Differential Gene Expression in Human Peripheral Blood Mononuclear Cells Induced by Cigarette Smoke and Its Constituents
http://www.100md.com
《毒物学科学杂志》
Department of Health Risk Analysis and Toxicology, Maastricht University, Maastricht, the Netherlands
ABSTRACT
In current molecular epidemiology studies, a wide range of methods are used to monitor early biological effects after exposure to xenobiotic agents. Gene expression profiling is considered a promising tool that may provide more sensitive, mechanism-based biomarkers. As a first step toward obtaining information on the applicability of gene expression profiles as a biomarker for early biological effects of carcinogen exposure, we conducted in vitro studies on human peripheral blood mononuclear cells (PBMC). We used cigarette smoke condensate (CSC) and a selection of its genotoxic constituents as model agents, applying cDNA microarray technology to investigate modulated gene expression. In independent experiments using cells from several donors, quiescent PBMC were exposed for 18 h, followed by gene expression analyses on a microarray containing 600 toxicologically relevant genes. The search for candidate biomarker genes was binomial: first we looked for genes responding similarly to all agents; second, for agent-specific genes. Many genes were significantly deregulated by all compounds, but as the direction of deregulation frequently differed per agent, they are not useful as generic biomarkers. Cigarette smoke condensate modulated the expression of many more genes than any of its constituents, with the largest effect in SERPINB2. The affected genes are involved in immune or stress responses, but surprisingly no genes involved in DNA damage response were modulated, and only a few in DNA repair. In conclusion, several genes have been identified as potential biomarkers for population studies on early biological effects caused by cigarette smoke exposure, but no genes were identified that represent a generic biomarker.
Key Words: gene expression profiling; cDNA microarray; peripheral blood mononuclear cells; cigarette smoke; biomarker early effect.
INTRODUCTION
Today, a range of biomonitoring methods is available for biomonitoring humans for the possible adverse effects caused by environmental exposure to hazardous compounds. Most of these methods have been developed to determine early biological effects due to exposure to xenobiotics. In general, these effects are detectable long before a diagnostic health effect appears (Bonassi and Au 2002; van Delft et al., 1998), but generally different methods have to be applied for each health effect and for each agent or class of agents. It would be of profound benefit to molecular epidemiology to develop methods that would give (mechanism-based) information on several health effects simultaneously, and that would be more suitable for monitoring effects at low exposure levels. Gene expression profiling is considered a promising tool to fulfill these needs. It is believed that the greatest potential for new biomarkers of early effect lies in toxicogenomics (Toraason et al., 2004), which can be defined as a field of study that integrates toxicology and genomics and examines how the entire genome responds to toxicants or other environmental hazards (Cunningham et al., 2003; Waters et al., 2003).
Chemical carcinogenesis by environmental and occupational pollutants, as well as by life-style factors like nutrition and tobacco smoke, is of major interest in molecular epidemiology. Very likely, cigarette smoke is the main external source for human exposure to carcinogens (Villard et al., 1998). Cigarette smoke contains over 4000 chemicals, including approximately 60 proven, probable, and possible carcinogens according to International Agency for Research on Cancer (IARC) standards (Smith et al., 2003). Cigarette smoke and a number of its chemical constituents are known to induce a broad range of (biomarkers of) genotoxic precarcinogenic effects—e.g., DNA adducts, micronuclei, chromosome aberrations and mutations—which have also been applied in molecular epidemiology studies using peripheral blood mononuclear cells (PBMC) as target cells (De Flora et al., 2003; Van Schooten et al., 1998).
In molecular epidemiology, target cells for toxic action are routinely not obtainable, because highly invasive procedures would be required. However, for more than 10 years, evidence has been available that accessible cells, such as PBMC, can be used as surrogate target cells to monitor effects in target tissues (Godschalk et al., 2000; Rockett et al., 2004). Thus, these surrogate target cells also seem to be best suited for a biomonitoring method based on gene expression profiling. Up until now, only a few studies on gene expression profiling in cells from exposed humans have been performed. Wu et al. (2003) described altered gene expressions in PBMC as result of exposure to arsenic, and their study demonstrated the potential of the technology to identify candidate biomarker genes. Recently, in a study on gene expression differences in peripheral leukocytes from human smokers versus non-smokers (Lampe et al., 2004), genes were identified that significantly correlated with plasma cotinine levels, and their expression profiles accurately distinguish smokers from non-smokers. Importantly, another recent study on radiation exposure found in vivo gene expression patterns in blood to be similar to previously observed in vitro patterns in quiescent lymphocytes (Amundson et al., 2004).
These human studies demonstrate the potential of gene expression profiling (in surrogate cells) in human biological monitoring studies. Furthermore, following an in vitro approach on gene expression profiling in human PBMC for the effects induced by ionizing radiation, Amundson et al. (2000) have suggested that estimates of environmental radiation exposure may be provided by gene expression analyses in PBMC.
The aim of the present study was to investigate in vitro the gene responses induced by genotoxic carcinogens present in cigarette smoke. To this end, we used cDNA microarray technology in a cellular model that best mimics the human in vivo situation as applied in biomonitoring studies (thus quiescent PBMC). Knowledge about these effects might aid in identifying potential biomarker genes for biomonitoring studies. Whether gene expression profiles are applicable for that purpose should be established in well-designed future studies on humans who have been exposed to genotoxic carcinogens.
MATERIALS AND METHODS
Study design.
The cell treatments were designed to represent the processes in vivo as much as possible. First, quiescent human PBMC were chosen rather than mitogen-activated cells. Human S9-mix was added to the treatment cultures. Because cigarette smoke compounds first pass the lung and only after absorption enter the blood and go through the liver, a mixture of S9 from human lung and liver was applied. Exceptions were made for treatments with H2O2, because S9 inactivates H2O2. Also, because smokers are chronically exposed to cigarette smoke and its compounds are deposited in the lung, we chose to apply a long exposure period (18 h) rather than a pulse treatment (4–6 h). The 18-h period seemed appropriate because other investigators have shown that, in general, deregulation of gene expression was larger and more genes were deregulated after approximately 24 h than after 4, 48, or 72 h (Amundson et al., 2000; Hu et al., 2004). Furthermore, in many in vitro studies, an 18–24-h exposure period is common for investigation of genotoxicity (Kirsch-Volders et al., 2003; Moore et al., 2002). Cigarette smoke condensate (CSC)–DNA adducts were measured to investigate the physiological relevance of the concentrations we applied in vitro, in comparison to adduct levels observed in smokers.
Test chemicals and materials.
All chemicals were purchased from Sigma-Aldrich (St. Louis, MO) unless stated otherwise. KCl and Mg2Cl were from Merck AG (Darmstadt, Germany). Glucose-6-phosphate was from Boehringer Mannheim (Indianapolis, IN), glucose-6-phosphate dehydrogenase and NADP were from Roche Diagnostics (Mannheim, Germany), L-glutamine, penicillin/streptomycin, and fetal calf serum were from Gibco (Paisley, UK). Trizol and RPMI1640 medium were from Life Technologies (Breda, the Netherlands), and Lymphoprep was from Nycomed (Oslo, Norway).
Chemical carcinogens used in this study were CSC, benzo(a)pyrene (BaP), 4-(methylnitrosamino)-1-(3-pyridinyl)-1-butanone (NNK, Toronto Research, North York, Canada), 4-aminobiphenyl (4-ABP), and hydrogen peroxide (H2O2). BaP, NNK, and 4-ABP are frequently investigated compounds representing three major classes of carcinogens present in cigarette smoke, i.e., polycyclic aromatic hydrocarbons, nitrosamines, and aromatic amines. Hydrogen peroxide was included as a surrogate for the oxidative stress induced by cigarette smoking. Cigarette smoke condensate was generated from 45 Kentucky 2R4F reference cigarettes on a 20-port Borgwaldt smoking machine under standardized ISO conditions, i.e., 35 ml per 2 sec puff, 1 puff/min (kindly provided by Dr. K. Rustemeier, INBIFO, Cologne, Germany; Rustemeier et al., 2002). The CSC stock solution was in dimethyl sulfoxide (DMSO) at 100 mg total particulate matter (TPM) per milliliter.
Composition of cigarette smoke condensate.
The composition of CSC with respect to polycyclic aromatic hydrocarbons was analyzed by high performance liquid chromatography (HPLC) according to the National Institute for Occupational Safety and Health (NIOSH) protocol issue 3 (Eller, 1984; NIOSH, 1998). The average concentration of BaP was 21.27 ng/ml (83 μM). The NNK and 4-ABP concentrations were determined by standard GC/MS (mass resolution 10,000) and were 161.0 ng/ml (770 μM) ± 11% and 57.33 ng/ml (338 μM) ± 32.2%, respectively.
Collection of blood samples.
Buffy coats of 500 ml whole blood from seven healthy non-smoking donors (age range: 20–50 years) were randomly obtained from Sanquin Blood Bank Limburg in Maastricht. Donors gave written informed consent. One buffy coat was sufficient for assessing gene expression profiles of two agents at three concentrations and a concomitant vehicle control.
PBMC isolation.
Human PBMC were isolated from buffy coat through gradient centrifugation using Lymphoprep in Leucosep filter tubes (Greiner Bio-One, Frickenhausen, Germany) according to the manufacturer's protocol. Finally, cells were resuspended in Roswell Park Memorial Institute (RPMI) 1640 medium, supplemented with 10% heat-inactivated fetal calf serum, L-glutamine (2 mM), and penicillin/streptomycin (penicillin 50 U/ml and streptomycin 50 μg/ml) at 6–8106 cells/ml. Cells were quiescent, meaning that no mitotic stimulation was applied.
Metabolic activation system.
S9 fractions of human liver and human lung were used for metabolic activation of the carcinogens. Pooled human liver S9 fraction from 15 donors (males and females) was purchased from In Vitro Technologies (Baltimore, MD). Human lung S9 fractions were prepared from healthy tissue of four different patients who had undergone lung tumor surgery at the Maastricht Academic Hospital. Liver and lung S9 fractions were mixed, with each covering 15% of the total S9 mix, which furthermore consisted of 32 μl 1M KCl, 31 μl 0.25 M MgCl26H2O, 24.3 μl 0.2 M glucose-6-phosphate, 26 μl 140 U/ml glucose-6-phosphate dehydrogenase, and 97.4 μl 0.04 M NADP with 204.5 μl MQ and 292 μl 1x (PBS) per milliliter (Gonzalez Borroto et al., 2001).
Toxicity testing.
Toxicity levels of the agents were determined by flow cytometry analysis. Briefly, PBMC were exposed for 18 h to the test agents in wide concentration ranges (CSC from 18 to 500 μg/ml, BaP from 0.1 to 100 μM, NNK from 2 to 2000 μM, 4-ABP from 0.1 to 100 μM, and H2O2 from 310–5 to 0.3% (v/v)), all in the presence (10% v/v) and absence of S9-mix, and at least in duplicate. After exposure, propidium iodide was added (20 μg/ml). Propidium iodide intercalates with DNA and only stains dead cells because of the membrane impermeability of viable cells. Samples were analyzed on a FACSort flow cytometer (Becton Dickinson, Heidelberg, Germany). Based on a forward-sideward scatter plot, 10,000 PBMC were analyzed for cytotoxicity (number of stained cells as percentage of total cell count).
