Content uploaded by Mario Capasso
Author content
All content in this area was uploaded by Mario Capasso on Feb 03, 2020
Content may be subject to copyright.
Somatic mutations in specific and connected subpathways
are associated with short neuroblastoma patients’survival
and indicate proteins targetable at onset of disease
Maria Rosaria Esposito
1
, Andrea Binatti
2
, Marcella Pantile
1
, Alessandro Coppe
3
, Katia Mazzocco
4
, Luca Longo
5
,
Mario Capasso
6,7,8
, Vito Alessandro Lasorsa
7
, Roberto Luksch
9
, Stefania Bortoluzzi
2
and Gian Paolo Tonini
1
1
Neuroblastoma Laboratory, Fondazione Istituto di Ricerca Pediatrica Città della Speranza, Padua, Italy
2
Department of Molecular Medicine, University of Padua, Padua, Italy
3
Department of Women’s and Children’s Health, University of Padova, Padua, Italy
4
Translational Research Department, Laboratory Medicine, Diagnostics and Services U.O.C. Pathological Anatomy, IRCCS Giannina Gaslini Institute,
Genoa, Italy
5
U.O.C. Bioterapie, Ospedale Policlinico San Martino, Genoa, Italy
6
Department of Molecular Medicine and Medical Biotechnology, University of Naples Federico II, Naples, Italy
7
CEINGE Biotecnologie Avanzate, Naples, Italy
8
IRCCS SDN, Istituto di Ricerca Diagnostica e Nucleare, Naples, Italy
9
Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy
Neuroblastoma (NB) is an embryonic malignancy of the sympathetic nervous system with heterogeneous biological,
morphological, genetic and clinical characteristics. Although genomic studies revealed the specific biological features of NB
pathogenesis useful for new therapeutic approaches, the improvement of high-risk (HR)-NB patients overall survival remains
unsatisfactory. To further clarify the biological basis of disease aggressiveness, we used whole-exome sequencing to examine
the genomic landscape of HR-NB patients at stage M with short survival (SS) and long survival (LS). Only a few genes,
including SMARCA4,SMO,ZNF44 and CHD2, were recurrently and specifically mutated in the SS group, confirming the low
recurrence of common mutations in this tumor. A systems biology approach revealed that in the two patient groups, mutations
occurred in different pathways. Mutated genes (ARHGEF11,CACNA1G, FGF4,PTPRA,PTK2,ANK3,SMO,NTNG2, VCL and NID2)
regulate the MAPK pathway associated with the organization of the extracellular matrix, cell motility through PTK2signaling
and matrix metalloproteinase activity. Moreover, we detected mutations in LAMA2,PTK2,LAMA4, and MMP14 genes, impairing
MET signaling, in SFI1and CHD2involved in centrosome maturation and chromosome remodeling, in AK7and SPTLC2, which
regulate the metabolism of nucleotides and lipoproteins, and in NALCN,SLC12A1,SLC9A9, which are involved in the transport
of small molecules. Notably, connected networks of somatically mutated genes specific for SS patients were identified. The
detection of mutated genes present at the onset of disease may help to address an early treatment of HR-NB patients using
FDA-approved compounds targeting the deregulated pathways.
Key words: neuroblastoma, whole-exome sequencing, somatic mutations, pathways, gene networks, target protein
Additional Supporting Information may be found in the online version of this article.
†
M.R.E. and A.B. contributed equally to this work
Maria Rosaria Esposito’s current address is: Fondazione Istituto di Ricerca Pediatrica Città della Speranza; Department of Industrial
Engineering (DII), University of Padua Corso Stati Uniti, 4 35127, Padua, Italy
Conflict of interest: The authors declare that they have no conflict of interest.
Grant sponsor: Fondazione Italiana per la Lotta al Neuroblastoma; Grant sponsor: Fondazione Cassa di Risparmio di Padova e Rovigo;
Grant sponsor: Ministero dell’Istruzione, dell’Università e della Ricerca; Grant numbers: 2010NYKNS7_002; Grant sponsor: University of
Padova
DOI: 10.1002/ijc.31748
History: Received 27 Oct 2017; Accepted 21 Jun 2018; Online 11 Jul 2018
Correspondence to: Maria Rosaria Esposito, Fondazione Istituto di Ricerca Pediatrica Città della Speranza, Neuroblastoma Laboratory, Corso
Stati Uniti, 4 35127, Padua, Italy, E-mail: mr.esposito@irpcds.org; or Stefania Bortoluzzi, Department of Molecular Medicine, University of
Padua, Padua, Italy, E-mail: stefania.bortoluzzi@unipd.it
International Journal of Cancer
IJC
Molecular Cancer Biology
Int. J. Cancer: 143, 2525–2536 (2018) ©2018 UICC
Introduction
Neuroblastoma (NB) is a pediatric cancer of the sympathetic
nervous system. Metastatic disease (stage M) usually
involves liver, skin, the bone marrow and/or skeleton.
1
Stage
M patients are classified as high-risk NB (HR-NB) and show
an overall survival lower than 40% at 5 years from diagno-
sis.
2
The majority of HR-NB stage M patients responds well
to the first-line therapy, but relapse occurs in the majority
of patients. Several studies demonstrated that both copy
number aberrations (CNAs)
3–5
and genomic variants
6–11
contribute to the tumor aggressiveness. In 2012, Stigliani
et al.
4
investigated the genomic features of HR-NB, dividing
patients into a short-survival (SS; with disease progression
and survival at most 5 years from diagnosis) and a long-
survival group (LS; responsive to the treatments and a sur-
vival over 5 years from diagnosis). The study showed that
tumor cells from the SS group are characterized by a high
number of structural CNAs and high chromosomal instabil-
ity. A previous study showed that, beyond CNAs, also muta-
tions hitting specific pathways could be implicated in HR-
NB progression.
9
Afterthislineofevidence,toelucidatethe
biology underlying differences between the SS and LS
groups in term of different outcome, our study characterized
by whole-exome sequencing (WES) the genomic landscape
of primary tumors, with focus to single nucleotide variants
(SNVs) and indels. Aberrant somatic mutations exclusive
for the SS or LS patients were found, as well as pathways
and subpathways that are specifically targeted in SS tumors,
which were confirmed by the analysis of a large independent
cohort.
8
Materials and Methods
Patients and tumor samples
A cohort of stage M NB patients from the Italian Neuro-
blastoma Registry with complete clinical data and follow-up
over 10 years was considered. Frozen tissue from the pri-
mary tumor at onset was available for each patient. Patients
were stratified into two groups according to their overall
survival: the SS group (n= 14), including patients with
rapid disease progression and rapid fatal outcome, all with a
survival time < 60 months, and the LS group (n= 15),
including patients who are responsive to therapy and sur-
vived at least 60 months from diagnosis. Five SS patients
(ID2475; ID2368; ID2181; ID1995; ID2100) were also
included in the previous NB report.
