Mutagenesis Advance Access originally published online on December 6, 2006
Mutagenesis 2007 22(1):55-62; doi:10.1093/mutage/gel058
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Modulation of gene expression and DNA-adduct formation in precision-cut liver slices exposed to polycyclic aromatic hydrocarbons of different carcinogenic potency
1 Department of Health Risk Analysis and Toxicology, Maastricht University PO Box 616 6200 MD Maastricht, The Netherlands 2 Molecular Toxicology Group, School of Biomedical and Molecular Sciences, University of Surrey Guildford, Surrey, GU2 7XH, UK
Polycyclic aromatic hydrocarbons (PAHs) differ markedly in their carcinogenic potencies. Differences in transcriptomic responses upon PAH exposures might improve our current understanding of the differences in carcinogenicity, and therefore gene expression modulation by six PAHs in precision-cut rat liver slices was investigated. Gene expression modulation by benzo[a]pyrene (B[a]P), dibenzo[a,l]pyrene (DB[a,l]P), benzo[b]fluoranthene (B[b]F), fluoranthene (FA), dibenzo[a,h]anthracene (DB[a,h]A) and 1-methylphenanthrene (1-MPA) was assessed after 6- (B[a]P, DB[a,l]P) and 24-h (all compounds) exposure, using oligonucleotide arrays. DNA-adduct formation was determined using 32P-post-labelling. The effects of PAHs on gene expression and on DNA-adduct formation were much more pronounced after 24-h exposure than after a 6-h exposure. Each compound induced gene expression changes dose-dependently and gene expression profiles were generally compound-specific. B[a]P, B[b]F and DB[a,h]A displayed comparable gene expression profiles, and so did DB[a,l]P, FA and 1-MPA. Only the carcinogenic PAHs (B[a]P, B[b]F, DB[a,l]P and DB[a,h]A) induced the oxidative stress pathway. DNA-adduct levels were: DB[a,l]P >> B[a]P > B[b]F 
DB[a,h]A > FA 1-MPA. The expression of only a few genes was found to correlate significantly with DNA-adduct formation, carcinogenic potency or Ah-receptor binding capacity (the last two taken from literature). These genes differed between the parameters. Our results indicate that PAHs generally induce a compound-specific response on gene expression and that discrimination of carcinogenic from non-carcinogenic compounds is partly feasible using this approach. Only at a specific pathway level, namely oxidative stress response, PAHs with high and low carcinogenic potency could be discriminated.
 |
Introduction |
|---|
 Top Introduction Materials and methodsResultsDiscussionConclusionSupplementary DataReferences |
|---|
Polycyclic aromatic hydrocarbons (PAHs) comprise a large and ubiquitous class of structurally related environmental chemicals that differ greatly in their carcinogenic potency (1
,2). An indication of their carcinogenic potential may be derived from a consideration of their structural properties, such as the presence of fjord and bay regions, in relation to their metabolism (3,4). Several other properties may help define the carcinogenic potency of PAHs, such as their ability to interact covalently with DNA following metabolic activation (5–9). Although DNA-adduct formation has been shown to correlate with the mutagenic potency in the case of PAHs (6), the various PAH–DNA adducts may be differently repaired by the cellular DNA-repair mechanisms, and thereby induction of mutations (5), and additional events are required to transform a normal cell into a malignant one.
In the liver, PAHs are metabolically activated by the CYP1 family of cytochromes P450, whose genes are under transcriptional control of the Aromatic hydrocarbon-receptor (Ah-receptor) (10). The principal catalysts of the activation of PAHs are CYP1A1 and CYP1B1 which, although poorly expressed in the liver, are highly inducible by these compounds (11). After metabolic activation, PAHs form adducts with DNA, induce mutations, and thereby initiate carcinogenesis (5,12). Many PAHs are ligands for the Ah-receptor, leading to transactivation of Ah-receptor mediated gene expression, including CYP1. Consequently, PAHs can increase their own metabolism including the activation pathways (13). The ability to bind to the Ah-receptor is considered to be a critical factor in determining the genotoxicity of PAHs (14). For comparison of PAHs in their ability to bind to the Ah-receptor, we used the data of Machala et al. (13), who expressed Ah-receptor binding of a PAH as an Induction Equivalency Factor (IEF).
Ranking of PAHs according to their carcinogenic potencies is currently based on several parameters, such as the ability to generate DNA adducts, mutagenicity and tumour forming potency. In order to compare the different PAHs, the toxicity equivalency factor (TEF) has been introduced (15,16), based on the formation of carcinomas, full carcinogenesis and DNA-adduct formation. Another ranking for carcinogenicity of PAHs is developed by Collins et al. (17), who estimated a potency equivalency factor (PEF) for PAHs relative to B[a]P, based on bioassay data. However, for many PAHs these data are not currently available. Since none of the short-term endpoints, e.g. DNA-adduct formation, mutagenicity, Ah-receptor binding, is solely indicative of PAH carcinogenicity, the development of additional methods that can improve their ranking would be advantageous.
Gene expression profiling by DNA microarray technology has proved to be a useful tool in revealing mechanisms of toxicity and in toxicity classification of compounds (18,19). Thus this technology can improve our appreciation of the carcinogenic potency of chemicals. By studying the effects of a PAH on the expression of a large set of genes, a compound-specific ‘transcriptomic fingerprint’ may be obtained that reveals mechanistic information related to its carcinogenicity. Data derived from studies with a large number of PAHs may eventually improve our understanding of the differences in carcinogenic potency between PAHs, and might enable us to discriminate better between the carcinogenic and less or non-carcinogenic compounds.