In vitro exposure.
Peripheral blood mononuclear cells were exposed continuously for 18 h to three concentrations of each agent at 37°C in a total volume of 5 ml (6–8106 cells/ml) in 15 ml polypropylene conical tubes (Greiner BioOne). Cigarette smoke condensate, BaP, 4-ABP, and NNK exposures took place in the presence of S9 mix (10% v/v), H2O2 exposure without metabolic activation (S9 inactivates H2O2). Control cells were exposed to the vehicle solvents only; i.e., DMSO (CSC, BaP, 4-ABP, and NNK) or PBS (H2O2). Control cells always originated from the same buffy coat as the treated cells. After 18 h, tubes were chilled on ice and centrifuged at 250x g for 15 min. Cells were lyzed and homogenized in 1 ml Trizol/107 cells for RNA stabilization and subsequent RNA isolation. Independent treatments with cells from different donors were performed twice for the low and medium concentrations of the individual agents, and three times for the high concentrations of the individual agents, and for all concentrations of CSC. We believe that by investigating effects in cells from three different donors, a reasonable estimation is obtained of the gross effects that may occur in the human population. This conclusion is based on variations observed in comparable studies on PBMC in vitro (Amundson et al., 2000; Ryder et al., 2004).
Total RNA isolation.
Total RNA was isolated by a combination of two methods. First, total RNA was isolated using the Trizol method, according to the manufacturer's protocol. Next, RNA was purified using the RNeasy Mini Kit (Qiagen) according to the manufacturer's protocol. RNA was dissolved in diethylpyrocarbonate (DEPC)-treated, RNase-free, water. Purity was tested spectrophotometrically and considered suitable for further processing at 260/280 ratios of >1.7. Integrity was tested by lab-on-a-chip technology on the BioAnalyzer (Agilent, Palo Alto, CA), and RNA was considered to be intact when showing two distinct bands for the 18S and 28S ribosomal RNA.
cDNA labeling and microarray hybridization.
Reverse transcription was carried out on 10 μg of total RNA with minor modifications according to the TIGR protocol for aminoallyl labeling for microarrays (Hasseman, 2002). Finally, the test and accompanying control samples were mixed and dried in vacuo. As technical replicates, each test sample was hybridized twice against its concomitant vehicle control; once Cy5 labeled (control Cy3) and once Cy3 labeled (control Cy5). Samples were hybridized under a coverslip to spotted glass PHASE-1 Human 600 cDNA microarrays (PHASE-1 Molecular Toxicology, Santa Fe, NM) according to the manufacturer's protocol. These microarrays contain 597 toxicologically relevant human genes in quadruplicate (in a 2 x 2 design) representing several gene categories, such as apoptosis, cell proliferation, cell cycle, DNA damage and repair, inflammation, metabolism, oxidative stress, peroxisome proliferators, transport, and cell environment. Each gene is represented on the array in quadruplicate, which adds to a lower inter- and intra-assay variability. Samples were dissolved in 30 μl hybridization buffer containing 50% formamide, 5x SSC; 0.1% sodium dodecyl sulfate (SDS), and 100 μg/ml salmon sperm DNA. Samples were denatured for 5 min at 95°C, centrifuged at maximal speed for 3 min, and kept at 70°C until hybridization. Then, 25 μl was placed on the microarrays and incubated in hybridization chambers (Corning Inc., Corning, NY) overnight (18 h) in a water bath at 42°C. After hybridization, slides were dipped in 2x SSC at 34°C to remove the coverslips. Thereupon, slides were washed in 2x SSC/0.1% SDS, 0.1x SSC/0.1% SDS, and 0.1x SSC for 5 min each. Finally, slides were washed in 0.1x SSC at 32°C and spun to dry. Microarrays were scanned within 3 days and stored in a dark, dry environment.
Image and data analysis.
Microarrays were scanned on the Affymetrix GMS 418 scanner (Santa Clara, CA) at a fixed photomultiplier gain of 65 and a variable laser power, such that no spots were saturated. Cy5 and Cy3 fluorescent signals were scanned in separate channels. TIFF graphics were imported pairwise (Cy5 and Cy3) in ImaGene 5.0 (BioDiscovery, Marina del Rey, CA) for image analysis, resulting in raw data files with average pixel intensities for every spot and its local background. Poor spots were manually flagged (these are spots that are not well processed by ImaGene, generally due to severe blurring or contamination with a dust particle). Raw data files were imported in GeneSight 4.0 (BioDiscovery) for further data analysis. First, for every spot, local background was subtracted, and then manually flagged poor spots and spots with low expression levels (<5 mean pixel intensities) were eliminated. After log2 transformation of the remaining data, LOWESS normalization was performed on all genes with a smoothing factor of 0.2, and the difference of each test relative to concomitant vehicle control samples was calculated. Finally, replicate spots and arrays were averaged for each gene, omitting outliers with >2 standard deviations. The transformed, normalized data can be found online at http://www.grat.unimaas.nl/MAdata-DvLetal-TS2005.htm.
To identify differentially expressed genes, t-tests were performed on the difference data of all genes for each concentration of each agent relative to a set of four self-self hybridizations. In a self–self hybridization, two aliquots of the same RNA sample were hybridized against each other, which resulted in the estimation of the experimental variation. The t-tests were performed with Holm's p value adjustment, to reduce the number of false positives, with significance levels of p < 0.001, p < 0.01, and p < 0.05. Further t-tests were performed without Holm's adjustment. Also, for every agent, an analysis of variance (ANOVA) test was performed, including all concentrations and the self-self hybridizations, to reveal genes with dose–response relationships. For these analyses, the same restrictions regarding the Holm's adjustments and p values were maintained as described above. Because this is an initial exploratory study, we did not target the aspect of interindividual variations; this will be the subject of future in vivo investigations. Therefore, the data of all donors were pooled for each concentration and each agent.
Annotation and ontology analyses were done using the Database for Annotation, Visualization and Integrated Discovery (DAVID, http://apps1.niaid.nih.gov/David) (Dennis et al., 2003). Analyses of overrepresentation of specific biological pathways by selected genes compared to all genes present on the microarrays was conducted with the online tool Expression Analysis Systematic Explorer (http://david.niaid.nih.gov/david) (Hosack et al., 2003).
Quantitative real-time PCR.
Quantitative real-time polymerase chain reaction (PCR—qPCR) was performed to quantitate mRNA levels of a selection of genes in order to verify expression changes resulting from the microarray experiments. cDNA was prepared from 1 μg total RNA using random hexamer primers and SuperScript II reverse transcriptase (Invitrogen, Life Technologies) and used for real-time qPCR with the qPCR Core Kit for SYBR Green I, according to the manufacturer's instructions (Eurogentec, Seraing, Belgium). All PCR reactions were performed in duplicate. -actin was used as reference in order to normalize expression levels and to quantitate changes in gene expressions between the control and treated samples. The qPCR was run on the ABI Prism 7500 Sequence Detector (Applied BioSystems, Foster City, CA): 2 min at 50°C, 10 min at 95°C, and 40 cycles of 95°C for 15 s and 60°C for 1 min. The following forward and reverse primers were used (OPERON, 5'-3' sequences): IL1 CTGAGCTCGCCAGTGAAATG (forward) and TTTAGGGCCATCAGCTTCAAA (reverse), IL6 TCCAGGAGCCCAGCTATGAA (forward) and GAGCAGCCCCAGGGAGAA (reverse), TGM2 GGGCTCGGCCAAGTTCAT (forward) and TCTAGAAGGATCAGGCAGATGTCTAG (reverse), SERPINB2 GGGCTTTATCCTTTCCGTGTAA (forward) and TTAGCTTTTCACGCAAGTACATCAT (reverse) and -actin CCTGGCACCCAGCACAAT (forward) and GCCGATCCACACGGAGTACT (reverse). Dissociation curve analysis was performed and "no template controls" were analyzed to check for nonspecific products in the reaction. For each sample, the quantity was derived from Ct = Ct(target gene) – Ct(-actin reference). These values were log2 transformed, and the difference of each test relative to their concomitant vehicle control sample was calculated.
DNA-adduct analysis.
Aromatic DNA adduct levels were assessed by 32P-postlabeling. DNA was isolated from the CSC Trizol samples of three donors according to the manufacturer's instructions. The 32P-postlabeling procedure was conducted as originally described by Reddy and Randerath (1986) with minor modifications as described by Godschalk et al. (1998). Two DNA samples with known levels of modification (1 adduct per 107 and 108 nucleotides) were included in the analyses for quantitation purposes. Radiolabeled adduct nucleotide biphosphates were separated by two-dimensional chromatography on PEI-cellulose sheets (Machery Nagel, Germany), using solvent systems described by Godschalk et al. (1998). Chromatograms were visualized and quantified using a FLA-3000 Fuji Phosphor Imager (Fuji, Paris, France) with AIDA/2D Densitometry software.
RESULTS
Toxicity Tests
Concentration levels for the gene expression studies were based on cytotoxicity assessment by flow cytometry analysis of treated cells. Except for H2O2 in the absence of S9, none of the agents showed substantial toxicity at the tested concentrations as cytotoxicity levels were always <10%. For H2O2, 110–3 % (v/v) showed a cytotoxicity of 34% (mean of 4 donors, ranging from 17% to 83%) whereas 310–4% (v/v) showed 23% toxicity (4 donors, ranging from 4% to 68%) and 110–4% showed 3% toxicity (average 2 donors).
Gene Expression Analyses
Human PBMC were in vitro exposed to CSC, BaP, 4-ABP, NNK, or H2O2 at three concentrations, followed by analyses of altered mRNA levels by cDNA microarrays. Differentially expressed genes per agent and concentration (to be viewed online at http://www.grat.unimaas.nl/MAdata-DvLetal-TS2005.htm) were identified by t-testing against self-self hybridizations and for concentration-related effects of each agent by ANOVA of the three concentrations combined with self-self hybridizations. The numbers of significant gene expression modulations in the various tests are shown in Table 1.
To identify candidate biomarker genes suitable for human biomonitoring studies, two approaches were followed: We first looked for genes that were modulated by all agents, as these genes, when affected in the same direction by all agents, may be specific for genotoxic stress to PBMC without discriminating the type of agent. Figure 1 shows an adaptation to the three-ring Venn diagram, presenting the numbers of genes that were differentially expressed by CSC and one or more of the other agents. The intersection of all areas, the center of the diagram, contains 16 genes that were differentially expressed by all agents (at p < 0.05). Interleukin-6 (IL6), platelet/endothelial cell adhesion molecule-1 (PECAM1), and spermidine/spermine N1-acteyltransferase (SAT) were most significantly modulated (p < 0.01 without Holm's adjustment). A heat map presenting expression differences for all 16 genes is shown in Figure 2. None of the genes appears to be affected in the same direction by all agents, but the direction of deregulation differs per agent. Therefore, these genes appear not useful as general biomarker for exposure to genotoxic carcinogens.
