Using a 3D tri-culture lung model, we found that non-crystalline SiNPs (Si10) and crystalline silica particles of micro-meter size (Min-U-Sil) induced gene expression and release of pro-inflammatory cytokines in the apical epithelial/macrophage compartment and the basolateral endothelial compartment. Effects were time- and concentration-dependent. Endothelial gene expression were mainly exerted at higher concentrations, delayed and of less magnitude when compared to the apical macrophage/epithelial cell compartment. Furthermore, Si10 seemed more potent than Min-U-Sil on a mass basis, but effects of Si10 were delayed. Similar findings were observed for gene expression of adhesion molecules. An IL-1R antagonist almost completely blocked silica particle-induced up-regulation of cytokine release, and cytokine and adhesion molecule gene expression in both compartments. Thus, the present study suggests that IL-1α/β are central mediators of pro-inflammatory responses in both compartments of the 3D tri-culture.
The findings with transfer of pro-inflammatory mediators across the cell barrier expand the knowledge of our previous studies with Min-U-Sil using a 2D co-culture system involving THP-1 monocytes, A549 cells and another endothelial cell line (HIVE-26) . Effects at lower concentrations and the higher magnitude of endothelial responses seen in the present study may be due to the 3D organization, which brings the endothelial cells in closer contact with the A549 cells and/or the addition of THP-1 macrophages. Use of THP-1 macrophages has been shown to have a large impact on the complexity of the system [44, 48, 59]. We previously observed that also non-differentiated THP-1 monocytes exposed to Min-U-Sil in the apical compartment (in absence of A549 cells) increased the gene expression of cytokines in endothelial cells in the basolateral compartment . Thus, apparently both addition of THP-1 monocytes and macrophages will result in co-culture models with increased cytokine responses.
Our present data, show that addition of an IL-R antagonist almost completely blocked effects of Min-U-Sil and Si10 on gene expression of cytokines and adhesion molecules in the apical epithelial cells and macrophages, as well as endothelial cells on the basolateral side. As increased IL-1α/β expression is required for an up-regulation of the gene expression of cytokines and adhesion molecules in both cell compartments, the data suggest that IL-1α/β cytokines from the apical side cross over the cell barrier to the basolateral endothelial cells. Notably, we have previously shown that IL-1β is released from mouse macrophage (RAW264.7) upon exposure to different types and sizes of silica particles . IL-1α is an alarmin, known to be released from both macrophages  as well as epithelial cells upon exposure to various agents [61, 62]. Also IL-1β is released from both macrophages and epithelial lung cells upon exposure to silica particles [12, 18]. IL-1β release is due to silica particle-induced activation of the inflammasome due to uptake of particles causing destabilization of the lysosomal membrane [12, 16, 18, 63, 64]. Furthermore, we have in previous studies found that SiNPs also are able to induce gene expression and release of IL-1α and IL-1β in lung epithelial cells at rather low concentrations . Interestingly, IL-1α and IL-1β are both known to be critical paracrine (and endocrine) mediators of inflammation linked to lung [34, 60, 65, 66] and cardiovascular disease [35, 36]. A number of in vitro studies have also illustrated that these cytokines may trigger other inflammatory responses. We have previously reported that an IL-R antagonist reduced Min-U-Sil-induced cytokine release in different co-culture systems, involving THP-1 monocytes, alveolar lung epithelial cells (A549) and endothelial cells (HIVE 26), suggesting that IL-1 is important for the interplay between the different cell types [49, 50, 67]. In the present study we show in a more complex 3D co-culture system that cytokine and adhesion molecule responses are due to IL-1-dependent gene expression.
In our 3D tri-culture, concentrations of IL-1 in the apical compartment was higher than in the basolateral compartment, suggesting that the IL-1 signalling may start on the apical side. This may also explain why enhanced gene expression of cytokines and adhesion molecules are seen at lower concentrations of silica particles in the apical macrophages/epithelial cells than compared to basolateral endothelial cells. Similarly, the up-regulations in endothelial cells are lower in magnitude and time-delayed compared to macrophages/epithelial cells. This may be related to time-points and concentrations at which sufficient IL-1 levels are reached in basolateral compartment to trigger effects. These findings presumably reflect what occurs in vivo upon inhalation exposure of particles, with systemic responses of much less magnitude and appearing at later time-points than in the airways. This suggests that such 3D co-culture systems may have properties that make them suitable to examine some aspects of particle toxicology, and might to some extent replace in vivo studies.