To better understand the difference in carcinogenic potency among PAHs, we investigated the transcriptomic fingerprints induced by PAHs with a wide range of carcinogenic activity in precision-cut rat liver slices. Precision-cut liver slices are frequently used in in vitro model for an increasing number of applications, including transcriptome profiling studies (20). In this model, cells are refrained in an environment with normal cell–cell and cell–matrix contacts, and remain to express high levels of metabolic enzymes that are important in PAH-mediated effects. The effects of PAHs on gene expression were examined for a total of 5700 genes simultaneously and compared with DNA-adduct formation, carcinogenic potency and Ah-receptor binding.
|
Materials and methods |
|---|
|
TopIntroductionMaterials and methodsResultsDiscussionConclusionSupplementary DataReferences |
|---|
Chemicals
Benzo[a]pyrene (B[a]P, purity 97%, CAS no. 50-32-8), benzo[b]fluoranthene (B[b]F, purity 98%, CAS no. 205-99-2), fluoranthene (FA, purity 99%, CAS no. 206-44-0), dibenzo[a,h]anthracene (DB[a,h]A, purity 97%, CAS no. 53-70-3) and dibenzo[a,l]pyrene (DB[a,l]P, purity 99,6%, CAS no. 191-30-0) were obtained from Sigma-Aldich (Zwijndrecht, The Netherlands). 1-Methylphenanthrene (1-MPA, purity 99%, CAS no. 832-69-9) was obtained from LGC Promochem (Teddington, UK). All chemicals were dissolved in DMSO.
Preparing and exposure of precision-cut liver slices
Rat livers were obtained from male Wistar rats (175–250 g) killed by cervical dislocation. Livers were immediately excised and slices (250 µm) were prepared using a Krumdieck tissue slicer (Alabama Research and Development Corp., Munsford, AL) as previously described (21). Slices were pre-incubated for 30 min at 37°C in RPMI supplemented with 5% foetal calf serum (FCS), 0.5 mM L-methionine, 1 µM insulin, 0.1 mM hydrocothsone-21-hemisuccinate and 50 µg/ml gesntomycin in 12-well plates on a shaking incubator (5% CO2 and 95% air). After pre-incubation, the slices were transferred to 12-well plates containing fresh media and to which was added 3, 10 or 30 µM of the PAHs or a solvent control (DMSO, 0.067% v/v). In each of two independent experiments, three slices were used for each treatment. After 6- or 24-h exposure, the slices were removed from the medium and immediately frozen in liquid nitrogen.
RNA isolation, cDNA synthesis and dye labelling
After crushing the liver slices under liquid nitrogen, RNA was stabilized by dissolving the crushed powder in Trizol (Gibco/BRL, Breda, The Netherlands) and isolated according to the manufacturer's manual. RNA was purified using the RNeasy mini kit (Qiagen Westburg bv., Leusden, The Netherlands) with DNase treatment, quantity was measured spectrophotometrically in 50 mM NaOH and quality was determined using a BioAnalyzer (Agilent Technologies, Breda, The Netherlands). OD 260/280 nm were 1.79 ± 0.13. Only RNA samples, which were not degraded (clear 18S and 28S peaks), were used for labelling and hybridization. To obtain sufficient RNA for microarray analysis, for each treatment two or three liver slices were pooled.
RNA samples were reverse transcribed into cDNA in triplicate with amino-allyl labelled dUTP (Sigma-Aldrich, St Louis, MO) and subsequently labelled with one of the three dyes, namely Cyanine 3 (Cy3), Cyanine 5 (Cy5) or Alexa 594 (A594), as was described previously (22). The labelling schedule is shown in Table I.
|
Microarray hybridization and data analysis
Dye-labelled samples were hybridized on an Operon rat oligonucleotide array containing 5700 oligonucleotides (v1.2.1, Operon, Qiagen, Venlo, The Netherlands) printed in triplicate on Corning UltraGAPS Coated Slides (Corning Life Sciences, New York, NY) by the Genome Centre Maastricht (Maastricht University, Maastricht, The Netherlands). Hybridization and washing were performed according to Corning's protocol for oligonucleotide arrays and as described previously (22
). The microarray slides were scanned on a ScanArrayExpress (Packard Biochip Technologies, PerkinElmer life sciences, Boston, MA). All three channels were scanned at 100% laser power and PMT gain was adjusted, such that the signal of the highest fluorescent spots was just below the maximum measurable level. The images (10 µ resolution; 16 bit tiff) were processed with ImaGene 5.0 software (BioDiscovery Inc., Los Angeles, USA) to quantify spot signals. Irregular spots were manually or automatically flagged and excluded from the data analysis.
Data from ImaGene were transported into GeneSight software version 4.1.5 (BioDiscovery Inc.) for further analysis. For each spot, local background was subtracted and flagged spots as well as spots with a net expression level <20 were omitted. Data were log base 2 transformed and expression differences between exposed and control calculated. Data normalization was carried out using LOWESS and centring expression differences by subtracting mean values. Data of replicate spots were combined while omitting outliers (>2 SD). To detect significantly modulated genes, first genes were selected using the confidence analysis tool from GeneSight (up or down regulation of 0.1 and 99% confidence limit) using the averaged data per treatment. This was followed by a Student's t-test analysis (P < 0.01) between the gene expression differences at each PAH concentration compared to self-hybridizations of RNA labelled with the same dyes. Genes were assumed significantly modulated if they were found in both the confidence analysis and the Student's t-test.
