Gene Expression Studies Demonstrate that the K- ras /Erk MAP Kinase Signal Transduction Pathway and Other Novel Pathways Contribute to the Pathogenesis of Cumene-induced Lung Tumors

Abstract

National Toxicology Program (NTP) inhalation studies demonstrated that cumene significantly increased the incidence of alveolar/bronchiolar adenomas and carcinomas in B6C3F1 mice. Cumene or isopropylbenzene is a component of crude oil used primarily in the production of phenol and acetone. The authors performed global gene expression analysis to distinguish patterns of gene regulation between cumene-induced tumors and normal lung tissue and to look for patterns based on the presence or absence of K-ras and p53 mutations in the tumors. Principal component analysis segregated the carcinomas into groups with and without K-ras mutations, but failed to separate the tumors based on p53 mutation status. Expression of genes associated with the Erk MAP kinase signaling pathway was significantly altered in carcinomas with K-ras mutations compared to tumors without K-ras mutations or normal lung. Gene expression analysis also suggested that cumene-induced carcinomas with K-ras mutations have greater malignant potential than those without mutations. In addition, significance analysis of function and expression (SAFE) demonstrated expression changes of genes regulated by histone modification in carcinomas with K-ras mutations. The gene expression analysis suggested the formation of alveolar/bronchiolar carcinomas in cumene-exposed mice typically involves mutation of K-ras, which results in increased Erk MAP kinase signaling and modification of histones.

Introduction

A companion paper demonstrated that cumene-induced lung tumors have mutations in both K-ras and p53. K-ras mutations were detected in 87% (45/52) of the cumene-induced lung neoplasms. Mutations in p53 were detected in 52% (27/52) of the tumors and increased p53 protein expression was detected in 56% (29/52) of the tumors (Hong et al. 2008). In the current study, we further evaluated cumene-induced lung carcinomas for additional factors that contribute to the carcinogenic properties of the chemical, including the potential roles of K-ras and p53 gene mutation.

Recent microarray experiments have enabled the analysis of global gene expression changes in both human (Bhattacharjee et al. 2001; Garber et al. 2001) and mouse (Bonner et al. 2004; Sweet-Cordero et al. 2005) lung tumors. Bonner et al. (2004) examined chemically induced mouse lung adenocarcinomas for global gene expression changes and found similarities with human lung tumors. A study by Sweet-Cordero et al. (2005) used a K-ras gene targeted approach to induce mouse lung tumors with oncogenic K-ras. They identified global gene expression changes in mouse lung tumors that could distinguish human adenocarcinomas with and without K-ras mutations, which was not possible using expression data from the human study alone.

In addition to using DNA microarray technology to further explore K-ras signal transduction in cumene-induced lung neoplasms, we also evaluated the potential involvement of epi-genetic mechanisms. It has been suggested that epigenetic alterations in gene expression are a more frequent mechanism of gene inactivation than mutation (Crews and McLachlan 2006; Stebbing et al. 2006). The potential involvement of epi-genetics was investigated using significance analysis of function and expression (SAFE).

The epigenetic regulation of gene transcription is controlled in part through the regulation of chromatin structure, most commonly by the posttranslational modification of histone tails. The most intensively studied histone modification is acetylation. The acetylation of conserved lysine residues in the histone tail neutralizes the positive charge of the histone and decreases its affinity for neighboring histone tails, regulatory proteins, and DNA. The decreased affinity promotes unfolding and facilitates gene transcription. Acetylation is carried out by histone acetyl-transferases (HATs), while the opposite function is carried out by a group of enzymes known as histone deacetylases (HDACs) (Eberharter and Becker 2002; Kristeleit et al. 2004).

