Abstract 3761: Distinct transcriptional programs drive phenotypic heterogeneity in small cell lung cancer

Abstract

Abstract Small-cell lung cancer (SCLC), a lethal neuroendocrine cancer, lacks targetable oncogenes and is monolithically treated with standard combination chemotherapy. Interestingly, distinct subpopulations of neuroendocrine and non-neuroendocrine cells have been identified in mouse models of SCLC, but their existence in human SCLC remains unknown. Phenotypic heterogeneity, an important phenomenon in cellular reprogramming and therapeutic resistance, can arise from cell state transitions (i.e. network-level changes dynamically controlled by transcriptional regulators). The motivation for this study is to determine whether phenotypically distinct cell states exist in human SCLC, and investigate transcription factor (TF) dynamics that maintain these states. We implemented a mixed bioinformatics and experimental approach that defines inter-tumor heterogeneity in SCLC patients and cell lines as a spectrum of neuroendocrine (NE) and mesenchymal (MC) differentiation delineated by two anti-correlated gene co-expression networks. Features such as adhesion, surface markers and kinases effectively summarize heterogeneity in SCLC cell lines and patients as three distinct phenotypic states: NE, MC and intermediate. To characterize the transcriptional influence that governs these distinct phenotypic states, we constructed a TF regulatory network using the mutual information based method ARACNE. Boolean network model dynamics of the top 19 predicted TFs of the NE and MC networks gives rise to 3 distinct clusters of stable phenotypic states or ‘attractors’, each given by a unique TF network configuration. These TF network states were validated at both gene and protein level in SCLC cell lines and patients, identifying specific network states governing NE, MC and intermediate attractors. In silico TF perturbation experiments (single or combination TF activation/knockdown) performed to explore the possibility of state transitions, indicated that the NE state is more easily reprogrammable via single TF manipulations than the MC state, which required a combination of 3-4 simultaneous TF manipulations. SOX2, FOXA2 and OVOL2 were identified as master regulators of MC → NE state transitions while NOTCH1, MYC, SMAD3, and NFKB1 were master regulators of the NE → MC state transitions. Experimentally, phenotypic state transitions such as NE → Intermediate and MC → Intermediate were successfully induced using HDAC inhibitors, but not with demethylating agents or cisplatin. Etoposide treatment could also induce NE → Intermediate state transitions in NE cell lines but MC cell lines were resistant. Drug rebound experiments reveal that the transitioned cells fall back into NE/MC differentiated states upon removal of the drug. Thus, classification of human SCLC into 3 distinct phenotypic states - NE, MC and intermediate serves as a useful mapping tool for defining heterogeneity that could lead into personalized treatment strategies. Citation Format: Akshata R. Udyavar, Megan Hoeksema, David J. Wooten, Mukesh Bansal, Andrea Califano, Lourdes Estrada, Jonathan Irish, Pierre Massion, Vito Quaranta. Distinct transcriptional programs drive phenotypic heterogeneity in small cell lung cancer. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 3761. doi:10.1158/1538-7445.AM2015-3761