Unsupervised clustering was performed by hierarchical clustering analysis and principal component analysis using GeneSight tools. Classification analysis by supervised clustering was achieved using the nearest-shrunken-centroid method using the PAM software tool [http://www-stat.stanford.edu/~tibs/PAM/, 27-09-2005 (23)]. Genes used for classification were selected by multiple cross-validations based on the leave-one-out procedure. At the lowest misclassification rate, the smallest number of genes for that rate was selected for classification analysis into carcinogenic or non-carcinogenic.
Pathway analysis
Pathway analysis was carried out using GenMAPP version 2.0 software (Gladstone Institutes, University of California, San Fransisco, USA) and local maps from GenMAPP (Gene database Rn_contributed_20041216 and maps Rn_std_20041102 from www.GenMAPP.org) (24,25). For each treatment, all modulated genes were included in the analysis, with the exclusion of flagged genes. Pathways with a Z-score >2 and with at least two genes affected, were assumed to be significantly affected by the treatment.
DNA-adduct analysis
After removal of the aqueous phase during RNA isolation using Trizol, the remaining phases were used for DNA isolation according to the manufacturer's protocol. DNA-adduct levels were determined according to the procedure originally described by Reddy and Randerath (26) with the modifications introduced by Godschalk et al. (27). By including samples with known DNA-adduct levels (1 adduct per 106, 107 or 108 nt), it was possible to quantify DNA-adduct levels (detection limit 1 adduct per 108 nt). Adduct spots on the chromatograms were located and quantified using a phosphor imager (FLA-3000, Fuji, Paris, France) and AIDA/2D densitometry software.
Correlation analysis
For correlation of gene expression changes with carcinogenic potency, TEF values from Nisbet and LaGoy (16) were used for B[a]P, B[b]F, FA and DB[a,h]A. These were complemented with the PEF values proposed by Collins et al. (17) for DB[a,l]P, and we estimated a TEF for 1-MPA based on IARC data (1,2) according to the method proposed by Nisbet and LaGoy (16). The resulting TEF values were; 1.0 for B[a]P, 0.1 for B[b]F, 0.001 for FA and 1-MPA, 5.0 for DB[a,h]A and 10 for DB[a,l]P. For correlation with Ah-receptor binding, IEF values determined by Machala et al. (13) were used. The values after 24 hours exposure were used and expressed relative to B[a]P. They were 1 for B[a]P, 0.24 for B[b]F, 0 for FA, 7.64 for DB[a,h]A and 0.01 for DB[a,l]P.
After log (base 2) transformation, DNA-adduct levels, TEF and IEF were correlated with the expression changes of modulated genes by at least one concentration of the compound. Since after log transformation, DNA-adduct formation, TEF and IEF were normally distributed (according to the Kolmogorov-Smirnov test in SPSS, P < 0.05), Pearson correlation coefficients of gene expression with TEF, IEF (both for each concentration separately) and DNA-adduct formation (for all concentrations) were calculated using SPSS for windows 11.5 (SPSS Inc., Chicago, IL).
All microarray data are stored in ArrayExpress (http://www.ebi.ac.uk/arrayexpress/). Accession numbers are E-TOXM-24 and E-TOXM-25 for the experiments and A-MEXP-350, 351 and 352 for the array designs.
|
Results |
|---|
|
TopIntroductionMaterials and methodsResultsDiscussionConclusionSupplementary DataReferences |
|---|
After 6 h exposure, B[a]P modulated 30 genes and DB[a,l]P modulated 41 genes. When exposure was extended to 24 h, the number of modulated genes rose to 56 for B[a]P and 111 for DB[a,l]P. Only three genes were found to overlap between both time points for B[a]P and for DB[a,l]P this was one gene. Hierarchical clustering as well as principal component analysis of the 83 genes modulated by B[a]P at either 6 or 24 h together with self-hybridizations showed that the 6 h exposure resembles the self-hybridizations more closely than the longer exposure (Figure 1); similar results were obtained for DB[a,l]P. Furthermore, at 6 h, DNA-adduct levels in B[a]P- or DB[a,l]P-treated liver slices were
6% of those following a 24 h exposure (data not shown). Collectively, these data led us to conclude that the PAHs have only limited effects on gene expression and DNA-adduct formation after 6 h exposure compared with 24 h, and therefore we focused on the 24 h time point only for the remaining four compounds.
|
All six compounds modulated gene expression in liver slices after 24 h exposure. B[b]F modulated the highest number of genes, namely 165 genes for all concentrations combined, followed by DB[a,h]A (115 genes), DB[a,l]P (111 genes), FA (77 genes), B[a]P (56 genes), and finally 1-MPA (27 genes). Overall, 425 genes were modulated by at least one treatment after 24 h exposure, and in general a dose-dependent effect on gene expression modulation was observed (see Supplementary Data, including information on probe ID, gene names, abbreviations, GenBank accession numbers and gene expression differences).