9
Informed consent was
received for the use of biological material from legal tutors,
and the study was approved by the Institutional Board of
the participating Institutions. Total genomic DNAs (gDNA)
from 29 tumors and matched constitutional DNA of
patients was purified according to the standard protocol
with Invisorb
®
Spin Tissue Mini Kit (SPA-Stratec molecu-
lar). The amount and quality of gDNA were assessed by
Nanodrop and Qubit Instruments (Invitrogen), respectively,
and only high-quality samples (DNA/protein ratio, A260/
A280: 1.8–2.0) were processed. All tumor samples were clas-
sified as NB Schwannian stroma-poor according to criteria
established by the International Neuroblastoma Pathology
Committee.
1
The presence of at least 60% of neuroblasts in
tumor samples was verified.
Exome library preparation and WES
For each sample, 100 ng of DNA (determined by Qubit
®
2.0
Fluorometer) was used for exome library preparation by the
AmpliSeq™exome kit (Life Technologies) targeting approxi-
mately 35 Mb of human exons. Briefly, the gDNA was ampli-
fied by oligo pools/primers to perform ultra-high multiplex
PCR enrichment of the exonic regions of the genome. Next,
the amplicons were ligated to adaptors with Ion Xpress Bar-
code Adapters Kit and purified with Agencourt AMPure XP
kit (Beckman Coulter Genomics). The library was quantified
with Quantitation RT-PCR with the Ion Library Quantitation
Kit (Life Technologies), diluted to 100 pM and loaded on a P1
chip for Ion Proton Sequencing according to the manufac-
turer’s protocol.
WES variant calling
Read mapping and variant calling were performed with Tor-
rent Suite and Ion ReporterTM software, provided by the Ion
ProtonTM System. The Proton Run Browser was used for
quality control metrics (percent bead loading, usable
sequences, read length, alignment metrics to hg19 reference
genome and mean raw accuracy). The samples were processed
using the workflow: “Somatic –Proton –High Stringency
Configuration”. Bam files of the tumor and blood samples of
each patient were uploaded to Ion ReporterTM (IR) software
using the available plug-in, IonReporterUploader_V1_2. Vari-
ant calling was done using Torrent Variant Caller (v. 5.0–9).
Next, the files were processed using a workflow AmpliSeq
What’s new?
Most patients with metastatic neuroblastoma don’t survive 5years from diagnosis, despite responding well to first-line
treatments. Previous work comparing short-survival and long-survival patients identified some key chromosomal
differences. These authors take the search deeper, conducting whole-exome sequencing to compare somatic mutations
between patients who survived at least 5years and those who did not. They determined that mutations among the short-
survival group affected different pathways than those afflicting the long-survival patients. In some cases, drugs already
exist that target these proteins, suggesting that testing for these mutations at the time of diagnosis could indicate specific
treatments.
Molecular Cancer Biology
2526 Mutated pathways in aggressive neuroblastoma
Int. J. Cancer: 143, 2525–2536 (2018) ©2018 UICC
Exome paired sample (tumor/normal) to subtract variants
[SNV, multiple nucleotide variant (MNV), indel and copy
number variant (CNV)] discovered in the peripheral blood
DNA against the tumor DNA.
Variant annotation and prioritization
The annotation of somatic variants was performed by
SnpSift
12
and SnpEff.
13
SnpSift’sannotate command provided
the association of known variants to dbSNP (v. 147) and
COSMIC (v. 77)
14
identifiers, clinical significance from Clin-
var (updated on May 02, 2016),
15
, as well as functional pre-
diction indicated by MetaSVM and MetaLR
16
from dbNSFP
database (v. 2.9.1).
17
The two algorithms predict whether the
variant is tolerated or deleterious, considering nine scores
present in dbNSFP (SIFT, Polyphen –2, GERP ++, Mutation-
Taster, MutationAssessor, FATHMM, LRT, SiPhy and Phy-
loP) and MMAF observed in different populations of 1,000
genomes. SnpEff predicts the functional and putative impact
of detected variants. Known variants annotated in Clinvar as
benign or likely benign were discarded, and only variants
with HIGH or MODERATE SnpEff predicted impact were
further considered. After integration of exome sequencing
and Exome Aggregation Consortium (ExAC, Cambridge,
MA, http://exac.broadinstitute.org)
18
data, variants with a
non-Finnish European population allele frequency > 5% were
discarded (Supporting Information Fig.S1). The interpreta-
tion of variants’impact was also obtained by mapping
selected variants to protein sequences and their domain
annotation using MutationMapper (http://www.cbioportal.
org/mutation_mapper.jsp). Additional structural predictions
and analysis on the mutated protein sequences were obtained
using Phyre2.
19
Pathway analyses
Genes mutated in SS and in LS were separately mapped to
the KEGG
20
and Reactome
21
pathways. The pathways with
at least three genes mutated in a group and none in the
other group were defined as “group-specificpathways.”
Significant pathway enrichment was calculated considering
the separately mutated genes in the SS and LS patient
groups. Significantly enriched pathways only in one group
and with a number of genes mutated in the group by at
least 1.5x the number of genes mutated in the other group
were considered to be “group-specifically enriched path-
ways.”The detected pathways were organized after the
architecture of the KEGG and Reactome databases to have
a less redundant description of altered molecular signaling
and biological functions, gathering pathways into func-
tional classes.
Gene network analysis
The R Graphite Bioconductor package (v. 1.20.1)
22
was used
to convert complex pathway topologies into Reactome
pathway-derived gene networks using appropriate biology-
driven rules to transform different types of direct and indirect
relations between genes and gene products annotated in path-
ways (i.e., regulatory relations, participation to molecular
complexes and biosynthetic pathways, also with compound
intermediates) into pairwise gene connections. Reactome net-
works were merged into a pathway-derived gene network of
186,808 pairwise interactions between 8,678 genes. We
applied the HotNet2 algorithm
23
to Reactome-derived gene
networks to statistically identify group-specificsubnetworks
of mutated protein-coding genes, defining groups of func-
tionally related genes in which mutations significantly con-
verge (Supporting Information Fig. S1). HotNet2 consists of
an insulated heat diffusion model to detect significantly
mutated gene subnetworks, evaluating both the heat score of
nodes and the local network topology. The heat score for
each node was calculated from the number of samples carry-
ing prioritized somatic variants in the corresponding gene.
Hotnet2 analysis was first conducted using all the somatically
mutated genes in the whole cohort of patients, and then the
two patient groups were considered separately. Due to the
cardinality of our cohort and the considerable dimension of
the considered network, multiple testing correction applied in
this analysis, considerably increases the p-value. Gene groups
emerging from Hotnet2 analysis were also linked to Gene
Ontology biological processes (The Gene Ontology Consor-
tium, 2015).
