(P 89) GENERATION OF CHONDROCYTE-DERIVED MOUSE OSTEOSARCOMA MODELS - TRACKABLE SYSTEMS TO DELINEATE ONCOGENIC AND METASTATIC ENVENTS PRIOR TO TUMOROGENESIS

Objective: Osteosarcoma (OS) incidence peaks during puberty and coincides with rapid longitudinal bone growth. Primary OS is frequently localized at the growing ends of long bones formed by endochondral ossification, a process driven by growth plates comprised of chondrocytes, which themselves are a significant source of osteoblasts contributing to bone formation (X Zhou et al PlosGen 2014, L Yang et al PNAS 2014). Given the temporal and spatial links, we propose that growth plate chondrocytes are the origin of osteosarcoma cells, and if proven, we aimed to delineate the primary oncogenic as well as the disseminating events leading to osteosarcoma genesis.

Methods: To explore growth plate chondrocytes as a potential OS cell of origin, we conditionally deleted p53 alleles alone, or p53 and Rb1 alleles specifically in chondrocytes through either Col10a1-Cre (hypertrophic chondrocytes) or Agc-CreERT2 (all chondrocytes)-driven recombination. These mutant mice also harbor a fluorescence reporter allele (ROSA26-Tomato), which is expressed upon Cre recombination, providing a visual tool for tracking the entire collection of mutant cells, including the Cre-active ancestries and their descendances from their origins to any subsequent disseminating sites. The derived mice were observed for tumor formation.

Results: The Col10p53flfl(tm) and [Col10p53flflRbfl+(tm)] mutants developed osteoblastic OS with 100% penetrance, the latter cohort with shortened latencies and increased metastases compared to that of the former.
The tamoxifen-induced AgcCreERT2p53flfl(tm) and AgcCreERT2p53flflRbfl+(tm) mutants developed OS in an age dependent manner correlating with periods of growth: the penetrance was inversely correlated with the age of the mice at the time of tamoxifen treatment, from 100% at younger than 2-week-old to nearly zero after 2 months of age, In both of the Cre-driven models, the tumors occurred exclusively in endochondral bones, often in metaphases adjacent to growth plates. The WGS of the primary and metastatic tumors revealed cytogenetic complexity resembling human OS.

The pulmonary metastases were detectable at euthanasia in ~10% of the Col10p53flflRbfl+tm and ~25% of the AgcCreERT2p53flflRbfl+tm mutant mice that developed OS. Fluorescence microscopy uncovered varied numbers of tm+ cell clusters with sizes below the visible limit ( < 0.5mm: micro) dispersed throughout the lungs of mOS mice, which didn’t present perceptible lung malignancies at autopsy. In comparison, large pulmonary diseases were present in nearly all mice who were permitted to survive longer through amputation of the primary OS site. The micro- tm+ cells were capable of robust sarcosphere-formation, as well as capability for osteo- and adipogenesis in vitro. The RNA-seq profiling of lung metastatic cells identified sets of differentially expressed genes like that of human OS lung metastases. A hierarchical relationship existed between control, primary tumors, micro-metastatic lung disease and bulk lung metastases. The data suggested that the micro-metastatic lung cells may yield clues as to events earlier in the process of dissemination. Additional work also demonstrated less genetically complex OS cells in the regional bone marrow stroma, suggesting early precursors exist in that environment as well.

Conclusion: The chondrocyte-derived murine OS resembles human OS in histology, cytogenetic complexity, expression profiles and drug responses. Compared to Osx-Cre-driven osteosarcoma models the skeletal distribution of the OS is more reflective of patterns seen in human OS. This is concordant with the known relationship of osteosarcoma to growth plates. The reporter transgene allowed us to detect, isolate and characterize micro-metastatic cells. These micro-metastatic cells may yield insights into key genetic pathways involved in metastases. OS precursor cells were present in the mutant bone stroma as well. We believe these chondrocyte-derived murine OS models are faithful tools with which to study OS derivation and dissemination.