In various studies ufTiO2 particles have been shown to possess increased inflammogenic potential in comparison to fTiO2 [2, 8, 40, 41]. Physiologic and systemic reactions towards NP exposure, including the ufTiO2 that was used in this study, have been shown in several in vivo investigations [6, 9, 42, 43]. However, for the investigation of underlying basic cellular mechanisms and pathways, in vitro studies are necessary with established cell lines, e.g. NR8383 cells . Responses of these cells to various toxicants such as PMA, endotoxin and DQ12 were shown to be highly comparable to those in primary AM obtained from rat lungs by bronchoalveolar lavage [44, 45].
In the present study we observed a rapid internalization of fTiO2 , ufTiO2 and DQ12 by AM in a clear dose-dependent manner. Direct comparison between the specific particles was not possible, because of intrinsic differences in their light scattering properties as observed under cell free testing conditions. However, our study demonstrates that, although all three particle types are taken up, the cellular responses of the AM are substantially different. DQ12 and ufTiO2 showed similar cytotoxicity, while significant effects for fTiO2 were absent. This confirms the accepted view in particle toxicology that uptake of inorganic poorly soluble particles does not necessarily culminate in a toxic response in AM. Our findings also show that particle uptake per se does not dictate oxidative stress and the induction of inflammatory mediators. DQ12 represented the most potent sample in inducing NF-κB activation and release of TNF-α and IL-1β from the AM. HO-1 and iNOS mRNA expression levels in AM were most pronounced after treatment with ufTiO2 . Remarkably also, both DQ12 and ufTiO2 triggered TNF-α release, while only DQ12 induced IL-1β release.
The contrasting abilities of DQ12, ufTiO2 and fTiO2 to induce IL-1β and/or TNF-α release can likely be explained by underlying differences in signaling pathways of activation of both inflammatory genes. NF-κB, a key regulator in the pathogenesis of particle-induced diseases [15, 46], controls the expression of cytokines, growth factors and distinct enzymes in response to ligation of many receptors involved in immunity . Indeed, in our current study, an association between TNF-α levels in the supernatants from AM upon particle treatment at equal mass (i.e. DQ12 > ufTiO2 > fTiO2 ) and their abilities to cause NF-κB p65 nuclear translocation was found. The exclusive effect of DQ12 on IL-1β release is likely to be explained by the recently unraveled mechanism of its cellular activation via the inflammasome. IL-1β is produced as the inactive cytoplasmic precursor proIL-1β which has to be cleaved by caspase-1 to generate the mature active form of the protein [48–50]. In turn, caspase-1 is regulated by the inflammasome protein complex NALP3 , which has been proposed to be activated by crystalline silica particles following lysosomal rupture  or by NADPH oxidase-generated ROS driven by phagocytosis . A recent study has revealed that upon priming with LPS (to induce proIL-1β), both DQ12 and ufTiO2 trigger IL-1β secretion from bone marrow derived dendritic cells from wild-type but not caspase-1 or NLRP3-deficient mice . This suggests that the contrasting IL-1β responses observed with NR8383 cells may be due to differences in the abilities of specific types of poorly soluble particles to act on proIL-1β activation, i.e. upstream of the inflammasome activation.
A further remarkable observation in our study concerned the mRNA expression of HO-1 and iNOS. The positive control DQ12 appeared to be less potent than ufTiO2 with regard to the induction of mRNA expression of both genes. HO-1 is considered as a sensitive marker of oxidative stress and has shown to be induced by inhaled ambient ultrafine particles  as well as by DQ12 quartz [45, 56]. The induction of iNOS in macrophages has been well-established in previous studies for crystalline silica particles, and this is considered to play a major role in its pulmonary toxicity . The contrasts in nuclear translocation of NF-κBp65, iNOS and HO-1 mRNA expression in NR8383 cells in response to ufTiO2 and DQ12 suggest that particle-induced iNOS activation in AM can occur in an NF-κB-independent manner. Thus, while ufTiO2 and DQ12 both trigger pro-inflammatory effects unlike fTiO2 , these particles likely activate AM through different mechanisms.
In our present study, calcium influx and intracellular ROS generation were observed in AM with all three particle types to a similar extent, although both are considered as key mechanisms for adverse particle effects . Tian and colleagues  recently demonstrated that ROS do not modulate [Ca2+]i in quartz-treated rat AM, however, calcium increase in the cytoplasm causes ROS generation after silica treatment. Enhanced [Ca2+]i in relation to pro-inflammatory signaling pathways has also been observed after treatment of macrophages with ultrafine carbon black (CB) particles in contrast to fine CB . To the best of our knowledge, a comparison of effects between fTiO2 and ufTiO2 on calcium homeostasis in macrophages has not yet been performed. We observed no clear difference between both particle types in terms of the number of activated cells or the intensity of activation. In line with this, intracellular ROS levels also did not differ after treatment with both types of TiO2 . Our findings are in contrast to observations with CB  and suggest that particle size- and/or surface area-dependent effects on calcium influx and ROS formation are (nano)particle type-specific.