Variants validation by ultra-deep sequencing
Validation of tumor variants was performed by ultra-deep
sequencing on Amplicon libraries using the 454 Junior Tita-
nium sequencer (Roche) according to the protocol for Ampli-
con amplification, Lib-A (Roche). Amplicons were obtained
by one-step PCR using the FastStart™High Fidelity PCR
System, dNTPack (Roche) and specific adaptors were ligated
for each patient. Amplicon lengths ranged from 200 to
400 bp, including forward and reverse Phusion primers,
intermediate patient-specific sequence MID and the target
template. The initial PCR was performed with 10 ng of
gDNA input according to the manufacturer’s protocol, the
FastStart™High Fidelity PCR System, and the dNTPack
(Roche). After PCR amplification, the Library with Agencourt
AMPure beads (Beckman Coulter) was purified and the
libraries were quantified according to the Quant-iT Pico-
Green dsDNA Assay Kit (Thermo Fisher Scientific). Finally,
the amount of library to be used in the emPCR was deter-
mined according to the Method Manual Lib-A (Roche). Tar-
get regions of the genome reference sequences corresponding
to the amplicons were obtained from the human reference
genome (GRCh37/hg19) using the getfasta command of Bed-
tools.
24
Reads were mapped to these sequences through the
bwa-sw command of Burrows-Wheeler Aligner
25
and vari-
ants were called using GATK –Genome Analysis Tool Kit.
26
Variants were also confirmed using IGV –Integrated
Genome Viewer.
27
Molecular Cancer Biology
Esposito et al.2527
Int. J. Cancer: 143, 2525–2536 (2018) ©2018 UICC
Results
Somatic mutations in HR-NB patients
To clarify the genomic features associated with rapid disease
progression in NB patients, we analyzed an Italian cohort of
29 tumors of HR-NB stage M patients classified in SS and LS
by WES (Table 1). The survival ranged from 6 to 48 months in
SS and from 62 to 159 months in LS. Two SS patients out of
14 died as a result of therapy toxicity. Out of the 15 LS patients,
11 were still in complete remission, two were alive with disease
and two died of disease at last follow-up. Alignment of
2,189,622,787 reads to the reference exome (37.1 million reads
per sample on average) yielded a 97x average coverage and a
76.29% of the target exome with at least 30x coverage, ranging
from 53% to 84% in different patients. Sequence coverage in
the SS and LS groups, considering both the tumor and periph-
eral blood cell samples, was homogeneous (Supporting Infor-
mation Fig. S2). A total of 2,301 and 1,805 high quality and
coverage somatic variants for the SS and LS groups, respec-
tively, were detected after read mapping, variant calling and
identification of somatic variants comparing the tumor and
control data. In accordance with the mutations types observed
before,
7–9
somatic variants resulted in enrichment in C > A
(LS = 32.3%, SS = 25.2%) transversions at TCT sites and in
C > T transitions (LS = 20.3%, SS = 22.6%) at GCG trinucleo-
tide substitution types, normally due to deamination of
5-methylcytosine (Supporting InformationFig. S3). Next, 1,288
high-quality, detrimental and rare somatic variants in 1,043
genes, 580 variants detected in LS patients (Supporting Infor-
mation Table S1A), and 708 in SS (Supporting Information
Table S1B) passed the variant effect- and frequency-based fil-
tering steps (Supporting Information Fig. S1). Variants were
later examined considering recurrence, hit gene and pathways,
and the possible impacts of mutations on disease progression.
Variant effects, group-exclusivity, intra-group recurrence, gene
Table 1. NB patient cohort description.
Patient ID Sex Age at diagnosis (months) MYCN statusDNA index Survival (months) Outcome Group
2,475 M 208 Gain No data 33 DOD SS
1,965 M 83 Not amplified 1.51 34 DOD
1,955 F 77 Not amplified No data 6 DOT
2,368 M 75 Not amplified No data 48 DOD
3,060 F 118 Unknown No data 33 DOD
1,900 M 37 Not amplified 1.14 24 DOD
2,181 M 47 Amplified No data 28 DOD
2,384 M 58 Gain No data 45 DOD
1,995 M 22 Amplified 2.37 23 DOD
1,920 M 14 Not amplified No data 9 DOT
2,100 M 27 Not amplified No data 12 DOD
2,513 M 52 Not amplified No data 20 DOD
2,852 M 50 Gain No data 43 DOD
2,578 F 23 Gain 1.07 42 DOD
1,409 F 34 Not amplified 1.00 159 CR LS
1,641 M 33 Not amplified 1.96 105 CR
2,121 M 61 Not amplified 1.88 144 CR
2,393 F 73 Gain No data 62 CR
2,140 M 55 Not amplified 1.00 75 CR
1,905 M 15 Not amplified 1.96 80 CR
2,528 M 61 Gain No data 86 CR
2,035 M 17 Not amplified 1.52 144 CR
2,488 M 47 Not amplified No data 65 AWD
2,951 M 68 Gain 1.00 64 DOD
2,251 F 12 Amplified No data 110 CR
2,576 F 32 Not amplified No data 71 AWD
2,426 F 7 Not amplified No data 53 CR
2,613 F 12 Not amplified No data 39 CR
2,828 M 8 Not amplified 1.92 71 CR
Abbreviations: M, male; F, Female; Unknown, Physician did not have data; No data, data was not made available; DOD, Dead of disease; DOT, Dead of
toxicity of the treatment; AWD, Alive with disease; CR, Complete remission
Molecular Cancer Biology
2528 Mutated pathways in aggressive neuroblastoma
Int. J. Cancer: 143, 2525–2536 (2018) ©2018 UICC
products biological function and relationships among the
mutated genes were used to prioritize group-specific variants
for validation. We confirmed 50 selected variants in 49 genes
(Supporting Information Table S2).
Mutation landscapes in SS and LS patients
We compared the number, type and effect of somatic muta-
tions observed in the two patient groups. No significant differ-
ence in the numbers of somatic variants per patient in the SS
(median 37) and LS (36) groups were observed (Wilcoxon test
pvalue = 1; Fig. 1a). Very close numbers of selected somatic
variants per Mb were observed in the two groups (median
values of 0.62 and 0.64, respectively). The number of somatic
variants per patient in our cohort was higher than previously
reported,
6–9
but a direct comparison between WES studies is
hampered by several factors, on top the different sequencing
depth or technology and the different analysis methods and
settings used.
Moreover, similar patterns in the SS and LS patients were
present considering the variant type (Supporting Information
Fig. S4A; G test p= 0.11) and predicted variant effect on the
protein sequences (Supporting Information Fig. S4B, G test
p= 0.06) without evident of differences in relation to different
therapy responses and outcomes.
