Aberrant basal cell clonal dynamics shape early lung carcinogenesis

Abstract

Preinvasive squamous lung lesions are precursors of lung squamous cell carcinoma (LUSC). The cellular events underlying lesion formation are unknown. Using a carcinogen-induced model of LUSC with no added genetic hits or cell type bias, we found that carcinogen exposure leads to non-neutral competition among basal cells, aberrant clonal expansions, and basal cell mobilization along the airways. Ultimately, preinvasive lesions developed from a few highly mutated clones that dominate most of the bronchial tree. Multisite sequencing in human patients confirmed the presence of clonally related preinvasive lesions across distinct airway regions. Our work identifies a transition in basal cell clonal dynamics, and an associated shift in basal cell fate, as drivers of field cancerization in the lung.

Figure 1. NTCU-induced squamous lung lesions develop from pre-existing basal cells.

(A) NTCU administration protocol for the induction of murine lung squamous cell carcinoma. (B) Immunofluorescence for the basal cell and squamous cell lesion and tumor marker KRT5 and the proliferation marker Ki67 on murine lung sections 18 weeks after NTCU treatment commencement. Tissue overview shows abnormal KRT5 expression in the bronchial tree. Scale bar, 1 mm. (I, II) Low-grade preinvasive lesions are shown. Scale bar, 50 μm. (C) Immunostaining for KRT5 and Ki67 on lung sections from NTCU-treated mice 23 weeks after treatment commencement. Tissue overview shows regions in the bronchial tree expressing KRT5. Scale bar, 1 mm. (III, IV) Invasive tumors filling intraparenchymal spaces are shown. Scale bar, 50 μm. (D) Strategy to track lineage-labeled airway basal cells during NTCU-induced lung carcinogenesis in KRT5-CreER;tdTomato mice. (E) 3D projection of dorsal trachea whole-mount showing expression of tdTomato in the great majority of KRT5⁺ basal cells following tamoxifen treatment. Arrow points to non-labeled KRT5⁺ cell. Scale bar, 100 μm. (F) Left lung lobe whole-mounts showing lineage-traced tdTomato⁺ cells in control lungs. Scale bar, 2 mm. (G) 3D projection of a lung whole-mount from a NTCU-treated mouse. Lineage-labeled tdTomato⁺ cells co-expressing KRT5 are detected throughout the bronchial tree. Scale bars, 2 mm (overview) and 500 μm (magnified region). (H) Immunostaining for tdTomato and KRT5 on bronchial section collected 18 weeks after NTCU start. Histology is indicative of squamous dysplasia, with partially disorganized layers of epithelial cells. Scale bar, 100 μm. (I) Bronchial tissue sections from control and NTCU-treated KRT5-CreER;tdTomato mice stained with hematoxylin and eosin (H&E, top), or processed for tdTomato immunohistochemistry (bottom). Differentiated luminal cells are seen in the control bronchial epithelium, and no lineage-labeled cells are detected. Following NTCU administration, tdTomato⁺ dysplastic lesions and loss of luminal cells are observed. Scale bars, 50 μm.

Figure 2. Carcinogen exposure induces non-neutral basal cell clonal expansions.

(A) Strategy to track basal-cell derived clones during NTCU-induced lung carcinogenesis using mosaic cell labeling in KRT5-CreER;tdTomato mice. (B) 3D projections of dorsal trachea whole-mounts from control and NTCU-treated mice at 4 days and 24 weeks post-tamoxifen. The boxed regions highlight the dorsal smooth muscle, running longitudinally between the open cartilage rings, whose dorsal ends can be seen at the lateral edges of the preparation. Scale bar, 500 μm. (C) Images of the dorsal tracheal epithelium in the regions indicated in dashed boxes in (B). EdU staining was performed on 24-week samples. Scale bar, 200 μm. (D) Schematics of the biophysical models describing the dynamic of the basal cell layer in the control (top) and NTCU-treated trachea (bottom). (E) Cumulative probability of observing voids larger than a given size in the control trachea 4 days post-tamoxifen. The black reference line corresponds to a power-law decay with exponent -1. (F) Cumulative probability of observing voids larger than a given size in the control trachea 24 weeks post-tamoxifen. Data from 4 individuals in shown (black and grey lines). The blue ‘theory’ line corresponds to the average with shaded standard deviation obtained from 200 numerical simulations of a neutral competition model. The black reference line corresponds to a power-law decay with exponent -1. (G) Cumulative probability of observing voids larger than a given size in the trachea of NTCU-treated mice, 24 weeks post-tamoxifen. Data from 4 NTCU-treated individuals is shown (black and grey lines). The blue ‘theory’ line corresponds to the average with shaded standard deviation obtained from 200 numerical simulations of non-neutral model (see Supplementary Text).

Figure 3. Single cell profiling reveals an epithelial cell fate shift following NTCU treatment.

