| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Research Articles |
1,4Authors' Affiliations: 1 Physiology and Biophysics, 2 Medicine, and 3 Otolaryngology, and 4 HRH Prince Alwaleed Bin Talal Bin Abdulaziz Alsaud Institute for Computational Biomedicine, Weill Medical College of Cornell University; and 5 Department of Surgery (Head and Neck Service), Memorial Sloan-Kettering Cancer Center, New York, New York
Requests for reprints: Harel Weinstein, Department of Physiology and Biophysics, Weill Medical College of Cornell University, Box 75, Room E-509, 1300 York Avenue, New York, NY 10065. Phone: 212-746-6358; Fax: 212-746-8690; E-mail: haw2002{at}med.cornell.edu.
 |
Abstract |
|---|
 TOP Abstract Materials and MethodsResultsDiscussionDisclosure of Potential...AcknowledgmentsReferences |
|---|
Key Words: microarray • tobacco smoke • gene expression • oral • aryl hydrocarbon receptor
|
|
TOPAbstract Materials and MethodsResultsDiscussionDisclosure of Potential...AcknowledgmentsReferences |
|---|
30% of all cancer-related deaths in the United States (1). Exposure to tobacco has been linked to a variety of malignancies including cancers of the lung, oral cavity and pharynx, esophagus, pancreas, liver, bladder, and cervix (2). More than 100 carcinogens, mutagens, and tumor promoters have been identified in tobacco smoke (TS; refs. 3, 4). In addition to being a major cause of cancer, smoking can alter the activity of chemopreventive agents (4, 5), stimulate the metabolic clearance of targeted anticancer therapies (6), and increase the risk of second primary tumors (7). Although cessation of tobacco use is highly desirable, it is not realistic for everyone. Hence, there is a significant interest in chemopreventive agents that could protect against the carcinogenic effects of TS. Moreover, a clearer understanding of the mechanisms that are modulated by TS should lead to more effective treatments resulting in an improved outcome for cancer patients. Given the significance of TS as both a cause of cancer and a potential modifier of treatment outcome (1, 8–10), we have investigated the effects of a TS extract on the transcriptome in a cellular model of oral leukoplakia. This model was chosen because the link between exposure to TS and head and neck squamous cell carcinoma is well established. Furthermore, smoking reduces the likelihood of treatment response in head and neck cancer patients (11) and increases the risk of second primary tumors in patients who have been successfully treated for their index head and neck malignancy (7). The transcriptome analysis involved the identification of genes differentially expressed due to TS exposure in this cell line, followed by a classification of these genes into domains of putative physiologic function. The classification involved the mapping of interactions among differentially expressed genes based on information from interaction databases. Several different databases and tools were employed in this analysis of the observed global transcriptome changes in terms of biological functions and pathways, with the results suggesting that pathways related to cell proliferation, inflammation, apoptosis, and tissue injury were affected by TS. Finally, network representations of these data led to identification of proteins in the differentially expressed cohort that have multiple interaction partners (interaction hubs) and of transcription factors affected by TS. The analyses identified an epidermal growth factor receptor (EGFR)–centered network composed of several ligands of the EGFR that were induced by TS. Notably, aryl hydrocarbon receptor (AhR)–dependent genes induced by TS included the enzymes CYP1A1 and CYP1B1, which are of special interest because each may contribute to both carcinogenesis of the aerodigestive tract and drug metabolism (12–14). Consequently, we extended our analysis of TS-related transcriptome changes to human volunteers. Consistent with the in vitro findings presented here, we found increased levels of both CYP1A1 and CYP1B1 in the oral mucosa of healthy human subjects who smoked cigarettes. Further comparison of our findings in the MSK-Leuk1 cell model to in vivo data on transcriptome differences in airway epithelial cells of smokers (versus never smokers; ref. 15) identified a canonical set of differentially expressed genes and perturbed pathways. In addition to providing new insights into the procarcinogenic effects of TS, these findings highlight the potential of TS to alter the efficacy of pharmaceutical agents by inducing the expression of xenobiotic metabolizing enzymes.
|
Materials and Methods |
|---|
|
TOPAbstract Materials and MethodsResultsDiscussionDisclosure of Potential...AcknowledgmentsReferences |
|---|
Tissue culture
The MSK-Leuk1 cell line was established from a dysplastic leukoplakia lesion adjacent to a squamous cell carcinoma of the tongue (16). Cells were routinely maintained in keratinocyte growth medium supplemented with bovine pituitary extract. Cells were grown in basal medium for 24 h before treatment. Treatment with vehicle (PBS) or TS was carried out on cells grown in growth factor–free basal medium. Cellular cytotoxicity was assessed by measurements of cell number, trypan blue exclusion, and release of lactate dehydrogenase. There was no evidence of cytotoxicity in our experiments.