The goal of our study was to use global gene expression analysis to distinguish patterns of gene regulation between lung tumors induced by cumene and control lung tissue. In addition, we wanted to establish a pattern of gene expression based on the presence or absence of K-ras and p53 mutations and examine the potential contribution of tumor suppressor genes and genes involved in malignancy and metastatic potential. The analysis of gene expression will add to our understanding of the molecular mechanisms of lung cancer, identify pathways that are important in tumor formation, and identify differences on the basis of gene mutation.

Materials and Methods

Selection of Samples

In the NTP study, six-week-old male and female B6C3F1 mice were exposed to 0, 125 (females only), 250, 500, or 1,000 (males only) ppm cumene by whole-body inhalation, six hours per day, five days per week, for two years (NTP 2007). Treatment of B6C3F1 mice with cumene significantly increased the incidence of lung tumors, specifically, alveolar bronchiolar adenomas and carcinomas that ranged in size from 3 to 10 mm in diameter. In NTP studies, normal and tumor tissues are collected at necropsy to ensure representative samples are available for adequate histopathology evaluation. Remaining tissues are frozen for molecular biology studies. A portion of each lung tumor was fixed in 10% neutral buffered formalin, embedded in paraffin, and cut into 5-μm sections for hematoxylin and eosin staining and routine histological examination. The remaining portion of each lung tumor was frozen in liquid nitrogen and stored at −80°C until subsequent RNA isolation. The criteria for selecting frozen tissues for the microarray study included absence of inflammatory cell infiltration and lack of necrosis in lung tumors from cumene-treated mice and in normal lung tissues from untreated mice. Based on the criteria, eight of twenty-three frozen cumene-induced tumors and four normal lung tissues from untreated mice were selected for the gene expression analysis.

K-ras and p53 Mutations

DNA isolated from the eight carcinomas selected for microarray analysis was evaluated for K-ras mutations in exons 1 and 2 (codons 12, 13, and 61) and p53 mutations in exons 5, 6, 7, and 8. Five unstained, 10-μm thick serial sections were used for DNA isolation and subsequent amplification by the polymerase chain reaction (PCR). Previously described primers were used for nested PCR of the K-ras gene and the p53 gene (Lambertini et al. 2005; Sills et al. 1995). Positive and negative DNA controls for mutations were run with all sets of reactions. PCR products were purified using a QIAquick Gel Extraction Kit (Qiagen, Valencia, CA). The purified samples were sequenced using a cycle sequencing kit (US Biochemical, Cleveland, OH), which incorporated [α-³³P]-dideoxynucleotide triphosphate (ddNTP) terminators (A, C, G, T) into the sequencing products. Detected mutations were confirmed by repeat analysis, starting from amplification of the original DNA extract.

Microarray Analysis

Affymetrix microarray technology was used to examine changes in gene expression in cumene-induced lung tumors. RNA was collected from the tissues by digestion with Trizol (Invitrogen, Carisbad, CA) and subsequent extraction with chloroform and isoamyl alcohol. RNA was further purified using the RNeasy Mini Kit (Qiagen, Valencia, CA). Quality of RNA was assessed by gel electrophoresis.

Gene expression analysis was conducted using Affymetrix Mouse Genome 430 2.0 GeneChip arrays (Mouse430 v2, Affymetrix, Santa Clara, CA). Total RNA (1 μg) was amplified using the Affymetrix One-Cycle cDNA Synthesis protocol. For each array, 15 μg of amplified biotin-cRNAs were fragmented and hybridized to the array for sixteen hours at 45ºC in a rotating hybridization oven using the Affymetrix Eukaryotic Target Hybridization Controls and protocol. Slides were stained with steptavidin/phycoerythrin using a double-antibody staining procedure and washed using the EukGE-WS2v5 protocol of the Affymetrix Fluidics Station FS450 for antibody amplification. Arrays were scanned with an Affymetrix Scanner 3000 and data obtained using the GeneChip Operating Software (GCOS; version 1.2.0.037). The resulting files (.dat, .cel, and .chp) were imported into the Rosetta Resolver system (version 6.0), which performed data preprocessing, normalization, and error modeling (Weng et al. 2006).

