The aim of this study was to assess the toxicology of aerosolized ZnO when exposed to a realistic triple cell co – culture model of the human epithelial airway barrier, and to provide a comparison with a dose – equivalent suspension scenario. A large number of in vitro models of the histologically different pulmonary epithelial barriers are currently available, and many have already been used to investigate particle – lung interactions. Traditionally, experiments have mainly been performed using monocultures of single cell lines or isolated primary cells
. Such monocultures can be taken as “simplistic” models of a tissue of interest such, in this case the lung. In reality, biology is much more complex. The airway barrier is a multidimensional structure and consists of many different specialized cell types, including epithelia, macrophages and dendritic cells among others
, wherefore increasing emphasis was put on the development of sophisticated 3D co - culture models. Against this background, an advanced, well established and characterized triple – cell in vitro model of the human respiratory tract, whose focus lies on cellular crosstalk, was chosen for this study
[46, 49, 50]. In order to simulate aerosol – inhalation as close as possible, the here used in vitro exposure system combines realistic particle production and defined aerosol aging (in terms of i. a. agglomeration) with a dose – controlled, passive mass deposition by means of diffusion and sedimentation. The combination of a sophisticated exposure system with a realistic cell culture model of the airway epithelium provides us with a reliable tool to investigate aerosol toxicology under “quasi – realistic” occupational conditions.
Nowadays, studies on workplace related exposure to airborne submicron particles are numerous
[5, 8]. Investigations on particle emissions from pilot scale flame – reactors, comparable to the one used in this study, revealed peak number concentrations in the range of 104 to 105 pt/ccm during regular particle production, what was about an order of magnitude above background concentrations
[9, 11]. These data represent routine working conditions. In hazardous situations however (neglected or missing worker health protection, leaks etc.), a substantial higher amount of particle may be released unintentionally into the environment during particle manufacture. Such a situation may occur especially within industrial settings, as simulated in this study, where huge quantities of particulate matter are being produced and handled. Deliberated particles undergo various modifications over time, as they are subject to aging processes like agglomeration, dissolution of the stabilizing shell and full or partial dissolution of the nanoparticle
. Agglomeration is a spontaneous and ever – present interparticle phenomenon for aerosols, resulting in a continuous decrease of number concentration, coupled with an increase in particle (agglomerate) size
, a pattern demonstrated for the ZnO aerosols used in this study (Figure
1B and Additional file
1: Figure S1). The obtained aerosol data were in good agreement with predictions made for monodisperse coagulation, such as a reduction of aerosol number concentration of approximately two orders of magnitude over an exposure time of 30 min for about 108 pt/ccm initial concentration
The deposited ZnO dose (mass per surface area) of 1.3 – 6.1 μg/cm2 measured in this study was found to be comparable with a previous ZnO aerosol – and suspension exposure study carried out by Lenz et al.
, where dose levels of 0.3 – 8.5 μg/cm2 were obtained for the aerosol scenario using an air-liquid exposure system by nebulizing a particle suspension over the surface of lung cells. To put the dose into perspective, it is important to consider, that these doses are within the maximum alveolar lifetime dose accumulated by a worker, which was calculated as 3.6 – 18 μg/cm2, based upon lung morphology, physiology and recommendations of the Occupational Safety and Health Administration (OSHA) (5 mg/m3 threshold value)
Dose – equivalence is a prerequisite for a comparison of aerosol – and suspension exposure scenarios. Under submerged conditions, the Zinc depletion of the cell culture’s supernatant was taken as indirect measure for particle deposition. It could be shown by AAS analysis that most of the administered zinc (on a mass base) has reached the cells after a 2 h exposure period (Figure
7). This fast particle transport can be explained by the high agglomeration tendency of ZnO particles, when dispersed in cell culture medium or water (non – stabilized suspensions)
 (Additional file
1: Figure S3). Over time, large clusters are formed which are deposited with high efficiency (large mass) due to gravitational settling (sedimentation)
Beside the administration of identical amounts of ZnO, it is also essential that suspended as well as aerosolized particles reach the cells in a comparable sate of agglomeration. However, the measured size distributions are generally difficult to compare, as the specific measuring principles of the different particle monitors differ fundamentally. However, the output metrics (mobility equivalent diameter in the given matrix), allow, from a pragmatic point of view, an (at least qualitative) correlation of the particle size data determined by FMPS and DLS.
At first sight, aerosol and suspension data seems to differ significantly, especially with regard to the maximal particle size over prolonged time, which appears much higher in the suspension scenario (Additional file
1: Figure S3). Considering the immanent limitations of FMPS such as the systematic underestimation of fractal particle size and the limited measuring range, the “real” particle diameter seems to far exceed the measuring range (560 nm) already soon after the extinction of the flame due to fast agglomeration (a result of the generally high particle mobility in the gas phase, the high aerosol concentration and the actively mixed atmosphere). Thus, Figure
1 and Additional file
1: Figure S1 show an adequate picture of the initial phase of the experiment (particle production, initial agglomeration) but the FMPS massively underestimate particle size over more prolonged time. In view of the above considerations, the particles size range in either gas or liquid seems to saturate in a comparable order of magnitude (micron range), thus excluding biological effects arising solely from differences in agglomeration state/particle size between the two exposure scenarios.