Hierarchical clustering of the PAH treatments and the 425 modulated genes shows that for each compound, except for DB[a,l]P, all doses cluster together, indicating a compound-specific response on gene expression (Figure 2). Higher in the dendrogram, it can be seen that B[a]P, DB[a,h]A and B[b]F elicit, to some extent similar responses on gene expression. Similar observations were made for DB[a,l]P, FA and 1-MPA. This finding is supported by the principal component analysis (PCA) using the 425 modulated genes (Figure 3). Moreover, using PCA the compounds are split in the same two groups, indicating a distinction between the gene expression profiles of both groups. However, PCA also shows a similarity on gene expression response after DB[a,l]P and DB[a,h]A treatment, which are both in the lower part of the plot. Thus using both of these unsupervised clustering methods, the most carcinogenic PAH (DB[a,l]P), is grouped together with the two least carcinogenic (FA and 1-MPA).
|
|
Furthermore, we investigated whether a sub-group of the 425 genes would be better classifiers for discrimination of the highly carcinogenic PAHs (B[a]P, DB[a,l]P, DB[a,h]A and B[b]F) from the weakly carcinogenic PAHs (FA and 1-MPA). In this approach the supervised clustering method, the nearest-shrunken-centroid method was employed. When all treatments for all PAHs are used, 16 classifier genes were selected to best discriminate carcinogenic from non-carcinogenic PAHs, namely CYP1A1, CYP1A2, NQO1, PGRMC1, SPP1, EPHX1, CYP1B1, MPL3, GSTM1, TTR, RT1-N1, GBP2, DIG1, AOX1, GARG16 and CTSK. Using these genes, it was noted that all treatments, except the two lowest doses of DB[a,l]P, were correctly classified as being non-carcinogenic or carcinogenic (Figure 4). In addition, this method was applied to predict the carcinogenic potency of a PAH based on data derived from the five other PAHs. This was done at all concentrations studied, and the results are shown in Table II. B[a]P, B[b]F and DB[a,h]A were always correctly classified, whereas DB[a,l]P was always misclassified. Classification of 1-MPA and FA based on data of the remaining 5 compounds was not possible since at least two compounds are required in each class for classification.
|
|
Pathway analysis by GenMAPP based on the modulated genes for each PAH and exposure concentration, revealed several affected pathways, which are listed in Table III. Only oxidative stress was affected by more than one compound, namely by DB[a,l]P, DB[a,h]A, B[a]P and B[b]F. It is noteworthy that these are all PAHs with a medium or high carcinogenic activity.
|
DNA-adducts were measured by 32P-post-labelling in the same samples used for gene expression profiling (Table IV). DB[a,l]P exposure resulted in the highest number of DNA adducts (up to 912 adducts per 108 nt), followed by B[a]P (up to 35 per 108 nt), B[b]F (up to 15 per 108 nt) and DB[a,h]A (up to 12 per 108 nt). In most cases dose-dependent DNA-adduct formation was observed. No or virtually no DNA-adducts above background were detected after exposure of the liver slices to FA or 1-MPA.
|
DNA-adduct levels (after log transformation) were correlated with the expression changes for each of the 425 modulated genes. Significantly correlating genes (P < 0.05) with an R2 > 0.4 are shown in Table IV. Only six genes satisfied these criteria.
The expression changes for the 425 significantly modulated genes were also correlated with carcinogenic potency (TEF) and Ah-receptor binding capacity (IEF). Significantly correlating genes (P < 0.05) with an R2 > 0.4 are shown in Table V. Two genes correlated with carcinogenic potency and 24 genes with Ah-receptor binding. No genes were found to correlate with more than one parameter. More detailed information on correlation coefficients and fold changes of the genes can be found in Supplementary Data.
|
|
Discussion |
|---|
|
TopIntroductionMaterials and methodsResultsDiscussionConclusionSupplementary DataReferences |
|---|
To better understand the difference in carcinogenic potency among PAHs, we examined the effect of six PAHs on gene expression in precision-cut rat liver slices in relation to carcinogenic potency, DNA binding capacity and Ah-receptor binding.
After 24 h exposure, all tested PAHs modulated gene expression in a dose-dependent manner. Although no genes were found to be modulated by every PAH, CYP1A1 and BEST5 were both up-regulated by five of six compounds (not by 1-MPA or DB[a,h]A respectively). Further, AFAR, DIG1, EPHX1 and GSTM1 were all up-regulated by the carcinogenic PAHs but not by the non-carcinogenic PAHs (FA and 1-MPA). Interestingly, these four genes are all involved in mechanisms to prevent carcinogenic damaging events and thereby protecting the cell, which may explain their modulation solely by carcinogenic PAHs (28–30).
Hierarchical clustering and principal component analyses using all 425 modulated genes both indicate that, in general, each compound has a specific gene expression pattern. Only the 3 µM treatment with DB[a,l]P had a different gene expression pattern than the higher concentration treatments, which is probably due to a marginal effect on gene expression at this low dose. Although the response on gene expression of DB[a,l]P seems to show some similarity with the response of DB[a,h]A treatment (Figure 3), all compounds can be divided into two groups, one consisting of B[a]P, DB[a,h]A and B[b]F and the other of 1-MPA, FA and DB[a,l]P. The first group comprises only carcinogenic PAHs, whereas the other group contains a highly carcinogenic compound, DB[a,l]P in addition to two relatively non-carcinogenic compounds. Thus, the two groups do not perfectly correspond with carcinogenicity based on studies carried out in precision-cut rat liver slices. The separation in these two groups might be explained by the poor induction of Ah-receptor dependent gene expression (CYP1A1, CYP1A2 and CYP1B1) by DB[a,l]P and the two relatively non-carcinogenic compounds, whereas B[a]P, B[b]F and DB[a,h]A are all strong inducers of Ah-receptor dependent gene expression. In contrast, in HepG2 cells, a human hepatoma cell line, exposed to identical PAHs at similar concentrations carcinogenic and non-carcinogenic PAHs could be discriminated in hierarchical clustering (31). This discrepancy most probably reflects differences in the in vitro system employed and the species differences.
