Osteosarcoma exhibits marked patient-to-patient heterogeneity, but little is known about heterogeneity within individual tumors. This study focuses on the roles that phenotypic plasticity and clonal selection play as tumors adapt to primary and metastatic microenvironments. We show that osteosarcomas have a high degree of transcriptional heterogeneity, like osteoblasts, that is retained even after prolonged cell culture or adaptation to changing microenvironments. We find that both cell lines and PDXs grown in cell culture or as flank tumors adopt markedly different transcriptional profiles when grown as primary bone tumors or metastatic lung lesions. By combining lineage tracing with single-cell transcriptomics, we find that very little clonal selection occurs when tumors grow in the tibia, but significant expansion of select clones occurs when grown as experimental metastases in the lung. Interestingly, the selective pressures that drive clonal expansion do not cause narrowing of transcriptional phenotypes. By comparing the phenotypes from transcriptional clusters in orthotopic/metastatic tumor pairs, we identify a transcriptional signature that is shared among clusters that become enriched during lung colonization. This includes responses to IFNG, TNF, PDGF, previously unidentified IL1B, and a shift away from genes associated with glycolysis and toward those associated with oxidative metabolism. The metastatic microenvironment enriches for phenotypically diverse clones that each display metabolic properties that engender fitness within the metastatic microenvironment. Together, these data suggest that an underlying program, possibly a developmental program retained from the tissue of origin, maintains phenotypic heterogeneity, even during adaptation to changing microenvironmental conditions.

Recent advances in the characterization of pioneer factors have led to new questions about domain structures and chromatin recognition. Seemingly separate fields have integrated common scientific problems with the identification of fusion pioneer factor (PF) pairs in childhood cancers. Excitingly, this represents a convergence of fields within science, of pediatric oncology, genomics, epigenetics, and biophysics. With PF-pairs emerging as driver oncogenes in childhood rhabdomyosarcoma (RMS), our studies investigate these fusion genes as a gestalt phenomenon, where the whole might equal more than the sum of the parts. We developed a new method for ChIP-seq with per-cell normalization (pc-ChIP-seq) to simultaneously define genome size and PAX3-FOXO1 localization in RMS. We report novel pioneer function for the major driver oncogene in RMS, with nucleosome-motif targeting and kinetic displacement of nucleosomes in human cells.