Small Cell Lung Cancer Plasticity Enables NFIB-Independent Metastasis

Abstract

Metastasis is a major cause of morbidity and mortality in patients with cancer, highlighting the need to identify improved treatment and prevention strategies. Previous observations in preclinical models and tumors from patients with small cell lung cancer (SCLC), a fatal form of lung cancer with high metastatic potential, identified the transcription factor NFIB as a driver of tumor growth and metastasis. However, investigation into the requirement for NFIB activity for tumor growth and metastasis in relevant in vivo models is needed to establish NFIB as a therapeutic target. Here, using conditional gene knockout strategies in genetically engineered mouse models of SCLC, we found that upregulation of NFIB contributes to tumor progression, but NFIB is not required for metastasis. Molecular studies in NFIB wild-type and knockout tumors identified the pioneer transcription factors FOXA1/2 as candidate drivers of metastatic progression. Thus, while NFIB upregulation is a frequent event in SCLC during tumor progression, SCLC tumors can employ NFIB-independent mechanisms for metastasis, further highlighting the plasticity of these tumors.

Significance: Small cell lung cancer cells overcome deficiency of the prometastatic oncogene NFIB to gain metastatic potential through various molecular mechanisms, which may represent targets to block progression of this fatal cancer type.

Fig. 1.. NFIB is not required for the development of SCLC tumors in the RPR2 mouse model

A. Representative images of serial sections from RPR2 and RPR2;Nfib^f/f mutant tumors stained with H&E (hematoxylin and eosin) or immunostained for GFP (brown signal, with blue hematoxylin counterstain), 4.5 months post Ad-CMV-infection. Scale bar, 100 μm. B. Representative images of the lungs of RPR2 and RPR2;Nfib^f/f mutant mice stained with H&E, 4.5 months post Ad-CMV-Cre infection. Arrows point to representative tumors. Scale bar, 1mm. C. Quantification of (B) for tumor number per mm². Here and in (D), tumors were confirmed by GFP immunostaining of serial sections as in (A). Point shapes represent lungs from n=2 cohorts of mice given virus at separate times. Outlined points represent the averages from each cohort. D. Quantification of (B) for tumor burden (% tumor area relative to lung area). E. Representative images of serial sections from RPR2 and RPR2;Nfib^f/f mutant tumors stained with H&E or immunostained for NFIB. Scale bar, 100 μm. F. Quantification of NFIB expression from (E). Here and below, point shapes represent tumors from two lung sections, and outlined points represent the averages from each lung. G. Representative images of RPR2 and RPR2;Nfib^f/f mutant tumors immunostained for ASCL1. Scale bar, 100μm. H. Quantification of ASCL1 expression from (G). I. Representative images of RPR2 and RPR2;Nfib^f/f mutant tumors immunostained for Ki67. Scale bar, 100 μm. J. Quantification of Ki67 expression from (I). K. Representative images of RPR2 and RPR2;Nfib^f/f mutant tumors immunostained for the neuroendocrine marker UCHL1 or the non-neuroendocrine marker HES1. Scale bar, 100 μm. L. Quantification of HES1 expression from (K). All p values calculated by two-tailed Mann-Whitney test. Error bars in (C) and (D) indicate mean ± SEM. Error bars in (F), (H), (J), and (L) indicate mean ± SD.

Fig. 2.. NFIB contributes to tumor development when upregulated but is not required for metastasis in the RPR2 mouse model

A. Representative images of RPR2 and RPR2;Nfib^f/f mutant lungs stained with H&E, 6.5 months post Ad-CMV-Cre infection. Scale bar, 1 mm. B. Quantification of (A) for tumor burden. Here and below, point shapes represent lungs from n=3 cohorts of mice given virus at separate times. Outlined points represent the averages from each cohort. C. Survival curves and median survival time of RPR2 and RPR2;Nfib^f/f mutant mice. D. Representative images of RPR2 and RPR2;Nfib^f/f mutant lungs stained with H&E at morbidity. Scale bar, 1 mm. E. Quantification of (D) for tumor burden. F. Proportion of RPR2 and RPR2;Nfib^f/f mutant livers with SCLC tumors at morbidity. G. Quantification of metastatic tumor burden in RPR2 and RPR2;Nfib^f/f mutant livers at morbidity. H. Quantification of metastatic tumor number per mm² in RPR2 and RPR2;Nfib^f/f mutant livers at morbidity. I. Representative images of livers (left) and metastatic tumors (right) from RPR2 and RPR2;Nfib^f/f mutant mice at morbidity, immunostained for NFIB. Scale bar, 1 mm (left), 100 μm (right). J. Representative images of serial sections of metastatic tumors as in (I), stained with H&E or immunostained for ASCL1. Scale bar, 100 μm. P values for (B), (E), (G), and (H) calculated by two-tailed Mann-Whitney test. P value for (C) calculated by logrank test. P value for (F) calculated by two-tailed Fisher’s exact test. Error bars indicate mean ± SEM.

