HAL Id: hal-01580787
https://hal.inria.fr/hal-01580787
Submitted on 26 Feb 2018
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-
entic research documents, whether they are pub-
lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diusion de documents
scientiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Germline CDKN2A/P16INK4A mutations contribute to
genetic determinism of sarcoma
Fanélie Jouenne, Isaure Chauvot de Beauchêne, Emeline Bollaert,
Marie-Francoise Avril, Olivier Caron, Olivier Ingster, Axel Lecesne, Patrick
Benuisiglio, Philippe Terrier, Vincent Caumette, et al.
To cite this version:
Fanélie Jouenne, Isaure Chauvot de Beauchêne, Emeline Bollaert, Marie-Francoise Avril, Olivier
Caron, et al.. Germline CDKN2A/P16INK4A mutations contribute to genetic determinism
of sarcoma. Journal of Medical Genetics, BMJ Publishing Group, 2017, 54 (9), pp.607-612.
�10.1136/jmedgenet-2016-104402�. �hal-01580787�
1
Jouenne F, et al. J Med Genet 2017;0:1–6. doi:10.1136/jmedgenet-2016-104402
SHORT REPORT
Germline CDKN2A/P16INK4A mutations contribute to
genetic determinism ofsarcoma
Fanélie Jouenne,
1,2
Isaure Chauvot de Beauchene,
3
Emeline Bollaert,
4
Marie-Françoise Avril,
5,6
Olivier Caron,
7
Olivier Ingster,
8
Axel Lecesne,
7
Patrick Benusiglio,
7
Philippe Terrier,
1
Vincent Caumette,
1
Daniel Pissaloux,
9
Arnaud de la Fouchardière,
9
Odile Cabaret,
1
Birama N'Diaye,
1
Amélie Velghe,
4
Gaelle Bougeard,
10,11
Graham J Mann,
12
Serge Koscielny,
13,14
Jennifer H Barrett,
15
Mark Harland,
15
Julia Newton- Bishop,
15
Nelleke Gruis,
16
Remco Van Doorn,
16
Marion Gauthier-Villars,
17
Gaelle Pierron,
17
Dominique Stoppa-Lyonnet,
17
Isabelle Coupier,
18,19
Rosine Guimbaud,
20
Capucine Delnatte,
21
Jean-Yves Scoazec,
1
Alexander M Eggermont,
22
Jean Feunteun,
23
Luba Tchertanov,
24
Jean-Baptiste Demoulin,
4
Thierry Frebourg,
10,11
Brigitte Bressac-de Paillerets
1,2
Cancer genetics
To cite: Jouenne F,
Chauvot de Beauchene I,
Bollaert E, et al. J Med
Genet Published Online
First: [please include Day
Month Year]. doi:10.1136/
jmedgenet-2016-104402
► Additional material is
published online only. To view
please visit the journal online
(http:// dx. doi. org/ 10. 1136/
jmedgenet- 2016- 104402)
For numbered affiliations see
end of article.
Correspondence to
Dr Brigitte Bressac-de Paillerets,
Département de Biopathologie,
Gustave Roussy, 114 rue
Edouard Vaillant, 94805 Villejuif,
France; brigitte. bressac@
gustaveroussy. fr
Received 25 October 2016
Revised 13 March 2017
Accepted 14 March 2017
ABSTRACT
Background Sarcomas are rare mesenchymal
malignancies whose pathogenesis is poorly understood;
both environmental and genetic risk factors could
contribute to their aetiology.
Methods and results We performed whole-exome
sequencing (WES) in a familial aggregation of three
individuals affected with soft-tissue sarcoma (STS)
without TP53 mutation (Li-Fraumeni-like, LFL) and found
a shared pathogenic mutation in CDKN2A tumour
suppressor gene. We searched for individuals with
sarcoma among 474 melanoma-prone families with a
CDKN2A-/+ genotype and for CDKN2A mutations in 190
TP53-negative LFL families where the index case was a
sarcoma. Including the initial family, eight independent
sarcoma cases carried a germline mutation in the
CDKN2A/p16
INK4A
gene. In five out of seven formalin-
fixed paraffin-embedded sarcomas, heterozygosity was
lost at germline CDKN2A mutations sites demonstrating
complete loss of function. As sarcomas are rare in
CDKN2A/p16
INK4A
carriers, we searched in constitutional
WES of nine carriers for potential modifying rare variants
and identified three in platelet-derived growth factor
receptor (PDGFRA) gene. Molecular modelling showed
that two never-described variants could impact the
PDGFRA extracellular domain structure.