Second, we searched for the most sensitive genes per agent, meaning that they responded at low or medium concentration, and with high significance (p < 0.05 with Holm's p value adjustment; see Table 1). This should yield agent-specific genes. The differential expression for these genes is shown in Figure 3. As is clear, CSC caused the largest effect on the gene expression profile. The effects induced by the other agents are less pronounced. For BaP, NNK, and H2O2 often a less than twofold induction or suppression is observed. Only 4-ABP caused clear concentration-related effects for several genes. Furthermore, the data show that the effects caused by CSC are far more complex and cannot be accounted for by a simple addition of the results for the other individual agents. For the agent-specific genes that showed a more than twofold difference (p < 0.05 with Holm's adjustment; Fig. 3), information about molecular function and biological process is shown in Table 2.
To validate expression changes resulting from the microarray experiments, real-time qPCR was performed to quantify mRNA levels for the main agent-specific deregulated genes, i.e., IL1, IL6, TGM2 and SERPINB2 for CSC, SERPINB2 for BaP, and IL1, IL6 for 4-ABP (Fig. 4). All these real-time-PCR data confirmed the microarray data.
DNA-Adduct Analysis
In addition to the gene expression analyses, PAH-DNA adduct levels were measured in the CSC-treated samples and the controls by 32P-postlabeling. In the exposed samples, elevated numbers of adducts were measured when compared to the control samples, showing a concentration–effect relationship (Table 3).
DISCUSSION
Our scope was to examine in vitro in a cellular model that best mimics the human in vivo situation as applied in biomonitoring studies (thus quiescent PBMC), the gene responses induced by genotoxic carcinogens present in cigarette smoke. To this end, we used a cDNA microarray technology. Knowledge about these effects may aid in identifying potential biomarker genes for which expression changes can be applied as a biomarker of early effect in biomonitoring studies. For this purpose, we evaluated effects of CSC and a selection of its major carcinogenic constituents on human quiescent PBMC in vitro. Here we report the induction of differential gene expression in this model and our approach to identifying candidate biomarker genes. The present study demonstrates that gene expression in PBMC can be significantly modulated in vitro by exposure to genotoxic chemicals. By exposing human PBMC in vitro to CSC, BaP, NNK, 4-ABP, and H2O2, we identified genes that were differentially expressed by all agents (the intersection for all agents from Figure 1), as well as differentially expressed genes for each agent individually (differentially expressed at the two lowest concentrations; see Figure 3). We hypothesize that these genes are most promising for future biomonitoring purposes in human population studies.
Genes that were modulated by all agents can be considered to represent a general response to genotoxic stress, as the genes responded to each agent investigated in this study. Thus, hypothetically, monitoring the expression levels for these (groups of) genes in a biomonitoring study may discriminate persons exposed to toxic agents from those who have not been exposed, without discriminating the type of toxicant. This requires that the deregulation must be robust and in the same direction for all agents. However, the heat map presenting expression changes for all 16 genes shows that this is not the case (Fig. 2). None of the genes appeared to be consistently affected in the same direction by all agents, but the direction of deregulation rather appeared to be agent-specific. Therefore, we conclude that these genes are not useful as a generic biomarker for exposure to genotoxic carcinogens in biomonitoring studies.
The genes differentially expressed at the lowest and/or medium concentrations for any agent can be regarded as the most relevant agent-specific genes. Of all agents studied, CSC induced alterations in the largest number of genes (Table 1, Fig. 3). For a number of genes, dose-related effects from low to high concentrations were found (e.g., for TGM2, ATF3, ENO1, and VEGF), whereas for other genes, constant expression changes were observed (CSF1R, SSP1, and SERPINB2). The latter finding may be due to saturation and would thus require exposure to lower concentrations of CSC to avoid this effect. For several genes, modulation of gene expression did not show a clear concentration–response relationship. Especially the highest concentration deviated, which in some cases caused a reversal of the expression change (e.g., for IL1, IL6, and PTGS2). As this reversal has been independently observed for PBMC from several donors, it does not reflect inaccuracy of the method. Furthermore, the qPCR analyses confirm the microarray data (compare Fig. 3, Fig. 4). Presumably, the change in the direction of deregulation is due to pre-cytotoxic processes that might have been induced by the highest CSC concentration, but that were not yet apparent in our cytotoxicity tests.
DNA-adduct levels in CSC-exposed cells showed a clear concentration-related effect. The DNA-adduct data confirm that the lowest CSC concentration was physiologically relevant, because the adduct level at this concentration is comparable to that in lymphocytes from smokers (Nia et al., 2000). Therefore, induced gene expression profiles can be considered physiologically relevant as well.
A response of the PBMC to CSC or BaP exposure that could have been expected in advance concerns the differential expression of genes regulated by the aryl hydrocarbon receptor (AhR). This receptor is known for its induction of several genes, such as cytochrome P450 1A1, 1A2, and 1B1, in response to exposure to these agents (Dertinger et al., 2001). In mitogen-activated lymphocytes, indeed CYP1A1 and CYP1B1 are inducible by TCDD, a very potent inducer of AhR activity (Landi et al., 2003). However, in quiescent lymphocytes this induction does not seem to occur, because the long-term presence of dioxin in the human body does not result in an increase in AhR pathway responsiveness (Landi et al., 2003). In our study, differential gene expression of AhR-related genes could not be demonstrated, which indicates that the AhR-signaling pathway is not very active in quiescent PBMC, a finding that agrees with the population study by Landi et al. (2003). In a comparable gene expression study in HepG2 cells, the AhR-related genes were activated by BaP (van Delft et al., 2004), thus showing that the technology is sufficiently sensitive. Furthermore, in our study DNA damage-response genes and DNA repair-related genes hardly showed modulations of their expression. None of the DNA damage-response genes regulated by p53 and induced in exposed HepG2 cells, such as BAX, CDKNA2, and GADD45 (van Delft et al., 2004), were significantly affected in PBMC. In contrast, CSC, BaP, and NNK exposed cells did show one significantly deregulated DNA repair gene each, namely XRCC1, XRCC5, and ERCC5, respectively (Fig. 3). In all cases, however, a downregulation was observed where upregulation would be expected, and the effects were marginal without obvious concentration-–response relationships.
From the subset of genes that most sensitively responded to the exposures, we specifically focused on the genes that showed at least a twofold (–1 expression difference 1; see Fig. 3) upregulation or downregulation. These are CSF1R, CXCL2, SPP1, SERPINB2, TGM2, PTGS2, ATF3, ACO1, ENO1 (for CSC), SERPINB2, PCK2 (for BaP), ERCC5, APG5L (for NNK), IL1, IL6, IL8, HAMP, SCP2, and ACO1 (for 4-ABP). Annotation analyses by DAVID (http://david.niaid.nih.gov/david) showed that 8 of these genes were involved in or associated with immune response, 10 in metabolism, 4 in apoptosis, 5 in cell proliferation, and 5 in signal transduction (see also Table 2). Analysis of overrepresentation of specific biological pathways in this selection of genes compared to all genes present on the microarrays by EASE (http://david.niaid.nih.gov/david), showed that biological processes involved in immune and defense response were most significantly affected (Fisher exact test p values 0.00031 and 0.00036). This emphasizes that in vitro exposure of PBMC to these agents particularly activates biological pathways that are involved in defense response implicating genes of the immune system. Whether this relates to the fact that we used PBMC, which are involved in immune response and therefore may be more responsive in these pathways, or alternatively that many genotoxicants are also immunotoxicants (Nordskog et al., 2003; van Grevenynghe et al., 2004), remains unanswered. The absence of modulation of DNA damage response pathways may indicate that rather nonspecific responses were induced, but it might also imply that DNA damage response pathways are not induced because quiescent cells were exposed. To investigate this matter further, it would be prudent to follow up our experiments with a selection of other classes of toxic agents (e.g., non-genotoxic agents) and with mitogen-stimulated PBMC.
EASE analyses per agent, showed that CSC significantly affected the immune/defense response (Fisher exact test p value 0.00016 to 0.004), and that 4-ABP affected immune/defense response and regulation of cell proliferation (p value 0.00016 to 0.002), whereas for BaP, NNK, and H2O2 no specific biological processes were significantly affected. It is remarkable that 4-ABP was consistently shown to influence expression of immune response–related genes, and that the change was in the opposite direction of that observed with CSC. It is likely that the suppressed expression has repercussions for the functionality of the cells. Yet, immunotoxicity of 4-ABP has not been reported. As such, these observations warrant a study of potential immunotoxicity of this agent. It is noteworthy that BaP failed to influence expression of immune response–related genes, whereas BaP is a known immunotoxic agent (De Jong et al., 1999). BaP possibly induces immunotoxicity through the Ah receptor, which is hardly expressed in lymphocytes, as discussed above for the absence of the AhR response.
The genes that were differentially expressed by a specific agent, especially at the lowest concentrations, can be considered to represent agent-specific responding genes (Fig. 3). Thus, hypothetically, monitoring the expression levels for these (groups of) genes in a biomonitoring study may discriminate persons exposed to the specific agents from those who are not exposed. Taking into account that only CSC and 4-ABP caused concentration-related effects for several genes, possibly only gene expression analyses related to exposure to these agents might be sufficiently achievable.
Taking into account the microarray and qPCR data, we believe that IL1, IL6, TGM2, and SERPINB2 are the most promising potential biomarker genes for effects caused by cigarette smoke exposure, where the effects on SERPINB2 may be ascribed to BaP or possibly other polycyclic aromatic hydrocarbons present in cigarette smoke. As mentioned earlier, it seems not possible to identify (profiles of) genes that can be used as general biomarker for genotoxic agent exposure, but before reaching a final conclusion on this, these in vitro experiments should be followed by studies on humans who are heavily exposed to cigarette smoke. Therefore, our current study must be seen as an initial, exploratory step in the development of gene expression profiles as biomarker of early effect.
To date, several other in vitro and in vivo studies on gene expression profiling in PBMC following exposure to a carcinogenic agents like cigarette smoke and ionizing radiation have been published (Amundson et al., 2000, 2004; Lampe et al., 2004; Ryder et al., 2004; Wu et al., 2003). These studies show that expressions of genes involved in inflammation or immune response were frequently modulated. Also, a study has been published on acute inflammatory response in human bronchial epithelial cells, a target cell for cigarette smoke exposure, which describes increased expression of IL1, IL6, and IL8 induced by in vitro exposure to CSC (Hellermann et al., 2002). This substantiates our observations in PBMC in vitro.
For SERPINB2, which has plasminogen activation regulation as its main function, induction by TCDD exposure has been observed in monocytic U937 cells (Gohl et al., 1996). TGM2 catalyzes protein cross linking, and its activity has been shown to be increased in apoptotic hepatocytes (Fesus et al., 1987), suggesting a role for the gene in regulation of cell death. However, we cannot address the relevance of this for our findings.
In our study, we used PBMC as surrogate targets cells for effects that may occur in the true targets organs in vivo, as this is also the approach in biomonitoring studies of human populations. In human volunteers it is, a priori, not possible to investigate biological effects in cells from target organs like lung and liver (Godschalk et al., 2000; Rockett et al., 2004). A critical question is whether effects in PBMC at gene expression level are comparable to effects in target organs. To our knowledge, this has not yet been investigated in humans or animals.