Preparation of TS condensate
Cigarettes (2R4F, Kentucky Tobacco Research Institute) were smoked in a Borgwaldt piston-controlled apparatus (model RG-1) using the Federal Trade Commission standard protocol (17). The protocol variables attempt to mimic a standardized human smoking pattern (duration, 2 s/puff; frequency, 1 puff/min; volume, 35 mL/puff). Cigarettes were smoked one at a time in the apparatus and the smoke drawn under sterile conditions into premeasured amounts of sterile PBS (pH 7.4). This smoke in PBS represents whole trapped mainstream smoke abbreviated as TS. Quantitation of smoke content is expressed in puffs per milliliter, with one cigarette yielding 8 puffs drawn into a 5-mL volume. The final TS concentration in cell culture medium is expressed as puffs per milliliter of medium.
Human tissue
Buccal mucosa specimens were obtained from nine never smokers and nine active smokers with a history of at least 10 pack-years. Subjects were excluded if they had gross evidence of oral inflammation, had a history of heavy alcohol consumption, or recently used nonsteroidal anti-inflammatory drugs or other anti-inflammatory medications. After topical anesthesia, 5-mm punch biopsies were obtained from grossly normal-appearing buccal mucosa. Tissue samples were immediately snap frozen in liquid nitrogen and stored at –80°C until analysis. H&E staining of representative formalin-fixed samples indicated that the biopsies were primarily composed of epithelium. This study was approved by the Committee on Human Rights in Research at Weill Cornell Medical College.
Reverse transcription-PCR
Total cellular RNA was isolated from cells using the RNeasy Mini Kit (Qiagen) according to the manufacturer's instructions. Reverse transcription was done with 2 µg of RNA per 50 µL of reaction. The reaction mixture contained 1x PCR buffer II, 2.5 mmol/L MgCl2, 0.5 mmol/L deoxynucleotide triphosphates, 2.5 µmol/L oligo(dT)16 primer, 50 units of RNase inhibitor, and 125 units of MuLV. Samples were amplified in a thermocycler at 25°C for 10 min, 42°C for 15 min, 99°C for 5 min, and 5°C for 5 min. The resulting cDNA was then used for amplification. The volume of the PCR reaction was 25 µL and contained 5 µL of cDNA, 1x PCR buffer II, 2 mmol/L MgCl2, 0.4 mmol/L deoxynucleotide triphosphates, 400 nmol/L forward primer, 400 nmol/L reverse primer, and 2.5 units of Taq polymerase. Samples were denatured at 95°C for 2 min and then amplified for 30 cycles in a thermocycler under the following conditions: 95°C for 30 s, 62°C for 30 s, and 70°C for 45 s. Subsequently, the extension was carried out at 70°C for 10 min. Primers were synthesized by Sigma Genosys, and the sequences are listed in Supplementary Table S1.
To determine the levels of mRNAs for CYP1A1 and CYP1B1 in buccal mucosa, total RNA was isolated from biopsy samples with the RNeasy Mini Kit. Analysis was carried out as described above. Thermal cycling conditions were 95°C for 2 min, followed by 30 s at 95°C, 30 s at 62°C, and 45 s at 72°C for 30 cycles and then 72°C for 10 min. PCR products were subjected to electrophoresis on a 1% agarose gel with 0.5 µg/mL ethidium bromide. The identity of each PCR product was confirmed by DNA sequencing. A computer densitometer (ChemDoc, Bio-Rad) was used to quantify the density of the different bands.
Microarray procedures
Biotinylated cRNAs were prepared according to the standard Affymetrix6 protocol from 2.5 µg of total RNA. Following fragmentation, 10 µg of cRNA were hybridized for 16 h at 45°C on GeneChip HG U133 Plus 2 arrays. GeneChips were washed and stained in Affymetrix Fluidics Station 450 and scanned with Affymetrix GeneChip Scanner 3000. At each of the five time points, six biological replicates were used for each TS treatment and another six biological replicates for vehicle-treated samples. In total, 60 chips were used.
Microarray data analysis
The gene annotations used for each probe set were from the February 2008 NetAffx HGU133 Plus 2 Annotation Files.
Preprocessing
Raw image data were background corrected, normalized, and summarized into probe set expression values using the Robust Multichip Average (RMA) algorithm. In our analysis, the largest variation in results arose from the method of preprocessing. Both the development of preprocessing methods and the assessment of results are active research areas and the choice of method affects the analysis outcome (18). RMA (19) and a modification, GC-RMA (20), have been shown to perform as well as, or better than, alternatives using Plasmode data sets. However, GC-RMA can be biased when outliers are not eliminated. Further, both GC-RMA and another commonly used method, Affymetrix Microarray Analysis Suite v5.0 (MAS5), may perform poorly with highly variable human data (21). To check for the robustness of our results to different preprocessing methods, raw image data were preprocessed using both the RMA algorithm (19, 22) within GeneSpring 7.2 Software (Agilent Technologies) and MAS5, and then analyzed statistically as described below. We found a 75% agreement between RMA-preprocessed results and those from MAS5 (data not shown). This indicates a high level of consistency, as the overlap between RMA and MAS cited in the literature ranges from 27% (23) to 70% (18). In light of evaluation of the currently applicable methods in recent reviews (24), we performed both the statistical and functional inference analyses from the RMA-preprocessed data.
