal.cgi? Team, 2008), and H3K4me3 and H3K27me3 maps were overlayed to generate data=rheinbay_GBMCSC. the final map. Cells exceeding 15% of maximum signal for H3K4me3 and H3K27me3 (bivalent) were additionally enhanced with orange. DNA Methylation Assay and Analysis For each sample, about 1 3 106 cells were harvested and genomic DNA was RNA Extraction and Gene Expression Analyses isolated using the QiaAMP DNA mini kit following the manufacturer’s instruc- Total RNA was isolated from cells using TRIzol Reagent (Invitrogen) and puri- tions. DNA was eluted in 100 ml water, treated with RNase (37C for 30 min), fied using the RNeasy Kit (QIAGEN). Gene expression data were acquired with and cleaned up again with the QiaAMP DNA mini kit. Affymetrix U133 2.0 arrays. CEL files were normalized with Data were processed using Illumina BeadStudio software. Probes with p robust multichip average and multiple probe sets per gene were collapsed value calls above 0.05 were discarded and b values for two replicates for by taking the maximum expression value using the GenePattern package each sample were averaged. Probes with b R 0.75 were classified as ‘‘hyper- (Reich et al., 2006). Gene expression data for NHA were included from Balani methylated.’’ Data are available through the Gene Expression Omnibus et al. (2009) (GSE12305). Normal brain and astrocytoma transcriptome profiles (GSE46016). from Sun et al. (2006) were used (GSE4290) and processed as described above. TCGA combined expression data and subtype information was ob- tained from Verhaak et al. (2010). Detection of Regions Enriched for Histone Modifications Genomic regions enriched for a histone mark were identified using a sliding window approach as previously described (Mikkelsen et al., 2007) with several Motif Analyses modifications. We adapted the previous approach for the highly copy-number We used the HOMER software package (Heinz et al., 2010) to search for de variant genomes of the GBM CSCs with the help of an unenriched input novo enriched motifs in TF peak regions. sequencing track generated from whole-cell extract (WCE). In short, a fixed size window of 1 kb was used to scan the genome in 100 bp steps for local Overexpression and Knockdown Experiments enrichment of ChIP signal. Significance of signal in each window was as- Human cDNA for ASCL1, OLIG1, OLIG2, HEY2, LHX2, and EN2 was cloned sessed based on the assumption that random read alignment would follow a into a lentiviral plasmid and sequence verified. Primers used are listed in Table

Poisson distribution with parameter lChIP. lChIP was adjusted for local variation S6. For knockdown experiments, an ASCL1 lentiviral shRNA set from Open in genome copy number by multiplication with the observed-to-expected ratio Biosystems was used (RHS4533-NM_004316), of which TRCN0000013551 (O/E) for unenriched input reads in this region (Mikkelsen et al., 2010). To in- (CCCGAACTGATGCGCTGCAAA) yielded sufficient knockdown. The same crease numeric stability in regions of heterozygous deletions, we calculated sequence was also used in vector pGIPZ (RHS4430-101103529) to allow for this O/E ratio based on input reads in the scoring window as well as in a 10 GFP sorting. Lentiviruses were produced as previously described (Barde kb and a 100 kb region centered at the current window and used the maximal et al., 2010). Briefly, cDNA coding or shRNA plasmids were cotransfected value of these three. When all three input O/E ratios were