Exome Sequencing of Familial MDS Reveals Novel Mutations and High Rates of False Positive Mutations in MLL3 Due to Pseudogene Effects

Abstract

IntroductionFamilial MDS is a rare disease and has been associated with mutations in multiple genes including GATA2. Mixed Lineage Leukemia 3 (MLL3) encodes a histone methylase that is a tumor suppressor and implicated in poor prognosis in MDS and AML. It occurs at high frequency and across multiple tissue types in genomic surveys of somatic mutations in cancer, including reports of childhood AML and MDS. To investigate the genetic basis of MDS we analyzed the sequence variation in familial and non familial cases of MDS.MethodsExome sequencing of blood and germline tissue of a young patient with MDS with her mother and brother was performed. The patient presented with pancytopenia at the age of 15 and had monolobated megakaryocytes, macrocytosis and dysplastic appearance on bone marrow exam. Her mother and younger brother have milder variants of similar disease and have unexplained cytopenias without any nutrient deficiencies or chronic diseases. Data was also compared to other exomes performed by us and by other groups, by Illumina next generation sequencing. The raw data was processed using standard techniques (Illumina preprocessing, Burroughs Wheeler Alignment using GATKv3 via Genesifter). The lists of high quality mutations were compared to each other and shared mutations were identified. The presence of a selection of these mutations in MLL3 were validated by PCR and direct Sanger sequencing. Predicted mutations were cross-referenced to the COSMIC and TCGA databases.ResultsExome sequencing revealed a intronic mutation in p53 (C>T at Chr 17 7579596) in all affected family members. We also identified a shared germline variant affecting BCL6B and a previously identified mutation in NOX5. BCL6B was homozygous for an insertion of an extra CAG at Chr 17:6928019 causing p.S235G change. BCL6B is a functional tumor suppressor, involved in early B cell development and COSMIC database lists for presence of mutations in breast, CNS, endometrium, large intestinal cancers. NOX5 was homozygous for a 6 base pair deletion at Chr 15:69,238,301 causing p.EP150del alteration. NOX5 encodes a NADPH oxidase that has been implicated in hairy cell leukemia, esophageal and prostate cancer and it thought to contribute to increased tumor proliferation. Functional studies are underway to understand the role of these variants.Interestingly, exome sequencing also identified missense mutations in the MLL3 gene (p.R284Q, p.R526P). Surprisingly, sequencing of unrelated MDS patients in our institution also identified many highly scoring, identical mutations in MLL3 (p.Y987*, p.T316S, p.P860S, p.G892R, p.D348N, p.E765G). Similar mutations have been reported at a high frequency in childhood leukemia (Leukemia. 2014 Jun;28(6); T insertion at chr7:151945072 (rs150073007)) as well as in a case of familial AML (Blood: 121 (8), 2013; T insertion at codon 817 in exon 14). These mutations occurred in clusters and match paralogous areas of the genome. Conventional PCR from within regions matching pseudogene copies appear to show heterozygous mutations. We then devised long range PCR (covering between 3 to 20 Kb) from non-conserved regions and determined that these mutant calls are false positives. Many of these mutations are identical those in catalogues of SNPs, somatic mutations in cancer are implicated in disease processes. Examination of the TCGA database reveals that MLL3 mutation has been reported at a high frequency in not only liquid tumors but also solid tumors such as melanoma (24%), lung adenocarcinoma (19%), pancreatic cancer (6%) and others. Examination of the mutations reveals that the false positive calls may be responsible for many of these calls.ConclusionsWe have determined novel genetic variants in a family with MDS and cytopenias. Functional studies are underway to determine the significance of these variants. More interestingly, we have determined that there is a high rate of false positive variation calls within the MLL3 gene. This is caused by systematically misattributing homologous or unassembled areas of the genome to MLL3. The significance of MLL3 as an oncogene has been misrepresented by this phenomenon. Long range PCRs from unique genomic regions are recommended to test for these mutations. The psuedogene effect can be the source of systematic bias that affects many other genes and has misrepresented their degree of natural polymorphism and correlation with malignancy. DisclosuresNo relevant conflicts of interest to declare.