Normalization and filtering
The data from each chip were normalized for interarray comparisons as follows: measurements of <0.01 were set to 0.01 and each chip was normalized to 50% of the measurements taken from that chip (a procedure considered appropriate for large arrays when most of the genes are unaffected by experimental parameters). We further applied a filter to remove probe sets that were not reliably detected. From the complete set of 54,675 probe sets on the HGU133 Plus 2 array, for every time point, we filtered out probe sets whose minimum raw expression level was not 50 in at least 2 of 12 conditions. This cutoff was chosen from the scatter plot distribution of expression values for TS versus vehicle-treated controls (marked as C). We further filtered out probe sets with low confidence if their t-test P value was not <0.05 in at least 2 of 12 conditions, using the Benjamini and Hochberg false discovery rate (FDR) criterion (25). Genes that passed these tests were defined as expressed and were statistically analyzed. This set of analyzed genes consists of the following numbers of probe sets: 20,791 (at 0.5 h), 20,443 (3 h), 20,979 (6 h), 19,584 (12 h), and 20,034 (24 h).
Statistical analysis
Probe sets were analyzed using both ANOVA and Significance Analysis of Microarrays (SAM). Genes that passed both SAM and ANOVA tests and had normalized expression values altered by a factor of 1.5-fold were deemed significant. ANOVA analysis was done using GeneSpring 7.2 software. SAM analysis was done using SAM Microsoft Excel plug-in. The details of this analysis are as follows: A one-way ANOVA was done at every time point using parametric test, not assuming variances equal (Welch's t test) with P < 0.05. To address the problem of multiple comparisons, Benjamini and Hochberg multiple testing correction was used to maintain FDR at 5% (25). The SAM method uses a set of gene-specific t tests, where each gene is assigned a score on the basis of its change in gene expression, relative to the SD of repeated measurements for that gene (26). Genes with scores greater than threshold 
values at FDR 4% to 5% were deemed significant (because the FDR is dependent on value, an exact 5% FDR is not possible using the delta-slider within the SAM Excel plug in). At this value, the fold change parameter was set to 1.5. For input to SAM, two-class unpaired response was chosen on normalized data with t-statistics, 200 permutations, and default SAM options.
In summary, the analysis steps described above in comparing TS versus control at every time point are as follows: preprocessing (RMA) 
normalization filtering for expression statistical analysis (SAM and ANOVA with multiple testing correction) filtering for fold change.
Comparative analysis
Raw data (Affymetrix HGU133A chips) were downloaded from GEO repository.7 Smoking status–related genes in airway epithelia that are significantly differentially expressed were assessed using GeneSpring 7.2 as described above and separated into up-regulated and down-regulated gene sets comprising 110 and 21 genes, respectively. Gene Set Enrichment Analysis (27) was used to compare the MSK-Leuk1 and airway epithelia gene expression results. The up-regulated and down-regulated gene sets in airway data were used to identify enrichment in the MSK-Leuk1 data. The MSK-Leuk1 expression data inputs were the normalized ratio values of all probe sets in HGU133 Plus 2 arrays and their collapsed HUGO symbols. Gene set enrichment analysis was done for every time point using default parameters, except the permutation type, which was gene set instead of phenotype (as recommended by the Gene Set Enrichment Analysis manual when number of replicates are <7). A reverse analysis was also done by using all the probe sets in airway epithelial data set as input and searching for enrichment on the significantly up-regulated or down-regulated gene sets at every time point of MSK-Leuk1 cell exposure to TS. Here, the differentially expressed gene sets in MSK-Leuk1 data were collapsed to the HUGO symbols of probe sets available in HGU133A microarrays.
Leading-edge subset
The leading-edge subset is defined as comprising those members of the gene set that appear in the ranked list at, or before, the point where the running sum reaches its maximum deviation from zero. The set is suggested to be the core of a gene set that accounts for the enrichment score (27).
