Since TiO2 NP tends to agglomerate in aerosols [9, 32–34], establishing the effect of these NP involves considering the primary particle size and that of the agglomerates . The agglomeration state of NP influences the site of particle deposition in the respiratory tract and affects lung clearance mechanisms, including endocytosis [9, 19, 45–53]. In general, the size distribution of a NP aerosol is composed of a less (< 100 nm) and highly (> 100 nm) agglomerated fraction, with the percentage varying from one aerosol to another. The less agglomerated fraction, for which the size of the agglomerates is closer to the size of the primary NP, could possibly induce effects related to their interaction with lung tissue and epithelial cells at the site of pulmonary deposition. These small particles are also more readily available for translocation to the lymph nodes or bloodstream . Larger agglomerates (> 100 nm) are thought to be more easily detected and removed by the lung macrophages [41, 54–57].
In our study, the cellular pulmonary response observed for all exposed groups denoted by the increases in leukocytes from BALF compared to the controls (Table 3, Figure 4) could be considered as a normal immediate response to particle aggression . The increase in the number of macrophages and neutrophils is thought to contribute to particle removal. Indeed, previous studies have shown that following inhalation of nano-TiO2 this type of response is temporary and resolves rapidly [9, 19, 30, 32]. Hence, after an acute exposure, this could be a defense mechanism .
NP aerosol characterization
For the LA aerosols, we observed that the 5 nm TiO2 produced larger agglomerates than the 10–30 and 50 nm particles (Table 2). This result could be partly explained by the fact that as particle size decreases, the attractive force per unit mass increases, which favors agglomeration . Indeed, small particles that coagulate into agglomerates larger than superior-sized counterparts is a common finding previously described in numerous inhalation studies, including ones using nano-TiO2
[11, 30, 38, 45, 61].
As shown in Figure 3, the estimated pulmonary deposition was different for the LA and SA aerosols. Also, given their different agglomeration states (Table 2), it can be assumed, as described by Oberdörster et al.
 as well as Geiser and Kreyling , that the penetration of these aerosols into the various regions of the respiratory tract is different. Considering their size distribution characteristics (D25 = 128 nm to D75 = 783 nm; Table 2), particles of the LA aerosols could be more easily detected by immune system cells, including alveolar macrophages (Table 3, Figures 7 and 8). Indeed, the estimation of NP endocytosis showed for all LA aerosols that 79 ± 2% of macrophages contains nano-TiO2 agglomerates (Table 3). In our study, activation of macrophages following phagocytosis of large NP agglomerates (Figure 7) is supported by a slight but significant increase in the total cell count, and number of macrophages and neutrophils compared to controls for two out of the three LA aerosols (10–30 and 50 nm) (Table 3 and Figure 4). However, the 50 nm LA aerosol did not increase the level of inflammatory cytokines, whereas interferon γ (INF-γ), chemokine C-X-C motif ligand 7 (CXCL7), interleukine-6 (IL-6), macrophage inflammatory protein 1α (MIP-1α) and MIP-3α were increased for the 10–30 nm LA aerosol when compared to the controls. These cytokines are produced by activated macrophages and act in host defense by promoting phagocytosis, resulting in chemotaxis, inflammatory cell recruitment and activation at the site of injury [63–65]. In particular, MIP-1α activates granulocytes (neutrophils, eosinophils, basophils) which can lead to acute neutrophilic inflammation [63–65]. The detection and phagocytosis of these agglomerates by macrophages can prevent their interaction with lung cells and tissue [41, 45, 66, 67]. This is also supported by our results, where no cytotoxicity or oxidative stress effects, evaluated through LDH activity and 8-isoprostane concentration, were observed for these LA aerosols (Figure 6). LDH is a cytoplasmic enzyme that is released by dead cells and is therefore a suitable marker of cell cytotoxicity, while 8-isoprostane is a biomarker of lipid peroxidation and thus an indicator of oxidative stress effects [68, 69].
An increase in the number of neutrophils supports the presence of an inflammatory reaction [12, 32, 39, 65]. Thus, a mild significant inflammatory (p < 0.05) response was observed following the 10–30 and 50 nm LA aerosols exposures. These results are consistent with the common finding of various nanotoxicological studies on the increases in the number of neutrophils following agglomerated nano-TiO2 exposures [9, 19, 32, 41, 45, 70–72]. Previous studies have also observed that inhalation of agglomerated nano-TiO2, in mice and rats, caused slight inflammatory responses and long-term pulmonary inflammation [9, 19, 30, 32, 72–75].