zero, ladjusted was set with GAG/POL and VSV plasmids into 293T packaging cells to produce the vi- to equal lChIP. Poisson p values were then calculated for each window with rus used to infect the target cells (NHA or GBM CSC). Viral supernatant was ladjusted. All p values were corrected for multiple-hypothesis testing using concentrated by ultracentrifugation using an SW41Ti rotor (Beckman Coulter) the Benjamini-Hochberg procedure (Benjamini and Hochberg, 1995), and at 28,000 rpm for 120 min. Using GFP control, efficiency of infection was esti- only windows with significance p < 10À5 were kept. Finally, adjacent (distance mated as greater than 90% (data not shown). For maximal homogeneity, NHA % 1 kb) enriched windows were merged into a single interval. For histone mod- were selected using 0.75 mg/ml puromycin for 5 days and GBM CSC were ifications in NSCs and NHAs, we applied the same algorithm and parameters either selected using 2 mg/ml puromycin or sorted for GFP depending on vec- albeit without background correction. Genomic regions enriched for ASCL1 or tor used. After selection, RNA was extracted (QIAGEN RNeasy kit) following LEF-1 were identified with MACS (Zhang et al., 2008). No background correc- the manufacturer’s instructions. tion as described above was applied; instead, peaks identified in the WCE track served to remove spurious TF peaks. Real-Time Quantitative RT-PCR TSS for genes from the hg19 human genome assembly were defined as re- cDNA was obtained using Moloney murine leukemia virus reverse transcrip- gion from 500 bp upstream to 2 kb downstream as previously described (Mik- tase and RNase H minus (Promega). Typically, 250 ng of template total RNA kelsen et al., 2007). Chromatin state calls for TSS were then made based on a and 250 ng of random hexamers were used per reaction. Real-time PCR ampli- minimal overlap of 500 bp of enriched intervals with the 2.5 kb TSS region. A fication was performed using Power SYBR mix and specific PCR primers in a consensus set of TSS chromatin states in the CSCs was generated with the 7500 Fast PCR instrument (Applied Biosystems). Relative quantification of chromatin state of the majority (at least three of four CSC lines) assigned to each target, normalized to an endogenous control (GAPDH), was performed summarized ‘‘CSC’’ cell type. 87% (n = 20,422) of TSS satisfied the majority using the comparative Ct method (Applied Biosystems). Error bars indicate criterion and were thus included in the consensus set. SEM. Primer sequences are listed in Table S6.

Functional Gene Enrichment Analysis Luciferase Assay Functional enrichment in gene sets was determined using the DAVID func- TOPFLASH-Firefly and FOPFLASH-Renilla plasmids were cotransfected with tional annotation tool with ‘‘FAT’’ terms (Dennis et al., 2003; ASCL1 lentivirus or control vector in 293T cells using Fugene6, as previously version 6.7). Benjamini-Hochberg p values correcting for multiple hypothesis described (Firestein et al., 2008; Veeman et al., 2003). When indicated, testing were used for further interpretation. Wnt3a was added at 100 ng/ml (R&D 5036-WN-010). Luciferase activity was

1576 Cell Reports 3, 1567–1579, May 30, 2013 ª2013 The Authors measured after 48 hr using a dual-luciferase reporter assay system (Promega Fund (B.E.B.), and the Howard Hughes Medical Institute (B.E.B.). M.N.R. E1910) according to the manufacturer’s instructions. and M.S. receive research funding from Affymetrix.