In order to investigate whether within the total of 425 genes a sub-group of genes exists that can be used to discriminate PAHs regarding carcinogenicity, classification analyses by the nearest-shrunken-centroid method was performed. Though the gene numbers could be reduced, still B[a]P, B[b]F and DB[a,h]A were constantly and correctly classified as being carcinogenic, whereas DB[a,l]P was wrongly classified as having a low carcinogenic potency. This incorrect classification of DB[a,l]P may be caused by the minor induction of the CYP1A1 gene by DB[a,l]P only at the highest concentration, whereas the other carcinogenic PAHs induce CYP1A1 to a higher extent and at all concentrations. Noteworthy, DB[a,l]P was found to induce CYP1A1 in HepG2 cells, which probably contributed to the more accurate classification of PAHs in these cells.
These results suggest that HepG2 cells are a more suitable model for classification of the carcinogenic potency of PAH based on gene expression profiling. However, we tested only six compounds and used arrays with only a limited part of all genes. A larger number of compounds and arrays covering the whole transcriptome, may improve the classification of PAHs in liver slices.
Pathway analysis can be useful in revealing the processes in which the modulated genes are involved and, thereby, may elucidate PAH induced carcinogenic mechanisms. All carcinogenic PAHs modified genes involved in oxidative stress (which implies an increased antioxidant production and decreased ROS formation), whereas the least carcinogenic PAHs, FA and 1-MPA, failed to do so. This observation would imply that affecting oxidative stress might be an important characteristic or classifier for carcinogenic PAHs. PAHs are known to induce oxidative stress following metabolism to yield compounds such as quinones [reviewed in ref. (4)]. Our observation might be a protective mechanism of the liver cells to oxidative damage induced by PAHs. Nuclear receptors were affected by DB[a,l]P as for instance the RXR receptor was down-regulated and the LXR (or oxysterol) receptor was up-regulated. Over-expression of the RXR receptor has been shown to inhibit growth of carcinoma cells (32), which means that our observed down-regulation may be seen as a tumour promoting activity of DB[a,l]P. Furthermore, DB[a,l]P seems to influence apoptosis and HSP70 suggesting a stress response of the liver slices. Accordingly, the regulation of actin cytoskeleton is down-regulated in response to DB[a,l]P which also indicate induction of apoptosis. Additionally, DB[a,l]P decreased the TGF-beta protein signalling pathway. TGF-beta serves as a tumour suppressor pathway in normal tissue and is known to be over-expressed in advanced stages of carcinogenesis (33). DB[a,h]A influenced the electron transport chain pathway which may be related to its carcinogenic mode of action. There is evidence that defects in electron transport activities is linked to carcinogenic outcome, as for instance was shown for brain tumours (34). Further, DB[a,h]A showed elevation of cholesterol biosynthesis pathway, which could be interpreted as cell growth promotion by DB[a,h]A, since in proliferating normal tissue and tumours cholesterol biosynthesis is enhanced (35). Treatment with B[b]F was found to down-regulate glycolysis and gluconeogenesis. Although the energy requirement of hepatic cancer cells increases, gluconeogenesis decreases in poorly differentiated cells (36). This seems to indicate a preliminary carcinogenic response to B[b]F treatment. Overall, some of the perturbed pathways point to either a protective mechanism of the cell against carcinogenic insult or promotion of the carcinogenic process. Thus, with the exception of effects on oxidative stress, all PAHs appear to have a different effect on liver slices.
At pathway level, some similarities between the current study on liver slices and the data from our previous study in HepG2 cells were found. In both cell systems, the apoptosis pathway, cholesterol biosynthesis and effects on fatty acid synthesis or degradation were observed. Only the HMGCR gene in the cholesterol biosynthesis pathway and the NF
B gene in the apoptosis pathway were modulated in liver slices and HepG2 cells. Other genes (caspases in the apoptosis pathway) were not affected in HepG2 cells or genes (others) were not present on the arrays used in the HepG2 experiment. Thus, modulation of these pathways appear to be important responses to PAH exposure.
The order of DNA binding capacity in liver slices was: DB[a,l]P >> B[a]P > B[b]F DB[a,h]A > FA 1-MPA. This order differs slightly from literature data (7–9), where it is in general: DB[a,l]P > DB[a,h]A > B[a]P > B[b]F. Most of these data, however, were obtained in in vivo studies conducted in the lung, whereas the current studies were carried out in vitro in the liver. DNA-adduct formation in rat hepatocytes in vitro, in the presence of rat S9-mix, resulted in the same order of DNA binding capacity for B[a]P, B[b]F, DB[a,h]A and FA as observed in the present studies (9). In HepG2 cells the order of DNA binding capacity was B[a]P >> DB[a,l]P > B[b]F > DB[a,h]A > 1-MPA FA showing that DB[a,l]P induces far more DNA adducts in liver slices than in HepG2 cells. DNA-adduct formation has been shown to be higher in CYP1A1 knockout mice after B[a]P exposure compared to wild types (37), and therefore CYP1A1 activation results in lower DNA-adduct levels. DB[a,l]P only marginally induced the expression of CYP1A1 in liver slices, but it caused a more markedly induction of CYP1A1 in HepG2 cells, and may therefore induce higher levels of DNA adducts in liver slices than in HepG2 cells.
