© 2014 Nature America, Inc. All rights reserved. A A full list of author affiliations appears at the end of the paper. tumor–related brain of cause main the now are DIPGs and decades, over static remained has survival treatments, improve to efforts tive fatal universally is is which almost DIPG, these cause of cancer-related deaths in childhood leading the and tumors solid of group largest the are tumors Brain pediatric subgroups targets and constitutively gene recurrent subgroups discovering with drives of still or with have DIPGs to can treatment that Diffuse Oren Becher Jason Cain Juliette Hukin Andrew Donson Javad Nazarian Uri Tabori Guillaume Bourque Sameer Agnihotri Misko Dzamba Pawel Buczkowicz activating identifies three molecular subgroups and recurrent Genomic analysis of intrinsic diffuse pontine gliomas Nature Ge Nature Received 24 April 2013; accepted 5 March 2014; published online 6 April 2014;

the the p.Lys27Met

be thought increased

arises

methylation, nearly unraveled ACVR1

genetic selection

DIPGs,

are attributed

ID1 intrinsic

cancer.

activating

n and in and (H3-K27M, molecularly

and etics 80%

that 2 21

of

in , the

8

activating,

drivers we expression , , Cameron Brennan

4 , , King Ching Ho near histone

novel as

20% , the of

ID2 25 17

DIPGs pontine

brainstem

found 11 6

to integrated one

& Cynthia Hawkins expression , Chris , Dunham Chris , , Andrew Morrison therapies

ADVANCE ONLINE PUBLICATION ONLINE ADVANCE 14 1 unique , , Jason R Fangusaro

. mutation the

2 1 00% of therapeutic

, , Julia V Hegert Our

, , Scott Ryall of –

disease, H3. silent

5

3

ACVR1

similar

DIPGs. comprise

to , , Alexandre Montpetit , delicate

these

24 glioma

leading

of

results

1 1

harbor fatality. , Christine , HoemanChristine genetic

of

alteration. the on whole-genome and

and

affecting tumors. children, with to

Mutations

the

downstream

(DIPG)

location

MYCN)

highlight to targets

adult 2 three copy a

The , , Annie Huang

makeup

basis 2 p.Lys27Met limited SMAD , , Mark Barszczyk 17 22

To

failure

15

disease. , , Katrin Scheinemann

However, number , , Mark M Souweidane molecularly the

1 with

is

mutations

. . The most of devastating for

2 of , , Amy Smith

and understand 1 of

, , Peter Lewis

12 in a

distinct –

phosphorylation

activin understanding

assumptions 2 of

3 fatal sequencing this

these , , Matthias A Karajannis ,

3 , ACVR1 no

activin

25 uncovering

. Despite . collabora Despite of this

Recent histone profiling,

effective incurable most

brain DIPGs

5 4

tumors

brain molecular , , Francisco Cordero , receptor 24

2

distinct

signaling ,

were 8 what

, , Patricia Rakopoulos therapies , Kathryn , R Kathryn Taylor

cancer

studies

15

H3.3

cancer, 2 7 that are

, a , , Eric Bouffet , , Jennifer Chan

3 and

, , Yevgen Chornenkyy new

-

18 22 doi:10.1038/ng.293 , , Jean Michaud cohort. Verification of somatic alterations was conducted on all 36 36 all on conducted was alterations somatic of Verification cohort. IDH1 in mutations new in listed are cohort combined the in identified (SNVs) variants single-nucleotide somatic coding (35 DIPGs; Illumina HT-12 v4) data ( data v4) HT-12 Illumina DIPGs; (35 and expression SNP6.0) Affymetrix DIPGs; (45 copy number array), methylation Infinium450k Illumina (28 DIPGs; methylation hensive exome sequencing (Applied Biosystems SOLiD 5500xl)) with compre and 16 whole- 2000) HiSeq (Illumina sequencing whole-genome (20 children. affected for therapies better of development the for critical are data These unknown. remain DIPG in have may landscapes of DIPG and the functional role that histone modifications ers of these tumors that may and mechanisms both genetic be epigenetic important driv suggested has DIPGs in p.Lys27Met) (encoding mutations gene H3 histone of frequent identification Recent strategies. therapeutic own indicating that they may be separate entities biological requiring their counterparts GBM supratentorial pediatric and adult their both from DIPGs pediatric distinguish that levels and expression number (GBMs) glioblastomas adult A number of recent studies have of reported differences at both the copy alterations genetic the targeting agents of use the is trials clinical DIPG of failure the to contributor benefit no shown has chemotherapy far, so and, trol, con symptom only offers but therapy of mainstay the is Radiation acquired. rarely is biopsy tissue a findings; radiological and clinical of combination a on based is DIPG of Diagnosis children. in death , Chris , JonesChris We integrated deep sequencing analysis of 36 tumor-normal pairs pairs of tumor-normal 36 analysis Wesequencing deep integrated 13 or structural rearrangements in in rearrangements structural or 8 4 , , David Zagzag , , Ute Bartels , , Pedro Castelo-Branco 16 9 , , Alan Mackay , Lucy , Lafay-Cousin Lucy 2 , 3 , , Sanja Pajovic 9 6 , , C David Allis 2 10– 19 ACVR1 , 3 , , Shayna Zelcer , , Mathieu Bourgey 1 2 8 . However, the complete genetic and epigenetic , , Jennifer Zuccaro 13 . We did not find mutations in in mutations find not did We . 9 , , Nicholas K Foreman , , Anne E Bendel Supplementary Table 2 Table Supplementary 2 , Louis , Letourneau Louis 7 , , Michael Brudno 16 2 , , Joshua Mangerel , , Sandra Dunn Supplementary Table 1 Table Supplementary FGFR1 20 , , David Ramsay 5 , 2 or or

,

10 MYB s r e t t e l ,

4 in our DIPG DIPG our in 14 . A potential potential A . 17 , including including , , 5 6 , , , 23 BRAF

2 20 , ,

). All All ). , 6– or or 5 9  - - - , .

© 2014 Nature America, Inc. All rights reserved. (median of 0.11 mutations per megabase (range of 0.02–0.19) in the the in 0.02–0.19) of (range megabase per mutations 0.11 of (median subgroups two other the in tumors than rate mutation lower a had genome and sequencing, structural SNP6.0 copy number analysis and whole- both of basis the on genomes silent had subgroup silent the Tumors in ID2. possibly or MYCN targeting therapies from benefit this in subgroup. of Rather, with subtype this potentially DIPG children will effective be not would modifications histone altered geting included group groups, respectively. The top most genes in overexpressed the MYCN and overexpressed group this in Tumors of fication on ampli mothripsis chromosome 2p to leading high-level recurrent chro and histology high-grade hypermethylation, by characterized tumors. these for therapy appropriate of design the for tions ( data clinical and expression structural, mutation, of integration the clustering ( ( ported by multiple analyses, including principal-components analysis ‘MYCN’, ‘silent’ and ( ‘H3-K27M’ subgroups; distinct three in resulted Methods) (Online methylation profiles. mutational and (CNA) alteration number copy expression, subgroups with methylation, different three distinct represents DIPG disease, one be to considered previously although validation cohort of an additional 25 tumors. This analysis shows that, sequencing chip Torrent Ion ( Array Fluidigm using pairs sample ( coefficient = 0.9934, coefficient (cophenetic ( clustering. hierarchical unsupervised by 1 Figure s r e t t e l  of (range megabase per mutations 0.99 to compared cohort silent c a Fig. Fig. Fig. 1 Fig. Supplementary Table 3 Table Supplementary The MYCN subgroup had no recurrent mutations but was instead but instead was mutations no recurrent had MYCN The subgroup Unsupervised subgrouping of DIPG cases on the basis of CpG island 0.5 ID2 0 1 values 2 ). The identification of have ). DIPG will subgroups The key identification implica b �

Cophenetic correlation ), non-negative matrix factorization ( factorization matrix non-negative ), by 2.5-fold and 5-fold relative to the H3-K27M and silent silent and H3-K27M the to relative 5-fold and 2.5-fold by 0.94 0.98 Methylation profiling identifies three molecular subgroups of DIPG. ( DIPG. of subgroups molecular three identifies profiling Methylation Unmethylated Hemimethylated Methylated MYCN d Fig. Fig. 1 ). PC, principal component. principal PC, ). 2.0 MYCN D01 FAP d D29 and and ). Subgroup-specific differences were differences ). supported bySubgroup-specific Silent 3.0 D52 k

groups D09 , , D18 HRSP12 ID2 D26

4.0 D63 ). Further, for key SNVs, we also tested a tested also we SNVs, key for Further, ).

( D19

Table D62 D60 and and 5.0 D06 k Fig. Fig. 1 = 3) ( = 3) D57

1 D28 and and

DYX2 D24 b D27 H3-K27M a c MYCN – ). ). This subgrouping was sup

) and consensus clustering represented by a cumulative distribution function (CDF) and change in Gini Gini in change and (CDF) function distribution a cumulative by represented clustering consensus ) and D25 d Supplementary Figs. 1 Figs. Supplementary

) Subrouping was supported by principal-components analysis ( analysis principal-components by supported was ) Subrouping D08 . Therapies aimed at tar at aimed Therapies . D30 D67 Fig. 1 Fig.

by 4-fold and 8-fold 8-fold and 4-fold by D58 D64 Samples D78 k =

c D05 ) and consensus consensus ) and 3 D07 D04 D56 D03 D61

Normal

= 1,0000 probes) 1,0000 = n ( DNA methylation methylation DNA brain – 3 ). ). - - - - - versus 33.3% in other groups; groups; other in 33.3% versus sive to this group. ( 4 Fig. significantly older at the time of diagnosis ( harbored subgroup this in tumors the of two only subgroup; H3-K27M the with associated exclusively is data) sequencing whole-genome from length telomere assay fragment restriction mere P group; silent the in genome CNAs per 300 versus genome CNAs per 497 data; of SNP6.0 analysis (segmentation genomes unstable highly ( H3.3 subtypes. DIPG these with patients in effective less be will kinases these that the group targeting of inhibitors suggesting amplification, gene kinase tyrosine receptor had subgroup silent the ( subgroups two as well as ( genes pathway WNT of overexpression K27M subgroups. At the expression level, the silent subgroup showed H3- and MYCN the in observed as changes number copy recurrent H3.3 in 33% and p.Lys27Met histone H3.1 in but 11%), there were no tone H3 was alteration present in 44% of tumors (p.Lys27Met histone years for not in those group; the silent 6.89 versus group silent the from children the for years 1.64 (4.81 age younger significantly a at diagnosed were group this in survival in comparison with the other subgroups. Affected individuals overall in difference no was there interestingly, group, although, this from were histology (LGA) astrocytoma low-grade with DIPGs All 0.19–24) in the other two groups; a Fig. 2 Fig. ) Heat map of methylation levels in three DIPG subgroups identified identified subgroups DIPG three in levels methylation of map ) Heat d b = 0.04). Alternative lengthening of telomeres (ALT; detected by telo PC3

H3-K27M subgroup DIPGs were highly mutated in either histone histone either in mutated highly were DIPGs subgroup H3-K27M PC2 –14 –20 10 16 –38 –32 –26 –44 –8 –2 4 ). Similarly, Similarly, ). 1835 H3F3A a

) and structural variants ( variants ) and structural CDF 0.2 0.4 0.6 0.8 1.0 MDM2 0 –29.3 Silent 0 ) ) or ( H3.1 Consensus indexvalue a –23.6 0.2 Table DVANCE ONLINE PUBLICATION ONLINE DVANCE TP53 , , PVT1 MSMP 0.4 –17.9 1 mutations were enriched in this group (67.9% 0.6 , , ). Notably, DIPGs in neither the MYCN nor nor MYCN the neither in Notably,DIPGs ). HIST1H3B ATRX MYC –12.2 and and 0.8 b 5 clusters 4 clusters 3 clusters 2 clusters ), non-negative matrix factorization factorization matrix non-negative ), MYCN and and P –6.5 1.0 mutations. ALT-positive cases were were cases ALT-positive mutations. PC1 ADAM33 1 = 0.007). Given the complexity and and complexity the Given 0.007). = 3 P Supplementary Fig. 5 Fig. Supplementary , the presence of C-circles , presence the = 0.05 ( and PDGFRA –0.8

P Change in Gini 0.2 0.4 0.6 0.8 1.0 1.2 = The p.Lys27Met 0.04). his HIST1H3C , compared with the other other the with compared , 4.9 Supplementary Table 5 Table Supplementary 2.0 Supplementary TableSupplementary 4 P < 0.001; gains or amplifications amplifications or gains 2.5 10.6 Number ofclusters

3.0 Nature Ge Nature 16.3 ). This group had ). This 3.5 Supplementary Supplementary 4.0 H3-K27M ) ) were exclu 22 4.5 1 4 n and/or and/or 5.0 ± etics 2.62 2.62 )). )). ), ), ± ­ - -

© 2014 Nature America, Inc. All rights reserved. c.983G>A (p.Gly328Glu) mutation and five DIPGs in our cohort had a Twop.Gln207Glu).a had (encoding DIPGs codon neighboring a ing p.Arg206His substitution ( of Mutations gene. DIPG was in provided is subgroups DIPG H3-K27M and silent MYCN, the in cases among observed features clinical and molecular = (OR 0.046, histology LGA status (OR = 8.3, of p.Lys27Met regardless GBM histology with in cases to likely occur 3.50 years at diagnosis; 4.11 (OR = 7.1, histology astrocytoma anaplastic with cases in occur to likely more 5.3, = (OR histology P GBM have to likely more were H3.3 histone 2 Fig. p.Lys27Met histone H3 alteration was present (OR = 0.019, MYCN ACVR1 P and H3.3 in histone p.Lys27Met encoding mutation co-occur—specifically, to likely therapies. multimodal require likely would subgroup this available, become mutations histone the targeting pies heterogeneity of genetic changes for the H3-K27M subgroup, if thera others. between correlation significant statistically shows OR of basis the on histology ( deletion. del, amplification; Amp, and alteration H3.3 histone p.Lys27Met between association significant statistically suggests OR of basis the on co-occurring DIPG of features structural or ( astrocytoma. low-grade LGA, astrocytoma; anaplastic AA, tumor; neuroectodermal PNET, primitive glioblastoma; GBM, applicable; not NA, manner. subgroup–specific a DIPG in represented ALT are CNAs and mutations, recurrent of frequency histology, sex, as 2 Figure Nature Ge Nature c.983G>T mutation that resulted in a p.Gly328Val substitution ( PVT1 a = 0.0127) and mutation encoding p.Lys27Met in histone H3.1 and and H3.1 histone in p.Lys27Met encoding mutation and 0.0127) = = 0.0035), whereas p.Lys27Met alterations in histone H3.1 were were H3.1 histone in alterations p.Lys27Met whereas 0.0035), = After After more significantly were variants structural and mutations Certain K27M H3.1 K27M H3.3 PDGFRA Histology PDGFRA

