The Concentration of Iron in Real-World Geogenic PM10 Is Associated with Increased Inflammation and Deficits in Lung Function in Mice

Background There are many communities around the world that are exposed to high levels of particulate matter <10 µm (PM10) of geogenic (earth derived) origin. Mineral dusts in the occupational setting are associated with poor lung health, however very little is known about the impact of heterogeneous community derived particles. We have preliminary evidence to suggest that the concentration of iron (Fe) may be associated with the lung inflammatory response to geogenic PM10. We aimed to determine which physico-chemical characteristics of community sampled geogenic PM10 are associated with adverse lung responses. Methods We collected geogenic PM10 from four towns in the arid regions of Western Australia. Adult female BALB/c mice were exposed to 100 µg of particles and assessed for inflammatory and lung function responses 6 hours, 24 hours and 7 days post-exposure. We assessed the physico-chemical characteristics of the particles and correlated these with lung outcomes in the mice using principal components analysis and multivariate linear regression. Results Geogenic particles induced an acute inflammatory response that peaked 6 hours post-exposure and a deficit in lung mechanics 7 days post-exposure. This deficit in lung mechanics was positively associated with the concentration of Fe and particle size variability and inversely associated with the concentration of Si. Conclusions The lung response to geogenic PM10 is complex and highly dependent on the physico-chemical characteristics of the particles. In particular, the concentration of Fe in the particles may be a key indicator of the potential population health consequences for inhaling geogenic PM10.


Introduction
The relative contribution of geogenic (earth derived) dusts to the total suspended particulates (TSPs) varies geographically. In some areas, such as urban centres, these dusts typically represent only a minor proportion of TSPs [1,2] whereas in other regions, notably those prone to aridity and high levels of wind erosion, they may predominate [1,3]. Importantly, the physico-chemical characteristics of suspended geogenic dusts in a particular area will vary depending on local geology and anthropogenic activities [4]. As such, the respiratory health implications for inhaling geogenic dusts are also likely to vary spatially.
It is well-established that mineral dusts in the occupational setting have pro-inflammatory properties [5,6,7,8]. However, much less is known about the effect of inhaled geogenic particles on lung health when exposure occurs at a community level. Such dusts are likely to be physically and chemically heterogeneous, in contrast to dusts in the occupational setting which are usually consistent in size and chemical composition. We have previously shown that community sampled geogenic particles less than 10 mm in aero-dynamic diameter (PM 10 ) have the capacity to induce a dose-dependent inflammatory response in the lung [9]. Importantly, the response varied between particles collected from the two sites that we studied. The most striking physico-chemical difference between the particle sources was the concentration of iron (Fe) with a higher concentration of Fe being associated with a greater influx of neutrophils and a higher production of proinflammatory cytokines.
While much of the research in this field has focused on the lung response to inhaled silica (Si) particles due to their importance in occupational lung disease, there is recognition that other metals can have adverse impacts on the lung when inhaled; including Fe. Iron homeostasis in the lung is tightly regulated due to its importance in cell function and host defence [10]. The application of exogenous iron to the system can result in the production of oxidative stress [11] and synergistically enhance the response to other particles [12]. Exposure to dust from iron oxide stockpiles has also been associated with increased hospitalisations for respiratory disease [13] and in particular increased respiratory infections in children [14].
While our previous study [9] anecdotally suggests that Fe may influence the magnitude of the response in heterogeneous particles the fact that we only used two particle sources and did not have measures of lung function limited our capacity to probe this observation. In this study we aimed to use real-world community sampled geogenic PM 10 to assess the impact of the association between the concentration of Fe, after controlling for other particle characteristics, and the lung response using an established mouse model.

Animals
Eight-week old females BALB/c mice were purchased from the Animal Resource Centre (ARC, Murdoch, Western Australia) and housed at the Telethon Institute for Child Health Research under a 12:12 hour light:dark cycle. Food and water were provided ad libitum. All experiments were approved by the Telethon Institute Animal Ethics Committee (AEC #249) and conformed to the guidelines of the National Health and Medical Research Council of Australia.