For the 5 and 10–30 nm SA aerosols, the cytological analysis showed a statistically significant increase in total cell count and number of macrophages (Table 3, Figure 4). Qualitatively, these results are consistent with the histopathological findings (Figure 8). The estimation of NP endocytosis showed that all SA aerosols had 57 ± 3% of macrophages containing nano-TiO2 agglomerates (Table 3). Despite the NMAD values that were below 100 nm for these aerosols, the D75 values ranged from 124 to 305 nm. Thus, the agglomerated (> 100 nm) fraction, which was encountered for 29 to 46% of these aerosols could explain the NP endocytosis observed. Nonetheless, for each primary NP size, a significant difference was observed for the percentage of particle-laden macrophages between LA and SA aerosols. Also, increases in the relative levels of CXCL7 and MIP-3α were observed in all SA aerosols, while it was also the case for the tissue inhibitor matrix proteinase 1 (TIMP-1), a glycoprotein involved in the degradation of the extracellular matrix, for the 10–30 nm SA aerosol. Considering the size distributions of the SA aerosols (D25 = 28 nm to D75 = 305 nm; Table 2) and as shown with the estimation of NP endocytosis, it can be assumed that these aerosols were not as well detected and phagocytized by alveolar macrophages as the LA aerosols. Thus, increased NP interaction with biological materials (lung cells and tissue) may have occurred compared to the LA aerosols and could be expressed as cytotoxicity and oxidative stress effects. In our study, statistically significant increases were observed in LDH activity and 8-isoprostane concentration for the 5 nm SA aerosol compared to the controls and its respective LA aerosol, while only 8-isoprostane was significantly increased for the 10–30 and 50 nm SA aerosols (Figure 6). Therefore, overall, the results for the SA aerosols indicate clear trends of NP interaction with lung cells and tissue through oxidative stress damage and suggestive slight cytotoxic effects (Figure 6).
Effect of the agglomeration state
Overall, these results confirm, at a higher mass concentration, what we had previously shown at 7 mg/m3
, namely that an acute inhalation of nano-TiO2 with two distinct agglomeration states, smaller or larger than 100 nm, induced different mild pulmonary effects. An acute inflammatory response measured by an increase in the number of neutrophils was induced by exposure to two out of three LA (> 100 nm) aerosols, while significant oxidative stress effects were observed after exposures to all of the SA (< 100 nm) aerosols. With respect to hazard identification, our results indicate that even though LA aerosols induced an acute inflammatory response, which is reversible according to the literature [9, 19, 32, 45, 70, 72], it cannot be concluded that these aerosols induce toxicity through the same mechanisms as SA aerosols, which showed clear oxidative stress damage in BALF.
Effect of primary nanoparticle size
For the three initial TiO2 NP sizes, we observed only one significant difference within the smaller than 100 nm agglomeration state aerosols. The significant difference was observed between the 5 nm and the two other SA aerosols for the 8-isoprostane concentration (Figure 6). This suggests that the larger 10–30 and 50 nm particles induced more lipid peroxidation and oxidative stress damage than the smaller 5 nm particles. Numerous inhalation studies have previously demonstrated that translocation of various type of NP, including TiO2, to extrapulmonary compartments occurred for small agglomerated NP (average diameter < 80 nm) in aerosols [37, 76–83]. Collectively, these studies indicate that the penetration efficiency of NP through cellular membranes increases as the NP size decreases and that the translocation time increases with particle size [84–86]. Thus, in our study, the smaller size of the 5 nm particles (D50 = 48 nm in aerosol) would facilitate their possible and rapid translocation from the lung epithelial cells, thereby reducing their availability and time to cause cellular membrane lipid peroxidation at the NP - cell interface. The larger size of the 10–30 and 50 nm particles (D50 = 65 and 85 nm in aerosols, respectively) may on the other hand promote translocation to a lesser extent and over a longer period of time, resulting in increased interaction of NP with the cellular membranes, which generates oxidative stress through membranolytic effects (Figure 6).