Functional analysis
To relate the results to cell physiologic mechanisms, the transcriptional data were integrated with available experimental signaling data for TS. The complex biological processes induced by TS were examined in the context of detailed protein-protein interaction maps (28) and molecular networks (29). The interaction networks shown in Figs. 1C and 2A were generated with Ingenuity Pathways Analysis (IPA), a web-delivered application used to discover, visualize, and explore relevant networks (29). Affymetrix probe identifiers were uploaded to IPA, each identifier was mapped to its corresponding gene object in the Ingenuity Pathways Knowledgebase, and only direct interactions were considered. For the network in Fig. 2A, interactions were queried between these gene objects and all other gene objects stored within IPA to generate a set of networks that were then merged. The only putative hubs considered in the merged network were transcription regulators that are expressed in MSK-Leuk1 cells and have at least six direct interactions with differentially expressed genes.
|
|
We reasoned that a small but coherent difference in the expression of a group of genes in a category or pathway can be more important than large differences in unrelated genes. Therefore, we carried out categorization and pathway-enrichment analyses that enabled identification and scoring of both significant and modest perturbations in corresponding gene groups. For a functional analysis that can capture modest perturbations in functional groups, we used all expressed genes (as defined above in "Microarray data analysis"). Hence, we identified groups of genes that correspond to specific enzymatic, metabolic, or signaling pathways within pathway databases of KEGG9 and BioCarta10 using the PLAGE tool (31) and the SAFE tool (32) within GO categories. Note that we did not use a FDR or family-wise error rate estimate for functional groups. Whereas this might result in an increase in type II estimates, it is acceptable because we focused particularly on pathways that seem to be consistently perturbed at multiple time points.
Clustering
An unsupervised hierarchical clustering analysis across all samples of the microarray data was done for the probe sets found to be differentially expressed between control and TS-treated cells at any of the five time points (using log-transformed normalized data). A Pearson correlation (uncentered) similarity metric and average linkage clustering was done with Cluster and TreeView software11 (see Supplementary Fig. S1). Our study was not designed to identify the dynamic effects of TS so that limitation to five time points results in a rather sparse matrix for such analysis under the commonly used tools [CAGED (33) and SSCLUST (34)]. Further results from the use of a newly described clustering algorithm, EP_GOS_Clust (35), were used to classify the time-dependent patterns of changes in gene expression and will be reported as a separate study.
Additional information
The complete list of differentially expressed genes and functional groups at every time point is made available through an interactive web site12 established as a resource of the Institute for Computational Biomedicine. Also available at that site are the results from the functional analyses of the in vivo data for comparative purposes. The microarray data have been deposited at the National Center for Biotechnology Information Gene Expression Omnibus (GEO)13 under GEO Series accession no. GSE10063.
|
Results |
|---|
|
TOPAbstract Materials and MethodsResultsDiscussionDisclosure of Potential...AcknowledgmentsReferences |
|---|
(TGFA; ref. 17).
|
Network analysis
The differentially expressed gene set was integrated with the set of signaling proteins previously shown to be activated by TS: ADAM17 (37), AhR (36, 38), cyclic AMP responsive element binding protein 1 (CREB1; ref. 36), mitogen-activated protein kinase [extracellular signal–regulated kinase (ERK)-1/2; ref. 39], protein kinase A catalytic subunit (PRKACA; ref. 36), and SRC.15 Assuming that highly interconnected networks likely represent significant biological function, we sought to identify the interrelations among these genes.
Protein-protein interactions. Direct physical relationships between proteins associated with the gene set were identified using the human protein-protein interaction map HiMAP (28), as shown in Fig. 1B. The proteins found to have four or more possible physical interactions are listed in Supplementary Table S2. Hubs with the most Human Protein Reference Database (40) connections (blue edges in Fig. 1B) are MAPK3, COL1A1, and EGFR. In fact, an EGFR-centered subnetwork was observed, where EGFR and its ligands EREG, AREG, DTR (HB-EGF), and TGFA are all significantly induced.
Interactions based on mammalian biology data. The interactions within the set of TS-modified genes were mapped in the context of the network of physical, transcriptional, and enzymatic interactions observed in mammals. The results (obtained using the Ingenuity Pathway Analysis Tool, IPA; ref. 29) are represented in Fig. 1C. The interactions considered were proteolysis, inhibition, protein-protein interaction, expression, protein-DNA interaction, activation, and transcription (29). Among hub proteins (listed in Supplementary Table S2), EGFR, PLAU, IL1R1, and related proteins are mainly induced (shown in red), whereas STAT1 and related proteins are repressed (shown in green). Interestingly, cytokines form a subnetwork composed of IL1R1, IL1R2, IL1A, IL1B, and IL1RN, suggesting a TS-induced inflammatory process. Another hub, formed around PLAU, involves genes that contribute to invasiveness (39). At the stringency level used for the initial analysis of TS-affected gene expression, AhR expression was not considered significantly altered, but the likely role of the AhR is evident from the effect of TS on genes known to be regulated by this receptor. Specific examples include the observed increased levels of CYP1A1, CYP1B1, and AREG, which are known targets of ligand-activated AhR (36, 38). In a similar manner, various transcription regulators connected to large numbers of differentially expressed genes can have important roles in signaling pathways perturbed by TS. Mapping the direct interaction network of differentially expressed genes and transcription regulators can reveal such involvement. Shown in Fig. 2A are highly connected putative transcription regulators in response to TS, including sp1 transcription factor (SP1), catenin β1 (CTNNB1), IFN regulatory factor 5 (IRF5), histone deacetylase 1 (HDAC1), tumor protein p53 (TP53), tumor protein p73-like (TP73L), hypoxia inducible factor 1, subunit (HIF1A), nuclear factor (erythroid-derived 2)-like 2 (NFE2L2), cellular oncogene c-fos (FOS), myc proto-oncogene protein (MYC), early growth response (EGR1), and a number of well-studied transcription factors and cancer-related genes that included BRCA1, CEBPB, CEBPD, and the AhR nuclear translocator (ARNT). We note that the AhR and hypoxia signaling pathways have been shown to cross-talk via the involvement of HIF1 (41), and whereas HIF1 does not seem to be significantly altered, our analysis suggests its involvement in signaling pathways as a transcription regulator hub.
