Analysis of biomarkers in a Czech population exposed to heavy air pollution. Part II: chromosomal aberrations and oxidative stress

Transcription

1 Mutagenesis vol. 28 no. 1 pp , 2013 Advance Access publication 10 October 2012 doi: /mutage/ges058 Analysis of biomarkers in a Czech population exposed to heavy air pollution. Part II: chromosomal aberrations and oxidative stress Pavel Rossner, Jr*,1, Andrea Rossnerova, 1 Milada Spatova 1, Olena Beskid 1, Katerina Uhlirova 1,2, Helena Libalova 1, Ivo Solansky 1, Jan Topinka 1 and Radim J. Sram 1 1 Department of Genetic Ecotoxicology, Institute of Experimental Medicine AS CR, Videnska 1083, Prague, Czech Republic 2 Department of Molecular Genetics, Institute of Plant Molecular Biology, Branisovska 31/1160, Ceske Budejovice, Czech Republic *To whom correspondence should be addressed. Tel: ; Fax: ; prossner@biomed.cas.cz Received on April 23, 2012; revised on July 25, 2012; accepted on August 9, 2012 Populations living in industrialised regions are at higher risk of a number of diseases and shortened life span. These negative effects are primarily brought about by damage to cells and macromolecules caused by environmental pollutants. In this study, we analysed the effect of exposure to benzo[a]pyrene, a particulate matter of aerodynamic diameter < 2.5 µm (PM2.5), and benzene on oxidative stress markers [including 8-oxo-7,8-dihydro-2 -deoxyguanosine (8-oxodG), 15-F 2t -isoprostane (15-F2t-IsoP) and protein carbonyls] and cytogenetic parameters (stable and unstable chromosomal aberrations). The samples were collected from subjects living in the Ostrava region characterised by very high levels of air pollution and in Prague with comparatively lower concentrations of pollutants in three seasons (winter 2009, summer 2009 and winter 2010). Despite several-fold higher concentrations of air pollutants in the Ostrava region, the levels of stable aberrations (genomic frequency of translocations per 100 cells, percentage of aberrant cells and frequency of acentric fragments) were mostly comparable in both locations. The frequency of unstable aberrations measured as the number of micronuclei was unexpectedly significantly lower in the Ostrava region subjects in both seasons of Urinary excretion of 8-oxodG did not differ between locations in either season. Lipid peroxidation measured as levels of 15-F2t-IsoP in blood plasma was elevated in the Ostrava subjects sampled in Protein oxidation was higher in Prague samples collected in summer Multivariate analyses conducted separately in subjects from Prague and Ostrava showed a negative association between the frequency of micronuclei and concentrations of benzo[a]pyrene and PM2.5 in both regions. A positive relationship was observed between lipid peroxidation and air pollution; protein oxidation seems to be positively affected by PM2.5 in both regions. Introduction Air pollution is associated with many negative health effects including pulmonary and cardiovascular diseases or cancer as well as increased mortality (1). Although underlying biological mechanisms of induction of negative health effects are mostly unknown, it has been demonstrated that pollutants cause damage to DNA and other important biomolecules (lipids, proteins) (2 4). Air pollutants belong to diverse groups of chemical compounds that are generally grouped into four categories: gaseous pollutants [e.g. volatile organic compounds (VOCs), including benzene], persistent organic compounds [e.g. polycyclic aromatic hydrocarbons (PAHs), including benzo[a]pyrene (B[a] P)], heavy metals and particulate matter (PM), onto which many chemicals are adsorbed (5). As PM of various compositions is ubiquitous in the environment, health consequences of exposure to PM are of great concern. Inhalation of PM, particularly PM of aerodynamic diameter < 2.5 µm (PM2.5) and smaller, leads to inflammation and subsequent production of reactive oxygen species (ROS) (6). The production of ROS, which include e.g. the hydroxyl radical, superoxide anion or hydrogen peroxide, is caused by both the physical effects of PM (PM is phagocyted by macrophages that consequently produce ROS) and the presence of various chemicals on the surface of PM (e.g. metals, PAHs) with pro-oxidant properties. Several metabolic pathways of PAH activation have been described (7), and one of them, activation through PAH-o-quinones, leads to ROS generation and oxidative stress (8). Benzene, a ubiquitous VOC, is a carcinogenic compound metabolised in the organism by CYP2E1 into benzene oxide, which is further metabolised into several products including reactive quinones that again cause oxidative stress by redox cycling (9). Although ROS play important roles in many physiological processes including gene expression regulation, if there is an imbalance between the levels of ROS and antioxidants, ROS cause oxidative damage to cellular macromolecules. The modification of DNA molecules represents the most serious form of impact of ROS on the organism because it may lead to base changes, mutations and/or DNA breaks. 8-Oxo-7,8- dihydro-2 -deoxyguanosine (8-oxodG) is the most abundant product of oxidation of bases in DNA and has become a widely used biomarker of oxidative damage in subjects both suffering from various diseases and exposed to xenobiotics (10). If ROS attack both DNA strands, double-strand DNA breaks may appear. These breaks may lead either to unstable chromosomal aberrations or, if homologous or non-homologous end-joining repair seals the breaks, to stable chromosomal translocations. Translocations are more serious because they are usually fixed in the genome and may lead to rearrangements of regulatory elements and genes, including oncogenes, thus increasing cancer risk (11). Another indirect mechanism of induction of DNA double-strand breaks is associated with formation of DNA adducts. Adducts may cause persistent blockage of one DNA strand during its synthesis and uncoupling of the other strand, which may result in the formation of double-strand breaks (12). The attack of ROS on lipids leads to lipid peroxidation. This reaction may have potentially serious consequences, as it may damage cellular membrane and inactivate membranebound receptors or enzymes. In addition, secondary products of lipid peroxidation, such as aldehydes, are highly reactive and may propagate oxidative stress by reacting with other cellular molecules including proteins (13). Currently, isoprostanes are The Author Published by Oxford University Press on behalf of the UK Environmental Mutagen Society. All rights reserved. For permissions, please journals.permissions@oup.com. 97