In summary, our study demonstrates that cigarette smoke condensate and its constituents caused gene expression changes in human PBMC in vitro. The identified differentially expressed genes have the potential to become marker genes for population studies on biological effects caused by cigarette smoke exposure. However, these findings need to be substantiated in vivo.
SUPPLEMENTARY DATA
Supplementary data are available online at www.toxsci.oupjournals.org.
ACKNOWLEDGMENTS
The authors thank Dr. K. Rustemeier (INBIFO, Cologne, Germany) for kindly generating and providing the cigarette smoke condensate, Dr. B. Schutte (Department of Genetics, Maastricht University, Netherlands) for assisting on flow cytometry, and Prof. H. van Loveren (Department of Health Risk Analysis & Toxicology, Maastricht University, Netherlands) for his comments. This study was supported by the research program "Environment and Health" of the Flemish Government (Belgium).
REFERENCES
Amundson, S. A., Do, K. T., Shahab, S., Bittner, M., Meltzer, P., Trent, J., Fornace, A. J., Jr. (2000). Identification of potential mRNA biomarkers in peripheral blood lymphocytes for human exposure to ionizing radiation. Radiat. Res. 154, 342–346.
Amundson, S. A., Grace, M. B., McLeland, C. B., Epperly, M. W., Yeager, A., Zhan, Q., Greenberger, J. S., and Fornace, A. J., Jr. (2004). Human in vivo radiation-induced biomarkers: Gene expression changes in radiotherapy patients. Cancer Res. 64, 6368–6371.
Bonassi, S., and Au, W. W. (2002). Biomarkers in molecular epidemiology studies for health risk prediction. Mutat. Res. 511, 73–86.
Cunningham, M. L., Bogdanffy, M. S., Zacharewski, T. R., and Hines, R. N. (2003). Workshop overview: Use of genomic data in risk assessment. Toxicol. Sci. 73, 209–215.
De Flora, S., D'Agostini, F., Balansky, R., Camoirano, A., Bennicelli, C., Bagnasco, M., Cartiglia, C., Tampa, E., Longobardi, M. G., Lubet, R. A. et al. (2003). Modulation of cigarette smoke–related end-points in mutagenesis and carcinogenesis. Mutat. Res. 523/524, 237–252.
De Jong, W. H., Kroese, E. D., Vos, J. G., and Van Loveren, H. (1999). Detection of immunotoxicity of benzo[a]pyrene in a subacute toxicity study after oral exposure in rats. Toxicol. Sci. 50, 214–220.
Dennis, G., Jr., Sherman, B. T., Hosack, D. A., Yang, J., Gao, W., Lane, H. C., and Lempicki, R. A. (2003). DAVID: Database for Annotation, Visualization, and Integrated Discovery. Genome Biol. 4, P3.
Dertinger, S. D., Nazarenko, D. A., Silverstone, A. E., and Gasiewicz, T. A. (2001). Aryl hydrocarbon receptor signaling plays a significant role in mediating benzo[a]pyrene- and cigarette smoke condensate-induced cytogenetic damage in vivo. Carcinogenesis 22, 171–177.
Eller, P. M. (1984). NIOSH Manual of Analytical Methods (Vol. 2). Cincinnati, OH, National Institute for Occupational Safety and Health, USA.
Fesus, L., Thomazy, V., and Falus, A. (1987). Induction and activation of tissue transglutaminase during programmed cell death. FEBS Lett. 224, 104–108.
Godschalk, R. W., Maas, L. M., Van Zandwijk, N., van't Veer, L. J., Breedijk, A., Borm, P. J., Verhaert, J., Kleinjans, J. C., and van Schooten, F. J. (1998). Differences in aromatic-DNA adduct levels between alveolar macrophages and subpopulations of white blood cells from smokers. Carcinogenesis 19, 819–825.
Godschalk, R. W., Moonen, E. J., Schilderman, P. A., Broekmans, W. M., Kleinjans, J. C., and Van Schooten, F. J. (2000). Exposure-route-dependent DNA adduct formation by polycyclic aromatic hydrocarbons. Carcinogenesis 21, 87–92.
Gohl, G., Lehmkoster, T., Munzel, P. A., Schrenk, D., Viebahn, R., and Bock, K. W. (1996). TCDD-inducible plasminogen activator inhibitor type 2 (PAI-2) in human hepatocytes, HepG2 and monocytic U937 cells. Carcinogenesis 17, 443–449.
Gonzalez Borroto, J. I., Creus, A., and Marcos, R. (2001). Genotoxic evaluation of the furylethylene derivative 2-furyl-1-nitroethene in cultured human lymphocytes. Mutat. Res. 497, 177–184.
Hasseman, J. (2002). Standard operating procedure: aminoallyl labeling of RNA for microarrays (eukaryotic). The Institute for Genomic Research. Available: http://pga.tigr.org/sop/M004_1a.pdf [effective date 3/4/02].
Hellermann, G. R., Nagy, S. B., Kong, X., Lockey, R. F., and Mohapatra, S. S. (2002). Mechanism of cigarette smoke condensate-induced acute inflammatory response in human bronchial epithelial cells. Respir. Res. 3, 22.
Hosack, D. A., Dennis, G., Jr., Sherman, B. T., Lane, H. C., and Lempicki, R. A. (2003). Identifying biological themes within lists of genes with EASE. Genome Biol. 4, R70.
Hu, T., Gibson, D. P., Carr, G. J., Torontali, S. M., Tiesman, J. P., Chaney, J. G., and Aardema, M. J. (2004). Identification of a gene expression profile that discriminates indirect-acting genotoxins from direct-acting genotoxins. Mutat. Res. 549, 5–27.
Kirsch-Volders, M., Sofuni, T., Aardema, M., Albertini, S., Eastmond, D., Fenech, M., Ishidate, M., Jr., Kirchner, S., Lorge, E., Morita, T. et al. (2003). Report from the In Vitro Micronucleus Assay Working Group. Mutat. Res. 540, 153–163.
Lampe, J. W., Stepaniants, S. B., Mao, M., Radich, J. P., Dai, H., Linsley, P. S., Friend, S. H., and Potter, J. D. (2004). Signatures of environmental exposures using peripheral leukocyte gene expression: Tobacco smoke. Cancer Epidemiol. Biomarkers Prev. 13, 445–453.
Landi, M. T., Bertazzi, P. A., Baccarelli, A., Consonni, D., Masten, S., Lucier, G., Mocarelli, P., Needham, L., Caporaso, N., and Grassman, J. (2003). TCDD-mediated alterations in the AhR-dependent pathway in Seveso, Italy, 20 years after the accident. Carcinogenesis 24, 673–680.
Moore, M. M., Honma, M., Clements, J., Harrington-Brock, K., Awogi, T., Bolcsfoldi, G., Cifone, M., Collard, D., Fellows, M., Flanders, K. et al. (2002). Mouse lymphoma thymidine kinase gene mutation assay: Follow-up International Workshop on Genotoxicity Test Procedures, New Orleans, Louisiana, April 2000. Environ. Mol. Mutagen. 40, 292–299.
Nia, A. B., Maas, L. M., Brouwer, E. M., Kleinjans, J. C., and Van Schooten, F. J. (2000). Comparison between smoking-related DNA adduct analysis in induced sputum and peripheral blood lymphocytes. Carcinogenesis 21, 1335–1340.
NIOSH (National Institute for Occupational Safety and Health). (1998). Polynuclear aromatic hydrocarbons by HPLC (issue 3), Washington, DC, The National Institute for Occupational Safety and Health, USA.
Nordskog, B. K., Blixt, A. D., Morgan, W. T., Fields, W. R., and Hellmann, G. M. (2003). Matrix-degrading and pro-inflammatory changes in human vascular endothelial cells exposed to cigarette smoke condensate. Cardiovasc. Toxicol. 3, 101–117.
Reddy, M. V., and Randerath K. (1986). Nuclease P1-mediated enhancement of sensitivity of 32P-postlabeling test for structurally diverse DNA adducts. Carcinogenesis 7, 1543–1551.
Rockett, J. C., Burczynski, M. E., Fornace, A. J., Herrmann, P. C., Krawetz, S. A., and Dix, D. J. (2004). Surrogate tissue analysis: Monitoring toxicant exposure and health status of inaccessible tissues through the analysis of accessible tissues and cells. Toxicol. Appl. Pharmacol. 194, 189–199.
Rustemeier, K., Stabbert, R., Haussmann, H. J., Roemer, E., and Carmines, E. L. (2002). Evaluation of the potential effects of ingredients added to cigarettes. Part 2: Chemical composition of mainstream smoke. Food Chem. Toxicol. 40, 93–104.
Ryder, M. I., Hyun, W., Loomer, P., and Haqq, C. (2004). Alteration of gene expression profiles of peripheral mononuclear blood cells by tobacco smoke: Implications for periodontal diseases. Oral Microbiol. Immunol. 19, 39–49.
Smith, C. J., Perfetti, T. A., Garg, R., and Hansch, C. (2003). IARC carcinogens reported in cigarette mainstream smoke and their calculated log P values. Food Chem. Toxicol. 41, 807–817.
Toraason, M., Albertini, R., Bayard, S., Bigbee, W., Blair, A., Boffetta, P., Bonassi, S., Chanock, S., Christiani, D., Eastmond, D. et al. (2004). Applying new biotechnologies to the study of occupational cancer—A workshop summary. Environ. Health Perspect. 112, 413–416.
van Delft, J. H., Baan, R. A., and Roza, L. (1998). Biological effect markers for exposure to carcinogenic compounds and their relevance for risk assessment. Crit. Rev. Toxicol. 28, 477–510.
van Delft, J. H., van Agen, E., van Breda, S. G., Herwijnen, M. H., Staal, Y. C., and Kleinjans, J. C. (2004). Discrimination of genotoxic from non-genotoxic carcinogens by gene expression profiling. Carcinogenesis 25, 1265–1276.
van Grevenynghe, J., Sparfel, L., Le Vee, M., Gilot, D., Drenou, B., Fauchet, R., and Fardel, O. (2004). Cytochrome P450-dependent toxicity of environmental polycyclic aromatic hydrocarbons towards human macrophages. Biochem. Biophys. Res. Commun. 317, 708–716.
Van Schooten, F. J., Hirvonen, A., Maas, L. M., De Mol, B. A., Kleinjans, J. C., Bell, D. A., and Durrer, J. D. (1998). Putative susceptibility markers of coronary artery disease: Association between VDR genotype, smoking, and aromatic DNA adduct levels in human right atrial tissue. FASEB J. 12, 1409–1417.
Villard, P. H., Seree, E. M., Re, J. L., De Meo, M., Barra, Y., Attolini, L., Dumenil, G., Catalin, J., Durand, A., and Lacarelle, B. (1998). Effects of tobacco smoke on the gene expression of the Cyp1a, Cyp2b, Cyp2e, and Cyp3a subfamilies in mouse liver and lung: Relation to single strand breaks of DNA. Toxicol. Appl. Pharmacol. 148, 195–204.
Waters, M. D., Olden, K., and Tennant, R. W. (2003). Toxicogenomic approach for assessing toxicant-related disease. Mutat. Res. 544, 415–424.
