Medulloblastoma exome sequencing uncovers subtype-specific somatic mutations.
Journal: 2012/September - Nature
ISSN: 1476-4687
Abstract:
Medulloblastomas are the most common malignant brain tumours in children. Identifying and understanding the genetic events that drive these tumours is critical for the development of more effective diagnostic, prognostic and therapeutic strategies. Recently, our group and others described distinct molecular subtypes of medulloblastoma on the basis of transcriptional and copy number profiles. Here we use whole-exome hybrid capture and deep sequencing to identify somatic mutations across the coding regions of 92 primary medulloblastoma/normal pairs. Overall, medulloblastomas have low mutation rates consistent with other paediatric tumours, with a median of 0.35 non-silent mutations per megabase. We identified twelve genes mutated at statistically significant frequencies, including previously known mutated genes in medulloblastoma such as CTNNB1, PTCH1, MLL2, SMARCA4 and TP53. Recurrent somatic mutations were newly identified in an RNA helicase gene, DDX3X, often concurrent with CTNNB1 mutations, and in the nuclear co-repressor (N-CoR) complex genes GPS2, BCOR and LDB1. We show that mutant DDX3X potentiates transactivation of a TCF promoter and enhances cell viability in combination with mutant, but not wild-type, β-catenin. Together, our study reveals the alteration of WNT, hedgehog, histone methyltransferase and now N-CoR pathways across medulloblastomas and within specific subtypes of this disease, and nominates the RNA helicase DDX3X as a component of pathogenic β-catenin signalling in medulloblastoma.
Relations:
Content
Citations
(260)
References
(27)
Grants
(1)
Diseases
(2)
Chemicals
(18)
Organisms
(1)
Processes
(6)
Similar articles
Articles by the same authors
Discussion board
Nature. Aug/1/2012; 488(7409): 106-110

MEDULLOBLASTOMA EXOME SEQUENCING UNCOVERS SUBTYPE-SPECIFIC SOMATIC MUTATIONS

+34 authors

Abstract

Medulloblastomas are the most common malignant brain tumors in children1. Identifying and understanding the genetic events that drive these tumors is critical for the development of more effective diagnostic, prognostic and therapeutic strategies. Recently, our group and others described distinct molecular subtypes of medulloblastoma based on transcriptional and copy number profiles25. Here, we utilized whole exome hybrid capture and deep sequencing to identify somatic mutations across the coding regions of 92 primary medulloblastoma/normal pairs. Overall, medulloblastomas exhibit low mutation rates consistent with other pediatric tumors, with a median of 0.35 non-silent mutations per megabase. We identified twelve genes mutated at statistically significant frequencies, including previously known mutated genes in medulloblastoma such as CTNNB1, PTCH1, MLL2, SMARCA4 and TP53. Recurrent somatic mutations were identified in an RNA helicase gene, DDX3X, often concurrent with CTNNB1 mutations, and in the nuclear co-repressor (N-CoR) complex genes GPS2, BCOR, and LDB1, novel findings in medulloblastoma. We show that mutant DDX3X potentiates transactivation of a TCF promoter and enhances cell viability in combination with mutant but not wild type beta-catenin. Together, our study reveals the alteration of Wnt, Hedgehog, histone methyltransferase and now N-CoR pathways across medulloblastomas and within specific subtypes of this disease, and nominates the RNA helicase DDX3X as a component of pathogenic beta-catenin signaling in medulloblastoma.

Medulloblastomas are aggressive tumors of primitive neuroectodermal origin. More than one third of patients diagnosed with medulloblastoma succumb to their disease within 5 years6 and surviving patients often have significant long-term adverse effects from current therapies. Identifying the underlying genetic events responsible for medulloblastomas can help guide the development of more effective therapies and refine the selection of currently available chemotherapy and radiotherapy. Recent efforts profiling transcriptional and DNA copy number changes in medulloblastoma have provided insights into the biological processes involved in these tumors and have underscored the molecular heterogeneity of this disease24. Based on these data, four broad subgroups have been established, known according to a consensus nomenclature as SHH, WNT, Group 3 and Group 45.