Besides intracellular ROS by DCFH-DA assay, we also determined extracellular ROS levels by means of EPR. Significant increases were observed after treatment with DQ12 and ufTiO2 , but not after fTiO2 . Previous studies indicate that fTiO2 and ufTiO2 samples do not markedly differ in their intrinsic ROS generating capacity, when measured in cell free assays in the absence of photosensitization [8, 60, 61]. In concordance with our current findings in NR8383 cells, we could previously also show that ufTiO2 , unlike fTiO2 , caused enhanced ROS formation in supernatants of A549 human lung epithelial cells. This suggests that ROS predominantly originate from interactions between ufTiO2 and cellular constituents and compartments rather than from the particles themselves. Potential relevant sources herein include NADPH oxidase enzyme family members as well as mitochondria . Our findings indicate that different ROS-generating mechanisms exist in AM, with a selective sensitivity towards particle size or chemical composition as already concluded by Dick and colleagues . At this stage however, it should be emphasized that the calcium imaging experiments and both ROS assays were not performed in complete culture medium, but in saline, or HBSS+/+, respectively. This was required to minimize potent radical scavenging properties of various (protein) constituents in the FCS-containing medium that can interfere with the assays. DLS measurements on unfiltered samples demonstrated that both fTiO2 and ufTiO2 , when suspended in HBSS, reside as large agglomerates with an average hydrodynamic diameter of 936.6 or 2018 nm, respectively, unlike in FCS-containing medium (see Table 1). Lacking differences in calcium influx and intracellular ROS between fTiO2 and ufTiO2 may therefore reflect an "agglomeration"-response of NR8383 cells. Interestingly though, increased extracellular ROS levels could be shown for ufTiO2 by EPR analysis, despite its agglomeration. All other parameters in our study were evaluated using FCS-containing culture medium, in which the number-average diameter of ufTiO2 sample was well within the nanosize range. However, apart from effects on agglomeration behavior, these treatment conditions also generate so-called (protein) coronas, most probably in a material specific manner . This should be taken into account with regard to the various effects described in our study.
The importance of agglomeration and (protein) coating effects can be demonstrated from the comparative mRNA expression measurement of HO-1 and iNOS in HBSS treated versus full medium treated NR8383 cells (Figure 8). When suspended in HBSS, ufTiO2 failed to cause a significant increase of HO-1 mRNA, which indicated that the agglomeration state (see Table 1) of this sample is crucial for its ability to induce this oxidative stress marker. Remarkably however, the induction of iNOS by ufTiO2 was not abrogated. This suggests that the activation of HO-1 and iNOS by particles involves, at least in part, different signaling pathways driven by different physico-chemical properties. In contrast to the ultrafine TiO2 , the crystalline silica sample induced HO-1 and iNOS under both treatment conditions. At the lower treatment concentration (10 μg/cm2), the effect of DQ12 tended to be even stronger in the HBSS than in the FCS-containing medium. Current observations are in concordance with previous investigations in our laboratory where chemical coating of DQ12 was shown to abrogate its pro-inflammatory properties in NR8383 cells [45, 64], as well as in vivo in the rat lung .
The goal of our study was to investigate the interactions between particles and AM and their associated pro-inflammatory effects in relation to particle size and physico-chemical properties. The contrasting cellular responses observed by the three types of particles could not be explained by uptake by the AM per se. Therefore, investigations were performed addressing the underlying cellular mechanisms of particle internalization. Herein, the importance of particle size and distribution as well as of agglomeration behavior in cell culture medium suspensions was taken into account. We also specifically compared the uptake mechanisms for both TiO2 samples with those that we previously investigated for DQ12 in NR8383 cells [20, 65]. Evaluation of uptake at 4°C indicates a passive, energy-independent entrance of particles into cells and/or their adherence to outer membranes of AM. For both TiO2 samples the proportion was found to be higher than for DQ12. However, no clear difference could be seen between fTiO2 and ufTiO2 despite the marked differences in their size distributions in complete culture medium used for the uptake experiments. These observations are in line with previous findings for both materials concerning their uptake into A549 human lung epithelial cells .