Processed microarray data were analyzed with a variety of unsupervised and supervised techniques. Principal component analysis (PCA) was performed on all samples and all probe sets to characterize the variability present in the data. In order to identify differentially expressed genes, a log base 10 error-weighted analysis of variance (ANOVA) using the Benjamini Hochberg false discovery rate was performed using Rosetta Resolver (www.rosettabio.com). Error weighting in Resolver takes advantage of the technology-specific error model that Affymetrix has generated for use with its arrays. This error model takes into account factors contributing to measurement error (sample prep, labeling, chip quality variation, etc.). Error weighting helps to increase statistical power despite low numbers of replicates. In addition to applying error weighting, we have used a stringent multiple test correction (Bonferroni) to reduce false positives. The ANOVA was used to compare the normal lung tissue, tumors with K-ras mutations, and tumors without K-ras mutations and to compare normal lung tissue with tumors with and without p53 mutations.

SAFE (Barry, Nobel, and Wright 2005) was used to test for gene sets demonstrating different activity between classes. SAFE is a permutation-based method for testing functional categories in gene expression experiments. SAFE analysis interprets data within the context of a larger set of functionally related genes, instead of interpreting genes in isolation. This analysis has the ability to detect changes in the expression of a set of genes that may otherwise be missed when considering the expression patterns of individual genes in isolation. SAFE computes a statistic for each probe, termed the local statistic. The local statistics are placed into sets based on biological associations given in a database. A global statistic is calculated for each set and compared to permuted data to assess significance. Default local (t-test) and global (Wilcoxon) statistics were used in this analysis. The database of gene sets was formed by combining the L2L database with Gene Ontology sets and the pathway sets in v1 of Molecular Signature Database (MSigDB) (the current version of MSigDB includes L2L). SAFE was used to compare tumor samples to normal samples and tumors with K-ras mutations against those without K-ras mutations. The Cluster and Resolver were used for visualization of microarray features.

Real-time PCR

Real-time PCR was performed with five genes (Clusterin, Map2k1, Dusp4, Ets1, and Akap12) that are known to play important roles in tumorigenesis and are significantly altered in at least two tumors by microarray analysis. Quantitative gene expression levels were detected using real-time PCR with the ABI PRISM 7900HT Sequence Detection System (Applied Biosystems, Foster City, CA) and TaqMan MGB probes (FAM dye labeled). Primers and probes for all genes analyzed were purchased from Applied Biosystems Assays-on-Demand Gene Expression products (Clusterin, assay ID no. Mm00442773_ml; Map2k1, mitogen-activated protein kinase1, assay ID no. Mm00435940_ml; Dusp4, dual specificity phosphatase 4, assay ID no. Mm00723761_ml; Ets1, E26 avian leukemia oncogene, assay ID no. Mm00468970_ml; Akap12, A kinase (PRKA) anchor protein (gravin) 12, assay ID no. Mm00513511_m1). For amplification, diluted cDNA was combined with a reaction mixture containing TaqMan^® universal PCR Master Mix (Applied Biosystems, catalog no. 4304437) according to manufacture’s instructions. Samples were analyzed in duplicate, and a sample without reverse transcriptase was included in each plate to detect contamination by genomic DNA. Amplification was carried out as follows: (1) 50°C for two minutes (for uracil-N-glycosylase incubation), (2) 95°C for ten minutes (denaturation), (3) 95°C for fifteen seconds and 60°C for thirty seconds (denaturation and amplification) for forty cycles. Fold increases or decreases in gene expression were determined by quantitation of cDNA from tumor samples relative to a pool of normal lung samples. The 18S RNA gene was used as the endogenous control for normalization of initial RNA levels. To determine this normalized value, 2^−(ΔΔCt) values were compared between tumors and normal lungs, where the changes in crossing threshold (ΔCt) = Ct_{Target gene} − Ct_{18S RNA}, and ΔΔCt = ΔCt_normal − ΔCt_tumor.