Gene classification
Classification of statistically significant genes into biological categories. The genes affected by TS were also classified into related diseases and molecular, cellular, and physiologic functions based on the information from the largest literature curated information database, IPA (Fig. 2B; ref. 29). Cancer is the most relevant disease and the most relevant functions are cell-to-cell signaling and interaction, cellular growth and proliferation, drug and lipid metabolism, immune response, and connective tissue development and function. We further categorized genes in terms of relevant functional categories using public databases of controlled vocabulary including GO (Fig. 2C; Supplementary Table S3). The information from these databases also confirms that genes related to cell proliferation and immune responses are overrepresented. As the examples above suggest, the classification of differentially expressed genes is inherently sensitive to the statistical analysis results and filtering criteria. We therefore identified an additional category of statistically significant modestly perturbed functional groups of genes among all expressed genes in the MSK-Leuk1 cells and defined their functional categories using the GO, KEGG, and BioCarta databases. Pathways consistently perturbed at multiple time points are given in Table 2.
|
|
|
The leading-edge subsets of the significant gene sets in airway epithelia include 50 up-regulated and 16 down-regulated genes. Of these, ALDH1A3, CYP1A1, CYP1B1, GCLM, GPX2, NQO1, PIR, SERPINB13, SLC7A11, TXNDR1, and UGT1A10 /// UGT1A8 are members of a consistently induced subset (in at least four time points). At the same time, KAL1, MMP10, NFKBIA, and TMEM45A are members of a consistently repressed subset (in at least four time points) in both sets of experiments. The highest number of leading-edge subset genes in the airway epithelial data are for the 0.5-hour time point, which is interesting because this time point does not have the largest number of differentially expressed genes. The reverse analysis, for the leading-edge subsets of the significantly up-regulated genes in MSK-Leuk1 data, results in a total of 20 genes enriched in airway epithelial cells. The complete set of results is given in Supplementary Table S6. Details of this analysis and the related html files are available online.16
|
Discussion |
|---|
|
TOPAbstract Materials and MethodsResultsDiscussionDisclosure of Potential...AcknowledgmentsReferences |
|---|
In addition to being able to activate carcinogens, CYP1A1 and CYP1B1 play a role in the metabolism of several anticancer drugs including docetaxel, tamoxifen, and erlotinib (6, 46). Recently, erlotinib, a small-molecule inhibitor of EGFR tyrosine kinase, was found to be more effective in the treatment of patients with non–small cell lung cancer who were never smokers compared with smokers (10). Reduced levels of erlotinib were found in the plasma of smokers compared with never smokers, suggesting increased metabolic clearance (6). Our finding that CYP1A1 and CYP1B1 levels are increased in the oral mucosa of smokers raises the distinct possibility that, in addition to systemic clearance, local clearance of erlotinib will be enhanced in smokers, leading to decreased clinical benefit. Collectively, these findings underscore the need for both a more careful monitoring of smoking status in clinical trials and increased smoking cessation efforts in cancer patients. Studies are under way or being planned to evaluate the efficacy of erlotinib in the prevention and treatment of head and neck squamous cell carcinoma.