2 P. Rossner et al. considered the most reliable markers of lipid peroxidation. These prostaglandin-like compounds, first described in the 1990s, are formed by free-radical induced peroxidation of arachidonic acid, independent of cyclooxygenase enzymes (14). Lipid peroxidation products, including isoprostanes, play a role in the pathogenesis of many diseases (15). Apart from reactive aldehydes, proteins may also be damaged directly by ROS or by reactive sugars. These reactions generate carbonyl groups (aldehyde or ketone groups) mostly on side chains of protein molecules. These modifications result in alterations of protein structure and/or function. Damaged proteins are recognised by the proteolytic system of the cell and degraded by proteasomes (16). However, this system may be inefficient and damaged proteins may accumulate in the organism. Carbonyl groups in damaged proteins may then be detected as a biomarker of protein oxidation. In order to monitor effects of heavy air pollution on levels of selected biomarkers, we collected samples in two regions of the Czech Republic that differ in concentrations of environmental pollutants: the Ostrava region where the population has been exposed to polluted air for several generations (see description in 17) and the capital city of Prague where concentrations of pollutants are substantially lower. To measure seasonal variability of pollutants and biomarkers we collected samples in the winter and summer of 2009 and winter of We hypothesised that long-term exposure to high concentrations of pollutants in subjects from the Ostrava region will result in elevated levels of the studied biomarkers. Because of higher concentrations of air pollutants caused by increased burn of fossil fuels, the presence of industrial sources and frequent inversions in winter, we also expected the levels of biomarkers to be higher in winter samples than in the summer season. In the first part of our study (17) we concentrated on bulky DNA adducts induced by reactive metabolites of carcinogenic polycyclic aromatic hydrocarbons (c-pahs). Despite higher B[a]P air pollution in the Ostrava region during all sampling periods, the levels of B[a]P-like DNA adducts/10 8 nucleotides were significantly higher in the Ostrava subjects only in winter During the other two sampling periods, the levels of Table 1. Basic characteristics of the study population (see 17 for details) B[a]P-like bulky DNA adducts were significantly higher in the Prague subjects (P < 0.001). Multivariate analyses conducted among subjects from Ostrava and Prague separately during all sampling periods revealed that exposure to B[a]P and PM2.5 significantly increased levels of B[a]P-like DNA adducts in the Ostrava subjects, but not in subjects from Prague. In the present study we investigated the effect of air pollutants on urinary excretion of 8-oxodG, lipid peroxidation and protein oxidation, as well as unstable and stable chromosomal aberrations in peripheral blood lymphocytes. Materials and methods Subjects, sampling and exposure assessment The number of subjects participating in the study varied from 61 to 65 (Prague) and from 98 to 149 (Ostrava region) depending on the sampling season. Other characteristics of the subjects including age, exposure to tobacco smoke, body mass index (BMI), plasma levels of vitamin C, A and E, cholesterol, LDLcholesterol, HDL-cholesterol and triglycerides, as well as their exposure to environmental pollutants, are detailed in the first part of the study (17). There we also discuss geographical and historical specifics of the Ostrava region. We briefly report some of subjects characteristics and exposure to environmental pollutants in Table 1. Analysis of oxidative stress markers 8-oxodG ELISA. Urinary 8-oxodG levels were analysed by competitive ELISA essentially as previously described (18,19). Wells were coated with 5 ng of 8-oxoG conjugated with bovine serum albumin (BSA; total volume, 50 µl/well) by drying the plates overnight at 37 C. Plates were washed with phosphate buffered saline (PBS)/Tween (0.05% Tween 20 in PBS) and blocked with 200 µl/well of blocking buffer (1% FCS in PBS/Tween) for 1 h at 37 C. After blocking, 50 µl of 8-oxodG standards (concentration range, ng/ ml) and urine samples (diluted 1:1 with PBS) were added followed by 50 µl of primary antibody (JaICA, Japan, clone N45.1, concentration 0.2 µg/ml). After incubation for 1.5 h at 37 C and washing, 100 µl of secondary antibody conjugated with alkaline phosphatase was added. Further incubation for 1.5 h at 37 C was followed by washing with PBS/Tween and with 0.01% diethanolamine in water. The colour was developed by adding 100 µl of p-nitrophenyl phosphate substrate (1 mg/ml of 1 mol/l diethanolamine) per well and incubating the plates for min at 37 C. The absorbance was measured with a microplate reader at 405 nm. Any samples with inhibition <20% or >80% were repeatedly analysed either without dilution or with further dilution, respectively. Each sample was analysed in triplicate. Urinary 8-oxodG concentration was expressed as nmol 8-oxodG/mmol creatinine. To control for the interassay Prague Ostrava region P Winter 2009 Cotinine (ng/mg creat.) 46.1 ± ± Vitamin C (mg/l) 8.03 ± ± 3.14 <0.001 Vitamin A (mg/l) 1.20 ± ± 0.42 <0.001 Vitamin E (mg/l) 14.3 ± ± 6.08 <0.001 Benzo[a]pyrene (ng/m 3 ) 0.78 ± ± 3.03 <0.001 Benzene (µg/m 3 ) 5.40 ± ± 10.0 <0.001 PM2.5 (µg/m 3 ) 13.8 ± ± 21.1 <0.001 Summer 2009 Cotinine (ng/mg creat.) 71.9 ± ± 21.7 <0.05 Vitamin C (mg/l) 8.59 ± ± 8.50 N.S. Vitamin A (mg/l) 0.93 ± ± 0.30 N.S. Vitamin E (mg/l) 10.9 ± ± 4.40 N.S. Benzo[a]pyrene (ng/m 3 ) 0.12 ± ± 0.24 <0.001 Benzene (µg/m 3 ) 3.36 ± ± 14.5 <0.001 PM2.5 (µg/m 3 ) 13.3 ± ± 4.06 N.S. Winter 2010 Cotinine (ng/mg creat.) 91.3 ± ± 66.3 <0.001 Vitamin C (mg/l) 8.40 ± ± 3.93 <0.001 Vitamin A (mg/l) 0.79 ± ± 0.42 <0.001 Vitamin E (mg/l) 9.84 ± ± 9.57 <0.001 Benzo[a]pyrene (ng/m 3 ) 2.80 ± ± 13.3 <0.001 Benzene (µg/m 3 ) 5.66 ± ± 10.7 <0.001 PM2.5 (µg/m 3 ) 42.6 ± ± 28.6 <0.001 PM2.5 = particulate matter of aerodynamic diameter < 2.5 µm. 98