The IL-1R antagonist used in this work is not specific, as its affects both IL-1α and IL-1β ligand binding. Since release of IL-1β strongly exceeds the amounts of IL-1α, this may suggest that IL-1β is the critical signalling molecule. This is supported by the fact that while IL-1β appears to have a controlled release at non-cytotoxic concentrations, release of IL-1α is more often parallel to cell death. However, IL-1α has also been reported to be an alarmin and a master cytokine that induces acute lung inflammation via pro-IL-1β synthesis and IL-1β release . Furthermore, as the responses will depend on the affinity of IL-1α and IL-1β towards the IL-1 receptor, we will not draw any final conclusion on the relative role of these IL-1R ligands. Further studies are required to elucidate this. Overall, our present findings support that IL-1α/β act as paracrine regulators, promoting interaction between THP-1 macrophages and A549 cells. THP-1 cells have also been reported to activate the endothelial cells separated from the macrophages/lung epithelial cells by a micro-porous membrane (pore size 1 μm), which these cells cannot penetrate . It is interesting to note that we have demonstrated a similar IL-1-dependent regulation of pro-inflammatory responses in co-cultures of primary cardiac fibroblasts and cardiomyocytes , indicating IL-1 to be a general mediator in paracrine interactions of different cell types. Although, the present evidence supports a paracrine role of IL-1, responsible for the interplay between THP-1 macrophages, A549 cells and endothelial cells, IL-1 may also have a potential role to act at longer distances from the releasing cell source. Accordingly, in our previous tri-cultures which also suggested a role for IL-1, the endothelial cells were localized at the bottom of the wells . Furthermore, we have shown that transfer of conditioned medium from primary rat lung epithelial type 2 cells exposed to ultrafine Printex particles enhanced release of IL-6 and CXCL8 from rat primary cardiac co-cultures in a partial IL-1-dependent manner . Thus, IL-1 released due to airway exposure to Min-U-Sil- and Si10 may potentially cause systemic effects. This will obviously depend on the concentration of IL-1 reaching different target organs.
In our 3D co-culture system rather high concentrations of silica particles are required for eliciting sufficient high concentrations of IL-1 to trigger endothelial cytokine gene expression. A critical question is whether sufficient high concentrations of IL-1 will be reached in plasma during in vivo silica particle exposure. Previously, it has been reported that amorphous silica particles (Si50, Si500) upon intra-tracheal administration (50 mg/kg) in mice increased plasma IL-β concentrations to 0.2–0.3 ng/ml . Such concentrations would not induce an endothelial gene expression in our co-culture system, however it should emphasized that such extrapolation from in vivo to in vitro settings, are difficult and should be carefully interpreted. The concentrations of silica particles used in the present study are in the concentrations range of previous in vitro studies [1, 70]. When using the estimation method of Li and coworkers 2003 , the concentrations of Min-U-Sil (24–192 μg/cm2) in our co-cultures (apically) seem to be in the upper range or somewhat above what should be expected from occupational exposure, with observed silica concentrations up to 0.6 mg/m3 and an EU occupational exposure limit (OEL) of 0.1 mg/m3. For SiNPs the knowledge of exposure levels are lacking , making it difficult to estimate the relevant concentrations for in vitro exposure.