Somatic variants and mutated genes exclusive of SS or LS
The above-cited 708 and 580 high-confidence damaging, and
rare variants detected in SS and LS patients, fell into 583 and
515 individual genes, respectively. Eighteen variants in 18 dif-
ferent genes occurred in both patient groups, resulting in
recurrence in NB-HR patients considered as a whole, whereas
there were 690 and 562 group-specific variants. Only
102 (9.8%) out of 1,043 genes mutated in the whole NB
cohort were recurrently mutated in two or more patients.
Fifty-five genes were recurrently mutated in patients of both
classes (Fig. 1b), including 17 genes detected in more than
two patients, with DPCR1 (mutated in nine patients),
AHNAK2 (6), and CBX4 and ZNF717 (both mutated in four
patients) being the most recurrently observed. Notably,
528 genes were specifically mutated in SS and 460 in LS
patients (Fig. 1b), including 21 and 17 recurrent and group-
specific genes, respectively, and six that carry particularly
damaging variants (Fig. 1c). Of these genes, only KMT2A
(Lysine Methyltransferase 2A; E2926Q in patient ID2426;
S3291C in patient ID1905) and NUPL1 (Nucleoporin 58;
N153 fs in patient ID2393 and ID2576) resulted in recurrently
mutated and group-specific pathways LS patients.
In SS patients, four genes (SMO,SMARCA4,ZNF44 and
CHD2), all known to be expressed in neural tissues, were recur-
rently mutated and group-specific and carried particularly
Figure 1.Comparison of mutation landscapes in SS and LS NB patients. (a) The number of variants in LS and SS groups were not significantly
different, as shown by the boxplot of the distribution of selected somatic variants per patient (p=0.98 of Wilcoxon test of median equality,
conducted after Shapiro–Wilk test of normal distribution pvalue = 3.79 ×10
−5
). (b) Venn chart of number of somatically mutated genes in
LS and SS groups, showing class-specifically mutated genes and their subset of genes being both class-specific and recurrent intra-class. (c)
Mutation matrix indicating in which class and patients are mutated class-specific and recurrent genes, hit by particularly deleterious
mutations.
Molecular Cancer Biology
Esposito et al.2529
Int. J. Cancer: 143, 2525–2536 (2018) ©2018 UICC
deleterious variants not described before, except SMARCA4
R906H. SMO (Frizzled Class Receptor Smoothened) encodes
nonclassical G-protein-coupled receptors that are highly
expressed in neural tissues and involved in Hedgehog signal-
ing. Two SMO variants (Fig. 2; Supporting Information
Fig. S5) were detected, R451G in ID2578 (Supporting Infor-
mation Table S2) in the Frizzled/Smoothened family mem-
brane region, and T640 fs (ID1955) that induces a premature
stop codon ending the protein 132 amino acids before the C-
terminus. The other recurrent SS-specificgenes,CHD2 and
SMARCA4, were transcriptional regulators. Chromodomain
Helicase DNA Binding Protein 2 (CHD2)isimportantfor
neurogenesis and de novo mutations in this gene were found
in neurodevelopmental disorders.
28
CHD2 is a tumor sup-
pressor chromatin remodeler, previously observed to be
mutated and proposed as a cancer driver in chronic lympho-
cytic leukemia.
29
Notably, the transcription co-activator and
tumor suppressor SMARCA4 (SWI/SNF Related, Matrix
Associated, Actin Dependent Regulator of Chromatin,
Subfamily A, Member 4) were recurrently mutated in the SS
group. SMARCA4 encodes a member of the SWI/SNF
nucleosome-remodeling complex whose mutations impact
growth control, differentiation, development and cell adhe-
sion.
30
SMARCA4 somatic variants (R468C in patient ID2513;
R906H in ID3060; Fig. 2; Supporting Information Fig. S5) in
the two SS patients were validated (Supporting Information
Table S2). The deleterious SMARCA4 R906H mutation was
annotated in the COSMIC database (COSM5576007), as pre-
viously being observed in gastric cancer,
31
whereas the puta-
tively damaging R468C variant was not described previously.
Both SMARCA4 variants were localized in the transcription
activator chain of the protein, falling, respectively, in the Heli-
case Sant-associated domain (HSA) and in the helicase ATP-
binding domain of the protein (Fig. 2). Phyre2 structural
analysis, in particular, suggested a possible strong impact of
R906H on ATP binding that is essential for transcriptional
activation. SS-specific FGFR1 N577 L (detected in ID2100)
and PTK2 R569L (detected in ID2181) variants were validated
and described in a previous NB study (Supporting Informa-
tion Fig. S5 and Table S2).
9
These two genes were already
associated with NB tumorigenesis
7,9
although not specifically
in relation to patient survival.
The PTK2/FAK1 (focal adhesion kinase) variant is located
close to the Tyr576 phosphorylation site of the kinase domain
Figure 2.Impact on proteins of somatic variants detected in NB patients in PTK2,PTPRA,SMARCA4and SMO genes. For each gene,
considering the protein encoded by the reference transcript, lollipop plots show the type and the position of somatic variants in relation to
the protein sequence and domains (colored portions) according to Pfam annotation (http://pfam.xfam.org/); different lollipop colors indicate
variant annotation types (See Supporting Information Fig. S5for additional protein plots).
Molecular Cancer Biology
2530 Mutated pathways in aggressive neuroblastoma
Int. J. Cancer: 143, 2525–2536 (2018) ©2018 UICC
“catalytic loop”(Fig. 2) required for PTK2 activation
32
and for
mediating NB progression and aggressiveness.
33
Different pathways and functions are hit by mutations in SS
and LS patients
Beyond the gene level analysis, somatic mutations falling into
different genes that co-participate in the same pathways
and/or are linked together in specific functional or interaction
networks were investigated, considering mutated genes anno-
tated in Reactome (48%) and KEGG (24%). Thirty-Four Reac-
tome and 12 KEGG pathways were specifically enriched in the
LS group, and 12 Reactome and 17 KEGG pathways were
recurrently mutated in LS patients and never mutated in SS
(LS-specific; Supporting Information Tables S3 and S4). In SS
patients, 17 Reactome and one KEGG pathways were specifi-
cally enriched, and 25 Reactome and 4 KEGG pathways were
SS-specific (Supporting Information Tables S3 and S4). Con-
sidering the hierarchical structure of pathway annotation and
redundancy and most deleterious mutations, the somatically
mutated genes in each of the two patient groups tend to par-
ticipate to different pathways and pathway classes (Fig. 3; Sup-
porting Information Figure S6) indicating a link between
disease aggressiveness and specific processes and functions hit
by mutations. Mutations in 15 cell cycle genes were present in
LS patients, with several deleterious mutations in genes linked
to mitosis, including CDC27,CDCA5,CENPC and AURKB.
Notch-related genes are mutated in both groups, but several
genes (TBL1X,CREBBP,NOTCH4,NOTCH3,TNRC6B and
TLE2) specifically belonging to Notch1 signaling are mutated
only in LS patients.