3 Chromosomal aberrations, oxidative stress and air pollution variability, a control sample was analysed on every plate, and the interassay coefficient of variability was calculated. For the analysis of 8-oxodG, the interassay coefficient of variability was 5.7%. 15-F2t-IsoP immunoassay. Plasma 15-F2t-IsoP levels were analysed using immunoassay kits from Cayman Chemical Company (Ann Arbor, MI, USA) according to the manufacturer s protocol. For the assay, 125 µl of plasma was used, samples were hydrolysed to remove isoprostanes esterified in lipoproteins and were purified using an affinity sorbent provided by the kit manufacturer. According to Cayman, purified plasma samples analysed by immunoassay give excellent correlation to GC/MS. Samples were further diluted 1:2 in the assay buffer, and the assay was then carried out. Each sample was analysed in duplicate. The 15-F2t-IsoP concentrations were expressed as pg 15-F2t-IsoP/ ml plasma. Protein carbonyl assay. The levels of protein carbonyl groups were assessed in blood plasma using a noncompetitive ELISA, as previously described (20), with some modifications (19,21). Briefly, the oxidised protein standards were prepared by incubation of BSA (50 mg/ml) with 0.73 M H 2 O 2 and 0.42 mm Fe 2+ for 1 h at 37 C. The reaction was stopped with 40 µm butylated hydroxytoluene. The carbonyl content of the oxidised BSA standard was measured spectrophotometrically. It was then diluted with native (unoxidised) BSA and PBS to give a final carbonyl content of 2.0 nmol/mg protein and a protein concentration of 4 mg/ml. Total protein concentration in the plasma samples was measured using a Bicinchoninic Acid kit, and the samples were diluted with PBS to a final protein concentration of 4 mg/ml. After the derivatisation with 2,4-dinitrophenylhydrazine, the plate was coated with 200 µl of sample and incubated overnight at 4 C in the dark. The plate was washed with PBS/ Tween (0.05% Tween 20 in PBS) and blocked with 0.1% BSA in PBS for 1.5 h. After another washing step, biotinylated primary anti-dnp antibody (Molecular Probes, OR, USA; diluted 1:1500 with 0.1% BSA, 0.1% Tween 20 in PBS), was added, and the plate was incubated at 37 C for 1 h. Another washing was followed by the addition of streptavidin biotinylated horseradish peroxidase conjugate (Amersham Biosciences, UK; diluted 1:4000 in 0.1% BSA, 0.1% Tween 20 in PBS) and incubated at room temperature for 1 h. Colour was developed by adding the tetramethyl benzidine liquid substrate system, and the reaction was stopped with H 2 SO 4 after min incubation in the dark. The absorbance was measured with a microplate reader at 450 nm. Each sample was analysed in triplicate. Plasma protein carbonyl concentration was expressed as nmol carbonyl/ml plasma. The interassay coefficient of variability was 3.1%. Cytogenetic analyses Whole venous blood cultures. Whole venous blood cultures were established within 24 h of blood collection, according to the method described by Rossner et al. (22). Cell cultures were cultivated at 37 C in RPMI 1640 medium supplemented with 20% calf serum and 1% phytohemagglutinin. Two duplicate cultures were set up from each sample, which were harvested after 72 h of incubation and processed for both FISH and MN analyses. In the case of FISH cultures, colchicine (Fluka) was added to a final concentration of 0.5 μg/ml 2 h before the end of incubation. In the case of MN cultures, cytochalasin B (Sigma) was added to a final concentration of 5 µg/ml at 44 h (23). After 72 h of incubation, both FISH and MN cultures were harvested using a similar technique that included centrifugation, treatment with a hypotonic solution of KCl (0.075 M) and repeated fixation with methanol/ acetic acid. Fluorescence in situ hybridisation. The cell suspensions were stored at 20 C until painting by FISH. One day before painting, the cell suspensions were dropped on cold and wet slides and left to dry at room temperature overnight. FISH analysis using commercial whole chromosome painting probes differing in colour (Cambio, UK) for chromosomes #1 (biotinylated) and #4 (FITC-labelled) was carried out according to the manufacturer s chromosome painting protocol adapted by Beskid et al. (24). Probes were placed on slides, sealed with rubber cement, and incubated at 37 C in a moisture chamber overnight. After washing and detecting the signal from chromosome #1 with Streptavidin-Cy3, the slides were mounted in antifade solution Vectashield (Vector Laboratories, Burlingame, CA, USA) with DAPI to counterstain unpainted chromosomes. The slides from each culture were coded and stored at 4 C in the dark until analysis. One thousand metaphases were examined for each subject under a fluorescent microscope equipped with a triple-band pass filter for visualisation of DAPI (blue), FITC (green) and Cy3 (red) signals. All aberrant cells were classified according to the Protocol for Aberration Identification and Nomenclature (25). The genomic frequencies (F G ) of stable chromosome exchanges were calculated according to Lucas et al. (26) using the equation: F G = F rg /2.05 [f r (1 f r ) + f g (1 f g ) f r f g ]. F rg is the translocation frequency measured by FISH after two-colour painting, while f r (red) and f g (green) are the fractions of the genome (27) (chromosomes #1 and #4 represent 8.28% and 6.39% of the human