Geogenic Particle Extraction (PM 10 )
Surface soil samples were collected from four towns throughout the arid regions of Western Australia (Table 1). No specific permissions were required for access to these sites or for these collection activities. At each town, two samples from areas exposed to wind erosion and lacking vegetation were collected and processed using an established technique that allows extraction of PM 10 at sufficient quantities for in vivo [15]. We have previously demonstrated that the chemical composition of PM 10 obtained using this technique correlates strongly with airborne particles [9]. In this particular study we were not seeking to assign specific health risks to particular towns. Rather we sought to obtain PM 10 samples with varying physico-chemical characteristics in order to establish how variation in these characteristics impacts on the lung response.
Two of the towns, Newman and Tom Price, are closely aligned with open cut mining for iron ore and have itinerant populations. Kalgoorlie has a stable population and is situated near open cut gold mining activities while Karratha is a coastal iron ore port. All of these towns are exposed to high levels of particulate matter that are largely geogenic in origin.

Particle Characterisation
Endotoxin levels were assessed using a limulus amebocyte lysate (LAL) assay (GenScript; New Jersey, U.S.A.). For this purpose samples were diluted 1:1000, adjusted for pH (6)(7)(8) and incubated in LAL at 37uC for 10 minutes. A chromogenic substrate was added and the absorbance was read at 405 nm and compared to an E.coli standard.
The chemical composition for a panel of 12 common metals (aluminium, Al; iron, Fe; cadmium, Cd; cobalt, Co; chromium, Cr; copper, Cu; arsenic, As; manganese, Mn; nickel, Ni; lead, Pb; uranium, U; zinc, Zn) was assessed by ICP-MS (Chem Centre, Western Australia). Samples were prepared according to the USEPA 3051A Method using acid digestion. The quantity of silica (SiO 2 ) in the samples was assessed by ICP-OES (University of Western Australia) which was used to estimate the amount of Si in the particles.
Particle size was assessed using an Andersen Cascade Impactor (Copley Scientific, UK) using a previously described technique [9]. The mass median aerodynamic diameter (MMAD) and geometric standard deviation (GSD) of the particle sizes were calculated using an online tool (http://mmadcalculator.com/andersenimpactor-mmad.html).

Exposure Protocol
Mice (n = 13 per group) were intranasally exposed to 100 mg of geogenic particles suspended in 50 mL of saline (+0.5% Tween-80) or vehicle (saline +0.5% Tween-80) alone under light methoxyflurane anaesthesia. Mice were studied for the outcomes (described below) at 6 hours, 24 hours or 7 days post-exposure. The limited quantity of PM 10 available prevented exposure by dry aerosol. We chose to use intranasal exposure over intratracheal instillation to avoid a biased deposition of the particles in the large airways [16] and to ensure that the particles were able to reach the alveoli we used an optimised protocol with a delivery volume of 50 mL and mild sedation [17]. The 100 mg dose was chosen based on the dose-response data from our previous study [9] as a dose that produces a robust and measurable inflammatory response. This dose (,5 mg.kg 21 ) is an order of magnitude smaller than typical rodent particle exposure models (,100 mg.kg 21 [18]) and is consistent with our prior work on diesel exhaust particles [19,20].

Lung Mechanics and Lung Volume
Mice were prepared for in vivo assessment of lung function as described previously [21]. Following anaesthesia, surgical tracheostomy and standardisation of lung volume history [21] baseline lung volume at 0 cmH 2 O transrespiratory pressure (P rs ) was determined [22]. Briefly, thoracic gas volume (TGV) at 0 cmH 2 O P rs (V 0 ) was calculated by the relationship between tracheal pressure and plethysmograph (box) pressure following electrical stimulation of the inspiratory muscles [22].
Lung mechanics were then measured using a modification of the forced oscillation technique. During pauses in ventilation, an oscillatory signal containing 9 frequencies (4-38 Hz) was generated by a loudspeaker and passed through a wavetube to the mouse via the tracheal cannula. The respiratory system impedance (Z rs ) was measured as the load impedance on the wavetube. A four parameter model with constant phase tissue impedance was fitted to the Z rs data [23] to obtain measures of R aw , the Newtonian resistance which is equivalent to airway resistance in the mouse due to the compliance of the chest wall, G (tissue damping) which is thought to represent the resistance of the small airways where air movement occurs primarily by diffusion and H (tissue elastance) the stiffness of the lung parenchyma. Hysteresivity (g = G/H), was calculated as a marker of lung heterogeneity [24]. Following the measurement of baseline mechanics a further TGV measurement was made prior to a slow inflation deflation manoeuvre from 0 to 20 cmH 2 O P rs . By integrating flow through the wavetube we estimated the maximum lung volume achieved at 20 cmH 2 O P rs (V 20 ) as an index of lung capacity [25].