Conclusion Germline mutations in CDKN2A/P16
INK4A
,
a gene known to predispose to hereditary melanoma,
pancreatic cancer and tobacco-related cancers, account
also for a subset of hereditary sarcoma. In addition, we
identified PDGFRA as a candidate modifier gene.
Sarcomas are a complex group of rare malignant
tumours derived from cells that originate from the
mesenchyma. These tumours, which can affect both
bone and soft tissue, include more than 50 different
subtypes. The annual incidence of soft-tissue
sarcomas (STS) is around five new cases per 100
000 population, whereas it is 0.8 for bone sarcomas
in Caucasians.
1
They account for nearly 20% of all
paediatric solid malignant cancers, but less than 1%
of all adult solid malignant cancers.
2
The pathogen-
esis of most sarcomas is still poorly understood, and
both environmental and genetic risk factors could
contribute to their aetiology. The main environ-
mental factors are carcinogens, viruses and ionising
radiation, particularly radiation therapy received
for a first cancer.
3
The risk of sarcoma is enhanced
in several hereditary cancer syndromes, including
Li-Fraumeni syndrome (LFS), a rare, dominant
Mendelian cancer syndrome linked to TP53 muta-
tions and possibly to POT1 mutations.
4 5
Beyond
these syndromes, there may be other complex
heritable predispositions and others not yet iden-
tified.
6
The potential for intrafamily exome-sequencing
approach to identify additional cancer suscepti-
bility genes has been demonstrated. Therefore,
we conducted germline whole-exome sequencing
(WES) in two affected members of a three sarco-
ma-cases family (Patients I-2 and II-1, Family 7389,
table 1, figure 1A). We performed data mining
applying classical filtering strategies using Inge-
nuity Variant Analysis (IVA) software (Qiagen).
7
With very stringent frequency filtering (MAF)
<0.001%, using a Biological Context of sarcoma,
three germline variants shared by both sarcoma-af-
fected relatives (uncle and nephew) were identified
in CDKN2A, PDGFRA and SKA3 genes (figure 1B).
Because of the loss of function mutation detected in
CDKN2A and the well-known role of CDKN2A in
somatic sarcomagenesis, both in humans and mice,
we focused first on this gene.
8
CDKN2A is a known
tumour suppressor gene and the first familial
melanoma gene identified; it encodes two distinct
proteins, p16
INK4A
and p14
ARF
, which both function
in cell cycle regulation.
9
We confirmed the germ-
line splice mutation (c.151-2A>G) with Sanger
sequencing and also in DNA extracted from forma-
lin-fixed, paraffin-embedded (FFPE) tumorous
tissue from the third case, deceased patient I-1
JMG Online First, published on June 7, 2017 as 10.1136/jmedgenet-2016-104402
Copyright Article author (or their employer) 2017. Produced by BMJ Publishing Group Ltd under licence.
group.bmj.com on June 8, 2017 - Published by http://jmg.bmj.com/Downloaded from
2
Jouenne F, et al. J Med Genet 2017;0:1–6. doi:10.1136/jmedgenet-2016-104402
Cancer genetics
Table 1 CDKN2A/P16INK4A germline mutations identified in eight families with members affected by sarcoma and candidate modifiers
Collections* Family ID Patient ID
Cases Clinical
context P16 exon p16
INK4A
mutation
c†
p14
ARF
AA change
Sarcoma LOH at
CDKN2A/P16
‡
Melanoma/ pancreatic
cancer reports
P16INK4A Loss of
function
PDGFRA
variants
Other candidate
modifiers
5 6
A 7389 I-1
I-2
II-1
LPS (57)
AGS (67)
STS (33)
IVS 1
IVS 1
IVS 1
c.151-2A>G
c.151-2A>G