genome, respectively) painted red and green, respectively. Micronuclei automated image analysis. After processing the cultures and preparation of slides, the slides were dried and stained by DAPI for automated image analysis. The automated scanning system Metafer 4, Version 3.2.1, from MetaSystems (Atlussheim, Germany) with a motorised Axio Imager Z1 microscope (Carl Zeiss, Germany) was used (28). Scanning of binucleated cells (BNC) was performed using a 10 objective (final magnification 100 ) and a DAPI filter, and 3000 BNC per subject were analysed. In this study, the MicroNuc Classifier, Version , was used for the nuclei and micronuclei image processing operation with the same settings as in our previous studies (29). All automated findings with one or more MN were checked and corrected if necessary by one person. The results were calculated as mean MN per 1000 BNC (MN/1000 BNC). Statistical analysis We first compared genomic frequencies of translocations (F G /100), percentage of aberrant cells (%AB.C.), acentric fragments (ace), frequencies of micronuclei per 1000 binucleated cells (MN/1000 BNC), levels of DNA oxidation (8-oxodG), lipid peroxidation (15-F2t-IsoP) and protein oxidation (protein carbonyls) for individual sampling periods and locations. We used the Mann Whitney U-test for those variables that did not follow a normal distribution and the t-test for data distributed normally. A relationship between percentiles of exposure to B[a]P and 15-F2t-IsoP concentrations in blood plasma was tested using the Jonckheere Terpstra test and the Mann Whitney U-test. Associations between environmental pollutants and biomarkers of chromosomal damage and oxidative stress were studied using generalised estimating equations (GEE); cytogenetic parameters were analysed using binary logistic models, and for oxidative stress markers, linear models were applied. Multivariate analyses were adjusted for age, cotinine, BMI, vitamin C, A and E, cholesterol, LDL-cholesterol, HDL-cholesterol and triglycerides. To normalise the distribution of 15-F2t-IsoP and protein carbonyl levels for linear models, log-transformed values were used. Statistical analyses were performed using SPSS 19.0 software. Results The basic characteristics of the study population and their comparison in subjects from both locations were discussed in the first part of our study (17); some of these characteristics are shown in Table 1. In this article we report descriptive statistics and a comparison of cytogenetic parameters and oxidative stress markers in the studied locations and their associations with exposure to environmental pollutants. Stable and unstable chromosomal aberrations In Table 2 we show mean and median levels of stable chromosomal translocations (F G /100 and %AB.C.) in subjects from Prague and the Ostrava region. Although inhabitants of the Ostrava region are constantly exposed to significantly higher levels of air pollutants than subjects from Prague, we did not observe an increase in the frequency of stable chromosomal aberrations. In fact, the frequencies were comparable in both locations. The winter 2010 sampling in Ostrava city was the only exception; however, here we observed a significantly lower genomic frequency of translocations/100 cells than in subjects from Prague [median (min, max): 0.75 (0, 5.23) and 1.12 (0, 4.11) for Ostrava city and Prague, respectively; P < 0.05]. It is also noteworthy that we did not find any significant seasonal variability for either cytogenetic parameter. The frequencies of unstable chromosomal aberrations (ace and MN/1000 BNC) are reported in Table 3. The number of acentric fragments was comparable in subjects from both locations for all seasons with only one exception: in the Ostrava region in winter 2009, the frequencies of ace were significantly lower (P < 0.05). The frequencies of MN had a similar trend, although the number of significant results was substantially higher than that for acentric fragments. Thus, in both seasons of 2009, the frequencies of MN/1000 BNC were significantly lower (P < or P < 0.05) in subjects from the Ostrava region than in subjects 99

4 P. Rossner et al. Table 2. Genomic frequency of translocations and percentage of aberrant cells in peripheral blood lymphocytes of study subjects F G /100 %AB.C. Mean ± SD Median (min, max) Mean ± SD Median (min, max) Winter 2009 Prague (N = 61) 1.43 ± (0.00, 5.60) 0.27 ± (0.00, 0.90) Ostrava region (N = 98) 1.44 ± (0.00, 6.72) 0.26 ± (0.00, 1.00) Ostrava city (N = 74) 1.46 ± (0.00, 6.72) 0.26 ± (0.00, 1.00) Karvina (N = 24) 1.38 ± (0.00, 4.11) 0.26 ± (0.00, 0.70) Summer 2009 Prague (N = 61) 1.34 ± (0.00, 4.85) 0.23 ± (0.00, 0.70) Ostrava region (N = 100) 1.39 ± (0.00, 6.72) 0.25 ± (0.00, 0.80) Ostrava city (N = 64) 1.44 ± (0.00, 6.72) 0.25 ± (0.00, 0.80) Karvina (N = 24) 1.31 ± (0.00, 3.36) 0.23 ± (0.00, 0.50) Havirov (N = 12) 1.34 ± (0.00, 2.99) 0.24 ± (0.00, 0.40) Winter 2010 Prague (N = 65) 1.39 ± (0.00, 4.11) 0.23 ± (0.00, 0.60) Ostrava region (N = 149) 1.25 ± (0.00, 6.35) 0.22 ± (0.00, 0.90) Ostrava city (N = 78) 1.11 ± (0.00, 5.23) a 0.20 ± (0.00, 0.70) Bartovice (N = 28) 1.21 ± (0.00, 6.35) 0.20 ± (0.00, 0.80) Karvina (N = 31) 1.44 ± (0.00, 4.48) 0.27 ± (0.00, 0.60) Havirov (N = 12) 1.82 ± (0.00, 4.48) 0.32 ± (0.00, 0.90) a P < 0.05; comparison of Ostrava city vs. Prague; %AB.C. = percentage of aberrant cells; F G /100 = genomic frequency of translocations/100 cells. from Prague. In the winter 2010 sampling, the frequencies of MN/1000 BNC were comparable for all locations. In summary, high levels of air pollutants in the Ostrava region failed to induce higher genetic damage in subjects from this location; moreover, the frequencies of chromosomal translocations were in some cases lower in subjects from the more polluted region. Oxidative damage to macromolecules The levels of oxidative stress markers in study subjects from all locations in individual sampling periods are compared in Table 4. Urinary excretion of 8-oxodG did not differ between locations