The Genomic Landscape of Diffuse Intrinsic Pontine Glioma And

© 2014 Nature America, Inc. All rights reserved. S.J.B. S.J.B. ( St. Jude Children’s Research Hospital, Memphis, Tennessee, USA. Hospital, Memphis, Tennessee, USA. Research, London, UK. Tennessee, USA. St. Biostatistics, Jude Children’s Research Hospital, Memphis, Tennessee, USA. Project, St. Jude Children’s Research Hospital, Memphis, Tennessee, USA. Tennessee, USA. Children’s Research Hospital, Memphis, Tennessee, USA. 1 the in differences substantial also are There locations. of spectrum broader a involve frequently more HGGs childhood whereas cortex, cerebral the in predominantly arise HGGs adult characteristics, cal histopathologi related share HGGs adult and childhood Although pediatric provides in 68%, chromatin tyrosine 40% neurotrophin DIPGs variants TP53 reported ACVR1 sequencing. by pontine We with Pediatric Hospital–Washington University Pediatric Cancer Genome Project James R Downing Cynthia Wetmore Kerri Ochoa WeiLei Arzu Onar-Thomas LuCharles Xiaoyan Zhu Gang Wu glioma and non-brainstempediatric glioma high-grade The genomic landscape of intrinsic diffuse pontine Nature Ge Nature Received 1 July 2013; accepted 6 March 2014; published online 6 April 2014; Department Department of Biology,Computational St. Jude Children’s Research Hospital, Memphis, Tennessee, USA.

DIPGs whole-genome,

analyzed

and [email protected] 73%

less and

exclusively

NBS-HGGs gliomas generating

kinase–RAS-PI3K 9

insights frequent high-grade

HGG and , , Heather L Mulder ATRX

1 remodeling,

than NBS-HGGs, and , etics

We 15 6 1

, , Panduka Nagahawatte 9 4 receptor NBS-HGGs. 6 Biostatistics and Biostatistics Roswell Bioinformatics, Park Cancer Institute, Buffalo, New York, USA. Department of Chemical Biology and St. Therapeutics, Jude Children’s Research Hospital, Memphis, Tennessee, USA. , , Alexander K Diaz 27

, Robert , S Robert Fulton

within ,

59% , , Chunxu Qu 20% (DIPGs)

identified

into

pediatric somatic

ADVANCE ONLINE PUBLICATION ONLINE ADVANCE whole-exome

both in

fusion in

Divison Divison of Cancer Institute Therapeutics, for Cancer Research, London, UK. 13 14

glioma

of survival DIPGs the

and infants. , , Amar Gajjar , , David W Ellison 7

genes

and , Robert , HuetherRobert

pediatric

with DIPGs and

unique

) ) or Jinghui Zhang (

signaling, genes This recurrent mutations outside

HGGs,

cell (HGG)

non-brainstem (32%), recurrent NTRK1, NTRK2

13 rate

Mutations 1 comprehensive Department Department of Oncology, St. Jude Children’s Research Hospital, Memphis, Tennessee, USA.

and 5 , , Xiang Chen

cycle

were , , Kristy Boggs

and and/or HGGs,

the 6 including

, , Lucinda L Fulton NBS-HGGs

years 2 histone somatic in

, in shared

regulation a 3

found 13

brainstem.

addition fusions

15 devastating histone

, , Li Ding 1 transcriptome

respectively, targeting

[email protected] , , Barbara S Paugh , , Erin Hedlund after

14 1 HGGs

modification

, , Richard Kriwacki in pathways

mutations diffuse , , Jinghui Zhang and 3

Integrated Integrated Biomedical Sciences Program, University of Tennessee Health Science Center, Memphis,

analysis involving 1 47%

diagnosis 2–

to , , Junyuan Zhang

were 5

5 NTRK3

, , Bhavin Vadodaria

. (NBS-HGGs),

6 previously

receptor

Structural disease genes, , Elaine , R Elaine Mardis

intrinsic 15 of

These These authors contributed equally to this work. should Correspondence be addressed to including

found

driving

1 the

. or

, Chris , JonesChris

1 6

The The Genome Institute, Washington University, St. Louis, Missouri, USA.

, , Michael Rusch

- 1 , , Sherri L Rankin 8 doi:10.1038/ng.293

Department Department of Structural Biology, St. Jude Children’s Research Hospital, Memphis, & Suzanne J Baker 8 tion tion rate of 9 × 10 muta background median showed cohort a respectively. the Overall, iations were found by exome sequencing and transcriptome sequencing,ing, and an additional 2,600 sequence mutations and 138 structural var and 2,039 structural variations were found by whole-genome sequenc nucleotide variations (SNVs) and small insertions or deletions (indels), Tables1 exome ( ( whole-genome by HGGs NBS 70 and DIPGs 57 of pediatric cases (127 tumors, 108 matched to germline DNA) consisting 118 from HGGs of landscape genomic the analyzed we glioma, hood site. anatomical and age with vary children in gliomagenesis within the population pediatric that suggesting there may subgroups be of age-dependent HGG, of even frequency lower a and prognosis better a have HGGs adult in cortex ing p.Gly34Arg or p.Gly34Val occur in pediatric HGGs of the cerebral thalamus and cerebellum, whereas histone H3 gene mutations encod the as such structures midline in HGGs pediatric in and children, in exclusively almost brainstem the in arise which DIPGs, in frequent ( gene HGGs adult and pediatric of features molecular 2 , , Matthew Parker , , John Easton To more comprehensively understand the pathways driving child driving pathwaysTo the comprehensively moreunderstand 6 5 3– H3F3A , , Richard K Wilson 10 , , Donald Yergeau n 5 – , = 80) and transcriptome ( , 11 1 9 7 ). A total of 39,590 sequence mutations, including single- including mutations, sequence 39,590 of total A ). . . In contrast, histone H3 gene mutations are extremely rare , , Frederick A Boop 1 , Stanley , PoundsStanley and and 8 12 3 . HGGs arising in infants younger than 3 years of age 3 years than younger infants in arising . HGGs Department Department of Surgery, St. Jude Children’s Research 2 −7 2 5 , , Bensheng Ju 10 HIST1H3B Department Department of Neurobiology,Developmental St. Jude , , Michael Edmonson mutations and base a per median of 22 structural Division Division of Molecular Pathology, Institute for Cancer 2 , 1 3 , , Pankaj Gupta for the St. Jude Children’s Research 5 , , Jake C Russell 6 ) mutations encoding p.Lys27Met are are p.Lys27Met encoding mutations ) 2 , , Michael R Taylor . . Thus, the pressuresselective driving n 7 = 75) sequencing ( 12 , , Tong Lin 4 , , Yongjin Li , Alberto , BroniscerAlberto 14 Department Department of Pathology, 1 , , Jared Becksfort 5 Pediatric Pediatric Cancer Genome 1 , , Xiaotu Ma 7 2 , , 3

1 , 6– TP53 s r e t t e l , n Supplementary

1 7 4 = 42), whole- 42), = 6 Department Department of , . Histone H3 H3 Histone .

mutations, mutations,

13 , 1

, , - - - - - © 2014 Nature America, Inc. All rights reserved. with retroviruses expressing Flag-tagged wild-type ACVR1 or the indicated mutants and serum starved for 2 h. Quantification of the ratio of phosphorylated cultures. Protein blots are shown for lysates of primary astrocytes isolated from the brainstem of neonatal mice with conditional knockout of (49%, without the formation of dorsal structures for embryos expressing the Gly356Asp (89%, the reduced severity of ventralization showing a decrease in the proportion of embryos with the V5 phenotype n embryos expressing the Arg258Gly (100%, rescue of dorsal structures (e.g., the head) in mutant ACVR1, as demonstrated by the partial 193189 reversed the ventralization effects of more severely affected. Treatment with LDN- the Gly356Asp and Gly328Val mutants are no dorsal structures, and embryos expressing Gly328Trp and Arg206His mutants have few to embryos expressing the Arg258Gly, Gly328Glu, with the indicated phenotype images of zebrafish embryos injected examined is shown on top. ( V5 ventralized phenotype. The number of embryos severe effect, with 90% of embryos showing the whereas the Gly328Val mutant had the most resulting only in the V3–V4 ventralized phenotype, Arg258Gly mutant had the least severe effect, (increasing severity from left to right). The ACVR1 Fig. 1 phenotype C1–C4 (as shown in ACVR1 phenotype. Embryos injected with wild-type embryos exhibiting a dorsalized or ventralized embryos. The graph shows the percentage of ( domain (TM) were not affected by mutations. extracellular domain (EC) and transmembrane germline mutations in individuals with FOP. The indicates alterations previously found as carrying the specified alteration, and an asterisk domains. Each red circle indicates a DIPG in DIPG were clustered in the GS or kinase signaling. ( Figure 2 mentally by independent sequencing methods (Online Methods). experi verified were sequencing whole-genome by found variations variations per genome ( were available were data in form RNA-seq only tabular in which shown for are Data samples 11 and summary. this samples from excluded hypermutator Four shown. are HGGs 112 for Mutations available. not information indicate grade tumor or location for boxes White indicated. are grade tumor (WHO) Organization Health World and hemisphere) cerebral versus (midline NBS-HGG of location NBS-HGG), or (DIPG subgroup Tumor together. CDK6 the of components of (CNVs) variants number copy involving variants Structural H3. histone category the in together grouped are mutations H3F3A for case male a or allele wild-type the of loss indicates line white diagonal A displayed. are left, the on indicated pathways the in mutated recurrently most genes the and including genes, 19 in detected alterations Genetic HGG. pediatric in 1 Figure s r e t t e l SMAD to total SMAD normalized to the empty vector control is shown below. b = 20), Gly328Asp (83%, ) Human ), whereas embryos injected with mutant G1 checkpoint complex are grouped grouped are complex checkpoint G1 CCND1 n mRNA showed a ventralized phenotype mRNA (WT) showed a dorsalized (H3.3) and and (H3.3) = 24) mutants. Scale bar, 200

ACVR1 Recurrent genetic alterations alterations genetic Recurrent NTRK1 ACVR1 a ) Missense ACVR1 substitutions , , CCND2, CCND3 CCND2, mutations in DIPG activate BMP ATRX , , ACVR1 mutants ventralize zebrafish NTRK2 HIST1H3B , which is X linked. linked. X is which , Supplementary Fig. 1 mRNA. Untreated or or n c = 23), Gly328Trp (100%, ) Representative NTRK3 S (H3.1) (H3.1) upplementary , , CDK4 ACVR1

and and

or or

m. (

)

ACVR1

). All SNVs and structural

mutations drive increased levels of phosphorylated SMAD1/5 (pSMAD1/5) in primary astrocyte b a d n Chromatin and = 20) and Arg206His (100%, transcription Percentage of embryos RB1 and TP53 RTK-RAS-PI3K

100 pathways regulation 10 20 30 40 50 60 70 80 90 1.00 Aberration 0 n

Vector Structural varian Missense Truncating Indel =

0.62 WT 22 WT R258G S 1.64 upplementary upplementary / s G328E 50 G328E / 3.87 ACVR

G328V // G328W 86 1.51 - t

R258G DIPG EC 1 47 / / mRNA

1.74 R206H / / mutated gene not previously identified in cancer was was cancer in identified previously not gene mutated G356D known as 64 2.82 Germline Deletion Amplification G356D / R206H Among recurrent mutations in HGG, pediatric the most frequently /// T / 48 able 9 able Tubulin Flag Total SMAD1/5/8 pSMAD1/5 pSMAD:total SMAD G328V n n 52 = 27) and Gly328Val = 27) mutants and ALK2 TM .