± Sex or ACVR1 MYCN Ploidy b 2.03 years at diagnosis versus wild-type histone H3.1: 6.50 6.50 H3.1: histone wild-type versus diagnosis at years 2.03 PTEN MY TP53 TP53 # mlfctos ee ttsial ls lkl t ocr f a if occur to likely less statistically were amplifications Female Male RB1 ). Affected individuals carrying p.Lys27Met alterations in in alterations p.Lys27Met carrying individuals Affected ).

mutations (OR = 15.8, 15.8, = (OR mutations ALT P Sex ID2 H3F3A Molecular subgroups of DIPG share common clinical features and recurrent genomic events. ( events. genomic recurrent and features clinical common share DIPG of subgroups Molecular < 0.05, * < 0.05, C ACVR1 P n amplifications (OR = 8.0, = 8.0, (OR amplifications = and 0.016) in younger (p.Lys27Metcases histone H3.1: MYCN etics DIPG01 DIPG29

and DIPG38 P Histology DIPG49 (encoding activin A receptor,activin (encoding I), a type new cancer P = 0.0039) and were unlikely to occur in cases with in and to cases = were occur 0.0039) unlikely

PDGFRA DIPG02 ACVR1 < 0.01, ** < 0.01, AA GBM ADVANCE ONLINE PUBLICATION ONLINE ADVANCE TP53 DIPG09 DIPG18 Silent P

= 0.040) ( DIPG26 DIPG37 , the next most frequently mutated gene in in gene mutated frequently most next the ,

Fig. 3 DIPG44 LGA PNET in four DIPGs (c.617G>A) resulted in a in resulted (c.617G>A) DIPGs four in

amplifications (odds ratio (OR) = 8.0, 8.0, = (OR) ratio (odds amplifications DIPG48 P DIPG52 = 0.0040) ( = 0.0040) P

< 0.001. DIPG55 a c

P DIPG03 ). One ). DIPG had a mutation affect ) Probability of mutations or structural events in DIPG occurring with a clinical feature such as sex or tumor tumor or sex as such feature a clinical with occurring DIPG in events structural or mutations of ) Probability Fig. 2

= 0.0004) ( 0.0004) = DIPG05 Mutations DIPG07 P DIPG08 = 0.0127) and between p.Lys27Met histone H3.1 alteration and and alteration H3.1 histone p.Lys27Met between and = 0.0127)

NA Wild type Mutated DIPG24 c

). DIPG27 Fig. 2 Fig.

TP53 DIPG46 DIPG50 DIPG57 DIPG60 Fig. 2 Fig. c mutations were more DIPG06 ). A summary of the A ). summary ALT status DIPG19 DIPG39 H3-K27M

Non-ALT ALT DIPG62 b Table DIPG79 ). Conversely, Conversely, ). DIPG80 DIPG28 P DIPG30 = 0.0103; 1 DIPG45 Fig. 3 . DIPG10 DIPG04 Ploidy DIPG41

a DIPG61 Silent Aneuploid ). ±

- - DIPG67

DIPG63 DIPG64 which are known targets of BMP signaling of BMP targets are known which PCR quantitative for we pathway performed activation, ( SMAD1/5 phosphorylated for tivity by protein blot. Only cells expressing Gly328Val ACVR1 showed posi SMAD1/5 pathway,for phosphorylated we tested signaling BMP the in Tocells. transfected the ACVR1 whether mutant determine activated detected expression ACVR1 total higher and Flag-ACVR1 with cells, all in ACVR1 of expression endogenous showed analysis blot Protein H3.3. histone Lys27Met and/or ACVR1 Gly328Val taining con vectors expression pCDH511b Flag–tagged 3× with (iNHAs) vitro in tions compared to cases with wild-type and SMAD5 (SMAD1/5) were higher in DIPG cases with constitutive activation of the receptor. Levels of phosphorylated regulatorsSMAD1 of ACVR1 negative such as FKBP12, with resulting domain in ligand-independentGS the of interaction the weaken to thought domain GS Gly328 residue map to the kinase domain and are spatially close to the reported in individuals with FOP or in cancer. Mutations affecting the als with FOP p.Gln207Glu substitutions are described as rare alterations in individu to BMP signaling, transitioning to osteoblasts progressivaare cells ossificans endothelial sensitized which (FOP), in an autosomal dominant disease of the connective tissue, fibrodysplasia serine-rich) domain and is the most common alteration responsible for signaling tein) activin A (ALK2) receptor involved in BMP (bone morphogenetic pro had DIPGs of 20% total, In

TP53 DIPG66 DIPG78 To determine the biological consequence of of consequence biological the determine To DIPG71 DIPG47 mutations and GBM histology (OR = 10.8, = 10.8, (OR histology GBM and mutations CNA DIPG51 , we transfected immortalized normal human astrocytes astrocytes human normal immortalized transfected we , DIPG56

Loss Gain DIPG58 DIPG11 1 7 16 DIPG25 , and, similar to the p.Arg206His substitution, they are they substitution, p.Arg206His the to similar and, , , 1 1 5 7 . The p.Arg206His substitution is in the GS (glycine- GS the in p.Arg206His is substitution The . . The p.Gly328Val substitution has not previously been GBM AA LGA PNET Male Female c b a ) Clinical and genomic features such such features genomic and ) Clinical H3 H3.3 # ACVR1 Male H3.1 p53 R273 ACVR1 # Female H3 p53 R175 p53

H3.3 ACVR1 G328 * H3 ACVR1 R206 mutations (OR = 15.8, = 15.8, (OR mutations H3.1 ** ACVR1 # H3.3 p53 # # mutations. #

p53 R273 ALT status H3.1 b PDGFRA Fig. 3 Fig. ** ACVR1 p53 R175 # ** ) Probability of two mutational mutational two of ) Probability p53 #

ACVR1 RB1 ** # * 18– ACVR1 ACVR1 G328 MYCN 1 del c 2

5 ACVR1 R206 am P ). To test downstream downstream test To ). ALT status . . The p.Gly328Glu and 0 ( < 0.005), among among < 0.005),

. Expression of either of either . Expression ALT status * am Fig. 3 ACVR1 RB1 del RB1 del p ACVR1 ** * PDGFRA amp PDGFRA amp p s r e t t e l # MYCN amp * b MYCN amp ACVR1 ). ID1 P encodes the the encodes 0.031–0.061 0.062–0.124 0.031–0.061 0.062–0.124 < 0.001). < 0.001). mutations mutations 0.125–0.24 0.125–0.24

0.25–0.49 0.25–0.49 16–31.99 16–31.99 0.5–1.99 and 0.5–1.99 Odds ratio 8–15.99 8–15.99 Odds ratio 2–3.99 4–7.99 2–3.99 4–7.99 muta ID2  - - - - - ,

© 2014 Nature America, Inc. All rights reserved. Error Error bars, s.e.m.; suggesting increased * proliferation. higher significantly BrdU incorporation, type brainstem progenitor cultures with wild- bars, s.d.; to empty vector controls. * ACVR1 have an increased growth rate compared ( empty vector control. Error bars, s.d.; in ACVR1 or a combination of both shows increase expressing Lys27Met histone H3.3, Gly328Val cells transfected with empty vector or with vector compared to control cells. ( with construct expressing Gly328Val ACVR1 mutant iNHA and DIPG58 cells transfected levels of SMAD1/5 phosphorylated in ACVR1 compared with DIPGs with wild-type (WT) levels of SMAD1/5 phosphorylated (pSMAD1/5) DIPGs with occur in the protein kinase domain. ( in the GS domain, and the Gly328 alterations p.Arg206His and p.Gln207Glu alterations occur were detected in 12 of 61 DIPG cases. The p.Gln207Glu, p.Gly328Gly and p.Gly328Val) DIPG. ( BMP signaling Figure 3 in increase fourfold nearly a in resulted proteins these in increase twofold a to led alone ACVR1 Gly328Val or H3.3 histone Lys27Met Age atdiagnosis Gene expression ALT phenotype CNAs CpG methylation Mutations Histology Sex molecular subgroups T s r e t t e l  at 320 nm. a Alternative lengtheningoftelomeres. e able able 1

