Given our previous success in using whole exome sequencing to identify somatically mutated driver genes for serous endometrial cancer (Nature Genetics 2012; 44:1310-1315) we have taken this same approach to identify somatically mutated driver genes for endometrial carcinosarcomas. Our study design consists of 4 major phases: Phase-1: Mutation discovery. Phase-2: Mutation validation. Phase-3: A two-tiered mutation prevalence screen. Phase-4: Nomination of candidate driver genes. In past reporting periods, we completed phase-1 and phase -2 of the project. Specifically, we procured a series of de-identified primary endometrial carcinosarcoma tumors and paired non-tumor tissues through the Cooperative Human Tissue Network, which is supported by the National Cancer Institute, and exome sequenced a subset of tumor-normal pairs (n=16) were subjected to whole exome sequencing at by the NIH Intramural Sequencing Center. Short sequence reads for each exome were aligned to the human reference sequence and sequence variants in the normal and tumor exomes by called by our NHGRI collaborators Dr. Nancy Hansen and Dr. James Mullikin. The variant calls consisted of germline variants (present in both tumor and normal DNAs), somatic variants (present in tumor and matched normal DNAs) and false positive calls. In work that was led by former Postdoctoral fellow Dr. Matthieu Le Gallo, my laboratory rigorously analyzed the exome variants to identify those that were somatic and exclude germline variants and false positive calls. Next, we orthogonally validated somatic variants calls using Sanger sequencing. We subsequently prioritized a subset of validated somatically mutated genes, referred to as genes-of-interest, for further analysis in a two-tiered mutation prevalence screen. The purpose of the prevalence screen is to rigorously define the frequency and spectrum of mutations among these genes in a large cohort (n=82) of primary endometrial carcinosarcomas. Tumor DNAs sequenced in tier-1 of the mutation prevalence screen were obtained from carcinosarcomas within our own tumor bank and also provided by extramural collaborators Dr. Paul Goodfellow and Dr. David Mutch; tumors in tier-2 of the mutation prevalence screen were provided by extramural collaborator Dr. Helga Salvesen*. In the previous reporting period we completed tier-1 of the mutation prevalence screen and identified any existing gaps in sequence coverage. In the current reporting period we filled coverage gaps existing in tier-1 of the mutation prevalence screen. This effort was lead by Meghan Rudd, a senior Biologist in the laboratory, and consisted of polymerase chain reaction (PCR) amplification of 4,000 targets (amplicons) followed by bidirectional Sanger sequencing of each amplicon. The resulting sequence data (8,000 sequence reads) were analyzed to identify potential somatic mutations. Putative somatic mutations were subsequently verified by PCR and sequencing of matched tumor-normal DNAs. Also in the current reporting period, we extended our study to conduct tier-2 of the mutation prevalence screen. This consisted of PCR amplification and bidirectional Sanger sequencing of 8,900 targets, followed by analysis of the sequence reads (17,756 reads) to identify somatic mutations. This phase of the project is nearing completion. Thus far, our genomic findings have allowed us to nominate a novel candidate driver gene for endometrial carcinosarcomas. In follow up studies on this gene, led by staff scientist Dr. Mary Ellen Urick, we used Western blotting and RT-PCR to evaluate protein and transcript expression levels for this gene in primary endometrial carcinosarcoma tumor tissues. Because endometrial carcinosarcomas are believed to originate from endometrial carcinomas that undergo a metaplastic transition, we have extended our study to search for somatic mutations in this gene and related genes from a large number of endometrial carcinomas within our tumor bank. A manuscript describing our findings is in preparation. *Deceased
