Effects of tobacco smoke on gene expression and cellular pathways in a cellular model of oral leukoplakia

Abstract

In addition to being causally linked to the formation of multiple tumor types, tobacco use has been associated with decreased efficacy of anticancer treatment and reduced survival time. A detailed understanding of the cellular mechanisms that are affected by tobacco smoke (TS) should facilitate the development of improved preventive and therapeutic strategies. We have investigated the effects of a TS extract on the transcriptome of MSK-Leuk1 cells, a cellular model of oral leukoplakia. Using Affymetrix HGU133 Plus 2 arrays, 411 differentially expressed probe sets were identified. The observed transcriptome changes were grouped according to functional information and translated into molecular interaction network maps and signaling pathways. Pathways related to cellular proliferation, inflammation, apoptosis, and tissue injury seemed to be perturbed. Analysis of networks connecting the affected genes identified specific modulated molecular interactions, hubs, and key transcription regulators. Thus, TS was found to induce several epidermal growth factor receptor (EGFR) ligands forming an EGFR-centered molecular interaction network, as well as several aryl hydrocarbon receptor-dependent genes, including the xenobiotic metabolizing enzymes CYP1A1 and CYP1B1. Notably, the latter findings in vitro are consistent with our parallel finding that CYP1A1 and CYP1B1 levels were increased in oral mucosa of smokers. Collectively, these results offer insights into the mechanisms underlying the procarcinogenic effects of TS and raise the possibility that inhibitors of EGFR or aryl hydrocarbon receptor signaling will prevent or delay the development of TS-related tumors. Moreover, the inductive effects of TS on xenobiotic metabolizing enzymes may help explain the reduced efficacy of chemotherapy, and suggest targets for chemopreventive agents in smokers.

Figure 1

A, Flowsheet describing the systematic analysis of TS induced changes in the transcriptome of MSK-Leuk1 cells, including hypothesis generation steps. B, Protein-protein interaction network of differentially expressed genes (www.physiology.med.cornell.edu/go/smoke) and signaling proteins known to be activated by TS (AhR, ADAM17, CREB1, MAPK3 (ERK1), MAPK1 (ERK2), PRKACA and SRC) within HiMAP database, which includes interactions that are literature-confirmed from the Human Protein Reference Database (40) (blue), yeast two-hybrid-defined (gray), or predicted based on function (gray). Hubs with 4 or more connections (red) are CCL5, COL1A1, CXCL10, EGFR, HMOX1, IL1A, IL1B, IL1RN, IL1R1, IL6, IL8, MAPK1, MAPK3, PTGS2, PLAU, PLSCR1, STAT1 are VEGF. C, Direct interaction network of differentially expressed genes and known TS-activated signaling proteins generated using IPA (29). At each edge, the interaction type is shown and the number of publications is indicated in parentheses. Hubs are listed in Supplementary Table 2.

Figure 2

A, Direct interaction network of differentially expressed genes in TS-treated MSK-Leuk1 cells, integrated with genes related to signaling proteins known to be activated by TS. The white nodes are genes with no significant expression change due to TS in our analysis, but which are expressed in MSK-Leuk1 cells (filtered as in methods, statistical analysis 2a,b) and interact with at least 6 differentially expressed genes, suggesting potential roles for these in TS perturbed signaling processes. These include ARNT, BRCA1, CEBPB, CEBPD, CREB1, CTNNB1, EGR1, FOS, HDAC1, HIF1A, IRF5, MYC, NFE2L2, SMAD3, SMARCA4, SP1, TP53, TP73L. IPA tool (29) was used to generate the panel A. B, Statistically significant diseases and physiological, cellular or molecular functions of the differentially expressed genes. The bars are sized according to the calculated -log (significance) score. C, Statistically overrepresented groups in GO Biological Process (GO BP), GO Molecular Function (GO MF), GO Cellular Components (GO CC) and organismal role, among the differentially expressed genes as compared to the set of genes in the microarray, generated with EASE software, based on overrepresented chromosomes, organismal roles, GO categories, and pathway categories within KEGG, GenMAPP (http://www.genmapp.org), and BBID (http://bbid.grc.nia.nih.gov) databases. Categories with Bonferroni p-value<0.05 were assumed significant, and Affymetrix identifiers are used.

Figure 3

Levels of CYP1A1 and CYP1B1 are increased in the oral mucosa of cigarette smokers. A, oral mucosal biopsies were obtained from both never smoking (never smokers, n=9) and smoking (smoker, n=9) human volunteers. Total cellular RNA was extracted from the oral mucosal biopsy samples and reverse transcribed. The expression of CYP1A1 and CYP1B1 mRNAs was assessed by RT-PCR. No bands were observed when cDNA was omitted from the PCR reaction or when the reverse transcriptase enzyme was not included in the reverse transcriptase reaction. B, Results of the data shown in panel A expressed in arbitrary units. Means and S.D. are shown; *P <0.05, **P<0.01 vs. never smokers.