Recently, Parsons et al. reported the first genome-scale sequencing of protein coding regions in medulloblastoma7. They identified alteration of genes encoding for histone modification proteins in 20% of cases, most notably MLL2 and MLL37. This initial survey was limited by a small discovery sample size (22 patients), lack of subtype-specific analysis, and use of Sanger sequencing technology insensitive to variants present at low allelic fraction. Here we survey coding somatic mutations at deeper coverage in a larger cohort of 92 medulloblastoma/normal pairs and assess these mutations in the context of specific molecular subtypes (Supplemental Table 1).

In total, 1,908 mutations were detected within 1,671 of 18,863 genes sequenced to a median of 106X coverage (Supplemental Table 2). Confirmation of 20 candidate mutations in selected genes (CTNNB1, DDX3X, SMARCA4, TP53, and CTDNEP1) was performed by amplification of 48 exons using a microfluidic PCR device (Fluidigm) followed by single-molecule real-time sequencing (SMRT, Pacific Biosciences) (Supplemental text). Sequence data was unavailable for one DDX3X mutation due to poor PCR amplification from the sample. All remaining 19 mutations were confirmed by this orthogonal method (median 73 redundant subreads, range 3–287, Supplemental Figure 1).

A median of 16 somatic mutations (12 non-silent, 4 silent) per tumor was identified, corresponding to a mutation rate of 0.35 non-silent mutations per megabase of callable sequence, less than most adult solid tumors and consistent with results from Parsons et al7. Six of the twelve most frequently mutated tumors were from the oldest patients (16–31 years at diagnosis), consistent with increased mutation frequency in adult versus childhood medulloblastomas (p=7.7×10−5, Wilcoxon rank-sum test, Supplemental Figure 2).

To identify genes mutated at statistically significant frequencies across our cohort, we utilized the MutSig algorithm8 which takes into account gene size, sample-specific mutation rate, non-silent to silent mutation ratios, clustering within genes, and base conservation across species. In our cohort of 92 samples, we identified 12 significantly mutated genes (q < 0.1, Table 1, Supplemental Table 3). Strikingly, these genes were not mutated in c5 (Group 3) and c4 (Group 4) tumors with extensive somatic copy number alteration (Figure 1), suggesting these subtypes are driven primarily by structural variation, rather than base mutation. Not unexpectedly, CTNNB1 (beta-catenin) and PTCH1 were the two most significantly mutated genes (see Table 1, Figure 1). Point mutations of CTNNB1 in combination with loss of chromosome 6 were found in all WNT subgroup tumors and were concurrent with several other recurrently mutated genes, namely CSNK2B, DDX3X, TP53 and SMARCA4. Mutations involving PTCH1 occurred exclusively in SHH subgroup tumors and mutations of genes associated with the Hedgehog (Hh) pathway were also restricted to this subgroup (p < 0.0001, Fisher’s exact test). All but one of the tumors with PTCH1 mutations had somatic loss of 9q, resulting in hemizygosity for the mutant allele. The remaining tumor had apparent copy neutral loss-of-heterozygosity of 9q22. Other somatic mutations of Hh pathway members include a splice site mutation in SUFU, an in-frame deletion in WNT6, and missense mutations in GLI2, SMO, PRKACA, WNT2, and WNT2B.

Two patients with SHH subgroup tumors had germline variants in PTCH1, one with somatic loss of 9q resulting in hemizygosity for a loss-of-function germline allele (MD-085, c.3030delC, p.Asn1011Thrfs*38) and the other with a substitution previously reported in patients with holoprosencephaly (MD-286, p.T1052M9). Two additional cases (MD-097 and MD-335) had loss-of-function variants in SUFU (1 frameshift deletion and 1 nonsense) that began as heterozygotes in the germline and became hemizygous in the tumor, due to somatic loss of chromosome 10 in one case and copy neutral loss-of-heterozygosity in the other.

MLL2 was also subject to recurrent inactivating mutations, consistent with findings by Parsons et al7 and providing further evidence for dysregulated histone modification in medulloblastoma. Indeed, six of the twelve most significantly mutated genes are involved in histone modification and/or related chromatin remodeling complexes (MLL2, GPS2, KDM6A, BCOR, SMARCA4, and LDB1; see Table 1). As a gene set, histone methyltransferases (HMTs) were enriched for somatic mutation with 21 tumors having apparent, predominantly loss-of-function, HMT mutations (q = 5.8×10−9; Figure 2, Supplemental Table 4).

