Cooperation between somatic mutations and germline susceptibility variants in tumorigenesis – a dangerous liaison

ABSTRACT

High-throughput genotyping and sequencing generate comprehensive catalogs of inherited genetic variations and acquired somatic mutations. However, their possible interactions and roles in tumorigenesis remain largely unexplored. We recently reported cooperation between the EWSR1-FLI1 (Ewing sarcoma breakpoint region 1 – Friend leukemia virus integration 1) fusion oncogene and a germline variant that regulates the EGR2 (early growth response 2) Ewing sarcoma susceptibility gene via a GGAA-microsatellite.

Genome sequencing has revealed that childhood cancers exhibit a relative paucity of recurrent somatic mutations at the time of diagnosis.^1–3 In contrast, genome-wide association studies (GWAS) and their functional continuative studies conducted in pediatric cancer to date hint at a relatively strong contribution of germline variants to tumorigenesis.^4–7 However, the possible oncogenic interaction of somatic driver mutations and germline susceptibility variants remains elusive.

We recently reported that, in Ewing sarcoma (EwS), the chimeric EWSR1-FLI1 (Ewing sarcoma breakpoint region 1–Friend leukemia virus integration 1) transcription factor cooperates with a regulatory germline variant residing in a GGAA-microsatellite to upregulate EGR2 (early growth response 2),⁷ a candidate EwS susceptibility gene identified by our preceding GWAS.⁴ The EGR2 susceptibility locus also contains the ADO (2-aminoethanethiol (cysteamine) dioxygenase) gene that encodes an enzyme involved in cysteamine metabolism, whereas EGR2 is a transcription factor promoting the proliferation and survival of different cell types.⁴ We initially observed that elevated expression of either gene was associated with risk alleles.⁴ Our recent study showed that EGR2 in particular was strongly and specifically overexpressed in EwS relative to normal tissues and other pediatric tumors, and that it was co-expressed with established EWSR1-FLI1 target genes.⁷ Interestingly, the observed correlation of EGR2 and ADO expression with the germline genotype at this locus appeared to be specific for EwS, because it was not observed in EWSR1-FLI1-negative tumor types and normal tissues. Moreover, the consistent regulation of EGR2 (but not ADO) by EWSR1-FLI1 in homologous and heterologous models further suggested that EGR2 expression in EwS is controlled by both the germline genotype and EWSR1-FLI1. In functional experiments we found that inhibition of EGR2 (but not ADO) impaired proliferation and clonogenicity of EwS cell lines. Similarly, doxycycline-induced short hairpin RNA-mediated EGR2 knockdown suppressed anchorage-independent spheroidal growth in vitro and induced regression of EwS xenografts in vivo, indicating a critical role of EGR2 in EwS tumorigenicity.

To fine-map the EGR2 susceptibility locus and to identify variants that potentially contribute to EGR2 expression, we performed targeted deep-sequencing in germline DNA of 343 EwS cases and 251 genetically matched controls, which identified 290 common single nucleotide polymorphisms (SNPs) that were significantly associated with EwS. We next sought to highlight potentially regulatory variants among these SNPs that might at least partially explain the observed association signal.

In addition to statistical association, we also took into account epigenetic profiles generated in EwS cells, as recent studies suggested that causal variants may cluster in epigenetically active and cell type-specific regulatory elements.⁸ Thus, we cross-referenced our sequencing data with published EwS-specific chromatin immunoprecipitation (ChIP)-seq, DNase-seq, and ENCODE (encyclopedia of DNA elements) data.⁷ Activating chromatin marks, signals for formaldehyde-assisted isolation of regulatory elements (FAIRE), and/or DNaseI hypersensitivity were only observed at a few loci: 2 corresponded to known EGR2 regulatory elements, and 2 to GGAA-microsatellites (mSat1 and mSat2) that overlaid with EWSR1-FLI1 ChIP-seq signals.⁷ Luciferase reporter assays indicated that the known EGR2 regulatory elements had no or weak activity in EwS cells whereas both GGAA-microsatellites exhibited strong EWSR1-FLI1-dependent enhancer-like activity,⁷ which is consistent with the unique ability of EWSR1-FLI1 to use GGAA-microsatellites de novo as enhancers.^9,10 Interestingly, earlier work by us and others suggested that EWSR1-FLI1 drives many of its target genes through GGAA-microsatellites, and that enhancer activity increases with the number of consecutive GGAA-motifs.^9,10

