Transition Metals in Personal Samples of PM_2.5 and Oxidative Stress in Human Volunteers

Epidemiologic studies have found exposure to high concentrations of ambient particulate matter (PM) to be associated with an increase in cancer, especially lung cancer (1, 2). Particles generated by combustion usually consist of a carbon core to which complex mixtures of compounds, such as transition metals, polyaromatic hydrocarbons, and endotoxins, adhere. One of the proposed mechanisms to the observed adverse health effects is that PM can induce oxidative stress mediated by transition metals on the particle surface, by a PM-induced inflammation causing phagocytes to release reactive oxygen species (ROS), and/or by quinones in the particles that produce ROS through redox cycling.

Different PM fractions have been found to generate ROS. Soluble transition metals are abundant in the water-soluble PM fraction and in vitro studies have shown that this fraction is able to induce ROS, specifically hydroxyl radicals, through the metal-dependent Fenton reaction both in cell-free systems and biological systems (3-5). In the Fenton reaction, hydroxyl radicals are generated through a transition metal–mediated reduction of hydrogen peroxide. Studies have shown that particle suspensions from which the soluble fraction has been removed also can generate ROS, suggesting that other fractions of PM than the transition metals can induce oxidative stress (5, 6).

A biomarker often used in assessing oxidative damage is 7-hydro-8-oxo-2′-deoxyguanosine (8-oxodG), a C8 hydroxylation of guanine, which is believed to be a major product of hydroxyl radical attack on DNA (7). Several experimental studies have found increased concentration of 8-oxodG following exposure to transition metals and PM both in vitro (4, 8, 9) and in vivo (10). PM-related induction of 8-oxodG has been shown to correlate significantly with induction of lung tumors in mice, suggesting that 8-oxodG could be a premutagenic lesion in PM-induced lung cancer (11). We have previously measured the personal exposure to PM_2.5 (mass of PM with an aerodynamic diameter <2.5 μm) in 49 persons four times during 1 year, and found that personal PM_2.5 exposure was associated with the 8-oxodG concentrations in lymphocyte DNA with an 11% increase in 8-oxodG per 10 μg/m³ increase in personal PM_2.5 exposure (12).

The aim of this study was to examine the relationship between the personal exposure to water-soluble transition metals in PM_2.5 and oxidative DNA damage. In 49 students from central Copenhagen, we collected personal PM_2.5 exposure twice in 1 year and measured the concentration of soluble transition metals in the PM_2.5. Collected lymphocytes and urine samples were analyzed for 8-oxodG.

The personal exposure to PM_2.5 was measured twice, once during summer and once during autumn, in 49 students. Measurements were conducted over 2-weekday periods for each subject as previously described (13). All participants were nonsmokers, living and studying in central Copenhagen. They were 20 to 33 years of age, with a median age of 24 years, and there was an even distribution of males and females. Not all of the 49 subjects could participate in both campaigns and new subjects were recruited so that in each campaign 49 subjects participated. In all, 66 subjects participated, of whom 32 subjects participated in both seasons and 34 subjects participated in only one campaign. All together we collected 98 measurements. Morning blood samples were collected at the end of each 2-day campaign and 24-hour urine samples were collected at the second day of the measuring campaign. The local ethics committee approved the study protocol and subjects gave written informed consent.

We measured lymphocyte and urinary concentrations of 8-oxodG by high-performance liquid chromatography with electrochemical detection as previously described (12, 14). Due to analytic problems, we only measured 8-oxodG levels in 52 of the 98 of the lymphocyte samples. The failure in measuring 8-oxodG in the remaining samples was mainly caused by high-performance liquid chromatography problems in terms of baseline drift that resulted in inaccurate 8-oxodG peaks and, therefore, we excluded these samples. We could not repeat the analysis of these samples because determination of 8-oxodG in lymphocytes has to be preformed on freshly isolated lymphocytes due to the fact that storage induces 8-oxodG damage to DNA (14).