In SS patients, the axon guidance pathway was possibly
hampered by mutations in 20 genes, involved particularly in
NCAM signaling for neurite outgrowth by MAPK2 and MAPK
activation (including ARHGEF11,CACNA1G,FGF4,PTPRA,
PTK2,ANK3,SMO and NTNG2) processes, which is important
for neurodevelopment and oncogenesis. The MAPK pathway
is linked through PTK2 signaling to ERBB4 (including FGF4,
PTPRA and PTK2), and MET (LAMA2,PTK2 and LAMA4).
ERBB4 signaling was specifically enriched, and both MET sig-
naling and the Cilium assembly pathway (BBS10,SMO,
INPP5E) were exclusively mutated in SS patients.
Genes mutated in SS patients cluster into specific pathway-
derived subnetworks
A further analysis of the topological structure of mutation
gene networks derived from the Reactome pathway annota-
tion, encoding direct relations among genes and their prod-
ucts, detected significant associations of somatically mutated
genes in NB SS patients belonging to functionally connected
gene networks specific to the worst survival group, which were
in accordance with previous observations at the pathway level.
Neither the 463 genes mutated in the whole cohort (adjusted
pvalue of global mutation clustering 0.43) nor the 213 genes
mutated in LS patients (adjusted pvalue 0.93) showed
significant clustering according to Hotnet2 analysis. Con-
versely, a more pronounced clustering was observed of the
268 genes mutated in SS patients, 79 of which converged into
18 subnetworks of at least three genes (adjusted pvalue of
global mutation clustering 0.24) (Supporting Information
Table S5). Figure 4 shows the six most relevant network com-
ponents, comprising 31 functionally connected genes that are
somatically mutated specifically in SS patients. The largest
component, which was recurrently identified in almost two
thirds of SS patients (9 of 14), included nine genes (NID2,
LAMA4,LAMA2,PTK2,PTPRA,FGG,VCL,MMP14 and
KSR2) of the RAF/MAPK signaling pathway and extracellular
matrix organization. In addition to the previously observed
PTK2 variant, we validated the D377Y variant, which fell into
the Y phosphatase domain of PTPRA (Fig. 2), closely con-
nected with PTK2 in the RAS/MAPK pathway, and the stop
gaining variant (E352*)ofLAMA2 (Supporting Information
Table S2 and Fig. S5). A second component of six genes
(NALCN,UNC79,SLC9A9,SLC12A1,SLC5A8 and SLC4A9)
that was linked to the transmembrane transport of small mol-
ecules was mutated in four SS patients. Two patients carried
mutations in two genes of the component (NALCN and
SLC9A9 co-mutated in ID2368; SCL5A8 and SLC4A9 in
ID1955). The third component, which was linked to centro-
some maturation, included five genes (CDK5RAP2,CDK11A,
CEP89,TUBGCP6 and SFI1) mutated in three different
patients (ID1965, ID2100, ID2384) (CEP89,TUBGCP6, and
CDK11A co-mutated in the patient ID1965). Four genes were
involved in lipid and lipoprotein (SPTLC2 and ACSL6)or
nucleotide (AK7,AK9 and ACSL6) metabolism, which were
mutated in three SS patients (ID2368, ID2513, ID2852), with
SPTLC2 and ACSL6 in the same patient. Two additional SS-
specific components were defined by SMO, recurrently
mutated in two patients and functionally connected BBS10
and GAS8 genes co-mutated in a third, and by KMT2C,
HOXB3, and HOXC4 mutated in ID1965 and ID2852
patients.
Specific mutated genes and deregulated pathways of SS
patients are confirmed by analysis of a large independent
cohort
To confirm our findings, we analyzed the largest available
group of stage M NB with survival data profiled by WES
(Pugh cohort).
8
The 4,120 genes with nonsilent somatic muta-
tions reported in the Pugh cohort were analyzed, separating
the 240 patients into SS (221; with overall survival ≤5 years)
and LS (19; with overall survival >5 years) according to our
classification.
SS-specific genes, pathways and component identified in
our study were compared to the findings in Pugh cohort
(Supporting Information Fig. S7). Of the genes with SS-
specific recurrence in our cohort, NFATC1 and OR14J1 were
recurrent with SS-specificity also in the Pugh cohort. Further-
more, five genes (CHD2,DIDO1,KRTAP4–8,ZNF44 and
Molecular Cancer Biology
Esposito et al.2531
Int. J. Cancer: 143, 2525–2536 (2018) ©2018 UICC
Figure 3.Summary of Reactome pathways exclusively mutated or exclusively enriched in LS or in SS NB patients. The figure depicts the
hierarchy of Reactome pathways that resulted or contained pathways exclusively mutated in LS (green fill) or in SS (light red fill) patients, or
that were enriched in a class-specific way (bold text); the gray fill indicates more general classes at high hierarchical level being not class-
specific; for the most high-level specific or specifically enriched class of each group, the corresponding mutated genes are indicated in the
right part of the figure (See Supporting Information Tables S3,S4and Figs. S4,S5for additional information).
Molecular Cancer Biology
2532 Mutated pathways in aggressive neuroblastoma
Int. J. Cancer: 143, 2525–2536 (2018) ©2018 UICC
ZNF91) with SS-specific recurrence in our cohort were
mutated with SS-specificity also in Pugh patients (Supporting
Information Fig. S7A). The SMARCA4 mutation was reported
in a patient with a survival of 61 months according to our
classification. Our analysis indicated 34 genes mutated in one
SS patient of our cohort and recurrently mutated with SS-
specificity in the Pugh cohort, including ABCA13 which was
mutated in seven patients.
Nineteen of the 78 genes prioritized because involved in
SS-specific pathways identified (Fig. 3) were mutated with SS-
specificity in the Pugh cohort, including five (ANK3,
COL11A1,COL12A1,COL1A1,PNPLA7) that were also recur-
rent (Supplementary Figure 7A). Five genes (AK7,NALCN,
PTK2,SLC5A8,TUBGCP6) included both in SS-specific path-
ways and in significant subnetworks (Fig. 4) identified in our
study were mutated only in SS patients of the Pugh cohort.
Furthermore, analysis with HotNet2 was performed con-
sidering the 1,810 genes somatically mutated in SS patients of
the Pugh cohort and mapped in the Reactome-derived
network, detecting 14 significant (adjusted pvalue 0.05)
pathway-derived subnetworks involving 143 genes
(Supporting Information Table S6). Extracellular matrix orga-
nization, carbohydrate and lipid metabolism emerged both
from our study and (Figs. 3 and 4) Pugh data. PTK2, which
was shown to be mutated with SS-specificity both in our data
and in the Pugh cases, in the network of mutations detected
in Pugh patients (Supporting Information Table S6), was
directly linked to two gene groups involved in ECM
(Supporting Information Fig. S7B).