Subtype-specific MutSig analysis identified additional significant mutations of histone modifying genes, MLL3 and HDAC2, in Group 4 tumors along with KDM6A mutations (q=0.039 and 0.066, see Supplemental Table 3). Mutations in KDM6A, interestingly, occurred exclusively in tumors with an i17q as the sole autosomal alteration (p = 0.0023, Fisher’s exact test) with the one female case with KDM6A mutation also having loss of a chromosome X. Notably, the two ‘i17q only’ tumors without KDM6A mutations had other histone modifying enzymes mutated, namely THUMPD3, ZMYM3 and MLL3, perhaps suggesting a distinct biology for tumors with this karyotype.

Mutations in several genes encoding components of the nuclear co-repressor (N-CoR) complex were observed at a statistically significant frequency: BCOR in 3 tumors, GPS2 in 3 tumors, and LDB1 in 2 tumors. BCOR mutations have recently been reported at high frequency in retinoblastoma10 and in ‘copy-neutral’ acute myelogenous leukemia11. BCOR is located on the X-chromosome and two hemizygous frameshift mutants were found in tumors from males (allele fractions 0.90 and 0.92). A third nonsense mutation was also found in a male but at low allelic fraction (0.12), suggesting a subclonal event. Two of three BCOR mutations occurred in SHH subgroup tumors. LDB1 missense and nonsense mutations were found in two additional SHH tumors, both appearing hemizygous due to loss of 10q and complete chromosome 10 loss, respectively (allele fractions 0.81 and 0.78). Both BCOR and LDB1 promote assembly of the repressive N-CoR complex12 and harbor apparent loss of function mutations. GPS2, which encodes a critical subunit of the N-CoR complex, a repressor of JNK/MAPK signaling through partnership with histone deacetylases12, was mutated in two Group 3 tumors. The GPS2 mutations cluster within amino acids 53 – 90, the domain critical for heterodimerization with NCOR2 (SMRT) and interacting with a TBL1-NTD tetramer to assemble the N-CoR repression complex12. Finally, an additional nonsense mutation in NCOR2 was identified in a single SHH subgroup tumor, underscoring the central role of N-CoR dysregulation in medulloblastoma development and particularly within the SHH subgroup.

Several genes encoding subunits of the SWI/SNF-like chromatin-remodeling complex were also mutated in our cohort, including significant recurrent mutations of SMARCA4 (Brg/BAF190), which encodes a DNA helicase with ATPase activity13 and has been reported to be mutated in lung, ovarian, and pancreatic cancers14 as well as medulloblastoma15. In our cohort, SMARCA4 (Brg/BAF190) mutations clustered in helicase domains and occurred in three Group 3 tumors (significant within the c1 subtype, q=0.019), and one WNT tumor. In addition, mutations were found in the alternative ATPase subunit SMARCD2 (Brm) (missense at a highly conserved residue) and two other members of the SWI/SNF complex, ARID1B (BAF250b) (2 bp frameshift deletion) and SMARCC2 (BAF170) (splice site). These were all apparent loss-of-function mutations and occurred in SHH tumors. Thus, it appears that disruption of this complex is frequent across medulloblastomas.

Novel and hemizygous mutations were found in CTDNEP1 (previously known as DULLARD), a phosphatase with roles in Xenopus neural development through regulation of BMP receptors16, and as a direct regulator of LIPIN, an integral component of the mTOR complex17. CTDNEP1 mutations were found in two Group 3 tumors (significant within the subtype, q=0.0087), a 2 bp frameshift deletion and a substitution disruptive of a splice site. Both tumors have i17q chromosomes resulting in loss of the wild type allele at 17p13.

Mutations in DDX3X, an ATP-dependent RNA helicase with functions in transcription, splicing, RNA transport and translation18, were found in seven tumors, including half of the WNT pathway tumors (p = 0.005, Fisher’s exact test) and several SHH subgroup tumors. DDX3X mutations have recently been reported at low frequency in five other tumor types (Catalogue of Somatic Mutations in Cancer, COSMIC15) but the significance of these mutations for DDX3X function remains unclear. To understand the consequence of observed point mutations on DDX3X’s physical structure, we mapped the mutations onto the previously reported crystal structure of DDX3X19 and its ortholog Vasa/DDX420 (Figure 3a; Supplemental Figure 3; Supplemental table 5). The mutations appear to cluster in two structural domains, a helicase ATP-binding domain (residues 211–403) and a helicase C-terminal domain (residues 414–575). The location of these mutations suggests that they may alter DDX3X-RNA interaction (Figure 3a; Supplementary table 5).