The induction of CYP1A1 and CYP1B1 by TS is consistent with evidence that the AhR plays a central role in regulating these genes. The AhR had been linked to carcinogenesis (47, 48), and the polycyclic aromatic hydrocarbons in TS bind to and activate the AhR, resulting in the induction of CYP1A1 and CYP1B1 (13, 38, 41). Following ligand-induced activation by polycyclic aromatic hydrocarbons, the AhR releases its chaperoning heat shock protein 90, translocates into the nucleus, and dimerizes with ARNT (49). The heterodimer binds to xenobiotic response elements present in the 5' flanking region of target genes and thereby modulates transcription. In addition to CYP1A1 and CYP1B1, polycyclic aromatic hydrocarbons induce the phase II xenobiotic metabolizing enzyme NAD(P)H:quinone oxidoreductase (NQO1) and the AhR repressor. The AhR and AhR repressor are believed to constitute a negative feedback loop of xenobiotic signal transduction. The liganded AhR induces AhR repressor transcription, whereas expressed AhR repressor, in turn, inhibits the function of AhR (49). Both NQO1 and AhR repressor were induced in the TS-treated MSK-Leuk1 cells although the magnitude of this induction was less than for CYP1A1 and CYP1B1. Clearly, these results suggest that it will be worthwhile to determine if increased levels of NQO1 and AhR repressor also occur in the buccal mucosa of smokers. Easily accessible buccal mucosa may serve as a surrogate tissue for understanding the effects of TS on the biology of difficult-to-obtain bronchial mucosa. In support of this notion, we found that TS induced changes in the transcriptome of MSK-Leuk1 cells that were mimicked by differences in bronchial mucosa of smokers versus nonsmokers. The AhR-driven gene induction observed in the MSK-Leuk1 model matches the in vivo results in the airway epithelial cells of smokers (Table 3). In this regard, the fact that a small subset (CYP1A1 and CYP1B1) of highly inducible genes was also overexpressed in the buccal mucosa of smokers is of interest. Future studies of the transcriptome of buccal mucosa of smokers versus never smokers will be needed to draw more definitive conclusions about how closely the biology of the buccal mucosa reflects changes in the lower respiratory tract.
Our pathway analysis showed that treatment of MSK-Leuk1 cells altered the expression of genes involved in cellular proliferation, raising the possibility that TS amplifies its own mutagenicity by stimulating the proliferation of cells because conversion of TS-induced DNA adducts to mutations can only occur in proliferating cells (50). Enhanced cell proliferation has been observed in the aerodigestive tracts of active smokers (51). Interestingly, intracellular levels of glutathione, an antioxidant, have been shown to modulate the effects of TS condensate on cell proliferation (52). Notably, we find that the glutathione metabolism pathway is induced both in MSK-Leuk1 cells and in the airway epithelial cells of smokers.
We and others found that TS-mediated activation of EGFR signaling led to increased cell proliferation (36, 37). In the current study, treatment with TS led to increased levels of both EGFR and its ligands including AREG, TGFA, HB-EGF, and EREG, forming an EGFR-centered hub within the interactome maps. Collectively, these findings strengthen the rationale for evaluating whether an inhibitor of EGFR tyrosine kinase can prevent or delay the onset of TS-related malignancies of the aerodigestive tract.
Our pathway analysis suggests that TS modulates cell movement, apoptosis, immune function, and coagulation. Both inflammation and immune suppression have been suggested to contribute to TS-induced carcinogenesis (53). Consistently, we identified the differential expression of several interleukins and their receptors, including IL1A, IL1B, IL1RN, IL1R1, IL1R2, IL6, IL7R, IL8, IL11, IL17RC, IL20, and IL24, in addition to TNFA, where IL1B, IL1R1, IL1RN, and IL8 are connected within a network. Our pathway-level analysis also links TS exposure to the induction of specific inflammatory pathways (Table 2).
Taken together, our results provide new insights into the potential mechanisms underlying procarcinogenic effects of TS and may help to explain the worse outcome of cancer patients who continue to smoke. Furthermore, our results suggest important targets for future studies designed to identify agents that could reduce or eliminate the detrimental effects of TS exposure indicated by the findings presented here.
|
Disclosure of Potential Conflicts of Interest |
|---|
|
TOPAbstract Materials and MethodsResultsDiscussionDisclosure of Potential...AcknowledgmentsReferences |
|---|
|
Acknowledgments |
|---|
|
TOPAbstract Materials and MethodsResultsDiscussionDisclosure of Potential...AcknowledgmentsReferences |
|---|
|
Footnotes |
|---|
Note: Supplementary data for this article are available at Cancer Prevention Research Online (http://cancerprevres.aacrjournals.org/).
7http://www.ncbi.nlm.nih.gov/geo/ 
10http://biocarta.com/genes/index.asp
11Cluster and TreeView were obtained at http://rana.lbl.gov/EisenSoftware.htm.
12http://physiology.med.cornell.edu/go/smoke
13http://www.ncbi.nlm.nih.gov/geo/
14http://physiology.med.cornell.edu/go/smoke
11http://physiology.med.cornell.edu/go/smoke
Received for publication January 9, 2008.
Revision received February 17, 2008.
Accepted February 22, 2008
|
References |
|---|
|
TOPAbstract Materials and MethodsResultsDiscussionDisclosure of Potential...AcknowledgmentsReferences |
|---|
2. Tobacco smoke and involuntary smoking. IARC Monogr Eval Carcinog Risks Hum 2004;83:1–1438.[Medline]
3. Hecht SSTobacco carcinogens, their biomarkers and tobacco-induced cancer. Nat Rev Cancer 2003;3:733–44.[CrossRef][Medline]
4. Mayne ST Lippman SMCigarettes: a smoking gun in cancer chemoprevention. J Natl Cancer Inst 2005;97:1319–21.
5. The -Tocopherol, β Carotene Cancer Prevention Study GroupThe effect of vitamin E and β carotene on the incidence of lung cancer and other cancers in male smokers. N Engl J Med 1994;330:1029–35.
