Human beings have always been exposed to airborne ultrafine particles (i.e. particles below 100 nm size) from e.g., forest fires, volcanic eruptions or indoor fire places. However, since the industrial revolution, exposures to ultrafine particles have increased dramatically. This is mainly due to the invention of the combustion engines [1]. Engineered nanomaterials and nanotechnologies are expected to have a profound impact on many aspects of society and economy. However, they also represent a new source of human exposures and awareness is growing that their unusual chemical and physical properties may lead to potential environmental and health risks [1,2]. The toxicity and carcinogenicity of low-soluble particles is thought to be exerted primarily through generation of inflammation and oxidative stress [3]. Pulmonary inflammation is central in lung diseases and also thought to be involved in risk of atherothrombosis [4,5]. Concern has been raised over nano-sized particles because the bronchio-alveolar deposition rate is great, they have a large surface area, are more reactive, and the clearance of them is slow. All these factors may contribute to more severe and prolonged inflammation and, consequently, increased risk for disease. Although it has been estimated that less than 1% of the deposited dose translocates from the lung into the circulation, this is suspected to be important in nanoparticle-induced cardiovascular effects, and may also produce adverse hepatic, developmental effects and other effects in organs remote from the lung [6]. Although many laboratories are currently investigating toxicological effects of nano-sized materials, few in vivo studies have been published in the field on nanotoxicology and only with a very limited panels of nanomaterials in the same experimental setup.

For an in vivo comparison study of toxicity we chose a panel of nanoparticles with carbon black, Printex 90 (CB), single walled carbon nanotubes (SWCNT), fullerenes C₆₀ (C₆₀), quantum dots (QDs) and nanosized gold particles. CB is a well-known ingredient in rubber, plastics, inks, and paints with an annual production about 10 million tonnes [7]. The toxic effects of CB have been well described in vitro and in vivo, making it an excellent benchmark material. SWCNT and C₆₀ have the potential to be among the most widely used carbonaceous engineered nanomaterials in the future. Material scientists have envisioned the use of these type of particles in wide range of applications (e.g., composite materials, disease treatment and electronics)[8-10]. QDs are currently applied in biomedical imaging and electronics, but have been suggested for use in computer memories, visual displays, solar cells and lasers [11] as well as a replacement for organic dyes due to superior quantum yield and resistance to photo bleaching [12]. Both QDs and gold particles additionally have properties which make them detectable in translocation and targeting studies [13-17]. The overall goal of this study was to test an expected susceptible animal model (apolipoprotein E knockout mouse, ApoE^-/-) as well as background strain (C57) for direct comparison of markers of inflammation, lung injury and genotoxicity in lung tissue and broncho-alveolar lavage (BAL) fluid following pulmonary exposure across the panel of nanoparticles. The ApoE^-/- mice were chosen since these experiments were part of a larger study which also includes cardiovascular research. However, these animals have previously shown increased permeability of particulates in blood vessels [18-20]. If this permeability is related to the modest elevated blood cholesterol found in ApoE^-/- mice, the model may closer resemble humans with elevated cholesterol levels. Experiments showed that the most sensitive of the tested exposure methods was a combination of intratracheal (i.t.) instillation of nanoparticles in apolipoprotein E knockout mice (ApoE^-/-). The chosen inflammatory markers were mRNA levels of macrophage inflammatory protein-2 (Mip-2), interleukin-6 (Il-6) and macrophages/monocyte chemoattractant protein-1 (Mcp-1) in lung tissue and BAL cell composition. MIP-2 is involved in the chemotactic recruitment of neutrophils to the pulmonary system following exposure [21,22]. MCP-1 is produced by numerous inflammatory cells including epithelium fibroblasts, monocytes and macrophages and is believed to promote the maturation of monocytes to macrophages as well as being a major chemoattractant for monocyte recruitment [23,24]. Il-6 is an important early mediator of inflammation, involved in fever and acute phase responses. It is secreted mainly by macrophages e.g. as a response in particle induced inflammation [25]. We have reported before that all of these markers are elevated at 24 h after inhalation of diesel particles [26]. Additionally, we determined DNA damage in BAL cells by the comet assay as a sensitive marker for particle genotoxicity related mainly to oxidative stress and cellular damage by leakage of protein and lactate dehydrogenase (LDH) to the BAL fluid.

We here present results that show that particle instillation induced a faster and stronger lung inflammatory response in hyperlipidemic ApoE^-/- mice compared to wild-type mice. Instillation produced stronger effects than inhalation. SWCNT, CB, C₆₀ and gold nanoparticles showed inflammatory effects corresponding to their surface areas after instillation, whereas QDs were highly toxic, possibly due to cadmium leakage.