Sphere-Formation Assay Received: October 15, 2012 GFP-sorted GBM CSC spheres, infected either with lentiviral control vector Revised: February 6, 2013 or with ASCL1 shRNA vector, were mechanically dissociated into single cells Accepted: April 20, 2013 and plated at a density of one cell per well in 96-well plates, in triplicate. Sphere Published: May 23, 2013 number was assessed 2 weeks later under a fluorescence microscope. For sphere diameter quantification, five pictures were taken per condition at REFERENCES 1003 magnification. At least 60 spheres per conditions were measured with Image J. Balani, P., Boulaire, J., Zhao, Y., Zeng, J., Lin, J., and Wang, S. (2009). High mobility group box2 promoter-controlled suicide gene expression enables tar- Tumorigenicity Study geted glioblastoma treatment. Mol. Ther. 17, 1003–1011. Intracranial injections of 5,000 cells from acutely dissociated gliomaspheres Bao, S., Wu, Q., McLendon, R.E., Hao, Y., Shi, Q., Hjelmeland, A.B., Dewhirst, were performed with a stereotactic apparatus (Kopf Instruments) at coordi- M.W., Bigner, D.D., and Rich, J.N. (2006). Glioma stem cells promote radiore- nates 2.2 mm lateral relative to Bregma point and 2.5 mm deep from dura sistance by preferential activation of the DNA damage response. Nature 444, mater. Four severe combined immunodeficient mice were used per condition. 756–760. Kaplan-Meier curves and statistical significance (log-rank test) were calcu- Barde, I., Salmon, P., and Trono, D. (2010). Production and titration of lentiviral lated with the R survival package (R Development Core Team, 2008). Animal vectors. Curr. Protoc. Neurosci. Chapter 4, Unit 4, 21. experiments were approved by protocol number 2009N000061. Baylin, S.B., and Jones, P.A. (2011). A decade of exploring the cancer epige- nome - biological and translational implications. Nat. Rev. Cancer 11, 726–734. RNA ISH Bell, O., Tiwari, V.K., Thoma¨ , N.H., and Schu¨ beler, D. (2011). Determinants and mRNA was detected in formalin-fixed, paraffin-embedded tissue sections dynamics of genome accessibility. Nat. Rev. Genet. 12, 554–564. using Quantigene ViewRNA (Affymetrix). Probes for ASCL1 (type 1, red, VA1-11908; Affymetrix) and Sox2 (type 6, blue, VA-11765) were used following Benjamini, Y., and Hochberg, Y. (1995). Controlling the false discovery rate: a the manufacturer’s instructions for two-color chromogenic ISH. Tissue practical and powerful approach to multiple testing. J. R. Stat. Soc. Series B sections were prepared for hybridization by fixation in 10% formaldehyde, de- Stat. Methodol. 57, 289–300. paraffinization, boiling for 10 min, and digestion with protease for 20 min. Hy- Bruggeman, S.W., Hulsman, D., Tanger, E., Buckle, T., Blom, M., Zevenhoven, bridization was performed for 2 hr at 40C. Signal amplification and detection J., van Tellingen, O., and van Lohuizen, M. (2007). Bmi1 controls tumor devel- were performed using standard conditions for fast red and fast blue sub- opment in an Ink4a/Arf-independent manner in a mouse model for glioma. strates. Tissues were counterstained with hematoxylin and visualized with a Cancer Cell 12, 328–341. standard bright-field microscope. Chen, J., Li, Y., Yu, T.S., McKay, R.M., Burns, D.K., Kernie, S.G., and Parada, L.F. (2012). A restricted cell population propagates glioblastoma growth after ACCESSION NUMBERS chemotherapy. Nature 488, 522–526. Clement, V., Sanchez, P., de Tribolet, N., Radovanovic, I., and Ruiz i Altaba, A. The