) ) iNHA cells expressing mutant Gly328Val profiling ALT Median ACVR1 histone H3.3 p.Lys27Met Grade IV Grade III Male:female Median age Grade II ID1 ACVR1 a and . . ( a

) ) Four mutations (encoding p.Arg206His, c ACVR1 Clinical, Clinical, genetic and epigenetic features of DIPG β ) ) Protein blot showing increased n value ID2 ACVR1 , , those with mutant = 3. ( ID1

gene expression compared to in in vitro mutations activate constitutively and and Overexpression of MYCN f n mutations have increased ) ) Compared to mouse DYX2 chr. 2pchromothripsis = 3. OD and in ID2 amplification, MYCN , 6.3 years HSPR12 P 100% 75% d expression levels, whereas coexpression of of coexpression whereas levels, expression = 0.0034. Error 1:3 0% 0% 0% 0% 0% 320 ) ) RT-PCR in iNHA ACVR1 , , optical density ACVR1 ,

FAP P -mutant -mutant b

< 0.05. ACVR1 n ) ) Human have = 3. Overexpression Few copynumber ADAM33 MSMP of genome changes, silent

- MDM2 Silent 5 years 13% 22% 35% 56% 11% 33% 2:1

0% , , G328V ACVR1 K27M H3.3 Empty vector c a e 3 1

Number of cells (×1,000) ACVR1 Q207E R206H 10 20 30 40 50 60 70 Underexpression of PVT1 3 0 0 0 0 0 0 0 0 37 kDa 57 kDa 57 kDa 52 kDa 52 kDa 52 kDa HOXA7 HOXA9 VAX2 deletion RB1 amplification, PDGFRA 4 2 0 ID1 H3-K27M / 7 years MYC 22% 15% 18% 97% 74% 26% and G328V ACVR1 Empty vector 1:1 0% , and and G328E G328V + – – – HOXC6 , , iNHA 105 and + + – DOXD4 gain/ TP53 + + – – ID2 – – 178 , DIPG58 Days

+ + + – receptor domain TGF-

+ H3.3 had a significantly reduced proliferation rate when grown grown when rate proliferation in DMEM ( adherently reduced histone significantly Lys27Met a had expressing H3.3 cells iNHA controls, ( to Compared nucleus the phenotype. in variations clonal remove to to pooled and selected were localize to staining ( and Lys27Met histone H3.3 protein, according to protein blot analysis wild-type of amounts comparable expressing Clones control. vector or or empty H3.3 with Lys27Met histone wild-type Flag-tagged nally we transfected iNHA cells with expression vectors encoding N-termi and methylation patterns as well as its potential transformative capacity, ( incorporation (bromodeoxyuridine) by BrdU mutant expressing with virus ACVR1 proliferation showed increased ACVR1 ( express wild-type to acceptor) compared ACVR1 mouse Gly328Val or splice Arg206His ing a with (LTR) repeat terminal long virus vitro in mouse brainstem progenitor cells from 3 (P) day postnatal in blot protein by seen also was SMAD1/5 ylated ( controls ACVR1 increased the growth rate of iNHAs compared to ( empty effect vector additive an suggesting levels, expression H3.3 compared to both wild-type and empty vector controls were ( were controls observed vector empty histone and Lys27Met wild-type both to expressing compared H3.3 cells the of profiles Infinium450k) in both the expression (Illumina HT-12v4) and methylation (Illumina GFAP, TUJ1 or nestin, O4 ( levels in changes marked no with staining of immunofluorescence by expression SOX2 higher had medium NSC in H3.3 expressing histone Lys27Met cells iNHA semi-adherent the H3.3, histone Lys27Met medium (NSC) cell ( stem neural in seeded when semi-adherently β 208 Supplementary Fig. 6e Supplementary Fig. 6a Fig. Supplementary /activin β Total ACVR1 Flag-ACVR1 G328V SMAD SMAD pSMAD1/5 (Ser463/Ser465) -actin To assess the impact of Lys27Met histone H3.3 on gene expression To on of H3.3 expression gene Lys27Met histone impact the assess 8 6 1 5 with RCAS (replication-competent avian sarcoma-leukosis sarcoma-leukosis avian (replication-competent RCAS with P P GS domain < 0.05; data not shown). Brainstem progenitor cells infected Supplementary Fig. 6f Fig. Supplementary * = 0.0034; 0.0034; = a DVANCE ONLINE PUBLICATION ONLINE DVANCE domain Protein kinase 493 d f OD Fold change relative to empty vector 3 Fig. P

320 ). Compared to adherent iNHA cells expressing 0. 1. 1. 2. 2. 3. 3. 4. 4. 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 < 0.0001; < 0.0001; ), that was verified by immun by verified was that ), 509 5 0 5 0 5 0 5 0 5 0 0 Empty vector e ). An increase in the levels of phosphor of levels the in increase An ). 37 kDa 52 kDa 52 kDa ACVR1 Case b K27M WT Supplementary Fig. 7 Supplementary ). The top pathways perturbed by perturbed pathways top The ). Supplementary Fig. 6c Fig. Supplementary DIPG4 K27M H3.3 WT Nes–tv-a P <0.00065 3 upeetr Fg 6b Fig. Supplementary R206 DIPG7 G328E H3.1 * G328V H 2 P ; ; DIPG5 R206H

H3.3 < 0.05; < 0.05; Tp53 Fig. 3 Fig. Nature Ge Nature K27M +G328V P <0.0002 D 3 G328 G328E fl H3.1 * IPG5 d o mice infected ). Alterations ). Alterations fluorescence fluorescence ). Gly328Val ). V Fig. 3 Fig. 4 ) ) and grew β pSMAD1/5 Total SMAD ID ID -actin 2 1 n f etics ). - - - )