H3 alteration V5 V3–V4 V1–V2 Normal C1–C4 //// / H3.3 G34R H3.3 K27M H3.1 K27M ) encoding a bone morphogenetic protein (BMP) type I a DVANCE ONLINE PUBLICATION ONLINE DVANCE Midlin GS

// / / e R206H* Clinical summary

NBS-HG / Hemispheri NBS-HGG midlin NBS-HGG DIPG

/ R258G / / / c // / / ACVR1 mRNA G Hemispheri c

G328V G356D R206H G328W G328E R258G WT Uninjected G328E* / e

Kinase G328V Untreated / / c G328W* / G356D* <3 yearsol WHO gradeIV WHO gradeIII PPM1D TP53 CDKN2A/CDKN2 CCND1/CCND2/CCND3/CDK4/CDK6 NF BRAF PIK3R1 PIK3CA EGFR NTRK1/NTRK2/NTRK3 MET PDGFR MYC MYC BCORL1 BCO Histone H3 <3 yearsol Tumor grade Location ATRX ACVR1

1 Nature Ge Nature N R d 2.5 A Tp53 µ d M LDN-193189 ACVR1 transduced B n etics (also (also

© 2014 Nature America, Inc. All rights reserved. Indeed, mutations affecting these residues induced a weak gain of of gain weak a induced residues these affecting mutations Indeed, ( conformation active an to pocket of the kinase domain and would be to expected shift the kinase ATP-binding the or domain (GS) glycine-serine-rich inhibitory the either around clustered DIPGs in mutation by affected residues All ossifications heterotopic progressive drives differentiation lular cel aberrant which in (FOP), progressiva ossificans fibrodysplasia mutations previously identified in the autosomal dominant syndrome somatic these of Four ( ( of presence the and time survival longer age, younger with associated significantly were and missense ( receptor tumor relative to surrounding normal tissue. Scale bar, 50 of AKT phosphorylated (pAKT) and 44/p42 MAPK (p-44/p42 MAPK) in showed expression of Flag-tagged NTRK fusion proteins and elevated levels cells reminiscent of giant cell analysis Immunohistochemical glioblastoma. Tumors induced by BTBD1-NTRK3 showed the frequent presence of giant many with features of astrocytic and differentiation high mitotic activity. (of seven independent mice per construct) shows pleomorphic tumor cells, implanted into mouse brain. hematoxylin Representative and eosin staining expressing Flag-tagged TPM3-NTRK1 (top) or BTBD1-NTRK3 (bottom) and isolated from neonatal cortex or brainstem were transduced with virus induce high-grade astrocytomas. protein is shown at the lower right of each fusion. ( breakpoint indicated. The number of amino acids in each full-length fusion amino acid positions of the N-terminal and C-terminal fusion partners at the each fusion protein, the dashed red line shows the fusion point, with the domain carboxypeptidase of AGBL4 is not present in the fusion protein. For binding protein AGBL4 (green) was fused to NTRK2. The functional (light blue; VCL) was fused to NTRK2, and the N terminus of the ATP/GTP- ETV6 fused to NTRK3. The N terminus of the protein actin-binding vinculin interaction domain protein-protein (purple) of the ETS factor transcription (orange) from the protein I–interacting topoisomerase BTBD1 or the pointed NTRK1, and the BTB/POZ dimerization domain (gray) and the Kelch domain tropomyosin domain (yellow) of TPM3, an protein actin-binding fused to NTRK1, NTRK2 or NTRK3 (blue). N-terminal fusion partners include the proteins. ( Figure 3 Nature Ge Nature Figs. 1 P < 0.0000001) or with with or 0.0000001) < NBS-HGG DIPG and

Structural Structural variants generate oncogenic chimeric NTRK fusion a ACVR1 Figs. 1 1 Figs. ) ) All fusions included the C-terminal kinase domain from

2 n , etics KDM5B KDM5C Supplementary Fig. 3 SETD1A SMYD MLL MLL MLL PRDM ML K4 L 4 3 2 mutations were found exclusively in DIPGs (32%) (32%) DIPGs in exclusively found were mutations and 3 9

ADVANCE ONLINE PUBLICATION ONLINE ADVANCE

2 JMJD1C KDM3B KDM4B ACVR1 KDM3A , and and , K9 HIST1H3B PIK3CA Fig. Fig. Supplementary Figs. 2 2 Figs. Supplementary Tp53 mutations were the same as germline germline as same the were mutations Writers or or 2 and -null -null mouse primary astrocytes and and SIRT mutation encoding p.Lys27Met p.Lys27Met encoding mutation K1 PIK3R1 Erasers Supplementary Tables 4 8 7 Supplementary Fig. 3c Fig. Supplementary mutations ( mutations K2 b ) ) NTRK fusion proteins 7

µ SETD SETD SETD ASH1L KDM2 m. and K3 6 3 2 2 A 3 P < 0.005) 0.005) < ). Clonal Clonal ). Chromatin remodelers and H2A H3 18 Other histoneornonspecificwriters SMARCD2 SMARCA4 ) 5 RAD54B ARID1B , 2 1 H2B ARID1B CHD8 CHD6 CHD CHD ATRX ATRX H4 0 9 ). - . .

STK4–phS14 H2B 7 2 NCOA1-acH3/ UHRF1-ubH3 UBR2-ubH2A sal head structures for the Arg258Gly, Gly328Glu, Gly328Trp and and Gly328Trp Gly328Glu, Arg258Gly, the for structures head sal by induced of pathway antagonist BMP the selective highly a LDN-193189, of dose moderate A activation pathway BMP of indicator an embryos, zebrafish in phenotype ventralized a caused p.Arg206His encoding function head and dorsal structures ( structures and dorsal head of loss complete to partial with ventralization, of degrees varying in six all for mRNA with injection whereas inhibition, pathway BMP with consistent phenotype dorsalized mild human wild-type with injected embryos the of b a BTBD1-NTRK3 TPM3-NTRK1 AGBL4 ETV6 TPM VCL BTBD1 UBR5-ubH2A UHRF1-ubH CBX4-ubH2A ACVR1 Hematoxylin and eosin 3 20 4 , 2 ACVR1 1 . A previous study showed that the the that showed study previous A . mutations found in DIPG using this assay. Zebrafish Zebrafish assay. this using DIPG in found mutations 3 22 222 mutants, as demonstrated by the rescue of dor of rescue the by demonstrated as mutants, 211 , 2 3 , partially reversed the ventralization effects effects ventralization the reversed partially , NTRK1 400 289 histone H4. histone H4, H2; histone H2, phosphorylation; ph, acetylation; ac, ubiquitination; Ub, NBS-HGG. in degree, a lesser to and, DIPG in frequency high at mutated directly is which Lys27, of erasers or writers in alterations no were there Notably, remodelers. chromatin or histones other of modifications in involved proteins as well as H3, histone of tail the in Lys36) Lys9, (Lys4, Lys27, Lys18, residues lysine of modifications post-translational below) (erasers; remove or above) (writers; attach that proteins in identified were alterations Genetic shown. are DIPG or NBS-HGG in identified Alterations regulators. chromatin or modifiers 4 Figure rylated SMAD1 and SMAD5 (SMAD1/5), (SMAD1/5), SMAD5 and SMAD1 rylated of phospho levels increased caused cultures ACVR1 astrocyte mutants in primary mouse ( mutants Gly328Val and Gly356Asp the for ventralization of ity sever reduced the and mutants Arg206His NTRK2 482 Flag NTRK3 Fig. 2b Fig. 468 338

NTRK3 529 Pediatric HGG alterations in histone histone in alterations HGG Pediatric 812 , c and NTRK2 388 pAKT S473 ACVR1 Supplementary Fig. 3d Fig. Supplementary ACVR1 Fig. 2 Fig. 1,262 aa mRNA displayed a displayed mRNA 567 aa 634 aa 646 aa 618 aa mutants resulted resulted mutants ACVR1 c s r e t t e l 2 ). Expression of of Expression ). 2 p-p44/42 MAPK . We tested all all tested We . SJHGG027_D SJHGG127_D SJHGG082_D SJHGG004_D SJHGG099_D SJHGG081_D SJHGG016_D SJHGG009_A mutation mutation