Immunohistochemistry for Map2k1

Immunohistochemical staining for Map2k1, which is one of ERK/MAPK pathway genes, was performed on selected for-malin-fixed paraffin-embedded alveolar/bronchiolar carcinomas and normal lung tissue sections. Reduction of nonspecific staining was achieved with Vectastain Elite ABC kit (Rabbit IgG, PK-6101; Vector Laboratories, Burlingame, CA). After deparaffinization, tissue sections were incubated with prediluted normal goat serum for twenty minutes at room temperature, followed by Vector Avidin/Biotin Block. Sections were incubated with the primary antibody, MEK-1 (sc-219; Santa Cruz Biotechnology, Santa Cruz, CA), at 1:1,000 dilution for sixty minutes at room temperature. Normal rabbit serum (Jackson ImmunoResearch Laboratories, West Grove, PA) at a 1:1,000 dilution was applied to the negative slides for an equivalent incubation period. After washing, sections were incubated with secondary antibody from the Vectastain Elite ABC Kit (Rabbit IgG, PK-6101; Vector Laboratories, Burlingame, CA), for thirty minutes at room temperature. Immunoreactivity was detected with 3′-diaminonbenzidine tetrahydrochloride (DakoCytomation Liquid DAB Substrate Chromogen System; DakoCytomation, Carpinteria, CA). Sections were counter-stained with hematoxylin and cover slipped.

Results

K-ras and p53 Mutations

Eight alveolar/bronchiolar carcinomas were chosen from twenty-three cumene-treated mice and used for microarray analysis. The carcinomas were chosen based on the absence of necrosis and inflammatory cell infiltration. Six of the selected tumors had K-ras mutations and four had p53 mutations (Table 1). Mutations were not detected in control samples.

Microarray Analysis

Principal component analysis (PCA)

Using genome-wide Affymetrix microarrays, changes in gene expression were monitored in cumene-induced lung tumors and in normal lung. PCA of the gene expression changes separated the samples into three groups: control lung tissue, tumors with K-ras mutations, and tumors without K-ras mutations (Figure 1A). Although there were only two tumors without K-ras mutations in our microarray study, PCA was able to provide evidence of a differential gene expression profile based on K-ras mutation status. PCA failed to separate the tumors into groups based on p53 mutation status (Figure 1B). Therefore, we focused our analysis on the gene expression profiles of tumors with and without K-ras mutations.

ANOVA and post hoc Tukey-Kramer

ANOVA was used to test for differential expression across the three groups (normal lung, tumors with K-ras mutations, and tumors without K-ras mutations). At p < .01, this analysis yielded 1,606 probe sets and was followed by a post hoc analysis (Tukey-Kramer, p < .01). The results of this analysis are summarized in Figure 2. ANOVA and post hoc analysis (Figure 2A and Supplemental Table 1; see http://tpx.sagepub.com/supplemental/ for supplemental tables) identified 1,061 probe sets (627 known Entrez gene IDs) that were differentially expressed between normal lung and tumors with K-ras mutations, while only 429 probe sets (281 genes) differed between normal lung and tumors without K-ras mutations. Additionally, there were 278 differentially expressed probe sets (181 genes) between tumors with and without K-ras mutations. An interesting and potentially important group of genes is the overlapping set of 105 probe sets (66 genes; Figure 2B) that were different between the tumors with K-ras mutations and both of the other groups (normal lung and tumors without K-ras mutations).

Analysis of Specific Pathways and Genes

Our analysis identified 1,061 probe sets that were differentially expressed between normal lung and tumors with a mutation in K-ras. These genes are listed in Supplemental Table 1 and were examined for consistent changes in gene expression. We decided to focus our study on the ras signal transduction pathway, specifically, the extracellular signal-regulated kinase mitogen activated pathway (Erk MAPK). Our initial clustering analyses identified genes associated with the Erk MAPK pathway as having significant expression changes between the normal and tumor tissue. In addition, cluster analysis identified a significant expression change in some of these genes between tumors with and without K-ras mutations.

