Mutational signature analysis identifies MUTYH deficiency in colorectal cancers and adrenocortical carcinomas
Camilla Pilati1*, Jayendra Shinde2,3,4,5*, Ludmil B. Alexandrov6*, Guillaume Assié7, Thierry André8, Zofia Hélias-Rodzewicz9, Romain Doucoudray9, Delphine Le Corre1, Jessica Zucman- Rossi2,3,4,5, Jean-François Emile9, Jérôme Bertherat79, Eric Letouzé2,3,4,55, Pierre Laurent- Puig1$@
Accepted Article
1 INSERM UMR-S1147, Personalized Medicine, Pharmacogenomics, Therapeutic Optimization, Université Paris Descartes, Paris, France.
2 INSERM, Unité Mixte de Recherche (UMR) 1162, Génomique Fonctionnelle des Tumeurs Solides, Equipe Labellisée Ligue contre le Cancer, Paris, France.
Université Paris Descartes, Labex Immuno-Oncology, Sorbonne Paris Cité, Paris, France.
4 Université Paris 13, Sorbonne Paris Cité, Unité de Formation et de Recherche (UFR) Santé, Médecine, Biologie Humaine (SMBH), Bobigny, France.
Université Paris Diderot, Institut Universitaire d’Hématologie, Paris, France.
6 [1] Theoretical Biology and Biophysics (T-6), Los Alamos National Laboratory, Los Alamos, New Mexico 87545,
USA. [2] Center for Nonlinear Studies, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, USA.
7 [1] INSERM U1016, CNRS UMR 8104, Paris Descartes University, Institut Cochin, Paris, France. [2] Center for Rare Adrenal Diseases, Department of Endocrinology, Assistance Publique-Hôpitaux de Paris, Hôpital Cochin, Paris, France.
8 [1] Department of Medical Oncology AP-HP, Hospital Saint-Antoine, Paris, France, [2] Université Pierre et Marie Curie (UMPC) Paris VI, Paris, France
9 [1] Department of Pathology AP-HP, Hôpital Ambroise Paré, Paris, France. [2] EA 4340, Université de Versailles, Versailles, France.
* These authors contributed equally to this work.
$ These authors also contributed equally to this work.
@ Correspondence to: Professor Pierre Laurent-Puig, 45, rue des Saints Pères 75006 Paris,
France Email: pierre.laurent-puig@parisdescartes.fr
Conflict of interest statement: The authors declare that there is no conflict of interest regarding the publication of this paper.
Mutational signature associated with MUTYH deficiency in cancers
Abstract
Germline alterations in DNA repair genes are implicated in cancer predisposition and can result in characteristic mutational signatures. However, specific mutational signatures
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1002/path.4880
Accepted Article
associated with base excision repair (BER) defects remain to be characterized. Here, by analysing a series of colorectal cancers (CRC) using exome sequencing, we identified a particular spectrum of somatic mutations characterized by an enrichment of C>A transversions in NpCpA or NpCpT contexts, in three tumours from a MUTYH-associated polyposis (MAP) patient and in two cases harbouring pathogenic germline MUTYH mutations. In two series of adrenocortical carcinomas (ACC), we identified four tumours with a similar signature also presenting germline MUTYH mutations. Taken together, these findings demonstrate that MUTYH inactivation results in a particular mutational signature, which may serve as a useful marker of BER-related genomic instability in new cancer types.
Keywords: Mutational signatures, MUTYH, Colorectal cancer, Adrenocortical carcinomas
Accepted Article
Introduction
Inherited genetic defects affecting several DNA repair pathways have been associated with cancer predisposition. Defects in mismatch repair (mainly MLH1, MSH2 or MSH6 mutations associated with hereditary non-polyposis colorectal cancer), base-excision repair (MUTYH mutations in MUTYH-associated polyposis (MAP)), nucleotide-excision repair (xeroderma pigmentosum genes in skin cancer), homologous repair (mainly BRCA1/2 in breast and other cancers), and the Fanconi anaemia pathway (FANC genes in acute myeloid leukaemia and squamous cell carcinomas) lead to different types of genetic instability that should be recognizable as distinct mutational signatures is cancer genomes [1]. At present, at least 30 mutational signatures have been identified by examining the cancers of more than 12,000 patients across 40 distinct human cancer types [2-4] and these are reported in the COSMIC database. Specific signatures have been related to BRCA1/2 mutations, mismatch repair and nucleotide-excision repair deficiency [4,5], but mutational signatures of defects affecting the other DNA repair pathways remain to be characterised.