As half of the beta-catenin mutated tumors contained concurrent DDX3X mutations, we investigated whether DDX3X could enhance beta-catenin’s ability to transactivate a TCF4-luciferase reporter (TOPflash) and if DDX3X/beta-catenin co-expression had a measureable effect on cell viability/proliferation. In combination with wild-type beta-catenin, neither wild type nor mutant DDX3X alone significantly transactivated the TOPflash reporter. However, in combination with mutant beta-catenin (S33Y substitution), the majority of DDX3X point mutants in our cohort potentiated reporter activity (Figure 3b; p < 0.05). This potentiation was also apparent in cell viability assays in both HeLa (data not shown) and D425 medulloblastoma cell lines (Figure 3c; p < 0.05).

Given the apparent importance of DDX3X mutations in medulloblastoma, we searched the genes listed in the RNA Helicase Database30 for low frequency mutations in medulloblastoma. We found five tumors with mutations within RNA helicase or RNA binding domains of DHX9, DHX32, DHX57, FANCM and SKIV2L (Figure 2, Supplemental Table 6). The missense mutations were at conserved residues and predicted to be deleterious by SIFT, AlignGVGD, and PolyPhen2. In addition, a frameshift insertion in SETX occurs upstream of, and likely disrupts, its RNA helicase domain. Overall, 15% of medulloblastomas appear to have some disruption of RNA helicase activity.

In summary, we report a next-generation sequencing analysis of medulloblastoma, the most common malignant brain tumor in children. Our results reveal mutations in several known pathways such as histone methylation (MLL2 and others), sonic hedgehog (PTCH1, SUFU, and others) and Wnt (CTNNB1 and others), and also mutations in novel genes including DDX3X, BCOR, LDB1, and GPS2. Our preliminary functional studies implicate DDX3X as a candidate component of pathogenic WNT/beta-catenin signaling. In a broader sense, DDX3X mutations have recently been reported in chronic lymphocytic leukemia21 and head and neck cancers22, both of which have subsets of tumors with dysregulated WNT signaling. Studies investigating whether mutant DDX3X functions complicit with beta-catenin in these contexts should provide additional insights into this multifaceted molecule and open potential avenues for novel therapies. Finally, the delineation of nuclear receptor co-repressor complex molecules as altered in medulloblastoma provides new insight into the pathogenesis of this deadly childhood disease.

Methods Summary

Informed consent was provided by families of medulloblastoma patients treated at Children’s Hospital Boston, The Hospital for Sick Children Toronto and institutions contributing to the Children’s Oncology Group/Cooperative Human Tissue Network, under approval and oversight by their respective Internal Review Boards. All tumors were obtained at the initial surgical resection and recurrent tumors were excluded from our analysis. Hematoxylin and eosin stained slides of tumor samples were pathologist reviewed to confirm the diagnosis of medulloblastoma, determine histological subtype when able, and assess tumor purity. DNA was isolated from tumor specimens and matched peripheral blood as previously described2. Exome sequencing of DNA from 92 tumor/normal pairs was performed using in-solution hybrid-capture of 193,094 exons from 18,863 miRNA- and protein-coding genes, followed by sequencing of 76 bp paired-end reads using Illumina sequencing-by-synthesis technology23. Reads were aligned to human genome build GRCh3724 using Burrows-Wheeler aligner25. The ~33 Mb target region was sequenced to 106X mean coverage in each sample (range 73 – 234). Gene expression data and copy number profiles (derived from SNP microarrays or sequence data) were used to assign each tumor to a subgroup using published criteria2. Our cohort consisted of 6 WNT(c6), 23 SHH(c3), 33 Group 3(12 c1, 21 c5), and 30 Group 4(12 c2, 18 c4) tumors (see Supplemental Table 1 for case annotations). Mutations were detected using muTect, annotated using Oncotator26, and manually reviewed using the Integrated Genomics Viewer (IGV)27. For validation, PCR on Access Array microfluidic chips (Fluidigm) was followed by single-molecule real-time sequencing (Pacific Biosciences) per manufacturer’s instructions. Subreads were extracted and assigned to samples using manufacturer’s and custom software, and aligned to the hg19 (GRCh37) build of the human reference genome sequence using BWA-SW25. Candidate mutations were confirmed by manual review using IGV27 (Supplemental Figure 1). See Supplementary Information and http://www.broadinstitute.org/cancer/cga/ for complete descriptions of materials and methods.