, e ). ). - - -

s r e t t e l

HDGFRP

AKAP1 CCDC113

CPEB1 BTBD1

NTRK

ABHD2

TP53

3 CDH2

16 16 16 16 1 16 16

14 BTBD BICD1 17

15 16 16 16 16 16 16 18

3 16 16 16 16 16 16 16

SJHGG004_D 14

ATM 14 6

19 19

ATM 19 13

SJHGG027_D 13 SJHGG003_D

12 18

20 20 20

19 19 19

12 12

21 21 21

20 20 20 20 20

11 20

22 22 22

21 21 21 21 21 21

11 11

22 22 22

10 X PTEN

10 X 10

Y Y NTRK2 Y

LINGO2 9 OR13C5 Y Y Y Y Y Y SLC24A

2 9

BEND5 9

KDM4

8 AGBL4

8 MMP1 8

1 6 1 PPP3CC

PT PIWIL2

7 7 POM121L12 H3F3A

2 6 2 2

AAK1

6 6

3 PMS2

HSPE1 C6orf94 5 HSPE1-MOBKL3 4 TTN

3 3

5 5 ENC1 4 4

PDLIM5

SCNN1 CLCN7

RASSF9

SH2B3 PLEK2

TMTC2

CLEC4D PPFIA2 HS3ST3A

PDPR SJHGG016_D

SHISA6

CCDC15 BM VMP1

BAHCC1 USP3 TP53 DNAH10 RNF214

C11orf67 SETD1B G

F SJHGG044_D CYP27B1 ACER3 2 SJHGG003_A

UVRAG 1 B4GALNT1

16 16 16 16 16 16

14 KIF5A 6

16 16 16 16 16 16 16 17

C2CD3 15 14

16 16 16 16 16 16 16

14 CHD6

17 13 18

13 17 a ARL2 18

NRXN2 18

19 19 19

DVANCE ONLINE PUBLICATION ONLINE DVANCE NRIP1

19 19 19 12

19 19

OR8U8 12 19 20 20

12 20

20 20 20

21 21 EXT2 21

21 21 21

21 21

ELP4 21 11 22 22

11 22

22 22 22 11

22 22 SBF2 22 GPR158 TOLLIP

MLLT10

10 X X

X 10 MAGEC1 10 TAF3 DIP2C

DIP2C

Y Y Y

Y Y Y 9

MELK 9

CCDC23 WDYHV1 PTPRZ1

8 SLC26A TESK2 1

1 8

3 1

SRPK GPR89A

2 LELP1

RELN TPM3

CALC NTRK1

H3F3A 7 R GNAT

MAGI2

3 2 2 PMS2

Nature Ge Nature 2 SLC3A1

AUTS KDM3A 6

6 NXPH 2 CCDC148

3 5 1 MFSD6 3

5 3 5 4

MYRIP 4 SGCD AP3B1

ABLIM2 AFF1

PDGFRA CLNK

NM PRDM

n U

etics 5 © 2014 Nature America, Inc. All rights reserved. BTBD1 driving chromothripsis with a DIPG is SJHGG004_D radiotherapy. before collected was sample tumor This genome. stable a relatively A-T,with showing PMS2 germline with associated mutations of complexity and range broad the demonstrating thus (SJHGG003_A), HGGs other in seen abnormalities structural and number copy genomic typical carried SNVs fewer 100-fold approximately with tumor second the whereas (SJHGG003_D), genome stable extremely an had mutations somatic 800,000 than more with tumor hypermutator The include. to numerous too were tumors two these for indels and SNVs The (SJHGG003_A). a later, DIPG 2 years and, (SJHGG003_D) tumor glioneuronal malignant a hemispheric first tumors, independent two developed and a germline carried SJHGG003 Subject others. blue, fusions; in-frame in involved gene pink, breakpoints: variation structural at Genes SNV. splice-site blue, SNV;indel; red, missense brown, genes: RefSeq in mutations Sequence deletion. blue, amplification; red, (LOH); of heterozygosity loss Orange, mutations. non-sequence and translocations, interchromosomal and intra- CNAs, DNA including variants, genetic structural depict and 5 Figure and astr high-grade induced fusions NTRK Both brain. mouse into NTRK3 or BTBD1- human TPM3-NTRK1 Flag-tagged expressing retrovirus implanted we oncogenic be to shown and types cancer other in cohort, glioblastomas adult fied at low frequency in low-grade pediatric astrocytomas as well as in death or survival and neuronal differentiation outgrowth, of neurite induction ing the includ system, nervous the in signals developmental of range wide 7 Tables Supplementary an contained old years 3 than younger children in of NBS-HGGs Notably, (4/10) 40% NBS-HGGs. N-terminal fusion partners were identified in 4% of DIPGs and 10% of ( receptor neurotrophin three the of each of domain kinase whole-genome the involving and fusions Gene data. sequencing transcriptome from NBS-HGGs, and of DIPGs proportions equal in HGGs, pediatric of 47% in identified were ( 0.26%) = (FDR) rate ery without and with DIPGs in expression different significantly showed cesses 3g Fig. by DIPGs of segregation no with location tumor by clustered samples HGG that showed sets the exclusive between association state of the cell epigenetic the to part in related cells, stem glioblastoma in response medulloblastoma progenitors driving astrocytic differentiation or the proliferation of early hindbrain on context, dependent effects contrasting with is associated signaling BMP shown). not (data brain the into DIPG- implanted when tumorigenic the of six all hypothesis, associated ACVR1 mutants this failed to with render Consistent tumor initiation. driving than rather mutations critical other of presence the indi that predisposition, cating cancer with associated not is FOP syndrome germline However, disease. this in that evidence strong provides DIPGs ( astrocytes phosphorylated SMAD1/5 downstream of mutant ACVR1 in primary ( nitude mag varying with of BMP signaling, active indication a downstream Nature Ge Nature display. to numerous too were variants structural by disrupted genes the of names the chromothripsis, of examples To test the ability of and common were genes fusion generating variants Structural probe most variable the 1,000 using analysis Principal-component of mutation activating clonal and Recurrent o Supplementary Fig. 4 Fig. Supplementary cytomas with very short latency and complete penetrance ( mutation. SJHGG016_D is an infant NBS-HGG with a with NBS-HGG infant an is SJHGG016_D mutation. - ). Genes involved in the regulation of immune system pro system immune of regulation the in involved Genes ). NTRK3

TPM3 Fig. 2 Fig. ACVR1 Circos plots showing the range of structural alterations in pediatric HGG. Circos plots display the genome by chromosome in a circular plot plot a circular in chromosome by genome the display plots Circos HGG. pediatric in alterations structural of range the showing plots Circos n 2 ACVR1 Supplementary Fig. 3f Fig. Supplementary 4 etics . In tumors, BMP signaling induced the differentiation of - 26 fusion, shown in more detail in in detail more in shown fusion, d Tp53 NTRK1 ). LDN-193189 also effectively blocked signaling to to signaling blocked effectively also LDN-193189 ). mutation (Fisher’s exact test, , 2 28 7 2 . These context-dependent consequences likely drive 5

, mutation likely provides a selective advantage in in advantage selective a provides likely mutation but drove either a differentiation or proliferation or differentiation a either drove but 2 -null primary mouse astrocytes transduced with with transduced astrocytes mouse primary -null 30– 9 ADVANCE ONLINE PUBLICATION ONLINE ADVANCE . . NTRK NTRK 3 and 2 . Two of the five Two. five the of Supplementary Table 10 Supplementary ACVR1 and ETV6 ). The resulting tumors showed elevated elevated showed tumors resulting The ). fusion genes have recently been identi been recently have genes fusion fusion genes fusion to drive formation, glioma 8 NTRK ). The NTRK receptors transduce a transduce receptors NTRK The ). - ACVR1 mutation status ( status mutation NTRK3 ). ACVR1 NTRK fusion gene ( gene fusion ACVR1 NTRK mutation and DIPG. , were previously identified , were identified previously Tp53 ) genes and five different different five and genes ) mutation in the genetic genetic the in mutation S P upplementary Figure 8a Figure upplementary is an oncogenic driver driver oncogenic an is = 0.0002, false discov -null mouse astrocytes fusions found in our our in found fusions 33– ACVR1 Figs. 1 1 Figs. ). 3 Supplementary Supplementary

7 . TPM3 n 2 of 32% in and Fig. 3 -

3 NTRK1 , and and , . SJHGG044_D is an NBS-HGG showing dramatic chromothripsis. For the the For chromothripsis. dramatic showing NBS-HGG an is . SJHGG044_D b - - - - -

fusion and a very stable genome. SJHGG027_D is an NBS-HGG from a case a case from NBS-HGG an is SJHGG027_D genome. stable a very and fusion ing histone H3 gene mutations ( mutations gene H3 histone ing targeted was by mutationregulation in 22% ofepigenetic DIPGs in and 48% involved of NBS-HGGs, genes exclud of group this lectively, ( tone erasers were not significantly different in DIPGs and NBS-HGGs ( HGGs histone affecting H3 writers were more significantly frequent in NBS- ( genes chromatin-remodeling of and erasers and writers histone other of tions reported previously (RNA-seq). sequencing RNA or sequencing whole-genome by analyzed tumors HGG atric In contrast to previous reports detecting EGFRvIII expression in pedi ( NBS-HGGs of 67% and DIPGs of 69% in occurring cohort, entire the across quent tyrosine was fre mutations other through receptor signaling of (RTK)-RAS-PI3K kinase activation infants, from tumors in frequency activation. pathway MAPK and (PI3K) kinase (MAPK), downstream indicators of phosphoinositide 3-kinase protein mitogen-activated AKT and p42/44 of phosphorylated levels was identified as one of only two non-silent alterations in SJHGG082, a this age group ( from tumors in required mutationsis driver of number small a only mutation rates than the rest of the cohort ( lowersignificantlyshowed old years 3 under children in NBS-HGGs ( HGGpediatric drivingtions genes corepressor genome across the genes expressed of levels the amplify to act that factors tion ing of amplifications focal regulators that affect the epigenetic landscape were also found, includ ( HGG pediatric in alteration p.Lys27Met of advantage selective unique the supporting genes, H3 histone wild-type with DIPGs in including cohort, HGG entire the types tumor KDM6A and KDM6B are found in or other demethylases the H3K27 of the Polycomb-repressive complex 2 (PRC2) that methylates H3K27 components by levels targeting H3K27me3 Mutations that modulate proteins H3 histone of pool cellular entire the in (H3K27me3) at in results a 27 lysine of loss dominant of H3 histone trimethylation alteration H3.1 or H3.3 histone p.Lys27Met regulators. epigenetic of subgroup this and/or genes histone in mutations contained HGGs NBS- and 48% of NBS-HGGs hemispheric of 70% of DIPGs, midline histone H3 histone H3 including gene Indeed, genes. mutations, 91% in mutations missense with concurrent often were mutations These P = 0.3). Although only = Although 0.3). In addition to recurrent histone H3 histone In to gene, recurrent addition Although There was an enormous range in the complexity of the somatic muta Fig. Fig. P = 0.007), whereas frequencies of mutations affecting his affecting mutations of frequencies whereas 0.007), = 38 4 51 , NTRK 45– 3 and and 9 , 5 Fig. , we only detected EGFRvIII expression in 1 of the 85 85 the of 1 in expression EGFRvIII detected only we , 2 5 Fig. Fig. , and truncating mutations in the transcriptional transcriptional the in mutations truncating and , 0 . However, there were no such mutations across across mutations such no were there However, . BCOR Supplementary Fig. 5 Fig. Supplementary 5 -activating fusions were specifically found at high were specifically fusions -activating and 1 , Online Methods and and Methods Online , ATRX Supplementary Fig. 7 3– and and MYC 5 , 17 Supplementary6 Figs. BCORL1 , 4 Fig. Fig. mutations were highly recurrent, col recurrent, mutations were highly 0 and , , we also detected frequent mutations Fig. Fig. 4 MYCN ). Mutations in transcriptional transcriptional in Mutations ). 4 ( and and Fig. Fig. P ). Interestingly, mutations mutations Interestingly, ). < 0.0001), suggesting that , which encode , transcrip which encode ATRX Supplementary Fig. 5 Fig. Supplementary Supplementary Fig. 5 Fig. Supplementary 1 ). ). ETV6 and s r e t t e l PMS2 and - NTRK3 SETD2 7 mutation mutation ). The ten The ). muta fusion 41– 4 4 ). ). ). ). ------.