Fig. 3.. NFIB loss is counterselected for at the time of tumor initiation in RPM mutant mice

A. Representative images of RPM and RPM;Nfib^f/f mutant lungs stained with H&E, 1 month post Ad-CGRP-Cre infection. Arrows point to representative tumors. Scale bar, 1 mm. B. Quantification of (A) for tumor number per mm². C. Representative images of RPM and RPM;Nfib^f/f mutant tumors immunostained for NFIB, 1 month post Ad-CGRP-Cre infection. The arrow points to a lesion with low/no NFIB expression. Scale bar, 100 μm. D. Staining intensity for NFIB expression from (C) from a scale of 0 (no signal) to 3 (high signal). E. Representative images of RPM and RPM;Nfib^f/f mutant lungs stained with H&E, 2 months post Ad-CGRP-Cre infection. Scale bar, 1 mm. F. Quantification of (E) for tumor burden. G. Survival curves and median survival time of RPM and RPM;Nfibf/f mutant mice. H. Representative images of RPM and RPM;Nfib^f/f mutant lungs stained with H&E at morbidity. Scale bar, 1 mm. I. Representative images of RPM and RPM;Nfib^f/f mutant lungs immunostained for NFIB at morbidity. The arrow points to a large tumor with low/no NFIB expression. Scale bar, 1 mm. J. Representative images of RPM and RPM;Nfib^f/f mutant tumors immunostained for NFIB and Ki67. The near-ubiquitous staining for Ki67 made it challenging to accurately quantify the number of positive cells. Scale bar, 100 μm. P values in (B) and (F) calculated by two tailed Mann-Whitney test. P value in (G) calculated by logrank test. Error bars indicate mean ± SEM.

Fig. 4.. NFIB^neg and NFIB^high SCLC metastases display distinct chromosomal and transcriptional features

A. Cell lines used for low-pass whole genome sequencing, their genotypes, and the type of tumor from which they were derived. B. Plots showing chromosome 4 gains and losses from three cell lines, as noted. Yellow lines indicate loci for Nfib, Mycl, and Nfia for 1399M2. N2N1G cells are derived from lymph node metastases from the RPR2 model and show Nfib genomic amplification as well as Mycl amplification. Cell lines derived from liver metastases from the RPR2;Nfib^f/f model show no genomic amplification of the Nfib locus with the exception of 1399M2. C. Immunoassay for NFIB levels in cell lines in (A). Iso1 and iso3 indicate protein isoforms 1 and 3 of NFIB, respectively. HSP90 serves as a loading control. D. Plots showing genome-wide gains and losses from low-pass whole genome sequencing of cell lines in (A). Note the loss of chr19 in two of the RPR2;Nfib^f/f cell lines (1398M2 and 1399M2). E. Principal component analysis of RNA sequencing from primary lung tumors and liver metastases macrodissected from the RPR2 and the RPR2;Nfib^f/f models. F, G. Top enriched terms from KEGG 2021 Human pathways in genes upregulated in liver metastases over primary lung tumors in the RPR2 (F) and RPR2;Nfib^f/f (G) models. Differentially expressed genes (log2 fold change > 1.5, adjusted p value < 0.05) were filtered for those that did not overlap with the liver gene set in the Mouse Gene Atlas to account for contaminating normal liver tissue. P values were calculated by Enrichr (Fisher’s exact test).

Fig. 5.. NFIB^neg SCLC metastases show distinct chromatin states enriched for FOXA1/2 and ASCL1-related motifs

A, B. Differential accessibility (log2 fold change in reads per peak) plotted against the mean reads per peak by ATAC sequencing. Comparisons are between cell lines derived from RPR2;Nfib^f/f mutant metastases (mets) and cell lines derived from (A) RPR2;Nfib^f/f mutant primary tumors and (B) RPR2 mutant metastases. Number of peaks that are differentially accessible are indicated. C. Correlation of ATAC sequencing with cell lines derived from RPR2;Nfib^f/f mutant liver metastases, RPR2;Nfib^f/f mutant primary lung tumors, RPR2 mutant liver metastases, and RPR2 mutant lymph node (LN) metastases. D, E. Motif enrichment in regions differentially open in RPR2;Nfib^f/f metastases from (A) and (B), respectively. Orange bars indicate motifs for transcription factors that are expressed in SCLC tumors as determined by RNA sequencing (see Figure S6C). Asterisks indicate basic helix-loop-helix (bHLH) transcription factors. Duplicate motifs identified for TCF12 (C), TCF3 (D), and FOXA1 (C,D) are not shown but can be found in Table S5.

Fig. 6.. FOXA1 binding sites are enriched in open chromatin regions in NFIB^neg SCLC cells

A. Tornado plot showing IgG and FOXA1 binding signal by CUT&RUN analysis at all accessible ATAC peaks in N2N1G (NFIB^high cell line derived from RPR2 metastasis) and 1399M2 (NFIB^neg cell line derived from RPR2;Nfib^f/f metastasis). B. Quantitation of genome-wide FOXA1 binding in (A) shows no significant difference between N2N1G and 1399M2. C, D. Heatmap and profile plots showing FOXA1 binding at ATAC peaks that lose (C) and gain (D) accessibility in 1399M2 compared to N2N1G. E, F. Motif enrichment in FOXA1 binding sites in N2N1G (E) and 1399M2 (F). Green bars in (E) indicate motifs for NFI family members. Duplicate motifs for FOXA1 and motifs for transcription factors without mouse orthologs are not shown but can be found in Table S6. G. Model for metastasis of SCLC-A tumors. When SCLC cells can upregulate NFIB (top), high NFIB activity acts as a switch that promotes tumor growth and metastasis. NFIB upregulation opens the chromatin at numerous loci in the genome, with FOXA1/2 as possible co-factors. When SCLC cells are unable to upregulate NFIB (bottom, for example when the gene is knocked out), tumor progression is slower, and cells that metastasize may rely on transcription factors such as ASCL1 and FOXA1/2, which may work in independent programs.