Gene Expression Omnibus accession number for the microarray gene (2007). HEDGEHOG-GLI1 signaling regulates human glioma growth, cancer expression data, DNA methylation profiles, and ChIP-seq density maps re- stem cell self-renewal, and tumorigenicity. Curr. Biol. 17, 165–172. ported in this paper is GSE46016. Clevers, H. (2006). Wnt/beta-catenin signaling in development and disease. Cell 127, 469–480. SUPPLEMENTAL INFORMATION Conti, L., Pollard, S.M., Gorba, T., Reitano, E., Toselli, M., Biella, G., Sun, Y., Sanzone, S., Ying, Q.L., Cattaneo, E., and Smith, A. (2005). Niche-independent Supplemental Information includes six figures and six tables and can be found symmetrical self-renewal of a mammalian tissue stem cell. PLoS Biol. 3, e283. with this article online at http://dx.doi.org/10.1016/j.celrep.2013.04.021. Dennis, G., Jr., Sherman, B.T., Hosack, D.A., Yang, J., Gao, W., Lane, H.C., and Lempicki, R.A. (2003). DAVID: Database for Annotation, Visualization, LICENSING INFORMATION and Integrated Discovery. Genome Biol. 4,3. Dunham, I., Kundaje, A., Aldred, S.F., Collins, P.J., Davis, C.A., Doyle, F., This is an open-access article distributed under the terms of the Creative Epstein, C.B., Frietze, S., Harrow, J., Kaul, R., et al.; ENCODE Project Con- Commons Attribution-NonCommercial-No Derivative Works License, which sortium. (2012). An integrated encyclopedia of DNA elements in the human permits non-commercial use, distribution, and reproduction in any medium, genome. Nature 489, 57–74. provided the original author and source are credited. Ernst, J., Kheradpour, P., Mikkelsen, T.S., Shoresh, N., Ward, L.D., Epstein, C.B., Zhang, X., Wang, L., Issner, R., Coyne, M., et al. (2011). Mapping and ACKNOWLEDGMENTS analysis of chromatin state dynamics in nine human cell types. Nature 473, 43–49. We thank Tarjei Mikkelsen, Timothy Durham, and Noam Shoresh for computa- Firestein, R., Bass, A.J., Kim, S.Y., Dunn, I.F., Silver, S.J., Guney, I., Freed, E., tional assistance and Chuck Epstein, Robbyn Issner, and the Broad Institute Ligon, A.H., Vena, N., Ogino, S., et al. (2008). CDK8 is a colorectal cancer Genome Sequencing Platform for help with data production. We thank David oncogene that regulates beta-catenin activity. Nature 455, 547–551. Dombkowski for help with FACS analysis and Lauren Solomon and Leslie Gaff- ney for assistance with graphics. A.S.C. is supported by a Joan Ambriz Amer- Galli, R., Binda, E., Orfanelli, U., Cipelletti, B., Gritti, A., De Vitis, S., Fiocco, R., ican Brain Tumor Association basic research fellowship and early career Foroni, C., Dimeco, F., and Vescovi, A. (2004). Isolation and characterization of research award from the Ben and Catherine Ivy Foundation. M.L.S. is sup- tumorigenic, stem-like neural precursors from human glioblastoma. Cancer ported by Oncosuisse grant BIL-KFS-02590-02-2010 and a Medic Foundation Res. 64, 7011–7021. grant. M.N.R. is supported by awards from the Burroughs Wellcome Fund and Gangemi, R.M., Griffero, F., Marubbi, D., Perera, M., Capra, M.C., Malatesta, the Howard Hughes Medical Institute. S.D.R. supported in part by NIH (RO1 P., Ravetti, G.L., Zona, G.L., Daga, A., and Corte, G. (2009). SOX2 silencing in CA160762). This research was supported by funds from the Starr Cancer Con- glioblastoma tumor-initiating cells causes stop of proliferation and loss of sortium (B.E.B.), the Harvard Stem Cell Institute, the Burroughs Wellcome tumorigenicity. Stem Cells 27, 40–48.