© 2014 Nature America, Inc. All rights reserved. Research) Research) and C17, through the Genome jointCanada/CIHR ATID Competition (cofunding from Genome Génome BC, CIHR-ICRQuébec, (Institute for Cancer MOP(CIHR, 115004) and was funded in part by a Genome grantCanada/CIHR research. This work was supported by the Canadian Institutes of Health Research We would like to thank all of the patients and families for donating tissue for this online version of the pape Note: Any Supplementary Information and Source Data files are available in the (accession Omnibus Expression Gene the (acces sion Omnibus Expression Gene the through accessible are data (accession Omnibus Expression Gene (accession Archive EGAS00001000 Genome-phenome European the through ble codes. Accession the th of in version available are references associated any and Methods Me children. affected for outcome improve to approaches peutic thera new designing when considered be to needs complexity This and tion altera H3.3 histone of p.Lys27Met effects the into investigation ther However, event. fur reprogramming a cellular suggest H3.3, histone tion and gene changes in expression iNHA Lys27Met expressing cells expression differential of stem cell markers, as well as methyla global with poor prognosis in other cancers regulating cell differentiation mechanism in DIPG tumorigenesis. ID2 may have a role in negatively ID2-related common, potentially a suggesting expression, increased subgroup cases had genomic amplification of overexpressed alterations H3.3 histone p.Lys27Met increasing mutant of combination the of effect additive wild-type had that mutant H3 ACVR1 action. its for pathway Phosphorylated SMAD1/5 was also only in observed DIPG alternate cases with an suggesting SMAD1/5, of tion increased shown). not (data manner dependent mutation- a in altered not were (H3K9ac) H3 histone of 9 lysine at acetylation and (H3K4me3) H3 (H3K27ac), histone of 4 H3 lysine at trimethylation histone of 27 lysine at acetylation of levels ( Global H3.3 histone wild-type to compared with H3.3 tumors histone Lys27Met for positive tumors in H3K27me3 levels global lower showed DIPGs of staining histochemical ( to controls compared levels global of at trimethylation lysine 27 of histone H3 (H3K27me3) reduced showed cells agar. These soft in colonies form not did H3.3 cell-to-cell signaling ( ( proliferation and growth cell decreased ( development embryonic were H3.3 histone Lys27Met mutant in changes expressing by cells methylation affected functions ( sion signaling cell-to-cell in increase ( an especially functions, cellular and molecular to corresponded expression H3.3 histone Lys27Met Nature Ge Nature (project title: The Canadian Paediatric Cancer Genome Consortium (CPCGC): A P ckno = 0.00882, 0.00882, = Our Our results highlight the many pathways to tumorigenesis in DIPG. to led also H3.3 histone Lys27Met of expression Interestingly, thods GSE5002 P < 0.00001, 0.00001, < w mutations and not in ones expressing the Lys27Met histone histone Lys27Met the expressing ones in not and mutations ACVR1 ledgments ID1 ID1 n e pape e etics and and 1 z and and score = 2.365) and a decrease in cell cycle progres cycle cell in decrease a and 2.365) = score ). SNP6.0 copy number data are accessible through through accessible are data number copy SNP6.0 ). 57 mutations in DIPG tumorigenesis is warranted. is tumorigenesis DIPG in mutations

5 hl-eoe eunig aa r accessi are data sequencing Whole-genome z r ID2 ). Methylation data are accessible through the the through accessible are data Methylation ). score = −1.347). The top molecular and cellular cellular and molecular top The −1.347). = score . ID2 r ADVANCE ONLINE PUBLICATION ONLINE ADVANCE . P < 0.00001). Cells expressing Lys27Met histone Supplementary Fig. 6d Fig. Supplementary expression yet did not induce phosphoryla induce not did yet expression levels. Further, a subset of DIPG cases with with cases DIPG of subset a Further, levels. 2 1 , , and its expression has been associated ACVR1 22– 2 . Our Our . GSE5002 4 upeetr Fg 6g Fig. Supplementary . Morphological changes and P GSE5002 = 0.00103) and increased increased and 0.00103) = ID2 in vitro in ACVR1 ). Similarly, immuno Similarly, ). 2 with corresponding ). Gene expression expression Gene ). 4 ID2 data suggest an an suggest data

). and and P , and MYCN and , < 0.00001), 0.00001), < H3F3A online online , h in in ). ). ------

20. 19. reprints/index. at online available is information permissions and Reprints The authors declare no competing interests.financial technical support and conceptual advice. All authors approved the manuscript. C. Hawkins wrote the manuscript. P.L., C.B., C.D.A., M. Brudno, A.H. and U.T. gave U.T. and U.B. provided reagents, tissue and mice. P.B., P.R., S.P., M.D., O.B. and J. Chan, L.L.-C., S.D., J.H., C.D., K.S., J. Michaud, S.Z., D.R., J. Cain, M.M.S., E.B., C.J., K.R.T., A. Mackay, A.E.B., J.N., J.R.F., M.A.K., D.Z., N.K.F., A.D., J.V.H., A.S., M. Bourgey, G.B. and A. Montpetit and collected analyzed data. O.B., C. Hawkins, J. Mangerel performed experiments. P.B., C. Hoeman, F.C., P.R., L.L., M.D., P.R., S.P., A. Morrison, J.Z., S.A., M. S.R., Y.C.,Barszczyk, P.C.-B., K.C.H. and E.B., U.B., P.B., O.B. and C. Hawkins the designed study. P.B., C. Hoeman, F.C., from 5P30CA016087-32 the National Cancer Institute. Center for Research Resources, US National Institutes of Health and by grant M.A.K. and D.Z. was supported in part by grant fromUL1TR000038 the National Centre and support from the Stavros Niarchos Foundation. Sample forcollection acknowledge National (NHS) funding Healthto Service the Biomedical Research Defense and the Pediatric Brain Tumor Foundation. C.J., A. Mackay and K.R.T. a Damon Runyon InvestigatorClinical and is supported by the US Department of Frederick Banting and Charles Canada Best Graduate Scholarships award. O.B. is for Cancer)). P.B. High-Risk Childhood is a recipient of a CIHR Doctoral Translating Next-Generation Sequencing Technologies into Improved Therapies 18. 17. 16. 15. 14. 13. 12. 11. 10. 9. 8. 7. 6. 5. 4. 3. 2. 1. COMPET AUTHOR CONTRiBUT

agon, M. Zarghooni, B.S. Paugh, B.S. Paugh, P.Buczkowicz, understanding an toward P.G.Advances Fisher, F. & Laningham, S.S., Donaldson, Hargrave, D., Bartels, U. & Bouffet, E. Diffuse brainstem glioma in children: critical B.L. Maria, Britain, in gliomas. stem brain cancer for Chemotherapy G. Perilongo, & childhood C.R. Freeman, for rates survival based Population C.A. Stiller, oyeae s oeta teaetc targets. (2010). therapeutic potential as polymerase receptor factor growth platelet-derived highlights gliomas disease. adult the with (2010). 3061–3068 differences key reveals gliomas glioma. pontine intrinsic diffuse in genes regulatory cell-cycle and kinases tyrosine receptor glioma. pontine intrinsic diffuse gliomas. brainstem of trials. clinical of review Neurol. Child J. Syst. 1980–1991. Shin, M. is signaling BMP Y. Yokota, & K. Sano, K., Mori, T., Nakahiro, H., Kurooka, Kowanetz, M., Valcourt, U., Bergstrom, R., Heldin, C.H. & Moustakas, A. Id2 and Id2 A. Moustakas, & C.H. Heldin, R., Bergstrom, U., Valcourt, M., Kowanetz, K.A. Petrie, F.S.Kaplan, G.A. Song, Henson, J.D. telomerase. and telomeres of Analysis W.E. Wright, & J.W. Shay, B.S., Herbert, G. Wu, J. Schwartzentruber, D.A. Khuong-Quang, elements in the Id genes. Id the in elements BMP-responsive conserved the utilizing by transcript-1, BMP-inducible transcript, 420 expression. Id2 serum-induced for responsible (2004). 4241–4254 Id3 define the potency of cell proliferation and differentiation responses to responses anddifferentiation proliferation ofcell factor growth transforming potency the define Id3 patients. progressiva ossificans (2009). fibrodysplasia two in ACVR1. receptor I phenotypes are caused by mutations in the bone morphogenetic protein (BMP) type progressiva. ossificans fibrodysplasia activity. lengthening-of-telomeres Biol. Cell Protoc. Curr. glioblastomas. non-brainstem and gliomas glioblastoma. paediatric in genes remodelling Neuropathol. gliomas. pontine intrinsic diffuse pediatric of subgroups distinct biologically , 281–287 (2012). 281–287 ,