There is evidence in the literature, that the toxicity of ZnO is mainly driven by the extracellular release of Zn(II) – Ions followed by their subsequent uptake into cells most likely through zinc transporters. Additionally, several ZnO particle - preparations were found to be highly soluble in cell culture medium
. To clarify the amount of dissolved zinc in our experimental system, the solubility (in cell culture medium) of the flame – made ZnO particles, used in this study, was analyzed (Additional file
1: Figure S4). Independent of timespan (up to 4 h) and incubation procedure, only about 4% of the initial ZnO dose (≥ as the maximal added dose in a biological experiment) was dissolved. This reflects the improved stability of the here used ZnO as it has been prepared with the corresponding industrially relevant high temperature process (see experimental section). A number of earlier reports on ZnO used nanoparticles from low temperature synthesis, later are not sintered and dissolve rapidly. In this study, we deliberately chose the industrially more relevant material, as to provide a test material as close as possible to eventual human exposure scenarios
[13, 15]. This finding demonstrates that zinc reaches the cells mainly as ZnO particles and not in an ionic state. This mechanism has been named a Trojan horse type uptake and was first experimentally observed using similar, industrially relevant ZnO nanoparticles (Brunner et al.
) and later confirmed by Xia et al.
A dose - dependent reduction of cell viability is described in the literature following contact of ZnO particles with cell lines derived from human lung (BEAS-2B, A549)
[31, 36], nasal mucosa
, aorta endothelium
, the kidney
 and the lymphatic system
. Additionally, increased cytotoxicity (reduced viability) was measured directly in BEAS-2B, RAW264.7 macrophage, MSTO-211H mesothelioma and 3T3 rodent fibroblast cell lines
[13, 30]. In line with these findings, a pronounced ZnO - cytotoxicity was observed in the 16HBE14o- triple cell co-culture model under aerosol and suspension exposure conditions, as demonstrated by measuring extracellular lactate dehydrogenase (LDH) activity.
Compared to either incubator - or gas – control, all aerosol concentrations induced an elevated, but statistically not significant, LDH release for 4 and 24 h timepoints (Figure
3), indicating, that the dose – response curve seems to reach its plateau already at or before a dose, corresponding to 22 sec particle production. Compared to the Triton positive – control, the values appears as to be comparably high. As a maximal cytotoxicity was already reached after 4 h, and no further cell death could be detected over prolonged time (no time – course of acute toxicity), a certain cell population of the cultures remains viable under all conditions. It can be hypothesized that the different cell types that the co – culture is composed of (epithelial cells, macrophages, dendritic cells) may differ in their susceptibility to zinc, and that the most responsive cell category would be eradicated first, followed by the other populations ordered along a gradient of their susceptibility to ZnO. In addition, ZnO may be deposited on local “hotspots”, where the particle or agglomerate define a radius within which cells may be subject to focal toxicity. Cells localized outside of this area would be protected of any adverse effects and therefore remain viable. In the suspension scenario, no LDH release was observed after 4 hours on the apical and basal side. However, after 24 h post – incubation the values appeared significantly increased in both compartments.
The immanent ability of particles to induced oxidative stress, a phenomenon resulting from an imbalance in the cellular oxidant - antioxidant equilibrium in favor of the oxidants, is a fundamental mechanistic paradigm in (nano) toxicology. Particles may cause oxidative stress by triggering excessive (biotic or abiotic) generation of reactive oxygen species (ROS) so that the cellular antioxidant defence mechanisms may be overwhelmed by these oxidants
[52, 53]. The ability of ZnO to cause elevated ROS levels was demonstrated in several studies e.g. for RAW264.7 mouse macrophage and BEAS-2B human alveolar epithelial cell lines
, A549 human bronchioalveolar cells
, human kidney cell lines (IP15, HK-2)
, a mouse macrophage line (Ana-1)
 and human lymphoplastoid cells (WIL2-NS)
. Furthermore, cell – free (“abiotic”) ROS generation was identified as a material – specific property of ZnO
[30, 54]. Under physiological conditions, the steady-state formation of cellular oxidants, an attribute of aerobic life, is balanced by a similar rate of their neutralization by antioxidants that are of enzymatic and non-enzymatic origin
. In contrast, oxidative stress causes a depletion of protective cellular antioxidants due to overburdened defence mechanisms. In this context, the essential, ubiquitous non-enzymatic radical scavenger glutathione (GSH)
 is of great importance and its depletion can be used as indirect measure for oxidative stress. Reduced cellular GSH concentrations upon ZnO exposure were already measured in several studies
The quantification of total reduced glutathione in aerosol – samples revealed a distinct loss of total reduced GSH when cells were either exposed to flame off-gases or to all different particle concentrations at any timepoint (Figure
4). This behavior indicates a strong effect of the gaseous components of the aerosol on the degree of cellular oxidative stress. Interestingly, also after removal of the factor “flame – gases” and 24 h post – incubation, the values do not recover to baseline. This may be evidence of the fact that the cells undergo persistent and irreversible alterations by the gas components. It is also conceivable, that the initial reduction of total GSH at 4 h was caused by gases, an effect which was later on (after 24 h) overtaken by the deposited zinc particles.
