An increasing body of evidence has linked exposure to indoor pollutants with increased rates of respiratory morbidity and mortality. Women and children exposed to high indoor PM concentrations have significantly increased rates of acute respiratory infections (ARI) [5,8-10], and women cooking over biomass fires for extensive periods have an enhanced risk of chronic obstructive pulmonary disease (COPD) and lung cancer [11,12]. In a recent study we found that lower airway cells from women and children exposed to biomass smoke contained significantly more carbonaceous material than age-matched subjects exposed to fossil fuel derived PM₁₀ [13]. Whilst enhanced particle deposition in the airways is clearly important, the chemical and physical characteristics of airborne particles that contribute to their toxicity have not been firmly established, though particle size and surface area [14], as well as acidity and composition [15,16] have all been mooted as important determinants in this regard.

The capacity of inhaled PM to elicit damaging oxidation reactions in the lung and systemic circulation may account for many of the toxic responses observed in exposed individuals [17]. Certain transition metal components of PM are capable of catalysing oxidation reactions [18], as are quinones [19] resulting in the production of reactive oxygen species. An excessive production of these species can activate redox sensitive transcription factors regulating the expression of pro-inflammatory cytokines, leading to inflammation and tissue injury [20]. Measuring the capacity of PM to oxidise physiologically relevant molecules can therefore be used to provide a biologically relevant index of activity, integrating PM size, surface area and composition.

GSH was also depleted in a dose dependent manner by the dung cake PM extracts, 49.6 ± 4.3 and 63.5 ± 22.4% at 50 and 100 μg/m³ respectively. These losses were significantly greater than those observed with ROFA, which actually demonstrated a reduced loss of GSH compared with the particle free control, or the traffic derived samples, which depleted GSH 10.2% (diesel) and 15.7% (gasoline) (Figure ​(Figure2).2). Notably, no loss in total glutathione was noted during the incubations suggesting that adsorption to the particle surface was not occurring to any great extent. Similar to the situation with AA, co-incubation with DTPA completely abolished the loss of GSH. Incubation with DTPA also prevented the background losses of both ascorbate (-20.8 ± 6.2 μM – Figure ​Figure1)1) and glutathione (-47.4 ± 6.1 μM – Figure ​Figure2)2) from the particle-free controls over the 4 h incubation. These losses were attributable to the presence of contaminating metal ions in the synthetic RTLF following Chelex-resin treatment. Urate was not depleted from the RTLF model by any of the particle types examined in line with our previous observations [21] (data not shown).

Further characterisation of the metal dependence of the oxidative reactions observed was performed using an AA only RTLF model. Individual AA consumption rates at 50 μg/mL for each of the three dung cake extracts were similar: 13.7 ± 0.15 (DC1), 13.5 ± 0.49 (DC2) and 13.3 ± 0.19 nM/s (DC3). Co-incubation with the metal chelator DTPA (100 μM) significantly reduced the rate of AA oxidation: 6.5 ± 0.36 versus 13.5 ± 0.27 nM/s for all three samples (Figure ​(Figure3).3). Notably, in this simplified model in the absence of GSH and UA, the protection afforded by DTPA was only approximately half of that seen using the synthetic RTLF. Increasing the DTPA concentration to 200 μM completely blocked AA depletion in this model (Figure ​(Figure4).4). Limited, though statistically significant protection was also seen using the antioxidant enzymes superoxide dismutase and catalase (11.9%), whilst no decrease in AA oxidation was seen with catalase alone or heat inactivated superoxide dismutase and catalase (Figure ​(Figure33).

When the dung cake aqueous and organic extracts were separated the vast majority (80.6%) of the oxidative activity was found to be associated with water-soluble components (Figure ​(Figure5).5). The residual activity associated with the hexane extract could be completely inhibited through the co-incubation of 100 μM DTPA indicating that this activity was not due to organic radicals. Consistent with these findings, dung cake aqueous extracts were shown to contain appreciable concentrations of the redox active metals Fe and Cu quantified using the chromogenic chelators bathophenantroline-disulphonate (BPS) and bathocuproine-disulphonate acid (BCS) in each of the three dung cake particle suspensions. These methods gave Fe concentrations of 7.1, 3.6, and 8.5 μg/mg in DC 1, 2 and 3 respectively and 26.2, 26.2 and 23.8 μg/mg for Cu. In comparison ROFA contained 23.7 ± 0.30 μg/mg Fe and 5.9 ± 0.73 μg/mg of Cu, whilst both metals were undetectable in the diesel and gasoline samples, using either chromogenic chelator.

