Whole-exome sequencing of a rare case of familial childhood acute lymphoblastic leukemia reveals putative predisposing mutations in Fanconi anemia genes

The sibs were diagnosed with nonsyndromic childhood ALL 3 years apart. We previously identified a rare PRDM9 allele segregating within the familly [11]. PRDM9 is a histone H3 methyltransferase involved in crossing-over at recombination hotspots and Holliday junctions. To further characterize the underlying inherited genetic contribution to this childhood ALL family in an unbiased manner, we performed whole exome sequencing of the siblings and both parents. Though the family was nonsyndromic and asymptomatic for FA, this recessive disorder is linked to hematopoietic dysfunction, chromosomal instability and increased susceptibility to childhood ALL. Based on the observed concordant hyperdiploid phenotype of both siblings, we postulated that inherited rare disadvantaging DNA variants in leukemia predisposing pathways like the FA pathway could affect overall genomic instability and, in combination with the rare allelic form of PRDM9, favour nondisjunction of chromosomes leading to increased risk of hyperdiploid pre-B ALL within this family. Under a recessive disease model, we interrogated our exome data and identified shared non-synonymous mutations that were either compound heterozygous or homozygous variant (Table 1) and specifically screened genes associated with the leukemia predisposing syndrome FA (FANCA, FANCB, FANCC, FANCD1/BRCA2, FANCD2, FANCE, FANCF, FANCG, FANCI, FANCJ, FANCL, FANCM, FANCN/PALB2, FANCO/RAD51C, FANCP/SLX4, FANCQ/XPF and FANCS/BRCA1). Among the identified variants, we identified a combination of missense variants in the FA gene FANCP/SLX4 (compound heterozygous at rs137976282 and rs79842542), corroborating the assumption of FA pathway destabilization (Fig. 2). A more thorough investigation of the other FA pathway genes led then to the identification of a rare heterozygous variant in FANCA (rs61753269) that was also shared by the sibs. Although this variant was heterozygous, restricting the analysis to extremely rare variants allowed us to identify potentially deleterious non-synonymous variations in FA genes that could be contributing to inherited susceptibility to ALL in the sibs. For FANCP/SLX4, both parents transmitted a putatively damaging allele to their affected offspring who were therefore compound heterozygous at rs137976282 (ESP and 1000 Genomes general population MAF <0.001) and rs79842542 (MAF =0.059 and 0.071 in 1000 Genomes and ESP general populations respectively). While two of the three in silico algorithms predicted that the compound heterozygous variants in FANCP/SLX4 were likely deleterious (Sift score =0 for both alleles and Polyphen2 score =1 and 0.964 for rs79842542 and rs137976282 respectively), only Fathmm predicted rs61753269 in FANCA to be damaging (Fathmm score = −1.78) (Table 1). Nevertheless, the high conservation score at FANCA rs61753269 (SiPhy =12.742), combined with its extreme rarity in the population (MAF <0.001 in 1000 Genomes and ESP), suggest that this variant is under strong functional constraint and therefore could have a specific role on protein conformation. Although not a Fanconi anemia gene per se, our exome data also revealed a rare non-synonymous homozygous variant in GEN1 (rs16981869, MAF =0.145394, ESP general population homozygous frequency q² = 0.025), that was predicted to be deleterious by all three algorithms. GEN1 is a member of the FANCP/SLX4 complex involved in Holliday junction resolution [27], and in conjunction with PRDM9 and the FA genes identified here, could be contributing to genomic instability in the sibs. Our in-house exome database of 369 individuals from our childhood ALL cohort (103 patient-mother-father trios and 60 patients) from the QcALL cohort [28] (whole exome sequencing performed on Life Technologies SOLiD System or Illumina HiSeq 2500; data available upon request), revealed a single heterozygote patient at both FANCP/SLX4 positions, 0/369 variant allele carriers at FANCA rs61753269 and 3/369 carriers of the homozygous allele at GEN1 rs16981869 (one patient and two parents). Interestingly, the only two other cases harbouring either both variants in FANCP/SLX4 or the homozygous variant in GEN1 were also diagnosed with hyperdiploid pre-B ALL, concordant with the sibship. Overall, the sibs were the only two individuals who carried this particularly rare combination of damaging alleles at FANCA rs61753269, FANCP/SLX4 rs137976282, rs79842542 and GEN1 rs16981869.