Supplementary Material

Figure 1

Demographic characteristics, molecular subtypes and selected copy number alterations and somatic mutations across 92 medulloblastoma cases

Data tracks describing 92 medulloblastoma cases. Identifier: Unique name used to denote each case. Identifiers also link samples to those analyzed by Cho et al. Sex: males in blue, females in pink. Age: years of age at diagnosis binned as infants, children, or adults. Histology: pathology review of primary tissue specimen. Subtypes: based on copy number profiles derived from sequence or microarray data. Consensus subtypes from Taylor et al. Cho et al subtypes as published. Copy number alterations: Selected copy number alterations used to assign tumors to subtypes. Losses are blue. Gains are red. Somatic mutations: Gene names (HUGO symbols) grouped by functional category. MutSig gene names are in bold. Missense mutations are black, nonsense/splice site/indel mutations are orange, silent mutations are purple, and germline variants are green.

Figure 2

Location of mutations in histone methyltransferases, RNA helicases, and N-CoR complex-associated genes

Location of somatic mutations on linear protein domain models of genes from sets frequently mutated in medulloblastoma. All domain annotations are from UniProt and InterPro annotations. Diagrams were constructed using Domain Graph (DOG)28, version 2.0. a. Histone methyltransferase domains: red = SET, green = coiled-coil, blue = zinc-finger, and cyan=other. b. N-CoR complex-associated domains: purple = anti-parallel coiled-coil domains required for GPS2/NCOR2 (SMRT) interaction12, yellow = other interaction domains as labeled: SANT domains binds DNA; CoRNR domains binds nuclear receptors; ANK repeats mediate a diversity of protein-protein interactions, and LIM-binding domains bind a common protein structural motif. c. RNA helicase domains: cyan = helicase and helicase-associated (InterPro), red = RNA-binding and RNA polymerase sigma factor (InterPro), blue = ATP binding site, and green = DEAD or DExH box motif. See Supplemental Table 1 for UniProt protein model identifiers.

Figure 3

Functional consequence of DDX3X point mutations

a, Three-dimensional model of the two recA-like domains of human DDX3X in complex with single-stranded RNA and a Mg-ATP analog. Displayed are the residues mutated in the N-terminal recA-like domain (R276K, D354H, R376C) and C-terminal recA-like domain (D506Y, R528H, R534H, P568L). Coloring: DDX3X residues 166–405 (light blue); DDX3X residues 406–582 (dark blue); single-stranded RNA (cyan); Mg-ATP analog (magenta and green). Molecular graphics images were produced using the University of San Francisco Chimera package29 (http://www.cgl.ucsf.edu/chimera). b, Mutant DDX3X potentiates mutant beta-catenin transactivation of TOPflash promoter. Represented is relative luciferase activity in 293T cells co-transfected with TOPflash reporter, FOPflash control, and either wild type or mutant DDX3Xs in combination with wild type or mutant beta-catenin. One-dimensional model of DDX3X displayed about bar graphs to illustrates the position of the mutations. c, Cell viability assays of medulloblastoma D425 cells stably transduced with either wild type or mutant DDX3X lentivirus in combination with either wild type or mutant beta-catenin lentivirus.

For b and c, error bars depict the standard deviation of the mean from 5 replicate experiments performed for each condition. Student’s t-tests were performed to evaluate significance of differences in TOPflash intensity or cell proliferation value distributions as follows: increases with DDX3X alone vs. empty vector, increases with wtBetaCat vs. DDX3X alone, increases with mutBetaCat vs. DDX3X alone, and increases with mutBetaCat vs. wtBetaCat.

