Numerous studies of health effects of wildfire smoke exposure have found increased prescription usage, emergency department and hospital admissions for asthma; and increases in bronchitis, pneumonia and cardiovascular symptoms during wildfire events [3, 6]. In contrast to forest fires, peat fires produce more smoke, persist for longer periods of time and are notoriously difficult to control [7–9, 33]. Little is known about the toxicity of these emissions from different combustion stages however, or if different particle sizes (and resultant chemistry) may influence the effect. Here we investigated pulmonary inflammatory responses and cardiotoxic effects of size-fractionated PM collected from a peat bog wildfire in Eastern North Carolina in the summer of 2008, and also explored the toxicity patterns between in vivo and ex vivo bioassay systems. The results showed that on an equi-mass basis, the coarse PM collected during the active smolder phase of the wildfire event induced the strongest pro-inflammatory responses in the lung in association with endotoxin content and ROS production, while the ultrafine PM collected from the same period caused significant cardiac effects. In addition, the pulmonary cytokine responses in mice correlated well with those in the lung tissue slice ex vivo system.
Chemical components of wildfire PM affected by wildfire combustion stages
Chemical components of wildfire PM have been investigated from various types of wildfires and during different combustion stages. Compared to emissions from different wildfire types (e.g., savanna, forest, and woodland fires), peat fires produce >100% more CO and >300% more CH4, suggesting that they may have a greater impact on human health and climate change . Robinson et al.  reported that flaming and smoldering combustion stages of wildfire produced distinctly different ionic and elemental compositions of fine PM, but other size fractions were not assessed. Here we found that the coarse PM from both sampling periods had similar elemental composition, while the fine and ultrafine PM (which form through nucleation, condensation, and coagulation processes) differed between the smoldering and the glowing phases of the burn . In particular, fine and ultrafine PM from ENCF-1 had higher OC content suggesting they were enriched with wildfire smoke, while the PM samples from the ENCF-4 period had higher concentrations of NH4+ and SO42− indicating that the pollution from this period reflected typical secondary particulate air pollution for the region . The OC mass fractions from ENCF-1 were in good agreement with reported data from the 2005 peat fires in Indonesia , while forest fires often produce even higher OC levels (typically > 90% by mass) , suggesting that OC levels are affected not only by the different combustion stages but also the type (i.e., fuel source) of wildfires. It should be noted that the wind speed during ENCF-1 was higher than during ENCF-4 and the prevailing wind was from the South-west, while ENCF-4 wind conditions were in general lighter and more southerly in nature with some periods of wind from other directions. Despite this caveat, it is clear from the up- and down-wind regional monitors, that the ENCF-1 samples were enriched for wildfire-type pollutants and provided the contrast with ENCF-4 that we intended to explore in this study.
Acute lung inflammation caused by coarse but not smaller size fractions
We and others have previously shown that coarse PM from a variety of different urban locations causes much more lung inflammation than fine or ultrafine fractions, and these effects have been attributed to numerous components including transition metals, crustal material as well as bioactive components such as LPS [39–41]. Similarly we found that on an equivalent mass basis (100 μg PM exposure), the wildfire coarse PM caused relatively more pulmonary toxicity than coarse PM collected from either traffic- or urban derived PM from North Carolina [30, 32], and the effect was at least, in part, explained by endotoxin levels. Studies from other wildfire events have similarly found that coarse PM derived from wildfires induce larger lung inflammatory responses compared with smaller-sized fractions . Whether the LPS originated from the smoke itself or was associated with differences in meteorological conditions of agricultural practices that varied across the two periods is difficult to discern. While it is clear that coarse PM collected from agricultural areas contain more LPS and is associated with increased toxic and pro-inflammatory effects , it is also known that cigarette smoke which is another example of smoldering combustion of agricultural products also contains significant levels of bacterial endotoxin and may, contribute to smoking-related pulmonary diseases . Furthermore, while forest fires (e.g., California wildfires) are dominated by flaming and higher temperature, which would normally degrade LPS [7, 16], peat fires normally burn at lower temperature and can be pyrolyzed between 200-250°C. It is also worth noting that peat is made up of organic materials partially decomposed by bacteria, fungi, insects, and worms, and when burned, produces massive amount of organic aerosol including endotoxins [17, 43]. While the chemical composition of wildfire PM is variable across forest, grassland, agricultural residue, and peat fires [44, 45], polycyclic aromatic hydrocarbons (PAH) are relatively low in abundance in most wildfire-derived PM, and particularly in relation to emissions from peat fires [14, 38, 46].