For determining the ranking order of the inflammatory potential of the particles, we intended to evoke a substantial pulmonary inflammation using a susceptible animal model. The exposure dose, 54 μg per mouse, is relatively large, but is well within the range used by others within the field of nanotoxicology [30-32]. Recently, hyperlipidemic ApoE^-/- mice have been used extensively to study the mechanisms involved in the vascular and particularly the atherothrombosis-inducing effects of exposure to both air pollution particles and SWCNT [33,4]. These studies have shown that such exposure accelerates vascular plaque formation [34,33,35] as well as cause hepatic oxidative stress and endothelial dysfunction even in 10–13 weeks old ApoE^-/- mice with less than 1% plaques in the aortic wall [36-38]. Although, ApoE^-/- and C57 mice responded to CB instillation, the response was much stronger in the former group particularly 3 h after exposure. This may indicate that ApoE^-/- are more sensitive or are primed to a response. The elevated susceptibility of the ApoE^-/- mice is important because these mice at young age display a modest dyslipidemia with elevated levels of cholesterol in plasma. Thus, they can be used as a experimental model because a large proportion of the human population in the Western World has elevated levels of lipids in the blood. In this respect it is interesting that the blood vessels in ApoE^-/- mice have increased permeability, and that ApoE^-/- mice have greater leakage of particulates across the blood-brain-barrier [18-20]. Probably the different lipid composition the cell membranes of ApoE^-/- mice affects permeability. It might be that the alveolar-blood barrier of ApoE^-/- mice also is more permeable to nanoparticles, or that inflammatory target cells are more permeable. Nevertheless, the recognition of an increased lung inflammation response is important for the interpretation of enhanced systemic effects in ApoE^-/- mice following pulmonary exposure. Especially because the suggested mechanisms of action for the atherothrombotic effects of nanoparticles include events secondary to pulmonary inflammation [4,5]. At present we are investigating the relationship of the pulmonary effects with cardiovascular effects in ApoE^-/- mice with some of the nanoparticles (data not shown).

In the past, there has been tremendous work done on comparisons of the pulmonary effect elicited by either instillation or inhalation (REFERENCES), whereas such comparisons are sparse in transgenic models representing susceptible human populations. We compared the lung inflammation elicited by instillation and inhalation of CB in presumed similar doses in ApoE^-/-mice. We instilled 18 μg and 54 μg and this is almost identical to the pulmonary deposited doses (17.6 and 52.7 μg) estimated by assuming 33.8% deposition of the mass during a CB mouse inhalation exposure. Despite the apparently similar deposited dose, inhalation of CB caused much less inflammatory response than instillation did. None of the cytokine markers increased significantly following inhalation (up to 2.1-fold), whereas all were significantly increased following instillation (up to 40-fold). The fraction of neutrophils in BAL cells reached 6% following inhalation and 76% following instillation. Only the protein content of BAL fluid increased significantly and in a dose dependent manner following inhalation. However, protein levels were still less than for instilled animals. This difference between i.t. instillation and inhalation exposure is in keeping with the general conclusion from the literature that the clearance of instilled particles from the lung is slower and inflammation is greater (reviewed by [39]).

We have earlier studied pulmonary inflammation after inhalation of CB and diesel exhaust particles [40,26,43]. Inhalation is the "gold standard" for determining the potential toxicity of inhalants. It is the closest to a normal route of entry and the distribution pattern may also correspond more closely to that of a true exposure scenario with particles being deposited through the pulmonary system dependent on their size and shape. Advantages with instillation are the small and very precise amount of test material used and deposited. The benefits and problems of delivering materials to the lung by inhalation and instillation have been discussed in the literature (although available data are almost exclusively based on rats) [39,44,45]. Generally, intratracheal instillation is a well accepted procedure which usually well reproduces the effects of inhalation. However, the bolus administration by instillation may produce less homogeneous distribution of the material, with more focal exposure than following inhalation. Instillation also forces material into the alveoli resulting in lesser deposition in bronchia or bronchioles and may overwhelm mucociliary clearence.

