Charting oncogenicity of genes and variants across lineages via multiplexed screens in teratomas

To deconstruct tissue-specific effects of genes and variants on proliferative advantage and transformation potential we developed a novel approach, harnessing single-cell cancer driver screens in teratomas coupled with hit enrichment by serial teratoma reinjection, to simultaneously screen drivers across multiple lineages in vivo. Using this system, we analyzed population shifts and lineage-specific enrichment for 51 cancer associated genes and gene variants, profiling over 100,000 cells spanning over 20 lineages, across two rounds of serially injected teratomas. We confirmed that c-MYC alone or combined with myristoylated AKT1 potently drives proliferation in progenitor neural lineages, demonstrating signatures of malignancy. These drivers directed teratoma development to lineages representative of pediatric tumors such as medulloblastoma and rhabdomyosarcoma. Additionally, mutant MEK1 (S218D/S222D) provides a proliferative advantage in mesenchymal lineages like fibroblasts. We first transduced H1 human embryonic stem cells (hESCs) with a library of 51 cancer drivers and injected them into immunodeficient mice to form round 1 teratomas in 4 mice. Cells prior to injection as well as each round 1 teratoma was dissociated and profiled via single cell RNA-seq. Cells from the dissociated round 1 teratomas were reinjected into immunodeficient mice to form round 2 tumors, 2 of which were profiled via single cell RNA-seq. We then removed the most potent drivers in the initial screen, c-MYC and myristoylated AKT1, and formed a driver sub-library. H1 hESCs were transduced with this sub-library and injected into immunodeficient mice to form round 1 teratomas. 3 of these round 1 teratomas were profiled via single cell RNA-seq. Cells from the dissociated round 1 teratomas were reinjected into immunodeficient mice to form round 2 tumors, 2 of which were profiled via single cell RNA-seq. For each of the single cell RNA-seq runs, driver barcodes were associated with cell barcodes by amplifying from unfragmented cDNA. Individual drivers which were top hits (c-MYC, myr-AKT1, c-MYC + myr-AKT1 and MEK1 (S218D/S222D)) were also profiled via bulk RNA-seq (2 tumors each) and barcode enrichment was assessed from genomic DNA.