SOX2 Drives Bronchial Dysplasia in a Novel Organotypic Model of Early Human Squamous Lung Cancer

Abstract

Rationale: Improving the early detection and chemoprevention of lung cancer are key to improving outcomes. The pathobiology of early squamous lung cancer is poorly understood. We have shown that amplification of sex-determining region Y-box 2 (SOX2) is an early and consistent event in the pathogenesis of this disease, but its functional oncogenic potential remains uncertain. We tested the impact of deregulated SOX2 expression in a novel organotypic system that recreates the molecular and microenvironmental context in which squamous carcinogenesis occurs.

Objectives: (1) To develop an in vitro model of bronchial dysplasia that recapitulates key molecular and phenotypic characteristics of the human disease; (2) to test the hypothesis that SOX2 deregulation is a key early event in the pathogenesis of bronchial dysplasia; and (3) to use the model for studies on pathogenesis and chemoprevention.

Methods: We engineered the inducible activation of oncogenes in immortalized bronchial epithelial cells. We used three-dimensional tissue culture to build an organotypic model of bronchial dysplasia.

Measurements and main results: We recapitulated human bronchial dysplasia in vitro. SOX2 deregulation drives dysplasia, and loss of tumor promoter 53 is a cooperating genetic event that potentiates the dysplastic phenotype. Deregulated SOX2 alters critical genes implicated in hallmarks of cancer progression. Targeted inhibition of AKT prevents the initiation of the dysplastic phenotype.

Conclusions: In the appropriate genetic and microenvironmental context, acute deregulation of SOX2 drives bronchial dysplasia. This confirms its oncogenic potential in human cells and affords novel insights into the impact of SOX2 deregulation. This model can be used to test therapeutic agents aimed at chemoprevention.

Keywords: bronchial dysplasia; early lung cancer; organotypic culture; sex-determining region Y-box 2 (SOX2); squamous lung cancer.

Figure 1.

(A) KT cells (human bronchial epithelial cells immortalized with hTERT and CDK4) were sequentially transduced with lentiviral constructs to create a series of cell lines with different genotypes. (B) A mean of 49% of cells were positive for sex-determining region Y-box 2 (SOX2) on induction with doxycycline. (C) Western blot with a range of the cell lines used with different genotypes. Scale bars = 100 μm. The expression of SOX2 was tightly controlled and only induced in the presence of M2 reverse tetracycline transactivator (M2rtTA) and doxycycline. Cells were positive for keratin 5 and p63 in the presence and absence of doxycycline. CDK4 = cyclin-dependent kinase 4; DAPI = 4′,6-diamidino-2-phenylindole; Dox = doxycycline; hTERT = human telomerase reverse transcriptase; iSOX2 = inducible SOX2; KRT5 = keratin 5; p53lo = TP53 low; shp53 = short hairpin RNA against TP53; TP53 = tumor promoter 53.

Figure 2.

(A) Human bronchial epithelial cells immortalized with human telomerase reverse transcriptase and cyclin-dependent kinase 4, known as KT cells, were established in short-term culture on top of a disc consisting of Type I collagen and pulmonary fibroblasts. (B) Supplementation of medium in the lower chamber with doxycycline induces the expression of red fluorescence in A549 cells transduced with an inducible red fluorescent protein reporter gene that are in established confluent culture at the air–liquid interface (ALI). Scale bars = 400 μm. Transduced human bronchial epithelial cells were established in the organotypic system described in A; (C) doxycycline supplementation in the lower chamber led to induction of sex-determining region Y-box 2 (SOX2) expression in the monolayer. (D–H) Genetically manipulated KT inducible SOX2 (iSOX2) and tumor promoter 53 knockdown (KTiSOX2p53lo) cells retain the capacity to differentiate into constituent cells of the human airway epithelium. Scale bars = 50 μm. ALI Transwell cultures using differentiation medium at Day 23 show a mixed population of cells with clear differentiation into ciliated cells (D and E) with acetylated tubulin representing mature cilia in red; (F) mucin-secreting (goblet) cells (MUC5AC, green; γ-tubulin/cilia basal bodies, red); (G) Club cells (CC10, green; acetylated tubulin, red); and (H) basal cells (KRT5, green; acetylated tubulin, red). (E–H) Scale bars = 50 μm. dox = doxycycline; CC10 = club cell 10; KRT5 = keratin 5; MUC5AC = mucin 5AC; TUB = tubulin.

