BMI1 is a context-dependent tumor suppressor that is a barrier to dedifferentiation in non-small cell lung adenocarcinoma

Abstract

Predictive value is expected when preclinical models of disease are used for research. However, not all models appropriately mimic the disease progression or the treatment paradigm in the clinic. This thesis addresses an epigenetic regulator, Bmi1, which acts in stem cells to maintain their proliferative and self-renewal capacity primarily through silencing of the Ink4a/Arf locus. Bmi1 has been proposed as a good therapeutic candidate in cancer because of its presumed role in maintaining tumor propagating cells (TPCs). This conclusion is based on the observed tumor suppressive effects of Bmi1 deletion in in vitro cell culture models, in vivo transplant models, and autochthonous models in which Bmi1 was absent throughout development. However, to date, no one has assessed the consequences of deleting Bmi1 in existing autochthonous tumors, to mimic patient treatment in the clinic.

To accomplish this, we have generated a mouse model that allows induction of autochthonous lung adenocarcinoma, driven by oncogenic Kras and Tp53 loss (KP LUAD), and subsequent deletion of Bmi1 specifically within the tumor cells once more than half the tumors progress to grade 3 or higher. We confirmed that this model yielded Bmi1 loss that was tumor-specific and almost complete. We then aged tumor bearing mice for up to seven weeks post Bmi1 deletion to determine the impact on LUAD. Unexpectedly, Bmi1 deletion did not yield significant tumor suppression. Instead, gene expression analyses of Bmi1 deficient tumor cells revealed upregulation of a gastric gene expression program that is a known marker of lung tumor progression towards a more aggressive state in the KP LUAD model. Additionally, single cell sequencing showed that Bmi1 deficient tumors contained a higher frequency of cells that expressed previously described markers of TPCs and metastasis.

We also extended these findings to colorectal cancer where we show that deletion of Bmi1 is not tumor suppressive in either in vitro organoids or orthotopic transplants. Given these findings, we conclude that deletion, or inhibition, of BMI1 in existing tumors will be ineffective for cancer treatment in the contexts examined, and potentially deleterious because it can enable acquisition of alternate differentiation states that promote tumor progression.