15 I NG NG F et al. et J. Clin. Oncol. Clin. J. , 545–553 (1999). 545–553 , et al. htm t al. et

Somatic histone H3 alterations in pediatric diffuse intrinsic pontine intrinsic diffuse pediatric in alterations H3 histone Somatic et al. et al. et t al. et t al. et I Br. Med. J. Med. Br. 124 t al. et Identification of a novel bone morphogenetic protein (BMP)-inducible t al. et NANC t al. et l . et al. et

8 rise gim: . ahlg, lncl etrs ad therapy. and features, clinical Pathology, I. glioma: Brainstem , 439–447 (2012). 439–447 , , 112–128 (1993). 112–128 , DNA C-circles are specific and quantifiable markers of alternative- Classic and atypical fibrodysplasia ossificans progressiva (FOP) progressiva ossificans fibrodysplasia atypical and Classic Integrated molecular genetic profiling of pediatric high-grade pediatric of profiling genetic molecular Integrated Hum. Mutat. Hum. oeua cneune o te ACVR1 the of consequences Molecular Genome-wide analyses identify recurrent amplifications of amplifications recurrent identify analyses Genome-wide Whole-genome profiling of pediatric diffuse intrinsic pontine intrinsic diffuse pediatric of profiling Whole-genome oe mttos n CR rsl i ayia features atypical in result ACVR1 in mutations Novel Aurora kinase B is a potential therapeutic target in pediatric in target therapeutic potential a is B kinase Aurora et al. et I

J. Clin. Oncol. Clin. J. t al. et Chapter I AL AL 29 ONS Lancet Oncol. Lancet

309 , 3999–4006 (2011). 3999–4006 , K27M mutation in histone H3.3 defines clinically and clinically defines H3.3 histone in mutation K27M I J. Bone Miner. Metab. Miner. Bone J. β NTERESTS and bone morphogenetic protein. morphogenetic bone and rvr uain i hsoe 33 n chromatin and H3.3 histone in mutations Driver , 1612–1616 (1994). 1612–1616 ,

18, Unit 18.6 (2003). 18.6 Unit 18, 30 Brain Pathol. Brain Nat. Biotechnol. Nat. , 379–390 (2009). 379–390 ,

J. Biol. Chem. Biol. J. 24

7 , 241–248 (2006). 241–248 , , 1266–1272 (2006). 1266–1272 , Nat. Genet. Nat.

Nature 23 Biochem. Biophys. Res. Commun. Res. Biophys. Biochem. . ln Oncol. Clin. J.

, 244–253 (2013). 244–253 , 31

27

285 , 34–43 (2013). 34–43 ,

, 1181–1185 (2009). 1181–1185 , 482

44 , 22542–22553 (2010). 22542–22553 , α http://www.nature , 226–231 (2012). 226–231 , , 251–253 (2012). 251–253 , LS ONE PLoS and poly (ADP-ribose) poly and s r e t t e l . ln Oncol. Clin. J. R206H Mol. Cell. Biol. Cell. Mol.

28 1337–1344 , uain of mutation

Childs Nerv. Childs 4 e5005 ,

.com/

Acta

24 28  , ,

© 2014 Nature America, Inc. All rights reserved. to to C. Hawkins ( Children, Toronto, Ontario, Canada. Weill Cornell Medical College and Memorial Cancer Sloan-Kettering Center, New York, New York, USA. Centre, London, Ontario, Canada. Medicine, Children’s Hospital of Eastern Ontario, Ottawa, Ontario, Canada. British Columbia, Canada. Faculty of Medicine, University of Calgary, Calgary, Alberta, Canada. Colorado, Denver, Colorado, USA. New York University Cancer Institute, New York University Langone Medical Center, New York, New York, USA. Hematology, Ann & Robert H. Lurie Children’s Hospital of Chicago, Chicago, Illinois, USA. Minnesota, Minneapolis, Minnesota, USA. 9 and Epigenetics, Rockefeller University, New York, New York, USA. McGill University, Montreal, Quebec, Canada. Ontario, Canada. The Hospital for Sick Children, Toronto, Ontario, Canada. 1 22. 21. s r e t t e l  Division Division of Cancer Institute Therapeutics, of Cancer Research, London, UK. Division of Pathology, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada.

i, Z. Liu, factor transcription negative a ID2: Z.X. Yao, Q.Y.& Y.H.,Cai, Zhang, X.S., Chen, rgoi fco pooe cl poieain n aohrnel carcinoma. nasopharyngeal (Shanghai) Sin. in Biophys. Biochim. proliferation cell promotes factor prognosis differentiation. (2012). oligodendroglia regulating t al. et [email protected] vrxrse DAbnig rti ihbtr a a unfavorable an as 2 inhibitor protein DNA-binding Overexpressed 4 Division Division of Pediatric Hematology/Oncology, Duke University Medical Center, Durham, North Carolina, USA. 18 Division Division of Hematology/Oncology, McMaster Children’s Hospital, Hamilton, Ontario, Canada. 15 21 24 Department Department of Pathology, Arnold Palmer Hospital for Children, Orlando, Florida, USA. Monash Monash Institute of Medical Research, Monash Medical Centre, Clayton, Victoria, Australia.

44 These These authors contributed equally to this work. , 503–512 (2012). 503–512 , 11 ) ) or O.B. ( Center Center for Genetic Medicine, Children’s National Medical Center, Washington, DC, USA. 6 . ersi Res. Neurosci. J. Department Department of Computer Science, University of Toronto, Toronto, Ontario, Canada. [email protected] 3 Department Department of Laboratory Medicine and Pathobiology, Faculty of Medicine, University of Toronto, Toronto,

8 90 Division Division of Haematology/Oncology, The Hospital for Sick Children, Toronto, Ontario, Canada. 17 925–932 , Department Department of Pediatric Neurology and Neuro-Oncology, BC Children’s Hospital, Vancouver, 20 ). 10 Acta Department Department of Hematology/Oncology, Children’s Hospital London Health Sciences Department Department of Pediatric Hematology-Oncology, Children’s Hospitals and Clinics of 24. 23. 25 These These authors jointly directed this work. should Correspondence be addressed

ai, . Jag WG, hra AK, ebl, .. Mke, . h mRNA The K. Mokbel, & R.F. Newbold, A.K., Sharma, W.G., Jiang, U., Wazir, J. Rollin, uor tg ad des ciia otoe n ua bes cancer. breast advanced human with Res. in associated outcome clinical is adverse -2 and and stage binding-1 tumour DNA of inhibitors of expression cancer. lung cell small non in value 13 Division Division of and Pediatric Hematology/Oncology Neuropathology,

33 , 2179–2183 (2013). 2179–2183 , t al. et 2 23 Arthur Arthur and Sonia Labatt Brain Tumour Research Centre, h itaellr oaiain f D epeso hs predictive a has expression ID2 of localization intracellular The a The The Centre for Applied Genomics, The Hospital for Sick DVANCE ONLINE PUBLICATION ONLINE DVANCE 14 Department Department of Pediatrics, Children’s Hospital 19 16 Department Department of Pathology and Laboratory PLoS ONE PLoS Clark Clark H. Smith Brain Tumour Centre, 5 Génome Génome Québec Innovation Centre, 22 7 Department Department of Neurological Surgery, Laboratory Laboratory of Chromatin Biology 12