One other major hypothesis for systemic effects of NPs, is that NPs may translocate from the airways into the circulation, thus reaching secondary organs, and triggering inflammation and pathogenic processes. Accordingly, gold-NPs have been reported at atherosclerotic plaques, however only small amounts of inhaled NPs seem to translocate into the blood . The critical question is whether these concentrations are sufficient to trigger inflammatory responses [39, 40]. Although not specifically addressed in the present in vitro study, a question is whether SiNPs may pass the epithelial barrier and the micro-porous membrane in 3D tri-cultures, and thereby exert direct effects on endothelial cells. Rhodamine-labelled Si50 added to the presently used 3D tri-culture was only located in THP-1 macrophages and not in A549 cells nor endothelial cells . In contrast, translocation of NPs across the filter/membrane barriers to endothelial cells has been reported in a rather similar 3D co-culture model . This may depend on the particle size and its agglomeration state as well as the pore size of the filter. In the current study we used filters with pore size of 1 μm, while SiNPs had a lower nominal size of 10 nm, and will penetrate the filter without cells. Thus we cannot exclude that Si10 could pass across the cell barrier, but we presume that this does not occur in sufficient amounts to trigger a response in the endothelial cells. In support of this, the passage of sodium fluorescein across the cell barrier in the 3D co-cultures was rather low. Furthermore, Min-U-Sil particles with a nominal median size of 1.4 μm that could not pass the filters, still caused very similar responses as Si10. Finally, upon assuming that a high amount of Si10 particles could pass the cell barrier on the insert membrane, we examined the potential of Si10 to directly activate the endothelial cells in mono-cultures. These studies showed that rather high concentrations of Si10 were required to induce cytokine responses, and most importantly, the cytokine response pattern in the mono-cultures was qualitatively different from the pattern in the 3D co-cultures; with no IL-1β and TNF-α responses in the monocultures. Overall this suggests that IL-1, and not Si10 pass the cell barrier, and is the critical determinant in eliciting endothelial responses in our co-culture model.
Our data show that Si10 was more potent than Min-U-Sil on the gene expression in both compartments when related to PM mass, whereas Min-U-Sil was more potent when related to PM surface area. (Fig. S3). The experiments were performed by using submerged culture conditions and different dispersion conditions. Conceivably, using ALI-exposure and similar dispersants could have influenced the relative potency. Furthermore, different mechanisms for uptake of the silica particles of different sizes would affect the potency for induction of pro-inflammatory responses and cytotoxicity [8,9,10]. The early up-regulation of genes in both compartments seen after exposure to Min-U-Sil compared to Si10 particles in the submerged co-culture is presumably due to the higher sedimentation rate and faster contact with the apical cells.
A role for both ROS and metalloprotease activity has been reported for particle- and NP-induced cytokine gene expression [17, 74, 75]. Here enhanced gene expressions of HO-1, MMP-1 and MMP-9 were only seen at later time-points. This suggests that the induced gene expression of HO-1 and metalloproteases are not necessary for the acute Si10- and Min-U-Sil-induced changes seen in pro-inflammatory genes. Up-regulation of HO-1, MMP-1 and -9 may possibly be related to downstream effects of earlier cytokine responses. Similarly, exposing the same 3D tri-culture to DEP, MMP-1 was only induced at the latest time point .
It is important to emphasize that our 3D co-culture is only a model, and that it needs to be further improved. Results from such models cannot be directly extrapolated to in vivo settings. A low trans-epithelial transport of sodium fluorescein across the cell barrier was observed in our model, indicating a relatively tight cell barrier. On the other hand, the trans-epithelial resistance (TEER)-value was rather low, indicating that A549 cells have no functional tight junctions . Theoretically, IL-1α/β may be actively transported from the apical epithelial cells/macrophages across the alveolar barrier to the basolateral endothelial cells, or passively diffuse across the barrier via non-functional cellular tight junctions. Several studies have also shown that exposure to toxicants may reduce TEER values . This may result in unspecific leakage of cytokines across the barrier. Presumably, this is not occurring in the present study, as neither Si10 nor Min-U-Sil induced any toxicity. However, to more properly address the importance of IL-1 in 3D co-culture systems, there is a need to optimize the models to have tight alveolar-endothelial cell barriers. To achieve a more functional lung cell barrier other epithelial lung cell lines need to be introduced to replace A549 cells, to better mimic the in vivo situation [73, 76,77,78,79]. As A549 cells also are known to be less responsive to different toxicants  than many other lung epithelial cell lines, and in particular compared to primary type 2 cells, replacement of A549 cells with other lung epithelial cell types, might enhance the sensitivity of the model system. The ratio between THP-1 cells and A549 cells in our model is approximately 1 to 2.5, which is in the upper range of what could be expected for the ratio between macrophages and type 2 epithelial lung cells in alveoli from humans . Thus, the number of macrophages in our co-culture model seems rather high, making it most representative for lung-disease states. Notably, it has been reported that the macrophage number in the human alveoli is approximately 100/ mm2 and 200/ mm2 in healthy, non-smoking individuals and COPD-patients, respectively . Presently, work is in progress to introduce other epithelial lung cell lines, and to examine the influence of the macrophage number in the co-culture model, to improve and make the models as representative as possible for in vivo conditions.