Table 1
Genes mutated at a statistically significant frequencyin 92 medulloblastomas
GeneDescriptionMutationsPatientsUnique sitesSilentMissenseIndel or nullDouble nullq
CTNNB1beta-catenin6640600<1.8×10−11
PTCH1patched homolog 1(Drosophila)77700704.0×10−9
MLL2myeloid/lymphoid or mixed-lineage leukemia 21081002444.0×10−9
DDX3XDEAD box polypeptide 3, X-linked77707002.3×10−8
GPS2G protein pathway suppressor 233301201.2×10−4
TP53tumor protein p5333303000.039
KDM6AUTX, lysine (K)-specific demethylase 6A33302100.042
BCORBCL6 co-repressor33300300.046
SMARCA4ATP-dependent helicase44304000.046
LDB1LIM domain binding 122201100.047
CTDNEP1(DULLARD)CTD nuclear envelope phosphatase 122200200.047
CSNK2Bcasein kinase 2, beta polypeptide22202000.071

null = nonsense, frameshift, or splice-site mutations; Double null = null mutations co-occurring in a single tumor; q = q-value, False Discovery Rate (Benjamini-Hochberg procedure). See Supplemental Table 3 for further statistics and subtype analysis.

Footnotes

Sequence data used for this analysis are available in dbGaP under accession phs000504.v1.p1. Reprints and permissions information is available at www.nature.com/reprints. MM is a paid consultant for and equity holder in Foundation Medicine, a genomics-based oncology diagnostics company, and is a paid consultant for Novartis. YJC has served on an advisory board for Novartis.

Author contributions

YJC, MM and SLP conceived the project. YJC, TJP, MM and SLP wrote the manuscript with input from coauthors. DW, TA, JPF, SS, NT, YJC, AK and FY performed functional characterization studies. DPK generated in silico structural modeling of DDX3X mutations. TJP conducted the bioinformatic analysis, supported by SLC, PS, KC, MSL, AM, AHR, AS, HG, PT, JPM, NJ, and DTWJ. DA, ES, SGB, and GG facilitated transfer, sequencing, and analysis of samples. PN and MDT provided tissues for analysis. YJC, JPF and VA processed tumor and blood samples for study. GC generated reagents used in functional characterization studies. PL, SMP, and TMR assisted with interpretation of results. JB, MOC, RE, NJL, JM, MR, CR and BS performed microfluidic PCR and single molecule real-time sequencing for validation analysis.

Acknowledgments

This work was supported by NIH grants NHGRI U54HG003067 to ES Lander (ES, DA, SBG, GG, MM); R01CA109467(SLP, JPM); R01CA105607(HG, TMR, MM, SLP); P30 HD18655(SLP); R01 CA030002 and CA050661 (TMR); R01 NS046789(GRC); R01 CA154480(PT); R25NS070682(SS) and R01CA148699(MDT); St. Baldrick’s Foundation Scholar Award and the Beirne Faculty Scholar endowment and Center for Children’s Brain Tumors at Stanford University (YJC); German Cancer Aid (109252) and the BMBF ICGC-PedBrain project (NJ, DTWJ, PL, SMP); HHMI (GRC); the Pediatric Brain Tumor Foundation(MDT); Canadian Institutes of Health Research Fellowship (TJP); Restracomp funding from the Hospital for Sick Children (PAN); and the Mullarkey Research Fund (SLP). We thank Children’s Oncology Group and the Cooperative Human Tissue Network for providing tumor samples, the staff of the Broad Institute Biological Samples, Genome Sequencing and Genetic Analysis Platforms for their assistance in genomic processing of samples and generating the sequencing data used in this analysis, K. Keho and M. Brown at Pacific Biosciences for technical support with sample barcoding methods, and L. Gaffney of Broad Institute Communications for assistance with figure layout and design.