MUTYH is a DNA glycosylase involved in the repair of oxidative DNA damage, in particular of 8-oxoG, playing a critical role in base excision repair (BER) pathway. Germline biallelic loss- of-function mutations in MUTYH gene cause a condition predisposing to familial colorectal cancer (CRC) termed MUTYH association polyposis (MAP). Targeted and whole exome sequencing (WES) studies revealed an excess of C>A mutations in MAP tumours [6,7]. Here, we describe a strong association between a particular spectrum of C>A single base substitutions, similar to the previously reported signature 18 in the COSMIC database, and MUTYH mutations in a series of CRC and in two series of adrenocortical carcinomas (ACC) analysed by WES.
Accepted Article
Materials and Methods
We performed whole exome sequencing from fresh frozen tumour samples from 37 patients with advanced CRC (stage 4, treated with anti-EGFR antibodies), and two adenomas (P1 and P2) and a carcinoma (TU) from one patient with MAP (Supplementary Table S1). Analysis of the triple knockout mouse model [8] and the ACC cohorts [9,10] were based on the mutation lists provided in the original publications. We used the Wellcome Trust Sanger Institute mutational signatures framework [2] to extract signatures from the catalogues of somatic mutations identified in CRC and ACC samples.
Results and Discussion
We performed a de novo analysis of the mutational spectrum of 37 CRC using the Wellcome Trust Sanger Institute (WTSI) mutational signatures framework [2]. Mutational signature analysis identified three distinct signatures closely related to the previously described signatures 1, 5, and 18 [3] (Supplementary Figure S1A). As previously reported [11], the clock-like signature 1 (but not signature 5) was positively correlated with age (Pearson coefficient= 0.2, P=0.0042, Supplementary Figure S1B). In contrast, the aetiology of signature 18 remains unknown. Signature 18, characterised by an enrichment of C>A transversions in NpCpA or NpCpT contexts (Figure 1A), has been described in neuroblastoma and in incidental samples of other cancer types, like ACC, breast and stomach carcinomas [11], but not in CRC. In our series, signature 18 was the dominant signature in two samples (CRC103 and CRC355, Figure 1B, C). Strikingly, these two cases were the only cases harbouring germline mutations (i.e., germline variants with a reported frequency ≤1% in the ExAC database [12]) in the MUTYH gene (P=0.0015, Fisher’s exact test). CRC103 displayed a nonsense mutation (rs762307622, W156*, ExAC frequency in non-Finnish Europeans = 0%) This article is protected by copyright. All rights reserved.
Accepted Article
potentially leading to a truncated protein lacking most functional domains (Supplementary Table S2). CRC355 harboured the known pathogenic variant rs36053993 (ExAC frequency = 0.4%) leading to the amino acid substitution G396D[13]. These two variants were heterozygous in the normal samples but became homozygous in the two tumours following loss of the wild-type allele (Fig. 1d). To confirm the link between signature 18 and MUTYH mutations, we analysed by WES two adenomas (P1 and P2) and a carcinoma (TU) from one patient with MAP harbouring two different missense mutations of MUTYH: a Y179C mutation (rs34612342, ExAC frequency = 0.25%) known to be associated with polyposis [14], and a R495P mutation (ExAC frequency = 0%) predicted to be possibly damaging by using SNP&Go and SIFT softwares (Supplementary Table S2). Signature 18 was the dominant signature in these three samples, accounting for >70% of mutations (Figure 1B). As a result, the mutational spectrum in MUTYH-mutated cases displayed a dramatic increase of C>A mutations at trinucleotide contexts characteristic of signature 18 (Figure 1C). The somatic mutations identified by Rashid et al. in 8 colorectal adenomas from two MAP patients [6] also displayed a trinucleotide spectrum very similar to signature 18 (cosine similarity=0.85, Supplementary Figure S2). KRAS G12C mutations are particularly frequent in MAP-related colorectal tumours [15] and were observed in our three MAP samples. These mutations result from C>A substitutions occurring in a CCA trinucleotide context showing a high mutation probability in signature 18 (Figure 1A). Thus, BER deficiency makes MUTYH- mutated colorectal cells particularly prone to acquiring C>A mutations at CCA sites, explaining the high frequency of KRAS G12C mutations in these patients.
Accepted Article
mouse model (TOY-KO), which accumulates 8-oxoG in the nuclear DNA of gonadal cells [8]. Using WES, Ohno et al. identified 262 germline mutations acquired by the offspring of these mice. The trinucleotide mutation pattern of these mutations was highly consistent with the mutational pattern observed in our MUTYH-mutated CRC cases (Figure 2A), and a hierarchical clustering with the 30 signatures described in COSMIC showed that the TOY-KO mutational signature clustered together with signature 18 (Figure 2B). Indeed, the trinucleotide context frequencies of TOY-KO mice mutations and signature 18 displayed a cosine similarity of 0.94, where a similarity of 1.00 reflects a perfect match (Figure 2C).