Correlation of gene expression changes with DNA-adduct formation, carcinogenic potency and Ah-receptor binding, revealed several highly correlating genes. As expected, many genes, known to be regulated via the Ah-receptor, were seen to correlate with IEF. These were genes from the cytochrome family (e.g. CYP1A1 and CYP1A2), from the UDP glucuronyltransferase 1 family (UGT1A1, UGT1A6 and UGT1A7), SLC21A5, NQO1 and GSTA2. These genes are also modulated in rats following exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin, the highest affinity ligand known for this receptor (38). In comparison with HepG2 cells only the Ah-receptor controlled genes CYP1A1 and CYP1A2 correlated in both cell systems with IEF. Other genes were either not significantly modulated in HepG2 cells, or were not present on the arrays used for these experiments. The functions of some significantly correlating genes could be related to carcinogenicity. For example, IEF correlated with CASP12 (involved with the terminal stage of apoptosis), FABP1 (may play a role in hepatocyte cell proliferation and may transport activated chemical carcinogens), SLCO1A4 (SLC21A4, mediates transport of organic anions) and the UDP glucuronyltransferase 1 family (involved in metabolism of, for example, PAHs). TEF correlated with AOX1 and NOS2, which are both enzymes involved in inflammation response, and thereby possibly indirectly with carcinogenesis [reviewed in ref. (39)]. Furthermore, NOS2 plays an important role in the carcinogenesis of the PAH 3-methylcholanthrene (40). Finally, in relation to carcinogenesis, DNA-adduct formation correlated significantly with MMP9. The transcription and expression of MMP9 is increased throughout the process of hepatocarcinogenesis (41). Although genes involved in DNA repair might be expected to correlate with DNA-adduct formation, those that were present on the array (for example XRCC1, XRCC5, OGG1, MPG and BRCA1) were not significantly modulated by either of the treatments and therefore not included in the correlation studies. From this study, but also from other studies with HepG2 cells, it appears that modulation of DNA repair genes by genotoxic carcinogens is not a pronounced effect (19,31). No gene was found to correlate with all parameters, indicating that either not all parameters used for correlation are highly relevant to carcinogenicity or no gene can be used as an indicator of carcinogenic potency of PAHs.
When analyzing the correlating genes for each parameter with GenMAPP, no pathways were significantly affected by genes correlating with DNA-adduct formation or TEF. Genes correlating with IEF, however, appear to affect the oxidative stress pathway. The relationship between oxidative stress and IEF might be explained by the ability of PAHs to interact with the Ah-receptor, which leads to induction of cytochrome P450 enzymes. PAHs are metabolized by these enzymes and can form reactive intermediates, which can give rise to oxidative stress [reviewed in ref. (4)].
|
Conclusion |
|---|
|
TopIntroductionMaterials and methodsResultsDiscussionConclusionSupplementary DataReferences |
|---|
It may be inferred that PAHs generally induce a compound-specific response on gene expression. This is reflected in both the clustering analyses as well as in the pathway analyses.
Using all modulated genes, discrimination of carcinogenic from non-carcinogenic compounds is, to some extent, feasible (only DB[a,l]P is incorrectly grouped). But reduction of the discriminating genes using a classification tool, does not improve this discrimination.
Only at a specific pathway level, namely that for oxidative stress response, PAHs with high and low carcinogenic potency could be discriminated.
|
Supplementary Data |
|---|
|
TopIntroductionMaterials and methodsResultsDiscussionConclusionSupplementary DataReferences |
|---|
The Supplementary Data are available at Mutagenesis Online. ‘Differences and standard deviations modulated genes’ contains information on probe ID, gene names, abbreviations, GenBank accession numbers and gene expression differences. ‘Correlations’ contains more detailed information on correlation coefficients and fold changes of the genes significantly correlating genes.
|
Acknowledgments |
|---|
The research was carried out as part of the AMBIPAH project (mechanism-based approaches to improved cancer risk assessment of ambient air polycyclic aromatic hydrocarbons), funded by the European Union (No. QRLT-2001-024202).
|
Notes |
|---|
*To whom correspondence should be addressed. Tel: +31 43 3881092; Fax: +31 43 3884146; Email: j.vandelft{at}grat.unimaas.nl
|
References |
|---|
|
TopIntroductionMaterials and methodsResultsDiscussionConclusionSupplementary DataReferences |
|---|
-
1. IARC. (1973) Certain polycyclic aromatic hydrocarbons and heterocyclic compounds. IARC Monogr. Eval. Carcinog. Risk Chem. Hum. 3:1–271.[Medline]
2. IARC. (1983) Polynuclear aromatic compounds, Part 1, Chemical, environmental and experimental data. IARC Monogr. Eval. Carcinog. Risk Chem. Hum. 32:1–453.[Medline]
3. Melendez-Colon V.J., Luch A., Seidel A., Baird W.M. (2000) Formation of stable DNA adducts and apurinic sites upon metabolic activation of bay and fjord region polycyclic aromatic hydrocarbons in human cell cultures. Chem. Res. Toxicol. 13:10–17.[CrossRef][ISI][Medline]
4. Xue W. and Warshawsky D. (2005) Metabolic activation of polycyclic and heterocyclic aromatic hydrocarbons and DNA damage: a review. Toxicol. Appl. Pharmacol. 206:73–93.[CrossRef][ISI][Medline]
5. Ross J.A., Nelson G.B., Wilson K.H., Rabinowitz J.R., Galati A., Stoner G.D., Nesnow S., Mass M.J. (1995) Adenomas induced by polycyclic aromatic hydrocarbons in strain A/J mouse lung correlate with time-integrated DNA adduct levels. Cancer Res. 55:1039–1044.