© 2014 Nature America, Inc. All rights reserved. found in HGG 59% of ( pediatric were pathways, RB and TP53 the including regulation, cycle cell ing was restricted to NBS-HGGs ( whereas DIPGs, in amplified nantly regulators TP53-dependent G1 cell cycle checkpoint ( phosphatase TP53-induced the in mutations truncating with exclusive mutually were and HGGs previous with reports consistent pathways, cancer common in mutations glioblastomas primary adult in fied of identi 86% to frequency the contrast in strong 3% of NBS-HGGs, Note Supplementary ( tumor autopsy or relapsed of opment devel the seed could tumor diagnosis the in clone founder a of ant descend a or clone founder A tumors. HGG all almost in identified ( ATM functional of absence the to due checkpoint DNA damage promised com a despite genome, stable relatively a had (A-T), telangiectasia ataxia with child a in arising NBS-HGG an SJHGG027_D, thera peutics. DNA-damaging of independent cases, some in least at was, occurrence their underlying mechanism the that therapy, indicating adjuvant before collected were chromothripsis showing samples all of tion including rearrangement, oncogenic in resulted chromothripsis cohort, In our Figs. 6 ( states number copy oscillating with ments to multiple breakpoints corresponding interconnecting genomic seg chromothripsis had PMS2 line mutations in known cancer predisposition genes, including from the evaluation of mutation frequency. Seven cases germ carried and ( genes repair DNA mismatch in mutation inherited with ated associ burden mutation tumor the in range potential the onstrating higher than in the DIPG that arose 2 years later (SJHGG003_A), dem 100-fold nearly was (SJHGG003_D), MGNT the case, this in arising remaining the in mutations inactivating somatic different acquired independently tumors both Although DIPG. separate a and (MGNT) tumor ronal glioneu malignant hemispheric IV a grade developed and mutation 7 Fig. which germline in biallelic SJHGG111 in mutations somatic 800,000 than more ing rate, includ mutation background high an with extremely mutators, oncogenic drivers in early postnatal brain tumor development. mental glioma model strongly suggest thatexperi ourthese in fusiononset tumorgenes rapid the areand tumorspotent these in mutations additional of paucity the age, of years 3 than younger children from disease congenital fibrosarcoma, as well as in papillary thyroid cancer,as an such adultchildren young very from those including types, tumor tiple mul in identified fusionsfoundhere,have been the of including two glioblastoma from a 1-month-old subject. Notably, s r e t t e l brain childhood of spectrum devastating a driving pathways critical This global view of the geneticlandscapedefinespediatrictheHGGof of view global This frequent includes also HGGs pediatric of landscape genomic The We identified sequencing whole-genome by analyzed (31%) tumors Thirteen hyper as DNA, scored were normal matched 3 with tumors, Four Supplementary Fig. 7 Fig. Supplementary , , ). Subject SJHGG003 carried a heterozygous germline germline heterozygous a carried SJHGG003 Subject ). , 30 Fig. Fig. MSH6 3 8 8 PDGFRA , , 7– 33– and CCND1 9 BTBD1 , 3 5 11 6 9 and and . The high frequency of and and , 12 , , PMS2 Supplementary Table Supplementary 12 , 15 TERT and and - , , PMS2 Supplementary Fig. 7 Fig. Supplementary NF1 1 NTRK3 CCND2 6 5 . . allele, the basal mutation rate in the first tumor tumor first the in rate mutation basal the allele, 3 TP53 ). EGFR , identified by complex rearrangements with with rearrangements complex by identified , promoter mutations in 2% only and of mutations promoter DIPGs ( mutations were identified ( identified were mutations Supplementary Table 11 Supplementary fusion and rearrangement or amplifica or rearrangement and fusion , mutations occurred in 42% of pediatric pediatric of 42% in occurred mutations ). Hypermutator tumors were excluded excluded were tumors Hypermutator ). CCND3 PPM1D ( Fig. Supplementary Fig. 8 Fig. Supplementary Fig. Fig. 1 ). Taken together, mutations affect 5 NTRK , , previously shown to impair the the impair to shown previously , 4 CDKN2A CDK4 1 . and Supplementary Fig. 10 Fig. Supplementary and Fig. ). Multiple subclones were were subclones Multiple ). fusion genes in NBS-HGGs and Supplementary Fig. 5 Supplementary Fig. Fig. 1 Supplementary Note Supplementary homozygous deletion deletion homozygous ) 5 CDK6 5 . The G1 checkpoint NTRK 5 ). , , Supplementary Supplementary Supplementary Supplementary ). Nearly half of of half Nearly ). were predomi fusion genes, PMS2 Fig. Fig. TP53 and and ). ). ). 5 ------,

E. E. Shore and A. Culbert (University of Pennsylvania) for helpful advice with Imaging Core for assistance expert with intracranial implantations. We thank B. Gordon, M. Johnson, S. Brown and C. Calabrese in the St. Jude Small Animal We thank the St. Jude Children’s Research Hospital Tissue Resource Facility and onli Note: Any Supplementary Information and Source Data files are available in the accession hosted under (EBI), is EGAS0000100019 Institute which Bioinformatics European (EGA), the the at by Archive deposited been Genome-phenome have European data transcriptome and sequencing codes. Accession the in v available are references associated any and Methods Me n p URLs. targets— tumors and identifies high-frequency mutations in potential therapeutic 9. 8. 7. 6. 5. 4. 3. 2. 1. reprints/index.htm at online available is information permissions and Reprints The authors declare no competing interests.financial wrote the manuscript with contributions from D.W.E., J.R.D. and M.R.T. completed all pathological evaluations. S.J.B., Jinghui Zhang, G.W. and A.K.D. performed experiments, analyzed data or prepared tables and figures. D.W.E. R.S.F., L.L.F., L.D., R.K.W.,E.R.M., M.R.T., J.R.D., D.W.E., Jinghui Zhang and S.J.B. A.O.-T., M.P.,R.K., R.H., P.G., J.B., L.W., H.L.M., K.B., B.V., D.Y., J.C.R., K.O., X.Z., C.Q., X.C., Junyuan Zhang, J.E., M.E., X.M., C.L., P.N., E.H., S.P.,M.R., T.L., and C.J. provided samples or clinical data. G.W., A.K.D., B.S.P., B.J.,S.L.R., Y.L., D.W.E. research.experiments designed or A.G., supervised A.B., C.W., F.A.B. S.J.B., Jinghui Zhang, A.K.D., B.S.P., J.E., L.D., R.K.W.,E.R.M., M.R.T., J.R.D. and Tyler’s Treehouse and Musicians AgainstCancer. Childhood Cure Starts Now Foundation and the Smile for Sophie Forever Foundation, and by CA096832 to S.J.B., Jinghui Zhang and D.W.E. and R01 CA135554 to S.J.B.), by the Research Hospital, by grants from the US National Institutes of Health (P01 American andLebanese Syrian Associated Charities (ALSAC) of St. Jude Children’s Hospital–Washington University Pediatric Cancer Genome Project, by the antibody This selection. work was supported by the St. Jude Children’s Research A COMPETING FINANCIAL INTERESTS FINANCIAL COMPETING AUTHOR CONTRIBUTIONS ersion of the pape the of ersion c c cknowledgments

g i ne version of the pape ag, B.S. Paugh, B.S. Paugh, H.Q. Qu, D. Faury, D.A. Khuong-Quang, G. Wu, J. Schwartzentruber, Pollack, I.F. childhood. of tumors brain malignant for Chemotherapy Gajjar,A. & N.G. Gottardo, loa rvas e dfeecs ih h aut disease. adult the with (2010). 3061–3068 differences key reveals gliomas glioma. pontine intrinsic diffuse in genes regulatory cell-cycle and kinases tyrosine receptor 12 glioblastomas. pediatric in imbalances chromosomal novel identifies (2007). 1196–1208 tumors. in expression YB-1 increased shows and glioblastoma Neuropathol. gliomas. pontine intrinsic diffuse pediatric of subgroups distinct biologically glioblastomas. non-brainstem and gliomas glioblastoma. paediatric in genes remodelling multi-institutionalcohort.a in results Neurol. Child J. f th p c e r , 153–163 (2010). 153–163 , x f Pediatric Cancer Genome Project (PCGP) Explorer, Explorer, (PCGP) Project Genome Cancer Pediatric . o p g l o ACVR1 ds o et al. et J. Clin. Oncol. Clin. J. v r et al. et / e t al. et . . et al. o