in any season. Lipid peroxidation measured as plasma levels of 15-F2t-IsoP was also comparable between locations, but only in the winter 2010 sampling when concentrations of air pollutants were highest. In both 2009 samplings, lipid peroxidation was elevated in the Ostrava subjects. The difference was more pronounced in winter 2009; median (min, max) values of 15-F2t-IsoP were as follows: summer 2009: (53.0, 262.1) vs (64.8, 376.5), P < 0.05; winter 2009: (101.3, ) vs (108.7, 321.2), P < for Ostrava region and Prague, respectively. Interestingly, it seems that there is a non-linear dose response relationship between percentiles of exposure to B[a]P and 15-F2t-IsoP concentrations in blood plasma: increasing exposure to B[a] P leads to a proportional increase in 15-F2t-IsoP levels up to the 83.3 percentile (B[a]P concentration of about 8.82 ng/m 3 ). Past this point, plasma levels of 15-F2t-IsoP do not increase (Figure 1). Multivariate analyses of factors affecting cytogenetic parameters and oxidative stress markers The role of air pollutants (B[a]P, PM2.5, benzene), measured by personal and stationary monitors for 24 and 48 h before sampling (17), in the modulation of levels of cytogenetic parameters and oxidative stress biomarkers was estimated by GEE after adjustment for round, age, BMI, cotinine, levels of vitamin C, A and E, total cholesterol, HDL-cholesterol, LDL-cholesterol and triglycerides. Because of the lack of significant differences between individual locations in the Ostrava region for most of the biomarkers we pooled the data together into the Ostrava region Table 3. Frequency of acentric fragments assessed by FISH and micronuclei in peripheral blood lymphocytes of study subjects Ace MN/1000 BNC Mean ± SD Median (min, max) Mean ± SD Median (min, max) Winter 2009 Prague (N = 61) 0.72 ± (0.00, 6.00) 7.24 ± (3.00, 12.0) Ostrava region (N = 98) 0.33 ± (0.00, 4.00) b 5.51 ± (0.04, 12.9) a Ostrava city (N = 74) 0.36 ± (0.00, 4.00) 3.83 ± (0.00, 7.00) a Karvina (N = 24) 0.21 ± (0.00, 1.00) 3.57 ± (1.00, 6.00) a Summer 2009 Prague (N=61) 0.41 ± (0.00, 4.00) 10.2 ± (4.00, 20.0) Ostrava region (N = 100) 0.49 ± (0.00, 5.00) 7.26 ± (2.00, 17.0) a Ostrava city (N = 64) 0.55 ± (0.00, 5.00) 7.15 ± (2.00, 16.0) a Karvina (N = 24) 0.41 ± (0.00, 3.00) 7.15 ± (3.00, 12.0) a Havirov (N = 12) 0.33 ± (0.00, 2.00) 8.08 ± (4.00, 17.0) b Winter 2010 Prague (N = 65) 0.30 ± (0.00, 4.00) 6.94 ± (3.00, 14.0) Ostrava region (N = 149) 0.39 ± (0.00, 8.00) 6.66 ± (2.00, 15.0) Ostrava city (N = 78) 0.29 ± (0.00, 3.00) 6.81 ± (3.00, 14.0) Bartovice (N = 28) 0.25 ± (0.00, 2.00) 6.80 ± (3.00, 15.0) Karvina (N = 31) 0.71 ± (0.00, 8.00) 6.26 ± (2.00, 15.0) Havirov (N = 12) 0.50 ± (0.00, 2.00) 6.44 ± (4.00, 9.00) a P < 0.001, b P < 0.05; comparison of the Ostrava groups vs. Prague; ace = frequency of acentric fragments; MN/1000 BNC = frequency of micronuclei per 1000 binucleated cells. 100

5 Chromosomal aberrations, oxidative stress and air pollution Table 4. Oxidative stress markers in urine and plasma of study subjects 8-oxodG (nmol/mmol creat.) 15-F2t-IsoP (pg/ml plasma) Protein carbonyl (nmol/ml plasma) Winter 2009 Summer 2009 Winter 2010 group. The results of multivariate analyses conducted among all subjects irrespective of the season, but separated by the location (Prague or Ostrava region), are reported in Tables 5 and 6. Environmental pollutants had no significant effect on stable chromosomal aberrations and the frequency of acentric fragments. In both regions, the frequency of micronuclei/1000 BNC was found to be negatively associated with exposure to B[a]P measured by both types of monitors and PM2.5; exposure to benzene measured by personal monitors did not affect micronuclei levels (Table 5). Mean ± SD Median (min, max) Mean ± SD Median (min, max) Mean ± SD Median (min, max) Prague (N = 61) 5.27 ± (0.20, 14.4) ± (108.7, 321.2) 23.1 ± (16.4, 35.4) Ostrava region (N = 98) 5.51 ± (0.04, 12.9) ± (101.3, ) a 23.2 ± (16.2, 34.9) Ostrava city (N = 74) 5.43 ± (0.04, 11.2) ± (101.3, ) a 22.8 ± (16.2, 34.9) Karvina (N = 24) 5.77 ± (0.18, 12.9) ± (112.3, ) a 24.5 ± (18.3, 31.4) c Prague (N = 61) 5.27 ± (0.10, 20.4) ± (64.8, 376.5) 22.3 ± (16.7, 27.8) Ostrava region (N = 100) 5.26 ± (0.09, 18.4) ± (53.0, 262.1) c 19.8 ± (13.7, 32.7) a Ostrava city (N = 64) 4.91 ± (0.09, 11.6) ± (60.9, 262.1) 20.2 ± (13.7, 32.7) a Karvina (N = 24) 5.63 ± (0.15, 18.4) ± (54.1, 247.1) b 18.3 ± (14.8, 24.4) a Havirov (N = 12) 6.40 ± (2.58, 10.3) ± (53.0, 155.0) 21.0 ± (18.5, 23.9) Prague (N = 65) 4.78 ± (0.52, 9.86) ± (119.6, 647.8) 23.6 ± (10.0, 39.1) Ostrava region (N = 149) 4.68 ± (0.19, 10.4) ± (90.6, 814.6) 22.0 ± (11.2, 43.3) Ostrava city (N = 78) 4.29 ± (0.19, 10.4) ± (90.6, 814.6) 22.4 ± (11.2, 43.3) Bartovice (N = 28) 5.51 ± (1.85, 9.08) ± (113.2, 487.4) 20.0 ± (12.9, 26.0) Karvina (N = 31) 4.61 ± (0.32, 7.17) ± (119.1, 655.2) 22.5 ± (14.4, 30.5) Havirov (N = 12) 5.47 ± (2.42, 7.41) ± (135.5, 667.5) 22.8 ± (19.8, 25.5) a P < 0.001, b P = 0.001, c P < 0.05; comparison of the Ostrava groups vs. Prague; 8-oxodG = 8-oxo-7,8-dihydro-2 -deoxyguanosine; 15-F2t-IsoP = 15-F 2t -isoprostane. Urinary excretion of 8-oxodG was not affected by any of the analysed air pollutants (Table 6). Lipid peroxidation was significantly increased after exposure to B[a]P, PM2.5 and benzene in subjects from Prague; in the Ostrava region, the results were similar with the exception of benzene that had no effect on 15-F2t-IsoP levels (Table 6). The effect of air pollution on protein oxidation in Prague was minor: PM2.5 was the only pollutant that significantly affected protein carbonyl levels. In contrast, in the Ostrava region, protein oxidation increased significantly after exposure to Figure 1. Boxplots showing associations between plasma levels of 15-F2t-IsoP and percentiles of B[a]P among all subjects ( B[a]P percentile: P for trend < 0.001; comparison of 15-F2t-IsoP levels between 83.3 and 100 percentile: P = 0.359). Two outliers (1640 pg IsoP/ml plasma in the 50 percentile and 2410 pg IsoP/ml plasma in the 100 percentile) are not shown. 15-F2t-IsoP = 15-F 2t -isoprostane; B[a]P = benzo[a]pyrene. 101