Wu, M. M., Chiou, H. Y., Ho, I. C., Chen, C. J., and Lee, T. C. (2003). Gene expression of inflammatory molecules in circulating lymphocytes from arsenic-exposed human subjects. Environ. Health Perspect. 111, 1429–1438.(Danitsja M. van Leeuwen, )
ABSTRACT
In current molecular epidemiology studies, a wide range of methods are used to monitor early biological effects after exposure to xenobiotic agents. Gene expression profiling is considered a promising tool that may provide more sensitive, mechanism-based biomarkers. As a first step toward obtaining information on the applicability of gene expression profiles as a biomarker for early biological effects of carcinogen exposure, we conducted in vitro studies on human peripheral blood mononuclear cells (PBMC). We used cigarette smoke condensate (CSC) and a selection of its genotoxic constituents as model agents, applying cDNA microarray technology to investigate modulated gene expression. In independent experiments using cells from several donors, quiescent PBMC were exposed for 18 h, followed by gene expression analyses on a microarray containing 600 toxicologically relevant genes. The search for candidate biomarker genes was binomial: first we looked for genes responding similarly to all agents; second, for agent-specific genes. Many genes were significantly deregulated by all compounds, but as the direction of deregulation frequently differed per agent, they are not useful as generic biomarkers. Cigarette smoke condensate modulated the expression of many more genes than any of its constituents, with the largest effect in SERPINB2. The affected genes are involved in immune or stress responses, but surprisingly no genes involved in DNA damage response were modulated, and only a few in DNA repair. In conclusion, several genes have been identified as potential biomarkers for population studies on early biological effects caused by cigarette smoke exposure, but no genes were identified that represent a generic biomarker.
Key Words: gene expression profiling; cDNA microarray; peripheral blood mononuclear cells; cigarette smoke; biomarker early effect.
INTRODUCTION
Today, a range of biomonitoring methods is available for biomonitoring humans for the possible adverse effects caused by environmental exposure to hazardous compounds. Most of these methods have been developed to determine early biological effects due to exposure to xenobiotics. In general, these effects are detectable long before a diagnostic health effect appears (Bonassi and Au 2002; van Delft et al., 1998), but generally different methods have to be applied for each health effect and for each agent or class of agents. It would be of profound benefit to molecular epidemiology to develop methods that would give (mechanism-based) information on several health effects simultaneously, and that would be more suitable for monitoring effects at low exposure levels. Gene expression profiling is considered a promising tool to fulfill these needs. It is believed that the greatest potential for new biomarkers of early effect lies in toxicogenomics (Toraason et al., 2004), which can be defined as a field of study that integrates toxicology and genomics and examines how the entire genome responds to toxicants or other environmental hazards (Cunningham et al., 2003; Waters et al., 2003).
Chemical carcinogenesis by environmental and occupational pollutants, as well as by life-style factors like nutrition and tobacco smoke, is of major interest in molecular epidemiology. Very likely, cigarette smoke is the main external source for human exposure to carcinogens (Villard et al., 1998). Cigarette smoke contains over 4000 chemicals, including approximately 60 proven, probable, and possible carcinogens according to International Agency for Research on Cancer (IARC) standards (Smith et al., 2003). Cigarette smoke and a number of its chemical constituents are known to induce a broad range of (biomarkers of) genotoxic precarcinogenic effects—e.g., DNA adducts, micronuclei, chromosome aberrations and mutations—which have also been applied in molecular epidemiology studies using peripheral blood mononuclear cells (PBMC) as target cells (De Flora et al., 2003; Van Schooten et al., 1998).
In molecular epidemiology, target cells for toxic action are routinely not obtainable, because highly invasive procedures would be required. However, for more than 10 years, evidence has been available that accessible cells, such as PBMC, can be used as surrogate target cells to monitor effects in target tissues (Godschalk et al., 2000; Rockett et al., 2004). Thus, these surrogate target cells also seem to be best suited for a biomonitoring method based on gene expression profiling. Up until now, only a few studies on gene expression profiling in cells from exposed humans have been performed. Wu et al. (2003) described altered gene expressions in PBMC as result of exposure to arsenic, and their study demonstrated the potential of the technology to identify candidate biomarker genes. Recently, in a study on gene expression differences in peripheral leukocytes from human smokers versus non-smokers (Lampe et al., 2004), genes were identified that significantly correlated with plasma cotinine levels, and their expression profiles accurately distinguish smokers from non-smokers. Importantly, another recent study on radiation exposure found in vivo gene expression patterns in blood to be similar to previously observed in vitro patterns in quiescent lymphocytes (Amundson et al., 2004).
These human studies demonstrate the potential of gene expression profiling (in surrogate cells) in human biological monitoring studies. Furthermore, following an in vitro approach on gene expression profiling in human PBMC for the effects induced by ionizing radiation, Amundson et al. (2000) have suggested that estimates of environmental radiation exposure may be provided by gene expression analyses in PBMC.
The aim of the present study was to investigate in vitro the gene responses induced by genotoxic carcinogens present in cigarette smoke. To this end, we used cDNA microarray technology in a cellular model that best mimics the human in vivo situation as applied in biomonitoring studies (thus quiescent PBMC). Knowledge about these effects might aid in identifying potential biomarker genes for biomonitoring studies. Whether gene expression profiles are applicable for that purpose should be established in well-designed future studies on humans who have been exposed to genotoxic carcinogens.
MATERIALS AND METHODS
Study design.
The cell treatments were designed to represent the processes in vivo as much as possible. First, quiescent human PBMC were chosen rather than mitogen-activated cells. Human S9-mix was added to the treatment cultures. Because cigarette smoke compounds first pass the lung and only after absorption enter the blood and go through the liver, a mixture of S9 from human lung and liver was applied. Exceptions were made for treatments with H2O2, because S9 inactivates H2O2. Also, because smokers are chronically exposed to cigarette smoke and its compounds are deposited in the lung, we chose to apply a long exposure period (18 h) rather than a pulse treatment (4–6 h). The 18-h period seemed appropriate because other investigators have shown that, in general, deregulation of gene expression was larger and more genes were deregulated after approximately 24 h than after 4, 48, or 72 h (Amundson et al., 2000; Hu et al., 2004). Furthermore, in many in vitro studies, an 18–24-h exposure period is common for investigation of genotoxicity (Kirsch-Volders et al., 2003; Moore et al., 2002). Cigarette smoke condensate (CSC)–DNA adducts were measured to investigate the physiological relevance of the concentrations we applied in vitro, in comparison to adduct levels observed in smokers.
Test chemicals and materials.
All chemicals were purchased from Sigma-Aldrich (St. Louis, MO) unless stated otherwise. KCl and Mg2Cl were from Merck AG (Darmstadt, Germany). Glucose-6-phosphate was from Boehringer Mannheim (Indianapolis, IN), glucose-6-phosphate dehydrogenase and NADP were from Roche Diagnostics (Mannheim, Germany), L-glutamine, penicillin/streptomycin, and fetal calf serum were from Gibco (Paisley, UK). Trizol and RPMI1640 medium were from Life Technologies (Breda, the Netherlands), and Lymphoprep was from Nycomed (Oslo, Norway).
Chemical carcinogens used in this study were CSC, benzo(a)pyrene (BaP), 4-(methylnitrosamino)-1-(3-pyridinyl)-1-butanone (NNK, Toronto Research, North York, Canada), 4-aminobiphenyl (4-ABP), and hydrogen peroxide (H2O2). BaP, NNK, and 4-ABP are frequently investigated compounds representing three major classes of carcinogens present in cigarette smoke, i.e., polycyclic aromatic hydrocarbons, nitrosamines, and aromatic amines. Hydrogen peroxide was included as a surrogate for the oxidative stress induced by cigarette smoking. Cigarette smoke condensate was generated from 45 Kentucky 2R4F reference cigarettes on a 20-port Borgwaldt smoking machine under standardized ISO conditions, i.e., 35 ml per 2 sec puff, 1 puff/min (kindly provided by Dr. K. Rustemeier, INBIFO, Cologne, Germany; Rustemeier et al., 2002). The CSC stock solution was in dimethyl sulfoxide (DMSO) at 100 mg total particulate matter (TPM) per milliliter.
Composition of cigarette smoke condensate.
The composition of CSC with respect to polycyclic aromatic hydrocarbons was analyzed by high performance liquid chromatography (HPLC) according to the National Institute for Occupational Safety and Health (NIOSH) protocol issue 3 (Eller, 1984; NIOSH, 1998). The average concentration of BaP was 21.27 ng/ml (83 μM). The NNK and 4-ABP concentrations were determined by standard GC/MS (mass resolution 10,000) and were 161.0 ng/ml (770 μM) ± 11% and 57.33 ng/ml (338 μM) ± 32.2%, respectively.
Collection of blood samples.
Buffy coats of 500 ml whole blood from seven healthy non-smoking donors (age range: 20–50 years) were randomly obtained from Sanquin Blood Bank Limburg in Maastricht. Donors gave written informed consent. One buffy coat was sufficient for assessing gene expression profiles of two agents at three concentrations and a concomitant vehicle control.
PBMC isolation.
Human PBMC were isolated from buffy coat through gradient centrifugation using Lymphoprep in Leucosep filter tubes (Greiner Bio-One, Frickenhausen, Germany) according to the manufacturer's protocol. Finally, cells were resuspended in Roswell Park Memorial Institute (RPMI) 1640 medium, supplemented with 10% heat-inactivated fetal calf serum, L-glutamine (2 mM), and penicillin/streptomycin (penicillin 50 U/ml and streptomycin 50 μg/ml) at 6–8106 cells/ml. Cells were quiescent, meaning that no mitotic stimulation was applied.
Metabolic activation system.
S9 fractions of human liver and human lung were used for metabolic activation of the carcinogens. Pooled human liver S9 fraction from 15 donors (males and females) was purchased from In Vitro Technologies (Baltimore, MD). Human lung S9 fractions were prepared from healthy tissue of four different patients who had undergone lung tumor surgery at the Maastricht Academic Hospital. Liver and lung S9 fractions were mixed, with each covering 15% of the total S9 mix, which furthermore consisted of 32 μl 1M KCl, 31 μl 0.25 M MgCl26H2O, 24.3 μl 0.2 M glucose-6-phosphate, 26 μl 140 U/ml glucose-6-phosphate dehydrogenase, and 97.4 μl 0.04 M NADP with 204.5 μl MQ and 292 μl 1x (PBS) per milliliter (Gonzalez Borroto et al., 2001).
Toxicity testing.
Toxicity levels of the agents were determined by flow cytometry analysis. Briefly, PBMC were exposed for 18 h to the test agents in wide concentration ranges (CSC from 18 to 500 μg/ml, BaP from 0.1 to 100 μM, NNK from 2 to 2000 μM, 4-ABP from 0.1 to 100 μM, and H2O2 from 310–5 to 0.3% (v/v)), all in the presence (10% v/v) and absence of S9-mix, and at least in duplicate. After exposure, propidium iodide was added (20 μg/ml). Propidium iodide intercalates with DNA and only stains dead cells because of the membrane impermeability of viable cells. Samples were analyzed on a FACSort flow cytometer (Becton Dickinson, Heidelberg, Germany). Based on a forward-sideward scatter plot, 10,000 PBMC were analyzed for cytotoxicity (number of stained cells as percentage of total cell count).
In vitro exposure.