Somatic histone H3 alterations in pediatric diffuse intrinsic pontine intrinsic diffuse pediatric in alterations H3 histone Somatic t al. et 124 r t al. et Genome-wide profiling using single-nucleotide polymorphism arrays polymorphism single-nucleotide using profiling Genome-wide g l . a in DIPGs and Molecular profiling identifies prognostic subgroups ofpediatric subgroups prognostic identifies profiling Molecular / 23 Age and ; ; DAVID Bioinformatics Resource, , 439–447 (2012). 439–447 , DVANCE ONLINE PUBLICATION ONLINE DVANCE Integrated molecular genetic profiling of pediatric high-grade pediatric of profiling genetic molecular Integrated 2 , 1149–1159 (2008). 1149–1159 , All whole-genome sequencing, whole-exome whole-exome sequencing, whole-genome All Genome-wide analyses identify recurrent amplifications of amplifications recurrent identify analyses Genome-wide r . . et al. et r

. t al. et 29 TP53 , 3999–4006 (2011). 3999–4006 , K27M mutation in histone H3.3 defines clinically and clinically defines H3.3 histone in mutation K27M rvr uain i hsoe 33 n chromatin and H3.3 histone in mutations Driver mutation frequency in childhood malignant gliomas: NTRK Cancer Res. Cancer fusions in NBS-HGGs in infants. Nat. Genet. Nat. Nature

, 7404–7407 (2001).7404–7407 , 482

, 226–231 (2012). 226–231 , , 251–253 (2012). 251–253 , http://www.nature.c Nature Ge Nature h . ln Oncol. Clin. J. t t J. Clin. Oncol. Clin. J. p : / / d Neuro-oncol. a v i n d online online h etics . t a

b p

Acta om/ 28 25 c : / c / . , ,

Jones, D.T.Jones, J. Zhang, pathway transduction signal TrkB Trk—the On A.E. McKee, & Z. Li, C.J., Thiele, Harel, L., Costa, B. & Fainzilber, M. On the death Trk. J. Lee, S.G. Piccirillo, Post-transcriptional M.F. Roussel, & J.H. Kim, F., Zindy, O., Ayrault, H., Zhao, morphogenetic bone of role dynamic Kessler,The & J.A. O.G. Bhalala, A.M., Bond, Cannon, J.E., Upton, P.D., Smith, J.C. & Morrell, N.W. Intersegmental vessel formation Q. Shen, J. Bagarova, A. Chaikuad, E.M. Shore, heterotopic in transduction signal altered of Role F.S. Kaplan, & E.M. Shore, D. Sturm, J. Barrow, D.A. Bax, characterization genomic Comprehensive Network. Research Atlas Genome Cancer D.W. Parsons, M. Zarghooni, S. Puget, J.D. Schiffman, astrocytoma. gliomas. low-grade biology. cancer in target (2009). 5962–5967 important increasingly an is (2010). (2008). 69–80 cells. glioblastoma-initiating of differentiation inhibits pathway cells. tumour–initiating brain suppresses human of proteins morphogenic bone by development. medulloblastoma Atoh1/Math1 of down-regulation (2012). maturation. and fate cell stem neural in proteins non-selectiveantagonists.BMP zebrafish:in requirement signalling BMP not but VEGF for revealed selective by and Invest. ventralization. Clin. J. embryo zebrafish and chondrogenesis BMP-independent activates cooperation. receptor progressiva. (2012). 36990–36998 ossificans fibrodysplasia for implications and progressiva. ossificans (2006). 525–527 fibrodysplasia sporadic and inherited (2011). 83–88 progressiva. ossificans fibrodysplasia and ossification glioblastoma. of subgroups biological and 13 gliomas. pontine intrinsic diffuse and glioma high-grade pediatric of gliomas. grade (2008). core pathways. and genes glioblastoma human defines multiforme. targets. (2010). therapeutic potential as polymerase receptor factor growth platelet-derived highlights gliomas ONE gliomas. pontine intrinsic diffuse pediatric in events oncogenic distinct key astrocytomas. (2010). 512–519 malignant pediatric of subset a of characteristic , 212–222 (2011). 212–222 ,

7 , e30313 (2012). e30313 , et al. et et al. et et al. et n et al. et t al. et et al. et t al. et et al. et etics Science et al. et Epigenetic-mediated dysfunction of the bone morphogenetic protein morphogenetic bone the of dysfunction Epigenetic-mediated Nat. Genet. Nat. t al. et t al. et

Whole-genome sequencing identifies genetic alterations in pediatric in alterations genetic identifies sequencing Whole-genome t al. et et al. et The fibrodysplasia ossificans progressiva R206H ACVR1 mutation ACVR1 R206H progressiva ossificans fibrodysplasia The Clin. Cancer Res. Cancer Clin. A distinct spectrum of copy number aberrations in pediatric high- pediatric in aberrations number copy of spectrum distinct A 119 Hotspot mutations in mutations Hotspot eecya tasto and transition Mesenchymal t al. et ooyos os of loss Homozygous Recurrent somatic alterations of alterations somatic Recurrent t al. et A recurrent mutation in the BMP type I receptor I type BMP the in mutation recurrent A

osiuiey cie L2 eetr uat rqie ye II type require mutants receptor ALK2 active Constitutively Nat. Genet. Nat. tutr o te oe opoeei poen eetr ALK2 receptor protein morphogenetic bone the of Structure , 3462–3472 (2009). 3462–3472 , Bone morphogenetic proteins inhibit the tumorigenic potential tumorigenic the inhibit proteins morphogenetic Bone

Whole-genome profiling of pediatric diffuse intrinsic pontine intrinsic diffuse pediatric of profiling Whole-genome 321 ADVANCE ONLINE PUBLICATION ONLINE ADVANCE n nertd eoi aayi o hmn glioblastoma human of analysis genomic integrated An Mol. Cell. Biol. Cell. Mol. Oncogenic

, 1807–1812 (2008). 1807–1812 , 45 , 927–932 (2013). 927–932 ,

Br. J. Pharmacol.Br.J. Genes Dev. Genes 45

16 , 602–612 (2013). 602–612 , BRAF , 3368–3377 (2010). 3368–3377 ,

33 ADAM3A H3F3A Nature , 2413–2424 (2013). 2413–2424 , uain with mutation Cancer Cell Cancer

22 and PDGFRA , 722–727 (2008). 722–727 ,

eeld y eoewd analysis genome-wide by revealed 444 . ln Oncol. Clin. J.

161 Dev. Neurobiol. Dev. FGFR1 IDH1 , 761–765 (2006). 761–765 , Dev. Neurobiol. , 140–149(2010).,

22 amplification/mutation are amplification/mutation ur Otooo. Rep. Osteoporos. Curr. define distinct epigenetic distinct define Nature , 425–437 (2012). 425–437 , and CDKN2A α ln Cne Res. Cancer Clin. and poly (ADP-ribose) poly and . il Chem. Biol. J.

NTRK2

455 a. Genet. Nat.

28 acr Res. Cancer acr Cell Cancer 72 nciain is inactivation ACVR1 1337–1344 ,

1061–1068 , , 1068–1084 , 70 Neuro-oncol. in pilocytic in , 298–303 causes

PLoS

287

15 13 38 70

9 , , , , , ,

45. 43. 42. 41. 40. 39. 38. 37. 36. 35. 34. 33. 32. 55. 54. 53. 52. 51. 50. 49. 48. 47. 46. 44.

ülrb, . Kvng, . Jsp, . itn onco-modifications. Histone B. Joseph, & E. Kavanagh, J., Füllgrabe, ent, S. Venneti, P.W.Lewis, K.M. Chan, A.M. Fontebasso, Li, G. D.A. Bax, N. Cetinbas, M.G. Butti, M. Eguchi, Z. Li, kinase tyrosine protein chimeric a ETV6-NTRK3: P.H. Sorensen, & C.L. Lannon, V. Frattini, lilv, P. Kleiblova, P.J. Killela, P.J. Stephens, C.Y. Lin, Z. Nie, T.J. Pugh, D.T.Jones, Dubuc, A.M. G. Robinson, J. Zhang, Bender,S. 660–672 (2013). 660–672 558–564 (2013). 558–564 glioblastomas. mutant shows K27M H3F3A tumors in glioneuronal H3K27me3 and decreased glial pediatric in (EZH2) 2 zest of enhancer and glioblastoma. pediatric in found expression. gene and methylation H3K27 gliomas. high-grade (2013). 659–669 hemispheric target methylation gliomas. pontine intrinsic diffuse (2009). 5753–5761 lines. cell glioma pediatric in therapy targeted to response transformation. cellular Chem. induced ETV6-NTRK3 blocks interface domain carcinomas. thyroid papillary between rearrangements t(12;15)(p13;q25). with leukemia (2007). complex. AP1 of activation via progenitors mammary lineages. (2005). cell 215–223 multiple in activity transformation with glioblastoma. p53-dependent G1 checkpoint. G1 p53-dependent USA Sci. self-renewal. of rates low with cells from derived tumors of subset development. cancer during event catastrophic Cell cells. stem embryonic mutations. somatic Nature (2013). medulloblastoma. in subgroups across occur Nature leukaemia. 30 gliomas. high-grade pediatric mutant K27M in expression gene of , 3391–3403 (2011). 3391–3403 ,