6 P. Rossner et al. Table 5. Association of selected variables with cytogenetic parameters Variable F G /100 %AB.C. B *, 95% CI, P B *, 95% CI, P Prague Ostrava region Prague Ostrava region B[a]P (ng/m 3 ) a ( 0.183, 0.231), ( 0.024, 0.029), ( 0.329, 0.116), ( 0.026, 0.022), Benzene (µg/m 3 ) a ( 0.079, 0.035), ( 0.043, 0.006), ( 0.057, 0.032), ( 0.032, 0.025), B[a]P (ng/m 3 ) b ( 0.446, 0.530), ( 0.008, 0.035), ( 0.694, 0.221), ( 0.004, 0.036), PM2.5 (µg/m 3 ) b ( 0.029, 0.035), ( 0.008, 0.009), ( 0.042, 0.015), ( 0.008, 0.008), Variable Ace MN/1000 BNC B[a]P and PM2.5 measured by stationary monitors, while benzene had an unexpected negative effect on protein carbonyl levels (Table 6). Because of the generally weak effects of air pollution exposure, measured immediately before the collection of biological material, on the levels of biomarkers (particularly cytogenetic parameters), we decided to further analyse in more detail whether biomarkers could be affected by exposure to air pollutants for longer time periods before sampling. The results of multivariate-adjusted GEE analysis of the effect of air pollutants on the percentage of aberrant cells in subjects from each region separately are reported in Table 7. These results indicate Table 6. Association of selected variables with oxidative stress markers Variable 8-oxodG log (15-F2t-IsoP) B*, 95% CI, P that no significant association was found in subjects from the Ostrava region. However, in Prague subjects the percentage of aberrant cells was significantly elevated after exposure to benzene over a 14-day interval, days before sampling, exposure to B[a]P over a 14-day interval, days before sampling and exposure to PM2.5 over a 14-day interval immediately preceding collection of the samples. The role of other factors in the modulation of frequencies of chromosomal aberrations and oxidative stress markers is reported in supplementary Tables 1 4, available at Mutagenesis Online. As expected, the frequencies of stable chromosomal aberrations and micronuclei/1000 BNC were B*, 95% CI, P Prague Ostrava region Prague Ostrava region B[a]P (ng/m 3 ) a ( 0.120, 0.148), ( 0.053, 0.011), (0.011, 0.057), (0.003, 0.007), <0.001 Benzene (µg/m 3 ) a ( 0.066, 0.015), ( 0.028, 0.011), (0.0, 0.004), (0.0, 0.003), B[a]P (ng/m 3 ) b ( 0.464, 0.507), ( 0.034, 0.013), (0.105, 0.157), < (0.004, 0.007), <0.001 PM2.5 (µg/m 3 ) b ( 0.026, 0.030), ( 0.009, 0.010), (0.007, 0.010), < (0.002, 0.003), <0.001 Variable B *, 95% CI, P log (protein carbonyl) B *, 95% CI, P Prague Ostrava region Prague Ostrava region B[a]P (ng/m 3 ) a ( 0.261, 0.231), ( 0.029, 0.026), ( 0.723, 0.150), ( 0.095, 0.029), <0.001 Benzene (µg/m 3 ) a ( 0.037, 0.050), ( 0.027, 0.023), ( 0.169, 0.025), ( 0.039, 0.023), B[a]P (ng/m 3 ) b ( 0.575, 0.493), ( 0.026, 0.017), (1.368, 0.307), ( 0.096, 0.038), <0.001 PM2.5 (µg/m 3 ) b ( 0.049, 0.022), ( 0.011, 0.005), ( 0.114, 0.034), < ( 0.042, 0.022), <0.001 %AB.C. = percentage of aberrant cells; ace = frequency of acentric fragments; B[a]P = benzo[a]pyrene; F G /100 = genomic frequency of translocations/100 cells; MN/1000 BNC = frequency of micronuclei per 1000 binucleated cells; PM2.5 = particulate matter of aerodynamic diameter < 2.5 µm. a Data from personal monitors. b Data from stationary monitors. *Adjusted to round, age, BMI, cotinine, vitamin C, A, E, total cholesterol, HDL-cholesterol, LDL-cholesterol, triglycerides. B *, 95% CI, P Prague Ostrava region B[a]P (ng/m 3 ) a ( 0.013, 0.004), ( 0.001, 0.001), Benzene (µg/m 3 ) a ( 0.002, 0.001), ( 0.001, 0.0), B[a]P (ng/m 3 ) b ( 0.001, 0.026), (0.001, 0.002), <0.001 PM2.5 (µg/m 3 ) c (0.001, 0.003), (0.0, , oxodG = 8-oxo-7,8-dihydro-2 -deoxyguanosine; 15-F2t-IsoP = 15-F 2t -isoprostane; B[a]P = benzo[a]pyrene; PM2.5 = particulate matter of aerodynamic diameter < 2.5 µm. a Data from personal monitors. b Data from stationary monitors.* Adjusted to round, age, BMI, cotinine, vitamin C, A, E, total cholesterol, HDL-cholesterol, LDL-cholesterol, triglycerides. 102

7 Chromosomal aberrations, oxidative stress and air pollution Table 7. Association of exposure to environmental pollutants measured by stationary monitors in various periods before sample collection with %AB.C Variable Period before sampling (days) %AB.C. Concentration of pollutant Mean ± SD Median (min, max) OR *, 95% CI, P; q Prague Ostrava Prague Ostrava Prague Ostrava Benzene (µg/m 3 ) ± ± ( ) 3.24 ( ) 2.43 ( ), 0.008; ( ), 0.36; 0.26 B[a]P (ng/m 3 ) ± ± ( ) 3.89 ( ) 2.24 ( ), 0.012; ( ), 0.27; 0.31 PM2.5 (µg/m 3 ) ± ± ( ) 55.6 ( ) 2.43 (1.26, 4.68), 0.008; ( ), 0.24; 0.13 %AB.C. = percentage of aberrant cells; B[a]P = benzo[a]pyrene; PM2.5 = particulate matter of aerodynamic diameter < 2.5 µm. *Adjusted to round, age, BMI, cotinine, vitamin C, A and E, total cholesterol, HDL-cholesterol, LDL-cholesterol and triglycerides. positively associated with the age of study participants (supplementary Tables 1 and 2, available at Mutagenesis Online). In the Ostrava region, the frequencies of stable chromosomal aberrations (F G /100 and %AB.C.) were negatively associated with the sampling period (the samples collected in winter 2010 tended to have lower levels of stable chromosomal aberrations), while the frequency of MN in subjects from both Prague and Ostrava region was positively associated (supplementary Tables 1 and 2, available at Mutagenesis Online). In subjects from both regions, positive associations between lipid peroxidation and plasma levels of HDL cholesterol and triglycerides were detected (supplementary Table 3, available at Mutagenesis Online). Interestingly, there was a negative association between levels of vitamin C and urinary excretion of 8-oxodG, as well as plasma levels of 15-F2t-IsoP; however, this association was observed only in subjects from the Ostrava region (supplementary Table 3, available at Mutagenesis Online). Discussion In this study, we analysed the effect of