It has been proposed that the capacity of inhaled particles to elicit inflammation and injury in the lung, as well as systemically, may be related to their capacity to cause oxidative stress [17]. In this working paradigm, inhaled particles generate oxidative stress through three inter-related pathways: first, by directly introducing oxidising species into the lung, such as redox active transition metals [16] or quinones [19] absorbed onto their surface. Second, by introducing surface absorbed PAHs that can undergo bio-transformation in vivo into quinones species through the action of the cytochrome P450, epoxide hydrolase and dihydrodial dehydrogenase detoxification pathway [26] and finally by stimulating inflammatory cells to undergo the oxidative burst. In this final case, activation of inflammatory cells may be triggered by endotoxin on the surface of inhaled particles [27], futile phagocytic processing of PM [28], or by the up-regulation of redox sensitive transcription factors directing the synthesis of pro-inflammatory cytokines [29]. The integrated sum of all these processes can be considered the 'total' oxidative activity of the particle.

In this study we measured PM oxidative activity using an in vitro screening procedure that assessed the capacity of PM associated pro-oxidant components (metals and quinones) to deplete physiologically relevant antioxidants, ascorbate, urate and reduced glutathione from a synthetic model of the RTLF [21,22]. This 'intrinsic' activity, measured in a cell free system, only reflects the oxidative activity attributable to redox active metals and quinone compounds and not 'latent' activities that may be associated with PAHs or endotoxin. With this caveat, we found PM samples derived from the combustion of dung cake to be significantly more active, on an equal mass basis, than either metal-rich ROFA or PAH-rich vehicle exhaust PM, despite the greater surface area of these samples. This activity was manifest by the capacity of the PM suspensions to deplete both AA and GSH from synthetic RTLF. Notably, the endotoxin and PAH content of dung cake and other biofuels have been shown to be high [24,30] suggesting that their 'total' oxidative activity is likely to be far in excess of that associated with traffic derived PM. This very high oxidative activity in animal dung combustion particles supports studies demonstrating increased pulmonary toxicity in mice following instillation of particles derived from dried municipal sewage combustion, relative to coal alone [31].

The depletion of both AA and GSH from the model was prevented by co-incubation with the metal chelator diethylenetriaminepentaacetate (DTPA) indicating that the losses observed were driven by redox active metals such as Fe, Cu, Ni, and Cr. DTPA has five acetate groups linked to a molecular backbone that permits it to form tight complexes with a broad range of metals, preventing them from catalysing damaging oxidation reactions. Desferoxamine (DFO) was not used in these studies as it has been reported to reduce chelated Cu that would have resulted in interpretive difficulties when examining its protective role in mixtures of soluble metals [32]. The contribution of superoxide and hydrogen peroxide to the observed antioxidant losses was examined in co-incubation experiments using the antioxidant enzymes Cu, Zn superoxide dismutase (SOD) and catalase (CAT). Limited protection was observed with these enzymatic antioxidants suggesting that the contribution of these reactive oxygen species to the ascorbate and glutathione losses was minor compared with those attributable to their direct oxidation during the reduction of Fe³⁺ and Cu²⁺.

Thus we conclude that AA and GSH oxidation occurred predominately by their direct reduction of Fe³⁺ to Fe²⁺ and Cu²⁺ to Cu⁺. The superoxide formed by the subsequent oxidation of ferric and cupric ions could undergo dismutation to hydrogen peroxide, reduce Fe³⁺ and Cu²⁺, or oxidise ascorbate, urate or glutathione within the synthetic RTLF. As the reaction rate between Fe³⁺ and superoxide (1.5 × 10⁸ M^-1s^-1) greatly exceeds that its dismutation at physiological pH (5.4 × 10⁵ M^-1s^-1, pH7.4) it seems likely that the former reaction predominated, especially as little loss of ascorbate or glutathione could be attributed to superoxide or hydrogen peroxide production. Interestingly, we saw no evidence of urate depletion, despite the importance of this antioxidant in protecting the airway against oxidant gases [33], peroxynitrite [34] and hydroxyl radicals [35]. The rate of the reaction of urate with hydroxyl radicals (7.2 × 10⁹ M^-1s^-1, pH 6–7) is broadly similar to that of both ascorbate (1.6 × 10⁹ – 1.1 × 10¹⁰ M^-1s^-1, pH 7–7.4) and glutathione (9.0 × 10⁹ – 1.3 × 10¹⁰ M^-1s^-1, pH 8 and 7.8, respectively). Thus the absence of UA depletion in this model supported the contention that superoxide dismutation to hydrogen peroxide was not occurring to any great extent, with little evidence of hydroxyl radical generation. Whilst the redox potentials of UA and AA at pH7 (E°' = 590 and 282 mV respectively) [36] may suggest that the urate radical could be reduced back to UA at the expense of AA we do not believe that this occurred, as removal of urate from the RTLF had no impact on the observed rate of ascorbate depletion (data not shown).