In a recent study, we analysed the whole exome sequences of 45 ACC, and reported two ACC with a hypermutator phenotype dominated by C>A transversions [9]. Mutational signature analysis of these 45 ACC identified signatures 1 and 5 in most samples, and signatures 2 and 13 (related to the activity of AID/APOBEC cytidine deaminases) in three cases. Strikingly, signature 18 accounted for almost all mutations in the two hypermutated tumours (>10 mutations/Mb as compared to a maximum of 1.2 mutation/Mb in the 43 others, Fig. 3a, b). These two tumours were the only ones in the cohort harbouring germline MUTYH mutations (P=0.0010, Fisher’s exact test): ACC33 harboured the missense G396D variant (rs36053993), whereas ACC39 harboured the R241W variant (rs34126013, ExAC frequency = 0.005%) (Supplementary Table S2). These two MUTYH mutations are also known to be pathogenic [13,16] and were found in cancers with deletion of the wild-type allele by LOH of chromosome 1 (Figure 3C). In an independent series of 91 ACC from the TCGA [10], we identified two tumours with signatures similar to signature 18 (Suppl. Fig. 3). Both tumours displayed rare pathogenic MUTYH mutations (Y179C and W156*, Supplementary Table S2)
with loss of the wild-type allele in the tumour (Supplementary Figure S3), versus none of the 89 other samples (P=0.00024, Fisher’s exact test).
To confirm the similarity of signatures related to MUTYH mutations in different biological contexts, we then performed an unsupervised classification of the mutational profiles observed in CRC and ACC cases, together with the 30 signatures currently described in COSMIC and the mutational signature of the TOY-KO mouse model. The hierarchical clustering (cosine distance, Ward method) revealed a MUTYH cluster comprising the 3 MAP tumours from our series, the two MAP patients from Rashid et al., CRC103, ACC33 and ACC39, together with the signature of TOY-KO mice mutations and COSMIC signature 18 (Figure 4). One MUTYH-mutated CRC (CRC355) clustered separately due to the high prevalence of signature 1 in addition to signature 18 in this sample.
Taken together, our data suggest that mutational signature 18 reflects the accumulation of C>A mutations after incorrect replication of 8-oxoG. These mutations may accumulate in the context of defective base excision repair, e.g. in MUTYH-deficient cancers, or potentially by excessive generation of reactive oxygen species. Signature 18 has also been described in breast, stomach cancer and neuroblastoma. We examined the whole genome sequences of 560 breast cancers [17], 478 stomach cancers (TCGA-STAD) and 56 neuroblastomas [18], but we did not find any association between signature 18 and MUTYH mutations in these cancer types. Increased generation of ROS or other genetic defects or epigenetic alterations of BER pathway genes may explain signature 18 in these cancer types. Consistently, signature 18 is highly prevalent in neuroblastoma, a cancer type in which reactive oxygen species are known biological stimuli [19]. Finally, this study shows that mutational signatures can be
Accepted Article
used to detect DNA damage defects in tumours, even before a causal mutation is identified. This approach led us to identify, for the first time to our knowledge, pathogenic MUTYH mutations in ACC patients. This finding has potential clinical utility since these patients may benefit from genetic counselling and treatments (e.g. immunotherapy) taking advantage of their high mutation load. Mutational signatures are thus a useful tool that may be used in the clinics to identify patients who may benefit from specific therapies [20].
Accepted Article
Acknowledgements
C.P. is supported by fellowship from ARC. J.S. is funded by the ICE (Interpretation of Clinical Exome) project (FUI, MEDICEN, Région Ile de France). L.B.A. is supported through a J. Robert Oppenheimer Fellowship at Los Alamos National Laboratory. The INSERM UMR-S1147 is supported by the Ministère de l’Enseignement Supérieur et de la Recherche, the Université Paris-Descartes, the Centre National de la Recherche Scientifique (CNRS), the Institut National de la Santé et de la Recherche Médicale (INSERM), the Agence Nationale de la Recherche (ANR Nanobiotechnologies; no. ANR-10-NANO-0002-09), the SIRIC CARPEM, the Canceropole funding (no. 2011-1-LABEL-UP5-2) and the Ligue Nationale contre le Cancer (Program “Equipe labellisée LIGUE”; no. EL2016.LNCC/VaT). The INSERM U1162 team is supported by the Ligue Nationale contre le Cancer. The INSERM U1016 team is supported by the PHRC COMETE-Tactic. This research used resources provided by the Los Alamos National Laboratory Institutional Computing Program, which is supported by the U.S. Department of Energy National Nuclear Security Administration under Contract No. DE-AC52-06NA25396. Research performed at Los Alamos National Laboratory was carried out under the auspices of the National Nuclear Security Administration of the United States Department of Energy.