exposure to environmental pollutants in subjects living in the industrial Ostrava region characterised by very high ambient air concentrations of polycyclic aromatic hydrocarbons, volatile organic compounds and other contaminants. The concentrations of pollutants in the Ostrava region are so high that we were able to use subjects from the capital city of Prague as a control group, although Prague is also considered a polluted location. We originally planned to study biomarkers in city policemen from Ostrava and Prague. Unfortunately, due to negative attitude of the municipal authority of Ostrava, a founder of city police, we were not allowed to collaborate with city police in Ostrava and were forced to seek other options (for details see 17). We selected city policemen from two polluted areas rather than general populations from Ostrava and a rural region particularly due to logistic reasons. Selecting subjects from organised police forces makes obtaining matched groups much easier. The same is true for collecting biological samples and filters from personal monitors. Theoretically, subjects from the rural region could have been exposed to lower levels of air pollution and could serve as a better control group, but our recent study suggests that even in rural areas concentrations of air pollutants can be quite high, even exceeding the levels in Prague (30). Analysis of stable or unstable chromosomal aberrations has been repeatedly used to monitor the effect of air pollutants on human health. However, the authors mostly focused on occupationally exposed subjects (31 36). These studies indicate that unstable chromosomal aberrations represent a suitable biomarker for evaluating the effects of air pollution on human health, at least in occupationally exposed populations (37). We previously analysed both stable and unstable chromosomal aberrations in environmentally exposed groups from various locations (24,29,30,38 41) including the Ostrava region (42,43); however, the present study is the first complex analysis of the seasonal variability of biomarkers in Ostrava subjects and the comparison of the two locations: Ostrava and Prague. Despite differences in the concentrations of air pollutants we did not observe a corresponding trend for the frequency of stable aberrations. Unstable aberrations, measured as the frequency of MN, were elevated in Prague subjects in 2009 although they were exposed to significantly lower concentrations of air pollutants. In winter 2010, we did not observe any difference in the frequency of MN between studied groups, even though concentrations of air pollutants were highest in this season in both locations. All of these observations were unexpected and did not correspond with the results of many of our previous studies. We demonstrated that the FISH method could be successfully used to analyse the impact of environmental air pollution on stable chromosomal translocations in a group of police officers and unexposed controls originating from three European cities (Prague, Kosice, Sofia) (24). Interestingly, although exposure to B[a]P in subjects from Sofia was higher than that in subjects from the two other cities, the frequency of chromosomal aberrations in Bulgarian subjects was lower or comparable to that in the subjects from Prague and Kosice. In another study of city policemen from Prague, sampled in winter and spring 2004, a higher frequency of chromosomal aberrations was associated with higher concentrations of B[a]P in the ambient air (38). We further observed seasonal variability of F G /100 in a study analysing biomarkers in Prague bus drivers and in controls. However, the variability was limited to the control group; in bus drivers, F G /100 was not associated with exposure to B[a]P (41). Finally, we failed to detect seasonal variability in city policemen sampled in winter and spring 2007, probably due to opposite trends in the concentrations of c-pahs and PM2.5 (high concentration of c-pahs in winter and PM2.5 in spring) (40). In the same population we studied the frequency of MN. Unlike stable aberrations, the frequency of MN was significantly elevated in subjects sampled in the winter season, indicating that the two types of chromosomal aberrations differ in their response to air pollution exposure (29). This was further confirmed by a comparison of the frequency of MN in bus drivers and controls mentioned above where we were able to detect a significant difference between the two groups (29). In another study, women exposed 103

8 P. Rossner et al. to higher concentrations of B[a]P exhibited elevated frequencies of MN (30). The present study further confirms that the effect of air pollutants on stable and unstable aberrations differs: the frequencies of MN were the only biomarker for which we observed differences between the two locations even though the differences had an unexpected opposite trend: in Prague, the location with lower concentrations of air pollutants, the frequencies of MN were higher than that in the more polluted Ostrava region. At present we have no concrete evidence to explain these observations. We may speculate that other air pollutants not measured in our study modify the frequency of chromosomal aberrations. Cytogenetic damage may be further affected by various factors including age, gender, diet and lifestyle (44,45); however, we controlled for these factors partly by selection of the study population (all subjects were males, non-smokers of a similar age) and partly by including the factors as covariates in multivariate analyses. Thus, they should not affect the results of our study. We may also speculate that an adaptive response may have developed in subjects from the Ostrava region that probably causes resistance to high concentrations of air pollutants. This hypothesis is also supported by the results of our previous small study consisting of eight participants. All subjects resided in Prague, but during our experiment four moved to the Ostrava region where they spent 3 weeks