We focused on mSat2 because of the observed high enhancer activity and its localization in a sub-haploblock containing some of the most significant EwS-associated SNPs. PCR-based targeted long-read deep-sequencing of mSat2 uncovered another EwS-associated A/T SNP, rs79965208, which was in linkage disequilibrium with a nearby sentinel-SNP from our GWAS.^4,7 The significant association of the A-allele of rs79965208 with EwS was replicated in 2 independent cohorts.⁷ Interestingly, rs79965208 converts a GGAT- into a GGAA-motif, thereby connecting 2 adjacent GGAA-repeats of mSat2. As the first GGAA-repeat contains a median number of 11 GGAA-motifs and the second contains 4 GGAA-motifs, the A-allele at rs79965208 increases the median number of consecutive GGAA-motifs from 11 to 16.⁷

Reporter assays using the reference sequence containing either the T- or A-allele at rs79965208 confirmed that the A-allele significantly increased the EWSR1-FLI1-dependent enhancer activity of mSat2.⁷ In accordance, the A-allele was associated with significantly higher global and allele-specific EGR2 expression in EwS tumors, which corresponded to preferential binding of EWSR1-FLI1 to the A-allele of rs79965208.⁷

Collectively, our data showed that EGR2 is an EwS susceptibility gene whose overexpression in tumors is mediated by EWSR1-FLI1, at least partly through a risk-conferring enhancer-like GGAA-microsatellite.⁷ To explore the possibility of multiple functional variants at this locus we repeated association testing conditional on rs79965208, which indicated that other SNPs may also have a regulatory effect on EGR2 expression, since some association signal was still observed after conditioning on rs79965208.⁷ The relatively low EGR2 expression observed in some EwS tumors also suggested that EGR2 might not always be absolutely necessary for EwS growth.⁷ As the incidence of EwS is higher in Europeans relative to Africans,⁴ we explored the allele frequencies at rs79965208 across human populations in the available 1000 Genomes project data. The significantly higher frequency of the A-risk-allele in Europeans relative to Africans suggested that rs79965208 might contribute to the variable susceptibility to EwS.⁷

To our knowledge, our findings constitute one of the first examples of how a germline susceptibility variant can inform our understanding of a cancer-specific acquired genetic abnormality.⁷ They also illustrate how cooperation of a common germline variant with a dominant oncogene can alter key biologic pathways and ultimately contribute to tumorigenicity, and possibly disease susceptibility.⁷ We hypothesize that such oncogenic cooperation may also contribute to disease susceptibility, tumorigenicity, and tumor progression of other cancer entities (Fig. 1).

Figure 1.

Putative mechanisms of cooperation between acquired somatic mutations and germline susceptibility variants jointly driving oncogene expression. (A) General principle. (B) Indirect cooperation through the simultaneous presence of somatically acquired copy number gains and germline risk alleles that improve the activity of a regulatory element, both of which contribute to increased expression of an oncogene. Such indirect cooperation was reported in neuroblastoma.⁶ (C) Direct cooperation through binding of a chimeric transcription factor (TF) generated via somatically acquired gene fusion to an enhancer whose activity is increased through germline risk-alleles. Such direct cooperation was reported in Ewing sarcoma.⁷

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

T.G.P.G. is supported by a grant from ‘Verein zur Förderung von Wissenschaft und Forschung an der Medizinischen Fakultät der LMU München (WiFoMed)', the Daimler and Benz Foundation in cooperation with the Reinhard Frank Foundation, by LMU Munich's Institutional Strategy LMUexcellent within the framework of the German Excellence Initiative, the ‘Mehr LEBEN für krebskranke Kinder – Bettina-Bräu-Stiftung', and by the German Cancer Aid (DKH-111886). This work was supported by grants from the ANR-10-EQPX-03 from the Agence Nationale de la Recherche (Investissements d'Avenir), ANR10-INBS-09–08 from the Canceropôle Ile-de-France, the Ligue Nationale Contre le Cancer (Equipe labellisée), the Institut National du Cancer (PLBIO14–237), the European PROVABES (ERA-649 NET TRANSCAN JTC-2011), ASSET (FP7-HEALTH-2010–259348), and EEC (HEALTH-F2–2013–602856).