Creating novel mouse models of lung cancer
Squamous cell carcinoma (SCC) of the lung is a subtype of NSCLC that leads to ~40,000 deaths each year in the US. To study SCC, we created a genetically-engineered mouse model (GEMM) combining Sox2 overexpression (one of the most frequently amplified genes in SCC) with loss of Lkb1, a regulator of the mTOR pathway (Mukhodpadhyay et al, Cell Reports, 2014). We have now created additional squamous GEMMs and demonstrated their similarity to human lung SCC at the level of biomarker expression, signaling pathway activation, and gene expression (Mollaoglu et al, Immunity, 2018). Our laboratory and others have found that squamous tumors have an altered tumor microenvironment with high infiltration of tumor-associated neutrophils (TANs) and an increase in neutrophil extracellular traps (NETs). In contrast, lung adenocarcinoma, another type of NSCLC driven by different oncogenes, typically display elevated levels of macrophages. These results suggest that the immune microenvironment has a conserved genetic basis, and may be dictated by tumor type-specific oncogenes and/or tumor suppressor genes. To test this hypothesis, we developed multiple new GEMMs to determine the role of lineage-specifiers SOX2 (associated with squamous lung cancer) and NKX2-1 (associated with adenocarcinoma) in the tumor immune microenvironment. SOX2 is highly expressed in lung SCC and largely absent in adenocarcinoma, whereas the converse is true for NKX2-1. We used lentiviral and genetic approaches to overexpress and deplete SOX2 and NKX2-1, respectively, to determine their role in TAN infiltration in vivo. We found that SOX2 is sufficient to recruit TANs and that NKX2-1 represses TAN infiltration. Furthermore, we find that both transcription factors reciprocally regulate neutrophil chemoattractant CXCL5, which is sufficient to recruit TANs, and whose homolog is upregulated in human squamous tumors (Mollaoglu et al, Immunity, 2018). Together, this work has defined how lineage specifiers dictate the tumor immune microenvironment. We are currently studying the functions of pro-tumor TANs in the tumor microenvironment, and testing therapeutic strategies that modulate neutrophil function.
SCLC has molecular subsets with unique therapeutic vulnerabilities
Small cell lung cancer (SCLC) is a highly metastatic neuroendocrine lung tumor, with the worst prognosis among lung cancer subtypes. SCLC has historically been treated as a single disease in the clinic, predominantly with combination chemotherapy and radiation. Patients with SCLC initially respond very well to chemotherapy, but treatment relapse is universal, rapid, and associated with cross-resistance to additional therapies. The vast majority of patients succumb to the disease within two years, which presents an urgency to identify novel therapeutic targets. Our laboratory developed the first GEMM of Myc-driven SCLC (Mollaoglu et al, Cancer Cell, 2017). Rb1/Trp53/Myc (RPM) mice develop highly aggressive metastatic SCLC with an average survival time of six weeks. We found that Myc overexpression is sufficient to drive a unique tumor histopathology called “variant” SCLC and to activate a NEUROD1+ transcriptional program that is not observed in SCLC driven by other MYC family members (i.e. L-Myc and N-Myc). Importantly, we found that Myc-driven SCLC was preferentially sensitive to Aurora kinase inhibition (Mollaoglu et al, Cancer Cell, 2017), which accurately predicted the results of clinical trials (Owonikoko et al, JTO, 2019). This led us to re-investigate SCLC using unbiased approaches in light of these newly-appreciated molecular subsets. Together with collaborators, we found that compared to L-Myc and N-Myc-driven SCLC, Myc-high SCLC is preferentially sensitive to multiple targeted therapies including IMPDH1/2 inhibitors, MCL1 inhibitors, and depletion of extracellular arginine (Huang et al, Cell Metab, 2018; Dammert et al, Nat Comm, 2019; Chalishazar et al, Clin Can Res, 2019). These findings have contributed to a new view of SCLC not as a single disease, but comprising multiple molecular subsets, each with unique therapeutic vulnerabilities (Rudin et al, Nat Rev Can, 2019; Poirier et al, JTO, 2020).
SCLC subsets exhibit remarkable plasticity
There are currently four recognized molecular subtypes of SCLC, each expressing a unique lineage-related transcription factor: ASCL1, NEUROD1, YAP1, or POU2F3. Recently, we have investigated what determines SCLC subtype, and we found that it is a combination of cell of origin, genetics, and cell fate plasticity. Using mouse models and human cell lines coupled with single-cell RNA-sequencing (scRNA-seq), we made the unexpected discovery that MYC can convert ASCL1+ SCLC to NEUROD1+ and YAP1+ states in a dynamic fashion (Ireland et al, Cancer Cell, 2020). Mechanistically, we found that MYC activates Notch signaling, a known driver of the neuroendocrine (NE) to non-NE transition (Lim et al, Nature, 2017), to promote SCLC subtype evolution. In addition to cell fate plasticity, we found that tumor genetics are an important determinant of SCLC subtype, because MYCL-associated mouse models only develop ASCL1+ SCLC, whereas MYC-driven tumors also express NEUROD1 and YAP1. Moreover, cell of origin is important as we observed POU2F3+ tumors only in MYC-driven models initiated from an unknown (possibly tuft) cell type, whereas the other three subtypes arose from a neuroendocrine cell of origin.
Our discovery of subtype plasticity in SCLC predicted that individual human tumors harbor cells of not only one, but multiple subtypes. We confirmed this prediction using scRNA-seq on a human SCLC biopsy and by staining for subtype markers in human tissue (Ireland et al, Cancer Cell, 2020). Altogether, these data suggest that SCLC is remarkably heterogeneous and plastic—and due to the unique therapeutic vulnerabilities in each subtype state, we now view SCLC as a moving therapeutic target.
Finally, we have found that chemotherapy-relapsed SCLC has reduced ASCL1 expression (Wagner et al, Nature Commun, 2018), suggesting that subtype dynamics may change during treatment. We genetically deleted ASCL1 in the RPM mouse and surprisingly found that mice no longer develop neuroendocrine tumors, but instead develop bone and cartilage tumors in the lung (Olsen et al, Genes Dev, 2021). Further investigation of these tumors led us to discover that ASCL1 loss leads to a SOX9+ neural crest-stem-like state that we hypothesize precedes bone and cartilage differentiation. Importantly, ASCL1 knockdown in human SCLC cell lines also led to SOX9 induction, suggesting a conserved (likely indirect) mechanism of regulation. Together, these studies reveal SCLC’s remarkable capacity to evolve to new transcriptional states in the face of various pressures.
We are actively working with clinicians to translate our preclinical findings to clinical trials. Many of our ongoing projects seek to understand the key differences between SCLC subtypes at the level of cell differentiation, metabolism, immune microenvironment, and drug resistance. We aim to determine the mechanisms that regulate cell fate plasticity with the goal of constraining plasticity or directing cells toward fates that we can more effectively treat. We also have significant efforts toward identifying mechanisms of chemotherapy resistance in SCLC. Many of our findings have similarities in other cancer types like the childhood brain tumor, medulloblastoma, and in neuroendocrine prostate cancer. Together, these studies will impact diagnostic and therapeutic strategies for lung cancer and ultimately help tailor therapy to the individual patient’s disease.