Evaluation of Rare Variants in the New Fanconi Anemia Gene ERCC4 (FANCQ) as Familial Breast/Ovarian Cancer Susceptibility Alleles

Recently, it has been reported that biallelic mutations in the ERCC4 (FANCQ) gene cause Fanconi anemia (FA) subtype FA‐Q. To investigate the possible role of ERCC4 in breast and ovarian cancer susceptibility, as occurs with other FA genes, we screened the 11 coding exons and exon–intron boundaries of ERCC4 in 1573 index cases from high‐risk Spanish familial breast and ovarian cancer pedigrees that had been tested negative for BRCA1 and BRCA2 mutations and 854 controls. The frequency of ERCC4 mutation carriers does not differ between cases and controls, suggesting that ERCC4 is not a cancer susceptibility gene. Interestingly, the prevalence of ERCC4 mutation carriers (one in 288) is similar to that reported for FANCA, whereas there are approximately 100‐fold more FA‐A than FA‐Q patients, indicating that most biallelic combinations of ERCC4 mutations are embryo lethal. Finally, we identified additional bone‐fide FA ERCC4 mutations specifically disrupting interstrand cross‐link repair.

ABSTRACT: Recently, it has been reported that biallelic mutations in the ERCC4 (FANCQ) gene cause Fanconi anemia (FA) subtype FA-Q. To investigate the possible role of ERCC4 in breast and ovarian cancer susceptibility, as occurs with other FA genes, we screened the 11 coding exons and exon-intron boundaries of ERCC4 in 1573 index cases from high-risk Spanish familial breast and ovarian cancer pedigrees that had been tested negative for BRCA1 and BRCA2 mutations and 854 controls. † These authors contributed equally to this work. * Correspondence to: Jordi Surrallés, Department of Genetics and Microbiology The ERCC4 (FANCQ) gene (MIM #133520) encodes for a DNA repair endonuclease (XPF) that plays essential roles in nucleotide excision repair (NER) and interstrand cross-link repair (ICLR) [Gregg et al., 2011]. Biallelic mutations in ERCC4 have been linked to Xeroderma Pigmetosum (XP; MIMs #278700, #610651, #278720, #278730, #278740, #278760, #278780, and #278750) [Sijbers et al., 1996] and progeria (XFE; MIM #610965) [Niedernhofer et al., 2006] diseases. Very recently, it has been published that biallelic mutations in the gene are also causative of disease in previously unclassified Fanconi anemia (FA) patients subtype FA-Q and, therefore, the use of FANCQ as an alias for ERCC4 was approved by the HUGO Gene Nomenclature Committee [Bogliolo et al., 2013]. Individuals harboring mutations in ERCC4 show high variability on their clinical manifestations, ranging from mild XP symptoms (sun sensitivity, freckling of the skin, and basal or squamous cell carcinomas) to the dramatic accelerated aging symptoms of a progeroid Apart from the 854 controls in which XPF/ERCC4 was fully sequenced, these variants were specifically analyzed in 300 additional controls. The mutation gave rise to skipping of exon 3 (r.389_584del) that was confirmed at the cDNA level (data not shown). c MAF (minor allele frequency) in European American Population as described in the NHLBI Exome Sequencing Project (ESP) (http://evs.gs.washington.edu/EVS/). d Mutation found in two independent controls. e All mutations giving rise to a PTC were considered as predicted to affect protein function. Misssense mutations were previously evaluated using the program CONDEL that predicts the pathogenicity of nonsynonymous variants using a consensus deleteriousness score that combines various tools such as SIFT, Polyphen2, and MutationAssessor (http://bg.upf.edu/condel/home). Those missense variants predicted to be deleterious by condel were considered as predicted to affect protein function and this was confirmed by functional analysis; those predicted by condel to be neutral were considered as such and not further evaluated. f Detailed in the text and in Figure 1.