1. Erk MAPK pathway genes

Mutations in exon 1 or 2 of K-ras were detected in 75% (6/8) of the cumene-induced lung tumors. The predominant mutations identified were codon 12 G to T transversions and codon 61 A to G transitions (Table 1). Point mutations in the K-ras gene result in constitutive activation and promotion of cellular transformation.

Many of the significantly altered genes in cumene-induced lung tumors were associated with the MAP kinase signaling pathway (Figure 3). A number of these genes were altered regardless of K-ras mutation status. The anti-apoptotic gene Clu was increased in both tumor types. Genes that were down-regulated regardless of K-ras mutation status included negative regulators of MAPK signaling (Reck, Dusp1, Dusp4, Cav1, and Loxl1). (H. C. Chang, Liu, and Hung 2004; Owens and Keyse 2007; Sasai et al. 2007; Wu et al. 2007).

The majority of the genes associated with the MAPK pathway were significantly altered only in the tumors with K-ras mutations. The induced genes included those known to promote MAP kinase activation (Mif, Avpi, Map2K1, Ereg, Mapbpip, and Klf5) (Lue et al. 2006; Nicod et al. 2002; Shaul and Seger 2007; Takahashi et al. 2003; Teis et al. 2006; Yang et al. 2007) and genes that are activated by MAP kinase signaling (Ccnd1, Ptges, and Areg) (Lavoie et al. 1996; Pillinger et al. 2005; Shimada et al. 2006). Additional genes involved in the inactivation of the MAPK pathway were significantly down-regulated (Dusp14 and Dusp3) (Owens and Keyse 2007).

2. Genes involved in tumor suppression, invasion, and metastasis

A number of the significantly altered genes identified in our global gene expression analysis are related to tumor malignancy (Figure 4). Tumors with and without mutations in the K-ras gene exhibited a significant decrease in tumor suppressors (Ptprd, Igsf4a, Fhl1, Pdzd2, Cdkn2d, Cdh5, Loxl1, and Akap12) (Jacob and Motiwala 2005; Mao et al. 2003; Shen et al. 2006; Tam et al. 2006; Thullberg et al. 2000; Umesako et al. 2007; Wu et al. 2007; Yoon et al. 2007) and genes known to inhibit invasion (Reck, Gsn, Lims2, Cav1, and Gpx3) (H. C. Chang, Liu, and Hung 2004; Fujita et al. 2001; S. K. Kim et al. 2006; Williams et al. 2004; Yu et al. 2007).

There were a number of genes that were significantly altered only in the cumene-induced tumors with K-ras mutations. There was decreased expression of additional tumor suppressor genes, genes known to inhibit cell motility, and genes known to inhibit cell proliferation (IGFBP4, Sod3, Rb1, Cebpd, Vwf, and Dlc1) (Durai et al. 2007; S. H. Kim et al. 2005; Kranenburg, van der Eb, and Zantema 1995; O’Rourke et al. 1999; Terraube et al. 2006; Yuan et al. 2004). There was also reduced expression of genes that correlate with patient survival (Cyr61 and Enpp2) (P. P. Chen et al. 2007; Lu et al. 2006).