By the cytokine mRNA levels, the inflammatory response in lung tissue was increased 52- to 195-fold following 3 h and 7- to 30-fold following 24 h after SWCNT instillations compared to controls. This indicates that the inflammation after SWCNT exposure is very strong, but that the primary inflammatory signalling ceases earlier, as has been suggested before [46,47,32]. CB also elicited a rapid (3 h) strong inflammation, although not as strong as SWCNT did. Interestingly, all markers of inflammation in lung tissue and BAL fluid continued to increase from 3 h to 24 h after the instillation. Li and co-workers found an influx of neutrophils following 6 and 24 h of more than 40-fold following a single 125-μg instillation dose in rats of CB. The neutrophil count was still elevated more than 10-fold after 7 days. Tumor necrosis factor-α activity measured ex vivo in BAL leukocytes from the above mentioned CB exposed rats continued to increase through the 7 days [48], suggesting a longer inflammation than SWCNT. C₆₀ overall caused much less inflammation, with only Mcp-1 increasing significantly after 3 h (2-fold). The mRNA levels increased from 3 to 24 h with 4- to 6-fold significant elevation at 24 h. In all but one study [49] C₆₀ was not toxic in rodents [50-55], even following intra-peritoneal delivery of up to 5 g/kg [52]. Although C₆₀ was the least toxic of the carbonaceous particles, there may be a rapid distribution across membranes if agglomerates disintegrate to the small primary size [56]. In general, instillation of gold particles caused a very low and transient inflammatory response by detectable increases in Mip-2 and Mcp-1. It should be noted that the mass of instilled gold particles was very low (0.54 μg) and it is not possible to estimate the effect of an instillation of 54 μg. However, because of the very small size of these particles, gold still had a high number concentration in the instilled fluid (0.675 × 10¹³/50 μl).

It has been shown previously that the inflammatory response of low toxicity-low solubility particles is proportional to the surface area of the instilled particles rather than the mass [57-60]. We have previously determined surface area of SWCNT, CB and C₆₀ by the BET N₂ adsorption method to be 731, 338 and <20 m²/g [61], indicating that we have instilled 395, 183 and 11 cm² of these three particles, respectively. The BET surface area of the C₆₀ sample has since been corrected to 0.4 m²/g using a custom made single point BET modified for low surface areas (Personal communication Giulio Pojana) indicating that we have instilled 0.2 cm². We also calculated the surface area of the instilled gold to be 0.85 cm². The relationship between the surface area of the three carbonaceous particles (also 18 μg CB at 24 h), gold and controls and induction of cytokine mRNA and increased neutrophil fraction is illustrated in Figure 3. There was a good correlation between mRNA induction and fraction neutrophils at 3 h and the surface area instilled (r² between 0.84 and 1). However, the mRNA levels decreased between 3 h and 24 h after SWCNT instillation and there was a smaller fraction of neutrophils compared to CB instilled mice. Thus, SWCNT was less inflammatory following 24 h than anticipated from surface area alone. This means that r²-values were below 0.60 for all inflammatory end-points. It should be noted that nanoparticles were delivered as agglomerates and the factual instilled surface area may therefore be lower than estimated from BET. But that it has also been suggested that action of surfactants and proteins in the lung may reduce van der Waals interactions between nanoparticles and disperse these agglomerates [46].

Recently, it was suggested that there is threshold for inflammatory effects of low-toxicity, low-solubility particles at 1 cm² deposited particle surface area/cm² epithelial surface in the proximal alveolar region of the lung [62]. If the particle burden within a day is concentrated to the proximal alveolar region, the effect on interleukin expression after 24 h (but not after 3 h) might be interpreted as a threshold at between 1 and 30 cm² particle surface area per mouse (figure 3D). If we assume that the mice had lung has a epithelial surface area of about 600 cm² and 5% of this was in the proximal alveolar region [62] figure 3D might be interpreted as showing a threshold between 0,03 and 1 cm² deposited particle surface area/cm² epithelial surface area in the proximal alveoli, which would be in concordance with the theory of Donaldson et al. However, more data are needed to confirm this.

Numerous physicochemical parameters have been suggested to influence the inflammatory effects nanoparticles, including agglomeration state, shape, composition, surface reactivity, radical formation capacity and more [63]. SWCNT have a needle like shape and strong tendency to agglomerate. These parameters may distinguish SWCNT from C₆₀ and CB and provide a possible explanation for the detected differences following 24 h. On the other hand, we have also previously found that the CB employed here caused more reactive oxygen species (ROS) formation than the SWCNT in cellular and cell-free system, whereas C₆₀ generated far less ROS than the other particles [61]. Additionally, recent evidence may suggest that SWCNT produces long term toxicity and may be different from low toxicity particles [30,46,47,64,32]. The BAL protein at 24 h showed the strongest correlation with surface area (r² = 0.98 and 0.86 at 3 h)(not shown).

