Comparative functional genomics analysis of NNK tobacco-carcinogen induced lung adenocarcinoma development in Gprc5a-knockout mice

Abstract

Background: Improved understanding of lung cancer development and progression, including insights from studies of animal models, are needed to combat this fatal disease. Previously, we found that mice with a knockout (KO) of G-protein coupled receptor 5A (Gprc5a) develop lung tumors after a long latent period (12 to 24 months).

Methodology/principal findings: To determine whether a tobacco carcinogen will enhance tumorigenesis in this model, we administered 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK) i.p. to 2-months old Gprc5a-KO mice and sacrificed groups (n=5) of mice at 6, 9, 12, and 18 months later. Compared to control Gprc5a-KO mice, NNK-treated mice developed lung tumors at least 6 months earlier, exhibited 2- to 4-fold increased tumor incidence and multiplicity, and showed a dramatic increase in lesion size. A gene expression signature, NNK-ADC, of differentially expressed genes derived by transcriptome analysis of epithelial cell lines from normal lungs of Gprc5a-KO mice and from NNK-induced adenocarcinoma was highly similar to differential expression patterns observed between normal and tumorigenic human lung cells. The NNK-ADC expression signature also separated both mouse and human adenocarcinomas from adjacent normal lung tissues based on publicly available microarray datasets. A key feature of the signature, up-regulation of Ube2c, Mcm2, and Fen1, was validated in mouse normal lung and adenocarcinoma tissues and cells by immunohistochemistry and western blotting, respectively.

Conclusions/significance: Our findings demonstrate that lung tumorigenesis in the Gprc5a-KO mouse model is augmented by NNK and that gene expression changes induced by tobacco carcinogen(s) may be conserved between mouse and human lung epithelial cells. Further experimentation to prove the reliability of the Gprc5a knockout mouse model for the study of tobacco-induced lung carcinogenesis is warranted.

Figure 1. NNK-induced lung carcinogenesis in Gprc5a-knockout mice.

Schematic illustration depicting the experimental plan where 2 months old Gprc5a-knockout mice were treated with 104 mg/kg body weight NNK or control saline intraperitonealy (i.p.) twice one week apart and followed up for lung lesion development at 6, 9, 12 and 18 months following treatment.

Figure 2. Increased incidence and multiplicity of lung tumors in Gprc5a-knockout mice exposed to NNK compared to control mice.

Graphs depicting the percentage of mice with lung tumors (left panels) and number of tumors per mouse as a measure of multiplicity (right panels) in the control (orange line) and NNK-exposed (black line) groups at 6, 9, 12 and 18 months following injection with the carcinogen.

Figure 3. Increased lung tumor burden in NNK-exposed Gprc5a-knockout mice.

A.Gross and macroscopic images representing total lungs excised from two mice each injected i.p with control saline (left) or NNK (right). Arrows indicate gross lung images representing selected H& E stained histological sections of total lungs from both groups and arrowheads indicate presence of a lung tumor (adenomas, 1; adenocarcinomas, 2). B. Lung tumor areas were quantified by Image-J software to assess tumor burden displayed in the graph as percentage occupied of total area.

Figure 4. Lung tumors in NNK-exposed Gprc5a-knockout mice display increased features of poor differentiation.

Representative photomicrographs of tissue histological sections following H&E staining of spontaneous adenomas and adenocarcinomas obtained from control mice (A) and from Gprc5a-knockout mice exposed to NNK (B). Bars, 100 microns.

Figure 5. The mouse NNK-ADC signature exhibits significant modulation of cancer-related gene sets and pathways that are similar in an in vitro model of human lung carcinogenesis.

A. The mouse NNK-ADC signature comprising 4981 gene feature differentially expressed between Gprc5a-knockout MDA-F471 adenocarcinoma cells and Gprc5a^−/− normal lung cells normal cells was derived based on selection criteria described in the Methods. The signature was functionally analyzed using global functional categories from by IPA®. The value of −log(significance) represents the inverse log of the p-values of the modulation of the depicted functional categories (left panel) and pathways (right panel) between the mouse normal and adenocarcinoma cells. The number of genes displaying more than 2 fold change is indicated above each bar. B. Human orthologs of the mouse NNK-ADC and also differentially expressed between the human lung tumorigenic 1170-I and normal NHBE cells (n = 523) were median-centered independently in the mouse and human lung epithelial cells, integrated and then analyzed by hierarchical cluster analysis. High or low gene expression levels are indicated by red or green color, respectively as indicated by the log2 transformed scale bar. Gene clusters are numbered 1 to 4. The clusters of up-regulated and down-regulated genes are highlighted in red and green, respectively. C. PCA analysis in three-dimensional space of the integrated data using metric centered correlation.

Figure 6. Gene-interaction network analysis of genes differentially expressed between both mouse and human normal lung epithelial and adenocarcinoma cells.

IPA® analysis of gene-interaction networks and neighborhoods involving the selected genes, Mcm2 (top) and Fen1 (bottom), differentially expressed between Gprc5a ^−/− normal lung cells and adenocarcinomas as well as between human normal NHBE cells and 1170-I tumorigenic cells (highlighted in inlet of 5B). Interaction nodes of the Mcm2 and Fen1 genes are highlighted by a blue border. Gene expression variation by at least 2-fold is depicted by color (red, up-regulated; green, down-regulated).

Figure 7. Elevated expression of Ube2c, Mcm2, Fen1 and cyclin D1 proteins in Gprc5a-knockout adenocarcinomas and normal lung tissue and cells.

A. Representative photomicrographs depicting increased expression of the protein products of Mcm2, Ube2c, Fen1 and cyclin D1 in NNK-induced adenocarcinomas and normal lung histological tissue specimens obtained from NNK-exposed Gprc5a-knockout mice. All four proteins exhibited mainly nuclear localization of expression. B. The number of cells exhibiting positive reactivity (nuclear staining) was counted in up to three separate microscopic fields per tumor or normal lung in five different NNK-exposed Gprc5a ^−/− mice. P-values were obtained by the students two-sample t-test (*, p<0.001). C. Western blotting analysis of the four antigens in three different culture dishes of each of the Gprc5a ^−/− MDA-F471 adenocarcinoma and normal lung cells. Approximately 10⁶ cells per 10 cm cell culture dish were harvested to prepare total protein extracts. Samples of these extracts (20 µg protein) were analyzed by western blotting using primary antibodies against the indicated proteins. Membrane blots were also analyzed with antibodies against β-actin to compare protein loading in the different lanes.

Figure 8. The mouse NNK-ADC expression signature effectively separates both mouse and human lung adenocarcinomas from their adjacent normal counterparts.

A. Dendogram (left) of hierarchical cluster analysis and PCA (right) of the gene features (n = 1452) of the NNK-ADC signature that are present in the Affymetrix MG-U74Av2 platform and gene expression data of mouse normal lung (n = 15) and adenocarcinoma (n = 29) tissue from the study by Stearman et al . Dendograms (left) of hierarchical cluster analysis and PCA (right) of human datasets including 20 lung adenocarcinomas and 19 adjacent normal lung samples from the study by Stearman et al (B) and 27 lung adenocarcinomas and 27 paired adjacent normal lung from the study by Su et al (C) and using only orthologous genes of the signature and found in the Affymetrix MG-U95Av2 (B, n = 1877) and HG-U133A (C, n = 2475) platforms.