4 Department Department of Pediatrics- , e4158 (2009). e4158 ,

Nature Ge Nature

Anticancer n etics

© 2014 Nature America, Inc. All rights reserved. Gene expression profiling. expression Gene (Bioconductor/R). by SigClust for significance validated were Subgroups of analysis Institute). Viewer,Cancer Dana-Farber MultiExperiment (SAM; significance microarrays and Bioconductor/R) (ConsensusClusterPlus, ing cluster silhouette Bioconductor/R), ConsensusClusterPlus, Institute; Broad Institute), Cancer Dana-Farber Viewer, MultiExperiment Bioconductor/R; ConsensusClusterPlus, Institute; Broad (GenePattern, Broad Institute), consensus hierarchicalfactorization clusteringmatrix (GenePattern,non-negative including algorithms, and programs tiple mul using validated and performed was methylation probe CpG differential of basis the on clustering and Subgrouping protocols. manufacturer’s the to out on this array. DNA was using extracted the DNeasy kit (Qiagen) according with vectors expressing Lys27Met histone and H3.3 wild-type was also carried manufacturer’s Methylation profiling of specifications. iNHA cells transfected out ried using the EZ DNA to kit Methylation according (Zymo the Research) car was conversion array. Bisulfite Infinium450k Illumina the used Canada) Toronto, Network, Health (University Ontario, Centre Microarray at the ples Methylation profiling. ( gene are not simply a of consequence random background mutation processes a in mutations observed the whether testing algorithm an Institute), (Broad MutSig with cohort combined the of analysis upon altered significantly be to and found being sequencing and/or whole-exome sequencing whole-genome SNVs for were on selected validation the basis of from variant frequency allele These Technologies). Ion Torrent and (Life chips arrays Fluidigm by fication ampli PCR using genes 128 in variants Wesequence 151 excluded. validated were mapping non-unique with reads and duplicates assembly; genome ence refer human GRCh37/hg19 the to aligned were data Sequence Biosystems). described previously as mined pairs. Maximum-likelihood copy counts of chromothriptic regions were read deter discordant and coverage of depth by informed (HMM) model Markov 8 Fig. σ × 15 of window sliding × 3 > size (frag ment pairs read Discordant clustering. read-pair discordant by performed regions was in variants chromothriptic of Viewer Prediction structural (v2.2). (v2.0.3) Savant in data sequencing genome Suite (v.6.6) as well whole- Genomics in data Partek and SNP6.0 algorithm DNAC the using validated visually were and tissue normal in found variants variants structural common subtracting after found were (ref. 1.0.1 CTX 6 Table ( Picard 1.77 using were duplicates removed from alignments parameters default using GRCh37/hg19 to reads all map to used was (BWA) tool Aligner Burrows-Wheeler The samples. normal for 28–74× and samples tumor for 35–67× was coverage Mean length. read 100-bp with size on an sequenced 2000 HiSeq using DNA Illumina of bp fragments 259–333 in paired-end were DNA samples technologies. Illumina next-generation using sequenced DNAwere pairs normal matched and tumor of DIPG 20 genomes variants. structural and alignment sequencing Whole-genome months. 10.4 of time survival median a with years, 6.37 was diagnosis of age median The samples. autopsy post-treatment represented samples 54 and from autopsy), patients (2 samples non-treated pretreatment represented of board Samples centers. for review contributing tutional 20 of patients these and consent was approved informed by receiving after the insti was collected material patient All diagnosis. DIPG for required was pons the of 50% least involvement of atwith diffuse A lesion (MRI). contrasting imaging resonance available. DIPGs were diagnosed by a neuroradiologist on the basis of magnetic if samples, 74 blood brain DIPG normal including samples, and/or peripheral and samples. Patients ONLIN doi: (University Centre Microarray the at controls brain normal 10 and samples Supplementary Table 6 Supplementary is the s.d. of the insert size and and size insert of the s.d. the is Exome capture was carried out with the SOLiD platform (Applied (Applied platform SOLiD the with out carried was capture Exome 10.1038/ng.2936 ). Localization of a chromothriptic event was predicted using a hidden hidden a using predicted was event chromothriptic a of Localization ). ). Structural variants were identified using PRISM 1.1.6 and PRISM PRISM and 1.1.6 PRISM using identified were variants Structural ). E E M E THODS 2 σ 6 ). Recurrent structural variants ( variants structural Recurrent ). + + µ Biological material and clinical data were gathered for were data gathered and clinical material Biological Comprehensive methylation profiling of 28 DIPG sam ) were initially selected by greedy clustering using a using clustering greedy by selected initially were ) σ ). and were refined by clipped-read mappings, where where mappings, clipped-read by refined were and Expression profiling was conducted on 35 DIPG DIPG on 35 conducted was profiling Expression 2 9 . µ is the average insert size ( size insert average the is k -means clustering (GenePattern, (GenePattern, clustering -means 2 8 and the Integrated Genomics Genomics Integrated the and Supplementary Table 7 Table Supplementary 2 7 and all structural structural all and Supplementary Supplementary Supplementary Supplementary The entire entire The 2 5 . PCR PCR . ------) at 80% confluence. Clones expressing wild-type and mutant (Lys27Met) (Lys27Met) mutant and ( wild-type H3.3 histone expressing Clones confluence. 80% passaged at were Cells mycoplasma. for negative be to found and tested were cells In vitro described previously as Partek Genomics Suite v6.6 and Genotyping Console 4.1 (GTC4.1; Affymetrix) in HMM and tool segmentation the CNAs using for analyzed were data CEL Network. Health University the at Centre Microarray the at or Children Sick DNA were by performed the Centre for Applied at Genomics The Hospital for of hybridization and labeling Digestion, (Affymetrix). SNP6.0 using samples analysis. number Copy rate. discovery false for corrected was among the subgroups were identified using one-way ANOVA, and significance expressed differentially Genes brain). to non-neoplastic (relative data change on was log conducted Data analysis normalization. ized in Partek Suite Genomics v6.6 quantile using per-probe median-centered Integrity Number (RIN) ranged from 1.3 to 8.2. Microarray data were normal Technologies). RNA (Agilent 2100 the Bioanalyzer using was assessed quality and to RNAaccording RNeasy kit QIAShredder the manufacturer’s(Qiagen) specifications. the using extracted was RNA array. this on assessed also was H3.3 histone wild-type or Lys27Met expressing vector with iNHA transfected in cells expression gene Differential array. BeadChip V4 HT-12 Illumina the for DASL protocol the using Toronto, Network, Canada) Health Ontario, cells was performed in triplicate using the automated Vi-Cell Viability Viability Vi-Cell BrdU automated by measured was the proliferation Cell using Beckman-Coulter. from triplicate Analyzer in performed was cells iNHA control as well as ACVR1 Gly328Val and H3.3 histone Lys27Met ing proliferation. cell and counting Cell (Thermo kit See Scientific). Synthesis cDNA First-Strand RevertAid the using conducted Technology, Signaling Cell dilution; was (1:1,000 transcription Reverse 4398). Signaling Technology, 6944), Flag (1:1,000 dilution; Sigma, F3165) and ACVR1 dilution; Cell Signaling Technology, 9516), total SMAD1 (1:1,000 dilution; Cell blotting was conducted using antibodies to phosphorylated SMAD1/5 (1:1,000 for protein blot respectively. extracted Protein and PCR quantitative analyses, and and protein RNA were 72 h transfection, after were harvested Cells H3.3. Lys27Met-H3.3 or pCDH511b-Gly328Val-ACVR1 and pCDH511b-Lys27Met- 3 the at pCDH511b- tag pCDH511b-Gly328Val-ACVR1, pCDH511b, Flag with cDNA the of 3× for sequence with frame in Biosciences) (System vector pCDH511b the into subcloned then was mutant ACVR1 Gly328Val the encoding sequence The change. acid amino p.Gly328Val the in resulting 8 Table (see kit (Stratagene) Mutagenesis Site-Directed QuikChange samples. patient for described to those HT-12 methods by v4) arrays similar (Illumina expression and 450k) (Infinium methylation on analyzed were 15 and 4 passages from cells iNHA kit. RNeasy by isolated Total was RNA kit. the Methylation DNA EZ using out carried was conversion bisulfite and (Qiagen), kit Tissue and for of the isolation RNA. DNA Genomic using the was DNeasy isolated Blood DNAof used genomic and other for the isolation one the used into two parts, passages at collected were 4 and 15 at about cells 80% Totalconfluence. cell for pellet each sample was respectively, divided arrays, methylation and sion expres for RNA used DNA of and isolation the For phenotype. in variations clones were with 300 selected transfected Stably gene. hygromycin the containing fragment DNA instructions. manufacturer’s the to according (Promega) reagent transfection 6 FuGENE using transfected were Cells sequencing. Sanger bidirectional by confirmed (see Aldrich) (Sigma- vectors p3xFLAG-CMV-10 into cDNAs corresponding the cloning and byLys27Met was H3F3A established wild-type of Expression Flag-tagged mouse containing (Addgene) plasmid SPORT6 pCMV using ACVR1 a with cotransfected were cells iNHA transfectants, stable generate To 3 0 were grown in DMEM supplemented with 10% FBS (Invitrogen). Cells Cells were grown 10% with in FBS (Invitrogen). DMEM supplemented modeling of Lys27Met histone H3.3 and Gly328Val of H3.3 ACVR1. Lys27Met histone modeling for primer sequences). The c.983G>T mutation was generated generated was mutation c.983G>T The sequences). primer for mutants were created from human human from created were mutants Supplementary Table 8 Table Supplementary H3F3A Supplementary Table 8 Supplementary ) were created by site-direction mutagenesis and PCR PCR and mutagenesis site-direction by created were ) µ 9 g/ml g/ml hygromycin and were to pooled eliminate clonal . Copy number analysis was conducted for 40 DIPG DIPG 40 for conducted was analysis number Copy Cell counting on iNHA cells express cells iNHA on counting Cell for primer sequences). Clones were were Clones sequences). primer for for RT-PCR primer pairs. RT-PCRfor primer ACVR1 2 -transformed data or fold -transformed Nature Ge Nature cDNA using the the using cDNA Supplementary Supplementary H3f3a n cDNA. cDNA. iNHA etics ′ end end - - -