The tumors with K-ras mutations exhibited increased expression of genes associated with metastatic potential. The increased expression of these genes was not observed in tumors lacking mutated K-ras. There was upregulation of genes known to increase the invasion and metastasis of tumor cells (Krt18, Krt8, Lasp1, Mif, MMP14, and Tacstd1) (Chu et al. 1997; Grunewald et al. 2007; Swant et al. 2005; Tsunezuka et al. 1996; Xi et al. 2006), genes that correlate with increased tumor malignancy and poor patient survival (Eno1, Gpr30, Srd5a1, and Slc2a1) (G. C. Chang et al. 2006; Smith et al. 2007; Wako et al. 2007; Younes et al. 1997), anti-apoptotic genes (Areg and Cks1b) (Hurbin et al. 2002; Tsai et al. 2005), genes that induce angiogenesis (Slc2a1, Gnb2l1, and Ptges) (Airley and Mobasheri 2007; Berns et al. 2000; von Rahden et al. 2006), and genes that are increased in metastatic tumors (Sdc1 and Ccnd1) (Hirabayashi et al. 1998; Volm et al. 2002). The gene expression analysis suggests that carcinomas with K-ras mutations may exhibit a greater degree of malignancy at the molecular level than carcinomas without K-ras mutations. However, at the cellular level, there were no differences in the morphology of carcinomas with or without K-ras mutations (data not shown).

3. Genes regulated by the HDAC complex

In our initial assessment, SAFE identified certain pathways or groups of genes that were altered between carcinomas and normal lungs. SAFE analysis is a permutation-based method for testing functional categories in gene expression experiments. This analysis interprets the data within the context of a set of functionally related genes and has the ability to detect changes in the expression of a set of genes that may have been missed when examining the expression of individual genes. Genes associated with the HDAC complex, which has been shown to play a role in human cancer, were significantly altered (p = .046) in the mouse carcinomas. Figure 5 illustrates the results of SAFE analysis on the gene set defined by GO:0000118.

The x-axis indicates the rank order of the local statistic (see Methods), and the y-axis gives the cumulative distribution of probes in the set (solid line) and the theoretical null cumulative distribution (dotted line). The plot displays the relative extent and direction of differential expression observed for the genes in GO:0000118, with departures from the theoretical null cumulative distribution being of interest. The gene set displays bidirectional differential expression and is enriched for genes that are altered in tumors with K-ras mutations relative to tumors without K-ras mutations, as illustrated by the piling of probes on the left-hand side of Figure 5, indicating high rank scores. The separation of the tumors based on K-ras mutation status detected using SAFE analysis is in agreement with the PCA shown in Figure 1A. Based on these findings, we analyzed genes that are regulated by this complex. The analyzed genes were chosen based on a literature search (Mariadason, Corner, and Augenlicht 2000) and the Affymetrix Web site (GO:0000118 probe set from the Affy MOE430v2 annotation, HDAC complex). Cluster analysis identified 440 probe sets encoding 274 genes associated with HDAC (Supplemental Table 2). Of the 274 genes, 37 genes are involved in the cell cycle (GO:0007049), 42 are involved in cell differentiation (GO:0030154), 75 are involved in development (GO:0007275), 28 are involved in apoptosis (GO:0006915; GO:0012501), and 12 are involved in vascular development (GO:0001944) (Supplemental Table 3). When the genes associated with HDAC were clustered based on direction of change, the K-ras mutation status of the tumor appeared to correlate with the up-regulated genes. The K-ras mutation-positive tumors clustered closer together and separately from the K-ras mutation-negative tumors when examining the up-regulated genes compared to the down-regulated genes.

Validation of Microarray Analysis (Real-time PCR)

To validate the microarray results, the expression of a subset of genes was examined by Taqman qPCR. These genes were identified by microarray as dysregulated in the cumene-induced tumors. The expression of Clu, MAP2K1, Dusp4, and Akap12 by real-time PCR correlated well with the microarray results (Figure 6). The fold change in Clu, Map2k1, and Akap12 was greater by qPCR, while Dusp4 exhibited a slightly lower fold change. Ets1 showed increased gene expression by qPCR when it had displayed a decrease in expression by microarray, likely due to the difference in assay sensitivity. In addition, the expression of Clu in tumors with mutated K-ras was more than twice that of the tumors without mutated K-ras, which corresponds with the microarray data (data not shown).