(A) Experimental overview of single-cell RNA (scRNA-seq) profiling of the tracheal epithelium of NTCU-treated and age-matched control mice. (B) UMAP visualizations colored according to condition (left) and cell type (right) for all mice, depicting 29,088 cells. (C) Dotplot depicting the expression of selected marker genes for cell types shown in B. (D) Barplot showing changes in relative abundance of tracheal epithelial cell types in the NTCU-treated and control groups, 15 weeks after treatment commencement. (E) Barplot showing log 2-fold changes (log₂FC) in abundance of each cell type between NTCU-treated and control mice calculated using scCODA. Statistical significance was determined by a false discovery rate (FDR) < 0.05 (Benjamini–Hochberg adjusted). (F) Trajectory analysis of basal, Krt4/Krt13⁺, and secretory cell clusters identified a bifurcated structure, with one major branching point and three cell major cell states. Pseudotime progression is shown in the center. The origin (cell state 1) is enriched in proliferative and basal Krt14⁺ cells. The two branches diverge into different cell fates: one dominated by Krt4/Krt13⁺ and secretory cells (cell state 2), and the other by basal and basal Tgm2⁺ cells (cell state 3). (G) Tree structure of the trajectory shown in F, visualizing the distribution and relative abundance of cell types over each branch of the tree structure in the control and NTCU-treated groups. Trajectories were reconstructed in four dimensions but are rendered in two dimensions. (H) Immunofluorescence for KRT13 and the secretory cell marker SCGB1A1 on trachea whole-mounts from control and NTCU-treated mice, 18 weeks after NTCU treatment commencement. A single longitudinal cut was done along the ventral tracheal wall to expose the entire epithelial surface. Scale bar, 500 μm. (I) Quantitative assessment of the tracheal epithelial surface expressing KRT13 (top) and the secretory marker SCGB1A1 (bottom) in control and NTCU-treated mice, 18 weeks after NTCU treatment commencement. Bars depict mean ± standard error of the mean (SEM). Each dot represents a different individual; p values are indicated (unpaired two-tailed t-tests with Welch’s correction).

Figure 4. Epithelial cell shift during early human lung squamous cell carcinogenesis.

(A) UMAP visualization of epithelial cells in the human trachea. Clusters are colored according to cell type. (B) Dotplot displaying expression of selected markers for identified human tracheal epithelial cell populations. SMG, submucosal gland. (C) Neighborhood graph displaying outcome of differential cell abundance test with MiloR. Neighborhoods (nodes) are colored according to log fold changes between smoking status. (D) Beeswarm plot showing differences in tracheal epithelial cell abundance in log fold change between non- and current-smokers. Neighborhoods with differential cell abundance at FDR < 0.1 are colored in blue or red, if enriched in non-smokers or current-smokers, respectively. (E) Cell lineage inference for basal, suprabasal, and secretory cell populations from the surface airway epithelium using Slingshot. Principal curves are depicted on the UMAP to the left. The tree on right shows the cell populations in each lineage. Note that KRT4/KRT13⁺ cells are identified as a transitional cell state. (F) Pseudotime distribution of the three cell lineages identified in the surface airway epithelium, displaying differences between non- and current-smokers. (G) Plot depicting mean lineage weights. Weights assignments denote the probability of each cell belonging to a particular lineage. Cell fate choice between lineage 2 and 3 varies between non- and current-smokers. (H) KRT4/KRT13 signature score in normal airway epithelium, increasing grades of lung preinvasive squamous cell lesions and LUSC. Median, upper, and lower quartiles are shown. Individual samples are represented as dots; p values are displayed (Wilcoxon test).

Figure 5. Mutagenic effect of NTCU on basal cells in the airway.

(A) Burden of single nucleotide variants (SNVs) and single base substitutions (SBSs) signatures, across clones detected in both NTCU-treated mice. Stacked bar plots showing the proportional contribution of each mutational signature to the SNVs, with purple highlighting the NTCU signature. The grey line highlights the average mutation burden across clones. (B) Trinucleotide context spectrum of the NTCU signature. (C) Phylogenetic tree for samples and clones located on the left lung of mouse MD7047. Clones are highlighted with individual numbers, and mutations colored according to the respective mutational signature contributing to each branch. The boxed numbers represent progenitors of the clones branching of the respective box. Where boxes overlap, the clone number is displayed above the box. Selected mutations in driver genes are annotated on some branches including the amino acid change.

Figure 6. NTCU-driven clonal expansions in the lung.

(A) 3D projections of whole-mount lungs collected at different time points from NTCU treatment commencement. KRT5 immunostaining was used to visualize basal cells and preinvasive lesions. Scale bar, 2 mm. (B) Integrative visualization of the location of a selected clone and respective microbiopsies in the trachea and lung of mouse MD7047. All microbiopsies from the trachea and the right lung containing the magenta clone (lineage) are shown as grey circles within the histological images. The trachea is seen in the tissue section on the left side of the image; the bronchial tree is displayed in the section to the right. The phylogenetic tree depicted on the right-hand side is scaled according to the number of mutations per clone. The magenta ancestor and all subclones related to this clade are highlighted. The small tree schematic surrounding the histological image is equivalent in structure, but not scaled to the mutation burden of each clone. Each dot on the small tree represents a clone and branching point within the phylogeny. (C) Equivalent to (B) but for all samples from the trachea and left lung of mouse MD7047. All samples containing the green clone (lineage) are depicted. A complete interactive visualization can be found in the supplementary data (Data S9-S10).

Figure 7. Clonal relatedness between anatomically distinct human preinvasive lung lesions.

(A) Summary of preinvasive samples and patient characteristics used for assessment of clonality. Schematic shows the anatomical location of biopsy sites. MoD, moderate dysplasia; CIS, carcinoma in situ; RUL, right upper lobe; RMB, right main bronchus; RIB, right intermediate bronchus; RLL, right lower lobe; LUL, left upper lobe; LMB, left main bronchus; LLL, left lower lobe. (B) Phylogenetic tree based on somatic mutations illustrating the clonal relationships and evolutionary history of two indolent lesions present in an ex-smoker patient. (C) Phylogenetic tree depicting clonal relationships between indolent (R1 & R3) and progressive (R2) lesions in a current smoker. In B-C shared clusters across two or more anatomical sites are colored in blue, while unique and site-specific clusters are colored in purple, orange, or green. Clonal relationships between regions and seeding clones are shown.