Peripheral blood mononuclear cells were exposed continuously for 18 h to three concentrations of each agent at 37°C in a total volume of 5 ml (6–8106 cells/ml) in 15 ml polypropylene conical tubes (Greiner BioOne). Cigarette smoke condensate, BaP, 4-ABP, and NNK exposures took place in the presence of S9 mix (10% v/v), H2O2 exposure without metabolic activation (S9 inactivates H2O2). Control cells were exposed to the vehicle solvents only; i.e., DMSO (CSC, BaP, 4-ABP, and NNK) or PBS (H2O2). Control cells always originated from the same buffy coat as the treated cells. After 18 h, tubes were chilled on ice and centrifuged at 250x g for 15 min. Cells were lyzed and homogenized in 1 ml Trizol/107 cells for RNA stabilization and subsequent RNA isolation. Independent treatments with cells from different donors were performed twice for the low and medium concentrations of the individual agents, and three times for the high concentrations of the individual agents, and for all concentrations of CSC. We believe that by investigating effects in cells from three different donors, a reasonable estimation is obtained of the gross effects that may occur in the human population. This conclusion is based on variations observed in comparable studies on PBMC in vitro (Amundson et al., 2000; Ryder et al., 2004).
Total RNA isolation.
Total RNA was isolated by a combination of two methods. First, total RNA was isolated using the Trizol method, according to the manufacturer's protocol. Next, RNA was purified using the RNeasy Mini Kit (Qiagen) according to the manufacturer's protocol. RNA was dissolved in diethylpyrocarbonate (DEPC)-treated, RNase-free, water. Purity was tested spectrophotometrically and considered suitable for further processing at 260/280 ratios of >1.7. Integrity was tested by lab-on-a-chip technology on the BioAnalyzer (Agilent, Palo Alto, CA), and RNA was considered to be intact when showing two distinct bands for the 18S and 28S ribosomal RNA.
cDNA labeling and microarray hybridization.
Reverse transcription was carried out on 10 μg of total RNA with minor modifications according to the TIGR protocol for aminoallyl labeling for microarrays (Hasseman, 2002). Finally, the test and accompanying control samples were mixed and dried in vacuo. As technical replicates, each test sample was hybridized twice against its concomitant vehicle control; once Cy5 labeled (control Cy3) and once Cy3 labeled (control Cy5). Samples were hybridized under a coverslip to spotted glass PHASE-1 Human 600 cDNA microarrays (PHASE-1 Molecular Toxicology, Santa Fe, NM) according to the manufacturer's protocol. These microarrays contain 597 toxicologically relevant human genes in quadruplicate (in a 2 x 2 design) representing several gene categories, such as apoptosis, cell proliferation, cell cycle, DNA damage and repair, inflammation, metabolism, oxidative stress, peroxisome proliferators, transport, and cell environment. Each gene is represented on the array in quadruplicate, which adds to a lower inter- and intra-assay variability. Samples were dissolved in 30 μl hybridization buffer containing 50% formamide, 5x SSC; 0.1% sodium dodecyl sulfate (SDS), and 100 μg/ml salmon sperm DNA. Samples were denatured for 5 min at 95°C, centrifuged at maximal speed for 3 min, and kept at 70°C until hybridization. Then, 25 μl was placed on the microarrays and incubated in hybridization chambers (Corning Inc., Corning, NY) overnight (18 h) in a water bath at 42°C. After hybridization, slides were dipped in 2x SSC at 34°C to remove the coverslips. Thereupon, slides were washed in 2x SSC/0.1% SDS, 0.1x SSC/0.1% SDS, and 0.1x SSC for 5 min each. Finally, slides were washed in 0.1x SSC at 32°C and spun to dry. Microarrays were scanned within 3 days and stored in a dark, dry environment.
Image and data analysis.
Microarrays were scanned on the Affymetrix GMS 418 scanner (Santa Clara, CA) at a fixed photomultiplier gain of 65 and a variable laser power, such that no spots were saturated. Cy5 and Cy3 fluorescent signals were scanned in separate channels. TIFF graphics were imported pairwise (Cy5 and Cy3) in ImaGene 5.0 (BioDiscovery, Marina del Rey, CA) for image analysis, resulting in raw data files with average pixel intensities for every spot and its local background. Poor spots were manually flagged (these are spots that are not well processed by ImaGene, generally due to severe blurring or contamination with a dust particle). Raw data files were imported in GeneSight 4.0 (BioDiscovery) for further data analysis. First, for every spot, local background was subtracted, and then manually flagged poor spots and spots with low expression levels (<5 mean pixel intensities) were eliminated. After log2 transformation of the remaining data, LOWESS normalization was performed on all genes with a smoothing factor of 0.2, and the difference of each test relative to concomitant vehicle control samples was calculated. Finally, replicate spots and arrays were averaged for each gene, omitting outliers with >2 standard deviations. The transformed, normalized data can be found online at http://www.grat.unimaas.nl/MAdata-DvLetal-TS2005.htm.
To identify differentially expressed genes, t-tests were performed on the difference data of all genes for each concentration of each agent relative to a set of four self-self hybridizations. In a self–self hybridization, two aliquots of the same RNA sample were hybridized against each other, which resulted in the estimation of the experimental variation. The t-tests were performed with Holm's p value adjustment, to reduce the number of false positives, with significance levels of p < 0.001, p < 0.01, and p < 0.05. Further t-tests were performed without Holm's adjustment. Also, for every agent, an analysis of variance (ANOVA) test was performed, including all concentrations and the self-self hybridizations, to reveal genes with dose–response relationships. For these analyses, the same restrictions regarding the Holm's adjustments and p values were maintained as described above. Because this is an initial exploratory study, we did not target the aspect of interindividual variations; this will be the subject of future in vivo investigations. Therefore, the data of all donors were pooled for each concentration and each agent.
Annotation and ontology analyses were done using the Database for Annotation, Visualization and Integrated Discovery (DAVID, http://apps1.niaid.nih.gov/David) (Dennis et al., 2003). Analyses of overrepresentation of specific biological pathways by selected genes compared to all genes present on the microarrays was conducted with the online tool Expression Analysis Systematic Explorer (http://david.niaid.nih.gov/david) (Hosack et al., 2003).
Quantitative real-time PCR.
Quantitative real-time polymerase chain reaction (PCR—qPCR) was performed to quantitate mRNA levels of a selection of genes in order to verify expression changes resulting from the microarray experiments. cDNA was prepared from 1 μg total RNA using random hexamer primers and SuperScript II reverse transcriptase (Invitrogen, Life Technologies) and used for real-time qPCR with the qPCR Core Kit for SYBR Green I, according to the manufacturer's instructions (Eurogentec, Seraing, Belgium). All PCR reactions were performed in duplicate. -actin was used as reference in order to normalize expression levels and to quantitate changes in gene expressions between the control and treated samples. The qPCR was run on the ABI Prism 7500 Sequence Detector (Applied BioSystems, Foster City, CA): 2 min at 50°C, 10 min at 95°C, and 40 cycles of 95°C for 15 s and 60°C for 1 min. The following forward and reverse primers were used (OPERON, 5'-3' sequences): IL1 CTGAGCTCGCCAGTGAAATG (forward) and TTTAGGGCCATCAGCTTCAAA (reverse), IL6 TCCAGGAGCCCAGCTATGAA (forward) and GAGCAGCCCCAGGGAGAA (reverse), TGM2 GGGCTCGGCCAAGTTCAT (forward) and TCTAGAAGGATCAGGCAGATGTCTAG (reverse), SERPINB2 GGGCTTTATCCTTTCCGTGTAA (forward) and TTAGCTTTTCACGCAAGTACATCAT (reverse) and -actin CCTGGCACCCAGCACAAT (forward) and GCCGATCCACACGGAGTACT (reverse). Dissociation curve analysis was performed and "no template controls" were analyzed to check for nonspecific products in the reaction. For each sample, the quantity was derived from Ct = Ct(target gene) – Ct(-actin reference). These values were log2 transformed, and the difference of each test relative to their concomitant vehicle control sample was calculated.
DNA-adduct analysis.
Aromatic DNA adduct levels were assessed by 32P-postlabeling. DNA was isolated from the CSC Trizol samples of three donors according to the manufacturer's instructions. The 32P-postlabeling procedure was conducted as originally described by Reddy and Randerath (1986) with minor modifications as described by Godschalk et al. (1998). Two DNA samples with known levels of modification (1 adduct per 107 and 108 nucleotides) were included in the analyses for quantitation purposes. Radiolabeled adduct nucleotide biphosphates were separated by two-dimensional chromatography on PEI-cellulose sheets (Machery Nagel, Germany), using solvent systems described by Godschalk et al. (1998). Chromatograms were visualized and quantified using a FLA-3000 Fuji Phosphor Imager (Fuji, Paris, France) with AIDA/2D Densitometry software.
RESULTS
Toxicity Tests
Concentration levels for the gene expression studies were based on cytotoxicity assessment by flow cytometry analysis of treated cells. Except for H2O2 in the absence of S9, none of the agents showed substantial toxicity at the tested concentrations as cytotoxicity levels were always <10%. For H2O2, 110–3 % (v/v) showed a cytotoxicity of 34% (mean of 4 donors, ranging from 17% to 83%) whereas 310–4% (v/v) showed 23% toxicity (4 donors, ranging from 4% to 68%) and 110–4% showed 3% toxicity (average 2 donors).
Gene Expression Analyses
Human PBMC were in vitro exposed to CSC, BaP, 4-ABP, NNK, or H2O2 at three concentrations, followed by analyses of altered mRNA levels by cDNA microarrays. Differentially expressed genes per agent and concentration (to be viewed online at http://www.grat.unimaas.nl/MAdata-DvLetal-TS2005.htm) were identified by t-testing against self-self hybridizations and for concentration-related effects of each agent by ANOVA of the three concentrations combined with self-self hybridizations. The numbers of significant gene expression modulations in the various tests are shown in Table 1.
To identify candidate biomarker genes suitable for human biomonitoring studies, two approaches were followed: We first looked for genes that were modulated by all agents, as these genes, when affected in the same direction by all agents, may be specific for genotoxic stress to PBMC without discriminating the type of agent. Figure 1 shows an adaptation to the three-ring Venn diagram, presenting the numbers of genes that were differentially expressed by CSC and one or more of the other agents. The intersection of all areas, the center of the diagram, contains 16 genes that were differentially expressed by all agents (at p < 0.05). Interleukin-6 (IL6), platelet/endothelial cell adhesion molecule-1 (PECAM1), and spermidine/spermine N1-acteyltransferase (SAT) were most significantly modulated (p < 0.01 without Holm's adjustment). A heat map presenting expression differences for all 16 genes is shown in Figure 2. None of the genes appears to be affected in the same direction by all agents, but the direction of deregulation differs per agent. Therefore, these genes appear not useful as general biomarker for exposure to genotoxic carcinogens.