6. Hamilton M, Wolf JL, Rusk J et al Effects of smoking on the pharmacokinetics of erlotinib. Clin Cancer Res 2006;12:2166–71.
7. Khuri FR, Lee JJ, Lippman SM et al Randomized phase III trial of low-dose isotretinoin for prevention of second primary tumors in stage I and II head and neck cancer patients. J Natl Cancer Inst 2006;98:441–50.
8. Fox JL, Rosenzweig KE, Ostroff JSThe effect of smoking status on survival following radiation therapy for non-small cell lung cancer. Lung Cancer 2004;44:287–93.[CrossRef][Medline]
9. Pantarotto J, Malone S, Dahrouge S, Gallant V, Eapen LSmoking is associated with worse outcomes in patients with prostate cancer treated by radical radiotherapy. BJU Int 2007;99:564–9.[CrossRef][Medline]
10. Shepherd FA, Rodrigues Pereira J, Ciuleanu T et al Erlotinib in previously treated non-small-cell lung cancer. N Engl J Med 2005;353:123–32.
11. Browman GP, Wong G, Hodson I et al Influence of cigarette smoking on the efficacy of radiation therapy in head and neck cancer. N Engl J Med 1993;328:159–63.
12. Guengerich FP Shimada TActivation of procarcinogens by human cytochrome P450 enzymes. Mutat Res 1998;400:201–13.[Medline]
13. Murray GI, Melvin WT, Greenlee WF, Burke MDRegulation, function, and tissue-specific expression of cytochrome P450 CYP1B1. Annu Rev Pharmacol Toxicol 2001;41:297–316.[CrossRef][Medline]
14. Ko Y, Abel J, Harth V et al Association of CYP1B1 codon 432 mutant allele in head and neck squamous cell cancer is reflected by somatic mutations of p53 in tumor tissue. Cancer Res 2001;61:4398–404.
15. Spira A, Beane J, Shah V et al Effects of cigarette smoke on the human airway epithelial cell transcriptome. Proc Natl Acad Sci U S A 2004;101:10143–8.
16. Sacks PGCell, tissue and organ culture as in vitro models to study the biology of squamous cell carcinomas of the head and neck. Cancer Metastasis Rev 1996;15:27–51.[CrossRef][Medline]
17. Moraitis D, Du B, De Lorenzo MS et al Levels of cyclooxygenase-2 are increased in the oral mucosa of smokers: evidence for the role of epidermal growth factor receptor and its ligands. Cancer Res 2005;65:664–70.
18. Choe SE, Boutros M, Michelson AM, Church GM, Halfon MSPreferred analysis methods for Affymetrix GeneChips revealed by a wholly defined control data set. Genome Biol 2005;6:R16[CrossRef][Medline]
19. Irizarry RA, Bolstad BM, Collin F et al Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Res 2003;31:e15
20. Wu Z, Irizarry RA, Gentleman RFM, Spencer FA model based background adjustment for oligonucleotide expression arrays. J Am Stat Assoc 2003;99:909–17.[CrossRef]
21. Shedden K, Chen W, Kuick R et al Comparison of seven methods for producing Affymetrix expression scores based on false discovery rates in disease profiling data. BMC Bioinformatics 2005;6:26[CrossRef][Medline]
22. Cope LM, Irizarry RA, Jaffee HA, Wu Z, Speed TPA benchmark for Affymetrix GeneChip expression measures. Bioinformatics 2004;20:323–31.
23. Millenaar FF, Okyere J, May ST et al How to decide? Different methods of calculating gene expression from short oligonucleotide array data will give different results. BMC Bioinformatics 2006;7:137[CrossRef][Medline]
24. Allison DB, Cui X, Page GP, Sabripour MMicroarray data analysis: from disarray to consolidation and consensus. Nat Rev Genet 2006;7:55–65.[CrossRef][Medline]
25. Benjamini Y Hochberg YControlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc B Met 1995;57:289–300.
26. Tusher VG, Tibshirani R, Chu GSignificance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci U S A 2001;98:5116–21.
27. Subramanian A, Tamayo P, Mootha VK et al Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 2005;102:15545–50.
28. Rhodes DR, Tomlins SA, Varambally S et al Probabilistic model of the human protein-protein interaction network. Nat Biotechnol 2005;23:951–9.[CrossRef][Medline]
29. Available from: http://www.ingenuity.com [cited 2007 Dec 15]
30. Hosack DA, Dennis G, Sherman BT, Lane HC, Lempicki RAIdentifying biological themes within lists of genes with EASE. Genome Biol 2003;4:R70[CrossRef][Medline]
31. Tomfohr J, Lu J, Kepler TBPathway level analysis of gene expression using singular value decomposition. BMC Bioinformatics 2005;6:225[CrossRef][Medline]
32. Barry WT, Nobel AB, Wright FASignificance analysis of functional categories in gene expression studies: a structured permutation approach. Bioinformatics 2005;21:1943–9.