Figure 3.

(A) A cell population growth assay was performed on KT cells (human bronchial epithelial cells immortalized with human telomerase reverse transcriptase and cyclin-dependent kinase 4) in the presence and absence of doxycycline. Induction of sex-determining region Y-box 2 (SOX2) expression did not lead to an increase in the number of cells over 120 hours. Induction of SOX2 expression did not alter cell cycle distribution in standard culture conditions but led to a significant increase in the proportion of cells in S-G2/M in the organotypic culture (OTC; B, paired Student’s t test, two tailed) and in total epithelial cell number (C, paired Student’s t test, two tailed). Data presented in A represents means (±SD) of three separate experiments in triplicate. (D and E) Control cell lines (KT-M2 reverse tetracycline transactivator [M2rtTA] [KTM2]) were treated with doxycycline with no significant alteration in morphology on phase-contrast or histology. (H) Loss of p53 led to modest focal outgrowths in the monolayer. Induction of SOX2 (inducible SOX2 [iSOX2]) led to an obvious phenotype with focal (KTiSOX2; F and G) and more diffuse (KTiSOX2 tumor promoter 53 [TP53] knockdown [KTiSOX2p53lo]) changes (H and I) in the epithelial monolayer. (J–O) Histological analysis was consistent with the changes seen using phase-contrast microscopy. There were diffuse outgrowths demonstrated in the KTiSOX2p53lo upon induction of SOX2 expression in the confluent monolayer. The histological features were typical of high-grade preinvasive lesions: loss of contact inhibition, a thickened epithelial layer with dysplasia, absence of maturation, and a high nuclear-to-cytoplasmic ratio. (D–F) Scale bars = 400 μm; (G–I) scale bars = 1,000 μm; (J–O) scale bars = 100 μm. Dox = doxycycline; NS = not significant. *P ≤ 0.05.

Figure 4.

Activation of sex-determining region Y-box 2 (SOX2) leads to focal dysplastic changes and alterations in key cell signaling cascades. (A and B) Dysplastic lesions in doxycycline-treated cultures were strongly positive for SOX2 with no SOX2 staining in the untreated cultures. (C–J) Serial sections from dysplastic lesions show a down-regulation in phospho–extracellular signal–regulated kinase (pERK) and an up-regulation in phosphoAKT (pAKT) in those cells that are SOX2 positive and in cycle (Ki67 positive). Scale bars = 100 μm. DAPI = 4′,6-diamidino-2-phenylindole; dox = doxycycline; H+E = hematoxylin and eosin stain; Ki67 = marker of proliferation Ki67; KT = human bronchial epithelial cells immortalized with human telomerase reverse transcriptase and cyclin-dependent kinase 4; KTiSOX2p53lo = KT inducible SOX2 and tumor promoter 53 (TP53) knockdown.

Figure 5.

Representative images from a tissue microarray with high-grade preinvasive bronchial lesions. (A and B) Basal cell hyperplasia has weakly positive staining for sex-determining region Y-box 2 (SOX2) and rare Ki67 (marker of proliferation Ki67)-positive cells. (C–F) High-grade dysplastic lesions exhibit much more marked SOX2 and Ki67 staining that extends throughout the epithelial layer. PhosphoAKT (pAKT) staining proved challenging on archived tissue microarray specimens; however, (G–I) there was a single sample in which the border between normal, ciliated, pseudostratified epithelium, and high-grade dysplasia was available that provided a robust internal control for the antibody performance. There was a clear up-regulation in pAKT staining in those dysplastic cells in which SOX2 and Ki67 were up-regulated. Scale bars = 100 μm.

Figure 6.