c.151-2A>G
c.194-2A>G
c.194-2A>G
c.194-2A>G
No
Yes
Yes
Yes
(unpublished data)
Splice p.Leu112Arg
p.Leu112Arg
p.Leu112Arg
None
ATM c.8584+1G>A
ND
A 18998 II-1
II-2
LPS (32)
OS (24)
1α
c.146T>G, p.Ile49Ser
ND
NA
ND
ND
ND
Yes
(Holland E, 1999,
22
Begg CB, 2005
23
)
Yes
(Lal G, 2000
24
)
None
ND
None
ND
B 14288 II-1 OS (16) 2 c.194T>C, p.Leu65Pro p.Ala79Ala Yes Yes
(Landi MT, 2004
25
)
Yes
(Landi MT, 2004)
25
None None
B 2225 II-1 UCS (43) 2 c.225_243del, p.Ala76fs p.Arg90fs Yes Yes
(Gruis N, 1995
26
)
ND None None
B 14289 I-1
FS (69)
2 c.301G>T, p.Gly101Trp
p.Arg115Leu
Yes
Yes
(Kannengiesser C, 2009,
27
Miller PJ, 2011
28
)
Yes
(Ranade K., 1995
29
,
Kannengiesser C,
2009
27
)
None
None
III-2 OS (12) ND ND ND ND ND
B 14291 II-1 OS (13)+ MM
(26,27)
2 c.301G>T, p.Gly101Trp p.Arg115Leu ND idem idem p.Asn76Ser POT1 p.Gln376Arg
B 15118 II-1 SVS (57)+ 2MM
(39,58)
2 c.301G>T, p.Gly101Trp p.Arg115Leu No idem idem None None
C 20473 I-4
I-2
OS (7)+8 MM
RMS (25) + MM
(36)
1α
c.95T>C, p.Leu32Pro
ND
NA
ND
ND
ND
Yes
(Walker GJ., 1995,
28
Goldstein A, 2007
30
)
Yes
(McKenzie HA, 2010
31
)
p.Thr463Ser
ND
None
ND
*All subjects provided written informed consent for participation in these oncogenetic research studies, which were approved by the local research ethics committees. Collection (A) comprised 190 families with suspected Li-Fraumeni syndrome
that included at least one member with sarcoma. Collection (B) comprised 296 melanoma-prone French families CDKN2A/p16INK4A+. Collection (C) comprised 178 CDKN2A/p16INK4A+ melanoma-prone families from the international
GenoMEL database.
†CDKN2A/P16
INK4A
exons 1alpha, 2 and 3 were sequenced by Sanger method based on transcript NM_000077.4 to screen for mutations in collections A, B and C. None were present in public controls databases such as 1000 genomes or ExAC.
‡LOH analyses were performed by Sanger sequencing at germline CDKN2A mutations sites in DNA extracted from sarcoma FFPE samples.
Ages (years) at diagnosis appears in parentheses
AGS, angiosarcoma; FS, fibrosarcoma; LPS, liposarcoma; MM, cutaneous melanoma; ; NA, not analysed; ND, not determined; OS, osteosarcoma; STS, soft-tissue sarcoma; SVS, synovialosarcoma; UCS, uterine carcinosarcoma; RMS,
rhabdomyosarcoma.
group.bmj.com on June 8, 2017 - Published by http://jmg.bmj.com/Downloaded from
3
Jouenne F, et al. J Med Genet 2017;0:1–6. doi:10.1136/jmedgenet-2016-104402
Cancer genetics
Figure 1 Whole-exome sequencing in a three sarcoma cases family without TP53 germline mutation: identification of CDKN2A and PDGFRA germline
mutations, co-segregating with sarcomas (A) Pedigree of the Li-Fraumeni-like family. Cancer diagnosis and age at onset are indicated for affected members;
hatched circles/squares indicate sarcoma: AGS, angiosarcoma; LPS, liposarcoma; STS, soft-tissue sarcoma. Genotypes of CDKN2A and PDGFRA for all
samples available for testing are shown. Patients with WES data are indicated with a black star. (B) Whole-exome sequencing (WES) germline SNV filtering
and interpretation, for two patients of Family 7389. We used Ingenuity Variant Analysis software (v.2.1.20130711, IVA, Qiagen) and predetermined filters
(see Bioinformatics analysis, online supplementary material). Starting with 307 690 variants spanning 17 673 genes, successive filters lead to 3 variants
spanning 3 genes (CDKN2A, PDGFRA and SKA3). (C) Structural properties of platelet-derived growth factor receptor α (PDGFRα) wild-type and variants.