We observed that the LDH activity for the SA aerosols compared to controls was only significant for the 5 nm particles (Figure 6). This same aerosol also showed a lower increase in 8-isoprostane concentration than the other two SA aerosols (Figure 6). This could possibly be explained by the higher cytotoxicity response observed in these animals. Indeed, LDH is an enzyme that leaks from damaged cells as a sign of membrane integrity lost  and as previously mentioned, is a suitable marker of cell death, particularly by necrosis. It could also be considered as evidence of NP penetration into cells . NP penetration into cells leading to interactions with intracellular components is size-dependent [85, 86]. Thus, the lower cytotoxicity observed for the larger NP (10–30 and 50 nm) could possibly be due to their less efficient penetration into cells. Interestingly, our data suggest that membrane damage by lipid peroxidation at the NP – cell membrane interface might not be the primary cause of cytotoxicity. Thus, the size-dependent effect of nano-TiO2 observed in our study in the smaller than 100 nm agglomeration state is supported by the literature. Also, these results are in line with Paulhun’s study  that reported that the clearance kinetics of NP was more dependent on their initial particle size.
50 nm TiO2 NP
In addition, for the SA aerosols, there may be a few reasons that explain the lack of cellular and histopathological changes with the 50 nm group compared to the 5 and 10–30 nm groups (Figures 4 and 8). First, considering the NP powder characterization, approximately 20% of the crystal phase of the 50 nm powder was in the rutile form, while it was 3% or less for the two other powders (Table 1). It has already been reported that the rutile form of TiO2 NP is less toxic than the anatase crystal phase [38, 40, 89–92]. Thus, in the less than 100 nm agglomeration state, the presence of the rutile phase in the 50 nm powder may be partly responsible for the lower cellular toxicity observed. At equal mass concentration, as the NP size decreases, the surface area per mass unit increases and leads to high surface to volume ratios, giving smaller NP enhanced surface reactivity . Studies have also shown that the surface adsorption and reactivity of smaller than 10 nm TiO2 NP were enhanced relatively to larger NP [19, 93]. Hence, the size effect of the initial 5 nm particle size (D50 = 48 nm in aerosol), which would be more toxic than the 50 nm particles (D50 = 85 nm in aerosol), may also contribute to the cytological effects observed for the SA aerosols. Also, for these three aerosols, the total particle number concentration was elevated (Table 2). However, the 50 nm SA aerosol had the lowest total particle number concentration by a factor of 2.4 and 1.4 compared to the 5 and 10–30 nm aerosols, respectively. Due to their small size, NP mainly contribute to number concentrations in aerosols and, to a much lesser degree, to mass concentration [94, 95]. For identical masses, a larger number of NP can occupy the same space, and thus, in theory, increase the interactions with biological material [12, 96]. Thus, all of these factors may also contribute to the toxicological results observed for the 50 nm SA aerosol.
Primary NP size-dependant effect
Overall, these results show that within a less than 100 nm agglomeration state, there may be a primary particle size-dependent effect of nano-TiO2. Even though the 10–30 and 50 nm particles induced significantly higher oxidative stress and pro-inflammatory damage than the 5 nm particles, it cannot be directly concluded that these larger TiO2 NP are more toxic. Our data, in line with the current literature, show that the smaller 5 nm particles may potentially pose greater health risks by causing more cytotoxicity through necrosis.
It is noteworthy that a limitation to our study can be attributed to the fact that only data 16 hours after a 6-hour exposure were collected. Therefore, conclusions on prolonged inflammation and cytotoxicity cannot be drawn from these data. However, our results indicate that for an acute exposure the 10–30 nm particles induced significant increases in the total cell count and number of macrophages in both the SA and LA aerosols, while the number of neutrophils was significantly increased in the LA aerosol (Table 3, Figure 4), which also showed the highest fold increases in pro-inflammatory cytokine (Figure 5). Moreover, qualitatively comparing the agglomerate structure of the LA aerosols (Figure 2) we noticed that the 10–30 nm particles agglomerated into loose structures with more void open spaces. The possibility of agglomeration and de-agglomeration of NP in physiological environments still remains an open question [3, 9, 38, 45, 47]. However, if de-agglomeration was to occur once deposited in the lungs, the loose agglomerate structure is thought to be more easily de-agglomerated . Moreover, when considering the geometry of similar agglomerates size, the loose type structure has a higher surface available to interact with biological materials and could hence increase their toxicity compared to the compact agglomerates. Therefore, the NP agglomerates structure may also play a role in toxicity. Overall, of the three NP sizes, the 10–30 nm TiO2 NP seemed to induce the most pronounced pro-inflammatory effects. These results are consistent with Grassian et al.
 inhalation study in mice at 7 mg/m3 where it was concluded, solely based on the inflammatory cell response, that the 21 nm nano-TiO2 particles (139 nm in aerosol) were slightly, but significantly more toxic than the 5 nm ones (120 nm in aerosol). Interestingly, the highest relative deposition efficiency of NP in the alveolar region occurs at approximately 20 nm [17, 18].