Cell Reports 3, 1567–1579, May 30, 2013 ª2013 The Authors 1577 Heinz, S., Benner, C., Spann, N., Bertolino, E., Lin, Y.C., Laslo, P., Cheng, J.X., Reich, M., Liefeld, T., Gould, J., Lerner, J., Tamayo, P., and Mesirov, J.P. Murre, C., Singh, H., and Glass, C.K. (2010). Simple combinations of lineage- (2006). GenePattern 2.0. Nat. Genet. 38, 500–501. determining transcription factors prime cis-regulatory elements required for Robinson, J.T., Thorvaldsdo´ ttir, H., Winckler, W., Guttman, M., Lander, E.S., macrophage and B cell identities. Mol. Cell 38, 576–589. Getz, G., and Mesirov, J.P. (2011). Integrative genomics viewer. Nat. Bio- Hulleman, E., Quarto, M., Vernell, R., Masserdotti, G., Colli, E., Kros, J.M., Levi, technol. 29, 24–26. D., Gaetani, P., Tunici, P., Finocchiaro, G., et al. (2009). A role for the transcrip- Schlesinger, Y., Straussman, R., Keshet, I., Farkash, S., Hecht, M., Zimmer- tion factor HEY1 in glioblastoma. J. Cell. Mol. Med. 13, 136–146. man, J., Eden, E., Yakhini, Z., Ben-Shushan, E., Reubinoff, B.E., et al. Huse, J.T., and Holland, E.C. (2010). Targeting brain cancer: advances in the (2007). Polycomb-mediated methylation on Lys27 of histone H3 pre-marks molecular pathology of malignant glioma and medulloblastoma. Nat. Rev. genes for de novo methylation in cancer. Nat. Genet. 39, 232–236. Cancer 10, 319–331. Shih, A.H., and Holland, E.C. (2006). Notch signaling enhances nestin expres- Ku, M., Koche, R.P., Rheinbay, E., Mendenhall, E.M., Endoh, M., Mikkelsen, sion in gliomas. Neoplasia 8, 1072–1082. T.S., Presser, A., Nusbaum, C., Xie, X., Chi, A.S., et al. (2008). Genomewide Singh, S.K., Clarke, I.D., Terasaki, M., Bonn, V.E., Hawkins, C., Squire, J., and analysis of PRC1 and PRC2 occupancy identifies two classes of bivalent do- Dirks, P.B. (2003). Identification of a cancer stem cell in human brain tumors. mains. PLoS Genet. 4, e1000242. Cancer Res. 63, 5821–5828. Langmead, B., Trapnell, C., Pop, M., and Salzberg, S.L. (2009). Ultrafast and Singh, S.K., Hawkins, C., Clarke, I.D., Squire, J.A., Bayani, J., Hide, T., Henkel- memory-efficient alignment of short DNA sequences to the human genome. man, R.M., Cusimano, M.D., and Dirks, P.B. (2004). Identification of human Genome Biol. 10, R25. brain tumour initiating cells. Nature 432, 396–401. Lee, J., Kotliarova, S., Kotliarov, Y., Li, A., Su, Q., Donin, N.M., Pastorino, S., Son, M.J., Woolard, K., Nam, D.H., Lee, J., and Fine, H.A. (2009). SSEA-1 is an Purow, B.W., Christopher, N., Zhang, W., et al. (2006). Tumor stem cells enrichment marker for tumor-initiating cells in human glioblastoma. Cell Stem derived from glioblastomas cultured in bFGF and EGF more closely mirror Cell 4, 440–452. the phenotype and genotype of primary tumors than do serum-cultured cell lines. Cancer Cell 9, 391–403. Stricker, S.H., Feber, A., Engstro¨ m, P.G., Care´ n, H., Kurian, K.M., Takashima, Y., Watts, C., Way, M., Dirks, P., Bertone, P., et al. (2013). Widespread Lee, J., Son, M.J., Woolard, K., Donin, N.M., Li, A., Cheng, C.H., Kotliarova, S., resetting of DNA methylation in glioblastoma-initiating cells suppresses Kotliarov, Y., Walling, J., Ahn, S., et al. (2008). Epigenetic-mediated dysfunc- malignant cellular behavior in a lineage-dependent manner. Genes Dev. 27, tion