In addition to the significant association of endotoxin with pulmonary inflammation, we also demonstrated that the observed pulmonary inflammatory effect of wildfire coarse PM increased intracellular ROS production in BALF cells. Although we did not determine which chemicals of the wildfire PM in this study were responsible to induce oxidative stress, the coarse fraction of wildfire PM has been reported to have higher concentrations of carbon-centered radicals that are capable of inducing redox cycling in cells, leading to high oxidative stress [14, 47, 48]. Although both ENCF-1 and −4 coarse PM had a similar organic component fraction (~45%) the ENCF-1 sample induced higher intracellular ROS production suggesting that it had different chemistry and higher concentrations of the carbon-radicals than ENCF-4 coarse PM. In addition to the size-dependent pro-inflammatory responses of wildfire PM, production of pro-inflammatory cytokines and activation of neutrophil recruitment occurred in a time-dependent manner. Airway (or alveolar) epithelial cells, macrophages, fibroblasts, and/or endothelial cells rapidly produce cytokines (e.g., IL-6 and TNF-α) and chemokines (e.g., MIP-2) after exposure to ambient PM which gradually diminishes as inflammatory cell number increases [42, 49–52]. Often macrophage numbers are initially decreased due to their tighter adhesion to the alveolar epithelium and an increased numbers are only seen later in demand for enhanced particle clearance . Here we found no significant changes in macrophage numbers up to 24 h post-exposure, but did observe a pronounced early cytokine response followed by a later influx of neutrophils which is consistent with other findings in studies on wildfire PM exposures [14, 15, 54]. Swiston et al.  reported increased levels of circulating cytokines (e.g., IL-6), and neutrophils (but not macrophages) in the sputum of healthy fire fighters after a day of fire-fighting. In an in vivo study on the toxicology of wildfire PM, Wegesser et al.  demonstrated that concentrations of inflammatory cytokines/chemokines (e.g., TNF-α) in BALF from mice were significantly increased at 6 h post-exposure to wildfire coarse and fine PM, followed by a sharp decrease at 24 h. Exposure of a mouse macrophage cell line (RAW 264.7) to wildfire PM at 24 h increased inflammatory cytokine responses (IL-6, TNF-α and MIP-2) in a dose-dependent manner, and the coarse PM was the most potent stimulator for the cytokine responses .
Cardiac dysfunction caused by ultrafine PM collected during the active wildfire
In addition to the pulmonary effects, cardiac function and injury induced by the wildfire PM were investigated at 24 h post-exposure. Ultrafine PM collected during smoldering stage (ENCF-1) but not glowing stage (ENCF-4), caused significant adverse cardiac effects (e.g., low recovery of post-ischemic LVDP and increased infarct size). Since this fraction had approximately three-fold greater organic matter content than ENCF-4, these adverse effects were likely due to source dependent differences in chemistry. In a similar vein, we have previously reported that near road, but not far road, ultrafine PM had similar effects in this same cardiac reperfusion model compared with the saline control group . Taken together, the results suggest that the size (i.e., ultrafine) as well as chemical composition of these particles influences cardiac toxicity. Although more studies are needed to understand the mechanism(s) of this effect, it is clear that cardiovascular function can be altered by various chemical components including organic compounds, trace metal oxides, and elemental carbon . It should be noted that several epidemiological studies report no significant association between cardiovascular effects and smoke exposure from forest wildfires [56, 57] which may possibly be explained by the difference in chemistry between forest and peat wildfire smoke. On the other hand, peat fires have been associated with increased ED visits for heart failure , consistent with the cardiac effects of ENCF-1 ultrafine PM presented here.
Ex vivo lung tissue slices as a good surrogate for in vivo toxicity screening and testing
The aims of this study were not only to assess toxicity of wildfire PM in vivo but also to investigate whether the in vivo effects could be predicted by ex vivo models (i.e., lung tissue slices). For the past 25 years, lung tissue slices have been successfully developed and applied in the research fields of physiology, pharmacology, pathology, and toxicology [23–28], however few studies have compared response patterns between in vivo and ex vivo models . In this respect, a promising finding was that the lung tissue slice model showed a similar pattern of responses to those seen in mice. Additionally, the lung tissue slices demonstrated very low baseline cytokine production but readily responded to bioactive PM samples and LPS. Exposure to wildfire PM induced differential cytokine responses in BALF of mice characterized by 3–5 times greater responses of cytokine IL-6 and MIP-2 than those of TNF-α and a similar differential cytokine response was observed in lung tissue slices. Besides showing concordance with these three cytokines, we also found that IL-1β was not increased in either the in vivo and ex vivo systems (data not shown), suggesting that the lung slices produced similar responses to the whole lung environment. Due to the fact that there are more than 40 different cell types in the mammalian lung , in vitro cell culture models of one or multiple cell types present challenges in terms of interpretation of cellular toxicity, physiology, and morphology in relation to complex cell-cell contacts and cell-matrix interactions in vivo. For example, it has been reported that slightly different genotoxic responses to wood smoke were observed in alveolar epithelial cells and monocytes . Furthermore, human alveolar epithelial type 2 cell line (A549) and primary rat type 2 cells showed different inflammatory responses to ambient PM exposures . From a practical point of view, it is relatively straightforward to prepare lung tissue slices and also easy to maintain high cell viability in culture, compared to primary cells or cell lines. In addition to the structural and morphological similarities to the whole organ, application of this technique greatly reduces the number of animals needed. In this study, we were able to generate more than 10 lung tissue slices from one mouse lung, thus requiring approximately 10-fold less number of animals. Ideally mouse lung tissue should be replaced possible by human cells and tissues to avoid species differences that complicate risk assessment, not only from a dosimetric perspective, but because of differences in biochemistry and (patho) physiological responses . Since there are many issues (e.g., ethical, legal, logistical, safety, and quality) relating to the use of human tissue for research  however, normal human lung tissue is not widely available for use in laboratories across the world.
The advantages of the lung tissue slice technique lead us to investigate the contribution of endotoxin to the observed inflammatory effect of wildfire PM in this study. The lung slices effectively responded to endotoxin and the coarse PM and this was significantly inhibited by PMB. Since the inflammatory responses to the particles were not completely inhibited by PMB other microbial components may also have played a role  or the binding of PMB may have been reduced by the presence of hydrophobic molecules . Overall however, the results show that the lung tissue slices could be used as a surrogate for in vivo toxicity screening and testing to identify and discriminate key factor(s) that induce pulmonary toxicity.