151 et al. et al. et

288 488 488 et al. et , 56–67 (2012). 56–67 ,

t al. et 110 t al. et t al. et ExpressionofepidermalgrowthfactorvariantIII(EGFRvIII)inpediatric et al. et , 27940–27950 (2013). 27940–27950 ,

et al. et Nature t al. et , 100–105 (2012). 100–105 , (2012). 43–48 , et al. et et al. et et al. et ETV6 t al. et t al. et c-Myc is a universal amplifier of expressed genes in lymphocytes and lymphocytes in genes expressed of amplifier universal a is c-Myc t al. et t al. et et al. et al. et et al. et Nat. Genet. Nat. , 6021–6026 (2013). 6021–6026 , t al. et Transcriptional amplification in tumor cells with elevated c-Myc. with elevated cells tumor in amplification Transcriptional GRII eein uain i pdarc ihgae loa and glioma high-grade pediatric in mutations deletion EGFRvIII h gntc ai o ery -el rcro aue lymphoblastic acute precursor T-cell early of basis genetic The Reduced H3K27me3 and DNA hypomethylation are major drivers major are hypomethylation DNA and H3K27me3 Reduced Fusion of Fusion t al. et Inhibition of PRC2 activity by a gain-of-function H3 H3 mutation H3 H3 gain-of-function a by activity PRC2 of Inhibition Dissecting the genomic complexity underlying medulloblastoma. underlying complexity genomic the Dissecting The histone H3.3K27M mutation in pediatric glioma reprograms glioma pediatric in mutation H3.3K27M histone The eulbatm eoe eunig noes subtype-specific uncovers sequencing exome Medulloblastoma - vlain f itn 3 yie 7 rmtyain (H3K27me3) trimethylation 27 lysine 3 histone of Evaluation Aberrant patterns of H3K4 and H3K27 histone lysine methylation NTRK3

et al. et sqec aayi o te eoi rgos novd n the in involved regions genomic the of analysis sequence A Mutation of the salt bridge–forming residues in the ETV6 SAM ETV6 the in residues bridge–forming salt the of Mutation h itgae lnsae f rvr eoi atrtos in alterations genomic driver of landscape integrated The Novel mutations target distinct subgroups of medulloblastoma. of subgroups distinct target mutations Novel 481 TERT Nature ano-ucin uain o PMDWp ipi the impair PPM1D/Wip1 of mutations Gain-of-function asv gnmc eragmn aqie i a single a in acquired rearrangement genomic Massive Cell , 157–163 (2012). 157–163 , Mutations in Mutations rmtr uain ocr rqety n loa ad a and gliomas in frequently occur mutations promoter fusion oncogene initiates breast cancer from committed from cancer breast initiates oncogene fusion

TPM3 45

ETV6 488 151 , 1141–1149 (2013). 1141–1149 , Genomics and , 106–110 (2012). 106–110 , , 68–79 (2012). 68–79 , to neurotrophin-3 receptor neurotrophin-3 to J. Cell Biol. Cell J. Science J. Neurooncol. J. Blood NTRK1 SETD2

340 93 Genes Dev. Genes , 15–24 (1995). 15–24 ,

, 1355–1363 (1999). 1355–1363 , ee producing genes 201 and genes affecting histone H3K36 histone affecting genes and , 857–861 (2013). 857–861 , ca Neuropathol. Acta Cell

, 511–521 (2013). 511–521 , 108

144 , 395–402 (2012). 395–402 ,

27 acr Cell Cancer , 27–40 (2011). 27–40 , ei. acr Biol. Cancer Semin. ca Neuropathol. Acta , 985–990 (2013). 985–990 , TRKC Clin. Cancer Res. Cancer Clin. s r e t t e l TRK Brain Pathol. Brain in acute myeloid acute in Proc. Natl. Acad. Natl. Proc.

Cancer Cell Cancer 125

noee in oncogenes 12 373–384 , 542–558 , Oncogene . Biol. J.

125

24 23 15 15 , , , , ,

© 2014 Nature America, Inc. All rights reserved. subjected to exome sequencing, and the overlapping SNVs and indels were were indels and SNVs overlapping the and sequencing, exome to subjected samples (SJHGG003_A, SJHGG008_A, SJHGG019_E and SJHGG022_D) were sequencing whole-genome non-hypermutator Four (167/182). 92% was rate samples, we validation performed with the MiSeq platform, and the validation by validated also custom capture. were For indels found sequencing in all gene-coding whole-exome cohort sequencing whole-exome the in SNVs 1 tier All rate sample. of per 95% validation a with median rate 93%, was validation by The overall a capture were platform. custom validated samples sequencing whole-genome non-hypermutator in discovered variations) structural and SNVs. remaining the 4: tier (iv) and regions; non-repeat-masked in variants 3: tier (iii) cpgIslandExt)); and firstEF, L1 TAF1 Valid, laminB1 Poly(A), switchDbTss, encodeUViennaRnaz, included are ORegAnno, targetScanS, eponine, vistaEnhancers, tfbsConsSites, annotations by (regulatory annotated regions annotation UCSC in regulatory phastConsElements17way table from the UCSC Genome Browser, and variants score conservation (cutoff: variants conserved 2: tier (ii) RNA variants; noncoding described ferent sequencing platforms are summarized in the mapping of all samples. Mapping statistics and previouslycoverage for each tumor on dif ous effects of missense mutations and identification of LOH have been described indel detection, tier annotation for sequence mutations, prediction of the deleteri 100-bp a with length. read platform HiSeq or IIx Analyzer Genome Illumina the using described previously as formed Whole-genome whole-exome sequencing, sequencing and RNA-seq were per and analysis. sequencing and transcriptome whole-exome Whole-genome, described previously as was cases DIPG reviewed byof a pediatric neuro-oncologist(MRI) (A.B.). DNA imaging and RNA were resonance extracted magnetic and ropathologist, (SJHGG003_A). DIPG independent subsequent, a and pendent tumors, a hemispheric malignant glioneuronal tumor (SJHGG003_D) inde two developed case One SJHGG093_D/A). and SJHGG002_D/A D/A, and and and from SJHGG115_D/R) autopsy diagnosis in (SJHGG001_ three cases SJHGG112_D/R SJHGG033_D/R, SJHGG024_D/R, (SJHGG019_E/S, discovery. variant by structural for only RNA-seq characterized were NBS-HGGs) 9 and DIPGs (3 tumors 12 addition, In RNA-seq. by characterized were NBS-HGGs) 44 and DIPGs (31 75 tumors, sequenced by both whole-genome non-hypermutator and whole-exome sequencing. Among these four and tumors (SJHGG003_A, SJHGG111_D) SJHGG008_A, SJHGG019_E and SJHGG022_D), were and (SJHGG003_D hypermutator two tumors including samples, germline matched their and tumors ( sequencing whole-exome using and 22 NBS-HGGs) and a cohort for evaluating the frequency of abnormalities viduals in 2 cohorts: a cohort for whole-genome ( sequencing indi 118 from samples) germline matching with NBS-HGGs 54 and DIPGs ( cases NBS-HGG male sex ciation and between tumor with subtype, 63% female DIPG and cases 63% mation are in provided infor sequencing and clinicopathological Detailed approval. ethics research the of Board Group Children’s UK Leukemia and the and Hospital Cancer Marsden Development Royal and Research Clinical the and (PCGP) Project Jude St. with Genome Cancer Pediatric accordance the for approval (IRB) in Board Review Institutional analysis sequence genome for approval with Hospital Marsden Research/Royal Cancer of Institute the from and facility core resource tissue Jude Children’sSt. Hospital the from Research requested details. sample and cohorts Subject O Nature Ge Nature thus regarded as validated. Validated and high-quality variations for mutations NLIN All tier 1, tier 2 and tier 3 sequence mutations (including SNVs, indels indels SNVs, (including mutations sequence 3 tier and 2 tier 1, tier All previously as tiers, four following the into classified were SNVs Whole-genome sequencing mapping, coverage and quality assessment, SNV and neu experienced an D.W.E., by reviewed centrally was Histopathology cases five in relapse and diagnosis from available was tissue Tumor 54 and 127 tumors (57 DIPGs 70 NBS-HGGs; comprised The study cohort ≥ 500, on the basis of either the phastConsElements28way table or the the or table phastConsElements28way the either of basis the on 500, E E 46 46 M , , 5 5 n 7 7 . The reference human genome assembly GRCh37-lite was used for : : (i) tier 1: synonymous, coding nonsynonymous, and splice-site E etics TH O DS Supplementary Table 1 Supplementary 5 P 6 = 0.004). = . 46 n , 5 = 80, 36 DIPGs and 44 NBS-HGGs). Six Six NBS-HGGs). 44 and DIPGs 36 80, = 7 . Paired-end sequencing was performed performed was sequencing Paired-end . HGGs (WHO grade III or IV) were were IV) or III grade (WHO HGGs Supplementary Table 2 . There was a significant asso . There was a significant n = 42, 20 DIPGs . ------