6. Kriek E., Rojas M., Alexandrov K., Bartsch H. (1998) Polycyclic aromatic hydrocarbon-DNA adducts in humans: relevance as biomarkers for exposure and cancer risk. Mutat Res. 400:215–231.[ISI][Medline]
7. Harrigan J.A., Vezina C.M., McGarrigle B.P., Ersing N., Box H.C., Maccubbin A.E., Olson J.R. (2004) DNA adduct formation in precision-cut rat liver and lung slices exposed to benzo(a)pyrene. Toxicol Sci. 77:307–314.
8. Arif J.M., Smith W.A., Gupta R.C. (1999) DNA adduct formation and persistence in rat tissues following exposure to the mammary carcinogen dibenzo[a,l]pyrene. Carcinogenesis 20:1147–1150.
9. Segerback D. and Vodicka P. (1993) Recoveries of DNA adducts of polycyclic aromatic hydrocarbons in the 32P-postlabelling assay. Carcinogenesis 14:2463–2469.
10. Shimada T. and Fujii-Kuriyama Y. (2004) Metabolic activation of polycyclic aromatic hydrocarbons to carcinogens by cytochromes P450 1A1 and 1B1. Cancer Sci. 95:1–6.[CrossRef][Medline]
11. Shimada T., Sugie A., Shindo M., Nakajima T., Azuma E., Hashimoto M., Inoue K. (2003) Tissue-specific induction of cytochromes P450 1A1 and 1B1 by polycyclic aromatic hydrocarbons and polychlorinated biphenyls in engineered C57BL/6J mice of arylhydrocarbon receptor gene. Toxicol. Appl. Pharmacol. 187:1–10.[CrossRef][ISI][Medline]
12. Melendez-Colon V.J., Luch A., Seidel A., Baird W.M. (1999) Cancer initiation by polycyclic aromatic hydrocarbons results from formation of stable DNA adducts rather than apurinic sites. Carcinogenesis 20:1885–1891.
13. Machala M., Vondracek J., Blaha L., Ciganek M., Neca J.V. (2001) Aryl hydrocarbon receptor-mediated activity of mutagenic polycyclic aromatic hydrocarbons determined using in vitro reporter gene assay. Mutat Res. 497:49–62.[ISI][Medline]
14. Solhaug A., Refsnes M., Lag M., Schwarze P.E., Husoy T., Holme J.A. (2004) Polycyclic aromatic hydrocarbons induce both apoptotic and anti-apoptotic signals in Hepa1c1c7 cells. Carcinogenesis 25:809–819.
15. Bostrom C.E., Gerde P., Hanberg A., Jernstrom B., Johansson C., Kyrklund T., Rannug A., Tornqvist M., Victorin K., Westerholm R. (2002) Cancer risk assessment, indicators, and guidelines for polycyclic aromatic hydrocarbons in the ambient air. Environ. Health Perspect. 110:Suppl 3, 451–488.
16. Nisbet I.C. and LaGoy P.K. (1992) Toxic equivalency factors (TEFs) for polycyclic aromatic hydrocarbons (PAHs). Regul. Toxicol. Pharmacol. 16:290–300.[CrossRef][ISI][Medline]
17. Collins J.F., Brown J.P., Alexeeff G.V., Salmon A.G. (1998) Potency equivalency factors for some polycyclic aromatic hydrocarbons and polycyclic aromatic hydrocarbon derivatives. Regul. Toxicol. Pharmacol. 28:45–54.[CrossRef][ISI][Medline]
18. Hu T., Gibson D.P., Carr G.J., Torontali S.M., Tiesman J.P., Chaney J.G., Aardema M.J. (2004) Identification of a gene expression profile that discriminates indirect-acting genotoxins from direct-acting genotoxins. Mutat. Res. 549:5–27.[ISI][Medline]
19. van Delft J.H., van Agen E., van Breda S.G., Herwijnen M.H., Staal Y.C., Kleinjans J.C. (2004) Discrimination of genotoxic from non-genotoxic carcinogens by gene expression profiling. Carcinogenesis 25:1265–1276.
20. Parrish A.R., Fisher R., Bral C.M., Burghardt R.C., Gandolfi A.J., Brendel K., Ramos K.S. (1998) Benzo(a)pyrene-induced alterations in growth-related gene expression and signaling in precision-cut adult rat liver and kidney slices. Toxicol. Appl. Pharmacol. 152:302–8.[CrossRef][ISI][Medline]
21. Hashemi E., Dobrota M., Till C., Ioannides C. (1999) Structural and functional integrity of precision-cut liver slices in xenobiotic metabolism: a comparison of the dynamic organ and multiwell plate culture procedures. Xenobiotica 29:11–25.[CrossRef][ISI][Medline]
22. Staal Y.C., van Herwijnen M.H., van Schooten F.J., van Delft J.H. (2005) Application of four dyes in gene expression analyses by microarrays. BMC Genomics 6:101.[CrossRef][Medline]
23. Tibshirani R., Hastie T., Narasimhan B., Chu G. (2002) Diagnosis of multiple cancer types by shrunken centroids of gene expression. Proc. Natl Acad. Sci. USA 99:6567–6572.