at the time of extreme air pollution. The other four participants stayed in Prague. All subjects donated blood at the beginning and at the end of the experiment, and MN analysis was performed. While the frequency of MN remained unchanged in subjects staying in Prague, in the participants who spent 3 weeks in the Ostrava region the levels of MN increased (29). We also analysed stable chromosomal aberrations and global gene expression in these subjects. While F G /100 and %AB.C. increased by about 60% in subjects staying in the Ostrava region, gene expression profiles remained unchanged in both groups (unpublished data). These findings are in contrast with the data obtained from the subjects living permanently in Ostrava who seemed to be resistant to changes in the concentrations of air pollutants. Although no associations between cytogenetic parameters and concentrations of air pollutants measured immediately before sample collection were observed in the present study, the percentage of aberrant cells in Prague samples was positively affected by exposure to benzene, B[a]P and PM2.5 at various periods up to 70 days before sampling. We have repeatedly shown that exposure to B[a]P several weeks before blood collection significantly increases the frequencies of aberrant cells (38,40). In the present study, a similar result was observed for Prague subjects. Despite higher concentrations of pollutants in Ostrava, no such associations between any of the analysed compounds and %AB.C. were detected. This again suggests the uniqueness of the Ostrava population. Contrary to our expectations, we did not observe any difference in urinary excretion of 8-oxodG either between locations or between seasons. Exposure to environmental pollutants is known to induce oxidative stress that may result in increased concentrations of 8-oxodG in urine (reviewed in 10). Numerous studies have reported elevated levels of 8-oxodG in urine after exposure to traffic exhaust (19,46 51), c-pahs (52 54) or metals (55,56). Apart from c-pahs and metals that are usually bound to particulate matter, PM itself is an important factor causing ROS generation and thus oxidative stress in organisms due to inflammation (3). However, it has been shown that the chemical composition of 104 PM affects its ability to induce oxidative stress in organisms (57). Although we analysed the content of c-pahs, VOC and PM in samples collected by personal and stationary monitors and found significant differences between both locations and seasons, we did not analyse the concentrations of metals or other compounds with potential pro-oxidant properties. Thus, we may speculate that the presence of other compounds increased oxidative stress levels in the summer season when concentrations of c-pahs, VOC and PM were relatively low. The urinary excretion of 8-oxodG may be further affected by lifestyle factors. A recent study among 361 healthy male subjects indicated that urinary 8-oxodG excretion is inversely correlated with fruit consumption, physical activity and total energy consumed per day (58). We measured plasma vitamin levels and although we saw some differences between the analysed groups and seasons, they could not explain the lack of differences in the levels of urinary 8-oxodG observed in our study. However, we do not have any information on physical activity or total energy consumption that could potentially differ between the groups and thus affect urinary 8-oxodG levels. Finally, the lack of variability in 8-oxodG urinary excretion may be related to the effectiveness of DNA repair that is believed to be a major source of 8-oxodG in the urine (59). Although we are not aware of any study reporting decreased DNA repair gene expression after exposure to c-pahs or VOC, it has been shown that the expression of nucleotide excision repair genes was down-regulated in subjects exposed to asbestos in drinking water (60). Similarly, genes participating in various processes including DNA damage and repair were down-regulated in smokers (61). Thus, exposure to high levels of air pollutants in the winter season may have resulted in decreased activities of genes participating in the excretion of 8-oxodG to urine. Unlike urinary 8-oxodG, levels of 15-F2t-IsoP are not dependent on any cellular repair mechanism. Therefore, lipid peroxidation may be regarded as a biomarker of oxidative damage that is less prone to bias caused by individual differences between analysed subjects. We and others have repeatedly shown the usefulness of 15-F2t-IsoP as a biomarker of oxidative damage after exposure to cigarette smoke (62), ozone (63), c-pahs and PM (19,40,64 66). In the present study, plasma levels of 15-F2t-IsoP increased linearly with increasing concentrations of B[a]P in the ambient air up to a B[a]P concentration of about 9 ng/m 3. Higher doses of B[a]P did not increase plasma 15-F2t-IsoP levels any further. This suggests that after B[a]P reaches a certain concentration in ambient air, 15-F2t- IsoP is no longer a reliable biomarker of oxidative stress. Because of their catalytic functions, proteins are possibly the most immediate vehicles propagating oxidative stress in the cell (67). However, protein carbonyl groups are relatively difficult to induce and they probably reflect more severe cases of oxidative stress usually associated with protein dysfunction (67). This may be the reason for conflicting results in studies analysing protein oxidation after exposure to air pollutants (4,19,40,66,68). Interestingly, protein carbonyl levels were unexpectedly higher in a group of bus drivers sampled in the summer season when compared with the corresponding group sampled in winter, suggesting the role of an unidentified summer-specific pollutant (66). Although we obtained similar results in the present study, in a multivariate analysis of the data from all seasons split by region, we found a positive association between PM2.5 exposure and protein oxidation in both locations; B[a]P exposure had stronger effects in the Ostrava