Since the QDs required thioglycolic acid as stabilizer, we exposed mice in the control group to a vehicle solution that contained this chemical. It is evident from the results that adding thioglycolic acid to the vehicle causes an inflammatory response with 2- to 6.5-fold increased mRNA cytokine levels after 3 h and 24 h, where the fraction of neutrophils was increased 4-fold. The pulmonary inflammation was quite severe following the instillation of QDs with increases of 25–250-fold of the cytokine mRNA levels over the vehicle control. Almost all cells (93–97%) in the BAL fluid were neutrophils at 24 h. However, there was a remarkably similar magnitude of the inflammatory response of both negatively and positively charged QDs. We are not aware of other in vivo studies on cytokine induction by QD exposure. However it has been shown that several different cytokines, including IL-6 and MCP-1 are induced 2- to 3-fold in human epidermal keratinocytes and rat mesenchymal stem cells following QD exposure [65,66]. The CdSe QDs used with the cells were coated with ZnS shell, which probably decreases the toxicity. Similarly mice have been injected i.v. with CdSe QDs, but no noticeable adverse effects or signs of necrosis at sites of deposition were reported [13,67]. It therefore appears likely that cadmium was leaking from the QDs that we tested. Indeed, Pearson and co-workers measured BAL protein following instillation of different concentrations of Cd²⁺ in Male CF-1 mice. If we expect effects of the same magnitude this would mean that 18–26% of the cadmium in the QD620 and QD621 delivered to the mice might have been dissolved [68]. However, this will be overestimated if intact QDs or free telluride also show toxicity.

We determined the level of DNA damage by the comet assay in BAL cells 3 h following pulmonary exposure of all particles. At this time point the fraction of neutrophils was small and the BAL fluid primarily contained macrophages. The DNA damage increased significantly following exposure to SWCNT, CB and QDs, but only marginally with C₆₀ and gold nanoparticles. QDs were also much more genotoxic than the other particles with a more than 3-fold increase in the level of DNA damage. Even when compared to the genotoxic carbonaceous particles (SWCNT and CB) we detected about 2-fold more DNA damage in the BAL cells. This may be a result of severe oxidative stress induced by cadmium leaked from the QDs. Indeed cadmium has been shown to induce oxidative stress with DNA damage and impair repair in several cell types [69]. The comet assay is considered to be a sensitive assay for detection of the genotoxic effect of particulate matter as has been shown by several studies on air pollution particles [70,43]. DNA damage is an important mechanism in carcinogenesis but the quantitative relationship of DNA damage and the development of cancer are not clear and it is probably different for different chemicals and for different cell populations. In this study we investigated DNA damage in BAL cells and not lung tissue, where lung disease would normally arise. We believe that the effects observed in BAL cells are similar in the epithelial cell of the lung surface because they are located closely together. However, detecting effects in lung tissue is difficult because only a fraction of cells in the lung tissue is exposed and the effects are diluted within the cell population. We have previously shown that CB exposure increased the level of strand breaks as well as mutation frequency in a mouse lung epithelial cell line [71]. We suspect a high ROS production to be the cause of these effects as repeatedly suggested [72-74].

Here we report that ApoE^-/- mice is a sensitive model for comparing inflammatory potential of (nano) particle instillation. CB and SWCNT caused more inflammation and DNA and cell damage than C₆₀ did. This inflammatory signalling appeared shorter for SWCNT than for CB and C₆₀. The instilled surface area of low toxicity low solubility particles appears to be a good predictor for inflammatory response in vivo. QDs were the most toxic particle likely because of a Cd effect. Gold particles, at a lower dose than the other particles, did not induce an inflammatory or toxic response in the mice.

BAL: Broncho-alveolar lavage; C₆₀: fullerenes C₆₀; CB: carbon black; Il-6: interleukin-6; LDH: lactate dehydrogenase; Mip-2: macrophage inflammatory protein-2; Mcp-1: macrophages/monocyte chemoattractant protein-1; ROS: reactive oxygen species; SWCNT: single walled carbon nanotubes; i.t.: intratracheal; QD: quantum dot

The authors declare that they have no competing interests.

All authors contributed to the idea and design of the study. NRJ carried out all exposures, toxicological analysis and drafted the manuscript. KAJ analysed particle size distribution and developed the deposition model. HWA and PM participated substantially in the inhalation exposure and the statistical analysis, respectively. OL conducted the histopathological analysis of organs following QD and vehicle exposure. All authors contributed, read and approved the final manuscript.