of the bone morphogenetic protein pathway inhibits differentiation of glio- 654–669. blastoma-initiating cells. Cancer Cell 13, 69–80. Sun, L., Hui, A.M., Su, Q., Vortmeyer, A., Kotliarov, Y., Pastorino, S., Passaniti, Ligon, K.L., Alberta, J.A., Kho, A.T., Weiss, J., Kwaan, M.R., Nutt, C.L., Louis, A., Menon, J., Walling, J., Bailey, R., et al. (2006). Neuronal and glioma-derived D.N., Stiles, C.D., and Rowitch, D.H. (2004). The oligodendroglial lineage stem cell factor induces angiogenesis within the brain. Cancer Cell 9, 287–300. marker OLIG2 is universally expressed in diffuse gliomas. J. Neuropathol. Exp. Neurol. 63, 499–509. Suva` , M.L., Riggi, N., Janiszewska, M., Radovanovic, I., Provero, P., Stehle, Lottaz, C., Beier, D., Meyer, K., Kumar, P., Hermann, A., Schwarz, J., Junker, J.C., Baumer, K., Le Bitoux, M.A., Marino, D., Cironi, L., et al. (2009). EZH2 M., Oefner, P.J., Bogdahn, U., Wischhusen, J., et al. (2010). Transcriptional is essential for glioblastoma cancer stem cell maintenance. Cancer Res. 69, profiles of CD133+ and CD133- glioblastoma-derived cancer stem cell lines 9211–9218. suggest different cells of origin. Cancer Res. 70, 2030–2040. Thorvaldsdo´ ttir, H., Robinson, J.T., and Mesirov, J.P. (2013). Integrative Geno- Meissner, A., Mikkelsen, T.S., Gu, H., Wernig, M., Hanna, J., Sivachenko, A., mics Viewer (IGV): high-performance genomics data visualization and explora- Zhang, X., Bernstein, B.E., Nusbaum, C., Jaffe, D.B., et al. (2008). Genome- tion. Brief. Bioinform. 14, 178–192. scale DNA methylation maps of pluripotent and differentiated cells. Nature Travis, A., Amsterdam, A., Belanger, C., and Grosschedl, R. (1991). LEF-1, a 454, 766–770. gene encoding a lymphoid-specific protein with an HMG domain, regulates Mikkelsen, T.S., Ku, M., Jaffe, D.B., Issac, B., Lieberman, E., Giannoukos, G., T-cell alpha enhancer function [corrected]. Genes Dev. 5, 880–894. Alvarez, P., Brockman, W., Kim, T.K., Koche, R.P., et al. (2007). Genome-wide Valent, P., Bonnet, D., De Maria, R., Lapidot, T., Copland, M., Melo, J.V., Cho- maps of chromatin state in pluripotent and lineage-committed cells. Nature mienne, C., Ishikawa, F., Schuringa, J.J., Stassi, G., et al. (2012). Cancer stem 448, 553–560. cell definitions and terminology: the devil is in the details. Nat. Rev. Cancer 12, Mikkelsen, T.S., Xu, Z., Zhang, X., Wang, L., Gimble, J.M., Lander, E.S., and 767–775. Rosen, E.D. (2010). Comparative epigenomic analysis of murine and human Vaquerizas, J.M., Kummerfeld, S.K., Teichmann, S.A., and Luscombe, N.M. adipogenesis. Cell 143, 156–169. (2009). A census of human transcription factors: function, expression and evo- Murat, A., Migliavacca, E., Gorlia, T., Lambiv, W.L., Shay, T., Hamou, M.F., de lution. Nat. Rev. Genet. 10, 252–263. Tribolet, N., Regli, L., Wick, W., Kouwenhoven, M.C., et al. (2008). Stem cell- Veeman, M.T., Slusarski, D.C., Kaykas, A., Louie, S.H., and Moon, R.T. (2003). related ‘‘self-renewal’’ signature and high epidermal growth factor receptor Zebrafish prickle, a modulator of noncanonical Wnt/Fz signaling, regulates expression associated with resistance to concomitant chemoradiotherapy in gastrulation movements. Curr. Biol. 13, 680–685. glioblastoma. J. Clin. Oncol. 