33. Eisen MB, Spellman PT, Brown PO, Botstein DCluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci U S A 1998;95:14863–8.
34. Ma P, Castillo-Davis CI, Zhong W, Liu JSA data-driven clustering method for time course gene expression data. Nucleic Acids Res 2006;34:1261–9.
35. Tan MP, Broach JR, Floudas CAA novel clustering approach and prediction of optimal number of clusteris: I. Global optimum search with enhanced positioning. J Global Optimization 2007;39:323–46.[CrossRef]
36. Du B, Altorki NK, Kopelovich L, Subbaramaiah K, Dannenberg AJTobacco smoke stimulates the transcription of amphiregulin in human oral epithelial cells: evidence of a cyclic AMP-responsive element binding protein-dependent mechanism. Cancer Res 2005;65:5982–8.
37. Lemjabbar H, Li D, Gallup M et al Tobacco smoke-induced lung cell proliferation mediated by tumor necrosis factor -converting enzyme and amphiregulin. J Biol Chem 2003;278:26202–7.
38. Port JL, Yamaguchi K, Du B et al Tobacco smoke induces CYP1B1 in the aerodigestive tract. Carcinogenesis 2004;25:2275–81.
39. Du B, Leung H, Khan KM et al Tobacco smoke induces urokinase-type plasminogen activator and cell invasiveness: evidence for an epidermal growth factor receptor dependent mechanism. Cancer Res 2007;67:8966–72.
40. Peri S, Navarro JD, Amanchy R et al Development of human protein reference database as an initial platform for approaching systems biology in humans. Genome Res 2003;13:2363–71.
41. Brauze D, Widerak M, Cwykiel J, Szyfter K, Baer-Dubowska WThe effect of aryl hydrocarbon receptor ligands on the expression of AhR, AhRR, ARNT, Hif1, CYP1A1 and NQO1 genes in rat liver. Toxicol Lett 2006;167:212–20.[CrossRef][Medline]
42. Spivack SD, Hurteau GJ, Jain R et al Gene-environment interaction signatures by quantitative mRNA profiling in exfoliated buccal mucosal cells. Cancer Res 2004;64:6805–13.
43. Josephy PD, Batty SM, Boverhof DRRecombinant human P450 forms 1A1, 1A2, and 1B1 catalyze the bioactivation of heterocyclic amine mutagens in Escherichia coli lacZ strains. Environ Mol Mutagen 2001;38:12–8.[CrossRef][Medline]
44. Shimada T, Hayes CL, Yamazaki H et al Activation of chemically diverse procarcinogens by human cytochrome P-450 1B1. Cancer Res 1996;56:2979–84.
45. Denissenko MF, Pao A, Tang M, Pfeifer GPPreferential formation of benzo[a]pyrene adducts at lung cancer mutational hotspots in P53. Science 1996;274:430–2.
46. McFadyen MC, McLeod HL, Jackson FC et al Cytochrome P450 CYP1B1 protein expression: a novel mechanism of anticancer drug resistance. Biochem Pharmacol 2001;62:207–12.[CrossRef][Medline]
47. Shimizu Y, Nakatsuru Y, Ichinose M et al Benzo[a]pyrene carcinogenicity is lost in mice lacking the aryl hydrocarbon receptor. Proc Natl Acad Sci U S A 2000;97:779–82.
48. Andersson P, McGuire J, Rubio C et al A constitutively active dioxin/aryl hydrocarbon receptor induces stomach tumors. Proc Natl Acad Sci U S A 2002;99:9990–5.
49. Mimura J Fujii-Kuriyama YFunctional role of AhR in the expression of toxic effects by TCDD. Biochim Biophys Acta 2003;1619:263–8.[Medline]
50. Cohen SM Ellwein LBCell proliferation in carcinogenesis. Science 1990;249:1007–11.
51. van Oijen MG, Gilsing MM, Rijksen G, Hordijk GJ, Slootweg PJIncreased number of proliferating cells in oral epithelium from smokers and ex-smokers. Oral Oncol 1998;34:297–303.[CrossRef][Medline]
52. Luppi F, Aarbiou J, vanWetering S et al Effects of cigarette smoke condensate on proliferation and wound closure of bronchial epithelial cells in vitro: role of glutathione. Respir Res 2005;6:140[CrossRef][Medline]
53. Ballaz S Mulshine JLThe potential contributions of chronic inflammation to lung carcinogenesis. Clin Lung Cancer 2003;5:46–62.[Medline]
Commentary
| ||||||||||||||||||||||||||||||||||||||||||||