Contributions
E.L. and P.L .- P. directed the research. C.P., L.B.A., E.L. and P.L .- P. wrote the manuscript. D.L.C., Z.H .- R. and R.D. performed the experiments. C.P., J.S., L.B.A., E.L. and P.L .- P. analysed and interpreted the data. J.S., L.B.A. and E.L. carried out signatures and/or statistical analyses. G.A., T.A., J.Z .- R., J .- F.E. and J.B. provided essential biological resources and collected clinical data. All authors approved the final manuscript and contributed to critical revisions to its intellectual context.
Accepted Article
Supplementary materials and methods YES
Supplementary figure legends YES
Figure S1. Mutational signatures and correlations for 38 CRC patients
Figure S2. Mutational profiling of MAP patient samples and their similarity to mutational signature 18
Figure S3. Comparison of mutational patterns in two ACC series and COSMIC mutational signature 18
Supplementary Table S1. Somatic single nucleotide variants (SNV) among the 40 colorectal tumours
Supplementary Table S2. MUTYH SNPs characteristics
Accepted Article
References
Accepted Article
1. Curtin NJ. DNA repair dysregulation from cancer driver to therapeutic target. Nat Rev Cancer 2012; 12: 801-817.
2. Alexandrov LB, Nik-Zainal S, Wedge DC, et al. Deciphering signatures of mutational processes operative in human cancer. Cell Rep 2013; 3: 246-259.
3. Alexandrov LB, Nik-Zainal S, Wedge DC, et al. Signatures of mutational processes in human cancer. Nature 2013; 500: 415-421.
4. Alexandrov LB. Understanding the origins of human cancer. Science 2015; 350: 1175.
5. Kim J, Mouw KW, Polak P, et al. Somatic ERCC2 mutations are associated with a distinct genomic signature in urothelial tumors. Nat Genet 2016; 48: 600-606.
6. Rashid M, Fischer A, Wilson CH, et al. Adenoma development in familial adenomatous polyposis and MUTYH-associated polyposis: somatic landscape and driver genes. J Pathol 2016; 238: 98-108.
7. Sampson JR, Jones S, Dolwani S, et al. MutYH (MYH) and colorectal cancer. Biochemical Society transactions 2005; 33: 679-683.
8. Ohno M, Sakumi K, Fukumura R, et al. 8-oxoguanine causes spontaneous de novo germline mutations in mice. Sci Rep 2014; 4: 4689.
9. Assie G, Letouze E, Fassnacht M, et al. Integrated genomic characterization of adrenocortical carcinoma. Nat Genet 2014; 46: 607-612.
10. Zheng S, Cherniack AD, Dewal N, et al. Comprehensive pan-genomic characterization of adrenocortical carcinoma. Cancer Cell 2016; 29: 723-736.
11. Alexandrov LB, Jones PH, Wedge DC, et al. Clock-like mutational processes in human somatic cells. Nat Genet 2015.
12. Lek M, Karczewski KJ, Minikel EV, et al. Analysis of protein-coding genetic variation in 60,706 humans. Nature 2016; 536: 285-291.
13. Lubbe SJ, Di Bernardo MC, Chandler IP, et al. Clinical implications of the colorectal cancer risk associated with MUTYH mutation. J Clin Oncol 2009; 27: 3975-3980.
14. Al-Tassan N, Chmiel NH, Maynard J, et al. Inherited variants of MYH associated with somatic G:C — >T:A mutations in colorectal tumors. Nat Genet 2002; 30: 227-232.
15. Jones S, Lambert S, Williams GT, et al. Increased frequency of the k-ras G12C mutation in MYH polyposis colorectal adenomas. Br J Cancer 2004; 90: 1591-1593.
16. Bai H, Jones S, Guan X, et al. Functional characterization of two human MutY homolog (hMYH) missense mutations (R227W and V232F) that lie within the putative hMSH6 binding domain and are associated with hMYH polyposis. Nucleic Acids Res 2005; 33: 597-604.
17. Nik-Zainal S, Davies H, Staaf J, et al. Landscape of somatic mutations in 560 breast cancer whole-genome sequences. Nature 2016; 534: 47-54.
18. Peifer M, Hertwig F, Roels F, et al. Telomerase activation by genomic rearrangements in high-risk neuroblastoma. Nature 2015; 526: 700-704.