Immunohistochemical Staining for Map2k1

Map2k1 (MEK1) is an essential component of the MAP kinase signal transduction pathway. Map2k1 lies upstream of erk1 and erk2 and stimulates their enzymatic activity upon a wide variety of extra- and intracellular signals (Shaul and Seger 2007), resulting in the activation of many cellular processes, including proliferation, differentiation, transcriptional regulation, and development.

Tumors with K-ras mutations exhibited cytoplasmic localization of Map2k1, while no staining was observed in the tumors without K-ras mutations or in normal lung tissues (Figure 7). The increased Map2k1 protein expression in tumors with mutated K-ras correlated with the increased gene expression observed by microarray analysis.

Discussion

This study demonstrates that the formation of mouse alveolar/bronchiolar carcinomas with K-ras mutations occurs differently both qualitatively and quantitatively from tumors without K-ras mutations. Tumorigenesis associated with K-ras activation seems to involve expression changes in genes involved in the Erk MAPK signaling pathway, invasion and metastasis, and histone modification. Numerous genes identified in this study play important roles in human carcinogenesis.

Altered expression of Erk MAPK pathway genes downstream of K-ras or associated with Ras activation were identified in the K-ras mutation-positive tumors and likely play important roles in mouse lung carcinogenesis. For example, Ccnd1 (Cyclin D1) was up-regulated in the cumene-induced tumors with K-ras mutations, while there was no significant change in the tumors without K-ras mutations. Cyclin D1 is an important cell cycle protein in the Rb pathway and is up-regulated by K-ras in tumors and in cells that exhibit overexpression of oncogenic K-ras (Peeper et al. 1997). Among the Erk MAPK genes with altered expression in tumors with mutated K-ras are those that encode proteins that activate MAPK signaling, are activated by MAPK signaling, and inactivate MAPK signaling. The majority of the MAPK-related genes were altered only in cumene-induced lung tumors with K-ras mutations. The results suggest that the Erk MAPK signaling pathway is more active in cumene-induced lung tumors that harbor a mutation in K-ras and support the idea that the K-ras-induced activation of the MAPK signaling pathway plays a major role the formation of cumene-induced lung tumors.

In addition, gene expression analysis suggested that the cumene-induced carcinomas with mutations in K-ras possess a greater degree of malignancy at the molecular level. While tumors with and without K-ras mutations have decreased expression of tumor suppressor genes, tumors with mutations in K-ras display downregulation of additional tumor suppressors. The tumors with mutated K-ras also displayed increased expression of genes known to increase invasion and metastasis, inhibit apoptosis, increase angiogenesis, and increase metastatic potential. The difference in gene expression suggests that cumene-induced carcinomas with mutations in K-ras exhibit a higher degree of malignancy and could potentially result in lower survival rates. The NTP study also suggested a decreased survival rate in animals with mutated K-ras. Of the animals that died prior to study completion, 96% had a K-ras mutation, compared to 74% of the animals that survived to the end of the study (NTP 2007). The tumors with and without K-ras mutations did not display histopathological differences. This suggests that while the tumors are histologically similar, at the molecular level, there are distinct differences, which may have implications for tumor malignancy and is supported by the lower survival rate in mice with K-ras mutations.

The relationship between genetic and epigenetic alterations during lung cancer is not clear, but studies have begun to reveal an association between oncogenic K-ras, DNA methylation, and histone modification in cancer (Toyooka et al. 2006, 2003). Cluster analysis of altered genes putatively associated with HDAC regulation showed a stronger association with K-ras mutation-positive tumors than tumors without K-ras mutations. Our results, in combination with previous studies, suggest that the activation of K-ras may have an effect on histone modification. For example, oncogenic Ras increased the expression of HDAC4 in the nucleus of myoblast cells (Zhou et al. 2000). In a microarray study of K-ras transfected adrenocortical cells, retinoblastoma binding protein 1 (RBBP1) was down-regulated fourfold and was correlated to cell proliferation. RBBP1 binds with RB/E2F to form the mSIN3-HDAC complex that induces cell cycle arrest (Y. F. Chen et al. 2003). The study also suggested that activated K-ras inhibits the formation of mSIN3-HDAC complex. In another study, an inhibitor for the Sirt1 class III HDAC was found to inhibit the Ras-MAPK pathway, suggesting that activated Ras contributes to the activation of Sirt1 (Y. F. Chen et al. 2003; Ota et al. 2006; Zhou et al. 2000).