To investigate the possible role of ERCC4 in breast and ovarian cancer susceptibility, we screened, by DHPLC (denaturing highperformance liquid chromatography) and direct sequencing, the 11 coding exons and exon-intron boundaries of the ERCC4 gene in 1573 index cases from high-risk Spanish familial breast and ovarian cancer pedigrees that had been tested negative for mutations in BRCA1 and BRCA2 and 854 controls without personal or familial antecedents of cancer. Criteria for inclusion of cases and controls, and methods of screening for mutations in BRCA1/2 have been previously published [Bonache et al., 2013;Fernandez-Rodriguez et al., 2012;Osorio et al., 2012;Romero et al., 2011]. We identified five and four unique variants among cases and controls among which three and four, respectively, were considered as putatively deleterious. Missense mutations were evaluated using the program CONDEL that predicts the pathogenicity of nonsynonymous variants using a consensus deleteriousness score that combines various tools such as SIFT, Polyphen2, and Mu-tationAssessor (http://bg.upf.edu/condel/home). Those missense variants predicted to be deleterious by CONDEL were considered as predicted to affect protein function and this was studied later by functional analysis; those predicted to be neutral were considered as such and not further evaluated (Table 1). Regarding cases, one of the mutations found, c.584+1G>A in intron 3, was confirmed to cause skipping of exon 3 of the gene and a premature stop codon (PTC) (data not shown). The other two mutations, p.Arg150Cys in exon 3 and p.Ser786Phe in exon 11, were later functionally investigated. Regarding controls, we identified two different frameshift mutations c.540 541delAG in exon 3 (found in two independent controls) and c.2291delG in exon 11, both predicted to cause PTCs. Even though this later PTC is very C-terminal and could potentially result in a shorter but partially functional protein, this mutation results in a truncated XPF protein that lacks the double helix-hairpin-helix (HhH2) domain involved in heterodimerization with ERCC1 and DNA binding [de Laat et al., 1998], very similar to a pathogenic ERCC4 mutation (c.2371 2398dup28; p.Ile800Thrfs * 24) that functionally disrupts NER and ICLR activities [Bogliolo et al., 2013]. The last deleterious variant found in controls was the missense p.Arg689Ser in exon 11 previously found in a FA patient, and demonstrated to cause abnormal nuclease activity and to specifically disrupt ICLR [Bogliolo et al., 2013]. No differences were found regarding the localization of mutations in the gene among cases and controls (Table 1). All variants reported have been submitted to the Leiden Open Variation Database (LOVD).
To evaluate the functional impact of the missense variants, we cloned a HA-tagged wild type (WT) ERCC4 cDNA in a pBABEpuro retroviral vector (Addgene plasmid 14430, kindly shared by Dr. L.M. Martins) in internal ribosome entry site (IRES) with the GFP cDNA, and the c.448C>T, c.2065C>A, and c.2357C>T variants were introduced by site directed mutagenesis [Bogliolo et al., 2013]. The resulting constructs were transduced in NER and ICLR deficient Ercc4 KO mouse embryonic fibroblasts (MEFs) and, after puromicin selection, the green cells were sorted to achieve a purity of over 98% by fluorescence activated cell sorting [Bogliolo et al., 2013]. Due to the bicistronic nature of the IRES construct, we were able to assess the stability of the mutant XPF proteins using GFP as a reference, since both proteins are encoded by the same mRNA [Pelletier and Sonenberg, 1988]. The p.Arg689Ser variant reduced by 40% the stability of XPF, whereas the p.Arg150Cys and p.Ser786Phe variants had no impact on protein stability ( Fig. 1A and B). Ultraviolet radiation subtype C (UVC) sensitivity of Xpf KO MEFs was complemented with the expression of both p.Arg150Cys and p.Ser786Phe-XPF (Fig. 1C), and only the Xpf KO MEFs expressing p.Ser786Phe-XPF or p.Arg689Ser showed a FA phenotype in terms of mitomycin C (MMC) sensitivity (Fig. 1D), MMC-induced cell cycle arrest at the G2/M phase (Fig. 1E), and DEB-induced chromosome fragility (Fig. 1F). These data confirm that, resembling p.Arg689Ser, p.Ser786Phe specifically disrupts ICLR and, therefore, is a bone-fide FA mutation. Interestingly, both mutations are located in the nuclease domain of XPF. Despite a mild MMC sensitivity ( Fig. 1D and E), Ercc4 KO MEFs expressing p.Arg150Cys-XPF did not show DEB-induced chromosome fragility (Fig. 1F). These data,  [Bogliolo et al., 2013;Trujillo et al., 2012].
together with the protein stability and UVC sensitivity data, indicate a null impact of the c.448C>T variant on XPF NER functions and a mild effect on ICLR activity.
In conclusion, the frequency of Spanish individuals heterozygous for pathogenic mutations in the ERCC4 gene is approximately 0.3%, and it does not differ between familial breast/ovarian cancer patients and healthy controls (p = 0.251), suggesting that monoallelic muta-tions in ERCC4 are not linked to cancer susceptibility in the general population. Similar results were found with SLX4 that, like ERCC4, acts downstream FANCD2 monoubiquitination but upstream the homologous recombination step of ICLR [Fernandez-Rodriguez et al., 2012]. The prevalence of ERCC4 mutation carriers (one in 288) is similar to that reported for FANCA. However, there are approximately 100-fold more FA-A than FA-Q individuals, suggesting HUMAN MUTATION, Vol. 34, No. 12, 1615-1618 that over 90% of biallelic combinations of ERCC4 mutations are embryo lethal in humans. All reported XP patients subtype XPF worldwide have at least one missense mutation disrupting NER, whereas all missense mutations found in 2.427 Spanish individuals have substantial NER activity explaining why there are no reported XPF families in Spain.