Second, we searched for the most sensitive genes per agent, meaning that they responded at low or medium concentration, and with high significance (p < 0.05 with Holm's p value adjustment; see Table 1). This should yield agent-specific genes. The differential expression for these genes is shown in Figure 3. As is clear, CSC caused the largest effect on the gene expression profile. The effects induced by the other agents are less pronounced. For BaP, NNK, and H2O2 often a less than twofold induction or suppression is observed. Only 4-ABP caused clear concentration-related effects for several genes. Furthermore, the data show that the effects caused by CSC are far more complex and cannot be accounted for by a simple addition of the results for the other individual agents. For the agent-specific genes that showed a more than twofold difference (p < 0.05 with Holm's adjustment; Fig. 3), information about molecular function and biological process is shown in Table 2.
To validate expression changes resulting from the microarray experiments, real-time qPCR was performed to quantify mRNA levels for the main agent-specific deregulated genes, i.e., IL1, IL6, TGM2 and SERPINB2 for CSC, SERPINB2 for BaP, and IL1, IL6 for 4-ABP (Fig. 4). All these real-time-PCR data confirmed the microarray data.
DNA-Adduct Analysis
In addition to the gene expression analyses, PAH-DNA adduct levels were measured in the CSC-treated samples and the controls by 32P-postlabeling. In the exposed samples, elevated numbers of adducts were measured when compared to the control samples, showing a concentration–effect relationship (Table 3).
DISCUSSION
Our scope was to examine in vitro in a cellular model that best mimics the human in vivo situation as applied in biomonitoring studies (thus quiescent PBMC), the gene responses induced by genotoxic carcinogens present in cigarette smoke. To this end, we used a cDNA microarray technology. Knowledge about these effects may aid in identifying potential biomarker genes for which expression changes can be applied as a biomarker of early effect in biomonitoring studies. For this purpose, we evaluated effects of CSC and a selection of its major carcinogenic constituents on human quiescent PBMC in vitro. Here we report the induction of differential gene expression in this model and our approach to identifying candidate biomarker genes. The present study demonstrates that gene expression in PBMC can be significantly modulated in vitro by exposure to genotoxic chemicals. By exposing human PBMC in vitro to CSC, BaP, NNK, 4-ABP, and H2O2, we identified genes that were differentially expressed by all agents (the intersection for all agents from Figure 1), as well as differentially expressed genes for each agent individually (differentially expressed at the two lowest concentrations; see Figure 3). We hypothesize that these genes are most promising for future biomonitoring purposes in human population studies.
Genes that were modulated by all agents can be considered to represent a general response to genotoxic stress, as the genes responded to each agent investigated in this study. Thus, hypothetically, monitoring the expression levels for these (groups of) genes in a biomonitoring study may discriminate persons exposed to toxic agents from those who have not been exposed, without discriminating the type of toxicant. This requires that the deregulation must be robust and in the same direction for all agents. However, the heat map presenting expression changes for all 16 genes shows that this is not the case (Fig. 2). None of the genes appeared to be consistently affected in the same direction by all agents, but the direction of deregulation rather appeared to be agent-specific. Therefore, we conclude that these genes are not useful as a generic biomarker for exposure to genotoxic carcinogens in biomonitoring studies.
The genes differentially expressed at the lowest and/or medium concentrations for any agent can be regarded as the most relevant agent-specific genes. Of all agents studied, CSC induced alterations in the largest number of genes (Table 1, Fig. 3). For a number of genes, dose-related effects from low to high concentrations were found (e.g., for TGM2, ATF3, ENO1, and VEGF), whereas for other genes, constant expression changes were observed (CSF1R, SSP1, and SERPINB2). The latter finding may be due to saturation and would thus require exposure to lower concentrations of CSC to avoid this effect. For several genes, modulation of gene expression did not show a clear concentration–response relationship. Especially the highest concentration deviated, which in some cases caused a reversal of the expression change (e.g., for IL1, IL6, and PTGS2). As this reversal has been independently observed for PBMC from several donors, it does not reflect inaccuracy of the method. Furthermore, the qPCR analyses confirm the microarray data (compare Fig. 3, Fig. 4). Presumably, the change in the direction of deregulation is due to pre-cytotoxic processes that might have been induced by the highest CSC concentration, but that were not yet apparent in our cytotoxicity tests.
DNA-adduct levels in CSC-exposed cells showed a clear concentration-related effect. The DNA-adduct data confirm that the lowest CSC concentration was physiologically relevant, because the adduct level at this concentration is comparable to that in lymphocytes from smokers (Nia et al., 2000). Therefore, induced gene expression profiles can be considered physiologically relevant as well.
A response of the PBMC to CSC or BaP exposure that could have been expected in advance concerns the differential expression of genes regulated by the aryl hydrocarbon receptor (AhR). This receptor is known for its induction of several genes, such as cytochrome P450 1A1, 1A2, and 1B1, in response to exposure to these agents (Dertinger et al., 2001). In mitogen-activated lymphocytes, indeed CYP1A1 and CYP1B1 are inducible by TCDD, a very potent inducer of AhR activity (Landi et al., 2003). However, in quiescent lymphocytes this induction does not seem to occur, because the long-term presence of dioxin in the human body does not result in an increase in AhR pathway responsiveness (Landi et al., 2003). In our study, differential gene expression of AhR-related genes could not be demonstrated, which indicates that the AhR-signaling pathway is not very active in quiescent PBMC, a finding that agrees with the population study by Landi et al. (2003). In a comparable gene expression study in HepG2 cells, the AhR-related genes were activated by BaP (van Delft et al., 2004), thus showing that the technology is sufficiently sensitive. Furthermore, in our study DNA damage-response genes and DNA repair-related genes hardly showed modulations of their expression. None of the DNA damage-response genes regulated by p53 and induced in exposed HepG2 cells, such as BAX, CDKNA2, and GADD45 (van Delft et al., 2004), were significantly affected in PBMC. In contrast, CSC, BaP, and NNK exposed cells did show one significantly deregulated DNA repair gene each, namely XRCC1, XRCC5, and ERCC5, respectively (Fig. 3). In all cases, however, a downregulation was observed where upregulation would be expected, and the effects were marginal without obvious concentration-–response relationships.
From the subset of genes that most sensitively responded to the exposures, we specifically focused on the genes that showed at least a twofold (–1 expression difference 1; see Fig. 3) upregulation or downregulation. These are CSF1R, CXCL2, SPP1, SERPINB2, TGM2, PTGS2, ATF3, ACO1, ENO1 (for CSC), SERPINB2, PCK2 (for BaP), ERCC5, APG5L (for NNK), IL1, IL6, IL8, HAMP, SCP2, and ACO1 (for 4-ABP). Annotation analyses by DAVID (http://david.niaid.nih.gov/david) showed that 8 of these genes were involved in or associated with immune response, 10 in metabolism, 4 in apoptosis, 5 in cell proliferation, and 5 in signal transduction (see also Table 2). Analysis of overrepresentation of specific biological pathways in this selection of genes compared to all genes present on the microarrays by EASE (http://david.niaid.nih.gov/david), showed that biological processes involved in immune and defense response were most significantly affected (Fisher exact test p values 0.00031 and 0.00036). This emphasizes that in vitro exposure of PBMC to these agents particularly activates biological pathways that are involved in defense response implicating genes of the immune system. Whether this relates to the fact that we used PBMC, which are involved in immune response and therefore may be more responsive in these pathways, or alternatively that many genotoxicants are also immunotoxicants (Nordskog et al., 2003; van Grevenynghe et al., 2004), remains unanswered. The absence of modulation of DNA damage response pathways may indicate that rather nonspecific responses were induced, but it might also imply that DNA damage response pathways are not induced because quiescent cells were exposed. To investigate this matter further, it would be prudent to follow up our experiments with a selection of other classes of toxic agents (e.g., non-genotoxic agents) and with mitogen-stimulated PBMC.
EASE analyses per agent, showed that CSC significantly affected the immune/defense response (Fisher exact test p value 0.00016 to 0.004), and that 4-ABP affected immune/defense response and regulation of cell proliferation (p value 0.00016 to 0.002), whereas for BaP, NNK, and H2O2 no specific biological processes were significantly affected. It is remarkable that 4-ABP was consistently shown to influence expression of immune response–related genes, and that the change was in the opposite direction of that observed with CSC. It is likely that the suppressed expression has repercussions for the functionality of the cells. Yet, immunotoxicity of 4-ABP has not been reported. As such, these observations warrant a study of potential immunotoxicity of this agent. It is noteworthy that BaP failed to influence expression of immune response–related genes, whereas BaP is a known immunotoxic agent (De Jong et al., 1999). BaP possibly induces immunotoxicity through the Ah receptor, which is hardly expressed in lymphocytes, as discussed above for the absence of the AhR response.
The genes that were differentially expressed by a specific agent, especially at the lowest concentrations, can be considered to represent agent-specific responding genes (Fig. 3). Thus, hypothetically, monitoring the expression levels for these (groups of) genes in a biomonitoring study may discriminate persons exposed to the specific agents from those who are not exposed. Taking into account that only CSC and 4-ABP caused concentration-related effects for several genes, possibly only gene expression analyses related to exposure to these agents might be sufficiently achievable.
Taking into account the microarray and qPCR data, we believe that IL1, IL6, TGM2, and SERPINB2 are the most promising potential biomarker genes for effects caused by cigarette smoke exposure, where the effects on SERPINB2 may be ascribed to BaP or possibly other polycyclic aromatic hydrocarbons present in cigarette smoke. As mentioned earlier, it seems not possible to identify (profiles of) genes that can be used as general biomarker for genotoxic agent exposure, but before reaching a final conclusion on this, these in vitro experiments should be followed by studies on humans who are heavily exposed to cigarette smoke. Therefore, our current study must be seen as an initial, exploratory step in the development of gene expression profiles as biomarker of early effect.
To date, several other in vitro and in vivo studies on gene expression profiling in PBMC following exposure to a carcinogenic agents like cigarette smoke and ionizing radiation have been published (Amundson et al., 2000, 2004; Lampe et al., 2004; Ryder et al., 2004; Wu et al., 2003). These studies show that expressions of genes involved in inflammation or immune response were frequently modulated. Also, a study has been published on acute inflammatory response in human bronchial epithelial cells, a target cell for cigarette smoke exposure, which describes increased expression of IL1, IL6, and IL8 induced by in vitro exposure to CSC (Hellermann et al., 2002). This substantiates our observations in PBMC in vitro.
For SERPINB2, which has plasminogen activation regulation as its main function, induction by TCDD exposure has been observed in monocytic U937 cells (Gohl et al., 1996). TGM2 catalyzes protein cross linking, and its activity has been shown to be increased in apoptotic hepatocytes (Fesus et al., 1987), suggesting a role for the gene in regulation of cell death. However, we cannot address the relevance of this for our findings.
In our study, we used PBMC as surrogate targets cells for effects that may occur in the true targets organs in vivo, as this is also the approach in biomonitoring studies of human populations. In human volunteers it is, a priori, not possible to investigate biological effects in cells from target organs like lung and liver (Godschalk et al., 2000; Rockett et al., 2004). A critical question is whether effects in PBMC at gene expression level are comparable to effects in target organs. To our knowledge, this has not yet been investigated in humans or animals.
In summary, our study demonstrates that cigarette smoke condensate and its constituents caused gene expression changes in human PBMC in vitro. The identified differentially expressed genes have the potential to become marker genes for population studies on biological effects caused by cigarette smoke exposure. However, these findings need to be substantiated in vivo.