Inflammatory Markers
Following the measurement of lung function, mice were euthanased by anaesthetic overdose. A bronchoalveolar lavage (BAL) was obtained as described previously [9] in order to assess inflammatory cells and cytokines. The concentrations of IL-6, MIP-2 (mouse IL-8 analogue) and IL-1b in the BAL were determined by ELISA according to the manufacturer's instructions (MIP-2, R&D Systems, Minnesota, U.S.A; IL-6, IL-1b, MCP-1, BD Biosciences, California, U.S.A.) from the lavage supernatant.

Statistical Analyses
In order to deal with the volume of data generated in this study we took a hierarchical approach to the analysis. Firstly, we compared the response for each outcome (within each time point) to the saline group (ANOVA following by Holm-Sidak post-hoc tests with the saline group as the control comparison). We used this analysis to identify which outcomes were altered by exposure to geogenic PM 10 .
In this study we were interested in exploring the key characteristics of the geogenic particles that were associated with these specific lung outcomes. In order to explore these associations a two stage analysis was conducted. Firstly, we conducted a principal components factor analysis of the lung outcomes that were identified in the first ANOVA as being significantly altered by exposure to geogenic PM 10 (macrophages, neutrophils, IL-6, MIP-2, IL-1b, G, H, g). In order to satisfy the assumption of normal distribution of the error terms the data were transformed where appropriate using Box-Cox or logarithmic transformations. We assessed the screeplot to determine the number of factors to include and used orthogonal rotation. The results of this factor analysis were then used to create ''scores'' for each mouse. These scores were used in multivariate linear regression analyses at each time using the concentrations of Fe, Si, Al (i.e. metals with . 0.015% representation in each sample), endotoxin, MMAD and GSD as variables.

Particle Characteristics
The physico-chemical characteristics of the particles varied considerably between sites ( Table 2). The particles were dominated by Si, Al and Fe. Endotoxin levels varied considerably between samples but were generally low with the highest concentration (,158 EU/50 mL) equivalent to approximately 16 ng in the exposure solution.

Responses Compared to Control (Saline) Mice
Data for control mice are presented in Table 3. The lung response to exposure to 100 mg of geogenic PM 10 was highly variable (Figure 1). Exposure to geogenic PM 10 caused a significant influx of neutrophils that peaked 24 hours post exposure and had returned to control levels at the 7 day timepoint ( Figure 1). There was also a macrophage response, although the  pattern was less consistent. There was a concomitant increase in inflammatory cytokines which, in the case of MIP-2 and IL-6, peaked 6 hours post-exposure, was still evident 24 hours postexposure and had returned to control levels by the 7 day timepoint. Interestingly, in some cases, the levels of inflammatory cytokines were below control levels at the later timepoints. There was a similar pattern in IL-1b, although this response was more variable (Figure 1). We did not detect significant levels of MCP-1 in the BAL. Responses in lung mechanics were highly variable between groups. Exposure to particles from Tom Price ''1'' caused a decrease (''improvement'') in tissue damping at all timepoints, while tissue elastance increased (''deficit'') 7 days post exposure in the same mice. Exposure to particles from Newman ''2'' and Tom Price ''2'' caused an increase in tissue elastance at the later timepoints, whereas exposure to particles from Kalgoorlie appeared to improve tissue elastance ( Figure 1). Hysteresivity varied between exposure groups (Figure 1). There was no evidence for an effect of exposure to these particles on R aw , V 0 or V 20 (data not shown).

Factor Analysis
Based on the results of the previous analyses the following variables were included in the principal components analysis: neutrophils, macrophages, MIP-2, IL-6, IL-1b, tissue damping, tissue elastance and hysteresivity. Examination of the screeplot revealed a distinct break point after two factors (,58.9% of the variance explained). Inclusion of additional factors only resulted in incremental improvements in the explained variance. The first factor that we identified was strongly (|.0.75|) loaded on (+) MIP-2 and (+) IL-6 and moderately (|.0.50|) loaded on (+) neutrophils and (+) IL-1b (Table 4). As such, this factor described the inflammatory response to the geogenic PM 10 . The second factor we identified was strongly (|.0.75|) loaded on (+) tissue elastance and moderately (|.0.50|) loaded on (+) tissue damping and (-) hysteresivity. As such, this factor described the lung function response to the geogenic PM 10 .
Using these factors we generated scores for each mouse. Pooling these scores across sites revealed the following patterns; 1) the general inflammatory response peaked 6 hours post-exposure but then (on average) fell below control levels 7 days post-exposure, and 2) the lung function response showed evidence of a paradoxical ''improvement'' in lung mechanics 24 hours postexposure followed by a deficit 7 days post-exposure ( Figure 2).

