Search

Advances in the fields of cancer initiating cells and high-throughput in vivo shRNA screens have highlighted a need to observe the growth of tumor cells in cancer models at the clonal level. While in vivo cancer cell growth heterogeneity in xenografts has been described, it has yet to be measured. Here, we tested an approach to quantify the clonal growth heterogeneity of cancer cells in subcutaneous xenograft mouse models. Using a high-throughput sequencing method, we followed the fate in vitro and in vivo of ten thousand HCT-116 cells individually tagged with a unique barcode delivered by lentiviral transduction. While growth in vitro was less homogeneous than anticipated, we still find that 95% of the final cells derived from 80% of the original cells. In xenografts, however, 95% of the retrieved barcoded cells originated from only 6% of the initially injected cells, an effect we term "clonal dominance". We observed this clonal dominance in two additional xenograft models (MDA-MB-468 and A2780(cis)) and in two different host strains (NSG and Nude). By precisely and reproducibly quantifying clonal cancer cell growth in vivo, we find that a small subset of clones accounts for the vast majority of the descendant cells, even with HCT-116, a cell line reported to lack a tumor-initiating compartment. The stochastic in vivo selection process we describe has important implications for the fields of in vivo shRNA screening and tumor initiating cells.

Endoglin (ENG), a co-receptor for several TGFβ-family cytokines, is expressed in dividing endothelial cells alongside ALK1, the ACVRL1 gene product. ENG and ACVRL1 are both required for angiogenesis and mutations in either gene are associated with Hereditary Hemorrhagic Telangectasia, a rare genetic vascular disorder. ENG and ALK1 function in the same genetic pathway but the relative contribution of TGFβ and BMP9 to SMAD1/5/8 activation and the requirement of ENG as a co-mediator of SMAD phosphorylation in endothelial cells remain debated. Here, we show that BMP9 and TGFβ1 induce distinct SMAD phosphorylation responses in primary human endothelial cells and that, unlike BMP9, TGFβ only induces SMAD1/5/8 phosphorylation in a subset of immortalized mouse endothelial cell lines, but not in primary human endothelial cells. We also demonstrate, using siRNA depletion of ENG and novel anti-ENG antibodies, that ENG is required for BMP9/pSMAD1 signaling in all human and mouse endothelial cells tested. Finally, anti-ENG antibodies that interfere with BMP9/pSMAD1 signaling, but not with TGFβ1/pSMAD3 signaling, also decrease in vitro HUVEC endothelial tube formation and inhibit BMP9 binding to recombinant ENG in vitro. Our data demonstrate that BMP9 signaling inhibition is a key and previously unreported mechanism of action of TRC105, an anti-angiogenic anti-Endoglin antibody currently evaluated in clinical trials.