assembly to identify structural variation in variation RNA-seq structural to assembly data (Y.L.,identify B. Tang, M.R., variation uses that algorithm Structural novel a CICERO, with out carried databases. was detection four the in alignment best the selecting by constructed was BAM file final The coordinates. genome to reference human were aligned from human ESTs.(ii)–(iv) from The databases mapping results database flat file downloaded from UCSC representing transcripts constructed AceView the (iv) and exons RefSeq in pairs non-sequential of combinations lite reference sequence, (ii) RefSeq, (iii) a sequence file representing all possible (BWA Aligner 0.5.5) Burrows-Wheeler using files in reported are results The samples. ing used to validate somatic structural variations found in whole-genome sequenc described as previously and were annotated CREST in reported are results The data). algorithm, M.P.,Patel novel S. Roberts, Wang,J. K. X.C., a Sequencing; Next- Gen in Tree using Regression by sample Segmentation normal Number (Copy matching CONSERTING its and tumor each website. Explorer PCGP at the deposited been have tumors hypermutator in mutations 1 tier for variations high-quality and Validated validated. been having as regarded were sequencing whole-exome and whole-genome both in found SNVs validation. as served sequencing exome SJHGG111_D), and Tables 3 in are summarized tumors in non-hypermutator 1–3 in tiers PKN2 PLCG2 ands: (PIP (3,4,5)-triphosphate FLT1 RTKs: (i) cohort: study limited our in mutation and somatic one least at had that genes Database following the to analysis Interaction Protein (NCI)-Nature Institute National Cancer Kyoto the and the Ingenuity in (KEGG), Genomes genes and Genes of relevant Encyclopedia the identified we pathways, RTK-RAS-PI3K pathways. in mutations of Frequency Resource. DAVIDthe Bioinformatics using evaluated was genes selected the among terms (GO) Ontology Gene of (limma/R; models linear with were selected score. (MAD) without and deviation with DIPGs in absolute genes expressed median Differentially of basis the on selected variable probes most 1,000 top the with GeneMaths using performed was analysis with 9 samples, DIPG 32 including data. number copy Figure of deletions as focal MYCN of amplifications Focal selected. were deletion or gain focal of targets candidate and genes, glioma-related and/or cancer- for inspected manually were genes 60 than fewer with regions remaining All known removed. with were CNVs associated Regions deletion. or gain focal single a as classified were that or tumors two least at in found <1.7) number (copy recurrent deletion or focal >2.3) number (copy gain focal recurrent for regions common described as loss or gain focal of targets date candi identified also we number samples, DIPG copy For 0.8. with than less or 5 regions than greater common minimum from identified were tion as described were analyzed Microarray copy number and expression analysis. Table 8 Supplementary in reported struc are RNA-seq by found variations putative structural The containing variations. tural regions genomic the bracket to Primer3) (with designed were pairs Primer sequencing. MiSeq with validated were data seq K.B. J.E., Paired-end reads from RNA-seq were aligned to the following four database Structural variations in whole-genome sequencing data were analyzed using between depth read in difference the evaluating by identified were CNVs (SJHGG003_D samples sequencing whole-genome hypermutator For two Affymetrix U133v2 expression array data were available for 71 HGGs, HGGs, 71 for available were data array expression U133v2 Affymetrix , , FGF3 , , IGF1R , , , , PPP2R2D 1 PDGFRA – PLEKHA2 because many samples sequenced only over the exome lacked these these lacked exome the over only sequenced samples many because et al et 5 . , , FGF5 , , ., unpublished data). The structural variants detected in RNA- in detected variants structural The data). unpublished ., INSR , , , , , , MET PRKAA2 FIGF , , , , CDKN2A YES1 NRTK1 , , PDGFRA , , . CDK4 PDGFA 3 , , ) regulation: regulation: ) 8 SGK1 , , . Candidate targets of focal amplification or dele amplification of focal targets . Candidate , , PRKCA and NTRK2 , , CDK6 and , , , , Supplementary Table 6 Supplementary KIT CDKN2B G6PC ACVR1 , , VEGFC and and , , PRKCZ , , CCND1 PTEN EGFR Supplementary Table 7 Supplementary , , To summarize samples altered in in altered samples summarize To GNB1 NTRK3 mutation P , from SNP data were included in , SNP from were included data 8 value < 0.001), and enrichment and enrichment < value 0.001), ; and (iv) downstream effectors: ; and effectors: (iv) downstream , , , , , , . Briefly, we derived minimum minimum derived we Briefly, . MET PIK3CA RAC1 , , , , CCND2 GNGT1 5 ; (ii) phosphatidylinositol phosphatidylinositol (ii) ; 9 Affymetrix SNP6.0 arrays Affymetrix 46 : (i) the human GRCh37- human the : (i) , , CSF1R 8 , , , 5 . Principal-component Principal-component . 8 RPS6 and and . Custom capture . was capture Custom and doi: , , . et al et MLST8 ACVR1 , , , , PIK3R1 Supplementary Supplementary 10.1038/ng.2938 FGFR1 CCND3 RPTOR ., unpublished unpublished ., . , , mutation mutation PHLPP1 ; (iii) lig (iii) ; , , , , , , as well FGFR3 de novo de SGK3 MYC - - - - - , , , ,

© 2014 Nature America, Inc. All rights reserved. mRNA at the one-cell stage. Embryos were inspected at 24 hours post-fertiliza strain AB or mutant of wild-type pg 20 the approximately with of microinjected were embryos mMESSAGE Zebrafish the Technologies). using (Life kit generated mMACHINE was mRNA and pCS2, into cloned was injections. embryo Zebrafish for mutations missense few a except indels, and SNVs splice-site or nonsense kept only (ii) and repair or kept genes listed in Cancer Gene Consensus or genes involved in DNA damage described analysis. mutation Germline to used be can it outliers. and such detect outliers, by LMS- influenced the be Thus, not will range. interval interquartile identified same the with distribution normal a to values data the comparing by detected are outliers and data, the of densest ‘bulk’ the represents interval This data. the of 50% least at covers that val of the data and outlier detection least median squares (LMS) as a method for the robust estimation of the center Pearson by quantified was NBS-HGG and writers H3 histone in mutations between ( test Wallis rate and tumors in less than children 3 years of age by was quantified Kruskal- 1 mutation tier between association The by test. Fisher’s exact quantified was mutations, test: log-rank by the quantified was DIPG with individuals of vival P test: Kruskal-Wallis by the analyzed was nosis analysis. Statistical Consed using performed was moter SNVs. Sequence was analyzed using SNPDetector in using by listed the primers samples and normal Sanger sequencing sequenced by was PCR amplified from tumors and matched chr. 5: 1,295,151–1,295,347) 1,295,228 whole- 5: (chr. and Therefore, in 1,295,250). a of portion sites the mutated average) recurrently on two the (~3× at sequencing coverage genome poor was there regions, promoter TERT conservative. are pathways these ing UBR2 CHD7 remodelers: matin KDM2A SMYD3 writers: histone (i) genes: following the to our we analysis limited regulators, and chromatin modifications in histone RAD54B RAD23A ATR CCND3 CCND2 genes: following the included we regulation, cycle cell NF1 STK11 doi: (classes dorsalized or V1–V5) (classes ventralized as scored and (h.p.f.) tion P ACVR1 ACVR1 ( cohort this within survival with associated significantly not was diagnosis at and variable) continuous a (as diagnosis at age where DIPG with individuals for freedom, of degrees 1 P = and 0.0000001 = 0.010 (adjusted for ties). Association of of Association ties). for (adjusted 0.010 = = 0.55), but Supplementary Table 13 Supplementary To identify samples with extremely high and low mutation rates, we used used we rates, mutation low and high extremely with samples Toidentify NBS-HGG) versus (DIPG and sex tumor subtype between The association 10.1038/ng.2938 ACVR1 , , , , ATM BRAF and promoter mutation analysis. mutation promoter , , , , mutation co-occurrence with mutations with other mutations was quantified by Fisher’s exact test: test: by Fisher’s exact quantified was mutations other with mutations CHD8 INPP5D , , , , , , , , χ , , CDK1 RBL1 , , SETD1A KDM3A 6 RUVBL2 2 XAB2 3 UBR5 , , (by Monte Carlo), Monte (by Carlo), P . . In this study, we implemented additional filters, where we (i) only mutation status were included simultaneously, it appeared that age , , BRCA1 H = 0.0049702. = RASGRF2 and and = 13.42, 1 degree of freedom, freedom, of degree 1 13.42, = ACVR1 and , , CDK6 , , , , . . Therefore, our estimates of the numbers of mutations affect MSH6 SOS1 , , , , ACVR1 , , KDM3B ATRX SETD2 CDC27. RAD54B , , SETX Association between between Association RBBP8 mutation status was ( P , , , , CDKN1B and and = 0.00149. On the basis of a two-variable Cox model model Cox two-variable a of basis the On 0.00149. = , , AKT3 LIG1 , , , , mutation co-occurrence with mutation co-occurrence SMARCA4 , , PMS2 , , For DNA we genes, repair the following: included , , SETD3 to check for two previously described described for to check two previously TP53 ; and (iv) other writers: writers: other (iv) and ; KDM4B RAD51 RASSFS 6 , , , , P We identified germline variants as previously previously as variants germline We identified cDNA for wild-type or cDNA mutant human for wild-type 1 LIG3 JAK1 . = 0.0007. = 6 2 , , and . . . Basically, the inter shortest LMS identifies CDKN2A and and , , , , , , HIST1HB , , Owing to the high GC content of of content GC high the to Owing . To summarize the samples altered in in altered samples the summarize To . LIG4 BRCA2 , , UVRAG KRAS KDM5B BRWD1 ASH1L MLL , , TERT ACVR1 , , POLD1 , , P CDKN2B MAP2K1 , , = of The 0.0046). co-occurrence , , ACVR1 . . To altered samples summarize , , ERCC2 mutation encoding p.Lys27Met, MLL2 H , , KDM5C ; (ii) histone eraser: eraser: histone (ii) ; P CBX4 promoter (hg19 coordinates, = 6.62, 1 degree of freedom, freedom, of degree 1 6.62, = < 0.0001). The association association The 0.0001). < mutation and age at diag at age and mutation , , POLE , , UHRF1 mutation with the sur the with mutation , , , , , , MLL3 , , ERCC3 CDKN2C CHD2 TSC2 TP53 and 6 PIK3CA 0 , , and manual review , , RAD23A , , SIRT7 , , , , , , MLL4 , , NCOA1 , , GAB2 TP73 CHD4 ERCC8 , , and RB1 ; (iii) ; chro (iii) TERT , , , , χ , , , , JMJD1C PRDM9 CCND1 2 , , PPM1L RAD50 , , PIK3R1 ACVR1 ACVR1 = 10.1, 10.1, = , , CHD6 , , CDK4 STK4 TERT LIG4 pro ------, , , , , , , , ,