19. Marengo B, Raffaghello L, Pistoia V, et al. Reactive oxygen species: biological stimuli of neuroblastoma cell response. Cancer Lett 2005; 228: 111-116.
20. Alexandrov LB, Nik-Zainal S, Siu HC, et al. A mutational signature in gastric cancer suggests therapeutic strategies. Nat Commun 2015; 6: 8683.
Accepted Article
A
D
Mutational signature 18
CRC103
Percentage of mutations
Normal
Tumour
2 9
Variant allele fraction
1.0
1.0
15762307622
1762307622 (W156*).
ON
(W156”)
0.5
0.5
B
Colorectal cancers
Mutations/signature in CRC
MUTYH mutation
Signature 1
0.0
0.0
Signature 5 Signature 18
0
100
200
0
100
200
Position on chromosome 1 (in Mbp)
200
CRC355
·
MAP-4PZ
2
CRCITZ
CRCS
p
CHAC12
Normal
Turrour
1.0
1.0
Variant allele fraction
IS36053993
C
(G396D)
1636053983
Colorectal cancers
(G396D)
0.5
0.5
học
THG
MUTYH-mutated (n=5)
Percentage of mutations
10%
0.0
0.0
2
0
100
200
0
100
200
Position on chromosome 1 (in Mbp)
10%%
MUTYH-wild-type (n=35)
Accepted Article
Figure 2. Mutations identified in a model of MUTYH-deficient mice are highly similar to mutational signature 18. (A) Mutational profile of 262 germline mutations identified by whole-exome sequencing in the offspring of Mth1/Ogg1/Mutyh triple knockout mice (TOY- KO mice). (B) Unsupervised classification of trinucleotide patterns corresponding to the 30 known mutational signatures [4] and mutations identified in TOY-KO mice [8]. The dendrogram (top) shows the grouping of signatures obtained by hierarchical clustering. The heatmap below shows the prevalence of the 6 substitution types (identified by a colour code) broken down according to the trinucleotide context. (C) Cosine similarity between the mutational patterns of TKO mice mutations and mutational signature 18. The proportion of mutations corresponding to each trinucleotide context in mutational signature 18 (x-axis) and TKO mice mutations (y-axis) is represented with the same colour code as in panel B (e.g., C>A mutations are indicated in blue).
A
Mutational signature of Mth1/Ogg1/Mutyh triple knock-out (TKO) mice
C>A
C>G
O>T
T>A
T>C
T>G
Percentage of mutations
15%
10%
5%
0%
AC
Trinucleotide context
B
00 18 38
C
-
59.24
5
OCT
C>A
0
Trinucleotide context
C>G
TKO mice
C>T
15
-
T>A
0
cosine=0.94
T>C
0
2
4
6
8
10
12
Mutational signature 18
T>G
Accepted Article
A
C
ACC33
Adrenocortical carcinomas
Mutations/signature in ACC
Normal
Tumour
C
100
MUTYH
Signature 1
1.0
1.0
mutation
Signature 5
Variant allele fraction
Signature 18
rs-36053993
800
Signature 2
(G396D)
Signature 13
05
0.5
-
1536053993
(G3960)
a
AOCH
0.0
0.0
0
100
200
0
100
200
B
Position on chromosome 1 (in Mbp)
Adrenocortical carcinomas
Of
TMA
T.C
ToG
ACC39
Normal
Tumour
1.0
20%
MUTYH-mutated (n=2)
1.0
Percentage of mutations
Variant allele fraction
r$34126013
(R241W)
1834126013
0.5
0.5
(R241W)
of
MUTYH-wild-type (n=43)
20%
0.0
0.0
0
100
200
0
100
200
Position on chromosome 1 (in Mbp)
Figure 4. Unsupervised classification of the mutational patterns of 40 CRC, two MAP patients (5 and 3 colorectal adenomas), 45 ACC, TKO mice mutations and COSMIC mutational signatures. The dendrogram (top) shows the result of hierarchical clustering, highlighting a
Accepted Article
cluster of MUTYH-related signatures (dashed box). The heatmap below shows the prevalence of the 6 substitution types (identified by a colour code) broken down according to the trinucleotide context.
00
Series
MUTYH
☒ Colorectal cancers (this study)
☒ Colorectal cancers (Rashid et al.)
☒ Mutated
6
☐ Not mutated
Height
☒ Adrenocortical carcinomas
COSMIC signatures ☐
☐ Not applicable
.
☒ TKO mice (Ohro et al.)
2
0
2
Series MUTYH
CHA
C>G
Trinucleotide context
C>T
T>A
T>C
T>G
MUTYH cluster