SAFE analysis demonstrated a trend in the expression of HDAC-associated genes between carcinomas and normal lungs. This analysis displayed bidirectional differential expression and was enriched in genes altered in K-ras+ tumors relative to K-ras−tumors. Analysis of the gene set identified a number of genes involved in histone regulation, including components of the HDAC complex (Sap30 and Sap18), HATs (Myst2, Pcaf, Gcn5l2, and Atf4), and HDACs (Hdac4, Sirt1, and Hdac10) (Eberharter and Becker 2002; Kristeleit et al. 2004). In addition, the analysis identified numerous genes whose expression can be altered by the action of HATs and HDACs (Nos3, Ctnna1, Clu, Cdkn1c, FosB, Dusp1, RhoB, Vwf, Prdx1, Id2, and Krt23) (Gan et al. 2005; Hajra and Fearon 2002; Hellebrekers et al. 2007; Kikuchi et al. 2002; Levine et al. 2005; Li et al. 2001; Mazieres et al. 2007; Peng et al. 2007; Sanda et al. 2007; Yoshikawa et al. 2007; Zhang et al. 2001; see Supplemental Table 2). The genes in each category did not display consistent alterations in expression pattern. It is likely that the action of the HDACs and HATs are gene dependent and vary from one gene to another. Mouse lung tumor models could be useful tools to study the interaction between activated K-ras and the HDAC complexes. Therefore, future work will be carried out to examine the extent of HAT and HDAC involvement in altered gene expression in cumene-induced lung tumors. We also examined the potential role of methylation in cumene-induced lung tumors by examining the methylation status of genes known to be methylated and exhibiting decreased expression in our microarray analysis. There was no evidence of methylation of Akap12, Gata2, and Timp3 in any of the cumene-induced lung tumors (data not shown).

In addition, our gene expression analysis identified potential alterations in the Notch and Wnt signaling pathways. These pathways are usually activated in NSCLC and inactivated in SCLC in humans and rodents (Collins, Kleeberger, and Ball 2004; Daniel, Peacock, and Watkins 2006; Tennis, Van Scoyk, and Winn 2007). However, our analysis suggests that these pathways may be inhibited in cumene-induced lung tumors with K-ras mutations. The Delta homologue, Dlk1, which functions as a Notch inhibitor, is significantly increased in tumors with K-ras mutations, while the ligand Notch3 is significantly decreased in the same tumors. Notch4 is decreased in tumors with and without K-ras mutations. Four members of the Wnt signaling pathway were also altered in cumene-induced tumors bearing K-ras mutations. The expression of a Wnt receptor (Fzd4), a Wnt ligand (Wnt2), a transcription factor (Tcf4), and a downstream target (Tbx3) were all significantly down-regulated in the tumors. These gene expression changes suggest that the Notch and Wnt signaling pathways may be inhibited in cumene-induced lung tumors, specifically in the tumors carrying a mutation in K-ras. Future work needs to be done to examine the activation of these signaling pathways in cumene-induced lung tumors.

Our studies demonstrate the usefulness of animal models in providing clues on both genetic and epigenetic factors that may contribute to the development of common cancers, including lung cancer. We were able to show that in addition to K-ras mutations, the gene expression profile of cumene-induced lung tumors was linked to the MAP kinase signaling pathway and histone modification. The significance of these findings is that many of the genes with altered expression in the mouse tumor model represent major genes that may play a role in lung cancer and other cancers in humans.