A series of specific inhibition experiments were performed to investigate the various active uptake routes in NR8383 cells. A combination between different uptake mechanisms in our study can be reasoned by the findings of Rothen-Rutishauser and colleagues  showing TiO2 particles in a three-dimensional cell culture model free in the cytoplasm as well as membrane-bound. Our own findings indicate that the active internalization of ufTiO2 particles in AM is mainly performed via a FcγRII-mediated mechanism and, to a lesser extent, by CCP which exhibit a vesicle diameter of 100 - 120 nm . Uptake of fTiO2 particles also took place via CCP, but in addition an actin-dependent uptake mechanism was equally involved. This may include macropinocytosis, by which large vesicles between 0.2 - 10 μm are formed spontaneously or upon stimulation . Actin-mediated endocytosis is connected to receptor activation like MARCO and SR-A mediated processes, as previously shown by Kobzik and co-workers for primary AM of different species [69–71]. The prominent receptor-mediated uptake mechanism for fTiO2 and silica by human macrophages via SR-A reported by Thakur et al.  is beyond all question for our study, since NR8383 cells lack this receptor as determined by PCR analysis (data not shown). As expected from their size distribution, the DQ12 particles were taken up by actin-dependent classical phagocytosis which is described to be mediated by FcγRII . Phagocytosis is the most effective clearance mechanism for particles between 1 - 5 μm in diameter . Inhibition experiments with filipin III were found to be unsuccessful in reducing particle uptake in NR8383 cells. For DQ12 and fTiO2 this could be anticipated in view of their size distributions and the typical diameter of 50 - 100 nm of the caveolae vesicles. However, filipin III was also ineffective for ufTiO2 , despite the fact that its number-average hydrodynamic diameter falls into the vesicle size range of caveolae. This suggests that even for these smaller particles/aggregates alternative uptake pathways such as CCP dominate. Notably, apart from differences in primary particle size and agglomeration behavior, both TiO2 samples also differ in their chemical composition. Therefore, a potential role of rutile vs. anatase in particle-macrophage interactions can not be ruled out. Another explanation may be related to the specific method of particle uptake used in the present study. In relation to the relative contribution of particle number and particle mass to changes in AM granularity it is possible that the uptake of the smallest particles is underestimated in the flow cytometry approach. However, in a recent study it was demonstrated that flow cytometry can detect organosilica nanoparticles as small as 58 nm in diameter via side scattering analysis and that the method is actually suitable for size distribution analysis . Comparison of the TEM distribution data of the samples used in our study with SSC histograms of cell free particle suspensions obtained with the same apparatus (FACS Calibur) indicates that this is also valid for TiO2 (data not shown).
A major conclusion that can be drawn from the uptake experiments is that fTiO2 and DQ12 which are both classified as "fine" particles show a specificity and size-dependency with regard to the tested cellular uptake mechanisms. Further studies are needed to investigate these alternative uptake pathways using independent inhibition strategies (e.g. siRNA). Because of the limitations of the cell line used in our present study (e.g. lack of SR), this should be done preferably with primary macrophages. These studies should also focus on the potential contribution of particle-specific coronas to kinetics and pathways of uptake and associated cellular responses.
Our observation that multiple uptake mechanisms may be relevant for one specific type of particle can be explained by the size-distribution of the specific samples and, likely of more importance, their agglomeration behavior when suspended in culture media. Obviously, for the identification of the in vivo relevant mechanisms of uptake of specific types of (nano)particles by AM, one should obviously take into account the role of the micro-environment of these cells, e.g. potential corona-forming constituents of the alveolar lining fluid. Nevertheless, our current in vitro findings emphasize on the aspect that the inflammatory properties are not driven by uptake per se. It appeared that specific physico-chemical properties of quartz, fTiO2 and ufTiO2 particles are responsible for qualitative as well as quantitative differences in their ability to induce oxidative stress and inflammatory responses in AM. In contrast to fTiO2 which was relatively inert, both DQ12 and ufTiO2 increased extracellular ROS and TNF-α release, while ufTiO2 predominantly enhanced iNOS mRNA expression, and DQ12 exclusively triggered IL-1β release. Our findings indicate that these dissimilar macrophage responses may be related to specific differences in uptake mechanisms, respectively involving actin cytoskeleton, CCP formation and FcγRII-mediated internalization. The findings with fTiO2 demonstrate that actin and CCP are not necessarily involved in pro-inflammatory cytokine release and extracellular ROS generation by macrophages after particle uptake. The data for DQ12 and ufTiO2 indicate a role for FcγRII in macrophage responsiveness. The contrasting activation profiles for both materials emphasize the need for further investigations, specifically regarding the inflammasome. The relative overlap of the investigated mechanisms for the various types of particles used in this study are likely explained by their size distributions. Indeed, studies with monodisperse particles are most appropriate to clarify individual mechanisms of uptake by macrophages. However, it should be emphasized that AM typically encounter size ranges of particles and their agglomerates of various types of respirable materials, including the fine and ultrafine TiO2 samples used in our current study.