(Top row) the PDGFRα protein has a modular structure composed of five Ig-like domains (D1, D2, D3, D4 and D5), a trans-membrane domain (TMD) and
a cytoplasmic region. The cytoplasmic region consists of a regulatory juxtamembrane region (JMR) and a catalytic kinase domain, with an N-lobe and a
C-lobe, which harbours a kinase insert domain (KID). (Bottom row) The X-ray analysis structures are represented as ribbon diagrams, based on the KIT
oncogene structural data. D1 and D5 are denoted as ovals. (Middle row) Schematic representations of D1 and D5 topologies. (Bottom figures) Superimposed
conformations of wild-type PDGFRα (blue) and PDGFRα germline variants (pink), obtained from molecular dynamics (MD) simulations. Representative
conformations were selected by RMSDs clustering and are presented as ribbon diagrams.
group.bmj.com on June 8, 2017 - Published by http://jmg.bmj.com/Downloaded from
4
Jouenne F, et al. J Med Genet 2017;0:1–6. doi:10.1136/jmedgenet-2016-104402
Cancer genetics
(online supplementary figure S2A). We had previously identified
this specific mutation in three independent, melanoma-prone
families. Transcript analysis was performed for a proband,
indicating that CDKN2A exon two had been skipped in both
p16
INK4A
and p14
ARF
transcripts, creating putative frameshifts
(online supplementary figure S3).
Next, we performed Sanger sequencing of CDKN2A for germ-
line mutations in full collection A (190 unrelated families with
suspected LFS or Li-Fraumeni-like, LFL, whose index case was a
sarcoma without detectable TP53 germline mutation). We iden-
tified a second carrier of a CDKN2A/P16
INK4A
germline mutation
(p.Ile49Ser), a patient diagnosed with a pleomorphic liposar-
coma at the age of 32 years whose brother died of osteosarcoma
at the age of 24 years (Family 18 998, table 1, online supplemen-
tary S1A and S2B).
To explore further the potential connection between CDKN2A
germline mutations and sarcoma, we reviewed the phenotypes
in our collection of 296 melanoma-prone French families with
CDKN2A/P16
INK4A
germline mutations (collection B; mutations
were partially described previously
10
and found eight kindreds that
contained at least one member with sarcoma. Among them, five
probands with sarcoma carried the pathogenic familial CDKN2A/
P16
INK4A
germline mutation (table 1, families 14 288, 2 225, 1
4289, 14 291 and 15 118, online supplementary figure S1B) and
three families had incomplete data (two untested index cases and
one unconfirmed STS; see online supplementary figure S4A,S4C;
material and methods). Overall, among the 296 families, the differ-
ence in sarcoma incidence between CDKN2A mutation carriers
(5/593; 0.84%; 95% CI 0.3% to 2%) and non-carriers (1/298;
0.34%; 95% CI 0.02% to 2.16%) did not reach statistical signif-
icance (p=0.67; Fisher’s exact test). Considering the yearly inci-
dence in Caucasians of 5.8 per 1 00 000
1
and the mean follow-up
duration in collection B of 46 years, the probability of observing
at least five sarcomas in the 593 CDKN2A carriers population was
0.02 (assuming a binomial distribution). In the 298 CDKN2A WT
populations, the mean follow-up was 39 years, and the probability
of observing at least 1 sarcoma was 49%.
Next, we searched for biological arguments. As loss of hetero-
zygosity (LOH) is considered in tumour’s biology as a strong indi-
cator to the causative role of a tumour suppressor, we performed
Sanger sequencing in seven FFPE sarcoma blocs available from
French patients. We identified LOH at the CDKN2A germline
mutation site in five out of seven samples (table 1; online supple-
mentary figure S2). These LOHs demonstrate the occurrence of a
second genetic hit on CDKN2A and, therefore, complete loss of
p16
INK4A
function in five sarcomas, in accordance with the driver
role of CDKN2A tumour suppressor gene in sarcomagenesis.
8
Finally, we interrogated the GenoMEL database containing
178 CDKN2A+ melanoma-prone families (collection C), after
removal of 60 French families already included in collection B.
We identified three additional independent CDKN2A mutation
carriers affected with a sarcoma. One family was from Australia
and carried a CDKN2A/p16
INK4A
p.Leu32Pro germline muta-
tion (Family 20 473, table 1; online supplementary figure S1C).
The second family was from the UK (21 kb deletion targeting
CDKN2A/p14
ARF
exon 1b) (data not shown), and the third
family was from the Netherlands, but the initial diagnosis of
fibrosarcoma case was revised to melanoma and, therefore, was
excluded.