Physico-chemical Characteristics and the Lung Response
We used the four primary responses we identified in the factor analysis (1: acute inflammatory response 6 hours post-exposure, 2: decrease in inflammatory markers 7 days post-exposure below control levels, 3: ''improvement'' in lung mechanics 24 hours postexposure, 4: deficit in lung mechanics 7 days post-exposure) in order to identify which physico-chemical characteristics of the geogenic PM 10 were associated with each lung response. These relationships were assessed using multivariate linear regression with the Fe, Si, Al (the most abundant metals), MMAD, GSD and endotoxin levels as the independent variables (Table 5).
We found that the concentration of Fe was the only characteristic associated (p = 0.007) with the early inflammatory response. This was an inverse relationship such that as the concentration of iron increased, the magnitude of this response decreased. In contrast, the decrease in inflammatory markers below control levels we observed 7 days post-exposure was associated with a number of characteristics including (-) Si  Table 3. Results for control (saline exposed) mice for the outcome measures of interest 6 hours, 24 hours and 7 days postexposure. straightforward interpretation such that as the concentration of Fe and the variability in size of the particles (GSD) increased, the magnitude of the deficit in lung mechanics also increased. In contrast, this deficit was inversely associated with the concentration of Si.

Discussion
The magnitude and pattern of the lung response to instilled geogenic PM 10 in mice varied considerably depending on the source of the particles. The response was characterised by 1) an acute inflammatory response including the production of proinflammatory cytokines, such as MIP-2 and IL-6 and an influx of neutrophils, 2) a reduction in inflammatory markers 7 days postexposure below control levels and, 3) an apparent improvement in lung mechanics 24 post-exposure, which was followed by a deficit in lung mechanics 7 days post-exposure. Increasing concentrations of Fe in the samples were inversely associated with the acute inflammatory response, but were positively associated with a reduction in the deficit in inflammatory markers and an increase in the lung impairment observed 7 days-post exposure. A similar pattern was observed for the variability in particle size (GSD). Conversely, higher concentrations of Si were associated with a greater reduction in the inflammatory response below control levels and a smaller impairment in lung mechanics 7 days postexposure.
We have previously shown that community-sampled geogenic PM 10 induces a potent inflammatory response in the lung [9]. This neutrophil dominated response is consistent with many previous studies examining the lung response to inhaled dusts (primarily SiO 2 ) of mineral origin [18,26]. While exposure to crystalline silica in mouse models elicits fibrotic changes [27,28], which presumably alters lung function, there appears to be very little data in the literature on the impact of silica on lung function in mouse models (although silicosis in humans is associated with impaired lung function [29]). It should be noted that the exposure doses used in our study (,5 mg.kg 21 ) are an order of magnitude lower than the doses typically used in mouse studies of pure crystalline silica exposure (e.g. ,100 mg.kg 21 , [18]) and silica only constituted a proportion of our particles. These observations suggest that the lung was particularly sensitive to the heterogeneous geogenic particulates used in our study.
The potent neutrophilia we observed has been linked to structural degradation of lung tissue in a variety of contexts [30,31,32] including exposure to mineral particles [33]. This neutrophil response was preceded by the production of proinflammatory cytokines. In addition to the potential for direct damage to the lung tissue resulting from the production of these cytokines, this response may also have important implications for an individual's response to a secondary insult such as a bacterial [34] or viral lung infection [20].
As discussed earlier, there is a substantial body of work on the influence of silica based particles on the lung due to their importance in occupational lung disease. Given the dominance of Si in our samples (in the form of SiO 2 ), it is likely that this mineral was a key driver of the primary response to the particles. Interestingly, an increase in the concentration of Si in our samples was associated with an increase in the magnitude of the deficit in inflammatory markers 7 days post-exposure (below control levels) yet a decrease in the magnitude of the deficit in lung mechanics we observed at the same timepoint. Given the large body of evidence  showing a positive association between silica and poor lung outcomes this was a surprising result. Absolute levels of SiO 2 in ambient particulates may therefore be a poor marker of the risk posed by geogenic particles to lung health. The only metal that was associated with all of the responses we observed was iron. The direction of the association between the responses we observed 7 days post-exposure and Fe was the opposite of that observed for Si: that is, elevated concentrations of Fe were linked to increases in the magnitude of the deficit in lung mechanics. Paradoxically, an increase in the concentration of Fe was associated with a smaller inflammatory response 6 hours postexposure. This observation, combined with the link between the 7 day response we also found in the Si analysis, implies that the initial inflammatory response may not be the primary determinant of the long term deficit in lung mechanics. We confirmed this association by examining the correlation between the two Table 5. Results of the multi-variate linear regression analysis examining the association between the physico-chemical characteristics of the geogenic PM 10 and the primary lung responses identified in the factor analysis.  ''factors'' we identified at the 7 day timepoint (Figure 3). This appears to indicate that the degree of ''overshoot'' of the inflammatory response 7 days post-exposure is inversely associated with the deficit in lung mechanics at that time. We postulate that this may reflect the lung attempting to ''protect'' itself from the adverse effect of over production of pro-inflammatory cytokines. In situations where this inflammatory ''overshoot'' is not triggered, for example when the concentration of Fe in the particles is high, then there were long lasting impacts on lung function as a consequence. This observation implies that it is the sustained production of cytokines, rather than the magnitude of the initial response, that is the key driver of the lung function impairments associated with exposure to geogenic PM 10 . Iron is well known for its role in modulating responses to additional lung insults [35] and local Fe levels are tightly regulated to maintain normal physiological function [10]. However, the direct impact of iron particles on lung inflammation is equivocal, with some studies showing no effect [12,36] and others showing some potentially detrimental responses [37]. It is possible that the negative response induced by Fe requires an additional insult such as co-exposure to silica [38], which was the situation in our study. Interestingly, this study did not support our anecdotal suggestion, based on our earlier study [9], that higher Fe concentrations may be associated with a greater acute inflammatory response to geogenic PM 10 . This discrepancy can be reconciled by the fact that in the earlier study we only used two samples whereas in this study we used eight samples which allowed us to identify the specific contribution of Fe while controlling for the potential impact of other particle characteristics on the lung response.
Particle size has been demonstrated to impact on the respiratory response to inhaled particles. For example, there is an inverse association between particle size and the lung inflammatory response to silver nanoparticles [39] while there is a positive association for crystalline silica in the fine to ultrafine range [40]. Interestingly we found no association between adverse lung outcomes and MMAD; although it should be noted that variation in MMAD was not large (1.52-4.5 mm). We did, however, find that the variability in particle size (GSD) had a significant impact on both the inflammatory response and lung function deficit 7 days post-exposure. There are very few, if any, studies examining the impact of GSD on the lung response to inhaled particles. One explanation for our data is that a collection of particles with a wide size range is likely to initiate responses down multiple pathways. For example, a sample with a high GSD is likely to have a high mass weighted representation of fine and ultrafine particles which are known to independently have potent impacts on the lung inflammatory response; as compared to a sample with the same MMAD but a lower GSD. Notwithstanding the mechanism behind this association, this observation suggests that additional work is required to elucidate the impact of GSD on lung responses in community settings.
The concentration of Al was also associated with the magnitude of the ''overshoot'' in inflammatory markers below control levels we observed at the 7 day timepont. This response was associated with almost every physico-chemical characteristic we examined (except MMAD). Interestingly, a previous study comparing responses in human epithelial cell lines (BEAS-2B) to soil derived particulate matter of varying composition found the ratio of Al:Si to be a key determinant of the IL-6 response in vitro [4]. While we found no such association in our study (p = 0.46, data not shown), together these data support the notion that mineral heterogenity is an important determinant of the lung response to geogenic particulates.
It should be acknowledged that this study had some limitations. This study was exploratory and observational in nature, and therefore further work will be required to identify the causal pathways for the biological responses we observed. It should also be acknowledged that this study involved a single acute exposure to geogenic PM 10 which may be unlike the typical situation of an individual in a community who is likely to be exposed chronically to these particles.
Notwithstanding these limitations, this study demonstrated that the lung response to a single dose of geogenic PM 10 is complex and highly dependent on specific particle characteristics. We have identified that the concentration of Fe, while not being the only important characteristic, is strongly associated with many of the adverse lung responses to community sampled geogenic PM 10 .