26, 3015–3024. Verhaak, R.G., Hoadley, K.A., Purdom, E., Wang, V., Qi, Y., Wilkerson, M.D., Ohm, J.E., McGarvey, K.M., Yu, X., Cheng, L., Schuebel, K.E., Cope, L., Miller, C.R., Ding, L., Golub, T., Mesirov, J.P., et al.; Cancer Genome Atlas Mohammad, H.P., Chen, W., Daniel, V.C., Yu, W., et al. (2007). A stem cell- Research Network. (2010). Integrated genomic analysis identifies clinically like chromatin pattern may predispose tumor suppressor genes to DNA hyper- relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, methylation and heritable silencing. Nat. Genet. 39, 237–242. IDH1, EGFR, and NF1. Cancer Cell 17, 98–110. Osada, H., Tomida, S., Yatabe, Y., Tatematsu, Y., Takeuchi, T., Murakami, H., Wakimoto, H., Kesari, S., Farrell, C.J., Curry, W.T., Jr., Zaupa, C., Aghi, M., Kondo, Y., Sekido, Y., and Takahashi, T. (2008). Roles of achaete-scute homo- Kuroda, T., Stemmer-Rachamimov, A., Shah, K., Liu, T.C., et al. (2009). Human logue 1 in DKK1 and E-cadherin repression and neuroendocrine differentiation glioblastoma-derived cancer stem cells: establishment of invasive glioma in lung cancer. Cancer Res. 68, 1647–1655. models and treatment with oncolytic herpes simplex virus vectors. Cancer Pujadas, E., and Feinberg, A.P. (2012). Regulated noise in the epigenetic land- Res. 69, 3472–3481. scape of development and disease. Cell 148, 1123–1131. Wakimoto, H., Mohapatra, G., Kanai, R., Curry, W.T., Jr., Yip, S., Nitta, M., R Development Core Team (2008). R: a language and environment for statis- Patel, A.P., Barnard, Z.R., Stemmer-Rachamimov, A.O., Louis, D.N., et al. tical computing. R Foundation for Statistical Computing, Vienna. ISBN 3- (2012). Maintenance of primary tumor phenotype and genotype in glioblas- 900051-07-0, http://www.R-project.org. toma stem cells. Neuro. Oncol. 14, 132–144.

1578 Cell Reports 3, 1567–1579, May 30, 2013 ª2013 The Authors Widschwendter, M., Fiegl, H., Egle, D., Mueller-Holzner, E., Spizzo, G., Marth, Zheng, H., Ying, H., Wiedemeyer, R., Yan, H., Quayle, S.N., Ivanova, E.V., Paik, C., Weisenberger, D.J., Campan, M., Young, J., Jacobs, I., and Laird, P.W. J.H., Zhang, H., Xiao, Y., Perry, S.R., et al. (2010). PLAGL2 regulates Wnt (2007). Epigenetic stem cell signature in cancer. Nat. Genet. 39, 157–158. signaling to impede differentiation in neural stem cells and gliomas. Cancer Cell 17, 497–509. Zhang, Y., Liu, T., Meyer, C.A., Eeckhoute, J., Johnson, D.S., Bernstein, B.E., Zhou, V.W., Goren, A., and Bernstein, B.E. (2011). Charting histone modifica- Nusbaum, C., Myers, R.M., Brown, M., Li, W., and Liu, X.S. (2008). Model- tions and the functional organization of mammalian genomes. Nat. Rev. Genet. based analysis of ChIP-Seq (MACS). Genome Biol. 9, R137. 12, 7–18. Zhang, N., Wei, P., Gong, A., Chiu, W.T., Lee, H.T., Colman, H., Huang, H., Zhu, J., Adli, M., Zou, J.Y., Verstappen, G., Coyne, M., Zhang, X., Durham, T., Xue, J., Liu, M., Wang, Y., et al. (2011). FoxM1 promotes b-catenin nuclear Miri, M., Deshpande, V., De Jager, P.L., et al. (2013). Genome-wide chromatin localization and controls Wnt target-gene expression and glioma tumorigen- state transitions associated with developmental and environmental cues. Cell esis. Cancer Cell 20, 427–442. 152, 642–654.

Cell Reports 3, 1567–1579, May 30, 2013 ª2013 The Authors 1579