Discussion
One of the major challenges for oncologists treating HR-NB is
the high percentage of patients showing rapid disease progres-
sion despite multimodal treatment. Of these, approximately
60% of HR-NBs have a fatal course within 5 years of diagno-
sis. To identify genetic abnormalities associated with disease
aggressiveness, we compared a somatic mutation profile of
HR-NB patients with SS and LS.
Figure 4.Reactome-derived network of genes somatically mutated in NB patients formed by six SS-specific components. (a) Colored nodes in
the net indicate genes mutated in SS patients, with six different components (groups of functionally connected genes somatically mutated in
NB patients with rapid disease progression identified by Hotnet2analysis) in different colors; gray nodes represent genes directly connecting
the components according to pathway topology and non mutated in the analyzed cohort (edges between gray nodes are omitted). (b) Each
component was recurrently mutated in different patients, and specific tumors carried mutations in multiple genes and components of the
network.
Molecular Cancer Biology
Esposito et al.2533
Int. J. Cancer: 143, 2525–2536 (2018) ©2018 UICC
General tumor genomic landscapes of HR-NB patients
with SS and LS were similar, exhibiting close frequencies of
variants and numbers of somatically mutated genes per
patient.
Nevertheless, few genes were recurrently mutated specifi-
cally in the SS group, including SMARCA4,SMO,ZNF44 and
CHD2. SMARCA4 also known as BRG1, is a tumor suppressor
gene of the SWI/SNF complex
34–37
that shows inactivating
mutations or overexpression in several adult cancers.
38,39
We
found two missense mutations, R468C and R906H, in the
Transcription Activator chain region responsible for DNA
and ATP binding and ATP hydrolysis, which are predicted to
be very dangerous. The loss of function of the SMARCA4 pro-
tein likely impaired its activity with damage of the SWI/SNF
that is involved in chromatin remodeling. Jubierre
et al. showed that the SMARCA4 gene has a role in the prolif-
eration of NB cells both in vitro and in vivo.
40
Matsubara
et al. observed a correlation between SMARCA4 mutations
and loss of function in lung cancer cell lines, indicating an
association with aggressive tumor behavior and worse patient
survival.
41
Thus, SMARCA4 mutations may determine the loss
of function associated with tumor aggressiveness and poor NB
patient survival. As it has been demonstrated that SMARCA4
and TERT are functionally linked,
42
SWI/SNF damage could
alter TERT function,
43
which one of the most important genes
rearranged in NB.
44
Among the nonrecurrent gene mutations, we validated the
FGFR1 N546 K variant, as well as novel deleterious variants
of CREBBP and OR5T1 (Supporting Information Fig.S5 and
Table S2), whose mutations were found to be associated with
NB aggressiveness.
7,8
Remarkably, somatic mutations occur-
ring in SS or LS patients hit different pathways. In addition,
functional gene networks, corresponding to sub pathways, hit
only SS patients. Numerous gene variants observed in the
tumor of SS patients affected the RAF/MAP kinase cascade, as
well as MET and ERBB4 pathways linked to PTK2 signaling.
MAP2K and MAPK activation, specific of SS tumors, are of
interest because they can be involved in cell motility by trig-
gering PTK2 signaling and Matrix Metalloproteinases activa-
tion. These results agree with previous data on the enrichment
of somatic mutations in FAK signaling and cell adhesion
signaling.
9
Furthermore, several genes connected to RAF/MAPK sig-
naling were mutated in SS (NID2,LAMA4,LAMA2,PTK2,
PTPRA,FGG,VCL,MMP14 and KSR2), impacting extracellu-
lar matrix organization, regulation of cell adhesion and migra-
tion. A previous observation of PTK2 mutation in a HR-NB
patient by Lasorsa et al., further strengthens the importance of
PTK2 signaling in aggressive tumors. Mutations of NELL1,
UNC79 and COL5A2 genes in one SS patient of our cohort
(Supporting Information Table S1B), were previously reported
in NB patients with SS, albeit with different variants.
9
Particu-
larly relevant groups of clustered genes mutated in SS were
involved in centrosome maturation, in the regulation of the
cell cycle, in ciliary basal body docking (CDK5RAP2,CDK11A,
CEP89,TUBGCP6 and SFI1) and in cilium assembly (the
recurrently and SS specifically mutated SMO, and GAS8,
BBS10 and AK7). Our observations on the mutations linked
to the chromosome remodeling pathway in SS tumors support
the role of chromosome instability in NB,
45
providing further
Table 2. Information on drugs available in relation to genes carrying deleterious mutations in NB patients.
Gene ID Gene Description
Variant
discovered
FDA approved
drugs Drug class References for drugs use
SDHB Succinate
dehydrogenase
complex, subunit B,
iron sulfur (Ip)
L7FS Succinic acid Small
molecule
He et al., 2004, Citric acid cycle intermediates as ligands
for orphan G-protein-coupled receptors., Nature; Southern
et al., 2013, Screening β-arrestin recruitment for the
identification of natural ligands for orphan G-protein-coupled
receptors., J Biomol Screen
SMO Smoothened, frizzled
family receptor
R451G;
T640FS
Vismodegib Small
molecule
inhibitor
Yauch et al., 2009, Smoothened mutation confers resistance
to a Hedgehog pathway inhibitor in medulloblastoma.,
Science; Wang et al., 2012, Identification of a novel
Smoothened antagonist that potently suppresses Hedgehog
signaling., Bioorg. Med. Chem.
Fluocinonide Small
molecule
Wang et al., 2010, Identification of select glucocorticoids as
Smoothened agonists: potential utility for regenerative
medicine., Proc. Natl. Acad. Sci. U.S.A.
Halcinonide Small
molecule
Wang et al., 2011, Glucocorticoid hedgehog agonists in
neurogenesis., Vitam. Horm.; Wojnar et al., 1986,
Androstene-17-thioketals. 1st communication: glucocorticoid
receptor binding, antiproliferative and antiinflammatory
activities of some novel 20-thiasteroids
(androstene-17-thioketals)., Arzneimittelforschung
PTK2 PTK2 protein tyrosine
kinase 2
R569L Masitinib Kit inhibitor Dubreuil et al., 2009, Masitinib (AB1010), a potent and
selective tyrosine kinase inhibitor targeting KIT., PLoS ONE
MMP14 Matrix metallopeptidase
14 (membrane-inserted)
P8FS Prinomastat Mmp inhibitor Abbenante et al., 2005, Protease inhibitors in the clinic.,
Med Chem
Molecular Cancer Biology
2534 Mutated pathways in aggressive neuroblastoma
Int. J. Cancer: 143, 2525–2536 (2018) ©2018 UICC
explanation for the observed CNA in patients with fatal
outcomes.
4,46
The clustering of somatic mutations observed in SS
patients reflected two phenomena: specific functions targeted
in several SS patients, as observed for the RAF/MAPK signal-
ing component, and co-occurrence in the same patient of
mutations in two or more functionally connected genes.