Fanconi anemia is a recessive genetic disorder and most frequent cause of inherited bone marrow failure. To date, 17 FA genes have been identified and mutations within these genes have been shown to cause DNA repair defects leading to genomic instability and aneuploidy, characteristic of FA [29]. Given cumulative hematopoietic dysfunction and excess chromosomal instability, FA patients are at higher risk of developing hematopoietic malignancies including leukemia [30]. Interestingly, the rare variants FANCP/SLX4 rs137976282 and FANCA rs61753269 have previously been identified in familial breast cancer cases [31‐34], however their pathological effects in cancer predisposition remain unknown. FANCA, mutated in over 60 % of FA cases, is an essential member of the FA core complex involved in monoubiquitination of the FANCI/D2 complex which in turn guides downstream activation of the DNA repair processes for resolving DNA interstrand cross-links during replication [35]. Mono-allelic deletion of FANCA has been suggested to promote genetic instabilities associated with acute myeloid leukemia [36]. FANCP/SLX4 on the other hand, is a downstream component of the FA pathway that codes for a Holliday junction resolvase. It acts as a docking platform for three structure-specific endonucleases XPF–ERCC1, MUS81–EME1 and SLX1 [37]. Recently identified as a FA gene, FANCP/SLX4 modulates DNA repair and cellular responses to replication fork failure [38]. GEN1 codes for an endonuclease, and is a member of the FANCP/SLX4 complex [27] shown to play a role in the maintenance of centrosome integrity [39]. Along with PRDM9, GEN1 and the FANCP/SLX4 complex are involved in the definition of Holliday junction branch migration boundaries and the cleavage of static and migrating Holliday junctions [12, 27, 37]. Efficient DNA damage repair and simultaneous regulation of cell cycle progression is critical for genomic stability. Interestingly, a rare recessive homozygous variant in GEN1 has been associated with bilateral breast cancer [40] and the depletion of GEN1 or FANCP/SLX4 in Bloom’s syndrome cells results in defects in chromosome condensation and severe chromosome abnormalities, such as nondisjunction of sister chromatids and abnormal mitosis leading to aneuploidy [41, 42], highlighting their important role in maintaining genome stability. Thus, mutated FANCP/SLX4 and GEN1, in combination with the rare allele of PRDM9 also segregating within this family, could alter Holliday junction resolution leading to nondisjunction of chromosomes and segregation defects.

While autosomal recessive FA patients are known to present with malformations [43], it has been reported that heterozygous carriers of a FA gene may be predisposed to some of the same congenital malformations or developmental abnormalities that are common among homozygotes [44]. Although the sibs had no apparent physical abnormalities, family history revealed a consanguineous marriage on the paternal side (Fig. 1) resulting in multiple miscarriages and polymalformation syndrome in surviving offspring. Given that both rare FANCP/SLX4 rs137976282 and FANCA rs61753269 variants were paternally inherited we could hypothesize an underlying recessive disorder affecting the FA pathway; however this remains highly speculative without further genotype information on the extended family. Overall, these data support a functional role for the rare variants identified in FANCA, FANCP/SLX4 and GEN1 in disrupting the FA pathway and Holliday junction resolution, and as a result, they could lead to genomic instability and hematopoietic dysfunction, and increased risk of ALL within this family. However functional assays are required to confirm these observations.

Despite the fact that both siblings were asymptomatic and were not diagnosed with an ALL-linked genetic disorder, the possibility of an underlying FA condition exists and an undiagnosed disorder, although rare, cannot be excluded. One may argue that pure, nonsyndromic ALL families are unlikely and that genetic interrogation of such families will ultimately reveal underlying inherited disorders associated with increased risk of ALL. Indeed, our results show that the study of familial or inherited forms of ALL can further our understanding of the genetic causes underlying more common, sporadic forms and shed light on otherwise asymptomatic genetic syndromes.

Finally, though our rare variant analysis strongly suggests FANCP/SLX4 and FANCA as the most likely candidates, we cannot exclude the possibility that additional inherited genetic variants, rare or common, outside of the FA pathway could contribute to ALL onset within the family. For example, we identified common non-synonymous variants in PDE4DIP and CEP55 (Table 1). Though these centrosomal proteins have been involved in myeloproliferative disorder [45] and carcinogenesis [46] and could promote abnormal cell division and hyperdiploidy, as evidenced recently by Paulsson et al. [47], the identified variants had high MAFs and were predicted to have benign effects on protein function, making them unlikely candidates here. Furthermore, the sibs carry common ALL susceptibility alleles at known GWAS loci [3‐6, 28] (Table 2), that under an additive effects model could lead up to a 2- to 10-fold increase in risk [9]. Given the male-specific inheritance, we also looked for shared deleterious variants on the X chromosome but found no evidence of X-linked genes contributing to ALL in this family. The exomes of the siblings were also screened for shared de novo mutations that could result from gonadal mosaicism. Putative de novo events were defined as private mutations shared by both siblings, and therefore unknown in public databases, and showing no evidence of heritability from either parent, i.e. no reads supporting the variation in the parental exomes considering a minimum coverage of 8X at the given position in the exome sequencing data. Although no candidate de novo mutation fitting our criteria was identified, the limited coverage of parental exomes may have hindered this analysis. The investigation of more complex genetic models including gene-gene and eventually gene-environment interactions could also reveal additional ALL risk factors.