nodeficient mice at 6–16 weeks of age as described as age of weeks 6–16 at mice nodeficient mutant or ACVR1, or TPM3-NTRK1 wild-type BTBD1-NTRK2 and implanted into human female immu expressing vectors MIC from generated ruses studies, Tp53 tumorigenesis For cultures. primary among variation potential for isolated from multiple mice were pooled for retr described as previously retroviruses with duced phosphorylated phosphorylated SMAD1/5 (Cell Signaling Technology, 9516; 1:1,000 dilution), transferred to nitrocellulose membranes and detected with antibodies targeting (20 Protein (Roche). tors (1 medium growth astrocyte standard in incubated were cells LDN-193189, with treatment For factors. growth or serum without for 2 h medium in incubated and then PBS with twice washed 2.5 with selected MIP with retroviruses one expressing wild-type or passage mutant at ACVR1. After transduced 48 h, and cells were brainstem neonatal from isolated were blotting. Protein cohort. entire the in of ACRV1-driven and evidence no tumor formation histological was detected tronic message with ACVR1. None of the mice showed brain tumor symptoms, for immunohistochemistry the marker mCherry expressed as part of the bicis by and site and implantation the surrounding hematoxylin area the in of sections analysis eosin–stained by both growth tumor brain ACVR1-driven for evaluated and eosin, ad hematoxylin with stained paraffin and sectioned, fixed embedded, formalin were Brains implantation. after d 222 and 219 between euthanized were mice remaining The implantation). after ACVR1 d 212 at wild-type expressing one and implantation, after d 190 and 139 at Gly356Asp expressing 2 implantation, after d 217 at ACVR1 Arg258Gly ing express one implantation, after d 97 at ACVR1 symptoms Gly328Glu tumor expressing (one brain showing without ill became they when euthanized Gly356Asp, six mice were implanted. A few of for the mice immunodeficient were and, construct, each for implanted or were mice Arg258Gly seven ACVR1, Arg206His Gly328Glu, or ACVR1 wild-type control, vector construct. empty each For for implanted were mice seven controls, vector empty and (VectorBTBD1-NTRK2 hematoxylin with For stained Labs). TPM3-NTRK1, counter and Labs) (Vector substrate NovaRED with detected Labs), ABC, Vector (Elite antibod streptavidin (HRP)-conjugated secondary peroxidase horseradish biotinylated and ies dilution), 1:100 F1804; Flag (Sigma-Aldrich, or dilution) 1:400 4270; Technology, Signaling (Cell MAPK (Thr202/Tyr204) p42/44 phosphorylated (Cell dilution), 1:50 (Ser473) 9271; AKT Technology, Signaling phosphorylated against antibodies primary antigen retrieval, microwave with performed was Immunohistochemistry (D.W.E.). neuropathologist by a clinical evaluated were tumors collected all from tions 5- embedded into cut and paraffin processed, in then were Tumors overnight. °C 4 at PBS in dehyde paraformal 4% in fixed and dissected were tumors symptoms, tumor brain srcts ee sltd rm h cre o bantm f otaa dy 2 day postnatal of brainstem or GFAP-cre cortex the from isolated were astrocytes guidelines. institutional and national with ance compli in are and Committee Use and Care Animal Institutional Hospital mycoplasma. for negative tested (ATCC) and previously as plasmids helper with described along constructs viral these with cells 293T HEK by transfecting was produced Retrovirus were verified. constructs (MIP) MSCV-IRES-Puro or site, cloning Gateway a erate to gen site EcoRI a into blunted an adaptor by inserting modified was which (MIC) ments were cloned into MSCV-IRES-mCherry vectors the retroviral or mutant wild-type human of ORF full the encoding cDNA of codon termination the before immediately PCR by incorporated was tag Flag a encoding and sequence transduction astrocyte Primary h.p.f. 3 at about embryos to added was (Stemgent) LDN-193189 criteria published of basis the on C1–C4) µ Mouse experiments were approved by the St. Jude Children Jude St. the by approved were experiments Mouse M). Cells M). were Cells lysed in RIPA buffer with protease and phosphatase inhibi -null primary astrocytes at passage one were transduced with retrovi with transduced were one passage at astrocytes primary -null ; ; 6 ACVR1 6 Trp53 . HEK 293T cells were from the American Tissue Type Collection TypeTissue Collection American the were from cells . 293T HEK loxP µ For protein blotting, blotting, protein For , , g/ml puromycin for 48 h. For serum starvation, cells were were cells starvation, serum For h. 48 for puromycin g/ml TPM3 / loxP ( - µ NTRK1 Tp53 µ g) was resolved on 4–12% NuPAGE Bis-Tris gels, NuPAGE gels, Bis-Tris 4–12% on resolved was g) m sections. Hematoxylin and eosin–stained sec eosin–stained and Hematoxylin sections. m conditional knockout) mice knockout) conditional and 6 9 with vehicle (DMSO) or LDN-193189 LDN-193189 or (DMSO) vehicle with BTBD1 Tp53 n vivo in 22 , 2 -null primary mouse astrocytes astrocytes mouse primary -null - 3 NTRK2 . For chemical rescue, 2.5 2.5 rescue, chemical For . o 6 viral viral transductions to control 9 uoieei assays. tumorigenesis . Primary astrocyte cultures cultures astrocyte . Primary Tp53 7 0 . When mice displayed displayed mice When . , , and the resulting frag -null primary mouse mouse primary -null Nature Ge Nature 6 5 . Sequences of all all of Sequences . 67 , 6 8 and trans and ’ s Research Research s n etics µ A 6 M 4 ------,

© 2014 Nature America, Inc. All rights reserved. inter- and intrachromosomal events with a minimum of five structural structural five of significant minimum A a with variations. events for separately test intrachromosomal and inter- goodness-of-fit tail- the or applied we rearrangement rearrangement), head-to-head to-tail type duplication, tandem variation any was deletion, structural by (i.e., there categorized whether joints evaluate fragment clustering ToDNA the in bias supports breakpoints. variation distribution random structural distribution. of from random of departure hypothesis in significant null A the breakpoints from variation departed tumor structural of each distribution distribution the whether exponential for assess test to goodness-of-fit Bartlett’s performed we in tumors 13 the For feature. this evaluate to was not decided we (iii)) analysis, own Campbell’s and (criterion Korbel in even order valid, entirely not fragment chromo DNA of derivative the randomness walk to Because some. ability (iv) and order; fragment DNA randomness of (iii) clustering joints; (i) fragment DNA of randomness chromothripsis: (ii) breakpoints; assessing of for criteria potential four Campbell posed and Korbel recently, Most study. this in states CNV tumors HGG main 15 2 between oscillation was initial The event. criterion single a in occurring repair and shattering by chromosome analyzed tumors HGG sequencing. in whole-genome chromothripsis of evaluation Statistical in listed are Primers sequencing. Sanger by analyzed were products PCR interest. of gene fusion the identify to PCR for used and RNA tumor from transcription reverse by RT-PCR of validation Biosciences). (LI-COR System Imaging Odyssey the with obtained were Images Scientific). Thermo West Dura, (SuperSignal substrate chemiluminescent to exposed were membranes antibody, ondary the with appropriate HRP incubation After dilution). sc-23948; 1:1,000 Biotechnology, Cruz (Santa tubulin and (Sigma-Aldrich, dilution) Flag 1:1,000 F1804; dilution), 1:1,000 9212; Technology, Signaling (Cell phorylated p38MAPK (Invitrogen, 44-684G; 1:1,000 dilution), total p38MAPK phos dilution), 1:500 sc-6031; Biotechnology, Cruz (Santa SMAD1/5/8 total Nature Ge Nature n etics Supplementary Table 13 Supplementary NTRK P value suggests biased fragment joints, which which joints, fragment biased suggests value fusions. Chromothripsis was described as localized localized as described was Chromothripsis Random-primed cDNA Random-primed was generated . Supplementary Table 12 Table Supplementary 5 3 and was found in in found was and - conjugated sec conjugated 7 1 pro - - - - , 66. 65. 71. 70. 69. 68. 67. 64. 63. 62. 61. 60. 59. 58. 57. 56. the in included are analysis evolution frag DNA for joints. distribution ment random the of assessment lower the conservative more reported we available, intrachromo and were inter- data both somal When chromothripsis. support not would Details of tumor purity and tumor heterogeneity estimations and of tumor estimations of and tumor tumor purity heterogeneity Details

Persons, D.A. Persons, D. Kawauchi, obl JO & apel PJ Ciei fr neec o crmtrpi i cancer in chromothripsis of inference for Criteria P.J. Campbell, & J.O. Korbel, Hay,D.W.Endersby,X., Ellison, Zhu, functions N., Baker,R., Nonredundant & S.J. M.M. Fraser, J. Jonkers, Bajenaru, M.L. zmzk A.L. Szymczak, L. Holmfeldt, Rousseeuw,P. Regression. Squares Median Least P.Green, & C. Abajian, D., Gordon, finishing. sequence for tool graphical a Consed: J. Zhang, Burrows-Wheeler with alignment read short accurate and Fast R. Durbin, & H. Li, J. Wang, J. Zhang, Torchia, E.C., Boyd, K., Rehg, J.E., Qu, C. & Baker, S.J. EWS/FLI-1 induces rapid onset medulloblastoma. (2004). o At sfrs n srct got ad loaeei i a orthotopic an in gliomagenesis genomes. and growth astrocyte in model. transplantation isoforms Akt for astrocytes cancer. breast for model mouse conditional insufficientis for astrocytoma formation. hematopoiesis. normal sl-laig 2 ppiebsd ervrl vector. retroviral peptide–based 2A ‘self-cleaving’ leukemia. (1984). Res. Genome detection. transform. resolution. base-pair analyses. myeloid/erythroidof leukemia mice.in et al. et Nature Cell Nat. Genet. Nat. in vivo in PLOS Comput. Biol. Comput. PLOS Bioinformatics t al. et t al. et t al. et

t al. et 8 et al. et al. et t al. et t al. et CREST maps somatic structural variation in cancer genomes with genomes cancer in variation structural somatic maps CREST 152 , 195–202 (1998). 195–202 ,

t al. et nvl eiolsoa hrp fo gnmc n epigenetic and genomic from therapy retinoblastoma novel A Ndtco: sfwr to fr estv ad cuae SNP accurate and sensitive for tool software a SNPdetector: . 481 yegsi tmr upesr ciiy f RA ad 5 i a in p53 and BRCA2 of activity suppressor tumor Synergistic Cancer Cell Cancer Astrocyte-specific inactivation of the neurofibromatosis 1 gene ( Cancer Res. Cancer , 1226–1236 (2013). 1226–1236 , Enforced expression of the GATA-2 transcription factor blocks factor GATA-2transcription the of expression Enforced tn os ass yetoh ad nrae poieain of proliferation increased and hypertrophy causes loss Pten mue oe o te ot grsie ugop f human of subgroup aggressive most the of model mouse A h gnmc adcp o hpdpod ct lymphoblastic acute hypodiploid of landscape genomic The Nat. Methods Nat. , 329–334 (2012). 329–334 , orcin f ut-ee deficiency multi-gene of Correction

45 Blood Cancer Res. Cancer

, 242–252 (2013). 242–252 , 25 , 1754–1760 (2009). 1754–1760 ,

93 21

64 1 , 488–499 (1999). 488–499 , , e53 (2005). e53 , , 168–180 (2012). 168–180 , , 7773–7779 (2004). 7773–7779 ,

8 71 , 652–654 (2011). 652–654 , Mol.Cell.Biol. Supplementary Note Supplementary Mol.Cell. Biol. , 4106–4116 (2011). 4106–4116 , Nat. Genet. Nat. J. Am. Stat. Assoc. Stat. Am. J. a. Biotechnol. Nat.

22 , 7918–7934, (2007). P

, 5100–5113, (2002). 29 value to represent a represent to value , 418–425 (2001). 418–425 , n vivo in doi: . 10.1038/ng.2938 sn a single a using

22 79 589–594 , , 871–880 , NF1 - - )