The analysis of a sizeable independent cohort of 240 stage
M NB patients
8
gave additional strength to our findings.
Mutations in Pugh SS patients targeting genes prioritized in
our cohort (ANK3,COL11A1,COL12A1,COL1A1,PNPLA7,
AK7,NALCN,PTK2,SLC5A8 and TUBGCP6) as SS-specific
based on recurrence, pathway enrichment and/or pathway-
derived network topology analysis, were particularly notewor-
thy and supported our results. The reconstruction and analy-
sis of pathway-derived mutation networks reported in Pugh
SS patients further backed the observations done in our cohort
about the deregulation of lipid metabolism and RAF/MAP sig-
naling in relation to ECM mutated genes.
Recent comparison of matched primary and relapsed NB
tumors revealed that disease progression is accompanied by
an increased mutational load in MAPK pathway genes, exhi-
biting new mutations in the MAPK pathway that were not
present at the onset of disease, and accumulated in tumors of
relapsing patients.
47,48
Our findings of specific MAPK signal-
ing pathway damages (also observed by Eleveld et al.
47
and
Schramm et al.
48
) may be relevant for more efficacious thera-
peutic management of patients at diagnosis. Specific genes
mutated at diagnosis exclusively in pathways belonging to the
SS group could be candidates for pharmacological targeting.
SMO,PTK2,MMP14 and SDHB are quite interesting as they
are targeted by FDA approved drugs according to the Drug
Gene Interaction Database (DGID: http://dgidb.genome.wustl.
edu/) (Table 2). Recently, Padovan-Merhar et al.
49
reported an
increased SMO mutation frequency in tumors of HR-NB
patients at relapse, showing that most of these new mutations
are targetable and give an additional tool to treat relapsing
patients. Functional investigation is mandatory to assess the
potential significance of mutated genes as therapeutic targets,
and further study is needed to evaluate drugs, such as Masiti-
nib and Vismodegib, for NB therapy.
In our study, two groups of HR-NB patients with different
outcome were characterized, providing new data on mutations
recurrently affecting specific pathways and functions in
patients with SS, informing the molecular features, beyond
well-defined CNA patterns, that are associated with high
tumor aggressiveness.
Author Contributions
GPT, MRE have designed, planned and performed the study.
MP realized DNA extraction from tumor biopsy, exome
library and sequencing on Ion proton sequencer. AB, AC and
SB performed bioinformatics analyses and contributed new
systems biology methods. KM and LL have contributed for
collection of matched tumor biopsies and peripheral blood
samples, DNA extraction from peripheral blood. MC, AVL
and RL revised the manuscript. MRE, AB, SB and GPT wrote
the manuscript that has been revised and approved by all
Authors.
Acknowledgements
The authors would like to thank Drs. Alberto Garaventa, Angela Rita
Sementa, Riccardo Haupt for providing patient data. The authors would
like also to thank Dr. Francesca Dal Pero of Roche Company for support
on 454 sequencing, Dr. Silvia Bresolin for technical support on Roche
454-GS Junior DNA sequencing platform and Prof. Giuseppe Basso for
fruitful result discussion. The present work was mainly supported by Fon-
dazione Italiana per la Lotta al Neuroblastoma. The authors thank for
financial support to SB Fondazione Cassa di Risparmio di Padova e
Rovigo Progetti di Eccellenza 2011/2012 by Ministero dell’Istruzione, del-
l’Università e della Ricerca PRIN 2010/11 (2010NYKNS7_002) and Uni-
versity of Padova. AB is recipient of a PhD fellowship from the University
of Padova (PhD in Biosciences)
References
1. Shimada H, Ambros IM, Dehner LP, et al. The
International Neuroblastoma Pathology Classifica-
tion (the Shimada system). Cancer 1999;86:364–72.
2. Luksch R, Castellani MR, Collini P,
et al. Neuroblastoma (peripheral neuroblastic
tumours). Crit Rev Oncol Hematol 2016;107:
163–81.
3. Coco S, Theissen J, ScaruffiP, et al. Age-
dependent accumulation of genomic aberrations
and deregulation of cell cycle and telomerase
genes in metastatic neuroblastoma. Int J Cancer
2012;131:1591–600.
4. Stigliani S, Coco S, Moretti S, et al. High genomic
instability predicts survival in metastatic high-risk
neuroblastoma. Neoplasia 2012;14:823–32.
5. Schleiermacher G, Mosseri V, London WB,
et al. Segmental chromosomal alterations have
prognostic impact in neuroblastoma: a report
from the INRG project. Br J Cancer 2012;107:
1418–22.
6. Molenaar JJ, Koster J, Zwijnenburg DA,
et al. Sequencing of neuroblastoma identifies
chromothripsis and defects in neuritogenesis
genes. Nature 2012;483:589–93.
7. Sausen M, Leary RJ, Jones S, et al. Integrated
genomic analyses identify ARID1A and ARID1B
alterations in the childhood cancer neuroblas-
toma. Nat Genet 2013;45:12–7.
8. Pugh TJ, Morozova O, Attiyeh EF, et al. The
genetic landscape of high-risk neuroblastoma. Nat
Genet 2013;45:279–84.
9. Lasorsa VA, Formicola D, Pignataro P,
et al. Exome and deep sequencing of clinically
aggressive neuroblastoma reveal somatic muta-
tions that affect key pathways involved in cancer
progression. Oncotarget 2016;7:21840–52.
10. Calabrese FM, Clima R, Pignataro P, et al. A com-
prehensive characterization of rare mitochondrial
DNA variants in neuroblastoma. Oncotarget 2016;
7:49246–58.
11. Capasso M, Diskin SJ. Genetics and genomics of
neuroblastoma. Cancer Treat Res 2010;155:
65–84.
12. Cingolani P, Patel VM, Coon M, et al. Using dro-
sophila melanogaster as a model for genotoxic
chemical mutational studies with a new program,
SnpSift. Front Genet 2012;3:35.
13. Cingolani P, Platts A, Wang le L, et al. A program
for annotating and predicting the effects of single
nucleotide polymorphisms, SnpEff: SNPs in the
genome of Drosophila melanogaster strain w1118;
iso-2; iso-3. Fly (Austin) 2012;6:80–92.
14. Forbes SA, Beare D, Gunasekaran P,
et al. COSMIC: exploring the world’s knowledge
of somatic mutations in human cancer. Nucleic
Acids Res 2015;43:D805–11.
15. Landrum MJ, Lee JM, Benson M, et al. ClinVar:
public archive of interpretations of clinically
relevant variants. Nucleic Acids Res 2016;44:
D862–8.
Molecular Cancer Biology
Esposito et al.2535
Int. J. Cancer: 143, 2525–2536 (2018) ©2018 UICC
16. Dong C, Wei P, Jian X, et al. Comparison and inte-
gration of deleteriousness prediction methods for
nonsynonymous SNVs in whole exome sequencing
studies. Hum Mol Genet 2015;24:2125–37.