Deregulated sex-determining region Y-box 2 (SOX2) targets key cancer-related genes in the organotypic culture system. Chromatin immunoprecipitation–polymerase chain reaction (PCR) was performed using input cross-linked DNA and the same DNA incubated with both a SOX2-specific antibody and an IgG control. There was enrichment for the B-cell lymphoma 2 (BCL2) and cyclinD1 (CCND1) promoters in the SOX2 pulldown treated with doxycycline (Dox), but not the untreated (No Dox) nor the IgG control, confirming that SOX2 binds both gene promoters and regulates their transcription in this system (A). (B and C) Quantitative PCR (qPCR) experiments confirmed that BCL2 (B) and CCND1 (C) were both up-regulated transcriptionally on induction of SOX2. The expression of a series of cell cycle genes was analyzed. (C) There was up-regulation of key cyclin-dependent kinases (CCNA2, CCNB1, CCNB2, and CCNE2), and down-regulation of cyclin-dependent kinase inhibitor 1 (CDKN1A or p21), a critical inhibitor of cell cycle progression. These alterations were seen on treatment with doxycycline compared with the untreated control. The following kinases were also tested but were not significantly altered: CCNA1, CCNE1, CCND2, and CCND3. The results shown are representative of three independent experiments, and data are presented as the mean (n = 3) (±SD). For the qPCR experiments, ΔΔ threshold cycle values were log (base 10) transformed, and results for each marker were compared with glyceraldehyde phosphate dehydrogenase. Statistical analysis: a Student’s t test (unpaired) was performed to test significance: *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; and ****P ≤ 0.0001. Ab = antibody.

Figure 7.

The most significantly differentially expressed genes (P < 1 × 10⁻⁶, n = 832) were analyzed using ingenuity pathway analysis. (A) The most significantly altered molecular and cellular functions were cellular movement and cellular proliferation. (B) A heatmap of 197 genes associated with cell migration illustrating up- and down-regulation of genes with induction of sex-determining region Y-box 2 (SOX2) is presented. The results show three paired replicates with (italics) and without (nonitalic font) SOX2 induction. A number of genes are highlighted, including reported direct targets of SOX2 (ETS translocation variant 4 [ETV4] and dickkopf-related protein 1 [DKK1]) and those directly implicated in epithelial–mesenchymal transition (EMT; CDH2, CEMIP). Further details are in Table E1. (C) Quantitative polymerase chain reaction (qPCR) of canonical mediators of EMT was performed to corroborate RNA-Seq results and confirmed their up-regulation by SOX2. In addition, a number of the most significantly altered genes in differential gene expression analyses (CEMIP, SERPINI1, and AGR2; all P < 1 × 10⁻¹³) were recently implicated in EMT in the literature but not annotated in the “cellular movement” group. (D) qPCR experiments confirmed significant up-regulation in each case. (E) Interrogation of The Cancer Genome Atlas database showed that SERPINI1 was the most commonly altered of these in early squamous lung cancer and commonly altered with SOX2. The corresponding protein, SERPINI1, was also overexpressed on SOX2 induction in the organotypic culture (F), and chromatin immunoprecipitation–PCR confirmed enrichment for SOX2 binding near the SERPINI1 transcriptional start site (G). (H) Further sections from the tissue microarray were stained for SERPINI1 and showed clear cytoplasmic staining in biopsies of high-grade dysplasia (upper panels) compared with normal epithelium. This is particularly reinforced by the section containing the transition from relatively normal pseudostratified epithelium to disorganized high-grade dysplasia (lower panel). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ****P < 0.0001. Ab = antibody; AGR2 = anterior gradient protein 2 homolog; CDH2 = cadherin 2; CEMIP = cell migration inducing hyaluronan binding protein; CPM = counts per million; Dox = doxycycline; LEF1 = lymphoid enhancer binding factor 1; MMP10 = matrix metallopeptidase 10; ns = not significant; SERPINI1 = serpin family I member 1; SNAI = snail family transcriptional repressor; TWIST = twist family basic helix-loop-helix transcription factor 1.

Figure 8.

Representative phase-contrast (A and B) and hematoxylin and eosin stains (C and D) of doxycycline-treated (A and C) and doxycycline (Dox) and AZD5363–treated (B and D) organotypic cultures. All test and control cultures were treated with dimethyl sulfoxide. As before, induction of sex-determining region Y-box 2 (SOX2) led to a diffuse dysplastic phenotype. (B and D) Treatment with a pan-AKT inhibitor (AZD5363, 10 μM) abrogated the development of the dysplastic lesions. Cells in the treated culture were alive and continued to express SOX2 (E) and Ki67 (F). Images are representative of three independent experiments. (A and B) Scale bars = 1,000 μm; (C–F) scale bars = 100 μm. KT = human bronchial epithelial cells immortalized with human telomerase reverse transcriptase and cyclin-dependent kinase 4; KTiSOX2 p53lo = KT inducible SOX2 and tumor promoter 53 (TP53) knockdown.