We measured personal exposure to PM_2.5 with portable equipment from BGI (Waltham, MA) as previously described (13). Inductively coupled plasma mass spectrometry was used to determine the concentrations of vanadium, chromium, iron, nickel, copper, and platinum in the aqueous suspensions of PM, as previously described (15, 16). The freshly prepared suspensions of PM were filtered through a 0.2 μm Millipore filter (Minisart RC15 syringe filter, Sartorius AG, Göttingen, Germany). The filtrate was diluted with deionized water (1:5) and filtered again. We used the above method to achieve determination exclusively of the fraction of metals that would be readily available in the aqueous environment (i.e., in the form of soluble salts and/or as particles of size in the nanometer range). The transition metals were analyzed by sector field inductively coupled plasma mass spectrometry (ELEMENT, Finnigan MAT, Bremen, Germany) in the medium resolution mode (m / Δm ≅ 4,000) using the standard addition procedure for calibration. We diluted 50 μL of the filtrate with 500 μL of 0.08 N HNO₃ and 2,000 μL of ultrapure water and added to 50 μL of standard solutions containing 5 to 20 μg/L Ni, Cu, V, and Cr and 50 to 200 μg/L Fe, respectively. Platinum was measured in the low-resolution mode (m / Δm = 3,000) with a detection limit of 0.01 ng/L.

All statistical analyses were carried out using SAS (version 8e). Mixed model repeated measures analysis (Proc mixed) was used to describe concentrations of the biomarkers (8-oxodG in lymphocytes and urine) as a function of the transition metals. We adjusted for the effect of PM_2.5 exposure and season by including them as additional variables in the models, and we included subject as a random factor. The dependent variables were transformed by the natural logarithm to obtain variance homogeneity and normal distribution of the residuals. The models are therefore not linear in original scale and model estimates represent slopes in the logarithmic analysis. To calculate the predictive value of an X unit increase in one of the transition metals, the following formula was used: [e (model estimate·X) − 1] × 100.

The median concentrations of 8-oxodG in lymphocyte DNA and urine, personal exposure to the six measured transition metals in PM_2.5, and personal exposure to PM_2.5 (mass) are shown in Table 1. We found weak correlations between the PM_2.5 mass and the concentration of chromium (R_Spearman = 0.22, P < 0.05), copper (R_Spearman = 0.33, P < 0.005), and iron (R_Spearman = 0.29, P < 0.01), whereas the PM_2.5 mass did not correlate with the vanadium, nickel, or platinum concentrations (P > 0.5, Spearman). The concentrations of the metals, in particular of nickel, seemed to increase in the summer.

In Table 2, associations between 8-oxodG concentration in lymphocytes and personal exposure to transition metals in PM_2.5 are shown. For all six transition metals we adjusted for the effect of mass PM_2.5 measurement and season by including them as variables (and thus as possible confounders of the main effect). We found that the vanadium and chromium concentrations were significantly associated with the 8-oxodG concentration, causing a 1.9% increase in 8-oxodG per 1 μg/L increase in the vanadium concentration and a 2.2% increase in 8-oxodG per 1 μg/L increase in the chromium concentration. Platinum, nickel, copper, and iron were not significantly associated with the 8-oxodG concentration. The mass concentration of PM_2.5 was independently and significantly associated with 8-oxodG in five of the six transition metal models (P < 0.02 in the models with vanadium, chromium, nickel, copper, and iron) and only in the case of platinum, the PM_2.5 concentration was not significantly associated with the 8-oxodG concentration when platinum was in the model (P = 0.07).

None of the six measured transition metals was significantly associated with the 8-oxodG concentration in urine in the whole data set or in the subset for whom 8-oxodG levels in lymphocyte DNA were available (P > 0.1 for all six transition metals; results not shown). There was no correlation between the 8-oxodG concentrations in urine and in lymphocytes (R_Spearman = −0.14, P = 0.31).