We also observed that the capacity of DTPA to inhibit ascorbate oxidation was significantly reduced in the ascorbate only incubation experiment: only 100 μM being required for full inhibition in the complete synthetic RTLF as opposed to 200 μM in the ascorbate only RTLF model. This finding may imply that either GSH or UA is limiting the bioavailability of Fe, either through chelation, as has been proposed for UA [37], or by interfering with the capacity of AA to solubilise ferric iron from the particle surface [16]. We are currently investigating these potential actions of UA and GSH. Irrespective of this, when the concentration of DTPA was increased in the AA only model, all AA oxidation was blocked indicating the absence of a quinone-dependent activity in the DC particles. This contention was supported by the observation that only a fraction of the measured oxidative activity was present in organic DC extracts. Whilst other groups have emphasised the importance of PM associated quinones/hydroquinones in the oxidative activity of ambient PM [19] our data would tend to emphasise metal content as the major determinant of DC oxidative activity. Clearly the contribution of metal and organic components to PM oxidative activity may vary depending on its source. In addition it is also likely that the age and storage conditions of the filters used in this study resulted in losses of potentially reactive organic species. These cautionary caveats only further emphasise that we are probably underestimating the 'true' oxidative capacity of freshly generated DC particulates.

As these findings implicated redox active metals in the oxidation process we measured the bioavailable Fe and Cu content of the DC particles. This measurement included reduced and oxidised forms of these metals, both water-soluble and surface mobilisable through ligation to the chromogenic chelators bathophenantroline-disulphonate (BPS) [38] and bathocuproine-disulphonate acid (BCS) [39]. Using these approaches we detected considerable Fe and Cu content. The especially high content of Cu is likely to explain extensive glutathione oxidation observed with the dung cake samples, due to copper's high reactivity toward this antioxidant [18], as well as the lack of reactivity of the ROFA sample toward GSH. It should be noted, however, that the variation in the content of these metals in the three DC samples did not match their observed variation in oxidative activity implying that Fe and Cu were not the sole determinates of the observed activity. The compositional data pertaining to the ROFA sample (PM_2.5), used in the current study have been described previously [23,40] using ICP-MS. These analyses have confirmed the relatively high concentrations of total Fe, and low concentrations of Cu in the ROFA sample derived from the burning of heavy fuel oil (N° 5). These metals are however less abundant in the ROFA sample than either vanadium (58.6 μg/g) or nickel (10.6 μg/g). In contrast, ICP-MS analysis of both the gasoline and diesel samples revealed these metals to be below detectable limits [41], which concurs with their low reactivity in this assay system and these measurements made using the chromogenic chelators. Parallel ICP-MS analysis of dung cake PM obtained under identical burn conditions has also subsequently revealed appreciable concentrations of the redox active metals Ni and Cr (45 and 40 μg/g PM, respectively) in these samples, but no detectable V. This elemental analysis will be described in detail in a subsequent manuscript.

AA Ascorbate

UA Urate

GSH Reduced glutathione

GSSG Glutathione disulphide

RTLF Respiratory tract lining fluid

DC Dung cake

DFO Desferoxamine

BPS Bathophenantroline-disulphonate

BCS Bathocuproine-disulphonate

DPTA Diethylene triamine pentaacetate

SOD Superoxide dismutase

hiSOD Heat inactivated superoxide dismutase

CAT Catalase

PAH Polyaromatic hydrocarbons

PM Particulate matter

Sean T Duggan: None

Chandra Venkataraman: None

Frank J Kelly: None

Sean T Duggan: Study design and experimental analysis

Chandra Venkataraman: Collection of particle samples from dung cake combustion

Gazala Habib: Collection of particle samples from dung cake combustion

Frank J Kelly: Study conception and design, manuscript review