References

  • 1. CBTRUS Statistical Report: Primary Brain and Central Nervous System Tumors Diagnosed in the United States in 2004–2007Central Brain Tumor Registry of the United StatesHinsdale, IL2011at <www.cbtrus.org>
  • 2. ChoYJIntegrative Genomic Analysis of Medulloblastoma Identifies a Molecular Subgroup That Drives Poor Clinical OutcomeJournal of Clinical Oncology29142414302011[PubMed][Google Scholar]
  • 3. KoolMIntegrated Genomics Identifies Five Medulloblastoma Subtypes with Distinct Genetic Profiles, Pathway Signatures and Clinicopathological FeaturesPLoS ONE3e30882008[PubMed][Google Scholar]
  • 4. RemkeMAdult medulloblastoma comprises three major molecular variantsJ Clin Oncol29271727232011[PubMed][Google Scholar]
  • 5. TaylorMDMolecular subgroups of medulloblastoma: the current consensusActa Neuropathologica2011[PubMed][Google Scholar]
  • 6. SmollNRRelative survival of childhood and adult medulloblastomas and primitive neuroectodermal tumors (PNETs)Cancer[PubMed][Google Scholar]
  • 7. ParsonsDWThe Genetic Landscape of the Childhood Cancer MedulloblastomaScience3314354392010[PubMed][Google Scholar]
  • 8. GetzGComment on “The Consensus Coding Sequences of Human Breast and Colorectal CancersScience31715002007[PubMed][Google Scholar]
  • 9. MingJMutations in PATCHED-1, the receptor for SONIC HEDGEHOG, are associated with holoprosencephalyHuman Genetics1102973012002[PubMed][Google Scholar]
  • 10. ZhangJA novel retinoblastoma therapy from genomic and epigenetic analysesNature4813293342012[PubMed][Google Scholar]
  • 11. GrossmannVWhole-exome sequencing identifies somatic mutations of BCOR in acute myeloid leukemia with normal karyotypeBlood118615361632011[PubMed][Google Scholar]
  • 12. OberoiJStructural basis for the assembly of the SMRT/NCoR core transcriptional repression machineryNat Struct Mol Biol181771842011[PubMed][Google Scholar]
  • 13. BaekSHRegulated subset of G1 growth-control genes in response to derepression by the Wnt pathwayProc Natl Acad Sci U S A100324532502003[PubMed][Google Scholar]
  • 14. WilsonBGRobertsCWMSWI/SNF nucleosome remodellers and cancerNat Rev Cancer114814922011[PubMed][Google Scholar]
  • 15. FutrealPAA census of human cancer genesNat Rev Cancer41771832004[PubMed][Google Scholar]
  • 16. SatowRKurisakiAChanT-chuanHamazakiTSAsashimaMDullard Promotes Degradation and Dephosphorylation of BMP Receptors and Is Required for Neural InductionDevelopmental Cell117637742006[PubMed][Google Scholar]
  • 17. PetersonTRmTOR Complex 1 Regulates Lipin 1 Localization to Control the SREBP PathwayCell1464084202011[PubMed][Google Scholar]
  • 18. GarbelliABeermannSDi CiccoGDietrichUMagaGA Motif Unique to the Human Dead-Box Protein DDX3 Is Important for Nucleic Acid Binding, ATP Hydrolysis, RNA/DNA Unwinding and HIV-1 ReplicationPLoS ONE6e198102011[PubMed][Google Scholar]
  • 19. HögbomMCrystal structure of conserved domains 1 and 2 of the human DEAD-box helicase DDX3X in complex with the mononucleotide AMPJ Mol Biol3721501592007[PubMed][Google Scholar]
  • 20. SengokuTNurekiONakamuraAKobayashiSYokoyamaSStructural basis for RNA unwinding by the DEAD-box protein Drosophila VasaCell1252873002006[PubMed][Google Scholar]
  • 21. WangLSF3B1 and Other Novel Cancer Genes in Chronic Lymphocytic LeukemiaNew England Journal of Medicine365249725062011[PubMed][Google Scholar]
  • 22. StranskyNThe Mutational Landscape of Head and Neck Squamous Cell CarcinomaScience333115711602011[PubMed][Google Scholar]
  • 23. BentleyDRAccurate whole human genome sequencing using reversible terminator chemistryNature45653592008[PubMed][Google Scholar]
  • 24. Genome Reference Consortiumat <http://www.ncbi.nlm.nih.gov.ezp-prod1.hul.harvard.edu/projects/genome/assembly/grc/index.shtml>
  • 25. LiHDurbinRFast and accurate short read alignment with Burrows-Wheeler transformBioinformatics25175417602009[PubMed][Google Scholar]
  • 26. RamosAlexOncotatorat <http://www.broadinstitute.org/oncotator/>
  • 27. RobinsonJTIntegrative genomics viewerNat Biotech2924262011[Google Scholar]
  • 28. RenJDOG 1.0: illustrator of protein domain structuresCell Res192712732009[PubMed][Google Scholar]
  • 29. PettersenEFUCSF Chimera--a visualization system for exploratory research and analysisJ Comput Chem25160516122004[PubMed][Google Scholar]
  • 30. JankowskyAnjaGuentherUlf-PeterJankowskyEckhard2011The RNA helicase databaseNucl Acids Res39suppl 1D338D341first published online November 25, 2010[PubMed][Google Scholar]
Collaboration tool especially designed for Life Science professionals.Drag-and-drop any entity to your messages.