17. Liu X, Jian X, Boerwinkle E. dbNSFP v2.0: a data-
base of human non-synonymous SNVs and their
functional predictions and annotations. Hum
Mutat 2013;34:E2393–402.
18. Lek M, Karczewski KJ, Minikel EV, et al. Analysis
of protein-coding genetic variation in 60,706
humans. Nature 2016;536:285–91.
19. Kelley LA, Mezulis S, Yates CM, et al. The Phyre2
web portal for protein modeling, prediction and
analysis. Nat Protoc 2015;10:845–58.
20. Kanehisa M, Goto S. KEGG: kyoto encyclopedia
of genes and genomes. Nucleic Acids Res 2000;28:
27–30.
21. Fabregat A, Sidiropoulos K, Garapati P, et al. The
Reactome pathway Knowledgebase. Nucleic Acids
Res 2016;44:D481–7.
22. Sales G, Calura E, Cavalieri D, et al. graphite - a Bio-
conductor package to convert pathway topology to
gene network. BMC Bioinformatics 2012;13:20.
23. Leiserson MD, Vandin F, Wu HT, et al. Pan-
cancer network analysis identifies combinations of
rare somatic mutations across pathways and pro-
tein complexes. Nat Genet 2015;47:106–4.
24. Quinlan AR, Hall IM. BEDTools: a flexible suite
of utilities for comparing genomic features. Bioin-
formatics 2010;26:841–2.
25. Li H, Durbin R. Fast and accurate long-read align-
ment with Burrows-Wheeler transform. Bioinfor-
matics 2010;26:589–95.
26. McKenna A, Hanna M, Banks E, et al. The
Genome Analysis Toolkit: a MapReduce frame-
work for analyzing next-generation DNA
sequencing data. Genome Res 2010;20:1297–303.
27. Robinson JT, Thorvaldsdottir H, Winckler W,
et al. Integrative genomics viewer. Nat Biotechnol
2011;29:24–6.
28. Carvill GL, Heavin SB, Yendle SC, et al. Targeted
resequencing in epileptic encephalopathies iden-
tifies de novo mutations in CHD2 and SYNGAP1.
Nat Genet 2013;45:825–30.
29. Rodriguez D, Bretones G, Quesada V,
et al. Mutations in CHD2 cause defective associa-
tion with active chromatin in chronic lymphocytic
leukemia. Blood 2015;126:195–202.
30. Reisman D, Glaros S, Thompson EA. The SWI/-
SNF complex and cancer. Oncogene 2009;28:
1653–68.
31. Kuboki Y, Yamashita S, Niwa T,
et al. Comprehensive analyses using next-
generation sequencing and immunohistochemistry
enable precise treatment in advanced gastric can-
cer. Ann Oncol 2016;27:127–33.
32. Owen JD, Ruest PJ, Fry DW, et al. Induced focal
adhesion kinase (FAK) expression in FAK-null
cells enhances cell spreading and migration
requiring both auto- and activation loop phos-
phorylation sites and inhibits adhesion-dependent
tyrosine phosphorylation of Pyk2. Mol Cell Biol
1999;19:4806–18.
33. Lee S, Qiao J, Paul P, et al. FAK is a critical regu-
lator of neuroblastoma liver metastasis. Oncotar-
get 2012;3(12):1576–87.
34. Decristofaro MF, Betz BL, Rorie CJ, et al. Charac-
terization of SWI/SNF protein expression in
human breast cancer cell lines and other malig-
nancies. J Cell Physiol 2001;186:136–45.
35. Schneppenheim R, Fruhwald MC, Gesk S,
et al. Germline nonsense mutation and somatic
inactivation of SMARCA4/BRG1 in a family with
rhabdoid tumor predisposition syndrome.
Am J Hum Genet 2010;86:279–84.
36. Wilson BG, Roberts CW. SWI/SNF nucleosome
remodellers and cancer. Nat Rev Cancer 2011;11:
481–92.
37. Robinson G, Parker M, Kranenburg TA,
et al. Novel mutations target distinct subgroups of
medulloblastoma. Nature 2012;488:43–8.
38. Sentani K, Oue N, Kondo H, et al. Increased
expression but not genetic alteration of BRG1, a
component of the SWI/SNF complex, is associated
with the advanced stage of human gastric carcino-
mas. Pathobiology 2001;69:315–20.
39. Bai J, Mei PJ, Liu H, et al. BRG1 expression is
increased in human glioma and controls glioma
cell proliferation, migration and invasion in vitro.
J Cancer Res Clin Oncol 2012;138:991–8.
40. Jubierre L, Soriano A, Planells-F errer L,
et al. BRG1/SMARCA4 is essential for neuroblas-
toma cell viability through modulation of cell
death and survival pathways. Oncogene 2016;35:
5179–90.
41. Matsubara D, Kishaba Y, Ishikawa S, et al. Lung
cancer with loss of BRG1/BRM, shows epithelial
mesenchymal transition phenotype and distinct
histologic and genetic features. Cancer Sci 2013;
104:266–73.
42. Wu S, Ge Y, Huang L, et al. BRG1, the ATPase
subunit of SWI/SNF chromatin remodeling com-
plex, interacts with HDAC2 to modulate telome-
rase expression in human cancer cells. Cell Cycle
2014;13:2869–78.
43. Maida Y, Yasukawa M, Okamoto N,
et al. Involvement of telomerase reverse transcrip-
tase in heterochromatin maintenance. Mol Cell
Biol 2014;34:1576–93.
44. Peifer M, Hertwig F, Roels F, et al. Telomerase
activation by genomic rearrangements in high-risk
neuroblastoma. Nature 2015;526:700–4.
45. Tonini GP. Growth, progression and chromosome
instability of Neuroblastoma: a new scenario of
tumorigenesis? BMC Cancer 2017;17:20.
46. Defferrari R, Mazzocco K, Ambros IM,
et al. Influence of segmental chromosome abnor-
malities on survival in children over the age of
12 months with unresectable localised peripheral
neuroblastic tumours without MYCN amplifica-
tion. Br J Cancer 2015;112:290–5.
47. Eleveld TF, Oldridge DA, Bernard V,
et al. Relapsed neuroblastomas show frequent
RAS-MAPK pathway mutations. Nat Genet 2015;
47:864–71.
48. Schramm A, Koster J, Assenov Y, et al. Muta-
tional dynamics between primary and relapse
neuroblastomas. Nat Genet 2015;47:872–7.
49. Padovan-Merhar OM, Raman P, Ostrovnaya I,
et al. Enrichment of Targetable Mutations in the
Relapsed Neuroblastoma Genome. PLoS Genet
2016;12:e1006501.
Molecular Cancer Biology
2536 Mutated pathways in aggressive neuroblastoma
Int. J. Cancer: 143, 2525–2536 (2018) ©2018 UICC