SUPPLEMENTARY DATA
Supplementary data are available online at www.toxsci.oupjournals.org.
ACKNOWLEDGMENTS
The authors thank Dr. K. Rustemeier (INBIFO, Cologne, Germany) for kindly generating and providing the cigarette smoke condensate, Dr. B. Schutte (Department of Genetics, Maastricht University, Netherlands) for assisting on flow cytometry, and Prof. H. van Loveren (Department of Health Risk Analysis & Toxicology, Maastricht University, Netherlands) for his comments. This study was supported by the research program "Environment and Health" of the Flemish Government (Belgium).
REFERENCES
Amundson, S. A., Do, K. T., Shahab, S., Bittner, M., Meltzer, P., Trent, J., Fornace, A. J., Jr. (2000). Identification of potential mRNA biomarkers in peripheral blood lymphocytes for human exposure to ionizing radiation. Radiat. Res. 154, 342–346.
Amundson, S. A., Grace, M. B., McLeland, C. B., Epperly, M. W., Yeager, A., Zhan, Q., Greenberger, J. S., and Fornace, A. J., Jr. (2004). Human in vivo radiation-induced biomarkers: Gene expression changes in radiotherapy patients. Cancer Res. 64, 6368–6371.
Bonassi, S., and Au, W. W. (2002). Biomarkers in molecular epidemiology studies for health risk prediction. Mutat. Res. 511, 73–86.
Cunningham, M. L., Bogdanffy, M. S., Zacharewski, T. R., and Hines, R. N. (2003). Workshop overview: Use of genomic data in risk assessment. Toxicol. Sci. 73, 209–215.
De Flora, S., D'Agostini, F., Balansky, R., Camoirano, A., Bennicelli, C., Bagnasco, M., Cartiglia, C., Tampa, E., Longobardi, M. G., Lubet, R. A. et al. (2003). Modulation of cigarette smoke–related end-points in mutagenesis and carcinogenesis. Mutat. Res. 523/524, 237–252.
De Jong, W. H., Kroese, E. D., Vos, J. G., and Van Loveren, H. (1999). Detection of immunotoxicity of benzo[a]pyrene in a subacute toxicity study after oral exposure in rats. Toxicol. Sci. 50, 214–220.
Dennis, G., Jr., Sherman, B. T., Hosack, D. A., Yang, J., Gao, W., Lane, H. C., and Lempicki, R. A. (2003). DAVID: Database for Annotation, Visualization, and Integrated Discovery. Genome Biol. 4, P3.
Dertinger, S. D., Nazarenko, D. A., Silverstone, A. E., and Gasiewicz, T. A. (2001). Aryl hydrocarbon receptor signaling plays a significant role in mediating benzo[a]pyrene- and cigarette smoke condensate-induced cytogenetic damage in vivo. Carcinogenesis 22, 171–177.
Eller, P. M. (1984). NIOSH Manual of Analytical Methods (Vol. 2). Cincinnati, OH, National Institute for Occupational Safety and Health, USA.
Fesus, L., Thomazy, V., and Falus, A. (1987). Induction and activation of tissue transglutaminase during programmed cell death. FEBS Lett. 224, 104–108.
Godschalk, R. W., Maas, L. M., Van Zandwijk, N., van't Veer, L. J., Breedijk, A., Borm, P. J., Verhaert, J., Kleinjans, J. C., and van Schooten, F. J. (1998). Differences in aromatic-DNA adduct levels between alveolar macrophages and subpopulations of white blood cells from smokers. Carcinogenesis 19, 819–825.
Godschalk, R. W., Moonen, E. J., Schilderman, P. A., Broekmans, W. M., Kleinjans, J. C., and Van Schooten, F. J. (2000). Exposure-route-dependent DNA adduct formation by polycyclic aromatic hydrocarbons. Carcinogenesis 21, 87–92.
Gohl, G., Lehmkoster, T., Munzel, P. A., Schrenk, D., Viebahn, R., and Bock, K. W. (1996). TCDD-inducible plasminogen activator inhibitor type 2 (PAI-2) in human hepatocytes, HepG2 and monocytic U937 cells. Carcinogenesis 17, 443–449.
Gonzalez Borroto, J. I., Creus, A., and Marcos, R. (2001). Genotoxic evaluation of the furylethylene derivative 2-furyl-1-nitroethene in cultured human lymphocytes. Mutat. Res. 497, 177–184.
Hasseman, J. (2002). Standard operating procedure: aminoallyl labeling of RNA for microarrays (eukaryotic). The Institute for Genomic Research. Available: http://pga.tigr.org/sop/M004_1a.pdf [effective date 3/4/02].
Hellermann, G. R., Nagy, S. B., Kong, X., Lockey, R. F., and Mohapatra, S. S. (2002). Mechanism of cigarette smoke condensate-induced acute inflammatory response in human bronchial epithelial cells. Respir. Res. 3, 22.
Hosack, D. A., Dennis, G., Jr., Sherman, B. T., Lane, H. C., and Lempicki, R. A. (2003). Identifying biological themes within lists of genes with EASE. Genome Biol. 4, R70.
Hu, T., Gibson, D. P., Carr, G. J., Torontali, S. M., Tiesman, J. P., Chaney, J. G., and Aardema, M. J. (2004). Identification of a gene expression profile that discriminates indirect-acting genotoxins from direct-acting genotoxins. Mutat. Res. 549, 5–27.
Kirsch-Volders, M., Sofuni, T., Aardema, M., Albertini, S., Eastmond, D., Fenech, M., Ishidate, M., Jr., Kirchner, S., Lorge, E., Morita, T. et al. (2003). Report from the In Vitro Micronucleus Assay Working Group. Mutat. Res. 540, 153–163.
Lampe, J. W., Stepaniants, S. B., Mao, M., Radich, J. P., Dai, H., Linsley, P. S., Friend, S. H., and Potter, J. D. (2004). Signatures of environmental exposures using peripheral leukocyte gene expression: Tobacco smoke. Cancer Epidemiol. Biomarkers Prev. 13, 445–453.
Landi, M. T., Bertazzi, P. A., Baccarelli, A., Consonni, D., Masten, S., Lucier, G., Mocarelli, P., Needham, L., Caporaso, N., and Grassman, J. (2003). TCDD-mediated alterations in the AhR-dependent pathway in Seveso, Italy, 20 years after the accident. Carcinogenesis 24, 673–680.
Moore, M. M., Honma, M., Clements, J., Harrington-Brock, K., Awogi, T., Bolcsfoldi, G., Cifone, M., Collard, D., Fellows, M., Flanders, K. et al. (2002). Mouse lymphoma thymidine kinase gene mutation assay: Follow-up International Workshop on Genotoxicity Test Procedures, New Orleans, Louisiana, April 2000. Environ. Mol. Mutagen. 40, 292–299.
Nia, A. B., Maas, L. M., Brouwer, E. M., Kleinjans, J. C., and Van Schooten, F. J. (2000). Comparison between smoking-related DNA adduct analysis in induced sputum and peripheral blood lymphocytes. Carcinogenesis 21, 1335–1340.
NIOSH (National Institute for Occupational Safety and Health). (1998). Polynuclear aromatic hydrocarbons by HPLC (issue 3), Washington, DC, The National Institute for Occupational Safety and Health, USA.
Nordskog, B. K., Blixt, A. D., Morgan, W. T., Fields, W. R., and Hellmann, G. M. (2003). Matrix-degrading and pro-inflammatory changes in human vascular endothelial cells exposed to cigarette smoke condensate. Cardiovasc. Toxicol. 3, 101–117.
Reddy, M. V., and Randerath K. (1986). Nuclease P1-mediated enhancement of sensitivity of 32P-postlabeling test for structurally diverse DNA adducts. Carcinogenesis 7, 1543–1551.
Rockett, J. C., Burczynski, M. E., Fornace, A. J., Herrmann, P. C., Krawetz, S. A., and Dix, D. J. (2004). Surrogate tissue analysis: Monitoring toxicant exposure and health status of inaccessible tissues through the analysis of accessible tissues and cells. Toxicol. Appl. Pharmacol. 194, 189–199.
Rustemeier, K., Stabbert, R., Haussmann, H. J., Roemer, E., and Carmines, E. L. (2002). Evaluation of the potential effects of ingredients added to cigarettes. Part 2: Chemical composition of mainstream smoke. Food Chem. Toxicol. 40, 93–104.
Ryder, M. I., Hyun, W., Loomer, P., and Haqq, C. (2004). Alteration of gene expression profiles of peripheral mononuclear blood cells by tobacco smoke: Implications for periodontal diseases. Oral Microbiol. Immunol. 19, 39–49.
Smith, C. J., Perfetti, T. A., Garg, R., and Hansch, C. (2003). IARC carcinogens reported in cigarette mainstream smoke and their calculated log P values. Food Chem. Toxicol. 41, 807–817.
Toraason, M., Albertini, R., Bayard, S., Bigbee, W., Blair, A., Boffetta, P., Bonassi, S., Chanock, S., Christiani, D., Eastmond, D. et al. (2004). Applying new biotechnologies to the study of occupational cancer—A workshop summary. Environ. Health Perspect. 112, 413–416.
van Delft, J. H., Baan, R. A., and Roza, L. (1998). Biological effect markers for exposure to carcinogenic compounds and their relevance for risk assessment. Crit. Rev. Toxicol. 28, 477–510.
van Delft, J. H., van Agen, E., van Breda, S. G., Herwijnen, M. H., Staal, Y. C., and Kleinjans, J. C. (2004). Discrimination of genotoxic from non-genotoxic carcinogens by gene expression profiling. Carcinogenesis 25, 1265–1276.
van Grevenynghe, J., Sparfel, L., Le Vee, M., Gilot, D., Drenou, B., Fauchet, R., and Fardel, O. (2004). Cytochrome P450-dependent toxicity of environmental polycyclic aromatic hydrocarbons towards human macrophages. Biochem. Biophys. Res. Commun. 317, 708–716.
Van Schooten, F. J., Hirvonen, A., Maas, L. M., De Mol, B. A., Kleinjans, J. C., Bell, D. A., and Durrer, J. D. (1998). Putative susceptibility markers of coronary artery disease: Association between VDR genotype, smoking, and aromatic DNA adduct levels in human right atrial tissue. FASEB J. 12, 1409–1417.
Villard, P. H., Seree, E. M., Re, J. L., De Meo, M., Barra, Y., Attolini, L., Dumenil, G., Catalin, J., Durand, A., and Lacarelle, B. (1998). Effects of tobacco smoke on the gene expression of the Cyp1a, Cyp2b, Cyp2e, and Cyp3a subfamilies in mouse liver and lung: Relation to single strand breaks of DNA. Toxicol. Appl. Pharmacol. 148, 195–204.
Waters, M. D., Olden, K., and Tennant, R. W. (2003). Toxicogenomic approach for assessing toxicant-related disease. Mutat. Res. 544, 415–424.
Wu, M. M., Chiou, H. Y., Ho, I. C., Chen, C. J., and Lee, T. C. (2003). Gene expression of inflammatory molecules in circulating lymphocytes from arsenic-exposed human subjects. Environ. Health Perspect. 111, 1429–1438.(Danitsja M. van Leeuwen, )