© 2014 Nature America, Inc. All rights reserved. arrays (TMAs) constructed in our laboratory or tissue sections were cut from sections or tissue in our laboratory constructed (TMAs) arrays (5 Sections ab4441). Abcam, dilution; (1:1,000 H3K9ac and 9727) Technology, Signaling Cell ab4729), dilution; Abcam, (1:1,000 dilution; H3K4me3 (1:1,000 H3K27ac 07-449), Millipore, dilution; (1:50 H3K27me3 against antibodies selected with stained nohistochemically Immunohistochemistry. for ALT-negativean as control. used were cells positive cancer cervical HeLa be to considered were C-circles. GMA47 ALT-positive DNA) fibroblasts were used as a positive control, and non-amplified and amplified greater than 0.2 (when comparing mean triplicate cycles of 95 °C for 15 s and 54 °C for 2 min. Samples with a delta mean was conducted using the following conditions: 95 qPCR °C for 15 min (Roche). followed by 480 35 LightCycler a on triplicate in run were ng) (2 DNA fied of the presence C-circles. of 5 units sample DNA, which was incubated for 8 h at 30 °C with master mix containing described previously as performed were assays C-circle DNA. sample with tandem in ALT the with kit Positive were and controls included run negative phenotype. the of indicative considered was kb) >50 to (<3 length telomere long of ence Briefly, 1.5 Assay kit (Roche). C-circles. TRF assays were using performed the TeloTAGGG Telomere of presence Length the or assay (TRF) fragment restriction telomere by determined of telomeres. lengthening of alternative Detection Versa Reader. Tunable Max Molecular Microplate Devices a using measured was Absorbance protocol. manufacturer’s the to according treated and BrdU with pulsed were cells Subsequently, h. 72 for ACVR1 Gly328Val or Arg206His wild-type, expressing virus with 10 incubated × (10 plate flat-bottom 96-well a in plated brief, In ­incorporation. Nature Ge Nature φ 14 29 polymerase. Quantitative PCR (qPCR) was employed to detect to detect employed was (qPCR) PCR Quantitative polymerase. 29 . Briefly, amplification of C-circles was conducted on 16 ng of of ng 16 on conducted was C-circles of amplification Briefly, . n etics Nes–tv-a φ Available surgical and autopsy material was immu and autopsy material Available surgical 29 polymerase–amplified product and product non-ampli 29 polymerase–amplified µ g g of DNAsample The lane. pres per was used ; ; Tp53 fl mouse brainstem progenitors were were progenitors brainstem mouse 3 cells/well) in triplicate and and triplicate in cells/well) C µ t m) of DIPG tissue micro tissue DIPG of m) values of The ALT phenotype was The ALT phenotype φ 29 polymerase– C t value ­ - - -

31. 28. 27. 26. 25. analysis. multivariate ( testing and significance model Student’s two-tailed unpaired by t test. calculated exact were Fisher’s data two-sided Continuous-scale by analyzed were comparisons Two-group software or SPSS v21 (IBM). OR values were as calculated previously shown analysis. Statistical (Ventana). kit II Hematoxylin the with counterstained were (Ventana).Slides kit Detection DAB Universal Ultraview auto the (Ventana)using the stainer XT with Benchmark mated performed was Immunodetection retrieval. antigen of pose pur in the for buffer citrate in ethanol treated heat were of sections water. Tissue distilled concentrations decreasing in immersion by hydration tissue by followed washes xylene several by removed Waxwas °C. 60 at overnight paraffin blocks and mounted on positively charged slides. wereSections baked 30. 29. test. test.

Bland, J.M.&Altman,D.G.Statisticsnotes.Theoddsratio. Fiume, M. R.E. Mills, mapping split-read informed pair-read PRISM: M. Brudno, & Y. Wang, Y., Jiang, Burrows-Wheeler with alignment long-read accurate and Fast R. Durbin, & H. Li, aa, K. Sasai, copy Detecting M. Brudno, & T. Smith, M., Dzamba, M., Fiume, P., Medvedev, (2000). (2007). 36 o bs-ar ee dtcin f neto, eein n srcua variants. structural genome. human the in and deletion insertion, of detection Bioinformatics level base-pair for transform. transformed astrocyte cells: implications for anti-glioma therapies. anti-glioma for implications cells: astrocyte transformed reads. short mated with variation number genomics. scale P values values et al. t al. et Bioinformatics et al. et ≤

0.05 were considered significant. Cox proportional hazards hazards proportional Cox significant. considered were 0.05 28 Savant Genome Browser 2: visualization and analysis for population- Natural genetic variation caused by small insertions and deletions and insertions small by caused variation genetic Natural 6mtygaieDA ehlrnfrs i dwrgltd in downregulated is methyltransferase O6-methylguanine-DNA Nucleic Acids Res. Acids Nucleic , 2576–2583 (2012). 2576–2583 , Statistical analysis was performed on GraphPad Prism 5 Prism GraphPad on performed was analysis Statistical Genome Res. Genome

26 , 589–595 (2010). 589–595 , α = 0.05) based on Wald the based for = 0.05) used was test

40

21 , W615–W621 (2012). W615–W621 , , 830–839 (2011). 830–839 , Genome Res. Genome

20 , 1613–1622 (2010). 1613–1622 , Br. Med.J. doi: 10.1038/ng.2936 Mol. Cancer Mol.

320 , 1468

3 6 1 - - , .