24. Doniger S.W., Salomonis N., Dahlquist K.D., Vranizan K., Lawlor S.C., Conklin B.R. (2003) MAPPFinder: using Gene Ontology and GenMAPP to create a global gene-expression profile from microarray data. Genome Biol. 4:R7.[CrossRef][Medline]
25. Dahlquist K.D., Salomonis N., Vranizan K., Lawlor S.C., Conklin B.R. (2002) GenMAPP, a new tool for viewing and analyzing microarray data on biological pathways. Nat. Genet. 31:19–20.[CrossRef][ISI][Medline]
26. 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.
27. Godschalk R.W., Maas L.M., Van Zandwijk N., van't Veer L.J., Breedijk A., Borm P.J., Verhaert J., Kleinjans J.C., 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.
28. Bardwell L., Cook J.G., Zhu-Shimoni J.X., Voora D., Thorner J. (1998) Differential regulation of transcription: repression by unactivated mitogen-activated protein kinase Kss1 requires the Dig1 and Dig2 proteins. Proc. Natl Acad. Sci. USA 95:15400–15405.
29. Knight L.P., Primiano T., Groopman J.D., Kensler T.W., Sutter T.R. (1999) cDNA cloning, expression and activity of a second human aflatoxin B1-metabolizing member of the aldo-keto reductase superfamily, AKR7A3. Carcinogenesis 20:1215–1223.
30. Spivack S.D., Hurteau G.J., Fasco M.J., Kaminsky L.S. (2003) Phase I and II carcinogen metabolism gene expression in human lung tissue and tumors. Clin Cancer Res. 9:6002–6011.
31. Staal Y.C., van Herwijnen M.H., van Schooten F.J., van Delft J.H. (2006) Modulation of gene expression and DNA adduct formation in HepG2 cells by polycyclic aromatic hydrocarbons with different carcinogenic potencies. Carcinogenesis 27:646–655.
32. Wu S., Zhang D., Zhang Z.P., Soprano D.R., Soprano K.J. (1998) Critical role of both retinoid nuclear receptors and retinoid-X-receptors in mediating growth inhibition of ovarian cancer cells by all-trans retinoic acid. Oncogene 17:2839–2849.[CrossRef][ISI][Medline]
33. Letterio J.J. (2005) Disruption of the TGF-beta pathway and modeling human cancer in mice. Mutat Res. 576:120–131.[ISI][Medline]
34. Lichtor T. and Dohrmann G.J. (1986) Respiratory patterns in human brain tumors. Neurosurgery 19:896–899.[ISI][Medline]
35. Rao K.N. (1995) The significance of the cholesterol biosynthetic pathway in cell growth and (review). Anticancer Res. 15:309–314.[ISI][Medline]
36. Bannasch P., Klimek F., Mayer D. (1997) Early bioenergetic changes in hepatocarcinogenesis: preneoplastic phenotypes mimic responses to insulin and thyroid hormone. J. Bioenerg. Biomembr. 29:303–313.[CrossRef][ISI][Medline]
37. Uno S., Dalton T.P., Dragin N., Curran C.P., Derkenne S., Miller M.L., Shertzer H.G., Gonzalez F.J., Nebert D.W. (2006) Oral benzo[a]pyrene in Cyp1 knockout mouse lines: CYP1A1 important in detoxication, CYP1B1 metabolism required for immune damage independent of total-body burden and clearance rate. Mol. Pharmacol. 69:1103–1114.
38. Fletcher N., Wahlstrom D., Lundberg R., Nilsson C.B., Nilsson K.C., Stockling K., Hellmold H., Hakansson H. (2005) 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) alters the mRNA expression of critical genes associated with cholesterol metabolism, bile acid biosynthesis, and bile transport in rat liver: a microarray study. Toxicol. Appl. Pharmacol. 207:1–24.[ISI][Medline]
39. Moss S.F. and Blaser M.J. (2005) Mechanisms of disease: Inflammation and the origins of cancer. Nat. Clin. Pract. Oncol. 2:90–97 and quiz 1 p. following 113.[CrossRef][ISI][Medline]
40. Li H., Chen F., Yu L., Liu M., Chen H., Zhang Y., Liu X. (2002) Expression of inflammation related enzymes during experimental rat lung carcinogenesis. Zhonghua Zhong Liu Za Zhi 24:316–319.[Medline]
41. Gao Y., Zhang Z., Jiang Z.S., Fang S.G., Sun E.W., Wang Y., Chen B.H. (2002) Dynamic changes of matrix metalloproteinases in rat liver during the development of diethylnitrosamine-induced hepatocarcinoma. Di Yi Jun Yi Da Xue Xue Bao 22:865–868.[Medline]
Received on August 17, 2006;
revised on October 23, 2006;
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  D. S. Pushparajah, M. Umachandran, K. E. Plant, N. Plant, and C. Ioannides Up-regulation of the glutathione S-transferase system in human liver by polycyclic aromatic hydrocarbons; comparison with rat liver and lung Mutagenesis, July 1, 2008; 23(4): 299 - 308. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. N. McDougal and C. M. Garrett Gene Expression and Target Tissue Dose in the Rat Epidermis after Brief JP-8 and JP-8 Aromatic and Aliphatic Component Exposures Toxicol. Sci., June 1, 2007; 97(2): 569 - 581. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||