In the suspension scenario, no alteration of total GSH was observed under any conditions (Figure
9). This finding supports the assumption that interaction of cell cultures at the air – liquid interface with the gas atmosphere may strongly influence the toxicological outcome. The gas components of an aerosol are adding an additional layer of complexity to this scenario, whose influence needs to be assessed for each individual factor. As combustion by product of flame spray pyrolysis, gaseous compounds such as CO2, CO, NOx, are being released
. It was shown in the literature that exposure to filtered (particle free) diesel exhaust induces an acute reduction of the cellular GSH content, whereby glutathione levels were restored to control levels 24 h after a 1 h exposure
. This finding is in contrast to the current study, as the reduction of intracellular glutathione was not reversible by placing the cultures for 24 h in an incubator. A prominent nitric oxide, namely the oxidant NO2, was found to directly interact with GSH, causing a reduction of glutathione
. This proves a mechanistic link between nitric oxide and the cellular antioxidant capacity.
In order to organize cellular responses to particles along a gradient of oxidative stress, a “hierarchical oxidative stress” model has been proposed
[30, 53]. Following this well-established paradigm, the lowest level of oxidative stress (Tier 1) is associated with the induction of a battery of antioxidant defence genes by redox – sensitive transcription factors, mainly Nrf2
, to restore redox homeostasis. A characteristic marker among this panel is HO-1, an antioxidant enzyme which was found to be up - regulated upon exposure to diesel exhaust
, the water insoluble fraction of fly ash
, cerium oxide
 and ZnO
[21, 30]. We have found that the pattern of HO-1 regulation differed among the two exposure scenarios. In aerosol exposed samples, gene induction correlates positively with applied dose for 4 and 24 h (Figure
5). In the suspension scenario, HO-1 appears elevated for 4 h and shows a significant dose – dependent trend for 24 h (Figure
10). Incubator and gas – control do not differ significantly, indicating that gas components alone are not able to induce heme oxygenase directly, although gas exposure causes oxidative stress as demonstrated by the reduction of intracellular reduced GSH. Consistently, also no indication for any oxidative stress can be derived from the GSH data in the suspension scenario, however a regulation of HO-1 was observed. The HO-1 induction is shaped mainly by ZnO exposure.
A further important line of antioxidant enzyme defence is the superoxide dismutase family (SOD), whereof, the SOD1 isoform (CuZn-SOD) was chosen as additional target in this study. In both scenarios, no regulation of SOD1 was observed (Figures
Further escalation of oxidative stress (Tier 2 level) is supposed to induce a number of pro - inflammatory pathways, such as mitogen – activated protein kinase (MAPK) - and NFκB cascades, resulting in the release of a wide array of cytokines and chemokines. Although the underlying regulatory mechanisms are little researched, ZnO was found to trigger the in vitro translation of inflammatory mediators such as TNFα
[30, 35] and IL-8
[30, 36] in several cell culture systems. For the latter, also transcriptional activation was observed
[21, 38]. Furthermore, a number of studies clearly demonstrated elevation of a set of inflammatory biomarkers (TNFα, Il-8, Il-6, IL-1β) in lung lavage fluid of human volunteers upon inhalation exposure of ZnO aerosols (“fumes”) at high concentrations. ZnO dose was found to correlate with TNFα, IL-8 and IL-6 response for several timepoints
[61, 62]. The observed temporal pattern lead to the hypothesis, that the early increased TNFα plays an important initial role in the pathology of metal fume fever, as it may lead the secondary release of IL-8 and IL-6 in the pulmonary environment.
To assess the pro – inflammatory potential of ZnO, a set of inflammatory mediators such as TNFα, IL – 8, IL – 6 and IL - 1β was assessed in this context. Only TNFα concentrations were found to be affected by zinc oxide. In the aerosol scenario, TNFα levels remained on gas – control baseline after 4 and 24 h (Figure
6, Additional file
1: Chart S1). However, gas – and incubator – control differed significantly, illustrating, that the basal TNFα level is elevated in glove box exposed cultures as a result of gas exposure. In the suspension scenario, a significant time and dose dependent increase of the extracellular TNFα concentration was observed for both timepoints in the apical, ZnO exposed, compartment. After 24 h, such an effect was also found in the basal well.