Overall, in collections A, B and C, ascertained for Li-Fraumeni
(A) or multiple cases of melanoma and/or pancreatic cancer (B
and C), we identified eight independent families in which at least
a CDKN2A/P16
INK4A
mutation carrier had a sarcoma (table 1).
Therefore, based on probabilistic and biological arguments,
CDKN2A/P16
INK4A
germline mutations can be strongly suspected
to increase sarcoma risk. Interestingly, in the literature, two
sarcoma cases in CDKN2A/P16
INK4A
mutation carriers were
identified in families with melanoma/pancreatic cancer and
very recently germline CDKN2A mutations were identified in
two independent patients presenting with LFS.
11 12
In addition
to the well-known role of CDKN2A in somatic sarcomagenesis,
other observations in animals suggested a germline effect.
8
First,
in a mouse model, deletion of the Cdkn2a locus could substi-
tute for mutations in Trp53 to generate STSs.
13
Second, in a
naturally occurring, canine breed-specific histiocytic sarcoma,
a genome-wide association study (GWAS) identified a haplo-
type near CDKN2A.
14
In conclusion to our work and published
data, germline mutations in CDKN2A/P16
INK4A
, a gene known
to predispose to hereditary melanoma, pancreatic cancer and
tobacco-related cancers, account also for a subset of hereditary
sarcoma.
9
As melanoma risk in CDKN2A mutation carriers is clearly asso-
ciated with MC1R frequent alleles acting as modifiers,
15
we formu-
lated the hypothesis that the very low frequency of sarcoma cases
observed in CDKN2A/P16
INK4A-
positive melanoma-prone families
could be explained by rare modifiers alleles. In a model of oligo-
genic inheritance, it is challenging to identify rare germline vari-
ants that act in synergy to initiate cancer, and GWAS is unable to
identify rare disease-predisposing variants.
16
Candidate pathogenic
variants for sarcoma risk in ATM, ATR, BRCA2 and ERCC2 genes
were identified recently in a large sarcoma case–control study as
well as POT1 variants in cardiac angiosarcoma, but other genes
not yet identified could also play a role.
5 6
To explore this hypoth-
esis, we considered the two additional germline variants identified
in PDGFRA and SKA3 genes in the WES data of patients I-2 and
II-1, both sarcoma-affected (initial Family 7389). In SKA3 gene, an
insertion of 2T was supposed to have occurred in a stretch of 12T
but was unconfirmed by Sanger sequencing (online supplementary
figure S5). The platelet-derived growth factor receptor alpha gene
(PDGFRA) harboured a germline missense mutation, c.335T>G,
p.Leu112Arg, located in the extracellular receptor domain and
predicted deleterious by two computational methods (GVGD and
SIFT). This mutation was verified by Sanger sequencing and was
also found in DNA extracted from FFPE-sarcoma tissue from the
third family member, patient I-1, therefore being present in the
three sarcoma-affected patients (Family 7389, figure 1A; online
supplementary S7A).
Next, we performed additional WES analyses in blood-ex-
tracted DNA from seven probands affected with sarcoma
that carried germline CDKN2A mutations (family 18 998-II.1
in collection A; families 14 288-II.1, 2 225-II.1, 14 289-I.1,
14 291-II.1 and 15 118-II.1 in collection B and family 20 473-I.4
in collection C). Subsequently, we data mined the WES avail-
able for a total of nine CDKN2A/P16
INK4A
carriers affected with
sarcoma, including two relatives. We applied classical filtering
strategies using the IVA software (Qiagen) (online supplemen-
tary figure S6).
7
For variant frequency, we defined rare variants
as those with a minor allele frequency (MAF) <0.5%.
16
The
outcome of our filtering strategy was the selection of 82 vari-
ants spanning 76 genes. Among previously published sarcoma
susceptibility genes, we found no mutations in TP53, ATR,
BRCA2 and ERCC2.
6
We found a c.8584+1G>A putative
splice site mutation in ATM gene in patient 7389-I.2, but this
variant was absent in the sarcoma-affected relative, II.1. We also
found, in patient 14 291-II.1, a POT1 c.1127A>G, p.Gln376Arg
missense variant, present at a frequency of 0.07% in Eur-Am
ESP and predicted deleterious by four prediction methods
(SIFT, MutationTaster, Polyphen 2 and Condel). This variant
group.bmj.com on June 8, 2017 - Published by http://jmg.bmj.com/Downloaded from