We have previously found that the personal exposure to PM_2.5 is associated with an increase in 8-oxodG concentration in lymphocytes (12). It has been suggested that 8-oxodG is a premutagenic lesion in PM-induced lung cancer because induction of 8-oxodG following PM exposure has been significantly correlated with induction of lung tumors in mice (11). We here report that the concentrations of water-soluble vanadium and chromium of PM_2.5 were associated with the 8-oxodG concentration independently of the PM_2.5 mass concentration. These data suggest that these soluble metals as well as other mechanisms are involved in PM-induced oxidative stress in vivo. This is supported by in vitro studies that have shown that the water-soluble metal fraction of PM and its insoluble fraction can produce ROS independent of each other (5, 6). Fenton-mediated generation of hydroxyl-radicals has been shown to play a particular important role in the induction of oxidative DNA damage by the water-soluble fraction of PM as well as by transition metals (5, 8, 9). Importantly, however, those transition metals which are particularly abundant in PM, such as iron and copper, were not associated with 8-oxodG in the present study. A possible explanation could be the fact that the concentrations of iron and copper are normally tightly controlled in the intact organism. Besides, through the soluble metals, PM can generate oxidative stress directly from the surface of particles (e.g., through redox cycling of quinones, through activation of ROS generation in target cells, and through formation of ROS and reactive nitrogen species by inflammatory cells; ref. 17).

Vanadium and chromium have both been shown to generate 8-oxodG in cell-free systems (8, 18, 19). Residual oil fly ash consists of particles with high concentrations of vanadium. Residual oil fly ash has been consistently shown to induce oxidative stress and inflammation both in vitro and in vivo as reviewed by Ghio et al. (20). Other studies have indicated that vanadium in PM has adverse effects in humans. A study on boilermakers found that personal exposure to vanadium in PM_2.5 increased the heart rate (21). Although vanadium is usually only a minor constituent of urban PM samples, a trend toward increased daily mortality has been linked to vanadium in some cities (22).

Anthropogenic emission of vanadium and chromium is thought to originate mainly from combustion of heavy oil, and for chromium also from wear of brake linings. In a study by Espinosa et al. (23) of metal distribution according to particle size in urban aerosols (total suspended particulate), they found 70% of the measured vanadium in the particle fraction with a diameter below 0.61 μm. In a similar Danish study of particles collected in a busy street in central Copenhagen, 64% of the vanadium in PM2.5 was found in the particle fraction with a diameter below 0.42 Am and 15% in the ultrafine fraction (24). Due to the high proportion of vanadium in particles with a relative small diameter, it seems likely that a considerable fraction of the vanadium reaches the alveoli where it comes in direct contact with the lung epithelium. It is possible, however, that vanadium is a marker of the smaller particles in PM_2.5 and thus not the causative agent. For chromium, the Danish study found that almost 90% of the chromium in PM_2.5 was found in the particle fraction with a diameter above 0.42 μm (24).

In a study by Espinosa et al. (25), they examined the concentration of metals in different compositional fractions of the particles below 0.61 μm, such as the soluble and the organic fractions. They found 55% of the vanadium in the soluble fraction, which was a higher percentage than the other measured metals such as iron and nickel (25). We only measured vanadium in the soluble fraction, which could explain why we find an association between oxidative DNA damage and this metal and not, e.g., iron, as only 4% of the iron was found in the soluble fraction (25). Unfortunately, the study by Espinosa et al. did not look at chromium in the different compositional fractions of the particles and we are not aware of other studies that have addressed this issue for chromium. We found a significant effect of season on the soluble nickel concentration, with highest concentrations during the summer. Daily measurements of urban background nickel concentrations in PM₁₀ in Copenhagen have shown higher levels during summer than during autumn and winter. This, together with more relevance of nickel-rich outdoor sources during summer, can possibly explain the difference in personal exposure to nickel between the two seasons.

The lack of correlation between the 8-oxodG concentrations in lymphocytes and urine is in agreement with other studies (26). Correlations may not be expected as 8-oxodG concentration in nuclear cell DNA reflects the balance between formation and repair, whereas the urinary excretion supposedly reflects the summed rates of formation and turnover of 8-oxodG in nuclear and mitochondrial DNA as well as the nucleotide pool in the whole body (14).

In conclusion, our study suggests that vanadium and chromium in ambient air PM_2.5 could play a major role in the induction of oxidative DNA damage and this could be relevant for the health effects associated with PM.

We thank Dr. Tingming Shi (Hubei Provincial Centre for Disease Control & Prevention, Wuhan, Hubei, P.R. China) and Dr. Jutta Begerow (Hygiene-Institut des Ruhrgebiets, Institute of Environmental Hygiene and Environmental Medicine, Gelsenkirchen, Germany) for sample processing and metal analysis, respectively.