Back

Opposing lineage specifiers induce a pro-tumor hybrid-identity state in lung adenocarcinoma

Fort, G.; Arnold, H.; Camolotto, S.; Tariq, R.; Waters, A.; O'Toole, K. T.; Snyder, E. L.

2024-12-05 cancer biology
10.1101/2024.12.02.626384 bioRxiv
Show abstract

The ability of cancer cells to alter their identity, known as lineage plasticity, is crucial for tumor progression and therapy resistance. In lung adenocarcinoma (LUAD), tumor progression is characterized by a gradual loss of lineage fidelity and the emergence of non-pulmonary identity programs. This can lead to hybrid-identity (hybrid-ID) states in which developmentally incompatible identity programs are co-activated within individual cells. However, the molecular mechanisms underlying these identity shifts remain incompletely understood. Here, we identify the gastrointestinal (GI) transcriptional regulator HNF4 as a critical driver of tumor growth and proliferation in KRAS-driven LUAD. In LUAD cells that express the lung lineage specifier NKX2-1, HNF4 can induce a GI/liver-like state by directly binding and activating its canonical targets. HNF4 also forms an aberrant protein complex with NKX2-1, which disrupts NKX2-1 localization and dampens pulmonary identity within hybrid-ID LUAD. Sustained signaling through the RAS/MEK pathway is critical for maintaining the hybrid-ID state. Moreover, RAS/MEK inhibition augments NKX2-1 chromatin binding at pulmonary-specific genes and induces resistance-associated pulmonary signatures. Finally, we demonstrate that HNF4 depletion enhances sensitivity to pharmacologic KRASG12D inhibition. Collectively, our data show that co-expression of opposing lineage specifiers leads to a hybrid identity state that can drive tumor progression and dictate response to targeted therapy in LUAD.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.