Infants with Leukemia Exhibit Compound Heterozygosity for Deleterious Germline Variation in MLL3 Independent of MLL-Rearrangements

Introduction: Infant Leukemia (IL) has a grim prognosis; fewer than half of cases survive, and survivors are often left with lifelong deficits in cognitive and other development as a result of treatment (Chow, Friedman, & Yasui, 2007; Linabery & Ross, 2008). Rearrangements in the Mixed Lineage Leukemia (MLL+) gene are frequent in IL (Ross, Davies, Potter, & Robison, 1994; Sam et al., 2012), but, on their own, are not sufficient to induce short-latency leukemia (Montes et al., 2011). Tumor sequencing studies have demonstrated that there are very few somatic mutations present in these malignancies (Andersson et al., 2012), suggesting that there are other factors contributing to IL pathogenesis.

To explore the possible contribution of rare, germline variation to early leukemogenesis, we previously completed germline exome sequencing of infants with leukemia and their mothers using samples collected by the Children’s Oncology Group. We showed that infants with wild-type MLL are born with an enrichment of rare, non-synonymous, deleterious variation in genes that were previously and independently shown to frequently harbor somatic mutations in leukemia (Valentine et al., 2013). This work is a complementary analysis of newly completed MLL+ IL germline exome sequencing.

Methods: We performed germline exome sequencing on 14 mother-infant pairs with MLL+ IL and 25 unaffected children. Variants were filtered as previously described (Valentine et al., 2013), and lists of 655 AML- and 125 ALL-related genes were obtained from COSMIC (http://www.sanger.ac.uk/genetics/CGP /cosmic). All frequently occurring variants in MLL3 were validated by Sanger sequencing.

Results: Like MLL-cases, MLL+ IL cases had a significant enrichment of rare, non-synonymous germline variation in leukemia-associated genes, relative to controls (Fisher’s Exact: p=0.002 for ALL, p=1.29E-10 for AML). A randomization test controlling for bias due to gene size or mutational heterogeneity in the candidate gene list remained highly significant (99.4 percentile for ALL and >99.99 percentile for AML). Independent of MLL status, there is a high prevalence of germline compound heterozygosity for rare, non-synonymous, deleterious variants in MLL3 (100% in infants with AML, ~60% in infants with ALL). Further, the variation clustered 5’ to MLL3’s two RING domains, important for recognizing H2A- and H2B-specific ubiquitylation and modulating the enzyme’s methyltransferase activity (Vethantham et al., 2012; Wu, Lee, Zhou, Nguyen, & Muir, 2013). This pattern of germline MLL3 variation was validated in an independent cohort undergoing whole genome sequencing (Andersson et al., 2012).

Discussion: There is mounting evidence that inherited germline variation plays an important role in IL. MLL3, a homolog of MLL, is important for developmental regulation of HOX genes and mesoderm (Bhagwat & Vakoc, 2014). Further, MLL3 was recently shown to act as a tumor suppressor in adult AML (Chen et al., 2014), and is among the most frequently mutated genes across all human cancers (Lawrence et al., 2014). Our observation of deleterious germline compound heterozygosity in MLL3 suggests that dysfunction of this gene may result in epigenetic dysregulation during hematopoietic differentiation, which could contribute to the in utero development of leukemia in these infants. Further studies to characterize the role of MLL3 in hematopoiesis and leukemogenesis are underway.